diff --git "a/checkpoint-585/trainer_state.json" "b/checkpoint-585/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-585/trainer_state.json"
@@ -0,0 +1,4128 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.963068181818182,
+  "eval_steps": 500,
+  "global_step": 585,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008522727272727272,
+      "grad_norm": 35.49958801269531,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 4.6143,
+      "step": 1
+    },
+    {
+      "epoch": 0.017045454545454544,
+      "grad_norm": 35.543556213378906,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 4.6719,
+      "step": 2
+    },
+    {
+      "epoch": 0.02556818181818182,
+      "grad_norm": 34.12852096557617,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 4.5546,
+      "step": 3
+    },
+    {
+      "epoch": 0.03409090909090909,
+      "grad_norm": 33.610572814941406,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.4919,
+      "step": 4
+    },
+    {
+      "epoch": 0.04261363636363636,
+      "grad_norm": 34.532169342041016,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 4.6095,
+      "step": 5
+    },
+    {
+      "epoch": 0.05113636363636364,
+      "grad_norm": 34.2357063293457,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 4.585,
+      "step": 6
+    },
+    {
+      "epoch": 0.05965909090909091,
+      "grad_norm": 34.94609832763672,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 4.5911,
+      "step": 7
+    },
+    {
+      "epoch": 0.06818181818181818,
+      "grad_norm": 34.79508590698242,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.6294,
+      "step": 8
+    },
+    {
+      "epoch": 0.07670454545454546,
+      "grad_norm": 35.18478775024414,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 4.6568,
+      "step": 9
+    },
+    {
+      "epoch": 0.08522727272727272,
+      "grad_norm": 33.75633239746094,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 4.4504,
+      "step": 10
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 34.20966339111328,
+      "learning_rate": 5.5e-07,
+      "loss": 4.5005,
+      "step": 11
+    },
+    {
+      "epoch": 0.10227272727272728,
+      "grad_norm": 33.20008087158203,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.4297,
+      "step": 12
+    },
+    {
+      "epoch": 0.11079545454545454,
+      "grad_norm": 33.53578567504883,
+      "learning_rate": 6.5e-07,
+      "loss": 4.4136,
+      "step": 13
+    },
+    {
+      "epoch": 0.11931818181818182,
+      "grad_norm": 31.951068878173828,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 4.3065,
+      "step": 14
+    },
+    {
+      "epoch": 0.1278409090909091,
+      "grad_norm": 30.890714645385742,
+      "learning_rate": 7.5e-07,
+      "loss": 4.2433,
+      "step": 15
+    },
+    {
+      "epoch": 0.13636363636363635,
+      "grad_norm": 29.448890686035156,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.0981,
+      "step": 16
+    },
+    {
+      "epoch": 0.14488636363636365,
+      "grad_norm": 26.769498825073242,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 3.9818,
+      "step": 17
+    },
+    {
+      "epoch": 0.1534090909090909,
+      "grad_norm": 25.418458938598633,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.8568,
+      "step": 18
+    },
+    {
+      "epoch": 0.16193181818181818,
+      "grad_norm": 24.099462509155273,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 3.7139,
+      "step": 19
+    },
+    {
+      "epoch": 0.17045454545454544,
+      "grad_norm": 22.487621307373047,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.6967,
+      "step": 20
+    },
+    {
+      "epoch": 0.17897727272727273,
+      "grad_norm": 20.72856330871582,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 3.5031,
+      "step": 21
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 19.552040100097656,
+      "learning_rate": 1.1e-06,
+      "loss": 3.4201,
+      "step": 22
+    },
+    {
+      "epoch": 0.19602272727272727,
+      "grad_norm": 18.033971786499023,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 3.1842,
+      "step": 23
+    },
+    {
+      "epoch": 0.20454545454545456,
+      "grad_norm": 18.865802764892578,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.1982,
+      "step": 24
+    },
+    {
+      "epoch": 0.21306818181818182,
+      "grad_norm": 18.849502563476562,
+      "learning_rate": 1.25e-06,
+      "loss": 2.9954,
+      "step": 25
+    },
+    {
+      "epoch": 0.2215909090909091,
+      "grad_norm": 19.714330673217773,
+      "learning_rate": 1.3e-06,
+      "loss": 2.8763,
+      "step": 26
+    },
+    {
+      "epoch": 0.23011363636363635,
+      "grad_norm": 20.26412010192871,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 2.7259,
+      "step": 27
+    },
+    {
+      "epoch": 0.23863636363636365,
+      "grad_norm": 19.212318420410156,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.6099,
+      "step": 28
+    },
+    {
+      "epoch": 0.2471590909090909,
+      "grad_norm": 16.80523681640625,
+      "learning_rate": 1.45e-06,
+      "loss": 2.4482,
+      "step": 29
+    },
+    {
+      "epoch": 0.2556818181818182,
+      "grad_norm": 14.624052047729492,
+      "learning_rate": 1.5e-06,
+      "loss": 2.2364,
+      "step": 30
+    },
+    {
+      "epoch": 0.26420454545454547,
+      "grad_norm": 14.759950637817383,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 2.2159,
+      "step": 31
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 14.904441833496094,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.0206,
+      "step": 32
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 15.03490161895752,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 1.8725,
+      "step": 33
+    },
+    {
+      "epoch": 0.2897727272727273,
+      "grad_norm": 15.70709228515625,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.8046,
+      "step": 34
+    },
+    {
+      "epoch": 0.29829545454545453,
+      "grad_norm": 15.637526512145996,
+      "learning_rate": 1.75e-06,
+      "loss": 1.6198,
+      "step": 35
+    },
+    {
+      "epoch": 0.3068181818181818,
+      "grad_norm": 14.345853805541992,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 1.4247,
+      "step": 36
+    },
+    {
+      "epoch": 0.3153409090909091,
+      "grad_norm": 14.281502723693848,
+      "learning_rate": 1.85e-06,
+      "loss": 1.2859,
+      "step": 37
+    },
+    {
+      "epoch": 0.32386363636363635,
+      "grad_norm": 13.567434310913086,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 1.1452,
+      "step": 38
+    },
+    {
+      "epoch": 0.33238636363636365,
+      "grad_norm": 13.128920555114746,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 1.0036,
+      "step": 39
+    },
+    {
+      "epoch": 0.3409090909090909,
+      "grad_norm": 12.954020500183105,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.851,
+      "step": 40
+    },
+    {
+      "epoch": 0.3494318181818182,
+      "grad_norm": 12.533946990966797,
+      "learning_rate": 2.05e-06,
+      "loss": 0.715,
+      "step": 41
+    },
+    {
+      "epoch": 0.35795454545454547,
+      "grad_norm": 11.564764022827148,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.5987,
+      "step": 42
+    },
+    {
+      "epoch": 0.3664772727272727,
+      "grad_norm": 10.383822441101074,
+      "learning_rate": 2.15e-06,
+      "loss": 0.4834,
+      "step": 43
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 8.901700973510742,
+      "learning_rate": 2.2e-06,
+      "loss": 0.3808,
+      "step": 44
+    },
+    {
+      "epoch": 0.3835227272727273,
+      "grad_norm": 7.737320423126221,
+      "learning_rate": 2.25e-06,
+      "loss": 0.2975,
+      "step": 45
+    },
+    {
+      "epoch": 0.39204545454545453,
+      "grad_norm": 5.334733963012695,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.2261,
+      "step": 46
+    },
+    {
+      "epoch": 0.4005681818181818,
+      "grad_norm": 3.3499741554260254,
+      "learning_rate": 2.35e-06,
+      "loss": 0.1854,
+      "step": 47
+    },
+    {
+      "epoch": 0.4090909090909091,
+      "grad_norm": 2.4037742614746094,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.1532,
+      "step": 48
+    },
+    {
+      "epoch": 0.41761363636363635,
+      "grad_norm": 1.8914185762405396,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.1372,
+      "step": 49
+    },
+    {
+      "epoch": 0.42613636363636365,
+      "grad_norm": 1.7124507427215576,
+      "learning_rate": 2.5e-06,
+      "loss": 0.136,
+      "step": 50
+    },
+    {
+      "epoch": 0.4346590909090909,
+      "grad_norm": 1.242527961730957,
+      "learning_rate": 2.55e-06,
+      "loss": 0.1238,
+      "step": 51
+    },
+    {
+      "epoch": 0.4431818181818182,
+      "grad_norm": 0.9835780262947083,
+      "learning_rate": 2.6e-06,
+      "loss": 0.1192,
+      "step": 52
+    },
+    {
+      "epoch": 0.45170454545454547,
+      "grad_norm": 1.0163497924804688,
+      "learning_rate": 2.6500000000000005e-06,
+      "loss": 0.1175,
+      "step": 53
+    },
+    {
+      "epoch": 0.4602272727272727,
+      "grad_norm": 0.8837094306945801,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.1104,
+      "step": 54
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.6102741956710815,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 0.0986,
+      "step": 55
+    },
+    {
+      "epoch": 0.4772727272727273,
+      "grad_norm": 0.85715252161026,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.1083,
+      "step": 56
+    },
+    {
+      "epoch": 0.48579545454545453,
+      "grad_norm": 0.9692059755325317,
+      "learning_rate": 2.85e-06,
+      "loss": 0.0994,
+      "step": 57
+    },
+    {
+      "epoch": 0.4943181818181818,
+      "grad_norm": 0.5620752573013306,
+      "learning_rate": 2.9e-06,
+      "loss": 0.0909,
+      "step": 58
+    },
+    {
+      "epoch": 0.5028409090909091,
+      "grad_norm": 0.5377550721168518,
+      "learning_rate": 2.95e-06,
+      "loss": 0.087,
+      "step": 59
+    },
+    {
+      "epoch": 0.5113636363636364,
+      "grad_norm": 0.7260486483573914,
+      "learning_rate": 3e-06,
+      "loss": 0.0949,
+      "step": 60
+    },
+    {
+      "epoch": 0.5198863636363636,
+      "grad_norm": 0.5636699199676514,
+      "learning_rate": 3.05e-06,
+      "loss": 0.0884,
+      "step": 61
+    },
+    {
+      "epoch": 0.5284090909090909,
+      "grad_norm": 0.3729614019393921,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.0771,
+      "step": 62
+    },
+    {
+      "epoch": 0.5369318181818182,
+      "grad_norm": 0.5472120046615601,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 0.0887,
+      "step": 63
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.4915490448474884,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.0826,
+      "step": 64
+    },
+    {
+      "epoch": 0.5539772727272727,
+      "grad_norm": 0.4123076796531677,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 0.0834,
+      "step": 65
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.32767826318740845,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.0765,
+      "step": 66
+    },
+    {
+      "epoch": 0.5710227272727273,
+      "grad_norm": 0.34970155358314514,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 0.0832,
+      "step": 67
+    },
+    {
+      "epoch": 0.5795454545454546,
+      "grad_norm": 0.38679587841033936,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.0766,
+      "step": 68
+    },
+    {
+      "epoch": 0.5880681818181818,
+      "grad_norm": 0.5464550852775574,
+      "learning_rate": 3.45e-06,
+      "loss": 0.0748,
+      "step": 69
+    },
+    {
+      "epoch": 0.5965909090909091,
+      "grad_norm": 0.3545376658439636,
+      "learning_rate": 3.5e-06,
+      "loss": 0.0759,
+      "step": 70
+    },
+    {
+      "epoch": 0.6051136363636364,
+      "grad_norm": 0.3532780110836029,
+      "learning_rate": 3.5500000000000003e-06,
+      "loss": 0.0728,
+      "step": 71
+    },
+    {
+      "epoch": 0.6136363636363636,
+      "grad_norm": 0.5024192929267883,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.081,
+      "step": 72
+    },
+    {
+      "epoch": 0.6221590909090909,
+      "grad_norm": 0.3845844268798828,
+      "learning_rate": 3.65e-06,
+      "loss": 0.0689,
+      "step": 73
+    },
+    {
+      "epoch": 0.6306818181818182,
+      "grad_norm": 0.46010249853134155,
+      "learning_rate": 3.7e-06,
+      "loss": 0.0757,
+      "step": 74
+    },
+    {
+      "epoch": 0.6392045454545454,
+      "grad_norm": 0.36048972606658936,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.0758,
+      "step": 75
+    },
+    {
+      "epoch": 0.6477272727272727,
+      "grad_norm": 0.2978876829147339,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.0747,
+      "step": 76
+    },
+    {
+      "epoch": 0.65625,
+      "grad_norm": 0.30899888277053833,
+      "learning_rate": 3.85e-06,
+      "loss": 0.0788,
+      "step": 77
+    },
+    {
+      "epoch": 0.6647727272727273,
+      "grad_norm": 0.45000651478767395,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.0672,
+      "step": 78
+    },
+    {
+      "epoch": 0.6732954545454546,
+      "grad_norm": 0.2621772587299347,
+      "learning_rate": 3.95e-06,
+      "loss": 0.0719,
+      "step": 79
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 0.28618085384368896,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0744,
+      "step": 80
+    },
+    {
+      "epoch": 0.6903409090909091,
+      "grad_norm": 0.40642479062080383,
+      "learning_rate": 4.05e-06,
+      "loss": 0.0743,
+      "step": 81
+    },
+    {
+      "epoch": 0.6988636363636364,
+      "grad_norm": 0.3514344096183777,
+      "learning_rate": 4.1e-06,
+      "loss": 0.0708,
+      "step": 82
+    },
+    {
+      "epoch": 0.7073863636363636,
+      "grad_norm": 0.426798939704895,
+      "learning_rate": 4.15e-06,
+      "loss": 0.073,
+      "step": 83
+    },
+    {
+      "epoch": 0.7159090909090909,
+      "grad_norm": 0.29413217306137085,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.0732,
+      "step": 84
+    },
+    {
+      "epoch": 0.7244318181818182,
+      "grad_norm": 0.37668099999427795,
+      "learning_rate": 4.25e-06,
+      "loss": 0.0664,
+      "step": 85
+    },
+    {
+      "epoch": 0.7329545454545454,
+      "grad_norm": 0.3696061372756958,
+      "learning_rate": 4.3e-06,
+      "loss": 0.0649,
+      "step": 86
+    },
+    {
+      "epoch": 0.7414772727272727,
+      "grad_norm": 0.311988890171051,
+      "learning_rate": 4.350000000000001e-06,
+      "loss": 0.0689,
+      "step": 87
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3180747628211975,
+      "learning_rate": 4.4e-06,
+      "loss": 0.0746,
+      "step": 88
+    },
+    {
+      "epoch": 0.7585227272727273,
+      "grad_norm": 0.46045729517936707,
+      "learning_rate": 4.450000000000001e-06,
+      "loss": 0.0763,
+      "step": 89
+    },
+    {
+      "epoch": 0.7670454545454546,
+      "grad_norm": 0.3566094934940338,
+      "learning_rate": 4.5e-06,
+      "loss": 0.0676,
+      "step": 90
+    },
+    {
+      "epoch": 0.7755681818181818,
+      "grad_norm": 0.35632985830307007,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 0.0677,
+      "step": 91
+    },
+    {
+      "epoch": 0.7840909090909091,
+      "grad_norm": 0.27962526679039,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.0689,
+      "step": 92
+    },
+    {
+      "epoch": 0.7926136363636364,
+      "grad_norm": 0.4532427191734314,
+      "learning_rate": 4.65e-06,
+      "loss": 0.0724,
+      "step": 93
+    },
+    {
+      "epoch": 0.8011363636363636,
+      "grad_norm": 0.363337904214859,
+      "learning_rate": 4.7e-06,
+      "loss": 0.0708,
+      "step": 94
+    },
+    {
+      "epoch": 0.8096590909090909,
+      "grad_norm": 0.3065521717071533,
+      "learning_rate": 4.75e-06,
+      "loss": 0.0713,
+      "step": 95
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 0.24705548584461212,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.0683,
+      "step": 96
+    },
+    {
+      "epoch": 0.8267045454545454,
+      "grad_norm": 0.5038250684738159,
+      "learning_rate": 4.85e-06,
+      "loss": 0.0738,
+      "step": 97
+    },
+    {
+      "epoch": 0.8352272727272727,
+      "grad_norm": 0.2972690761089325,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.0684,
+      "step": 98
+    },
+    {
+      "epoch": 0.84375,
+      "grad_norm": 0.40811270475387573,
+      "learning_rate": 4.95e-06,
+      "loss": 0.066,
+      "step": 99
+    },
+    {
+      "epoch": 0.8522727272727273,
+      "grad_norm": 0.3925577998161316,
+      "learning_rate": 5e-06,
+      "loss": 0.0692,
+      "step": 100
+    },
+    {
+      "epoch": 0.8607954545454546,
+      "grad_norm": 0.32043716311454773,
+      "learning_rate": 4.999965957943338e-06,
+      "loss": 0.0652,
+      "step": 101
+    },
+    {
+      "epoch": 0.8693181818181818,
+      "grad_norm": 0.26025497913360596,
+      "learning_rate": 4.999863832700438e-06,
+      "loss": 0.0634,
+      "step": 102
+    },
+    {
+      "epoch": 0.8778409090909091,
+      "grad_norm": 0.5605457425117493,
+      "learning_rate": 4.999693627052545e-06,
+      "loss": 0.0686,
+      "step": 103
+    },
+    {
+      "epoch": 0.8863636363636364,
+      "grad_norm": 0.48003220558166504,
+      "learning_rate": 4.9994553456349785e-06,
+      "loss": 0.0707,
+      "step": 104
+    },
+    {
+      "epoch": 0.8948863636363636,
+      "grad_norm": 0.22346197068691254,
+      "learning_rate": 4.99914899493701e-06,
+      "loss": 0.067,
+      "step": 105
+    },
+    {
+      "epoch": 0.9034090909090909,
+      "grad_norm": 0.45966312289237976,
+      "learning_rate": 4.998774583301685e-06,
+      "loss": 0.0663,
+      "step": 106
+    },
+    {
+      "epoch": 0.9119318181818182,
+      "grad_norm": 0.38695356249809265,
+      "learning_rate": 4.998332120925598e-06,
+      "loss": 0.0714,
+      "step": 107
+    },
+    {
+      "epoch": 0.9204545454545454,
+      "grad_norm": 0.35708868503570557,
+      "learning_rate": 4.997821619858614e-06,
+      "loss": 0.0632,
+      "step": 108
+    },
+    {
+      "epoch": 0.9289772727272727,
+      "grad_norm": 0.2699528932571411,
+      "learning_rate": 4.9972430940035355e-06,
+      "loss": 0.0649,
+      "step": 109
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 0.5043486952781677,
+      "learning_rate": 4.9965965591157314e-06,
+      "loss": 0.0715,
+      "step": 110
+    },
+    {
+      "epoch": 0.9460227272727273,
+      "grad_norm": 0.4046338200569153,
+      "learning_rate": 4.995882032802703e-06,
+      "loss": 0.0657,
+      "step": 111
+    },
+    {
+      "epoch": 0.9545454545454546,
+      "grad_norm": 0.3543775677680969,
+      "learning_rate": 4.995099534523608e-06,
+      "loss": 0.0698,
+      "step": 112
+    },
+    {
+      "epoch": 0.9630681818181818,
+      "grad_norm": 0.3295219838619232,
+      "learning_rate": 4.994249085588725e-06,
+      "loss": 0.0719,
+      "step": 113
+    },
+    {
+      "epoch": 0.9715909090909091,
+      "grad_norm": 0.3691340982913971,
+      "learning_rate": 4.993330709158879e-06,
+      "loss": 0.0689,
+      "step": 114
+    },
+    {
+      "epoch": 0.9801136363636364,
+      "grad_norm": 0.4219862222671509,
+      "learning_rate": 4.9923444302448095e-06,
+      "loss": 0.0644,
+      "step": 115
+    },
+    {
+      "epoch": 0.9886363636363636,
+      "grad_norm": 0.33437392115592957,
+      "learning_rate": 4.991290275706486e-06,
+      "loss": 0.074,
+      "step": 116
+    },
+    {
+      "epoch": 0.9971590909090909,
+      "grad_norm": 0.22970478236675262,
+      "learning_rate": 4.990168274252379e-06,
+      "loss": 0.0639,
+      "step": 117
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.22970478236675262,
+      "learning_rate": 4.988978456438678e-06,
+      "loss": 0.0718,
+      "step": 118
+    },
+    {
+      "epoch": 1.0085227272727273,
+      "grad_norm": 0.7316553592681885,
+      "learning_rate": 4.98772085466846e-06,
+      "loss": 0.062,
+      "step": 119
+    },
+    {
+      "epoch": 1.0170454545454546,
+      "grad_norm": 0.27226850390434265,
+      "learning_rate": 4.986395503190805e-06,
+      "loss": 0.0583,
+      "step": 120
+    },
+    {
+      "epoch": 1.0255681818181819,
+      "grad_norm": 0.20066767930984497,
+      "learning_rate": 4.9850024380998655e-06,
+      "loss": 0.0624,
+      "step": 121
+    },
+    {
+      "epoch": 1.0340909090909092,
+      "grad_norm": 0.26274779438972473,
+      "learning_rate": 4.9835416973338815e-06,
+      "loss": 0.0641,
+      "step": 122
+    },
+    {
+      "epoch": 1.0426136363636365,
+      "grad_norm": 0.3893499970436096,
+      "learning_rate": 4.982013320674149e-06,
+      "loss": 0.066,
+      "step": 123
+    },
+    {
+      "epoch": 1.0511363636363635,
+      "grad_norm": 0.3454163670539856,
+      "learning_rate": 4.980417349743936e-06,
+      "loss": 0.061,
+      "step": 124
+    },
+    {
+      "epoch": 1.0596590909090908,
+      "grad_norm": 0.26956701278686523,
+      "learning_rate": 4.97875382800735e-06,
+      "loss": 0.0641,
+      "step": 125
+    },
+    {
+      "epoch": 1.0681818181818181,
+      "grad_norm": 0.4147774279117584,
+      "learning_rate": 4.97702280076815e-06,
+      "loss": 0.0599,
+      "step": 126
+    },
+    {
+      "epoch": 1.0767045454545454,
+      "grad_norm": 0.2839421033859253,
+      "learning_rate": 4.9752243151685185e-06,
+      "loss": 0.0669,
+      "step": 127
+    },
+    {
+      "epoch": 1.0852272727272727,
+      "grad_norm": 0.41760239005088806,
+      "learning_rate": 4.973358420187776e-06,
+      "loss": 0.0661,
+      "step": 128
+    },
+    {
+      "epoch": 1.09375,
+      "grad_norm": 0.29833659529685974,
+      "learning_rate": 4.9714251666410426e-06,
+      "loss": 0.0672,
+      "step": 129
+    },
+    {
+      "epoch": 1.1022727272727273,
+      "grad_norm": 0.2806449830532074,
+      "learning_rate": 4.969424607177861e-06,
+      "loss": 0.0589,
+      "step": 130
+    },
+    {
+      "epoch": 1.1107954545454546,
+      "grad_norm": 0.33876338601112366,
+      "learning_rate": 4.967356796280756e-06,
+      "loss": 0.0647,
+      "step": 131
+    },
+    {
+      "epoch": 1.1193181818181819,
+      "grad_norm": 0.2331860363483429,
+      "learning_rate": 4.96522179026376e-06,
+      "loss": 0.0633,
+      "step": 132
+    },
+    {
+      "epoch": 1.1278409090909092,
+      "grad_norm": 0.22244933247566223,
+      "learning_rate": 4.963019647270866e-06,
+      "loss": 0.0618,
+      "step": 133
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 0.24649232625961304,
+      "learning_rate": 4.960750427274458e-06,
+      "loss": 0.0565,
+      "step": 134
+    },
+    {
+      "epoch": 1.1448863636363638,
+      "grad_norm": 0.23454727232456207,
+      "learning_rate": 4.958414192073665e-06,
+      "loss": 0.0607,
+      "step": 135
+    },
+    {
+      "epoch": 1.1534090909090908,
+      "grad_norm": 0.28441867232322693,
+      "learning_rate": 4.956011005292693e-06,
+      "loss": 0.0548,
+      "step": 136
+    },
+    {
+      "epoch": 1.1619318181818181,
+      "grad_norm": 0.26571962237358093,
+      "learning_rate": 4.9535409323790726e-06,
+      "loss": 0.0596,
+      "step": 137
+    },
+    {
+      "epoch": 1.1704545454545454,
+      "grad_norm": 0.2692795991897583,
+      "learning_rate": 4.951004040601898e-06,
+      "loss": 0.0572,
+      "step": 138
+    },
+    {
+      "epoch": 1.1789772727272727,
+      "grad_norm": 0.35209110379219055,
+      "learning_rate": 4.9484003990499785e-06,
+      "loss": 0.0657,
+      "step": 139
+    },
+    {
+      "epoch": 1.1875,
+      "grad_norm": 0.2726614773273468,
+      "learning_rate": 4.945730078629965e-06,
+      "loss": 0.0561,
+      "step": 140
+    },
+    {
+      "epoch": 1.1960227272727273,
+      "grad_norm": 0.2439083456993103,
+      "learning_rate": 4.942993152064415e-06,
+      "loss": 0.0594,
+      "step": 141
+    },
+    {
+      "epoch": 1.2045454545454546,
+      "grad_norm": 0.23282821476459503,
+      "learning_rate": 4.940189693889819e-06,
+      "loss": 0.0598,
+      "step": 142
+    },
+    {
+      "epoch": 1.2130681818181819,
+      "grad_norm": 0.22507373988628387,
+      "learning_rate": 4.937319780454559e-06,
+      "loss": 0.0632,
+      "step": 143
+    },
+    {
+      "epoch": 1.2215909090909092,
+      "grad_norm": 0.35792839527130127,
+      "learning_rate": 4.934383489916843e-06,
+      "loss": 0.0626,
+      "step": 144
+    },
+    {
+      "epoch": 1.2301136363636362,
+      "grad_norm": 0.3073280453681946,
+      "learning_rate": 4.931380902242564e-06,
+      "loss": 0.0558,
+      "step": 145
+    },
+    {
+      "epoch": 1.2386363636363638,
+      "grad_norm": 0.2743082642555237,
+      "learning_rate": 4.928312099203131e-06,
+      "loss": 0.0638,
+      "step": 146
+    },
+    {
+      "epoch": 1.2471590909090908,
+      "grad_norm": 0.2522992491722107,
+      "learning_rate": 4.925177164373237e-06,
+      "loss": 0.0599,
+      "step": 147
+    },
+    {
+      "epoch": 1.2556818181818181,
+      "grad_norm": 0.25230202078819275,
+      "learning_rate": 4.921976183128585e-06,
+      "loss": 0.0579,
+      "step": 148
+    },
+    {
+      "epoch": 1.2642045454545454,
+      "grad_norm": 0.24467583000659943,
+      "learning_rate": 4.918709242643563e-06,
+      "loss": 0.0607,
+      "step": 149
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 0.2191130518913269,
+      "learning_rate": 4.915376431888871e-06,
+      "loss": 0.0622,
+      "step": 150
+    },
+    {
+      "epoch": 1.28125,
+      "grad_norm": 0.26959559321403503,
+      "learning_rate": 4.911977841629092e-06,
+      "loss": 0.0568,
+      "step": 151
+    },
+    {
+      "epoch": 1.2897727272727273,
+      "grad_norm": 0.25139617919921875,
+      "learning_rate": 4.908513564420231e-06,
+      "loss": 0.0607,
+      "step": 152
+    },
+    {
+      "epoch": 1.2982954545454546,
+      "grad_norm": 0.27705878019332886,
+      "learning_rate": 4.904983694607183e-06,
+      "loss": 0.0607,
+      "step": 153
+    },
+    {
+      "epoch": 1.3068181818181819,
+      "grad_norm": 0.2860431373119354,
+      "learning_rate": 4.9013883283211705e-06,
+      "loss": 0.0572,
+      "step": 154
+    },
+    {
+      "epoch": 1.3153409090909092,
+      "grad_norm": 0.27946561574935913,
+      "learning_rate": 4.897727563477123e-06,
+      "loss": 0.054,
+      "step": 155
+    },
+    {
+      "epoch": 1.3238636363636362,
+      "grad_norm": 0.25392627716064453,
+      "learning_rate": 4.894001499771015e-06,
+      "loss": 0.0623,
+      "step": 156
+    },
+    {
+      "epoch": 1.3323863636363638,
+      "grad_norm": 0.24209532141685486,
+      "learning_rate": 4.890210238677141e-06,
+      "loss": 0.0592,
+      "step": 157
+    },
+    {
+      "epoch": 1.3409090909090908,
+      "grad_norm": 0.23799730837345123,
+      "learning_rate": 4.886353883445363e-06,
+      "loss": 0.0545,
+      "step": 158
+    },
+    {
+      "epoch": 1.3494318181818181,
+      "grad_norm": 0.20330385863780975,
+      "learning_rate": 4.88243253909829e-06,
+      "loss": 0.0582,
+      "step": 159
+    },
+    {
+      "epoch": 1.3579545454545454,
+      "grad_norm": 0.2390514761209488,
+      "learning_rate": 4.878446312428424e-06,
+      "loss": 0.0558,
+      "step": 160
+    },
+    {
+      "epoch": 1.3664772727272727,
+      "grad_norm": 0.3132871687412262,
+      "learning_rate": 4.874395311995248e-06,
+      "loss": 0.0522,
+      "step": 161
+    },
+    {
+      "epoch": 1.375,
+      "grad_norm": 0.2441999763250351,
+      "learning_rate": 4.870279648122271e-06,
+      "loss": 0.0632,
+      "step": 162
+    },
+    {
+      "epoch": 1.3835227272727273,
+      "grad_norm": 0.3468320369720459,
+      "learning_rate": 4.866099432894023e-06,
+      "loss": 0.0544,
+      "step": 163
+    },
+    {
+      "epoch": 1.3920454545454546,
+      "grad_norm": 0.2637532949447632,
+      "learning_rate": 4.8618547801530045e-06,
+      "loss": 0.0508,
+      "step": 164
+    },
+    {
+      "epoch": 1.4005681818181819,
+      "grad_norm": 0.25950244069099426,
+      "learning_rate": 4.857545805496581e-06,
+      "loss": 0.0535,
+      "step": 165
+    },
+    {
+      "epoch": 1.4090909090909092,
+      "grad_norm": 0.24120447039604187,
+      "learning_rate": 4.853172626273841e-06,
+      "loss": 0.0563,
+      "step": 166
+    },
+    {
+      "epoch": 1.4176136363636362,
+      "grad_norm": 0.2234521359205246,
+      "learning_rate": 4.848735361582396e-06,
+      "loss": 0.0526,
+      "step": 167
+    },
+    {
+      "epoch": 1.4261363636363638,
+      "grad_norm": 0.24657292664051056,
+      "learning_rate": 4.844234132265139e-06,
+      "loss": 0.0596,
+      "step": 168
+    },
+    {
+      "epoch": 1.4346590909090908,
+      "grad_norm": 0.4169978201389313,
+      "learning_rate": 4.839669060906952e-06,
+      "loss": 0.0569,
+      "step": 169
+    },
+    {
+      "epoch": 1.4431818181818181,
+      "grad_norm": 0.24876588582992554,
+      "learning_rate": 4.835040271831371e-06,
+      "loss": 0.0555,
+      "step": 170
+    },
+    {
+      "epoch": 1.4517045454545454,
+      "grad_norm": 0.2905561923980713,
+      "learning_rate": 4.830347891097193e-06,
+      "loss": 0.0518,
+      "step": 171
+    },
+    {
+      "epoch": 1.4602272727272727,
+      "grad_norm": 0.32290127873420715,
+      "learning_rate": 4.8255920464950545e-06,
+      "loss": 0.0544,
+      "step": 172
+    },
+    {
+      "epoch": 1.46875,
+      "grad_norm": 0.3065219223499298,
+      "learning_rate": 4.820772867543939e-06,
+      "loss": 0.0563,
+      "step": 173
+    },
+    {
+      "epoch": 1.4772727272727273,
+      "grad_norm": 0.3081221878528595,
+      "learning_rate": 4.8158904854876555e-06,
+      "loss": 0.0595,
+      "step": 174
+    },
+    {
+      "epoch": 1.4857954545454546,
+      "grad_norm": 0.3171551823616028,
+      "learning_rate": 4.810945033291267e-06,
+      "loss": 0.0509,
+      "step": 175
+    },
+    {
+      "epoch": 1.4943181818181819,
+      "grad_norm": 0.46660706400871277,
+      "learning_rate": 4.805936645637463e-06,
+      "loss": 0.0575,
+      "step": 176
+    },
+    {
+      "epoch": 1.5028409090909092,
+      "grad_norm": 0.3173319399356842,
+      "learning_rate": 4.800865458922899e-06,
+      "loss": 0.0579,
+      "step": 177
+    },
+    {
+      "epoch": 1.5113636363636362,
+      "grad_norm": 0.2553943693637848,
+      "learning_rate": 4.795731611254473e-06,
+      "loss": 0.0602,
+      "step": 178
+    },
+    {
+      "epoch": 1.5198863636363638,
+      "grad_norm": 0.39231568574905396,
+      "learning_rate": 4.790535242445573e-06,
+      "loss": 0.0606,
+      "step": 179
+    },
+    {
+      "epoch": 1.5284090909090908,
+      "grad_norm": 0.2721359133720398,
+      "learning_rate": 4.7852764940122636e-06,
+      "loss": 0.0545,
+      "step": 180
+    },
+    {
+      "epoch": 1.5369318181818183,
+      "grad_norm": 0.2545195519924164,
+      "learning_rate": 4.779955509169435e-06,
+      "loss": 0.0546,
+      "step": 181
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 0.3155561685562134,
+      "learning_rate": 4.7745724328269e-06,
+      "loss": 0.0543,
+      "step": 182
+    },
+    {
+      "epoch": 1.5539772727272727,
+      "grad_norm": 0.24933341145515442,
+      "learning_rate": 4.769127411585452e-06,
+      "loss": 0.0571,
+      "step": 183
+    },
+    {
+      "epoch": 1.5625,
+      "grad_norm": 0.2886108458042145,
+      "learning_rate": 4.763620593732867e-06,
+      "loss": 0.0521,
+      "step": 184
+    },
+    {
+      "epoch": 1.5710227272727273,
+      "grad_norm": 0.23579014837741852,
+      "learning_rate": 4.75805212923987e-06,
+      "loss": 0.0551,
+      "step": 185
+    },
+    {
+      "epoch": 1.5795454545454546,
+      "grad_norm": 0.2821877896785736,
+      "learning_rate": 4.752422169756048e-06,
+      "loss": 0.0494,
+      "step": 186
+    },
+    {
+      "epoch": 1.5880681818181817,
+      "grad_norm": 0.28472304344177246,
+      "learning_rate": 4.746730868605721e-06,
+      "loss": 0.0524,
+      "step": 187
+    },
+    {
+      "epoch": 1.5965909090909092,
+      "grad_norm": 0.30284813046455383,
+      "learning_rate": 4.7409783807837654e-06,
+      "loss": 0.0519,
+      "step": 188
+    },
+    {
+      "epoch": 1.6051136363636362,
+      "grad_norm": 0.2568075656890869,
+      "learning_rate": 4.735164862951395e-06,
+      "loss": 0.0532,
+      "step": 189
+    },
+    {
+      "epoch": 1.6136363636363638,
+      "grad_norm": 0.28199082612991333,
+      "learning_rate": 4.729290473431892e-06,
+      "loss": 0.0578,
+      "step": 190
+    },
+    {
+      "epoch": 1.6221590909090908,
+      "grad_norm": 0.3818947970867157,
+      "learning_rate": 4.723355372206297e-06,
+      "loss": 0.0552,
+      "step": 191
+    },
+    {
+      "epoch": 1.6306818181818183,
+      "grad_norm": 0.26314595341682434,
+      "learning_rate": 4.717359720909053e-06,
+      "loss": 0.0505,
+      "step": 192
+    },
+    {
+      "epoch": 1.6392045454545454,
+      "grad_norm": 0.2616359293460846,
+      "learning_rate": 4.7113036828235995e-06,
+      "loss": 0.0511,
+      "step": 193
+    },
+    {
+      "epoch": 1.6477272727272727,
+      "grad_norm": 0.33692309260368347,
+      "learning_rate": 4.705187422877931e-06,
+      "loss": 0.0554,
+      "step": 194
+    },
+    {
+      "epoch": 1.65625,
+      "grad_norm": 0.30861467123031616,
+      "learning_rate": 4.6990111076401e-06,
+      "loss": 0.0485,
+      "step": 195
+    },
+    {
+      "epoch": 1.6647727272727273,
+      "grad_norm": 0.251370370388031,
+      "learning_rate": 4.692774905313687e-06,
+      "loss": 0.0454,
+      "step": 196
+    },
+    {
+      "epoch": 1.6732954545454546,
+      "grad_norm": 0.33005693554878235,
+      "learning_rate": 4.686478985733212e-06,
+      "loss": 0.0509,
+      "step": 197
+    },
+    {
+      "epoch": 1.6818181818181817,
+      "grad_norm": 0.39172685146331787,
+      "learning_rate": 4.68012352035952e-06,
+      "loss": 0.0552,
+      "step": 198
+    },
+    {
+      "epoch": 1.6903409090909092,
+      "grad_norm": 0.5113612413406372,
+      "learning_rate": 4.673708682275097e-06,
+      "loss": 0.0537,
+      "step": 199
+    },
+    {
+      "epoch": 1.6988636363636362,
+      "grad_norm": 0.27660125494003296,
+      "learning_rate": 4.667234646179368e-06,
+      "loss": 0.0463,
+      "step": 200
+    },
+    {
+      "epoch": 1.7073863636363638,
+      "grad_norm": 0.38760706782341003,
+      "learning_rate": 4.660701588383939e-06,
+      "loss": 0.0496,
+      "step": 201
+    },
+    {
+      "epoch": 1.7159090909090908,
+      "grad_norm": 0.3487601578235626,
+      "learning_rate": 4.654109686807787e-06,
+      "loss": 0.0498,
+      "step": 202
+    },
+    {
+      "epoch": 1.7244318181818183,
+      "grad_norm": 0.2964484393596649,
+      "learning_rate": 4.647459120972422e-06,
+      "loss": 0.0501,
+      "step": 203
+    },
+    {
+      "epoch": 1.7329545454545454,
+      "grad_norm": 0.3318927586078644,
+      "learning_rate": 4.640750071996995e-06,
+      "loss": 0.0535,
+      "step": 204
+    },
+    {
+      "epoch": 1.7414772727272727,
+      "grad_norm": 0.3152824640274048,
+      "learning_rate": 4.633982722593367e-06,
+      "loss": 0.0537,
+      "step": 205
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.3267940878868103,
+      "learning_rate": 4.62715725706113e-06,
+      "loss": 0.0434,
+      "step": 206
+    },
+    {
+      "epoch": 1.7585227272727273,
+      "grad_norm": 0.32321202754974365,
+      "learning_rate": 4.620273861282593e-06,
+      "loss": 0.055,
+      "step": 207
+    },
+    {
+      "epoch": 1.7670454545454546,
+      "grad_norm": 0.3501484990119934,
+      "learning_rate": 4.613332722717714e-06,
+      "loss": 0.0458,
+      "step": 208
+    },
+    {
+      "epoch": 1.7755681818181817,
+      "grad_norm": 0.29107236862182617,
+      "learning_rate": 4.606334030399e-06,
+      "loss": 0.0496,
+      "step": 209
+    },
+    {
+      "epoch": 1.7840909090909092,
+      "grad_norm": 0.3194922208786011,
+      "learning_rate": 4.599277974926355e-06,
+      "loss": 0.0532,
+      "step": 210
+    },
+    {
+      "epoch": 1.7926136363636362,
+      "grad_norm": 0.2896871864795685,
+      "learning_rate": 4.592164748461892e-06,
+      "loss": 0.0482,
+      "step": 211
+    },
+    {
+      "epoch": 1.8011363636363638,
+      "grad_norm": 0.2559587061405182,
+      "learning_rate": 4.584994544724695e-06,
+      "loss": 0.0474,
+      "step": 212
+    },
+    {
+      "epoch": 1.8096590909090908,
+      "grad_norm": 0.3188915252685547,
+      "learning_rate": 4.577767558985552e-06,
+      "loss": 0.0454,
+      "step": 213
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.31643223762512207,
+      "learning_rate": 4.57048398806163e-06,
+      "loss": 0.0503,
+      "step": 214
+    },
+    {
+      "epoch": 1.8267045454545454,
+      "grad_norm": 0.3251838684082031,
+      "learning_rate": 4.563144030311114e-06,
+      "loss": 0.0484,
+      "step": 215
+    },
+    {
+      "epoch": 1.8352272727272727,
+      "grad_norm": 0.26603975892066956,
+      "learning_rate": 4.555747885627812e-06,
+      "loss": 0.0442,
+      "step": 216
+    },
+    {
+      "epoch": 1.84375,
+      "grad_norm": 0.4656536877155304,
+      "learning_rate": 4.548295755435703e-06,
+      "loss": 0.0428,
+      "step": 217
+    },
+    {
+      "epoch": 1.8522727272727273,
+      "grad_norm": 0.32816216349601746,
+      "learning_rate": 4.540787842683459e-06,
+      "loss": 0.0455,
+      "step": 218
+    },
+    {
+      "epoch": 1.8607954545454546,
+      "grad_norm": 0.31960371136665344,
+      "learning_rate": 4.533224351838914e-06,
+      "loss": 0.0454,
+      "step": 219
+    },
+    {
+      "epoch": 1.8693181818181817,
+      "grad_norm": 0.33228686451911926,
+      "learning_rate": 4.525605488883493e-06,
+      "loss": 0.0413,
+      "step": 220
+    },
+    {
+      "epoch": 1.8778409090909092,
+      "grad_norm": 0.271839439868927,
+      "learning_rate": 4.517931461306609e-06,
+      "loss": 0.0462,
+      "step": 221
+    },
+    {
+      "epoch": 1.8863636363636362,
+      "grad_norm": 0.3333345949649811,
+      "learning_rate": 4.510202478100008e-06,
+      "loss": 0.0458,
+      "step": 222
+    },
+    {
+      "epoch": 1.8948863636363638,
+      "grad_norm": 0.2923465371131897,
+      "learning_rate": 4.502418749752076e-06,
+      "loss": 0.0407,
+      "step": 223
+    },
+    {
+      "epoch": 1.9034090909090908,
+      "grad_norm": 0.34088781476020813,
+      "learning_rate": 4.494580488242109e-06,
+      "loss": 0.0422,
+      "step": 224
+    },
+    {
+      "epoch": 1.9119318181818183,
+      "grad_norm": 0.3535335958003998,
+      "learning_rate": 4.486687907034544e-06,
+      "loss": 0.0454,
+      "step": 225
+    },
+    {
+      "epoch": 1.9204545454545454,
+      "grad_norm": 0.32973766326904297,
+      "learning_rate": 4.478741221073136e-06,
+      "loss": 0.0429,
+      "step": 226
+    },
+    {
+      "epoch": 1.9289772727272727,
+      "grad_norm": 0.3509384095668793,
+      "learning_rate": 4.470740646775113e-06,
+      "loss": 0.0436,
+      "step": 227
+    },
+    {
+      "epoch": 1.9375,
+      "grad_norm": 0.2931392788887024,
+      "learning_rate": 4.462686402025277e-06,
+      "loss": 0.0466,
+      "step": 228
+    },
+    {
+      "epoch": 1.9460227272727273,
+      "grad_norm": 0.30415666103363037,
+      "learning_rate": 4.454578706170075e-06,
+      "loss": 0.0416,
+      "step": 229
+    },
+    {
+      "epoch": 1.9545454545454546,
+      "grad_norm": 0.3433127701282501,
+      "learning_rate": 4.446417780011618e-06,
+      "loss": 0.0424,
+      "step": 230
+    },
+    {
+      "epoch": 1.9630681818181817,
+      "grad_norm": 0.352159708738327,
+      "learning_rate": 4.4382038458016764e-06,
+      "loss": 0.0404,
+      "step": 231
+    },
+    {
+      "epoch": 1.9715909090909092,
+      "grad_norm": 0.27307572960853577,
+      "learning_rate": 4.42993712723562e-06,
+      "loss": 0.0429,
+      "step": 232
+    },
+    {
+      "epoch": 1.9801136363636362,
+      "grad_norm": 0.2662625312805176,
+      "learning_rate": 4.4216178494463305e-06,
+      "loss": 0.0425,
+      "step": 233
+    },
+    {
+      "epoch": 1.9886363636363638,
+      "grad_norm": 0.29612037539482117,
+      "learning_rate": 4.413246238998069e-06,
+      "loss": 0.0463,
+      "step": 234
+    },
+    {
+      "epoch": 1.9971590909090908,
+      "grad_norm": 0.33151501417160034,
+      "learning_rate": 4.404822523880305e-06,
+      "loss": 0.0441,
+      "step": 235
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.33151501417160034,
+      "learning_rate": 4.396346933501508e-06,
+      "loss": 0.0494,
+      "step": 236
+    },
+    {
+      "epoch": 2.008522727272727,
+      "grad_norm": 0.8211016058921814,
+      "learning_rate": 4.3878196986829015e-06,
+      "loss": 0.0394,
+      "step": 237
+    },
+    {
+      "epoch": 2.0170454545454546,
+      "grad_norm": 0.28689032793045044,
+      "learning_rate": 4.379241051652174e-06,
+      "loss": 0.0379,
+      "step": 238
+    },
+    {
+      "epoch": 2.0255681818181817,
+      "grad_norm": 0.3541617691516876,
+      "learning_rate": 4.370611226037155e-06,
+      "loss": 0.0363,
+      "step": 239
+    },
+    {
+      "epoch": 2.034090909090909,
+      "grad_norm": 0.32633963227272034,
+      "learning_rate": 4.361930456859455e-06,
+      "loss": 0.0403,
+      "step": 240
+    },
+    {
+      "epoch": 2.0426136363636362,
+      "grad_norm": 0.30213046073913574,
+      "learning_rate": 4.353198980528063e-06,
+      "loss": 0.0393,
+      "step": 241
+    },
+    {
+      "epoch": 2.0511363636363638,
+      "grad_norm": 0.3540145456790924,
+      "learning_rate": 4.3444170348329095e-06,
+      "loss": 0.035,
+      "step": 242
+    },
+    {
+      "epoch": 2.059659090909091,
+      "grad_norm": 0.2990974485874176,
+      "learning_rate": 4.335584858938388e-06,
+      "loss": 0.0346,
+      "step": 243
+    },
+    {
+      "epoch": 2.0681818181818183,
+      "grad_norm": 0.4325309991836548,
+      "learning_rate": 4.326702693376844e-06,
+      "loss": 0.0351,
+      "step": 244
+    },
+    {
+      "epoch": 2.0767045454545454,
+      "grad_norm": 0.31114035844802856,
+      "learning_rate": 4.317770780042025e-06,
+      "loss": 0.0341,
+      "step": 245
+    },
+    {
+      "epoch": 2.085227272727273,
+      "grad_norm": 0.30371102690696716,
+      "learning_rate": 4.308789362182492e-06,
+      "loss": 0.0373,
+      "step": 246
+    },
+    {
+      "epoch": 2.09375,
+      "grad_norm": 0.40639927983283997,
+      "learning_rate": 4.2997586843949905e-06,
+      "loss": 0.0333,
+      "step": 247
+    },
+    {
+      "epoch": 2.102272727272727,
+      "grad_norm": 0.33129221200942993,
+      "learning_rate": 4.290678992617797e-06,
+      "loss": 0.0323,
+      "step": 248
+    },
+    {
+      "epoch": 2.1107954545454546,
+      "grad_norm": 0.3084648549556732,
+      "learning_rate": 4.28155053412402e-06,
+      "loss": 0.0355,
+      "step": 249
+    },
+    {
+      "epoch": 2.1193181818181817,
+      "grad_norm": 0.3743211328983307,
+      "learning_rate": 4.2723735575148585e-06,
+      "loss": 0.0367,
+      "step": 250
+    },
+    {
+      "epoch": 2.127840909090909,
+      "grad_norm": 0.354165256023407,
+      "learning_rate": 4.263148312712838e-06,
+      "loss": 0.0293,
+      "step": 251
+    },
+    {
+      "epoch": 2.1363636363636362,
+      "grad_norm": 0.37607696652412415,
+      "learning_rate": 4.253875050955005e-06,
+      "loss": 0.0318,
+      "step": 252
+    },
+    {
+      "epoch": 2.1448863636363638,
+      "grad_norm": 0.33469027280807495,
+      "learning_rate": 4.2445540247860805e-06,
+      "loss": 0.038,
+      "step": 253
+    },
+    {
+      "epoch": 2.153409090909091,
+      "grad_norm": 0.32785046100616455,
+      "learning_rate": 4.2351854880515856e-06,
+      "loss": 0.0303,
+      "step": 254
+    },
+    {
+      "epoch": 2.1619318181818183,
+      "grad_norm": 0.41809284687042236,
+      "learning_rate": 4.2257696958909255e-06,
+      "loss": 0.0339,
+      "step": 255
+    },
+    {
+      "epoch": 2.1704545454545454,
+      "grad_norm": 0.3557347059249878,
+      "learning_rate": 4.216306904730448e-06,
+      "loss": 0.0347,
+      "step": 256
+    },
+    {
+      "epoch": 2.178977272727273,
+      "grad_norm": 0.3383937478065491,
+      "learning_rate": 4.206797372276447e-06,
+      "loss": 0.0384,
+      "step": 257
+    },
+    {
+      "epoch": 2.1875,
+      "grad_norm": 0.339476615190506,
+      "learning_rate": 4.197241357508159e-06,
+      "loss": 0.0351,
+      "step": 258
+    },
+    {
+      "epoch": 2.196022727272727,
+      "grad_norm": 0.4095429480075836,
+      "learning_rate": 4.187639120670702e-06,
+      "loss": 0.0339,
+      "step": 259
+    },
+    {
+      "epoch": 2.2045454545454546,
+      "grad_norm": 0.3317232131958008,
+      "learning_rate": 4.177990923267986e-06,
+      "loss": 0.0363,
+      "step": 260
+    },
+    {
+      "epoch": 2.2130681818181817,
+      "grad_norm": 0.4111759066581726,
+      "learning_rate": 4.168297028055599e-06,
+      "loss": 0.0369,
+      "step": 261
+    },
+    {
+      "epoch": 2.221590909090909,
+      "grad_norm": 0.4489494860172272,
+      "learning_rate": 4.158557699033644e-06,
+      "loss": 0.0312,
+      "step": 262
+    },
+    {
+      "epoch": 2.2301136363636362,
+      "grad_norm": 0.3993566930294037,
+      "learning_rate": 4.148773201439553e-06,
+      "loss": 0.0306,
+      "step": 263
+    },
+    {
+      "epoch": 2.2386363636363638,
+      "grad_norm": 0.32782599329948425,
+      "learning_rate": 4.138943801740865e-06,
+      "loss": 0.0329,
+      "step": 264
+    },
+    {
+      "epoch": 2.247159090909091,
+      "grad_norm": 0.26705044507980347,
+      "learning_rate": 4.129069767627963e-06,
+      "loss": 0.0341,
+      "step": 265
+    },
+    {
+      "epoch": 2.2556818181818183,
+      "grad_norm": 0.32875779271125793,
+      "learning_rate": 4.119151368006793e-06,
+      "loss": 0.0301,
+      "step": 266
+    },
+    {
+      "epoch": 2.2642045454545454,
+      "grad_norm": 0.42058393359184265,
+      "learning_rate": 4.109188872991529e-06,
+      "loss": 0.0346,
+      "step": 267
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 0.44719040393829346,
+      "learning_rate": 4.099182553897228e-06,
+      "loss": 0.0297,
+      "step": 268
+    },
+    {
+      "epoch": 2.28125,
+      "grad_norm": 0.35350751876831055,
+      "learning_rate": 4.089132683232437e-06,
+      "loss": 0.0285,
+      "step": 269
+    },
+    {
+      "epoch": 2.2897727272727275,
+      "grad_norm": 0.5518717169761658,
+      "learning_rate": 4.0790395346917674e-06,
+      "loss": 0.0333,
+      "step": 270
+    },
+    {
+      "epoch": 2.2982954545454546,
+      "grad_norm": 0.4138563573360443,
+      "learning_rate": 4.068903383148448e-06,
+      "loss": 0.0385,
+      "step": 271
+    },
+    {
+      "epoch": 2.3068181818181817,
+      "grad_norm": 0.406649112701416,
+      "learning_rate": 4.058724504646834e-06,
+      "loss": 0.0298,
+      "step": 272
+    },
+    {
+      "epoch": 2.315340909090909,
+      "grad_norm": 0.3303579092025757,
+      "learning_rate": 4.0485031763948935e-06,
+      "loss": 0.0339,
+      "step": 273
+    },
+    {
+      "epoch": 2.3238636363636362,
+      "grad_norm": 0.35349851846694946,
+      "learning_rate": 4.038239676756654e-06,
+      "loss": 0.0308,
+      "step": 274
+    },
+    {
+      "epoch": 2.3323863636363638,
+      "grad_norm": 0.4232323467731476,
+      "learning_rate": 4.027934285244624e-06,
+      "loss": 0.0307,
+      "step": 275
+    },
+    {
+      "epoch": 2.340909090909091,
+      "grad_norm": 0.35764482617378235,
+      "learning_rate": 4.017587282512181e-06,
+      "loss": 0.0296,
+      "step": 276
+    },
+    {
+      "epoch": 2.3494318181818183,
+      "grad_norm": 0.3608599603176117,
+      "learning_rate": 4.007198950345929e-06,
+      "loss": 0.0346,
+      "step": 277
+    },
+    {
+      "epoch": 2.3579545454545454,
+      "grad_norm": 0.530514121055603,
+      "learning_rate": 3.996769571658022e-06,
+      "loss": 0.0293,
+      "step": 278
+    },
+    {
+      "epoch": 2.3664772727272725,
+      "grad_norm": 0.3813832998275757,
+      "learning_rate": 3.9862994304784585e-06,
+      "loss": 0.0279,
+      "step": 279
+    },
+    {
+      "epoch": 2.375,
+      "grad_norm": 0.48923951387405396,
+      "learning_rate": 3.975788811947351e-06,
+      "loss": 0.0284,
+      "step": 280
+    },
+    {
+      "epoch": 2.3835227272727275,
+      "grad_norm": 0.3939470946788788,
+      "learning_rate": 3.965238002307156e-06,
+      "loss": 0.0315,
+      "step": 281
+    },
+    {
+      "epoch": 2.3920454545454546,
+      "grad_norm": 0.2854841649532318,
+      "learning_rate": 3.9546472888948825e-06,
+      "loss": 0.0288,
+      "step": 282
+    },
+    {
+      "epoch": 2.4005681818181817,
+      "grad_norm": 0.39930015802383423,
+      "learning_rate": 3.944016960134262e-06,
+      "loss": 0.0304,
+      "step": 283
+    },
+    {
+      "epoch": 2.409090909090909,
+      "grad_norm": 0.4333280026912689,
+      "learning_rate": 3.933347305527898e-06,
+      "loss": 0.0282,
+      "step": 284
+    },
+    {
+      "epoch": 2.4176136363636362,
+      "grad_norm": 0.34740084409713745,
+      "learning_rate": 3.922638615649381e-06,
+      "loss": 0.0313,
+      "step": 285
+    },
+    {
+      "epoch": 2.4261363636363638,
+      "grad_norm": 0.4125167727470398,
+      "learning_rate": 3.911891182135371e-06,
+      "loss": 0.0301,
+      "step": 286
+    },
+    {
+      "epoch": 2.434659090909091,
+      "grad_norm": 0.350911945104599,
+      "learning_rate": 3.9011052976776635e-06,
+      "loss": 0.0255,
+      "step": 287
+    },
+    {
+      "epoch": 2.4431818181818183,
+      "grad_norm": 0.2700820863246918,
+      "learning_rate": 3.890281256015207e-06,
+      "loss": 0.0296,
+      "step": 288
+    },
+    {
+      "epoch": 2.4517045454545454,
+      "grad_norm": 0.3823346197605133,
+      "learning_rate": 3.879419351926115e-06,
+      "loss": 0.0279,
+      "step": 289
+    },
+    {
+      "epoch": 2.4602272727272725,
+      "grad_norm": 0.4576987326145172,
+      "learning_rate": 3.868519881219631e-06,
+      "loss": 0.0298,
+      "step": 290
+    },
+    {
+      "epoch": 2.46875,
+      "grad_norm": 0.319879412651062,
+      "learning_rate": 3.857583140728075e-06,
+      "loss": 0.0313,
+      "step": 291
+    },
+    {
+      "epoch": 2.4772727272727275,
+      "grad_norm": 0.36349669098854065,
+      "learning_rate": 3.8466094282987575e-06,
+      "loss": 0.0266,
+      "step": 292
+    },
+    {
+      "epoch": 2.4857954545454546,
+      "grad_norm": 0.3291328251361847,
+      "learning_rate": 3.835599042785872e-06,
+      "loss": 0.0292,
+      "step": 293
+    },
+    {
+      "epoch": 2.4943181818181817,
+      "grad_norm": 0.36020705103874207,
+      "learning_rate": 3.824552284042351e-06,
+      "loss": 0.0295,
+      "step": 294
+    },
+    {
+      "epoch": 2.502840909090909,
+      "grad_norm": 0.3137778341770172,
+      "learning_rate": 3.8134694529117045e-06,
+      "loss": 0.0244,
+      "step": 295
+    },
+    {
+      "epoch": 2.5113636363636362,
+      "grad_norm": 0.36281993985176086,
+      "learning_rate": 3.802350851219826e-06,
+      "loss": 0.0272,
+      "step": 296
+    },
+    {
+      "epoch": 2.5198863636363638,
+      "grad_norm": 0.3342132270336151,
+      "learning_rate": 3.79119678176677e-06,
+      "loss": 0.0288,
+      "step": 297
+    },
+    {
+      "epoch": 2.528409090909091,
+      "grad_norm": 0.45444363355636597,
+      "learning_rate": 3.7800075483185073e-06,
+      "loss": 0.0282,
+      "step": 298
+    },
+    {
+      "epoch": 2.5369318181818183,
+      "grad_norm": 0.4871613085269928,
+      "learning_rate": 3.7687834555986537e-06,
+      "loss": 0.0334,
+      "step": 299
+    },
+    {
+      "epoch": 2.5454545454545454,
+      "grad_norm": 0.49083876609802246,
+      "learning_rate": 3.7575248092801686e-06,
+      "loss": 0.0229,
+      "step": 300
+    },
+    {
+      "epoch": 2.5539772727272725,
+      "grad_norm": 0.3823300004005432,
+      "learning_rate": 3.7462319159770344e-06,
+      "loss": 0.023,
+      "step": 301
+    },
+    {
+      "epoch": 2.5625,
+      "grad_norm": 0.38027557730674744,
+      "learning_rate": 3.734905083235901e-06,
+      "loss": 0.025,
+      "step": 302
+    },
+    {
+      "epoch": 2.5710227272727275,
+      "grad_norm": 0.42017418146133423,
+      "learning_rate": 3.723544619527714e-06,
+      "loss": 0.03,
+      "step": 303
+    },
+    {
+      "epoch": 2.5795454545454546,
+      "grad_norm": 0.4550953209400177,
+      "learning_rate": 3.712150834239313e-06,
+      "loss": 0.0301,
+      "step": 304
+    },
+    {
+      "epoch": 2.5880681818181817,
+      "grad_norm": 0.4527112543582916,
+      "learning_rate": 3.7007240376650054e-06,
+      "loss": 0.0235,
+      "step": 305
+    },
+    {
+      "epoch": 2.596590909090909,
+      "grad_norm": 0.36789700388908386,
+      "learning_rate": 3.6892645409981166e-06,
+      "loss": 0.0231,
+      "step": 306
+    },
+    {
+      "epoch": 2.6051136363636362,
+      "grad_norm": 0.6067688465118408,
+      "learning_rate": 3.6777726563225147e-06,
+      "loss": 0.0256,
+      "step": 307
+    },
+    {
+      "epoch": 2.6136363636363638,
+      "grad_norm": 0.35337361693382263,
+      "learning_rate": 3.6662486966041104e-06,
+      "loss": 0.0244,
+      "step": 308
+    },
+    {
+      "epoch": 2.622159090909091,
+      "grad_norm": 0.4097234010696411,
+      "learning_rate": 3.654692975682336e-06,
+      "loss": 0.0262,
+      "step": 309
+    },
+    {
+      "epoch": 2.6306818181818183,
+      "grad_norm": 0.5017573833465576,
+      "learning_rate": 3.6431058082615966e-06,
+      "loss": 0.0245,
+      "step": 310
+    },
+    {
+      "epoch": 2.6392045454545454,
+      "grad_norm": 0.40696877241134644,
+      "learning_rate": 3.6314875099027e-06,
+      "loss": 0.023,
+      "step": 311
+    },
+    {
+      "epoch": 2.6477272727272725,
+      "grad_norm": 0.35261037945747375,
+      "learning_rate": 3.619838397014263e-06,
+      "loss": 0.0232,
+      "step": 312
+    },
+    {
+      "epoch": 2.65625,
+      "grad_norm": 0.3590414524078369,
+      "learning_rate": 3.6081587868440944e-06,
+      "loss": 0.0242,
+      "step": 313
+    },
+    {
+      "epoch": 2.6647727272727275,
+      "grad_norm": 0.37196213006973267,
+      "learning_rate": 3.5964489974705553e-06,
+      "loss": 0.0222,
+      "step": 314
+    },
+    {
+      "epoch": 2.6732954545454546,
+      "grad_norm": 0.3922335207462311,
+      "learning_rate": 3.5847093477938955e-06,
+      "loss": 0.0236,
+      "step": 315
+    },
+    {
+      "epoch": 2.6818181818181817,
+      "grad_norm": 0.3914213180541992,
+      "learning_rate": 3.5729401575275724e-06,
+      "loss": 0.025,
+      "step": 316
+    },
+    {
+      "epoch": 2.690340909090909,
+      "grad_norm": 0.4622058570384979,
+      "learning_rate": 3.561141747189538e-06,
+      "loss": 0.0252,
+      "step": 317
+    },
+    {
+      "epoch": 2.6988636363636362,
+      "grad_norm": 0.5096047520637512,
+      "learning_rate": 3.5493144380935155e-06,
+      "loss": 0.0242,
+      "step": 318
+    },
+    {
+      "epoch": 2.7073863636363638,
+      "grad_norm": 0.574635922908783,
+      "learning_rate": 3.537458552340247e-06,
+      "loss": 0.0224,
+      "step": 319
+    },
+    {
+      "epoch": 2.715909090909091,
+      "grad_norm": 0.6871437430381775,
+      "learning_rate": 3.5255744128087175e-06,
+      "loss": 0.0271,
+      "step": 320
+    },
+    {
+      "epoch": 2.7244318181818183,
+      "grad_norm": 0.48472297191619873,
+      "learning_rate": 3.5136623431473704e-06,
+      "loss": 0.0246,
+      "step": 321
+    },
+    {
+      "epoch": 2.7329545454545454,
+      "grad_norm": 0.5469712018966675,
+      "learning_rate": 3.501722667765286e-06,
+      "loss": 0.0252,
+      "step": 322
+    },
+    {
+      "epoch": 2.7414772727272725,
+      "grad_norm": 0.4378078877925873,
+      "learning_rate": 3.489755711823348e-06,
+      "loss": 0.0277,
+      "step": 323
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.4401378035545349,
+      "learning_rate": 3.47776180122539e-06,
+      "loss": 0.0235,
+      "step": 324
+    },
+    {
+      "epoch": 2.7585227272727275,
+      "grad_norm": 0.6726410388946533,
+      "learning_rate": 3.4657412626093204e-06,
+      "loss": 0.0207,
+      "step": 325
+    },
+    {
+      "epoch": 2.7670454545454546,
+      "grad_norm": 0.3676104247570038,
+      "learning_rate": 3.4536944233382248e-06,
+      "loss": 0.0196,
+      "step": 326
+    },
+    {
+      "epoch": 2.7755681818181817,
+      "grad_norm": 0.4041842222213745,
+      "learning_rate": 3.4416216114914493e-06,
+      "loss": 0.0275,
+      "step": 327
+    },
+    {
+      "epoch": 2.784090909090909,
+      "grad_norm": 0.7207781076431274,
+      "learning_rate": 3.429523155855672e-06,
+      "loss": 0.0243,
+      "step": 328
+    },
+    {
+      "epoch": 2.7926136363636362,
+      "grad_norm": 0.6210029125213623,
+      "learning_rate": 3.417399385915941e-06,
+      "loss": 0.0207,
+      "step": 329
+    },
+    {
+      "epoch": 2.8011363636363638,
+      "grad_norm": 0.49545082449913025,
+      "learning_rate": 3.405250631846708e-06,
+      "loss": 0.0243,
+      "step": 330
+    },
+    {
+      "epoch": 2.809659090909091,
+      "grad_norm": 0.5395477414131165,
+      "learning_rate": 3.393077224502832e-06,
+      "loss": 0.0227,
+      "step": 331
+    },
+    {
+      "epoch": 2.8181818181818183,
+      "grad_norm": 0.4645976126194,
+      "learning_rate": 3.3808794954105716e-06,
+      "loss": 0.0192,
+      "step": 332
+    },
+    {
+      "epoch": 2.8267045454545454,
+      "grad_norm": 0.3877430558204651,
+      "learning_rate": 3.368657776758555e-06,
+      "loss": 0.0219,
+      "step": 333
+    },
+    {
+      "epoch": 2.8352272727272725,
+      "grad_norm": 0.5611310005187988,
+      "learning_rate": 3.3564124013887324e-06,
+      "loss": 0.0241,
+      "step": 334
+    },
+    {
+      "epoch": 2.84375,
+      "grad_norm": 0.4889032244682312,
+      "learning_rate": 3.3441437027873135e-06,
+      "loss": 0.0181,
+      "step": 335
+    },
+    {
+      "epoch": 2.8522727272727275,
+      "grad_norm": 0.47838833928108215,
+      "learning_rate": 3.331852015075685e-06,
+      "loss": 0.0192,
+      "step": 336
+    },
+    {
+      "epoch": 2.8607954545454546,
+      "grad_norm": 0.4922442138195038,
+      "learning_rate": 3.3195376730013107e-06,
+      "loss": 0.0197,
+      "step": 337
+    },
+    {
+      "epoch": 2.8693181818181817,
+      "grad_norm": 0.3534923493862152,
+      "learning_rate": 3.3072010119286156e-06,
+      "loss": 0.0221,
+      "step": 338
+    },
+    {
+      "epoch": 2.877840909090909,
+      "grad_norm": 0.4476766884326935,
+      "learning_rate": 3.294842367829851e-06,
+      "loss": 0.0167,
+      "step": 339
+    },
+    {
+      "epoch": 2.8863636363636362,
+      "grad_norm": 0.4423317313194275,
+      "learning_rate": 3.2824620772759475e-06,
+      "loss": 0.0209,
+      "step": 340
+    },
+    {
+      "epoch": 2.8948863636363638,
+      "grad_norm": 0.3954547345638275,
+      "learning_rate": 3.270060477427347e-06,
+      "loss": 0.0209,
+      "step": 341
+    },
+    {
+      "epoch": 2.903409090909091,
+      "grad_norm": 0.47028860449790955,
+      "learning_rate": 3.257637906024822e-06,
+      "loss": 0.0189,
+      "step": 342
+    },
+    {
+      "epoch": 2.9119318181818183,
+      "grad_norm": 0.42390963435173035,
+      "learning_rate": 3.2451947013802746e-06,
+      "loss": 0.015,
+      "step": 343
+    },
+    {
+      "epoch": 2.9204545454545454,
+      "grad_norm": 0.4463144838809967,
+      "learning_rate": 3.2327312023675287e-06,
+      "loss": 0.0177,
+      "step": 344
+    },
+    {
+      "epoch": 2.9289772727272725,
+      "grad_norm": 0.43879300355911255,
+      "learning_rate": 3.2202477484130947e-06,
+      "loss": 0.0145,
+      "step": 345
+    },
+    {
+      "epoch": 2.9375,
+      "grad_norm": 0.36773353815078735,
+      "learning_rate": 3.20774467948693e-06,
+      "loss": 0.0186,
+      "step": 346
+    },
+    {
+      "epoch": 2.9460227272727275,
+      "grad_norm": 0.45109617710113525,
+      "learning_rate": 3.19522233609318e-06,
+      "loss": 0.0153,
+      "step": 347
+    },
+    {
+      "epoch": 2.9545454545454546,
+      "grad_norm": 0.48943161964416504,
+      "learning_rate": 3.1826810592609036e-06,
+      "loss": 0.0186,
+      "step": 348
+    },
+    {
+      "epoch": 2.9630681818181817,
+      "grad_norm": 0.5407648682594299,
+      "learning_rate": 3.170121190534785e-06,
+      "loss": 0.0173,
+      "step": 349
+    },
+    {
+      "epoch": 2.971590909090909,
+      "grad_norm": 0.5848131775856018,
+      "learning_rate": 3.157543071965835e-06,
+      "loss": 0.013,
+      "step": 350
+    },
+    {
+      "epoch": 2.9801136363636362,
+      "grad_norm": 0.46287980675697327,
+      "learning_rate": 3.1449470461020738e-06,
+      "loss": 0.0194,
+      "step": 351
+    },
+    {
+      "epoch": 2.9886363636363638,
+      "grad_norm": 0.655013382434845,
+      "learning_rate": 3.132333455979202e-06,
+      "loss": 0.0176,
+      "step": 352
+    },
+    {
+      "epoch": 2.997159090909091,
+      "grad_norm": 0.7326741218566895,
+      "learning_rate": 3.11970264511126e-06,
+      "loss": 0.0154,
+      "step": 353
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.9463636875152588,
+      "learning_rate": 3.107054957481271e-06,
+      "loss": 0.0142,
+      "step": 354
+    },
+    {
+      "epoch": 3.008522727272727,
+      "grad_norm": 0.8092179894447327,
+      "learning_rate": 3.0943907375318733e-06,
+      "loss": 0.0178,
+      "step": 355
+    },
+    {
+      "epoch": 3.0170454545454546,
+      "grad_norm": 0.43350231647491455,
+      "learning_rate": 3.0817103301559422e-06,
+      "loss": 0.0163,
+      "step": 356
+    },
+    {
+      "epoch": 3.0255681818181817,
+      "grad_norm": 0.3987981379032135,
+      "learning_rate": 3.069014080687195e-06,
+      "loss": 0.0153,
+      "step": 357
+    },
+    {
+      "epoch": 3.034090909090909,
+      "grad_norm": 0.3733361065387726,
+      "learning_rate": 3.056302334890786e-06,
+      "loss": 0.012,
+      "step": 358
+    },
+    {
+      "epoch": 3.0426136363636362,
+      "grad_norm": 0.5638909339904785,
+      "learning_rate": 3.043575438953893e-06,
+      "loss": 0.0128,
+      "step": 359
+    },
+    {
+      "epoch": 3.0511363636363638,
+      "grad_norm": 0.5339016914367676,
+      "learning_rate": 3.030833739476285e-06,
+      "loss": 0.0142,
+      "step": 360
+    },
+    {
+      "epoch": 3.059659090909091,
+      "grad_norm": 0.5681723356246948,
+      "learning_rate": 3.018077583460886e-06,
+      "loss": 0.0144,
+      "step": 361
+    },
+    {
+      "epoch": 3.0681818181818183,
+      "grad_norm": 0.5017768144607544,
+      "learning_rate": 3.0053073183043257e-06,
+      "loss": 0.0137,
+      "step": 362
+    },
+    {
+      "epoch": 3.0767045454545454,
+      "grad_norm": 1.0726104974746704,
+      "learning_rate": 2.9925232917874764e-06,
+      "loss": 0.013,
+      "step": 363
+    },
+    {
+      "epoch": 3.085227272727273,
+      "grad_norm": 0.5261201858520508,
+      "learning_rate": 2.979725852065981e-06,
+      "loss": 0.0122,
+      "step": 364
+    },
+    {
+      "epoch": 3.09375,
+      "grad_norm": 0.5007719993591309,
+      "learning_rate": 2.966915347660775e-06,
+      "loss": 0.0123,
+      "step": 365
+    },
+    {
+      "epoch": 3.102272727272727,
+      "grad_norm": 0.5095813870429993,
+      "learning_rate": 2.9540921274485913e-06,
+      "loss": 0.0133,
+      "step": 366
+    },
+    {
+      "epoch": 3.1107954545454546,
+      "grad_norm": 0.4466419816017151,
+      "learning_rate": 2.941256540652462e-06,
+      "loss": 0.0094,
+      "step": 367
+    },
+    {
+      "epoch": 3.1193181818181817,
+      "grad_norm": 0.9082984328269958,
+      "learning_rate": 2.9284089368322044e-06,
+      "loss": 0.0155,
+      "step": 368
+    },
+    {
+      "epoch": 3.127840909090909,
+      "grad_norm": 0.4232461750507355,
+      "learning_rate": 2.915549665874905e-06,
+      "loss": 0.0103,
+      "step": 369
+    },
+    {
+      "epoch": 3.1363636363636362,
+      "grad_norm": 0.736997663974762,
+      "learning_rate": 2.9026790779853877e-06,
+      "loss": 0.0153,
+      "step": 370
+    },
+    {
+      "epoch": 3.1448863636363638,
+      "grad_norm": 0.5333961844444275,
+      "learning_rate": 2.8897975236766784e-06,
+      "loss": 0.0122,
+      "step": 371
+    },
+    {
+      "epoch": 3.153409090909091,
+      "grad_norm": 0.610040545463562,
+      "learning_rate": 2.876905353760459e-06,
+      "loss": 0.0121,
+      "step": 372
+    },
+    {
+      "epoch": 3.1619318181818183,
+      "grad_norm": 0.3609560430049896,
+      "learning_rate": 2.864002919337513e-06,
+      "loss": 0.0106,
+      "step": 373
+    },
+    {
+      "epoch": 3.1704545454545454,
+      "grad_norm": 0.5127677917480469,
+      "learning_rate": 2.8510905717881615e-06,
+      "loss": 0.0146,
+      "step": 374
+    },
+    {
+      "epoch": 3.178977272727273,
+      "grad_norm": 0.6335748434066772,
+      "learning_rate": 2.838168662762699e-06,
+      "loss": 0.0133,
+      "step": 375
+    },
+    {
+      "epoch": 3.1875,
+      "grad_norm": 0.5320702195167542,
+      "learning_rate": 2.8252375441718137e-06,
+      "loss": 0.0108,
+      "step": 376
+    },
+    {
+      "epoch": 3.196022727272727,
+      "grad_norm": 0.49305084347724915,
+      "learning_rate": 2.812297568177002e-06,
+      "loss": 0.0107,
+      "step": 377
+    },
+    {
+      "epoch": 3.2045454545454546,
+      "grad_norm": 0.5496407747268677,
+      "learning_rate": 2.7993490871809808e-06,
+      "loss": 0.0109,
+      "step": 378
+    },
+    {
+      "epoch": 3.2130681818181817,
+      "grad_norm": 0.4906885623931885,
+      "learning_rate": 2.7863924538180904e-06,
+      "loss": 0.0123,
+      "step": 379
+    },
+    {
+      "epoch": 3.221590909090909,
+      "grad_norm": 0.5412434339523315,
+      "learning_rate": 2.773428020944687e-06,
+      "loss": 0.0128,
+      "step": 380
+    },
+    {
+      "epoch": 3.2301136363636362,
+      "grad_norm": 0.6251363754272461,
+      "learning_rate": 2.7604561416295366e-06,
+      "loss": 0.0117,
+      "step": 381
+    },
+    {
+      "epoch": 3.2386363636363638,
+      "grad_norm": 0.517668604850769,
+      "learning_rate": 2.747477169144202e-06,
+      "loss": 0.0113,
+      "step": 382
+    },
+    {
+      "epoch": 3.247159090909091,
+      "grad_norm": 0.5211403965950012,
+      "learning_rate": 2.7344914569534154e-06,
+      "loss": 0.0103,
+      "step": 383
+    },
+    {
+      "epoch": 3.2556818181818183,
+      "grad_norm": 0.797484278678894,
+      "learning_rate": 2.721499358705458e-06,
+      "loss": 0.0111,
+      "step": 384
+    },
+    {
+      "epoch": 3.2642045454545454,
+      "grad_norm": 0.6931593418121338,
+      "learning_rate": 2.708501228222523e-06,
+      "loss": 0.0115,
+      "step": 385
+    },
+    {
+      "epoch": 3.2727272727272725,
+      "grad_norm": 0.515566349029541,
+      "learning_rate": 2.695497419491089e-06,
+      "loss": 0.0114,
+      "step": 386
+    },
+    {
+      "epoch": 3.28125,
+      "grad_norm": 0.48849251866340637,
+      "learning_rate": 2.682488286652269e-06,
+      "loss": 0.01,
+      "step": 387
+    },
+    {
+      "epoch": 3.2897727272727275,
+      "grad_norm": 0.6838280558586121,
+      "learning_rate": 2.6694741839921734e-06,
+      "loss": 0.011,
+      "step": 388
+    },
+    {
+      "epoch": 3.2982954545454546,
+      "grad_norm": 0.4109753966331482,
+      "learning_rate": 2.656455465932259e-06,
+      "loss": 0.007,
+      "step": 389
+    },
+    {
+      "epoch": 3.3068181818181817,
+      "grad_norm": 0.4408889412879944,
+      "learning_rate": 2.6434324870196746e-06,
+      "loss": 0.0096,
+      "step": 390
+    },
+    {
+      "epoch": 3.315340909090909,
+      "grad_norm": 0.4267784059047699,
+      "learning_rate": 2.63040560191761e-06,
+      "loss": 0.009,
+      "step": 391
+    },
+    {
+      "epoch": 3.3238636363636362,
+      "grad_norm": 0.42103955149650574,
+      "learning_rate": 2.617375165395634e-06,
+      "loss": 0.0084,
+      "step": 392
+    },
+    {
+      "epoch": 3.3323863636363638,
+      "grad_norm": 0.4938065707683563,
+      "learning_rate": 2.6043415323200333e-06,
+      "loss": 0.0084,
+      "step": 393
+    },
+    {
+      "epoch": 3.340909090909091,
+      "grad_norm": 0.9540116190910339,
+      "learning_rate": 2.591305057644148e-06,
+      "loss": 0.0123,
+      "step": 394
+    },
+    {
+      "epoch": 3.3494318181818183,
+      "grad_norm": 0.5479457974433899,
+      "learning_rate": 2.5782660963987054e-06,
+      "loss": 0.01,
+      "step": 395
+    },
+    {
+      "epoch": 3.3579545454545454,
+      "grad_norm": 0.44472524523735046,
+      "learning_rate": 2.5652250036821522e-06,
+      "loss": 0.0066,
+      "step": 396
+    },
+    {
+      "epoch": 3.3664772727272725,
+      "grad_norm": 1.0710281133651733,
+      "learning_rate": 2.552182134650982e-06,
+      "loss": 0.0109,
+      "step": 397
+    },
+    {
+      "epoch": 3.375,
+      "grad_norm": 0.6171219944953918,
+      "learning_rate": 2.5391378445100646e-06,
+      "loss": 0.0054,
+      "step": 398
+    },
+    {
+      "epoch": 3.3835227272727275,
+      "grad_norm": 0.5634967684745789,
+      "learning_rate": 2.526092488502971e-06,
+      "loss": 0.0091,
+      "step": 399
+    },
+    {
+      "epoch": 3.3920454545454546,
+      "grad_norm": 0.5629643797874451,
+      "learning_rate": 2.5130464219022994e-06,
+      "loss": 0.0079,
+      "step": 400
+    },
+    {
+      "epoch": 3.4005681818181817,
+      "grad_norm": 0.45441868901252747,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0084,
+      "step": 401
+    },
+    {
+      "epoch": 3.409090909090909,
+      "grad_norm": 0.7561835646629333,
+      "learning_rate": 2.4869535780977023e-06,
+      "loss": 0.0148,
+      "step": 402
+    },
+    {
+      "epoch": 3.4176136363636362,
+      "grad_norm": 0.42049774527549744,
+      "learning_rate": 2.47390751149703e-06,
+      "loss": 0.0071,
+      "step": 403
+    },
+    {
+      "epoch": 3.4261363636363638,
+      "grad_norm": 0.4444538652896881,
+      "learning_rate": 2.460862155489936e-06,
+      "loss": 0.0075,
+      "step": 404
+    },
+    {
+      "epoch": 3.434659090909091,
+      "grad_norm": 0.6667734980583191,
+      "learning_rate": 2.447817865349018e-06,
+      "loss": 0.0098,
+      "step": 405
+    },
+    {
+      "epoch": 3.4431818181818183,
+      "grad_norm": 0.6603801250457764,
+      "learning_rate": 2.4347749963178486e-06,
+      "loss": 0.0079,
+      "step": 406
+    },
+    {
+      "epoch": 3.4517045454545454,
+      "grad_norm": 0.5178464651107788,
+      "learning_rate": 2.421733903601296e-06,
+      "loss": 0.0065,
+      "step": 407
+    },
+    {
+      "epoch": 3.4602272727272725,
+      "grad_norm": 0.47047677636146545,
+      "learning_rate": 2.408694942355853e-06,
+      "loss": 0.0084,
+      "step": 408
+    },
+    {
+      "epoch": 3.46875,
+      "grad_norm": 0.4323576092720032,
+      "learning_rate": 2.3956584676799676e-06,
+      "loss": 0.0068,
+      "step": 409
+    },
+    {
+      "epoch": 3.4772727272727275,
+      "grad_norm": 0.45331141352653503,
+      "learning_rate": 2.3826248346043664e-06,
+      "loss": 0.008,
+      "step": 410
+    },
+    {
+      "epoch": 3.4857954545454546,
+      "grad_norm": 0.4945763051509857,
+      "learning_rate": 2.369594398082391e-06,
+      "loss": 0.0088,
+      "step": 411
+    },
+    {
+      "epoch": 3.4943181818181817,
+      "grad_norm": 0.4799608588218689,
+      "learning_rate": 2.356567512980326e-06,
+      "loss": 0.0046,
+      "step": 412
+    },
+    {
+      "epoch": 3.502840909090909,
+      "grad_norm": 0.5917950868606567,
+      "learning_rate": 2.343544534067742e-06,
+      "loss": 0.0064,
+      "step": 413
+    },
+    {
+      "epoch": 3.5113636363636362,
+      "grad_norm": 0.5370076894760132,
+      "learning_rate": 2.3305258160078274e-06,
+      "loss": 0.0061,
+      "step": 414
+    },
+    {
+      "epoch": 3.5198863636363638,
+      "grad_norm": 0.5539790391921997,
+      "learning_rate": 2.317511713347731e-06,
+      "loss": 0.0057,
+      "step": 415
+    },
+    {
+      "epoch": 3.528409090909091,
+      "grad_norm": 0.5244786739349365,
+      "learning_rate": 2.304502580508912e-06,
+      "loss": 0.0077,
+      "step": 416
+    },
+    {
+      "epoch": 3.5369318181818183,
+      "grad_norm": 0.5492494106292725,
+      "learning_rate": 2.291498771777478e-06,
+      "loss": 0.008,
+      "step": 417
+    },
+    {
+      "epoch": 3.5454545454545454,
+      "grad_norm": 0.5254555940628052,
+      "learning_rate": 2.278500641294543e-06,
+      "loss": 0.007,
+      "step": 418
+    },
+    {
+      "epoch": 3.5539772727272725,
+      "grad_norm": 0.7219369411468506,
+      "learning_rate": 2.2655085430465855e-06,
+      "loss": 0.0055,
+      "step": 419
+    },
+    {
+      "epoch": 3.5625,
+      "grad_norm": 0.8314915895462036,
+      "learning_rate": 2.252522830855798e-06,
+      "loss": 0.0068,
+      "step": 420
+    },
+    {
+      "epoch": 3.5710227272727275,
+      "grad_norm": 0.4091542363166809,
+      "learning_rate": 2.239543858370464e-06,
+      "loss": 0.004,
+      "step": 421
+    },
+    {
+      "epoch": 3.5795454545454546,
+      "grad_norm": 0.6792344450950623,
+      "learning_rate": 2.2265719790553147e-06,
+      "loss": 0.0059,
+      "step": 422
+    },
+    {
+      "epoch": 3.5880681818181817,
+      "grad_norm": 0.4202214777469635,
+      "learning_rate": 2.2136075461819104e-06,
+      "loss": 0.0077,
+      "step": 423
+    },
+    {
+      "epoch": 3.596590909090909,
+      "grad_norm": 0.6968201398849487,
+      "learning_rate": 2.2006509128190196e-06,
+      "loss": 0.0086,
+      "step": 424
+    },
+    {
+      "epoch": 3.6051136363636362,
+      "grad_norm": 0.6859596371650696,
+      "learning_rate": 2.1877024318229987e-06,
+      "loss": 0.0055,
+      "step": 425
+    },
+    {
+      "epoch": 3.6136363636363638,
+      "grad_norm": 0.5254549980163574,
+      "learning_rate": 2.1747624558281867e-06,
+      "loss": 0.0076,
+      "step": 426
+    },
+    {
+      "epoch": 3.622159090909091,
+      "grad_norm": 0.38539767265319824,
+      "learning_rate": 2.1618313372373016e-06,
+      "loss": 0.0058,
+      "step": 427
+    },
+    {
+      "epoch": 3.6306818181818183,
+      "grad_norm": 0.5018851161003113,
+      "learning_rate": 2.1489094282118393e-06,
+      "loss": 0.0063,
+      "step": 428
+    },
+    {
+      "epoch": 3.6392045454545454,
+      "grad_norm": 0.2640945017337799,
+      "learning_rate": 2.1359970806624886e-06,
+      "loss": 0.0033,
+      "step": 429
+    },
+    {
+      "epoch": 3.6477272727272725,
+      "grad_norm": 0.5288704037666321,
+      "learning_rate": 2.1230946462395412e-06,
+      "loss": 0.0081,
+      "step": 430
+    },
+    {
+      "epoch": 3.65625,
+      "grad_norm": 0.5245815515518188,
+      "learning_rate": 2.1102024763233224e-06,
+      "loss": 0.0053,
+      "step": 431
+    },
+    {
+      "epoch": 3.6647727272727275,
+      "grad_norm": 0.4336767792701721,
+      "learning_rate": 2.0973209220146135e-06,
+      "loss": 0.0057,
+      "step": 432
+    },
+    {
+      "epoch": 3.6732954545454546,
+      "grad_norm": 0.6342284679412842,
+      "learning_rate": 2.084450334125096e-06,
+      "loss": 0.0066,
+      "step": 433
+    },
+    {
+      "epoch": 3.6818181818181817,
+      "grad_norm": 0.44606292247772217,
+      "learning_rate": 2.071591063167797e-06,
+      "loss": 0.0067,
+      "step": 434
+    },
+    {
+      "epoch": 3.690340909090909,
+      "grad_norm": 0.519180953502655,
+      "learning_rate": 2.0587434593475385e-06,
+      "loss": 0.0046,
+      "step": 435
+    },
+    {
+      "epoch": 3.6988636363636362,
+      "grad_norm": 0.44359397888183594,
+      "learning_rate": 2.045907872551409e-06,
+      "loss": 0.0052,
+      "step": 436
+    },
+    {
+      "epoch": 3.7073863636363638,
+      "grad_norm": 0.43811115622520447,
+      "learning_rate": 2.033084652339226e-06,
+      "loss": 0.0048,
+      "step": 437
+    },
+    {
+      "epoch": 3.715909090909091,
+      "grad_norm": 0.5960409045219421,
+      "learning_rate": 2.0202741479340193e-06,
+      "loss": 0.0063,
+      "step": 438
+    },
+    {
+      "epoch": 3.7244318181818183,
+      "grad_norm": 0.4575386345386505,
+      "learning_rate": 2.0074767082125244e-06,
+      "loss": 0.0052,
+      "step": 439
+    },
+    {
+      "epoch": 3.7329545454545454,
+      "grad_norm": 0.6704768538475037,
+      "learning_rate": 1.9946926816956743e-06,
+      "loss": 0.0075,
+      "step": 440
+    },
+    {
+      "epoch": 3.7414772727272725,
+      "grad_norm": 0.44066664576530457,
+      "learning_rate": 1.9819224165391145e-06,
+      "loss": 0.004,
+      "step": 441
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.5848504304885864,
+      "learning_rate": 1.969166260523717e-06,
+      "loss": 0.005,
+      "step": 442
+    },
+    {
+      "epoch": 3.7585227272727275,
+      "grad_norm": 0.43681785464286804,
+      "learning_rate": 1.956424561046108e-06,
+      "loss": 0.0041,
+      "step": 443
+    },
+    {
+      "epoch": 3.7670454545454546,
+      "grad_norm": 0.31628116965293884,
+      "learning_rate": 1.9436976651092143e-06,
+      "loss": 0.0026,
+      "step": 444
+    },
+    {
+      "epoch": 3.7755681818181817,
+      "grad_norm": 0.4530532658100128,
+      "learning_rate": 1.9309859193128055e-06,
+      "loss": 0.0052,
+      "step": 445
+    },
+    {
+      "epoch": 3.784090909090909,
+      "grad_norm": 0.258119136095047,
+      "learning_rate": 1.918289669844058e-06,
+      "loss": 0.0025,
+      "step": 446
+    },
+    {
+      "epoch": 3.7926136363636362,
+      "grad_norm": 0.40686145424842834,
+      "learning_rate": 1.905609262468128e-06,
+      "loss": 0.005,
+      "step": 447
+    },
+    {
+      "epoch": 3.8011363636363638,
+      "grad_norm": 0.4893140196800232,
+      "learning_rate": 1.8929450425187298e-06,
+      "loss": 0.0033,
+      "step": 448
+    },
+    {
+      "epoch": 3.809659090909091,
+      "grad_norm": 0.8415848612785339,
+      "learning_rate": 1.880297354888741e-06,
+      "loss": 0.0044,
+      "step": 449
+    },
+    {
+      "epoch": 3.8181818181818183,
+      "grad_norm": 0.21848903596401215,
+      "learning_rate": 1.8676665440207982e-06,
+      "loss": 0.0015,
+      "step": 450
+    },
+    {
+      "epoch": 3.8267045454545454,
+      "grad_norm": 0.7620615363121033,
+      "learning_rate": 1.8550529538979273e-06,
+      "loss": 0.0069,
+      "step": 451
+    },
+    {
+      "epoch": 3.8352272727272725,
+      "grad_norm": 0.35021477937698364,
+      "learning_rate": 1.8424569280341653e-06,
+      "loss": 0.004,
+      "step": 452
+    },
+    {
+      "epoch": 3.84375,
+      "grad_norm": 0.477241575717926,
+      "learning_rate": 1.8298788094652158e-06,
+      "loss": 0.0052,
+      "step": 453
+    },
+    {
+      "epoch": 3.8522727272727275,
+      "grad_norm": 0.3984035849571228,
+      "learning_rate": 1.817318940739098e-06,
+      "loss": 0.0029,
+      "step": 454
+    },
+    {
+      "epoch": 3.8607954545454546,
+      "grad_norm": 0.3424110412597656,
+      "learning_rate": 1.8047776639068204e-06,
+      "loss": 0.0044,
+      "step": 455
+    },
+    {
+      "epoch": 3.8693181818181817,
+      "grad_norm": 1.0182089805603027,
+      "learning_rate": 1.7922553205130708e-06,
+      "loss": 0.0027,
+      "step": 456
+    },
+    {
+      "epoch": 3.877840909090909,
+      "grad_norm": 0.4957254230976105,
+      "learning_rate": 1.7797522515869062e-06,
+      "loss": 0.0028,
+      "step": 457
+    },
+    {
+      "epoch": 3.8863636363636362,
+      "grad_norm": 0.42411452531814575,
+      "learning_rate": 1.767268797632472e-06,
+      "loss": 0.0032,
+      "step": 458
+    },
+    {
+      "epoch": 3.8948863636363638,
+      "grad_norm": 0.3485550582408905,
+      "learning_rate": 1.7548052986197258e-06,
+      "loss": 0.0032,
+      "step": 459
+    },
+    {
+      "epoch": 3.903409090909091,
+      "grad_norm": 0.3125132918357849,
+      "learning_rate": 1.7423620939751787e-06,
+      "loss": 0.0017,
+      "step": 460
+    },
+    {
+      "epoch": 3.9119318181818183,
+      "grad_norm": 0.7568768262863159,
+      "learning_rate": 1.7299395225726533e-06,
+      "loss": 0.006,
+      "step": 461
+    },
+    {
+      "epoch": 3.9204545454545454,
+      "grad_norm": 0.7975472807884216,
+      "learning_rate": 1.7175379227240524e-06,
+      "loss": 0.0044,
+      "step": 462
+    },
+    {
+      "epoch": 3.9289772727272725,
+      "grad_norm": 0.5542371273040771,
+      "learning_rate": 1.7051576321701493e-06,
+      "loss": 0.0034,
+      "step": 463
+    },
+    {
+      "epoch": 3.9375,
+      "grad_norm": 0.3784295916557312,
+      "learning_rate": 1.6927989880713852e-06,
+      "loss": 0.0024,
+      "step": 464
+    },
+    {
+      "epoch": 3.9460227272727275,
+      "grad_norm": 0.7081852555274963,
+      "learning_rate": 1.680462326998689e-06,
+      "loss": 0.0047,
+      "step": 465
+    },
+    {
+      "epoch": 3.9545454545454546,
+      "grad_norm": 0.268210232257843,
+      "learning_rate": 1.6681479849243153e-06,
+      "loss": 0.0019,
+      "step": 466
+    },
+    {
+      "epoch": 3.9630681818181817,
+      "grad_norm": 0.8834506273269653,
+      "learning_rate": 1.6558562972126869e-06,
+      "loss": 0.0025,
+      "step": 467
+    },
+    {
+      "epoch": 3.971590909090909,
+      "grad_norm": 0.8946535587310791,
+      "learning_rate": 1.6435875986112685e-06,
+      "loss": 0.0035,
+      "step": 468
+    },
+    {
+      "epoch": 3.9801136363636362,
+      "grad_norm": 0.37636420130729675,
+      "learning_rate": 1.631342223241446e-06,
+      "loss": 0.0035,
+      "step": 469
+    },
+    {
+      "epoch": 3.9886363636363638,
+      "grad_norm": 0.25573107600212097,
+      "learning_rate": 1.6191205045894283e-06,
+      "loss": 0.0021,
+      "step": 470
+    },
+    {
+      "epoch": 3.997159090909091,
+      "grad_norm": 0.6776405572891235,
+      "learning_rate": 1.6069227754971683e-06,
+      "loss": 0.0048,
+      "step": 471
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.6776405572891235,
+      "learning_rate": 1.594749368153292e-06,
+      "loss": 0.0021,
+      "step": 472
+    },
+    {
+      "epoch": 4.0085227272727275,
+      "grad_norm": 0.7230031490325928,
+      "learning_rate": 1.5826006140840593e-06,
+      "loss": 0.0017,
+      "step": 473
+    },
+    {
+      "epoch": 4.017045454545454,
+      "grad_norm": 0.28904300928115845,
+      "learning_rate": 1.570476844144329e-06,
+      "loss": 0.0012,
+      "step": 474
+    },
+    {
+      "epoch": 4.025568181818182,
+      "grad_norm": 0.2574220895767212,
+      "learning_rate": 1.5583783885085513e-06,
+      "loss": 0.0015,
+      "step": 475
+    },
+    {
+      "epoch": 4.034090909090909,
+      "grad_norm": 0.27897921204566956,
+      "learning_rate": 1.5463055766617763e-06,
+      "loss": 0.002,
+      "step": 476
+    },
+    {
+      "epoch": 4.042613636363637,
+      "grad_norm": 0.32508671283721924,
+      "learning_rate": 1.5342587373906798e-06,
+      "loss": 0.0009,
+      "step": 477
+    },
+    {
+      "epoch": 4.051136363636363,
+      "grad_norm": 0.30587589740753174,
+      "learning_rate": 1.5222381987746104e-06,
+      "loss": 0.0019,
+      "step": 478
+    },
+    {
+      "epoch": 4.059659090909091,
+      "grad_norm": 0.5016082525253296,
+      "learning_rate": 1.5102442881766533e-06,
+      "loss": 0.0012,
+      "step": 479
+    },
+    {
+      "epoch": 4.068181818181818,
+      "grad_norm": 0.28716009855270386,
+      "learning_rate": 1.4982773322347144e-06,
+      "loss": 0.0015,
+      "step": 480
+    },
+    {
+      "epoch": 4.076704545454546,
+      "grad_norm": 0.1832437366247177,
+      "learning_rate": 1.48633765685263e-06,
+      "loss": 0.0023,
+      "step": 481
+    },
+    {
+      "epoch": 4.0852272727272725,
+      "grad_norm": 0.35543736815452576,
+      "learning_rate": 1.4744255871912825e-06,
+      "loss": 0.0024,
+      "step": 482
+    },
+    {
+      "epoch": 4.09375,
+      "grad_norm": 0.38490474224090576,
+      "learning_rate": 1.4625414476597545e-06,
+      "loss": 0.003,
+      "step": 483
+    },
+    {
+      "epoch": 4.1022727272727275,
+      "grad_norm": 0.17410258948802948,
+      "learning_rate": 1.4506855619064847e-06,
+      "loss": 0.0009,
+      "step": 484
+    },
+    {
+      "epoch": 4.110795454545454,
+      "grad_norm": 0.30948376655578613,
+      "learning_rate": 1.4388582528104628e-06,
+      "loss": 0.0012,
+      "step": 485
+    },
+    {
+      "epoch": 4.119318181818182,
+      "grad_norm": 0.41439351439476013,
+      "learning_rate": 1.4270598424724291e-06,
+      "loss": 0.0017,
+      "step": 486
+    },
+    {
+      "epoch": 4.127840909090909,
+      "grad_norm": 1.3294129371643066,
+      "learning_rate": 1.415290652206105e-06,
+      "loss": 0.0019,
+      "step": 487
+    },
+    {
+      "epoch": 4.136363636363637,
+      "grad_norm": 0.2371089905500412,
+      "learning_rate": 1.4035510025294463e-06,
+      "loss": 0.0023,
+      "step": 488
+    },
+    {
+      "epoch": 4.144886363636363,
+      "grad_norm": 0.9173463582992554,
+      "learning_rate": 1.3918412131559073e-06,
+      "loss": 0.0019,
+      "step": 489
+    },
+    {
+      "epoch": 4.153409090909091,
+      "grad_norm": 0.9328559041023254,
+      "learning_rate": 1.380161602985738e-06,
+      "loss": 0.002,
+      "step": 490
+    },
+    {
+      "epoch": 4.161931818181818,
+      "grad_norm": 0.2602047026157379,
+      "learning_rate": 1.3685124900973013e-06,
+      "loss": 0.0013,
+      "step": 491
+    },
+    {
+      "epoch": 4.170454545454546,
+      "grad_norm": 0.19252748787403107,
+      "learning_rate": 1.3568941917384038e-06,
+      "loss": 0.0008,
+      "step": 492
+    },
+    {
+      "epoch": 4.1789772727272725,
+      "grad_norm": 0.259917289018631,
+      "learning_rate": 1.3453070243176646e-06,
+      "loss": 0.002,
+      "step": 493
+    },
+    {
+      "epoch": 4.1875,
+      "grad_norm": 0.4336180090904236,
+      "learning_rate": 1.3337513033958904e-06,
+      "loss": 0.0012,
+      "step": 494
+    },
+    {
+      "epoch": 4.1960227272727275,
+      "grad_norm": 0.3246527910232544,
+      "learning_rate": 1.3222273436774857e-06,
+      "loss": 0.0009,
+      "step": 495
+    },
+    {
+      "epoch": 4.204545454545454,
+      "grad_norm": 0.6794546246528625,
+      "learning_rate": 1.310735459001884e-06,
+      "loss": 0.001,
+      "step": 496
+    },
+    {
+      "epoch": 4.213068181818182,
+      "grad_norm": 0.3761320114135742,
+      "learning_rate": 1.2992759623349946e-06,
+      "loss": 0.0015,
+      "step": 497
+    },
+    {
+      "epoch": 4.221590909090909,
+      "grad_norm": 0.37112557888031006,
+      "learning_rate": 1.2878491657606874e-06,
+      "loss": 0.002,
+      "step": 498
+    },
+    {
+      "epoch": 4.230113636363637,
+      "grad_norm": 0.29353010654449463,
+      "learning_rate": 1.2764553804722868e-06,
+      "loss": 0.0005,
+      "step": 499
+    },
+    {
+      "epoch": 4.238636363636363,
+      "grad_norm": 0.16569066047668457,
+      "learning_rate": 1.2650949167640997e-06,
+      "loss": 0.0006,
+      "step": 500
+    },
+    {
+      "epoch": 4.247159090909091,
+      "grad_norm": 0.2684345543384552,
+      "learning_rate": 1.2537680840229665e-06,
+      "loss": 0.0018,
+      "step": 501
+    },
+    {
+      "epoch": 4.255681818181818,
+      "grad_norm": 1.2519632577896118,
+      "learning_rate": 1.2424751907198312e-06,
+      "loss": 0.001,
+      "step": 502
+    },
+    {
+      "epoch": 4.264204545454546,
+      "grad_norm": 0.4186049997806549,
+      "learning_rate": 1.2312165444013473e-06,
+      "loss": 0.0013,
+      "step": 503
+    },
+    {
+      "epoch": 4.2727272727272725,
+      "grad_norm": 0.32475754618644714,
+      "learning_rate": 1.219992451681494e-06,
+      "loss": 0.0009,
+      "step": 504
+    },
+    {
+      "epoch": 4.28125,
+      "grad_norm": 1.013559341430664,
+      "learning_rate": 1.2088032182332307e-06,
+      "loss": 0.0019,
+      "step": 505
+    },
+    {
+      "epoch": 4.2897727272727275,
+      "grad_norm": 0.459043949842453,
+      "learning_rate": 1.1976491487801747e-06,
+      "loss": 0.001,
+      "step": 506
+    },
+    {
+      "epoch": 4.298295454545454,
+      "grad_norm": 0.5227071642875671,
+      "learning_rate": 1.1865305470882955e-06,
+      "loss": 0.0014,
+      "step": 507
+    },
+    {
+      "epoch": 4.306818181818182,
+      "grad_norm": 0.23663417994976044,
+      "learning_rate": 1.17544771595765e-06,
+      "loss": 0.001,
+      "step": 508
+    },
+    {
+      "epoch": 4.315340909090909,
+      "grad_norm": 0.2578096091747284,
+      "learning_rate": 1.1644009572141293e-06,
+      "loss": 0.0012,
+      "step": 509
+    },
+    {
+      "epoch": 4.323863636363637,
+      "grad_norm": 0.9903965592384338,
+      "learning_rate": 1.1533905717012425e-06,
+      "loss": 0.002,
+      "step": 510
+    },
+    {
+      "epoch": 4.332386363636363,
+      "grad_norm": 0.34067901968955994,
+      "learning_rate": 1.1424168592719257e-06,
+      "loss": 0.0012,
+      "step": 511
+    },
+    {
+      "epoch": 4.340909090909091,
+      "grad_norm": 0.11968725919723511,
+      "learning_rate": 1.1314801187803687e-06,
+      "loss": 0.0003,
+      "step": 512
+    },
+    {
+      "epoch": 4.349431818181818,
+      "grad_norm": 0.12175023555755615,
+      "learning_rate": 1.1205806480738852e-06,
+      "loss": 0.0008,
+      "step": 513
+    },
+    {
+      "epoch": 4.357954545454546,
+      "grad_norm": 0.18433789908885956,
+      "learning_rate": 1.109718743984794e-06,
+      "loss": 0.0014,
+      "step": 514
+    },
+    {
+      "epoch": 4.3664772727272725,
+      "grad_norm": 0.17771068215370178,
+      "learning_rate": 1.0988947023223374e-06,
+      "loss": 0.0008,
+      "step": 515
+    },
+    {
+      "epoch": 4.375,
+      "grad_norm": 0.26217710971832275,
+      "learning_rate": 1.0881088178646291e-06,
+      "loss": 0.0011,
+      "step": 516
+    },
+    {
+      "epoch": 4.3835227272727275,
+      "grad_norm": 0.22049932181835175,
+      "learning_rate": 1.0773613843506201e-06,
+      "loss": 0.0007,
+      "step": 517
+    },
+    {
+      "epoch": 4.392045454545454,
+      "grad_norm": 0.3353421688079834,
+      "learning_rate": 1.0666526944721017e-06,
+      "loss": 0.0012,
+      "step": 518
+    },
+    {
+      "epoch": 4.400568181818182,
+      "grad_norm": 0.5478553771972656,
+      "learning_rate": 1.0559830398657387e-06,
+      "loss": 0.002,
+      "step": 519
+    },
+    {
+      "epoch": 4.409090909090909,
+      "grad_norm": 0.1836828738451004,
+      "learning_rate": 1.0453527111051183e-06,
+      "loss": 0.0018,
+      "step": 520
+    },
+    {
+      "epoch": 4.417613636363637,
+      "grad_norm": 2.7253522872924805,
+      "learning_rate": 1.0347619976928447e-06,
+      "loss": 0.0004,
+      "step": 521
+    },
+    {
+      "epoch": 4.426136363636363,
+      "grad_norm": 0.1142822802066803,
+      "learning_rate": 1.0242111880526495e-06,
+      "loss": 0.0007,
+      "step": 522
+    },
+    {
+      "epoch": 4.434659090909091,
+      "grad_norm": 0.27105048298835754,
+      "learning_rate": 1.013700569521542e-06,
+      "loss": 0.0008,
+      "step": 523
+    },
+    {
+      "epoch": 4.443181818181818,
+      "grad_norm": 0.3742014467716217,
+      "learning_rate": 1.0032304283419792e-06,
+      "loss": 0.0007,
+      "step": 524
+    },
+    {
+      "epoch": 4.451704545454546,
+      "grad_norm": 0.2388819307088852,
+      "learning_rate": 9.92801049654071e-07,
+      "loss": 0.002,
+      "step": 525
+    },
+    {
+      "epoch": 4.4602272727272725,
+      "grad_norm": 0.2067682445049286,
+      "learning_rate": 9.824127174878196e-07,
+      "loss": 0.0009,
+      "step": 526
+    },
+    {
+      "epoch": 4.46875,
+      "grad_norm": 0.2894744277000427,
+      "learning_rate": 9.72065714755377e-07,
+      "loss": 0.0009,
+      "step": 527
+    },
+    {
+      "epoch": 4.4772727272727275,
+      "grad_norm": 0.07446195930242538,
+      "learning_rate": 9.617603232433475e-07,
+      "loss": 0.0006,
+      "step": 528
+    },
+    {
+      "epoch": 4.485795454545454,
+      "grad_norm": 1.2312819957733154,
+      "learning_rate": 9.514968236051081e-07,
+      "loss": 0.0011,
+      "step": 529
+    },
+    {
+      "epoch": 4.494318181818182,
+      "grad_norm": 0.15863823890686035,
+      "learning_rate": 9.412754953531664e-07,
+      "loss": 0.0009,
+      "step": 530
+    },
+    {
+      "epoch": 4.502840909090909,
+      "grad_norm": 0.6933857798576355,
+      "learning_rate": 9.310966168515528e-07,
+      "loss": 0.0007,
+      "step": 531
+    },
+    {
+      "epoch": 4.511363636363637,
+      "grad_norm": 0.09022935479879379,
+      "learning_rate": 9.209604653082326e-07,
+      "loss": 0.0007,
+      "step": 532
+    },
+    {
+      "epoch": 4.519886363636363,
+      "grad_norm": 0.07025769352912903,
+      "learning_rate": 9.108673167675635e-07,
+      "loss": 0.0003,
+      "step": 533
+    },
+    {
+      "epoch": 4.528409090909091,
+      "grad_norm": 0.10223375260829926,
+      "learning_rate": 9.008174461027724e-07,
+      "loss": 0.0005,
+      "step": 534
+    },
+    {
+      "epoch": 4.536931818181818,
+      "grad_norm": 0.15842217206954956,
+      "learning_rate": 8.908111270084718e-07,
+      "loss": 0.0003,
+      "step": 535
+    },
+    {
+      "epoch": 4.545454545454545,
+      "grad_norm": 0.16270683705806732,
+      "learning_rate": 8.808486319932083e-07,
+      "loss": 0.0008,
+      "step": 536
+    },
+    {
+      "epoch": 4.5539772727272725,
+      "grad_norm": 0.07918278872966766,
+      "learning_rate": 8.709302323720367e-07,
+      "loss": 0.0003,
+      "step": 537
+    },
+    {
+      "epoch": 4.5625,
+      "grad_norm": 0.09812464565038681,
+      "learning_rate": 8.610561982591356e-07,
+      "loss": 0.0005,
+      "step": 538
+    },
+    {
+      "epoch": 4.5710227272727275,
+      "grad_norm": 0.25634539127349854,
+      "learning_rate": 8.512267985604475e-07,
+      "loss": 0.0009,
+      "step": 539
+    },
+    {
+      "epoch": 4.579545454545455,
+      "grad_norm": 0.13367760181427002,
+      "learning_rate": 8.414423009663564e-07,
+      "loss": 0.0005,
+      "step": 540
+    },
+    {
+      "epoch": 4.588068181818182,
+      "grad_norm": 0.09154052287340164,
+      "learning_rate": 8.317029719444017e-07,
+      "loss": 0.0006,
+      "step": 541
+    },
+    {
+      "epoch": 4.596590909090909,
+      "grad_norm": 0.18064351379871368,
+      "learning_rate": 8.220090767320138e-07,
+      "loss": 0.0005,
+      "step": 542
+    },
+    {
+      "epoch": 4.605113636363637,
+      "grad_norm": 0.10706845670938492,
+      "learning_rate": 8.123608793292987e-07,
+      "loss": 0.0008,
+      "step": 543
+    },
+    {
+      "epoch": 4.613636363636363,
+      "grad_norm": 0.1352279931306839,
+      "learning_rate": 8.027586424918413e-07,
+      "loss": 0.0005,
+      "step": 544
+    },
+    {
+      "epoch": 4.622159090909091,
+      "grad_norm": 0.23175151646137238,
+      "learning_rate": 7.932026277235533e-07,
+      "loss": 0.0008,
+      "step": 545
+    },
+    {
+      "epoch": 4.630681818181818,
+      "grad_norm": 0.13066157698631287,
+      "learning_rate": 7.836930952695535e-07,
+      "loss": 0.0008,
+      "step": 546
+    },
+    {
+      "epoch": 4.639204545454545,
+      "grad_norm": 0.1292756050825119,
+      "learning_rate": 7.74230304109074e-07,
+      "loss": 0.0005,
+      "step": 547
+    },
+    {
+      "epoch": 4.6477272727272725,
+      "grad_norm": 0.07149559259414673,
+      "learning_rate": 7.648145119484152e-07,
+      "loss": 0.0003,
+      "step": 548
+    },
+    {
+      "epoch": 4.65625,
+      "grad_norm": 0.1683271825313568,
+      "learning_rate": 7.554459752139204e-07,
+      "loss": 0.0009,
+      "step": 549
+    },
+    {
+      "epoch": 4.6647727272727275,
+      "grad_norm": 0.10963233560323715,
+      "learning_rate": 7.461249490449954e-07,
+      "loss": 0.0007,
+      "step": 550
+    },
+    {
+      "epoch": 4.673295454545455,
+      "grad_norm": 0.1800011694431305,
+      "learning_rate": 7.368516872871622e-07,
+      "loss": 0.0008,
+      "step": 551
+    },
+    {
+      "epoch": 4.681818181818182,
+      "grad_norm": 0.12762947380542755,
+      "learning_rate": 7.276264424851426e-07,
+      "loss": 0.0004,
+      "step": 552
+    },
+    {
+      "epoch": 4.690340909090909,
+      "grad_norm": 0.08830232918262482,
+      "learning_rate": 7.184494658759808e-07,
+      "loss": 0.0004,
+      "step": 553
+    },
+    {
+      "epoch": 4.698863636363637,
+      "grad_norm": 0.149992436170578,
+      "learning_rate": 7.093210073822027e-07,
+      "loss": 0.001,
+      "step": 554
+    },
+    {
+      "epoch": 4.707386363636363,
+      "grad_norm": 0.09241579473018646,
+      "learning_rate": 7.002413156050109e-07,
+      "loss": 0.0007,
+      "step": 555
+    },
+    {
+      "epoch": 4.715909090909091,
+      "grad_norm": 0.11096753925085068,
+      "learning_rate": 6.912106378175098e-07,
+      "loss": 0.0007,
+      "step": 556
+    },
+    {
+      "epoch": 4.724431818181818,
+      "grad_norm": 0.10129871964454651,
+      "learning_rate": 6.82229219957975e-07,
+      "loss": 0.0004,
+      "step": 557
+    },
+    {
+      "epoch": 4.732954545454545,
+      "grad_norm": 0.06802051514387131,
+      "learning_rate": 6.732973066231563e-07,
+      "loss": 0.0004,
+      "step": 558
+    },
+    {
+      "epoch": 4.7414772727272725,
+      "grad_norm": 0.12237358093261719,
+      "learning_rate": 6.644151410616123e-07,
+      "loss": 0.0007,
+      "step": 559
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.15545684099197388,
+      "learning_rate": 6.555829651670912e-07,
+      "loss": 0.0013,
+      "step": 560
+    },
+    {
+      "epoch": 4.7585227272727275,
+      "grad_norm": 0.1454661637544632,
+      "learning_rate": 6.468010194719376e-07,
+      "loss": 0.0005,
+      "step": 561
+    },
+    {
+      "epoch": 4.767045454545455,
+      "grad_norm": 0.25689929723739624,
+      "learning_rate": 6.380695431405453e-07,
+      "loss": 0.0011,
+      "step": 562
+    },
+    {
+      "epoch": 4.775568181818182,
+      "grad_norm": 0.15160903334617615,
+      "learning_rate": 6.29388773962846e-07,
+      "loss": 0.0004,
+      "step": 563
+    },
+    {
+      "epoch": 4.784090909090909,
+      "grad_norm": 0.1834772229194641,
+      "learning_rate": 6.207589483478266e-07,
+      "loss": 0.0003,
+      "step": 564
+    },
+    {
+      "epoch": 4.792613636363637,
+      "grad_norm": 0.17777854204177856,
+      "learning_rate": 6.121803013170988e-07,
+      "loss": 0.0013,
+      "step": 565
+    },
+    {
+      "epoch": 4.801136363636363,
+      "grad_norm": 0.17234425246715546,
+      "learning_rate": 6.036530664984922e-07,
+      "loss": 0.0006,
+      "step": 566
+    },
+    {
+      "epoch": 4.809659090909091,
+      "grad_norm": 0.16570787131786346,
+      "learning_rate": 5.951774761196955e-07,
+      "loss": 0.0008,
+      "step": 567
+    },
+    {
+      "epoch": 4.818181818181818,
+      "grad_norm": 0.20983879268169403,
+      "learning_rate": 5.867537610019317e-07,
+      "loss": 0.0007,
+      "step": 568
+    },
+    {
+      "epoch": 4.826704545454545,
+      "grad_norm": 0.13165390491485596,
+      "learning_rate": 5.783821505536696e-07,
+      "loss": 0.0008,
+      "step": 569
+    },
+    {
+      "epoch": 4.8352272727272725,
+      "grad_norm": 0.06456772238016129,
+      "learning_rate": 5.700628727643806e-07,
+      "loss": 0.0004,
+      "step": 570
+    },
+    {
+      "epoch": 4.84375,
+      "grad_norm": 0.14952756464481354,
+      "learning_rate": 5.617961541983244e-07,
+      "loss": 0.0007,
+      "step": 571
+    },
+    {
+      "epoch": 4.8522727272727275,
+      "grad_norm": 0.1199275404214859,
+      "learning_rate": 5.53582219988382e-07,
+      "loss": 0.0005,
+      "step": 572
+    },
+    {
+      "epoch": 4.860795454545455,
+      "grad_norm": 0.18303653597831726,
+      "learning_rate": 5.454212938299256e-07,
+      "loss": 0.0003,
+      "step": 573
+    },
+    {
+      "epoch": 4.869318181818182,
+      "grad_norm": 0.1395275741815567,
+      "learning_rate": 5.373135979747226e-07,
+      "loss": 0.0004,
+      "step": 574
+    },
+    {
+      "epoch": 4.877840909090909,
+      "grad_norm": 0.18017546832561493,
+      "learning_rate": 5.292593532248877e-07,
+      "loss": 0.0004,
+      "step": 575
+    },
+    {
+      "epoch": 4.886363636363637,
+      "grad_norm": 0.08687930554151535,
+      "learning_rate": 5.21258778926865e-07,
+      "loss": 0.0003,
+      "step": 576
+    },
+    {
+      "epoch": 4.894886363636363,
+      "grad_norm": 0.206526979804039,
+      "learning_rate": 5.133120929654567e-07,
+      "loss": 0.001,
+      "step": 577
+    },
+    {
+      "epoch": 4.903409090909091,
+      "grad_norm": 0.09521065652370453,
+      "learning_rate": 5.054195117578914e-07,
+      "loss": 0.0008,
+      "step": 578
+    },
+    {
+      "epoch": 4.911931818181818,
+      "grad_norm": 0.08821254968643188,
+      "learning_rate": 4.975812502479249e-07,
+      "loss": 0.0007,
+      "step": 579
+    },
+    {
+      "epoch": 4.920454545454545,
+      "grad_norm": 0.10224071145057678,
+      "learning_rate": 4.897975218999926e-07,
+      "loss": 0.0011,
+      "step": 580
+    },
+    {
+      "epoch": 4.9289772727272725,
+      "grad_norm": 0.22748161852359772,
+      "learning_rate": 4.82068538693391e-07,
+      "loss": 0.0006,
+      "step": 581
+    },
+    {
+      "epoch": 4.9375,
+      "grad_norm": 0.07090436667203903,
+      "learning_rate": 4.7439451111650685e-07,
+      "loss": 0.0003,
+      "step": 582
+    },
+    {
+      "epoch": 4.9460227272727275,
+      "grad_norm": 0.05179463326931,
+      "learning_rate": 4.667756481610866e-07,
+      "loss": 0.0002,
+      "step": 583
+    },
+    {
+      "epoch": 4.954545454545455,
+      "grad_norm": 0.04754333943128586,
+      "learning_rate": 4.5921215731654144e-07,
+      "loss": 0.0003,
+      "step": 584
+    },
+    {
+      "epoch": 4.963068181818182,
+      "grad_norm": 0.08957211673259735,
+      "learning_rate": 4.51704244564298e-07,
+      "loss": 0.0005,
+      "step": 585
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 702,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 117,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3035765497604342e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}