diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,34 +1,6798 @@
 {
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "best_metric": 0.007014184258878231,
+  "best_model_checkpoint": "/home/paperspace/Data/models/centime/llm3br256/checkpoint-800",
+  "epoch": 4.468599033816425,
   "eval_steps": 25,
-  "global_step": 1,
+  "global_step": 925,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.0,
-      "grad_norm": 0.8199809193611145,
-      "learning_rate": 0.0001,
-      "loss": 0.2239,
+      "epoch": 0.004830917874396135,
+      "grad_norm": 0.631982147693634,
+      "learning_rate": 1.9305019305019306e-07,
+      "loss": 0.1809,
       "step": 1
     },
     {
-      "epoch": 1.0,
-      "step": 1,
-      "total_flos": 467769719488512.0,
-      "train_loss": 0.2238682359457016,
-      "train_runtime": 24.0578,
-      "train_samples_per_second": 0.291,
-      "train_steps_per_second": 0.042
+      "epoch": 0.00966183574879227,
+      "grad_norm": 0.5283491015434265,
+      "learning_rate": 3.8610038610038613e-07,
+      "loss": 0.17,
+      "step": 2
+    },
+    {
+      "epoch": 0.014492753623188406,
+      "grad_norm": 0.5738242864608765,
+      "learning_rate": 5.791505791505791e-07,
+      "loss": 0.1893,
+      "step": 3
+    },
+    {
+      "epoch": 0.01932367149758454,
+      "grad_norm": 0.5831884741783142,
+      "learning_rate": 7.722007722007723e-07,
+      "loss": 0.1764,
+      "step": 4
+    },
+    {
+      "epoch": 0.024154589371980676,
+      "grad_norm": 0.5098283886909485,
+      "learning_rate": 9.652509652509653e-07,
+      "loss": 0.1701,
+      "step": 5
+    },
+    {
+      "epoch": 0.028985507246376812,
+      "grad_norm": 0.5736549496650696,
+      "learning_rate": 1.1583011583011583e-06,
+      "loss": 0.1891,
+      "step": 6
+    },
+    {
+      "epoch": 0.033816425120772944,
+      "grad_norm": 0.5005951523780823,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 0.164,
+      "step": 7
+    },
+    {
+      "epoch": 0.03864734299516908,
+      "grad_norm": 0.591622531414032,
+      "learning_rate": 1.5444015444015445e-06,
+      "loss": 0.1737,
+      "step": 8
+    },
+    {
+      "epoch": 0.043478260869565216,
+      "grad_norm": 0.5805624127388,
+      "learning_rate": 1.7374517374517375e-06,
+      "loss": 0.1965,
+      "step": 9
+    },
+    {
+      "epoch": 0.04830917874396135,
+      "grad_norm": 0.491947740316391,
+      "learning_rate": 1.9305019305019305e-06,
+      "loss": 0.1741,
+      "step": 10
+    },
+    {
+      "epoch": 0.05314009661835749,
+      "grad_norm": 0.431071937084198,
+      "learning_rate": 2.1235521235521236e-06,
+      "loss": 0.1531,
+      "step": 11
+    },
+    {
+      "epoch": 0.057971014492753624,
+      "grad_norm": 0.46217894554138184,
+      "learning_rate": 2.3166023166023166e-06,
+      "loss": 0.1544,
+      "step": 12
+    },
+    {
+      "epoch": 0.06280193236714976,
+      "grad_norm": 0.4673991799354553,
+      "learning_rate": 2.5096525096525096e-06,
+      "loss": 0.1599,
+      "step": 13
+    },
+    {
+      "epoch": 0.06763285024154589,
+      "grad_norm": 0.4075062572956085,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 0.1733,
+      "step": 14
+    },
+    {
+      "epoch": 0.07246376811594203,
+      "grad_norm": 0.38460931181907654,
+      "learning_rate": 2.895752895752896e-06,
+      "loss": 0.1754,
+      "step": 15
+    },
+    {
+      "epoch": 0.07729468599033816,
+      "grad_norm": 0.3275028169155121,
+      "learning_rate": 3.088803088803089e-06,
+      "loss": 0.1412,
+      "step": 16
+    },
+    {
+      "epoch": 0.0821256038647343,
+      "grad_norm": 0.3470560908317566,
+      "learning_rate": 3.2818532818532816e-06,
+      "loss": 0.1558,
+      "step": 17
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "grad_norm": 0.3038932979106903,
+      "learning_rate": 3.474903474903475e-06,
+      "loss": 0.1467,
+      "step": 18
+    },
+    {
+      "epoch": 0.09178743961352658,
+      "grad_norm": 0.34106558561325073,
+      "learning_rate": 3.6679536679536685e-06,
+      "loss": 0.1603,
+      "step": 19
+    },
+    {
+      "epoch": 0.0966183574879227,
+      "grad_norm": 0.27034294605255127,
+      "learning_rate": 3.861003861003861e-06,
+      "loss": 0.1334,
+      "step": 20
+    },
+    {
+      "epoch": 0.10144927536231885,
+      "grad_norm": 0.2735786437988281,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 0.1411,
+      "step": 21
+    },
+    {
+      "epoch": 0.10628019323671498,
+      "grad_norm": 0.26440420746803284,
+      "learning_rate": 4.247104247104247e-06,
+      "loss": 0.1448,
+      "step": 22
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 0.2545959949493408,
+      "learning_rate": 4.4401544401544405e-06,
+      "loss": 0.1352,
+      "step": 23
+    },
+    {
+      "epoch": 0.11594202898550725,
+      "grad_norm": 0.31827524304389954,
+      "learning_rate": 4.633204633204633e-06,
+      "loss": 0.1509,
+      "step": 24
+    },
+    {
+      "epoch": 0.12077294685990338,
+      "grad_norm": 0.2550322115421295,
+      "learning_rate": 4.8262548262548266e-06,
+      "loss": 0.1159,
+      "step": 25
+    },
+    {
+      "epoch": 0.12077294685990338,
+      "eval_loss": 0.10044483840465546,
+      "eval_runtime": 21.3847,
+      "eval_samples_per_second": 4.676,
+      "eval_steps_per_second": 0.14,
+      "step": 25
+    },
+    {
+      "epoch": 0.12560386473429952,
+      "grad_norm": 0.312794029712677,
+      "learning_rate": 5.019305019305019e-06,
+      "loss": 0.1705,
+      "step": 26
+    },
+    {
+      "epoch": 0.13043478260869565,
+      "grad_norm": 0.2570686936378479,
+      "learning_rate": 5.212355212355213e-06,
+      "loss": 0.1173,
+      "step": 27
+    },
+    {
+      "epoch": 0.13526570048309178,
+      "grad_norm": 0.1603628247976303,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 0.0852,
+      "step": 28
+    },
+    {
+      "epoch": 0.14009661835748793,
+      "grad_norm": 0.19329562783241272,
+      "learning_rate": 5.598455598455599e-06,
+      "loss": 0.1189,
+      "step": 29
+    },
+    {
+      "epoch": 0.14492753623188406,
+      "grad_norm": 0.19762450456619263,
+      "learning_rate": 5.791505791505792e-06,
+      "loss": 0.1149,
+      "step": 30
+    },
+    {
+      "epoch": 0.1497584541062802,
+      "grad_norm": 0.1888991743326187,
+      "learning_rate": 5.984555984555985e-06,
+      "loss": 0.1176,
+      "step": 31
+    },
+    {
+      "epoch": 0.15458937198067632,
+      "grad_norm": 0.20226870477199554,
+      "learning_rate": 6.177606177606178e-06,
+      "loss": 0.1362,
+      "step": 32
+    },
+    {
+      "epoch": 0.15942028985507245,
+      "grad_norm": 0.1509280502796173,
+      "learning_rate": 6.370656370656371e-06,
+      "loss": 0.0817,
+      "step": 33
+    },
+    {
+      "epoch": 0.1642512077294686,
+      "grad_norm": 0.1972505748271942,
+      "learning_rate": 6.563706563706563e-06,
+      "loss": 0.134,
+      "step": 34
+    },
+    {
+      "epoch": 0.16908212560386474,
+      "grad_norm": 0.18301893770694733,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 0.1113,
+      "step": 35
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 0.15621688961982727,
+      "learning_rate": 6.94980694980695e-06,
+      "loss": 0.106,
+      "step": 36
+    },
+    {
+      "epoch": 0.178743961352657,
+      "grad_norm": 0.16429874300956726,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.1061,
+      "step": 37
+    },
+    {
+      "epoch": 0.18357487922705315,
+      "grad_norm": 0.1622830629348755,
+      "learning_rate": 7.335907335907337e-06,
+      "loss": 0.0986,
+      "step": 38
+    },
+    {
+      "epoch": 0.18840579710144928,
+      "grad_norm": 0.1767168790102005,
+      "learning_rate": 7.52895752895753e-06,
+      "loss": 0.1014,
+      "step": 39
+    },
+    {
+      "epoch": 0.1932367149758454,
+      "grad_norm": 0.1411493867635727,
+      "learning_rate": 7.722007722007722e-06,
+      "loss": 0.0862,
+      "step": 40
+    },
+    {
+      "epoch": 0.19806763285024154,
+      "grad_norm": 0.15203985571861267,
+      "learning_rate": 7.915057915057915e-06,
+      "loss": 0.0955,
+      "step": 41
+    },
+    {
+      "epoch": 0.2028985507246377,
+      "grad_norm": 0.13702470064163208,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.0921,
+      "step": 42
+    },
+    {
+      "epoch": 0.20772946859903382,
+      "grad_norm": 0.14217808842658997,
+      "learning_rate": 8.301158301158302e-06,
+      "loss": 0.0855,
+      "step": 43
+    },
+    {
+      "epoch": 0.21256038647342995,
+      "grad_norm": 0.1266465187072754,
+      "learning_rate": 8.494208494208494e-06,
+      "loss": 0.091,
+      "step": 44
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 0.1271313726902008,
+      "learning_rate": 8.687258687258689e-06,
+      "loss": 0.0759,
+      "step": 45
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.15804509818553925,
+      "learning_rate": 8.880308880308881e-06,
+      "loss": 0.0966,
+      "step": 46
+    },
+    {
+      "epoch": 0.22705314009661837,
+      "grad_norm": 0.12065310776233673,
+      "learning_rate": 9.073359073359074e-06,
+      "loss": 0.0736,
+      "step": 47
+    },
+    {
+      "epoch": 0.2318840579710145,
+      "grad_norm": 0.13479556143283844,
+      "learning_rate": 9.266409266409266e-06,
+      "loss": 0.0885,
+      "step": 48
+    },
+    {
+      "epoch": 0.23671497584541062,
+      "grad_norm": 0.12735073268413544,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 0.059,
+      "step": 49
+    },
+    {
+      "epoch": 0.24154589371980675,
+      "grad_norm": 0.13692739605903625,
+      "learning_rate": 9.652509652509653e-06,
+      "loss": 0.0843,
+      "step": 50
+    },
+    {
+      "epoch": 0.24154589371980675,
+      "eval_loss": 0.06345108151435852,
+      "eval_runtime": 20.328,
+      "eval_samples_per_second": 4.919,
+      "eval_steps_per_second": 0.148,
+      "step": 50
+    },
+    {
+      "epoch": 0.2463768115942029,
+      "grad_norm": 0.1594228297472,
+      "learning_rate": 9.845559845559846e-06,
+      "loss": 0.092,
+      "step": 51
+    },
+    {
+      "epoch": 0.25120772946859904,
+      "grad_norm": 0.11453366279602051,
+      "learning_rate": 1.0038610038610038e-05,
+      "loss": 0.0802,
+      "step": 52
+    },
+    {
+      "epoch": 0.2560386473429952,
+      "grad_norm": 0.12839704751968384,
+      "learning_rate": 1.0231660231660233e-05,
+      "loss": 0.0787,
+      "step": 53
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "grad_norm": 0.10960253328084946,
+      "learning_rate": 1.0424710424710425e-05,
+      "loss": 0.0776,
+      "step": 54
+    },
+    {
+      "epoch": 0.26570048309178745,
+      "grad_norm": 0.1108475923538208,
+      "learning_rate": 1.0617760617760618e-05,
+      "loss": 0.0655,
+      "step": 55
+    },
+    {
+      "epoch": 0.27053140096618356,
+      "grad_norm": 0.12586434185504913,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.0841,
+      "step": 56
+    },
+    {
+      "epoch": 0.2753623188405797,
+      "grad_norm": 0.1350884884595871,
+      "learning_rate": 1.1003861003861005e-05,
+      "loss": 0.0649,
+      "step": 57
+    },
+    {
+      "epoch": 0.28019323671497587,
+      "grad_norm": 0.12750767171382904,
+      "learning_rate": 1.1196911196911197e-05,
+      "loss": 0.1061,
+      "step": 58
+    },
+    {
+      "epoch": 0.28502415458937197,
+      "grad_norm": 0.1260327398777008,
+      "learning_rate": 1.138996138996139e-05,
+      "loss": 0.0819,
+      "step": 59
+    },
+    {
+      "epoch": 0.2898550724637681,
+      "grad_norm": 0.12157632410526276,
+      "learning_rate": 1.1583011583011584e-05,
+      "loss": 0.0744,
+      "step": 60
+    },
+    {
+      "epoch": 0.2946859903381642,
+      "grad_norm": 0.10264753550291061,
+      "learning_rate": 1.1776061776061777e-05,
+      "loss": 0.0695,
+      "step": 61
+    },
+    {
+      "epoch": 0.2995169082125604,
+      "grad_norm": 0.1185842826962471,
+      "learning_rate": 1.196911196911197e-05,
+      "loss": 0.0825,
+      "step": 62
+    },
+    {
+      "epoch": 0.30434782608695654,
+      "grad_norm": 0.14522415399551392,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 0.0796,
+      "step": 63
+    },
+    {
+      "epoch": 0.30917874396135264,
+      "grad_norm": 0.1346604973077774,
+      "learning_rate": 1.2355212355212356e-05,
+      "loss": 0.0782,
+      "step": 64
+    },
+    {
+      "epoch": 0.3140096618357488,
+      "grad_norm": 0.10661154985427856,
+      "learning_rate": 1.2548262548262549e-05,
+      "loss": 0.0848,
+      "step": 65
+    },
+    {
+      "epoch": 0.3188405797101449,
+      "grad_norm": 0.09096452593803406,
+      "learning_rate": 1.2741312741312741e-05,
+      "loss": 0.0666,
+      "step": 66
+    },
+    {
+      "epoch": 0.32367149758454106,
+      "grad_norm": 0.11585734784603119,
+      "learning_rate": 1.2934362934362934e-05,
+      "loss": 0.0518,
+      "step": 67
+    },
+    {
+      "epoch": 0.3285024154589372,
+      "grad_norm": 0.102788545191288,
+      "learning_rate": 1.3127413127413127e-05,
+      "loss": 0.0664,
+      "step": 68
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.1031409502029419,
+      "learning_rate": 1.3320463320463322e-05,
+      "loss": 0.071,
+      "step": 69
+    },
+    {
+      "epoch": 0.33816425120772947,
+      "grad_norm": 0.1045219823718071,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.0658,
+      "step": 70
+    },
+    {
+      "epoch": 0.34299516908212563,
+      "grad_norm": 0.09023850411176682,
+      "learning_rate": 1.3706563706563708e-05,
+      "loss": 0.0517,
+      "step": 71
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 0.08952303230762482,
+      "learning_rate": 1.38996138996139e-05,
+      "loss": 0.0534,
+      "step": 72
+    },
+    {
+      "epoch": 0.3526570048309179,
+      "grad_norm": 0.09404554218053818,
+      "learning_rate": 1.4092664092664093e-05,
+      "loss": 0.0652,
+      "step": 73
+    },
+    {
+      "epoch": 0.357487922705314,
+      "grad_norm": 0.09435431659221649,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.0598,
+      "step": 74
+    },
+    {
+      "epoch": 0.36231884057971014,
+      "grad_norm": 0.13001886010169983,
+      "learning_rate": 1.4478764478764478e-05,
+      "loss": 0.0763,
+      "step": 75
+    },
+    {
+      "epoch": 0.36231884057971014,
+      "eval_loss": 0.04738979786634445,
+      "eval_runtime": 20.3171,
+      "eval_samples_per_second": 4.922,
+      "eval_steps_per_second": 0.148,
+      "step": 75
+    },
+    {
+      "epoch": 0.3671497584541063,
+      "grad_norm": 0.10421048104763031,
+      "learning_rate": 1.4671814671814674e-05,
+      "loss": 0.0686,
+      "step": 76
+    },
+    {
+      "epoch": 0.3719806763285024,
+      "grad_norm": 0.0982724204659462,
+      "learning_rate": 1.4864864864864867e-05,
+      "loss": 0.052,
+      "step": 77
+    },
+    {
+      "epoch": 0.37681159420289856,
+      "grad_norm": 0.11460539698600769,
+      "learning_rate": 1.505791505791506e-05,
+      "loss": 0.0632,
+      "step": 78
+    },
+    {
+      "epoch": 0.38164251207729466,
+      "grad_norm": 0.0933605507016182,
+      "learning_rate": 1.5250965250965252e-05,
+      "loss": 0.0479,
+      "step": 79
+    },
+    {
+      "epoch": 0.3864734299516908,
+      "grad_norm": 0.11033626645803452,
+      "learning_rate": 1.5444015444015444e-05,
+      "loss": 0.059,
+      "step": 80
+    },
+    {
+      "epoch": 0.391304347826087,
+      "grad_norm": 0.1291951686143875,
+      "learning_rate": 1.5637065637065637e-05,
+      "loss": 0.073,
+      "step": 81
+    },
+    {
+      "epoch": 0.3961352657004831,
+      "grad_norm": 0.11451836675405502,
+      "learning_rate": 1.583011583011583e-05,
+      "loss": 0.06,
+      "step": 82
+    },
+    {
+      "epoch": 0.40096618357487923,
+      "grad_norm": 0.1261017769575119,
+      "learning_rate": 1.6023166023166026e-05,
+      "loss": 0.0587,
+      "step": 83
+    },
+    {
+      "epoch": 0.4057971014492754,
+      "grad_norm": 0.1151406541466713,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.0551,
+      "step": 84
+    },
+    {
+      "epoch": 0.4106280193236715,
+      "grad_norm": 0.10128024965524673,
+      "learning_rate": 1.640926640926641e-05,
+      "loss": 0.0511,
+      "step": 85
+    },
+    {
+      "epoch": 0.41545893719806765,
+      "grad_norm": 0.1041695848107338,
+      "learning_rate": 1.6602316602316603e-05,
+      "loss": 0.0676,
+      "step": 86
+    },
+    {
+      "epoch": 0.42028985507246375,
+      "grad_norm": 0.1372576802968979,
+      "learning_rate": 1.6795366795366796e-05,
+      "loss": 0.0656,
+      "step": 87
+    },
+    {
+      "epoch": 0.4251207729468599,
+      "grad_norm": 0.11239906400442123,
+      "learning_rate": 1.698841698841699e-05,
+      "loss": 0.0438,
+      "step": 88
+    },
+    {
+      "epoch": 0.42995169082125606,
+      "grad_norm": 0.10844457149505615,
+      "learning_rate": 1.718146718146718e-05,
+      "loss": 0.0567,
+      "step": 89
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 0.11423434317111969,
+      "learning_rate": 1.7374517374517377e-05,
+      "loss": 0.0669,
+      "step": 90
+    },
+    {
+      "epoch": 0.4396135265700483,
+      "grad_norm": 0.108824722468853,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 0.0582,
+      "step": 91
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.1085321456193924,
+      "learning_rate": 1.7760617760617762e-05,
+      "loss": 0.0518,
+      "step": 92
+    },
+    {
+      "epoch": 0.4492753623188406,
+      "grad_norm": 0.16060896217823029,
+      "learning_rate": 1.7953667953667955e-05,
+      "loss": 0.0853,
+      "step": 93
+    },
+    {
+      "epoch": 0.45410628019323673,
+      "grad_norm": 0.10322371125221252,
+      "learning_rate": 1.8146718146718147e-05,
+      "loss": 0.0434,
+      "step": 94
+    },
+    {
+      "epoch": 0.45893719806763283,
+      "grad_norm": 0.1057286411523819,
+      "learning_rate": 1.833976833976834e-05,
+      "loss": 0.0488,
+      "step": 95
+    },
+    {
+      "epoch": 0.463768115942029,
+      "grad_norm": 0.1063227504491806,
+      "learning_rate": 1.8532818532818533e-05,
+      "loss": 0.0521,
+      "step": 96
+    },
+    {
+      "epoch": 0.46859903381642515,
+      "grad_norm": 0.12127968668937683,
+      "learning_rate": 1.8725868725868725e-05,
+      "loss": 0.0548,
+      "step": 97
+    },
+    {
+      "epoch": 0.47342995169082125,
+      "grad_norm": 0.10702498257160187,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.0481,
+      "step": 98
+    },
+    {
+      "epoch": 0.4782608695652174,
+      "grad_norm": 0.10290008783340454,
+      "learning_rate": 1.9111969111969114e-05,
+      "loss": 0.0516,
+      "step": 99
+    },
+    {
+      "epoch": 0.4830917874396135,
+      "grad_norm": 0.10148600488901138,
+      "learning_rate": 1.9305019305019306e-05,
+      "loss": 0.0496,
+      "step": 100
+    },
+    {
+      "epoch": 0.4830917874396135,
+      "eval_loss": 0.036481305956840515,
+      "eval_runtime": 20.2995,
+      "eval_samples_per_second": 4.926,
+      "eval_steps_per_second": 0.148,
+      "step": 100
+    },
+    {
+      "epoch": 0.48792270531400966,
+      "grad_norm": 0.10877996683120728,
+      "learning_rate": 1.94980694980695e-05,
+      "loss": 0.0523,
+      "step": 101
+    },
+    {
+      "epoch": 0.4927536231884058,
+      "grad_norm": 0.11746326833963394,
+      "learning_rate": 1.969111969111969e-05,
+      "loss": 0.0448,
+      "step": 102
+    },
+    {
+      "epoch": 0.4975845410628019,
+      "grad_norm": 0.1287139356136322,
+      "learning_rate": 1.9884169884169884e-05,
+      "loss": 0.054,
+      "step": 103
+    },
+    {
+      "epoch": 0.5024154589371981,
+      "grad_norm": 0.10861372202634811,
+      "learning_rate": 2.0077220077220077e-05,
+      "loss": 0.0486,
+      "step": 104
+    },
+    {
+      "epoch": 0.5072463768115942,
+      "grad_norm": 0.10173708945512772,
+      "learning_rate": 2.0270270270270273e-05,
+      "loss": 0.0413,
+      "step": 105
+    },
+    {
+      "epoch": 0.5120772946859904,
+      "grad_norm": 0.09587957710027695,
+      "learning_rate": 2.0463320463320465e-05,
+      "loss": 0.0429,
+      "step": 106
+    },
+    {
+      "epoch": 0.5169082125603864,
+      "grad_norm": 0.10803103446960449,
+      "learning_rate": 2.0656370656370658e-05,
+      "loss": 0.0376,
+      "step": 107
+    },
+    {
+      "epoch": 0.5217391304347826,
+      "grad_norm": 0.12559597194194794,
+      "learning_rate": 2.084942084942085e-05,
+      "loss": 0.0435,
+      "step": 108
+    },
+    {
+      "epoch": 0.5265700483091788,
+      "grad_norm": 0.12464939057826996,
+      "learning_rate": 2.1042471042471043e-05,
+      "loss": 0.0527,
+      "step": 109
+    },
+    {
+      "epoch": 0.5314009661835749,
+      "grad_norm": 0.11706342548131943,
+      "learning_rate": 2.1235521235521236e-05,
+      "loss": 0.0451,
+      "step": 110
+    },
+    {
+      "epoch": 0.5362318840579711,
+      "grad_norm": 0.10209579020738602,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.0405,
+      "step": 111
+    },
+    {
+      "epoch": 0.5410628019323671,
+      "grad_norm": 0.12944906949996948,
+      "learning_rate": 2.1621621621621624e-05,
+      "loss": 0.0536,
+      "step": 112
+    },
+    {
+      "epoch": 0.5458937198067633,
+      "grad_norm": 0.10440570116043091,
+      "learning_rate": 2.1814671814671817e-05,
+      "loss": 0.0353,
+      "step": 113
+    },
+    {
+      "epoch": 0.5507246376811594,
+      "grad_norm": 0.23940707743167877,
+      "learning_rate": 2.200772200772201e-05,
+      "loss": 0.0507,
+      "step": 114
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.11977086216211319,
+      "learning_rate": 2.2200772200772202e-05,
+      "loss": 0.0538,
+      "step": 115
+    },
+    {
+      "epoch": 0.5603864734299517,
+      "grad_norm": 0.15066705644130707,
+      "learning_rate": 2.2393822393822394e-05,
+      "loss": 0.0525,
+      "step": 116
+    },
+    {
+      "epoch": 0.5652173913043478,
+      "grad_norm": 0.12231211364269257,
+      "learning_rate": 2.2586872586872587e-05,
+      "loss": 0.0441,
+      "step": 117
+    },
+    {
+      "epoch": 0.5700483091787439,
+      "grad_norm": 0.09432463347911835,
+      "learning_rate": 2.277992277992278e-05,
+      "loss": 0.0345,
+      "step": 118
+    },
+    {
+      "epoch": 0.5748792270531401,
+      "grad_norm": 0.15186284482479095,
+      "learning_rate": 2.2972972972972976e-05,
+      "loss": 0.0495,
+      "step": 119
+    },
+    {
+      "epoch": 0.5797101449275363,
+      "grad_norm": 0.10990628600120544,
+      "learning_rate": 2.3166023166023168e-05,
+      "loss": 0.0387,
+      "step": 120
+    },
+    {
+      "epoch": 0.5845410628019324,
+      "grad_norm": 0.12975062429904938,
+      "learning_rate": 2.335907335907336e-05,
+      "loss": 0.0416,
+      "step": 121
+    },
+    {
+      "epoch": 0.5893719806763285,
+      "grad_norm": 0.11566514521837234,
+      "learning_rate": 2.3552123552123553e-05,
+      "loss": 0.0386,
+      "step": 122
+    },
+    {
+      "epoch": 0.5942028985507246,
+      "grad_norm": 0.11288753151893616,
+      "learning_rate": 2.3745173745173746e-05,
+      "loss": 0.0435,
+      "step": 123
+    },
+    {
+      "epoch": 0.5990338164251208,
+      "grad_norm": 0.12189905345439911,
+      "learning_rate": 2.393822393822394e-05,
+      "loss": 0.0452,
+      "step": 124
+    },
+    {
+      "epoch": 0.6038647342995169,
+      "grad_norm": 0.11823577433824539,
+      "learning_rate": 2.413127413127413e-05,
+      "loss": 0.046,
+      "step": 125
+    },
+    {
+      "epoch": 0.6038647342995169,
+      "eval_loss": 0.03164202719926834,
+      "eval_runtime": 20.309,
+      "eval_samples_per_second": 4.924,
+      "eval_steps_per_second": 0.148,
+      "step": 125
+    },
+    {
+      "epoch": 0.6086956521739131,
+      "grad_norm": 0.11446941643953323,
+      "learning_rate": 2.4324324324324327e-05,
+      "loss": 0.0399,
+      "step": 126
+    },
+    {
+      "epoch": 0.6135265700483091,
+      "grad_norm": 0.10811971127986908,
+      "learning_rate": 2.451737451737452e-05,
+      "loss": 0.0269,
+      "step": 127
+    },
+    {
+      "epoch": 0.6183574879227053,
+      "grad_norm": 0.10582166165113449,
+      "learning_rate": 2.4710424710424712e-05,
+      "loss": 0.0408,
+      "step": 128
+    },
+    {
+      "epoch": 0.6231884057971014,
+      "grad_norm": 0.10825195163488388,
+      "learning_rate": 2.4903474903474905e-05,
+      "loss": 0.0439,
+      "step": 129
+    },
+    {
+      "epoch": 0.6280193236714976,
+      "grad_norm": 0.11372285336256027,
+      "learning_rate": 2.5096525096525097e-05,
+      "loss": 0.0345,
+      "step": 130
+    },
+    {
+      "epoch": 0.6328502415458938,
+      "grad_norm": 0.12226233631372452,
+      "learning_rate": 2.528957528957529e-05,
+      "loss": 0.045,
+      "step": 131
+    },
+    {
+      "epoch": 0.6376811594202898,
+      "grad_norm": 0.125160351395607,
+      "learning_rate": 2.5482625482625483e-05,
+      "loss": 0.0508,
+      "step": 132
+    },
+    {
+      "epoch": 0.642512077294686,
+      "grad_norm": 0.10091613233089447,
+      "learning_rate": 2.5675675675675675e-05,
+      "loss": 0.0361,
+      "step": 133
+    },
+    {
+      "epoch": 0.6473429951690821,
+      "grad_norm": 0.09511641412973404,
+      "learning_rate": 2.5868725868725868e-05,
+      "loss": 0.0327,
+      "step": 134
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 0.11392970383167267,
+      "learning_rate": 2.606177606177606e-05,
+      "loss": 0.0427,
+      "step": 135
+    },
+    {
+      "epoch": 0.6570048309178744,
+      "grad_norm": 0.11419506371021271,
+      "learning_rate": 2.6254826254826253e-05,
+      "loss": 0.0395,
+      "step": 136
+    },
+    {
+      "epoch": 0.6618357487922706,
+      "grad_norm": 0.11490583419799805,
+      "learning_rate": 2.6447876447876452e-05,
+      "loss": 0.0297,
+      "step": 137
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.12591791152954102,
+      "learning_rate": 2.6640926640926645e-05,
+      "loss": 0.0346,
+      "step": 138
+    },
+    {
+      "epoch": 0.6714975845410628,
+      "grad_norm": 0.13922667503356934,
+      "learning_rate": 2.6833976833976838e-05,
+      "loss": 0.0407,
+      "step": 139
+    },
+    {
+      "epoch": 0.6763285024154589,
+      "grad_norm": 0.12055017054080963,
+      "learning_rate": 2.702702702702703e-05,
+      "loss": 0.0307,
+      "step": 140
+    },
+    {
+      "epoch": 0.6811594202898551,
+      "grad_norm": 0.10987816751003265,
+      "learning_rate": 2.7220077220077223e-05,
+      "loss": 0.0261,
+      "step": 141
+    },
+    {
+      "epoch": 0.6859903381642513,
+      "grad_norm": 0.14209432899951935,
+      "learning_rate": 2.7413127413127415e-05,
+      "loss": 0.0456,
+      "step": 142
+    },
+    {
+      "epoch": 0.6908212560386473,
+      "grad_norm": 0.11652625352144241,
+      "learning_rate": 2.7606177606177608e-05,
+      "loss": 0.0358,
+      "step": 143
+    },
+    {
+      "epoch": 0.6956521739130435,
+      "grad_norm": 0.12266818434000015,
+      "learning_rate": 2.77992277992278e-05,
+      "loss": 0.0378,
+      "step": 144
+    },
+    {
+      "epoch": 0.7004830917874396,
+      "grad_norm": 0.11468692868947983,
+      "learning_rate": 2.7992277992277993e-05,
+      "loss": 0.0346,
+      "step": 145
+    },
+    {
+      "epoch": 0.7053140096618358,
+      "grad_norm": 0.1078886017203331,
+      "learning_rate": 2.8185328185328186e-05,
+      "loss": 0.034,
+      "step": 146
+    },
+    {
+      "epoch": 0.7101449275362319,
+      "grad_norm": 0.12835174798965454,
+      "learning_rate": 2.8378378378378378e-05,
+      "loss": 0.0444,
+      "step": 147
+    },
+    {
+      "epoch": 0.714975845410628,
+      "grad_norm": 0.12754280865192413,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.0364,
+      "step": 148
+    },
+    {
+      "epoch": 0.7198067632850241,
+      "grad_norm": 0.10432615876197815,
+      "learning_rate": 2.8764478764478763e-05,
+      "loss": 0.0275,
+      "step": 149
+    },
+    {
+      "epoch": 0.7246376811594203,
+      "grad_norm": 0.1269528716802597,
+      "learning_rate": 2.8957528957528956e-05,
+      "loss": 0.0368,
+      "step": 150
+    },
+    {
+      "epoch": 0.7246376811594203,
+      "eval_loss": 0.02661709673702717,
+      "eval_runtime": 20.3433,
+      "eval_samples_per_second": 4.916,
+      "eval_steps_per_second": 0.147,
+      "step": 150
+    },
+    {
+      "epoch": 0.7294685990338164,
+      "grad_norm": 0.10792283713817596,
+      "learning_rate": 2.915057915057915e-05,
+      "loss": 0.0349,
+      "step": 151
+    },
+    {
+      "epoch": 0.7342995169082126,
+      "grad_norm": 0.11990821361541748,
+      "learning_rate": 2.9343629343629348e-05,
+      "loss": 0.0401,
+      "step": 152
+    },
+    {
+      "epoch": 0.7391304347826086,
+      "grad_norm": 0.12300258129835129,
+      "learning_rate": 2.953667953667954e-05,
+      "loss": 0.0332,
+      "step": 153
+    },
+    {
+      "epoch": 0.7439613526570048,
+      "grad_norm": 0.1269589215517044,
+      "learning_rate": 2.9729729729729733e-05,
+      "loss": 0.0348,
+      "step": 154
+    },
+    {
+      "epoch": 0.748792270531401,
+      "grad_norm": 0.1516416221857071,
+      "learning_rate": 2.9922779922779926e-05,
+      "loss": 0.0425,
+      "step": 155
+    },
+    {
+      "epoch": 0.7536231884057971,
+      "grad_norm": 0.11869463324546814,
+      "learning_rate": 3.011583011583012e-05,
+      "loss": 0.0252,
+      "step": 156
+    },
+    {
+      "epoch": 0.7584541062801933,
+      "grad_norm": 0.11784904450178146,
+      "learning_rate": 3.030888030888031e-05,
+      "loss": 0.0295,
+      "step": 157
+    },
+    {
+      "epoch": 0.7632850241545893,
+      "grad_norm": 0.14304758608341217,
+      "learning_rate": 3.0501930501930504e-05,
+      "loss": 0.0348,
+      "step": 158
+    },
+    {
+      "epoch": 0.7681159420289855,
+      "grad_norm": 0.13934855163097382,
+      "learning_rate": 3.0694980694980696e-05,
+      "loss": 0.0314,
+      "step": 159
+    },
+    {
+      "epoch": 0.7729468599033816,
+      "grad_norm": 0.15357963740825653,
+      "learning_rate": 3.088803088803089e-05,
+      "loss": 0.037,
+      "step": 160
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.17761190235614777,
+      "learning_rate": 3.108108108108108e-05,
+      "loss": 0.0331,
+      "step": 161
+    },
+    {
+      "epoch": 0.782608695652174,
+      "grad_norm": 0.11713641881942749,
+      "learning_rate": 3.1274131274131274e-05,
+      "loss": 0.0321,
+      "step": 162
+    },
+    {
+      "epoch": 0.7874396135265701,
+      "grad_norm": 0.14508825540542603,
+      "learning_rate": 3.1467181467181466e-05,
+      "loss": 0.031,
+      "step": 163
+    },
+    {
+      "epoch": 0.7922705314009661,
+      "grad_norm": 0.1814957559108734,
+      "learning_rate": 3.166023166023166e-05,
+      "loss": 0.0345,
+      "step": 164
+    },
+    {
+      "epoch": 0.7971014492753623,
+      "grad_norm": 0.10010110586881638,
+      "learning_rate": 3.185328185328185e-05,
+      "loss": 0.0426,
+      "step": 165
+    },
+    {
+      "epoch": 0.8019323671497585,
+      "grad_norm": 0.1516270786523819,
+      "learning_rate": 3.204633204633205e-05,
+      "loss": 0.0434,
+      "step": 166
+    },
+    {
+      "epoch": 0.8067632850241546,
+      "grad_norm": 0.131612628698349,
+      "learning_rate": 3.2239382239382244e-05,
+      "loss": 0.0354,
+      "step": 167
+    },
+    {
+      "epoch": 0.8115942028985508,
+      "grad_norm": 0.14485694468021393,
+      "learning_rate": 3.2432432432432436e-05,
+      "loss": 0.0322,
+      "step": 168
+    },
+    {
+      "epoch": 0.8164251207729468,
+      "grad_norm": 0.18271975219249725,
+      "learning_rate": 3.262548262548263e-05,
+      "loss": 0.0338,
+      "step": 169
+    },
+    {
+      "epoch": 0.821256038647343,
+      "grad_norm": 0.11214082688093185,
+      "learning_rate": 3.281853281853282e-05,
+      "loss": 0.0345,
+      "step": 170
+    },
+    {
+      "epoch": 0.8260869565217391,
+      "grad_norm": 0.1456298530101776,
+      "learning_rate": 3.3011583011583014e-05,
+      "loss": 0.0355,
+      "step": 171
+    },
+    {
+      "epoch": 0.8309178743961353,
+      "grad_norm": 0.17988459765911102,
+      "learning_rate": 3.3204633204633207e-05,
+      "loss": 0.0314,
+      "step": 172
+    },
+    {
+      "epoch": 0.8357487922705314,
+      "grad_norm": 0.1143067479133606,
+      "learning_rate": 3.33976833976834e-05,
+      "loss": 0.035,
+      "step": 173
+    },
+    {
+      "epoch": 0.8405797101449275,
+      "grad_norm": 0.18110711872577667,
+      "learning_rate": 3.359073359073359e-05,
+      "loss": 0.0296,
+      "step": 174
+    },
+    {
+      "epoch": 0.8454106280193237,
+      "grad_norm": 0.13397149741649628,
+      "learning_rate": 3.3783783783783784e-05,
+      "loss": 0.0283,
+      "step": 175
+    },
+    {
+      "epoch": 0.8454106280193237,
+      "eval_loss": 0.023197688162326813,
+      "eval_runtime": 20.5122,
+      "eval_samples_per_second": 4.875,
+      "eval_steps_per_second": 0.146,
+      "step": 175
+    },
+    {
+      "epoch": 0.8502415458937198,
+      "grad_norm": 0.14032766222953796,
+      "learning_rate": 3.397683397683398e-05,
+      "loss": 0.0397,
+      "step": 176
+    },
+    {
+      "epoch": 0.855072463768116,
+      "grad_norm": 0.12052756547927856,
+      "learning_rate": 3.416988416988417e-05,
+      "loss": 0.0269,
+      "step": 177
+    },
+    {
+      "epoch": 0.8599033816425121,
+      "grad_norm": 0.14283765852451324,
+      "learning_rate": 3.436293436293436e-05,
+      "loss": 0.0328,
+      "step": 178
+    },
+    {
+      "epoch": 0.8647342995169082,
+      "grad_norm": 0.12751182913780212,
+      "learning_rate": 3.4555984555984555e-05,
+      "loss": 0.0284,
+      "step": 179
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.09834396094083786,
+      "learning_rate": 3.4749034749034754e-05,
+      "loss": 0.0232,
+      "step": 180
+    },
+    {
+      "epoch": 0.8743961352657005,
+      "grad_norm": 0.17338038980960846,
+      "learning_rate": 3.4942084942084947e-05,
+      "loss": 0.0354,
+      "step": 181
+    },
+    {
+      "epoch": 0.8792270531400966,
+      "grad_norm": 0.1350838541984558,
+      "learning_rate": 3.513513513513514e-05,
+      "loss": 0.0354,
+      "step": 182
+    },
+    {
+      "epoch": 0.8840579710144928,
+      "grad_norm": 0.12952224910259247,
+      "learning_rate": 3.532818532818533e-05,
+      "loss": 0.0294,
+      "step": 183
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.1229347437620163,
+      "learning_rate": 3.5521235521235524e-05,
+      "loss": 0.0303,
+      "step": 184
+    },
+    {
+      "epoch": 0.893719806763285,
+      "grad_norm": 0.11522738635540009,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.0296,
+      "step": 185
+    },
+    {
+      "epoch": 0.8985507246376812,
+      "grad_norm": 0.1309077888727188,
+      "learning_rate": 3.590733590733591e-05,
+      "loss": 0.0338,
+      "step": 186
+    },
+    {
+      "epoch": 0.9033816425120773,
+      "grad_norm": 0.1434311419725418,
+      "learning_rate": 3.61003861003861e-05,
+      "loss": 0.0375,
+      "step": 187
+    },
+    {
+      "epoch": 0.9082125603864735,
+      "grad_norm": 0.13514019548892975,
+      "learning_rate": 3.6293436293436295e-05,
+      "loss": 0.0344,
+      "step": 188
+    },
+    {
+      "epoch": 0.9130434782608695,
+      "grad_norm": 0.13007153570652008,
+      "learning_rate": 3.648648648648649e-05,
+      "loss": 0.0293,
+      "step": 189
+    },
+    {
+      "epoch": 0.9178743961352657,
+      "grad_norm": 0.10133199393749237,
+      "learning_rate": 3.667953667953668e-05,
+      "loss": 0.0196,
+      "step": 190
+    },
+    {
+      "epoch": 0.9227053140096618,
+      "grad_norm": 0.15782149136066437,
+      "learning_rate": 3.687258687258687e-05,
+      "loss": 0.0294,
+      "step": 191
+    },
+    {
+      "epoch": 0.927536231884058,
+      "grad_norm": 0.10956969112157822,
+      "learning_rate": 3.7065637065637065e-05,
+      "loss": 0.0274,
+      "step": 192
+    },
+    {
+      "epoch": 0.9323671497584541,
+      "grad_norm": 0.1855483055114746,
+      "learning_rate": 3.725868725868726e-05,
+      "loss": 0.0326,
+      "step": 193
+    },
+    {
+      "epoch": 0.9371980676328503,
+      "grad_norm": 0.14759743213653564,
+      "learning_rate": 3.745173745173745e-05,
+      "loss": 0.0281,
+      "step": 194
+    },
+    {
+      "epoch": 0.9420289855072463,
+      "grad_norm": 0.19224123656749725,
+      "learning_rate": 3.764478764478765e-05,
+      "loss": 0.0235,
+      "step": 195
+    },
+    {
+      "epoch": 0.9468599033816425,
+      "grad_norm": 0.12383207678794861,
+      "learning_rate": 3.783783783783784e-05,
+      "loss": 0.0227,
+      "step": 196
+    },
+    {
+      "epoch": 0.9516908212560387,
+      "grad_norm": 0.13436304032802582,
+      "learning_rate": 3.8030888030888035e-05,
+      "loss": 0.0248,
+      "step": 197
+    },
+    {
+      "epoch": 0.9565217391304348,
+      "grad_norm": 0.11358322203159332,
+      "learning_rate": 3.822393822393823e-05,
+      "loss": 0.0272,
+      "step": 198
+    },
+    {
+      "epoch": 0.961352657004831,
+      "grad_norm": 0.09752532094717026,
+      "learning_rate": 3.841698841698842e-05,
+      "loss": 0.0264,
+      "step": 199
+    },
+    {
+      "epoch": 0.966183574879227,
+      "grad_norm": 0.11581085622310638,
+      "learning_rate": 3.861003861003861e-05,
+      "loss": 0.0237,
+      "step": 200
+    },
+    {
+      "epoch": 0.966183574879227,
+      "eval_loss": 0.021242819726467133,
+      "eval_runtime": 20.3025,
+      "eval_samples_per_second": 4.925,
+      "eval_steps_per_second": 0.148,
+      "step": 200
+    },
+    {
+      "epoch": 0.9710144927536232,
+      "grad_norm": 0.11759760975837708,
+      "learning_rate": 3.8803088803088805e-05,
+      "loss": 0.0357,
+      "step": 201
+    },
+    {
+      "epoch": 0.9758454106280193,
+      "grad_norm": 0.1146010234951973,
+      "learning_rate": 3.8996138996139e-05,
+      "loss": 0.0293,
+      "step": 202
+    },
+    {
+      "epoch": 0.9806763285024155,
+      "grad_norm": 0.08743493258953094,
+      "learning_rate": 3.918918918918919e-05,
+      "loss": 0.0233,
+      "step": 203
+    },
+    {
+      "epoch": 0.9855072463768116,
+      "grad_norm": 0.09883563965559006,
+      "learning_rate": 3.938223938223938e-05,
+      "loss": 0.0277,
+      "step": 204
+    },
+    {
+      "epoch": 0.9903381642512077,
+      "grad_norm": 0.1223803460597992,
+      "learning_rate": 3.9575289575289576e-05,
+      "loss": 0.033,
+      "step": 205
+    },
+    {
+      "epoch": 0.9951690821256038,
+      "grad_norm": 0.10520818829536438,
+      "learning_rate": 3.976833976833977e-05,
+      "loss": 0.0285,
+      "step": 206
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.19492344558238983,
+      "learning_rate": 3.996138996138996e-05,
+      "loss": 0.0287,
+      "step": 207
+    },
+    {
+      "epoch": 1.0048309178743962,
+      "grad_norm": 0.1388760209083557,
+      "learning_rate": 4.015444015444015e-05,
+      "loss": 0.0276,
+      "step": 208
+    },
+    {
+      "epoch": 1.0096618357487923,
+      "grad_norm": 0.10627097636461258,
+      "learning_rate": 4.034749034749035e-05,
+      "loss": 0.0263,
+      "step": 209
+    },
+    {
+      "epoch": 1.0144927536231885,
+      "grad_norm": 0.13346248865127563,
+      "learning_rate": 4.0540540540540545e-05,
+      "loss": 0.0222,
+      "step": 210
+    },
+    {
+      "epoch": 1.0193236714975846,
+      "grad_norm": 0.14749185740947723,
+      "learning_rate": 4.073359073359074e-05,
+      "loss": 0.0313,
+      "step": 211
+    },
+    {
+      "epoch": 1.0241545893719808,
+      "grad_norm": 0.14564776420593262,
+      "learning_rate": 4.092664092664093e-05,
+      "loss": 0.0282,
+      "step": 212
+    },
+    {
+      "epoch": 1.0289855072463767,
+      "grad_norm": 0.12179957330226898,
+      "learning_rate": 4.111969111969112e-05,
+      "loss": 0.0223,
+      "step": 213
+    },
+    {
+      "epoch": 1.0338164251207729,
+      "grad_norm": 0.12672066688537598,
+      "learning_rate": 4.1312741312741316e-05,
+      "loss": 0.0264,
+      "step": 214
+    },
+    {
+      "epoch": 1.038647342995169,
+      "grad_norm": 0.13626433908939362,
+      "learning_rate": 4.150579150579151e-05,
+      "loss": 0.0299,
+      "step": 215
+    },
+    {
+      "epoch": 1.0434782608695652,
+      "grad_norm": 0.09698859602212906,
+      "learning_rate": 4.16988416988417e-05,
+      "loss": 0.0268,
+      "step": 216
+    },
+    {
+      "epoch": 1.0483091787439613,
+      "grad_norm": 0.16582411527633667,
+      "learning_rate": 4.189189189189189e-05,
+      "loss": 0.0251,
+      "step": 217
+    },
+    {
+      "epoch": 1.0531400966183575,
+      "grad_norm": 0.12611252069473267,
+      "learning_rate": 4.2084942084942086e-05,
+      "loss": 0.0289,
+      "step": 218
+    },
+    {
+      "epoch": 1.0579710144927537,
+      "grad_norm": 0.12253251671791077,
+      "learning_rate": 4.227799227799228e-05,
+      "loss": 0.028,
+      "step": 219
+    },
+    {
+      "epoch": 1.0628019323671498,
+      "grad_norm": 0.12088047713041306,
+      "learning_rate": 4.247104247104247e-05,
+      "loss": 0.0296,
+      "step": 220
+    },
+    {
+      "epoch": 1.067632850241546,
+      "grad_norm": 0.13391311466693878,
+      "learning_rate": 4.2664092664092664e-05,
+      "loss": 0.0229,
+      "step": 221
+    },
+    {
+      "epoch": 1.0724637681159421,
+      "grad_norm": 0.12342043966054916,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.0209,
+      "step": 222
+    },
+    {
+      "epoch": 1.077294685990338,
+      "grad_norm": 0.117287777364254,
+      "learning_rate": 4.305019305019305e-05,
+      "loss": 0.0255,
+      "step": 223
+    },
+    {
+      "epoch": 1.0821256038647342,
+      "grad_norm": 0.15881749987602234,
+      "learning_rate": 4.324324324324325e-05,
+      "loss": 0.0191,
+      "step": 224
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 0.14633190631866455,
+      "learning_rate": 4.343629343629344e-05,
+      "loss": 0.0234,
+      "step": 225
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "eval_loss": 0.019356004893779755,
+      "eval_runtime": 20.3141,
+      "eval_samples_per_second": 4.923,
+      "eval_steps_per_second": 0.148,
+      "step": 225
+    },
+    {
+      "epoch": 1.0917874396135265,
+      "grad_norm": 0.14119146764278412,
+      "learning_rate": 4.3629343629343633e-05,
+      "loss": 0.0262,
+      "step": 226
+    },
+    {
+      "epoch": 1.0966183574879227,
+      "grad_norm": 0.15874817967414856,
+      "learning_rate": 4.3822393822393826e-05,
+      "loss": 0.0219,
+      "step": 227
+    },
+    {
+      "epoch": 1.1014492753623188,
+      "grad_norm": 0.10163708031177521,
+      "learning_rate": 4.401544401544402e-05,
+      "loss": 0.0239,
+      "step": 228
+    },
+    {
+      "epoch": 1.106280193236715,
+      "grad_norm": 0.1453448086977005,
+      "learning_rate": 4.420849420849421e-05,
+      "loss": 0.03,
+      "step": 229
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.11025876551866531,
+      "learning_rate": 4.4401544401544404e-05,
+      "loss": 0.0256,
+      "step": 230
+    },
+    {
+      "epoch": 1.1159420289855073,
+      "grad_norm": 0.08861984312534332,
+      "learning_rate": 4.4594594594594596e-05,
+      "loss": 0.0271,
+      "step": 231
+    },
+    {
+      "epoch": 1.1207729468599035,
+      "grad_norm": 0.08897382766008377,
+      "learning_rate": 4.478764478764479e-05,
+      "loss": 0.024,
+      "step": 232
+    },
+    {
+      "epoch": 1.1256038647342996,
+      "grad_norm": 0.10531286150217056,
+      "learning_rate": 4.498069498069498e-05,
+      "loss": 0.0273,
+      "step": 233
+    },
+    {
+      "epoch": 1.1304347826086956,
+      "grad_norm": 0.10693822056055069,
+      "learning_rate": 4.5173745173745174e-05,
+      "loss": 0.024,
+      "step": 234
+    },
+    {
+      "epoch": 1.1352657004830917,
+      "grad_norm": 0.08394042402505875,
+      "learning_rate": 4.536679536679537e-05,
+      "loss": 0.0177,
+      "step": 235
+    },
+    {
+      "epoch": 1.1400966183574879,
+      "grad_norm": 0.13688276708126068,
+      "learning_rate": 4.555984555984556e-05,
+      "loss": 0.0292,
+      "step": 236
+    },
+    {
+      "epoch": 1.144927536231884,
+      "grad_norm": 0.11339515447616577,
+      "learning_rate": 4.575289575289575e-05,
+      "loss": 0.0171,
+      "step": 237
+    },
+    {
+      "epoch": 1.1497584541062802,
+      "grad_norm": 0.12284926325082779,
+      "learning_rate": 4.594594594594595e-05,
+      "loss": 0.0211,
+      "step": 238
+    },
+    {
+      "epoch": 1.1545893719806763,
+      "grad_norm": 0.16894559562206268,
+      "learning_rate": 4.6138996138996144e-05,
+      "loss": 0.0369,
+      "step": 239
+    },
+    {
+      "epoch": 1.1594202898550725,
+      "grad_norm": 0.09520669281482697,
+      "learning_rate": 4.6332046332046336e-05,
+      "loss": 0.0211,
+      "step": 240
+    },
+    {
+      "epoch": 1.1642512077294687,
+      "grad_norm": 0.13406948745250702,
+      "learning_rate": 4.652509652509653e-05,
+      "loss": 0.0232,
+      "step": 241
+    },
+    {
+      "epoch": 1.1690821256038648,
+      "grad_norm": 0.09464199095964432,
+      "learning_rate": 4.671814671814672e-05,
+      "loss": 0.0267,
+      "step": 242
+    },
+    {
+      "epoch": 1.1739130434782608,
+      "grad_norm": 0.10455941408872604,
+      "learning_rate": 4.6911196911196914e-05,
+      "loss": 0.0199,
+      "step": 243
+    },
+    {
+      "epoch": 1.178743961352657,
+      "grad_norm": 0.13815800845623016,
+      "learning_rate": 4.710424710424711e-05,
+      "loss": 0.0224,
+      "step": 244
+    },
+    {
+      "epoch": 1.183574879227053,
+      "grad_norm": 0.1540479212999344,
+      "learning_rate": 4.72972972972973e-05,
+      "loss": 0.0321,
+      "step": 245
+    },
+    {
+      "epoch": 1.1884057971014492,
+      "grad_norm": 0.129620760679245,
+      "learning_rate": 4.749034749034749e-05,
+      "loss": 0.0184,
+      "step": 246
+    },
+    {
+      "epoch": 1.1932367149758454,
+      "grad_norm": 0.14373724162578583,
+      "learning_rate": 4.7683397683397685e-05,
+      "loss": 0.0279,
+      "step": 247
+    },
+    {
+      "epoch": 1.1980676328502415,
+      "grad_norm": 0.09800299257040024,
+      "learning_rate": 4.787644787644788e-05,
+      "loss": 0.0149,
+      "step": 248
+    },
+    {
+      "epoch": 1.2028985507246377,
+      "grad_norm": 0.11203493922948837,
+      "learning_rate": 4.806949806949807e-05,
+      "loss": 0.026,
+      "step": 249
+    },
+    {
+      "epoch": 1.2077294685990339,
+      "grad_norm": 0.12114997208118439,
+      "learning_rate": 4.826254826254826e-05,
+      "loss": 0.0232,
+      "step": 250
+    },
+    {
+      "epoch": 1.2077294685990339,
+      "eval_loss": 0.01758512854576111,
+      "eval_runtime": 20.3081,
+      "eval_samples_per_second": 4.924,
+      "eval_steps_per_second": 0.148,
+      "step": 250
+    },
+    {
+      "epoch": 1.21256038647343,
+      "grad_norm": 0.08669663965702057,
+      "learning_rate": 4.8455598455598455e-05,
+      "loss": 0.02,
+      "step": 251
+    },
+    {
+      "epoch": 1.2173913043478262,
+      "grad_norm": 0.08553537726402283,
+      "learning_rate": 4.8648648648648654e-05,
+      "loss": 0.0129,
+      "step": 252
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.0841841846704483,
+      "learning_rate": 4.884169884169885e-05,
+      "loss": 0.0191,
+      "step": 253
+    },
+    {
+      "epoch": 1.2270531400966185,
+      "grad_norm": 0.10962934792041779,
+      "learning_rate": 4.903474903474904e-05,
+      "loss": 0.0218,
+      "step": 254
+    },
+    {
+      "epoch": 1.2318840579710144,
+      "grad_norm": 0.0924534946680069,
+      "learning_rate": 4.922779922779923e-05,
+      "loss": 0.0284,
+      "step": 255
+    },
+    {
+      "epoch": 1.2367149758454106,
+      "grad_norm": 0.10882215946912766,
+      "learning_rate": 4.9420849420849425e-05,
+      "loss": 0.0215,
+      "step": 256
+    },
+    {
+      "epoch": 1.2415458937198067,
+      "grad_norm": 0.11583147943019867,
+      "learning_rate": 4.961389961389962e-05,
+      "loss": 0.0235,
+      "step": 257
+    },
+    {
+      "epoch": 1.2463768115942029,
+      "grad_norm": 0.10939005017280579,
+      "learning_rate": 4.980694980694981e-05,
+      "loss": 0.0291,
+      "step": 258
+    },
+    {
+      "epoch": 1.251207729468599,
+      "grad_norm": 0.10021679848432541,
+      "learning_rate": 5e-05,
+      "loss": 0.022,
+      "step": 259
+    },
+    {
+      "epoch": 1.2560386473429952,
+      "grad_norm": 0.10946795344352722,
+      "learning_rate": 5.0193050193050195e-05,
+      "loss": 0.0224,
+      "step": 260
+    },
+    {
+      "epoch": 1.2608695652173914,
+      "grad_norm": 0.10062968730926514,
+      "learning_rate": 5.038610038610039e-05,
+      "loss": 0.0246,
+      "step": 261
+    },
+    {
+      "epoch": 1.2657004830917875,
+      "grad_norm": 0.09000770002603531,
+      "learning_rate": 5.057915057915058e-05,
+      "loss": 0.0214,
+      "step": 262
+    },
+    {
+      "epoch": 1.2705314009661834,
+      "grad_norm": 0.09875966608524323,
+      "learning_rate": 5.077220077220077e-05,
+      "loss": 0.0224,
+      "step": 263
+    },
+    {
+      "epoch": 1.2753623188405796,
+      "grad_norm": 0.14588375389575958,
+      "learning_rate": 5.0965250965250965e-05,
+      "loss": 0.0233,
+      "step": 264
+    },
+    {
+      "epoch": 1.2801932367149758,
+      "grad_norm": 0.08824263513088226,
+      "learning_rate": 5.115830115830116e-05,
+      "loss": 0.0183,
+      "step": 265
+    },
+    {
+      "epoch": 1.285024154589372,
+      "grad_norm": 0.1460726410150528,
+      "learning_rate": 5.135135135135135e-05,
+      "loss": 0.0296,
+      "step": 266
+    },
+    {
+      "epoch": 1.289855072463768,
+      "grad_norm": 0.10470607876777649,
+      "learning_rate": 5.154440154440154e-05,
+      "loss": 0.025,
+      "step": 267
+    },
+    {
+      "epoch": 1.2946859903381642,
+      "grad_norm": 0.1622585654258728,
+      "learning_rate": 5.1737451737451736e-05,
+      "loss": 0.0241,
+      "step": 268
+    },
+    {
+      "epoch": 1.2995169082125604,
+      "grad_norm": 0.11767762899398804,
+      "learning_rate": 5.193050193050193e-05,
+      "loss": 0.0272,
+      "step": 269
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 0.154914990067482,
+      "learning_rate": 5.212355212355212e-05,
+      "loss": 0.0268,
+      "step": 270
+    },
+    {
+      "epoch": 1.3091787439613527,
+      "grad_norm": 0.1314283013343811,
+      "learning_rate": 5.2316602316602313e-05,
+      "loss": 0.0232,
+      "step": 271
+    },
+    {
+      "epoch": 1.3140096618357489,
+      "grad_norm": 0.10727723687887192,
+      "learning_rate": 5.2509652509652506e-05,
+      "loss": 0.0241,
+      "step": 272
+    },
+    {
+      "epoch": 1.318840579710145,
+      "grad_norm": 0.13747063279151917,
+      "learning_rate": 5.27027027027027e-05,
+      "loss": 0.0239,
+      "step": 273
+    },
+    {
+      "epoch": 1.3236714975845412,
+      "grad_norm": 0.08335089683532715,
+      "learning_rate": 5.2895752895752905e-05,
+      "loss": 0.0182,
+      "step": 274
+    },
+    {
+      "epoch": 1.3285024154589373,
+      "grad_norm": 0.12808988988399506,
+      "learning_rate": 5.30888030888031e-05,
+      "loss": 0.0307,
+      "step": 275
+    },
+    {
+      "epoch": 1.3285024154589373,
+      "eval_loss": 0.01781173050403595,
+      "eval_runtime": 20.3141,
+      "eval_samples_per_second": 4.923,
+      "eval_steps_per_second": 0.148,
+      "step": 275
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.10613222420215607,
+      "learning_rate": 5.328185328185329e-05,
+      "loss": 0.0228,
+      "step": 276
+    },
+    {
+      "epoch": 1.3381642512077294,
+      "grad_norm": 0.09993978589773178,
+      "learning_rate": 5.347490347490348e-05,
+      "loss": 0.0239,
+      "step": 277
+    },
+    {
+      "epoch": 1.3429951690821256,
+      "grad_norm": 0.08622147142887115,
+      "learning_rate": 5.3667953667953675e-05,
+      "loss": 0.0192,
+      "step": 278
+    },
+    {
+      "epoch": 1.3478260869565217,
+      "grad_norm": 0.08509330451488495,
+      "learning_rate": 5.386100386100387e-05,
+      "loss": 0.0153,
+      "step": 279
+    },
+    {
+      "epoch": 1.3526570048309179,
+      "grad_norm": 0.08358469605445862,
+      "learning_rate": 5.405405405405406e-05,
+      "loss": 0.0168,
+      "step": 280
+    },
+    {
+      "epoch": 1.357487922705314,
+      "grad_norm": 0.09522559493780136,
+      "learning_rate": 5.424710424710425e-05,
+      "loss": 0.0233,
+      "step": 281
+    },
+    {
+      "epoch": 1.3623188405797102,
+      "grad_norm": 0.07689143717288971,
+      "learning_rate": 5.4440154440154445e-05,
+      "loss": 0.0181,
+      "step": 282
+    },
+    {
+      "epoch": 1.3671497584541064,
+      "grad_norm": 0.08445203304290771,
+      "learning_rate": 5.463320463320464e-05,
+      "loss": 0.0176,
+      "step": 283
+    },
+    {
+      "epoch": 1.3719806763285023,
+      "grad_norm": 0.13829359412193298,
+      "learning_rate": 5.482625482625483e-05,
+      "loss": 0.0212,
+      "step": 284
+    },
+    {
+      "epoch": 1.3768115942028984,
+      "grad_norm": 0.09589879214763641,
+      "learning_rate": 5.501930501930502e-05,
+      "loss": 0.0195,
+      "step": 285
+    },
+    {
+      "epoch": 1.3816425120772946,
+      "grad_norm": 0.08803611993789673,
+      "learning_rate": 5.5212355212355216e-05,
+      "loss": 0.0194,
+      "step": 286
+    },
+    {
+      "epoch": 1.3864734299516908,
+      "grad_norm": 0.08383037894964218,
+      "learning_rate": 5.540540540540541e-05,
+      "loss": 0.0198,
+      "step": 287
+    },
+    {
+      "epoch": 1.391304347826087,
+      "grad_norm": 0.1102117970585823,
+      "learning_rate": 5.55984555984556e-05,
+      "loss": 0.0246,
+      "step": 288
+    },
+    {
+      "epoch": 1.396135265700483,
+      "grad_norm": 0.09469425678253174,
+      "learning_rate": 5.5791505791505794e-05,
+      "loss": 0.0266,
+      "step": 289
+    },
+    {
+      "epoch": 1.4009661835748792,
+      "grad_norm": 0.10489413887262344,
+      "learning_rate": 5.5984555984555986e-05,
+      "loss": 0.0319,
+      "step": 290
+    },
+    {
+      "epoch": 1.4057971014492754,
+      "grad_norm": 0.0961574912071228,
+      "learning_rate": 5.617760617760618e-05,
+      "loss": 0.0171,
+      "step": 291
+    },
+    {
+      "epoch": 1.4106280193236715,
+      "grad_norm": 0.09252733737230301,
+      "learning_rate": 5.637065637065637e-05,
+      "loss": 0.0222,
+      "step": 292
+    },
+    {
+      "epoch": 1.4154589371980677,
+      "grad_norm": 0.08882331103086472,
+      "learning_rate": 5.6563706563706564e-05,
+      "loss": 0.0233,
+      "step": 293
+    },
+    {
+      "epoch": 1.4202898550724639,
+      "grad_norm": 0.0826525092124939,
+      "learning_rate": 5.6756756756756757e-05,
+      "loss": 0.0184,
+      "step": 294
+    },
+    {
+      "epoch": 1.42512077294686,
+      "grad_norm": 0.06665649265050888,
+      "learning_rate": 5.694980694980695e-05,
+      "loss": 0.0154,
+      "step": 295
+    },
+    {
+      "epoch": 1.4299516908212562,
+      "grad_norm": 0.11430980265140533,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 0.0182,
+      "step": 296
+    },
+    {
+      "epoch": 1.434782608695652,
+      "grad_norm": 0.10717899352312088,
+      "learning_rate": 5.7335907335907334e-05,
+      "loss": 0.0189,
+      "step": 297
+    },
+    {
+      "epoch": 1.4396135265700483,
+      "grad_norm": 0.11982893198728561,
+      "learning_rate": 5.752895752895753e-05,
+      "loss": 0.0186,
+      "step": 298
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.10576062649488449,
+      "learning_rate": 5.772200772200772e-05,
+      "loss": 0.019,
+      "step": 299
+    },
+    {
+      "epoch": 1.4492753623188406,
+      "grad_norm": 0.12444712966680527,
+      "learning_rate": 5.791505791505791e-05,
+      "loss": 0.0228,
+      "step": 300
+    },
+    {
+      "epoch": 1.4492753623188406,
+      "eval_loss": 0.014667052775621414,
+      "eval_runtime": 20.3168,
+      "eval_samples_per_second": 4.922,
+      "eval_steps_per_second": 0.148,
+      "step": 300
+    },
+    {
+      "epoch": 1.4541062801932367,
+      "grad_norm": 0.08958641439676285,
+      "learning_rate": 5.8108108108108105e-05,
+      "loss": 0.0201,
+      "step": 301
+    },
+    {
+      "epoch": 1.458937198067633,
+      "grad_norm": 0.10375913977622986,
+      "learning_rate": 5.83011583011583e-05,
+      "loss": 0.0262,
+      "step": 302
+    },
+    {
+      "epoch": 1.463768115942029,
+      "grad_norm": 0.09301627427339554,
+      "learning_rate": 5.8494208494208503e-05,
+      "loss": 0.0208,
+      "step": 303
+    },
+    {
+      "epoch": 1.4685990338164252,
+      "grad_norm": 0.08783616125583649,
+      "learning_rate": 5.8687258687258696e-05,
+      "loss": 0.0248,
+      "step": 304
+    },
+    {
+      "epoch": 1.4734299516908211,
+      "grad_norm": 0.07414194941520691,
+      "learning_rate": 5.888030888030889e-05,
+      "loss": 0.019,
+      "step": 305
+    },
+    {
+      "epoch": 1.4782608695652173,
+      "grad_norm": 0.07132302969694138,
+      "learning_rate": 5.907335907335908e-05,
+      "loss": 0.0173,
+      "step": 306
+    },
+    {
+      "epoch": 1.4830917874396135,
+      "grad_norm": 0.09330809116363525,
+      "learning_rate": 5.9266409266409274e-05,
+      "loss": 0.0228,
+      "step": 307
+    },
+    {
+      "epoch": 1.4879227053140096,
+      "grad_norm": 0.06476512551307678,
+      "learning_rate": 5.9459459459459466e-05,
+      "loss": 0.0178,
+      "step": 308
+    },
+    {
+      "epoch": 1.4927536231884058,
+      "grad_norm": 0.08418332040309906,
+      "learning_rate": 5.965250965250966e-05,
+      "loss": 0.0176,
+      "step": 309
+    },
+    {
+      "epoch": 1.497584541062802,
+      "grad_norm": 0.0991380512714386,
+      "learning_rate": 5.984555984555985e-05,
+      "loss": 0.0207,
+      "step": 310
+    },
+    {
+      "epoch": 1.502415458937198,
+      "grad_norm": 0.0704076737165451,
+      "learning_rate": 6.0038610038610044e-05,
+      "loss": 0.0159,
+      "step": 311
+    },
+    {
+      "epoch": 1.5072463768115942,
+      "grad_norm": 0.12095512449741364,
+      "learning_rate": 6.023166023166024e-05,
+      "loss": 0.0208,
+      "step": 312
+    },
+    {
+      "epoch": 1.5120772946859904,
+      "grad_norm": 0.0958373174071312,
+      "learning_rate": 6.042471042471043e-05,
+      "loss": 0.0165,
+      "step": 313
+    },
+    {
+      "epoch": 1.5169082125603865,
+      "grad_norm": 0.07881686091423035,
+      "learning_rate": 6.061776061776062e-05,
+      "loss": 0.018,
+      "step": 314
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 0.10672390460968018,
+      "learning_rate": 6.0810810810810814e-05,
+      "loss": 0.0241,
+      "step": 315
+    },
+    {
+      "epoch": 1.5265700483091789,
+      "grad_norm": 0.09105593711137772,
+      "learning_rate": 6.100386100386101e-05,
+      "loss": 0.0173,
+      "step": 316
+    },
+    {
+      "epoch": 1.531400966183575,
+      "grad_norm": 0.07651592046022415,
+      "learning_rate": 6.11969111969112e-05,
+      "loss": 0.0183,
+      "step": 317
+    },
+    {
+      "epoch": 1.5362318840579712,
+      "grad_norm": 0.12212449312210083,
+      "learning_rate": 6.138996138996139e-05,
+      "loss": 0.0206,
+      "step": 318
+    },
+    {
+      "epoch": 1.541062801932367,
+      "grad_norm": 0.08930546045303345,
+      "learning_rate": 6.158301158301159e-05,
+      "loss": 0.0195,
+      "step": 319
+    },
+    {
+      "epoch": 1.5458937198067633,
+      "grad_norm": 0.18044796586036682,
+      "learning_rate": 6.177606177606178e-05,
+      "loss": 0.0227,
+      "step": 320
+    },
+    {
+      "epoch": 1.5507246376811594,
+      "grad_norm": 0.13932865858078003,
+      "learning_rate": 6.196911196911198e-05,
+      "loss": 0.0458,
+      "step": 321
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 0.13868370652198792,
+      "learning_rate": 6.216216216216216e-05,
+      "loss": 0.0163,
+      "step": 322
+    },
+    {
+      "epoch": 1.5603864734299517,
+      "grad_norm": 0.09221283346414566,
+      "learning_rate": 6.235521235521236e-05,
+      "loss": 0.0237,
+      "step": 323
+    },
+    {
+      "epoch": 1.5652173913043477,
+      "grad_norm": 0.12691855430603027,
+      "learning_rate": 6.254826254826255e-05,
+      "loss": 0.0205,
+      "step": 324
+    },
+    {
+      "epoch": 1.5700483091787438,
+      "grad_norm": 0.0665147602558136,
+      "learning_rate": 6.274131274131275e-05,
+      "loss": 0.0167,
+      "step": 325
+    },
+    {
+      "epoch": 1.5700483091787438,
+      "eval_loss": 0.01550750620663166,
+      "eval_runtime": 20.3076,
+      "eval_samples_per_second": 4.924,
+      "eval_steps_per_second": 0.148,
+      "step": 325
+    },
+    {
+      "epoch": 1.57487922705314,
+      "grad_norm": 0.08152501285076141,
+      "learning_rate": 6.293436293436293e-05,
+      "loss": 0.0188,
+      "step": 326
+    },
+    {
+      "epoch": 1.5797101449275361,
+      "grad_norm": 0.08206786960363388,
+      "learning_rate": 6.312741312741313e-05,
+      "loss": 0.0217,
+      "step": 327
+    },
+    {
+      "epoch": 1.5845410628019323,
+      "grad_norm": 0.06393270939588547,
+      "learning_rate": 6.332046332046332e-05,
+      "loss": 0.0257,
+      "step": 328
+    },
+    {
+      "epoch": 1.5893719806763285,
+      "grad_norm": 0.08677301555871964,
+      "learning_rate": 6.351351351351352e-05,
+      "loss": 0.0239,
+      "step": 329
+    },
+    {
+      "epoch": 1.5942028985507246,
+      "grad_norm": 0.06595584750175476,
+      "learning_rate": 6.37065637065637e-05,
+      "loss": 0.0141,
+      "step": 330
+    },
+    {
+      "epoch": 1.5990338164251208,
+      "grad_norm": 0.07364228367805481,
+      "learning_rate": 6.38996138996139e-05,
+      "loss": 0.0225,
+      "step": 331
+    },
+    {
+      "epoch": 1.603864734299517,
+      "grad_norm": 0.0809311717748642,
+      "learning_rate": 6.40926640926641e-05,
+      "loss": 0.0165,
+      "step": 332
+    },
+    {
+      "epoch": 1.608695652173913,
+      "grad_norm": 0.08457648009061813,
+      "learning_rate": 6.428571428571429e-05,
+      "loss": 0.0233,
+      "step": 333
+    },
+    {
+      "epoch": 1.6135265700483092,
+      "grad_norm": 0.10034564137458801,
+      "learning_rate": 6.447876447876449e-05,
+      "loss": 0.0208,
+      "step": 334
+    },
+    {
+      "epoch": 1.6183574879227054,
+      "grad_norm": 0.07684129476547241,
+      "learning_rate": 6.467181467181467e-05,
+      "loss": 0.0189,
+      "step": 335
+    },
+    {
+      "epoch": 1.6231884057971016,
+      "grad_norm": 0.12758824229240417,
+      "learning_rate": 6.486486486486487e-05,
+      "loss": 0.0361,
+      "step": 336
+    },
+    {
+      "epoch": 1.6280193236714977,
+      "grad_norm": 0.07505548000335693,
+      "learning_rate": 6.505791505791506e-05,
+      "loss": 0.0179,
+      "step": 337
+    },
+    {
+      "epoch": 1.6328502415458939,
+      "grad_norm": 0.08614417910575867,
+      "learning_rate": 6.525096525096526e-05,
+      "loss": 0.0228,
+      "step": 338
+    },
+    {
+      "epoch": 1.6376811594202898,
+      "grad_norm": 0.09482710063457489,
+      "learning_rate": 6.544401544401544e-05,
+      "loss": 0.0233,
+      "step": 339
+    },
+    {
+      "epoch": 1.642512077294686,
+      "grad_norm": 0.09455744177103043,
+      "learning_rate": 6.563706563706564e-05,
+      "loss": 0.023,
+      "step": 340
+    },
+    {
+      "epoch": 1.6473429951690821,
+      "grad_norm": 0.07523331791162491,
+      "learning_rate": 6.583011583011583e-05,
+      "loss": 0.0184,
+      "step": 341
+    },
+    {
+      "epoch": 1.6521739130434783,
+      "grad_norm": 0.08421140909194946,
+      "learning_rate": 6.602316602316603e-05,
+      "loss": 0.0248,
+      "step": 342
+    },
+    {
+      "epoch": 1.6570048309178744,
+      "grad_norm": 0.07453511655330658,
+      "learning_rate": 6.621621621621621e-05,
+      "loss": 0.0205,
+      "step": 343
+    },
+    {
+      "epoch": 1.6618357487922706,
+      "grad_norm": 0.08574186265468597,
+      "learning_rate": 6.640926640926641e-05,
+      "loss": 0.0163,
+      "step": 344
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.07299024611711502,
+      "learning_rate": 6.66023166023166e-05,
+      "loss": 0.0253,
+      "step": 345
+    },
+    {
+      "epoch": 1.6714975845410627,
+      "grad_norm": 0.07789258658885956,
+      "learning_rate": 6.67953667953668e-05,
+      "loss": 0.0102,
+      "step": 346
+    },
+    {
+      "epoch": 1.6763285024154588,
+      "grad_norm": 0.10461621731519699,
+      "learning_rate": 6.6988416988417e-05,
+      "loss": 0.0209,
+      "step": 347
+    },
+    {
+      "epoch": 1.681159420289855,
+      "grad_norm": 0.09494093060493469,
+      "learning_rate": 6.718146718146718e-05,
+      "loss": 0.0232,
+      "step": 348
+    },
+    {
+      "epoch": 1.6859903381642511,
+      "grad_norm": 0.06846007704734802,
+      "learning_rate": 6.737451737451738e-05,
+      "loss": 0.0124,
+      "step": 349
+    },
+    {
+      "epoch": 1.6908212560386473,
+      "grad_norm": 0.1044989824295044,
+      "learning_rate": 6.756756756756757e-05,
+      "loss": 0.0238,
+      "step": 350
+    },
+    {
+      "epoch": 1.6908212560386473,
+      "eval_loss": 0.012536454014480114,
+      "eval_runtime": 20.2976,
+      "eval_samples_per_second": 4.927,
+      "eval_steps_per_second": 0.148,
+      "step": 350
+    },
+    {
+      "epoch": 1.6956521739130435,
+      "grad_norm": 0.08567062020301819,
+      "learning_rate": 6.776061776061777e-05,
+      "loss": 0.0143,
+      "step": 351
+    },
+    {
+      "epoch": 1.7004830917874396,
+      "grad_norm": 0.08636616915464401,
+      "learning_rate": 6.795366795366795e-05,
+      "loss": 0.0185,
+      "step": 352
+    },
+    {
+      "epoch": 1.7053140096618358,
+      "grad_norm": 0.1519186943769455,
+      "learning_rate": 6.814671814671815e-05,
+      "loss": 0.0374,
+      "step": 353
+    },
+    {
+      "epoch": 1.710144927536232,
+      "grad_norm": 0.11869102716445923,
+      "learning_rate": 6.833976833976834e-05,
+      "loss": 0.0202,
+      "step": 354
+    },
+    {
+      "epoch": 1.714975845410628,
+      "grad_norm": 0.07147668302059174,
+      "learning_rate": 6.853281853281854e-05,
+      "loss": 0.0259,
+      "step": 355
+    },
+    {
+      "epoch": 1.7198067632850242,
+      "grad_norm": 0.08442330360412598,
+      "learning_rate": 6.872586872586872e-05,
+      "loss": 0.0155,
+      "step": 356
+    },
+    {
+      "epoch": 1.7246376811594204,
+      "grad_norm": 0.10003894567489624,
+      "learning_rate": 6.891891891891892e-05,
+      "loss": 0.0219,
+      "step": 357
+    },
+    {
+      "epoch": 1.7294685990338166,
+      "grad_norm": 0.08303581923246384,
+      "learning_rate": 6.911196911196911e-05,
+      "loss": 0.0137,
+      "step": 358
+    },
+    {
+      "epoch": 1.7342995169082127,
+      "grad_norm": 0.06369996815919876,
+      "learning_rate": 6.930501930501931e-05,
+      "loss": 0.0145,
+      "step": 359
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 0.09714517742395401,
+      "learning_rate": 6.949806949806951e-05,
+      "loss": 0.0179,
+      "step": 360
+    },
+    {
+      "epoch": 1.7439613526570048,
+      "grad_norm": 0.10487079620361328,
+      "learning_rate": 6.96911196911197e-05,
+      "loss": 0.023,
+      "step": 361
+    },
+    {
+      "epoch": 1.748792270531401,
+      "grad_norm": 0.1678726226091385,
+      "learning_rate": 6.988416988416989e-05,
+      "loss": 0.0181,
+      "step": 362
+    },
+    {
+      "epoch": 1.7536231884057971,
+      "grad_norm": 0.13542933762073517,
+      "learning_rate": 7.007722007722008e-05,
+      "loss": 0.0259,
+      "step": 363
+    },
+    {
+      "epoch": 1.7584541062801933,
+      "grad_norm": 0.10919655114412308,
+      "learning_rate": 7.027027027027028e-05,
+      "loss": 0.0206,
+      "step": 364
+    },
+    {
+      "epoch": 1.7632850241545892,
+      "grad_norm": 0.08248183876276016,
+      "learning_rate": 7.046332046332046e-05,
+      "loss": 0.0131,
+      "step": 365
+    },
+    {
+      "epoch": 1.7681159420289854,
+      "grad_norm": 0.10803995281457901,
+      "learning_rate": 7.065637065637066e-05,
+      "loss": 0.0179,
+      "step": 366
+    },
+    {
+      "epoch": 1.7729468599033815,
+      "grad_norm": 0.12460337579250336,
+      "learning_rate": 7.084942084942085e-05,
+      "loss": 0.0186,
+      "step": 367
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.08366953581571579,
+      "learning_rate": 7.104247104247105e-05,
+      "loss": 0.0161,
+      "step": 368
+    },
+    {
+      "epoch": 1.7826086956521738,
+      "grad_norm": 0.12464456260204315,
+      "learning_rate": 7.123552123552123e-05,
+      "loss": 0.0263,
+      "step": 369
+    },
+    {
+      "epoch": 1.78743961352657,
+      "grad_norm": 0.06674820184707642,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 0.0124,
+      "step": 370
+    },
+    {
+      "epoch": 1.7922705314009661,
+      "grad_norm": 0.13090987503528595,
+      "learning_rate": 7.162162162162162e-05,
+      "loss": 0.0239,
+      "step": 371
+    },
+    {
+      "epoch": 1.7971014492753623,
+      "grad_norm": 0.08615607768297195,
+      "learning_rate": 7.181467181467182e-05,
+      "loss": 0.0192,
+      "step": 372
+    },
+    {
+      "epoch": 1.8019323671497585,
+      "grad_norm": 0.07857555896043777,
+      "learning_rate": 7.2007722007722e-05,
+      "loss": 0.0181,
+      "step": 373
+    },
+    {
+      "epoch": 1.8067632850241546,
+      "grad_norm": 0.08106198161840439,
+      "learning_rate": 7.22007722007722e-05,
+      "loss": 0.016,
+      "step": 374
+    },
+    {
+      "epoch": 1.8115942028985508,
+      "grad_norm": 0.17997094988822937,
+      "learning_rate": 7.23938223938224e-05,
+      "loss": 0.0191,
+      "step": 375
+    },
+    {
+      "epoch": 1.8115942028985508,
+      "eval_loss": 0.013815541751682758,
+      "eval_runtime": 20.2875,
+      "eval_samples_per_second": 4.929,
+      "eval_steps_per_second": 0.148,
+      "step": 375
+    },
+    {
+      "epoch": 1.816425120772947,
+      "grad_norm": 0.0769357979297638,
+      "learning_rate": 7.258687258687259e-05,
+      "loss": 0.0268,
+      "step": 376
+    },
+    {
+      "epoch": 1.821256038647343,
+      "grad_norm": 0.07703779637813568,
+      "learning_rate": 7.277992277992279e-05,
+      "loss": 0.0208,
+      "step": 377
+    },
+    {
+      "epoch": 1.8260869565217392,
+      "grad_norm": 0.09476115554571152,
+      "learning_rate": 7.297297297297297e-05,
+      "loss": 0.0247,
+      "step": 378
+    },
+    {
+      "epoch": 1.8309178743961354,
+      "grad_norm": 0.06667820364236832,
+      "learning_rate": 7.316602316602317e-05,
+      "loss": 0.0166,
+      "step": 379
+    },
+    {
+      "epoch": 1.8357487922705316,
+      "grad_norm": 0.06834971159696579,
+      "learning_rate": 7.335907335907336e-05,
+      "loss": 0.0146,
+      "step": 380
+    },
+    {
+      "epoch": 1.8405797101449275,
+      "grad_norm": 0.0759090781211853,
+      "learning_rate": 7.355212355212356e-05,
+      "loss": 0.0191,
+      "step": 381
+    },
+    {
+      "epoch": 1.8454106280193237,
+      "grad_norm": 0.0922810360789299,
+      "learning_rate": 7.374517374517374e-05,
+      "loss": 0.0175,
+      "step": 382
+    },
+    {
+      "epoch": 1.8502415458937198,
+      "grad_norm": 0.08224644511938095,
+      "learning_rate": 7.393822393822394e-05,
+      "loss": 0.0184,
+      "step": 383
+    },
+    {
+      "epoch": 1.855072463768116,
+      "grad_norm": 0.07002474367618561,
+      "learning_rate": 7.413127413127413e-05,
+      "loss": 0.0148,
+      "step": 384
+    },
+    {
+      "epoch": 1.8599033816425121,
+      "grad_norm": 0.06282170861959457,
+      "learning_rate": 7.432432432432433e-05,
+      "loss": 0.0146,
+      "step": 385
+    },
+    {
+      "epoch": 1.864734299516908,
+      "grad_norm": 0.0824386402964592,
+      "learning_rate": 7.451737451737452e-05,
+      "loss": 0.0177,
+      "step": 386
+    },
+    {
+      "epoch": 1.8695652173913042,
+      "grad_norm": 0.10285218060016632,
+      "learning_rate": 7.471042471042471e-05,
+      "loss": 0.0266,
+      "step": 387
+    },
+    {
+      "epoch": 1.8743961352657004,
+      "grad_norm": 0.06601214408874512,
+      "learning_rate": 7.49034749034749e-05,
+      "loss": 0.0143,
+      "step": 388
+    },
+    {
+      "epoch": 1.8792270531400965,
+      "grad_norm": 0.08264856040477753,
+      "learning_rate": 7.50965250965251e-05,
+      "loss": 0.0211,
+      "step": 389
+    },
+    {
+      "epoch": 1.8840579710144927,
+      "grad_norm": 0.06638066470623016,
+      "learning_rate": 7.52895752895753e-05,
+      "loss": 0.0182,
+      "step": 390
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.08071152865886688,
+      "learning_rate": 7.548262548262549e-05,
+      "loss": 0.02,
+      "step": 391
+    },
+    {
+      "epoch": 1.893719806763285,
+      "grad_norm": 0.0696544498205185,
+      "learning_rate": 7.567567567567568e-05,
+      "loss": 0.0193,
+      "step": 392
+    },
+    {
+      "epoch": 1.8985507246376812,
+      "grad_norm": 0.0809270590543747,
+      "learning_rate": 7.586872586872587e-05,
+      "loss": 0.0192,
+      "step": 393
+    },
+    {
+      "epoch": 1.9033816425120773,
+      "grad_norm": 0.06880292296409607,
+      "learning_rate": 7.606177606177607e-05,
+      "loss": 0.0171,
+      "step": 394
+    },
+    {
+      "epoch": 1.9082125603864735,
+      "grad_norm": 0.10304336249828339,
+      "learning_rate": 7.625482625482626e-05,
+      "loss": 0.0154,
+      "step": 395
+    },
+    {
+      "epoch": 1.9130434782608696,
+      "grad_norm": 0.07314369082450867,
+      "learning_rate": 7.644787644787645e-05,
+      "loss": 0.0165,
+      "step": 396
+    },
+    {
+      "epoch": 1.9178743961352658,
+      "grad_norm": 0.08954203128814697,
+      "learning_rate": 7.664092664092664e-05,
+      "loss": 0.017,
+      "step": 397
+    },
+    {
+      "epoch": 1.922705314009662,
+      "grad_norm": 0.11527922749519348,
+      "learning_rate": 7.683397683397684e-05,
+      "loss": 0.0175,
+      "step": 398
+    },
+    {
+      "epoch": 1.927536231884058,
+      "grad_norm": 0.14367862045764923,
+      "learning_rate": 7.702702702702703e-05,
+      "loss": 0.0201,
+      "step": 399
+    },
+    {
+      "epoch": 1.9323671497584543,
+      "grad_norm": 0.09122224897146225,
+      "learning_rate": 7.722007722007723e-05,
+      "loss": 0.0273,
+      "step": 400
+    },
+    {
+      "epoch": 1.9323671497584543,
+      "eval_loss": 0.011990789324045181,
+      "eval_runtime": 20.3107,
+      "eval_samples_per_second": 4.924,
+      "eval_steps_per_second": 0.148,
+      "step": 400
+    },
+    {
+      "epoch": 1.9371980676328504,
+      "grad_norm": 0.07356519997119904,
+      "learning_rate": 7.741312741312741e-05,
+      "loss": 0.024,
+      "step": 401
+    },
+    {
+      "epoch": 1.9420289855072463,
+      "grad_norm": 0.11204323172569275,
+      "learning_rate": 7.760617760617761e-05,
+      "loss": 0.0271,
+      "step": 402
+    },
+    {
+      "epoch": 1.9468599033816425,
+      "grad_norm": 0.056930117309093475,
+      "learning_rate": 7.779922779922781e-05,
+      "loss": 0.0177,
+      "step": 403
+    },
+    {
+      "epoch": 1.9516908212560387,
+      "grad_norm": 0.07158982753753662,
+      "learning_rate": 7.7992277992278e-05,
+      "loss": 0.0191,
+      "step": 404
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 0.09313391894102097,
+      "learning_rate": 7.81853281853282e-05,
+      "loss": 0.0183,
+      "step": 405
+    },
+    {
+      "epoch": 1.961352657004831,
+      "grad_norm": 0.06606443971395493,
+      "learning_rate": 7.837837837837838e-05,
+      "loss": 0.0164,
+      "step": 406
+    },
+    {
+      "epoch": 1.966183574879227,
+      "grad_norm": 0.07535674422979355,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 0.0182,
+      "step": 407
+    },
+    {
+      "epoch": 1.971014492753623,
+      "grad_norm": 0.061097193509340286,
+      "learning_rate": 7.876447876447877e-05,
+      "loss": 0.015,
+      "step": 408
+    },
+    {
+      "epoch": 1.9758454106280192,
+      "grad_norm": 0.05617830529808998,
+      "learning_rate": 7.895752895752897e-05,
+      "loss": 0.0144,
+      "step": 409
+    },
+    {
+      "epoch": 1.9806763285024154,
+      "grad_norm": 0.12362464517354965,
+      "learning_rate": 7.915057915057915e-05,
+      "loss": 0.0154,
+      "step": 410
+    },
+    {
+      "epoch": 1.9855072463768115,
+      "grad_norm": 0.08519264310598373,
+      "learning_rate": 7.934362934362935e-05,
+      "loss": 0.0227,
+      "step": 411
+    },
+    {
+      "epoch": 1.9903381642512077,
+      "grad_norm": 0.0802370011806488,
+      "learning_rate": 7.953667953667954e-05,
+      "loss": 0.0185,
+      "step": 412
+    },
+    {
+      "epoch": 1.9951690821256038,
+      "grad_norm": 0.07258555293083191,
+      "learning_rate": 7.972972972972974e-05,
+      "loss": 0.024,
+      "step": 413
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.14007550477981567,
+      "learning_rate": 7.992277992277992e-05,
+      "loss": 0.0156,
+      "step": 414
+    },
+    {
+      "epoch": 2.004830917874396,
+      "grad_norm": 0.049884360283613205,
+      "learning_rate": 8.011583011583012e-05,
+      "loss": 0.0109,
+      "step": 415
+    },
+    {
+      "epoch": 2.0096618357487923,
+      "grad_norm": 0.06829845905303955,
+      "learning_rate": 8.03088803088803e-05,
+      "loss": 0.013,
+      "step": 416
+    },
+    {
+      "epoch": 2.0144927536231885,
+      "grad_norm": 0.08212131261825562,
+      "learning_rate": 8.05019305019305e-05,
+      "loss": 0.0209,
+      "step": 417
+    },
+    {
+      "epoch": 2.0193236714975846,
+      "grad_norm": 0.07331026345491409,
+      "learning_rate": 8.06949806949807e-05,
+      "loss": 0.0104,
+      "step": 418
+    },
+    {
+      "epoch": 2.024154589371981,
+      "grad_norm": 0.06080864369869232,
+      "learning_rate": 8.088803088803089e-05,
+      "loss": 0.0142,
+      "step": 419
+    },
+    {
+      "epoch": 2.028985507246377,
+      "grad_norm": 0.07056247442960739,
+      "learning_rate": 8.108108108108109e-05,
+      "loss": 0.0155,
+      "step": 420
+    },
+    {
+      "epoch": 2.033816425120773,
+      "grad_norm": 0.0810440331697464,
+      "learning_rate": 8.127413127413128e-05,
+      "loss": 0.0148,
+      "step": 421
+    },
+    {
+      "epoch": 2.0386473429951693,
+      "grad_norm": 0.08997571468353271,
+      "learning_rate": 8.146718146718148e-05,
+      "loss": 0.0126,
+      "step": 422
+    },
+    {
+      "epoch": 2.0434782608695654,
+      "grad_norm": 0.07606981694698334,
+      "learning_rate": 8.166023166023166e-05,
+      "loss": 0.0153,
+      "step": 423
+    },
+    {
+      "epoch": 2.0483091787439616,
+      "grad_norm": 0.06378842890262604,
+      "learning_rate": 8.185328185328186e-05,
+      "loss": 0.0115,
+      "step": 424
+    },
+    {
+      "epoch": 2.0531400966183573,
+      "grad_norm": 0.06914151459932327,
+      "learning_rate": 8.204633204633205e-05,
+      "loss": 0.0194,
+      "step": 425
+    },
+    {
+      "epoch": 2.0531400966183573,
+      "eval_loss": 0.012465307489037514,
+      "eval_runtime": 20.2973,
+      "eval_samples_per_second": 4.927,
+      "eval_steps_per_second": 0.148,
+      "step": 425
+    },
+    {
+      "epoch": 2.0579710144927534,
+      "grad_norm": 0.07839097827672958,
+      "learning_rate": 8.223938223938225e-05,
+      "loss": 0.0158,
+      "step": 426
+    },
+    {
+      "epoch": 2.0628019323671496,
+      "grad_norm": 0.06898388266563416,
+      "learning_rate": 8.243243243243243e-05,
+      "loss": 0.016,
+      "step": 427
+    },
+    {
+      "epoch": 2.0676328502415457,
+      "grad_norm": 0.058221787214279175,
+      "learning_rate": 8.262548262548263e-05,
+      "loss": 0.0143,
+      "step": 428
+    },
+    {
+      "epoch": 2.072463768115942,
+      "grad_norm": 0.05164722725749016,
+      "learning_rate": 8.281853281853282e-05,
+      "loss": 0.0113,
+      "step": 429
+    },
+    {
+      "epoch": 2.077294685990338,
+      "grad_norm": 0.08281679451465607,
+      "learning_rate": 8.301158301158302e-05,
+      "loss": 0.0144,
+      "step": 430
+    },
+    {
+      "epoch": 2.082125603864734,
+      "grad_norm": 0.06709595769643784,
+      "learning_rate": 8.32046332046332e-05,
+      "loss": 0.0179,
+      "step": 431
+    },
+    {
+      "epoch": 2.0869565217391304,
+      "grad_norm": 0.07612422853708267,
+      "learning_rate": 8.33976833976834e-05,
+      "loss": 0.0139,
+      "step": 432
+    },
+    {
+      "epoch": 2.0917874396135265,
+      "grad_norm": 0.057307854294776917,
+      "learning_rate": 8.35907335907336e-05,
+      "loss": 0.0099,
+      "step": 433
+    },
+    {
+      "epoch": 2.0966183574879227,
+      "grad_norm": 0.06918705999851227,
+      "learning_rate": 8.378378378378379e-05,
+      "loss": 0.0145,
+      "step": 434
+    },
+    {
+      "epoch": 2.101449275362319,
+      "grad_norm": 0.07992324978113174,
+      "learning_rate": 8.397683397683399e-05,
+      "loss": 0.0133,
+      "step": 435
+    },
+    {
+      "epoch": 2.106280193236715,
+      "grad_norm": 0.05948076397180557,
+      "learning_rate": 8.416988416988417e-05,
+      "loss": 0.0139,
+      "step": 436
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 0.08342638611793518,
+      "learning_rate": 8.436293436293437e-05,
+      "loss": 0.0106,
+      "step": 437
+    },
+    {
+      "epoch": 2.1159420289855073,
+      "grad_norm": 0.10372233390808105,
+      "learning_rate": 8.455598455598456e-05,
+      "loss": 0.0171,
+      "step": 438
+    },
+    {
+      "epoch": 2.1207729468599035,
+      "grad_norm": 0.10722784698009491,
+      "learning_rate": 8.474903474903476e-05,
+      "loss": 0.0153,
+      "step": 439
+    },
+    {
+      "epoch": 2.1256038647342996,
+      "grad_norm": 0.05321025848388672,
+      "learning_rate": 8.494208494208494e-05,
+      "loss": 0.0157,
+      "step": 440
+    },
+    {
+      "epoch": 2.130434782608696,
+      "grad_norm": 0.0834476426243782,
+      "learning_rate": 8.513513513513514e-05,
+      "loss": 0.0158,
+      "step": 441
+    },
+    {
+      "epoch": 2.135265700483092,
+      "grad_norm": 0.04624385014176369,
+      "learning_rate": 8.532818532818533e-05,
+      "loss": 0.0101,
+      "step": 442
+    },
+    {
+      "epoch": 2.140096618357488,
+      "grad_norm": 0.09156984090805054,
+      "learning_rate": 8.552123552123553e-05,
+      "loss": 0.0139,
+      "step": 443
+    },
+    {
+      "epoch": 2.1449275362318843,
+      "grad_norm": 0.11084246635437012,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 0.0172,
+      "step": 444
+    },
+    {
+      "epoch": 2.14975845410628,
+      "grad_norm": 0.05298859253525734,
+      "learning_rate": 8.590733590733591e-05,
+      "loss": 0.0116,
+      "step": 445
+    },
+    {
+      "epoch": 2.154589371980676,
+      "grad_norm": 0.07835599780082703,
+      "learning_rate": 8.61003861003861e-05,
+      "loss": 0.0191,
+      "step": 446
+    },
+    {
+      "epoch": 2.1594202898550723,
+      "grad_norm": 0.09636026620864868,
+      "learning_rate": 8.62934362934363e-05,
+      "loss": 0.0181,
+      "step": 447
+    },
+    {
+      "epoch": 2.1642512077294684,
+      "grad_norm": 0.10710686445236206,
+      "learning_rate": 8.64864864864865e-05,
+      "loss": 0.015,
+      "step": 448
+    },
+    {
+      "epoch": 2.1690821256038646,
+      "grad_norm": 0.06177045777440071,
+      "learning_rate": 8.667953667953668e-05,
+      "loss": 0.0165,
+      "step": 449
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 0.05858482047915459,
+      "learning_rate": 8.687258687258688e-05,
+      "loss": 0.0125,
+      "step": 450
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "eval_loss": 0.012776307761669159,
+      "eval_runtime": 20.2754,
+      "eval_samples_per_second": 4.932,
+      "eval_steps_per_second": 0.148,
+      "step": 450
+    },
+    {
+      "epoch": 2.178743961352657,
+      "grad_norm": 0.07402500510215759,
+      "learning_rate": 8.706563706563707e-05,
+      "loss": 0.0169,
+      "step": 451
+    },
+    {
+      "epoch": 2.183574879227053,
+      "grad_norm": 0.08418232947587967,
+      "learning_rate": 8.725868725868727e-05,
+      "loss": 0.0167,
+      "step": 452
+    },
+    {
+      "epoch": 2.1884057971014492,
+      "grad_norm": 0.05793262645602226,
+      "learning_rate": 8.745173745173745e-05,
+      "loss": 0.0103,
+      "step": 453
+    },
+    {
+      "epoch": 2.1932367149758454,
+      "grad_norm": 0.08266332000494003,
+      "learning_rate": 8.764478764478765e-05,
+      "loss": 0.0164,
+      "step": 454
+    },
+    {
+      "epoch": 2.1980676328502415,
+      "grad_norm": 0.09661264717578888,
+      "learning_rate": 8.783783783783784e-05,
+      "loss": 0.0165,
+      "step": 455
+    },
+    {
+      "epoch": 2.2028985507246377,
+      "grad_norm": 0.06425665318965912,
+      "learning_rate": 8.803088803088804e-05,
+      "loss": 0.0148,
+      "step": 456
+    },
+    {
+      "epoch": 2.207729468599034,
+      "grad_norm": 0.09541848301887512,
+      "learning_rate": 8.822393822393822e-05,
+      "loss": 0.0139,
+      "step": 457
+    },
+    {
+      "epoch": 2.21256038647343,
+      "grad_norm": 0.0549871027469635,
+      "learning_rate": 8.841698841698842e-05,
+      "loss": 0.0145,
+      "step": 458
+    },
+    {
+      "epoch": 2.217391304347826,
+      "grad_norm": 0.0751703754067421,
+      "learning_rate": 8.861003861003861e-05,
+      "loss": 0.012,
+      "step": 459
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.061911825090646744,
+      "learning_rate": 8.880308880308881e-05,
+      "loss": 0.0165,
+      "step": 460
+    },
+    {
+      "epoch": 2.2270531400966185,
+      "grad_norm": 0.07648997008800507,
+      "learning_rate": 8.899613899613901e-05,
+      "loss": 0.0168,
+      "step": 461
+    },
+    {
+      "epoch": 2.2318840579710146,
+      "grad_norm": 0.06206296756863594,
+      "learning_rate": 8.918918918918919e-05,
+      "loss": 0.0135,
+      "step": 462
+    },
+    {
+      "epoch": 2.236714975845411,
+      "grad_norm": 0.06858911365270615,
+      "learning_rate": 8.938223938223939e-05,
+      "loss": 0.015,
+      "step": 463
+    },
+    {
+      "epoch": 2.241545893719807,
+      "grad_norm": 0.0879889577627182,
+      "learning_rate": 8.957528957528958e-05,
+      "loss": 0.0133,
+      "step": 464
+    },
+    {
+      "epoch": 2.246376811594203,
+      "grad_norm": 0.06427818536758423,
+      "learning_rate": 8.976833976833978e-05,
+      "loss": 0.018,
+      "step": 465
+    },
+    {
+      "epoch": 2.2512077294685993,
+      "grad_norm": 0.06355290859937668,
+      "learning_rate": 8.996138996138996e-05,
+      "loss": 0.0112,
+      "step": 466
+    },
+    {
+      "epoch": 2.2560386473429954,
+      "grad_norm": 0.06012749299407005,
+      "learning_rate": 9.015444015444016e-05,
+      "loss": 0.0154,
+      "step": 467
+    },
+    {
+      "epoch": 2.260869565217391,
+      "grad_norm": 0.08245503157377243,
+      "learning_rate": 9.034749034749035e-05,
+      "loss": 0.0168,
+      "step": 468
+    },
+    {
+      "epoch": 2.2657004830917873,
+      "grad_norm": 0.07163789123296738,
+      "learning_rate": 9.054054054054055e-05,
+      "loss": 0.0171,
+      "step": 469
+    },
+    {
+      "epoch": 2.2705314009661834,
+      "grad_norm": 0.11928809434175491,
+      "learning_rate": 9.073359073359073e-05,
+      "loss": 0.0202,
+      "step": 470
+    },
+    {
+      "epoch": 2.2753623188405796,
+      "grad_norm": 0.05260397493839264,
+      "learning_rate": 9.092664092664093e-05,
+      "loss": 0.012,
+      "step": 471
+    },
+    {
+      "epoch": 2.2801932367149758,
+      "grad_norm": 0.12959927320480347,
+      "learning_rate": 9.111969111969112e-05,
+      "loss": 0.0231,
+      "step": 472
+    },
+    {
+      "epoch": 2.285024154589372,
+      "grad_norm": 0.06374755501747131,
+      "learning_rate": 9.131274131274132e-05,
+      "loss": 0.0117,
+      "step": 473
+    },
+    {
+      "epoch": 2.289855072463768,
+      "grad_norm": 0.08595716953277588,
+      "learning_rate": 9.15057915057915e-05,
+      "loss": 0.0158,
+      "step": 474
+    },
+    {
+      "epoch": 2.2946859903381642,
+      "grad_norm": 0.05997462198138237,
+      "learning_rate": 9.16988416988417e-05,
+      "loss": 0.0132,
+      "step": 475
+    },
+    {
+      "epoch": 2.2946859903381642,
+      "eval_loss": 0.011733826249837875,
+      "eval_runtime": 20.2984,
+      "eval_samples_per_second": 4.927,
+      "eval_steps_per_second": 0.148,
+      "step": 475
+    },
+    {
+      "epoch": 2.2995169082125604,
+      "grad_norm": 0.0654623806476593,
+      "learning_rate": 9.18918918918919e-05,
+      "loss": 0.0141,
+      "step": 476
+    },
+    {
+      "epoch": 2.3043478260869565,
+      "grad_norm": 0.09593897312879562,
+      "learning_rate": 9.208494208494209e-05,
+      "loss": 0.0171,
+      "step": 477
+    },
+    {
+      "epoch": 2.3091787439613527,
+      "grad_norm": 0.05213898792862892,
+      "learning_rate": 9.227799227799229e-05,
+      "loss": 0.0125,
+      "step": 478
+    },
+    {
+      "epoch": 2.314009661835749,
+      "grad_norm": 0.06953102350234985,
+      "learning_rate": 9.247104247104247e-05,
+      "loss": 0.016,
+      "step": 479
+    },
+    {
+      "epoch": 2.318840579710145,
+      "grad_norm": 0.07360464334487915,
+      "learning_rate": 9.266409266409267e-05,
+      "loss": 0.0143,
+      "step": 480
+    },
+    {
+      "epoch": 2.323671497584541,
+      "grad_norm": 0.07077494263648987,
+      "learning_rate": 9.285714285714286e-05,
+      "loss": 0.0329,
+      "step": 481
+    },
+    {
+      "epoch": 2.3285024154589373,
+      "grad_norm": 0.08831976354122162,
+      "learning_rate": 9.305019305019306e-05,
+      "loss": 0.0131,
+      "step": 482
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.06258786469697952,
+      "learning_rate": 9.324324324324324e-05,
+      "loss": 0.0178,
+      "step": 483
+    },
+    {
+      "epoch": 2.3381642512077296,
+      "grad_norm": 0.0674007460474968,
+      "learning_rate": 9.343629343629344e-05,
+      "loss": 0.0206,
+      "step": 484
+    },
+    {
+      "epoch": 2.342995169082126,
+      "grad_norm": 0.09977665543556213,
+      "learning_rate": 9.362934362934363e-05,
+      "loss": 0.0118,
+      "step": 485
+    },
+    {
+      "epoch": 2.3478260869565215,
+      "grad_norm": 0.057383470237255096,
+      "learning_rate": 9.382239382239383e-05,
+      "loss": 0.0157,
+      "step": 486
+    },
+    {
+      "epoch": 2.3526570048309177,
+      "grad_norm": 0.07880475372076035,
+      "learning_rate": 9.401544401544401e-05,
+      "loss": 0.0108,
+      "step": 487
+    },
+    {
+      "epoch": 2.357487922705314,
+      "grad_norm": 0.06846434623003006,
+      "learning_rate": 9.420849420849421e-05,
+      "loss": 0.0128,
+      "step": 488
+    },
+    {
+      "epoch": 2.36231884057971,
+      "grad_norm": 0.07583679258823395,
+      "learning_rate": 9.44015444015444e-05,
+      "loss": 0.0166,
+      "step": 489
+    },
+    {
+      "epoch": 2.367149758454106,
+      "grad_norm": 0.0955696627497673,
+      "learning_rate": 9.45945945945946e-05,
+      "loss": 0.0209,
+      "step": 490
+    },
+    {
+      "epoch": 2.3719806763285023,
+      "grad_norm": 0.08694009482860565,
+      "learning_rate": 9.47876447876448e-05,
+      "loss": 0.0164,
+      "step": 491
+    },
+    {
+      "epoch": 2.3768115942028984,
+      "grad_norm": 0.04880060255527496,
+      "learning_rate": 9.498069498069498e-05,
+      "loss": 0.0119,
+      "step": 492
+    },
+    {
+      "epoch": 2.3816425120772946,
+      "grad_norm": 0.07593809068202972,
+      "learning_rate": 9.517374517374518e-05,
+      "loss": 0.0214,
+      "step": 493
+    },
+    {
+      "epoch": 2.3864734299516908,
+      "grad_norm": 0.07582546025514603,
+      "learning_rate": 9.536679536679537e-05,
+      "loss": 0.0147,
+      "step": 494
+    },
+    {
+      "epoch": 2.391304347826087,
+      "grad_norm": 0.06178034842014313,
+      "learning_rate": 9.555984555984557e-05,
+      "loss": 0.0165,
+      "step": 495
+    },
+    {
+      "epoch": 2.396135265700483,
+      "grad_norm": 0.05816735699772835,
+      "learning_rate": 9.575289575289575e-05,
+      "loss": 0.011,
+      "step": 496
+    },
+    {
+      "epoch": 2.4009661835748792,
+      "grad_norm": 0.0542505607008934,
+      "learning_rate": 9.594594594594595e-05,
+      "loss": 0.0164,
+      "step": 497
+    },
+    {
+      "epoch": 2.4057971014492754,
+      "grad_norm": 0.07111211121082306,
+      "learning_rate": 9.613899613899614e-05,
+      "loss": 0.016,
+      "step": 498
+    },
+    {
+      "epoch": 2.4106280193236715,
+      "grad_norm": 0.04318169131875038,
+      "learning_rate": 9.633204633204634e-05,
+      "loss": 0.0152,
+      "step": 499
+    },
+    {
+      "epoch": 2.4154589371980677,
+      "grad_norm": 0.05651891976594925,
+      "learning_rate": 9.652509652509652e-05,
+      "loss": 0.0142,
+      "step": 500
+    },
+    {
+      "epoch": 2.4154589371980677,
+      "eval_loss": 0.009894417598843575,
+      "eval_runtime": 20.2924,
+      "eval_samples_per_second": 4.928,
+      "eval_steps_per_second": 0.148,
+      "step": 500
+    },
+    {
+      "epoch": 2.420289855072464,
+      "grad_norm": 0.04565607011318207,
+      "learning_rate": 9.671814671814672e-05,
+      "loss": 0.0101,
+      "step": 501
+    },
+    {
+      "epoch": 2.42512077294686,
+      "grad_norm": 0.05563889071345329,
+      "learning_rate": 9.691119691119691e-05,
+      "loss": 0.0181,
+      "step": 502
+    },
+    {
+      "epoch": 2.429951690821256,
+      "grad_norm": 0.07889113575220108,
+      "learning_rate": 9.710424710424711e-05,
+      "loss": 0.0126,
+      "step": 503
+    },
+    {
+      "epoch": 2.4347826086956523,
+      "grad_norm": 0.05971178784966469,
+      "learning_rate": 9.729729729729731e-05,
+      "loss": 0.0104,
+      "step": 504
+    },
+    {
+      "epoch": 2.4396135265700485,
+      "grad_norm": 0.06959031522274017,
+      "learning_rate": 9.74903474903475e-05,
+      "loss": 0.0169,
+      "step": 505
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.059511762112379074,
+      "learning_rate": 9.76833976833977e-05,
+      "loss": 0.0095,
+      "step": 506
+    },
+    {
+      "epoch": 2.449275362318841,
+      "grad_norm": 0.07739004492759705,
+      "learning_rate": 9.787644787644788e-05,
+      "loss": 0.0142,
+      "step": 507
+    },
+    {
+      "epoch": 2.454106280193237,
+      "grad_norm": 0.06242101639509201,
+      "learning_rate": 9.806949806949808e-05,
+      "loss": 0.013,
+      "step": 508
+    },
+    {
+      "epoch": 2.4589371980676327,
+      "grad_norm": 0.07871799916028976,
+      "learning_rate": 9.826254826254826e-05,
+      "loss": 0.0198,
+      "step": 509
+    },
+    {
+      "epoch": 2.463768115942029,
+      "grad_norm": 0.06508443504571915,
+      "learning_rate": 9.845559845559846e-05,
+      "loss": 0.0125,
+      "step": 510
+    },
+    {
+      "epoch": 2.468599033816425,
+      "grad_norm": 0.049432434141635895,
+      "learning_rate": 9.864864864864865e-05,
+      "loss": 0.0105,
+      "step": 511
+    },
+    {
+      "epoch": 2.473429951690821,
+      "grad_norm": 0.08035603910684586,
+      "learning_rate": 9.884169884169885e-05,
+      "loss": 0.0128,
+      "step": 512
+    },
+    {
+      "epoch": 2.4782608695652173,
+      "grad_norm": 0.05681634321808815,
+      "learning_rate": 9.903474903474904e-05,
+      "loss": 0.0147,
+      "step": 513
+    },
+    {
+      "epoch": 2.4830917874396135,
+      "grad_norm": 0.09850388020277023,
+      "learning_rate": 9.922779922779923e-05,
+      "loss": 0.0159,
+      "step": 514
+    },
+    {
+      "epoch": 2.4879227053140096,
+      "grad_norm": 0.06364881247282028,
+      "learning_rate": 9.942084942084942e-05,
+      "loss": 0.0128,
+      "step": 515
+    },
+    {
+      "epoch": 2.4927536231884058,
+      "grad_norm": 0.08354095369577408,
+      "learning_rate": 9.961389961389962e-05,
+      "loss": 0.0123,
+      "step": 516
+    },
+    {
+      "epoch": 2.497584541062802,
+      "grad_norm": 0.047128040343523026,
+      "learning_rate": 9.98069498069498e-05,
+      "loss": 0.0104,
+      "step": 517
+    },
+    {
+      "epoch": 2.502415458937198,
+      "grad_norm": 0.07560840994119644,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 518
+    },
+    {
+      "epoch": 2.5072463768115942,
+      "grad_norm": 0.08767189085483551,
+      "learning_rate": 9.9999988623013e-05,
+      "loss": 0.02,
+      "step": 519
+    },
+    {
+      "epoch": 2.5120772946859904,
+      "grad_norm": 0.06940112262964249,
+      "learning_rate": 9.999995449205719e-05,
+      "loss": 0.0128,
+      "step": 520
+    },
+    {
+      "epoch": 2.5169082125603865,
+      "grad_norm": 0.08003146946430206,
+      "learning_rate": 9.999989760714809e-05,
+      "loss": 0.0156,
+      "step": 521
+    },
+    {
+      "epoch": 2.5217391304347827,
+      "grad_norm": 0.06202618032693863,
+      "learning_rate": 9.999981796831159e-05,
+      "loss": 0.0152,
+      "step": 522
+    },
+    {
+      "epoch": 2.526570048309179,
+      "grad_norm": 0.05850287154316902,
+      "learning_rate": 9.999971557558395e-05,
+      "loss": 0.0134,
+      "step": 523
+    },
+    {
+      "epoch": 2.531400966183575,
+      "grad_norm": 0.05508386343717575,
+      "learning_rate": 9.999959042901174e-05,
+      "loss": 0.0162,
+      "step": 524
+    },
+    {
+      "epoch": 2.536231884057971,
+      "grad_norm": 0.05612890422344208,
+      "learning_rate": 9.999944252865192e-05,
+      "loss": 0.0119,
+      "step": 525
+    },
+    {
+      "epoch": 2.536231884057971,
+      "eval_loss": 0.010469529777765274,
+      "eval_runtime": 20.2698,
+      "eval_samples_per_second": 4.933,
+      "eval_steps_per_second": 0.148,
+      "step": 525
+    },
+    {
+      "epoch": 2.541062801932367,
+      "grad_norm": 0.06656938046216965,
+      "learning_rate": 9.999927187457181e-05,
+      "loss": 0.0124,
+      "step": 526
+    },
+    {
+      "epoch": 2.545893719806763,
+      "grad_norm": 0.05605817586183548,
+      "learning_rate": 9.999907846684906e-05,
+      "loss": 0.0152,
+      "step": 527
+    },
+    {
+      "epoch": 2.550724637681159,
+      "grad_norm": 0.055510830134153366,
+      "learning_rate": 9.999886230557167e-05,
+      "loss": 0.0182,
+      "step": 528
+    },
+    {
+      "epoch": 2.5555555555555554,
+      "grad_norm": 0.047456130385398865,
+      "learning_rate": 9.999862339083804e-05,
+      "loss": 0.012,
+      "step": 529
+    },
+    {
+      "epoch": 2.5603864734299515,
+      "grad_norm": 0.08240535855293274,
+      "learning_rate": 9.999836172275688e-05,
+      "loss": 0.015,
+      "step": 530
+    },
+    {
+      "epoch": 2.5652173913043477,
+      "grad_norm": 0.06938566267490387,
+      "learning_rate": 9.999807730144728e-05,
+      "loss": 0.0131,
+      "step": 531
+    },
+    {
+      "epoch": 2.570048309178744,
+      "grad_norm": 0.054002463817596436,
+      "learning_rate": 9.999777012703866e-05,
+      "loss": 0.0117,
+      "step": 532
+    },
+    {
+      "epoch": 2.57487922705314,
+      "grad_norm": 0.06356227397918701,
+      "learning_rate": 9.999744019967081e-05,
+      "loss": 0.01,
+      "step": 533
+    },
+    {
+      "epoch": 2.579710144927536,
+      "grad_norm": 0.0544712208211422,
+      "learning_rate": 9.999708751949389e-05,
+      "loss": 0.0103,
+      "step": 534
+    },
+    {
+      "epoch": 2.5845410628019323,
+      "grad_norm": 0.06751914322376251,
+      "learning_rate": 9.999671208666838e-05,
+      "loss": 0.0142,
+      "step": 535
+    },
+    {
+      "epoch": 2.5893719806763285,
+      "grad_norm": 0.04503876715898514,
+      "learning_rate": 9.999631390136513e-05,
+      "loss": 0.0115,
+      "step": 536
+    },
+    {
+      "epoch": 2.5942028985507246,
+      "grad_norm": 0.07345236837863922,
+      "learning_rate": 9.999589296376537e-05,
+      "loss": 0.0115,
+      "step": 537
+    },
+    {
+      "epoch": 2.5990338164251208,
+      "grad_norm": 0.07252614200115204,
+      "learning_rate": 9.999544927406063e-05,
+      "loss": 0.0146,
+      "step": 538
+    },
+    {
+      "epoch": 2.603864734299517,
+      "grad_norm": 0.07327065616846085,
+      "learning_rate": 9.999498283245284e-05,
+      "loss": 0.0125,
+      "step": 539
+    },
+    {
+      "epoch": 2.608695652173913,
+      "grad_norm": 0.04617929458618164,
+      "learning_rate": 9.999449363915427e-05,
+      "loss": 0.0103,
+      "step": 540
+    },
+    {
+      "epoch": 2.6135265700483092,
+      "grad_norm": 0.05602256953716278,
+      "learning_rate": 9.999398169438754e-05,
+      "loss": 0.0124,
+      "step": 541
+    },
+    {
+      "epoch": 2.6183574879227054,
+      "grad_norm": 0.07086073607206345,
+      "learning_rate": 9.999344699838562e-05,
+      "loss": 0.0138,
+      "step": 542
+    },
+    {
+      "epoch": 2.6231884057971016,
+      "grad_norm": 0.07867629081010818,
+      "learning_rate": 9.999288955139183e-05,
+      "loss": 0.0154,
+      "step": 543
+    },
+    {
+      "epoch": 2.6280193236714977,
+      "grad_norm": 0.06904333829879761,
+      "learning_rate": 9.999230935365989e-05,
+      "loss": 0.0161,
+      "step": 544
+    },
+    {
+      "epoch": 2.632850241545894,
+      "grad_norm": 0.04477986320853233,
+      "learning_rate": 9.999170640545378e-05,
+      "loss": 0.0141,
+      "step": 545
+    },
+    {
+      "epoch": 2.63768115942029,
+      "grad_norm": 0.07836603373289108,
+      "learning_rate": 9.999108070704795e-05,
+      "loss": 0.0111,
+      "step": 546
+    },
+    {
+      "epoch": 2.642512077294686,
+      "grad_norm": 0.07152969390153885,
+      "learning_rate": 9.99904322587271e-05,
+      "loss": 0.0124,
+      "step": 547
+    },
+    {
+      "epoch": 2.6473429951690823,
+      "grad_norm": 0.08956319838762283,
+      "learning_rate": 9.998976106078634e-05,
+      "loss": 0.0189,
+      "step": 548
+    },
+    {
+      "epoch": 2.6521739130434785,
+      "grad_norm": 0.05870571359992027,
+      "learning_rate": 9.99890671135311e-05,
+      "loss": 0.0177,
+      "step": 549
+    },
+    {
+      "epoch": 2.6570048309178746,
+      "grad_norm": 0.05460686236619949,
+      "learning_rate": 9.998835041727723e-05,
+      "loss": 0.0131,
+      "step": 550
+    },
+    {
+      "epoch": 2.6570048309178746,
+      "eval_loss": 0.011807677336037159,
+      "eval_runtime": 20.2717,
+      "eval_samples_per_second": 4.933,
+      "eval_steps_per_second": 0.148,
+      "step": 550
+    },
+    {
+      "epoch": 2.661835748792271,
+      "grad_norm": 0.08161139488220215,
+      "learning_rate": 9.998761097235083e-05,
+      "loss": 0.0124,
+      "step": 551
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.05696139857172966,
+      "learning_rate": 9.998684877908844e-05,
+      "loss": 0.0149,
+      "step": 552
+    },
+    {
+      "epoch": 2.6714975845410627,
+      "grad_norm": 0.10040969401597977,
+      "learning_rate": 9.998606383783691e-05,
+      "loss": 0.017,
+      "step": 553
+    },
+    {
+      "epoch": 2.676328502415459,
+      "grad_norm": 0.05457184091210365,
+      "learning_rate": 9.998525614895343e-05,
+      "loss": 0.0142,
+      "step": 554
+    },
+    {
+      "epoch": 2.681159420289855,
+      "grad_norm": 0.08906134963035583,
+      "learning_rate": 9.99844257128056e-05,
+      "loss": 0.0198,
+      "step": 555
+    },
+    {
+      "epoch": 2.685990338164251,
+      "grad_norm": 0.06964429467916489,
+      "learning_rate": 9.99835725297713e-05,
+      "loss": 0.0118,
+      "step": 556
+    },
+    {
+      "epoch": 2.6908212560386473,
+      "grad_norm": 0.03187815472483635,
+      "learning_rate": 9.998269660023882e-05,
+      "loss": 0.0067,
+      "step": 557
+    },
+    {
+      "epoch": 2.6956521739130435,
+      "grad_norm": 0.0665697306394577,
+      "learning_rate": 9.998179792460676e-05,
+      "loss": 0.012,
+      "step": 558
+    },
+    {
+      "epoch": 2.7004830917874396,
+      "grad_norm": 0.05260172113776207,
+      "learning_rate": 9.99808765032841e-05,
+      "loss": 0.0126,
+      "step": 559
+    },
+    {
+      "epoch": 2.7053140096618358,
+      "grad_norm": 0.048168253153562546,
+      "learning_rate": 9.997993233669014e-05,
+      "loss": 0.013,
+      "step": 560
+    },
+    {
+      "epoch": 2.710144927536232,
+      "grad_norm": 0.07857447117567062,
+      "learning_rate": 9.997896542525459e-05,
+      "loss": 0.0097,
+      "step": 561
+    },
+    {
+      "epoch": 2.714975845410628,
+      "grad_norm": 0.042889662086963654,
+      "learning_rate": 9.997797576941744e-05,
+      "loss": 0.0122,
+      "step": 562
+    },
+    {
+      "epoch": 2.7198067632850242,
+      "grad_norm": 0.056598689407110214,
+      "learning_rate": 9.997696336962907e-05,
+      "loss": 0.0115,
+      "step": 563
+    },
+    {
+      "epoch": 2.7246376811594204,
+      "grad_norm": 0.06313615292310715,
+      "learning_rate": 9.997592822635021e-05,
+      "loss": 0.0152,
+      "step": 564
+    },
+    {
+      "epoch": 2.7294685990338166,
+      "grad_norm": 0.05057726800441742,
+      "learning_rate": 9.997487034005193e-05,
+      "loss": 0.0157,
+      "step": 565
+    },
+    {
+      "epoch": 2.7342995169082127,
+      "grad_norm": 0.044472839683294296,
+      "learning_rate": 9.997378971121564e-05,
+      "loss": 0.0098,
+      "step": 566
+    },
+    {
+      "epoch": 2.7391304347826084,
+      "grad_norm": 0.07471047341823578,
+      "learning_rate": 9.997268634033312e-05,
+      "loss": 0.0172,
+      "step": 567
+    },
+    {
+      "epoch": 2.7439613526570046,
+      "grad_norm": 0.053372822701931,
+      "learning_rate": 9.99715602279065e-05,
+      "loss": 0.014,
+      "step": 568
+    },
+    {
+      "epoch": 2.7487922705314007,
+      "grad_norm": 0.0864279493689537,
+      "learning_rate": 9.997041137444823e-05,
+      "loss": 0.0163,
+      "step": 569
+    },
+    {
+      "epoch": 2.753623188405797,
+      "grad_norm": 0.0629257932305336,
+      "learning_rate": 9.996923978048115e-05,
+      "loss": 0.0129,
+      "step": 570
+    },
+    {
+      "epoch": 2.758454106280193,
+      "grad_norm": 0.05703673139214516,
+      "learning_rate": 9.996804544653842e-05,
+      "loss": 0.0098,
+      "step": 571
+    },
+    {
+      "epoch": 2.763285024154589,
+      "grad_norm": 0.04491772875189781,
+      "learning_rate": 9.996682837316356e-05,
+      "loss": 0.0101,
+      "step": 572
+    },
+    {
+      "epoch": 2.7681159420289854,
+      "grad_norm": 0.06901387870311737,
+      "learning_rate": 9.996558856091043e-05,
+      "loss": 0.0198,
+      "step": 573
+    },
+    {
+      "epoch": 2.7729468599033815,
+      "grad_norm": 0.07164058834314346,
+      "learning_rate": 9.996432601034324e-05,
+      "loss": 0.0125,
+      "step": 574
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 0.059621043503284454,
+      "learning_rate": 9.996304072203657e-05,
+      "loss": 0.0089,
+      "step": 575
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "eval_loss": 0.00997437909245491,
+      "eval_runtime": 20.2741,
+      "eval_samples_per_second": 4.932,
+      "eval_steps_per_second": 0.148,
+      "step": 575
+    },
+    {
+      "epoch": 2.782608695652174,
+      "grad_norm": 0.07000112533569336,
+      "learning_rate": 9.99617326965753e-05,
+      "loss": 0.0105,
+      "step": 576
+    },
+    {
+      "epoch": 2.78743961352657,
+      "grad_norm": 0.08453238010406494,
+      "learning_rate": 9.996040193455472e-05,
+      "loss": 0.015,
+      "step": 577
+    },
+    {
+      "epoch": 2.792270531400966,
+      "grad_norm": 0.07728136330842972,
+      "learning_rate": 9.99590484365804e-05,
+      "loss": 0.017,
+      "step": 578
+    },
+    {
+      "epoch": 2.7971014492753623,
+      "grad_norm": 0.11687036603689194,
+      "learning_rate": 9.995767220326829e-05,
+      "loss": 0.0235,
+      "step": 579
+    },
+    {
+      "epoch": 2.8019323671497585,
+      "grad_norm": 0.07454598695039749,
+      "learning_rate": 9.99562732352447e-05,
+      "loss": 0.0115,
+      "step": 580
+    },
+    {
+      "epoch": 2.8067632850241546,
+      "grad_norm": 0.049887824803590775,
+      "learning_rate": 9.995485153314628e-05,
+      "loss": 0.0111,
+      "step": 581
+    },
+    {
+      "epoch": 2.8115942028985508,
+      "grad_norm": 0.05184588208794594,
+      "learning_rate": 9.995340709762002e-05,
+      "loss": 0.0149,
+      "step": 582
+    },
+    {
+      "epoch": 2.816425120772947,
+      "grad_norm": 0.05257965624332428,
+      "learning_rate": 9.995193992932321e-05,
+      "loss": 0.0132,
+      "step": 583
+    },
+    {
+      "epoch": 2.821256038647343,
+      "grad_norm": 0.05391634255647659,
+      "learning_rate": 9.995045002892358e-05,
+      "loss": 0.0159,
+      "step": 584
+    },
+    {
+      "epoch": 2.8260869565217392,
+      "grad_norm": 0.052167993038892746,
+      "learning_rate": 9.994893739709912e-05,
+      "loss": 0.0128,
+      "step": 585
+    },
+    {
+      "epoch": 2.8309178743961354,
+      "grad_norm": 0.03560145944356918,
+      "learning_rate": 9.994740203453821e-05,
+      "loss": 0.0094,
+      "step": 586
+    },
+    {
+      "epoch": 2.8357487922705316,
+      "grad_norm": 0.05537371709942818,
+      "learning_rate": 9.994584394193957e-05,
+      "loss": 0.0116,
+      "step": 587
+    },
+    {
+      "epoch": 2.8405797101449277,
+      "grad_norm": 0.05569041520357132,
+      "learning_rate": 9.994426312001223e-05,
+      "loss": 0.0121,
+      "step": 588
+    },
+    {
+      "epoch": 2.845410628019324,
+      "grad_norm": 0.04385409504175186,
+      "learning_rate": 9.994265956947563e-05,
+      "loss": 0.0078,
+      "step": 589
+    },
+    {
+      "epoch": 2.85024154589372,
+      "grad_norm": 0.056397683918476105,
+      "learning_rate": 9.994103329105947e-05,
+      "loss": 0.0147,
+      "step": 590
+    },
+    {
+      "epoch": 2.855072463768116,
+      "grad_norm": 0.06379224359989166,
+      "learning_rate": 9.993938428550387e-05,
+      "loss": 0.0147,
+      "step": 591
+    },
+    {
+      "epoch": 2.8599033816425123,
+      "grad_norm": 0.06637546420097351,
+      "learning_rate": 9.993771255355921e-05,
+      "loss": 0.0144,
+      "step": 592
+    },
+    {
+      "epoch": 2.864734299516908,
+      "grad_norm": 0.05549359321594238,
+      "learning_rate": 9.993601809598634e-05,
+      "loss": 0.0149,
+      "step": 593
+    },
+    {
+      "epoch": 2.869565217391304,
+      "grad_norm": 0.06366102397441864,
+      "learning_rate": 9.99343009135563e-05,
+      "loss": 0.0119,
+      "step": 594
+    },
+    {
+      "epoch": 2.8743961352657004,
+      "grad_norm": 0.047170039266347885,
+      "learning_rate": 9.993256100705058e-05,
+      "loss": 0.0126,
+      "step": 595
+    },
+    {
+      "epoch": 2.8792270531400965,
+      "grad_norm": 0.07046766579151154,
+      "learning_rate": 9.993079837726096e-05,
+      "loss": 0.0121,
+      "step": 596
+    },
+    {
+      "epoch": 2.8840579710144927,
+      "grad_norm": 0.06048014014959335,
+      "learning_rate": 9.992901302498959e-05,
+      "loss": 0.0125,
+      "step": 597
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.09012962877750397,
+      "learning_rate": 9.992720495104895e-05,
+      "loss": 0.0142,
+      "step": 598
+    },
+    {
+      "epoch": 2.893719806763285,
+      "grad_norm": 0.05745493993163109,
+      "learning_rate": 9.992537415626183e-05,
+      "loss": 0.0144,
+      "step": 599
+    },
+    {
+      "epoch": 2.898550724637681,
+      "grad_norm": 0.08338668942451477,
+      "learning_rate": 9.992352064146142e-05,
+      "loss": 0.0158,
+      "step": 600
+    },
+    {
+      "epoch": 2.898550724637681,
+      "eval_loss": 0.009580705314874649,
+      "eval_runtime": 20.3022,
+      "eval_samples_per_second": 4.926,
+      "eval_steps_per_second": 0.148,
+      "step": 600
+    },
+    {
+      "epoch": 2.9033816425120773,
+      "grad_norm": 0.04696241393685341,
+      "learning_rate": 9.992164440749119e-05,
+      "loss": 0.0104,
+      "step": 601
+    },
+    {
+      "epoch": 2.9082125603864735,
+      "grad_norm": 0.0753055065870285,
+      "learning_rate": 9.9919745455205e-05,
+      "loss": 0.0133,
+      "step": 602
+    },
+    {
+      "epoch": 2.9130434782608696,
+      "grad_norm": 0.054660920053720474,
+      "learning_rate": 9.991782378546702e-05,
+      "loss": 0.016,
+      "step": 603
+    },
+    {
+      "epoch": 2.917874396135266,
+      "grad_norm": 0.05713624507188797,
+      "learning_rate": 9.991587939915173e-05,
+      "loss": 0.0128,
+      "step": 604
+    },
+    {
+      "epoch": 2.922705314009662,
+      "grad_norm": 0.06272771209478378,
+      "learning_rate": 9.991391229714401e-05,
+      "loss": 0.016,
+      "step": 605
+    },
+    {
+      "epoch": 2.927536231884058,
+      "grad_norm": 0.059268005192279816,
+      "learning_rate": 9.991192248033908e-05,
+      "loss": 0.0201,
+      "step": 606
+    },
+    {
+      "epoch": 2.9323671497584543,
+      "grad_norm": 0.05580710619688034,
+      "learning_rate": 9.990990994964239e-05,
+      "loss": 0.0099,
+      "step": 607
+    },
+    {
+      "epoch": 2.9371980676328504,
+      "grad_norm": 0.05673280358314514,
+      "learning_rate": 9.990787470596985e-05,
+      "loss": 0.0138,
+      "step": 608
+    },
+    {
+      "epoch": 2.942028985507246,
+      "grad_norm": 0.05470005422830582,
+      "learning_rate": 9.990581675024763e-05,
+      "loss": 0.0109,
+      "step": 609
+    },
+    {
+      "epoch": 2.9468599033816423,
+      "grad_norm": 0.04339778795838356,
+      "learning_rate": 9.99037360834123e-05,
+      "loss": 0.0114,
+      "step": 610
+    },
+    {
+      "epoch": 2.9516908212560384,
+      "grad_norm": 0.04861651733517647,
+      "learning_rate": 9.99016327064107e-05,
+      "loss": 0.0105,
+      "step": 611
+    },
+    {
+      "epoch": 2.9565217391304346,
+      "grad_norm": 0.05664124712347984,
+      "learning_rate": 9.989950662020007e-05,
+      "loss": 0.014,
+      "step": 612
+    },
+    {
+      "epoch": 2.9613526570048307,
+      "grad_norm": 0.06411837786436081,
+      "learning_rate": 9.98973578257479e-05,
+      "loss": 0.0148,
+      "step": 613
+    },
+    {
+      "epoch": 2.966183574879227,
+      "grad_norm": 0.06767908483743668,
+      "learning_rate": 9.989518632403208e-05,
+      "loss": 0.0121,
+      "step": 614
+    },
+    {
+      "epoch": 2.971014492753623,
+      "grad_norm": 0.07915518432855606,
+      "learning_rate": 9.989299211604082e-05,
+      "loss": 0.0131,
+      "step": 615
+    },
+    {
+      "epoch": 2.975845410628019,
+      "grad_norm": 0.06088368594646454,
+      "learning_rate": 9.989077520277264e-05,
+      "loss": 0.0098,
+      "step": 616
+    },
+    {
+      "epoch": 2.9806763285024154,
+      "grad_norm": 0.0866197720170021,
+      "learning_rate": 9.988853558523646e-05,
+      "loss": 0.0148,
+      "step": 617
+    },
+    {
+      "epoch": 2.9855072463768115,
+      "grad_norm": 0.05279546231031418,
+      "learning_rate": 9.988627326445143e-05,
+      "loss": 0.0095,
+      "step": 618
+    },
+    {
+      "epoch": 2.9903381642512077,
+      "grad_norm": 0.06465419381856918,
+      "learning_rate": 9.988398824144714e-05,
+      "loss": 0.0214,
+      "step": 619
+    },
+    {
+      "epoch": 2.995169082125604,
+      "grad_norm": 0.06091202422976494,
+      "learning_rate": 9.98816805172634e-05,
+      "loss": 0.0131,
+      "step": 620
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.0471409372985363,
+      "learning_rate": 9.987935009295044e-05,
+      "loss": 0.0073,
+      "step": 621
+    },
+    {
+      "epoch": 3.004830917874396,
+      "grad_norm": 0.05190076678991318,
+      "learning_rate": 9.987699696956878e-05,
+      "loss": 0.0091,
+      "step": 622
+    },
+    {
+      "epoch": 3.0096618357487923,
+      "grad_norm": 0.03487321734428406,
+      "learning_rate": 9.987462114818928e-05,
+      "loss": 0.0091,
+      "step": 623
+    },
+    {
+      "epoch": 3.0144927536231885,
+      "grad_norm": 0.05589701607823372,
+      "learning_rate": 9.987222262989315e-05,
+      "loss": 0.0095,
+      "step": 624
+    },
+    {
+      "epoch": 3.0193236714975846,
+      "grad_norm": 0.06497800350189209,
+      "learning_rate": 9.986980141577187e-05,
+      "loss": 0.0119,
+      "step": 625
+    },
+    {
+      "epoch": 3.0193236714975846,
+      "eval_loss": 0.009575610049068928,
+      "eval_runtime": 20.3012,
+      "eval_samples_per_second": 4.926,
+      "eval_steps_per_second": 0.148,
+      "step": 625
+    },
+    {
+      "epoch": 3.024154589371981,
+      "grad_norm": 0.041647836565971375,
+      "learning_rate": 9.98673575069273e-05,
+      "loss": 0.0079,
+      "step": 626
+    },
+    {
+      "epoch": 3.028985507246377,
+      "grad_norm": 0.05941716581583023,
+      "learning_rate": 9.98648909044716e-05,
+      "loss": 0.0091,
+      "step": 627
+    },
+    {
+      "epoch": 3.033816425120773,
+      "grad_norm": 0.0788661390542984,
+      "learning_rate": 9.986240160952732e-05,
+      "loss": 0.0105,
+      "step": 628
+    },
+    {
+      "epoch": 3.0386473429951693,
+      "grad_norm": 0.04788121581077576,
+      "learning_rate": 9.985988962322721e-05,
+      "loss": 0.0075,
+      "step": 629
+    },
+    {
+      "epoch": 3.0434782608695654,
+      "grad_norm": 0.07558400183916092,
+      "learning_rate": 9.985735494671448e-05,
+      "loss": 0.0119,
+      "step": 630
+    },
+    {
+      "epoch": 3.0483091787439616,
+      "grad_norm": 0.06005953997373581,
+      "learning_rate": 9.985479758114259e-05,
+      "loss": 0.0122,
+      "step": 631
+    },
+    {
+      "epoch": 3.0531400966183573,
+      "grad_norm": 0.0517011322081089,
+      "learning_rate": 9.985221752767535e-05,
+      "loss": 0.0089,
+      "step": 632
+    },
+    {
+      "epoch": 3.0579710144927534,
+      "grad_norm": 0.045838914811611176,
+      "learning_rate": 9.984961478748688e-05,
+      "loss": 0.0066,
+      "step": 633
+    },
+    {
+      "epoch": 3.0628019323671496,
+      "grad_norm": 0.05188421905040741,
+      "learning_rate": 9.984698936176164e-05,
+      "loss": 0.0151,
+      "step": 634
+    },
+    {
+      "epoch": 3.0676328502415457,
+      "grad_norm": 0.04941697418689728,
+      "learning_rate": 9.984434125169441e-05,
+      "loss": 0.0132,
+      "step": 635
+    },
+    {
+      "epoch": 3.072463768115942,
+      "grad_norm": 0.07087542861700058,
+      "learning_rate": 9.98416704584903e-05,
+      "loss": 0.0143,
+      "step": 636
+    },
+    {
+      "epoch": 3.077294685990338,
+      "grad_norm": 0.03977886214852333,
+      "learning_rate": 9.983897698336471e-05,
+      "loss": 0.0117,
+      "step": 637
+    },
+    {
+      "epoch": 3.082125603864734,
+      "grad_norm": 0.04413791000843048,
+      "learning_rate": 9.98362608275434e-05,
+      "loss": 0.0084,
+      "step": 638
+    },
+    {
+      "epoch": 3.0869565217391304,
+      "grad_norm": 0.053000308573246,
+      "learning_rate": 9.983352199226243e-05,
+      "loss": 0.0111,
+      "step": 639
+    },
+    {
+      "epoch": 3.0917874396135265,
+      "grad_norm": 0.050405777990818024,
+      "learning_rate": 9.98307604787682e-05,
+      "loss": 0.0076,
+      "step": 640
+    },
+    {
+      "epoch": 3.0966183574879227,
+      "grad_norm": 0.041119880974292755,
+      "learning_rate": 9.982797628831739e-05,
+      "loss": 0.0102,
+      "step": 641
+    },
+    {
+      "epoch": 3.101449275362319,
+      "grad_norm": 0.06324031949043274,
+      "learning_rate": 9.982516942217705e-05,
+      "loss": 0.0105,
+      "step": 642
+    },
+    {
+      "epoch": 3.106280193236715,
+      "grad_norm": 0.06366854906082153,
+      "learning_rate": 9.982233988162455e-05,
+      "loss": 0.0106,
+      "step": 643
+    },
+    {
+      "epoch": 3.111111111111111,
+      "grad_norm": 0.06035742536187172,
+      "learning_rate": 9.981948766794752e-05,
+      "loss": 0.0109,
+      "step": 644
+    },
+    {
+      "epoch": 3.1159420289855073,
+      "grad_norm": 0.04412363842129707,
+      "learning_rate": 9.981661278244394e-05,
+      "loss": 0.0106,
+      "step": 645
+    },
+    {
+      "epoch": 3.1207729468599035,
+      "grad_norm": 0.07694416493177414,
+      "learning_rate": 9.981371522642212e-05,
+      "loss": 0.0102,
+      "step": 646
+    },
+    {
+      "epoch": 3.1256038647342996,
+      "grad_norm": 0.04476933553814888,
+      "learning_rate": 9.98107950012007e-05,
+      "loss": 0.009,
+      "step": 647
+    },
+    {
+      "epoch": 3.130434782608696,
+      "grad_norm": 0.06581299751996994,
+      "learning_rate": 9.980785210810859e-05,
+      "loss": 0.0094,
+      "step": 648
+    },
+    {
+      "epoch": 3.135265700483092,
+      "grad_norm": 0.05227496474981308,
+      "learning_rate": 9.980488654848505e-05,
+      "loss": 0.0127,
+      "step": 649
+    },
+    {
+      "epoch": 3.140096618357488,
+      "grad_norm": 0.043910861015319824,
+      "learning_rate": 9.980189832367966e-05,
+      "loss": 0.0097,
+      "step": 650
+    },
+    {
+      "epoch": 3.140096618357488,
+      "eval_loss": 0.009869378991425037,
+      "eval_runtime": 20.298,
+      "eval_samples_per_second": 4.927,
+      "eval_steps_per_second": 0.148,
+      "step": 650
+    },
+    {
+      "epoch": 3.1449275362318843,
+      "grad_norm": 0.05177716165781021,
+      "learning_rate": 9.979888743505225e-05,
+      "loss": 0.0095,
+      "step": 651
+    },
+    {
+      "epoch": 3.14975845410628,
+      "grad_norm": 0.07282808423042297,
+      "learning_rate": 9.979585388397308e-05,
+      "loss": 0.0145,
+      "step": 652
+    },
+    {
+      "epoch": 3.154589371980676,
+      "grad_norm": 0.05156973749399185,
+      "learning_rate": 9.979279767182262e-05,
+      "loss": 0.0098,
+      "step": 653
+    },
+    {
+      "epoch": 3.1594202898550723,
+      "grad_norm": 0.055524859577417374,
+      "learning_rate": 9.978971879999169e-05,
+      "loss": 0.0109,
+      "step": 654
+    },
+    {
+      "epoch": 3.1642512077294684,
+      "grad_norm": 0.06012894958257675,
+      "learning_rate": 9.97866172698814e-05,
+      "loss": 0.0087,
+      "step": 655
+    },
+    {
+      "epoch": 3.1690821256038646,
+      "grad_norm": 0.05175028741359711,
+      "learning_rate": 9.978349308290325e-05,
+      "loss": 0.0072,
+      "step": 656
+    },
+    {
+      "epoch": 3.1739130434782608,
+      "grad_norm": 0.04955373331904411,
+      "learning_rate": 9.978034624047895e-05,
+      "loss": 0.0092,
+      "step": 657
+    },
+    {
+      "epoch": 3.178743961352657,
+      "grad_norm": 0.05167277157306671,
+      "learning_rate": 9.977717674404056e-05,
+      "loss": 0.0088,
+      "step": 658
+    },
+    {
+      "epoch": 3.183574879227053,
+      "grad_norm": 0.04037657380104065,
+      "learning_rate": 9.977398459503049e-05,
+      "loss": 0.009,
+      "step": 659
+    },
+    {
+      "epoch": 3.1884057971014492,
+      "grad_norm": 0.04062899202108383,
+      "learning_rate": 9.977076979490138e-05,
+      "loss": 0.007,
+      "step": 660
+    },
+    {
+      "epoch": 3.1932367149758454,
+      "grad_norm": 0.06958416849374771,
+      "learning_rate": 9.976753234511627e-05,
+      "loss": 0.0119,
+      "step": 661
+    },
+    {
+      "epoch": 3.1980676328502415,
+      "grad_norm": 0.05316910520195961,
+      "learning_rate": 9.97642722471484e-05,
+      "loss": 0.0108,
+      "step": 662
+    },
+    {
+      "epoch": 3.2028985507246377,
+      "grad_norm": 0.050726763904094696,
+      "learning_rate": 9.976098950248141e-05,
+      "loss": 0.0088,
+      "step": 663
+    },
+    {
+      "epoch": 3.207729468599034,
+      "grad_norm": 0.047353360801935196,
+      "learning_rate": 9.975768411260917e-05,
+      "loss": 0.0078,
+      "step": 664
+    },
+    {
+      "epoch": 3.21256038647343,
+      "grad_norm": 0.0337323397397995,
+      "learning_rate": 9.975435607903596e-05,
+      "loss": 0.0085,
+      "step": 665
+    },
+    {
+      "epoch": 3.217391304347826,
+      "grad_norm": 0.051542479544878006,
+      "learning_rate": 9.975100540327624e-05,
+      "loss": 0.0085,
+      "step": 666
+    },
+    {
+      "epoch": 3.2222222222222223,
+      "grad_norm": 0.07729382812976837,
+      "learning_rate": 9.974763208685487e-05,
+      "loss": 0.0126,
+      "step": 667
+    },
+    {
+      "epoch": 3.2270531400966185,
+      "grad_norm": 0.06189757585525513,
+      "learning_rate": 9.974423613130697e-05,
+      "loss": 0.009,
+      "step": 668
+    },
+    {
+      "epoch": 3.2318840579710146,
+      "grad_norm": 0.04578680172562599,
+      "learning_rate": 9.974081753817795e-05,
+      "loss": 0.0075,
+      "step": 669
+    },
+    {
+      "epoch": 3.236714975845411,
+      "grad_norm": 0.04176221042871475,
+      "learning_rate": 9.973737630902356e-05,
+      "loss": 0.0083,
+      "step": 670
+    },
+    {
+      "epoch": 3.241545893719807,
+      "grad_norm": 0.08334269374608994,
+      "learning_rate": 9.973391244540983e-05,
+      "loss": 0.0106,
+      "step": 671
+    },
+    {
+      "epoch": 3.246376811594203,
+      "grad_norm": 0.06806229054927826,
+      "learning_rate": 9.973042594891309e-05,
+      "loss": 0.0134,
+      "step": 672
+    },
+    {
+      "epoch": 3.2512077294685993,
+      "grad_norm": 0.04712403565645218,
+      "learning_rate": 9.972691682111997e-05,
+      "loss": 0.01,
+      "step": 673
+    },
+    {
+      "epoch": 3.2560386473429954,
+      "grad_norm": 0.04654091224074364,
+      "learning_rate": 9.972338506362742e-05,
+      "loss": 0.01,
+      "step": 674
+    },
+    {
+      "epoch": 3.260869565217391,
+      "grad_norm": 0.05907722935080528,
+      "learning_rate": 9.971983067804265e-05,
+      "loss": 0.0089,
+      "step": 675
+    },
+    {
+      "epoch": 3.260869565217391,
+      "eval_loss": 0.009203135967254639,
+      "eval_runtime": 20.2973,
+      "eval_samples_per_second": 4.927,
+      "eval_steps_per_second": 0.148,
+      "step": 675
+    },
+    {
+      "epoch": 3.2657004830917873,
+      "grad_norm": 0.06332624703645706,
+      "learning_rate": 9.971625366598319e-05,
+      "loss": 0.0099,
+      "step": 676
+    },
+    {
+      "epoch": 3.2705314009661834,
+      "grad_norm": 0.06277251988649368,
+      "learning_rate": 9.971265402907688e-05,
+      "loss": 0.0085,
+      "step": 677
+    },
+    {
+      "epoch": 3.2753623188405796,
+      "grad_norm": 0.05057535693049431,
+      "learning_rate": 9.970903176896183e-05,
+      "loss": 0.011,
+      "step": 678
+    },
+    {
+      "epoch": 3.2801932367149758,
+      "grad_norm": 0.04588181897997856,
+      "learning_rate": 9.970538688728644e-05,
+      "loss": 0.0086,
+      "step": 679
+    },
+    {
+      "epoch": 3.285024154589372,
+      "grad_norm": 0.05869683250784874,
+      "learning_rate": 9.970171938570946e-05,
+      "loss": 0.0081,
+      "step": 680
+    },
+    {
+      "epoch": 3.289855072463768,
+      "grad_norm": 0.050425752997398376,
+      "learning_rate": 9.969802926589986e-05,
+      "loss": 0.0121,
+      "step": 681
+    },
+    {
+      "epoch": 3.2946859903381642,
+      "grad_norm": 0.060307394713163376,
+      "learning_rate": 9.969431652953695e-05,
+      "loss": 0.0089,
+      "step": 682
+    },
+    {
+      "epoch": 3.2995169082125604,
+      "grad_norm": 0.07875365763902664,
+      "learning_rate": 9.969058117831034e-05,
+      "loss": 0.011,
+      "step": 683
+    },
+    {
+      "epoch": 3.3043478260869565,
+      "grad_norm": 0.07338164746761322,
+      "learning_rate": 9.968682321391986e-05,
+      "loss": 0.0112,
+      "step": 684
+    },
+    {
+      "epoch": 3.3091787439613527,
+      "grad_norm": 0.09912315756082535,
+      "learning_rate": 9.968304263807574e-05,
+      "loss": 0.0113,
+      "step": 685
+    },
+    {
+      "epoch": 3.314009661835749,
+      "grad_norm": 0.10973814129829407,
+      "learning_rate": 9.96792394524984e-05,
+      "loss": 0.0096,
+      "step": 686
+    },
+    {
+      "epoch": 3.318840579710145,
+      "grad_norm": 0.06479693949222565,
+      "learning_rate": 9.967541365891863e-05,
+      "loss": 0.0109,
+      "step": 687
+    },
+    {
+      "epoch": 3.323671497584541,
+      "grad_norm": 0.05435623601078987,
+      "learning_rate": 9.967156525907743e-05,
+      "loss": 0.0065,
+      "step": 688
+    },
+    {
+      "epoch": 3.3285024154589373,
+      "grad_norm": 0.04574191942811012,
+      "learning_rate": 9.966769425472616e-05,
+      "loss": 0.0066,
+      "step": 689
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.04818063601851463,
+      "learning_rate": 9.966380064762642e-05,
+      "loss": 0.0064,
+      "step": 690
+    },
+    {
+      "epoch": 3.3381642512077296,
+      "grad_norm": 0.046589504927396774,
+      "learning_rate": 9.96598844395501e-05,
+      "loss": 0.007,
+      "step": 691
+    },
+    {
+      "epoch": 3.342995169082126,
+      "grad_norm": 0.05948101729154587,
+      "learning_rate": 9.96559456322794e-05,
+      "loss": 0.012,
+      "step": 692
+    },
+    {
+      "epoch": 3.3478260869565215,
+      "grad_norm": 0.06270439177751541,
+      "learning_rate": 9.96519842276068e-05,
+      "loss": 0.0118,
+      "step": 693
+    },
+    {
+      "epoch": 3.3526570048309177,
+      "grad_norm": 0.047514282166957855,
+      "learning_rate": 9.964800022733504e-05,
+      "loss": 0.0094,
+      "step": 694
+    },
+    {
+      "epoch": 3.357487922705314,
+      "grad_norm": 0.04377506673336029,
+      "learning_rate": 9.964399363327716e-05,
+      "loss": 0.0109,
+      "step": 695
+    },
+    {
+      "epoch": 3.36231884057971,
+      "grad_norm": 0.04509812965989113,
+      "learning_rate": 9.963996444725647e-05,
+      "loss": 0.0112,
+      "step": 696
+    },
+    {
+      "epoch": 3.367149758454106,
+      "grad_norm": 0.03941660374403,
+      "learning_rate": 9.96359126711066e-05,
+      "loss": 0.0099,
+      "step": 697
+    },
+    {
+      "epoch": 3.3719806763285023,
+      "grad_norm": 0.04203685000538826,
+      "learning_rate": 9.963183830667138e-05,
+      "loss": 0.0088,
+      "step": 698
+    },
+    {
+      "epoch": 3.3768115942028984,
+      "grad_norm": 0.05573079362511635,
+      "learning_rate": 9.9627741355805e-05,
+      "loss": 0.0095,
+      "step": 699
+    },
+    {
+      "epoch": 3.3816425120772946,
+      "grad_norm": 0.04746829718351364,
+      "learning_rate": 9.96236218203719e-05,
+      "loss": 0.0087,
+      "step": 700
+    },
+    {
+      "epoch": 3.3816425120772946,
+      "eval_loss": 0.00878093857318163,
+      "eval_runtime": 20.3171,
+      "eval_samples_per_second": 4.922,
+      "eval_steps_per_second": 0.148,
+      "step": 700
+    },
+    {
+      "epoch": 3.3864734299516908,
+      "grad_norm": 0.056552231311798096,
+      "learning_rate": 9.96194797022468e-05,
+      "loss": 0.0129,
+      "step": 701
+    },
+    {
+      "epoch": 3.391304347826087,
+      "grad_norm": 0.0671212300658226,
+      "learning_rate": 9.961531500331469e-05,
+      "loss": 0.0108,
+      "step": 702
+    },
+    {
+      "epoch": 3.396135265700483,
+      "grad_norm": 0.04948098585009575,
+      "learning_rate": 9.961112772547083e-05,
+      "loss": 0.0126,
+      "step": 703
+    },
+    {
+      "epoch": 3.4009661835748792,
+      "grad_norm": 0.058139242231845856,
+      "learning_rate": 9.960691787062076e-05,
+      "loss": 0.0089,
+      "step": 704
+    },
+    {
+      "epoch": 3.4057971014492754,
+      "grad_norm": 0.045286696404218674,
+      "learning_rate": 9.960268544068032e-05,
+      "loss": 0.0083,
+      "step": 705
+    },
+    {
+      "epoch": 3.4106280193236715,
+      "grad_norm": 0.07364138215780258,
+      "learning_rate": 9.959843043757557e-05,
+      "loss": 0.0134,
+      "step": 706
+    },
+    {
+      "epoch": 3.4154589371980677,
+      "grad_norm": 0.052311353385448456,
+      "learning_rate": 9.959415286324289e-05,
+      "loss": 0.0076,
+      "step": 707
+    },
+    {
+      "epoch": 3.420289855072464,
+      "grad_norm": 0.048165835440158844,
+      "learning_rate": 9.958985271962896e-05,
+      "loss": 0.0085,
+      "step": 708
+    },
+    {
+      "epoch": 3.42512077294686,
+      "grad_norm": 0.039163313806056976,
+      "learning_rate": 9.958553000869061e-05,
+      "loss": 0.0084,
+      "step": 709
+    },
+    {
+      "epoch": 3.429951690821256,
+      "grad_norm": 0.07590556144714355,
+      "learning_rate": 9.958118473239507e-05,
+      "loss": 0.0089,
+      "step": 710
+    },
+    {
+      "epoch": 3.4347826086956523,
+      "grad_norm": 0.05805916339159012,
+      "learning_rate": 9.957681689271977e-05,
+      "loss": 0.0091,
+      "step": 711
+    },
+    {
+      "epoch": 3.4396135265700485,
+      "grad_norm": 0.05946679040789604,
+      "learning_rate": 9.957242649165241e-05,
+      "loss": 0.0111,
+      "step": 712
+    },
+    {
+      "epoch": 3.4444444444444446,
+      "grad_norm": 0.04727727919816971,
+      "learning_rate": 9.956801353119099e-05,
+      "loss": 0.0097,
+      "step": 713
+    },
+    {
+      "epoch": 3.449275362318841,
+      "grad_norm": 0.07006451487541199,
+      "learning_rate": 9.956357801334375e-05,
+      "loss": 0.0104,
+      "step": 714
+    },
+    {
+      "epoch": 3.454106280193237,
+      "grad_norm": 0.04728998243808746,
+      "learning_rate": 9.955911994012923e-05,
+      "loss": 0.0103,
+      "step": 715
+    },
+    {
+      "epoch": 3.4589371980676327,
+      "grad_norm": 0.03930573910474777,
+      "learning_rate": 9.955463931357616e-05,
+      "loss": 0.0087,
+      "step": 716
+    },
+    {
+      "epoch": 3.463768115942029,
+      "grad_norm": 0.04594094678759575,
+      "learning_rate": 9.955013613572362e-05,
+      "loss": 0.0069,
+      "step": 717
+    },
+    {
+      "epoch": 3.468599033816425,
+      "grad_norm": 0.05382443219423294,
+      "learning_rate": 9.954561040862088e-05,
+      "loss": 0.008,
+      "step": 718
+    },
+    {
+      "epoch": 3.473429951690821,
+      "grad_norm": 0.05387111380696297,
+      "learning_rate": 9.954106213432755e-05,
+      "loss": 0.0079,
+      "step": 719
+    },
+    {
+      "epoch": 3.4782608695652173,
+      "grad_norm": 0.0623965822160244,
+      "learning_rate": 9.95364913149134e-05,
+      "loss": 0.012,
+      "step": 720
+    },
+    {
+      "epoch": 3.4830917874396135,
+      "grad_norm": 0.04346054419875145,
+      "learning_rate": 9.953189795245857e-05,
+      "loss": 0.0098,
+      "step": 721
+    },
+    {
+      "epoch": 3.4879227053140096,
+      "grad_norm": 0.05038152635097504,
+      "learning_rate": 9.952728204905338e-05,
+      "loss": 0.0098,
+      "step": 722
+    },
+    {
+      "epoch": 3.4927536231884058,
+      "grad_norm": 0.04060295969247818,
+      "learning_rate": 9.952264360679844e-05,
+      "loss": 0.0069,
+      "step": 723
+    },
+    {
+      "epoch": 3.497584541062802,
+      "grad_norm": 0.04667475074529648,
+      "learning_rate": 9.951798262780458e-05,
+      "loss": 0.0098,
+      "step": 724
+    },
+    {
+      "epoch": 3.502415458937198,
+      "grad_norm": 0.062279459089040756,
+      "learning_rate": 9.951329911419298e-05,
+      "loss": 0.0083,
+      "step": 725
+    },
+    {
+      "epoch": 3.502415458937198,
+      "eval_loss": 0.00877303909510374,
+      "eval_runtime": 20.3131,
+      "eval_samples_per_second": 4.923,
+      "eval_steps_per_second": 0.148,
+      "step": 725
+    },
+    {
+      "epoch": 3.5072463768115942,
+      "grad_norm": 0.07258505374193192,
+      "learning_rate": 9.950859306809494e-05,
+      "loss": 0.0086,
+      "step": 726
+    },
+    {
+      "epoch": 3.5120772946859904,
+      "grad_norm": 0.06038128584623337,
+      "learning_rate": 9.950386449165212e-05,
+      "loss": 0.0103,
+      "step": 727
+    },
+    {
+      "epoch": 3.5169082125603865,
+      "grad_norm": 0.04698696732521057,
+      "learning_rate": 9.94991133870164e-05,
+      "loss": 0.0086,
+      "step": 728
+    },
+    {
+      "epoch": 3.5217391304347827,
+      "grad_norm": 0.038896944373846054,
+      "learning_rate": 9.949433975634992e-05,
+      "loss": 0.0105,
+      "step": 729
+    },
+    {
+      "epoch": 3.526570048309179,
+      "grad_norm": 0.05197020620107651,
+      "learning_rate": 9.948954360182503e-05,
+      "loss": 0.009,
+      "step": 730
+    },
+    {
+      "epoch": 3.531400966183575,
+      "grad_norm": 0.05059540644288063,
+      "learning_rate": 9.948472492562438e-05,
+      "loss": 0.0114,
+      "step": 731
+    },
+    {
+      "epoch": 3.536231884057971,
+      "grad_norm": 0.04194273799657822,
+      "learning_rate": 9.947988372994086e-05,
+      "loss": 0.0106,
+      "step": 732
+    },
+    {
+      "epoch": 3.541062801932367,
+      "grad_norm": 0.04211822524666786,
+      "learning_rate": 9.947502001697757e-05,
+      "loss": 0.0098,
+      "step": 733
+    },
+    {
+      "epoch": 3.545893719806763,
+      "grad_norm": 0.06714488565921783,
+      "learning_rate": 9.947013378894792e-05,
+      "loss": 0.0114,
+      "step": 734
+    },
+    {
+      "epoch": 3.550724637681159,
+      "grad_norm": 0.038067255169153214,
+      "learning_rate": 9.946522504807551e-05,
+      "loss": 0.0087,
+      "step": 735
+    },
+    {
+      "epoch": 3.5555555555555554,
+      "grad_norm": 0.03965510427951813,
+      "learning_rate": 9.94602937965942e-05,
+      "loss": 0.0087,
+      "step": 736
+    },
+    {
+      "epoch": 3.5603864734299515,
+      "grad_norm": 0.05775628611445427,
+      "learning_rate": 9.945534003674812e-05,
+      "loss": 0.0095,
+      "step": 737
+    },
+    {
+      "epoch": 3.5652173913043477,
+      "grad_norm": 0.06312840431928635,
+      "learning_rate": 9.945036377079164e-05,
+      "loss": 0.0121,
+      "step": 738
+    },
+    {
+      "epoch": 3.570048309178744,
+      "grad_norm": 0.058142922818660736,
+      "learning_rate": 9.94453650009893e-05,
+      "loss": 0.0104,
+      "step": 739
+    },
+    {
+      "epoch": 3.57487922705314,
+      "grad_norm": 0.09870404750108719,
+      "learning_rate": 9.9440343729616e-05,
+      "loss": 0.0065,
+      "step": 740
+    },
+    {
+      "epoch": 3.579710144927536,
+      "grad_norm": 0.07584329694509506,
+      "learning_rate": 9.943529995895679e-05,
+      "loss": 0.0097,
+      "step": 741
+    },
+    {
+      "epoch": 3.5845410628019323,
+      "grad_norm": 0.09112714231014252,
+      "learning_rate": 9.943023369130698e-05,
+      "loss": 0.0115,
+      "step": 742
+    },
+    {
+      "epoch": 3.5893719806763285,
+      "grad_norm": 0.032687995582818985,
+      "learning_rate": 9.942514492897212e-05,
+      "loss": 0.0072,
+      "step": 743
+    },
+    {
+      "epoch": 3.5942028985507246,
+      "grad_norm": 0.047948360443115234,
+      "learning_rate": 9.942003367426803e-05,
+      "loss": 0.0084,
+      "step": 744
+    },
+    {
+      "epoch": 3.5990338164251208,
+      "grad_norm": 0.05047725886106491,
+      "learning_rate": 9.941489992952071e-05,
+      "loss": 0.0082,
+      "step": 745
+    },
+    {
+      "epoch": 3.603864734299517,
+      "grad_norm": 0.045492023229599,
+      "learning_rate": 9.940974369706642e-05,
+      "loss": 0.0085,
+      "step": 746
+    },
+    {
+      "epoch": 3.608695652173913,
+      "grad_norm": 0.042984895408153534,
+      "learning_rate": 9.940456497925168e-05,
+      "loss": 0.0082,
+      "step": 747
+    },
+    {
+      "epoch": 3.6135265700483092,
+      "grad_norm": 0.06112075597047806,
+      "learning_rate": 9.939936377843321e-05,
+      "loss": 0.0097,
+      "step": 748
+    },
+    {
+      "epoch": 3.6183574879227054,
+      "grad_norm": 0.06138254702091217,
+      "learning_rate": 9.939414009697795e-05,
+      "loss": 0.0073,
+      "step": 749
+    },
+    {
+      "epoch": 3.6231884057971016,
+      "grad_norm": 0.05048954114317894,
+      "learning_rate": 9.938889393726314e-05,
+      "loss": 0.0088,
+      "step": 750
+    },
+    {
+      "epoch": 3.6231884057971016,
+      "eval_loss": 0.007979786954820156,
+      "eval_runtime": 20.3088,
+      "eval_samples_per_second": 4.924,
+      "eval_steps_per_second": 0.148,
+      "step": 750
+    },
+    {
+      "epoch": 3.6280193236714977,
+      "grad_norm": 0.05653483420610428,
+      "learning_rate": 9.938362530167613e-05,
+      "loss": 0.0061,
+      "step": 751
+    },
+    {
+      "epoch": 3.632850241545894,
+      "grad_norm": 0.046065203845500946,
+      "learning_rate": 9.937833419261462e-05,
+      "loss": 0.0091,
+      "step": 752
+    },
+    {
+      "epoch": 3.63768115942029,
+      "grad_norm": 0.053545523434877396,
+      "learning_rate": 9.937302061248646e-05,
+      "loss": 0.008,
+      "step": 753
+    },
+    {
+      "epoch": 3.642512077294686,
+      "grad_norm": 0.06943827867507935,
+      "learning_rate": 9.936768456370977e-05,
+      "loss": 0.0239,
+      "step": 754
+    },
+    {
+      "epoch": 3.6473429951690823,
+      "grad_norm": 0.04048341512680054,
+      "learning_rate": 9.936232604871285e-05,
+      "loss": 0.0092,
+      "step": 755
+    },
+    {
+      "epoch": 3.6521739130434785,
+      "grad_norm": 0.05726747587323189,
+      "learning_rate": 9.935694506993427e-05,
+      "loss": 0.0163,
+      "step": 756
+    },
+    {
+      "epoch": 3.6570048309178746,
+      "grad_norm": 0.039764195680618286,
+      "learning_rate": 9.935154162982281e-05,
+      "loss": 0.0073,
+      "step": 757
+    },
+    {
+      "epoch": 3.661835748792271,
+      "grad_norm": 0.05742327868938446,
+      "learning_rate": 9.934611573083744e-05,
+      "loss": 0.0091,
+      "step": 758
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 0.0707569345831871,
+      "learning_rate": 9.934066737544741e-05,
+      "loss": 0.009,
+      "step": 759
+    },
+    {
+      "epoch": 3.6714975845410627,
+      "grad_norm": 0.04180197790265083,
+      "learning_rate": 9.93351965661321e-05,
+      "loss": 0.0109,
+      "step": 760
+    },
+    {
+      "epoch": 3.676328502415459,
+      "grad_norm": 0.05421733483672142,
+      "learning_rate": 9.932970330538123e-05,
+      "loss": 0.0106,
+      "step": 761
+    },
+    {
+      "epoch": 3.681159420289855,
+      "grad_norm": 0.03703097254037857,
+      "learning_rate": 9.932418759569462e-05,
+      "loss": 0.0083,
+      "step": 762
+    },
+    {
+      "epoch": 3.685990338164251,
+      "grad_norm": 0.04790617153048515,
+      "learning_rate": 9.931864943958238e-05,
+      "loss": 0.0092,
+      "step": 763
+    },
+    {
+      "epoch": 3.6908212560386473,
+      "grad_norm": 0.057676345109939575,
+      "learning_rate": 9.931308883956479e-05,
+      "loss": 0.0085,
+      "step": 764
+    },
+    {
+      "epoch": 3.6956521739130435,
+      "grad_norm": 0.03428034111857414,
+      "learning_rate": 9.930750579817239e-05,
+      "loss": 0.0066,
+      "step": 765
+    },
+    {
+      "epoch": 3.7004830917874396,
+      "grad_norm": 0.05824226886034012,
+      "learning_rate": 9.93019003179459e-05,
+      "loss": 0.0113,
+      "step": 766
+    },
+    {
+      "epoch": 3.7053140096618358,
+      "grad_norm": 0.032023362815380096,
+      "learning_rate": 9.929627240143625e-05,
+      "loss": 0.0073,
+      "step": 767
+    },
+    {
+      "epoch": 3.710144927536232,
+      "grad_norm": 0.04681350663304329,
+      "learning_rate": 9.92906220512046e-05,
+      "loss": 0.011,
+      "step": 768
+    },
+    {
+      "epoch": 3.714975845410628,
+      "grad_norm": 0.05592908337712288,
+      "learning_rate": 9.92849492698223e-05,
+      "loss": 0.0105,
+      "step": 769
+    },
+    {
+      "epoch": 3.7198067632850242,
+      "grad_norm": 0.042512789368629456,
+      "learning_rate": 9.927925405987093e-05,
+      "loss": 0.0094,
+      "step": 770
+    },
+    {
+      "epoch": 3.7246376811594204,
+      "grad_norm": 0.04821282997727394,
+      "learning_rate": 9.927353642394224e-05,
+      "loss": 0.0094,
+      "step": 771
+    },
+    {
+      "epoch": 3.7294685990338166,
+      "grad_norm": 0.05211241543292999,
+      "learning_rate": 9.926779636463824e-05,
+      "loss": 0.0116,
+      "step": 772
+    },
+    {
+      "epoch": 3.7342995169082127,
+      "grad_norm": 0.04904523119330406,
+      "learning_rate": 9.926203388457107e-05,
+      "loss": 0.0119,
+      "step": 773
+    },
+    {
+      "epoch": 3.7391304347826084,
+      "grad_norm": 0.061542410403490067,
+      "learning_rate": 9.925624898636317e-05,
+      "loss": 0.009,
+      "step": 774
+    },
+    {
+      "epoch": 3.7439613526570046,
+      "grad_norm": 0.04940609633922577,
+      "learning_rate": 9.925044167264708e-05,
+      "loss": 0.0058,
+      "step": 775
+    },
+    {
+      "epoch": 3.7439613526570046,
+      "eval_loss": 0.006909946445375681,
+      "eval_runtime": 20.3264,
+      "eval_samples_per_second": 4.92,
+      "eval_steps_per_second": 0.148,
+      "step": 775
+    },
+    {
+      "epoch": 3.7487922705314007,
+      "grad_norm": 0.07155938446521759,
+      "learning_rate": 9.924461194606561e-05,
+      "loss": 0.0092,
+      "step": 776
+    },
+    {
+      "epoch": 3.753623188405797,
+      "grad_norm": 0.04605531692504883,
+      "learning_rate": 9.923875980927175e-05,
+      "loss": 0.0065,
+      "step": 777
+    },
+    {
+      "epoch": 3.758454106280193,
+      "grad_norm": 0.0463431291282177,
+      "learning_rate": 9.923288526492869e-05,
+      "loss": 0.0062,
+      "step": 778
+    },
+    {
+      "epoch": 3.763285024154589,
+      "grad_norm": 0.04100046679377556,
+      "learning_rate": 9.922698831570982e-05,
+      "loss": 0.0102,
+      "step": 779
+    },
+    {
+      "epoch": 3.7681159420289854,
+      "grad_norm": 0.06026915833353996,
+      "learning_rate": 9.92210689642987e-05,
+      "loss": 0.0109,
+      "step": 780
+    },
+    {
+      "epoch": 3.7729468599033815,
+      "grad_norm": 0.03291574865579605,
+      "learning_rate": 9.921512721338912e-05,
+      "loss": 0.0066,
+      "step": 781
+    },
+    {
+      "epoch": 3.7777777777777777,
+      "grad_norm": 0.04719441756606102,
+      "learning_rate": 9.920916306568504e-05,
+      "loss": 0.007,
+      "step": 782
+    },
+    {
+      "epoch": 3.782608695652174,
+      "grad_norm": 0.06509106606245041,
+      "learning_rate": 9.920317652390063e-05,
+      "loss": 0.0092,
+      "step": 783
+    },
+    {
+      "epoch": 3.78743961352657,
+      "grad_norm": 0.06969207525253296,
+      "learning_rate": 9.919716759076025e-05,
+      "loss": 0.0157,
+      "step": 784
+    },
+    {
+      "epoch": 3.792270531400966,
+      "grad_norm": 0.044033829122781754,
+      "learning_rate": 9.919113626899841e-05,
+      "loss": 0.0064,
+      "step": 785
+    },
+    {
+      "epoch": 3.7971014492753623,
+      "grad_norm": 0.05721135437488556,
+      "learning_rate": 9.918508256135988e-05,
+      "loss": 0.0112,
+      "step": 786
+    },
+    {
+      "epoch": 3.8019323671497585,
+      "grad_norm": 0.07257521897554398,
+      "learning_rate": 9.917900647059955e-05,
+      "loss": 0.0084,
+      "step": 787
+    },
+    {
+      "epoch": 3.8067632850241546,
+      "grad_norm": 0.04365171492099762,
+      "learning_rate": 9.917290799948253e-05,
+      "loss": 0.008,
+      "step": 788
+    },
+    {
+      "epoch": 3.8115942028985508,
+      "grad_norm": 0.05915320664644241,
+      "learning_rate": 9.916678715078411e-05,
+      "loss": 0.0075,
+      "step": 789
+    },
+    {
+      "epoch": 3.816425120772947,
+      "grad_norm": 0.06364689022302628,
+      "learning_rate": 9.916064392728979e-05,
+      "loss": 0.0052,
+      "step": 790
+    },
+    {
+      "epoch": 3.821256038647343,
+      "grad_norm": 0.0655299499630928,
+      "learning_rate": 9.915447833179519e-05,
+      "loss": 0.0149,
+      "step": 791
+    },
+    {
+      "epoch": 3.8260869565217392,
+      "grad_norm": 0.03559907153248787,
+      "learning_rate": 9.914829036710614e-05,
+      "loss": 0.0067,
+      "step": 792
+    },
+    {
+      "epoch": 3.8309178743961354,
+      "grad_norm": 0.05308472737669945,
+      "learning_rate": 9.914208003603869e-05,
+      "loss": 0.0088,
+      "step": 793
+    },
+    {
+      "epoch": 3.8357487922705316,
+      "grad_norm": 0.044624749571084976,
+      "learning_rate": 9.913584734141901e-05,
+      "loss": 0.0076,
+      "step": 794
+    },
+    {
+      "epoch": 3.8405797101449277,
+      "grad_norm": 0.08039918541908264,
+      "learning_rate": 9.912959228608348e-05,
+      "loss": 0.0079,
+      "step": 795
+    },
+    {
+      "epoch": 3.845410628019324,
+      "grad_norm": 0.055277448147535324,
+      "learning_rate": 9.912331487287864e-05,
+      "loss": 0.0134,
+      "step": 796
+    },
+    {
+      "epoch": 3.85024154589372,
+      "grad_norm": 0.047117929905653,
+      "learning_rate": 9.911701510466124e-05,
+      "loss": 0.0089,
+      "step": 797
+    },
+    {
+      "epoch": 3.855072463768116,
+      "grad_norm": 0.04270472377538681,
+      "learning_rate": 9.911069298429814e-05,
+      "loss": 0.0085,
+      "step": 798
+    },
+    {
+      "epoch": 3.8599033816425123,
+      "grad_norm": 0.060901861637830734,
+      "learning_rate": 9.910434851466642e-05,
+      "loss": 0.011,
+      "step": 799
+    },
+    {
+      "epoch": 3.864734299516908,
+      "grad_norm": 0.0431562140583992,
+      "learning_rate": 9.90979816986533e-05,
+      "loss": 0.008,
+      "step": 800
+    },
+    {
+      "epoch": 3.864734299516908,
+      "eval_loss": 0.007014184258878231,
+      "eval_runtime": 20.3118,
+      "eval_samples_per_second": 4.923,
+      "eval_steps_per_second": 0.148,
+      "step": 800
+    },
+    {
+      "epoch": 3.869565217391304,
+      "grad_norm": 0.03784063830971718,
+      "learning_rate": 9.909159253915623e-05,
+      "loss": 0.009,
+      "step": 801
+    },
+    {
+      "epoch": 3.8743961352657004,
+      "grad_norm": 0.046868495643138885,
+      "learning_rate": 9.908518103908274e-05,
+      "loss": 0.0112,
+      "step": 802
+    },
+    {
+      "epoch": 3.8792270531400965,
+      "grad_norm": 0.034640517085790634,
+      "learning_rate": 9.907874720135061e-05,
+      "loss": 0.0082,
+      "step": 803
+    },
+    {
+      "epoch": 3.8840579710144927,
+      "grad_norm": 0.03212960809469223,
+      "learning_rate": 9.907229102888772e-05,
+      "loss": 0.0073,
+      "step": 804
+    },
+    {
+      "epoch": 3.888888888888889,
+      "grad_norm": 0.038106419146060944,
+      "learning_rate": 9.906581252463216e-05,
+      "loss": 0.0083,
+      "step": 805
+    },
+    {
+      "epoch": 3.893719806763285,
+      "grad_norm": 0.04272741824388504,
+      "learning_rate": 9.905931169153215e-05,
+      "loss": 0.0126,
+      "step": 806
+    },
+    {
+      "epoch": 3.898550724637681,
+      "grad_norm": 0.05206140875816345,
+      "learning_rate": 9.905278853254609e-05,
+      "loss": 0.0095,
+      "step": 807
+    },
+    {
+      "epoch": 3.9033816425120773,
+      "grad_norm": 0.04624548926949501,
+      "learning_rate": 9.904624305064255e-05,
+      "loss": 0.0088,
+      "step": 808
+    },
+    {
+      "epoch": 3.9082125603864735,
+      "grad_norm": 0.04489309340715408,
+      "learning_rate": 9.903967524880022e-05,
+      "loss": 0.0075,
+      "step": 809
+    },
+    {
+      "epoch": 3.9130434782608696,
+      "grad_norm": 0.05215749889612198,
+      "learning_rate": 9.903308513000798e-05,
+      "loss": 0.0128,
+      "step": 810
+    },
+    {
+      "epoch": 3.917874396135266,
+      "grad_norm": 0.05710824206471443,
+      "learning_rate": 9.902647269726489e-05,
+      "loss": 0.0084,
+      "step": 811
+    },
+    {
+      "epoch": 3.922705314009662,
+      "grad_norm": 0.052041713148355484,
+      "learning_rate": 9.901983795358008e-05,
+      "loss": 0.0122,
+      "step": 812
+    },
+    {
+      "epoch": 3.927536231884058,
+      "grad_norm": 0.043076708912849426,
+      "learning_rate": 9.901318090197291e-05,
+      "loss": 0.008,
+      "step": 813
+    },
+    {
+      "epoch": 3.9323671497584543,
+      "grad_norm": 0.03354997932910919,
+      "learning_rate": 9.900650154547286e-05,
+      "loss": 0.0078,
+      "step": 814
+    },
+    {
+      "epoch": 3.9371980676328504,
+      "grad_norm": 0.04225970059633255,
+      "learning_rate": 9.89997998871196e-05,
+      "loss": 0.0066,
+      "step": 815
+    },
+    {
+      "epoch": 3.942028985507246,
+      "grad_norm": 0.047922004014253616,
+      "learning_rate": 9.899307592996287e-05,
+      "loss": 0.0075,
+      "step": 816
+    },
+    {
+      "epoch": 3.9468599033816423,
+      "grad_norm": 0.06206347793340683,
+      "learning_rate": 9.898632967706264e-05,
+      "loss": 0.0135,
+      "step": 817
+    },
+    {
+      "epoch": 3.9516908212560384,
+      "grad_norm": 0.04242934286594391,
+      "learning_rate": 9.897956113148899e-05,
+      "loss": 0.0085,
+      "step": 818
+    },
+    {
+      "epoch": 3.9565217391304346,
+      "grad_norm": 0.05467364192008972,
+      "learning_rate": 9.897277029632212e-05,
+      "loss": 0.0091,
+      "step": 819
+    },
+    {
+      "epoch": 3.9613526570048307,
+      "grad_norm": 0.05071014538407326,
+      "learning_rate": 9.896595717465243e-05,
+      "loss": 0.0062,
+      "step": 820
+    },
+    {
+      "epoch": 3.966183574879227,
+      "grad_norm": 0.05355323106050491,
+      "learning_rate": 9.89591217695804e-05,
+      "loss": 0.007,
+      "step": 821
+    },
+    {
+      "epoch": 3.971014492753623,
+      "grad_norm": 0.04782043769955635,
+      "learning_rate": 9.895226408421669e-05,
+      "loss": 0.0089,
+      "step": 822
+    },
+    {
+      "epoch": 3.975845410628019,
+      "grad_norm": 0.07460252195596695,
+      "learning_rate": 9.894538412168213e-05,
+      "loss": 0.0076,
+      "step": 823
+    },
+    {
+      "epoch": 3.9806763285024154,
+      "grad_norm": 0.0481514111161232,
+      "learning_rate": 9.89384818851076e-05,
+      "loss": 0.0115,
+      "step": 824
+    },
+    {
+      "epoch": 3.9855072463768115,
+      "grad_norm": 0.045337025076150894,
+      "learning_rate": 9.893155737763419e-05,
+      "loss": 0.0099,
+      "step": 825
+    },
+    {
+      "epoch": 3.9855072463768115,
+      "eval_loss": 0.007348967250436544,
+      "eval_runtime": 20.3101,
+      "eval_samples_per_second": 4.924,
+      "eval_steps_per_second": 0.148,
+      "step": 825
+    },
+    {
+      "epoch": 3.9903381642512077,
+      "grad_norm": 0.07469549030065536,
+      "learning_rate": 9.892461060241312e-05,
+      "loss": 0.0115,
+      "step": 826
+    },
+    {
+      "epoch": 3.995169082125604,
+      "grad_norm": 0.05278422683477402,
+      "learning_rate": 9.891764156260568e-05,
+      "loss": 0.0076,
+      "step": 827
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.09103444963693619,
+      "learning_rate": 9.891065026138338e-05,
+      "loss": 0.0118,
+      "step": 828
+    },
+    {
+      "epoch": 4.004830917874396,
+      "grad_norm": 0.07645177841186523,
+      "learning_rate": 9.890363670192776e-05,
+      "loss": 0.0072,
+      "step": 829
+    },
+    {
+      "epoch": 4.009661835748792,
+      "grad_norm": 0.051178693771362305,
+      "learning_rate": 9.889660088743063e-05,
+      "loss": 0.0074,
+      "step": 830
+    },
+    {
+      "epoch": 4.0144927536231885,
+      "grad_norm": 0.1458490490913391,
+      "learning_rate": 9.888954282109377e-05,
+      "loss": 0.0082,
+      "step": 831
+    },
+    {
+      "epoch": 4.019323671497585,
+      "grad_norm": 0.06586052477359772,
+      "learning_rate": 9.888246250612921e-05,
+      "loss": 0.0073,
+      "step": 832
+    },
+    {
+      "epoch": 4.024154589371981,
+      "grad_norm": 0.12695728242397308,
+      "learning_rate": 9.887535994575902e-05,
+      "loss": 0.0103,
+      "step": 833
+    },
+    {
+      "epoch": 4.028985507246377,
+      "grad_norm": 0.036658771336078644,
+      "learning_rate": 9.886823514321548e-05,
+      "loss": 0.0063,
+      "step": 834
+    },
+    {
+      "epoch": 4.033816425120773,
+      "grad_norm": 0.0739186480641365,
+      "learning_rate": 9.886108810174088e-05,
+      "loss": 0.0096,
+      "step": 835
+    },
+    {
+      "epoch": 4.038647342995169,
+      "grad_norm": 0.0768413096666336,
+      "learning_rate": 9.885391882458773e-05,
+      "loss": 0.0103,
+      "step": 836
+    },
+    {
+      "epoch": 4.043478260869565,
+      "grad_norm": 0.03403015434741974,
+      "learning_rate": 9.88467273150186e-05,
+      "loss": 0.008,
+      "step": 837
+    },
+    {
+      "epoch": 4.048309178743962,
+      "grad_norm": 0.05291101709008217,
+      "learning_rate": 9.883951357630622e-05,
+      "loss": 0.0084,
+      "step": 838
+    },
+    {
+      "epoch": 4.053140096618358,
+      "grad_norm": 0.08055385947227478,
+      "learning_rate": 9.88322776117334e-05,
+      "loss": 0.0105,
+      "step": 839
+    },
+    {
+      "epoch": 4.057971014492754,
+      "grad_norm": 0.04251623898744583,
+      "learning_rate": 9.882501942459308e-05,
+      "loss": 0.0091,
+      "step": 840
+    },
+    {
+      "epoch": 4.06280193236715,
+      "grad_norm": 0.03910037502646446,
+      "learning_rate": 9.881773901818832e-05,
+      "loss": 0.0066,
+      "step": 841
+    },
+    {
+      "epoch": 4.067632850241546,
+      "grad_norm": 0.04807236045598984,
+      "learning_rate": 9.881043639583227e-05,
+      "loss": 0.0089,
+      "step": 842
+    },
+    {
+      "epoch": 4.072463768115942,
+      "grad_norm": 0.030411390587687492,
+      "learning_rate": 9.880311156084823e-05,
+      "loss": 0.0048,
+      "step": 843
+    },
+    {
+      "epoch": 4.0772946859903385,
+      "grad_norm": 0.03571973741054535,
+      "learning_rate": 9.879576451656955e-05,
+      "loss": 0.0085,
+      "step": 844
+    },
+    {
+      "epoch": 4.082125603864735,
+      "grad_norm": 0.04853428527712822,
+      "learning_rate": 9.878839526633974e-05,
+      "loss": 0.009,
+      "step": 845
+    },
+    {
+      "epoch": 4.086956521739131,
+      "grad_norm": 0.04168358072638512,
+      "learning_rate": 9.878100381351239e-05,
+      "loss": 0.0096,
+      "step": 846
+    },
+    {
+      "epoch": 4.091787439613527,
+      "grad_norm": 0.06781593710184097,
+      "learning_rate": 9.877359016145117e-05,
+      "loss": 0.006,
+      "step": 847
+    },
+    {
+      "epoch": 4.096618357487923,
+      "grad_norm": 0.06934653222560883,
+      "learning_rate": 9.876615431352994e-05,
+      "loss": 0.0075,
+      "step": 848
+    },
+    {
+      "epoch": 4.101449275362318,
+      "grad_norm": 0.04383888468146324,
+      "learning_rate": 9.875869627313255e-05,
+      "loss": 0.0078,
+      "step": 849
+    },
+    {
+      "epoch": 4.106280193236715,
+      "grad_norm": 0.06643518805503845,
+      "learning_rate": 9.875121604365302e-05,
+      "loss": 0.0072,
+      "step": 850
+    },
+    {
+      "epoch": 4.106280193236715,
+      "eval_loss": 0.011321146041154861,
+      "eval_runtime": 20.2974,
+      "eval_samples_per_second": 4.927,
+      "eval_steps_per_second": 0.148,
+      "step": 850
+    },
+    {
+      "epoch": 4.111111111111111,
+      "grad_norm": 0.06936809420585632,
+      "learning_rate": 9.874371362849548e-05,
+      "loss": 0.0155,
+      "step": 851
+    },
+    {
+      "epoch": 4.115942028985507,
+      "grad_norm": 0.06398960947990417,
+      "learning_rate": 9.873618903107406e-05,
+      "loss": 0.0071,
+      "step": 852
+    },
+    {
+      "epoch": 4.120772946859903,
+      "grad_norm": 0.05030336230993271,
+      "learning_rate": 9.87286422548131e-05,
+      "loss": 0.0103,
+      "step": 853
+    },
+    {
+      "epoch": 4.125603864734299,
+      "grad_norm": 0.0524723194539547,
+      "learning_rate": 9.872107330314696e-05,
+      "loss": 0.0075,
+      "step": 854
+    },
+    {
+      "epoch": 4.130434782608695,
+      "grad_norm": 0.05538473650813103,
+      "learning_rate": 9.871348217952012e-05,
+      "loss": 0.0069,
+      "step": 855
+    },
+    {
+      "epoch": 4.1352657004830915,
+      "grad_norm": 0.04896744713187218,
+      "learning_rate": 9.870586888738715e-05,
+      "loss": 0.007,
+      "step": 856
+    },
+    {
+      "epoch": 4.140096618357488,
+      "grad_norm": 0.03770635277032852,
+      "learning_rate": 9.869823343021271e-05,
+      "loss": 0.0071,
+      "step": 857
+    },
+    {
+      "epoch": 4.144927536231884,
+      "grad_norm": 0.023480841889977455,
+      "learning_rate": 9.869057581147152e-05,
+      "loss": 0.005,
+      "step": 858
+    },
+    {
+      "epoch": 4.14975845410628,
+      "grad_norm": 0.039378829300403595,
+      "learning_rate": 9.868289603464842e-05,
+      "loss": 0.0063,
+      "step": 859
+    },
+    {
+      "epoch": 4.154589371980676,
+      "grad_norm": 0.03883344307541847,
+      "learning_rate": 9.86751941032383e-05,
+      "loss": 0.0078,
+      "step": 860
+    },
+    {
+      "epoch": 4.159420289855072,
+      "grad_norm": 0.03314659744501114,
+      "learning_rate": 9.866747002074617e-05,
+      "loss": 0.0069,
+      "step": 861
+    },
+    {
+      "epoch": 4.164251207729468,
+      "grad_norm": 0.05089738592505455,
+      "learning_rate": 9.865972379068711e-05,
+      "loss": 0.0097,
+      "step": 862
+    },
+    {
+      "epoch": 4.169082125603865,
+      "grad_norm": 0.05422361195087433,
+      "learning_rate": 9.865195541658623e-05,
+      "loss": 0.0071,
+      "step": 863
+    },
+    {
+      "epoch": 4.173913043478261,
+      "grad_norm": 0.0533452071249485,
+      "learning_rate": 9.86441649019788e-05,
+      "loss": 0.0099,
+      "step": 864
+    },
+    {
+      "epoch": 4.178743961352657,
+      "grad_norm": 0.04176698252558708,
+      "learning_rate": 9.86363522504101e-05,
+      "loss": 0.007,
+      "step": 865
+    },
+    {
+      "epoch": 4.183574879227053,
+      "grad_norm": 0.034925658255815506,
+      "learning_rate": 9.862851746543554e-05,
+      "loss": 0.007,
+      "step": 866
+    },
+    {
+      "epoch": 4.188405797101449,
+      "grad_norm": 0.06717666238546371,
+      "learning_rate": 9.862066055062051e-05,
+      "loss": 0.0058,
+      "step": 867
+    },
+    {
+      "epoch": 4.193236714975845,
+      "grad_norm": 0.05857114866375923,
+      "learning_rate": 9.861278150954059e-05,
+      "loss": 0.0057,
+      "step": 868
+    },
+    {
+      "epoch": 4.1980676328502415,
+      "grad_norm": 0.07392554730176926,
+      "learning_rate": 9.860488034578132e-05,
+      "loss": 0.0081,
+      "step": 869
+    },
+    {
+      "epoch": 4.202898550724638,
+      "grad_norm": 0.0408996120095253,
+      "learning_rate": 9.85969570629384e-05,
+      "loss": 0.0058,
+      "step": 870
+    },
+    {
+      "epoch": 4.207729468599034,
+      "grad_norm": 0.058850813657045364,
+      "learning_rate": 9.858901166461754e-05,
+      "loss": 0.0086,
+      "step": 871
+    },
+    {
+      "epoch": 4.21256038647343,
+      "grad_norm": 0.04763633757829666,
+      "learning_rate": 9.85810441544345e-05,
+      "loss": 0.0049,
+      "step": 872
+    },
+    {
+      "epoch": 4.217391304347826,
+      "grad_norm": 0.04741189256310463,
+      "learning_rate": 9.857305453601517e-05,
+      "loss": 0.0069,
+      "step": 873
+    },
+    {
+      "epoch": 4.222222222222222,
+      "grad_norm": 0.04383055865764618,
+      "learning_rate": 9.856504281299546e-05,
+      "loss": 0.0061,
+      "step": 874
+    },
+    {
+      "epoch": 4.2270531400966185,
+      "grad_norm": 0.047976478934288025,
+      "learning_rate": 9.85570089890213e-05,
+      "loss": 0.0065,
+      "step": 875
+    },
+    {
+      "epoch": 4.2270531400966185,
+      "eval_loss": 0.01065537054091692,
+      "eval_runtime": 20.3101,
+      "eval_samples_per_second": 4.924,
+      "eval_steps_per_second": 0.148,
+      "step": 875
+    },
+    {
+      "epoch": 4.231884057971015,
+      "grad_norm": 0.060130525380373,
+      "learning_rate": 9.854895306774876e-05,
+      "loss": 0.009,
+      "step": 876
+    },
+    {
+      "epoch": 4.236714975845411,
+      "grad_norm": 0.03845841810107231,
+      "learning_rate": 9.854087505284391e-05,
+      "loss": 0.0053,
+      "step": 877
+    },
+    {
+      "epoch": 4.241545893719807,
+      "grad_norm": 0.05535529926419258,
+      "learning_rate": 9.853277494798287e-05,
+      "loss": 0.0067,
+      "step": 878
+    },
+    {
+      "epoch": 4.246376811594203,
+      "grad_norm": 0.036337871104478836,
+      "learning_rate": 9.852465275685187e-05,
+      "loss": 0.0068,
+      "step": 879
+    },
+    {
+      "epoch": 4.251207729468599,
+      "grad_norm": 0.05601054057478905,
+      "learning_rate": 9.851650848314713e-05,
+      "loss": 0.0136,
+      "step": 880
+    },
+    {
+      "epoch": 4.256038647342995,
+      "grad_norm": 0.04372893646359444,
+      "learning_rate": 9.850834213057494e-05,
+      "loss": 0.0062,
+      "step": 881
+    },
+    {
+      "epoch": 4.260869565217392,
+      "grad_norm": 0.03550177812576294,
+      "learning_rate": 9.850015370285164e-05,
+      "loss": 0.0051,
+      "step": 882
+    },
+    {
+      "epoch": 4.265700483091788,
+      "grad_norm": 0.04691091179847717,
+      "learning_rate": 9.84919432037036e-05,
+      "loss": 0.0082,
+      "step": 883
+    },
+    {
+      "epoch": 4.270531400966184,
+      "grad_norm": 0.04239903762936592,
+      "learning_rate": 9.84837106368673e-05,
+      "loss": 0.0061,
+      "step": 884
+    },
+    {
+      "epoch": 4.27536231884058,
+      "grad_norm": 0.05059480294585228,
+      "learning_rate": 9.847545600608917e-05,
+      "loss": 0.0054,
+      "step": 885
+    },
+    {
+      "epoch": 4.280193236714976,
+      "grad_norm": 0.03358094394207001,
+      "learning_rate": 9.846717931512573e-05,
+      "loss": 0.0081,
+      "step": 886
+    },
+    {
+      "epoch": 4.285024154589372,
+      "grad_norm": 0.03535458818078041,
+      "learning_rate": 9.845888056774354e-05,
+      "loss": 0.0054,
+      "step": 887
+    },
+    {
+      "epoch": 4.2898550724637685,
+      "grad_norm": 0.03758447989821434,
+      "learning_rate": 9.845055976771919e-05,
+      "loss": 0.0057,
+      "step": 888
+    },
+    {
+      "epoch": 4.294685990338165,
+      "grad_norm": 0.03534655645489693,
+      "learning_rate": 9.844221691883929e-05,
+      "loss": 0.0086,
+      "step": 889
+    },
+    {
+      "epoch": 4.29951690821256,
+      "grad_norm": 0.05712832883000374,
+      "learning_rate": 9.843385202490051e-05,
+      "loss": 0.0086,
+      "step": 890
+    },
+    {
+      "epoch": 4.304347826086957,
+      "grad_norm": 0.047709014266729355,
+      "learning_rate": 9.842546508970955e-05,
+      "loss": 0.0113,
+      "step": 891
+    },
+    {
+      "epoch": 4.309178743961352,
+      "grad_norm": 0.04342706501483917,
+      "learning_rate": 9.841705611708311e-05,
+      "loss": 0.009,
+      "step": 892
+    },
+    {
+      "epoch": 4.314009661835748,
+      "grad_norm": 0.04739096015691757,
+      "learning_rate": 9.840862511084798e-05,
+      "loss": 0.0113,
+      "step": 893
+    },
+    {
+      "epoch": 4.318840579710145,
+      "grad_norm": 0.03753174841403961,
+      "learning_rate": 9.840017207484089e-05,
+      "loss": 0.0081,
+      "step": 894
+    },
+    {
+      "epoch": 4.323671497584541,
+      "grad_norm": 0.06840533018112183,
+      "learning_rate": 9.839169701290868e-05,
+      "loss": 0.0084,
+      "step": 895
+    },
+    {
+      "epoch": 4.328502415458937,
+      "grad_norm": 0.052864886820316315,
+      "learning_rate": 9.838319992890816e-05,
+      "loss": 0.0069,
+      "step": 896
+    },
+    {
+      "epoch": 4.333333333333333,
+      "grad_norm": 0.04285022243857384,
+      "learning_rate": 9.837468082670617e-05,
+      "loss": 0.0085,
+      "step": 897
+    },
+    {
+      "epoch": 4.338164251207729,
+      "grad_norm": 0.029472695663571358,
+      "learning_rate": 9.836613971017961e-05,
+      "loss": 0.0057,
+      "step": 898
+    },
+    {
+      "epoch": 4.342995169082125,
+      "grad_norm": 0.05671832337975502,
+      "learning_rate": 9.835757658321533e-05,
+      "loss": 0.0114,
+      "step": 899
+    },
+    {
+      "epoch": 4.3478260869565215,
+      "grad_norm": 0.06673644483089447,
+      "learning_rate": 9.834899144971025e-05,
+      "loss": 0.0079,
+      "step": 900
+    },
+    {
+      "epoch": 4.3478260869565215,
+      "eval_loss": 0.009663479402661324,
+      "eval_runtime": 20.3004,
+      "eval_samples_per_second": 4.926,
+      "eval_steps_per_second": 0.148,
+      "step": 900
+    },
+    {
+      "epoch": 4.352657004830918,
+      "grad_norm": 0.046011269092559814,
+      "learning_rate": 9.834038431357129e-05,
+      "loss": 0.0067,
+      "step": 901
+    },
+    {
+      "epoch": 4.357487922705314,
+      "grad_norm": 0.05855799466371536,
+      "learning_rate": 9.833175517871538e-05,
+      "loss": 0.0062,
+      "step": 902
+    },
+    {
+      "epoch": 4.36231884057971,
+      "grad_norm": 0.04528895020484924,
+      "learning_rate": 9.832310404906946e-05,
+      "loss": 0.0087,
+      "step": 903
+    },
+    {
+      "epoch": 4.367149758454106,
+      "grad_norm": 0.07774829864501953,
+      "learning_rate": 9.831443092857049e-05,
+      "loss": 0.0069,
+      "step": 904
+    },
+    {
+      "epoch": 4.371980676328502,
+      "grad_norm": 0.06823405623435974,
+      "learning_rate": 9.830573582116542e-05,
+      "loss": 0.0146,
+      "step": 905
+    },
+    {
+      "epoch": 4.3768115942028984,
+      "grad_norm": 0.05296029895544052,
+      "learning_rate": 9.829701873081122e-05,
+      "loss": 0.0094,
+      "step": 906
+    },
+    {
+      "epoch": 4.381642512077295,
+      "grad_norm": 0.05371993035078049,
+      "learning_rate": 9.828827966147485e-05,
+      "loss": 0.0075,
+      "step": 907
+    },
+    {
+      "epoch": 4.386473429951691,
+      "grad_norm": 0.04041058197617531,
+      "learning_rate": 9.827951861713329e-05,
+      "loss": 0.0063,
+      "step": 908
+    },
+    {
+      "epoch": 4.391304347826087,
+      "grad_norm": 0.04803326725959778,
+      "learning_rate": 9.827073560177351e-05,
+      "loss": 0.0068,
+      "step": 909
+    },
+    {
+      "epoch": 4.396135265700483,
+      "grad_norm": 0.05994880944490433,
+      "learning_rate": 9.826193061939249e-05,
+      "loss": 0.0074,
+      "step": 910
+    },
+    {
+      "epoch": 4.400966183574879,
+      "grad_norm": 0.04290630295872688,
+      "learning_rate": 9.825310367399716e-05,
+      "loss": 0.0097,
+      "step": 911
+    },
+    {
+      "epoch": 4.405797101449275,
+      "grad_norm": 0.0505148209631443,
+      "learning_rate": 9.824425476960453e-05,
+      "loss": 0.0069,
+      "step": 912
+    },
+    {
+      "epoch": 4.4106280193236715,
+      "grad_norm": 0.04741643741726875,
+      "learning_rate": 9.823538391024151e-05,
+      "loss": 0.006,
+      "step": 913
+    },
+    {
+      "epoch": 4.415458937198068,
+      "grad_norm": 0.0431365892291069,
+      "learning_rate": 9.822649109994508e-05,
+      "loss": 0.0064,
+      "step": 914
+    },
+    {
+      "epoch": 4.420289855072464,
+      "grad_norm": 0.05312202498316765,
+      "learning_rate": 9.821757634276217e-05,
+      "loss": 0.0074,
+      "step": 915
+    },
+    {
+      "epoch": 4.42512077294686,
+      "grad_norm": 0.05456024035811424,
+      "learning_rate": 9.820863964274969e-05,
+      "loss": 0.0053,
+      "step": 916
+    },
+    {
+      "epoch": 4.429951690821256,
+      "grad_norm": 0.043725527822971344,
+      "learning_rate": 9.819968100397455e-05,
+      "loss": 0.008,
+      "step": 917
+    },
+    {
+      "epoch": 4.434782608695652,
+      "grad_norm": 0.05771199241280556,
+      "learning_rate": 9.819070043051366e-05,
+      "loss": 0.0071,
+      "step": 918
+    },
+    {
+      "epoch": 4.4396135265700485,
+      "grad_norm": 0.07217767834663391,
+      "learning_rate": 9.818169792645388e-05,
+      "loss": 0.005,
+      "step": 919
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 0.14544545114040375,
+      "learning_rate": 9.817267349589205e-05,
+      "loss": 0.0075,
+      "step": 920
+    },
+    {
+      "epoch": 4.449275362318841,
+      "grad_norm": 0.034144334495067596,
+      "learning_rate": 9.816362714293504e-05,
+      "loss": 0.0056,
+      "step": 921
+    },
+    {
+      "epoch": 4.454106280193237,
+      "grad_norm": 0.04955586418509483,
+      "learning_rate": 9.815455887169965e-05,
+      "loss": 0.006,
+      "step": 922
+    },
+    {
+      "epoch": 4.458937198067633,
+      "grad_norm": 0.04086367413401604,
+      "learning_rate": 9.814546868631264e-05,
+      "loss": 0.0063,
+      "step": 923
+    },
+    {
+      "epoch": 4.463768115942029,
+      "grad_norm": 0.03800726681947708,
+      "learning_rate": 9.813635659091078e-05,
+      "loss": 0.007,
+      "step": 924
+    },
+    {
+      "epoch": 4.468599033816425,
+      "grad_norm": 0.06027813255786896,
+      "learning_rate": 9.81272225896408e-05,
+      "loss": 0.0081,
+      "step": 925
+    },
+    {
+      "epoch": 4.468599033816425,
+      "eval_loss": 0.010265327990055084,
+      "eval_runtime": 20.306,
+      "eval_samples_per_second": 4.925,
+      "eval_steps_per_second": 0.148,
+      "step": 925
+    },
+    {
+      "epoch": 4.468599033816425,
+      "step": 925,
+      "total_flos": 3.488842745919701e+18,
+      "train_loss": 0.025700105609926017,
+      "train_runtime": 24639.0069,
+      "train_samples_per_second": 10.045,
+      "train_steps_per_second": 0.21
     }
   ],
   "logging_steps": 1,
-  "max_steps": 1,
+  "max_steps": 5175,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 25,
   "save_steps": 100,
   "stateful_callbacks": {
     "EarlyStoppingCallback": {
@@ -37,7 +6801,7 @@
         "early_stopping_threshold": 0.0
       },
       "attributes": {
-        "early_stopping_patience_counter": 0
+        "early_stopping_patience_counter": 4
       }
     },
     "TrainerControl": {
@@ -46,12 +6810,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": true
+        "should_training_stop": false
       },
       "attributes": {}
     }
   },
-  "total_flos": 467769719488512.0,
+  "total_flos": 3.488842745919701e+18,
   "train_batch_size": 48,
   "trial_name": null,
   "trial_params": null