{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.7476476710350247,
  "eval_steps": 100,
  "global_step": 2200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 18.108646392822266,
      "learning_rate": 1e-05,
      "loss": 21.0115,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.891128540039062,
      "learning_rate": 1e-05,
      "loss": 20.6659,
      "step": 10
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.71453285217285,
      "learning_rate": 1e-05,
      "loss": 20.2163,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.3765926361084,
      "learning_rate": 1e-05,
      "loss": 18.8949,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.546733856201172,
      "learning_rate": 1e-05,
      "loss": 17.7048,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.177675247192383,
      "learning_rate": 1e-05,
      "loss": 16.1373,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.215681076049805,
      "learning_rate": 1e-05,
      "loss": 15.0542,
      "step": 60
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.990703582763672,
      "learning_rate": 1e-05,
      "loss": 13.6239,
      "step": 70
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.313385009765625,
      "learning_rate": 1e-05,
      "loss": 12.5539,
      "step": 80
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.427459716796875,
      "learning_rate": 1e-05,
      "loss": 11.1956,
      "step": 90
    },
    {
      "epoch": 0.03,
      "grad_norm": 38.51707077026367,
      "learning_rate": 1e-05,
      "loss": 9.8649,
      "step": 100
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.18704605102539,
      "learning_rate": 1e-05,
      "loss": 8.7252,
      "step": 110
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.252742767333984,
      "learning_rate": 1e-05,
      "loss": 7.8294,
      "step": 120
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.109561920166016,
      "learning_rate": 1e-05,
      "loss": 6.7392,
      "step": 130
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.344636917114258,
      "learning_rate": 1e-05,
      "loss": 5.7289,
      "step": 140
    },
    {
      "epoch": 0.05,
      "grad_norm": 73.92993927001953,
      "learning_rate": 1e-05,
      "loss": 4.6879,
      "step": 150
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.572184562683105,
      "learning_rate": 1e-05,
      "loss": 3.5502,
      "step": 160
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.667272567749023,
      "learning_rate": 1e-05,
      "loss": 2.801,
      "step": 170
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.249564170837402,
      "learning_rate": 1e-05,
      "loss": 2.3337,
      "step": 180
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.55443000793457,
      "learning_rate": 1e-05,
      "loss": 1.9857,
      "step": 190
    },
    {
      "epoch": 0.07,
      "grad_norm": 248.29791259765625,
      "learning_rate": 1e-05,
      "loss": 1.7209,
      "step": 200
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2151737213134766,
      "learning_rate": 1e-05,
      "loss": 1.5336,
      "step": 210
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.544109582901001,
      "learning_rate": 1e-05,
      "loss": 1.3586,
      "step": 220
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0416696071624756,
      "learning_rate": 1e-05,
      "loss": 1.3511,
      "step": 230
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0984798669815063,
      "learning_rate": 1e-05,
      "loss": 1.2404,
      "step": 240
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1240887641906738,
      "learning_rate": 1e-05,
      "loss": 1.1461,
      "step": 250
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9642849564552307,
      "learning_rate": 1e-05,
      "loss": 1.1479,
      "step": 260
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.870434761047363,
      "learning_rate": 1e-05,
      "loss": 1.0765,
      "step": 270
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7763110399246216,
      "learning_rate": 1e-05,
      "loss": 1.0369,
      "step": 280
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.542849063873291,
      "learning_rate": 1e-05,
      "loss": 1.0358,
      "step": 290
    },
    {
      "epoch": 0.1,
      "grad_norm": 41.260128021240234,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 300
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45591601729393005,
      "learning_rate": 1e-05,
      "loss": 1.0068,
      "step": 310
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4061857759952545,
      "learning_rate": 1e-05,
      "loss": 0.9619,
      "step": 320
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6108375787734985,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 330
    },
    {
      "epoch": 0.12,
      "grad_norm": 54.73992156982422,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 340
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1031421422958374,
      "learning_rate": 1e-05,
      "loss": 0.9131,
      "step": 350
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8087090849876404,
      "learning_rate": 1e-05,
      "loss": 0.8975,
      "step": 360
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.830881655216217,
      "learning_rate": 1e-05,
      "loss": 0.8793,
      "step": 370
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6168428659439087,
      "learning_rate": 1e-05,
      "loss": 0.8614,
      "step": 380
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6438579559326172,
      "learning_rate": 1e-05,
      "loss": 0.844,
      "step": 390
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7063890099525452,
      "learning_rate": 1e-05,
      "loss": 0.8112,
      "step": 400
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6345189809799194,
      "learning_rate": 1e-05,
      "loss": 0.7793,
      "step": 410
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7021321654319763,
      "learning_rate": 1e-05,
      "loss": 0.7494,
      "step": 420
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6041214466094971,
      "learning_rate": 1e-05,
      "loss": 0.7391,
      "step": 430
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7526829242706299,
      "learning_rate": 1e-05,
      "loss": 0.6812,
      "step": 440
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9282905459403992,
      "learning_rate": 1e-05,
      "loss": 0.7046,
      "step": 450
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7867602109909058,
      "learning_rate": 1e-05,
      "loss": 0.6556,
      "step": 460
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7166017889976501,
      "learning_rate": 1e-05,
      "loss": 0.6598,
      "step": 470
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48004817962646484,
      "learning_rate": 1e-05,
      "loss": 0.6207,
      "step": 480
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5272298455238342,
      "learning_rate": 1e-05,
      "loss": 0.6203,
      "step": 490
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7496746182441711,
      "learning_rate": 1e-05,
      "loss": 0.5562,
      "step": 500
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7099195718765259,
      "learning_rate": 1e-05,
      "loss": 0.532,
      "step": 510
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6780660152435303,
      "learning_rate": 1e-05,
      "loss": 0.5432,
      "step": 520
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4947499632835388,
      "learning_rate": 1e-05,
      "loss": 0.5428,
      "step": 530
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.102908968925476,
      "learning_rate": 1e-05,
      "loss": 0.5605,
      "step": 540
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.204094886779785,
      "learning_rate": 1e-05,
      "loss": 0.5213,
      "step": 550
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5231891870498657,
      "learning_rate": 1e-05,
      "loss": 0.4716,
      "step": 560
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3716018199920654,
      "learning_rate": 1e-05,
      "loss": 0.4907,
      "step": 570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6284762620925903,
      "learning_rate": 1e-05,
      "loss": 0.4753,
      "step": 580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4310474097728729,
      "learning_rate": 1e-05,
      "loss": 0.4351,
      "step": 590
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1138970851898193,
      "learning_rate": 1e-05,
      "loss": 0.4556,
      "step": 600
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0587685108184814,
      "learning_rate": 1e-05,
      "loss": 0.4312,
      "step": 610
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4586256742477417,
      "learning_rate": 1e-05,
      "loss": 0.4448,
      "step": 620
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5893386006355286,
      "learning_rate": 1e-05,
      "loss": 0.4097,
      "step": 630
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6643804311752319,
      "learning_rate": 1e-05,
      "loss": 0.3823,
      "step": 640
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.758120059967041,
      "learning_rate": 1e-05,
      "loss": 0.3963,
      "step": 650
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0745816230773926,
      "learning_rate": 1e-05,
      "loss": 0.3846,
      "step": 660
    },
    {
      "epoch": 0.23,
      "grad_norm": 53.08203887939453,
      "learning_rate": 1e-05,
      "loss": 0.3815,
      "step": 670
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9295670390129089,
      "learning_rate": 1e-05,
      "loss": 0.3577,
      "step": 680
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4713806211948395,
      "learning_rate": 1e-05,
      "loss": 0.3429,
      "step": 690
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.426747590303421,
      "learning_rate": 1e-05,
      "loss": 0.3568,
      "step": 700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.32080256938934326,
      "learning_rate": 1e-05,
      "loss": 0.3549,
      "step": 710
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3621009290218353,
      "learning_rate": 1e-05,
      "loss": 0.3292,
      "step": 720
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7618722915649414,
      "learning_rate": 1e-05,
      "loss": 0.3325,
      "step": 730
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.679786205291748,
      "learning_rate": 1e-05,
      "loss": 0.3231,
      "step": 740
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5011792182922363,
      "learning_rate": 1e-05,
      "loss": 0.3287,
      "step": 750
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.47687655687332153,
      "learning_rate": 1e-05,
      "loss": 0.302,
      "step": 760
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5612205862998962,
      "learning_rate": 1e-05,
      "loss": 0.3053,
      "step": 770
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2981828451156616,
      "learning_rate": 1e-05,
      "loss": 0.2853,
      "step": 780
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4281419813632965,
      "learning_rate": 1e-05,
      "loss": 0.2746,
      "step": 790
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6802565455436707,
      "learning_rate": 1e-05,
      "loss": 0.2708,
      "step": 800
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.3523566722869873,
      "learning_rate": 1e-05,
      "loss": 0.5448,
      "step": 810
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0021793842315674,
      "learning_rate": 1e-05,
      "loss": 0.3757,
      "step": 820
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0229527950286865,
      "learning_rate": 1e-05,
      "loss": 0.3256,
      "step": 830
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5520893931388855,
      "learning_rate": 1e-05,
      "loss": 0.292,
      "step": 840
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5874707698822021,
      "learning_rate": 1e-05,
      "loss": 0.2686,
      "step": 850
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38817650079727173,
      "learning_rate": 1e-05,
      "loss": 0.2527,
      "step": 860
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7592063546180725,
      "learning_rate": 1e-05,
      "loss": 0.243,
      "step": 870
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42729002237319946,
      "learning_rate": 1e-05,
      "loss": 0.2297,
      "step": 880
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3907771706581116,
      "learning_rate": 1e-05,
      "loss": 0.223,
      "step": 890
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38824358582496643,
      "learning_rate": 1e-05,
      "loss": 0.2173,
      "step": 900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2964054048061371,
      "learning_rate": 1e-05,
      "loss": 0.2081,
      "step": 910
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37019163370132446,
      "learning_rate": 1e-05,
      "loss": 0.2032,
      "step": 920
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2846088111400604,
      "learning_rate": 1e-05,
      "loss": 0.1971,
      "step": 930
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2604421377182007,
      "learning_rate": 1e-05,
      "loss": 0.1936,
      "step": 940
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3167709410190582,
      "learning_rate": 1e-05,
      "loss": 0.1894,
      "step": 950
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4469132423400879,
      "learning_rate": 1e-05,
      "loss": 0.1817,
      "step": 960
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3273838758468628,
      "learning_rate": 1e-05,
      "loss": 0.1769,
      "step": 970
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3062090277671814,
      "learning_rate": 1e-05,
      "loss": 0.1758,
      "step": 980
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1933787316083908,
      "learning_rate": 1e-05,
      "loss": 0.1701,
      "step": 990
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2302626371383667,
      "learning_rate": 1e-05,
      "loss": 0.1655,
      "step": 1000
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1817045956850052,
      "learning_rate": 1e-05,
      "loss": 0.1649,
      "step": 1010
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2034452259540558,
      "learning_rate": 1e-05,
      "loss": 0.1596,
      "step": 1020
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.393022745847702,
      "learning_rate": 1e-05,
      "loss": 0.1565,
      "step": 1030
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.45332619547843933,
      "learning_rate": 1e-05,
      "loss": 0.1525,
      "step": 1040
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3920886516571045,
      "learning_rate": 1e-05,
      "loss": 0.1519,
      "step": 1050
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2492988407611847,
      "learning_rate": 1e-05,
      "loss": 0.1486,
      "step": 1060
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.18808089196681976,
      "learning_rate": 1e-05,
      "loss": 0.144,
      "step": 1070
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17545698583126068,
      "learning_rate": 1e-05,
      "loss": 0.1436,
      "step": 1080
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39826542139053345,
      "learning_rate": 1e-05,
      "loss": 0.142,
      "step": 1090
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25920161604881287,
      "learning_rate": 1e-05,
      "loss": 0.1377,
      "step": 1100
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24598653614521027,
      "learning_rate": 1e-05,
      "loss": 0.1343,
      "step": 1110
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3771628439426422,
      "learning_rate": 1e-05,
      "loss": 0.1308,
      "step": 1120
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.17581653594970703,
      "learning_rate": 1e-05,
      "loss": 0.1304,
      "step": 1130
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29346442222595215,
      "learning_rate": 1e-05,
      "loss": 0.1272,
      "step": 1140
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.32129135727882385,
      "learning_rate": 1e-05,
      "loss": 0.1249,
      "step": 1150
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24236874282360077,
      "learning_rate": 1e-05,
      "loss": 0.1246,
      "step": 1160
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2639343738555908,
      "learning_rate": 1e-05,
      "loss": 0.1221,
      "step": 1170
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1657143533229828,
      "learning_rate": 1e-05,
      "loss": 0.1196,
      "step": 1180
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20446737110614777,
      "learning_rate": 1e-05,
      "loss": 0.1159,
      "step": 1190
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.20889534056186676,
      "learning_rate": 1e-05,
      "loss": 0.1147,
      "step": 1200
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34177979826927185,
      "learning_rate": 1e-05,
      "loss": 0.1141,
      "step": 1210
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.12536485493183136,
      "learning_rate": 1e-05,
      "loss": 0.1126,
      "step": 1220
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20015142858028412,
      "learning_rate": 1e-05,
      "loss": 0.1099,
      "step": 1230
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.29639932513237,
      "learning_rate": 1e-05,
      "loss": 0.1102,
      "step": 1240
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.41865894198417664,
      "learning_rate": 1e-05,
      "loss": 0.1085,
      "step": 1250
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.14246885478496552,
      "learning_rate": 1e-05,
      "loss": 0.1054,
      "step": 1260
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.29295045137405396,
      "learning_rate": 1e-05,
      "loss": 0.1048,
      "step": 1270
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5125443339347839,
      "learning_rate": 1e-05,
      "loss": 0.1038,
      "step": 1280
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7473035454750061,
      "learning_rate": 1e-05,
      "loss": 0.1025,
      "step": 1290
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7184324264526367,
      "learning_rate": 1e-05,
      "loss": 0.101,
      "step": 1300
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2857365906238556,
      "learning_rate": 1e-05,
      "loss": 0.0988,
      "step": 1310
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.15471303462982178,
      "learning_rate": 1e-05,
      "loss": 0.0987,
      "step": 1320
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5794267654418945,
      "learning_rate": 1e-05,
      "loss": 0.0986,
      "step": 1330
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3085970878601074,
      "learning_rate": 1e-05,
      "loss": 0.0964,
      "step": 1340
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38039007782936096,
      "learning_rate": 1e-05,
      "loss": 0.0939,
      "step": 1350
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20592401921749115,
      "learning_rate": 1e-05,
      "loss": 0.0932,
      "step": 1360
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20163756608963013,
      "learning_rate": 1e-05,
      "loss": 0.0931,
      "step": 1370
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3857424855232239,
      "learning_rate": 1e-05,
      "loss": 0.0913,
      "step": 1380
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.13959680497646332,
      "learning_rate": 1e-05,
      "loss": 0.092,
      "step": 1390
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4001765251159668,
      "learning_rate": 1e-05,
      "loss": 0.0902,
      "step": 1400
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28235796093940735,
      "learning_rate": 1e-05,
      "loss": 0.0896,
      "step": 1410
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4206916391849518,
      "learning_rate": 1e-05,
      "loss": 0.0874,
      "step": 1420
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20532837510108948,
      "learning_rate": 1e-05,
      "loss": 0.0859,
      "step": 1430
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5433242917060852,
      "learning_rate": 1e-05,
      "loss": 0.0868,
      "step": 1440
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19398020207881927,
      "learning_rate": 1e-05,
      "loss": 0.0862,
      "step": 1450
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8159850835800171,
      "learning_rate": 1e-05,
      "loss": 0.0848,
      "step": 1460
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.18659061193466187,
      "learning_rate": 1e-05,
      "loss": 0.0839,
      "step": 1470
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2018926590681076,
      "learning_rate": 1e-05,
      "loss": 0.0826,
      "step": 1480
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2839762568473816,
      "learning_rate": 1e-05,
      "loss": 0.0824,
      "step": 1490
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1700347363948822,
      "learning_rate": 1e-05,
      "loss": 0.0814,
      "step": 1500
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.13129307329654694,
      "learning_rate": 1e-05,
      "loss": 0.0815,
      "step": 1510
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.397320032119751,
      "learning_rate": 1e-05,
      "loss": 0.0805,
      "step": 1520
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2305113822221756,
      "learning_rate": 1e-05,
      "loss": 0.0789,
      "step": 1530
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37059059739112854,
      "learning_rate": 1e-05,
      "loss": 0.0775,
      "step": 1540
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2454056739807129,
      "learning_rate": 1e-05,
      "loss": 0.0781,
      "step": 1550
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.340133398771286,
      "learning_rate": 1e-05,
      "loss": 0.0778,
      "step": 1560
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.250377893447876,
      "learning_rate": 1e-05,
      "loss": 0.0753,
      "step": 1570
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18379095196723938,
      "learning_rate": 1e-05,
      "loss": 0.0755,
      "step": 1580
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.16987499594688416,
      "learning_rate": 1e-05,
      "loss": 0.0743,
      "step": 1590
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1600923091173172,
      "learning_rate": 1e-05,
      "loss": 0.0739,
      "step": 1600
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.15362544357776642,
      "learning_rate": 1e-05,
      "loss": 0.0732,
      "step": 1610
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.40313267707824707,
      "learning_rate": 1e-05,
      "loss": 0.0726,
      "step": 1620
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43576550483703613,
      "learning_rate": 1e-05,
      "loss": 0.0722,
      "step": 1630
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.255543977022171,
      "learning_rate": 1e-05,
      "loss": 0.0709,
      "step": 1640
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.23503994941711426,
      "learning_rate": 1e-05,
      "loss": 0.0719,
      "step": 1650
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.27073901891708374,
      "learning_rate": 1e-05,
      "loss": 0.0702,
      "step": 1660
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.15940971672534943,
      "learning_rate": 1e-05,
      "loss": 0.0699,
      "step": 1670
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3071956932544708,
      "learning_rate": 1e-05,
      "loss": 0.0701,
      "step": 1680
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.13690294325351715,
      "learning_rate": 1e-05,
      "loss": 0.0687,
      "step": 1690
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11188670992851257,
      "learning_rate": 1e-05,
      "loss": 0.0685,
      "step": 1700
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.22904923558235168,
      "learning_rate": 1e-05,
      "loss": 0.069,
      "step": 1710
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.15047454833984375,
      "learning_rate": 1e-05,
      "loss": 0.0676,
      "step": 1720
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1788366287946701,
      "learning_rate": 1e-05,
      "loss": 0.0663,
      "step": 1730
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1737486571073532,
      "learning_rate": 1e-05,
      "loss": 0.0665,
      "step": 1740
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18163177371025085,
      "learning_rate": 1e-05,
      "loss": 0.066,
      "step": 1750
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.15237963199615479,
      "learning_rate": 1e-05,
      "loss": 0.0647,
      "step": 1760
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1539287269115448,
      "learning_rate": 1e-05,
      "loss": 0.0655,
      "step": 1770
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.13951052725315094,
      "learning_rate": 1e-05,
      "loss": 0.0649,
      "step": 1780
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1300867348909378,
      "learning_rate": 1e-05,
      "loss": 0.0641,
      "step": 1790
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12030315399169922,
      "learning_rate": 1e-05,
      "loss": 0.0636,
      "step": 1800
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.10913581401109695,
      "learning_rate": 1e-05,
      "loss": 0.0636,
      "step": 1810
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.17842526733875275,
      "learning_rate": 1e-05,
      "loss": 0.0637,
      "step": 1820
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20770031213760376,
      "learning_rate": 1e-05,
      "loss": 0.0632,
      "step": 1830
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12772506475448608,
      "learning_rate": 1e-05,
      "loss": 0.0625,
      "step": 1840
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10420256853103638,
      "learning_rate": 1e-05,
      "loss": 0.0625,
      "step": 1850
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2480468451976776,
      "learning_rate": 1e-05,
      "loss": 0.062,
      "step": 1860
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17825984954833984,
      "learning_rate": 1e-05,
      "loss": 0.0605,
      "step": 1870
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.12127891182899475,
      "learning_rate": 1e-05,
      "loss": 0.0608,
      "step": 1880
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.09041409939527512,
      "learning_rate": 1e-05,
      "loss": 0.0602,
      "step": 1890
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.10765266418457031,
      "learning_rate": 1e-05,
      "loss": 0.0604,
      "step": 1900
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12556296586990356,
      "learning_rate": 1e-05,
      "loss": 0.0592,
      "step": 1910
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2558988928794861,
      "learning_rate": 1e-05,
      "loss": 0.0598,
      "step": 1920
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12372375279664993,
      "learning_rate": 1e-05,
      "loss": 0.0582,
      "step": 1930
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.0909271314740181,
      "learning_rate": 1e-05,
      "loss": 0.0586,
      "step": 1940
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19711463153362274,
      "learning_rate": 1e-05,
      "loss": 0.058,
      "step": 1950
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.13144218921661377,
      "learning_rate": 1e-05,
      "loss": 0.0592,
      "step": 1960
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.10313715785741806,
      "learning_rate": 1e-05,
      "loss": 0.0577,
      "step": 1970
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.15561428666114807,
      "learning_rate": 1e-05,
      "loss": 0.0581,
      "step": 1980
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1571255624294281,
      "learning_rate": 1e-05,
      "loss": 0.0576,
      "step": 1990
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34297963976860046,
      "learning_rate": 1e-05,
      "loss": 0.0574,
      "step": 2000
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3346453011035919,
      "learning_rate": 1e-05,
      "loss": 0.0573,
      "step": 2010
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.09346891939640045,
      "learning_rate": 1e-05,
      "loss": 0.0558,
      "step": 2020
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3849794566631317,
      "learning_rate": 1e-05,
      "loss": 0.0576,
      "step": 2030
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3006366789340973,
      "learning_rate": 1e-05,
      "loss": 0.0561,
      "step": 2040
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.08292259275913239,
      "learning_rate": 1e-05,
      "loss": 0.0567,
      "step": 2050
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37851089239120483,
      "learning_rate": 1e-05,
      "loss": 0.0556,
      "step": 2060
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2993626892566681,
      "learning_rate": 1e-05,
      "loss": 0.0551,
      "step": 2070
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3539237976074219,
      "learning_rate": 1e-05,
      "loss": 0.0556,
      "step": 2080
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6489169597625732,
      "learning_rate": 1e-05,
      "loss": 0.0556,
      "step": 2090
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20589084923267365,
      "learning_rate": 1e-05,
      "loss": 0.0545,
      "step": 2100
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.08394920825958252,
      "learning_rate": 1e-05,
      "loss": 0.0542,
      "step": 2110
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.0861760601401329,
      "learning_rate": 1e-05,
      "loss": 0.0534,
      "step": 2120
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.22998709976673126,
      "learning_rate": 1e-05,
      "loss": 0.0538,
      "step": 2130
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1493302583694458,
      "learning_rate": 1e-05,
      "loss": 0.0542,
      "step": 2140
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11739397794008255,
      "learning_rate": 1e-05,
      "loss": 0.0539,
      "step": 2150
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.10503773391246796,
      "learning_rate": 1e-05,
      "loss": 0.053,
      "step": 2160
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2653915286064148,
      "learning_rate": 1e-05,
      "loss": 0.0533,
      "step": 2170
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1554844081401825,
      "learning_rate": 1e-05,
      "loss": 0.0528,
      "step": 2180
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20956221222877502,
      "learning_rate": 1e-05,
      "loss": 0.0524,
      "step": 2190
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.07734131067991257,
      "learning_rate": 1e-05,
      "loss": 0.0524,
      "step": 2200
    }
  ],
  "logging_steps": 10,
  "max_steps": 17652,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 200,
  "total_flos": 3.689592339116851e+18,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}