diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7971 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.988118811881188,
+  "eval_steps": 500,
+  "global_step": 1134,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005280528052805281,
+      "grad_norm": 7.677456378936768,
+      "learning_rate": 4.385964912280702e-07,
+      "loss": 0.9639,
+      "step": 1
+    },
+    {
+      "epoch": 0.010561056105610561,
+      "grad_norm": 6.208090305328369,
+      "learning_rate": 8.771929824561404e-07,
+      "loss": 0.8832,
+      "step": 2
+    },
+    {
+      "epoch": 0.015841584158415842,
+      "grad_norm": 7.966090202331543,
+      "learning_rate": 1.3157894736842106e-06,
+      "loss": 0.9749,
+      "step": 3
+    },
+    {
+      "epoch": 0.021122112211221122,
+      "grad_norm": 7.089982986450195,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 0.9166,
+      "step": 4
+    },
+    {
+      "epoch": 0.026402640264026403,
+      "grad_norm": 9.171483039855957,
+      "learning_rate": 2.1929824561403507e-06,
+      "loss": 1.02,
+      "step": 5
+    },
+    {
+      "epoch": 0.031683168316831684,
+      "grad_norm": 7.198622703552246,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 0.9205,
+      "step": 6
+    },
+    {
+      "epoch": 0.036963696369636964,
+      "grad_norm": 6.673768997192383,
+      "learning_rate": 3.070175438596491e-06,
+      "loss": 0.9179,
+      "step": 7
+    },
+    {
+      "epoch": 0.042244224422442245,
+      "grad_norm": 5.332284450531006,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 0.7845,
+      "step": 8
+    },
+    {
+      "epoch": 0.047524752475247525,
+      "grad_norm": 5.0864458084106445,
+      "learning_rate": 3.9473684210526315e-06,
+      "loss": 0.7676,
+      "step": 9
+    },
+    {
+      "epoch": 0.052805280528052806,
+      "grad_norm": 4.446084022521973,
+      "learning_rate": 4.3859649122807014e-06,
+      "loss": 0.6778,
+      "step": 10
+    },
+    {
+      "epoch": 0.058085808580858087,
+      "grad_norm": 4.680896759033203,
+      "learning_rate": 4.824561403508772e-06,
+      "loss": 0.6388,
+      "step": 11
+    },
+    {
+      "epoch": 0.06336633663366337,
+      "grad_norm": 3.745345115661621,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.7111,
+      "step": 12
+    },
+    {
+      "epoch": 0.06864686468646865,
+      "grad_norm": 3.1842403411865234,
+      "learning_rate": 5.701754385964912e-06,
+      "loss": 0.6327,
+      "step": 13
+    },
+    {
+      "epoch": 0.07392739273927393,
+      "grad_norm": 2.8893682956695557,
+      "learning_rate": 6.140350877192982e-06,
+      "loss": 0.6334,
+      "step": 14
+    },
+    {
+      "epoch": 0.07920792079207921,
+      "grad_norm": 2.5611019134521484,
+      "learning_rate": 6.578947368421053e-06,
+      "loss": 0.6284,
+      "step": 15
+    },
+    {
+      "epoch": 0.08448844884488449,
+      "grad_norm": 2.8038105964660645,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 0.6224,
+      "step": 16
+    },
+    {
+      "epoch": 0.08976897689768977,
+      "grad_norm": 2.51741099357605,
+      "learning_rate": 7.456140350877193e-06,
+      "loss": 0.6035,
+      "step": 17
+    },
+    {
+      "epoch": 0.09504950495049505,
+      "grad_norm": 2.309906482696533,
+      "learning_rate": 7.894736842105263e-06,
+      "loss": 0.4823,
+      "step": 18
+    },
+    {
+      "epoch": 0.10033003300330033,
+      "grad_norm": 2.719104528427124,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.5194,
+      "step": 19
+    },
+    {
+      "epoch": 0.10561056105610561,
+      "grad_norm": 2.3248131275177,
+      "learning_rate": 8.771929824561403e-06,
+      "loss": 0.5342,
+      "step": 20
+    },
+    {
+      "epoch": 0.11089108910891089,
+      "grad_norm": 2.2312052249908447,
+      "learning_rate": 9.210526315789474e-06,
+      "loss": 0.5275,
+      "step": 21
+    },
+    {
+      "epoch": 0.11617161716171617,
+      "grad_norm": 2.045426368713379,
+      "learning_rate": 9.649122807017545e-06,
+      "loss": 0.5499,
+      "step": 22
+    },
+    {
+      "epoch": 0.12145214521452145,
+      "grad_norm": 2.17621111869812,
+      "learning_rate": 1.0087719298245614e-05,
+      "loss": 0.5891,
+      "step": 23
+    },
+    {
+      "epoch": 0.12673267326732673,
+      "grad_norm": 2.363523244857788,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.667,
+      "step": 24
+    },
+    {
+      "epoch": 0.132013201320132,
+      "grad_norm": 2.3387694358825684,
+      "learning_rate": 1.0964912280701754e-05,
+      "loss": 0.5842,
+      "step": 25
+    },
+    {
+      "epoch": 0.1372937293729373,
+      "grad_norm": 2.1026556491851807,
+      "learning_rate": 1.1403508771929824e-05,
+      "loss": 0.6199,
+      "step": 26
+    },
+    {
+      "epoch": 0.14257425742574256,
+      "grad_norm": 2.3579792976379395,
+      "learning_rate": 1.1842105263157895e-05,
+      "loss": 0.5985,
+      "step": 27
+    },
+    {
+      "epoch": 0.14785478547854786,
+      "grad_norm": 2.136988639831543,
+      "learning_rate": 1.2280701754385964e-05,
+      "loss": 0.5759,
+      "step": 28
+    },
+    {
+      "epoch": 0.15313531353135312,
+      "grad_norm": 1.9653126001358032,
+      "learning_rate": 1.2719298245614037e-05,
+      "loss": 0.5132,
+      "step": 29
+    },
+    {
+      "epoch": 0.15841584158415842,
+      "grad_norm": 2.0811052322387695,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 0.514,
+      "step": 30
+    },
+    {
+      "epoch": 0.16369636963696368,
+      "grad_norm": 2.0110175609588623,
+      "learning_rate": 1.3596491228070177e-05,
+      "loss": 0.588,
+      "step": 31
+    },
+    {
+      "epoch": 0.16897689768976898,
+      "grad_norm": 2.020909070968628,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.5896,
+      "step": 32
+    },
+    {
+      "epoch": 0.17425742574257425,
+      "grad_norm": 2.001084804534912,
+      "learning_rate": 1.4473684210526317e-05,
+      "loss": 0.5028,
+      "step": 33
+    },
+    {
+      "epoch": 0.17953795379537954,
+      "grad_norm": 1.9398471117019653,
+      "learning_rate": 1.4912280701754386e-05,
+      "loss": 0.5199,
+      "step": 34
+    },
+    {
+      "epoch": 0.1848184818481848,
+      "grad_norm": 2.3714287281036377,
+      "learning_rate": 1.5350877192982457e-05,
+      "loss": 0.6112,
+      "step": 35
+    },
+    {
+      "epoch": 0.1900990099009901,
+      "grad_norm": 2.054084062576294,
+      "learning_rate": 1.5789473684210526e-05,
+      "loss": 0.4697,
+      "step": 36
+    },
+    {
+      "epoch": 0.19537953795379537,
+      "grad_norm": 2.1482019424438477,
+      "learning_rate": 1.62280701754386e-05,
+      "loss": 0.442,
+      "step": 37
+    },
+    {
+      "epoch": 0.20066006600660066,
+      "grad_norm": 2.005889892578125,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4276,
+      "step": 38
+    },
+    {
+      "epoch": 0.20594059405940593,
+      "grad_norm": 2.1889655590057373,
+      "learning_rate": 1.7105263157894737e-05,
+      "loss": 0.5114,
+      "step": 39
+    },
+    {
+      "epoch": 0.21122112211221122,
+      "grad_norm": 1.9033912420272827,
+      "learning_rate": 1.7543859649122806e-05,
+      "loss": 0.5266,
+      "step": 40
+    },
+    {
+      "epoch": 0.2165016501650165,
+      "grad_norm": 2.01960825920105,
+      "learning_rate": 1.7982456140350878e-05,
+      "loss": 0.485,
+      "step": 41
+    },
+    {
+      "epoch": 0.22178217821782178,
+      "grad_norm": 2.0285496711730957,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.4915,
+      "step": 42
+    },
+    {
+      "epoch": 0.22706270627062705,
+      "grad_norm": 1.9628126621246338,
+      "learning_rate": 1.885964912280702e-05,
+      "loss": 0.4059,
+      "step": 43
+    },
+    {
+      "epoch": 0.23234323432343235,
+      "grad_norm": 2.2826972007751465,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.5476,
+      "step": 44
+    },
+    {
+      "epoch": 0.2376237623762376,
+      "grad_norm": 2.3612606525421143,
+      "learning_rate": 1.9736842105263158e-05,
+      "loss": 0.5466,
+      "step": 45
+    },
+    {
+      "epoch": 0.2429042904290429,
+      "grad_norm": 1.8648459911346436,
+      "learning_rate": 2.0175438596491227e-05,
+      "loss": 0.505,
+      "step": 46
+    },
+    {
+      "epoch": 0.24818481848184817,
+      "grad_norm": 1.9400116205215454,
+      "learning_rate": 2.06140350877193e-05,
+      "loss": 0.4488,
+      "step": 47
+    },
+    {
+      "epoch": 0.25346534653465347,
+      "grad_norm": 1.8791626691818237,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.4311,
+      "step": 48
+    },
+    {
+      "epoch": 0.25874587458745874,
+      "grad_norm": 1.8579607009887695,
+      "learning_rate": 2.149122807017544e-05,
+      "loss": 0.4188,
+      "step": 49
+    },
+    {
+      "epoch": 0.264026402640264,
+      "grad_norm": 1.8994585275650024,
+      "learning_rate": 2.1929824561403507e-05,
+      "loss": 0.4455,
+      "step": 50
+    },
+    {
+      "epoch": 0.2693069306930693,
+      "grad_norm": 1.9908592700958252,
+      "learning_rate": 2.236842105263158e-05,
+      "loss": 0.4684,
+      "step": 51
+    },
+    {
+      "epoch": 0.2745874587458746,
+      "grad_norm": 2.282810926437378,
+      "learning_rate": 2.280701754385965e-05,
+      "loss": 0.6328,
+      "step": 52
+    },
+    {
+      "epoch": 0.27986798679867986,
+      "grad_norm": 2.017083168029785,
+      "learning_rate": 2.324561403508772e-05,
+      "loss": 0.4561,
+      "step": 53
+    },
+    {
+      "epoch": 0.2851485148514851,
+      "grad_norm": 2.0290071964263916,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.4749,
+      "step": 54
+    },
+    {
+      "epoch": 0.29042904290429045,
+      "grad_norm": 2.1724143028259277,
+      "learning_rate": 2.412280701754386e-05,
+      "loss": 0.6216,
+      "step": 55
+    },
+    {
+      "epoch": 0.2957095709570957,
+      "grad_norm": 1.8445512056350708,
+      "learning_rate": 2.456140350877193e-05,
+      "loss": 0.4473,
+      "step": 56
+    },
+    {
+      "epoch": 0.300990099009901,
+      "grad_norm": 1.9536579847335815,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4746,
+      "step": 57
+    },
+    {
+      "epoch": 0.30627062706270625,
+      "grad_norm": 2.0544443130493164,
+      "learning_rate": 2.5438596491228074e-05,
+      "loss": 0.4069,
+      "step": 58
+    },
+    {
+      "epoch": 0.31155115511551157,
+      "grad_norm": 2.315953016281128,
+      "learning_rate": 2.5877192982456143e-05,
+      "loss": 0.4296,
+      "step": 59
+    },
+    {
+      "epoch": 0.31683168316831684,
+      "grad_norm": 2.234273672103882,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.4941,
+      "step": 60
+    },
+    {
+      "epoch": 0.3221122112211221,
+      "grad_norm": 2.1946239471435547,
+      "learning_rate": 2.675438596491228e-05,
+      "loss": 0.4416,
+      "step": 61
+    },
+    {
+      "epoch": 0.32739273927392737,
+      "grad_norm": 2.0075912475585938,
+      "learning_rate": 2.7192982456140354e-05,
+      "loss": 0.4202,
+      "step": 62
+    },
+    {
+      "epoch": 0.3326732673267327,
+      "grad_norm": 4.797417640686035,
+      "learning_rate": 2.7631578947368426e-05,
+      "loss": 0.452,
+      "step": 63
+    },
+    {
+      "epoch": 0.33795379537953796,
+      "grad_norm": 3.3362960815429688,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.5405,
+      "step": 64
+    },
+    {
+      "epoch": 0.3432343234323432,
+      "grad_norm": 1.9122159481048584,
+      "learning_rate": 2.850877192982456e-05,
+      "loss": 0.3242,
+      "step": 65
+    },
+    {
+      "epoch": 0.3485148514851485,
+      "grad_norm": 1.8941394090652466,
+      "learning_rate": 2.8947368421052634e-05,
+      "loss": 0.4896,
+      "step": 66
+    },
+    {
+      "epoch": 0.3537953795379538,
+      "grad_norm": 1.8021477460861206,
+      "learning_rate": 2.9385964912280706e-05,
+      "loss": 0.3874,
+      "step": 67
+    },
+    {
+      "epoch": 0.3590759075907591,
+      "grad_norm": 2.036555767059326,
+      "learning_rate": 2.9824561403508772e-05,
+      "loss": 0.4886,
+      "step": 68
+    },
+    {
+      "epoch": 0.36435643564356435,
+      "grad_norm": 2.1943323612213135,
+      "learning_rate": 3.0263157894736844e-05,
+      "loss": 0.421,
+      "step": 69
+    },
+    {
+      "epoch": 0.3696369636963696,
+      "grad_norm": 2.077173948287964,
+      "learning_rate": 3.0701754385964913e-05,
+      "loss": 0.5161,
+      "step": 70
+    },
+    {
+      "epoch": 0.37491749174917494,
+      "grad_norm": 2.0704095363616943,
+      "learning_rate": 3.1140350877192986e-05,
+      "loss": 0.4241,
+      "step": 71
+    },
+    {
+      "epoch": 0.3801980198019802,
+      "grad_norm": 2.082000970840454,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.4373,
+      "step": 72
+    },
+    {
+      "epoch": 0.38547854785478547,
+      "grad_norm": 1.8969218730926514,
+      "learning_rate": 3.2017543859649124e-05,
+      "loss": 0.4705,
+      "step": 73
+    },
+    {
+      "epoch": 0.39075907590759074,
+      "grad_norm": 2.206298351287842,
+      "learning_rate": 3.24561403508772e-05,
+      "loss": 0.4938,
+      "step": 74
+    },
+    {
+      "epoch": 0.39603960396039606,
+      "grad_norm": 2.0572750568389893,
+      "learning_rate": 3.289473684210527e-05,
+      "loss": 0.5007,
+      "step": 75
+    },
+    {
+      "epoch": 0.4013201320132013,
+      "grad_norm": 1.94302237033844,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.3689,
+      "step": 76
+    },
+    {
+      "epoch": 0.4066006600660066,
+      "grad_norm": 2.0126149654388428,
+      "learning_rate": 3.377192982456141e-05,
+      "loss": 0.4881,
+      "step": 77
+    },
+    {
+      "epoch": 0.41188118811881186,
+      "grad_norm": 1.887984037399292,
+      "learning_rate": 3.421052631578947e-05,
+      "loss": 0.4761,
+      "step": 78
+    },
+    {
+      "epoch": 0.4171617161716172,
+      "grad_norm": 1.9020264148712158,
+      "learning_rate": 3.4649122807017546e-05,
+      "loss": 0.4343,
+      "step": 79
+    },
+    {
+      "epoch": 0.42244224422442245,
+      "grad_norm": 1.942435622215271,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 0.4563,
+      "step": 80
+    },
+    {
+      "epoch": 0.4277227722772277,
+      "grad_norm": 2.269737482070923,
+      "learning_rate": 3.5526315789473684e-05,
+      "loss": 0.4508,
+      "step": 81
+    },
+    {
+      "epoch": 0.433003300330033,
+      "grad_norm": 2.0216665267944336,
+      "learning_rate": 3.5964912280701756e-05,
+      "loss": 0.4971,
+      "step": 82
+    },
+    {
+      "epoch": 0.4382838283828383,
+      "grad_norm": 2.1765635013580322,
+      "learning_rate": 3.640350877192983e-05,
+      "loss": 0.452,
+      "step": 83
+    },
+    {
+      "epoch": 0.44356435643564357,
+      "grad_norm": 2.25856876373291,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.4824,
+      "step": 84
+    },
+    {
+      "epoch": 0.44884488448844884,
+      "grad_norm": 2.2144601345062256,
+      "learning_rate": 3.728070175438597e-05,
+      "loss": 0.4479,
+      "step": 85
+    },
+    {
+      "epoch": 0.4541254125412541,
+      "grad_norm": 1.97480309009552,
+      "learning_rate": 3.771929824561404e-05,
+      "loss": 0.3071,
+      "step": 86
+    },
+    {
+      "epoch": 0.4594059405940594,
+      "grad_norm": 1.9352009296417236,
+      "learning_rate": 3.815789473684211e-05,
+      "loss": 0.4513,
+      "step": 87
+    },
+    {
+      "epoch": 0.4646864686468647,
+      "grad_norm": 2.055535316467285,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 0.4508,
+      "step": 88
+    },
+    {
+      "epoch": 0.46996699669966996,
+      "grad_norm": 1.93705415725708,
+      "learning_rate": 3.9035087719298244e-05,
+      "loss": 0.4083,
+      "step": 89
+    },
+    {
+      "epoch": 0.4752475247524752,
+      "grad_norm": 1.9412288665771484,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.3715,
+      "step": 90
+    },
+    {
+      "epoch": 0.48052805280528055,
+      "grad_norm": 2.098421335220337,
+      "learning_rate": 3.991228070175439e-05,
+      "loss": 0.414,
+      "step": 91
+    },
+    {
+      "epoch": 0.4858085808580858,
+      "grad_norm": 2.2177186012268066,
+      "learning_rate": 4.0350877192982455e-05,
+      "loss": 0.4736,
+      "step": 92
+    },
+    {
+      "epoch": 0.4910891089108911,
+      "grad_norm": 2.056107759475708,
+      "learning_rate": 4.078947368421053e-05,
+      "loss": 0.4177,
+      "step": 93
+    },
+    {
+      "epoch": 0.49636963696369635,
+      "grad_norm": 1.9581352472305298,
+      "learning_rate": 4.12280701754386e-05,
+      "loss": 0.3688,
+      "step": 94
+    },
+    {
+      "epoch": 0.5016501650165016,
+      "grad_norm": 2.2061662673950195,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4705,
+      "step": 95
+    },
+    {
+      "epoch": 0.5069306930693069,
+      "grad_norm": 1.7467211484909058,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.3303,
+      "step": 96
+    },
+    {
+      "epoch": 0.5122112211221123,
+      "grad_norm": 1.9702417850494385,
+      "learning_rate": 4.254385964912281e-05,
+      "loss": 0.5031,
+      "step": 97
+    },
+    {
+      "epoch": 0.5174917491749175,
+      "grad_norm": 2.079378604888916,
+      "learning_rate": 4.298245614035088e-05,
+      "loss": 0.456,
+      "step": 98
+    },
+    {
+      "epoch": 0.5227722772277228,
+      "grad_norm": 1.8181231021881104,
+      "learning_rate": 4.342105263157895e-05,
+      "loss": 0.4231,
+      "step": 99
+    },
+    {
+      "epoch": 0.528052805280528,
+      "grad_norm": 2.1575117111206055,
+      "learning_rate": 4.3859649122807014e-05,
+      "loss": 0.457,
+      "step": 100
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 2.3540713787078857,
+      "learning_rate": 4.429824561403509e-05,
+      "loss": 0.397,
+      "step": 101
+    },
+    {
+      "epoch": 0.5386138613861386,
+      "grad_norm": 2.3277106285095215,
+      "learning_rate": 4.473684210526316e-05,
+      "loss": 0.474,
+      "step": 102
+    },
+    {
+      "epoch": 0.5438943894389439,
+      "grad_norm": 2.0837771892547607,
+      "learning_rate": 4.517543859649123e-05,
+      "loss": 0.4911,
+      "step": 103
+    },
+    {
+      "epoch": 0.5491749174917492,
+      "grad_norm": 2.315387487411499,
+      "learning_rate": 4.56140350877193e-05,
+      "loss": 0.5138,
+      "step": 104
+    },
+    {
+      "epoch": 0.5544554455445545,
+      "grad_norm": 1.9372241497039795,
+      "learning_rate": 4.605263157894737e-05,
+      "loss": 0.402,
+      "step": 105
+    },
+    {
+      "epoch": 0.5597359735973597,
+      "grad_norm": 2.0722286701202393,
+      "learning_rate": 4.649122807017544e-05,
+      "loss": 0.3484,
+      "step": 106
+    },
+    {
+      "epoch": 0.565016501650165,
+      "grad_norm": 1.8825434446334839,
+      "learning_rate": 4.6929824561403515e-05,
+      "loss": 0.4057,
+      "step": 107
+    },
+    {
+      "epoch": 0.5702970297029702,
+      "grad_norm": 2.0918331146240234,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 0.4073,
+      "step": 108
+    },
+    {
+      "epoch": 0.5755775577557756,
+      "grad_norm": 2.246974468231201,
+      "learning_rate": 4.780701754385965e-05,
+      "loss": 0.4104,
+      "step": 109
+    },
+    {
+      "epoch": 0.5808580858085809,
+      "grad_norm": 1.8505111932754517,
+      "learning_rate": 4.824561403508772e-05,
+      "loss": 0.351,
+      "step": 110
+    },
+    {
+      "epoch": 0.5861386138613861,
+      "grad_norm": 2.2233192920684814,
+      "learning_rate": 4.868421052631579e-05,
+      "loss": 0.4916,
+      "step": 111
+    },
+    {
+      "epoch": 0.5914191419141914,
+      "grad_norm": 2.3233530521392822,
+      "learning_rate": 4.912280701754386e-05,
+      "loss": 0.4886,
+      "step": 112
+    },
+    {
+      "epoch": 0.5966996699669967,
+      "grad_norm": 2.298288345336914,
+      "learning_rate": 4.956140350877193e-05,
+      "loss": 0.497,
+      "step": 113
+    },
+    {
+      "epoch": 0.601980198019802,
+      "grad_norm": 1.9848483800888062,
+      "learning_rate": 5e-05,
+      "loss": 0.3982,
+      "step": 114
+    },
+    {
+      "epoch": 0.6072607260726073,
+      "grad_norm": 1.7758945226669312,
+      "learning_rate": 4.995098039215686e-05,
+      "loss": 0.4284,
+      "step": 115
+    },
+    {
+      "epoch": 0.6125412541254125,
+      "grad_norm": 2.1073226928710938,
+      "learning_rate": 4.990196078431373e-05,
+      "loss": 0.4741,
+      "step": 116
+    },
+    {
+      "epoch": 0.6178217821782178,
+      "grad_norm": 2.1385958194732666,
+      "learning_rate": 4.985294117647059e-05,
+      "loss": 0.3853,
+      "step": 117
+    },
+    {
+      "epoch": 0.6231023102310231,
+      "grad_norm": 2.053973436355591,
+      "learning_rate": 4.980392156862745e-05,
+      "loss": 0.4385,
+      "step": 118
+    },
+    {
+      "epoch": 0.6283828382838283,
+      "grad_norm": 1.7011091709136963,
+      "learning_rate": 4.975490196078432e-05,
+      "loss": 0.3604,
+      "step": 119
+    },
+    {
+      "epoch": 0.6336633663366337,
+      "grad_norm": 2.1312694549560547,
+      "learning_rate": 4.970588235294118e-05,
+      "loss": 0.4636,
+      "step": 120
+    },
+    {
+      "epoch": 0.638943894389439,
+      "grad_norm": 1.9020744562149048,
+      "learning_rate": 4.9656862745098046e-05,
+      "loss": 0.3795,
+      "step": 121
+    },
+    {
+      "epoch": 0.6442244224422442,
+      "grad_norm": 1.9798043966293335,
+      "learning_rate": 4.960784313725491e-05,
+      "loss": 0.382,
+      "step": 122
+    },
+    {
+      "epoch": 0.6495049504950495,
+      "grad_norm": 1.8981302976608276,
+      "learning_rate": 4.955882352941177e-05,
+      "loss": 0.4038,
+      "step": 123
+    },
+    {
+      "epoch": 0.6547854785478547,
+      "grad_norm": 1.9499566555023193,
+      "learning_rate": 4.9509803921568634e-05,
+      "loss": 0.434,
+      "step": 124
+    },
+    {
+      "epoch": 0.6600660066006601,
+      "grad_norm": 1.912457823753357,
+      "learning_rate": 4.9460784313725495e-05,
+      "loss": 0.3187,
+      "step": 125
+    },
+    {
+      "epoch": 0.6653465346534654,
+      "grad_norm": 2.0483133792877197,
+      "learning_rate": 4.9411764705882355e-05,
+      "loss": 0.3916,
+      "step": 126
+    },
+    {
+      "epoch": 0.6706270627062706,
+      "grad_norm": 1.924310564994812,
+      "learning_rate": 4.936274509803922e-05,
+      "loss": 0.378,
+      "step": 127
+    },
+    {
+      "epoch": 0.6759075907590759,
+      "grad_norm": 2.1889538764953613,
+      "learning_rate": 4.931372549019608e-05,
+      "loss": 0.4093,
+      "step": 128
+    },
+    {
+      "epoch": 0.6811881188118812,
+      "grad_norm": 1.8973898887634277,
+      "learning_rate": 4.9264705882352944e-05,
+      "loss": 0.3571,
+      "step": 129
+    },
+    {
+      "epoch": 0.6864686468646864,
+      "grad_norm": 2.0125250816345215,
+      "learning_rate": 4.9215686274509804e-05,
+      "loss": 0.359,
+      "step": 130
+    },
+    {
+      "epoch": 0.6917491749174918,
+      "grad_norm": 1.9622538089752197,
+      "learning_rate": 4.9166666666666665e-05,
+      "loss": 0.3238,
+      "step": 131
+    },
+    {
+      "epoch": 0.697029702970297,
+      "grad_norm": 1.7294894456863403,
+      "learning_rate": 4.911764705882353e-05,
+      "loss": 0.3555,
+      "step": 132
+    },
+    {
+      "epoch": 0.7023102310231023,
+      "grad_norm": 2.0299930572509766,
+      "learning_rate": 4.906862745098039e-05,
+      "loss": 0.4667,
+      "step": 133
+    },
+    {
+      "epoch": 0.7075907590759076,
+      "grad_norm": 1.813370704650879,
+      "learning_rate": 4.901960784313725e-05,
+      "loss": 0.3004,
+      "step": 134
+    },
+    {
+      "epoch": 0.7128712871287128,
+      "grad_norm": 2.090129852294922,
+      "learning_rate": 4.897058823529412e-05,
+      "loss": 0.3845,
+      "step": 135
+    },
+    {
+      "epoch": 0.7181518151815182,
+      "grad_norm": 2.3778114318847656,
+      "learning_rate": 4.892156862745098e-05,
+      "loss": 0.4731,
+      "step": 136
+    },
+    {
+      "epoch": 0.7234323432343235,
+      "grad_norm": 2.1827681064605713,
+      "learning_rate": 4.887254901960784e-05,
+      "loss": 0.4078,
+      "step": 137
+    },
+    {
+      "epoch": 0.7287128712871287,
+      "grad_norm": 2.18556547164917,
+      "learning_rate": 4.882352941176471e-05,
+      "loss": 0.4042,
+      "step": 138
+    },
+    {
+      "epoch": 0.733993399339934,
+      "grad_norm": 1.9759682416915894,
+      "learning_rate": 4.877450980392157e-05,
+      "loss": 0.3694,
+      "step": 139
+    },
+    {
+      "epoch": 0.7392739273927392,
+      "grad_norm": 1.8205828666687012,
+      "learning_rate": 4.872549019607843e-05,
+      "loss": 0.244,
+      "step": 140
+    },
+    {
+      "epoch": 0.7445544554455445,
+      "grad_norm": 1.8210268020629883,
+      "learning_rate": 4.86764705882353e-05,
+      "loss": 0.4084,
+      "step": 141
+    },
+    {
+      "epoch": 0.7498349834983499,
+      "grad_norm": 2.2197041511535645,
+      "learning_rate": 4.862745098039216e-05,
+      "loss": 0.5236,
+      "step": 142
+    },
+    {
+      "epoch": 0.7551155115511551,
+      "grad_norm": 2.137676239013672,
+      "learning_rate": 4.8578431372549024e-05,
+      "loss": 0.3302,
+      "step": 143
+    },
+    {
+      "epoch": 0.7603960396039604,
+      "grad_norm": 2.126865863800049,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 0.4265,
+      "step": 144
+    },
+    {
+      "epoch": 0.7656765676567657,
+      "grad_norm": 1.828809380531311,
+      "learning_rate": 4.8480392156862745e-05,
+      "loss": 0.4363,
+      "step": 145
+    },
+    {
+      "epoch": 0.7709570957095709,
+      "grad_norm": 1.7918983697891235,
+      "learning_rate": 4.843137254901961e-05,
+      "loss": 0.3568,
+      "step": 146
+    },
+    {
+      "epoch": 0.7762376237623763,
+      "grad_norm": 1.7884886264801025,
+      "learning_rate": 4.838235294117647e-05,
+      "loss": 0.3338,
+      "step": 147
+    },
+    {
+      "epoch": 0.7815181518151815,
+      "grad_norm": 1.8494501113891602,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.3311,
+      "step": 148
+    },
+    {
+      "epoch": 0.7867986798679868,
+      "grad_norm": 2.0265438556671143,
+      "learning_rate": 4.82843137254902e-05,
+      "loss": 0.4381,
+      "step": 149
+    },
+    {
+      "epoch": 0.7920792079207921,
+      "grad_norm": 2.0035383701324463,
+      "learning_rate": 4.823529411764706e-05,
+      "loss": 0.4291,
+      "step": 150
+    },
+    {
+      "epoch": 0.7973597359735973,
+      "grad_norm": 1.8268975019454956,
+      "learning_rate": 4.818627450980392e-05,
+      "loss": 0.464,
+      "step": 151
+    },
+    {
+      "epoch": 0.8026402640264027,
+      "grad_norm": 1.9028264284133911,
+      "learning_rate": 4.813725490196079e-05,
+      "loss": 0.3807,
+      "step": 152
+    },
+    {
+      "epoch": 0.807920792079208,
+      "grad_norm": 1.8994662761688232,
+      "learning_rate": 4.808823529411765e-05,
+      "loss": 0.3904,
+      "step": 153
+    },
+    {
+      "epoch": 0.8132013201320132,
+      "grad_norm": 1.832381248474121,
+      "learning_rate": 4.803921568627452e-05,
+      "loss": 0.3856,
+      "step": 154
+    },
+    {
+      "epoch": 0.8184818481848185,
+      "grad_norm": 1.877752661705017,
+      "learning_rate": 4.799019607843138e-05,
+      "loss": 0.3185,
+      "step": 155
+    },
+    {
+      "epoch": 0.8237623762376237,
+      "grad_norm": 2.0872726440429688,
+      "learning_rate": 4.794117647058824e-05,
+      "loss": 0.3217,
+      "step": 156
+    },
+    {
+      "epoch": 0.829042904290429,
+      "grad_norm": 1.8779263496398926,
+      "learning_rate": 4.7892156862745105e-05,
+      "loss": 0.3654,
+      "step": 157
+    },
+    {
+      "epoch": 0.8343234323432344,
+      "grad_norm": 1.996422529220581,
+      "learning_rate": 4.7843137254901966e-05,
+      "loss": 0.3524,
+      "step": 158
+    },
+    {
+      "epoch": 0.8396039603960396,
+      "grad_norm": 1.5910488367080688,
+      "learning_rate": 4.7794117647058826e-05,
+      "loss": 0.2859,
+      "step": 159
+    },
+    {
+      "epoch": 0.8448844884488449,
+      "grad_norm": 1.9424618482589722,
+      "learning_rate": 4.774509803921569e-05,
+      "loss": 0.3422,
+      "step": 160
+    },
+    {
+      "epoch": 0.8501650165016502,
+      "grad_norm": 1.9187934398651123,
+      "learning_rate": 4.7696078431372554e-05,
+      "loss": 0.3807,
+      "step": 161
+    },
+    {
+      "epoch": 0.8554455445544554,
+      "grad_norm": 1.7809456586837769,
+      "learning_rate": 4.7647058823529414e-05,
+      "loss": 0.2991,
+      "step": 162
+    },
+    {
+      "epoch": 0.8607260726072608,
+      "grad_norm": 1.9575221538543701,
+      "learning_rate": 4.7598039215686275e-05,
+      "loss": 0.4688,
+      "step": 163
+    },
+    {
+      "epoch": 0.866006600660066,
+      "grad_norm": 1.8438433408737183,
+      "learning_rate": 4.7549019607843135e-05,
+      "loss": 0.3691,
+      "step": 164
+    },
+    {
+      "epoch": 0.8712871287128713,
+      "grad_norm": 1.9522879123687744,
+      "learning_rate": 4.75e-05,
+      "loss": 0.3987,
+      "step": 165
+    },
+    {
+      "epoch": 0.8765676567656766,
+      "grad_norm": 2.243354320526123,
+      "learning_rate": 4.745098039215686e-05,
+      "loss": 0.4836,
+      "step": 166
+    },
+    {
+      "epoch": 0.8818481848184818,
+      "grad_norm": 1.8762164115905762,
+      "learning_rate": 4.7401960784313724e-05,
+      "loss": 0.4691,
+      "step": 167
+    },
+    {
+      "epoch": 0.8871287128712871,
+      "grad_norm": 1.8055609464645386,
+      "learning_rate": 4.735294117647059e-05,
+      "loss": 0.2506,
+      "step": 168
+    },
+    {
+      "epoch": 0.8924092409240925,
+      "grad_norm": 1.8521029949188232,
+      "learning_rate": 4.730392156862745e-05,
+      "loss": 0.3291,
+      "step": 169
+    },
+    {
+      "epoch": 0.8976897689768977,
+      "grad_norm": 1.939030647277832,
+      "learning_rate": 4.725490196078431e-05,
+      "loss": 0.4104,
+      "step": 170
+    },
+    {
+      "epoch": 0.902970297029703,
+      "grad_norm": 1.8853607177734375,
+      "learning_rate": 4.720588235294118e-05,
+      "loss": 0.3797,
+      "step": 171
+    },
+    {
+      "epoch": 0.9082508250825082,
+      "grad_norm": 2.0953316688537598,
+      "learning_rate": 4.715686274509804e-05,
+      "loss": 0.324,
+      "step": 172
+    },
+    {
+      "epoch": 0.9135313531353135,
+      "grad_norm": 1.9342799186706543,
+      "learning_rate": 4.71078431372549e-05,
+      "loss": 0.4306,
+      "step": 173
+    },
+    {
+      "epoch": 0.9188118811881189,
+      "grad_norm": 1.8248006105422974,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 0.3388,
+      "step": 174
+    },
+    {
+      "epoch": 0.9240924092409241,
+      "grad_norm": 1.9689913988113403,
+      "learning_rate": 4.700980392156863e-05,
+      "loss": 0.3591,
+      "step": 175
+    },
+    {
+      "epoch": 0.9293729372937294,
+      "grad_norm": 2.017063856124878,
+      "learning_rate": 4.6960784313725495e-05,
+      "loss": 0.4723,
+      "step": 176
+    },
+    {
+      "epoch": 0.9346534653465347,
+      "grad_norm": 1.9692254066467285,
+      "learning_rate": 4.6911764705882356e-05,
+      "loss": 0.3893,
+      "step": 177
+    },
+    {
+      "epoch": 0.9399339933993399,
+      "grad_norm": 1.9935567378997803,
+      "learning_rate": 4.6862745098039216e-05,
+      "loss": 0.3938,
+      "step": 178
+    },
+    {
+      "epoch": 0.9452145214521452,
+      "grad_norm": 1.7153037786483765,
+      "learning_rate": 4.681372549019608e-05,
+      "loss": 0.2368,
+      "step": 179
+    },
+    {
+      "epoch": 0.9504950495049505,
+      "grad_norm": 1.4944133758544922,
+      "learning_rate": 4.6764705882352944e-05,
+      "loss": 0.2181,
+      "step": 180
+    },
+    {
+      "epoch": 0.9557755775577558,
+      "grad_norm": 1.9143524169921875,
+      "learning_rate": 4.6715686274509804e-05,
+      "loss": 0.413,
+      "step": 181
+    },
+    {
+      "epoch": 0.9610561056105611,
+      "grad_norm": 2.162576675415039,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.3745,
+      "step": 182
+    },
+    {
+      "epoch": 0.9663366336633663,
+      "grad_norm": 1.8236726522445679,
+      "learning_rate": 4.661764705882353e-05,
+      "loss": 0.3056,
+      "step": 183
+    },
+    {
+      "epoch": 0.9716171617161716,
+      "grad_norm": 1.9680614471435547,
+      "learning_rate": 4.656862745098039e-05,
+      "loss": 0.3875,
+      "step": 184
+    },
+    {
+      "epoch": 0.976897689768977,
+      "grad_norm": 1.575900912284851,
+      "learning_rate": 4.651960784313726e-05,
+      "loss": 0.1831,
+      "step": 185
+    },
+    {
+      "epoch": 0.9821782178217822,
+      "grad_norm": 2.6015613079071045,
+      "learning_rate": 4.647058823529412e-05,
+      "loss": 0.4227,
+      "step": 186
+    },
+    {
+      "epoch": 0.9874587458745875,
+      "grad_norm": 2.066946268081665,
+      "learning_rate": 4.642156862745098e-05,
+      "loss": 0.4256,
+      "step": 187
+    },
+    {
+      "epoch": 0.9927392739273927,
+      "grad_norm": 2.1683449745178223,
+      "learning_rate": 4.637254901960785e-05,
+      "loss": 0.3943,
+      "step": 188
+    },
+    {
+      "epoch": 0.998019801980198,
+      "grad_norm": 1.876991629600525,
+      "learning_rate": 4.632352941176471e-05,
+      "loss": 0.4049,
+      "step": 189
+    },
+    {
+      "epoch": 1.0033003300330032,
+      "grad_norm": 2.2598772048950195,
+      "learning_rate": 4.6274509803921576e-05,
+      "loss": 0.3749,
+      "step": 190
+    },
+    {
+      "epoch": 1.0085808580858087,
+      "grad_norm": 1.8292592763900757,
+      "learning_rate": 4.6225490196078436e-05,
+      "loss": 0.2654,
+      "step": 191
+    },
+    {
+      "epoch": 1.0138613861386139,
+      "grad_norm": 2.6986138820648193,
+      "learning_rate": 4.61764705882353e-05,
+      "loss": 0.3662,
+      "step": 192
+    },
+    {
+      "epoch": 1.019141914191419,
+      "grad_norm": 1.9084346294403076,
+      "learning_rate": 4.6127450980392164e-05,
+      "loss": 0.2929,
+      "step": 193
+    },
+    {
+      "epoch": 1.0244224422442245,
+      "grad_norm": 1.6963775157928467,
+      "learning_rate": 4.607843137254902e-05,
+      "loss": 0.2481,
+      "step": 194
+    },
+    {
+      "epoch": 1.0297029702970297,
+      "grad_norm": 2.524332046508789,
+      "learning_rate": 4.6029411764705885e-05,
+      "loss": 0.2213,
+      "step": 195
+    },
+    {
+      "epoch": 1.034983498349835,
+      "grad_norm": 1.945142388343811,
+      "learning_rate": 4.5980392156862746e-05,
+      "loss": 0.2892,
+      "step": 196
+    },
+    {
+      "epoch": 1.0402640264026402,
+      "grad_norm": 2.3402678966522217,
+      "learning_rate": 4.5931372549019606e-05,
+      "loss": 0.2232,
+      "step": 197
+    },
+    {
+      "epoch": 1.0455445544554456,
+      "grad_norm": 1.7755571603775024,
+      "learning_rate": 4.588235294117647e-05,
+      "loss": 0.2416,
+      "step": 198
+    },
+    {
+      "epoch": 1.0508250825082508,
+      "grad_norm": 2.110517740249634,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.2557,
+      "step": 199
+    },
+    {
+      "epoch": 1.056105610561056,
+      "grad_norm": 1.7219949960708618,
+      "learning_rate": 4.5784313725490194e-05,
+      "loss": 0.2274,
+      "step": 200
+    },
+    {
+      "epoch": 1.0613861386138614,
+      "grad_norm": 1.9697656631469727,
+      "learning_rate": 4.573529411764706e-05,
+      "loss": 0.2588,
+      "step": 201
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 1.5107744932174683,
+      "learning_rate": 4.568627450980392e-05,
+      "loss": 0.2175,
+      "step": 202
+    },
+    {
+      "epoch": 1.0719471947194719,
+      "grad_norm": 1.8557658195495605,
+      "learning_rate": 4.563725490196078e-05,
+      "loss": 0.2901,
+      "step": 203
+    },
+    {
+      "epoch": 1.0772277227722773,
+      "grad_norm": 1.764145851135254,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 0.176,
+      "step": 204
+    },
+    {
+      "epoch": 1.0825082508250825,
+      "grad_norm": 1.8334012031555176,
+      "learning_rate": 4.553921568627451e-05,
+      "loss": 0.2572,
+      "step": 205
+    },
+    {
+      "epoch": 1.0877887788778877,
+      "grad_norm": 1.8206666707992554,
+      "learning_rate": 4.549019607843137e-05,
+      "loss": 0.248,
+      "step": 206
+    },
+    {
+      "epoch": 1.0930693069306932,
+      "grad_norm": 2.0776381492614746,
+      "learning_rate": 4.544117647058824e-05,
+      "loss": 0.3113,
+      "step": 207
+    },
+    {
+      "epoch": 1.0983498349834984,
+      "grad_norm": 1.8429386615753174,
+      "learning_rate": 4.53921568627451e-05,
+      "loss": 0.2849,
+      "step": 208
+    },
+    {
+      "epoch": 1.1036303630363036,
+      "grad_norm": 1.7320504188537598,
+      "learning_rate": 4.5343137254901966e-05,
+      "loss": 0.2145,
+      "step": 209
+    },
+    {
+      "epoch": 1.108910891089109,
+      "grad_norm": 1.7207646369934082,
+      "learning_rate": 4.5294117647058826e-05,
+      "loss": 0.2013,
+      "step": 210
+    },
+    {
+      "epoch": 1.1141914191419142,
+      "grad_norm": 2.134873151779175,
+      "learning_rate": 4.524509803921569e-05,
+      "loss": 0.3293,
+      "step": 211
+    },
+    {
+      "epoch": 1.1194719471947194,
+      "grad_norm": 1.7931280136108398,
+      "learning_rate": 4.5196078431372554e-05,
+      "loss": 0.2292,
+      "step": 212
+    },
+    {
+      "epoch": 1.1247524752475249,
+      "grad_norm": 2.1878650188446045,
+      "learning_rate": 4.5147058823529415e-05,
+      "loss": 0.3173,
+      "step": 213
+    },
+    {
+      "epoch": 1.13003300330033,
+      "grad_norm": 1.8994349241256714,
+      "learning_rate": 4.5098039215686275e-05,
+      "loss": 0.2742,
+      "step": 214
+    },
+    {
+      "epoch": 1.1353135313531353,
+      "grad_norm": 1.9204659461975098,
+      "learning_rate": 4.504901960784314e-05,
+      "loss": 0.2713,
+      "step": 215
+    },
+    {
+      "epoch": 1.1405940594059407,
+      "grad_norm": 1.9061977863311768,
+      "learning_rate": 4.5e-05,
+      "loss": 0.2433,
+      "step": 216
+    },
+    {
+      "epoch": 1.145874587458746,
+      "grad_norm": 2.254232168197632,
+      "learning_rate": 4.495098039215686e-05,
+      "loss": 0.3523,
+      "step": 217
+    },
+    {
+      "epoch": 1.1511551155115511,
+      "grad_norm": 1.9071446657180786,
+      "learning_rate": 4.490196078431373e-05,
+      "loss": 0.2196,
+      "step": 218
+    },
+    {
+      "epoch": 1.1564356435643564,
+      "grad_norm": 1.8089710474014282,
+      "learning_rate": 4.485294117647059e-05,
+      "loss": 0.2907,
+      "step": 219
+    },
+    {
+      "epoch": 1.1617161716171618,
+      "grad_norm": 1.9056932926177979,
+      "learning_rate": 4.480392156862745e-05,
+      "loss": 0.2564,
+      "step": 220
+    },
+    {
+      "epoch": 1.166996699669967,
+      "grad_norm": 1.9336401224136353,
+      "learning_rate": 4.475490196078432e-05,
+      "loss": 0.205,
+      "step": 221
+    },
+    {
+      "epoch": 1.1722772277227722,
+      "grad_norm": 1.9126192331314087,
+      "learning_rate": 4.470588235294118e-05,
+      "loss": 0.2647,
+      "step": 222
+    },
+    {
+      "epoch": 1.1775577557755776,
+      "grad_norm": 1.8508714437484741,
+      "learning_rate": 4.4656862745098047e-05,
+      "loss": 0.262,
+      "step": 223
+    },
+    {
+      "epoch": 1.1828382838283829,
+      "grad_norm": 1.478278398513794,
+      "learning_rate": 4.460784313725491e-05,
+      "loss": 0.1711,
+      "step": 224
+    },
+    {
+      "epoch": 1.188118811881188,
+      "grad_norm": 1.7818694114685059,
+      "learning_rate": 4.455882352941177e-05,
+      "loss": 0.2043,
+      "step": 225
+    },
+    {
+      "epoch": 1.1933993399339933,
+      "grad_norm": 1.916344404220581,
+      "learning_rate": 4.450980392156863e-05,
+      "loss": 0.3064,
+      "step": 226
+    },
+    {
+      "epoch": 1.1986798679867987,
+      "grad_norm": 1.8637932538986206,
+      "learning_rate": 4.446078431372549e-05,
+      "loss": 0.3247,
+      "step": 227
+    },
+    {
+      "epoch": 1.203960396039604,
+      "grad_norm": 1.7257781028747559,
+      "learning_rate": 4.4411764705882356e-05,
+      "loss": 0.1981,
+      "step": 228
+    },
+    {
+      "epoch": 1.2092409240924091,
+      "grad_norm": 1.9121214151382446,
+      "learning_rate": 4.4362745098039216e-05,
+      "loss": 0.2963,
+      "step": 229
+    },
+    {
+      "epoch": 1.2145214521452146,
+      "grad_norm": 1.5968807935714722,
+      "learning_rate": 4.431372549019608e-05,
+      "loss": 0.1802,
+      "step": 230
+    },
+    {
+      "epoch": 1.2198019801980198,
+      "grad_norm": 1.7256313562393188,
+      "learning_rate": 4.4264705882352944e-05,
+      "loss": 0.2442,
+      "step": 231
+    },
+    {
+      "epoch": 1.225082508250825,
+      "grad_norm": 1.7865731716156006,
+      "learning_rate": 4.4215686274509805e-05,
+      "loss": 0.2683,
+      "step": 232
+    },
+    {
+      "epoch": 1.2303630363036304,
+      "grad_norm": 1.7361854314804077,
+      "learning_rate": 4.4166666666666665e-05,
+      "loss": 0.2285,
+      "step": 233
+    },
+    {
+      "epoch": 1.2356435643564356,
+      "grad_norm": 1.8758944272994995,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 0.3081,
+      "step": 234
+    },
+    {
+      "epoch": 1.2409240924092408,
+      "grad_norm": 2.000033140182495,
+      "learning_rate": 4.406862745098039e-05,
+      "loss": 0.2405,
+      "step": 235
+    },
+    {
+      "epoch": 1.2462046204620463,
+      "grad_norm": 1.8750522136688232,
+      "learning_rate": 4.401960784313725e-05,
+      "loss": 0.2778,
+      "step": 236
+    },
+    {
+      "epoch": 1.2514851485148515,
+      "grad_norm": 1.7535063028335571,
+      "learning_rate": 4.397058823529412e-05,
+      "loss": 0.2045,
+      "step": 237
+    },
+    {
+      "epoch": 1.2567656765676567,
+      "grad_norm": 1.9849064350128174,
+      "learning_rate": 4.392156862745098e-05,
+      "loss": 0.3018,
+      "step": 238
+    },
+    {
+      "epoch": 1.2620462046204621,
+      "grad_norm": 1.8400393724441528,
+      "learning_rate": 4.387254901960784e-05,
+      "loss": 0.2555,
+      "step": 239
+    },
+    {
+      "epoch": 1.2673267326732673,
+      "grad_norm": 1.8575385808944702,
+      "learning_rate": 4.382352941176471e-05,
+      "loss": 0.2783,
+      "step": 240
+    },
+    {
+      "epoch": 1.2726072607260726,
+      "grad_norm": 1.8452024459838867,
+      "learning_rate": 4.377450980392157e-05,
+      "loss": 0.3091,
+      "step": 241
+    },
+    {
+      "epoch": 1.277887788778878,
+      "grad_norm": 1.9682793617248535,
+      "learning_rate": 4.3725490196078437e-05,
+      "loss": 0.2169,
+      "step": 242
+    },
+    {
+      "epoch": 1.2831683168316832,
+      "grad_norm": 1.7926579713821411,
+      "learning_rate": 4.36764705882353e-05,
+      "loss": 0.27,
+      "step": 243
+    },
+    {
+      "epoch": 1.2884488448844884,
+      "grad_norm": 1.690425157546997,
+      "learning_rate": 4.362745098039216e-05,
+      "loss": 0.2336,
+      "step": 244
+    },
+    {
+      "epoch": 1.2937293729372938,
+      "grad_norm": 1.775240421295166,
+      "learning_rate": 4.3578431372549025e-05,
+      "loss": 0.1981,
+      "step": 245
+    },
+    {
+      "epoch": 1.299009900990099,
+      "grad_norm": 1.7951467037200928,
+      "learning_rate": 4.3529411764705885e-05,
+      "loss": 0.2195,
+      "step": 246
+    },
+    {
+      "epoch": 1.3042904290429043,
+      "grad_norm": 1.5444797277450562,
+      "learning_rate": 4.3480392156862746e-05,
+      "loss": 0.2264,
+      "step": 247
+    },
+    {
+      "epoch": 1.3095709570957097,
+      "grad_norm": 1.731550931930542,
+      "learning_rate": 4.343137254901961e-05,
+      "loss": 0.1676,
+      "step": 248
+    },
+    {
+      "epoch": 1.314851485148515,
+      "grad_norm": 1.747083306312561,
+      "learning_rate": 4.3382352941176474e-05,
+      "loss": 0.28,
+      "step": 249
+    },
+    {
+      "epoch": 1.3201320132013201,
+      "grad_norm": 1.7961376905441284,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.2003,
+      "step": 250
+    },
+    {
+      "epoch": 1.3254125412541253,
+      "grad_norm": 1.8366891145706177,
+      "learning_rate": 4.32843137254902e-05,
+      "loss": 0.2093,
+      "step": 251
+    },
+    {
+      "epoch": 1.3306930693069308,
+      "grad_norm": 1.4352390766143799,
+      "learning_rate": 4.323529411764706e-05,
+      "loss": 0.1597,
+      "step": 252
+    },
+    {
+      "epoch": 1.335973597359736,
+      "grad_norm": 1.9146888256072998,
+      "learning_rate": 4.318627450980392e-05,
+      "loss": 0.2425,
+      "step": 253
+    },
+    {
+      "epoch": 1.3412541254125412,
+      "grad_norm": 1.4383189678192139,
+      "learning_rate": 4.313725490196079e-05,
+      "loss": 0.1943,
+      "step": 254
+    },
+    {
+      "epoch": 1.3465346534653464,
+      "grad_norm": 1.5246001482009888,
+      "learning_rate": 4.308823529411765e-05,
+      "loss": 0.1543,
+      "step": 255
+    },
+    {
+      "epoch": 1.3518151815181518,
+      "grad_norm": 1.4863159656524658,
+      "learning_rate": 4.303921568627452e-05,
+      "loss": 0.2441,
+      "step": 256
+    },
+    {
+      "epoch": 1.357095709570957,
+      "grad_norm": 1.6645705699920654,
+      "learning_rate": 4.299019607843138e-05,
+      "loss": 0.2329,
+      "step": 257
+    },
+    {
+      "epoch": 1.3623762376237623,
+      "grad_norm": 1.946554183959961,
+      "learning_rate": 4.294117647058823e-05,
+      "loss": 0.1952,
+      "step": 258
+    },
+    {
+      "epoch": 1.3676567656765677,
+      "grad_norm": 2.0372443199157715,
+      "learning_rate": 4.28921568627451e-05,
+      "loss": 0.332,
+      "step": 259
+    },
+    {
+      "epoch": 1.372937293729373,
+      "grad_norm": 1.846138834953308,
+      "learning_rate": 4.284313725490196e-05,
+      "loss": 0.1408,
+      "step": 260
+    },
+    {
+      "epoch": 1.378217821782178,
+      "grad_norm": 1.5724695920944214,
+      "learning_rate": 4.2794117647058827e-05,
+      "loss": 0.1926,
+      "step": 261
+    },
+    {
+      "epoch": 1.3834983498349835,
+      "grad_norm": 2.1506614685058594,
+      "learning_rate": 4.274509803921569e-05,
+      "loss": 0.217,
+      "step": 262
+    },
+    {
+      "epoch": 1.3887788778877888,
+      "grad_norm": 2.0763325691223145,
+      "learning_rate": 4.269607843137255e-05,
+      "loss": 0.2871,
+      "step": 263
+    },
+    {
+      "epoch": 1.394059405940594,
+      "grad_norm": 1.9296153783798218,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 0.2655,
+      "step": 264
+    },
+    {
+      "epoch": 1.3993399339933994,
+      "grad_norm": 1.7979801893234253,
+      "learning_rate": 4.2598039215686275e-05,
+      "loss": 0.2715,
+      "step": 265
+    },
+    {
+      "epoch": 1.4046204620462046,
+      "grad_norm": 1.4527943134307861,
+      "learning_rate": 4.2549019607843136e-05,
+      "loss": 0.203,
+      "step": 266
+    },
+    {
+      "epoch": 1.4099009900990098,
+      "grad_norm": 1.8454203605651855,
+      "learning_rate": 4.25e-05,
+      "loss": 0.198,
+      "step": 267
+    },
+    {
+      "epoch": 1.4151815181518153,
+      "grad_norm": 1.6438169479370117,
+      "learning_rate": 4.2450980392156864e-05,
+      "loss": 0.2056,
+      "step": 268
+    },
+    {
+      "epoch": 1.4204620462046205,
+      "grad_norm": 1.5819754600524902,
+      "learning_rate": 4.2401960784313724e-05,
+      "loss": 0.2154,
+      "step": 269
+    },
+    {
+      "epoch": 1.4257425742574257,
+      "grad_norm": 1.800973653793335,
+      "learning_rate": 4.235294117647059e-05,
+      "loss": 0.2536,
+      "step": 270
+    },
+    {
+      "epoch": 1.431023102310231,
+      "grad_norm": 1.6425402164459229,
+      "learning_rate": 4.230392156862745e-05,
+      "loss": 0.2111,
+      "step": 271
+    },
+    {
+      "epoch": 1.4363036303630363,
+      "grad_norm": 1.908632755279541,
+      "learning_rate": 4.225490196078431e-05,
+      "loss": 0.2822,
+      "step": 272
+    },
+    {
+      "epoch": 1.4415841584158415,
+      "grad_norm": 2.028026580810547,
+      "learning_rate": 4.220588235294118e-05,
+      "loss": 0.3118,
+      "step": 273
+    },
+    {
+      "epoch": 1.446864686468647,
+      "grad_norm": 1.9891923666000366,
+      "learning_rate": 4.215686274509804e-05,
+      "loss": 0.2469,
+      "step": 274
+    },
+    {
+      "epoch": 1.4521452145214522,
+      "grad_norm": 1.771210789680481,
+      "learning_rate": 4.210784313725491e-05,
+      "loss": 0.2057,
+      "step": 275
+    },
+    {
+      "epoch": 1.4574257425742574,
+      "grad_norm": 1.8335461616516113,
+      "learning_rate": 4.205882352941177e-05,
+      "loss": 0.203,
+      "step": 276
+    },
+    {
+      "epoch": 1.4627062706270628,
+      "grad_norm": 1.4403390884399414,
+      "learning_rate": 4.200980392156863e-05,
+      "loss": 0.1652,
+      "step": 277
+    },
+    {
+      "epoch": 1.467986798679868,
+      "grad_norm": 2.0193352699279785,
+      "learning_rate": 4.1960784313725496e-05,
+      "loss": 0.2914,
+      "step": 278
+    },
+    {
+      "epoch": 1.4732673267326732,
+      "grad_norm": 1.5873808860778809,
+      "learning_rate": 4.1911764705882356e-05,
+      "loss": 0.1871,
+      "step": 279
+    },
+    {
+      "epoch": 1.4785478547854787,
+      "grad_norm": 2.0619425773620605,
+      "learning_rate": 4.1862745098039217e-05,
+      "loss": 0.2449,
+      "step": 280
+    },
+    {
+      "epoch": 1.4838283828382839,
+      "grad_norm": 1.777978539466858,
+      "learning_rate": 4.1813725490196084e-05,
+      "loss": 0.2585,
+      "step": 281
+    },
+    {
+      "epoch": 1.489108910891089,
+      "grad_norm": 1.7166889905929565,
+      "learning_rate": 4.1764705882352944e-05,
+      "loss": 0.2648,
+      "step": 282
+    },
+    {
+      "epoch": 1.4943894389438943,
+      "grad_norm": 1.7009400129318237,
+      "learning_rate": 4.1715686274509805e-05,
+      "loss": 0.1847,
+      "step": 283
+    },
+    {
+      "epoch": 1.4996699669966997,
+      "grad_norm": 2.067512035369873,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.2803,
+      "step": 284
+    },
+    {
+      "epoch": 1.504950495049505,
+      "grad_norm": 1.6885766983032227,
+      "learning_rate": 4.161764705882353e-05,
+      "loss": 0.2037,
+      "step": 285
+    },
+    {
+      "epoch": 1.5102310231023104,
+      "grad_norm": 1.9202988147735596,
+      "learning_rate": 4.156862745098039e-05,
+      "loss": 0.2791,
+      "step": 286
+    },
+    {
+      "epoch": 1.5155115511551154,
+      "grad_norm": 1.6683584451675415,
+      "learning_rate": 4.151960784313726e-05,
+      "loss": 0.2228,
+      "step": 287
+    },
+    {
+      "epoch": 1.5207920792079208,
+      "grad_norm": 1.783361792564392,
+      "learning_rate": 4.147058823529412e-05,
+      "loss": 0.2524,
+      "step": 288
+    },
+    {
+      "epoch": 1.526072607260726,
+      "grad_norm": 1.9562329053878784,
+      "learning_rate": 4.142156862745099e-05,
+      "loss": 0.2773,
+      "step": 289
+    },
+    {
+      "epoch": 1.5313531353135312,
+      "grad_norm": 1.4846049547195435,
+      "learning_rate": 4.137254901960784e-05,
+      "loss": 0.2528,
+      "step": 290
+    },
+    {
+      "epoch": 1.5366336633663367,
+      "grad_norm": 1.6068270206451416,
+      "learning_rate": 4.13235294117647e-05,
+      "loss": 0.1513,
+      "step": 291
+    },
+    {
+      "epoch": 1.5419141914191419,
+      "grad_norm": 1.8563951253890991,
+      "learning_rate": 4.127450980392157e-05,
+      "loss": 0.2518,
+      "step": 292
+    },
+    {
+      "epoch": 1.547194719471947,
+      "grad_norm": 1.9122540950775146,
+      "learning_rate": 4.122549019607843e-05,
+      "loss": 0.2235,
+      "step": 293
+    },
+    {
+      "epoch": 1.5524752475247525,
+      "grad_norm": 1.619687557220459,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 0.236,
+      "step": 294
+    },
+    {
+      "epoch": 1.5577557755775577,
+      "grad_norm": 1.9006292819976807,
+      "learning_rate": 4.112745098039216e-05,
+      "loss": 0.2688,
+      "step": 295
+    },
+    {
+      "epoch": 1.563036303630363,
+      "grad_norm": 1.5912319421768188,
+      "learning_rate": 4.107843137254902e-05,
+      "loss": 0.2036,
+      "step": 296
+    },
+    {
+      "epoch": 1.5683168316831684,
+      "grad_norm": 1.9365366697311401,
+      "learning_rate": 4.1029411764705886e-05,
+      "loss": 0.2708,
+      "step": 297
+    },
+    {
+      "epoch": 1.5735973597359736,
+      "grad_norm": 1.535831332206726,
+      "learning_rate": 4.0980392156862746e-05,
+      "loss": 0.1875,
+      "step": 298
+    },
+    {
+      "epoch": 1.5788778877887788,
+      "grad_norm": 2.117027997970581,
+      "learning_rate": 4.0931372549019607e-05,
+      "loss": 0.3118,
+      "step": 299
+    },
+    {
+      "epoch": 1.5841584158415842,
+      "grad_norm": 1.8837215900421143,
+      "learning_rate": 4.0882352941176474e-05,
+      "loss": 0.2084,
+      "step": 300
+    },
+    {
+      "epoch": 1.5894389438943894,
+      "grad_norm": 1.502886176109314,
+      "learning_rate": 4.0833333333333334e-05,
+      "loss": 0.1348,
+      "step": 301
+    },
+    {
+      "epoch": 1.5947194719471947,
+      "grad_norm": 1.6588914394378662,
+      "learning_rate": 4.0784313725490195e-05,
+      "loss": 0.1775,
+      "step": 302
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.6253366470336914,
+      "learning_rate": 4.073529411764706e-05,
+      "loss": 0.2719,
+      "step": 303
+    },
+    {
+      "epoch": 1.6052805280528053,
+      "grad_norm": 1.9440994262695312,
+      "learning_rate": 4.068627450980392e-05,
+      "loss": 0.2278,
+      "step": 304
+    },
+    {
+      "epoch": 1.6105610561056105,
+      "grad_norm": 1.7787673473358154,
+      "learning_rate": 4.063725490196078e-05,
+      "loss": 0.2513,
+      "step": 305
+    },
+    {
+      "epoch": 1.615841584158416,
+      "grad_norm": 1.3408194780349731,
+      "learning_rate": 4.058823529411765e-05,
+      "loss": 0.1134,
+      "step": 306
+    },
+    {
+      "epoch": 1.6211221122112212,
+      "grad_norm": 1.739343523979187,
+      "learning_rate": 4.053921568627451e-05,
+      "loss": 0.194,
+      "step": 307
+    },
+    {
+      "epoch": 1.6264026402640264,
+      "grad_norm": 1.7391927242279053,
+      "learning_rate": 4.049019607843138e-05,
+      "loss": 0.2273,
+      "step": 308
+    },
+    {
+      "epoch": 1.6316831683168318,
+      "grad_norm": 2.134809732437134,
+      "learning_rate": 4.044117647058824e-05,
+      "loss": 0.349,
+      "step": 309
+    },
+    {
+      "epoch": 1.636963696369637,
+      "grad_norm": 1.8628054857254028,
+      "learning_rate": 4.03921568627451e-05,
+      "loss": 0.2687,
+      "step": 310
+    },
+    {
+      "epoch": 1.6422442244224422,
+      "grad_norm": 1.8895047903060913,
+      "learning_rate": 4.0343137254901966e-05,
+      "loss": 0.2684,
+      "step": 311
+    },
+    {
+      "epoch": 1.6475247524752477,
+      "grad_norm": 1.784513235092163,
+      "learning_rate": 4.029411764705883e-05,
+      "loss": 0.2728,
+      "step": 312
+    },
+    {
+      "epoch": 1.6528052805280526,
+      "grad_norm": 1.505031704902649,
+      "learning_rate": 4.024509803921569e-05,
+      "loss": 0.1806,
+      "step": 313
+    },
+    {
+      "epoch": 1.658085808580858,
+      "grad_norm": 1.9005438089370728,
+      "learning_rate": 4.0196078431372555e-05,
+      "loss": 0.3178,
+      "step": 314
+    },
+    {
+      "epoch": 1.6633663366336635,
+      "grad_norm": 1.769161581993103,
+      "learning_rate": 4.0147058823529415e-05,
+      "loss": 0.2392,
+      "step": 315
+    },
+    {
+      "epoch": 1.6686468646864685,
+      "grad_norm": 1.8065416812896729,
+      "learning_rate": 4.0098039215686276e-05,
+      "loss": 0.2972,
+      "step": 316
+    },
+    {
+      "epoch": 1.673927392739274,
+      "grad_norm": 1.6391319036483765,
+      "learning_rate": 4.004901960784314e-05,
+      "loss": 0.2132,
+      "step": 317
+    },
+    {
+      "epoch": 1.6792079207920794,
+      "grad_norm": 1.3760650157928467,
+      "learning_rate": 4e-05,
+      "loss": 0.1106,
+      "step": 318
+    },
+    {
+      "epoch": 1.6844884488448844,
+      "grad_norm": 1.5832698345184326,
+      "learning_rate": 3.9950980392156864e-05,
+      "loss": 0.1965,
+      "step": 319
+    },
+    {
+      "epoch": 1.6897689768976898,
+      "grad_norm": 1.830043911933899,
+      "learning_rate": 3.990196078431373e-05,
+      "loss": 0.2658,
+      "step": 320
+    },
+    {
+      "epoch": 1.695049504950495,
+      "grad_norm": 1.8273866176605225,
+      "learning_rate": 3.985294117647059e-05,
+      "loss": 0.2284,
+      "step": 321
+    },
+    {
+      "epoch": 1.7003300330033002,
+      "grad_norm": 1.6484299898147583,
+      "learning_rate": 3.980392156862745e-05,
+      "loss": 0.1503,
+      "step": 322
+    },
+    {
+      "epoch": 1.7056105610561056,
+      "grad_norm": 1.7297075986862183,
+      "learning_rate": 3.975490196078431e-05,
+      "loss": 0.1818,
+      "step": 323
+    },
+    {
+      "epoch": 1.7108910891089109,
+      "grad_norm": 1.9557067155838013,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 0.3179,
+      "step": 324
+    },
+    {
+      "epoch": 1.716171617161716,
+      "grad_norm": 1.8135654926300049,
+      "learning_rate": 3.965686274509804e-05,
+      "loss": 0.1871,
+      "step": 325
+    },
+    {
+      "epoch": 1.7214521452145215,
+      "grad_norm": 1.9970617294311523,
+      "learning_rate": 3.96078431372549e-05,
+      "loss": 0.2435,
+      "step": 326
+    },
+    {
+      "epoch": 1.7267326732673267,
+      "grad_norm": 1.584839105606079,
+      "learning_rate": 3.955882352941177e-05,
+      "loss": 0.1813,
+      "step": 327
+    },
+    {
+      "epoch": 1.732013201320132,
+      "grad_norm": 1.5239087343215942,
+      "learning_rate": 3.950980392156863e-05,
+      "loss": 0.1193,
+      "step": 328
+    },
+    {
+      "epoch": 1.7372937293729374,
+      "grad_norm": 1.514541506767273,
+      "learning_rate": 3.946078431372549e-05,
+      "loss": 0.1658,
+      "step": 329
+    },
+    {
+      "epoch": 1.7425742574257426,
+      "grad_norm": 1.8841454982757568,
+      "learning_rate": 3.9411764705882356e-05,
+      "loss": 0.3376,
+      "step": 330
+    },
+    {
+      "epoch": 1.7478547854785478,
+      "grad_norm": 1.5418506860733032,
+      "learning_rate": 3.936274509803922e-05,
+      "loss": 0.2179,
+      "step": 331
+    },
+    {
+      "epoch": 1.7531353135313532,
+      "grad_norm": 1.637239933013916,
+      "learning_rate": 3.931372549019608e-05,
+      "loss": 0.1976,
+      "step": 332
+    },
+    {
+      "epoch": 1.7584158415841584,
+      "grad_norm": 1.8015220165252686,
+      "learning_rate": 3.9264705882352945e-05,
+      "loss": 0.2629,
+      "step": 333
+    },
+    {
+      "epoch": 1.7636963696369636,
+      "grad_norm": 1.7710020542144775,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 0.2208,
+      "step": 334
+    },
+    {
+      "epoch": 1.768976897689769,
+      "grad_norm": 1.9169963598251343,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 0.2811,
+      "step": 335
+    },
+    {
+      "epoch": 1.7742574257425743,
+      "grad_norm": 1.738980770111084,
+      "learning_rate": 3.911764705882353e-05,
+      "loss": 0.2009,
+      "step": 336
+    },
+    {
+      "epoch": 1.7795379537953795,
+      "grad_norm": 1.888925313949585,
+      "learning_rate": 3.906862745098039e-05,
+      "loss": 0.2615,
+      "step": 337
+    },
+    {
+      "epoch": 1.784818481848185,
+      "grad_norm": 1.8248200416564941,
+      "learning_rate": 3.9019607843137254e-05,
+      "loss": 0.2284,
+      "step": 338
+    },
+    {
+      "epoch": 1.7900990099009901,
+      "grad_norm": 1.6162784099578857,
+      "learning_rate": 3.897058823529412e-05,
+      "loss": 0.2206,
+      "step": 339
+    },
+    {
+      "epoch": 1.7953795379537953,
+      "grad_norm": 1.441273808479309,
+      "learning_rate": 3.892156862745098e-05,
+      "loss": 0.1806,
+      "step": 340
+    },
+    {
+      "epoch": 1.8006600660066008,
+      "grad_norm": 1.815674066543579,
+      "learning_rate": 3.887254901960785e-05,
+      "loss": 0.2169,
+      "step": 341
+    },
+    {
+      "epoch": 1.805940594059406,
+      "grad_norm": 1.6441361904144287,
+      "learning_rate": 3.882352941176471e-05,
+      "loss": 0.2177,
+      "step": 342
+    },
+    {
+      "epoch": 1.8112211221122112,
+      "grad_norm": 1.923663854598999,
+      "learning_rate": 3.877450980392157e-05,
+      "loss": 0.2942,
+      "step": 343
+    },
+    {
+      "epoch": 1.8165016501650166,
+      "grad_norm": 1.978797197341919,
+      "learning_rate": 3.872549019607844e-05,
+      "loss": 0.2693,
+      "step": 344
+    },
+    {
+      "epoch": 1.8217821782178216,
+      "grad_norm": 1.8134146928787231,
+      "learning_rate": 3.86764705882353e-05,
+      "loss": 0.2713,
+      "step": 345
+    },
+    {
+      "epoch": 1.827062706270627,
+      "grad_norm": 1.7703922986984253,
+      "learning_rate": 3.862745098039216e-05,
+      "loss": 0.2413,
+      "step": 346
+    },
+    {
+      "epoch": 1.8323432343234325,
+      "grad_norm": 1.7030301094055176,
+      "learning_rate": 3.8578431372549025e-05,
+      "loss": 0.1866,
+      "step": 347
+    },
+    {
+      "epoch": 1.8376237623762375,
+      "grad_norm": 1.521941065788269,
+      "learning_rate": 3.8529411764705886e-05,
+      "loss": 0.174,
+      "step": 348
+    },
+    {
+      "epoch": 1.842904290429043,
+      "grad_norm": 1.8277267217636108,
+      "learning_rate": 3.8480392156862746e-05,
+      "loss": 0.2077,
+      "step": 349
+    },
+    {
+      "epoch": 1.8481848184818483,
+      "grad_norm": 2.028367042541504,
+      "learning_rate": 3.8431372549019614e-05,
+      "loss": 0.2073,
+      "step": 350
+    },
+    {
+      "epoch": 1.8534653465346533,
+      "grad_norm": 1.373708724975586,
+      "learning_rate": 3.8382352941176474e-05,
+      "loss": 0.1383,
+      "step": 351
+    },
+    {
+      "epoch": 1.8587458745874588,
+      "grad_norm": 1.6797735691070557,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.1971,
+      "step": 352
+    },
+    {
+      "epoch": 1.864026402640264,
+      "grad_norm": 1.4328402280807495,
+      "learning_rate": 3.82843137254902e-05,
+      "loss": 0.1601,
+      "step": 353
+    },
+    {
+      "epoch": 1.8693069306930692,
+      "grad_norm": 1.8246557712554932,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 0.2408,
+      "step": 354
+    },
+    {
+      "epoch": 1.8745874587458746,
+      "grad_norm": 1.6448115110397339,
+      "learning_rate": 3.818627450980392e-05,
+      "loss": 0.2677,
+      "step": 355
+    },
+    {
+      "epoch": 1.8798679867986798,
+      "grad_norm": 1.6840052604675293,
+      "learning_rate": 3.813725490196078e-05,
+      "loss": 0.1955,
+      "step": 356
+    },
+    {
+      "epoch": 1.885148514851485,
+      "grad_norm": 1.989355206489563,
+      "learning_rate": 3.8088235294117644e-05,
+      "loss": 0.2761,
+      "step": 357
+    },
+    {
+      "epoch": 1.8904290429042905,
+      "grad_norm": 1.5983843803405762,
+      "learning_rate": 3.803921568627451e-05,
+      "loss": 0.2057,
+      "step": 358
+    },
+    {
+      "epoch": 1.8957095709570957,
+      "grad_norm": 1.5310300588607788,
+      "learning_rate": 3.799019607843137e-05,
+      "loss": 0.1787,
+      "step": 359
+    },
+    {
+      "epoch": 1.900990099009901,
+      "grad_norm": 1.835742712020874,
+      "learning_rate": 3.794117647058824e-05,
+      "loss": 0.2972,
+      "step": 360
+    },
+    {
+      "epoch": 1.9062706270627063,
+      "grad_norm": 1.890306830406189,
+      "learning_rate": 3.78921568627451e-05,
+      "loss": 0.2972,
+      "step": 361
+    },
+    {
+      "epoch": 1.9115511551155115,
+      "grad_norm": 2.0529651641845703,
+      "learning_rate": 3.784313725490196e-05,
+      "loss": 0.3286,
+      "step": 362
+    },
+    {
+      "epoch": 1.9168316831683168,
+      "grad_norm": 1.742149829864502,
+      "learning_rate": 3.779411764705883e-05,
+      "loss": 0.2106,
+      "step": 363
+    },
+    {
+      "epoch": 1.9221122112211222,
+      "grad_norm": 1.8192955255508423,
+      "learning_rate": 3.774509803921569e-05,
+      "loss": 0.2289,
+      "step": 364
+    },
+    {
+      "epoch": 1.9273927392739274,
+      "grad_norm": 1.7159464359283447,
+      "learning_rate": 3.769607843137255e-05,
+      "loss": 0.2324,
+      "step": 365
+    },
+    {
+      "epoch": 1.9326732673267326,
+      "grad_norm": 1.7470611333847046,
+      "learning_rate": 3.7647058823529415e-05,
+      "loss": 0.1381,
+      "step": 366
+    },
+    {
+      "epoch": 1.937953795379538,
+      "grad_norm": 1.7340490818023682,
+      "learning_rate": 3.7598039215686276e-05,
+      "loss": 0.2136,
+      "step": 367
+    },
+    {
+      "epoch": 1.9432343234323433,
+      "grad_norm": 1.6391620635986328,
+      "learning_rate": 3.7549019607843136e-05,
+      "loss": 0.2178,
+      "step": 368
+    },
+    {
+      "epoch": 1.9485148514851485,
+      "grad_norm": 1.5655213594436646,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.2091,
+      "step": 369
+    },
+    {
+      "epoch": 1.953795379537954,
+      "grad_norm": 1.7240495681762695,
+      "learning_rate": 3.7450980392156864e-05,
+      "loss": 0.1779,
+      "step": 370
+    },
+    {
+      "epoch": 1.9590759075907591,
+      "grad_norm": 1.5939252376556396,
+      "learning_rate": 3.7401960784313724e-05,
+      "loss": 0.1863,
+      "step": 371
+    },
+    {
+      "epoch": 1.9643564356435643,
+      "grad_norm": 1.5324146747589111,
+      "learning_rate": 3.735294117647059e-05,
+      "loss": 0.1767,
+      "step": 372
+    },
+    {
+      "epoch": 1.9696369636963698,
+      "grad_norm": 1.689475417137146,
+      "learning_rate": 3.730392156862745e-05,
+      "loss": 0.2135,
+      "step": 373
+    },
+    {
+      "epoch": 1.974917491749175,
+      "grad_norm": 1.9409862756729126,
+      "learning_rate": 3.725490196078432e-05,
+      "loss": 0.2633,
+      "step": 374
+    },
+    {
+      "epoch": 1.9801980198019802,
+      "grad_norm": 1.743085265159607,
+      "learning_rate": 3.720588235294118e-05,
+      "loss": 0.2098,
+      "step": 375
+    },
+    {
+      "epoch": 1.9854785478547856,
+      "grad_norm": 1.6699271202087402,
+      "learning_rate": 3.715686274509804e-05,
+      "loss": 0.2288,
+      "step": 376
+    },
+    {
+      "epoch": 1.9907590759075906,
+      "grad_norm": 1.662705421447754,
+      "learning_rate": 3.710784313725491e-05,
+      "loss": 0.2385,
+      "step": 377
+    },
+    {
+      "epoch": 1.996039603960396,
+      "grad_norm": 1.4365413188934326,
+      "learning_rate": 3.705882352941177e-05,
+      "loss": 0.1793,
+      "step": 378
+    },
+    {
+      "epoch": 2.0013201320132015,
+      "grad_norm": 1.5893417596817017,
+      "learning_rate": 3.700980392156863e-05,
+      "loss": 0.2293,
+      "step": 379
+    },
+    {
+      "epoch": 2.0066006600660065,
+      "grad_norm": 1.2844809293746948,
+      "learning_rate": 3.6960784313725496e-05,
+      "loss": 0.1413,
+      "step": 380
+    },
+    {
+      "epoch": 2.011881188118812,
+      "grad_norm": 1.3603469133377075,
+      "learning_rate": 3.6911764705882356e-05,
+      "loss": 0.1292,
+      "step": 381
+    },
+    {
+      "epoch": 2.0171617161716173,
+      "grad_norm": 1.5470837354660034,
+      "learning_rate": 3.686274509803922e-05,
+      "loss": 0.1572,
+      "step": 382
+    },
+    {
+      "epoch": 2.0224422442244223,
+      "grad_norm": 1.8591777086257935,
+      "learning_rate": 3.6813725490196084e-05,
+      "loss": 0.1545,
+      "step": 383
+    },
+    {
+      "epoch": 2.0277227722772277,
+      "grad_norm": 1.2977832555770874,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 0.0742,
+      "step": 384
+    },
+    {
+      "epoch": 2.033003300330033,
+      "grad_norm": 1.9429755210876465,
+      "learning_rate": 3.6715686274509805e-05,
+      "loss": 0.1088,
+      "step": 385
+    },
+    {
+      "epoch": 2.038283828382838,
+      "grad_norm": 2.0210816860198975,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.1492,
+      "step": 386
+    },
+    {
+      "epoch": 2.0435643564356436,
+      "grad_norm": 1.8192780017852783,
+      "learning_rate": 3.6617647058823526e-05,
+      "loss": 0.1585,
+      "step": 387
+    },
+    {
+      "epoch": 2.048844884488449,
+      "grad_norm": 1.5634256601333618,
+      "learning_rate": 3.6568627450980393e-05,
+      "loss": 0.1152,
+      "step": 388
+    },
+    {
+      "epoch": 2.054125412541254,
+      "grad_norm": 1.7004332542419434,
+      "learning_rate": 3.6519607843137254e-05,
+      "loss": 0.1629,
+      "step": 389
+    },
+    {
+      "epoch": 2.0594059405940595,
+      "grad_norm": 1.7340906858444214,
+      "learning_rate": 3.6470588235294114e-05,
+      "loss": 0.1523,
+      "step": 390
+    },
+    {
+      "epoch": 2.064686468646865,
+      "grad_norm": 1.4663294553756714,
+      "learning_rate": 3.642156862745098e-05,
+      "loss": 0.1138,
+      "step": 391
+    },
+    {
+      "epoch": 2.06996699669967,
+      "grad_norm": 1.6033658981323242,
+      "learning_rate": 3.637254901960784e-05,
+      "loss": 0.1485,
+      "step": 392
+    },
+    {
+      "epoch": 2.0752475247524753,
+      "grad_norm": 1.3706963062286377,
+      "learning_rate": 3.632352941176471e-05,
+      "loss": 0.0883,
+      "step": 393
+    },
+    {
+      "epoch": 2.0805280528052803,
+      "grad_norm": 1.0890157222747803,
+      "learning_rate": 3.627450980392157e-05,
+      "loss": 0.0526,
+      "step": 394
+    },
+    {
+      "epoch": 2.0858085808580857,
+      "grad_norm": 1.5241326093673706,
+      "learning_rate": 3.622549019607843e-05,
+      "loss": 0.1417,
+      "step": 395
+    },
+    {
+      "epoch": 2.091089108910891,
+      "grad_norm": 1.389540433883667,
+      "learning_rate": 3.61764705882353e-05,
+      "loss": 0.0972,
+      "step": 396
+    },
+    {
+      "epoch": 2.096369636963696,
+      "grad_norm": 1.8035510778427124,
+      "learning_rate": 3.612745098039216e-05,
+      "loss": 0.1539,
+      "step": 397
+    },
+    {
+      "epoch": 2.1016501650165016,
+      "grad_norm": 1.5949468612670898,
+      "learning_rate": 3.607843137254902e-05,
+      "loss": 0.1086,
+      "step": 398
+    },
+    {
+      "epoch": 2.106930693069307,
+      "grad_norm": 1.5921865701675415,
+      "learning_rate": 3.6029411764705886e-05,
+      "loss": 0.1227,
+      "step": 399
+    },
+    {
+      "epoch": 2.112211221122112,
+      "grad_norm": 1.6403027772903442,
+      "learning_rate": 3.5980392156862746e-05,
+      "loss": 0.1315,
+      "step": 400
+    },
+    {
+      "epoch": 2.1174917491749174,
+      "grad_norm": 1.7026506662368774,
+      "learning_rate": 3.593137254901961e-05,
+      "loss": 0.1178,
+      "step": 401
+    },
+    {
+      "epoch": 2.122772277227723,
+      "grad_norm": 1.5574462413787842,
+      "learning_rate": 3.5882352941176474e-05,
+      "loss": 0.1316,
+      "step": 402
+    },
+    {
+      "epoch": 2.128052805280528,
+      "grad_norm": 1.6316189765930176,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 0.1351,
+      "step": 403
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 1.2187312841415405,
+      "learning_rate": 3.5784313725490195e-05,
+      "loss": 0.0708,
+      "step": 404
+    },
+    {
+      "epoch": 2.1386138613861387,
+      "grad_norm": 1.619545817375183,
+      "learning_rate": 3.573529411764706e-05,
+      "loss": 0.1369,
+      "step": 405
+    },
+    {
+      "epoch": 2.1438943894389437,
+      "grad_norm": 1.4654717445373535,
+      "learning_rate": 3.568627450980392e-05,
+      "loss": 0.1499,
+      "step": 406
+    },
+    {
+      "epoch": 2.149174917491749,
+      "grad_norm": 1.393074870109558,
+      "learning_rate": 3.563725490196079e-05,
+      "loss": 0.1277,
+      "step": 407
+    },
+    {
+      "epoch": 2.1544554455445546,
+      "grad_norm": 1.7896983623504639,
+      "learning_rate": 3.558823529411765e-05,
+      "loss": 0.166,
+      "step": 408
+    },
+    {
+      "epoch": 2.1597359735973596,
+      "grad_norm": 1.5691279172897339,
+      "learning_rate": 3.553921568627451e-05,
+      "loss": 0.0936,
+      "step": 409
+    },
+    {
+      "epoch": 2.165016501650165,
+      "grad_norm": 1.6163969039916992,
+      "learning_rate": 3.549019607843138e-05,
+      "loss": 0.1234,
+      "step": 410
+    },
+    {
+      "epoch": 2.1702970297029704,
+      "grad_norm": 1.7184455394744873,
+      "learning_rate": 3.544117647058824e-05,
+      "loss": 0.1287,
+      "step": 411
+    },
+    {
+      "epoch": 2.1755775577557754,
+      "grad_norm": 1.8023498058319092,
+      "learning_rate": 3.53921568627451e-05,
+      "loss": 0.1433,
+      "step": 412
+    },
+    {
+      "epoch": 2.180858085808581,
+      "grad_norm": 1.6063473224639893,
+      "learning_rate": 3.534313725490197e-05,
+      "loss": 0.1158,
+      "step": 413
+    },
+    {
+      "epoch": 2.1861386138613863,
+      "grad_norm": 1.6521129608154297,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 0.115,
+      "step": 414
+    },
+    {
+      "epoch": 2.1914191419141913,
+      "grad_norm": 1.5446815490722656,
+      "learning_rate": 3.524509803921569e-05,
+      "loss": 0.115,
+      "step": 415
+    },
+    {
+      "epoch": 2.1966996699669967,
+      "grad_norm": 1.4110487699508667,
+      "learning_rate": 3.5196078431372555e-05,
+      "loss": 0.132,
+      "step": 416
+    },
+    {
+      "epoch": 2.201980198019802,
+      "grad_norm": 1.732271671295166,
+      "learning_rate": 3.514705882352941e-05,
+      "loss": 0.116,
+      "step": 417
+    },
+    {
+      "epoch": 2.207260726072607,
+      "grad_norm": 1.8033490180969238,
+      "learning_rate": 3.5098039215686276e-05,
+      "loss": 0.1675,
+      "step": 418
+    },
+    {
+      "epoch": 2.2125412541254126,
+      "grad_norm": 1.6612602472305298,
+      "learning_rate": 3.5049019607843136e-05,
+      "loss": 0.1549,
+      "step": 419
+    },
+    {
+      "epoch": 2.217821782178218,
+      "grad_norm": 1.4840703010559082,
+      "learning_rate": 3.5e-05,
+      "loss": 0.116,
+      "step": 420
+    },
+    {
+      "epoch": 2.223102310231023,
+      "grad_norm": 1.5208748579025269,
+      "learning_rate": 3.4950980392156864e-05,
+      "loss": 0.1286,
+      "step": 421
+    },
+    {
+      "epoch": 2.2283828382838284,
+      "grad_norm": 1.6537951231002808,
+      "learning_rate": 3.4901960784313725e-05,
+      "loss": 0.0995,
+      "step": 422
+    },
+    {
+      "epoch": 2.233663366336634,
+      "grad_norm": 1.8060580492019653,
+      "learning_rate": 3.4852941176470585e-05,
+      "loss": 0.1264,
+      "step": 423
+    },
+    {
+      "epoch": 2.238943894389439,
+      "grad_norm": 1.8712666034698486,
+      "learning_rate": 3.480392156862745e-05,
+      "loss": 0.1722,
+      "step": 424
+    },
+    {
+      "epoch": 2.2442244224422443,
+      "grad_norm": 1.5156561136245728,
+      "learning_rate": 3.475490196078431e-05,
+      "loss": 0.117,
+      "step": 425
+    },
+    {
+      "epoch": 2.2495049504950497,
+      "grad_norm": 1.6250231266021729,
+      "learning_rate": 3.470588235294118e-05,
+      "loss": 0.1299,
+      "step": 426
+    },
+    {
+      "epoch": 2.2547854785478547,
+      "grad_norm": 2.060224771499634,
+      "learning_rate": 3.465686274509804e-05,
+      "loss": 0.1552,
+      "step": 427
+    },
+    {
+      "epoch": 2.26006600660066,
+      "grad_norm": 1.6682344675064087,
+      "learning_rate": 3.46078431372549e-05,
+      "loss": 0.137,
+      "step": 428
+    },
+    {
+      "epoch": 2.2653465346534656,
+      "grad_norm": 1.5238713026046753,
+      "learning_rate": 3.455882352941177e-05,
+      "loss": 0.0956,
+      "step": 429
+    },
+    {
+      "epoch": 2.2706270627062706,
+      "grad_norm": 1.7260777950286865,
+      "learning_rate": 3.450980392156863e-05,
+      "loss": 0.127,
+      "step": 430
+    },
+    {
+      "epoch": 2.275907590759076,
+      "grad_norm": 1.5617260932922363,
+      "learning_rate": 3.446078431372549e-05,
+      "loss": 0.1495,
+      "step": 431
+    },
+    {
+      "epoch": 2.2811881188118814,
+      "grad_norm": 1.6972553730010986,
+      "learning_rate": 3.441176470588236e-05,
+      "loss": 0.1521,
+      "step": 432
+    },
+    {
+      "epoch": 2.2864686468646864,
+      "grad_norm": 1.4397108554840088,
+      "learning_rate": 3.436274509803922e-05,
+      "loss": 0.1218,
+      "step": 433
+    },
+    {
+      "epoch": 2.291749174917492,
+      "grad_norm": 1.099647045135498,
+      "learning_rate": 3.431372549019608e-05,
+      "loss": 0.0629,
+      "step": 434
+    },
+    {
+      "epoch": 2.297029702970297,
+      "grad_norm": 1.1678043603897095,
+      "learning_rate": 3.4264705882352945e-05,
+      "loss": 0.0905,
+      "step": 435
+    },
+    {
+      "epoch": 2.3023102310231023,
+      "grad_norm": 1.3565727472305298,
+      "learning_rate": 3.4215686274509805e-05,
+      "loss": 0.1028,
+      "step": 436
+    },
+    {
+      "epoch": 2.3075907590759077,
+      "grad_norm": 1.846362590789795,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 0.1715,
+      "step": 437
+    },
+    {
+      "epoch": 2.3128712871287127,
+      "grad_norm": 1.637338399887085,
+      "learning_rate": 3.411764705882353e-05,
+      "loss": 0.131,
+      "step": 438
+    },
+    {
+      "epoch": 2.318151815181518,
+      "grad_norm": 1.5920330286026,
+      "learning_rate": 3.4068627450980394e-05,
+      "loss": 0.1279,
+      "step": 439
+    },
+    {
+      "epoch": 2.3234323432343236,
+      "grad_norm": 1.6633886098861694,
+      "learning_rate": 3.401960784313726e-05,
+      "loss": 0.1284,
+      "step": 440
+    },
+    {
+      "epoch": 2.3287128712871286,
+      "grad_norm": 1.677240252494812,
+      "learning_rate": 3.397058823529412e-05,
+      "loss": 0.132,
+      "step": 441
+    },
+    {
+      "epoch": 2.333993399339934,
+      "grad_norm": 1.6058099269866943,
+      "learning_rate": 3.392156862745098e-05,
+      "loss": 0.116,
+      "step": 442
+    },
+    {
+      "epoch": 2.3392739273927394,
+      "grad_norm": 1.586955189704895,
+      "learning_rate": 3.387254901960785e-05,
+      "loss": 0.1201,
+      "step": 443
+    },
+    {
+      "epoch": 2.3445544554455444,
+      "grad_norm": 2.02844500541687,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 0.1732,
+      "step": 444
+    },
+    {
+      "epoch": 2.34983498349835,
+      "grad_norm": 1.858375072479248,
+      "learning_rate": 3.377450980392157e-05,
+      "loss": 0.1591,
+      "step": 445
+    },
+    {
+      "epoch": 2.3551155115511553,
+      "grad_norm": 1.9212247133255005,
+      "learning_rate": 3.372549019607844e-05,
+      "loss": 0.2203,
+      "step": 446
+    },
+    {
+      "epoch": 2.3603960396039603,
+      "grad_norm": 1.327735424041748,
+      "learning_rate": 3.36764705882353e-05,
+      "loss": 0.0983,
+      "step": 447
+    },
+    {
+      "epoch": 2.3656765676567657,
+      "grad_norm": 1.6712234020233154,
+      "learning_rate": 3.362745098039216e-05,
+      "loss": 0.1282,
+      "step": 448
+    },
+    {
+      "epoch": 2.370957095709571,
+      "grad_norm": 1.812563180923462,
+      "learning_rate": 3.357843137254902e-05,
+      "loss": 0.1417,
+      "step": 449
+    },
+    {
+      "epoch": 2.376237623762376,
+      "grad_norm": 1.4844976663589478,
+      "learning_rate": 3.352941176470588e-05,
+      "loss": 0.0892,
+      "step": 450
+    },
+    {
+      "epoch": 2.3815181518151816,
+      "grad_norm": 1.7959266901016235,
+      "learning_rate": 3.348039215686275e-05,
+      "loss": 0.1753,
+      "step": 451
+    },
+    {
+      "epoch": 2.3867986798679866,
+      "grad_norm": 1.4532088041305542,
+      "learning_rate": 3.343137254901961e-05,
+      "loss": 0.1154,
+      "step": 452
+    },
+    {
+      "epoch": 2.392079207920792,
+      "grad_norm": 1.618485927581787,
+      "learning_rate": 3.338235294117647e-05,
+      "loss": 0.1268,
+      "step": 453
+    },
+    {
+      "epoch": 2.3973597359735974,
+      "grad_norm": 1.537070870399475,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.1304,
+      "step": 454
+    },
+    {
+      "epoch": 2.4026402640264024,
+      "grad_norm": 1.4429882764816284,
+      "learning_rate": 3.3284313725490195e-05,
+      "loss": 0.1521,
+      "step": 455
+    },
+    {
+      "epoch": 2.407920792079208,
+      "grad_norm": 1.5398533344268799,
+      "learning_rate": 3.3235294117647056e-05,
+      "loss": 0.1237,
+      "step": 456
+    },
+    {
+      "epoch": 2.4132013201320133,
+      "grad_norm": 1.7345346212387085,
+      "learning_rate": 3.318627450980392e-05,
+      "loss": 0.1765,
+      "step": 457
+    },
+    {
+      "epoch": 2.4184818481848183,
+      "grad_norm": 1.6664462089538574,
+      "learning_rate": 3.3137254901960784e-05,
+      "loss": 0.1363,
+      "step": 458
+    },
+    {
+      "epoch": 2.4237623762376237,
+      "grad_norm": 1.2778750658035278,
+      "learning_rate": 3.308823529411765e-05,
+      "loss": 0.0798,
+      "step": 459
+    },
+    {
+      "epoch": 2.429042904290429,
+      "grad_norm": 1.5715030431747437,
+      "learning_rate": 3.303921568627451e-05,
+      "loss": 0.1376,
+      "step": 460
+    },
+    {
+      "epoch": 2.434323432343234,
+      "grad_norm": 1.3661701679229736,
+      "learning_rate": 3.299019607843137e-05,
+      "loss": 0.1179,
+      "step": 461
+    },
+    {
+      "epoch": 2.4396039603960396,
+      "grad_norm": 1.6188709735870361,
+      "learning_rate": 3.294117647058824e-05,
+      "loss": 0.102,
+      "step": 462
+    },
+    {
+      "epoch": 2.444884488448845,
+      "grad_norm": 1.5044273138046265,
+      "learning_rate": 3.28921568627451e-05,
+      "loss": 0.1048,
+      "step": 463
+    },
+    {
+      "epoch": 2.45016501650165,
+      "grad_norm": 1.8329862356185913,
+      "learning_rate": 3.284313725490196e-05,
+      "loss": 0.1501,
+      "step": 464
+    },
+    {
+      "epoch": 2.4554455445544554,
+      "grad_norm": 1.4117523431777954,
+      "learning_rate": 3.279411764705883e-05,
+      "loss": 0.0899,
+      "step": 465
+    },
+    {
+      "epoch": 2.460726072607261,
+      "grad_norm": 1.4140467643737793,
+      "learning_rate": 3.274509803921569e-05,
+      "loss": 0.1143,
+      "step": 466
+    },
+    {
+      "epoch": 2.466006600660066,
+      "grad_norm": 1.5512601137161255,
+      "learning_rate": 3.269607843137255e-05,
+      "loss": 0.1294,
+      "step": 467
+    },
+    {
+      "epoch": 2.4712871287128713,
+      "grad_norm": 1.366589069366455,
+      "learning_rate": 3.2647058823529416e-05,
+      "loss": 0.1044,
+      "step": 468
+    },
+    {
+      "epoch": 2.4765676567656767,
+      "grad_norm": 1.4923369884490967,
+      "learning_rate": 3.2598039215686276e-05,
+      "loss": 0.1209,
+      "step": 469
+    },
+    {
+      "epoch": 2.4818481848184817,
+      "grad_norm": 1.4540033340454102,
+      "learning_rate": 3.254901960784314e-05,
+      "loss": 0.1094,
+      "step": 470
+    },
+    {
+      "epoch": 2.487128712871287,
+      "grad_norm": 1.7414342164993286,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.2082,
+      "step": 471
+    },
+    {
+      "epoch": 2.4924092409240926,
+      "grad_norm": 1.3162891864776611,
+      "learning_rate": 3.2450980392156864e-05,
+      "loss": 0.0972,
+      "step": 472
+    },
+    {
+      "epoch": 2.4976897689768975,
+      "grad_norm": 1.5640320777893066,
+      "learning_rate": 3.240196078431373e-05,
+      "loss": 0.1221,
+      "step": 473
+    },
+    {
+      "epoch": 2.502970297029703,
+      "grad_norm": 1.3759536743164062,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 0.1026,
+      "step": 474
+    },
+    {
+      "epoch": 2.5082508250825084,
+      "grad_norm": 1.3398675918579102,
+      "learning_rate": 3.230392156862745e-05,
+      "loss": 0.0904,
+      "step": 475
+    },
+    {
+      "epoch": 2.5135313531353134,
+      "grad_norm": 1.1896995306015015,
+      "learning_rate": 3.225490196078432e-05,
+      "loss": 0.0847,
+      "step": 476
+    },
+    {
+      "epoch": 2.518811881188119,
+      "grad_norm": 1.3623279333114624,
+      "learning_rate": 3.220588235294118e-05,
+      "loss": 0.1159,
+      "step": 477
+    },
+    {
+      "epoch": 2.5240924092409243,
+      "grad_norm": 1.3050968647003174,
+      "learning_rate": 3.215686274509804e-05,
+      "loss": 0.0967,
+      "step": 478
+    },
+    {
+      "epoch": 2.5293729372937293,
+      "grad_norm": 1.8732277154922485,
+      "learning_rate": 3.210784313725491e-05,
+      "loss": 0.1273,
+      "step": 479
+    },
+    {
+      "epoch": 2.5346534653465347,
+      "grad_norm": 1.5116699934005737,
+      "learning_rate": 3.205882352941177e-05,
+      "loss": 0.1245,
+      "step": 480
+    },
+    {
+      "epoch": 2.53993399339934,
+      "grad_norm": 1.6075270175933838,
+      "learning_rate": 3.200980392156863e-05,
+      "loss": 0.1215,
+      "step": 481
+    },
+    {
+      "epoch": 2.545214521452145,
+      "grad_norm": 1.6938683986663818,
+      "learning_rate": 3.196078431372549e-05,
+      "loss": 0.1226,
+      "step": 482
+    },
+    {
+      "epoch": 2.5504950495049505,
+      "grad_norm": 1.361075520515442,
+      "learning_rate": 3.191176470588235e-05,
+      "loss": 0.108,
+      "step": 483
+    },
+    {
+      "epoch": 2.555775577557756,
+      "grad_norm": 1.388642430305481,
+      "learning_rate": 3.186274509803922e-05,
+      "loss": 0.1223,
+      "step": 484
+    },
+    {
+      "epoch": 2.561056105610561,
+      "grad_norm": 1.637742519378662,
+      "learning_rate": 3.181372549019608e-05,
+      "loss": 0.1395,
+      "step": 485
+    },
+    {
+      "epoch": 2.5663366336633664,
+      "grad_norm": 1.377528429031372,
+      "learning_rate": 3.176470588235294e-05,
+      "loss": 0.1064,
+      "step": 486
+    },
+    {
+      "epoch": 2.571617161716172,
+      "grad_norm": 1.3829468488693237,
+      "learning_rate": 3.1715686274509806e-05,
+      "loss": 0.1101,
+      "step": 487
+    },
+    {
+      "epoch": 2.576897689768977,
+      "grad_norm": 1.142531394958496,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.0885,
+      "step": 488
+    },
+    {
+      "epoch": 2.5821782178217823,
+      "grad_norm": 1.226916790008545,
+      "learning_rate": 3.161764705882353e-05,
+      "loss": 0.0713,
+      "step": 489
+    },
+    {
+      "epoch": 2.5874587458745877,
+      "grad_norm": 1.3948824405670166,
+      "learning_rate": 3.1568627450980394e-05,
+      "loss": 0.0945,
+      "step": 490
+    },
+    {
+      "epoch": 2.5927392739273927,
+      "grad_norm": 1.679543375968933,
+      "learning_rate": 3.1519607843137254e-05,
+      "loss": 0.1122,
+      "step": 491
+    },
+    {
+      "epoch": 2.598019801980198,
+      "grad_norm": 1.4996442794799805,
+      "learning_rate": 3.147058823529412e-05,
+      "loss": 0.1502,
+      "step": 492
+    },
+    {
+      "epoch": 2.6033003300330035,
+      "grad_norm": 1.6479462385177612,
+      "learning_rate": 3.142156862745098e-05,
+      "loss": 0.1335,
+      "step": 493
+    },
+    {
+      "epoch": 2.6085808580858085,
+      "grad_norm": 1.9503371715545654,
+      "learning_rate": 3.137254901960784e-05,
+      "loss": 0.1378,
+      "step": 494
+    },
+    {
+      "epoch": 2.613861386138614,
+      "grad_norm": 1.6987338066101074,
+      "learning_rate": 3.132352941176471e-05,
+      "loss": 0.174,
+      "step": 495
+    },
+    {
+      "epoch": 2.6191419141914194,
+      "grad_norm": 1.4869428873062134,
+      "learning_rate": 3.127450980392157e-05,
+      "loss": 0.1444,
+      "step": 496
+    },
+    {
+      "epoch": 2.6244224422442244,
+      "grad_norm": 1.4220130443572998,
+      "learning_rate": 3.122549019607843e-05,
+      "loss": 0.1146,
+      "step": 497
+    },
+    {
+      "epoch": 2.62970297029703,
+      "grad_norm": 1.8738462924957275,
+      "learning_rate": 3.11764705882353e-05,
+      "loss": 0.1999,
+      "step": 498
+    },
+    {
+      "epoch": 2.6349834983498353,
+      "grad_norm": 1.6982347965240479,
+      "learning_rate": 3.112745098039216e-05,
+      "loss": 0.1241,
+      "step": 499
+    },
+    {
+      "epoch": 2.6402640264026402,
+      "grad_norm": 1.5183193683624268,
+      "learning_rate": 3.107843137254902e-05,
+      "loss": 0.1198,
+      "step": 500
+    },
+    {
+      "epoch": 2.6455445544554457,
+      "grad_norm": 1.4872636795043945,
+      "learning_rate": 3.1029411764705886e-05,
+      "loss": 0.0927,
+      "step": 501
+    },
+    {
+      "epoch": 2.6508250825082507,
+      "grad_norm": 1.6051157712936401,
+      "learning_rate": 3.098039215686275e-05,
+      "loss": 0.1524,
+      "step": 502
+    },
+    {
+      "epoch": 2.656105610561056,
+      "grad_norm": 1.5139557123184204,
+      "learning_rate": 3.093137254901961e-05,
+      "loss": 0.1237,
+      "step": 503
+    },
+    {
+      "epoch": 2.6613861386138615,
+      "grad_norm": 1.5860090255737305,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 0.1409,
+      "step": 504
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 1.7378284931182861,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 0.1551,
+      "step": 505
+    },
+    {
+      "epoch": 2.671947194719472,
+      "grad_norm": 1.5908591747283936,
+      "learning_rate": 3.07843137254902e-05,
+      "loss": 0.1169,
+      "step": 506
+    },
+    {
+      "epoch": 2.6772277227722774,
+      "grad_norm": 1.4210619926452637,
+      "learning_rate": 3.073529411764706e-05,
+      "loss": 0.1217,
+      "step": 507
+    },
+    {
+      "epoch": 2.6825082508250824,
+      "grad_norm": 1.4068129062652588,
+      "learning_rate": 3.0686274509803923e-05,
+      "loss": 0.1484,
+      "step": 508
+    },
+    {
+      "epoch": 2.687788778877888,
+      "grad_norm": 1.4808945655822754,
+      "learning_rate": 3.063725490196079e-05,
+      "loss": 0.1402,
+      "step": 509
+    },
+    {
+      "epoch": 2.693069306930693,
+      "grad_norm": 1.342729926109314,
+      "learning_rate": 3.058823529411765e-05,
+      "loss": 0.0826,
+      "step": 510
+    },
+    {
+      "epoch": 2.6983498349834982,
+      "grad_norm": 1.457270860671997,
+      "learning_rate": 3.053921568627451e-05,
+      "loss": 0.0809,
+      "step": 511
+    },
+    {
+      "epoch": 2.7036303630363037,
+      "grad_norm": 1.3564640283584595,
+      "learning_rate": 3.0490196078431376e-05,
+      "loss": 0.1333,
+      "step": 512
+    },
+    {
+      "epoch": 2.7089108910891087,
+      "grad_norm": 1.5419988632202148,
+      "learning_rate": 3.0441176470588233e-05,
+      "loss": 0.1145,
+      "step": 513
+    },
+    {
+      "epoch": 2.714191419141914,
+      "grad_norm": 1.4028230905532837,
+      "learning_rate": 3.0392156862745097e-05,
+      "loss": 0.1156,
+      "step": 514
+    },
+    {
+      "epoch": 2.7194719471947195,
+      "grad_norm": 1.5217773914337158,
+      "learning_rate": 3.034313725490196e-05,
+      "loss": 0.0962,
+      "step": 515
+    },
+    {
+      "epoch": 2.7247524752475245,
+      "grad_norm": 1.2986435890197754,
+      "learning_rate": 3.0294117647058824e-05,
+      "loss": 0.1039,
+      "step": 516
+    },
+    {
+      "epoch": 2.73003300330033,
+      "grad_norm": 1.3331955671310425,
+      "learning_rate": 3.0245098039215685e-05,
+      "loss": 0.1077,
+      "step": 517
+    },
+    {
+      "epoch": 2.7353135313531354,
+      "grad_norm": 1.7780873775482178,
+      "learning_rate": 3.019607843137255e-05,
+      "loss": 0.2033,
+      "step": 518
+    },
+    {
+      "epoch": 2.7405940594059404,
+      "grad_norm": 1.1954641342163086,
+      "learning_rate": 3.0147058823529413e-05,
+      "loss": 0.1134,
+      "step": 519
+    },
+    {
+      "epoch": 2.745874587458746,
+      "grad_norm": 1.465927004814148,
+      "learning_rate": 3.0098039215686273e-05,
+      "loss": 0.1329,
+      "step": 520
+    },
+    {
+      "epoch": 2.7511551155115512,
+      "grad_norm": 1.359671711921692,
+      "learning_rate": 3.0049019607843137e-05,
+      "loss": 0.0876,
+      "step": 521
+    },
+    {
+      "epoch": 2.756435643564356,
+      "grad_norm": 1.570184350013733,
+      "learning_rate": 3e-05,
+      "loss": 0.1125,
+      "step": 522
+    },
+    {
+      "epoch": 2.7617161716171617,
+      "grad_norm": 1.286363959312439,
+      "learning_rate": 2.9950980392156865e-05,
+      "loss": 0.0978,
+      "step": 523
+    },
+    {
+      "epoch": 2.766996699669967,
+      "grad_norm": 1.6705877780914307,
+      "learning_rate": 2.9901960784313725e-05,
+      "loss": 0.1546,
+      "step": 524
+    },
+    {
+      "epoch": 2.772277227722772,
+      "grad_norm": 1.475122332572937,
+      "learning_rate": 2.985294117647059e-05,
+      "loss": 0.1083,
+      "step": 525
+    },
+    {
+      "epoch": 2.7775577557755775,
+      "grad_norm": 1.4407463073730469,
+      "learning_rate": 2.9803921568627453e-05,
+      "loss": 0.1123,
+      "step": 526
+    },
+    {
+      "epoch": 2.782838283828383,
+      "grad_norm": 1.4154635667800903,
+      "learning_rate": 2.9754901960784313e-05,
+      "loss": 0.1232,
+      "step": 527
+    },
+    {
+      "epoch": 2.788118811881188,
+      "grad_norm": 1.6671222448349,
+      "learning_rate": 2.9705882352941177e-05,
+      "loss": 0.1767,
+      "step": 528
+    },
+    {
+      "epoch": 2.7933993399339934,
+      "grad_norm": 1.4719637632369995,
+      "learning_rate": 2.965686274509804e-05,
+      "loss": 0.1325,
+      "step": 529
+    },
+    {
+      "epoch": 2.798679867986799,
+      "grad_norm": 0.979076087474823,
+      "learning_rate": 2.9607843137254905e-05,
+      "loss": 0.0692,
+      "step": 530
+    },
+    {
+      "epoch": 2.803960396039604,
+      "grad_norm": 1.507076621055603,
+      "learning_rate": 2.9558823529411766e-05,
+      "loss": 0.1275,
+      "step": 531
+    },
+    {
+      "epoch": 2.809240924092409,
+      "grad_norm": 1.6182982921600342,
+      "learning_rate": 2.950980392156863e-05,
+      "loss": 0.1528,
+      "step": 532
+    },
+    {
+      "epoch": 2.8145214521452147,
+      "grad_norm": 1.5272071361541748,
+      "learning_rate": 2.9460784313725493e-05,
+      "loss": 0.1344,
+      "step": 533
+    },
+    {
+      "epoch": 2.8198019801980196,
+      "grad_norm": 1.3018370866775513,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.101,
+      "step": 534
+    },
+    {
+      "epoch": 2.825082508250825,
+      "grad_norm": 1.2394366264343262,
+      "learning_rate": 2.9362745098039218e-05,
+      "loss": 0.0893,
+      "step": 535
+    },
+    {
+      "epoch": 2.8303630363036305,
+      "grad_norm": 1.5351279973983765,
+      "learning_rate": 2.931372549019608e-05,
+      "loss": 0.1226,
+      "step": 536
+    },
+    {
+      "epoch": 2.8356435643564355,
+      "grad_norm": 1.7062324285507202,
+      "learning_rate": 2.9264705882352945e-05,
+      "loss": 0.1363,
+      "step": 537
+    },
+    {
+      "epoch": 2.840924092409241,
+      "grad_norm": 1.666043996810913,
+      "learning_rate": 2.9215686274509806e-05,
+      "loss": 0.137,
+      "step": 538
+    },
+    {
+      "epoch": 2.8462046204620464,
+      "grad_norm": 1.3965767621994019,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.1018,
+      "step": 539
+    },
+    {
+      "epoch": 2.8514851485148514,
+      "grad_norm": 1.4355634450912476,
+      "learning_rate": 2.9117647058823534e-05,
+      "loss": 0.1009,
+      "step": 540
+    },
+    {
+      "epoch": 2.856765676567657,
+      "grad_norm": 1.369983434677124,
+      "learning_rate": 2.9068627450980394e-05,
+      "loss": 0.0964,
+      "step": 541
+    },
+    {
+      "epoch": 2.862046204620462,
+      "grad_norm": 1.5841052532196045,
+      "learning_rate": 2.9019607843137258e-05,
+      "loss": 0.148,
+      "step": 542
+    },
+    {
+      "epoch": 2.867326732673267,
+      "grad_norm": 1.360392451286316,
+      "learning_rate": 2.8970588235294122e-05,
+      "loss": 0.0744,
+      "step": 543
+    },
+    {
+      "epoch": 2.8726072607260726,
+      "grad_norm": 1.379198431968689,
+      "learning_rate": 2.8921568627450986e-05,
+      "loss": 0.1047,
+      "step": 544
+    },
+    {
+      "epoch": 2.877887788778878,
+      "grad_norm": 1.3699851036071777,
+      "learning_rate": 2.8872549019607843e-05,
+      "loss": 0.0813,
+      "step": 545
+    },
+    {
+      "epoch": 2.883168316831683,
+      "grad_norm": 1.162501335144043,
+      "learning_rate": 2.8823529411764703e-05,
+      "loss": 0.0848,
+      "step": 546
+    },
+    {
+      "epoch": 2.8884488448844885,
+      "grad_norm": 1.6580568552017212,
+      "learning_rate": 2.8774509803921567e-05,
+      "loss": 0.1252,
+      "step": 547
+    },
+    {
+      "epoch": 2.893729372937294,
+      "grad_norm": 1.8794211149215698,
+      "learning_rate": 2.872549019607843e-05,
+      "loss": 0.1581,
+      "step": 548
+    },
+    {
+      "epoch": 2.899009900990099,
+      "grad_norm": 1.7686625719070435,
+      "learning_rate": 2.8676470588235295e-05,
+      "loss": 0.172,
+      "step": 549
+    },
+    {
+      "epoch": 2.9042904290429044,
+      "grad_norm": 1.371221899986267,
+      "learning_rate": 2.8627450980392155e-05,
+      "loss": 0.0977,
+      "step": 550
+    },
+    {
+      "epoch": 2.90957095709571,
+      "grad_norm": 1.4571088552474976,
+      "learning_rate": 2.857843137254902e-05,
+      "loss": 0.1251,
+      "step": 551
+    },
+    {
+      "epoch": 2.9148514851485148,
+      "grad_norm": 1.1055774688720703,
+      "learning_rate": 2.8529411764705883e-05,
+      "loss": 0.0761,
+      "step": 552
+    },
+    {
+      "epoch": 2.92013201320132,
+      "grad_norm": 1.4305754899978638,
+      "learning_rate": 2.8480392156862744e-05,
+      "loss": 0.143,
+      "step": 553
+    },
+    {
+      "epoch": 2.9254125412541256,
+      "grad_norm": 1.3293156623840332,
+      "learning_rate": 2.8431372549019608e-05,
+      "loss": 0.1218,
+      "step": 554
+    },
+    {
+      "epoch": 2.9306930693069306,
+      "grad_norm": 1.7073545455932617,
+      "learning_rate": 2.838235294117647e-05,
+      "loss": 0.1495,
+      "step": 555
+    },
+    {
+      "epoch": 2.935973597359736,
+      "grad_norm": 1.0066826343536377,
+      "learning_rate": 2.8333333333333335e-05,
+      "loss": 0.0698,
+      "step": 556
+    },
+    {
+      "epoch": 2.9412541254125415,
+      "grad_norm": 1.5967711210250854,
+      "learning_rate": 2.8284313725490196e-05,
+      "loss": 0.1436,
+      "step": 557
+    },
+    {
+      "epoch": 2.9465346534653465,
+      "grad_norm": 1.7498648166656494,
+      "learning_rate": 2.823529411764706e-05,
+      "loss": 0.1793,
+      "step": 558
+    },
+    {
+      "epoch": 2.951815181518152,
+      "grad_norm": 1.4582027196884155,
+      "learning_rate": 2.8186274509803924e-05,
+      "loss": 0.1152,
+      "step": 559
+    },
+    {
+      "epoch": 2.9570957095709574,
+      "grad_norm": 1.1277716159820557,
+      "learning_rate": 2.8137254901960784e-05,
+      "loss": 0.0807,
+      "step": 560
+    },
+    {
+      "epoch": 2.9623762376237623,
+      "grad_norm": 1.4396110773086548,
+      "learning_rate": 2.8088235294117648e-05,
+      "loss": 0.1141,
+      "step": 561
+    },
+    {
+      "epoch": 2.9676567656765678,
+      "grad_norm": 1.3722310066223145,
+      "learning_rate": 2.8039215686274512e-05,
+      "loss": 0.0952,
+      "step": 562
+    },
+    {
+      "epoch": 2.972937293729373,
+      "grad_norm": 1.6217355728149414,
+      "learning_rate": 2.7990196078431376e-05,
+      "loss": 0.1329,
+      "step": 563
+    },
+    {
+      "epoch": 2.978217821782178,
+      "grad_norm": 1.3327069282531738,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 0.109,
+      "step": 564
+    },
+    {
+      "epoch": 2.9834983498349836,
+      "grad_norm": 1.6068836450576782,
+      "learning_rate": 2.78921568627451e-05,
+      "loss": 0.1136,
+      "step": 565
+    },
+    {
+      "epoch": 2.9887788778877886,
+      "grad_norm": 1.6134989261627197,
+      "learning_rate": 2.7843137254901964e-05,
+      "loss": 0.1452,
+      "step": 566
+    },
+    {
+      "epoch": 2.994059405940594,
+      "grad_norm": 1.301640272140503,
+      "learning_rate": 2.7794117647058824e-05,
+      "loss": 0.1692,
+      "step": 567
+    },
+    {
+      "epoch": 2.9993399339933995,
+      "grad_norm": 1.5887155532836914,
+      "learning_rate": 2.774509803921569e-05,
+      "loss": 0.1529,
+      "step": 568
+    },
+    {
+      "epoch": 3.0046204620462045,
+      "grad_norm": 1.1964720487594604,
+      "learning_rate": 2.7696078431372552e-05,
+      "loss": 0.05,
+      "step": 569
+    },
+    {
+      "epoch": 3.00990099009901,
+      "grad_norm": 1.26799476146698,
+      "learning_rate": 2.7647058823529416e-05,
+      "loss": 0.0582,
+      "step": 570
+    },
+    {
+      "epoch": 3.0151815181518153,
+      "grad_norm": 1.0527023077011108,
+      "learning_rate": 2.7598039215686277e-05,
+      "loss": 0.0661,
+      "step": 571
+    },
+    {
+      "epoch": 3.0204620462046203,
+      "grad_norm": 1.0215531587600708,
+      "learning_rate": 2.754901960784314e-05,
+      "loss": 0.0487,
+      "step": 572
+    },
+    {
+      "epoch": 3.0257425742574258,
+      "grad_norm": 0.9712955951690674,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.0533,
+      "step": 573
+    },
+    {
+      "epoch": 3.031023102310231,
+      "grad_norm": 0.7851030826568604,
+      "learning_rate": 2.7450980392156865e-05,
+      "loss": 0.0253,
+      "step": 574
+    },
+    {
+      "epoch": 3.036303630363036,
+      "grad_norm": 1.6078637838363647,
+      "learning_rate": 2.740196078431373e-05,
+      "loss": 0.0841,
+      "step": 575
+    },
+    {
+      "epoch": 3.0415841584158416,
+      "grad_norm": 1.459649920463562,
+      "learning_rate": 2.7352941176470593e-05,
+      "loss": 0.0654,
+      "step": 576
+    },
+    {
+      "epoch": 3.046864686468647,
+      "grad_norm": 1.6587159633636475,
+      "learning_rate": 2.730392156862745e-05,
+      "loss": 0.0524,
+      "step": 577
+    },
+    {
+      "epoch": 3.052145214521452,
+      "grad_norm": 1.8531984090805054,
+      "learning_rate": 2.7254901960784314e-05,
+      "loss": 0.0682,
+      "step": 578
+    },
+    {
+      "epoch": 3.0574257425742575,
+      "grad_norm": 2.0687427520751953,
+      "learning_rate": 2.7205882352941174e-05,
+      "loss": 0.0809,
+      "step": 579
+    },
+    {
+      "epoch": 3.062706270627063,
+      "grad_norm": 1.7597553730010986,
+      "learning_rate": 2.7156862745098038e-05,
+      "loss": 0.0906,
+      "step": 580
+    },
+    {
+      "epoch": 3.067986798679868,
+      "grad_norm": 1.593573808670044,
+      "learning_rate": 2.7107843137254902e-05,
+      "loss": 0.0798,
+      "step": 581
+    },
+    {
+      "epoch": 3.0732673267326733,
+      "grad_norm": 1.6800479888916016,
+      "learning_rate": 2.7058823529411766e-05,
+      "loss": 0.0769,
+      "step": 582
+    },
+    {
+      "epoch": 3.0785478547854783,
+      "grad_norm": 1.2938110828399658,
+      "learning_rate": 2.7009803921568626e-05,
+      "loss": 0.0588,
+      "step": 583
+    },
+    {
+      "epoch": 3.0838283828382838,
+      "grad_norm": 0.9052571654319763,
+      "learning_rate": 2.696078431372549e-05,
+      "loss": 0.0343,
+      "step": 584
+    },
+    {
+      "epoch": 3.089108910891089,
+      "grad_norm": 1.4364440441131592,
+      "learning_rate": 2.6911764705882354e-05,
+      "loss": 0.0659,
+      "step": 585
+    },
+    {
+      "epoch": 3.094389438943894,
+      "grad_norm": 1.2938659191131592,
+      "learning_rate": 2.6862745098039214e-05,
+      "loss": 0.0653,
+      "step": 586
+    },
+    {
+      "epoch": 3.0996699669966996,
+      "grad_norm": 1.2038064002990723,
+      "learning_rate": 2.681372549019608e-05,
+      "loss": 0.0731,
+      "step": 587
+    },
+    {
+      "epoch": 3.104950495049505,
+      "grad_norm": 1.1313140392303467,
+      "learning_rate": 2.6764705882352942e-05,
+      "loss": 0.0488,
+      "step": 588
+    },
+    {
+      "epoch": 3.11023102310231,
+      "grad_norm": 1.426941990852356,
+      "learning_rate": 2.6715686274509806e-05,
+      "loss": 0.0724,
+      "step": 589
+    },
+    {
+      "epoch": 3.1155115511551155,
+      "grad_norm": 1.415529489517212,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.0902,
+      "step": 590
+    },
+    {
+      "epoch": 3.120792079207921,
+      "grad_norm": 1.4976341724395752,
+      "learning_rate": 2.661764705882353e-05,
+      "loss": 0.0837,
+      "step": 591
+    },
+    {
+      "epoch": 3.126072607260726,
+      "grad_norm": 1.3381704092025757,
+      "learning_rate": 2.6568627450980394e-05,
+      "loss": 0.0735,
+      "step": 592
+    },
+    {
+      "epoch": 3.1313531353135313,
+      "grad_norm": 1.0349838733673096,
+      "learning_rate": 2.6519607843137255e-05,
+      "loss": 0.0482,
+      "step": 593
+    },
+    {
+      "epoch": 3.1366336633663368,
+      "grad_norm": 1.4465690851211548,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 0.0823,
+      "step": 594
+    },
+    {
+      "epoch": 3.1419141914191417,
+      "grad_norm": 1.113172173500061,
+      "learning_rate": 2.6421568627450983e-05,
+      "loss": 0.0608,
+      "step": 595
+    },
+    {
+      "epoch": 3.147194719471947,
+      "grad_norm": 1.3824833631515503,
+      "learning_rate": 2.6372549019607846e-05,
+      "loss": 0.0566,
+      "step": 596
+    },
+    {
+      "epoch": 3.1524752475247526,
+      "grad_norm": 1.2780243158340454,
+      "learning_rate": 2.6323529411764707e-05,
+      "loss": 0.0566,
+      "step": 597
+    },
+    {
+      "epoch": 3.1577557755775576,
+      "grad_norm": 1.7640819549560547,
+      "learning_rate": 2.627450980392157e-05,
+      "loss": 0.0627,
+      "step": 598
+    },
+    {
+      "epoch": 3.163036303630363,
+      "grad_norm": 1.8267886638641357,
+      "learning_rate": 2.6225490196078435e-05,
+      "loss": 0.1135,
+      "step": 599
+    },
+    {
+      "epoch": 3.1683168316831685,
+      "grad_norm": 1.4951374530792236,
+      "learning_rate": 2.6176470588235295e-05,
+      "loss": 0.0824,
+      "step": 600
+    },
+    {
+      "epoch": 3.1735973597359735,
+      "grad_norm": 1.1276224851608276,
+      "learning_rate": 2.612745098039216e-05,
+      "loss": 0.0481,
+      "step": 601
+    },
+    {
+      "epoch": 3.178877887788779,
+      "grad_norm": 1.3539289236068726,
+      "learning_rate": 2.6078431372549023e-05,
+      "loss": 0.0774,
+      "step": 602
+    },
+    {
+      "epoch": 3.1841584158415843,
+      "grad_norm": 1.1666077375411987,
+      "learning_rate": 2.6029411764705887e-05,
+      "loss": 0.0484,
+      "step": 603
+    },
+    {
+      "epoch": 3.1894389438943893,
+      "grad_norm": 1.4730373620986938,
+      "learning_rate": 2.5980392156862747e-05,
+      "loss": 0.0576,
+      "step": 604
+    },
+    {
+      "epoch": 3.1947194719471947,
+      "grad_norm": 1.3001021146774292,
+      "learning_rate": 2.593137254901961e-05,
+      "loss": 0.0657,
+      "step": 605
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.2780375480651855,
+      "learning_rate": 2.5882352941176475e-05,
+      "loss": 0.0538,
+      "step": 606
+    },
+    {
+      "epoch": 3.205280528052805,
+      "grad_norm": 1.1565543413162231,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 0.0505,
+      "step": 607
+    },
+    {
+      "epoch": 3.2105610561056106,
+      "grad_norm": 1.4991666078567505,
+      "learning_rate": 2.57843137254902e-05,
+      "loss": 0.058,
+      "step": 608
+    },
+    {
+      "epoch": 3.215841584158416,
+      "grad_norm": 1.1679130792617798,
+      "learning_rate": 2.5735294117647057e-05,
+      "loss": 0.0525,
+      "step": 609
+    },
+    {
+      "epoch": 3.221122112211221,
+      "grad_norm": 1.251451849937439,
+      "learning_rate": 2.568627450980392e-05,
+      "loss": 0.0525,
+      "step": 610
+    },
+    {
+      "epoch": 3.2264026402640265,
+      "grad_norm": 1.4389278888702393,
+      "learning_rate": 2.5637254901960784e-05,
+      "loss": 0.0776,
+      "step": 611
+    },
+    {
+      "epoch": 3.231683168316832,
+      "grad_norm": 1.2829740047454834,
+      "learning_rate": 2.5588235294117645e-05,
+      "loss": 0.0529,
+      "step": 612
+    },
+    {
+      "epoch": 3.236963696369637,
+      "grad_norm": 1.0701279640197754,
+      "learning_rate": 2.553921568627451e-05,
+      "loss": 0.0409,
+      "step": 613
+    },
+    {
+      "epoch": 3.2422442244224423,
+      "grad_norm": 1.363618016242981,
+      "learning_rate": 2.5490196078431373e-05,
+      "loss": 0.0797,
+      "step": 614
+    },
+    {
+      "epoch": 3.2475247524752477,
+      "grad_norm": 1.2213047742843628,
+      "learning_rate": 2.5441176470588236e-05,
+      "loss": 0.0621,
+      "step": 615
+    },
+    {
+      "epoch": 3.2528052805280527,
+      "grad_norm": 1.4697519540786743,
+      "learning_rate": 2.5392156862745097e-05,
+      "loss": 0.0779,
+      "step": 616
+    },
+    {
+      "epoch": 3.258085808580858,
+      "grad_norm": 1.4408804178237915,
+      "learning_rate": 2.534313725490196e-05,
+      "loss": 0.0642,
+      "step": 617
+    },
+    {
+      "epoch": 3.2633663366336636,
+      "grad_norm": 1.207322120666504,
+      "learning_rate": 2.5294117647058825e-05,
+      "loss": 0.0588,
+      "step": 618
+    },
+    {
+      "epoch": 3.2686468646864686,
+      "grad_norm": 1.399373173713684,
+      "learning_rate": 2.5245098039215685e-05,
+      "loss": 0.0789,
+      "step": 619
+    },
+    {
+      "epoch": 3.273927392739274,
+      "grad_norm": 1.1810945272445679,
+      "learning_rate": 2.519607843137255e-05,
+      "loss": 0.0574,
+      "step": 620
+    },
+    {
+      "epoch": 3.2792079207920795,
+      "grad_norm": 1.0468858480453491,
+      "learning_rate": 2.5147058823529413e-05,
+      "loss": 0.0466,
+      "step": 621
+    },
+    {
+      "epoch": 3.2844884488448844,
+      "grad_norm": 1.3326268196105957,
+      "learning_rate": 2.5098039215686277e-05,
+      "loss": 0.0674,
+      "step": 622
+    },
+    {
+      "epoch": 3.28976897689769,
+      "grad_norm": 1.3134809732437134,
+      "learning_rate": 2.5049019607843137e-05,
+      "loss": 0.0588,
+      "step": 623
+    },
+    {
+      "epoch": 3.295049504950495,
+      "grad_norm": 1.1663881540298462,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0465,
+      "step": 624
+    },
+    {
+      "epoch": 3.3003300330033003,
+      "grad_norm": 0.9541448950767517,
+      "learning_rate": 2.4950980392156865e-05,
+      "loss": 0.0434,
+      "step": 625
+    },
+    {
+      "epoch": 3.3056105610561057,
+      "grad_norm": 1.1596111059188843,
+      "learning_rate": 2.4901960784313726e-05,
+      "loss": 0.0573,
+      "step": 626
+    },
+    {
+      "epoch": 3.3108910891089107,
+      "grad_norm": 1.3220415115356445,
+      "learning_rate": 2.485294117647059e-05,
+      "loss": 0.058,
+      "step": 627
+    },
+    {
+      "epoch": 3.316171617161716,
+      "grad_norm": 1.2258095741271973,
+      "learning_rate": 2.4803921568627453e-05,
+      "loss": 0.0471,
+      "step": 628
+    },
+    {
+      "epoch": 3.3214521452145216,
+      "grad_norm": 1.2251251935958862,
+      "learning_rate": 2.4754901960784317e-05,
+      "loss": 0.0647,
+      "step": 629
+    },
+    {
+      "epoch": 3.3267326732673266,
+      "grad_norm": 1.299035906791687,
+      "learning_rate": 2.4705882352941178e-05,
+      "loss": 0.056,
+      "step": 630
+    },
+    {
+      "epoch": 3.332013201320132,
+      "grad_norm": 1.5529186725616455,
+      "learning_rate": 2.465686274509804e-05,
+      "loss": 0.084,
+      "step": 631
+    },
+    {
+      "epoch": 3.3372937293729374,
+      "grad_norm": 1.0892446041107178,
+      "learning_rate": 2.4607843137254902e-05,
+      "loss": 0.0453,
+      "step": 632
+    },
+    {
+      "epoch": 3.3425742574257424,
+      "grad_norm": 1.1661828756332397,
+      "learning_rate": 2.4558823529411766e-05,
+      "loss": 0.0485,
+      "step": 633
+    },
+    {
+      "epoch": 3.347854785478548,
+      "grad_norm": 1.0938224792480469,
+      "learning_rate": 2.4509803921568626e-05,
+      "loss": 0.0469,
+      "step": 634
+    },
+    {
+      "epoch": 3.3531353135313533,
+      "grad_norm": 1.2503447532653809,
+      "learning_rate": 2.446078431372549e-05,
+      "loss": 0.0777,
+      "step": 635
+    },
+    {
+      "epoch": 3.3584158415841583,
+      "grad_norm": 1.069814682006836,
+      "learning_rate": 2.4411764705882354e-05,
+      "loss": 0.049,
+      "step": 636
+    },
+    {
+      "epoch": 3.3636963696369637,
+      "grad_norm": 1.321007251739502,
+      "learning_rate": 2.4362745098039215e-05,
+      "loss": 0.0596,
+      "step": 637
+    },
+    {
+      "epoch": 3.368976897689769,
+      "grad_norm": 1.5562186241149902,
+      "learning_rate": 2.431372549019608e-05,
+      "loss": 0.0883,
+      "step": 638
+    },
+    {
+      "epoch": 3.374257425742574,
+      "grad_norm": 1.3883391618728638,
+      "learning_rate": 2.4264705882352942e-05,
+      "loss": 0.0777,
+      "step": 639
+    },
+    {
+      "epoch": 3.3795379537953796,
+      "grad_norm": 1.6028392314910889,
+      "learning_rate": 2.4215686274509806e-05,
+      "loss": 0.0769,
+      "step": 640
+    },
+    {
+      "epoch": 3.384818481848185,
+      "grad_norm": 1.777230978012085,
+      "learning_rate": 2.4166666666666667e-05,
+      "loss": 0.0943,
+      "step": 641
+    },
+    {
+      "epoch": 3.39009900990099,
+      "grad_norm": 1.0344568490982056,
+      "learning_rate": 2.411764705882353e-05,
+      "loss": 0.0452,
+      "step": 642
+    },
+    {
+      "epoch": 3.3953795379537954,
+      "grad_norm": 1.2247084379196167,
+      "learning_rate": 2.4068627450980395e-05,
+      "loss": 0.0516,
+      "step": 643
+    },
+    {
+      "epoch": 3.4006600660066004,
+      "grad_norm": 1.2037266492843628,
+      "learning_rate": 2.401960784313726e-05,
+      "loss": 0.0486,
+      "step": 644
+    },
+    {
+      "epoch": 3.405940594059406,
+      "grad_norm": 1.3176463842391968,
+      "learning_rate": 2.397058823529412e-05,
+      "loss": 0.0522,
+      "step": 645
+    },
+    {
+      "epoch": 3.4112211221122113,
+      "grad_norm": 1.6888436079025269,
+      "learning_rate": 2.3921568627450983e-05,
+      "loss": 0.1001,
+      "step": 646
+    },
+    {
+      "epoch": 3.4165016501650163,
+      "grad_norm": 1.296751856803894,
+      "learning_rate": 2.3872549019607847e-05,
+      "loss": 0.0656,
+      "step": 647
+    },
+    {
+      "epoch": 3.4217821782178217,
+      "grad_norm": 1.126638650894165,
+      "learning_rate": 2.3823529411764707e-05,
+      "loss": 0.0477,
+      "step": 648
+    },
+    {
+      "epoch": 3.427062706270627,
+      "grad_norm": 1.394085168838501,
+      "learning_rate": 2.3774509803921568e-05,
+      "loss": 0.0761,
+      "step": 649
+    },
+    {
+      "epoch": 3.432343234323432,
+      "grad_norm": 1.4828298091888428,
+      "learning_rate": 2.372549019607843e-05,
+      "loss": 0.0683,
+      "step": 650
+    },
+    {
+      "epoch": 3.4376237623762376,
+      "grad_norm": 1.4716849327087402,
+      "learning_rate": 2.3676470588235295e-05,
+      "loss": 0.0855,
+      "step": 651
+    },
+    {
+      "epoch": 3.442904290429043,
+      "grad_norm": 1.3064403533935547,
+      "learning_rate": 2.3627450980392156e-05,
+      "loss": 0.0644,
+      "step": 652
+    },
+    {
+      "epoch": 3.448184818481848,
+      "grad_norm": 1.4186874628067017,
+      "learning_rate": 2.357843137254902e-05,
+      "loss": 0.0556,
+      "step": 653
+    },
+    {
+      "epoch": 3.4534653465346534,
+      "grad_norm": 1.3081494569778442,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 0.0663,
+      "step": 654
+    },
+    {
+      "epoch": 3.458745874587459,
+      "grad_norm": 1.174717664718628,
+      "learning_rate": 2.3480392156862748e-05,
+      "loss": 0.0704,
+      "step": 655
+    },
+    {
+      "epoch": 3.464026402640264,
+      "grad_norm": 1.310571551322937,
+      "learning_rate": 2.3431372549019608e-05,
+      "loss": 0.0555,
+      "step": 656
+    },
+    {
+      "epoch": 3.4693069306930693,
+      "grad_norm": 1.3356480598449707,
+      "learning_rate": 2.3382352941176472e-05,
+      "loss": 0.0565,
+      "step": 657
+    },
+    {
+      "epoch": 3.4745874587458747,
+      "grad_norm": 1.5645304918289185,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.0885,
+      "step": 658
+    },
+    {
+      "epoch": 3.4798679867986797,
+      "grad_norm": 1.6264077425003052,
+      "learning_rate": 2.3284313725490196e-05,
+      "loss": 0.0936,
+      "step": 659
+    },
+    {
+      "epoch": 3.485148514851485,
+      "grad_norm": 1.1637386083602905,
+      "learning_rate": 2.323529411764706e-05,
+      "loss": 0.0499,
+      "step": 660
+    },
+    {
+      "epoch": 3.4904290429042906,
+      "grad_norm": 1.3853108882904053,
+      "learning_rate": 2.3186274509803924e-05,
+      "loss": 0.0781,
+      "step": 661
+    },
+    {
+      "epoch": 3.4957095709570956,
+      "grad_norm": 1.1978479623794556,
+      "learning_rate": 2.3137254901960788e-05,
+      "loss": 0.0524,
+      "step": 662
+    },
+    {
+      "epoch": 3.500990099009901,
+      "grad_norm": 1.1041313409805298,
+      "learning_rate": 2.308823529411765e-05,
+      "loss": 0.054,
+      "step": 663
+    },
+    {
+      "epoch": 3.5062706270627064,
+      "grad_norm": 0.9631710052490234,
+      "learning_rate": 2.303921568627451e-05,
+      "loss": 0.0393,
+      "step": 664
+    },
+    {
+      "epoch": 3.5115511551155114,
+      "grad_norm": 1.166167974472046,
+      "learning_rate": 2.2990196078431373e-05,
+      "loss": 0.0401,
+      "step": 665
+    },
+    {
+      "epoch": 3.516831683168317,
+      "grad_norm": 1.7196093797683716,
+      "learning_rate": 2.2941176470588237e-05,
+      "loss": 0.0686,
+      "step": 666
+    },
+    {
+      "epoch": 3.5221122112211223,
+      "grad_norm": 1.4491196870803833,
+      "learning_rate": 2.2892156862745097e-05,
+      "loss": 0.0721,
+      "step": 667
+    },
+    {
+      "epoch": 3.5273927392739273,
+      "grad_norm": 1.724544882774353,
+      "learning_rate": 2.284313725490196e-05,
+      "loss": 0.0789,
+      "step": 668
+    },
+    {
+      "epoch": 3.5326732673267327,
+      "grad_norm": 1.2871899604797363,
+      "learning_rate": 2.2794117647058825e-05,
+      "loss": 0.0502,
+      "step": 669
+    },
+    {
+      "epoch": 3.537953795379538,
+      "grad_norm": 1.7234700918197632,
+      "learning_rate": 2.2745098039215685e-05,
+      "loss": 0.0896,
+      "step": 670
+    },
+    {
+      "epoch": 3.543234323432343,
+      "grad_norm": 1.6124573945999146,
+      "learning_rate": 2.269607843137255e-05,
+      "loss": 0.0899,
+      "step": 671
+    },
+    {
+      "epoch": 3.5485148514851486,
+      "grad_norm": 1.1992857456207275,
+      "learning_rate": 2.2647058823529413e-05,
+      "loss": 0.0514,
+      "step": 672
+    },
+    {
+      "epoch": 3.553795379537954,
+      "grad_norm": 1.2639023065567017,
+      "learning_rate": 2.2598039215686277e-05,
+      "loss": 0.0584,
+      "step": 673
+    },
+    {
+      "epoch": 3.559075907590759,
+      "grad_norm": 1.1662899255752563,
+      "learning_rate": 2.2549019607843138e-05,
+      "loss": 0.0551,
+      "step": 674
+    },
+    {
+      "epoch": 3.5643564356435644,
+      "grad_norm": 1.27886962890625,
+      "learning_rate": 2.25e-05,
+      "loss": 0.0706,
+      "step": 675
+    },
+    {
+      "epoch": 3.56963696369637,
+      "grad_norm": 1.487029790878296,
+      "learning_rate": 2.2450980392156865e-05,
+      "loss": 0.0973,
+      "step": 676
+    },
+    {
+      "epoch": 3.574917491749175,
+      "grad_norm": 1.0966662168502808,
+      "learning_rate": 2.2401960784313726e-05,
+      "loss": 0.0483,
+      "step": 677
+    },
+    {
+      "epoch": 3.5801980198019803,
+      "grad_norm": 1.592883825302124,
+      "learning_rate": 2.235294117647059e-05,
+      "loss": 0.097,
+      "step": 678
+    },
+    {
+      "epoch": 3.5854785478547857,
+      "grad_norm": 1.1297260522842407,
+      "learning_rate": 2.2303921568627454e-05,
+      "loss": 0.052,
+      "step": 679
+    },
+    {
+      "epoch": 3.5907590759075907,
+      "grad_norm": 1.3549542427062988,
+      "learning_rate": 2.2254901960784314e-05,
+      "loss": 0.0636,
+      "step": 680
+    },
+    {
+      "epoch": 3.596039603960396,
+      "grad_norm": 1.4858061075210571,
+      "learning_rate": 2.2205882352941178e-05,
+      "loss": 0.0823,
+      "step": 681
+    },
+    {
+      "epoch": 3.6013201320132016,
+      "grad_norm": 1.2688394784927368,
+      "learning_rate": 2.215686274509804e-05,
+      "loss": 0.0621,
+      "step": 682
+    },
+    {
+      "epoch": 3.6066006600660065,
+      "grad_norm": 1.2265433073043823,
+      "learning_rate": 2.2107843137254902e-05,
+      "loss": 0.0521,
+      "step": 683
+    },
+    {
+      "epoch": 3.611881188118812,
+      "grad_norm": 1.5389267206192017,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 0.0644,
+      "step": 684
+    },
+    {
+      "epoch": 3.6171617161716174,
+      "grad_norm": 1.0617897510528564,
+      "learning_rate": 2.2009803921568627e-05,
+      "loss": 0.0487,
+      "step": 685
+    },
+    {
+      "epoch": 3.6224422442244224,
+      "grad_norm": 1.5505889654159546,
+      "learning_rate": 2.196078431372549e-05,
+      "loss": 0.0727,
+      "step": 686
+    },
+    {
+      "epoch": 3.627722772277228,
+      "grad_norm": 1.6774746179580688,
+      "learning_rate": 2.1911764705882354e-05,
+      "loss": 0.0763,
+      "step": 687
+    },
+    {
+      "epoch": 3.6330033003300333,
+      "grad_norm": 1.253771185874939,
+      "learning_rate": 2.1862745098039218e-05,
+      "loss": 0.0629,
+      "step": 688
+    },
+    {
+      "epoch": 3.6382838283828383,
+      "grad_norm": 1.324569821357727,
+      "learning_rate": 2.181372549019608e-05,
+      "loss": 0.0673,
+      "step": 689
+    },
+    {
+      "epoch": 3.6435643564356437,
+      "grad_norm": 1.7756513357162476,
+      "learning_rate": 2.1764705882352943e-05,
+      "loss": 0.1118,
+      "step": 690
+    },
+    {
+      "epoch": 3.6488448844884487,
+      "grad_norm": 1.2172956466674805,
+      "learning_rate": 2.1715686274509807e-05,
+      "loss": 0.0553,
+      "step": 691
+    },
+    {
+      "epoch": 3.654125412541254,
+      "grad_norm": 1.201130747795105,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.0654,
+      "step": 692
+    },
+    {
+      "epoch": 3.6594059405940595,
+      "grad_norm": 1.3230106830596924,
+      "learning_rate": 2.161764705882353e-05,
+      "loss": 0.0647,
+      "step": 693
+    },
+    {
+      "epoch": 3.6646864686468645,
+      "grad_norm": 1.3373692035675049,
+      "learning_rate": 2.1568627450980395e-05,
+      "loss": 0.0691,
+      "step": 694
+    },
+    {
+      "epoch": 3.66996699669967,
+      "grad_norm": 1.423130750656128,
+      "learning_rate": 2.151960784313726e-05,
+      "loss": 0.0683,
+      "step": 695
+    },
+    {
+      "epoch": 3.6752475247524754,
+      "grad_norm": 1.2763397693634033,
+      "learning_rate": 2.1470588235294116e-05,
+      "loss": 0.0589,
+      "step": 696
+    },
+    {
+      "epoch": 3.6805280528052804,
+      "grad_norm": 1.749027967453003,
+      "learning_rate": 2.142156862745098e-05,
+      "loss": 0.0918,
+      "step": 697
+    },
+    {
+      "epoch": 3.685808580858086,
+      "grad_norm": 1.3163336515426636,
+      "learning_rate": 2.1372549019607844e-05,
+      "loss": 0.0585,
+      "step": 698
+    },
+    {
+      "epoch": 3.691089108910891,
+      "grad_norm": 1.6769047975540161,
+      "learning_rate": 2.1323529411764707e-05,
+      "loss": 0.0967,
+      "step": 699
+    },
+    {
+      "epoch": 3.6963696369636962,
+      "grad_norm": 1.2727911472320557,
+      "learning_rate": 2.1274509803921568e-05,
+      "loss": 0.0713,
+      "step": 700
+    },
+    {
+      "epoch": 3.7016501650165017,
+      "grad_norm": 1.4433225393295288,
+      "learning_rate": 2.1225490196078432e-05,
+      "loss": 0.0783,
+      "step": 701
+    },
+    {
+      "epoch": 3.7069306930693067,
+      "grad_norm": 1.3980076313018799,
+      "learning_rate": 2.1176470588235296e-05,
+      "loss": 0.0677,
+      "step": 702
+    },
+    {
+      "epoch": 3.712211221122112,
+      "grad_norm": 1.3446606397628784,
+      "learning_rate": 2.1127450980392156e-05,
+      "loss": 0.0746,
+      "step": 703
+    },
+    {
+      "epoch": 3.7174917491749175,
+      "grad_norm": 1.2781853675842285,
+      "learning_rate": 2.107843137254902e-05,
+      "loss": 0.0545,
+      "step": 704
+    },
+    {
+      "epoch": 3.7227722772277225,
+      "grad_norm": 1.4710532426834106,
+      "learning_rate": 2.1029411764705884e-05,
+      "loss": 0.0612,
+      "step": 705
+    },
+    {
+      "epoch": 3.728052805280528,
+      "grad_norm": 1.2814794778823853,
+      "learning_rate": 2.0980392156862748e-05,
+      "loss": 0.0718,
+      "step": 706
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 1.4661047458648682,
+      "learning_rate": 2.0931372549019608e-05,
+      "loss": 0.0791,
+      "step": 707
+    },
+    {
+      "epoch": 3.7386138613861384,
+      "grad_norm": 1.37482750415802,
+      "learning_rate": 2.0882352941176472e-05,
+      "loss": 0.0737,
+      "step": 708
+    },
+    {
+      "epoch": 3.743894389438944,
+      "grad_norm": 1.3684672117233276,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.0743,
+      "step": 709
+    },
+    {
+      "epoch": 3.7491749174917492,
+      "grad_norm": 0.859553873538971,
+      "learning_rate": 2.0784313725490197e-05,
+      "loss": 0.0356,
+      "step": 710
+    },
+    {
+      "epoch": 3.7544554455445542,
+      "grad_norm": 1.2247698307037354,
+      "learning_rate": 2.073529411764706e-05,
+      "loss": 0.0671,
+      "step": 711
+    },
+    {
+      "epoch": 3.7597359735973597,
+      "grad_norm": 1.2793521881103516,
+      "learning_rate": 2.068627450980392e-05,
+      "loss": 0.0692,
+      "step": 712
+    },
+    {
+      "epoch": 3.765016501650165,
+      "grad_norm": 1.26449453830719,
+      "learning_rate": 2.0637254901960785e-05,
+      "loss": 0.0622,
+      "step": 713
+    },
+    {
+      "epoch": 3.77029702970297,
+      "grad_norm": 1.1848664283752441,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 0.0535,
+      "step": 714
+    },
+    {
+      "epoch": 3.7755775577557755,
+      "grad_norm": 1.034454107284546,
+      "learning_rate": 2.053921568627451e-05,
+      "loss": 0.0403,
+      "step": 715
+    },
+    {
+      "epoch": 3.780858085808581,
+      "grad_norm": 1.005811333656311,
+      "learning_rate": 2.0490196078431373e-05,
+      "loss": 0.0489,
+      "step": 716
+    },
+    {
+      "epoch": 3.786138613861386,
+      "grad_norm": 1.294252872467041,
+      "learning_rate": 2.0441176470588237e-05,
+      "loss": 0.0599,
+      "step": 717
+    },
+    {
+      "epoch": 3.7914191419141914,
+      "grad_norm": 1.2683149576187134,
+      "learning_rate": 2.0392156862745097e-05,
+      "loss": 0.0527,
+      "step": 718
+    },
+    {
+      "epoch": 3.796699669966997,
+      "grad_norm": 1.4859176874160767,
+      "learning_rate": 2.034313725490196e-05,
+      "loss": 0.0748,
+      "step": 719
+    },
+    {
+      "epoch": 3.801980198019802,
+      "grad_norm": 1.5484555959701538,
+      "learning_rate": 2.0294117647058825e-05,
+      "loss": 0.0694,
+      "step": 720
+    },
+    {
+      "epoch": 3.8072607260726072,
+      "grad_norm": 1.0861327648162842,
+      "learning_rate": 2.024509803921569e-05,
+      "loss": 0.0489,
+      "step": 721
+    },
+    {
+      "epoch": 3.8125412541254127,
+      "grad_norm": 2.1083528995513916,
+      "learning_rate": 2.019607843137255e-05,
+      "loss": 0.0996,
+      "step": 722
+    },
+    {
+      "epoch": 3.8178217821782177,
+      "grad_norm": 1.1636090278625488,
+      "learning_rate": 2.0147058823529413e-05,
+      "loss": 0.0439,
+      "step": 723
+    },
+    {
+      "epoch": 3.823102310231023,
+      "grad_norm": 1.4956183433532715,
+      "learning_rate": 2.0098039215686277e-05,
+      "loss": 0.0771,
+      "step": 724
+    },
+    {
+      "epoch": 3.8283828382838285,
+      "grad_norm": 1.4790761470794678,
+      "learning_rate": 2.0049019607843138e-05,
+      "loss": 0.0561,
+      "step": 725
+    },
+    {
+      "epoch": 3.8336633663366335,
+      "grad_norm": 1.4751077890396118,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 726
+    },
+    {
+      "epoch": 3.838943894389439,
+      "grad_norm": 1.2702194452285767,
+      "learning_rate": 1.9950980392156866e-05,
+      "loss": 0.07,
+      "step": 727
+    },
+    {
+      "epoch": 3.8442244224422444,
+      "grad_norm": 0.9263429045677185,
+      "learning_rate": 1.9901960784313726e-05,
+      "loss": 0.0359,
+      "step": 728
+    },
+    {
+      "epoch": 3.8495049504950494,
+      "grad_norm": 1.4288864135742188,
+      "learning_rate": 1.9852941176470586e-05,
+      "loss": 0.0869,
+      "step": 729
+    },
+    {
+      "epoch": 3.854785478547855,
+      "grad_norm": 1.3054994344711304,
+      "learning_rate": 1.980392156862745e-05,
+      "loss": 0.0755,
+      "step": 730
+    },
+    {
+      "epoch": 3.8600660066006602,
+      "grad_norm": 1.422020435333252,
+      "learning_rate": 1.9754901960784314e-05,
+      "loss": 0.072,
+      "step": 731
+    },
+    {
+      "epoch": 3.8653465346534652,
+      "grad_norm": 1.2863701581954956,
+      "learning_rate": 1.9705882352941178e-05,
+      "loss": 0.0676,
+      "step": 732
+    },
+    {
+      "epoch": 3.8706270627062707,
+      "grad_norm": 1.4083125591278076,
+      "learning_rate": 1.965686274509804e-05,
+      "loss": 0.0805,
+      "step": 733
+    },
+    {
+      "epoch": 3.875907590759076,
+      "grad_norm": 1.275930404663086,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 0.0737,
+      "step": 734
+    },
+    {
+      "epoch": 3.881188118811881,
+      "grad_norm": 1.2777963876724243,
+      "learning_rate": 1.9558823529411766e-05,
+      "loss": 0.0674,
+      "step": 735
+    },
+    {
+      "epoch": 3.8864686468646865,
+      "grad_norm": 1.145695686340332,
+      "learning_rate": 1.9509803921568627e-05,
+      "loss": 0.0593,
+      "step": 736
+    },
+    {
+      "epoch": 3.891749174917492,
+      "grad_norm": 1.13607919216156,
+      "learning_rate": 1.946078431372549e-05,
+      "loss": 0.0481,
+      "step": 737
+    },
+    {
+      "epoch": 3.897029702970297,
+      "grad_norm": 1.5105829238891602,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 0.1056,
+      "step": 738
+    },
+    {
+      "epoch": 3.9023102310231024,
+      "grad_norm": 1.080869197845459,
+      "learning_rate": 1.936274509803922e-05,
+      "loss": 0.0469,
+      "step": 739
+    },
+    {
+      "epoch": 3.907590759075908,
+      "grad_norm": 1.6012327671051025,
+      "learning_rate": 1.931372549019608e-05,
+      "loss": 0.0883,
+      "step": 740
+    },
+    {
+      "epoch": 3.912871287128713,
+      "grad_norm": 1.3881632089614868,
+      "learning_rate": 1.9264705882352943e-05,
+      "loss": 0.0736,
+      "step": 741
+    },
+    {
+      "epoch": 3.9181518151815182,
+      "grad_norm": 1.2698092460632324,
+      "learning_rate": 1.9215686274509807e-05,
+      "loss": 0.061,
+      "step": 742
+    },
+    {
+      "epoch": 3.9234323432343237,
+      "grad_norm": 1.7859208583831787,
+      "learning_rate": 1.9166666666666667e-05,
+      "loss": 0.0981,
+      "step": 743
+    },
+    {
+      "epoch": 3.9287128712871286,
+      "grad_norm": 1.1902238130569458,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 0.0489,
+      "step": 744
+    },
+    {
+      "epoch": 3.933993399339934,
+      "grad_norm": 1.3295652866363525,
+      "learning_rate": 1.906862745098039e-05,
+      "loss": 0.0634,
+      "step": 745
+    },
+    {
+      "epoch": 3.9392739273927395,
+      "grad_norm": 1.2844423055648804,
+      "learning_rate": 1.9019607843137255e-05,
+      "loss": 0.0759,
+      "step": 746
+    },
+    {
+      "epoch": 3.9445544554455445,
+      "grad_norm": 1.062388300895691,
+      "learning_rate": 1.897058823529412e-05,
+      "loss": 0.0449,
+      "step": 747
+    },
+    {
+      "epoch": 3.94983498349835,
+      "grad_norm": 1.51730477809906,
+      "learning_rate": 1.892156862745098e-05,
+      "loss": 0.1383,
+      "step": 748
+    },
+    {
+      "epoch": 3.9551155115511554,
+      "grad_norm": 1.5683430433273315,
+      "learning_rate": 1.8872549019607844e-05,
+      "loss": 0.0915,
+      "step": 749
+    },
+    {
+      "epoch": 3.9603960396039604,
+      "grad_norm": 1.20012366771698,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.057,
+      "step": 750
+    },
+    {
+      "epoch": 3.965676567656766,
+      "grad_norm": 1.351365089416504,
+      "learning_rate": 1.8774509803921568e-05,
+      "loss": 0.0663,
+      "step": 751
+    },
+    {
+      "epoch": 3.9709570957095712,
+      "grad_norm": 1.5210295915603638,
+      "learning_rate": 1.8725490196078432e-05,
+      "loss": 0.0832,
+      "step": 752
+    },
+    {
+      "epoch": 3.976237623762376,
+      "grad_norm": 1.5740433931350708,
+      "learning_rate": 1.8676470588235296e-05,
+      "loss": 0.0951,
+      "step": 753
+    },
+    {
+      "epoch": 3.9815181518151816,
+      "grad_norm": 1.2916804552078247,
+      "learning_rate": 1.862745098039216e-05,
+      "loss": 0.0638,
+      "step": 754
+    },
+    {
+      "epoch": 3.9867986798679866,
+      "grad_norm": 1.4912750720977783,
+      "learning_rate": 1.857843137254902e-05,
+      "loss": 0.0718,
+      "step": 755
+    },
+    {
+      "epoch": 3.992079207920792,
+      "grad_norm": 1.2898180484771729,
+      "learning_rate": 1.8529411764705884e-05,
+      "loss": 0.0672,
+      "step": 756
+    },
+    {
+      "epoch": 3.9973597359735975,
+      "grad_norm": 0.9569465517997742,
+      "learning_rate": 1.8480392156862748e-05,
+      "loss": 0.0425,
+      "step": 757
+    },
+    {
+      "epoch": 4.002640264026403,
+      "grad_norm": 1.1869324445724487,
+      "learning_rate": 1.843137254901961e-05,
+      "loss": 0.0527,
+      "step": 758
+    },
+    {
+      "epoch": 4.007920792079208,
+      "grad_norm": 0.7702716588973999,
+      "learning_rate": 1.8382352941176472e-05,
+      "loss": 0.0358,
+      "step": 759
+    },
+    {
+      "epoch": 4.013201320132013,
+      "grad_norm": 0.6784669160842896,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.0268,
+      "step": 760
+    },
+    {
+      "epoch": 4.018481848184819,
+      "grad_norm": 0.8795797228813171,
+      "learning_rate": 1.8284313725490197e-05,
+      "loss": 0.0363,
+      "step": 761
+    },
+    {
+      "epoch": 4.023762376237624,
+      "grad_norm": 0.9152675271034241,
+      "learning_rate": 1.8235294117647057e-05,
+      "loss": 0.0303,
+      "step": 762
+    },
+    {
+      "epoch": 4.029042904290429,
+      "grad_norm": 0.685607373714447,
+      "learning_rate": 1.818627450980392e-05,
+      "loss": 0.0249,
+      "step": 763
+    },
+    {
+      "epoch": 4.034323432343235,
+      "grad_norm": 0.8284955620765686,
+      "learning_rate": 1.8137254901960785e-05,
+      "loss": 0.0257,
+      "step": 764
+    },
+    {
+      "epoch": 4.03960396039604,
+      "grad_norm": 0.8242619037628174,
+      "learning_rate": 1.808823529411765e-05,
+      "loss": 0.027,
+      "step": 765
+    },
+    {
+      "epoch": 4.044884488448845,
+      "grad_norm": 1.0114595890045166,
+      "learning_rate": 1.803921568627451e-05,
+      "loss": 0.0222,
+      "step": 766
+    },
+    {
+      "epoch": 4.0501650165016505,
+      "grad_norm": 0.6319472789764404,
+      "learning_rate": 1.7990196078431373e-05,
+      "loss": 0.0182,
+      "step": 767
+    },
+    {
+      "epoch": 4.0554455445544555,
+      "grad_norm": 0.9475066661834717,
+      "learning_rate": 1.7941176470588237e-05,
+      "loss": 0.0269,
+      "step": 768
+    },
+    {
+      "epoch": 4.0607260726072605,
+      "grad_norm": 1.0184468030929565,
+      "learning_rate": 1.7892156862745098e-05,
+      "loss": 0.0258,
+      "step": 769
+    },
+    {
+      "epoch": 4.066006600660066,
+      "grad_norm": 1.226219654083252,
+      "learning_rate": 1.784313725490196e-05,
+      "loss": 0.036,
+      "step": 770
+    },
+    {
+      "epoch": 4.071287128712871,
+      "grad_norm": 0.839829683303833,
+      "learning_rate": 1.7794117647058825e-05,
+      "loss": 0.027,
+      "step": 771
+    },
+    {
+      "epoch": 4.076567656765676,
+      "grad_norm": 1.2068742513656616,
+      "learning_rate": 1.774509803921569e-05,
+      "loss": 0.0345,
+      "step": 772
+    },
+    {
+      "epoch": 4.081848184818482,
+      "grad_norm": 1.1739568710327148,
+      "learning_rate": 1.769607843137255e-05,
+      "loss": 0.0334,
+      "step": 773
+    },
+    {
+      "epoch": 4.087128712871287,
+      "grad_norm": 1.167466640472412,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.0308,
+      "step": 774
+    },
+    {
+      "epoch": 4.092409240924092,
+      "grad_norm": 1.0854191780090332,
+      "learning_rate": 1.7598039215686277e-05,
+      "loss": 0.0288,
+      "step": 775
+    },
+    {
+      "epoch": 4.097689768976898,
+      "grad_norm": 0.8785208463668823,
+      "learning_rate": 1.7549019607843138e-05,
+      "loss": 0.0298,
+      "step": 776
+    },
+    {
+      "epoch": 4.102970297029703,
+      "grad_norm": 1.252079725265503,
+      "learning_rate": 1.75e-05,
+      "loss": 0.044,
+      "step": 777
+    },
+    {
+      "epoch": 4.108250825082508,
+      "grad_norm": 1.1048779487609863,
+      "learning_rate": 1.7450980392156862e-05,
+      "loss": 0.0285,
+      "step": 778
+    },
+    {
+      "epoch": 4.113531353135314,
+      "grad_norm": 1.3396317958831787,
+      "learning_rate": 1.7401960784313726e-05,
+      "loss": 0.0457,
+      "step": 779
+    },
+    {
+      "epoch": 4.118811881188119,
+      "grad_norm": 1.165590763092041,
+      "learning_rate": 1.735294117647059e-05,
+      "loss": 0.0319,
+      "step": 780
+    },
+    {
+      "epoch": 4.124092409240924,
+      "grad_norm": 0.9108593463897705,
+      "learning_rate": 1.730392156862745e-05,
+      "loss": 0.0286,
+      "step": 781
+    },
+    {
+      "epoch": 4.12937293729373,
+      "grad_norm": 0.9101956486701965,
+      "learning_rate": 1.7254901960784314e-05,
+      "loss": 0.0291,
+      "step": 782
+    },
+    {
+      "epoch": 4.134653465346535,
+      "grad_norm": 1.055116891860962,
+      "learning_rate": 1.720588235294118e-05,
+      "loss": 0.0326,
+      "step": 783
+    },
+    {
+      "epoch": 4.13993399339934,
+      "grad_norm": 0.9137353897094727,
+      "learning_rate": 1.715686274509804e-05,
+      "loss": 0.0312,
+      "step": 784
+    },
+    {
+      "epoch": 4.145214521452146,
+      "grad_norm": 0.9962388277053833,
+      "learning_rate": 1.7107843137254903e-05,
+      "loss": 0.0317,
+      "step": 785
+    },
+    {
+      "epoch": 4.150495049504951,
+      "grad_norm": 0.891434907913208,
+      "learning_rate": 1.7058823529411767e-05,
+      "loss": 0.0357,
+      "step": 786
+    },
+    {
+      "epoch": 4.155775577557756,
+      "grad_norm": 0.9022417068481445,
+      "learning_rate": 1.700980392156863e-05,
+      "loss": 0.0346,
+      "step": 787
+    },
+    {
+      "epoch": 4.161056105610561,
+      "grad_norm": 0.886080265045166,
+      "learning_rate": 1.696078431372549e-05,
+      "loss": 0.0308,
+      "step": 788
+    },
+    {
+      "epoch": 4.1663366336633665,
+      "grad_norm": 1.155617356300354,
+      "learning_rate": 1.6911764705882355e-05,
+      "loss": 0.0904,
+      "step": 789
+    },
+    {
+      "epoch": 4.1716171617161715,
+      "grad_norm": 1.101927399635315,
+      "learning_rate": 1.686274509803922e-05,
+      "loss": 0.0289,
+      "step": 790
+    },
+    {
+      "epoch": 4.1768976897689765,
+      "grad_norm": 1.2381272315979004,
+      "learning_rate": 1.681372549019608e-05,
+      "loss": 0.0473,
+      "step": 791
+    },
+    {
+      "epoch": 4.182178217821782,
+      "grad_norm": 1.2841603755950928,
+      "learning_rate": 1.676470588235294e-05,
+      "loss": 0.0392,
+      "step": 792
+    },
+    {
+      "epoch": 4.187458745874587,
+      "grad_norm": 0.9901617765426636,
+      "learning_rate": 1.6715686274509804e-05,
+      "loss": 0.0274,
+      "step": 793
+    },
+    {
+      "epoch": 4.192739273927392,
+      "grad_norm": 1.011318325996399,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.0329,
+      "step": 794
+    },
+    {
+      "epoch": 4.198019801980198,
+      "grad_norm": 0.8173012733459473,
+      "learning_rate": 1.6617647058823528e-05,
+      "loss": 0.0272,
+      "step": 795
+    },
+    {
+      "epoch": 4.203300330033003,
+      "grad_norm": 1.3152134418487549,
+      "learning_rate": 1.6568627450980392e-05,
+      "loss": 0.0309,
+      "step": 796
+    },
+    {
+      "epoch": 4.208580858085808,
+      "grad_norm": 0.9203322529792786,
+      "learning_rate": 1.6519607843137256e-05,
+      "loss": 0.0287,
+      "step": 797
+    },
+    {
+      "epoch": 4.213861386138614,
+      "grad_norm": 0.8306743502616882,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.0291,
+      "step": 798
+    },
+    {
+      "epoch": 4.219141914191419,
+      "grad_norm": 1.1384245157241821,
+      "learning_rate": 1.642156862745098e-05,
+      "loss": 0.0295,
+      "step": 799
+    },
+    {
+      "epoch": 4.224422442244224,
+      "grad_norm": 1.0017738342285156,
+      "learning_rate": 1.6372549019607844e-05,
+      "loss": 0.0347,
+      "step": 800
+    },
+    {
+      "epoch": 4.22970297029703,
+      "grad_norm": 0.7871435284614563,
+      "learning_rate": 1.6323529411764708e-05,
+      "loss": 0.0277,
+      "step": 801
+    },
+    {
+      "epoch": 4.234983498349835,
+      "grad_norm": 0.9366597533226013,
+      "learning_rate": 1.627450980392157e-05,
+      "loss": 0.0323,
+      "step": 802
+    },
+    {
+      "epoch": 4.24026402640264,
+      "grad_norm": 0.9586142301559448,
+      "learning_rate": 1.6225490196078432e-05,
+      "loss": 0.0212,
+      "step": 803
+    },
+    {
+      "epoch": 4.245544554455446,
+      "grad_norm": 1.21640145778656,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 0.0357,
+      "step": 804
+    },
+    {
+      "epoch": 4.250825082508251,
+      "grad_norm": 0.9384037852287292,
+      "learning_rate": 1.612745098039216e-05,
+      "loss": 0.0315,
+      "step": 805
+    },
+    {
+      "epoch": 4.256105610561056,
+      "grad_norm": 1.0717084407806396,
+      "learning_rate": 1.607843137254902e-05,
+      "loss": 0.0337,
+      "step": 806
+    },
+    {
+      "epoch": 4.261386138613862,
+      "grad_norm": 0.9708730578422546,
+      "learning_rate": 1.6029411764705884e-05,
+      "loss": 0.0319,
+      "step": 807
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 0.7535268068313599,
+      "learning_rate": 1.5980392156862745e-05,
+      "loss": 0.0294,
+      "step": 808
+    },
+    {
+      "epoch": 4.271947194719472,
+      "grad_norm": 1.0513904094696045,
+      "learning_rate": 1.593137254901961e-05,
+      "loss": 0.0262,
+      "step": 809
+    },
+    {
+      "epoch": 4.2772277227722775,
+      "grad_norm": 1.4030704498291016,
+      "learning_rate": 1.588235294117647e-05,
+      "loss": 0.0436,
+      "step": 810
+    },
+    {
+      "epoch": 4.2825082508250825,
+      "grad_norm": 0.5738725662231445,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.0201,
+      "step": 811
+    },
+    {
+      "epoch": 4.2877887788778875,
+      "grad_norm": 1.305862307548523,
+      "learning_rate": 1.5784313725490197e-05,
+      "loss": 0.043,
+      "step": 812
+    },
+    {
+      "epoch": 4.293069306930693,
+      "grad_norm": 0.9697719812393188,
+      "learning_rate": 1.573529411764706e-05,
+      "loss": 0.0306,
+      "step": 813
+    },
+    {
+      "epoch": 4.298349834983498,
+      "grad_norm": 0.8352718353271484,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 0.0269,
+      "step": 814
+    },
+    {
+      "epoch": 4.303630363036303,
+      "grad_norm": 1.000013828277588,
+      "learning_rate": 1.5637254901960785e-05,
+      "loss": 0.0272,
+      "step": 815
+    },
+    {
+      "epoch": 4.308910891089109,
+      "grad_norm": 0.6394527554512024,
+      "learning_rate": 1.558823529411765e-05,
+      "loss": 0.0224,
+      "step": 816
+    },
+    {
+      "epoch": 4.314191419141914,
+      "grad_norm": 0.9506754875183105,
+      "learning_rate": 1.553921568627451e-05,
+      "loss": 0.0293,
+      "step": 817
+    },
+    {
+      "epoch": 4.319471947194719,
+      "grad_norm": 0.9406654834747314,
+      "learning_rate": 1.5490196078431373e-05,
+      "loss": 0.0261,
+      "step": 818
+    },
+    {
+      "epoch": 4.324752475247525,
+      "grad_norm": 1.1264249086380005,
+      "learning_rate": 1.5441176470588237e-05,
+      "loss": 0.0347,
+      "step": 819
+    },
+    {
+      "epoch": 4.33003300330033,
+      "grad_norm": 0.9461225271224976,
+      "learning_rate": 1.53921568627451e-05,
+      "loss": 0.0261,
+      "step": 820
+    },
+    {
+      "epoch": 4.335313531353135,
+      "grad_norm": 1.1620982885360718,
+      "learning_rate": 1.5343137254901962e-05,
+      "loss": 0.0309,
+      "step": 821
+    },
+    {
+      "epoch": 4.340594059405941,
+      "grad_norm": 0.8539828658103943,
+      "learning_rate": 1.5294117647058826e-05,
+      "loss": 0.028,
+      "step": 822
+    },
+    {
+      "epoch": 4.345874587458746,
+      "grad_norm": 1.1822220087051392,
+      "learning_rate": 1.5245098039215688e-05,
+      "loss": 0.0352,
+      "step": 823
+    },
+    {
+      "epoch": 4.351155115511551,
+      "grad_norm": 1.000964641571045,
+      "learning_rate": 1.5196078431372548e-05,
+      "loss": 0.0325,
+      "step": 824
+    },
+    {
+      "epoch": 4.356435643564357,
+      "grad_norm": 1.211593508720398,
+      "learning_rate": 1.5147058823529412e-05,
+      "loss": 0.0394,
+      "step": 825
+    },
+    {
+      "epoch": 4.361716171617162,
+      "grad_norm": 1.0759446620941162,
+      "learning_rate": 1.5098039215686274e-05,
+      "loss": 0.0299,
+      "step": 826
+    },
+    {
+      "epoch": 4.366996699669967,
+      "grad_norm": 0.7947413921356201,
+      "learning_rate": 1.5049019607843137e-05,
+      "loss": 0.0269,
+      "step": 827
+    },
+    {
+      "epoch": 4.372277227722773,
+      "grad_norm": 1.283584713935852,
+      "learning_rate": 1.5e-05,
+      "loss": 0.0396,
+      "step": 828
+    },
+    {
+      "epoch": 4.377557755775578,
+      "grad_norm": 1.0424344539642334,
+      "learning_rate": 1.4950980392156863e-05,
+      "loss": 0.0303,
+      "step": 829
+    },
+    {
+      "epoch": 4.382838283828383,
+      "grad_norm": 1.1539040803909302,
+      "learning_rate": 1.4901960784313726e-05,
+      "loss": 0.0369,
+      "step": 830
+    },
+    {
+      "epoch": 4.3881188118811885,
+      "grad_norm": 0.9597378373146057,
+      "learning_rate": 1.4852941176470589e-05,
+      "loss": 0.0341,
+      "step": 831
+    },
+    {
+      "epoch": 4.3933993399339935,
+      "grad_norm": 1.133381724357605,
+      "learning_rate": 1.4803921568627453e-05,
+      "loss": 0.032,
+      "step": 832
+    },
+    {
+      "epoch": 4.398679867986798,
+      "grad_norm": 0.8584814071655273,
+      "learning_rate": 1.4754901960784315e-05,
+      "loss": 0.0274,
+      "step": 833
+    },
+    {
+      "epoch": 4.403960396039604,
+      "grad_norm": 0.7152499556541443,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 0.0228,
+      "step": 834
+    },
+    {
+      "epoch": 4.409240924092409,
+      "grad_norm": 1.0020297765731812,
+      "learning_rate": 1.465686274509804e-05,
+      "loss": 0.0405,
+      "step": 835
+    },
+    {
+      "epoch": 4.414521452145214,
+      "grad_norm": 1.236913800239563,
+      "learning_rate": 1.4607843137254903e-05,
+      "loss": 0.0399,
+      "step": 836
+    },
+    {
+      "epoch": 4.41980198019802,
+      "grad_norm": 1.0684503316879272,
+      "learning_rate": 1.4558823529411767e-05,
+      "loss": 0.0375,
+      "step": 837
+    },
+    {
+      "epoch": 4.425082508250825,
+      "grad_norm": 1.2516707181930542,
+      "learning_rate": 1.4509803921568629e-05,
+      "loss": 0.0379,
+      "step": 838
+    },
+    {
+      "epoch": 4.43036303630363,
+      "grad_norm": 1.063377857208252,
+      "learning_rate": 1.4460784313725493e-05,
+      "loss": 0.0389,
+      "step": 839
+    },
+    {
+      "epoch": 4.435643564356436,
+      "grad_norm": 0.740829586982727,
+      "learning_rate": 1.4411764705882352e-05,
+      "loss": 0.0258,
+      "step": 840
+    },
+    {
+      "epoch": 4.440924092409241,
+      "grad_norm": 0.9484118223190308,
+      "learning_rate": 1.4362745098039216e-05,
+      "loss": 0.0339,
+      "step": 841
+    },
+    {
+      "epoch": 4.446204620462046,
+      "grad_norm": 1.024625539779663,
+      "learning_rate": 1.4313725490196078e-05,
+      "loss": 0.0329,
+      "step": 842
+    },
+    {
+      "epoch": 4.451485148514852,
+      "grad_norm": 0.8846271634101868,
+      "learning_rate": 1.4264705882352942e-05,
+      "loss": 0.0293,
+      "step": 843
+    },
+    {
+      "epoch": 4.456765676567657,
+      "grad_norm": 0.6733059883117676,
+      "learning_rate": 1.4215686274509804e-05,
+      "loss": 0.0224,
+      "step": 844
+    },
+    {
+      "epoch": 4.462046204620462,
+      "grad_norm": 1.042506456375122,
+      "learning_rate": 1.4166666666666668e-05,
+      "loss": 0.0273,
+      "step": 845
+    },
+    {
+      "epoch": 4.467326732673268,
+      "grad_norm": 0.9040712118148804,
+      "learning_rate": 1.411764705882353e-05,
+      "loss": 0.0293,
+      "step": 846
+    },
+    {
+      "epoch": 4.472607260726073,
+      "grad_norm": 0.8816120028495789,
+      "learning_rate": 1.4068627450980392e-05,
+      "loss": 0.0264,
+      "step": 847
+    },
+    {
+      "epoch": 4.477887788778878,
+      "grad_norm": 0.9503970146179199,
+      "learning_rate": 1.4019607843137256e-05,
+      "loss": 0.0386,
+      "step": 848
+    },
+    {
+      "epoch": 4.483168316831684,
+      "grad_norm": 0.6163874268531799,
+      "learning_rate": 1.3970588235294118e-05,
+      "loss": 0.0179,
+      "step": 849
+    },
+    {
+      "epoch": 4.488448844884489,
+      "grad_norm": 1.139955759048462,
+      "learning_rate": 1.3921568627450982e-05,
+      "loss": 0.0364,
+      "step": 850
+    },
+    {
+      "epoch": 4.493729372937294,
+      "grad_norm": 0.8715453147888184,
+      "learning_rate": 1.3872549019607844e-05,
+      "loss": 0.0254,
+      "step": 851
+    },
+    {
+      "epoch": 4.4990099009900995,
+      "grad_norm": 0.9393417835235596,
+      "learning_rate": 1.3823529411764708e-05,
+      "loss": 0.033,
+      "step": 852
+    },
+    {
+      "epoch": 4.504290429042904,
+      "grad_norm": 1.1215251684188843,
+      "learning_rate": 1.377450980392157e-05,
+      "loss": 0.0357,
+      "step": 853
+    },
+    {
+      "epoch": 4.509570957095709,
+      "grad_norm": 0.9020299911499023,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 0.0257,
+      "step": 854
+    },
+    {
+      "epoch": 4.514851485148515,
+      "grad_norm": 1.0256519317626953,
+      "learning_rate": 1.3676470588235296e-05,
+      "loss": 0.0271,
+      "step": 855
+    },
+    {
+      "epoch": 4.52013201320132,
+      "grad_norm": 0.7362510561943054,
+      "learning_rate": 1.3627450980392157e-05,
+      "loss": 0.0214,
+      "step": 856
+    },
+    {
+      "epoch": 4.525412541254125,
+      "grad_norm": 1.1379399299621582,
+      "learning_rate": 1.3578431372549019e-05,
+      "loss": 0.0337,
+      "step": 857
+    },
+    {
+      "epoch": 4.530693069306931,
+      "grad_norm": 1.2522259950637817,
+      "learning_rate": 1.3529411764705883e-05,
+      "loss": 0.0401,
+      "step": 858
+    },
+    {
+      "epoch": 4.535973597359736,
+      "grad_norm": 1.050032138824463,
+      "learning_rate": 1.3480392156862745e-05,
+      "loss": 0.0346,
+      "step": 859
+    },
+    {
+      "epoch": 4.541254125412541,
+      "grad_norm": 0.9979643821716309,
+      "learning_rate": 1.3431372549019607e-05,
+      "loss": 0.0283,
+      "step": 860
+    },
+    {
+      "epoch": 4.546534653465347,
+      "grad_norm": 0.9979017972946167,
+      "learning_rate": 1.3382352941176471e-05,
+      "loss": 0.0302,
+      "step": 861
+    },
+    {
+      "epoch": 4.551815181518152,
+      "grad_norm": 0.9042947292327881,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0299,
+      "step": 862
+    },
+    {
+      "epoch": 4.557095709570957,
+      "grad_norm": 1.0963839292526245,
+      "learning_rate": 1.3284313725490197e-05,
+      "loss": 0.0331,
+      "step": 863
+    },
+    {
+      "epoch": 4.562376237623763,
+      "grad_norm": 0.9561034440994263,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 0.025,
+      "step": 864
+    },
+    {
+      "epoch": 4.567656765676568,
+      "grad_norm": 1.147443175315857,
+      "learning_rate": 1.3186274509803923e-05,
+      "loss": 0.0363,
+      "step": 865
+    },
+    {
+      "epoch": 4.572937293729373,
+      "grad_norm": 1.173801064491272,
+      "learning_rate": 1.3137254901960785e-05,
+      "loss": 0.0384,
+      "step": 866
+    },
+    {
+      "epoch": 4.578217821782178,
+      "grad_norm": 1.3326085805892944,
+      "learning_rate": 1.3088235294117648e-05,
+      "loss": 0.0428,
+      "step": 867
+    },
+    {
+      "epoch": 4.583498349834984,
+      "grad_norm": 0.8449905514717102,
+      "learning_rate": 1.3039215686274511e-05,
+      "loss": 0.02,
+      "step": 868
+    },
+    {
+      "epoch": 4.588778877887789,
+      "grad_norm": 0.9651418924331665,
+      "learning_rate": 1.2990196078431374e-05,
+      "loss": 0.03,
+      "step": 869
+    },
+    {
+      "epoch": 4.594059405940594,
+      "grad_norm": 0.7175216674804688,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.0243,
+      "step": 870
+    },
+    {
+      "epoch": 4.5993399339934,
+      "grad_norm": 0.6259102821350098,
+      "learning_rate": 1.28921568627451e-05,
+      "loss": 0.0212,
+      "step": 871
+    },
+    {
+      "epoch": 4.604620462046205,
+      "grad_norm": 0.7783799171447754,
+      "learning_rate": 1.284313725490196e-05,
+      "loss": 0.0261,
+      "step": 872
+    },
+    {
+      "epoch": 4.6099009900990096,
+      "grad_norm": 1.1052485704421997,
+      "learning_rate": 1.2794117647058822e-05,
+      "loss": 0.0413,
+      "step": 873
+    },
+    {
+      "epoch": 4.615181518151815,
+      "grad_norm": 1.0588634014129639,
+      "learning_rate": 1.2745098039215686e-05,
+      "loss": 0.0324,
+      "step": 874
+    },
+    {
+      "epoch": 4.62046204620462,
+      "grad_norm": 0.8361983299255371,
+      "learning_rate": 1.2696078431372548e-05,
+      "loss": 0.0263,
+      "step": 875
+    },
+    {
+      "epoch": 4.625742574257425,
+      "grad_norm": 0.9796653985977173,
+      "learning_rate": 1.2647058823529412e-05,
+      "loss": 0.0357,
+      "step": 876
+    },
+    {
+      "epoch": 4.631023102310231,
+      "grad_norm": 1.612053394317627,
+      "learning_rate": 1.2598039215686275e-05,
+      "loss": 0.0404,
+      "step": 877
+    },
+    {
+      "epoch": 4.636303630363036,
+      "grad_norm": 0.853406548500061,
+      "learning_rate": 1.2549019607843138e-05,
+      "loss": 0.0266,
+      "step": 878
+    },
+    {
+      "epoch": 4.641584158415841,
+      "grad_norm": 1.0649423599243164,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0373,
+      "step": 879
+    },
+    {
+      "epoch": 4.646864686468647,
+      "grad_norm": 1.1256874799728394,
+      "learning_rate": 1.2450980392156863e-05,
+      "loss": 0.0251,
+      "step": 880
+    },
+    {
+      "epoch": 4.652145214521452,
+      "grad_norm": 0.8546039462089539,
+      "learning_rate": 1.2401960784313727e-05,
+      "loss": 0.0272,
+      "step": 881
+    },
+    {
+      "epoch": 4.657425742574257,
+      "grad_norm": 0.7499862313270569,
+      "learning_rate": 1.2352941176470589e-05,
+      "loss": 0.0227,
+      "step": 882
+    },
+    {
+      "epoch": 4.662706270627063,
+      "grad_norm": 1.2291607856750488,
+      "learning_rate": 1.2303921568627451e-05,
+      "loss": 0.0416,
+      "step": 883
+    },
+    {
+      "epoch": 4.667986798679868,
+      "grad_norm": 0.919152021408081,
+      "learning_rate": 1.2254901960784313e-05,
+      "loss": 0.0325,
+      "step": 884
+    },
+    {
+      "epoch": 4.673267326732673,
+      "grad_norm": 0.8878404498100281,
+      "learning_rate": 1.2205882352941177e-05,
+      "loss": 0.0296,
+      "step": 885
+    },
+    {
+      "epoch": 4.678547854785479,
+      "grad_norm": 1.1350431442260742,
+      "learning_rate": 1.215686274509804e-05,
+      "loss": 0.0372,
+      "step": 886
+    },
+    {
+      "epoch": 4.683828382838284,
+      "grad_norm": 0.9311625957489014,
+      "learning_rate": 1.2107843137254903e-05,
+      "loss": 0.0324,
+      "step": 887
+    },
+    {
+      "epoch": 4.689108910891089,
+      "grad_norm": 1.2680948972702026,
+      "learning_rate": 1.2058823529411765e-05,
+      "loss": 0.0427,
+      "step": 888
+    },
+    {
+      "epoch": 4.694389438943895,
+      "grad_norm": 1.0892651081085205,
+      "learning_rate": 1.200980392156863e-05,
+      "loss": 0.0289,
+      "step": 889
+    },
+    {
+      "epoch": 4.6996699669967,
+      "grad_norm": 0.9546079635620117,
+      "learning_rate": 1.1960784313725491e-05,
+      "loss": 0.0355,
+      "step": 890
+    },
+    {
+      "epoch": 4.704950495049505,
+      "grad_norm": 1.0489920377731323,
+      "learning_rate": 1.1911764705882354e-05,
+      "loss": 0.0287,
+      "step": 891
+    },
+    {
+      "epoch": 4.710231023102311,
+      "grad_norm": 0.8627532720565796,
+      "learning_rate": 1.1862745098039216e-05,
+      "loss": 0.03,
+      "step": 892
+    },
+    {
+      "epoch": 4.7155115511551156,
+      "grad_norm": 1.1037335395812988,
+      "learning_rate": 1.1813725490196078e-05,
+      "loss": 0.0369,
+      "step": 893
+    },
+    {
+      "epoch": 4.7207920792079205,
+      "grad_norm": 0.8661030530929565,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.0267,
+      "step": 894
+    },
+    {
+      "epoch": 4.726072607260726,
+      "grad_norm": 0.7932984828948975,
+      "learning_rate": 1.1715686274509804e-05,
+      "loss": 0.0254,
+      "step": 895
+    },
+    {
+      "epoch": 4.731353135313531,
+      "grad_norm": 1.1760293245315552,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.0408,
+      "step": 896
+    },
+    {
+      "epoch": 4.736633663366336,
+      "grad_norm": 1.098027229309082,
+      "learning_rate": 1.161764705882353e-05,
+      "loss": 0.0304,
+      "step": 897
+    },
+    {
+      "epoch": 4.741914191419142,
+      "grad_norm": 1.1287803649902344,
+      "learning_rate": 1.1568627450980394e-05,
+      "loss": 0.0305,
+      "step": 898
+    },
+    {
+      "epoch": 4.747194719471947,
+      "grad_norm": 1.1614177227020264,
+      "learning_rate": 1.1519607843137254e-05,
+      "loss": 0.0436,
+      "step": 899
+    },
+    {
+      "epoch": 4.752475247524752,
+      "grad_norm": 0.8995096683502197,
+      "learning_rate": 1.1470588235294118e-05,
+      "loss": 0.027,
+      "step": 900
+    },
+    {
+      "epoch": 4.757755775577558,
+      "grad_norm": 1.0893324613571167,
+      "learning_rate": 1.142156862745098e-05,
+      "loss": 0.0304,
+      "step": 901
+    },
+    {
+      "epoch": 4.763036303630363,
+      "grad_norm": 0.8086807131767273,
+      "learning_rate": 1.1372549019607843e-05,
+      "loss": 0.0304,
+      "step": 902
+    },
+    {
+      "epoch": 4.768316831683168,
+      "grad_norm": 0.6953551769256592,
+      "learning_rate": 1.1323529411764707e-05,
+      "loss": 0.0248,
+      "step": 903
+    },
+    {
+      "epoch": 4.773597359735973,
+      "grad_norm": 0.707534670829773,
+      "learning_rate": 1.1274509803921569e-05,
+      "loss": 0.0219,
+      "step": 904
+    },
+    {
+      "epoch": 4.778877887788779,
+      "grad_norm": 0.9743644595146179,
+      "learning_rate": 1.1225490196078433e-05,
+      "loss": 0.0327,
+      "step": 905
+    },
+    {
+      "epoch": 4.784158415841584,
+      "grad_norm": 0.9171704053878784,
+      "learning_rate": 1.1176470588235295e-05,
+      "loss": 0.0277,
+      "step": 906
+    },
+    {
+      "epoch": 4.789438943894389,
+      "grad_norm": 1.019524097442627,
+      "learning_rate": 1.1127450980392157e-05,
+      "loss": 0.0308,
+      "step": 907
+    },
+    {
+      "epoch": 4.794719471947195,
+      "grad_norm": 0.8753800392150879,
+      "learning_rate": 1.107843137254902e-05,
+      "loss": 0.0282,
+      "step": 908
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 1.0594887733459473,
+      "learning_rate": 1.1029411764705883e-05,
+      "loss": 0.0283,
+      "step": 909
+    },
+    {
+      "epoch": 4.805280528052805,
+      "grad_norm": 0.9229967594146729,
+      "learning_rate": 1.0980392156862745e-05,
+      "loss": 0.0298,
+      "step": 910
+    },
+    {
+      "epoch": 4.810561056105611,
+      "grad_norm": 0.8133540153503418,
+      "learning_rate": 1.0931372549019609e-05,
+      "loss": 0.0276,
+      "step": 911
+    },
+    {
+      "epoch": 4.815841584158416,
+      "grad_norm": 0.8808728456497192,
+      "learning_rate": 1.0882352941176471e-05,
+      "loss": 0.0304,
+      "step": 912
+    },
+    {
+      "epoch": 4.821122112211221,
+      "grad_norm": 1.0457746982574463,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.0313,
+      "step": 913
+    },
+    {
+      "epoch": 4.8264026402640265,
+      "grad_norm": 1.004420518875122,
+      "learning_rate": 1.0784313725490197e-05,
+      "loss": 0.0341,
+      "step": 914
+    },
+    {
+      "epoch": 4.8316831683168315,
+      "grad_norm": 0.8167664408683777,
+      "learning_rate": 1.0735294117647058e-05,
+      "loss": 0.0254,
+      "step": 915
+    },
+    {
+      "epoch": 4.8369636963696365,
+      "grad_norm": 0.591806948184967,
+      "learning_rate": 1.0686274509803922e-05,
+      "loss": 0.0191,
+      "step": 916
+    },
+    {
+      "epoch": 4.842244224422442,
+      "grad_norm": 1.1454893350601196,
+      "learning_rate": 1.0637254901960784e-05,
+      "loss": 0.0394,
+      "step": 917
+    },
+    {
+      "epoch": 4.847524752475247,
+      "grad_norm": 0.9010074734687805,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.0287,
+      "step": 918
+    },
+    {
+      "epoch": 4.852805280528052,
+      "grad_norm": 0.9918347597122192,
+      "learning_rate": 1.053921568627451e-05,
+      "loss": 0.0288,
+      "step": 919
+    },
+    {
+      "epoch": 4.858085808580858,
+      "grad_norm": 0.9720319509506226,
+      "learning_rate": 1.0490196078431374e-05,
+      "loss": 0.0356,
+      "step": 920
+    },
+    {
+      "epoch": 4.863366336633663,
+      "grad_norm": 0.8841050267219543,
+      "learning_rate": 1.0441176470588236e-05,
+      "loss": 0.0301,
+      "step": 921
+    },
+    {
+      "epoch": 4.868646864686468,
+      "grad_norm": 1.0269768238067627,
+      "learning_rate": 1.0392156862745098e-05,
+      "loss": 0.0315,
+      "step": 922
+    },
+    {
+      "epoch": 4.873927392739274,
+      "grad_norm": 0.7187484502792358,
+      "learning_rate": 1.034313725490196e-05,
+      "loss": 0.0198,
+      "step": 923
+    },
+    {
+      "epoch": 4.879207920792079,
+      "grad_norm": 0.8568077087402344,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 0.0298,
+      "step": 924
+    },
+    {
+      "epoch": 4.884488448844884,
+      "grad_norm": 1.0901175737380981,
+      "learning_rate": 1.0245098039215687e-05,
+      "loss": 0.0321,
+      "step": 925
+    },
+    {
+      "epoch": 4.88976897689769,
+      "grad_norm": 0.7445207834243774,
+      "learning_rate": 1.0196078431372549e-05,
+      "loss": 0.0328,
+      "step": 926
+    },
+    {
+      "epoch": 4.895049504950495,
+      "grad_norm": 1.0356990098953247,
+      "learning_rate": 1.0147058823529413e-05,
+      "loss": 0.0301,
+      "step": 927
+    },
+    {
+      "epoch": 4.9003300330033,
+      "grad_norm": 1.0626378059387207,
+      "learning_rate": 1.0098039215686275e-05,
+      "loss": 0.0317,
+      "step": 928
+    },
+    {
+      "epoch": 4.905610561056106,
+      "grad_norm": 0.9375091195106506,
+      "learning_rate": 1.0049019607843139e-05,
+      "loss": 0.0306,
+      "step": 929
+    },
+    {
+      "epoch": 4.910891089108911,
+      "grad_norm": 0.7572767734527588,
+      "learning_rate": 1e-05,
+      "loss": 0.0233,
+      "step": 930
+    },
+    {
+      "epoch": 4.916171617161716,
+      "grad_norm": 0.7957236170768738,
+      "learning_rate": 9.950980392156863e-06,
+      "loss": 0.0286,
+      "step": 931
+    },
+    {
+      "epoch": 4.921452145214522,
+      "grad_norm": 0.723996639251709,
+      "learning_rate": 9.901960784313725e-06,
+      "loss": 0.0221,
+      "step": 932
+    },
+    {
+      "epoch": 4.926732673267327,
+      "grad_norm": 0.7637007236480713,
+      "learning_rate": 9.852941176470589e-06,
+      "loss": 0.0222,
+      "step": 933
+    },
+    {
+      "epoch": 4.932013201320132,
+      "grad_norm": 1.145386815071106,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.0401,
+      "step": 934
+    },
+    {
+      "epoch": 4.9372937293729375,
+      "grad_norm": 1.1786422729492188,
+      "learning_rate": 9.754901960784313e-06,
+      "loss": 0.04,
+      "step": 935
+    },
+    {
+      "epoch": 4.9425742574257425,
+      "grad_norm": 0.881608784198761,
+      "learning_rate": 9.705882352941177e-06,
+      "loss": 0.0292,
+      "step": 936
+    },
+    {
+      "epoch": 4.9478547854785475,
+      "grad_norm": 1.056412935256958,
+      "learning_rate": 9.65686274509804e-06,
+      "loss": 0.0325,
+      "step": 937
+    },
+    {
+      "epoch": 4.953135313531353,
+      "grad_norm": 1.0825217962265015,
+      "learning_rate": 9.607843137254903e-06,
+      "loss": 0.039,
+      "step": 938
+    },
+    {
+      "epoch": 4.958415841584158,
+      "grad_norm": 0.6380865573883057,
+      "learning_rate": 9.558823529411764e-06,
+      "loss": 0.0231,
+      "step": 939
+    },
+    {
+      "epoch": 4.963696369636963,
+      "grad_norm": 0.8444051742553711,
+      "learning_rate": 9.509803921568628e-06,
+      "loss": 0.0284,
+      "step": 940
+    },
+    {
+      "epoch": 4.968976897689769,
+      "grad_norm": 0.8642618656158447,
+      "learning_rate": 9.46078431372549e-06,
+      "loss": 0.0265,
+      "step": 941
+    },
+    {
+      "epoch": 4.974257425742574,
+      "grad_norm": 1.0572503805160522,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.0341,
+      "step": 942
+    },
+    {
+      "epoch": 4.979537953795379,
+      "grad_norm": 0.8778902888298035,
+      "learning_rate": 9.362745098039216e-06,
+      "loss": 0.0304,
+      "step": 943
+    },
+    {
+      "epoch": 4.984818481848185,
+      "grad_norm": 0.8647822737693787,
+      "learning_rate": 9.31372549019608e-06,
+      "loss": 0.0289,
+      "step": 944
+    },
+    {
+      "epoch": 4.99009900990099,
+      "grad_norm": 1.0456162691116333,
+      "learning_rate": 9.264705882352942e-06,
+      "loss": 0.0322,
+      "step": 945
+    },
+    {
+      "epoch": 4.995379537953795,
+      "grad_norm": 1.451906681060791,
+      "learning_rate": 9.215686274509804e-06,
+      "loss": 0.0414,
+      "step": 946
+    },
+    {
+      "epoch": 5.000660066006601,
+      "grad_norm": 1.2711623907089233,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.0378,
+      "step": 947
+    },
+    {
+      "epoch": 5.005940594059406,
+      "grad_norm": 0.5490008592605591,
+      "learning_rate": 9.117647058823529e-06,
+      "loss": 0.0236,
+      "step": 948
+    },
+    {
+      "epoch": 5.011221122112211,
+      "grad_norm": 0.4675862789154053,
+      "learning_rate": 9.068627450980392e-06,
+      "loss": 0.0193,
+      "step": 949
+    },
+    {
+      "epoch": 5.016501650165017,
+      "grad_norm": 0.5535048246383667,
+      "learning_rate": 9.019607843137255e-06,
+      "loss": 0.0198,
+      "step": 950
+    },
+    {
+      "epoch": 5.021782178217822,
+      "grad_norm": 0.759076714515686,
+      "learning_rate": 8.970588235294119e-06,
+      "loss": 0.024,
+      "step": 951
+    },
+    {
+      "epoch": 5.027062706270627,
+      "grad_norm": 0.551156222820282,
+      "learning_rate": 8.92156862745098e-06,
+      "loss": 0.0182,
+      "step": 952
+    },
+    {
+      "epoch": 5.032343234323433,
+      "grad_norm": 0.561464786529541,
+      "learning_rate": 8.872549019607845e-06,
+      "loss": 0.0211,
+      "step": 953
+    },
+    {
+      "epoch": 5.037623762376238,
+      "grad_norm": 0.3056110441684723,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.0146,
+      "step": 954
+    },
+    {
+      "epoch": 5.042904290429043,
+      "grad_norm": 0.6087129712104797,
+      "learning_rate": 8.774509803921569e-06,
+      "loss": 0.0186,
+      "step": 955
+    },
+    {
+      "epoch": 5.0481848184818485,
+      "grad_norm": 0.37718823552131653,
+      "learning_rate": 8.725490196078431e-06,
+      "loss": 0.0157,
+      "step": 956
+    },
+    {
+      "epoch": 5.0534653465346535,
+      "grad_norm": 0.5187584161758423,
+      "learning_rate": 8.676470588235295e-06,
+      "loss": 0.0206,
+      "step": 957
+    },
+    {
+      "epoch": 5.0587458745874585,
+      "grad_norm": 0.4911420941352844,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 0.0198,
+      "step": 958
+    },
+    {
+      "epoch": 5.064026402640264,
+      "grad_norm": 0.5944388508796692,
+      "learning_rate": 8.57843137254902e-06,
+      "loss": 0.0199,
+      "step": 959
+    },
+    {
+      "epoch": 5.069306930693069,
+      "grad_norm": 0.4171479344367981,
+      "learning_rate": 8.529411764705883e-06,
+      "loss": 0.018,
+      "step": 960
+    },
+    {
+      "epoch": 5.074587458745874,
+      "grad_norm": 0.4937235414981842,
+      "learning_rate": 8.480392156862745e-06,
+      "loss": 0.0182,
+      "step": 961
+    },
+    {
+      "epoch": 5.07986798679868,
+      "grad_norm": 0.44039833545684814,
+      "learning_rate": 8.43137254901961e-06,
+      "loss": 0.0166,
+      "step": 962
+    },
+    {
+      "epoch": 5.085148514851485,
+      "grad_norm": 0.5266954302787781,
+      "learning_rate": 8.38235294117647e-06,
+      "loss": 0.0199,
+      "step": 963
+    },
+    {
+      "epoch": 5.09042904290429,
+      "grad_norm": 0.5795379877090454,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.0168,
+      "step": 964
+    },
+    {
+      "epoch": 5.095709570957096,
+      "grad_norm": 0.7336843013763428,
+      "learning_rate": 8.284313725490196e-06,
+      "loss": 0.0265,
+      "step": 965
+    },
+    {
+      "epoch": 5.100990099009901,
+      "grad_norm": 0.4279429018497467,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.0177,
+      "step": 966
+    },
+    {
+      "epoch": 5.106270627062706,
+      "grad_norm": 0.6605322957038879,
+      "learning_rate": 8.186274509803922e-06,
+      "loss": 0.0211,
+      "step": 967
+    },
+    {
+      "epoch": 5.111551155115512,
+      "grad_norm": 0.6620057225227356,
+      "learning_rate": 8.137254901960784e-06,
+      "loss": 0.0186,
+      "step": 968
+    },
+    {
+      "epoch": 5.116831683168317,
+      "grad_norm": 0.36124753952026367,
+      "learning_rate": 8.088235294117648e-06,
+      "loss": 0.015,
+      "step": 969
+    },
+    {
+      "epoch": 5.122112211221122,
+      "grad_norm": 0.64441978931427,
+      "learning_rate": 8.03921568627451e-06,
+      "loss": 0.0176,
+      "step": 970
+    },
+    {
+      "epoch": 5.127392739273928,
+      "grad_norm": 0.45318105816841125,
+      "learning_rate": 7.990196078431372e-06,
+      "loss": 0.0155,
+      "step": 971
+    },
+    {
+      "epoch": 5.132673267326733,
+      "grad_norm": 0.528228223323822,
+      "learning_rate": 7.941176470588235e-06,
+      "loss": 0.0169,
+      "step": 972
+    },
+    {
+      "epoch": 5.137953795379538,
+      "grad_norm": 0.5010206699371338,
+      "learning_rate": 7.892156862745098e-06,
+      "loss": 0.018,
+      "step": 973
+    },
+    {
+      "epoch": 5.143234323432344,
+      "grad_norm": 0.4574146568775177,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 0.0172,
+      "step": 974
+    },
+    {
+      "epoch": 5.148514851485149,
+      "grad_norm": 0.5487357974052429,
+      "learning_rate": 7.794117647058825e-06,
+      "loss": 0.0173,
+      "step": 975
+    },
+    {
+      "epoch": 5.153795379537954,
+      "grad_norm": 0.7028687596321106,
+      "learning_rate": 7.745098039215687e-06,
+      "loss": 0.0219,
+      "step": 976
+    },
+    {
+      "epoch": 5.1590759075907595,
+      "grad_norm": 0.5689717531204224,
+      "learning_rate": 7.69607843137255e-06,
+      "loss": 0.0167,
+      "step": 977
+    },
+    {
+      "epoch": 5.1643564356435645,
+      "grad_norm": 0.4024108052253723,
+      "learning_rate": 7.647058823529413e-06,
+      "loss": 0.0139,
+      "step": 978
+    },
+    {
+      "epoch": 5.1696369636963695,
+      "grad_norm": 0.81581050157547,
+      "learning_rate": 7.598039215686274e-06,
+      "loss": 0.0156,
+      "step": 979
+    },
+    {
+      "epoch": 5.174917491749175,
+      "grad_norm": 0.4765579104423523,
+      "learning_rate": 7.549019607843137e-06,
+      "loss": 0.0155,
+      "step": 980
+    },
+    {
+      "epoch": 5.18019801980198,
+      "grad_norm": 0.43915316462516785,
+      "learning_rate": 7.5e-06,
+      "loss": 0.0188,
+      "step": 981
+    },
+    {
+      "epoch": 5.185478547854785,
+      "grad_norm": 0.5227847099304199,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 0.0206,
+      "step": 982
+    },
+    {
+      "epoch": 5.19075907590759,
+      "grad_norm": 0.6932447552680969,
+      "learning_rate": 7.401960784313726e-06,
+      "loss": 0.0215,
+      "step": 983
+    },
+    {
+      "epoch": 5.196039603960396,
+      "grad_norm": 0.6111219525337219,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 0.0212,
+      "step": 984
+    },
+    {
+      "epoch": 5.201320132013201,
+      "grad_norm": 0.4277690649032593,
+      "learning_rate": 7.3039215686274515e-06,
+      "loss": 0.0159,
+      "step": 985
+    },
+    {
+      "epoch": 5.206600660066006,
+      "grad_norm": 0.47906622290611267,
+      "learning_rate": 7.2549019607843145e-06,
+      "loss": 0.0148,
+      "step": 986
+    },
+    {
+      "epoch": 5.211881188118812,
+      "grad_norm": 0.565126359462738,
+      "learning_rate": 7.205882352941176e-06,
+      "loss": 0.0195,
+      "step": 987
+    },
+    {
+      "epoch": 5.217161716171617,
+      "grad_norm": 0.46082907915115356,
+      "learning_rate": 7.156862745098039e-06,
+      "loss": 0.0183,
+      "step": 988
+    },
+    {
+      "epoch": 5.222442244224422,
+      "grad_norm": 0.6407860517501831,
+      "learning_rate": 7.107843137254902e-06,
+      "loss": 0.0186,
+      "step": 989
+    },
+    {
+      "epoch": 5.227722772277228,
+      "grad_norm": 0.4975033104419708,
+      "learning_rate": 7.058823529411765e-06,
+      "loss": 0.018,
+      "step": 990
+    },
+    {
+      "epoch": 5.233003300330033,
+      "grad_norm": 0.8421338200569153,
+      "learning_rate": 7.009803921568628e-06,
+      "loss": 0.0236,
+      "step": 991
+    },
+    {
+      "epoch": 5.238283828382838,
+      "grad_norm": 0.5029832124710083,
+      "learning_rate": 6.960784313725491e-06,
+      "loss": 0.017,
+      "step": 992
+    },
+    {
+      "epoch": 5.243564356435644,
+      "grad_norm": 0.6490949988365173,
+      "learning_rate": 6.911764705882354e-06,
+      "loss": 0.0225,
+      "step": 993
+    },
+    {
+      "epoch": 5.248844884488449,
+      "grad_norm": 0.4616677165031433,
+      "learning_rate": 6.862745098039216e-06,
+      "loss": 0.0141,
+      "step": 994
+    },
+    {
+      "epoch": 5.254125412541254,
+      "grad_norm": 0.5079948306083679,
+      "learning_rate": 6.813725490196078e-06,
+      "loss": 0.0201,
+      "step": 995
+    },
+    {
+      "epoch": 5.25940594059406,
+      "grad_norm": 0.6577730178833008,
+      "learning_rate": 6.7647058823529414e-06,
+      "loss": 0.0166,
+      "step": 996
+    },
+    {
+      "epoch": 5.264686468646865,
+      "grad_norm": 0.4667012691497803,
+      "learning_rate": 6.715686274509804e-06,
+      "loss": 0.0142,
+      "step": 997
+    },
+    {
+      "epoch": 5.26996699669967,
+      "grad_norm": 0.4957946836948395,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.0183,
+      "step": 998
+    },
+    {
+      "epoch": 5.2752475247524755,
+      "grad_norm": 0.5439987182617188,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 0.0203,
+      "step": 999
+    },
+    {
+      "epoch": 5.2805280528052805,
+      "grad_norm": 0.6871292591094971,
+      "learning_rate": 6.568627450980393e-06,
+      "loss": 0.0187,
+      "step": 1000
+    },
+    {
+      "epoch": 5.2858085808580855,
+      "grad_norm": 1.2314088344573975,
+      "learning_rate": 6.519607843137256e-06,
+      "loss": 0.0218,
+      "step": 1001
+    },
+    {
+      "epoch": 5.291089108910891,
+      "grad_norm": 0.9941577911376953,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.0231,
+      "step": 1002
+    },
+    {
+      "epoch": 5.296369636963696,
+      "grad_norm": 0.5083587765693665,
+      "learning_rate": 6.42156862745098e-06,
+      "loss": 0.0175,
+      "step": 1003
+    },
+    {
+      "epoch": 5.301650165016501,
+      "grad_norm": 0.548037052154541,
+      "learning_rate": 6.372549019607843e-06,
+      "loss": 0.019,
+      "step": 1004
+    },
+    {
+      "epoch": 5.306930693069307,
+      "grad_norm": 0.5824829339981079,
+      "learning_rate": 6.323529411764706e-06,
+      "loss": 0.0229,
+      "step": 1005
+    },
+    {
+      "epoch": 5.312211221122112,
+      "grad_norm": 0.40190955996513367,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 0.0177,
+      "step": 1006
+    },
+    {
+      "epoch": 5.317491749174917,
+      "grad_norm": 0.5402151346206665,
+      "learning_rate": 6.225490196078431e-06,
+      "loss": 0.0204,
+      "step": 1007
+    },
+    {
+      "epoch": 5.322772277227723,
+      "grad_norm": 0.5250375866889954,
+      "learning_rate": 6.1764705882352944e-06,
+      "loss": 0.0179,
+      "step": 1008
+    },
+    {
+      "epoch": 5.328052805280528,
+      "grad_norm": 0.823615550994873,
+      "learning_rate": 6.127450980392157e-06,
+      "loss": 0.018,
+      "step": 1009
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 0.5749222636222839,
+      "learning_rate": 6.07843137254902e-06,
+      "loss": 0.0179,
+      "step": 1010
+    },
+    {
+      "epoch": 5.338613861386139,
+      "grad_norm": 0.44801655411720276,
+      "learning_rate": 6.029411764705883e-06,
+      "loss": 0.0196,
+      "step": 1011
+    },
+    {
+      "epoch": 5.343894389438944,
+      "grad_norm": 0.41333383321762085,
+      "learning_rate": 5.980392156862746e-06,
+      "loss": 0.0174,
+      "step": 1012
+    },
+    {
+      "epoch": 5.349174917491749,
+      "grad_norm": 0.6496670842170715,
+      "learning_rate": 5.931372549019608e-06,
+      "loss": 0.0185,
+      "step": 1013
+    },
+    {
+      "epoch": 5.354455445544555,
+      "grad_norm": 0.5372949838638306,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.0196,
+      "step": 1014
+    },
+    {
+      "epoch": 5.35973597359736,
+      "grad_norm": 0.6921015381813049,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.0208,
+      "step": 1015
+    },
+    {
+      "epoch": 5.365016501650165,
+      "grad_norm": 0.40831905603408813,
+      "learning_rate": 5.784313725490197e-06,
+      "loss": 0.0156,
+      "step": 1016
+    },
+    {
+      "epoch": 5.370297029702971,
+      "grad_norm": 0.746163010597229,
+      "learning_rate": 5.735294117647059e-06,
+      "loss": 0.0247,
+      "step": 1017
+    },
+    {
+      "epoch": 5.375577557755776,
+      "grad_norm": 0.3688511252403259,
+      "learning_rate": 5.686274509803921e-06,
+      "loss": 0.0149,
+      "step": 1018
+    },
+    {
+      "epoch": 5.380858085808581,
+      "grad_norm": 0.6566423773765564,
+      "learning_rate": 5.637254901960784e-06,
+      "loss": 0.0207,
+      "step": 1019
+    },
+    {
+      "epoch": 5.3861386138613865,
+      "grad_norm": 0.5603022575378418,
+      "learning_rate": 5.588235294117647e-06,
+      "loss": 0.015,
+      "step": 1020
+    },
+    {
+      "epoch": 5.3914191419141915,
+      "grad_norm": 0.48195961117744446,
+      "learning_rate": 5.53921568627451e-06,
+      "loss": 0.0196,
+      "step": 1021
+    },
+    {
+      "epoch": 5.3966996699669965,
+      "grad_norm": 0.8574143052101135,
+      "learning_rate": 5.490196078431373e-06,
+      "loss": 0.0213,
+      "step": 1022
+    },
+    {
+      "epoch": 5.401980198019802,
+      "grad_norm": 0.6029947996139526,
+      "learning_rate": 5.441176470588236e-06,
+      "loss": 0.0157,
+      "step": 1023
+    },
+    {
+      "epoch": 5.407260726072607,
+      "grad_norm": 0.6817245483398438,
+      "learning_rate": 5.392156862745099e-06,
+      "loss": 0.021,
+      "step": 1024
+    },
+    {
+      "epoch": 5.412541254125412,
+      "grad_norm": 0.5209661722183228,
+      "learning_rate": 5.343137254901961e-06,
+      "loss": 0.0174,
+      "step": 1025
+    },
+    {
+      "epoch": 5.417821782178218,
+      "grad_norm": 0.637236475944519,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.0227,
+      "step": 1026
+    },
+    {
+      "epoch": 5.423102310231023,
+      "grad_norm": 0.4161434471607208,
+      "learning_rate": 5.245098039215687e-06,
+      "loss": 0.0186,
+      "step": 1027
+    },
+    {
+      "epoch": 5.428382838283828,
+      "grad_norm": 0.4232980012893677,
+      "learning_rate": 5.196078431372549e-06,
+      "loss": 0.0157,
+      "step": 1028
+    },
+    {
+      "epoch": 5.433663366336634,
+      "grad_norm": 0.8480959534645081,
+      "learning_rate": 5.147058823529412e-06,
+      "loss": 0.0236,
+      "step": 1029
+    },
+    {
+      "epoch": 5.438943894389439,
+      "grad_norm": 0.7071532011032104,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 0.0232,
+      "step": 1030
+    },
+    {
+      "epoch": 5.444224422442244,
+      "grad_norm": 0.46706661581993103,
+      "learning_rate": 5.049019607843137e-06,
+      "loss": 0.0154,
+      "step": 1031
+    },
+    {
+      "epoch": 5.44950495049505,
+      "grad_norm": 0.38799214363098145,
+      "learning_rate": 5e-06,
+      "loss": 0.0153,
+      "step": 1032
+    },
+    {
+      "epoch": 5.454785478547855,
+      "grad_norm": 0.6109775900840759,
+      "learning_rate": 4.950980392156863e-06,
+      "loss": 0.0185,
+      "step": 1033
+    },
+    {
+      "epoch": 5.46006600660066,
+      "grad_norm": 0.4341484308242798,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 0.0167,
+      "step": 1034
+    },
+    {
+      "epoch": 5.465346534653466,
+      "grad_norm": 0.7449372410774231,
+      "learning_rate": 4.852941176470589e-06,
+      "loss": 0.0198,
+      "step": 1035
+    },
+    {
+      "epoch": 5.470627062706271,
+      "grad_norm": 0.36592140793800354,
+      "learning_rate": 4.803921568627452e-06,
+      "loss": 0.0157,
+      "step": 1036
+    },
+    {
+      "epoch": 5.475907590759076,
+      "grad_norm": 0.5300789475440979,
+      "learning_rate": 4.754901960784314e-06,
+      "loss": 0.018,
+      "step": 1037
+    },
+    {
+      "epoch": 5.481188118811881,
+      "grad_norm": 0.5365732908248901,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.0192,
+      "step": 1038
+    },
+    {
+      "epoch": 5.486468646864687,
+      "grad_norm": 0.5415321588516235,
+      "learning_rate": 4.65686274509804e-06,
+      "loss": 0.0176,
+      "step": 1039
+    },
+    {
+      "epoch": 5.491749174917492,
+      "grad_norm": 0.4487341344356537,
+      "learning_rate": 4.607843137254902e-06,
+      "loss": 0.0149,
+      "step": 1040
+    },
+    {
+      "epoch": 5.497029702970297,
+      "grad_norm": 0.5181805491447449,
+      "learning_rate": 4.558823529411764e-06,
+      "loss": 0.0167,
+      "step": 1041
+    },
+    {
+      "epoch": 5.5023102310231025,
+      "grad_norm": 0.7573541402816772,
+      "learning_rate": 4.509803921568627e-06,
+      "loss": 0.0192,
+      "step": 1042
+    },
+    {
+      "epoch": 5.5075907590759074,
+      "grad_norm": 2.3085265159606934,
+      "learning_rate": 4.46078431372549e-06,
+      "loss": 0.0912,
+      "step": 1043
+    },
+    {
+      "epoch": 5.512871287128712,
+      "grad_norm": 0.5817523002624512,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 0.015,
+      "step": 1044
+    },
+    {
+      "epoch": 5.518151815181518,
+      "grad_norm": 0.6805261969566345,
+      "learning_rate": 4.362745098039216e-06,
+      "loss": 0.0215,
+      "step": 1045
+    },
+    {
+      "epoch": 5.523432343234323,
+      "grad_norm": 0.597396969795227,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 0.0206,
+      "step": 1046
+    },
+    {
+      "epoch": 5.528712871287128,
+      "grad_norm": 0.934479296207428,
+      "learning_rate": 4.264705882352942e-06,
+      "loss": 0.0207,
+      "step": 1047
+    },
+    {
+      "epoch": 5.533993399339934,
+      "grad_norm": 0.4807792603969574,
+      "learning_rate": 4.215686274509805e-06,
+      "loss": 0.0158,
+      "step": 1048
+    },
+    {
+      "epoch": 5.539273927392739,
+      "grad_norm": 0.6328734159469604,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.0202,
+      "step": 1049
+    },
+    {
+      "epoch": 5.544554455445544,
+      "grad_norm": 0.4848245084285736,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.0199,
+      "step": 1050
+    },
+    {
+      "epoch": 5.54983498349835,
+      "grad_norm": 0.4785003066062927,
+      "learning_rate": 4.068627450980392e-06,
+      "loss": 0.0188,
+      "step": 1051
+    },
+    {
+      "epoch": 5.555115511551155,
+      "grad_norm": 0.4844956696033478,
+      "learning_rate": 4.019607843137255e-06,
+      "loss": 0.0175,
+      "step": 1052
+    },
+    {
+      "epoch": 5.56039603960396,
+      "grad_norm": 0.40522444248199463,
+      "learning_rate": 3.970588235294117e-06,
+      "loss": 0.0159,
+      "step": 1053
+    },
+    {
+      "epoch": 5.565676567656766,
+      "grad_norm": 0.40739139914512634,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 0.0165,
+      "step": 1054
+    },
+    {
+      "epoch": 5.570957095709571,
+      "grad_norm": 0.42678526043891907,
+      "learning_rate": 3.872549019607843e-06,
+      "loss": 0.0152,
+      "step": 1055
+    },
+    {
+      "epoch": 5.576237623762376,
+      "grad_norm": 0.5168190598487854,
+      "learning_rate": 3.823529411764706e-06,
+      "loss": 0.0164,
+      "step": 1056
+    },
+    {
+      "epoch": 5.581518151815182,
+      "grad_norm": 0.4231308400630951,
+      "learning_rate": 3.7745098039215686e-06,
+      "loss": 0.016,
+      "step": 1057
+    },
+    {
+      "epoch": 5.586798679867987,
+      "grad_norm": 0.604710578918457,
+      "learning_rate": 3.7254901960784316e-06,
+      "loss": 0.0225,
+      "step": 1058
+    },
+    {
+      "epoch": 5.592079207920792,
+      "grad_norm": 0.6866090893745422,
+      "learning_rate": 3.6764705882352942e-06,
+      "loss": 0.0195,
+      "step": 1059
+    },
+    {
+      "epoch": 5.597359735973598,
+      "grad_norm": 0.34970760345458984,
+      "learning_rate": 3.6274509803921573e-06,
+      "loss": 0.014,
+      "step": 1060
+    },
+    {
+      "epoch": 5.602640264026403,
+      "grad_norm": 0.4601968228816986,
+      "learning_rate": 3.5784313725490194e-06,
+      "loss": 0.018,
+      "step": 1061
+    },
+    {
+      "epoch": 5.607920792079208,
+      "grad_norm": 0.3815479874610901,
+      "learning_rate": 3.5294117647058825e-06,
+      "loss": 0.0152,
+      "step": 1062
+    },
+    {
+      "epoch": 5.6132013201320134,
+      "grad_norm": 0.36140522360801697,
+      "learning_rate": 3.4803921568627455e-06,
+      "loss": 0.0149,
+      "step": 1063
+    },
+    {
+      "epoch": 5.618481848184818,
+      "grad_norm": 0.485866904258728,
+      "learning_rate": 3.431372549019608e-06,
+      "loss": 0.0171,
+      "step": 1064
+    },
+    {
+      "epoch": 5.623762376237623,
+      "grad_norm": 0.5746606588363647,
+      "learning_rate": 3.3823529411764707e-06,
+      "loss": 0.0168,
+      "step": 1065
+    },
+    {
+      "epoch": 5.629042904290429,
+      "grad_norm": 0.5221585035324097,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.0199,
+      "step": 1066
+    },
+    {
+      "epoch": 5.634323432343234,
+      "grad_norm": 0.538971483707428,
+      "learning_rate": 3.2843137254901964e-06,
+      "loss": 0.0225,
+      "step": 1067
+    },
+    {
+      "epoch": 5.639603960396039,
+      "grad_norm": 0.4141756594181061,
+      "learning_rate": 3.2352941176470594e-06,
+      "loss": 0.0171,
+      "step": 1068
+    },
+    {
+      "epoch": 5.644884488448845,
+      "grad_norm": 0.3699265420436859,
+      "learning_rate": 3.1862745098039216e-06,
+      "loss": 0.0127,
+      "step": 1069
+    },
+    {
+      "epoch": 5.65016501650165,
+      "grad_norm": 0.8097654581069946,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 0.0168,
+      "step": 1070
+    },
+    {
+      "epoch": 5.655445544554455,
+      "grad_norm": 0.8609626889228821,
+      "learning_rate": 3.0882352941176472e-06,
+      "loss": 0.0202,
+      "step": 1071
+    },
+    {
+      "epoch": 5.660726072607261,
+      "grad_norm": 0.7239216566085815,
+      "learning_rate": 3.03921568627451e-06,
+      "loss": 0.0173,
+      "step": 1072
+    },
+    {
+      "epoch": 5.666006600660066,
+      "grad_norm": 0.5894525647163391,
+      "learning_rate": 2.990196078431373e-06,
+      "loss": 0.0248,
+      "step": 1073
+    },
+    {
+      "epoch": 5.671287128712871,
+      "grad_norm": 0.39227187633514404,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.0173,
+      "step": 1074
+    },
+    {
+      "epoch": 5.676567656765677,
+      "grad_norm": 0.5740650296211243,
+      "learning_rate": 2.8921568627450985e-06,
+      "loss": 0.0196,
+      "step": 1075
+    },
+    {
+      "epoch": 5.681848184818482,
+      "grad_norm": 0.5081339478492737,
+      "learning_rate": 2.8431372549019607e-06,
+      "loss": 0.0178,
+      "step": 1076
+    },
+    {
+      "epoch": 5.687128712871287,
+      "grad_norm": 0.45653530955314636,
+      "learning_rate": 2.7941176470588237e-06,
+      "loss": 0.017,
+      "step": 1077
+    },
+    {
+      "epoch": 5.692409240924093,
+      "grad_norm": 0.5196064710617065,
+      "learning_rate": 2.7450980392156863e-06,
+      "loss": 0.0158,
+      "step": 1078
+    },
+    {
+      "epoch": 5.697689768976898,
+      "grad_norm": 0.5605891942977905,
+      "learning_rate": 2.6960784313725493e-06,
+      "loss": 0.0172,
+      "step": 1079
+    },
+    {
+      "epoch": 5.702970297029703,
+      "grad_norm": 0.6851004958152771,
+      "learning_rate": 2.647058823529412e-06,
+      "loss": 0.0172,
+      "step": 1080
+    },
+    {
+      "epoch": 5.708250825082509,
+      "grad_norm": 0.4508073031902313,
+      "learning_rate": 2.5980392156862746e-06,
+      "loss": 0.0171,
+      "step": 1081
+    },
+    {
+      "epoch": 5.713531353135314,
+      "grad_norm": 0.4393492639064789,
+      "learning_rate": 2.549019607843137e-06,
+      "loss": 0.0167,
+      "step": 1082
+    },
+    {
+      "epoch": 5.718811881188119,
+      "grad_norm": 0.5042679309844971,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0197,
+      "step": 1083
+    },
+    {
+      "epoch": 5.724092409240924,
+      "grad_norm": 0.40561428666114807,
+      "learning_rate": 2.450980392156863e-06,
+      "loss": 0.0175,
+      "step": 1084
+    },
+    {
+      "epoch": 5.729372937293729,
+      "grad_norm": 0.390462189912796,
+      "learning_rate": 2.401960784313726e-06,
+      "loss": 0.0166,
+      "step": 1085
+    },
+    {
+      "epoch": 5.734653465346534,
+      "grad_norm": 0.44142773747444153,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.0161,
+      "step": 1086
+    },
+    {
+      "epoch": 5.73993399339934,
+      "grad_norm": 0.6280815601348877,
+      "learning_rate": 2.303921568627451e-06,
+      "loss": 0.0191,
+      "step": 1087
+    },
+    {
+      "epoch": 5.745214521452145,
+      "grad_norm": 0.6288333535194397,
+      "learning_rate": 2.2549019607843137e-06,
+      "loss": 0.0169,
+      "step": 1088
+    },
+    {
+      "epoch": 5.75049504950495,
+      "grad_norm": 0.39587247371673584,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 0.0154,
+      "step": 1089
+    },
+    {
+      "epoch": 5.755775577557756,
+      "grad_norm": 0.8085327744483948,
+      "learning_rate": 2.1568627450980393e-06,
+      "loss": 0.0214,
+      "step": 1090
+    },
+    {
+      "epoch": 5.761056105610561,
+      "grad_norm": 0.31861352920532227,
+      "learning_rate": 2.1078431372549023e-06,
+      "loss": 0.0113,
+      "step": 1091
+    },
+    {
+      "epoch": 5.766336633663366,
+      "grad_norm": 0.6090717315673828,
+      "learning_rate": 2.058823529411765e-06,
+      "loss": 0.0176,
+      "step": 1092
+    },
+    {
+      "epoch": 5.771617161716172,
+      "grad_norm": 0.37864407896995544,
+      "learning_rate": 2.0098039215686276e-06,
+      "loss": 0.013,
+      "step": 1093
+    },
+    {
+      "epoch": 5.776897689768977,
+      "grad_norm": 0.39943984150886536,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 0.0183,
+      "step": 1094
+    },
+    {
+      "epoch": 5.782178217821782,
+      "grad_norm": 0.3896962106227875,
+      "learning_rate": 1.911764705882353e-06,
+      "loss": 0.0147,
+      "step": 1095
+    },
+    {
+      "epoch": 5.787458745874588,
+      "grad_norm": 0.8300476670265198,
+      "learning_rate": 1.8627450980392158e-06,
+      "loss": 0.0219,
+      "step": 1096
+    },
+    {
+      "epoch": 5.792739273927393,
+      "grad_norm": 0.7217696905136108,
+      "learning_rate": 1.8137254901960786e-06,
+      "loss": 0.0193,
+      "step": 1097
+    },
+    {
+      "epoch": 5.798019801980198,
+      "grad_norm": 0.7046459317207336,
+      "learning_rate": 1.7647058823529412e-06,
+      "loss": 0.019,
+      "step": 1098
+    },
+    {
+      "epoch": 5.803300330033004,
+      "grad_norm": 0.5788043141365051,
+      "learning_rate": 1.715686274509804e-06,
+      "loss": 0.0197,
+      "step": 1099
+    },
+    {
+      "epoch": 5.808580858085809,
+      "grad_norm": 0.3641752004623413,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.0139,
+      "step": 1100
+    },
+    {
+      "epoch": 5.813861386138614,
+      "grad_norm": 0.5467818379402161,
+      "learning_rate": 1.6176470588235297e-06,
+      "loss": 0.0185,
+      "step": 1101
+    },
+    {
+      "epoch": 5.81914191419142,
+      "grad_norm": 0.7358901500701904,
+      "learning_rate": 1.5686274509803923e-06,
+      "loss": 0.0218,
+      "step": 1102
+    },
+    {
+      "epoch": 5.824422442244225,
+      "grad_norm": 0.39868056774139404,
+      "learning_rate": 1.519607843137255e-06,
+      "loss": 0.0168,
+      "step": 1103
+    },
+    {
+      "epoch": 5.8297029702970296,
+      "grad_norm": 0.5790492296218872,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 0.0154,
+      "step": 1104
+    },
+    {
+      "epoch": 5.834983498349835,
+      "grad_norm": 0.6561235189437866,
+      "learning_rate": 1.4215686274509803e-06,
+      "loss": 0.0184,
+      "step": 1105
+    },
+    {
+      "epoch": 5.84026402640264,
+      "grad_norm": 0.4628670811653137,
+      "learning_rate": 1.3725490196078432e-06,
+      "loss": 0.0178,
+      "step": 1106
+    },
+    {
+      "epoch": 5.845544554455445,
+      "grad_norm": 0.5217213034629822,
+      "learning_rate": 1.323529411764706e-06,
+      "loss": 0.0179,
+      "step": 1107
+    },
+    {
+      "epoch": 5.850825082508251,
+      "grad_norm": 0.6204583644866943,
+      "learning_rate": 1.2745098039215686e-06,
+      "loss": 0.0197,
+      "step": 1108
+    },
+    {
+      "epoch": 5.856105610561056,
+      "grad_norm": 0.48575830459594727,
+      "learning_rate": 1.2254901960784314e-06,
+      "loss": 0.0191,
+      "step": 1109
+    },
+    {
+      "epoch": 5.861386138613861,
+      "grad_norm": 0.8481599688529968,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.023,
+      "step": 1110
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 0.5180693864822388,
+      "learning_rate": 1.1274509803921568e-06,
+      "loss": 0.016,
+      "step": 1111
+    },
+    {
+      "epoch": 5.871947194719472,
+      "grad_norm": 0.47926902770996094,
+      "learning_rate": 1.0784313725490197e-06,
+      "loss": 0.018,
+      "step": 1112
+    },
+    {
+      "epoch": 5.877227722772277,
+      "grad_norm": 0.4460168182849884,
+      "learning_rate": 1.0294117647058825e-06,
+      "loss": 0.0155,
+      "step": 1113
+    },
+    {
+      "epoch": 5.882508250825083,
+      "grad_norm": 0.6543490886688232,
+      "learning_rate": 9.80392156862745e-07,
+      "loss": 0.0193,
+      "step": 1114
+    },
+    {
+      "epoch": 5.887788778877888,
+      "grad_norm": 0.5319867134094238,
+      "learning_rate": 9.313725490196079e-07,
+      "loss": 0.0164,
+      "step": 1115
+    },
+    {
+      "epoch": 5.893069306930693,
+      "grad_norm": 0.5237565636634827,
+      "learning_rate": 8.823529411764706e-07,
+      "loss": 0.0143,
+      "step": 1116
+    },
+    {
+      "epoch": 5.898349834983498,
+      "grad_norm": 0.43923959136009216,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 0.0172,
+      "step": 1117
+    },
+    {
+      "epoch": 5.903630363036304,
+      "grad_norm": 0.5296475291252136,
+      "learning_rate": 7.843137254901962e-07,
+      "loss": 0.0182,
+      "step": 1118
+    },
+    {
+      "epoch": 5.908910891089109,
+      "grad_norm": 0.38380494713783264,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 0.0159,
+      "step": 1119
+    },
+    {
+      "epoch": 5.914191419141914,
+      "grad_norm": 0.6031242609024048,
+      "learning_rate": 6.862745098039216e-07,
+      "loss": 0.0152,
+      "step": 1120
+    },
+    {
+      "epoch": 5.91947194719472,
+      "grad_norm": 0.5882181525230408,
+      "learning_rate": 6.372549019607843e-07,
+      "loss": 0.0171,
+      "step": 1121
+    },
+    {
+      "epoch": 5.924752475247525,
+      "grad_norm": 0.41396641731262207,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 0.0162,
+      "step": 1122
+    },
+    {
+      "epoch": 5.93003300330033,
+      "grad_norm": 0.595112144947052,
+      "learning_rate": 5.392156862745098e-07,
+      "loss": 0.0174,
+      "step": 1123
+    },
+    {
+      "epoch": 5.9353135313531356,
+      "grad_norm": 0.6171261072158813,
+      "learning_rate": 4.901960784313725e-07,
+      "loss": 0.022,
+      "step": 1124
+    },
+    {
+      "epoch": 5.9405940594059405,
+      "grad_norm": 0.40438172221183777,
+      "learning_rate": 4.411764705882353e-07,
+      "loss": 0.0144,
+      "step": 1125
+    },
+    {
+      "epoch": 5.9458745874587455,
+      "grad_norm": 0.6478520035743713,
+      "learning_rate": 3.921568627450981e-07,
+      "loss": 0.0169,
+      "step": 1126
+    },
+    {
+      "epoch": 5.951155115511551,
+      "grad_norm": 0.3631436824798584,
+      "learning_rate": 3.431372549019608e-07,
+      "loss": 0.0149,
+      "step": 1127
+    },
+    {
+      "epoch": 5.956435643564356,
+      "grad_norm": 0.7141023278236389,
+      "learning_rate": 2.9411764705882356e-07,
+      "loss": 0.0169,
+      "step": 1128
+    },
+    {
+      "epoch": 5.961716171617161,
+      "grad_norm": 0.6038995981216431,
+      "learning_rate": 2.4509803921568627e-07,
+      "loss": 0.0197,
+      "step": 1129
+    },
+    {
+      "epoch": 5.966996699669967,
+      "grad_norm": 0.5263422727584839,
+      "learning_rate": 1.9607843137254904e-07,
+      "loss": 0.0201,
+      "step": 1130
+    },
+    {
+      "epoch": 5.972277227722772,
+      "grad_norm": 0.604172945022583,
+      "learning_rate": 1.4705882352941178e-07,
+      "loss": 0.0181,
+      "step": 1131
+    },
+    {
+      "epoch": 5.977557755775577,
+      "grad_norm": 0.6950435638427734,
+      "learning_rate": 9.803921568627452e-08,
+      "loss": 0.0211,
+      "step": 1132
+    },
+    {
+      "epoch": 5.982838283828383,
+      "grad_norm": 0.5159468650817871,
+      "learning_rate": 4.901960784313726e-08,
+      "loss": 0.0153,
+      "step": 1133
+    },
+    {
+      "epoch": 5.988118811881188,
+      "grad_norm": 0.7329381108283997,
+      "learning_rate": 0.0,
+      "loss": 0.0224,
+      "step": 1134
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1134,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.176433819923251e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}