{
  "best_metric": 1.6775344610214233,
  "best_model_checkpoint": "miner_id_24/checkpoint-500",
  "epoch": 4.582546494992847,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0022889842632331904,
      "grad_norm": 0.09733663499355316,
      "learning_rate": 2e-05,
      "loss": 1.8105,
      "step": 1
    },
    {
      "epoch": 0.0022889842632331904,
      "eval_loss": 2.077899694442749,
      "eval_runtime": 24.8713,
      "eval_samples_per_second": 29.592,
      "eval_steps_per_second": 14.796,
      "step": 1
    },
    {
      "epoch": 0.004577968526466381,
      "grad_norm": 0.1035127118229866,
      "learning_rate": 4e-05,
      "loss": 1.9178,
      "step": 2
    },
    {
      "epoch": 0.0068669527896995704,
      "grad_norm": 0.1238258108496666,
      "learning_rate": 6e-05,
      "loss": 1.9348,
      "step": 3
    },
    {
      "epoch": 0.009155937052932762,
      "grad_norm": 0.12123893946409225,
      "learning_rate": 8e-05,
      "loss": 1.8745,
      "step": 4
    },
    {
      "epoch": 0.011444921316165951,
      "grad_norm": 0.15163357555866241,
      "learning_rate": 0.0001,
      "loss": 1.9241,
      "step": 5
    },
    {
      "epoch": 0.013733905579399141,
      "grad_norm": 0.16310647130012512,
      "learning_rate": 0.00012,
      "loss": 1.9244,
      "step": 6
    },
    {
      "epoch": 0.01602288984263233,
      "grad_norm": 0.18320989608764648,
      "learning_rate": 0.00014,
      "loss": 1.8611,
      "step": 7
    },
    {
      "epoch": 0.018311874105865523,
      "grad_norm": 0.19893847405910492,
      "learning_rate": 0.00016,
      "loss": 1.8367,
      "step": 8
    },
    {
      "epoch": 0.020600858369098713,
      "grad_norm": 0.20384521782398224,
      "learning_rate": 0.00018,
      "loss": 1.8406,
      "step": 9
    },
    {
      "epoch": 0.022889842632331903,
      "grad_norm": 0.2371494024991989,
      "learning_rate": 0.0002,
      "loss": 1.7277,
      "step": 10
    },
    {
      "epoch": 0.025178826895565092,
      "grad_norm": 0.2869950532913208,
      "learning_rate": 0.00019999998018159753,
      "loss": 1.786,
      "step": 11
    },
    {
      "epoch": 0.027467811158798282,
      "grad_norm": 0.2994249761104584,
      "learning_rate": 0.00019999992072639792,
      "loss": 1.7567,
      "step": 12
    },
    {
      "epoch": 0.029756795422031475,
      "grad_norm": 0.3566282391548157,
      "learning_rate": 0.00019999982163442477,
      "loss": 1.9092,
      "step": 13
    },
    {
      "epoch": 0.03204577968526466,
      "grad_norm": 0.3094356656074524,
      "learning_rate": 0.00019999968290571732,
      "loss": 1.718,
      "step": 14
    },
    {
      "epoch": 0.034334763948497854,
      "grad_norm": 0.36112821102142334,
      "learning_rate": 0.00019999950454033062,
      "loss": 1.6908,
      "step": 15
    },
    {
      "epoch": 0.03662374821173105,
      "grad_norm": 0.35511747002601624,
      "learning_rate": 0.00019999928653833532,
      "loss": 1.6965,
      "step": 16
    },
    {
      "epoch": 0.03891273247496423,
      "grad_norm": 0.37556490302085876,
      "learning_rate": 0.00019999902889981784,
      "loss": 1.7111,
      "step": 17
    },
    {
      "epoch": 0.041201716738197426,
      "grad_norm": 0.4161621034145355,
      "learning_rate": 0.00019999873162488029,
      "loss": 1.7206,
      "step": 18
    },
    {
      "epoch": 0.04349070100143061,
      "grad_norm": 0.43308475613594055,
      "learning_rate": 0.0001999983947136405,
      "loss": 1.7491,
      "step": 19
    },
    {
      "epoch": 0.045779685264663805,
      "grad_norm": 0.4341544210910797,
      "learning_rate": 0.00019999801816623203,
      "loss": 1.605,
      "step": 20
    },
    {
      "epoch": 0.048068669527897,
      "grad_norm": 0.44212228059768677,
      "learning_rate": 0.00019999760198280412,
      "loss": 1.8277,
      "step": 21
    },
    {
      "epoch": 0.050357653791130184,
      "grad_norm": 0.40694648027420044,
      "learning_rate": 0.00019999714616352176,
      "loss": 1.5795,
      "step": 22
    },
    {
      "epoch": 0.05264663805436338,
      "grad_norm": 0.4032455384731293,
      "learning_rate": 0.0001999966507085656,
      "loss": 1.6449,
      "step": 23
    },
    {
      "epoch": 0.054935622317596564,
      "grad_norm": 0.45054155588150024,
      "learning_rate": 0.00019999611561813199,
      "loss": 1.7238,
      "step": 24
    },
    {
      "epoch": 0.05722460658082976,
      "grad_norm": 0.4439268112182617,
      "learning_rate": 0.00019999554089243304,
      "loss": 1.5288,
      "step": 25
    },
    {
      "epoch": 0.05951359084406295,
      "grad_norm": 0.49793151021003723,
      "learning_rate": 0.00019999492653169658,
      "loss": 1.6311,
      "step": 26
    },
    {
      "epoch": 0.061802575107296136,
      "grad_norm": 0.48038819432258606,
      "learning_rate": 0.00019999427253616613,
      "loss": 1.7252,
      "step": 27
    },
    {
      "epoch": 0.06409155937052932,
      "grad_norm": 0.5030732750892639,
      "learning_rate": 0.00019999357890610088,
      "loss": 1.5894,
      "step": 28
    },
    {
      "epoch": 0.06638054363376251,
      "grad_norm": 0.5061493515968323,
      "learning_rate": 0.00019999284564177575,
      "loss": 1.4707,
      "step": 29
    },
    {
      "epoch": 0.06866952789699571,
      "grad_norm": 0.474224328994751,
      "learning_rate": 0.00019999207274348142,
      "loss": 1.6381,
      "step": 30
    },
    {
      "epoch": 0.0709585121602289,
      "grad_norm": 0.5087363123893738,
      "learning_rate": 0.00019999126021152423,
      "loss": 1.6322,
      "step": 31
    },
    {
      "epoch": 0.0732474964234621,
      "grad_norm": 0.7440306544303894,
      "learning_rate": 0.00019999040804622627,
      "loss": 1.4512,
      "step": 32
    },
    {
      "epoch": 0.07553648068669527,
      "grad_norm": 0.5521988868713379,
      "learning_rate": 0.00019998951624792524,
      "loss": 1.526,
      "step": 33
    },
    {
      "epoch": 0.07782546494992847,
      "grad_norm": 0.6378911733627319,
      "learning_rate": 0.00019998858481697467,
      "loss": 1.4424,
      "step": 34
    },
    {
      "epoch": 0.08011444921316166,
      "grad_norm": 0.7784550189971924,
      "learning_rate": 0.00019998761375374376,
      "loss": 1.4575,
      "step": 35
    },
    {
      "epoch": 0.08240343347639485,
      "grad_norm": 0.7092223167419434,
      "learning_rate": 0.00019998660305861737,
      "loss": 1.4501,
      "step": 36
    },
    {
      "epoch": 0.08469241773962805,
      "grad_norm": 0.7746505737304688,
      "learning_rate": 0.00019998555273199612,
      "loss": 1.808,
      "step": 37
    },
    {
      "epoch": 0.08698140200286122,
      "grad_norm": 0.8514325022697449,
      "learning_rate": 0.00019998446277429634,
      "loss": 1.481,
      "step": 38
    },
    {
      "epoch": 0.08927038626609442,
      "grad_norm": 0.8290690779685974,
      "learning_rate": 0.00019998333318595006,
      "loss": 1.6,
      "step": 39
    },
    {
      "epoch": 0.09155937052932761,
      "grad_norm": 0.9082464575767517,
      "learning_rate": 0.00019998216396740497,
      "loss": 1.563,
      "step": 40
    },
    {
      "epoch": 0.0938483547925608,
      "grad_norm": 1.1187679767608643,
      "learning_rate": 0.00019998095511912457,
      "loss": 1.3736,
      "step": 41
    },
    {
      "epoch": 0.096137339055794,
      "grad_norm": 0.8268250823020935,
      "learning_rate": 0.00019997970664158797,
      "loss": 1.3761,
      "step": 42
    },
    {
      "epoch": 0.09842632331902718,
      "grad_norm": 1.087795615196228,
      "learning_rate": 0.00019997841853529002,
      "loss": 1.6233,
      "step": 43
    },
    {
      "epoch": 0.10071530758226037,
      "grad_norm": 1.2439289093017578,
      "learning_rate": 0.00019997709080074133,
      "loss": 1.7651,
      "step": 44
    },
    {
      "epoch": 0.10300429184549356,
      "grad_norm": 1.3559924364089966,
      "learning_rate": 0.00019997572343846814,
      "loss": 1.885,
      "step": 45
    },
    {
      "epoch": 0.10529327610872675,
      "grad_norm": 1.2400606870651245,
      "learning_rate": 0.0001999743164490124,
      "loss": 1.5031,
      "step": 46
    },
    {
      "epoch": 0.10758226037195995,
      "grad_norm": 1.1934120655059814,
      "learning_rate": 0.0001999728698329318,
      "loss": 1.703,
      "step": 47
    },
    {
      "epoch": 0.10987124463519313,
      "grad_norm": 1.5033018589019775,
      "learning_rate": 0.00019997138359079984,
      "loss": 1.3202,
      "step": 48
    },
    {
      "epoch": 0.11216022889842632,
      "grad_norm": 2.2490179538726807,
      "learning_rate": 0.00019996985772320549,
      "loss": 2.048,
      "step": 49
    },
    {
      "epoch": 0.11444921316165951,
      "grad_norm": 2.415254592895508,
      "learning_rate": 0.00019996829223075361,
      "loss": 2.3641,
      "step": 50
    },
    {
      "epoch": 0.1167381974248927,
      "grad_norm": 0.74568110704422,
      "learning_rate": 0.00019996668711406472,
      "loss": 2.0548,
      "step": 51
    },
    {
      "epoch": 0.1190271816881259,
      "grad_norm": 0.6647669672966003,
      "learning_rate": 0.00019996504237377498,
      "loss": 1.9391,
      "step": 52
    },
    {
      "epoch": 0.12131616595135908,
      "grad_norm": 0.5252391695976257,
      "learning_rate": 0.00019996335801053641,
      "loss": 1.9836,
      "step": 53
    },
    {
      "epoch": 0.12360515021459227,
      "grad_norm": 0.36091306805610657,
      "learning_rate": 0.00019996163402501653,
      "loss": 1.8583,
      "step": 54
    },
    {
      "epoch": 0.12589413447782546,
      "grad_norm": 0.3017626702785492,
      "learning_rate": 0.00019995987041789874,
      "loss": 1.8193,
      "step": 55
    },
    {
      "epoch": 0.12818311874105864,
      "grad_norm": 0.27102571725845337,
      "learning_rate": 0.00019995806718988205,
      "loss": 1.8637,
      "step": 56
    },
    {
      "epoch": 0.13047210300429185,
      "grad_norm": 0.268863707780838,
      "learning_rate": 0.00019995622434168124,
      "loss": 1.8262,
      "step": 57
    },
    {
      "epoch": 0.13276108726752503,
      "grad_norm": 0.2693628966808319,
      "learning_rate": 0.00019995434187402672,
      "loss": 1.8127,
      "step": 58
    },
    {
      "epoch": 0.13505007153075824,
      "grad_norm": 0.3138720989227295,
      "learning_rate": 0.00019995241978766463,
      "loss": 1.8128,
      "step": 59
    },
    {
      "epoch": 0.13733905579399142,
      "grad_norm": 0.2930191457271576,
      "learning_rate": 0.00019995045808335687,
      "loss": 1.8679,
      "step": 60
    },
    {
      "epoch": 0.1396280400572246,
      "grad_norm": 0.31221839785575867,
      "learning_rate": 0.00019994845676188095,
      "loss": 1.7848,
      "step": 61
    },
    {
      "epoch": 0.1419170243204578,
      "grad_norm": 0.28518036007881165,
      "learning_rate": 0.00019994641582403016,
      "loss": 1.6757,
      "step": 62
    },
    {
      "epoch": 0.14420600858369098,
      "grad_norm": 0.31122198700904846,
      "learning_rate": 0.00019994433527061346,
      "loss": 1.8134,
      "step": 63
    },
    {
      "epoch": 0.1464949928469242,
      "grad_norm": 0.312227725982666,
      "learning_rate": 0.00019994221510245548,
      "loss": 1.7325,
      "step": 64
    },
    {
      "epoch": 0.14878397711015737,
      "grad_norm": 0.3113989233970642,
      "learning_rate": 0.00019994005532039663,
      "loss": 1.6979,
      "step": 65
    },
    {
      "epoch": 0.15107296137339055,
      "grad_norm": 0.3362973928451538,
      "learning_rate": 0.00019993785592529298,
      "loss": 1.7646,
      "step": 66
    },
    {
      "epoch": 0.15336194563662375,
      "grad_norm": 0.3250870108604431,
      "learning_rate": 0.00019993561691801625,
      "loss": 1.5356,
      "step": 67
    },
    {
      "epoch": 0.15565092989985693,
      "grad_norm": 0.34120023250579834,
      "learning_rate": 0.00019993333829945397,
      "loss": 1.6087,
      "step": 68
    },
    {
      "epoch": 0.15793991416309014,
      "grad_norm": 0.33193376660346985,
      "learning_rate": 0.00019993102007050928,
      "loss": 1.7436,
      "step": 69
    },
    {
      "epoch": 0.16022889842632332,
      "grad_norm": 0.399433434009552,
      "learning_rate": 0.00019992866223210104,
      "loss": 1.679,
      "step": 70
    },
    {
      "epoch": 0.1625178826895565,
      "grad_norm": 0.37392327189445496,
      "learning_rate": 0.00019992626478516386,
      "loss": 1.6859,
      "step": 71
    },
    {
      "epoch": 0.1648068669527897,
      "grad_norm": 0.3868749737739563,
      "learning_rate": 0.00019992382773064798,
      "loss": 1.8012,
      "step": 72
    },
    {
      "epoch": 0.16709585121602288,
      "grad_norm": 0.4253980815410614,
      "learning_rate": 0.0001999213510695194,
      "loss": 1.7289,
      "step": 73
    },
    {
      "epoch": 0.1693848354792561,
      "grad_norm": 0.4231208264827728,
      "learning_rate": 0.00019991883480275974,
      "loss": 1.5526,
      "step": 74
    },
    {
      "epoch": 0.17167381974248927,
      "grad_norm": 0.41427287459373474,
      "learning_rate": 0.0001999162789313664,
      "loss": 1.5526,
      "step": 75
    },
    {
      "epoch": 0.17396280400572245,
      "grad_norm": 0.4733915627002716,
      "learning_rate": 0.00019991368345635246,
      "loss": 1.7993,
      "step": 76
    },
    {
      "epoch": 0.17625178826895566,
      "grad_norm": 0.4497396945953369,
      "learning_rate": 0.00019991104837874666,
      "loss": 1.571,
      "step": 77
    },
    {
      "epoch": 0.17854077253218884,
      "grad_norm": 0.47930455207824707,
      "learning_rate": 0.00019990837369959348,
      "loss": 1.5495,
      "step": 78
    },
    {
      "epoch": 0.18082975679542204,
      "grad_norm": 0.4698319137096405,
      "learning_rate": 0.00019990565941995305,
      "loss": 1.572,
      "step": 79
    },
    {
      "epoch": 0.18311874105865522,
      "grad_norm": 0.4709617495536804,
      "learning_rate": 0.00019990290554090123,
      "loss": 1.4932,
      "step": 80
    },
    {
      "epoch": 0.1854077253218884,
      "grad_norm": 0.5004727244377136,
      "learning_rate": 0.00019990011206352958,
      "loss": 1.6006,
      "step": 81
    },
    {
      "epoch": 0.1876967095851216,
      "grad_norm": 0.5551359057426453,
      "learning_rate": 0.00019989727898894538,
      "loss": 1.6886,
      "step": 82
    },
    {
      "epoch": 0.1899856938483548,
      "grad_norm": 0.5628920197486877,
      "learning_rate": 0.0001998944063182715,
      "loss": 1.6126,
      "step": 83
    },
    {
      "epoch": 0.192274678111588,
      "grad_norm": 0.5695984363555908,
      "learning_rate": 0.0001998914940526466,
      "loss": 1.5413,
      "step": 84
    },
    {
      "epoch": 0.19456366237482117,
      "grad_norm": 0.605058491230011,
      "learning_rate": 0.00019988854219322506,
      "loss": 1.6111,
      "step": 85
    },
    {
      "epoch": 0.19685264663805435,
      "grad_norm": 0.61869215965271,
      "learning_rate": 0.00019988555074117683,
      "loss": 1.4727,
      "step": 86
    },
    {
      "epoch": 0.19914163090128756,
      "grad_norm": 0.6340058445930481,
      "learning_rate": 0.00019988251969768766,
      "loss": 1.541,
      "step": 87
    },
    {
      "epoch": 0.20143061516452074,
      "grad_norm": 0.7175170183181763,
      "learning_rate": 0.00019987944906395898,
      "loss": 1.6055,
      "step": 88
    },
    {
      "epoch": 0.20371959942775394,
      "grad_norm": 0.7348681688308716,
      "learning_rate": 0.0001998763388412078,
      "loss": 1.6315,
      "step": 89
    },
    {
      "epoch": 0.20600858369098712,
      "grad_norm": 0.7640527486801147,
      "learning_rate": 0.00019987318903066706,
      "loss": 1.6544,
      "step": 90
    },
    {
      "epoch": 0.2082975679542203,
      "grad_norm": 0.7852384448051453,
      "learning_rate": 0.00019986999963358512,
      "loss": 1.6871,
      "step": 91
    },
    {
      "epoch": 0.2105865522174535,
      "grad_norm": 0.8368479609489441,
      "learning_rate": 0.0001998667706512262,
      "loss": 1.45,
      "step": 92
    },
    {
      "epoch": 0.2128755364806867,
      "grad_norm": 0.8695632219314575,
      "learning_rate": 0.00019986350208487017,
      "loss": 1.5045,
      "step": 93
    },
    {
      "epoch": 0.2151645207439199,
      "grad_norm": 0.9876943826675415,
      "learning_rate": 0.00019986019393581256,
      "loss": 1.2662,
      "step": 94
    },
    {
      "epoch": 0.21745350500715308,
      "grad_norm": 1.4304019212722778,
      "learning_rate": 0.00019985684620536467,
      "loss": 1.5947,
      "step": 95
    },
    {
      "epoch": 0.21974248927038625,
      "grad_norm": 1.2126617431640625,
      "learning_rate": 0.00019985345889485338,
      "loss": 1.7813,
      "step": 96
    },
    {
      "epoch": 0.22203147353361946,
      "grad_norm": 1.27859628200531,
      "learning_rate": 0.0001998500320056213,
      "loss": 1.6291,
      "step": 97
    },
    {
      "epoch": 0.22432045779685264,
      "grad_norm": 1.4013851881027222,
      "learning_rate": 0.00019984656553902682,
      "loss": 1.8084,
      "step": 98
    },
    {
      "epoch": 0.22660944206008585,
      "grad_norm": 2.2598812580108643,
      "learning_rate": 0.00019984305949644385,
      "loss": 2.0362,
      "step": 99
    },
    {
      "epoch": 0.22889842632331903,
      "grad_norm": 3.096783399581909,
      "learning_rate": 0.00019983951387926214,
      "loss": 2.2111,
      "step": 100
    },
    {
      "epoch": 0.2311874105865522,
      "grad_norm": 1.1480828523635864,
      "learning_rate": 0.00019983592868888698,
      "loss": 1.974,
      "step": 101
    },
    {
      "epoch": 0.2334763948497854,
      "grad_norm": 1.306321144104004,
      "learning_rate": 0.00019983230392673954,
      "loss": 1.988,
      "step": 102
    },
    {
      "epoch": 0.2357653791130186,
      "grad_norm": 1.2203303575515747,
      "learning_rate": 0.00019982863959425645,
      "loss": 1.9348,
      "step": 103
    },
    {
      "epoch": 0.2380543633762518,
      "grad_norm": 0.874756932258606,
      "learning_rate": 0.00019982493569289017,
      "loss": 1.9535,
      "step": 104
    },
    {
      "epoch": 0.24034334763948498,
      "grad_norm": 0.613598108291626,
      "learning_rate": 0.0001998211922241088,
      "loss": 1.8154,
      "step": 105
    },
    {
      "epoch": 0.24263233190271816,
      "grad_norm": 0.41530901193618774,
      "learning_rate": 0.0001998174091893962,
      "loss": 1.8012,
      "step": 106
    },
    {
      "epoch": 0.24492131616595136,
      "grad_norm": 0.34850218892097473,
      "learning_rate": 0.00019981358659025173,
      "loss": 1.8313,
      "step": 107
    },
    {
      "epoch": 0.24721030042918454,
      "grad_norm": 0.3121732473373413,
      "learning_rate": 0.00019980972442819065,
      "loss": 1.8425,
      "step": 108
    },
    {
      "epoch": 0.24949928469241775,
      "grad_norm": 0.3218090236186981,
      "learning_rate": 0.00019980582270474374,
      "loss": 1.788,
      "step": 109
    },
    {
      "epoch": 0.25178826895565093,
      "grad_norm": 0.3283252418041229,
      "learning_rate": 0.00019980188142145754,
      "loss": 1.8021,
      "step": 110
    },
    {
      "epoch": 0.25407725321888414,
      "grad_norm": 0.3523484468460083,
      "learning_rate": 0.00019979790057989423,
      "loss": 1.759,
      "step": 111
    },
    {
      "epoch": 0.2563662374821173,
      "grad_norm": 0.3959892690181732,
      "learning_rate": 0.0001997938801816317,
      "loss": 1.6912,
      "step": 112
    },
    {
      "epoch": 0.2586552217453505,
      "grad_norm": 0.4004235565662384,
      "learning_rate": 0.00019978982022826353,
      "loss": 1.7836,
      "step": 113
    },
    {
      "epoch": 0.2609442060085837,
      "grad_norm": 0.43743371963500977,
      "learning_rate": 0.00019978572072139891,
      "loss": 1.8488,
      "step": 114
    },
    {
      "epoch": 0.2632331902718169,
      "grad_norm": 0.451616495847702,
      "learning_rate": 0.00019978158166266278,
      "loss": 1.8067,
      "step": 115
    },
    {
      "epoch": 0.26552217453505006,
      "grad_norm": 0.3634856939315796,
      "learning_rate": 0.00019977740305369575,
      "loss": 1.7808,
      "step": 116
    },
    {
      "epoch": 0.26781115879828327,
      "grad_norm": 0.38346609473228455,
      "learning_rate": 0.00019977318489615404,
      "loss": 1.6491,
      "step": 117
    },
    {
      "epoch": 0.2701001430615165,
      "grad_norm": 0.37397632002830505,
      "learning_rate": 0.00019976892719170963,
      "loss": 1.7784,
      "step": 118
    },
    {
      "epoch": 0.2723891273247496,
      "grad_norm": 0.37594476342201233,
      "learning_rate": 0.0001997646299420501,
      "loss": 1.5781,
      "step": 119
    },
    {
      "epoch": 0.27467811158798283,
      "grad_norm": 0.3883626461029053,
      "learning_rate": 0.00019976029314887881,
      "loss": 1.6927,
      "step": 120
    },
    {
      "epoch": 0.27696709585121604,
      "grad_norm": 0.38355129957199097,
      "learning_rate": 0.00019975591681391466,
      "loss": 1.7584,
      "step": 121
    },
    {
      "epoch": 0.2792560801144492,
      "grad_norm": 0.3981020152568817,
      "learning_rate": 0.0001997515009388923,
      "loss": 1.6489,
      "step": 122
    },
    {
      "epoch": 0.2815450643776824,
      "grad_norm": 0.45527127385139465,
      "learning_rate": 0.00019974704552556206,
      "loss": 1.6834,
      "step": 123
    },
    {
      "epoch": 0.2838340486409156,
      "grad_norm": 0.4475543200969696,
      "learning_rate": 0.00019974255057568992,
      "loss": 1.5066,
      "step": 124
    },
    {
      "epoch": 0.2861230329041488,
      "grad_norm": 0.4641362428665161,
      "learning_rate": 0.00019973801609105755,
      "loss": 1.7381,
      "step": 125
    },
    {
      "epoch": 0.28841201716738196,
      "grad_norm": 0.4138704836368561,
      "learning_rate": 0.00019973344207346226,
      "loss": 1.396,
      "step": 126
    },
    {
      "epoch": 0.29070100143061517,
      "grad_norm": 0.48385947942733765,
      "learning_rate": 0.00019972882852471702,
      "loss": 1.5532,
      "step": 127
    },
    {
      "epoch": 0.2929899856938484,
      "grad_norm": 0.5530406832695007,
      "learning_rate": 0.0001997241754466505,
      "loss": 1.7044,
      "step": 128
    },
    {
      "epoch": 0.2952789699570815,
      "grad_norm": 0.47684353590011597,
      "learning_rate": 0.0001997194828411071,
      "loss": 1.6817,
      "step": 129
    },
    {
      "epoch": 0.29756795422031473,
      "grad_norm": 0.5050346851348877,
      "learning_rate": 0.00019971475070994674,
      "loss": 1.451,
      "step": 130
    },
    {
      "epoch": 0.29985693848354794,
      "grad_norm": 0.5037690997123718,
      "learning_rate": 0.0001997099790550451,
      "loss": 1.5342,
      "step": 131
    },
    {
      "epoch": 0.3021459227467811,
      "grad_norm": 0.5959687232971191,
      "learning_rate": 0.00019970516787829357,
      "loss": 1.7144,
      "step": 132
    },
    {
      "epoch": 0.3044349070100143,
      "grad_norm": 0.5528478026390076,
      "learning_rate": 0.00019970031718159905,
      "loss": 1.498,
      "step": 133
    },
    {
      "epoch": 0.3067238912732475,
      "grad_norm": 0.6057027578353882,
      "learning_rate": 0.00019969542696688427,
      "loss": 1.6538,
      "step": 134
    },
    {
      "epoch": 0.3090128755364807,
      "grad_norm": 0.6749109625816345,
      "learning_rate": 0.00019969049723608752,
      "loss": 1.7115,
      "step": 135
    },
    {
      "epoch": 0.31130185979971386,
      "grad_norm": 0.6191538572311401,
      "learning_rate": 0.00019968552799116282,
      "loss": 1.5072,
      "step": 136
    },
    {
      "epoch": 0.31359084406294707,
      "grad_norm": 0.7352018356323242,
      "learning_rate": 0.00019968051923407982,
      "loss": 1.483,
      "step": 137
    },
    {
      "epoch": 0.3158798283261803,
      "grad_norm": 0.8472992181777954,
      "learning_rate": 0.00019967547096682377,
      "loss": 1.3482,
      "step": 138
    },
    {
      "epoch": 0.31816881258941343,
      "grad_norm": 0.6694830656051636,
      "learning_rate": 0.00019967038319139573,
      "loss": 1.1871,
      "step": 139
    },
    {
      "epoch": 0.32045779685264664,
      "grad_norm": 0.7963240146636963,
      "learning_rate": 0.00019966525590981227,
      "loss": 1.4613,
      "step": 140
    },
    {
      "epoch": 0.32274678111587984,
      "grad_norm": 0.8364604711532593,
      "learning_rate": 0.0001996600891241057,
      "loss": 1.3999,
      "step": 141
    },
    {
      "epoch": 0.325035765379113,
      "grad_norm": 0.7502745389938354,
      "learning_rate": 0.00019965488283632397,
      "loss": 1.2849,
      "step": 142
    },
    {
      "epoch": 0.3273247496423462,
      "grad_norm": 0.9014572501182556,
      "learning_rate": 0.0001996496370485307,
      "loss": 1.4747,
      "step": 143
    },
    {
      "epoch": 0.3296137339055794,
      "grad_norm": 0.9040302038192749,
      "learning_rate": 0.0001996443517628051,
      "loss": 1.6115,
      "step": 144
    },
    {
      "epoch": 0.3319027181688126,
      "grad_norm": 1.038684606552124,
      "learning_rate": 0.00019963902698124213,
      "loss": 1.4857,
      "step": 145
    },
    {
      "epoch": 0.33419170243204577,
      "grad_norm": 1.227838158607483,
      "learning_rate": 0.00019963366270595237,
      "loss": 2.0123,
      "step": 146
    },
    {
      "epoch": 0.336480686695279,
      "grad_norm": 1.1172336339950562,
      "learning_rate": 0.00019962825893906204,
      "loss": 1.7551,
      "step": 147
    },
    {
      "epoch": 0.3387696709585122,
      "grad_norm": 1.5263012647628784,
      "learning_rate": 0.00019962281568271304,
      "loss": 1.7541,
      "step": 148
    },
    {
      "epoch": 0.34105865522174533,
      "grad_norm": 1.9110766649246216,
      "learning_rate": 0.00019961733293906284,
      "loss": 1.8836,
      "step": 149
    },
    {
      "epoch": 0.34334763948497854,
      "grad_norm": 2.7583436965942383,
      "learning_rate": 0.00019961181071028467,
      "loss": 2.2889,
      "step": 150
    },
    {
      "epoch": 0.34563662374821175,
      "grad_norm": 1.0225343704223633,
      "learning_rate": 0.00019960624899856736,
      "loss": 1.824,
      "step": 151
    },
    {
      "epoch": 0.3479256080114449,
      "grad_norm": 0.8948195576667786,
      "learning_rate": 0.00019960064780611542,
      "loss": 1.9521,
      "step": 152
    },
    {
      "epoch": 0.3502145922746781,
      "grad_norm": 0.8883471488952637,
      "learning_rate": 0.00019959500713514895,
      "loss": 1.8698,
      "step": 153
    },
    {
      "epoch": 0.3525035765379113,
      "grad_norm": 0.6403118371963501,
      "learning_rate": 0.0001995893269879037,
      "loss": 1.8536,
      "step": 154
    },
    {
      "epoch": 0.3547925608011445,
      "grad_norm": 0.47938281297683716,
      "learning_rate": 0.00019958360736663117,
      "loss": 1.8996,
      "step": 155
    },
    {
      "epoch": 0.35708154506437767,
      "grad_norm": 0.33188125491142273,
      "learning_rate": 0.0001995778482735984,
      "loss": 1.7523,
      "step": 156
    },
    {
      "epoch": 0.3593705293276109,
      "grad_norm": 0.27183088660240173,
      "learning_rate": 0.0001995720497110881,
      "loss": 1.7282,
      "step": 157
    },
    {
      "epoch": 0.3616595135908441,
      "grad_norm": 0.27325254678726196,
      "learning_rate": 0.00019956621168139865,
      "loss": 1.9362,
      "step": 158
    },
    {
      "epoch": 0.36394849785407724,
      "grad_norm": 0.3127312958240509,
      "learning_rate": 0.00019956033418684407,
      "loss": 1.7945,
      "step": 159
    },
    {
      "epoch": 0.36623748211731044,
      "grad_norm": 0.33686363697052,
      "learning_rate": 0.000199554417229754,
      "loss": 1.7877,
      "step": 160
    },
    {
      "epoch": 0.36852646638054365,
      "grad_norm": 0.3611038327217102,
      "learning_rate": 0.0001995484608124737,
      "loss": 1.7242,
      "step": 161
    },
    {
      "epoch": 0.3708154506437768,
      "grad_norm": 0.36596477031707764,
      "learning_rate": 0.0001995424649373642,
      "loss": 1.6384,
      "step": 162
    },
    {
      "epoch": 0.37310443490701,
      "grad_norm": 0.39796876907348633,
      "learning_rate": 0.00019953642960680195,
      "loss": 1.6544,
      "step": 163
    },
    {
      "epoch": 0.3753934191702432,
      "grad_norm": 0.3859604001045227,
      "learning_rate": 0.00019953035482317923,
      "loss": 1.6687,
      "step": 164
    },
    {
      "epoch": 0.3776824034334764,
      "grad_norm": 0.4175392687320709,
      "learning_rate": 0.00019952424058890388,
      "loss": 1.8111,
      "step": 165
    },
    {
      "epoch": 0.3799713876967096,
      "grad_norm": 0.4093855023384094,
      "learning_rate": 0.0001995180869063994,
      "loss": 1.7979,
      "step": 166
    },
    {
      "epoch": 0.3822603719599428,
      "grad_norm": 0.4287075102329254,
      "learning_rate": 0.00019951189377810492,
      "loss": 1.7192,
      "step": 167
    },
    {
      "epoch": 0.384549356223176,
      "grad_norm": 0.410826712846756,
      "learning_rate": 0.00019950566120647515,
      "loss": 1.718,
      "step": 168
    },
    {
      "epoch": 0.38683834048640914,
      "grad_norm": 0.43523624539375305,
      "learning_rate": 0.0001994993891939805,
      "loss": 1.6765,
      "step": 169
    },
    {
      "epoch": 0.38912732474964234,
      "grad_norm": 0.42419037222862244,
      "learning_rate": 0.000199493077743107,
      "loss": 1.5921,
      "step": 170
    },
    {
      "epoch": 0.39141630901287555,
      "grad_norm": 0.422701895236969,
      "learning_rate": 0.00019948672685635632,
      "loss": 1.6893,
      "step": 171
    },
    {
      "epoch": 0.3937052932761087,
      "grad_norm": 0.41500985622406006,
      "learning_rate": 0.00019948033653624576,
      "loss": 1.6299,
      "step": 172
    },
    {
      "epoch": 0.3959942775393419,
      "grad_norm": 0.4861249327659607,
      "learning_rate": 0.0001994739067853082,
      "loss": 1.7727,
      "step": 173
    },
    {
      "epoch": 0.3982832618025751,
      "grad_norm": 0.4738679528236389,
      "learning_rate": 0.00019946743760609222,
      "loss": 1.575,
      "step": 174
    },
    {
      "epoch": 0.4005722460658083,
      "grad_norm": 0.44409942626953125,
      "learning_rate": 0.000199460929001162,
      "loss": 1.5484,
      "step": 175
    },
    {
      "epoch": 0.4028612303290415,
      "grad_norm": 0.48912644386291504,
      "learning_rate": 0.0001994543809730973,
      "loss": 1.6657,
      "step": 176
    },
    {
      "epoch": 0.4051502145922747,
      "grad_norm": 0.4906540513038635,
      "learning_rate": 0.00019944779352449362,
      "loss": 1.5919,
      "step": 177
    },
    {
      "epoch": 0.4074391988555079,
      "grad_norm": 0.4717342257499695,
      "learning_rate": 0.00019944116665796192,
      "loss": 1.5567,
      "step": 178
    },
    {
      "epoch": 0.40972818311874104,
      "grad_norm": 0.5700221657752991,
      "learning_rate": 0.00019943450037612895,
      "loss": 1.7091,
      "step": 179
    },
    {
      "epoch": 0.41201716738197425,
      "grad_norm": 0.5210567712783813,
      "learning_rate": 0.00019942779468163697,
      "loss": 1.6209,
      "step": 180
    },
    {
      "epoch": 0.41430615164520745,
      "grad_norm": 0.5747235417366028,
      "learning_rate": 0.00019942104957714396,
      "loss": 1.7068,
      "step": 181
    },
    {
      "epoch": 0.4165951359084406,
      "grad_norm": 0.5356692671775818,
      "learning_rate": 0.00019941426506532342,
      "loss": 1.548,
      "step": 182
    },
    {
      "epoch": 0.4188841201716738,
      "grad_norm": 0.6216205954551697,
      "learning_rate": 0.0001994074411488645,
      "loss": 1.6682,
      "step": 183
    },
    {
      "epoch": 0.421173104434907,
      "grad_norm": 0.5503804087638855,
      "learning_rate": 0.000199400577830472,
      "loss": 1.365,
      "step": 184
    },
    {
      "epoch": 0.4234620886981402,
      "grad_norm": 0.6388237476348877,
      "learning_rate": 0.00019939367511286635,
      "loss": 1.657,
      "step": 185
    },
    {
      "epoch": 0.4257510729613734,
      "grad_norm": 0.6503097414970398,
      "learning_rate": 0.00019938673299878351,
      "loss": 1.6333,
      "step": 186
    },
    {
      "epoch": 0.4280400572246066,
      "grad_norm": 0.7155764698982239,
      "learning_rate": 0.00019937975149097515,
      "loss": 1.6563,
      "step": 187
    },
    {
      "epoch": 0.4303290414878398,
      "grad_norm": 0.7112058401107788,
      "learning_rate": 0.00019937273059220854,
      "loss": 1.3623,
      "step": 188
    },
    {
      "epoch": 0.43261802575107294,
      "grad_norm": 0.7021076083183289,
      "learning_rate": 0.00019936567030526647,
      "loss": 1.2104,
      "step": 189
    },
    {
      "epoch": 0.43490701001430615,
      "grad_norm": 0.8481869697570801,
      "learning_rate": 0.00019935857063294748,
      "loss": 1.5795,
      "step": 190
    },
    {
      "epoch": 0.43719599427753936,
      "grad_norm": 0.6988086104393005,
      "learning_rate": 0.00019935143157806559,
      "loss": 1.5838,
      "step": 191
    },
    {
      "epoch": 0.4394849785407725,
      "grad_norm": 0.7628345489501953,
      "learning_rate": 0.00019934425314345055,
      "loss": 1.443,
      "step": 192
    },
    {
      "epoch": 0.4417739628040057,
      "grad_norm": 0.9695887565612793,
      "learning_rate": 0.00019933703533194766,
      "loss": 1.6091,
      "step": 193
    },
    {
      "epoch": 0.4440629470672389,
      "grad_norm": 1.1328017711639404,
      "learning_rate": 0.0001993297781464178,
      "loss": 1.7478,
      "step": 194
    },
    {
      "epoch": 0.44635193133047213,
      "grad_norm": 1.1095105409622192,
      "learning_rate": 0.00019932248158973745,
      "loss": 1.6539,
      "step": 195
    },
    {
      "epoch": 0.4486409155937053,
      "grad_norm": 1.0009630918502808,
      "learning_rate": 0.0001993151456647988,
      "loss": 1.2786,
      "step": 196
    },
    {
      "epoch": 0.4509298998569385,
      "grad_norm": 1.1418497562408447,
      "learning_rate": 0.0001993077703745096,
      "loss": 1.3044,
      "step": 197
    },
    {
      "epoch": 0.4532188841201717,
      "grad_norm": 1.251988410949707,
      "learning_rate": 0.00019930035572179308,
      "loss": 1.3441,
      "step": 198
    },
    {
      "epoch": 0.45550786838340485,
      "grad_norm": 2.516533374786377,
      "learning_rate": 0.00019929290170958824,
      "loss": 1.988,
      "step": 199
    },
    {
      "epoch": 0.45779685264663805,
      "grad_norm": 3.2890355587005615,
      "learning_rate": 0.0001992854083408496,
      "loss": 2.669,
      "step": 200
    },
    {
      "epoch": 0.46008583690987126,
      "grad_norm": 0.9220067858695984,
      "learning_rate": 0.0001992778756185473,
      "loss": 1.9532,
      "step": 201
    },
    {
      "epoch": 0.4623748211731044,
      "grad_norm": 1.0087976455688477,
      "learning_rate": 0.00019927030354566704,
      "loss": 1.951,
      "step": 202
    },
    {
      "epoch": 0.4646638054363376,
      "grad_norm": 0.8728040456771851,
      "learning_rate": 0.0001992626921252102,
      "loss": 1.8667,
      "step": 203
    },
    {
      "epoch": 0.4669527896995708,
      "grad_norm": 0.7662880420684814,
      "learning_rate": 0.00019925504136019363,
      "loss": 1.9252,
      "step": 204
    },
    {
      "epoch": 0.46924177396280403,
      "grad_norm": 0.5666810870170593,
      "learning_rate": 0.00019924735125364992,
      "loss": 1.887,
      "step": 205
    },
    {
      "epoch": 0.4715307582260372,
      "grad_norm": 0.46456143260002136,
      "learning_rate": 0.0001992396218086271,
      "loss": 1.8399,
      "step": 206
    },
    {
      "epoch": 0.4738197424892704,
      "grad_norm": 0.3381153643131256,
      "learning_rate": 0.000199231853028189,
      "loss": 1.8387,
      "step": 207
    },
    {
      "epoch": 0.4761087267525036,
      "grad_norm": 0.3180887699127197,
      "learning_rate": 0.0001992240449154148,
      "loss": 1.7023,
      "step": 208
    },
    {
      "epoch": 0.47839771101573675,
      "grad_norm": 0.2701670229434967,
      "learning_rate": 0.00019921619747339945,
      "loss": 1.8039,
      "step": 209
    },
    {
      "epoch": 0.48068669527896996,
      "grad_norm": 0.2869170904159546,
      "learning_rate": 0.00019920831070525342,
      "loss": 1.7957,
      "step": 210
    },
    {
      "epoch": 0.48297567954220316,
      "grad_norm": 0.28238070011138916,
      "learning_rate": 0.00019920038461410274,
      "loss": 1.6985,
      "step": 211
    },
    {
      "epoch": 0.4852646638054363,
      "grad_norm": 0.29044851660728455,
      "learning_rate": 0.00019919241920308907,
      "loss": 1.7362,
      "step": 212
    },
    {
      "epoch": 0.4875536480686695,
      "grad_norm": 0.3139536380767822,
      "learning_rate": 0.00019918441447536967,
      "loss": 1.7513,
      "step": 213
    },
    {
      "epoch": 0.4898426323319027,
      "grad_norm": 0.3140701949596405,
      "learning_rate": 0.00019917637043411732,
      "loss": 1.7322,
      "step": 214
    },
    {
      "epoch": 0.49213161659513593,
      "grad_norm": 0.33344417810440063,
      "learning_rate": 0.00019916828708252044,
      "loss": 1.5568,
      "step": 215
    },
    {
      "epoch": 0.4944206008583691,
      "grad_norm": 0.3508111834526062,
      "learning_rate": 0.00019916016442378303,
      "loss": 1.6284,
      "step": 216
    },
    {
      "epoch": 0.4967095851216023,
      "grad_norm": 0.38611069321632385,
      "learning_rate": 0.00019915200246112464,
      "loss": 1.7981,
      "step": 217
    },
    {
      "epoch": 0.4989985693848355,
      "grad_norm": 0.37997177243232727,
      "learning_rate": 0.0001991438011977804,
      "loss": 1.7162,
      "step": 218
    },
    {
      "epoch": 0.5012875536480687,
      "grad_norm": 0.3982301950454712,
      "learning_rate": 0.00019913556063700104,
      "loss": 1.6278,
      "step": 219
    },
    {
      "epoch": 0.5035765379113019,
      "grad_norm": 0.38773295283317566,
      "learning_rate": 0.00019912728078205283,
      "loss": 1.6726,
      "step": 220
    },
    {
      "epoch": 0.5058655221745351,
      "grad_norm": 0.38734155893325806,
      "learning_rate": 0.00019911896163621768,
      "loss": 1.5793,
      "step": 221
    },
    {
      "epoch": 0.5081545064377683,
      "grad_norm": 0.43666213750839233,
      "learning_rate": 0.000199110603202793,
      "loss": 1.6361,
      "step": 222
    },
    {
      "epoch": 0.5104434907010015,
      "grad_norm": 0.44197845458984375,
      "learning_rate": 0.00019910220548509185,
      "loss": 1.5188,
      "step": 223
    },
    {
      "epoch": 0.5127324749642346,
      "grad_norm": 0.4263112545013428,
      "learning_rate": 0.00019909376848644276,
      "loss": 1.5403,
      "step": 224
    },
    {
      "epoch": 0.5150214592274678,
      "grad_norm": 0.4494774043560028,
      "learning_rate": 0.00019908529221018993,
      "loss": 1.5897,
      "step": 225
    },
    {
      "epoch": 0.517310443490701,
      "grad_norm": 0.44257616996765137,
      "learning_rate": 0.0001990767766596931,
      "loss": 1.618,
      "step": 226
    },
    {
      "epoch": 0.5195994277539342,
      "grad_norm": 0.47845131158828735,
      "learning_rate": 0.00019906822183832746,
      "loss": 1.6991,
      "step": 227
    },
    {
      "epoch": 0.5218884120171674,
      "grad_norm": 0.490477591753006,
      "learning_rate": 0.00019905962774948398,
      "loss": 1.5725,
      "step": 228
    },
    {
      "epoch": 0.5241773962804006,
      "grad_norm": 0.5202469825744629,
      "learning_rate": 0.00019905099439656903,
      "loss": 1.5827,
      "step": 229
    },
    {
      "epoch": 0.5264663805436338,
      "grad_norm": 0.49094030261039734,
      "learning_rate": 0.00019904232178300464,
      "loss": 1.6101,
      "step": 230
    },
    {
      "epoch": 0.5287553648068669,
      "grad_norm": 0.5398317575454712,
      "learning_rate": 0.0001990336099122283,
      "loss": 1.5257,
      "step": 231
    },
    {
      "epoch": 0.5310443490701001,
      "grad_norm": 0.5775597095489502,
      "learning_rate": 0.00019902485878769315,
      "loss": 1.714,
      "step": 232
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.5621138215065002,
      "learning_rate": 0.0001990160684128678,
      "loss": 1.5823,
      "step": 233
    },
    {
      "epoch": 0.5356223175965665,
      "grad_norm": 0.541064441204071,
      "learning_rate": 0.00019900723879123658,
      "loss": 1.523,
      "step": 234
    },
    {
      "epoch": 0.5379113018597997,
      "grad_norm": 0.6561056971549988,
      "learning_rate": 0.0001989983699262992,
      "loss": 1.7816,
      "step": 235
    },
    {
      "epoch": 0.540200286123033,
      "grad_norm": 0.7181664705276489,
      "learning_rate": 0.000198989461821571,
      "loss": 1.8239,
      "step": 236
    },
    {
      "epoch": 0.542489270386266,
      "grad_norm": 0.6391181945800781,
      "learning_rate": 0.0001989805144805829,
      "loss": 1.463,
      "step": 237
    },
    {
      "epoch": 0.5447782546494992,
      "grad_norm": 0.6614975333213806,
      "learning_rate": 0.00019897152790688126,
      "loss": 1.4514,
      "step": 238
    },
    {
      "epoch": 0.5470672389127325,
      "grad_norm": 0.645230233669281,
      "learning_rate": 0.00019896250210402815,
      "loss": 1.3928,
      "step": 239
    },
    {
      "epoch": 0.5493562231759657,
      "grad_norm": 0.6474555134773254,
      "learning_rate": 0.00019895343707560108,
      "loss": 1.2767,
      "step": 240
    },
    {
      "epoch": 0.5516452074391989,
      "grad_norm": 0.7385896444320679,
      "learning_rate": 0.00019894433282519316,
      "loss": 1.5652,
      "step": 241
    },
    {
      "epoch": 0.5539341917024321,
      "grad_norm": 0.8108466863632202,
      "learning_rate": 0.000198935189356413,
      "loss": 1.3871,
      "step": 242
    },
    {
      "epoch": 0.5562231759656653,
      "grad_norm": 0.7885132431983948,
      "learning_rate": 0.00019892600667288478,
      "loss": 1.284,
      "step": 243
    },
    {
      "epoch": 0.5585121602288984,
      "grad_norm": 0.8143578767776489,
      "learning_rate": 0.0001989167847782482,
      "loss": 1.2018,
      "step": 244
    },
    {
      "epoch": 0.5608011444921316,
      "grad_norm": 0.8399714231491089,
      "learning_rate": 0.0001989075236761586,
      "loss": 1.5894,
      "step": 245
    },
    {
      "epoch": 0.5630901287553648,
      "grad_norm": 1.1949816942214966,
      "learning_rate": 0.0001988982233702867,
      "loss": 1.5344,
      "step": 246
    },
    {
      "epoch": 0.565379113018598,
      "grad_norm": 1.1354163885116577,
      "learning_rate": 0.0001988888838643189,
      "loss": 1.6228,
      "step": 247
    },
    {
      "epoch": 0.5676680972818312,
      "grad_norm": 1.588686227798462,
      "learning_rate": 0.00019887950516195706,
      "loss": 1.9686,
      "step": 248
    },
    {
      "epoch": 0.5699570815450644,
      "grad_norm": 1.8737907409667969,
      "learning_rate": 0.0001988700872669186,
      "loss": 1.8753,
      "step": 249
    },
    {
      "epoch": 0.5722460658082976,
      "grad_norm": 2.0973799228668213,
      "learning_rate": 0.0001988606301829365,
      "loss": 1.7227,
      "step": 250
    },
    {
      "epoch": 0.5745350500715307,
      "grad_norm": 1.042942762374878,
      "learning_rate": 0.00019885113391375918,
      "loss": 2.0237,
      "step": 251
    },
    {
      "epoch": 0.5768240343347639,
      "grad_norm": 1.1076831817626953,
      "learning_rate": 0.0001988415984631507,
      "loss": 2.026,
      "step": 252
    },
    {
      "epoch": 0.5791130185979971,
      "grad_norm": 1.0394678115844727,
      "learning_rate": 0.0001988320238348906,
      "loss": 1.9717,
      "step": 253
    },
    {
      "epoch": 0.5814020028612303,
      "grad_norm": 0.6771242618560791,
      "learning_rate": 0.00019882241003277398,
      "loss": 1.8525,
      "step": 254
    },
    {
      "epoch": 0.5836909871244635,
      "grad_norm": 0.6341869235038757,
      "learning_rate": 0.00019881275706061145,
      "loss": 1.8066,
      "step": 255
    },
    {
      "epoch": 0.5859799713876968,
      "grad_norm": 0.47391000390052795,
      "learning_rate": 0.00019880306492222908,
      "loss": 1.9094,
      "step": 256
    },
    {
      "epoch": 0.5882689556509298,
      "grad_norm": 0.3714869022369385,
      "learning_rate": 0.00019879333362146854,
      "loss": 1.7436,
      "step": 257
    },
    {
      "epoch": 0.590557939914163,
      "grad_norm": 0.297178715467453,
      "learning_rate": 0.00019878356316218706,
      "loss": 1.8322,
      "step": 258
    },
    {
      "epoch": 0.5928469241773963,
      "grad_norm": 0.30122897028923035,
      "learning_rate": 0.0001987737535482573,
      "loss": 1.8111,
      "step": 259
    },
    {
      "epoch": 0.5951359084406295,
      "grad_norm": 0.3071204721927643,
      "learning_rate": 0.00019876390478356748,
      "loss": 1.6817,
      "step": 260
    },
    {
      "epoch": 0.5974248927038627,
      "grad_norm": 0.310671329498291,
      "learning_rate": 0.00019875401687202135,
      "loss": 1.6248,
      "step": 261
    },
    {
      "epoch": 0.5997138769670959,
      "grad_norm": 0.30385780334472656,
      "learning_rate": 0.0001987440898175381,
      "loss": 1.6728,
      "step": 262
    },
    {
      "epoch": 0.6020028612303291,
      "grad_norm": 0.3122836649417877,
      "learning_rate": 0.00019873412362405262,
      "loss": 1.7415,
      "step": 263
    },
    {
      "epoch": 0.6042918454935622,
      "grad_norm": 0.46503645181655884,
      "learning_rate": 0.00019872411829551507,
      "loss": 1.6421,
      "step": 264
    },
    {
      "epoch": 0.6065808297567954,
      "grad_norm": 0.33438631892204285,
      "learning_rate": 0.0001987140738358913,
      "loss": 1.6463,
      "step": 265
    },
    {
      "epoch": 0.6088698140200286,
      "grad_norm": 0.3826068043708801,
      "learning_rate": 0.0001987039902491626,
      "loss": 1.6841,
      "step": 266
    },
    {
      "epoch": 0.6111587982832618,
      "grad_norm": 0.4045211672782898,
      "learning_rate": 0.00019869386753932578,
      "loss": 1.7039,
      "step": 267
    },
    {
      "epoch": 0.613447782546495,
      "grad_norm": 0.42447629570961,
      "learning_rate": 0.00019868370571039315,
      "loss": 1.8075,
      "step": 268
    },
    {
      "epoch": 0.6157367668097282,
      "grad_norm": 0.4018721878528595,
      "learning_rate": 0.00019867350476639257,
      "loss": 1.6227,
      "step": 269
    },
    {
      "epoch": 0.6180257510729614,
      "grad_norm": 0.4125591516494751,
      "learning_rate": 0.00019866326471136735,
      "loss": 1.6218,
      "step": 270
    },
    {
      "epoch": 0.6203147353361945,
      "grad_norm": 0.44766101241111755,
      "learning_rate": 0.0001986529855493763,
      "loss": 1.6192,
      "step": 271
    },
    {
      "epoch": 0.6226037195994277,
      "grad_norm": 0.43519511818885803,
      "learning_rate": 0.0001986426672844938,
      "loss": 1.6468,
      "step": 272
    },
    {
      "epoch": 0.6248927038626609,
      "grad_norm": 0.44375982880592346,
      "learning_rate": 0.0001986323099208096,
      "loss": 1.5353,
      "step": 273
    },
    {
      "epoch": 0.6271816881258941,
      "grad_norm": 0.3923269510269165,
      "learning_rate": 0.0001986219134624291,
      "loss": 1.7342,
      "step": 274
    },
    {
      "epoch": 0.6294706723891274,
      "grad_norm": 0.43816524744033813,
      "learning_rate": 0.00019861147791347313,
      "loss": 1.4718,
      "step": 275
    },
    {
      "epoch": 0.6317596566523606,
      "grad_norm": 0.4516178369522095,
      "learning_rate": 0.00019860100327807794,
      "loss": 1.6438,
      "step": 276
    },
    {
      "epoch": 0.6340486409155937,
      "grad_norm": 0.4520825147628784,
      "learning_rate": 0.00019859048956039538,
      "loss": 1.3999,
      "step": 277
    },
    {
      "epoch": 0.6363376251788269,
      "grad_norm": 0.48634177446365356,
      "learning_rate": 0.00019857993676459276,
      "loss": 1.601,
      "step": 278
    },
    {
      "epoch": 0.6386266094420601,
      "grad_norm": 0.5089531540870667,
      "learning_rate": 0.00019856934489485286,
      "loss": 1.5641,
      "step": 279
    },
    {
      "epoch": 0.6409155937052933,
      "grad_norm": 0.5448296666145325,
      "learning_rate": 0.00019855871395537395,
      "loss": 1.6257,
      "step": 280
    },
    {
      "epoch": 0.6432045779685265,
      "grad_norm": 0.5490623712539673,
      "learning_rate": 0.00019854804395036984,
      "loss": 1.5203,
      "step": 281
    },
    {
      "epoch": 0.6454935622317597,
      "grad_norm": 0.575749397277832,
      "learning_rate": 0.0001985373348840697,
      "loss": 1.7867,
      "step": 282
    },
    {
      "epoch": 0.6477825464949929,
      "grad_norm": 0.5910357236862183,
      "learning_rate": 0.00019852658676071833,
      "loss": 1.5229,
      "step": 283
    },
    {
      "epoch": 0.650071530758226,
      "grad_norm": 0.5715771913528442,
      "learning_rate": 0.00019851579958457592,
      "loss": 1.3046,
      "step": 284
    },
    {
      "epoch": 0.6523605150214592,
      "grad_norm": 0.5898134112358093,
      "learning_rate": 0.00019850497335991815,
      "loss": 1.5476,
      "step": 285
    },
    {
      "epoch": 0.6546494992846924,
      "grad_norm": 0.5905246734619141,
      "learning_rate": 0.00019849410809103622,
      "loss": 1.3776,
      "step": 286
    },
    {
      "epoch": 0.6569384835479256,
      "grad_norm": 0.6722264289855957,
      "learning_rate": 0.00019848320378223675,
      "loss": 1.5266,
      "step": 287
    },
    {
      "epoch": 0.6592274678111588,
      "grad_norm": 0.7278958559036255,
      "learning_rate": 0.00019847226043784186,
      "loss": 1.6041,
      "step": 288
    },
    {
      "epoch": 0.661516452074392,
      "grad_norm": 0.7815399765968323,
      "learning_rate": 0.00019846127806218914,
      "loss": 1.3236,
      "step": 289
    },
    {
      "epoch": 0.6638054363376252,
      "grad_norm": 0.7282021641731262,
      "learning_rate": 0.00019845025665963167,
      "loss": 1.1546,
      "step": 290
    },
    {
      "epoch": 0.6660944206008583,
      "grad_norm": 0.7975531816482544,
      "learning_rate": 0.000198439196234538,
      "loss": 1.3412,
      "step": 291
    },
    {
      "epoch": 0.6683834048640915,
      "grad_norm": 0.8583757877349854,
      "learning_rate": 0.00019842809679129205,
      "loss": 1.5696,
      "step": 292
    },
    {
      "epoch": 0.6706723891273247,
      "grad_norm": 0.8434293270111084,
      "learning_rate": 0.00019841695833429336,
      "loss": 1.4097,
      "step": 293
    },
    {
      "epoch": 0.672961373390558,
      "grad_norm": 0.8686861991882324,
      "learning_rate": 0.00019840578086795683,
      "loss": 1.4627,
      "step": 294
    },
    {
      "epoch": 0.6752503576537912,
      "grad_norm": 1.067740559577942,
      "learning_rate": 0.00019839456439671286,
      "loss": 1.5768,
      "step": 295
    },
    {
      "epoch": 0.6775393419170244,
      "grad_norm": 1.0355494022369385,
      "learning_rate": 0.0001983833089250073,
      "loss": 1.4593,
      "step": 296
    },
    {
      "epoch": 0.6798283261802575,
      "grad_norm": 1.3659474849700928,
      "learning_rate": 0.00019837201445730145,
      "loss": 2.0614,
      "step": 297
    },
    {
      "epoch": 0.6821173104434907,
      "grad_norm": 1.3352608680725098,
      "learning_rate": 0.00019836068099807208,
      "loss": 1.3051,
      "step": 298
    },
    {
      "epoch": 0.6844062947067239,
      "grad_norm": 1.860592007637024,
      "learning_rate": 0.00019834930855181141,
      "loss": 1.9475,
      "step": 299
    },
    {
      "epoch": 0.6866952789699571,
      "grad_norm": 2.1413514614105225,
      "learning_rate": 0.00019833789712302713,
      "loss": 2.3464,
      "step": 300
    },
    {
      "epoch": 0.6889842632331903,
      "grad_norm": 0.9661771655082703,
      "learning_rate": 0.00019832644671624234,
      "loss": 1.8456,
      "step": 301
    },
    {
      "epoch": 0.6912732474964235,
      "grad_norm": 0.9252067804336548,
      "learning_rate": 0.00019831495733599563,
      "loss": 1.9849,
      "step": 302
    },
    {
      "epoch": 0.6935622317596567,
      "grad_norm": 0.7364763021469116,
      "learning_rate": 0.00019830342898684102,
      "loss": 1.9118,
      "step": 303
    },
    {
      "epoch": 0.6958512160228898,
      "grad_norm": 0.650516927242279,
      "learning_rate": 0.00019829186167334798,
      "loss": 1.8714,
      "step": 304
    },
    {
      "epoch": 0.698140200286123,
      "grad_norm": 0.5958560109138489,
      "learning_rate": 0.00019828025540010144,
      "loss": 1.779,
      "step": 305
    },
    {
      "epoch": 0.7004291845493562,
      "grad_norm": 0.45483529567718506,
      "learning_rate": 0.00019826861017170172,
      "loss": 1.869,
      "step": 306
    },
    {
      "epoch": 0.7027181688125894,
      "grad_norm": 0.4068349003791809,
      "learning_rate": 0.00019825692599276465,
      "loss": 1.7763,
      "step": 307
    },
    {
      "epoch": 0.7050071530758226,
      "grad_norm": 0.30319955945014954,
      "learning_rate": 0.00019824520286792146,
      "loss": 1.8633,
      "step": 308
    },
    {
      "epoch": 0.7072961373390558,
      "grad_norm": 0.3124970495700836,
      "learning_rate": 0.0001982334408018188,
      "loss": 1.6706,
      "step": 309
    },
    {
      "epoch": 0.709585121602289,
      "grad_norm": 0.3324763774871826,
      "learning_rate": 0.0001982216397991188,
      "loss": 1.734,
      "step": 310
    },
    {
      "epoch": 0.7118741058655221,
      "grad_norm": 0.3057507872581482,
      "learning_rate": 0.00019820979986449896,
      "loss": 1.6463,
      "step": 311
    },
    {
      "epoch": 0.7141630901287553,
      "grad_norm": 0.30498653650283813,
      "learning_rate": 0.00019819792100265234,
      "loss": 1.7173,
      "step": 312
    },
    {
      "epoch": 0.7164520743919885,
      "grad_norm": 0.30107173323631287,
      "learning_rate": 0.00019818600321828723,
      "loss": 1.5885,
      "step": 313
    },
    {
      "epoch": 0.7187410586552218,
      "grad_norm": 0.3551604747772217,
      "learning_rate": 0.00019817404651612753,
      "loss": 1.826,
      "step": 314
    },
    {
      "epoch": 0.721030042918455,
      "grad_norm": 0.36133766174316406,
      "learning_rate": 0.0001981620509009125,
      "loss": 1.7346,
      "step": 315
    },
    {
      "epoch": 0.7233190271816882,
      "grad_norm": 0.3880329132080078,
      "learning_rate": 0.00019815001637739677,
      "loss": 1.6575,
      "step": 316
    },
    {
      "epoch": 0.7256080114449213,
      "grad_norm": 0.44181787967681885,
      "learning_rate": 0.00019813794295035048,
      "loss": 1.7522,
      "step": 317
    },
    {
      "epoch": 0.7278969957081545,
      "grad_norm": 0.40254104137420654,
      "learning_rate": 0.00019812583062455913,
      "loss": 1.6956,
      "step": 318
    },
    {
      "epoch": 0.7301859799713877,
      "grad_norm": 0.4149926006793976,
      "learning_rate": 0.0001981136794048237,
      "loss": 1.6434,
      "step": 319
    },
    {
      "epoch": 0.7324749642346209,
      "grad_norm": 0.379864364862442,
      "learning_rate": 0.00019810148929596047,
      "loss": 1.7092,
      "step": 320
    },
    {
      "epoch": 0.7347639484978541,
      "grad_norm": 0.4309485852718353,
      "learning_rate": 0.00019808926030280127,
      "loss": 1.6969,
      "step": 321
    },
    {
      "epoch": 0.7370529327610873,
      "grad_norm": 0.38200846314430237,
      "learning_rate": 0.00019807699243019325,
      "loss": 1.5015,
      "step": 322
    },
    {
      "epoch": 0.7393419170243205,
      "grad_norm": 0.3941330313682556,
      "learning_rate": 0.00019806468568299903,
      "loss": 1.7079,
      "step": 323
    },
    {
      "epoch": 0.7416309012875536,
      "grad_norm": 0.41064247488975525,
      "learning_rate": 0.00019805234006609661,
      "loss": 1.5983,
      "step": 324
    },
    {
      "epoch": 0.7439198855507868,
      "grad_norm": 0.428229421377182,
      "learning_rate": 0.00019803995558437934,
      "loss": 1.5719,
      "step": 325
    },
    {
      "epoch": 0.74620886981402,
      "grad_norm": 0.4445763826370239,
      "learning_rate": 0.00019802753224275614,
      "loss": 1.5251,
      "step": 326
    },
    {
      "epoch": 0.7484978540772532,
      "grad_norm": 0.4788428843021393,
      "learning_rate": 0.00019801507004615112,
      "loss": 1.4068,
      "step": 327
    },
    {
      "epoch": 0.7507868383404864,
      "grad_norm": 0.45525220036506653,
      "learning_rate": 0.00019800256899950397,
      "loss": 1.443,
      "step": 328
    },
    {
      "epoch": 0.7530758226037196,
      "grad_norm": 0.47794467210769653,
      "learning_rate": 0.00019799002910776968,
      "loss": 1.4274,
      "step": 329
    },
    {
      "epoch": 0.7553648068669528,
      "grad_norm": 0.5474439263343811,
      "learning_rate": 0.00019797745037591864,
      "loss": 1.5661,
      "step": 330
    },
    {
      "epoch": 0.7576537911301859,
      "grad_norm": 0.49069586396217346,
      "learning_rate": 0.00019796483280893668,
      "loss": 1.4501,
      "step": 331
    },
    {
      "epoch": 0.7599427753934191,
      "grad_norm": 0.5156876444816589,
      "learning_rate": 0.000197952176411825,
      "loss": 1.6807,
      "step": 332
    },
    {
      "epoch": 0.7622317596566524,
      "grad_norm": 0.5316874980926514,
      "learning_rate": 0.00019793948118960019,
      "loss": 1.459,
      "step": 333
    },
    {
      "epoch": 0.7645207439198856,
      "grad_norm": 0.5263762474060059,
      "learning_rate": 0.00019792674714729422,
      "loss": 1.4987,
      "step": 334
    },
    {
      "epoch": 0.7668097281831188,
      "grad_norm": 0.5568097829818726,
      "learning_rate": 0.0001979139742899545,
      "loss": 1.5967,
      "step": 335
    },
    {
      "epoch": 0.769098712446352,
      "grad_norm": 0.5508828163146973,
      "learning_rate": 0.0001979011626226437,
      "loss": 1.3767,
      "step": 336
    },
    {
      "epoch": 0.7713876967095851,
      "grad_norm": 0.6211968660354614,
      "learning_rate": 0.00019788831215044003,
      "loss": 1.4997,
      "step": 337
    },
    {
      "epoch": 0.7736766809728183,
      "grad_norm": 0.6698896884918213,
      "learning_rate": 0.000197875422878437,
      "loss": 1.59,
      "step": 338
    },
    {
      "epoch": 0.7759656652360515,
      "grad_norm": 0.6504703760147095,
      "learning_rate": 0.00019786249481174346,
      "loss": 1.6004,
      "step": 339
    },
    {
      "epoch": 0.7782546494992847,
      "grad_norm": 0.7118754982948303,
      "learning_rate": 0.00019784952795548372,
      "loss": 1.5852,
      "step": 340
    },
    {
      "epoch": 0.7805436337625179,
      "grad_norm": 0.7549391984939575,
      "learning_rate": 0.0001978365223147974,
      "loss": 1.2979,
      "step": 341
    },
    {
      "epoch": 0.7828326180257511,
      "grad_norm": 0.7815965414047241,
      "learning_rate": 0.00019782347789483958,
      "loss": 1.4829,
      "step": 342
    },
    {
      "epoch": 0.7851216022889843,
      "grad_norm": 0.6953243017196655,
      "learning_rate": 0.00019781039470078058,
      "loss": 1.4213,
      "step": 343
    },
    {
      "epoch": 0.7874105865522174,
      "grad_norm": 0.8302815556526184,
      "learning_rate": 0.00019779727273780618,
      "loss": 1.4489,
      "step": 344
    },
    {
      "epoch": 0.7896995708154506,
      "grad_norm": 0.9122905731201172,
      "learning_rate": 0.00019778411201111753,
      "loss": 1.5006,
      "step": 345
    },
    {
      "epoch": 0.7919885550786838,
      "grad_norm": 1.0813095569610596,
      "learning_rate": 0.0001977709125259311,
      "loss": 1.6602,
      "step": 346
    },
    {
      "epoch": 0.794277539341917,
      "grad_norm": 1.0793122053146362,
      "learning_rate": 0.00019775767428747876,
      "loss": 1.5739,
      "step": 347
    },
    {
      "epoch": 0.7965665236051502,
      "grad_norm": 1.4634814262390137,
      "learning_rate": 0.0001977443973010077,
      "loss": 1.5278,
      "step": 348
    },
    {
      "epoch": 0.7988555078683834,
      "grad_norm": 1.7866122722625732,
      "learning_rate": 0.00019773108157178056,
      "loss": 1.7888,
      "step": 349
    },
    {
      "epoch": 0.8011444921316166,
      "grad_norm": 2.7323102951049805,
      "learning_rate": 0.00019771772710507516,
      "loss": 2.1539,
      "step": 350
    },
    {
      "epoch": 0.8034334763948497,
      "grad_norm": 0.7558505535125732,
      "learning_rate": 0.00019770433390618486,
      "loss": 1.8877,
      "step": 351
    },
    {
      "epoch": 0.805722460658083,
      "grad_norm": 0.8223341703414917,
      "learning_rate": 0.0001976909019804183,
      "loss": 1.8084,
      "step": 352
    },
    {
      "epoch": 0.8080114449213162,
      "grad_norm": 0.8140975832939148,
      "learning_rate": 0.00019767743133309944,
      "loss": 1.9931,
      "step": 353
    },
    {
      "epoch": 0.8103004291845494,
      "grad_norm": 0.6950765252113342,
      "learning_rate": 0.0001976639219695676,
      "loss": 1.8119,
      "step": 354
    },
    {
      "epoch": 0.8125894134477826,
      "grad_norm": 0.6089228391647339,
      "learning_rate": 0.0001976503738951775,
      "loss": 1.7749,
      "step": 355
    },
    {
      "epoch": 0.8148783977110158,
      "grad_norm": 0.5156071186065674,
      "learning_rate": 0.0001976367871152991,
      "loss": 1.914,
      "step": 356
    },
    {
      "epoch": 0.8171673819742489,
      "grad_norm": 0.41553398966789246,
      "learning_rate": 0.00019762316163531786,
      "loss": 1.7222,
      "step": 357
    },
    {
      "epoch": 0.8194563662374821,
      "grad_norm": 0.38983991742134094,
      "learning_rate": 0.0001976094974606344,
      "loss": 1.8405,
      "step": 358
    },
    {
      "epoch": 0.8217453505007153,
      "grad_norm": 0.33772942423820496,
      "learning_rate": 0.00019759579459666484,
      "loss": 1.8195,
      "step": 359
    },
    {
      "epoch": 0.8240343347639485,
      "grad_norm": 0.3214252293109894,
      "learning_rate": 0.0001975820530488405,
      "loss": 1.7164,
      "step": 360
    },
    {
      "epoch": 0.8263233190271817,
      "grad_norm": 0.3165682256221771,
      "learning_rate": 0.00019756827282260808,
      "loss": 1.7964,
      "step": 361
    },
    {
      "epoch": 0.8286123032904149,
      "grad_norm": 0.30793261528015137,
      "learning_rate": 0.00019755445392342967,
      "loss": 1.7392,
      "step": 362
    },
    {
      "epoch": 0.8309012875536481,
      "grad_norm": 0.31366050243377686,
      "learning_rate": 0.0001975405963567826,
      "loss": 1.637,
      "step": 363
    },
    {
      "epoch": 0.8331902718168812,
      "grad_norm": 0.3474258482456207,
      "learning_rate": 0.00019752670012815958,
      "loss": 1.7027,
      "step": 364
    },
    {
      "epoch": 0.8354792560801144,
      "grad_norm": 0.38621070981025696,
      "learning_rate": 0.00019751276524306862,
      "loss": 1.6995,
      "step": 365
    },
    {
      "epoch": 0.8377682403433476,
      "grad_norm": 0.3972114622592926,
      "learning_rate": 0.0001974987917070331,
      "loss": 1.8308,
      "step": 366
    },
    {
      "epoch": 0.8400572246065808,
      "grad_norm": 0.4872405529022217,
      "learning_rate": 0.00019748477952559167,
      "loss": 1.677,
      "step": 367
    },
    {
      "epoch": 0.842346208869814,
      "grad_norm": 0.4697360396385193,
      "learning_rate": 0.00019747072870429828,
      "loss": 1.6135,
      "step": 368
    },
    {
      "epoch": 0.8446351931330472,
      "grad_norm": 0.38687723875045776,
      "learning_rate": 0.00019745663924872228,
      "loss": 1.7738,
      "step": 369
    },
    {
      "epoch": 0.8469241773962805,
      "grad_norm": 0.3866192102432251,
      "learning_rate": 0.00019744251116444818,
      "loss": 1.654,
      "step": 370
    },
    {
      "epoch": 0.8492131616595135,
      "grad_norm": 0.4005778431892395,
      "learning_rate": 0.00019742834445707604,
      "loss": 1.6451,
      "step": 371
    },
    {
      "epoch": 0.8515021459227468,
      "grad_norm": 0.46752938628196716,
      "learning_rate": 0.00019741413913222097,
      "loss": 1.5222,
      "step": 372
    },
    {
      "epoch": 0.85379113018598,
      "grad_norm": 0.44197577238082886,
      "learning_rate": 0.0001973998951955136,
      "loss": 1.768,
      "step": 373
    },
    {
      "epoch": 0.8560801144492132,
      "grad_norm": 0.44156143069267273,
      "learning_rate": 0.0001973856126525997,
      "loss": 1.6501,
      "step": 374
    },
    {
      "epoch": 0.8583690987124464,
      "grad_norm": 0.4552460014820099,
      "learning_rate": 0.00019737129150914045,
      "loss": 1.62,
      "step": 375
    },
    {
      "epoch": 0.8606580829756796,
      "grad_norm": 0.4581928551197052,
      "learning_rate": 0.00019735693177081227,
      "loss": 1.5722,
      "step": 376
    },
    {
      "epoch": 0.8629470672389127,
      "grad_norm": 0.47525301575660706,
      "learning_rate": 0.00019734253344330692,
      "loss": 1.5152,
      "step": 377
    },
    {
      "epoch": 0.8652360515021459,
      "grad_norm": 0.4734356105327606,
      "learning_rate": 0.00019732809653233145,
      "loss": 1.4712,
      "step": 378
    },
    {
      "epoch": 0.8675250357653791,
      "grad_norm": 0.4875081181526184,
      "learning_rate": 0.00019731362104360818,
      "loss": 1.6434,
      "step": 379
    },
    {
      "epoch": 0.8698140200286123,
      "grad_norm": 0.5375803709030151,
      "learning_rate": 0.0001972991069828747,
      "loss": 1.7196,
      "step": 380
    },
    {
      "epoch": 0.8721030042918455,
      "grad_norm": 0.5160648822784424,
      "learning_rate": 0.00019728455435588394,
      "loss": 1.4908,
      "step": 381
    },
    {
      "epoch": 0.8743919885550787,
      "grad_norm": 0.5890928506851196,
      "learning_rate": 0.0001972699631684041,
      "loss": 1.6377,
      "step": 382
    },
    {
      "epoch": 0.8766809728183119,
      "grad_norm": 0.5135856866836548,
      "learning_rate": 0.0001972553334262187,
      "loss": 1.7282,
      "step": 383
    },
    {
      "epoch": 0.878969957081545,
      "grad_norm": 0.7008813619613647,
      "learning_rate": 0.00019724066513512645,
      "loss": 1.5348,
      "step": 384
    },
    {
      "epoch": 0.8812589413447782,
      "grad_norm": 0.5755431652069092,
      "learning_rate": 0.00019722595830094138,
      "loss": 1.3093,
      "step": 385
    },
    {
      "epoch": 0.8835479256080114,
      "grad_norm": 0.663464367389679,
      "learning_rate": 0.00019721121292949284,
      "loss": 1.283,
      "step": 386
    },
    {
      "epoch": 0.8858369098712446,
      "grad_norm": 0.597924530506134,
      "learning_rate": 0.00019719642902662544,
      "loss": 1.3398,
      "step": 387
    },
    {
      "epoch": 0.8881258941344778,
      "grad_norm": 0.6721014380455017,
      "learning_rate": 0.00019718160659819902,
      "loss": 1.4372,
      "step": 388
    },
    {
      "epoch": 0.890414878397711,
      "grad_norm": 0.7276092171669006,
      "learning_rate": 0.00019716674565008871,
      "loss": 1.3964,
      "step": 389
    },
    {
      "epoch": 0.8927038626609443,
      "grad_norm": 1.1685963869094849,
      "learning_rate": 0.00019715184618818493,
      "loss": 1.6498,
      "step": 390
    },
    {
      "epoch": 0.8949928469241774,
      "grad_norm": 0.9710104465484619,
      "learning_rate": 0.00019713690821839336,
      "loss": 1.2143,
      "step": 391
    },
    {
      "epoch": 0.8972818311874106,
      "grad_norm": 0.8027259111404419,
      "learning_rate": 0.00019712193174663492,
      "loss": 1.3245,
      "step": 392
    },
    {
      "epoch": 0.8995708154506438,
      "grad_norm": 0.7331583499908447,
      "learning_rate": 0.0001971069167788458,
      "loss": 1.4017,
      "step": 393
    },
    {
      "epoch": 0.901859799713877,
      "grad_norm": 1.003387689590454,
      "learning_rate": 0.00019709186332097747,
      "loss": 1.4934,
      "step": 394
    },
    {
      "epoch": 0.9041487839771102,
      "grad_norm": 1.0797269344329834,
      "learning_rate": 0.00019707677137899662,
      "loss": 1.6799,
      "step": 395
    },
    {
      "epoch": 0.9064377682403434,
      "grad_norm": 1.0520235300064087,
      "learning_rate": 0.00019706164095888525,
      "loss": 1.4325,
      "step": 396
    },
    {
      "epoch": 0.9087267525035765,
      "grad_norm": 1.1153618097305298,
      "learning_rate": 0.0001970464720666405,
      "loss": 1.5667,
      "step": 397
    },
    {
      "epoch": 0.9110157367668097,
      "grad_norm": 1.4900482892990112,
      "learning_rate": 0.00019703126470827496,
      "loss": 1.7062,
      "step": 398
    },
    {
      "epoch": 0.9133047210300429,
      "grad_norm": 1.7424468994140625,
      "learning_rate": 0.0001970160188898162,
      "loss": 1.8459,
      "step": 399
    },
    {
      "epoch": 0.9155937052932761,
      "grad_norm": 2.2614736557006836,
      "learning_rate": 0.00019700073461730727,
      "loss": 2.2238,
      "step": 400
    },
    {
      "epoch": 0.9178826895565093,
      "grad_norm": 1.1257100105285645,
      "learning_rate": 0.00019698541189680632,
      "loss": 2.017,
      "step": 401
    },
    {
      "epoch": 0.9201716738197425,
      "grad_norm": 1.1153912544250488,
      "learning_rate": 0.0001969700507343868,
      "loss": 1.9208,
      "step": 402
    },
    {
      "epoch": 0.9224606580829757,
      "grad_norm": 1.0910239219665527,
      "learning_rate": 0.0001969546511361374,
      "loss": 1.9502,
      "step": 403
    },
    {
      "epoch": 0.9247496423462088,
      "grad_norm": 0.6753953099250793,
      "learning_rate": 0.00019693921310816204,
      "loss": 1.8156,
      "step": 404
    },
    {
      "epoch": 0.927038626609442,
      "grad_norm": 0.6514906883239746,
      "learning_rate": 0.00019692373665657978,
      "loss": 1.8235,
      "step": 405
    },
    {
      "epoch": 0.9293276108726752,
      "grad_norm": 0.4767726957798004,
      "learning_rate": 0.0001969082217875251,
      "loss": 1.7842,
      "step": 406
    },
    {
      "epoch": 0.9316165951359084,
      "grad_norm": 0.41879355907440186,
      "learning_rate": 0.00019689266850714753,
      "loss": 1.9024,
      "step": 407
    },
    {
      "epoch": 0.9339055793991416,
      "grad_norm": 0.33241036534309387,
      "learning_rate": 0.0001968770768216119,
      "loss": 1.7242,
      "step": 408
    },
    {
      "epoch": 0.9361945636623749,
      "grad_norm": 0.34479382634162903,
      "learning_rate": 0.0001968614467370983,
      "loss": 1.6996,
      "step": 409
    },
    {
      "epoch": 0.9384835479256081,
      "grad_norm": 0.33567941188812256,
      "learning_rate": 0.00019684577825980192,
      "loss": 1.648,
      "step": 410
    },
    {
      "epoch": 0.9407725321888412,
      "grad_norm": 0.3407452702522278,
      "learning_rate": 0.00019683007139593333,
      "loss": 1.5817,
      "step": 411
    },
    {
      "epoch": 0.9430615164520744,
      "grad_norm": 0.35921311378479004,
      "learning_rate": 0.00019681432615171814,
      "loss": 1.8467,
      "step": 412
    },
    {
      "epoch": 0.9453505007153076,
      "grad_norm": 0.36816340684890747,
      "learning_rate": 0.00019679854253339737,
      "loss": 1.6223,
      "step": 413
    },
    {
      "epoch": 0.9476394849785408,
      "grad_norm": 0.39302071928977966,
      "learning_rate": 0.00019678272054722703,
      "loss": 1.6172,
      "step": 414
    },
    {
      "epoch": 0.949928469241774,
      "grad_norm": 0.40473082661628723,
      "learning_rate": 0.0001967668601994785,
      "loss": 1.7205,
      "step": 415
    },
    {
      "epoch": 0.9522174535050072,
      "grad_norm": 0.43281951546669006,
      "learning_rate": 0.00019675096149643833,
      "loss": 1.4452,
      "step": 416
    },
    {
      "epoch": 0.9545064377682403,
      "grad_norm": 0.5113329887390137,
      "learning_rate": 0.00019673502444440821,
      "loss": 1.6463,
      "step": 417
    },
    {
      "epoch": 0.9567954220314735,
      "grad_norm": 0.45617660880088806,
      "learning_rate": 0.00019671904904970515,
      "loss": 1.4523,
      "step": 418
    },
    {
      "epoch": 0.9590844062947067,
      "grad_norm": 0.4716404676437378,
      "learning_rate": 0.00019670303531866122,
      "loss": 1.7247,
      "step": 419
    },
    {
      "epoch": 0.9613733905579399,
      "grad_norm": 0.5593003630638123,
      "learning_rate": 0.0001966869832576238,
      "loss": 1.5577,
      "step": 420
    },
    {
      "epoch": 0.9636623748211731,
      "grad_norm": 0.513721227645874,
      "learning_rate": 0.00019667089287295537,
      "loss": 1.6073,
      "step": 421
    },
    {
      "epoch": 0.9659513590844063,
      "grad_norm": 0.6354144215583801,
      "learning_rate": 0.00019665476417103368,
      "loss": 1.6619,
      "step": 422
    },
    {
      "epoch": 0.9682403433476395,
      "grad_norm": 0.5961793661117554,
      "learning_rate": 0.00019663859715825162,
      "loss": 1.4519,
      "step": 423
    },
    {
      "epoch": 0.9705293276108726,
      "grad_norm": 0.6090859770774841,
      "learning_rate": 0.00019662239184101726,
      "loss": 1.4414,
      "step": 424
    },
    {
      "epoch": 0.9728183118741058,
      "grad_norm": 0.6027433276176453,
      "learning_rate": 0.0001966061482257539,
      "loss": 1.466,
      "step": 425
    },
    {
      "epoch": 0.975107296137339,
      "grad_norm": 0.6649966239929199,
      "learning_rate": 0.00019658986631889997,
      "loss": 1.6139,
      "step": 426
    },
    {
      "epoch": 0.9773962804005722,
      "grad_norm": 0.6814236044883728,
      "learning_rate": 0.0001965735461269091,
      "loss": 1.4921,
      "step": 427
    },
    {
      "epoch": 0.9796852646638055,
      "grad_norm": 0.731559157371521,
      "learning_rate": 0.0001965571876562501,
      "loss": 1.4538,
      "step": 428
    },
    {
      "epoch": 0.9819742489270387,
      "grad_norm": 0.7567864060401917,
      "learning_rate": 0.00019654079091340696,
      "loss": 1.3168,
      "step": 429
    },
    {
      "epoch": 0.9842632331902719,
      "grad_norm": 0.7701776623725891,
      "learning_rate": 0.0001965243559048788,
      "loss": 1.2846,
      "step": 430
    },
    {
      "epoch": 0.986552217453505,
      "grad_norm": 0.8128049969673157,
      "learning_rate": 0.00019650788263717993,
      "loss": 1.4284,
      "step": 431
    },
    {
      "epoch": 0.9888412017167382,
      "grad_norm": 0.9019891619682312,
      "learning_rate": 0.00019649137111683981,
      "loss": 1.6122,
      "step": 432
    },
    {
      "epoch": 0.9911301859799714,
      "grad_norm": 1.0137925148010254,
      "learning_rate": 0.00019647482135040314,
      "loss": 1.4949,
      "step": 433
    },
    {
      "epoch": 0.9934191702432046,
      "grad_norm": 1.0654815435409546,
      "learning_rate": 0.00019645823334442968,
      "loss": 1.4488,
      "step": 434
    },
    {
      "epoch": 0.9957081545064378,
      "grad_norm": 1.2546265125274658,
      "learning_rate": 0.00019644160710549437,
      "loss": 1.7037,
      "step": 435
    },
    {
      "epoch": 0.997997138769671,
      "grad_norm": 1.871845006942749,
      "learning_rate": 0.00019642494264018737,
      "loss": 1.5653,
      "step": 436
    },
    {
      "epoch": 1.0014306151645207,
      "grad_norm": 2.864191770553589,
      "learning_rate": 0.00019640823995511388,
      "loss": 2.9079,
      "step": 437
    },
    {
      "epoch": 1.003719599427754,
      "grad_norm": 0.775518000125885,
      "learning_rate": 0.00019639149905689433,
      "loss": 1.8569,
      "step": 438
    },
    {
      "epoch": 1.0060085836909871,
      "grad_norm": 0.8842045664787292,
      "learning_rate": 0.00019637471995216432,
      "loss": 2.0388,
      "step": 439
    },
    {
      "epoch": 1.0082975679542203,
      "grad_norm": 0.8512495756149292,
      "learning_rate": 0.0001963579026475745,
      "loss": 1.7849,
      "step": 440
    },
    {
      "epoch": 1.0105865522174535,
      "grad_norm": 0.5596434473991394,
      "learning_rate": 0.00019634104714979077,
      "loss": 1.6806,
      "step": 441
    },
    {
      "epoch": 1.0128755364806867,
      "grad_norm": 0.5454633235931396,
      "learning_rate": 0.00019632415346549403,
      "loss": 1.8488,
      "step": 442
    },
    {
      "epoch": 1.01516452074392,
      "grad_norm": 0.3954569399356842,
      "learning_rate": 0.00019630722160138045,
      "loss": 1.7624,
      "step": 443
    },
    {
      "epoch": 1.0174535050071531,
      "grad_norm": 0.3752392530441284,
      "learning_rate": 0.00019629025156416127,
      "loss": 1.7142,
      "step": 444
    },
    {
      "epoch": 1.0197424892703864,
      "grad_norm": 0.35623693466186523,
      "learning_rate": 0.00019627324336056289,
      "loss": 1.6697,
      "step": 445
    },
    {
      "epoch": 1.0220314735336196,
      "grad_norm": 0.3730758726596832,
      "learning_rate": 0.00019625619699732675,
      "loss": 1.7098,
      "step": 446
    },
    {
      "epoch": 1.0243204577968525,
      "grad_norm": 0.3285910487174988,
      "learning_rate": 0.00019623911248120956,
      "loss": 1.6269,
      "step": 447
    },
    {
      "epoch": 1.0266094420600858,
      "grad_norm": 0.3648667335510254,
      "learning_rate": 0.00019622198981898302,
      "loss": 1.6538,
      "step": 448
    },
    {
      "epoch": 1.028898426323319,
      "grad_norm": 0.341791033744812,
      "learning_rate": 0.00019620482901743408,
      "loss": 1.668,
      "step": 449
    },
    {
      "epoch": 1.0311874105865522,
      "grad_norm": 0.34298187494277954,
      "learning_rate": 0.00019618763008336467,
      "loss": 1.5905,
      "step": 450
    },
    {
      "epoch": 1.0334763948497854,
      "grad_norm": 0.326028972864151,
      "learning_rate": 0.00019617039302359193,
      "loss": 1.496,
      "step": 451
    },
    {
      "epoch": 1.0357653791130186,
      "grad_norm": 0.3477809429168701,
      "learning_rate": 0.00019615311784494806,
      "loss": 1.664,
      "step": 452
    },
    {
      "epoch": 1.0380543633762518,
      "grad_norm": 0.36962223052978516,
      "learning_rate": 0.00019613580455428035,
      "loss": 1.6236,
      "step": 453
    },
    {
      "epoch": 1.040343347639485,
      "grad_norm": 0.3837951719760895,
      "learning_rate": 0.0001961184531584513,
      "loss": 1.5608,
      "step": 454
    },
    {
      "epoch": 1.0426323319027182,
      "grad_norm": 0.3928857147693634,
      "learning_rate": 0.00019610106366433846,
      "loss": 1.5334,
      "step": 455
    },
    {
      "epoch": 1.0449213161659514,
      "grad_norm": 0.4714345335960388,
      "learning_rate": 0.00019608363607883443,
      "loss": 1.5982,
      "step": 456
    },
    {
      "epoch": 1.0472103004291846,
      "grad_norm": 0.4616788923740387,
      "learning_rate": 0.00019606617040884695,
      "loss": 1.4667,
      "step": 457
    },
    {
      "epoch": 1.0494992846924178,
      "grad_norm": 0.447256863117218,
      "learning_rate": 0.00019604866666129885,
      "loss": 1.3407,
      "step": 458
    },
    {
      "epoch": 1.051788268955651,
      "grad_norm": 0.4581814110279083,
      "learning_rate": 0.0001960311248431281,
      "loss": 1.472,
      "step": 459
    },
    {
      "epoch": 1.054077253218884,
      "grad_norm": 0.46417236328125,
      "learning_rate": 0.00019601354496128764,
      "loss": 1.4983,
      "step": 460
    },
    {
      "epoch": 1.0563662374821172,
      "grad_norm": 0.46364760398864746,
      "learning_rate": 0.00019599592702274562,
      "loss": 1.3284,
      "step": 461
    },
    {
      "epoch": 1.0586552217453504,
      "grad_norm": 0.5039281845092773,
      "learning_rate": 0.00019597827103448525,
      "loss": 1.5006,
      "step": 462
    },
    {
      "epoch": 1.0609442060085836,
      "grad_norm": 0.4904272258281708,
      "learning_rate": 0.00019596057700350475,
      "loss": 1.296,
      "step": 463
    },
    {
      "epoch": 1.0632331902718168,
      "grad_norm": 0.5611284971237183,
      "learning_rate": 0.0001959428449368175,
      "loss": 1.3408,
      "step": 464
    },
    {
      "epoch": 1.06552217453505,
      "grad_norm": 0.5019165873527527,
      "learning_rate": 0.00019592507484145192,
      "loss": 1.3584,
      "step": 465
    },
    {
      "epoch": 1.0678111587982833,
      "grad_norm": 0.5802716612815857,
      "learning_rate": 0.00019590726672445152,
      "loss": 1.4471,
      "step": 466
    },
    {
      "epoch": 1.0701001430615165,
      "grad_norm": 0.6091588735580444,
      "learning_rate": 0.00019588942059287483,
      "loss": 1.3627,
      "step": 467
    },
    {
      "epoch": 1.0723891273247497,
      "grad_norm": 0.679551899433136,
      "learning_rate": 0.00019587153645379552,
      "loss": 1.4522,
      "step": 468
    },
    {
      "epoch": 1.0746781115879829,
      "grad_norm": 0.6207999587059021,
      "learning_rate": 0.00019585361431430226,
      "loss": 1.1738,
      "step": 469
    },
    {
      "epoch": 1.076967095851216,
      "grad_norm": 0.7162473201751709,
      "learning_rate": 0.00019583565418149887,
      "loss": 1.3295,
      "step": 470
    },
    {
      "epoch": 1.0792560801144493,
      "grad_norm": 0.7932853698730469,
      "learning_rate": 0.0001958176560625041,
      "loss": 1.3477,
      "step": 471
    },
    {
      "epoch": 1.0815450643776825,
      "grad_norm": 0.7633782625198364,
      "learning_rate": 0.00019579961996445187,
      "loss": 1.2986,
      "step": 472
    },
    {
      "epoch": 1.0838340486409157,
      "grad_norm": 0.8132832646369934,
      "learning_rate": 0.00019578154589449117,
      "loss": 1.2373,
      "step": 473
    },
    {
      "epoch": 1.0861230329041487,
      "grad_norm": 0.9217791557312012,
      "learning_rate": 0.00019576343385978588,
      "loss": 1.6054,
      "step": 474
    },
    {
      "epoch": 1.088412017167382,
      "grad_norm": 0.9242892861366272,
      "learning_rate": 0.00019574528386751507,
      "loss": 1.2478,
      "step": 475
    },
    {
      "epoch": 1.090701001430615,
      "grad_norm": 0.8688008189201355,
      "learning_rate": 0.00019572709592487284,
      "loss": 0.9066,
      "step": 476
    },
    {
      "epoch": 1.0929899856938483,
      "grad_norm": 1.0369139909744263,
      "learning_rate": 0.00019570887003906825,
      "loss": 1.0952,
      "step": 477
    },
    {
      "epoch": 1.0952789699570815,
      "grad_norm": 0.9730405211448669,
      "learning_rate": 0.00019569060621732554,
      "loss": 0.999,
      "step": 478
    },
    {
      "epoch": 1.0975679542203147,
      "grad_norm": 1.1482276916503906,
      "learning_rate": 0.00019567230446688387,
      "loss": 1.14,
      "step": 479
    },
    {
      "epoch": 1.099856938483548,
      "grad_norm": 1.2615000009536743,
      "learning_rate": 0.00019565396479499745,
      "loss": 0.9726,
      "step": 480
    },
    {
      "epoch": 1.1021459227467811,
      "grad_norm": 1.3929799795150757,
      "learning_rate": 0.00019563558720893555,
      "loss": 0.9384,
      "step": 481
    },
    {
      "epoch": 1.1044349070100143,
      "grad_norm": 1.6762079000473022,
      "learning_rate": 0.00019561717171598248,
      "loss": 1.145,
      "step": 482
    },
    {
      "epoch": 1.1067238912732476,
      "grad_norm": 1.7647395133972168,
      "learning_rate": 0.00019559871832343752,
      "loss": 1.2415,
      "step": 483
    },
    {
      "epoch": 1.1090128755364808,
      "grad_norm": 1.9230117797851562,
      "learning_rate": 0.00019558022703861503,
      "loss": 0.9165,
      "step": 484
    },
    {
      "epoch": 1.111301859799714,
      "grad_norm": 3.1017725467681885,
      "learning_rate": 0.00019556169786884436,
      "loss": 1.6217,
      "step": 485
    },
    {
      "epoch": 1.113590844062947,
      "grad_norm": 3.4964425563812256,
      "learning_rate": 0.00019554313082146988,
      "loss": 1.3126,
      "step": 486
    },
    {
      "epoch": 1.1158798283261802,
      "grad_norm": 2.408186912536621,
      "learning_rate": 0.00019552452590385097,
      "loss": 1.6793,
      "step": 487
    },
    {
      "epoch": 1.1181688125894134,
      "grad_norm": 1.1658360958099365,
      "learning_rate": 0.000195505883123362,
      "loss": 2.0857,
      "step": 488
    },
    {
      "epoch": 1.1204577968526466,
      "grad_norm": 1.1001274585723877,
      "learning_rate": 0.0001954872024873924,
      "loss": 1.8623,
      "step": 489
    },
    {
      "epoch": 1.1227467811158798,
      "grad_norm": 1.0129964351654053,
      "learning_rate": 0.0001954684840033466,
      "loss": 2.0119,
      "step": 490
    },
    {
      "epoch": 1.125035765379113,
      "grad_norm": 0.8631079792976379,
      "learning_rate": 0.00019544972767864398,
      "loss": 1.8473,
      "step": 491
    },
    {
      "epoch": 1.1273247496423462,
      "grad_norm": 0.6965784430503845,
      "learning_rate": 0.00019543093352071894,
      "loss": 1.9179,
      "step": 492
    },
    {
      "epoch": 1.1296137339055794,
      "grad_norm": 0.5939742922782898,
      "learning_rate": 0.00019541210153702093,
      "loss": 1.7868,
      "step": 493
    },
    {
      "epoch": 1.1319027181688126,
      "grad_norm": 0.501757800579071,
      "learning_rate": 0.00019539323173501427,
      "loss": 1.6728,
      "step": 494
    },
    {
      "epoch": 1.1341917024320458,
      "grad_norm": 0.48351621627807617,
      "learning_rate": 0.00019537432412217843,
      "loss": 1.6041,
      "step": 495
    },
    {
      "epoch": 1.136480686695279,
      "grad_norm": 0.4757158160209656,
      "learning_rate": 0.00019535537870600772,
      "loss": 1.7114,
      "step": 496
    },
    {
      "epoch": 1.1387696709585122,
      "grad_norm": 0.49289125204086304,
      "learning_rate": 0.0001953363954940115,
      "loss": 1.7379,
      "step": 497
    },
    {
      "epoch": 1.1410586552217454,
      "grad_norm": 0.44323959946632385,
      "learning_rate": 0.00019531737449371417,
      "loss": 1.5315,
      "step": 498
    },
    {
      "epoch": 1.1433476394849786,
      "grad_norm": 0.47619810700416565,
      "learning_rate": 0.00019529831571265497,
      "loss": 1.652,
      "step": 499
    },
    {
      "epoch": 1.1456366237482118,
      "grad_norm": 0.4648391604423523,
      "learning_rate": 0.00019527921915838827,
      "loss": 1.6536,
      "step": 500
    },
    {
      "epoch": 1.1456366237482118,
      "eval_loss": 1.6775344610214233,
      "eval_runtime": 25.8892,
      "eval_samples_per_second": 28.429,
      "eval_steps_per_second": 14.214,
      "step": 500
    },
    {
      "epoch": 1.1479256080114448,
      "grad_norm": 0.45624008774757385,
      "learning_rate": 0.00019526008483848327,
      "loss": 1.6079,
      "step": 501
    },
    {
      "epoch": 1.150214592274678,
      "grad_norm": 0.494411826133728,
      "learning_rate": 0.00019524091276052423,
      "loss": 1.5547,
      "step": 502
    },
    {
      "epoch": 1.1525035765379112,
      "grad_norm": 0.4686802923679352,
      "learning_rate": 0.00019522170293211034,
      "loss": 1.5446,
      "step": 503
    },
    {
      "epoch": 1.1547925608011445,
      "grad_norm": 0.46277472376823425,
      "learning_rate": 0.00019520245536085574,
      "loss": 1.6861,
      "step": 504
    },
    {
      "epoch": 1.1570815450643777,
      "grad_norm": 0.4636980891227722,
      "learning_rate": 0.00019518317005438963,
      "loss": 1.4744,
      "step": 505
    },
    {
      "epoch": 1.1593705293276109,
      "grad_norm": 0.4736781120300293,
      "learning_rate": 0.00019516384702035602,
      "loss": 1.5869,
      "step": 506
    },
    {
      "epoch": 1.161659513590844,
      "grad_norm": 0.5312774777412415,
      "learning_rate": 0.00019514448626641395,
      "loss": 1.7566,
      "step": 507
    },
    {
      "epoch": 1.1639484978540773,
      "grad_norm": 0.4982118606567383,
      "learning_rate": 0.00019512508780023744,
      "loss": 1.4993,
      "step": 508
    },
    {
      "epoch": 1.1662374821173105,
      "grad_norm": 0.5388206839561462,
      "learning_rate": 0.00019510565162951537,
      "loss": 1.5231,
      "step": 509
    },
    {
      "epoch": 1.1685264663805437,
      "grad_norm": 0.54203861951828,
      "learning_rate": 0.00019508617776195167,
      "loss": 1.3626,
      "step": 510
    },
    {
      "epoch": 1.170815450643777,
      "grad_norm": 0.5578249096870422,
      "learning_rate": 0.0001950666662052651,
      "loss": 1.5756,
      "step": 511
    },
    {
      "epoch": 1.17310443490701,
      "grad_norm": 0.557388186454773,
      "learning_rate": 0.00019504711696718947,
      "loss": 1.348,
      "step": 512
    },
    {
      "epoch": 1.175393419170243,
      "grad_norm": 0.6537132859230042,
      "learning_rate": 0.00019502753005547346,
      "loss": 1.6266,
      "step": 513
    },
    {
      "epoch": 1.1776824034334763,
      "grad_norm": 0.5653935074806213,
      "learning_rate": 0.00019500790547788072,
      "loss": 1.2826,
      "step": 514
    },
    {
      "epoch": 1.1799713876967095,
      "grad_norm": 0.6156082153320312,
      "learning_rate": 0.0001949882432421897,
      "loss": 1.3668,
      "step": 515
    },
    {
      "epoch": 1.1822603719599427,
      "grad_norm": 0.6442597508430481,
      "learning_rate": 0.00019496854335619403,
      "loss": 1.6094,
      "step": 516
    },
    {
      "epoch": 1.184549356223176,
      "grad_norm": 0.6525452136993408,
      "learning_rate": 0.00019494880582770203,
      "loss": 1.3442,
      "step": 517
    },
    {
      "epoch": 1.1868383404864091,
      "grad_norm": 0.7586766481399536,
      "learning_rate": 0.000194929030664537,
      "loss": 1.3797,
      "step": 518
    },
    {
      "epoch": 1.1891273247496423,
      "grad_norm": 0.7124922275543213,
      "learning_rate": 0.00019490921787453726,
      "loss": 1.406,
      "step": 519
    },
    {
      "epoch": 1.1914163090128755,
      "grad_norm": 0.8126552700996399,
      "learning_rate": 0.00019488936746555593,
      "loss": 1.3408,
      "step": 520
    },
    {
      "epoch": 1.1937052932761087,
      "grad_norm": 0.7620358467102051,
      "learning_rate": 0.00019486947944546106,
      "loss": 1.2841,
      "step": 521
    },
    {
      "epoch": 1.195994277539342,
      "grad_norm": 0.8049248456954956,
      "learning_rate": 0.00019484955382213564,
      "loss": 1.2457,
      "step": 522
    },
    {
      "epoch": 1.1982832618025752,
      "grad_norm": 0.8273380994796753,
      "learning_rate": 0.0001948295906034776,
      "loss": 1.1049,
      "step": 523
    },
    {
      "epoch": 1.2005722460658084,
      "grad_norm": 0.8063210248947144,
      "learning_rate": 0.0001948095897973996,
      "loss": 1.1276,
      "step": 524
    },
    {
      "epoch": 1.2028612303290416,
      "grad_norm": 0.9249168038368225,
      "learning_rate": 0.00019478955141182945,
      "loss": 1.1685,
      "step": 525
    },
    {
      "epoch": 1.2051502145922748,
      "grad_norm": 0.9692345857620239,
      "learning_rate": 0.00019476947545470967,
      "loss": 1.0298,
      "step": 526
    },
    {
      "epoch": 1.207439198855508,
      "grad_norm": 1.1224550008773804,
      "learning_rate": 0.00019474936193399772,
      "loss": 0.9859,
      "step": 527
    },
    {
      "epoch": 1.209728183118741,
      "grad_norm": 1.019800066947937,
      "learning_rate": 0.00019472921085766598,
      "loss": 1.0048,
      "step": 528
    },
    {
      "epoch": 1.2120171673819742,
      "grad_norm": 1.1730016469955444,
      "learning_rate": 0.00019470902223370166,
      "loss": 1.1061,
      "step": 529
    },
    {
      "epoch": 1.2143061516452074,
      "grad_norm": 1.1287165880203247,
      "learning_rate": 0.0001946887960701069,
      "loss": 1.0994,
      "step": 530
    },
    {
      "epoch": 1.2165951359084406,
      "grad_norm": 1.3811784982681274,
      "learning_rate": 0.00019466853237489872,
      "loss": 1.2046,
      "step": 531
    },
    {
      "epoch": 1.2188841201716738,
      "grad_norm": 1.5227758884429932,
      "learning_rate": 0.000194648231156109,
      "loss": 1.3211,
      "step": 532
    },
    {
      "epoch": 1.221173104434907,
      "grad_norm": 1.41282057762146,
      "learning_rate": 0.00019462789242178448,
      "loss": 1.0281,
      "step": 533
    },
    {
      "epoch": 1.2234620886981402,
      "grad_norm": 1.4078859090805054,
      "learning_rate": 0.0001946075161799868,
      "loss": 0.9925,
      "step": 534
    },
    {
      "epoch": 1.2257510729613734,
      "grad_norm": 2.359815835952759,
      "learning_rate": 0.00019458710243879242,
      "loss": 1.312,
      "step": 535
    },
    {
      "epoch": 1.2280400572246066,
      "grad_norm": 3.5413224697113037,
      "learning_rate": 0.00019456665120629274,
      "loss": 1.307,
      "step": 536
    },
    {
      "epoch": 1.2303290414878398,
      "grad_norm": 2.105780839920044,
      "learning_rate": 0.00019454616249059395,
      "loss": 1.8708,
      "step": 537
    },
    {
      "epoch": 1.232618025751073,
      "grad_norm": 0.6073758006095886,
      "learning_rate": 0.00019452563629981712,
      "loss": 2.1078,
      "step": 538
    },
    {
      "epoch": 1.234907010014306,
      "grad_norm": 0.5563433766365051,
      "learning_rate": 0.00019450507264209813,
      "loss": 1.9778,
      "step": 539
    },
    {
      "epoch": 1.2371959942775392,
      "grad_norm": 0.5401256680488586,
      "learning_rate": 0.0001944844715255879,
      "loss": 1.8105,
      "step": 540
    },
    {
      "epoch": 1.2394849785407724,
      "grad_norm": 0.5743811726570129,
      "learning_rate": 0.00019446383295845189,
      "loss": 1.8653,
      "step": 541
    },
    {
      "epoch": 1.2417739628040056,
      "grad_norm": 0.5073881149291992,
      "learning_rate": 0.00019444315694887068,
      "loss": 1.9549,
      "step": 542
    },
    {
      "epoch": 1.2440629470672389,
      "grad_norm": 0.5103323459625244,
      "learning_rate": 0.0001944224435050395,
      "loss": 1.8139,
      "step": 543
    },
    {
      "epoch": 1.246351931330472,
      "grad_norm": 0.49275580048561096,
      "learning_rate": 0.00019440169263516856,
      "loss": 1.6349,
      "step": 544
    },
    {
      "epoch": 1.2486409155937053,
      "grad_norm": 0.5180409550666809,
      "learning_rate": 0.00019438090434748285,
      "loss": 1.7747,
      "step": 545
    },
    {
      "epoch": 1.2509298998569385,
      "grad_norm": 0.5536676645278931,
      "learning_rate": 0.00019436007865022213,
      "loss": 1.6815,
      "step": 546
    },
    {
      "epoch": 1.2532188841201717,
      "grad_norm": 0.5071956515312195,
      "learning_rate": 0.00019433921555164104,
      "loss": 1.6834,
      "step": 547
    },
    {
      "epoch": 1.255507868383405,
      "grad_norm": 0.5304872989654541,
      "learning_rate": 0.00019431831506000908,
      "loss": 1.5739,
      "step": 548
    },
    {
      "epoch": 1.257796852646638,
      "grad_norm": 0.47479552030563354,
      "learning_rate": 0.00019429737718361053,
      "loss": 1.6643,
      "step": 549
    },
    {
      "epoch": 1.2600858369098713,
      "grad_norm": 0.4885936677455902,
      "learning_rate": 0.0001942764019307445,
      "loss": 1.8093,
      "step": 550
    },
    {
      "epoch": 1.2623748211731045,
      "grad_norm": 0.47408992052078247,
      "learning_rate": 0.0001942553893097249,
      "loss": 1.5076,
      "step": 551
    },
    {
      "epoch": 1.2646638054363377,
      "grad_norm": 0.5412789583206177,
      "learning_rate": 0.00019423433932888045,
      "loss": 1.7633,
      "step": 552
    },
    {
      "epoch": 1.266952789699571,
      "grad_norm": 0.5159979462623596,
      "learning_rate": 0.0001942132519965547,
      "loss": 1.4465,
      "step": 553
    },
    {
      "epoch": 1.2692417739628041,
      "grad_norm": 0.5230864882469177,
      "learning_rate": 0.00019419212732110598,
      "loss": 1.6595,
      "step": 554
    },
    {
      "epoch": 1.2715307582260371,
      "grad_norm": 0.5039998292922974,
      "learning_rate": 0.0001941709653109075,
      "loss": 1.4004,
      "step": 555
    },
    {
      "epoch": 1.2738197424892703,
      "grad_norm": 0.5318018794059753,
      "learning_rate": 0.00019414976597434714,
      "loss": 1.5194,
      "step": 556
    },
    {
      "epoch": 1.2761087267525035,
      "grad_norm": 0.5358762145042419,
      "learning_rate": 0.00019412852931982767,
      "loss": 1.6229,
      "step": 557
    },
    {
      "epoch": 1.2783977110157367,
      "grad_norm": 0.5722163319587708,
      "learning_rate": 0.00019410725535576656,
      "loss": 1.5816,
      "step": 558
    },
    {
      "epoch": 1.28068669527897,
      "grad_norm": 0.6073668599128723,
      "learning_rate": 0.0001940859440905962,
      "loss": 1.332,
      "step": 559
    },
    {
      "epoch": 1.2829756795422032,
      "grad_norm": 0.603227972984314,
      "learning_rate": 0.0001940645955327637,
      "loss": 1.3413,
      "step": 560
    },
    {
      "epoch": 1.2852646638054364,
      "grad_norm": 0.6240841150283813,
      "learning_rate": 0.00019404320969073092,
      "loss": 1.4942,
      "step": 561
    },
    {
      "epoch": 1.2875536480686696,
      "grad_norm": 0.6479021310806274,
      "learning_rate": 0.0001940217865729745,
      "loss": 1.3553,
      "step": 562
    },
    {
      "epoch": 1.2898426323319028,
      "grad_norm": 0.6595821380615234,
      "learning_rate": 0.00019400032618798594,
      "loss": 1.4632,
      "step": 563
    },
    {
      "epoch": 1.292131616595136,
      "grad_norm": 0.6678495407104492,
      "learning_rate": 0.0001939788285442714,
      "loss": 1.3536,
      "step": 564
    },
    {
      "epoch": 1.294420600858369,
      "grad_norm": 0.6886152625083923,
      "learning_rate": 0.00019395729365035184,
      "loss": 1.4667,
      "step": 565
    },
    {
      "epoch": 1.2967095851216022,
      "grad_norm": 0.7226461172103882,
      "learning_rate": 0.00019393572151476306,
      "loss": 1.3286,
      "step": 566
    },
    {
      "epoch": 1.2989985693848354,
      "grad_norm": 0.7189615368843079,
      "learning_rate": 0.00019391411214605557,
      "loss": 1.479,
      "step": 567
    },
    {
      "epoch": 1.3012875536480686,
      "grad_norm": 0.7047454118728638,
      "learning_rate": 0.0001938924655527946,
      "loss": 1.2467,
      "step": 568
    },
    {
      "epoch": 1.3035765379113018,
      "grad_norm": 0.7491849064826965,
      "learning_rate": 0.00019387078174356014,
      "loss": 1.3162,
      "step": 569
    },
    {
      "epoch": 1.305865522174535,
      "grad_norm": 0.8178296089172363,
      "learning_rate": 0.00019384906072694702,
      "loss": 1.334,
      "step": 570
    },
    {
      "epoch": 1.3081545064377682,
      "grad_norm": 0.8624579310417175,
      "learning_rate": 0.00019382730251156472,
      "loss": 1.3066,
      "step": 571
    },
    {
      "epoch": 1.3104434907010014,
      "grad_norm": 0.8471712470054626,
      "learning_rate": 0.0001938055071060375,
      "loss": 1.1664,
      "step": 572
    },
    {
      "epoch": 1.3127324749642346,
      "grad_norm": 0.9259144067764282,
      "learning_rate": 0.0001937836745190044,
      "loss": 1.2577,
      "step": 573
    },
    {
      "epoch": 1.3150214592274678,
      "grad_norm": 0.9440982937812805,
      "learning_rate": 0.0001937618047591191,
      "loss": 1.1096,
      "step": 574
    },
    {
      "epoch": 1.317310443490701,
      "grad_norm": 0.9747737050056458,
      "learning_rate": 0.00019373989783505016,
      "loss": 1.0136,
      "step": 575
    },
    {
      "epoch": 1.3195994277539342,
      "grad_norm": 1.135636329650879,
      "learning_rate": 0.0001937179537554807,
      "loss": 1.1614,
      "step": 576
    },
    {
      "epoch": 1.3218884120171674,
      "grad_norm": 1.0675294399261475,
      "learning_rate": 0.00019369597252910867,
      "loss": 1.0435,
      "step": 577
    },
    {
      "epoch": 1.3241773962804007,
      "grad_norm": 1.1525951623916626,
      "learning_rate": 0.00019367395416464676,
      "loss": 1.1846,
      "step": 578
    },
    {
      "epoch": 1.3264663805436339,
      "grad_norm": 1.0978894233703613,
      "learning_rate": 0.0001936518986708223,
      "loss": 1.0494,
      "step": 579
    },
    {
      "epoch": 1.328755364806867,
      "grad_norm": 1.1870222091674805,
      "learning_rate": 0.00019362980605637747,
      "loss": 1.2362,
      "step": 580
    },
    {
      "epoch": 1.3310443490701003,
      "grad_norm": 1.2120845317840576,
      "learning_rate": 0.00019360767633006898,
      "loss": 1.1418,
      "step": 581
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 1.383846402168274,
      "learning_rate": 0.0001935855095006684,
      "loss": 1.0872,
      "step": 582
    },
    {
      "epoch": 1.3356223175965665,
      "grad_norm": 1.712051272392273,
      "learning_rate": 0.00019356330557696193,
      "loss": 1.458,
      "step": 583
    },
    {
      "epoch": 1.3379113018597997,
      "grad_norm": 1.710627794265747,
      "learning_rate": 0.0001935410645677505,
      "loss": 1.0651,
      "step": 584
    },
    {
      "epoch": 1.3402002861230329,
      "grad_norm": 2.2404611110687256,
      "learning_rate": 0.00019351878648184972,
      "loss": 1.0784,
      "step": 585
    },
    {
      "epoch": 1.342489270386266,
      "grad_norm": 2.795382261276245,
      "learning_rate": 0.00019349647132808995,
      "loss": 1.4641,
      "step": 586
    },
    {
      "epoch": 1.3447782546494993,
      "grad_norm": 1.5983691215515137,
      "learning_rate": 0.00019347411911531617,
      "loss": 1.8025,
      "step": 587
    },
    {
      "epoch": 1.3470672389127325,
      "grad_norm": 0.42542368173599243,
      "learning_rate": 0.00019345172985238808,
      "loss": 1.808,
      "step": 588
    },
    {
      "epoch": 1.3493562231759657,
      "grad_norm": 0.4543238580226898,
      "learning_rate": 0.00019342930354818014,
      "loss": 2.0425,
      "step": 589
    },
    {
      "epoch": 1.351645207439199,
      "grad_norm": 0.4951767325401306,
      "learning_rate": 0.0001934068402115813,
      "loss": 1.8235,
      "step": 590
    },
    {
      "epoch": 1.3539341917024321,
      "grad_norm": 0.5128828883171082,
      "learning_rate": 0.00019338433985149542,
      "loss": 1.8303,
      "step": 591
    },
    {
      "epoch": 1.356223175965665,
      "grad_norm": 0.5125733017921448,
      "learning_rate": 0.00019336180247684089,
      "loss": 1.7866,
      "step": 592
    },
    {
      "epoch": 1.3585121602288983,
      "grad_norm": 0.5207964181900024,
      "learning_rate": 0.00019333922809655073,
      "loss": 1.788,
      "step": 593
    },
    {
      "epoch": 1.3608011444921315,
      "grad_norm": 0.4681180715560913,
      "learning_rate": 0.0001933166167195728,
      "loss": 1.7011,
      "step": 594
    },
    {
      "epoch": 1.3630901287553647,
      "grad_norm": 0.5094600319862366,
      "learning_rate": 0.0001932939683548695,
      "loss": 1.854,
      "step": 595
    },
    {
      "epoch": 1.365379113018598,
      "grad_norm": 0.4620515704154968,
      "learning_rate": 0.00019327128301141792,
      "loss": 1.7655,
      "step": 596
    },
    {
      "epoch": 1.3676680972818311,
      "grad_norm": 0.4834456741809845,
      "learning_rate": 0.0001932485606982098,
      "loss": 1.6682,
      "step": 597
    },
    {
      "epoch": 1.3699570815450643,
      "grad_norm": 0.4840511381626129,
      "learning_rate": 0.00019322580142425147,
      "loss": 1.7881,
      "step": 598
    },
    {
      "epoch": 1.3722460658082976,
      "grad_norm": 0.4967823922634125,
      "learning_rate": 0.00019320300519856413,
      "loss": 1.5658,
      "step": 599
    },
    {
      "epoch": 1.3745350500715308,
      "grad_norm": 0.5048989653587341,
      "learning_rate": 0.00019318017203018336,
      "loss": 1.6951,
      "step": 600
    },
    {
      "epoch": 1.376824034334764,
      "grad_norm": 0.477652370929718,
      "learning_rate": 0.00019315730192815954,
      "loss": 1.5982,
      "step": 601
    },
    {
      "epoch": 1.3791130185979972,
      "grad_norm": 0.5057041645050049,
      "learning_rate": 0.0001931343949015576,
      "loss": 1.7799,
      "step": 602
    },
    {
      "epoch": 1.3814020028612304,
      "grad_norm": 0.48348110914230347,
      "learning_rate": 0.00019311145095945718,
      "loss": 1.5,
      "step": 603
    },
    {
      "epoch": 1.3836909871244636,
      "grad_norm": 0.5346238613128662,
      "learning_rate": 0.00019308847011095259,
      "loss": 1.6912,
      "step": 604
    },
    {
      "epoch": 1.3859799713876968,
      "grad_norm": 0.5346901416778564,
      "learning_rate": 0.00019306545236515263,
      "loss": 1.5991,
      "step": 605
    },
    {
      "epoch": 1.38826895565093,
      "grad_norm": 0.5398430824279785,
      "learning_rate": 0.00019304239773118082,
      "loss": 1.5279,
      "step": 606
    },
    {
      "epoch": 1.3905579399141632,
      "grad_norm": 0.5098668932914734,
      "learning_rate": 0.00019301930621817528,
      "loss": 1.5044,
      "step": 607
    },
    {
      "epoch": 1.3928469241773964,
      "grad_norm": 0.5303160548210144,
      "learning_rate": 0.00019299617783528872,
      "loss": 1.5106,
      "step": 608
    },
    {
      "epoch": 1.3951359084406294,
      "grad_norm": 0.5479273200035095,
      "learning_rate": 0.00019297301259168855,
      "loss": 1.5519,
      "step": 609
    },
    {
      "epoch": 1.3974248927038626,
      "grad_norm": 0.5518145561218262,
      "learning_rate": 0.00019294981049655668,
      "loss": 1.4529,
      "step": 610
    },
    {
      "epoch": 1.3997138769670958,
      "grad_norm": 0.5993743538856506,
      "learning_rate": 0.00019292657155908972,
      "loss": 1.5277,
      "step": 611
    },
    {
      "epoch": 1.402002861230329,
      "grad_norm": 0.5985834002494812,
      "learning_rate": 0.00019290329578849883,
      "loss": 1.4888,
      "step": 612
    },
    {
      "epoch": 1.4042918454935622,
      "grad_norm": 0.5919128656387329,
      "learning_rate": 0.00019287998319400976,
      "loss": 1.4202,
      "step": 613
    },
    {
      "epoch": 1.4065808297567954,
      "grad_norm": 0.6324314475059509,
      "learning_rate": 0.00019285663378486287,
      "loss": 1.5408,
      "step": 614
    },
    {
      "epoch": 1.4088698140200286,
      "grad_norm": 0.6207531094551086,
      "learning_rate": 0.00019283324757031318,
      "loss": 1.3136,
      "step": 615
    },
    {
      "epoch": 1.4111587982832619,
      "grad_norm": 0.6301514506340027,
      "learning_rate": 0.0001928098245596302,
      "loss": 1.2727,
      "step": 616
    },
    {
      "epoch": 1.413447782546495,
      "grad_norm": 0.6654230356216431,
      "learning_rate": 0.00019278636476209804,
      "loss": 1.4756,
      "step": 617
    },
    {
      "epoch": 1.4157367668097283,
      "grad_norm": 0.7277924418449402,
      "learning_rate": 0.00019276286818701544,
      "loss": 1.3866,
      "step": 618
    },
    {
      "epoch": 1.4180257510729612,
      "grad_norm": 0.7223805785179138,
      "learning_rate": 0.0001927393348436957,
      "loss": 1.3548,
      "step": 619
    },
    {
      "epoch": 1.4203147353361945,
      "grad_norm": 0.7625775337219238,
      "learning_rate": 0.00019271576474146667,
      "loss": 1.4261,
      "step": 620
    },
    {
      "epoch": 1.4226037195994277,
      "grad_norm": 0.8419500589370728,
      "learning_rate": 0.00019269215788967077,
      "loss": 1.4048,
      "step": 621
    },
    {
      "epoch": 1.4248927038626609,
      "grad_norm": 0.7392059564590454,
      "learning_rate": 0.00019266851429766506,
      "loss": 1.2556,
      "step": 622
    },
    {
      "epoch": 1.427181688125894,
      "grad_norm": 0.9068576693534851,
      "learning_rate": 0.00019264483397482104,
      "loss": 1.3219,
      "step": 623
    },
    {
      "epoch": 1.4294706723891273,
      "grad_norm": 0.8360294699668884,
      "learning_rate": 0.00019262111693052486,
      "loss": 1.2682,
      "step": 624
    },
    {
      "epoch": 1.4317596566523605,
      "grad_norm": 0.9712021350860596,
      "learning_rate": 0.00019259736317417723,
      "loss": 1.3746,
      "step": 625
    },
    {
      "epoch": 1.4340486409155937,
      "grad_norm": 0.8988853693008423,
      "learning_rate": 0.0001925735727151933,
      "loss": 0.9284,
      "step": 626
    },
    {
      "epoch": 1.436337625178827,
      "grad_norm": 1.0043084621429443,
      "learning_rate": 0.00019254974556300294,
      "loss": 1.0586,
      "step": 627
    },
    {
      "epoch": 1.4386266094420601,
      "grad_norm": 0.9843996167182922,
      "learning_rate": 0.0001925258817270504,
      "loss": 1.2283,
      "step": 628
    },
    {
      "epoch": 1.4409155937052933,
      "grad_norm": 1.0277281999588013,
      "learning_rate": 0.00019250198121679456,
      "loss": 0.9889,
      "step": 629
    },
    {
      "epoch": 1.4432045779685265,
      "grad_norm": 1.1969608068466187,
      "learning_rate": 0.0001924780440417089,
      "loss": 0.9705,
      "step": 630
    },
    {
      "epoch": 1.4454935622317597,
      "grad_norm": 1.2368069887161255,
      "learning_rate": 0.00019245407021128118,
      "loss": 1.2699,
      "step": 631
    },
    {
      "epoch": 1.447782546494993,
      "grad_norm": 1.2611585855484009,
      "learning_rate": 0.00019243005973501403,
      "loss": 1.0745,
      "step": 632
    },
    {
      "epoch": 1.4500715307582261,
      "grad_norm": 1.5850770473480225,
      "learning_rate": 0.00019240601262242435,
      "loss": 1.2468,
      "step": 633
    },
    {
      "epoch": 1.4523605150214594,
      "grad_norm": 1.6457610130310059,
      "learning_rate": 0.00019238192888304366,
      "loss": 1.1383,
      "step": 634
    },
    {
      "epoch": 1.4546494992846923,
      "grad_norm": 2.165480375289917,
      "learning_rate": 0.00019235780852641798,
      "loss": 1.117,
      "step": 635
    },
    {
      "epoch": 1.4569384835479255,
      "grad_norm": 3.2665672302246094,
      "learning_rate": 0.0001923336515621079,
      "loss": 1.4645,
      "step": 636
    },
    {
      "epoch": 1.4592274678111588,
      "grad_norm": 1.9498387575149536,
      "learning_rate": 0.00019230945799968835,
      "loss": 1.8131,
      "step": 637
    },
    {
      "epoch": 1.461516452074392,
      "grad_norm": 0.5578173398971558,
      "learning_rate": 0.00019228522784874903,
      "loss": 1.9865,
      "step": 638
    },
    {
      "epoch": 1.4638054363376252,
      "grad_norm": 0.49856436252593994,
      "learning_rate": 0.00019226096111889387,
      "loss": 1.8909,
      "step": 639
    },
    {
      "epoch": 1.4660944206008584,
      "grad_norm": 0.5002744197845459,
      "learning_rate": 0.00019223665781974153,
      "loss": 1.9093,
      "step": 640
    },
    {
      "epoch": 1.4683834048640916,
      "grad_norm": 0.519328236579895,
      "learning_rate": 0.000192212317960925,
      "loss": 1.8239,
      "step": 641
    },
    {
      "epoch": 1.4706723891273248,
      "grad_norm": 0.49472370743751526,
      "learning_rate": 0.00019218794155209182,
      "loss": 1.8014,
      "step": 642
    },
    {
      "epoch": 1.472961373390558,
      "grad_norm": 0.48085641860961914,
      "learning_rate": 0.00019216352860290407,
      "loss": 1.7982,
      "step": 643
    },
    {
      "epoch": 1.4752503576537912,
      "grad_norm": 0.46093374490737915,
      "learning_rate": 0.00019213907912303822,
      "loss": 1.7495,
      "step": 644
    },
    {
      "epoch": 1.4775393419170242,
      "grad_norm": 0.4830629229545593,
      "learning_rate": 0.0001921145931221853,
      "loss": 1.8659,
      "step": 645
    },
    {
      "epoch": 1.4798283261802574,
      "grad_norm": 0.48863735795021057,
      "learning_rate": 0.0001920900706100507,
      "loss": 1.5145,
      "step": 646
    },
    {
      "epoch": 1.4821173104434906,
      "grad_norm": 0.4871852397918701,
      "learning_rate": 0.00019206551159635447,
      "loss": 1.6192,
      "step": 647
    },
    {
      "epoch": 1.4844062947067238,
      "grad_norm": 0.5043070316314697,
      "learning_rate": 0.00019204091609083092,
      "loss": 1.6239,
      "step": 648
    },
    {
      "epoch": 1.486695278969957,
      "grad_norm": 0.5216080546379089,
      "learning_rate": 0.00019201628410322897,
      "loss": 1.7139,
      "step": 649
    },
    {
      "epoch": 1.4889842632331902,
      "grad_norm": 0.4889988899230957,
      "learning_rate": 0.00019199161564331196,
      "loss": 1.6822,
      "step": 650
    },
    {
      "epoch": 1.4912732474964234,
      "grad_norm": 0.4753411114215851,
      "learning_rate": 0.00019196691072085765,
      "loss": 1.5634,
      "step": 651
    },
    {
      "epoch": 1.4935622317596566,
      "grad_norm": 0.4950065016746521,
      "learning_rate": 0.0001919421693456583,
      "loss": 1.6218,
      "step": 652
    },
    {
      "epoch": 1.4958512160228898,
      "grad_norm": 0.5170650482177734,
      "learning_rate": 0.0001919173915275206,
      "loss": 1.515,
      "step": 653
    },
    {
      "epoch": 1.498140200286123,
      "grad_norm": 0.572637140750885,
      "learning_rate": 0.00019189257727626567,
      "loss": 1.5218,
      "step": 654
    },
    {
      "epoch": 1.5004291845493563,
      "grad_norm": 0.5462222695350647,
      "learning_rate": 0.00019186772660172917,
      "loss": 1.7429,
      "step": 655
    },
    {
      "epoch": 1.5027181688125895,
      "grad_norm": 0.5567011833190918,
      "learning_rate": 0.00019184283951376096,
      "loss": 1.5633,
      "step": 656
    },
    {
      "epoch": 1.5050071530758227,
      "grad_norm": 0.535118043422699,
      "learning_rate": 0.0001918179160222256,
      "loss": 1.445,
      "step": 657
    },
    {
      "epoch": 1.5072961373390559,
      "grad_norm": 0.6310518383979797,
      "learning_rate": 0.00019179295613700194,
      "loss": 1.6857,
      "step": 658
    },
    {
      "epoch": 1.509585121602289,
      "grad_norm": 0.5791051387786865,
      "learning_rate": 0.00019176795986798327,
      "loss": 1.4243,
      "step": 659
    },
    {
      "epoch": 1.5118741058655223,
      "grad_norm": 0.6098392605781555,
      "learning_rate": 0.00019174292722507735,
      "loss": 1.6685,
      "step": 660
    },
    {
      "epoch": 1.5141630901287555,
      "grad_norm": 0.6138945817947388,
      "learning_rate": 0.00019171785821820624,
      "loss": 1.4922,
      "step": 661
    },
    {
      "epoch": 1.5164520743919887,
      "grad_norm": 0.5886393189430237,
      "learning_rate": 0.00019169275285730656,
      "loss": 1.3013,
      "step": 662
    },
    {
      "epoch": 1.518741058655222,
      "grad_norm": 0.6358413696289062,
      "learning_rate": 0.00019166761115232928,
      "loss": 1.2984,
      "step": 663
    },
    {
      "epoch": 1.521030042918455,
      "grad_norm": 0.6745731234550476,
      "learning_rate": 0.00019164243311323972,
      "loss": 1.4055,
      "step": 664
    },
    {
      "epoch": 1.523319027181688,
      "grad_norm": 0.7150712609291077,
      "learning_rate": 0.00019161721875001768,
      "loss": 1.5421,
      "step": 665
    },
    {
      "epoch": 1.5256080114449213,
      "grad_norm": 0.7404883503913879,
      "learning_rate": 0.00019159196807265732,
      "loss": 1.4154,
      "step": 666
    },
    {
      "epoch": 1.5278969957081545,
      "grad_norm": 0.7011783719062805,
      "learning_rate": 0.0001915666810911672,
      "loss": 1.3576,
      "step": 667
    },
    {
      "epoch": 1.5301859799713877,
      "grad_norm": 0.743200957775116,
      "learning_rate": 0.0001915413578155703,
      "loss": 1.2993,
      "step": 668
    },
    {
      "epoch": 1.532474964234621,
      "grad_norm": 0.7773009538650513,
      "learning_rate": 0.00019151599825590392,
      "loss": 1.2259,
      "step": 669
    },
    {
      "epoch": 1.534763948497854,
      "grad_norm": 0.7853114604949951,
      "learning_rate": 0.0001914906024222198,
      "loss": 1.4076,
      "step": 670
    },
    {
      "epoch": 1.5370529327610871,
      "grad_norm": 0.8292219638824463,
      "learning_rate": 0.000191465170324584,
      "loss": 1.3584,
      "step": 671
    },
    {
      "epoch": 1.5393419170243203,
      "grad_norm": 0.8770540952682495,
      "learning_rate": 0.000191439701973077,
      "loss": 1.3482,
      "step": 672
    },
    {
      "epoch": 1.5416309012875535,
      "grad_norm": 0.9225857853889465,
      "learning_rate": 0.00019141419737779374,
      "loss": 1.3806,
      "step": 673
    },
    {
      "epoch": 1.5439198855507867,
      "grad_norm": 0.8930622339248657,
      "learning_rate": 0.00019138865654884328,
      "loss": 1.1772,
      "step": 674
    },
    {
      "epoch": 1.54620886981402,
      "grad_norm": 0.8268585801124573,
      "learning_rate": 0.00019136307949634929,
      "loss": 0.9256,
      "step": 675
    },
    {
      "epoch": 1.5484978540772532,
      "grad_norm": 0.9346323609352112,
      "learning_rate": 0.00019133746623044965,
      "loss": 1.3986,
      "step": 676
    },
    {
      "epoch": 1.5507868383404864,
      "grad_norm": 0.9572579860687256,
      "learning_rate": 0.00019131181676129667,
      "loss": 1.0866,
      "step": 677
    },
    {
      "epoch": 1.5530758226037196,
      "grad_norm": 1.038373589515686,
      "learning_rate": 0.00019128613109905694,
      "loss": 1.1096,
      "step": 678
    },
    {
      "epoch": 1.5553648068669528,
      "grad_norm": 1.0466547012329102,
      "learning_rate": 0.00019126040925391148,
      "loss": 1.2318,
      "step": 679
    },
    {
      "epoch": 1.557653791130186,
      "grad_norm": 1.1050053834915161,
      "learning_rate": 0.00019123465123605557,
      "loss": 1.057,
      "step": 680
    },
    {
      "epoch": 1.5599427753934192,
      "grad_norm": 1.4505033493041992,
      "learning_rate": 0.00019120885705569887,
      "loss": 1.3467,
      "step": 681
    },
    {
      "epoch": 1.5622317596566524,
      "grad_norm": 1.4782710075378418,
      "learning_rate": 0.00019118302672306543,
      "loss": 1.3132,
      "step": 682
    },
    {
      "epoch": 1.5645207439198856,
      "grad_norm": 1.589423656463623,
      "learning_rate": 0.00019115716024839348,
      "loss": 1.2524,
      "step": 683
    },
    {
      "epoch": 1.5668097281831188,
      "grad_norm": 1.5407938957214355,
      "learning_rate": 0.00019113125764193568,
      "loss": 0.9281,
      "step": 684
    },
    {
      "epoch": 1.569098712446352,
      "grad_norm": 2.313833236694336,
      "learning_rate": 0.00019110531891395905,
      "loss": 1.2519,
      "step": 685
    },
    {
      "epoch": 1.5713876967095852,
      "grad_norm": 2.507740020751953,
      "learning_rate": 0.00019107934407474486,
      "loss": 1.1084,
      "step": 686
    },
    {
      "epoch": 1.5736766809728184,
      "grad_norm": 1.6293509006500244,
      "learning_rate": 0.00019105333313458868,
      "loss": 1.5332,
      "step": 687
    },
    {
      "epoch": 1.5759656652360516,
      "grad_norm": 0.4493796229362488,
      "learning_rate": 0.0001910272861038004,
      "loss": 1.8966,
      "step": 688
    },
    {
      "epoch": 1.5782546494992848,
      "grad_norm": 0.49884214997291565,
      "learning_rate": 0.00019100120299270427,
      "loss": 1.9927,
      "step": 689
    },
    {
      "epoch": 1.5805436337625178,
      "grad_norm": 0.4980679154396057,
      "learning_rate": 0.00019097508381163877,
      "loss": 1.783,
      "step": 690
    },
    {
      "epoch": 1.582832618025751,
      "grad_norm": 0.4812343716621399,
      "learning_rate": 0.00019094892857095675,
      "loss": 1.5191,
      "step": 691
    },
    {
      "epoch": 1.5851216022889842,
      "grad_norm": 0.5427776575088501,
      "learning_rate": 0.0001909227372810253,
      "loss": 1.9404,
      "step": 692
    },
    {
      "epoch": 1.5874105865522175,
      "grad_norm": 0.5218622088432312,
      "learning_rate": 0.00019089650995222576,
      "loss": 1.6748,
      "step": 693
    },
    {
      "epoch": 1.5896995708154507,
      "grad_norm": 0.4879237413406372,
      "learning_rate": 0.00019087024659495387,
      "loss": 1.8253,
      "step": 694
    },
    {
      "epoch": 1.5919885550786839,
      "grad_norm": 0.5004391670227051,
      "learning_rate": 0.00019084394721961956,
      "loss": 1.8175,
      "step": 695
    },
    {
      "epoch": 1.594277539341917,
      "grad_norm": 0.48198503255844116,
      "learning_rate": 0.00019081761183664705,
      "loss": 1.7907,
      "step": 696
    },
    {
      "epoch": 1.59656652360515,
      "grad_norm": 0.4696299433708191,
      "learning_rate": 0.00019079124045647485,
      "loss": 1.6107,
      "step": 697
    },
    {
      "epoch": 1.5988555078683833,
      "grad_norm": 0.4958980679512024,
      "learning_rate": 0.00019076483308955573,
      "loss": 1.7122,
      "step": 698
    },
    {
      "epoch": 1.6011444921316165,
      "grad_norm": 0.4773041903972626,
      "learning_rate": 0.00019073838974635675,
      "loss": 1.6884,
      "step": 699
    },
    {
      "epoch": 1.6034334763948497,
      "grad_norm": 0.4853622317314148,
      "learning_rate": 0.0001907119104373592,
      "loss": 1.6907,
      "step": 700
    },
    {
      "epoch": 1.6057224606580829,
      "grad_norm": 0.4870537221431732,
      "learning_rate": 0.0001906853951730586,
      "loss": 1.5317,
      "step": 701
    },
    {
      "epoch": 1.608011444921316,
      "grad_norm": 0.5046525001525879,
      "learning_rate": 0.00019065884396396476,
      "loss": 1.5933,
      "step": 702
    },
    {
      "epoch": 1.6103004291845493,
      "grad_norm": 0.47611844539642334,
      "learning_rate": 0.00019063225682060177,
      "loss": 1.6262,
      "step": 703
    },
    {
      "epoch": 1.6125894134477825,
      "grad_norm": 0.4944547414779663,
      "learning_rate": 0.0001906056337535079,
      "loss": 1.3557,
      "step": 704
    },
    {
      "epoch": 1.6148783977110157,
      "grad_norm": 0.5307270884513855,
      "learning_rate": 0.00019057897477323568,
      "loss": 1.6978,
      "step": 705
    },
    {
      "epoch": 1.617167381974249,
      "grad_norm": 0.5555788278579712,
      "learning_rate": 0.00019055227989035192,
      "loss": 1.4591,
      "step": 706
    },
    {
      "epoch": 1.6194563662374821,
      "grad_norm": 0.543749213218689,
      "learning_rate": 0.00019052554911543754,
      "loss": 1.5933,
      "step": 707
    },
    {
      "epoch": 1.6217453505007153,
      "grad_norm": 0.5415577292442322,
      "learning_rate": 0.0001904987824590878,
      "loss": 1.4175,
      "step": 708
    },
    {
      "epoch": 1.6240343347639485,
      "grad_norm": 0.5744507312774658,
      "learning_rate": 0.00019047197993191223,
      "loss": 1.4308,
      "step": 709
    },
    {
      "epoch": 1.6263233190271817,
      "grad_norm": 0.5747923254966736,
      "learning_rate": 0.00019044514154453434,
      "loss": 1.4975,
      "step": 710
    },
    {
      "epoch": 1.628612303290415,
      "grad_norm": 0.614585280418396,
      "learning_rate": 0.00019041826730759208,
      "loss": 1.428,
      "step": 711
    },
    {
      "epoch": 1.6309012875536482,
      "grad_norm": 0.6449160575866699,
      "learning_rate": 0.00019039135723173759,
      "loss": 1.4738,
      "step": 712
    },
    {
      "epoch": 1.6331902718168814,
      "grad_norm": 0.6278111934661865,
      "learning_rate": 0.00019036441132763712,
      "loss": 1.2803,
      "step": 713
    },
    {
      "epoch": 1.6354792560801146,
      "grad_norm": 0.6389395594596863,
      "learning_rate": 0.00019033742960597114,
      "loss": 1.2128,
      "step": 714
    },
    {
      "epoch": 1.6377682403433478,
      "grad_norm": 0.6629028916358948,
      "learning_rate": 0.0001903104120774344,
      "loss": 1.312,
      "step": 715
    },
    {
      "epoch": 1.640057224606581,
      "grad_norm": 0.7149224281311035,
      "learning_rate": 0.00019028335875273574,
      "loss": 1.5486,
      "step": 716
    },
    {
      "epoch": 1.642346208869814,
      "grad_norm": 0.7043018937110901,
      "learning_rate": 0.00019025626964259824,
      "loss": 1.3197,
      "step": 717
    },
    {
      "epoch": 1.6446351931330472,
      "grad_norm": 0.6970218420028687,
      "learning_rate": 0.00019022914475775918,
      "loss": 1.2391,
      "step": 718
    },
    {
      "epoch": 1.6469241773962804,
      "grad_norm": 0.8296447992324829,
      "learning_rate": 0.00019020198410896996,
      "loss": 1.5892,
      "step": 719
    },
    {
      "epoch": 1.6492131616595136,
      "grad_norm": 0.7861882448196411,
      "learning_rate": 0.00019017478770699624,
      "loss": 1.1756,
      "step": 720
    },
    {
      "epoch": 1.6515021459227468,
      "grad_norm": 0.8338953852653503,
      "learning_rate": 0.0001901475555626178,
      "loss": 1.3851,
      "step": 721
    },
    {
      "epoch": 1.65379113018598,
      "grad_norm": 0.8367292881011963,
      "learning_rate": 0.00019012028768662854,
      "loss": 1.2315,
      "step": 722
    },
    {
      "epoch": 1.6560801144492132,
      "grad_norm": 0.9444144368171692,
      "learning_rate": 0.0001900929840898366,
      "loss": 1.3334,
      "step": 723
    },
    {
      "epoch": 1.6583690987124462,
      "grad_norm": 0.9217041730880737,
      "learning_rate": 0.0001900656447830643,
      "loss": 1.2192,
      "step": 724
    },
    {
      "epoch": 1.6606580829756794,
      "grad_norm": 0.9736611843109131,
      "learning_rate": 0.00019003826977714798,
      "loss": 1.1086,
      "step": 725
    },
    {
      "epoch": 1.6629470672389126,
      "grad_norm": 1.0112082958221436,
      "learning_rate": 0.0001900108590829383,
      "loss": 1.2009,
      "step": 726
    },
    {
      "epoch": 1.6652360515021458,
      "grad_norm": 1.0199767351150513,
      "learning_rate": 0.00018998341271129996,
      "loss": 1.2564,
      "step": 727
    },
    {
      "epoch": 1.667525035765379,
      "grad_norm": 0.9764334559440613,
      "learning_rate": 0.0001899559306731118,
      "loss": 0.9171,
      "step": 728
    },
    {
      "epoch": 1.6698140200286122,
      "grad_norm": 1.163103461265564,
      "learning_rate": 0.00018992841297926685,
      "loss": 1.2957,
      "step": 729
    },
    {
      "epoch": 1.6721030042918454,
      "grad_norm": 1.1633704900741577,
      "learning_rate": 0.0001899008596406722,
      "loss": 1.0627,
      "step": 730
    },
    {
      "epoch": 1.6743919885550786,
      "grad_norm": 1.284395456314087,
      "learning_rate": 0.00018987327066824917,
      "loss": 1.1982,
      "step": 731
    },
    {
      "epoch": 1.6766809728183119,
      "grad_norm": 1.268510103225708,
      "learning_rate": 0.00018984564607293314,
      "loss": 1.0235,
      "step": 732
    },
    {
      "epoch": 1.678969957081545,
      "grad_norm": 1.399523138999939,
      "learning_rate": 0.00018981798586567358,
      "loss": 0.94,
      "step": 733
    },
    {
      "epoch": 1.6812589413447783,
      "grad_norm": 1.9600815773010254,
      "learning_rate": 0.0001897902900574341,
      "loss": 0.9426,
      "step": 734
    },
    {
      "epoch": 1.6835479256080115,
      "grad_norm": 2.2276952266693115,
      "learning_rate": 0.0001897625586591925,
      "loss": 1.1555,
      "step": 735
    },
    {
      "epoch": 1.6858369098712447,
      "grad_norm": 3.050342082977295,
      "learning_rate": 0.00018973479168194062,
      "loss": 1.2365,
      "step": 736
    },
    {
      "epoch": 1.688125894134478,
      "grad_norm": 1.8580230474472046,
      "learning_rate": 0.00018970698913668433,
      "loss": 1.4868,
      "step": 737
    },
    {
      "epoch": 1.690414878397711,
      "grad_norm": 0.517766535282135,
      "learning_rate": 0.00018967915103444368,
      "loss": 1.9625,
      "step": 738
    },
    {
      "epoch": 1.6927038626609443,
      "grad_norm": 0.5281224250793457,
      "learning_rate": 0.00018965127738625287,
      "loss": 1.9024,
      "step": 739
    },
    {
      "epoch": 1.6949928469241775,
      "grad_norm": 0.5033324360847473,
      "learning_rate": 0.00018962336820316006,
      "loss": 1.7628,
      "step": 740
    },
    {
      "epoch": 1.6972818311874107,
      "grad_norm": 0.4653541147708893,
      "learning_rate": 0.00018959542349622762,
      "loss": 1.7784,
      "step": 741
    },
    {
      "epoch": 1.699570815450644,
      "grad_norm": 0.5263776779174805,
      "learning_rate": 0.00018956744327653188,
      "loss": 1.8423,
      "step": 742
    },
    {
      "epoch": 1.7018597997138771,
      "grad_norm": 0.484857976436615,
      "learning_rate": 0.00018953942755516333,
      "loss": 1.66,
      "step": 743
    },
    {
      "epoch": 1.7041487839771101,
      "grad_norm": 0.46540528535842896,
      "learning_rate": 0.00018951137634322653,
      "loss": 1.7435,
      "step": 744
    },
    {
      "epoch": 1.7064377682403433,
      "grad_norm": 0.4492217004299164,
      "learning_rate": 0.00018948328965184003,
      "loss": 1.7099,
      "step": 745
    },
    {
      "epoch": 1.7087267525035765,
      "grad_norm": 0.4283190965652466,
      "learning_rate": 0.00018945516749213656,
      "loss": 1.6823,
      "step": 746
    },
    {
      "epoch": 1.7110157367668097,
      "grad_norm": 0.4640548527240753,
      "learning_rate": 0.00018942700987526276,
      "loss": 1.6032,
      "step": 747
    },
    {
      "epoch": 1.713304721030043,
      "grad_norm": 0.4854135513305664,
      "learning_rate": 0.00018939881681237953,
      "loss": 1.662,
      "step": 748
    },
    {
      "epoch": 1.7155937052932761,
      "grad_norm": 0.49598628282546997,
      "learning_rate": 0.00018937058831466157,
      "loss": 1.7955,
      "step": 749
    },
    {
      "epoch": 1.7178826895565091,
      "grad_norm": 0.4726034998893738,
      "learning_rate": 0.00018934232439329787,
      "loss": 1.4149,
      "step": 750
    },
    {
      "epoch": 1.7201716738197423,
      "grad_norm": 0.5018281936645508,
      "learning_rate": 0.00018931402505949126,
      "loss": 1.6037,
      "step": 751
    },
    {
      "epoch": 1.7224606580829755,
      "grad_norm": 0.5033923983573914,
      "learning_rate": 0.00018928569032445873,
      "loss": 1.7919,
      "step": 752
    },
    {
      "epoch": 1.7247496423462088,
      "grad_norm": 0.5036657452583313,
      "learning_rate": 0.00018925732019943125,
      "loss": 1.6631,
      "step": 753
    },
    {
      "epoch": 1.727038626609442,
      "grad_norm": 0.48502838611602783,
      "learning_rate": 0.00018922891469565385,
      "loss": 1.6094,
      "step": 754
    },
    {
      "epoch": 1.7293276108726752,
      "grad_norm": 0.4992237985134125,
      "learning_rate": 0.00018920047382438553,
      "loss": 1.6122,
      "step": 755
    },
    {
      "epoch": 1.7316165951359084,
      "grad_norm": 0.5455563068389893,
      "learning_rate": 0.0001891719975968994,
      "loss": 1.5948,
      "step": 756
    },
    {
      "epoch": 1.7339055793991416,
      "grad_norm": 0.5290236473083496,
      "learning_rate": 0.00018914348602448248,
      "loss": 1.5871,
      "step": 757
    },
    {
      "epoch": 1.7361945636623748,
      "grad_norm": 0.5112497210502625,
      "learning_rate": 0.00018911493911843584,
      "loss": 1.479,
      "step": 758
    },
    {
      "epoch": 1.738483547925608,
      "grad_norm": 0.564210832118988,
      "learning_rate": 0.00018908635689007458,
      "loss": 1.3729,
      "step": 759
    },
    {
      "epoch": 1.7407725321888412,
      "grad_norm": 0.5638823509216309,
      "learning_rate": 0.0001890577393507278,
      "loss": 1.5368,
      "step": 760
    },
    {
      "epoch": 1.7430615164520744,
      "grad_norm": 0.6375520825386047,
      "learning_rate": 0.00018902908651173856,
      "loss": 1.6355,
      "step": 761
    },
    {
      "epoch": 1.7453505007153076,
      "grad_norm": 0.6201082468032837,
      "learning_rate": 0.00018900039838446394,
      "loss": 1.3423,
      "step": 762
    },
    {
      "epoch": 1.7476394849785408,
      "grad_norm": 0.6683928370475769,
      "learning_rate": 0.00018897167498027495,
      "loss": 1.678,
      "step": 763
    },
    {
      "epoch": 1.749928469241774,
      "grad_norm": 0.6207825541496277,
      "learning_rate": 0.00018894291631055667,
      "loss": 1.3583,
      "step": 764
    },
    {
      "epoch": 1.7522174535050072,
      "grad_norm": 0.6242700815200806,
      "learning_rate": 0.00018891412238670815,
      "loss": 1.3729,
      "step": 765
    },
    {
      "epoch": 1.7545064377682404,
      "grad_norm": 0.7048245668411255,
      "learning_rate": 0.00018888529322014233,
      "loss": 1.4936,
      "step": 766
    },
    {
      "epoch": 1.7567954220314737,
      "grad_norm": 0.6496549248695374,
      "learning_rate": 0.0001888564288222862,
      "loss": 1.3397,
      "step": 767
    },
    {
      "epoch": 1.7590844062947069,
      "grad_norm": 0.6626686453819275,
      "learning_rate": 0.00018882752920458066,
      "loss": 1.3271,
      "step": 768
    },
    {
      "epoch": 1.76137339055794,
      "grad_norm": 0.8030514121055603,
      "learning_rate": 0.0001887985943784806,
      "loss": 1.445,
      "step": 769
    },
    {
      "epoch": 1.763662374821173,
      "grad_norm": 0.7380093932151794,
      "learning_rate": 0.00018876962435545488,
      "loss": 1.296,
      "step": 770
    },
    {
      "epoch": 1.7659513590844063,
      "grad_norm": 0.8091381192207336,
      "learning_rate": 0.00018874061914698627,
      "loss": 1.2228,
      "step": 771
    },
    {
      "epoch": 1.7682403433476395,
      "grad_norm": 0.7839142084121704,
      "learning_rate": 0.0001887115787645715,
      "loss": 1.1097,
      "step": 772
    },
    {
      "epoch": 1.7705293276108727,
      "grad_norm": 0.7487571835517883,
      "learning_rate": 0.00018868250321972128,
      "loss": 1.1244,
      "step": 773
    },
    {
      "epoch": 1.7728183118741059,
      "grad_norm": 0.917154848575592,
      "learning_rate": 0.00018865339252396022,
      "loss": 1.4103,
      "step": 774
    },
    {
      "epoch": 1.775107296137339,
      "grad_norm": 0.9280546307563782,
      "learning_rate": 0.00018862424668882687,
      "loss": 1.1329,
      "step": 775
    },
    {
      "epoch": 1.7773962804005723,
      "grad_norm": 0.8897112011909485,
      "learning_rate": 0.00018859506572587367,
      "loss": 0.9447,
      "step": 776
    },
    {
      "epoch": 1.7796852646638053,
      "grad_norm": 0.9497140049934387,
      "learning_rate": 0.0001885658496466671,
      "loss": 1.1904,
      "step": 777
    },
    {
      "epoch": 1.7819742489270385,
      "grad_norm": 1.0200343132019043,
      "learning_rate": 0.00018853659846278738,
      "loss": 1.2004,
      "step": 778
    },
    {
      "epoch": 1.7842632331902717,
      "grad_norm": 0.9385619759559631,
      "learning_rate": 0.0001885073121858288,
      "loss": 0.9457,
      "step": 779
    },
    {
      "epoch": 1.786552217453505,
      "grad_norm": 1.110852837562561,
      "learning_rate": 0.00018847799082739956,
      "loss": 0.9548,
      "step": 780
    },
    {
      "epoch": 1.788841201716738,
      "grad_norm": 1.2065496444702148,
      "learning_rate": 0.0001884486343991216,
      "loss": 1.1064,
      "step": 781
    },
    {
      "epoch": 1.7911301859799713,
      "grad_norm": 1.461957335472107,
      "learning_rate": 0.00018841924291263098,
      "loss": 1.1241,
      "step": 782
    },
    {
      "epoch": 1.7934191702432045,
      "grad_norm": 1.6875280141830444,
      "learning_rate": 0.00018838981637957743,
      "loss": 1.4554,
      "step": 783
    },
    {
      "epoch": 1.7957081545064377,
      "grad_norm": 1.9972717761993408,
      "learning_rate": 0.00018836035481162476,
      "loss": 1.2699,
      "step": 784
    },
    {
      "epoch": 1.797997138769671,
      "grad_norm": 1.956983208656311,
      "learning_rate": 0.0001883308582204506,
      "loss": 1.01,
      "step": 785
    },
    {
      "epoch": 1.8002861230329041,
      "grad_norm": 2.715808629989624,
      "learning_rate": 0.00018830132661774642,
      "loss": 1.3565,
      "step": 786
    },
    {
      "epoch": 1.8025751072961373,
      "grad_norm": 2.0629868507385254,
      "learning_rate": 0.0001882717600152176,
      "loss": 1.7408,
      "step": 787
    },
    {
      "epoch": 1.8048640915593706,
      "grad_norm": 0.5920199155807495,
      "learning_rate": 0.00018824215842458343,
      "loss": 1.7481,
      "step": 788
    },
    {
      "epoch": 1.8071530758226038,
      "grad_norm": 0.5530115365982056,
      "learning_rate": 0.00018821252185757703,
      "loss": 1.9706,
      "step": 789
    },
    {
      "epoch": 1.809442060085837,
      "grad_norm": 0.5147140026092529,
      "learning_rate": 0.00018818285032594538,
      "loss": 1.712,
      "step": 790
    },
    {
      "epoch": 1.8117310443490702,
      "grad_norm": 0.563723087310791,
      "learning_rate": 0.00018815314384144934,
      "loss": 1.7015,
      "step": 791
    },
    {
      "epoch": 1.8140200286123034,
      "grad_norm": 0.5451656579971313,
      "learning_rate": 0.00018812340241586357,
      "loss": 1.8385,
      "step": 792
    },
    {
      "epoch": 1.8163090128755366,
      "grad_norm": 0.4850448668003082,
      "learning_rate": 0.00018809362606097668,
      "loss": 1.6491,
      "step": 793
    },
    {
      "epoch": 1.8185979971387698,
      "grad_norm": 0.4890015721321106,
      "learning_rate": 0.000188063814788591,
      "loss": 1.779,
      "step": 794
    },
    {
      "epoch": 1.820886981402003,
      "grad_norm": 0.4644920825958252,
      "learning_rate": 0.00018803396861052284,
      "loss": 1.7255,
      "step": 795
    },
    {
      "epoch": 1.8231759656652362,
      "grad_norm": 0.4727228283882141,
      "learning_rate": 0.00018800408753860218,
      "loss": 1.6042,
      "step": 796
    },
    {
      "epoch": 1.8254649499284692,
      "grad_norm": 0.4832257628440857,
      "learning_rate": 0.000187974171584673,
      "loss": 1.7279,
      "step": 797
    },
    {
      "epoch": 1.8277539341917024,
      "grad_norm": 0.4614594876766205,
      "learning_rate": 0.00018794422076059302,
      "loss": 1.618,
      "step": 798
    },
    {
      "epoch": 1.8300429184549356,
      "grad_norm": 0.45358842611312866,
      "learning_rate": 0.00018791423507823372,
      "loss": 1.5979,
      "step": 799
    },
    {
      "epoch": 1.8323319027181688,
      "grad_norm": 0.5268234014511108,
      "learning_rate": 0.00018788421454948056,
      "loss": 1.7868,
      "step": 800
    },
    {
      "epoch": 1.834620886981402,
      "grad_norm": 0.4693477749824524,
      "learning_rate": 0.0001878541591862327,
      "loss": 1.5261,
      "step": 801
    },
    {
      "epoch": 1.8369098712446352,
      "grad_norm": 0.8950780630111694,
      "learning_rate": 0.00018782406900040308,
      "loss": 1.6545,
      "step": 802
    },
    {
      "epoch": 1.8391988555078684,
      "grad_norm": 0.488389253616333,
      "learning_rate": 0.0001877939440039185,
      "loss": 1.4573,
      "step": 803
    },
    {
      "epoch": 1.8414878397711014,
      "grad_norm": 0.5442367792129517,
      "learning_rate": 0.00018776378420871955,
      "loss": 1.646,
      "step": 804
    },
    {
      "epoch": 1.8437768240343346,
      "grad_norm": 0.5337203145027161,
      "learning_rate": 0.00018773358962676065,
      "loss": 1.6527,
      "step": 805
    },
    {
      "epoch": 1.8460658082975678,
      "grad_norm": 0.5353984832763672,
      "learning_rate": 0.0001877033602700099,
      "loss": 1.5666,
      "step": 806
    },
    {
      "epoch": 1.848354792560801,
      "grad_norm": 0.5341973304748535,
      "learning_rate": 0.00018767309615044927,
      "loss": 1.3956,
      "step": 807
    },
    {
      "epoch": 1.8506437768240342,
      "grad_norm": 0.6028610467910767,
      "learning_rate": 0.00018764279728007455,
      "loss": 1.5369,
      "step": 808
    },
    {
      "epoch": 1.8529327610872675,
      "grad_norm": 0.603038489818573,
      "learning_rate": 0.00018761246367089517,
      "loss": 1.5677,
      "step": 809
    },
    {
      "epoch": 1.8552217453505007,
      "grad_norm": 0.6117138862609863,
      "learning_rate": 0.00018758209533493444,
      "loss": 1.4475,
      "step": 810
    },
    {
      "epoch": 1.8575107296137339,
      "grad_norm": 0.6099836230278015,
      "learning_rate": 0.0001875516922842294,
      "loss": 1.2522,
      "step": 811
    },
    {
      "epoch": 1.859799713876967,
      "grad_norm": 0.6215634942054749,
      "learning_rate": 0.00018752125453083086,
      "loss": 1.3362,
      "step": 812
    },
    {
      "epoch": 1.8620886981402003,
      "grad_norm": 0.6947351098060608,
      "learning_rate": 0.00018749078208680331,
      "loss": 1.3849,
      "step": 813
    },
    {
      "epoch": 1.8643776824034335,
      "grad_norm": 0.634660005569458,
      "learning_rate": 0.0001874602749642251,
      "loss": 1.3116,
      "step": 814
    },
    {
      "epoch": 1.8666666666666667,
      "grad_norm": 0.6666608452796936,
      "learning_rate": 0.0001874297331751883,
      "loss": 1.4252,
      "step": 815
    },
    {
      "epoch": 1.8689556509299,
      "grad_norm": 0.7307305335998535,
      "learning_rate": 0.00018739915673179867,
      "loss": 1.4582,
      "step": 816
    },
    {
      "epoch": 1.871244635193133,
      "grad_norm": 0.7657129764556885,
      "learning_rate": 0.0001873685456461757,
      "loss": 1.4531,
      "step": 817
    },
    {
      "epoch": 1.8735336194563663,
      "grad_norm": 0.8193492293357849,
      "learning_rate": 0.0001873378999304527,
      "loss": 1.4915,
      "step": 818
    },
    {
      "epoch": 1.8758226037195995,
      "grad_norm": 0.8312426209449768,
      "learning_rate": 0.00018730721959677666,
      "loss": 1.6111,
      "step": 819
    },
    {
      "epoch": 1.8781115879828327,
      "grad_norm": 0.7933733463287354,
      "learning_rate": 0.00018727650465730827,
      "loss": 1.2889,
      "step": 820
    },
    {
      "epoch": 1.880400572246066,
      "grad_norm": 0.7646512985229492,
      "learning_rate": 0.0001872457551242219,
      "loss": 1.287,
      "step": 821
    },
    {
      "epoch": 1.8826895565092991,
      "grad_norm": 0.8263788819313049,
      "learning_rate": 0.00018721497100970575,
      "loss": 1.3444,
      "step": 822
    },
    {
      "epoch": 1.8849785407725324,
      "grad_norm": 0.8783066868782043,
      "learning_rate": 0.00018718415232596162,
      "loss": 1.0587,
      "step": 823
    },
    {
      "epoch": 1.8872675250357653,
      "grad_norm": 0.943548321723938,
      "learning_rate": 0.00018715329908520508,
      "loss": 1.3713,
      "step": 824
    },
    {
      "epoch": 1.8895565092989985,
      "grad_norm": 0.8338328003883362,
      "learning_rate": 0.00018712241129966532,
      "loss": 0.9961,
      "step": 825
    },
    {
      "epoch": 1.8918454935622318,
      "grad_norm": 0.9351692199707031,
      "learning_rate": 0.00018709148898158535,
      "loss": 1.0726,
      "step": 826
    },
    {
      "epoch": 1.894134477825465,
      "grad_norm": 1.006306767463684,
      "learning_rate": 0.00018706053214322173,
      "loss": 1.0892,
      "step": 827
    },
    {
      "epoch": 1.8964234620886982,
      "grad_norm": 0.9210097193717957,
      "learning_rate": 0.00018702954079684477,
      "loss": 0.9979,
      "step": 828
    },
    {
      "epoch": 1.8987124463519314,
      "grad_norm": 1.0254311561584473,
      "learning_rate": 0.00018699851495473845,
      "loss": 1.0705,
      "step": 829
    },
    {
      "epoch": 1.9010014306151644,
      "grad_norm": 1.0799310207366943,
      "learning_rate": 0.00018696745462920046,
      "loss": 1.0158,
      "step": 830
    },
    {
      "epoch": 1.9032904148783976,
      "grad_norm": 1.0019152164459229,
      "learning_rate": 0.00018693635983254203,
      "loss": 0.9876,
      "step": 831
    },
    {
      "epoch": 1.9055793991416308,
      "grad_norm": 1.3597310781478882,
      "learning_rate": 0.00018690523057708825,
      "loss": 1.2822,
      "step": 832
    },
    {
      "epoch": 1.907868383404864,
      "grad_norm": 1.388640284538269,
      "learning_rate": 0.0001868740668751777,
      "loss": 1.0117,
      "step": 833
    },
    {
      "epoch": 1.9101573676680972,
      "grad_norm": 1.498296856880188,
      "learning_rate": 0.0001868428687391627,
      "loss": 0.9964,
      "step": 834
    },
    {
      "epoch": 1.9124463519313304,
      "grad_norm": 1.9695405960083008,
      "learning_rate": 0.00018681163618140917,
      "loss": 0.9319,
      "step": 835
    },
    {
      "epoch": 1.9147353361945636,
      "grad_norm": 3.282263994216919,
      "learning_rate": 0.00018678036921429672,
      "loss": 1.6949,
      "step": 836
    },
    {
      "epoch": 1.9170243204577968,
      "grad_norm": 1.874495029449463,
      "learning_rate": 0.00018674906785021858,
      "loss": 1.6669,
      "step": 837
    },
    {
      "epoch": 1.91931330472103,
      "grad_norm": 0.5985230207443237,
      "learning_rate": 0.00018671773210158157,
      "loss": 1.8544,
      "step": 838
    },
    {
      "epoch": 1.9216022889842632,
      "grad_norm": 0.6386774182319641,
      "learning_rate": 0.00018668636198080623,
      "loss": 1.7688,
      "step": 839
    },
    {
      "epoch": 1.9238912732474964,
      "grad_norm": 0.7015856504440308,
      "learning_rate": 0.00018665495750032663,
      "loss": 1.898,
      "step": 840
    },
    {
      "epoch": 1.9261802575107296,
      "grad_norm": 0.6125592589378357,
      "learning_rate": 0.00018662351867259055,
      "loss": 1.6589,
      "step": 841
    },
    {
      "epoch": 1.9284692417739628,
      "grad_norm": 0.6526002883911133,
      "learning_rate": 0.00018659204551005934,
      "loss": 1.7954,
      "step": 842
    },
    {
      "epoch": 1.930758226037196,
      "grad_norm": 0.5852531790733337,
      "learning_rate": 0.00018656053802520788,
      "loss": 1.8123,
      "step": 843
    },
    {
      "epoch": 1.9330472103004293,
      "grad_norm": 0.5367109179496765,
      "learning_rate": 0.00018652899623052482,
      "loss": 1.7088,
      "step": 844
    },
    {
      "epoch": 1.9353361945636625,
      "grad_norm": 0.531198263168335,
      "learning_rate": 0.00018649742013851226,
      "loss": 1.5857,
      "step": 845
    },
    {
      "epoch": 1.9376251788268957,
      "grad_norm": 0.5513063669204712,
      "learning_rate": 0.00018646580976168595,
      "loss": 1.5867,
      "step": 846
    },
    {
      "epoch": 1.9399141630901289,
      "grad_norm": 0.5283839702606201,
      "learning_rate": 0.0001864341651125753,
      "loss": 1.6109,
      "step": 847
    },
    {
      "epoch": 1.942203147353362,
      "grad_norm": 0.5148706436157227,
      "learning_rate": 0.00018640248620372315,
      "loss": 1.662,
      "step": 848
    },
    {
      "epoch": 1.9444921316165953,
      "grad_norm": 0.5697648525238037,
      "learning_rate": 0.0001863707730476861,
      "loss": 1.686,
      "step": 849
    },
    {
      "epoch": 1.9467811158798283,
      "grad_norm": 0.5709057450294495,
      "learning_rate": 0.00018633902565703413,
      "loss": 1.4639,
      "step": 850
    },
    {
      "epoch": 1.9490701001430615,
      "grad_norm": 0.6249383687973022,
      "learning_rate": 0.000186307244044351,
      "loss": 1.5752,
      "step": 851
    },
    {
      "epoch": 1.9513590844062947,
      "grad_norm": 0.6226382255554199,
      "learning_rate": 0.00018627542822223384,
      "loss": 1.5567,
      "step": 852
    },
    {
      "epoch": 1.953648068669528,
      "grad_norm": 0.6388505697250366,
      "learning_rate": 0.00018624357820329346,
      "loss": 1.5492,
      "step": 853
    },
    {
      "epoch": 1.955937052932761,
      "grad_norm": 0.6010354161262512,
      "learning_rate": 0.00018621169400015418,
      "loss": 1.308,
      "step": 854
    },
    {
      "epoch": 1.9582260371959943,
      "grad_norm": 0.6577657461166382,
      "learning_rate": 0.00018617977562545389,
      "loss": 1.5635,
      "step": 855
    },
    {
      "epoch": 1.9605150214592275,
      "grad_norm": 0.7468054890632629,
      "learning_rate": 0.000186147823091844,
      "loss": 1.4909,
      "step": 856
    },
    {
      "epoch": 1.9628040057224605,
      "grad_norm": 0.6937854290008545,
      "learning_rate": 0.0001861158364119895,
      "loss": 1.4194,
      "step": 857
    },
    {
      "epoch": 1.9650929899856937,
      "grad_norm": 0.7472777366638184,
      "learning_rate": 0.00018608381559856888,
      "loss": 1.5602,
      "step": 858
    },
    {
      "epoch": 1.967381974248927,
      "grad_norm": 0.7680912613868713,
      "learning_rate": 0.00018605176066427412,
      "loss": 1.3623,
      "step": 859
    },
    {
      "epoch": 1.9696709585121601,
      "grad_norm": 0.7665932774543762,
      "learning_rate": 0.00018601967162181081,
      "loss": 1.2643,
      "step": 860
    },
    {
      "epoch": 1.9719599427753933,
      "grad_norm": 0.7979941368103027,
      "learning_rate": 0.00018598754848389806,
      "loss": 1.1944,
      "step": 861
    },
    {
      "epoch": 1.9742489270386265,
      "grad_norm": 0.827511191368103,
      "learning_rate": 0.00018595539126326838,
      "loss": 1.2876,
      "step": 862
    },
    {
      "epoch": 1.9765379113018597,
      "grad_norm": 0.812494158744812,
      "learning_rate": 0.00018592319997266793,
      "loss": 1.1597,
      "step": 863
    },
    {
      "epoch": 1.978826895565093,
      "grad_norm": 0.8327008485794067,
      "learning_rate": 0.00018589097462485627,
      "loss": 1.2395,
      "step": 864
    },
    {
      "epoch": 1.9811158798283262,
      "grad_norm": 1.0873643159866333,
      "learning_rate": 0.00018585871523260652,
      "loss": 1.6081,
      "step": 865
    },
    {
      "epoch": 1.9834048640915594,
      "grad_norm": 0.9741319417953491,
      "learning_rate": 0.00018582642180870523,
      "loss": 1.1003,
      "step": 866
    },
    {
      "epoch": 1.9856938483547926,
      "grad_norm": 1.1315044164657593,
      "learning_rate": 0.00018579409436595256,
      "loss": 1.1537,
      "step": 867
    },
    {
      "epoch": 1.9879828326180258,
      "grad_norm": 1.063059687614441,
      "learning_rate": 0.00018576173291716196,
      "loss": 1.0723,
      "step": 868
    },
    {
      "epoch": 1.990271816881259,
      "grad_norm": 1.0340529680252075,
      "learning_rate": 0.0001857293374751606,
      "loss": 0.8886,
      "step": 869
    },
    {
      "epoch": 1.9925608011444922,
      "grad_norm": 1.2680200338363647,
      "learning_rate": 0.00018569690805278892,
      "loss": 1.1097,
      "step": 870
    },
    {
      "epoch": 1.9948497854077254,
      "grad_norm": 1.6370130777359009,
      "learning_rate": 0.00018566444466290092,
      "loss": 1.1095,
      "step": 871
    },
    {
      "epoch": 1.9971387696709586,
      "grad_norm": 1.8714702129364014,
      "learning_rate": 0.00018563194731836404,
      "loss": 1.3104,
      "step": 872
    },
    {
      "epoch": 2.000572246065808,
      "grad_norm": 4.319216728210449,
      "learning_rate": 0.00018559941603205925,
      "loss": 2.3024,
      "step": 873
    },
    {
      "epoch": 2.0028612303290414,
      "grad_norm": 0.29445138573646545,
      "learning_rate": 0.00018556685081688086,
      "loss": 1.9599,
      "step": 874
    },
    {
      "epoch": 2.0051502145922746,
      "grad_norm": 0.2998034656047821,
      "learning_rate": 0.00018553425168573665,
      "loss": 1.8428,
      "step": 875
    },
    {
      "epoch": 2.007439198855508,
      "grad_norm": 0.32872092723846436,
      "learning_rate": 0.00018550161865154796,
      "loss": 1.7557,
      "step": 876
    },
    {
      "epoch": 2.009728183118741,
      "grad_norm": 0.34338611364364624,
      "learning_rate": 0.00018546895172724944,
      "loss": 1.4802,
      "step": 877
    },
    {
      "epoch": 2.0120171673819742,
      "grad_norm": 0.4170514941215515,
      "learning_rate": 0.00018543625092578918,
      "loss": 1.7753,
      "step": 878
    },
    {
      "epoch": 2.0143061516452074,
      "grad_norm": 0.4340693950653076,
      "learning_rate": 0.00018540351626012878,
      "loss": 1.6754,
      "step": 879
    },
    {
      "epoch": 2.0165951359084406,
      "grad_norm": 0.43999239802360535,
      "learning_rate": 0.0001853707477432432,
      "loss": 1.5503,
      "step": 880
    },
    {
      "epoch": 2.018884120171674,
      "grad_norm": 0.44529375433921814,
      "learning_rate": 0.0001853379453881208,
      "loss": 1.5,
      "step": 881
    },
    {
      "epoch": 2.021173104434907,
      "grad_norm": 0.48419424891471863,
      "learning_rate": 0.00018530510920776347,
      "loss": 1.4875,
      "step": 882
    },
    {
      "epoch": 2.0234620886981403,
      "grad_norm": 0.47460728883743286,
      "learning_rate": 0.00018527223921518634,
      "loss": 1.4873,
      "step": 883
    },
    {
      "epoch": 2.0257510729613735,
      "grad_norm": 0.4447152316570282,
      "learning_rate": 0.0001852393354234181,
      "loss": 1.2712,
      "step": 884
    },
    {
      "epoch": 2.0280400572246067,
      "grad_norm": 0.494338721036911,
      "learning_rate": 0.00018520639784550067,
      "loss": 1.5571,
      "step": 885
    },
    {
      "epoch": 2.03032904148784,
      "grad_norm": 0.5120548605918884,
      "learning_rate": 0.0001851734264944895,
      "loss": 1.4393,
      "step": 886
    },
    {
      "epoch": 2.032618025751073,
      "grad_norm": 0.5385511517524719,
      "learning_rate": 0.0001851404213834534,
      "loss": 1.5148,
      "step": 887
    },
    {
      "epoch": 2.0349070100143063,
      "grad_norm": 0.5163120031356812,
      "learning_rate": 0.00018510738252547451,
      "loss": 1.3462,
      "step": 888
    },
    {
      "epoch": 2.0371959942775395,
      "grad_norm": 0.563328742980957,
      "learning_rate": 0.0001850743099336484,
      "loss": 1.4456,
      "step": 889
    },
    {
      "epoch": 2.0394849785407727,
      "grad_norm": 0.5758668780326843,
      "learning_rate": 0.000185041203621084,
      "loss": 1.4787,
      "step": 890
    },
    {
      "epoch": 2.041773962804006,
      "grad_norm": 0.5854480862617493,
      "learning_rate": 0.00018500806360090356,
      "loss": 1.407,
      "step": 891
    },
    {
      "epoch": 2.044062947067239,
      "grad_norm": 0.5908710956573486,
      "learning_rate": 0.00018497488988624273,
      "loss": 1.3326,
      "step": 892
    },
    {
      "epoch": 2.046351931330472,
      "grad_norm": 0.6088721752166748,
      "learning_rate": 0.00018494168249025053,
      "loss": 1.25,
      "step": 893
    },
    {
      "epoch": 2.048640915593705,
      "grad_norm": 0.6382647156715393,
      "learning_rate": 0.00018490844142608927,
      "loss": 1.3686,
      "step": 894
    },
    {
      "epoch": 2.0509298998569383,
      "grad_norm": 0.6775152683258057,
      "learning_rate": 0.0001848751667069347,
      "loss": 1.3369,
      "step": 895
    },
    {
      "epoch": 2.0532188841201715,
      "grad_norm": 0.727466881275177,
      "learning_rate": 0.00018484185834597585,
      "loss": 1.2302,
      "step": 896
    },
    {
      "epoch": 2.0555078683834047,
      "grad_norm": 0.6658772230148315,
      "learning_rate": 0.00018480851635641505,
      "loss": 1.1617,
      "step": 897
    },
    {
      "epoch": 2.057796852646638,
      "grad_norm": 0.6813735961914062,
      "learning_rate": 0.00018477514075146802,
      "loss": 0.9391,
      "step": 898
    },
    {
      "epoch": 2.060085836909871,
      "grad_norm": 0.7125244140625,
      "learning_rate": 0.0001847417315443638,
      "loss": 1.0611,
      "step": 899
    },
    {
      "epoch": 2.0623748211731043,
      "grad_norm": 0.7523341774940491,
      "learning_rate": 0.0001847082887483447,
      "loss": 1.1344,
      "step": 900
    },
    {
      "epoch": 2.0646638054363375,
      "grad_norm": 0.7590739130973816,
      "learning_rate": 0.00018467481237666639,
      "loss": 1.0337,
      "step": 901
    },
    {
      "epoch": 2.0669527896995707,
      "grad_norm": 0.9229395985603333,
      "learning_rate": 0.00018464130244259786,
      "loss": 1.1902,
      "step": 902
    },
    {
      "epoch": 2.069241773962804,
      "grad_norm": 0.8496418595314026,
      "learning_rate": 0.00018460775895942134,
      "loss": 0.9731,
      "step": 903
    },
    {
      "epoch": 2.071530758226037,
      "grad_norm": 0.941224217414856,
      "learning_rate": 0.00018457418194043238,
      "loss": 1.1123,
      "step": 904
    },
    {
      "epoch": 2.0738197424892704,
      "grad_norm": 0.9761719107627869,
      "learning_rate": 0.00018454057139893993,
      "loss": 0.9648,
      "step": 905
    },
    {
      "epoch": 2.0761087267525036,
      "grad_norm": 0.8790442943572998,
      "learning_rate": 0.00018450692734826602,
      "loss": 0.9444,
      "step": 906
    },
    {
      "epoch": 2.078397711015737,
      "grad_norm": 0.9426433444023132,
      "learning_rate": 0.00018447324980174615,
      "loss": 0.8087,
      "step": 907
    },
    {
      "epoch": 2.08068669527897,
      "grad_norm": 0.8919757008552551,
      "learning_rate": 0.000184439538772729,
      "loss": 0.7485,
      "step": 908
    },
    {
      "epoch": 2.082975679542203,
      "grad_norm": 1.074262261390686,
      "learning_rate": 0.00018440579427457654,
      "loss": 0.7152,
      "step": 909
    },
    {
      "epoch": 2.0852646638054364,
      "grad_norm": 1.3287183046340942,
      "learning_rate": 0.000184372016320664,
      "loss": 0.9428,
      "step": 910
    },
    {
      "epoch": 2.0875536480686696,
      "grad_norm": 1.219948172569275,
      "learning_rate": 0.00018433820492437994,
      "loss": 0.5857,
      "step": 911
    },
    {
      "epoch": 2.089842632331903,
      "grad_norm": 1.2993091344833374,
      "learning_rate": 0.00018430436009912607,
      "loss": 0.6467,
      "step": 912
    },
    {
      "epoch": 2.092131616595136,
      "grad_norm": 1.2168344259262085,
      "learning_rate": 0.0001842704818583174,
      "loss": 0.5802,
      "step": 913
    },
    {
      "epoch": 2.0944206008583692,
      "grad_norm": 1.1992031335830688,
      "learning_rate": 0.00018423657021538216,
      "loss": 0.5578,
      "step": 914
    },
    {
      "epoch": 2.0967095851216024,
      "grad_norm": 1.432856798171997,
      "learning_rate": 0.0001842026251837619,
      "loss": 0.5725,
      "step": 915
    },
    {
      "epoch": 2.0989985693848356,
      "grad_norm": 1.5111587047576904,
      "learning_rate": 0.00018416864677691131,
      "loss": 0.4993,
      "step": 916
    },
    {
      "epoch": 2.101287553648069,
      "grad_norm": 1.711698055267334,
      "learning_rate": 0.00018413463500829833,
      "loss": 0.7523,
      "step": 917
    },
    {
      "epoch": 2.103576537911302,
      "grad_norm": 1.723205804824829,
      "learning_rate": 0.00018410058989140415,
      "loss": 0.561,
      "step": 918
    },
    {
      "epoch": 2.1058655221745353,
      "grad_norm": 1.83237886428833,
      "learning_rate": 0.00018406651143972318,
      "loss": 0.6931,
      "step": 919
    },
    {
      "epoch": 2.108154506437768,
      "grad_norm": 1.7761093378067017,
      "learning_rate": 0.00018403239966676305,
      "loss": 0.525,
      "step": 920
    },
    {
      "epoch": 2.1104434907010012,
      "grad_norm": 2.166442394256592,
      "learning_rate": 0.00018399825458604452,
      "loss": 0.502,
      "step": 921
    },
    {
      "epoch": 2.1127324749642344,
      "grad_norm": 3.568873882293701,
      "learning_rate": 0.00018396407621110164,
      "loss": 0.6335,
      "step": 922
    },
    {
      "epoch": 2.1150214592274676,
      "grad_norm": 5.127477645874023,
      "learning_rate": 0.00018392986455548164,
      "loss": 1.0344,
      "step": 923
    },
    {
      "epoch": 2.117310443490701,
      "grad_norm": 0.8031274676322937,
      "learning_rate": 0.0001838956196327449,
      "loss": 1.9795,
      "step": 924
    },
    {
      "epoch": 2.119599427753934,
      "grad_norm": 0.7841911911964417,
      "learning_rate": 0.00018386134145646503,
      "loss": 2.018,
      "step": 925
    },
    {
      "epoch": 2.1218884120171673,
      "grad_norm": 0.7021698355674744,
      "learning_rate": 0.0001838270300402288,
      "loss": 1.9867,
      "step": 926
    },
    {
      "epoch": 2.1241773962804005,
      "grad_norm": 0.6819107532501221,
      "learning_rate": 0.00018379268539763617,
      "loss": 1.8327,
      "step": 927
    },
    {
      "epoch": 2.1264663805436337,
      "grad_norm": 0.5971208810806274,
      "learning_rate": 0.00018375830754230026,
      "loss": 1.86,
      "step": 928
    },
    {
      "epoch": 2.128755364806867,
      "grad_norm": 0.56781005859375,
      "learning_rate": 0.00018372389648784728,
      "loss": 1.7978,
      "step": 929
    },
    {
      "epoch": 2.1310443490701,
      "grad_norm": 0.5251395106315613,
      "learning_rate": 0.0001836894522479168,
      "loss": 1.6111,
      "step": 930
    },
    {
      "epoch": 2.1333333333333333,
      "grad_norm": 0.5138212442398071,
      "learning_rate": 0.00018365497483616128,
      "loss": 1.6242,
      "step": 931
    },
    {
      "epoch": 2.1356223175965665,
      "grad_norm": 0.5583555102348328,
      "learning_rate": 0.0001836204642662466,
      "loss": 1.716,
      "step": 932
    },
    {
      "epoch": 2.1379113018597997,
      "grad_norm": 0.506164014339447,
      "learning_rate": 0.00018358592055185154,
      "loss": 1.6194,
      "step": 933
    },
    {
      "epoch": 2.140200286123033,
      "grad_norm": 0.5287511944770813,
      "learning_rate": 0.00018355134370666817,
      "loss": 1.4283,
      "step": 934
    },
    {
      "epoch": 2.142489270386266,
      "grad_norm": 0.5438607335090637,
      "learning_rate": 0.00018351673374440169,
      "loss": 1.5023,
      "step": 935
    },
    {
      "epoch": 2.1447782546494993,
      "grad_norm": 0.5360522270202637,
      "learning_rate": 0.0001834820906787703,
      "loss": 1.4852,
      "step": 936
    },
    {
      "epoch": 2.1470672389127325,
      "grad_norm": 0.5553703904151917,
      "learning_rate": 0.0001834474145235054,
      "loss": 1.6005,
      "step": 937
    },
    {
      "epoch": 2.1493562231759658,
      "grad_norm": 0.5676612854003906,
      "learning_rate": 0.0001834127052923516,
      "loss": 1.5788,
      "step": 938
    },
    {
      "epoch": 2.151645207439199,
      "grad_norm": 0.5647737383842468,
      "learning_rate": 0.00018337796299906648,
      "loss": 1.6008,
      "step": 939
    },
    {
      "epoch": 2.153934191702432,
      "grad_norm": 0.5388182997703552,
      "learning_rate": 0.00018334318765742078,
      "loss": 1.3566,
      "step": 940
    },
    {
      "epoch": 2.1562231759656654,
      "grad_norm": 0.5783641934394836,
      "learning_rate": 0.0001833083792811983,
      "loss": 1.4206,
      "step": 941
    },
    {
      "epoch": 2.1585121602288986,
      "grad_norm": 0.5811404585838318,
      "learning_rate": 0.00018327353788419602,
      "loss": 1.3912,
      "step": 942
    },
    {
      "epoch": 2.160801144492132,
      "grad_norm": 0.5994364023208618,
      "learning_rate": 0.00018323866348022395,
      "loss": 1.2235,
      "step": 943
    },
    {
      "epoch": 2.163090128755365,
      "grad_norm": 0.6608631014823914,
      "learning_rate": 0.00018320375608310514,
      "loss": 1.3675,
      "step": 944
    },
    {
      "epoch": 2.165379113018598,
      "grad_norm": 0.6767647862434387,
      "learning_rate": 0.0001831688157066758,
      "loss": 1.3954,
      "step": 945
    },
    {
      "epoch": 2.1676680972818314,
      "grad_norm": 0.6873164772987366,
      "learning_rate": 0.00018313384236478522,
      "loss": 1.1337,
      "step": 946
    },
    {
      "epoch": 2.169957081545064,
      "grad_norm": 0.7771623134613037,
      "learning_rate": 0.00018309883607129567,
      "loss": 1.1328,
      "step": 947
    },
    {
      "epoch": 2.1722460658082974,
      "grad_norm": 0.707826554775238,
      "learning_rate": 0.0001830637968400825,
      "loss": 1.3041,
      "step": 948
    },
    {
      "epoch": 2.1745350500715306,
      "grad_norm": 0.7644858360290527,
      "learning_rate": 0.00018302872468503416,
      "loss": 1.1299,
      "step": 949
    },
    {
      "epoch": 2.176824034334764,
      "grad_norm": 0.7727388739585876,
      "learning_rate": 0.00018299361962005218,
      "loss": 1.0293,
      "step": 950
    },
    {
      "epoch": 2.179113018597997,
      "grad_norm": 0.7909637689590454,
      "learning_rate": 0.00018295848165905104,
      "loss": 1.1994,
      "step": 951
    },
    {
      "epoch": 2.18140200286123,
      "grad_norm": 0.8658162355422974,
      "learning_rate": 0.0001829233108159583,
      "loss": 1.0583,
      "step": 952
    },
    {
      "epoch": 2.1836909871244634,
      "grad_norm": 0.8348693251609802,
      "learning_rate": 0.0001828881071047146,
      "loss": 0.9745,
      "step": 953
    },
    {
      "epoch": 2.1859799713876966,
      "grad_norm": 0.9874469637870789,
      "learning_rate": 0.00018285287053927354,
      "loss": 1.1132,
      "step": 954
    },
    {
      "epoch": 2.18826895565093,
      "grad_norm": 0.9581871032714844,
      "learning_rate": 0.00018281760113360177,
      "loss": 0.9918,
      "step": 955
    },
    {
      "epoch": 2.190557939914163,
      "grad_norm": 1.0167126655578613,
      "learning_rate": 0.00018278229890167892,
      "loss": 0.8966,
      "step": 956
    },
    {
      "epoch": 2.1928469241773962,
      "grad_norm": 0.9477719068527222,
      "learning_rate": 0.00018274696385749776,
      "loss": 0.8691,
      "step": 957
    },
    {
      "epoch": 2.1951359084406294,
      "grad_norm": 1.0375372171401978,
      "learning_rate": 0.00018271159601506386,
      "loss": 0.8534,
      "step": 958
    },
    {
      "epoch": 2.1974248927038627,
      "grad_norm": 1.0636913776397705,
      "learning_rate": 0.00018267619538839597,
      "loss": 0.8486,
      "step": 959
    },
    {
      "epoch": 2.199713876967096,
      "grad_norm": 1.0690898895263672,
      "learning_rate": 0.0001826407619915258,
      "loss": 0.7082,
      "step": 960
    },
    {
      "epoch": 2.202002861230329,
      "grad_norm": 1.2873176336288452,
      "learning_rate": 0.00018260529583849792,
      "loss": 0.8513,
      "step": 961
    },
    {
      "epoch": 2.2042918454935623,
      "grad_norm": 1.218408226966858,
      "learning_rate": 0.00018256979694337004,
      "loss": 0.651,
      "step": 962
    },
    {
      "epoch": 2.2065808297567955,
      "grad_norm": 1.4870026111602783,
      "learning_rate": 0.0001825342653202128,
      "loss": 0.751,
      "step": 963
    },
    {
      "epoch": 2.2088698140200287,
      "grad_norm": 1.2464792728424072,
      "learning_rate": 0.00018249870098310976,
      "loss": 0.6667,
      "step": 964
    },
    {
      "epoch": 2.211158798283262,
      "grad_norm": 1.353868842124939,
      "learning_rate": 0.0001824631039461575,
      "loss": 0.6421,
      "step": 965
    },
    {
      "epoch": 2.213447782546495,
      "grad_norm": 1.5224379301071167,
      "learning_rate": 0.0001824274742234656,
      "loss": 0.6203,
      "step": 966
    },
    {
      "epoch": 2.2157367668097283,
      "grad_norm": 1.3299599885940552,
      "learning_rate": 0.00018239181182915645,
      "loss": 0.4879,
      "step": 967
    },
    {
      "epoch": 2.2180257510729615,
      "grad_norm": 1.8070518970489502,
      "learning_rate": 0.00018235611677736556,
      "loss": 0.6045,
      "step": 968
    },
    {
      "epoch": 2.2203147353361947,
      "grad_norm": 2.1411795616149902,
      "learning_rate": 0.00018232038908224128,
      "loss": 0.6178,
      "step": 969
    },
    {
      "epoch": 2.222603719599428,
      "grad_norm": 2.281730890274048,
      "learning_rate": 0.00018228462875794488,
      "loss": 0.7351,
      "step": 970
    },
    {
      "epoch": 2.224892703862661,
      "grad_norm": 2.9870357513427734,
      "learning_rate": 0.0001822488358186507,
      "loss": 0.4734,
      "step": 971
    },
    {
      "epoch": 2.227181688125894,
      "grad_norm": 3.60560941696167,
      "learning_rate": 0.00018221301027854586,
      "loss": 0.5749,
      "step": 972
    },
    {
      "epoch": 2.2294706723891276,
      "grad_norm": 3.028756856918335,
      "learning_rate": 0.00018217715215183047,
      "loss": 0.8701,
      "step": 973
    },
    {
      "epoch": 2.2317596566523603,
      "grad_norm": 0.7292337417602539,
      "learning_rate": 0.00018214126145271758,
      "loss": 1.9832,
      "step": 974
    },
    {
      "epoch": 2.2340486409155935,
      "grad_norm": 0.7123857736587524,
      "learning_rate": 0.00018210533819543306,
      "loss": 1.9325,
      "step": 975
    },
    {
      "epoch": 2.2363376251788267,
      "grad_norm": 0.6750438213348389,
      "learning_rate": 0.00018206938239421575,
      "loss": 1.8835,
      "step": 976
    },
    {
      "epoch": 2.23862660944206,
      "grad_norm": 0.6163578033447266,
      "learning_rate": 0.00018203339406331744,
      "loss": 1.7117,
      "step": 977
    },
    {
      "epoch": 2.240915593705293,
      "grad_norm": 0.6570593118667603,
      "learning_rate": 0.00018199737321700267,
      "loss": 1.7142,
      "step": 978
    },
    {
      "epoch": 2.2432045779685263,
      "grad_norm": 0.575005054473877,
      "learning_rate": 0.00018196131986954902,
      "loss": 1.6074,
      "step": 979
    },
    {
      "epoch": 2.2454935622317596,
      "grad_norm": 0.5999249815940857,
      "learning_rate": 0.00018192523403524685,
      "loss": 1.5512,
      "step": 980
    },
    {
      "epoch": 2.2477825464949928,
      "grad_norm": 0.5682541728019714,
      "learning_rate": 0.00018188911572839943,
      "loss": 1.5361,
      "step": 981
    },
    {
      "epoch": 2.250071530758226,
      "grad_norm": 0.5664795637130737,
      "learning_rate": 0.00018185296496332293,
      "loss": 1.7084,
      "step": 982
    },
    {
      "epoch": 2.252360515021459,
      "grad_norm": 0.5817411541938782,
      "learning_rate": 0.00018181678175434635,
      "loss": 1.6734,
      "step": 983
    },
    {
      "epoch": 2.2546494992846924,
      "grad_norm": 0.5698080062866211,
      "learning_rate": 0.00018178056611581155,
      "loss": 1.5067,
      "step": 984
    },
    {
      "epoch": 2.2569384835479256,
      "grad_norm": 0.5515428781509399,
      "learning_rate": 0.00018174431806207325,
      "loss": 1.3399,
      "step": 985
    },
    {
      "epoch": 2.259227467811159,
      "grad_norm": 0.6005468368530273,
      "learning_rate": 0.00018170803760749903,
      "loss": 1.6026,
      "step": 986
    },
    {
      "epoch": 2.261516452074392,
      "grad_norm": 0.6267259120941162,
      "learning_rate": 0.00018167172476646928,
      "loss": 1.6313,
      "step": 987
    },
    {
      "epoch": 2.263805436337625,
      "grad_norm": 0.5913515090942383,
      "learning_rate": 0.00018163537955337726,
      "loss": 1.4842,
      "step": 988
    },
    {
      "epoch": 2.2660944206008584,
      "grad_norm": 0.6300111413002014,
      "learning_rate": 0.00018159900198262907,
      "loss": 1.5439,
      "step": 989
    },
    {
      "epoch": 2.2683834048640916,
      "grad_norm": 0.6443419456481934,
      "learning_rate": 0.00018156259206864362,
      "loss": 1.3522,
      "step": 990
    },
    {
      "epoch": 2.270672389127325,
      "grad_norm": 0.6766319870948792,
      "learning_rate": 0.0001815261498258526,
      "loss": 1.4922,
      "step": 991
    },
    {
      "epoch": 2.272961373390558,
      "grad_norm": 0.6364059448242188,
      "learning_rate": 0.00018148967526870059,
      "loss": 1.3963,
      "step": 992
    },
    {
      "epoch": 2.2752503576537912,
      "grad_norm": 0.71284419298172,
      "learning_rate": 0.00018145316841164489,
      "loss": 1.3948,
      "step": 993
    },
    {
      "epoch": 2.2775393419170245,
      "grad_norm": 0.6994620561599731,
      "learning_rate": 0.0001814166292691557,
      "loss": 1.3376,
      "step": 994
    },
    {
      "epoch": 2.2798283261802577,
      "grad_norm": 0.6798747181892395,
      "learning_rate": 0.00018138005785571593,
      "loss": 1.1566,
      "step": 995
    },
    {
      "epoch": 2.282117310443491,
      "grad_norm": 0.7776591181755066,
      "learning_rate": 0.00018134345418582137,
      "loss": 1.2501,
      "step": 996
    },
    {
      "epoch": 2.284406294706724,
      "grad_norm": 0.9156054854393005,
      "learning_rate": 0.0001813068182739805,
      "loss": 1.1123,
      "step": 997
    },
    {
      "epoch": 2.2866952789699573,
      "grad_norm": 0.7901226878166199,
      "learning_rate": 0.00018127015013471467,
      "loss": 1.1591,
      "step": 998
    },
    {
      "epoch": 2.28898426323319,
      "grad_norm": 0.7805190086364746,
      "learning_rate": 0.0001812334497825579,
      "loss": 1.0669,
      "step": 999
    },
    {
      "epoch": 2.2912732474964237,
      "grad_norm": 0.836260199546814,
      "learning_rate": 0.00018119671723205706,
      "loss": 0.9799,
      "step": 1000
    },
    {
      "epoch": 2.2912732474964237,
      "eval_loss": 1.7065112590789795,
      "eval_runtime": 24.7546,
      "eval_samples_per_second": 29.732,
      "eval_steps_per_second": 14.866,
      "step": 1000
    },
    {
      "epoch": 2.2935622317596565,
      "grad_norm": 0.8386034369468689,
      "learning_rate": 0.0001811599524977718,
      "loss": 1.0359,
      "step": 1001
    },
    {
      "epoch": 2.2958512160228897,
      "grad_norm": 0.9080078601837158,
      "learning_rate": 0.0001811231555942744,
      "loss": 1.1002,
      "step": 1002
    },
    {
      "epoch": 2.298140200286123,
      "grad_norm": 0.8837623596191406,
      "learning_rate": 0.00018108632653615007,
      "loss": 1.1061,
      "step": 1003
    },
    {
      "epoch": 2.300429184549356,
      "grad_norm": 0.9862326979637146,
      "learning_rate": 0.00018104946533799665,
      "loss": 0.9935,
      "step": 1004
    },
    {
      "epoch": 2.3027181688125893,
      "grad_norm": 0.9343245625495911,
      "learning_rate": 0.00018101257201442469,
      "loss": 0.9425,
      "step": 1005
    },
    {
      "epoch": 2.3050071530758225,
      "grad_norm": 0.9747886061668396,
      "learning_rate": 0.00018097564658005756,
      "loss": 0.9542,
      "step": 1006
    },
    {
      "epoch": 2.3072961373390557,
      "grad_norm": 1.003598690032959,
      "learning_rate": 0.0001809386890495313,
      "loss": 0.9832,
      "step": 1007
    },
    {
      "epoch": 2.309585121602289,
      "grad_norm": 1.2396725416183472,
      "learning_rate": 0.00018090169943749476,
      "loss": 0.832,
      "step": 1008
    },
    {
      "epoch": 2.311874105865522,
      "grad_norm": 1.0458011627197266,
      "learning_rate": 0.00018086467775860936,
      "loss": 0.8028,
      "step": 1009
    },
    {
      "epoch": 2.3141630901287553,
      "grad_norm": 1.2435302734375,
      "learning_rate": 0.00018082762402754936,
      "loss": 0.976,
      "step": 1010
    },
    {
      "epoch": 2.3164520743919885,
      "grad_norm": 1.1295289993286133,
      "learning_rate": 0.00018079053825900164,
      "loss": 0.7791,
      "step": 1011
    },
    {
      "epoch": 2.3187410586552217,
      "grad_norm": 1.2901514768600464,
      "learning_rate": 0.00018075342046766586,
      "loss": 0.7847,
      "step": 1012
    },
    {
      "epoch": 2.321030042918455,
      "grad_norm": 1.3471083641052246,
      "learning_rate": 0.00018071627066825428,
      "loss": 0.6205,
      "step": 1013
    },
    {
      "epoch": 2.323319027181688,
      "grad_norm": 1.3079593181610107,
      "learning_rate": 0.00018067908887549195,
      "loss": 0.5735,
      "step": 1014
    },
    {
      "epoch": 2.3256080114449214,
      "grad_norm": 1.449722170829773,
      "learning_rate": 0.00018064187510411644,
      "loss": 0.6728,
      "step": 1015
    },
    {
      "epoch": 2.3278969957081546,
      "grad_norm": 1.5137460231781006,
      "learning_rate": 0.00018060462936887823,
      "loss": 0.668,
      "step": 1016
    },
    {
      "epoch": 2.3301859799713878,
      "grad_norm": 1.8287323713302612,
      "learning_rate": 0.00018056735168454026,
      "loss": 0.7445,
      "step": 1017
    },
    {
      "epoch": 2.332474964234621,
      "grad_norm": 1.9629091024398804,
      "learning_rate": 0.00018053004206587821,
      "loss": 0.6927,
      "step": 1018
    },
    {
      "epoch": 2.334763948497854,
      "grad_norm": 1.8040260076522827,
      "learning_rate": 0.00018049270052768047,
      "loss": 0.5723,
      "step": 1019
    },
    {
      "epoch": 2.3370529327610874,
      "grad_norm": 1.9704588651657104,
      "learning_rate": 0.000180455327084748,
      "loss": 0.6148,
      "step": 1020
    },
    {
      "epoch": 2.3393419170243206,
      "grad_norm": 2.597782611846924,
      "learning_rate": 0.00018041792175189442,
      "loss": 0.5364,
      "step": 1021
    },
    {
      "epoch": 2.341630901287554,
      "grad_norm": 3.8625471591949463,
      "learning_rate": 0.00018038048454394607,
      "loss": 0.5843,
      "step": 1022
    },
    {
      "epoch": 2.343919885550787,
      "grad_norm": 3.2378756999969482,
      "learning_rate": 0.0001803430154757418,
      "loss": 0.8814,
      "step": 1023
    },
    {
      "epoch": 2.34620886981402,
      "grad_norm": 0.6680899858474731,
      "learning_rate": 0.00018030551456213315,
      "loss": 2.0763,
      "step": 1024
    },
    {
      "epoch": 2.3484978540772534,
      "grad_norm": 0.5856165289878845,
      "learning_rate": 0.00018026798181798435,
      "loss": 1.8341,
      "step": 1025
    },
    {
      "epoch": 2.350786838340486,
      "grad_norm": 0.5857422947883606,
      "learning_rate": 0.00018023041725817213,
      "loss": 1.694,
      "step": 1026
    },
    {
      "epoch": 2.35307582260372,
      "grad_norm": 0.6255279779434204,
      "learning_rate": 0.0001801928208975859,
      "loss": 1.679,
      "step": 1027
    },
    {
      "epoch": 2.3553648068669526,
      "grad_norm": 0.6410579085350037,
      "learning_rate": 0.00018015519275112765,
      "loss": 1.6533,
      "step": 1028
    },
    {
      "epoch": 2.357653791130186,
      "grad_norm": 0.6084139943122864,
      "learning_rate": 0.00018011753283371195,
      "loss": 1.5843,
      "step": 1029
    },
    {
      "epoch": 2.359942775393419,
      "grad_norm": 0.6109862327575684,
      "learning_rate": 0.00018007984116026604,
      "loss": 1.7548,
      "step": 1030
    },
    {
      "epoch": 2.362231759656652,
      "grad_norm": 0.5964168906211853,
      "learning_rate": 0.00018004211774572967,
      "loss": 1.6514,
      "step": 1031
    },
    {
      "epoch": 2.3645207439198854,
      "grad_norm": 0.5902396440505981,
      "learning_rate": 0.00018000436260505516,
      "loss": 1.6264,
      "step": 1032
    },
    {
      "epoch": 2.3668097281831186,
      "grad_norm": 0.572144627571106,
      "learning_rate": 0.0001799665757532075,
      "loss": 1.6393,
      "step": 1033
    },
    {
      "epoch": 2.369098712446352,
      "grad_norm": 0.6173785924911499,
      "learning_rate": 0.00017992875720516415,
      "loss": 1.6018,
      "step": 1034
    },
    {
      "epoch": 2.371387696709585,
      "grad_norm": 0.5906036496162415,
      "learning_rate": 0.00017989090697591518,
      "loss": 1.536,
      "step": 1035
    },
    {
      "epoch": 2.3736766809728183,
      "grad_norm": 0.6066197752952576,
      "learning_rate": 0.00017985302508046323,
      "loss": 1.3841,
      "step": 1036
    },
    {
      "epoch": 2.3759656652360515,
      "grad_norm": 0.6423449516296387,
      "learning_rate": 0.0001798151115338235,
      "loss": 1.6754,
      "step": 1037
    },
    {
      "epoch": 2.3782546494992847,
      "grad_norm": 0.613100528717041,
      "learning_rate": 0.0001797771663510236,
      "loss": 1.3935,
      "step": 1038
    },
    {
      "epoch": 2.380543633762518,
      "grad_norm": 0.6120763421058655,
      "learning_rate": 0.00017973918954710395,
      "loss": 1.3653,
      "step": 1039
    },
    {
      "epoch": 2.382832618025751,
      "grad_norm": 0.6863808035850525,
      "learning_rate": 0.00017970118113711715,
      "loss": 1.4394,
      "step": 1040
    },
    {
      "epoch": 2.3851216022889843,
      "grad_norm": 0.631229043006897,
      "learning_rate": 0.0001796631411361287,
      "loss": 1.2361,
      "step": 1041
    },
    {
      "epoch": 2.3874105865522175,
      "grad_norm": 0.6643753051757812,
      "learning_rate": 0.0001796250695592163,
      "loss": 1.5303,
      "step": 1042
    },
    {
      "epoch": 2.3896995708154507,
      "grad_norm": 0.6451334357261658,
      "learning_rate": 0.00017958696642147044,
      "loss": 1.1401,
      "step": 1043
    },
    {
      "epoch": 2.391988555078684,
      "grad_norm": 0.7480964064598083,
      "learning_rate": 0.00017954883173799388,
      "loss": 1.3991,
      "step": 1044
    },
    {
      "epoch": 2.394277539341917,
      "grad_norm": 0.6968662142753601,
      "learning_rate": 0.00017951066552390204,
      "loss": 1.1074,
      "step": 1045
    },
    {
      "epoch": 2.3965665236051503,
      "grad_norm": 0.7453241944313049,
      "learning_rate": 0.00017947246779432275,
      "loss": 1.236,
      "step": 1046
    },
    {
      "epoch": 2.3988555078683835,
      "grad_norm": 0.8179067373275757,
      "learning_rate": 0.00017943423856439642,
      "loss": 1.3621,
      "step": 1047
    },
    {
      "epoch": 2.4011444921316167,
      "grad_norm": 0.7809953689575195,
      "learning_rate": 0.00017939597784927583,
      "loss": 1.1341,
      "step": 1048
    },
    {
      "epoch": 2.40343347639485,
      "grad_norm": 0.8376346826553345,
      "learning_rate": 0.00017935768566412639,
      "loss": 1.1102,
      "step": 1049
    },
    {
      "epoch": 2.405722460658083,
      "grad_norm": 0.8247206807136536,
      "learning_rate": 0.00017931936202412581,
      "loss": 1.0338,
      "step": 1050
    },
    {
      "epoch": 2.4080114449213164,
      "grad_norm": 0.8612480759620667,
      "learning_rate": 0.00017928100694446442,
      "loss": 1.1738,
      "step": 1051
    },
    {
      "epoch": 2.4103004291845496,
      "grad_norm": 0.9406076073646545,
      "learning_rate": 0.0001792426204403449,
      "loss": 1.0237,
      "step": 1052
    },
    {
      "epoch": 2.4125894134477823,
      "grad_norm": 0.9248893857002258,
      "learning_rate": 0.00017920420252698254,
      "loss": 1.0914,
      "step": 1053
    },
    {
      "epoch": 2.414878397711016,
      "grad_norm": 0.9275763630867004,
      "learning_rate": 0.0001791657532196048,
      "loss": 0.9426,
      "step": 1054
    },
    {
      "epoch": 2.4171673819742487,
      "grad_norm": 0.9981589317321777,
      "learning_rate": 0.0001791272725334519,
      "loss": 0.9617,
      "step": 1055
    },
    {
      "epoch": 2.419456366237482,
      "grad_norm": 1.026535153388977,
      "learning_rate": 0.00017908876048377628,
      "loss": 0.9323,
      "step": 1056
    },
    {
      "epoch": 2.421745350500715,
      "grad_norm": 1.0345085859298706,
      "learning_rate": 0.00017905021708584294,
      "loss": 0.7822,
      "step": 1057
    },
    {
      "epoch": 2.4240343347639484,
      "grad_norm": 1.1605072021484375,
      "learning_rate": 0.00017901164235492923,
      "loss": 0.9703,
      "step": 1058
    },
    {
      "epoch": 2.4263233190271816,
      "grad_norm": 1.1480755805969238,
      "learning_rate": 0.00017897303630632489,
      "loss": 0.7877,
      "step": 1059
    },
    {
      "epoch": 2.4286123032904148,
      "grad_norm": 1.1079845428466797,
      "learning_rate": 0.0001789343989553322,
      "loss": 0.675,
      "step": 1060
    },
    {
      "epoch": 2.430901287553648,
      "grad_norm": 1.2436599731445312,
      "learning_rate": 0.00017889573031726572,
      "loss": 0.654,
      "step": 1061
    },
    {
      "epoch": 2.433190271816881,
      "grad_norm": 1.3957083225250244,
      "learning_rate": 0.00017885703040745253,
      "loss": 0.6844,
      "step": 1062
    },
    {
      "epoch": 2.4354792560801144,
      "grad_norm": 1.386443853378296,
      "learning_rate": 0.00017881829924123194,
      "loss": 0.6211,
      "step": 1063
    },
    {
      "epoch": 2.4377682403433476,
      "grad_norm": 1.4803379774093628,
      "learning_rate": 0.00017877953683395582,
      "loss": 0.7822,
      "step": 1064
    },
    {
      "epoch": 2.440057224606581,
      "grad_norm": 1.2971986532211304,
      "learning_rate": 0.0001787407432009883,
      "loss": 0.5583,
      "step": 1065
    },
    {
      "epoch": 2.442346208869814,
      "grad_norm": 1.462306261062622,
      "learning_rate": 0.00017870191835770598,
      "loss": 0.6908,
      "step": 1066
    },
    {
      "epoch": 2.4446351931330472,
      "grad_norm": 1.6145541667938232,
      "learning_rate": 0.00017866306231949775,
      "loss": 0.6788,
      "step": 1067
    },
    {
      "epoch": 2.4469241773962804,
      "grad_norm": 1.935477614402771,
      "learning_rate": 0.00017862417510176496,
      "loss": 0.6175,
      "step": 1068
    },
    {
      "epoch": 2.4492131616595136,
      "grad_norm": 1.959228277206421,
      "learning_rate": 0.00017858525671992119,
      "loss": 0.5429,
      "step": 1069
    },
    {
      "epoch": 2.451502145922747,
      "grad_norm": 1.948767066001892,
      "learning_rate": 0.00017854630718939252,
      "loss": 0.6042,
      "step": 1070
    },
    {
      "epoch": 2.45379113018598,
      "grad_norm": 2.4999332427978516,
      "learning_rate": 0.0001785073265256172,
      "loss": 0.511,
      "step": 1071
    },
    {
      "epoch": 2.4560801144492133,
      "grad_norm": 3.59029483795166,
      "learning_rate": 0.00017846831474404602,
      "loss": 0.7382,
      "step": 1072
    },
    {
      "epoch": 2.4583690987124465,
      "grad_norm": 3.1788058280944824,
      "learning_rate": 0.00017842927186014192,
      "loss": 0.8983,
      "step": 1073
    },
    {
      "epoch": 2.4606580829756797,
      "grad_norm": 0.5815398693084717,
      "learning_rate": 0.00017839019788938032,
      "loss": 1.8965,
      "step": 1074
    },
    {
      "epoch": 2.462947067238913,
      "grad_norm": 0.5826651453971863,
      "learning_rate": 0.00017835109284724886,
      "loss": 1.8977,
      "step": 1075
    },
    {
      "epoch": 2.465236051502146,
      "grad_norm": 0.641663134098053,
      "learning_rate": 0.00017831195674924754,
      "loss": 1.8514,
      "step": 1076
    },
    {
      "epoch": 2.4675250357653793,
      "grad_norm": 0.595970094203949,
      "learning_rate": 0.00017827278961088865,
      "loss": 1.6289,
      "step": 1077
    },
    {
      "epoch": 2.469814020028612,
      "grad_norm": 0.6265684962272644,
      "learning_rate": 0.00017823359144769678,
      "loss": 1.7979,
      "step": 1078
    },
    {
      "epoch": 2.4721030042918457,
      "grad_norm": 0.6197585463523865,
      "learning_rate": 0.00017819436227520886,
      "loss": 1.6857,
      "step": 1079
    },
    {
      "epoch": 2.4743919885550785,
      "grad_norm": 0.6327326893806458,
      "learning_rate": 0.00017815510210897407,
      "loss": 1.8233,
      "step": 1080
    },
    {
      "epoch": 2.476680972818312,
      "grad_norm": 0.5528489947319031,
      "learning_rate": 0.00017811581096455385,
      "loss": 1.5704,
      "step": 1081
    },
    {
      "epoch": 2.478969957081545,
      "grad_norm": 0.6038867831230164,
      "learning_rate": 0.00017807648885752203,
      "loss": 1.6922,
      "step": 1082
    },
    {
      "epoch": 2.481258941344778,
      "grad_norm": 0.6389109492301941,
      "learning_rate": 0.00017803713580346456,
      "loss": 1.7321,
      "step": 1083
    },
    {
      "epoch": 2.4835479256080113,
      "grad_norm": 0.6025187373161316,
      "learning_rate": 0.00017799775181797975,
      "loss": 1.701,
      "step": 1084
    },
    {
      "epoch": 2.4858369098712445,
      "grad_norm": 0.5835283994674683,
      "learning_rate": 0.0001779583369166782,
      "loss": 1.4868,
      "step": 1085
    },
    {
      "epoch": 2.4881258941344777,
      "grad_norm": 0.637869119644165,
      "learning_rate": 0.0001779188911151827,
      "loss": 1.5836,
      "step": 1086
    },
    {
      "epoch": 2.490414878397711,
      "grad_norm": 0.6257422566413879,
      "learning_rate": 0.00017787941442912826,
      "loss": 1.4761,
      "step": 1087
    },
    {
      "epoch": 2.492703862660944,
      "grad_norm": 0.5837921500205994,
      "learning_rate": 0.00017783990687416222,
      "loss": 1.4234,
      "step": 1088
    },
    {
      "epoch": 2.4949928469241773,
      "grad_norm": 0.6109277009963989,
      "learning_rate": 0.00017780036846594407,
      "loss": 1.4361,
      "step": 1089
    },
    {
      "epoch": 2.4972818311874105,
      "grad_norm": 0.6292274594306946,
      "learning_rate": 0.00017776079922014564,
      "loss": 1.4884,
      "step": 1090
    },
    {
      "epoch": 2.4995708154506437,
      "grad_norm": 0.6379925608634949,
      "learning_rate": 0.00017772119915245087,
      "loss": 1.4197,
      "step": 1091
    },
    {
      "epoch": 2.501859799713877,
      "grad_norm": 0.6590442657470703,
      "learning_rate": 0.00017768156827855598,
      "loss": 1.2775,
      "step": 1092
    },
    {
      "epoch": 2.50414878397711,
      "grad_norm": 0.7020080089569092,
      "learning_rate": 0.00017764190661416932,
      "loss": 1.2984,
      "step": 1093
    },
    {
      "epoch": 2.5064377682403434,
      "grad_norm": 0.6748842000961304,
      "learning_rate": 0.0001776022141750116,
      "loss": 1.3618,
      "step": 1094
    },
    {
      "epoch": 2.5087267525035766,
      "grad_norm": 0.7022404074668884,
      "learning_rate": 0.0001775624909768156,
      "loss": 1.4772,
      "step": 1095
    },
    {
      "epoch": 2.51101573676681,
      "grad_norm": 0.7599904537200928,
      "learning_rate": 0.00017752273703532632,
      "loss": 1.3722,
      "step": 1096
    },
    {
      "epoch": 2.513304721030043,
      "grad_norm": 0.7587012648582458,
      "learning_rate": 0.00017748295236630094,
      "loss": 1.2397,
      "step": 1097
    },
    {
      "epoch": 2.515593705293276,
      "grad_norm": 0.7206519842147827,
      "learning_rate": 0.00017744313698550882,
      "loss": 1.1942,
      "step": 1098
    },
    {
      "epoch": 2.5178826895565094,
      "grad_norm": 0.7684889435768127,
      "learning_rate": 0.0001774032909087316,
      "loss": 1.2748,
      "step": 1099
    },
    {
      "epoch": 2.5201716738197426,
      "grad_norm": 0.780749499797821,
      "learning_rate": 0.00017736341415176284,
      "loss": 1.1898,
      "step": 1100
    },
    {
      "epoch": 2.522460658082976,
      "grad_norm": 0.8183916211128235,
      "learning_rate": 0.0001773235067304085,
      "loss": 1.3704,
      "step": 1101
    },
    {
      "epoch": 2.524749642346209,
      "grad_norm": 0.9648556709289551,
      "learning_rate": 0.00017728356866048665,
      "loss": 1.222,
      "step": 1102
    },
    {
      "epoch": 2.527038626609442,
      "grad_norm": 0.8041174411773682,
      "learning_rate": 0.00017724359995782737,
      "loss": 1.0203,
      "step": 1103
    },
    {
      "epoch": 2.5293276108726754,
      "grad_norm": 0.9442490339279175,
      "learning_rate": 0.00017720360063827305,
      "loss": 1.2035,
      "step": 1104
    },
    {
      "epoch": 2.531616595135908,
      "grad_norm": 0.8448560237884521,
      "learning_rate": 0.0001771635707176781,
      "loss": 0.9913,
      "step": 1105
    },
    {
      "epoch": 2.533905579399142,
      "grad_norm": 0.9782529473304749,
      "learning_rate": 0.0001771235102119091,
      "loss": 1.1012,
      "step": 1106
    },
    {
      "epoch": 2.5361945636623746,
      "grad_norm": 0.9973860383033752,
      "learning_rate": 0.0001770834191368448,
      "loss": 0.9734,
      "step": 1107
    },
    {
      "epoch": 2.5384835479256083,
      "grad_norm": 1.0277292728424072,
      "learning_rate": 0.00017704329750837597,
      "loss": 0.8498,
      "step": 1108
    },
    {
      "epoch": 2.540772532188841,
      "grad_norm": 1.102047324180603,
      "learning_rate": 0.00017700314534240555,
      "loss": 0.9516,
      "step": 1109
    },
    {
      "epoch": 2.5430615164520742,
      "grad_norm": 1.0190706253051758,
      "learning_rate": 0.00017696296265484862,
      "loss": 0.7872,
      "step": 1110
    },
    {
      "epoch": 2.5453505007153074,
      "grad_norm": 1.0096639394760132,
      "learning_rate": 0.00017692274946163225,
      "loss": 0.7742,
      "step": 1111
    },
    {
      "epoch": 2.5476394849785406,
      "grad_norm": 1.2105683088302612,
      "learning_rate": 0.0001768825057786957,
      "loss": 0.66,
      "step": 1112
    },
    {
      "epoch": 2.549928469241774,
      "grad_norm": 1.3456848859786987,
      "learning_rate": 0.0001768422316219903,
      "loss": 0.7485,
      "step": 1113
    },
    {
      "epoch": 2.552217453505007,
      "grad_norm": 1.259321689605713,
      "learning_rate": 0.0001768019270074794,
      "loss": 0.628,
      "step": 1114
    },
    {
      "epoch": 2.5545064377682403,
      "grad_norm": 1.3102505207061768,
      "learning_rate": 0.0001767615919511385,
      "loss": 0.5815,
      "step": 1115
    },
    {
      "epoch": 2.5567954220314735,
      "grad_norm": 1.193208932876587,
      "learning_rate": 0.00017672122646895507,
      "loss": 0.5724,
      "step": 1116
    },
    {
      "epoch": 2.5590844062947067,
      "grad_norm": 1.395561933517456,
      "learning_rate": 0.00017668083057692874,
      "loss": 0.6288,
      "step": 1117
    },
    {
      "epoch": 2.56137339055794,
      "grad_norm": 1.6082841157913208,
      "learning_rate": 0.00017664040429107115,
      "loss": 0.797,
      "step": 1118
    },
    {
      "epoch": 2.563662374821173,
      "grad_norm": 1.6661381721496582,
      "learning_rate": 0.00017659994762740598,
      "loss": 0.5327,
      "step": 1119
    },
    {
      "epoch": 2.5659513590844063,
      "grad_norm": 2.2218363285064697,
      "learning_rate": 0.00017655946060196896,
      "loss": 0.5872,
      "step": 1120
    },
    {
      "epoch": 2.5682403433476395,
      "grad_norm": 2.8575551509857178,
      "learning_rate": 0.00017651894323080782,
      "loss": 0.6397,
      "step": 1121
    },
    {
      "epoch": 2.5705293276108727,
      "grad_norm": 2.5842721462249756,
      "learning_rate": 0.00017647839552998242,
      "loss": 0.5578,
      "step": 1122
    },
    {
      "epoch": 2.572818311874106,
      "grad_norm": 3.7809016704559326,
      "learning_rate": 0.00017643781751556454,
      "loss": 1.0117,
      "step": 1123
    },
    {
      "epoch": 2.575107296137339,
      "grad_norm": 0.5767310857772827,
      "learning_rate": 0.000176397209203638,
      "loss": 1.9668,
      "step": 1124
    },
    {
      "epoch": 2.5773962804005723,
      "grad_norm": 0.5222098231315613,
      "learning_rate": 0.0001763565706102986,
      "loss": 1.8954,
      "step": 1125
    },
    {
      "epoch": 2.5796852646638055,
      "grad_norm": 0.56516432762146,
      "learning_rate": 0.0001763159017516543,
      "loss": 1.8069,
      "step": 1126
    },
    {
      "epoch": 2.5819742489270388,
      "grad_norm": 0.6584917902946472,
      "learning_rate": 0.00017627520264382483,
      "loss": 1.7881,
      "step": 1127
    },
    {
      "epoch": 2.584263233190272,
      "grad_norm": 0.6168023943901062,
      "learning_rate": 0.00017623447330294203,
      "loss": 1.8305,
      "step": 1128
    },
    {
      "epoch": 2.586552217453505,
      "grad_norm": 0.6040744185447693,
      "learning_rate": 0.00017619371374514973,
      "loss": 1.5601,
      "step": 1129
    },
    {
      "epoch": 2.588841201716738,
      "grad_norm": 0.5929082632064819,
      "learning_rate": 0.00017615292398660373,
      "loss": 1.5202,
      "step": 1130
    },
    {
      "epoch": 2.5911301859799716,
      "grad_norm": 0.644721508026123,
      "learning_rate": 0.00017611210404347172,
      "loss": 1.6525,
      "step": 1131
    },
    {
      "epoch": 2.5934191702432043,
      "grad_norm": 0.6615864634513855,
      "learning_rate": 0.0001760712539319335,
      "loss": 1.6018,
      "step": 1132
    },
    {
      "epoch": 2.595708154506438,
      "grad_norm": 0.6303272247314453,
      "learning_rate": 0.00017603037366818071,
      "loss": 1.4965,
      "step": 1133
    },
    {
      "epoch": 2.5979971387696708,
      "grad_norm": 0.5902919173240662,
      "learning_rate": 0.00017598946326841702,
      "loss": 1.6197,
      "step": 1134
    },
    {
      "epoch": 2.6002861230329044,
      "grad_norm": 0.6038529276847839,
      "learning_rate": 0.00017594852274885794,
      "loss": 1.5555,
      "step": 1135
    },
    {
      "epoch": 2.602575107296137,
      "grad_norm": 0.6354286074638367,
      "learning_rate": 0.00017590755212573103,
      "loss": 1.4009,
      "step": 1136
    },
    {
      "epoch": 2.6048640915593704,
      "grad_norm": 0.6249221563339233,
      "learning_rate": 0.00017586655141527572,
      "loss": 1.5373,
      "step": 1137
    },
    {
      "epoch": 2.6071530758226036,
      "grad_norm": 0.6424983739852905,
      "learning_rate": 0.0001758255206337434,
      "loss": 1.3605,
      "step": 1138
    },
    {
      "epoch": 2.609442060085837,
      "grad_norm": 0.6414629817008972,
      "learning_rate": 0.00017578445979739735,
      "loss": 1.3734,
      "step": 1139
    },
    {
      "epoch": 2.61173104434907,
      "grad_norm": 0.7130486369132996,
      "learning_rate": 0.00017574336892251275,
      "loss": 1.4739,
      "step": 1140
    },
    {
      "epoch": 2.614020028612303,
      "grad_norm": 0.6644805073738098,
      "learning_rate": 0.0001757022480253767,
      "loss": 1.2916,
      "step": 1141
    },
    {
      "epoch": 2.6163090128755364,
      "grad_norm": 0.6536266803741455,
      "learning_rate": 0.0001756610971222883,
      "loss": 1.1959,
      "step": 1142
    },
    {
      "epoch": 2.6185979971387696,
      "grad_norm": 0.7095915079116821,
      "learning_rate": 0.00017561991622955838,
      "loss": 1.3816,
      "step": 1143
    },
    {
      "epoch": 2.620886981402003,
      "grad_norm": 0.7044734358787537,
      "learning_rate": 0.0001755787053635097,
      "loss": 1.242,
      "step": 1144
    },
    {
      "epoch": 2.623175965665236,
      "grad_norm": 0.697364091873169,
      "learning_rate": 0.00017553746454047705,
      "loss": 1.2702,
      "step": 1145
    },
    {
      "epoch": 2.6254649499284692,
      "grad_norm": 0.714318573474884,
      "learning_rate": 0.00017549619377680683,
      "loss": 1.276,
      "step": 1146
    },
    {
      "epoch": 2.6277539341917024,
      "grad_norm": 0.7835359573364258,
      "learning_rate": 0.00017545489308885755,
      "loss": 1.2941,
      "step": 1147
    },
    {
      "epoch": 2.6300429184549357,
      "grad_norm": 0.6969245076179504,
      "learning_rate": 0.00017541356249299944,
      "loss": 1.1615,
      "step": 1148
    },
    {
      "epoch": 2.632331902718169,
      "grad_norm": 0.8150187730789185,
      "learning_rate": 0.00017537220200561466,
      "loss": 1.2655,
      "step": 1149
    },
    {
      "epoch": 2.634620886981402,
      "grad_norm": 0.879152774810791,
      "learning_rate": 0.00017533081164309718,
      "loss": 1.2389,
      "step": 1150
    },
    {
      "epoch": 2.6369098712446353,
      "grad_norm": 0.8398150205612183,
      "learning_rate": 0.00017528939142185277,
      "loss": 1.0878,
      "step": 1151
    },
    {
      "epoch": 2.6391988555078685,
      "grad_norm": 0.8729166984558105,
      "learning_rate": 0.00017524794135829913,
      "loss": 1.1215,
      "step": 1152
    },
    {
      "epoch": 2.6414878397711017,
      "grad_norm": 0.9199173450469971,
      "learning_rate": 0.0001752064614688657,
      "loss": 1.2671,
      "step": 1153
    },
    {
      "epoch": 2.643776824034335,
      "grad_norm": 0.9060810208320618,
      "learning_rate": 0.00017516495176999385,
      "loss": 1.0512,
      "step": 1154
    },
    {
      "epoch": 2.646065808297568,
      "grad_norm": 0.9893232583999634,
      "learning_rate": 0.00017512341227813666,
      "loss": 1.0284,
      "step": 1155
    },
    {
      "epoch": 2.6483547925608013,
      "grad_norm": 0.9738585352897644,
      "learning_rate": 0.00017508184300975903,
      "loss": 1.0106,
      "step": 1156
    },
    {
      "epoch": 2.650643776824034,
      "grad_norm": 1.0360881090164185,
      "learning_rate": 0.00017504024398133774,
      "loss": 1.1363,
      "step": 1157
    },
    {
      "epoch": 2.6529327610872677,
      "grad_norm": 1.0261789560317993,
      "learning_rate": 0.00017499861520936126,
      "loss": 0.8798,
      "step": 1158
    },
    {
      "epoch": 2.6552217453505005,
      "grad_norm": 1.15342378616333,
      "learning_rate": 0.00017495695671032997,
      "loss": 0.8741,
      "step": 1159
    },
    {
      "epoch": 2.657510729613734,
      "grad_norm": 1.1546741724014282,
      "learning_rate": 0.00017491526850075593,
      "loss": 0.9994,
      "step": 1160
    },
    {
      "epoch": 2.659799713876967,
      "grad_norm": 1.1052244901657104,
      "learning_rate": 0.00017487355059716303,
      "loss": 0.7856,
      "step": 1161
    },
    {
      "epoch": 2.6620886981402005,
      "grad_norm": 1.4282923936843872,
      "learning_rate": 0.00017483180301608686,
      "loss": 0.6653,
      "step": 1162
    },
    {
      "epoch": 2.6643776824034333,
      "grad_norm": 1.3041657209396362,
      "learning_rate": 0.00017479002577407488,
      "loss": 0.7794,
      "step": 1163
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 1.3872514963150024,
      "learning_rate": 0.00017474821888768628,
      "loss": 0.7402,
      "step": 1164
    },
    {
      "epoch": 2.6689556509298997,
      "grad_norm": 1.4282207489013672,
      "learning_rate": 0.00017470638237349192,
      "loss": 0.6714,
      "step": 1165
    },
    {
      "epoch": 2.671244635193133,
      "grad_norm": 1.3409379720687866,
      "learning_rate": 0.00017466451624807447,
      "loss": 0.5722,
      "step": 1166
    },
    {
      "epoch": 2.673533619456366,
      "grad_norm": 1.5510210990905762,
      "learning_rate": 0.00017462262052802833,
      "loss": 0.7159,
      "step": 1167
    },
    {
      "epoch": 2.6758226037195993,
      "grad_norm": 1.7008919715881348,
      "learning_rate": 0.00017458069522995965,
      "loss": 0.6618,
      "step": 1168
    },
    {
      "epoch": 2.6781115879828326,
      "grad_norm": 1.945682406425476,
      "learning_rate": 0.00017453874037048623,
      "loss": 0.6938,
      "step": 1169
    },
    {
      "epoch": 2.6804005722460658,
      "grad_norm": 2.1040713787078857,
      "learning_rate": 0.00017449675596623764,
      "loss": 0.6935,
      "step": 1170
    },
    {
      "epoch": 2.682689556509299,
      "grad_norm": 2.7217857837677,
      "learning_rate": 0.00017445474203385523,
      "loss": 0.6595,
      "step": 1171
    },
    {
      "epoch": 2.684978540772532,
      "grad_norm": 3.4915311336517334,
      "learning_rate": 0.0001744126985899919,
      "loss": 0.584,
      "step": 1172
    },
    {
      "epoch": 2.6872675250357654,
      "grad_norm": 3.028200387954712,
      "learning_rate": 0.00017437062565131236,
      "loss": 1.0606,
      "step": 1173
    },
    {
      "epoch": 2.6895565092989986,
      "grad_norm": 0.5180421471595764,
      "learning_rate": 0.00017432852323449297,
      "loss": 1.8777,
      "step": 1174
    },
    {
      "epoch": 2.691845493562232,
      "grad_norm": 0.5247154235839844,
      "learning_rate": 0.00017428639135622177,
      "loss": 1.8795,
      "step": 1175
    },
    {
      "epoch": 2.694134477825465,
      "grad_norm": 0.5039599537849426,
      "learning_rate": 0.0001742442300331985,
      "loss": 1.7689,
      "step": 1176
    },
    {
      "epoch": 2.696423462088698,
      "grad_norm": 0.5855678915977478,
      "learning_rate": 0.0001742020392821346,
      "loss": 1.7705,
      "step": 1177
    },
    {
      "epoch": 2.6987124463519314,
      "grad_norm": 0.5479508638381958,
      "learning_rate": 0.0001741598191197531,
      "loss": 1.6392,
      "step": 1178
    },
    {
      "epoch": 2.7010014306151646,
      "grad_norm": 0.6287244558334351,
      "learning_rate": 0.0001741175695627887,
      "loss": 1.8446,
      "step": 1179
    },
    {
      "epoch": 2.703290414878398,
      "grad_norm": 0.5668229460716248,
      "learning_rate": 0.00017407529062798784,
      "loss": 1.561,
      "step": 1180
    },
    {
      "epoch": 2.705579399141631,
      "grad_norm": 0.5956827998161316,
      "learning_rate": 0.00017403298233210847,
      "loss": 1.6266,
      "step": 1181
    },
    {
      "epoch": 2.7078683834048642,
      "grad_norm": 0.5945736765861511,
      "learning_rate": 0.0001739906446919203,
      "loss": 1.7204,
      "step": 1182
    },
    {
      "epoch": 2.7101573676680975,
      "grad_norm": 0.6281150579452515,
      "learning_rate": 0.00017394827772420458,
      "loss": 1.5635,
      "step": 1183
    },
    {
      "epoch": 2.71244635193133,
      "grad_norm": 0.6402949094772339,
      "learning_rate": 0.00017390588144575425,
      "loss": 1.4419,
      "step": 1184
    },
    {
      "epoch": 2.714735336194564,
      "grad_norm": 0.6174708008766174,
      "learning_rate": 0.0001738634558733738,
      "loss": 1.5809,
      "step": 1185
    },
    {
      "epoch": 2.7170243204577966,
      "grad_norm": 0.6724423170089722,
      "learning_rate": 0.00017382100102387944,
      "loss": 1.3466,
      "step": 1186
    },
    {
      "epoch": 2.7193133047210303,
      "grad_norm": 0.6254955530166626,
      "learning_rate": 0.00017377851691409885,
      "loss": 1.5758,
      "step": 1187
    },
    {
      "epoch": 2.721602288984263,
      "grad_norm": 0.6141995787620544,
      "learning_rate": 0.0001737360035608714,
      "loss": 1.4968,
      "step": 1188
    },
    {
      "epoch": 2.7238912732474967,
      "grad_norm": 0.6651355624198914,
      "learning_rate": 0.00017369346098104803,
      "loss": 1.5747,
      "step": 1189
    },
    {
      "epoch": 2.7261802575107295,
      "grad_norm": 0.6363756656646729,
      "learning_rate": 0.00017365088919149124,
      "loss": 1.5275,
      "step": 1190
    },
    {
      "epoch": 2.7284692417739627,
      "grad_norm": 0.6541668176651001,
      "learning_rate": 0.00017360828820907516,
      "loss": 1.411,
      "step": 1191
    },
    {
      "epoch": 2.730758226037196,
      "grad_norm": 0.7034496665000916,
      "learning_rate": 0.00017356565805068543,
      "loss": 1.5428,
      "step": 1192
    },
    {
      "epoch": 2.733047210300429,
      "grad_norm": 0.7091159224510193,
      "learning_rate": 0.0001735229987332193,
      "loss": 1.4255,
      "step": 1193
    },
    {
      "epoch": 2.7353361945636623,
      "grad_norm": 0.6773807406425476,
      "learning_rate": 0.00017348031027358552,
      "loss": 1.4102,
      "step": 1194
    },
    {
      "epoch": 2.7376251788268955,
      "grad_norm": 0.6584862470626831,
      "learning_rate": 0.00017343759268870448,
      "loss": 1.1923,
      "step": 1195
    },
    {
      "epoch": 2.7399141630901287,
      "grad_norm": 0.7572290301322937,
      "learning_rate": 0.00017339484599550805,
      "loss": 1.3278,
      "step": 1196
    },
    {
      "epoch": 2.742203147353362,
      "grad_norm": 0.7460923790931702,
      "learning_rate": 0.00017335207021093965,
      "loss": 1.4353,
      "step": 1197
    },
    {
      "epoch": 2.744492131616595,
      "grad_norm": 0.7679999470710754,
      "learning_rate": 0.00017330926535195427,
      "loss": 1.4133,
      "step": 1198
    },
    {
      "epoch": 2.7467811158798283,
      "grad_norm": 0.7645053863525391,
      "learning_rate": 0.0001732664314355183,
      "loss": 1.1983,
      "step": 1199
    },
    {
      "epoch": 2.7490701001430615,
      "grad_norm": 0.8316181302070618,
      "learning_rate": 0.00017322356847860983,
      "loss": 1.2437,
      "step": 1200
    },
    {
      "epoch": 2.7513590844062947,
      "grad_norm": 0.7712181210517883,
      "learning_rate": 0.00017318067649821833,
      "loss": 1.1768,
      "step": 1201
    },
    {
      "epoch": 2.753648068669528,
      "grad_norm": 0.8656412959098816,
      "learning_rate": 0.0001731377555113448,
      "loss": 1.2009,
      "step": 1202
    },
    {
      "epoch": 2.755937052932761,
      "grad_norm": 0.8733013272285461,
      "learning_rate": 0.00017309480553500173,
      "loss": 1.2116,
      "step": 1203
    },
    {
      "epoch": 2.7582260371959944,
      "grad_norm": 0.8486379981040955,
      "learning_rate": 0.00017305182658621315,
      "loss": 1.2083,
      "step": 1204
    },
    {
      "epoch": 2.7605150214592276,
      "grad_norm": 0.9145635962486267,
      "learning_rate": 0.00017300881868201456,
      "loss": 1.1671,
      "step": 1205
    },
    {
      "epoch": 2.7628040057224608,
      "grad_norm": 0.9506994485855103,
      "learning_rate": 0.0001729657818394529,
      "loss": 1.0751,
      "step": 1206
    },
    {
      "epoch": 2.765092989985694,
      "grad_norm": 1.0189785957336426,
      "learning_rate": 0.0001729227160755866,
      "loss": 1.0213,
      "step": 1207
    },
    {
      "epoch": 2.767381974248927,
      "grad_norm": 1.0177013874053955,
      "learning_rate": 0.00017287962140748552,
      "loss": 1.1028,
      "step": 1208
    },
    {
      "epoch": 2.7696709585121604,
      "grad_norm": 1.0275033712387085,
      "learning_rate": 0.00017283649785223104,
      "loss": 0.9253,
      "step": 1209
    },
    {
      "epoch": 2.7719599427753936,
      "grad_norm": 1.1066482067108154,
      "learning_rate": 0.00017279334542691596,
      "loss": 0.9672,
      "step": 1210
    },
    {
      "epoch": 2.7742489270386264,
      "grad_norm": 1.1275511980056763,
      "learning_rate": 0.00017275016414864454,
      "loss": 0.8078,
      "step": 1211
    },
    {
      "epoch": 2.77653791130186,
      "grad_norm": 1.1769921779632568,
      "learning_rate": 0.00017270695403453243,
      "loss": 0.6926,
      "step": 1212
    },
    {
      "epoch": 2.7788268955650928,
      "grad_norm": 1.2011395692825317,
      "learning_rate": 0.00017266371510170673,
      "loss": 0.6458,
      "step": 1213
    },
    {
      "epoch": 2.7811158798283264,
      "grad_norm": 1.3434557914733887,
      "learning_rate": 0.000172620447367306,
      "loss": 0.6473,
      "step": 1214
    },
    {
      "epoch": 2.783404864091559,
      "grad_norm": 1.3743470907211304,
      "learning_rate": 0.00017257715084848018,
      "loss": 0.7497,
      "step": 1215
    },
    {
      "epoch": 2.785693848354793,
      "grad_norm": 1.2919795513153076,
      "learning_rate": 0.00017253382556239064,
      "loss": 0.5724,
      "step": 1216
    },
    {
      "epoch": 2.7879828326180256,
      "grad_norm": 1.2559446096420288,
      "learning_rate": 0.00017249047152621008,
      "loss": 0.5841,
      "step": 1217
    },
    {
      "epoch": 2.790271816881259,
      "grad_norm": 1.7144775390625,
      "learning_rate": 0.00017244708875712274,
      "loss": 0.7369,
      "step": 1218
    },
    {
      "epoch": 2.792560801144492,
      "grad_norm": 1.819254994392395,
      "learning_rate": 0.0001724036772723241,
      "loss": 0.6058,
      "step": 1219
    },
    {
      "epoch": 2.794849785407725,
      "grad_norm": 2.884782075881958,
      "learning_rate": 0.00017236023708902111,
      "loss": 0.7573,
      "step": 1220
    },
    {
      "epoch": 2.7971387696709584,
      "grad_norm": 2.429008960723877,
      "learning_rate": 0.00017231676822443206,
      "loss": 0.5119,
      "step": 1221
    },
    {
      "epoch": 2.7994277539341916,
      "grad_norm": 3.006176710128784,
      "learning_rate": 0.00017227327069578664,
      "loss": 0.6519,
      "step": 1222
    },
    {
      "epoch": 2.801716738197425,
      "grad_norm": 3.4379966259002686,
      "learning_rate": 0.00017222974452032585,
      "loss": 1.0953,
      "step": 1223
    },
    {
      "epoch": 2.804005722460658,
      "grad_norm": 0.47699180245399475,
      "learning_rate": 0.0001721861897153021,
      "loss": 1.8881,
      "step": 1224
    },
    {
      "epoch": 2.8062947067238913,
      "grad_norm": 0.5090044736862183,
      "learning_rate": 0.00017214260629797913,
      "loss": 1.7905,
      "step": 1225
    },
    {
      "epoch": 2.8085836909871245,
      "grad_norm": 0.5802180767059326,
      "learning_rate": 0.00017209899428563197,
      "loss": 1.7121,
      "step": 1226
    },
    {
      "epoch": 2.8108726752503577,
      "grad_norm": 0.5835916996002197,
      "learning_rate": 0.00017205535369554706,
      "loss": 1.6605,
      "step": 1227
    },
    {
      "epoch": 2.813161659513591,
      "grad_norm": 0.595698356628418,
      "learning_rate": 0.00017201168454502214,
      "loss": 1.6849,
      "step": 1228
    },
    {
      "epoch": 2.815450643776824,
      "grad_norm": 0.617529034614563,
      "learning_rate": 0.00017196798685136625,
      "loss": 1.641,
      "step": 1229
    },
    {
      "epoch": 2.8177396280400573,
      "grad_norm": 0.6735660433769226,
      "learning_rate": 0.00017192426063189983,
      "loss": 1.7236,
      "step": 1230
    },
    {
      "epoch": 2.8200286123032905,
      "grad_norm": 0.6534542441368103,
      "learning_rate": 0.0001718805059039544,
      "loss": 1.5511,
      "step": 1231
    },
    {
      "epoch": 2.8223175965665237,
      "grad_norm": 0.621945858001709,
      "learning_rate": 0.00017183672268487313,
      "loss": 1.5118,
      "step": 1232
    },
    {
      "epoch": 2.824606580829757,
      "grad_norm": 0.6952318549156189,
      "learning_rate": 0.0001717929109920101,
      "loss": 1.5235,
      "step": 1233
    },
    {
      "epoch": 2.82689556509299,
      "grad_norm": 0.6558645367622375,
      "learning_rate": 0.000171749070842731,
      "loss": 1.5219,
      "step": 1234
    },
    {
      "epoch": 2.8291845493562233,
      "grad_norm": 0.6748717427253723,
      "learning_rate": 0.00017170520225441262,
      "loss": 1.568,
      "step": 1235
    },
    {
      "epoch": 2.8314735336194565,
      "grad_norm": 0.6606290340423584,
      "learning_rate": 0.00017166130524444307,
      "loss": 1.5027,
      "step": 1236
    },
    {
      "epoch": 2.8337625178826897,
      "grad_norm": 0.6707291007041931,
      "learning_rate": 0.0001716173798302217,
      "loss": 1.4001,
      "step": 1237
    },
    {
      "epoch": 2.8360515021459225,
      "grad_norm": 0.7039106488227844,
      "learning_rate": 0.00017157342602915917,
      "loss": 1.3474,
      "step": 1238
    },
    {
      "epoch": 2.838340486409156,
      "grad_norm": 0.7407739758491516,
      "learning_rate": 0.00017152944385867737,
      "loss": 1.6713,
      "step": 1239
    },
    {
      "epoch": 2.840629470672389,
      "grad_norm": 0.6728647351264954,
      "learning_rate": 0.0001714854333362094,
      "loss": 1.3375,
      "step": 1240
    },
    {
      "epoch": 2.8429184549356226,
      "grad_norm": 0.6941105127334595,
      "learning_rate": 0.00017144139447919962,
      "loss": 1.4543,
      "step": 1241
    },
    {
      "epoch": 2.8452074391988553,
      "grad_norm": 0.7132325768470764,
      "learning_rate": 0.00017139732730510362,
      "loss": 1.5217,
      "step": 1242
    },
    {
      "epoch": 2.8474964234620885,
      "grad_norm": 0.7110108733177185,
      "learning_rate": 0.00017135323183138827,
      "loss": 1.3023,
      "step": 1243
    },
    {
      "epoch": 2.8497854077253217,
      "grad_norm": 0.7270708084106445,
      "learning_rate": 0.00017130910807553157,
      "loss": 1.3384,
      "step": 1244
    },
    {
      "epoch": 2.852074391988555,
      "grad_norm": 0.7716068029403687,
      "learning_rate": 0.0001712649560550228,
      "loss": 1.4268,
      "step": 1245
    },
    {
      "epoch": 2.854363376251788,
      "grad_norm": 0.8135033249855042,
      "learning_rate": 0.00017122077578736236,
      "loss": 1.2755,
      "step": 1246
    },
    {
      "epoch": 2.8566523605150214,
      "grad_norm": 0.7812631726264954,
      "learning_rate": 0.00017117656729006192,
      "loss": 1.2313,
      "step": 1247
    },
    {
      "epoch": 2.8589413447782546,
      "grad_norm": 0.7625741958618164,
      "learning_rate": 0.0001711323305806443,
      "loss": 1.1652,
      "step": 1248
    },
    {
      "epoch": 2.8612303290414878,
      "grad_norm": 0.8287955522537231,
      "learning_rate": 0.0001710880656766436,
      "loss": 1.0599,
      "step": 1249
    },
    {
      "epoch": 2.863519313304721,
      "grad_norm": 0.847343921661377,
      "learning_rate": 0.00017104377259560488,
      "loss": 1.0513,
      "step": 1250
    },
    {
      "epoch": 2.865808297567954,
      "grad_norm": 0.906118631362915,
      "learning_rate": 0.0001709994513550846,
      "loss": 1.2882,
      "step": 1251
    },
    {
      "epoch": 2.8680972818311874,
      "grad_norm": 0.8588566780090332,
      "learning_rate": 0.00017095510197265025,
      "loss": 1.0636,
      "step": 1252
    },
    {
      "epoch": 2.8703862660944206,
      "grad_norm": 0.9555876851081848,
      "learning_rate": 0.00017091072446588052,
      "loss": 1.1668,
      "step": 1253
    },
    {
      "epoch": 2.872675250357654,
      "grad_norm": 0.9155262112617493,
      "learning_rate": 0.0001708663188523652,
      "loss": 1.1219,
      "step": 1254
    },
    {
      "epoch": 2.874964234620887,
      "grad_norm": 1.0060838460922241,
      "learning_rate": 0.0001708218851497053,
      "loss": 1.3486,
      "step": 1255
    },
    {
      "epoch": 2.8772532188841202,
      "grad_norm": 1.0023645162582397,
      "learning_rate": 0.00017077742337551292,
      "loss": 0.9168,
      "step": 1256
    },
    {
      "epoch": 2.8795422031473534,
      "grad_norm": 1.0334773063659668,
      "learning_rate": 0.00017073293354741127,
      "loss": 1.011,
      "step": 1257
    },
    {
      "epoch": 2.8818311874105866,
      "grad_norm": 1.1294324398040771,
      "learning_rate": 0.00017068841568303468,
      "loss": 0.9188,
      "step": 1258
    },
    {
      "epoch": 2.88412017167382,
      "grad_norm": 1.1465139389038086,
      "learning_rate": 0.00017064386980002863,
      "loss": 0.9653,
      "step": 1259
    },
    {
      "epoch": 2.886409155937053,
      "grad_norm": 1.200527548789978,
      "learning_rate": 0.0001705992959160497,
      "loss": 0.8836,
      "step": 1260
    },
    {
      "epoch": 2.8886981402002863,
      "grad_norm": 1.1738464832305908,
      "learning_rate": 0.00017055469404876554,
      "loss": 0.8924,
      "step": 1261
    },
    {
      "epoch": 2.8909871244635195,
      "grad_norm": 1.2068474292755127,
      "learning_rate": 0.00017051006421585488,
      "loss": 0.799,
      "step": 1262
    },
    {
      "epoch": 2.8932761087267522,
      "grad_norm": 1.1777546405792236,
      "learning_rate": 0.00017046540643500757,
      "loss": 0.6786,
      "step": 1263
    },
    {
      "epoch": 2.895565092989986,
      "grad_norm": 1.3901464939117432,
      "learning_rate": 0.00017042072072392456,
      "loss": 0.7717,
      "step": 1264
    },
    {
      "epoch": 2.8978540772532186,
      "grad_norm": 1.1867256164550781,
      "learning_rate": 0.00017037600710031783,
      "loss": 0.669,
      "step": 1265
    },
    {
      "epoch": 2.9001430615164523,
      "grad_norm": 1.3812681436538696,
      "learning_rate": 0.0001703312655819104,
      "loss": 0.6903,
      "step": 1266
    },
    {
      "epoch": 2.902432045779685,
      "grad_norm": 1.7685177326202393,
      "learning_rate": 0.00017028649618643638,
      "loss": 0.5462,
      "step": 1267
    },
    {
      "epoch": 2.9047210300429187,
      "grad_norm": 1.9904850721359253,
      "learning_rate": 0.00017024169893164098,
      "loss": 0.8015,
      "step": 1268
    },
    {
      "epoch": 2.9070100143061515,
      "grad_norm": 1.666465163230896,
      "learning_rate": 0.00017019687383528032,
      "loss": 0.6191,
      "step": 1269
    },
    {
      "epoch": 2.9092989985693847,
      "grad_norm": 1.923681616783142,
      "learning_rate": 0.0001701520209151217,
      "loss": 0.6328,
      "step": 1270
    },
    {
      "epoch": 2.911587982832618,
      "grad_norm": 2.1182587146759033,
      "learning_rate": 0.00017010714018894337,
      "loss": 0.4895,
      "step": 1271
    },
    {
      "epoch": 2.913876967095851,
      "grad_norm": 2.672013998031616,
      "learning_rate": 0.0001700622316745346,
      "loss": 0.6496,
      "step": 1272
    },
    {
      "epoch": 2.9161659513590843,
      "grad_norm": 2.9380340576171875,
      "learning_rate": 0.00017001729538969573,
      "loss": 1.2208,
      "step": 1273
    },
    {
      "epoch": 2.9184549356223175,
      "grad_norm": 0.4684350788593292,
      "learning_rate": 0.00016997233135223802,
      "loss": 1.8687,
      "step": 1274
    },
    {
      "epoch": 2.9207439198855507,
      "grad_norm": 0.5193700790405273,
      "learning_rate": 0.0001699273395799838,
      "loss": 1.7657,
      "step": 1275
    },
    {
      "epoch": 2.923032904148784,
      "grad_norm": 0.5797861814498901,
      "learning_rate": 0.00016988232009076634,
      "loss": 1.7559,
      "step": 1276
    },
    {
      "epoch": 2.925321888412017,
      "grad_norm": 0.6416584253311157,
      "learning_rate": 0.00016983727290242998,
      "loss": 1.8304,
      "step": 1277
    },
    {
      "epoch": 2.9276108726752503,
      "grad_norm": 0.6223170161247253,
      "learning_rate": 0.00016979219803282998,
      "loss": 1.6377,
      "step": 1278
    },
    {
      "epoch": 2.9298998569384835,
      "grad_norm": 0.6889511942863464,
      "learning_rate": 0.00016974709549983256,
      "loss": 1.8109,
      "step": 1279
    },
    {
      "epoch": 2.9321888412017167,
      "grad_norm": 0.6475473046302795,
      "learning_rate": 0.00016970196532131487,
      "loss": 1.4537,
      "step": 1280
    },
    {
      "epoch": 2.93447782546495,
      "grad_norm": 0.6533897519111633,
      "learning_rate": 0.00016965680751516519,
      "loss": 1.4356,
      "step": 1281
    },
    {
      "epoch": 2.936766809728183,
      "grad_norm": 0.6685423851013184,
      "learning_rate": 0.0001696116220992825,
      "loss": 1.4711,
      "step": 1282
    },
    {
      "epoch": 2.9390557939914164,
      "grad_norm": 0.7833460569381714,
      "learning_rate": 0.00016956640909157695,
      "loss": 1.6696,
      "step": 1283
    },
    {
      "epoch": 2.9413447782546496,
      "grad_norm": 0.712066113948822,
      "learning_rate": 0.0001695211685099695,
      "loss": 1.2808,
      "step": 1284
    },
    {
      "epoch": 2.943633762517883,
      "grad_norm": 0.7387949228286743,
      "learning_rate": 0.00016947590037239208,
      "loss": 1.5425,
      "step": 1285
    },
    {
      "epoch": 2.945922746781116,
      "grad_norm": 0.7197155356407166,
      "learning_rate": 0.00016943060469678753,
      "loss": 1.3838,
      "step": 1286
    },
    {
      "epoch": 2.948211731044349,
      "grad_norm": 0.7758193612098694,
      "learning_rate": 0.0001693852815011096,
      "loss": 1.4557,
      "step": 1287
    },
    {
      "epoch": 2.9505007153075824,
      "grad_norm": 0.7598605751991272,
      "learning_rate": 0.00016933993080332297,
      "loss": 1.2277,
      "step": 1288
    },
    {
      "epoch": 2.9527896995708156,
      "grad_norm": 0.8117440938949585,
      "learning_rate": 0.00016929455262140316,
      "loss": 1.4811,
      "step": 1289
    },
    {
      "epoch": 2.9550786838340484,
      "grad_norm": 0.7775629162788391,
      "learning_rate": 0.0001692491469733367,
      "loss": 1.1106,
      "step": 1290
    },
    {
      "epoch": 2.957367668097282,
      "grad_norm": 0.7877098321914673,
      "learning_rate": 0.0001692037138771209,
      "loss": 1.2132,
      "step": 1291
    },
    {
      "epoch": 2.959656652360515,
      "grad_norm": 0.8923436403274536,
      "learning_rate": 0.00016915825335076402,
      "loss": 1.2882,
      "step": 1292
    },
    {
      "epoch": 2.9619456366237484,
      "grad_norm": 0.8452033996582031,
      "learning_rate": 0.00016911276541228508,
      "loss": 1.1631,
      "step": 1293
    },
    {
      "epoch": 2.964234620886981,
      "grad_norm": 0.9006503224372864,
      "learning_rate": 0.00016906725007971417,
      "loss": 1.2336,
      "step": 1294
    },
    {
      "epoch": 2.966523605150215,
      "grad_norm": 0.9380308389663696,
      "learning_rate": 0.000169021707371092,
      "loss": 1.1738,
      "step": 1295
    },
    {
      "epoch": 2.9688125894134476,
      "grad_norm": 0.9270153045654297,
      "learning_rate": 0.00016897613730447033,
      "loss": 1.052,
      "step": 1296
    },
    {
      "epoch": 2.971101573676681,
      "grad_norm": 0.9935706853866577,
      "learning_rate": 0.0001689305398979116,
      "loss": 0.9949,
      "step": 1297
    },
    {
      "epoch": 2.973390557939914,
      "grad_norm": 1.0165584087371826,
      "learning_rate": 0.0001688849151694892,
      "loss": 1.0326,
      "step": 1298
    },
    {
      "epoch": 2.9756795422031472,
      "grad_norm": 1.0113788843154907,
      "learning_rate": 0.00016883926313728735,
      "loss": 0.9761,
      "step": 1299
    },
    {
      "epoch": 2.9779685264663804,
      "grad_norm": 1.0460736751556396,
      "learning_rate": 0.00016879358381940103,
      "loss": 0.9182,
      "step": 1300
    },
    {
      "epoch": 2.9802575107296136,
      "grad_norm": 1.3358728885650635,
      "learning_rate": 0.00016874787723393603,
      "loss": 0.9974,
      "step": 1301
    },
    {
      "epoch": 2.982546494992847,
      "grad_norm": 1.1663224697113037,
      "learning_rate": 0.00016870214339900902,
      "loss": 0.7741,
      "step": 1302
    },
    {
      "epoch": 2.98483547925608,
      "grad_norm": 1.225078821182251,
      "learning_rate": 0.00016865638233274743,
      "loss": 0.4967,
      "step": 1303
    },
    {
      "epoch": 2.9871244635193133,
      "grad_norm": 1.1760244369506836,
      "learning_rate": 0.00016861059405328944,
      "loss": 0.6543,
      "step": 1304
    },
    {
      "epoch": 2.9894134477825465,
      "grad_norm": 1.4124011993408203,
      "learning_rate": 0.00016856477857878413,
      "loss": 0.801,
      "step": 1305
    },
    {
      "epoch": 2.9917024320457797,
      "grad_norm": 1.5243903398513794,
      "learning_rate": 0.00016851893592739124,
      "loss": 0.696,
      "step": 1306
    },
    {
      "epoch": 2.993991416309013,
      "grad_norm": 1.8234519958496094,
      "learning_rate": 0.00016847306611728137,
      "loss": 0.749,
      "step": 1307
    },
    {
      "epoch": 2.996280400572246,
      "grad_norm": 1.7365727424621582,
      "learning_rate": 0.00016842716916663577,
      "loss": 0.5216,
      "step": 1308
    },
    {
      "epoch": 2.9985693848354793,
      "grad_norm": 4.100002765655518,
      "learning_rate": 0.0001683812450936466,
      "loss": 0.7455,
      "step": 1309
    },
    {
      "epoch": 3.002002861230329,
      "grad_norm": 3.1343367099761963,
      "learning_rate": 0.0001683352939165167,
      "loss": 1.7378,
      "step": 1310
    },
    {
      "epoch": 3.004291845493562,
      "grad_norm": 0.3502930700778961,
      "learning_rate": 0.0001682893156534596,
      "loss": 1.7408,
      "step": 1311
    },
    {
      "epoch": 3.0065808297567953,
      "grad_norm": 0.3897700011730194,
      "learning_rate": 0.00016824331032269963,
      "loss": 1.7803,
      "step": 1312
    },
    {
      "epoch": 3.0088698140200285,
      "grad_norm": 0.419120192527771,
      "learning_rate": 0.00016819727794247186,
      "loss": 1.6434,
      "step": 1313
    },
    {
      "epoch": 3.0111587982832617,
      "grad_norm": 0.47108617424964905,
      "learning_rate": 0.00016815121853102204,
      "loss": 1.6733,
      "step": 1314
    },
    {
      "epoch": 3.013447782546495,
      "grad_norm": 0.5061365365982056,
      "learning_rate": 0.00016810513210660662,
      "loss": 1.5977,
      "step": 1315
    },
    {
      "epoch": 3.015736766809728,
      "grad_norm": 0.5216453075408936,
      "learning_rate": 0.00016805901868749282,
      "loss": 1.5067,
      "step": 1316
    },
    {
      "epoch": 3.0180257510729613,
      "grad_norm": 0.5895505547523499,
      "learning_rate": 0.00016801287829195852,
      "loss": 1.631,
      "step": 1317
    },
    {
      "epoch": 3.0203147353361945,
      "grad_norm": 0.6296285390853882,
      "learning_rate": 0.0001679667109382923,
      "loss": 1.4671,
      "step": 1318
    },
    {
      "epoch": 3.0226037195994278,
      "grad_norm": 0.6245955228805542,
      "learning_rate": 0.0001679205166447934,
      "loss": 1.3516,
      "step": 1319
    },
    {
      "epoch": 3.024892703862661,
      "grad_norm": 0.6855911612510681,
      "learning_rate": 0.0001678742954297718,
      "loss": 1.3897,
      "step": 1320
    },
    {
      "epoch": 3.027181688125894,
      "grad_norm": 0.6471125483512878,
      "learning_rate": 0.0001678280473115481,
      "loss": 1.2822,
      "step": 1321
    },
    {
      "epoch": 3.0294706723891274,
      "grad_norm": 0.7257754802703857,
      "learning_rate": 0.00016778177230845353,
      "loss": 1.3251,
      "step": 1322
    },
    {
      "epoch": 3.0317596566523606,
      "grad_norm": 0.7416722774505615,
      "learning_rate": 0.00016773547043883008,
      "loss": 1.4536,
      "step": 1323
    },
    {
      "epoch": 3.034048640915594,
      "grad_norm": 0.7330904603004456,
      "learning_rate": 0.00016768914172103036,
      "loss": 1.2802,
      "step": 1324
    },
    {
      "epoch": 3.036337625178827,
      "grad_norm": 0.7336977124214172,
      "learning_rate": 0.0001676427861734175,
      "loss": 1.2513,
      "step": 1325
    },
    {
      "epoch": 3.03862660944206,
      "grad_norm": 0.7748559713363647,
      "learning_rate": 0.0001675964038143654,
      "loss": 1.1544,
      "step": 1326
    },
    {
      "epoch": 3.0409155937052934,
      "grad_norm": 0.7851191759109497,
      "learning_rate": 0.00016754999466225858,
      "loss": 1.1701,
      "step": 1327
    },
    {
      "epoch": 3.0432045779685266,
      "grad_norm": 0.8110692501068115,
      "learning_rate": 0.00016750355873549212,
      "loss": 1.1024,
      "step": 1328
    },
    {
      "epoch": 3.04549356223176,
      "grad_norm": 0.7968434691429138,
      "learning_rate": 0.00016745709605247173,
      "loss": 1.0253,
      "step": 1329
    },
    {
      "epoch": 3.047782546494993,
      "grad_norm": 0.8266029953956604,
      "learning_rate": 0.00016741060663161373,
      "loss": 1.0633,
      "step": 1330
    },
    {
      "epoch": 3.0500715307582262,
      "grad_norm": 0.8626382946968079,
      "learning_rate": 0.00016736409049134503,
      "loss": 0.9754,
      "step": 1331
    },
    {
      "epoch": 3.0523605150214594,
      "grad_norm": 0.8473793864250183,
      "learning_rate": 0.00016731754765010316,
      "loss": 1.0026,
      "step": 1332
    },
    {
      "epoch": 3.054649499284692,
      "grad_norm": 0.8965425491333008,
      "learning_rate": 0.00016727097812633624,
      "loss": 0.9487,
      "step": 1333
    },
    {
      "epoch": 3.0569384835479254,
      "grad_norm": 0.9592859745025635,
      "learning_rate": 0.00016722438193850293,
      "loss": 0.9187,
      "step": 1334
    },
    {
      "epoch": 3.0592274678111586,
      "grad_norm": 1.032496452331543,
      "learning_rate": 0.0001671777591050724,
      "loss": 0.7986,
      "step": 1335
    },
    {
      "epoch": 3.061516452074392,
      "grad_norm": 0.9474940299987793,
      "learning_rate": 0.00016713110964452455,
      "loss": 0.7529,
      "step": 1336
    },
    {
      "epoch": 3.063805436337625,
      "grad_norm": 1.010635256767273,
      "learning_rate": 0.00016708443357534968,
      "loss": 0.7473,
      "step": 1337
    },
    {
      "epoch": 3.0660944206008582,
      "grad_norm": 0.9460374116897583,
      "learning_rate": 0.00016703773091604872,
      "loss": 0.7117,
      "step": 1338
    },
    {
      "epoch": 3.0683834048640914,
      "grad_norm": 1.0257648229599,
      "learning_rate": 0.0001669910016851331,
      "loss": 0.6653,
      "step": 1339
    },
    {
      "epoch": 3.0706723891273247,
      "grad_norm": 1.151338815689087,
      "learning_rate": 0.00016694424590112477,
      "loss": 0.7707,
      "step": 1340
    },
    {
      "epoch": 3.072961373390558,
      "grad_norm": 1.1915266513824463,
      "learning_rate": 0.0001668974635825563,
      "loss": 0.7319,
      "step": 1341
    },
    {
      "epoch": 3.075250357653791,
      "grad_norm": 1.9450591802597046,
      "learning_rate": 0.0001668506547479706,
      "loss": 0.5746,
      "step": 1342
    },
    {
      "epoch": 3.0775393419170243,
      "grad_norm": 1.206840991973877,
      "learning_rate": 0.00016680381941592127,
      "loss": 0.5839,
      "step": 1343
    },
    {
      "epoch": 3.0798283261802575,
      "grad_norm": 1.155497670173645,
      "learning_rate": 0.00016675695760497234,
      "loss": 0.449,
      "step": 1344
    },
    {
      "epoch": 3.0821173104434907,
      "grad_norm": 1.4850190877914429,
      "learning_rate": 0.0001667100693336983,
      "loss": 0.4633,
      "step": 1345
    },
    {
      "epoch": 3.084406294706724,
      "grad_norm": 1.267003059387207,
      "learning_rate": 0.0001666631546206842,
      "loss": 0.4105,
      "step": 1346
    },
    {
      "epoch": 3.086695278969957,
      "grad_norm": 1.6786085367202759,
      "learning_rate": 0.0001666162134845255,
      "loss": 0.4886,
      "step": 1347
    },
    {
      "epoch": 3.0889842632331903,
      "grad_norm": 1.5634887218475342,
      "learning_rate": 0.0001665692459438282,
      "loss": 0.392,
      "step": 1348
    },
    {
      "epoch": 3.0912732474964235,
      "grad_norm": 1.584593415260315,
      "learning_rate": 0.00016652225201720867,
      "loss": 0.3327,
      "step": 1349
    },
    {
      "epoch": 3.0935622317596567,
      "grad_norm": 1.9362475872039795,
      "learning_rate": 0.0001664752317232939,
      "loss": 0.3755,
      "step": 1350
    },
    {
      "epoch": 3.09585121602289,
      "grad_norm": 1.6230665445327759,
      "learning_rate": 0.00016642818508072113,
      "loss": 0.3716,
      "step": 1351
    },
    {
      "epoch": 3.098140200286123,
      "grad_norm": 1.6583707332611084,
      "learning_rate": 0.00016638111210813823,
      "loss": 0.2806,
      "step": 1352
    },
    {
      "epoch": 3.1004291845493563,
      "grad_norm": 1.4343819618225098,
      "learning_rate": 0.00016633401282420337,
      "loss": 0.2608,
      "step": 1353
    },
    {
      "epoch": 3.1027181688125895,
      "grad_norm": 2.14277720451355,
      "learning_rate": 0.00016628688724758522,
      "loss": 0.4055,
      "step": 1354
    },
    {
      "epoch": 3.1050071530758228,
      "grad_norm": 2.157909393310547,
      "learning_rate": 0.00016623973539696283,
      "loss": 0.3326,
      "step": 1355
    },
    {
      "epoch": 3.107296137339056,
      "grad_norm": 1.855731725692749,
      "learning_rate": 0.00016619255729102573,
      "loss": 0.3196,
      "step": 1356
    },
    {
      "epoch": 3.109585121602289,
      "grad_norm": 2.3847789764404297,
      "learning_rate": 0.00016614535294847375,
      "loss": 0.2748,
      "step": 1357
    },
    {
      "epoch": 3.1118741058655224,
      "grad_norm": 2.631718635559082,
      "learning_rate": 0.00016609812238801725,
      "loss": 0.2233,
      "step": 1358
    },
    {
      "epoch": 3.1141630901287556,
      "grad_norm": 2.3659348487854004,
      "learning_rate": 0.00016605086562837692,
      "loss": 0.1479,
      "step": 1359
    },
    {
      "epoch": 3.1164520743919883,
      "grad_norm": 1.285439133644104,
      "learning_rate": 0.00016600358268828376,
      "loss": 1.708,
      "step": 1360
    },
    {
      "epoch": 3.1187410586552216,
      "grad_norm": 0.6165206432342529,
      "learning_rate": 0.00016595627358647924,
      "loss": 1.821,
      "step": 1361
    },
    {
      "epoch": 3.1210300429184548,
      "grad_norm": 0.654448926448822,
      "learning_rate": 0.0001659089383417152,
      "loss": 1.6915,
      "step": 1362
    },
    {
      "epoch": 3.123319027181688,
      "grad_norm": 0.6725746393203735,
      "learning_rate": 0.00016586157697275384,
      "loss": 1.6973,
      "step": 1363
    },
    {
      "epoch": 3.125608011444921,
      "grad_norm": 0.6777236461639404,
      "learning_rate": 0.00016581418949836765,
      "loss": 1.5506,
      "step": 1364
    },
    {
      "epoch": 3.1278969957081544,
      "grad_norm": 0.6863792538642883,
      "learning_rate": 0.00016576677593733953,
      "loss": 1.6642,
      "step": 1365
    },
    {
      "epoch": 3.1301859799713876,
      "grad_norm": 0.6477636098861694,
      "learning_rate": 0.00016571933630846268,
      "loss": 1.5271,
      "step": 1366
    },
    {
      "epoch": 3.132474964234621,
      "grad_norm": 0.6331400275230408,
      "learning_rate": 0.00016567187063054066,
      "loss": 1.5405,
      "step": 1367
    },
    {
      "epoch": 3.134763948497854,
      "grad_norm": 0.6536558866500854,
      "learning_rate": 0.00016562437892238738,
      "loss": 1.5363,
      "step": 1368
    },
    {
      "epoch": 3.137052932761087,
      "grad_norm": 0.684052050113678,
      "learning_rate": 0.000165576861202827,
      "loss": 1.4938,
      "step": 1369
    },
    {
      "epoch": 3.1393419170243204,
      "grad_norm": 0.6870331764221191,
      "learning_rate": 0.00016552931749069403,
      "loss": 1.1879,
      "step": 1370
    },
    {
      "epoch": 3.1416309012875536,
      "grad_norm": 0.6792276501655579,
      "learning_rate": 0.00016548174780483328,
      "loss": 1.3665,
      "step": 1371
    },
    {
      "epoch": 3.143919885550787,
      "grad_norm": 0.697238564491272,
      "learning_rate": 0.00016543415216409987,
      "loss": 1.3141,
      "step": 1372
    },
    {
      "epoch": 3.14620886981402,
      "grad_norm": 0.6724427342414856,
      "learning_rate": 0.0001653865305873592,
      "loss": 1.3512,
      "step": 1373
    },
    {
      "epoch": 3.1484978540772532,
      "grad_norm": 0.7242927551269531,
      "learning_rate": 0.0001653388830934869,
      "loss": 1.2451,
      "step": 1374
    },
    {
      "epoch": 3.1507868383404865,
      "grad_norm": 0.7219815254211426,
      "learning_rate": 0.00016529120970136893,
      "loss": 1.1888,
      "step": 1375
    },
    {
      "epoch": 3.1530758226037197,
      "grad_norm": 0.728657066822052,
      "learning_rate": 0.00016524351042990153,
      "loss": 1.2222,
      "step": 1376
    },
    {
      "epoch": 3.155364806866953,
      "grad_norm": 0.7278046011924744,
      "learning_rate": 0.00016519578529799114,
      "loss": 1.1217,
      "step": 1377
    },
    {
      "epoch": 3.157653791130186,
      "grad_norm": 0.801794171333313,
      "learning_rate": 0.00016514803432455448,
      "loss": 1.2055,
      "step": 1378
    },
    {
      "epoch": 3.1599427753934193,
      "grad_norm": 0.7983406782150269,
      "learning_rate": 0.0001651002575285185,
      "loss": 1.2447,
      "step": 1379
    },
    {
      "epoch": 3.1622317596566525,
      "grad_norm": 0.8168611526489258,
      "learning_rate": 0.00016505245492882042,
      "loss": 1.0192,
      "step": 1380
    },
    {
      "epoch": 3.1645207439198857,
      "grad_norm": 0.8270989060401917,
      "learning_rate": 0.00016500462654440762,
      "loss": 1.1999,
      "step": 1381
    },
    {
      "epoch": 3.166809728183119,
      "grad_norm": 0.9000014066696167,
      "learning_rate": 0.0001649567723942378,
      "loss": 1.177,
      "step": 1382
    },
    {
      "epoch": 3.169098712446352,
      "grad_norm": 0.8611709475517273,
      "learning_rate": 0.0001649088924972788,
      "loss": 0.9349,
      "step": 1383
    },
    {
      "epoch": 3.1713876967095853,
      "grad_norm": 0.8958805203437805,
      "learning_rate": 0.00016486098687250864,
      "loss": 0.9474,
      "step": 1384
    },
    {
      "epoch": 3.173676680972818,
      "grad_norm": 0.9219009280204773,
      "learning_rate": 0.00016481305553891563,
      "loss": 0.8948,
      "step": 1385
    },
    {
      "epoch": 3.1759656652360517,
      "grad_norm": 0.9911636114120483,
      "learning_rate": 0.00016476509851549818,
      "loss": 0.8673,
      "step": 1386
    },
    {
      "epoch": 3.1782546494992845,
      "grad_norm": 0.9257792234420776,
      "learning_rate": 0.00016471711582126495,
      "loss": 0.8542,
      "step": 1387
    },
    {
      "epoch": 3.1805436337625177,
      "grad_norm": 0.9667647480964661,
      "learning_rate": 0.00016466910747523476,
      "loss": 0.7311,
      "step": 1388
    },
    {
      "epoch": 3.182832618025751,
      "grad_norm": 1.0631122589111328,
      "learning_rate": 0.00016462107349643655,
      "loss": 0.7766,
      "step": 1389
    },
    {
      "epoch": 3.185121602288984,
      "grad_norm": 1.1407530307769775,
      "learning_rate": 0.00016457301390390943,
      "loss": 0.912,
      "step": 1390
    },
    {
      "epoch": 3.1874105865522173,
      "grad_norm": 1.1690483093261719,
      "learning_rate": 0.00016452492871670278,
      "loss": 0.7225,
      "step": 1391
    },
    {
      "epoch": 3.1896995708154505,
      "grad_norm": 1.168144941329956,
      "learning_rate": 0.00016447681795387593,
      "loss": 0.7633,
      "step": 1392
    },
    {
      "epoch": 3.1919885550786837,
      "grad_norm": 1.1078940629959106,
      "learning_rate": 0.0001644286816344985,
      "loss": 0.5566,
      "step": 1393
    },
    {
      "epoch": 3.194277539341917,
      "grad_norm": 1.2715566158294678,
      "learning_rate": 0.00016438051977765022,
      "loss": 0.5951,
      "step": 1394
    },
    {
      "epoch": 3.19656652360515,
      "grad_norm": 1.2896710634231567,
      "learning_rate": 0.00016433233240242085,
      "loss": 0.5042,
      "step": 1395
    },
    {
      "epoch": 3.1988555078683834,
      "grad_norm": 1.5294461250305176,
      "learning_rate": 0.00016428411952791033,
      "loss": 0.579,
      "step": 1396
    },
    {
      "epoch": 3.2011444921316166,
      "grad_norm": 1.5561789274215698,
      "learning_rate": 0.00016423588117322877,
      "loss": 0.5071,
      "step": 1397
    },
    {
      "epoch": 3.2034334763948498,
      "grad_norm": 1.4642733335494995,
      "learning_rate": 0.00016418761735749623,
      "loss": 0.4201,
      "step": 1398
    },
    {
      "epoch": 3.205722460658083,
      "grad_norm": 1.7246230840682983,
      "learning_rate": 0.000164139328099843,
      "loss": 0.4069,
      "step": 1399
    },
    {
      "epoch": 3.208011444921316,
      "grad_norm": 1.4964900016784668,
      "learning_rate": 0.00016409101341940938,
      "loss": 0.3283,
      "step": 1400
    },
    {
      "epoch": 3.2103004291845494,
      "grad_norm": 1.59528386592865,
      "learning_rate": 0.00016404267333534577,
      "loss": 0.3679,
      "step": 1401
    },
    {
      "epoch": 3.2125894134477826,
      "grad_norm": 1.2055912017822266,
      "learning_rate": 0.0001639943078668126,
      "loss": 0.2304,
      "step": 1402
    },
    {
      "epoch": 3.214878397711016,
      "grad_norm": 1.648876428604126,
      "learning_rate": 0.00016394591703298048,
      "loss": 0.4268,
      "step": 1403
    },
    {
      "epoch": 3.217167381974249,
      "grad_norm": 1.8570979833602905,
      "learning_rate": 0.0001638975008530299,
      "loss": 0.3473,
      "step": 1404
    },
    {
      "epoch": 3.219456366237482,
      "grad_norm": 2.2466492652893066,
      "learning_rate": 0.0001638490593461515,
      "loss": 0.306,
      "step": 1405
    },
    {
      "epoch": 3.2217453505007154,
      "grad_norm": 1.954963207244873,
      "learning_rate": 0.00016380059253154602,
      "loss": 0.2592,
      "step": 1406
    },
    {
      "epoch": 3.2240343347639486,
      "grad_norm": 2.2373297214508057,
      "learning_rate": 0.0001637521004284241,
      "loss": 0.2749,
      "step": 1407
    },
    {
      "epoch": 3.226323319027182,
      "grad_norm": 2.0633468627929688,
      "learning_rate": 0.00016370358305600646,
      "loss": 0.1285,
      "step": 1408
    },
    {
      "epoch": 3.228612303290415,
      "grad_norm": 4.076617240905762,
      "learning_rate": 0.00016365504043352384,
      "loss": 0.3565,
      "step": 1409
    },
    {
      "epoch": 3.2309012875536482,
      "grad_norm": 0.5410581231117249,
      "learning_rate": 0.00016360647258021696,
      "loss": 1.6154,
      "step": 1410
    },
    {
      "epoch": 3.2331902718168815,
      "grad_norm": 0.5588700771331787,
      "learning_rate": 0.00016355787951533664,
      "loss": 1.7761,
      "step": 1411
    },
    {
      "epoch": 3.235479256080114,
      "grad_norm": 0.597781240940094,
      "learning_rate": 0.00016350926125814353,
      "loss": 1.6697,
      "step": 1412
    },
    {
      "epoch": 3.237768240343348,
      "grad_norm": 0.6258123517036438,
      "learning_rate": 0.00016346061782790842,
      "loss": 1.7055,
      "step": 1413
    },
    {
      "epoch": 3.2400572246065806,
      "grad_norm": 0.6659576892852783,
      "learning_rate": 0.000163411949243912,
      "loss": 1.6098,
      "step": 1414
    },
    {
      "epoch": 3.242346208869814,
      "grad_norm": 0.6856834888458252,
      "learning_rate": 0.00016336325552544488,
      "loss": 1.5784,
      "step": 1415
    },
    {
      "epoch": 3.244635193133047,
      "grad_norm": 0.7154250144958496,
      "learning_rate": 0.0001633145366918078,
      "loss": 1.6037,
      "step": 1416
    },
    {
      "epoch": 3.2469241773962803,
      "grad_norm": 0.6752023100852966,
      "learning_rate": 0.00016326579276231123,
      "loss": 1.3723,
      "step": 1417
    },
    {
      "epoch": 3.2492131616595135,
      "grad_norm": 0.7235175967216492,
      "learning_rate": 0.00016321702375627584,
      "loss": 1.5084,
      "step": 1418
    },
    {
      "epoch": 3.2515021459227467,
      "grad_norm": 0.698215663433075,
      "learning_rate": 0.000163168229693032,
      "loss": 1.246,
      "step": 1419
    },
    {
      "epoch": 3.25379113018598,
      "grad_norm": 0.7117575407028198,
      "learning_rate": 0.00016311941059192012,
      "loss": 1.4004,
      "step": 1420
    },
    {
      "epoch": 3.256080114449213,
      "grad_norm": 0.7212780714035034,
      "learning_rate": 0.0001630705664722906,
      "loss": 1.2367,
      "step": 1421
    },
    {
      "epoch": 3.2583690987124463,
      "grad_norm": 0.7310657501220703,
      "learning_rate": 0.00016302169735350365,
      "loss": 1.4142,
      "step": 1422
    },
    {
      "epoch": 3.2606580829756795,
      "grad_norm": 0.7708021998405457,
      "learning_rate": 0.00016297280325492945,
      "loss": 1.3222,
      "step": 1423
    },
    {
      "epoch": 3.2629470672389127,
      "grad_norm": 0.7900006771087646,
      "learning_rate": 0.000162923884195948,
      "loss": 1.2299,
      "step": 1424
    },
    {
      "epoch": 3.265236051502146,
      "grad_norm": 0.770846426486969,
      "learning_rate": 0.00016287494019594928,
      "loss": 1.2129,
      "step": 1425
    },
    {
      "epoch": 3.267525035765379,
      "grad_norm": 0.7942619919776917,
      "learning_rate": 0.00016282597127433317,
      "loss": 1.3292,
      "step": 1426
    },
    {
      "epoch": 3.2698140200286123,
      "grad_norm": 0.7913793325424194,
      "learning_rate": 0.00016277697745050934,
      "loss": 1.0872,
      "step": 1427
    },
    {
      "epoch": 3.2721030042918455,
      "grad_norm": 0.8320474624633789,
      "learning_rate": 0.0001627279587438974,
      "loss": 1.1262,
      "step": 1428
    },
    {
      "epoch": 3.2743919885550787,
      "grad_norm": 0.8214536309242249,
      "learning_rate": 0.0001626789151739268,
      "loss": 1.1206,
      "step": 1429
    },
    {
      "epoch": 3.276680972818312,
      "grad_norm": 0.8584417104721069,
      "learning_rate": 0.00016262984676003678,
      "loss": 1.1037,
      "step": 1430
    },
    {
      "epoch": 3.278969957081545,
      "grad_norm": 0.8927283883094788,
      "learning_rate": 0.0001625807535216766,
      "loss": 1.006,
      "step": 1431
    },
    {
      "epoch": 3.2812589413447784,
      "grad_norm": 0.9014906883239746,
      "learning_rate": 0.00016253163547830517,
      "loss": 1.1194,
      "step": 1432
    },
    {
      "epoch": 3.2835479256080116,
      "grad_norm": 0.9218708872795105,
      "learning_rate": 0.00016248249264939134,
      "loss": 0.9428,
      "step": 1433
    },
    {
      "epoch": 3.2858369098712448,
      "grad_norm": 0.952469527721405,
      "learning_rate": 0.00016243332505441375,
      "loss": 0.9644,
      "step": 1434
    },
    {
      "epoch": 3.288125894134478,
      "grad_norm": 0.9540469646453857,
      "learning_rate": 0.0001623841327128609,
      "loss": 1.0011,
      "step": 1435
    },
    {
      "epoch": 3.290414878397711,
      "grad_norm": 0.9375476837158203,
      "learning_rate": 0.000162334915644231,
      "loss": 0.8595,
      "step": 1436
    },
    {
      "epoch": 3.2927038626609444,
      "grad_norm": 1.1172115802764893,
      "learning_rate": 0.0001622856738680322,
      "loss": 0.7108,
      "step": 1437
    },
    {
      "epoch": 3.2949928469241776,
      "grad_norm": 1.0877395868301392,
      "learning_rate": 0.00016223640740378227,
      "loss": 0.9256,
      "step": 1438
    },
    {
      "epoch": 3.2972818311874104,
      "grad_norm": 1.0282500982284546,
      "learning_rate": 0.00016218711627100896,
      "loss": 0.7162,
      "step": 1439
    },
    {
      "epoch": 3.299570815450644,
      "grad_norm": 1.1520159244537354,
      "learning_rate": 0.00016213780048924962,
      "loss": 0.7184,
      "step": 1440
    },
    {
      "epoch": 3.3018597997138768,
      "grad_norm": 1.2279778718948364,
      "learning_rate": 0.00016208846007805149,
      "loss": 0.6585,
      "step": 1441
    },
    {
      "epoch": 3.30414878397711,
      "grad_norm": 1.2272692918777466,
      "learning_rate": 0.00016203909505697154,
      "loss": 0.64,
      "step": 1442
    },
    {
      "epoch": 3.306437768240343,
      "grad_norm": 1.3210211992263794,
      "learning_rate": 0.0001619897054455765,
      "loss": 0.5938,
      "step": 1443
    },
    {
      "epoch": 3.3087267525035764,
      "grad_norm": 1.2005237340927124,
      "learning_rate": 0.00016194029126344276,
      "loss": 0.4121,
      "step": 1444
    },
    {
      "epoch": 3.3110157367668096,
      "grad_norm": 1.2610912322998047,
      "learning_rate": 0.00016189085253015656,
      "loss": 0.4052,
      "step": 1445
    },
    {
      "epoch": 3.313304721030043,
      "grad_norm": 1.3167498111724854,
      "learning_rate": 0.00016184138926531388,
      "loss": 0.4806,
      "step": 1446
    },
    {
      "epoch": 3.315593705293276,
      "grad_norm": 1.4835947751998901,
      "learning_rate": 0.00016179190148852039,
      "loss": 0.49,
      "step": 1447
    },
    {
      "epoch": 3.3178826895565092,
      "grad_norm": 1.5051236152648926,
      "learning_rate": 0.00016174238921939135,
      "loss": 0.4697,
      "step": 1448
    },
    {
      "epoch": 3.3201716738197424,
      "grad_norm": 1.2012327909469604,
      "learning_rate": 0.00016169285247755194,
      "loss": 0.3358,
      "step": 1449
    },
    {
      "epoch": 3.3224606580829756,
      "grad_norm": 1.6092313528060913,
      "learning_rate": 0.0001616432912826369,
      "loss": 0.3421,
      "step": 1450
    },
    {
      "epoch": 3.324749642346209,
      "grad_norm": 1.6912235021591187,
      "learning_rate": 0.0001615937056542907,
      "loss": 0.3551,
      "step": 1451
    },
    {
      "epoch": 3.327038626609442,
      "grad_norm": 1.7515792846679688,
      "learning_rate": 0.00016154409561216758,
      "loss": 0.3097,
      "step": 1452
    },
    {
      "epoch": 3.3293276108726753,
      "grad_norm": 1.6860718727111816,
      "learning_rate": 0.00016149446117593122,
      "loss": 0.2895,
      "step": 1453
    },
    {
      "epoch": 3.3316165951359085,
      "grad_norm": 2.208627700805664,
      "learning_rate": 0.00016144480236525527,
      "loss": 0.3486,
      "step": 1454
    },
    {
      "epoch": 3.3339055793991417,
      "grad_norm": 1.5682424306869507,
      "learning_rate": 0.00016139511919982283,
      "loss": 0.2364,
      "step": 1455
    },
    {
      "epoch": 3.336194563662375,
      "grad_norm": 2.0854272842407227,
      "learning_rate": 0.0001613454116993267,
      "loss": 0.2979,
      "step": 1456
    },
    {
      "epoch": 3.338483547925608,
      "grad_norm": 1.9732942581176758,
      "learning_rate": 0.0001612956798834694,
      "loss": 0.1769,
      "step": 1457
    },
    {
      "epoch": 3.3407725321888413,
      "grad_norm": 2.396601438522339,
      "learning_rate": 0.00016124592377196302,
      "loss": 0.1852,
      "step": 1458
    },
    {
      "epoch": 3.3430615164520745,
      "grad_norm": 4.118147373199463,
      "learning_rate": 0.00016119614338452927,
      "loss": 0.509,
      "step": 1459
    },
    {
      "epoch": 3.3453505007153077,
      "grad_norm": 0.6210325956344604,
      "learning_rate": 0.00016114633874089954,
      "loss": 1.661,
      "step": 1460
    },
    {
      "epoch": 3.347639484978541,
      "grad_norm": 0.655790388584137,
      "learning_rate": 0.00016109650986081473,
      "loss": 1.87,
      "step": 1461
    },
    {
      "epoch": 3.349928469241774,
      "grad_norm": 0.6583636403083801,
      "learning_rate": 0.00016104665676402552,
      "loss": 1.8289,
      "step": 1462
    },
    {
      "epoch": 3.3522174535050073,
      "grad_norm": 0.7265445590019226,
      "learning_rate": 0.00016099677947029197,
      "loss": 1.6803,
      "step": 1463
    },
    {
      "epoch": 3.3545064377682405,
      "grad_norm": 0.706066906452179,
      "learning_rate": 0.00016094687799938392,
      "loss": 1.6104,
      "step": 1464
    },
    {
      "epoch": 3.3567954220314737,
      "grad_norm": 0.7544640302658081,
      "learning_rate": 0.00016089695237108068,
      "loss": 1.5576,
      "step": 1465
    },
    {
      "epoch": 3.3590844062947065,
      "grad_norm": 0.7512106895446777,
      "learning_rate": 0.0001608470026051712,
      "loss": 1.3705,
      "step": 1466
    },
    {
      "epoch": 3.36137339055794,
      "grad_norm": 0.7720670700073242,
      "learning_rate": 0.00016079702872145402,
      "loss": 1.4436,
      "step": 1467
    },
    {
      "epoch": 3.363662374821173,
      "grad_norm": 0.817127525806427,
      "learning_rate": 0.0001607470307397371,
      "loss": 1.4738,
      "step": 1468
    },
    {
      "epoch": 3.365951359084406,
      "grad_norm": 0.7341871857643127,
      "learning_rate": 0.00016069700867983805,
      "loss": 1.4624,
      "step": 1469
    },
    {
      "epoch": 3.3682403433476393,
      "grad_norm": 0.8181737065315247,
      "learning_rate": 0.00016064696256158406,
      "loss": 1.4542,
      "step": 1470
    },
    {
      "epoch": 3.3705293276108725,
      "grad_norm": 0.7523680329322815,
      "learning_rate": 0.0001605968924048118,
      "loss": 1.1351,
      "step": 1471
    },
    {
      "epoch": 3.3728183118741057,
      "grad_norm": 0.8069499731063843,
      "learning_rate": 0.0001605467982293675,
      "loss": 1.3879,
      "step": 1472
    },
    {
      "epoch": 3.375107296137339,
      "grad_norm": 0.7637686133384705,
      "learning_rate": 0.00016049668005510686,
      "loss": 1.2487,
      "step": 1473
    },
    {
      "epoch": 3.377396280400572,
      "grad_norm": 0.7612592577934265,
      "learning_rate": 0.00016044653790189513,
      "loss": 1.2787,
      "step": 1474
    },
    {
      "epoch": 3.3796852646638054,
      "grad_norm": 0.771624743938446,
      "learning_rate": 0.00016039637178960704,
      "loss": 1.245,
      "step": 1475
    },
    {
      "epoch": 3.3819742489270386,
      "grad_norm": 0.8291980624198914,
      "learning_rate": 0.00016034618173812687,
      "loss": 1.3004,
      "step": 1476
    },
    {
      "epoch": 3.384263233190272,
      "grad_norm": 0.8648401498794556,
      "learning_rate": 0.00016029596776734832,
      "loss": 1.2606,
      "step": 1477
    },
    {
      "epoch": 3.386552217453505,
      "grad_norm": 0.8369148969650269,
      "learning_rate": 0.00016024572989717462,
      "loss": 1.2021,
      "step": 1478
    },
    {
      "epoch": 3.388841201716738,
      "grad_norm": 0.8293965458869934,
      "learning_rate": 0.0001601954681475185,
      "loss": 1.113,
      "step": 1479
    },
    {
      "epoch": 3.3911301859799714,
      "grad_norm": 0.7760304808616638,
      "learning_rate": 0.000160145182538302,
      "loss": 0.8969,
      "step": 1480
    },
    {
      "epoch": 3.3934191702432046,
      "grad_norm": 0.894138514995575,
      "learning_rate": 0.00016009487308945685,
      "loss": 1.193,
      "step": 1481
    },
    {
      "epoch": 3.395708154506438,
      "grad_norm": 0.887180745601654,
      "learning_rate": 0.00016004453982092405,
      "loss": 1.0186,
      "step": 1482
    },
    {
      "epoch": 3.397997138769671,
      "grad_norm": 0.9309510588645935,
      "learning_rate": 0.00015999418275265407,
      "loss": 1.0727,
      "step": 1483
    },
    {
      "epoch": 3.4002861230329042,
      "grad_norm": 0.8975167274475098,
      "learning_rate": 0.00015994380190460688,
      "loss": 0.8729,
      "step": 1484
    },
    {
      "epoch": 3.4025751072961374,
      "grad_norm": 0.9632381796836853,
      "learning_rate": 0.00015989339729675188,
      "loss": 0.9396,
      "step": 1485
    },
    {
      "epoch": 3.4048640915593706,
      "grad_norm": 1.0273927450180054,
      "learning_rate": 0.00015984296894906778,
      "loss": 0.91,
      "step": 1486
    },
    {
      "epoch": 3.407153075822604,
      "grad_norm": 1.0678201913833618,
      "learning_rate": 0.0001597925168815428,
      "loss": 0.756,
      "step": 1487
    },
    {
      "epoch": 3.409442060085837,
      "grad_norm": 1.0562642812728882,
      "learning_rate": 0.00015974204111417447,
      "loss": 0.6905,
      "step": 1488
    },
    {
      "epoch": 3.4117310443490703,
      "grad_norm": 1.1230536699295044,
      "learning_rate": 0.00015969154166696985,
      "loss": 0.7499,
      "step": 1489
    },
    {
      "epoch": 3.4140200286123035,
      "grad_norm": 1.1529197692871094,
      "learning_rate": 0.00015964101855994527,
      "loss": 0.6784,
      "step": 1490
    },
    {
      "epoch": 3.4163090128755362,
      "grad_norm": 1.1707793474197388,
      "learning_rate": 0.00015959047181312647,
      "loss": 0.6062,
      "step": 1491
    },
    {
      "epoch": 3.41859799713877,
      "grad_norm": 1.1724509000778198,
      "learning_rate": 0.00015953990144654858,
      "loss": 0.5441,
      "step": 1492
    },
    {
      "epoch": 3.4208869814020026,
      "grad_norm": 1.247757077217102,
      "learning_rate": 0.0001594893074802561,
      "loss": 0.561,
      "step": 1493
    },
    {
      "epoch": 3.4231759656652363,
      "grad_norm": 1.2895193099975586,
      "learning_rate": 0.00015943868993430278,
      "loss": 0.4022,
      "step": 1494
    },
    {
      "epoch": 3.425464949928469,
      "grad_norm": 1.2842702865600586,
      "learning_rate": 0.00015938804882875189,
      "loss": 0.4317,
      "step": 1495
    },
    {
      "epoch": 3.4277539341917023,
      "grad_norm": 1.4491361379623413,
      "learning_rate": 0.00015933738418367593,
      "loss": 0.4495,
      "step": 1496
    },
    {
      "epoch": 3.4300429184549355,
      "grad_norm": 1.4111093282699585,
      "learning_rate": 0.00015928669601915666,
      "loss": 0.4913,
      "step": 1497
    },
    {
      "epoch": 3.4323319027181687,
      "grad_norm": 1.2714898586273193,
      "learning_rate": 0.00015923598435528536,
      "loss": 0.2881,
      "step": 1498
    },
    {
      "epoch": 3.434620886981402,
      "grad_norm": 1.4608423709869385,
      "learning_rate": 0.00015918524921216246,
      "loss": 0.4037,
      "step": 1499
    },
    {
      "epoch": 3.436909871244635,
      "grad_norm": 1.6259890794754028,
      "learning_rate": 0.00015913449060989774,
      "loss": 0.303,
      "step": 1500
    },
    {
      "epoch": 3.436909871244635,
      "eval_loss": 1.9215168952941895,
      "eval_runtime": 25.1379,
      "eval_samples_per_second": 29.278,
      "eval_steps_per_second": 14.639,
      "step": 1500
    },
    {
      "epoch": 3.4391988555078683,
      "grad_norm": 1.7406142950057983,
      "learning_rate": 0.00015908370856861033,
      "loss": 0.3547,
      "step": 1501
    },
    {
      "epoch": 3.4414878397711015,
      "grad_norm": 1.3736141920089722,
      "learning_rate": 0.00015903290310842857,
      "loss": 0.293,
      "step": 1502
    },
    {
      "epoch": 3.4437768240343347,
      "grad_norm": 1.7873742580413818,
      "learning_rate": 0.00015898207424949013,
      "loss": 0.3333,
      "step": 1503
    },
    {
      "epoch": 3.446065808297568,
      "grad_norm": 1.976028323173523,
      "learning_rate": 0.00015893122201194197,
      "loss": 0.3112,
      "step": 1504
    },
    {
      "epoch": 3.448354792560801,
      "grad_norm": 1.6089595556259155,
      "learning_rate": 0.00015888034641594026,
      "loss": 0.197,
      "step": 1505
    },
    {
      "epoch": 3.4506437768240343,
      "grad_norm": 2.2680089473724365,
      "learning_rate": 0.00015882944748165048,
      "loss": 0.324,
      "step": 1506
    },
    {
      "epoch": 3.4529327610872675,
      "grad_norm": 1.6552183628082275,
      "learning_rate": 0.00015877852522924732,
      "loss": 0.1868,
      "step": 1507
    },
    {
      "epoch": 3.4552217453505008,
      "grad_norm": 2.8168227672576904,
      "learning_rate": 0.00015872757967891476,
      "loss": 0.2664,
      "step": 1508
    },
    {
      "epoch": 3.457510729613734,
      "grad_norm": 2.649996042251587,
      "learning_rate": 0.00015867661085084596,
      "loss": 0.2186,
      "step": 1509
    },
    {
      "epoch": 3.459799713876967,
      "grad_norm": 1.109015703201294,
      "learning_rate": 0.00015862561876524338,
      "loss": 1.6589,
      "step": 1510
    },
    {
      "epoch": 3.4620886981402004,
      "grad_norm": 0.5797591805458069,
      "learning_rate": 0.00015857460344231863,
      "loss": 1.8554,
      "step": 1511
    },
    {
      "epoch": 3.4643776824034336,
      "grad_norm": 0.6286052465438843,
      "learning_rate": 0.00015852356490229252,
      "loss": 1.7645,
      "step": 1512
    },
    {
      "epoch": 3.466666666666667,
      "grad_norm": 0.7255878448486328,
      "learning_rate": 0.00015847250316539514,
      "loss": 1.701,
      "step": 1513
    },
    {
      "epoch": 3.4689556509299,
      "grad_norm": 0.7296729683876038,
      "learning_rate": 0.0001584214182518657,
      "loss": 1.5291,
      "step": 1514
    },
    {
      "epoch": 3.471244635193133,
      "grad_norm": 0.7586151361465454,
      "learning_rate": 0.00015837031018195268,
      "loss": 1.6909,
      "step": 1515
    },
    {
      "epoch": 3.4735336194563664,
      "grad_norm": 0.7578348517417908,
      "learning_rate": 0.00015831917897591365,
      "loss": 1.4752,
      "step": 1516
    },
    {
      "epoch": 3.4758226037195996,
      "grad_norm": 0.7664616703987122,
      "learning_rate": 0.00015826802465401536,
      "loss": 1.4779,
      "step": 1517
    },
    {
      "epoch": 3.4781115879828324,
      "grad_norm": 0.7997162342071533,
      "learning_rate": 0.00015821684723653378,
      "loss": 1.5449,
      "step": 1518
    },
    {
      "epoch": 3.480400572246066,
      "grad_norm": 0.8170934319496155,
      "learning_rate": 0.000158165646743754,
      "loss": 1.3985,
      "step": 1519
    },
    {
      "epoch": 3.482689556509299,
      "grad_norm": 0.7675970792770386,
      "learning_rate": 0.00015811442319597026,
      "loss": 1.4542,
      "step": 1520
    },
    {
      "epoch": 3.484978540772532,
      "grad_norm": 0.7991522550582886,
      "learning_rate": 0.00015806317661348594,
      "loss": 1.4358,
      "step": 1521
    },
    {
      "epoch": 3.487267525035765,
      "grad_norm": 0.7838619351387024,
      "learning_rate": 0.00015801190701661357,
      "loss": 1.2911,
      "step": 1522
    },
    {
      "epoch": 3.4895565092989984,
      "grad_norm": 0.7992157936096191,
      "learning_rate": 0.0001579606144256747,
      "loss": 1.3085,
      "step": 1523
    },
    {
      "epoch": 3.4918454935622316,
      "grad_norm": 0.8065170049667358,
      "learning_rate": 0.00015790929886100018,
      "loss": 1.3501,
      "step": 1524
    },
    {
      "epoch": 3.494134477825465,
      "grad_norm": 0.7681014537811279,
      "learning_rate": 0.0001578579603429298,
      "loss": 1.1413,
      "step": 1525
    },
    {
      "epoch": 3.496423462088698,
      "grad_norm": 0.7987495064735413,
      "learning_rate": 0.00015780659889181253,
      "loss": 1.4301,
      "step": 1526
    },
    {
      "epoch": 3.4987124463519312,
      "grad_norm": 0.8664441108703613,
      "learning_rate": 0.00015775521452800634,
      "loss": 1.1792,
      "step": 1527
    },
    {
      "epoch": 3.5010014306151644,
      "grad_norm": 0.8405476212501526,
      "learning_rate": 0.00015770380727187845,
      "loss": 1.1298,
      "step": 1528
    },
    {
      "epoch": 3.5032904148783977,
      "grad_norm": 0.896719753742218,
      "learning_rate": 0.000157652377143805,
      "loss": 1.3248,
      "step": 1529
    },
    {
      "epoch": 3.505579399141631,
      "grad_norm": 0.825574517250061,
      "learning_rate": 0.00015760092416417125,
      "loss": 1.0444,
      "step": 1530
    },
    {
      "epoch": 3.507868383404864,
      "grad_norm": 0.8801305890083313,
      "learning_rate": 0.00015754944835337157,
      "loss": 1.1188,
      "step": 1531
    },
    {
      "epoch": 3.5101573676680973,
      "grad_norm": 0.8957611322402954,
      "learning_rate": 0.00015749794973180921,
      "loss": 1.0719,
      "step": 1532
    },
    {
      "epoch": 3.5124463519313305,
      "grad_norm": 1.020003080368042,
      "learning_rate": 0.0001574464283198967,
      "loss": 1.0326,
      "step": 1533
    },
    {
      "epoch": 3.5147353361945637,
      "grad_norm": 0.9494165778160095,
      "learning_rate": 0.00015739488413805543,
      "loss": 0.9371,
      "step": 1534
    },
    {
      "epoch": 3.517024320457797,
      "grad_norm": 0.9374694228172302,
      "learning_rate": 0.00015734331720671584,
      "loss": 1.0078,
      "step": 1535
    },
    {
      "epoch": 3.51931330472103,
      "grad_norm": 0.984219491481781,
      "learning_rate": 0.00015729172754631749,
      "loss": 0.7388,
      "step": 1536
    },
    {
      "epoch": 3.5216022889842633,
      "grad_norm": 1.0056235790252686,
      "learning_rate": 0.00015724011517730878,
      "loss": 0.7902,
      "step": 1537
    },
    {
      "epoch": 3.5238912732474965,
      "grad_norm": 1.028628945350647,
      "learning_rate": 0.00015718848012014725,
      "loss": 0.8631,
      "step": 1538
    },
    {
      "epoch": 3.5261802575107297,
      "grad_norm": 1.0650653839111328,
      "learning_rate": 0.00015713682239529942,
      "loss": 0.8271,
      "step": 1539
    },
    {
      "epoch": 3.528469241773963,
      "grad_norm": 1.0517405271530151,
      "learning_rate": 0.00015708514202324075,
      "loss": 0.5618,
      "step": 1540
    },
    {
      "epoch": 3.530758226037196,
      "grad_norm": 1.1985256671905518,
      "learning_rate": 0.00015703343902445562,
      "loss": 0.7335,
      "step": 1541
    },
    {
      "epoch": 3.5330472103004293,
      "grad_norm": 1.4088890552520752,
      "learning_rate": 0.00015698171341943753,
      "loss": 0.637,
      "step": 1542
    },
    {
      "epoch": 3.535336194563662,
      "grad_norm": 1.1967028379440308,
      "learning_rate": 0.0001569299652286888,
      "loss": 0.6781,
      "step": 1543
    },
    {
      "epoch": 3.5376251788268958,
      "grad_norm": 1.2229645252227783,
      "learning_rate": 0.00015687819447272078,
      "loss": 0.576,
      "step": 1544
    },
    {
      "epoch": 3.5399141630901285,
      "grad_norm": 1.3130781650543213,
      "learning_rate": 0.00015682640117205377,
      "loss": 0.4962,
      "step": 1545
    },
    {
      "epoch": 3.542203147353362,
      "grad_norm": 1.6849275827407837,
      "learning_rate": 0.00015677458534721692,
      "loss": 0.5366,
      "step": 1546
    },
    {
      "epoch": 3.544492131616595,
      "grad_norm": 1.2390133142471313,
      "learning_rate": 0.0001567227470187484,
      "loss": 0.3497,
      "step": 1547
    },
    {
      "epoch": 3.5467811158798286,
      "grad_norm": 1.5223252773284912,
      "learning_rate": 0.00015667088620719528,
      "loss": 0.3136,
      "step": 1548
    },
    {
      "epoch": 3.5490701001430613,
      "grad_norm": 1.772728443145752,
      "learning_rate": 0.0001566190029331135,
      "loss": 0.4348,
      "step": 1549
    },
    {
      "epoch": 3.5513590844062946,
      "grad_norm": 2.023331880569458,
      "learning_rate": 0.00015656709721706798,
      "loss": 0.412,
      "step": 1550
    },
    {
      "epoch": 3.5536480686695278,
      "grad_norm": 1.3482370376586914,
      "learning_rate": 0.00015651516907963246,
      "loss": 0.2864,
      "step": 1551
    },
    {
      "epoch": 3.555937052932761,
      "grad_norm": 1.777756690979004,
      "learning_rate": 0.0001564632185413896,
      "loss": 0.3162,
      "step": 1552
    },
    {
      "epoch": 3.558226037195994,
      "grad_norm": 1.7310400009155273,
      "learning_rate": 0.0001564112456229309,
      "loss": 0.3019,
      "step": 1553
    },
    {
      "epoch": 3.5605150214592274,
      "grad_norm": 2.4951937198638916,
      "learning_rate": 0.00015635925034485678,
      "loss": 0.2914,
      "step": 1554
    },
    {
      "epoch": 3.5628040057224606,
      "grad_norm": 1.6481578350067139,
      "learning_rate": 0.00015630723272777654,
      "loss": 0.2832,
      "step": 1555
    },
    {
      "epoch": 3.565092989985694,
      "grad_norm": 1.9817842245101929,
      "learning_rate": 0.00015625519279230827,
      "loss": 0.2218,
      "step": 1556
    },
    {
      "epoch": 3.567381974248927,
      "grad_norm": 2.244994640350342,
      "learning_rate": 0.00015620313055907892,
      "loss": 0.2811,
      "step": 1557
    },
    {
      "epoch": 3.56967095851216,
      "grad_norm": 1.7713173627853394,
      "learning_rate": 0.00015615104604872437,
      "loss": 0.2204,
      "step": 1558
    },
    {
      "epoch": 3.5719599427753934,
      "grad_norm": 3.4250166416168213,
      "learning_rate": 0.0001560989392818892,
      "loss": 0.3418,
      "step": 1559
    },
    {
      "epoch": 3.5742489270386266,
      "grad_norm": 0.8141402006149292,
      "learning_rate": 0.00015604681027922684,
      "loss": 1.5305,
      "step": 1560
    },
    {
      "epoch": 3.57653791130186,
      "grad_norm": 0.576050877571106,
      "learning_rate": 0.0001559946590613996,
      "loss": 1.7689,
      "step": 1561
    },
    {
      "epoch": 3.578826895565093,
      "grad_norm": 0.6083397269248962,
      "learning_rate": 0.00015594248564907857,
      "loss": 1.7663,
      "step": 1562
    },
    {
      "epoch": 3.5811158798283262,
      "grad_norm": 0.6765815615653992,
      "learning_rate": 0.00015589029006294363,
      "loss": 1.68,
      "step": 1563
    },
    {
      "epoch": 3.5834048640915594,
      "grad_norm": 0.684815526008606,
      "learning_rate": 0.00015583807232368336,
      "loss": 1.5431,
      "step": 1564
    },
    {
      "epoch": 3.5856938483547927,
      "grad_norm": 0.7227407693862915,
      "learning_rate": 0.00015578583245199533,
      "loss": 1.4256,
      "step": 1565
    },
    {
      "epoch": 3.587982832618026,
      "grad_norm": 0.7657136917114258,
      "learning_rate": 0.00015573357046858567,
      "loss": 1.5605,
      "step": 1566
    },
    {
      "epoch": 3.590271816881259,
      "grad_norm": 0.715350866317749,
      "learning_rate": 0.00015568128639416933,
      "loss": 1.6071,
      "step": 1567
    },
    {
      "epoch": 3.5925608011444923,
      "grad_norm": 0.7893856167793274,
      "learning_rate": 0.00015562898024947013,
      "loss": 1.5667,
      "step": 1568
    },
    {
      "epoch": 3.5948497854077255,
      "grad_norm": 0.8078209161758423,
      "learning_rate": 0.00015557665205522052,
      "loss": 1.3805,
      "step": 1569
    },
    {
      "epoch": 3.5971387696709582,
      "grad_norm": 0.7596533298492432,
      "learning_rate": 0.0001555243018321617,
      "loss": 1.3577,
      "step": 1570
    },
    {
      "epoch": 3.599427753934192,
      "grad_norm": 0.8347642421722412,
      "learning_rate": 0.00015547192960104367,
      "loss": 1.2966,
      "step": 1571
    },
    {
      "epoch": 3.6017167381974247,
      "grad_norm": 0.8208310604095459,
      "learning_rate": 0.0001554195353826251,
      "loss": 1.4773,
      "step": 1572
    },
    {
      "epoch": 3.6040057224606583,
      "grad_norm": 0.8004641532897949,
      "learning_rate": 0.00015536711919767338,
      "loss": 1.3247,
      "step": 1573
    },
    {
      "epoch": 3.606294706723891,
      "grad_norm": 0.8222883939743042,
      "learning_rate": 0.00015531468106696455,
      "loss": 1.2533,
      "step": 1574
    },
    {
      "epoch": 3.6085836909871247,
      "grad_norm": 0.7890996932983398,
      "learning_rate": 0.00015526222101128355,
      "loss": 1.1291,
      "step": 1575
    },
    {
      "epoch": 3.6108726752503575,
      "grad_norm": 0.7932547330856323,
      "learning_rate": 0.00015520973905142372,
      "loss": 1.2697,
      "step": 1576
    },
    {
      "epoch": 3.6131616595135907,
      "grad_norm": 0.7988429665565491,
      "learning_rate": 0.00015515723520818732,
      "loss": 1.2013,
      "step": 1577
    },
    {
      "epoch": 3.615450643776824,
      "grad_norm": 0.824824869632721,
      "learning_rate": 0.00015510470950238516,
      "loss": 1.0999,
      "step": 1578
    },
    {
      "epoch": 3.617739628040057,
      "grad_norm": 0.8311036229133606,
      "learning_rate": 0.00015505216195483675,
      "loss": 1.1142,
      "step": 1579
    },
    {
      "epoch": 3.6200286123032903,
      "grad_norm": 0.906527578830719,
      "learning_rate": 0.0001549995925863703,
      "loss": 1.1211,
      "step": 1580
    },
    {
      "epoch": 3.6223175965665235,
      "grad_norm": 0.8750422596931458,
      "learning_rate": 0.00015494700141782263,
      "loss": 1.0368,
      "step": 1581
    },
    {
      "epoch": 3.6246065808297567,
      "grad_norm": 0.87028968334198,
      "learning_rate": 0.0001548943884700391,
      "loss": 0.8723,
      "step": 1582
    },
    {
      "epoch": 3.62689556509299,
      "grad_norm": 0.9463768005371094,
      "learning_rate": 0.0001548417537638739,
      "loss": 0.9306,
      "step": 1583
    },
    {
      "epoch": 3.629184549356223,
      "grad_norm": 0.9717691540718079,
      "learning_rate": 0.0001547890973201897,
      "loss": 0.9884,
      "step": 1584
    },
    {
      "epoch": 3.6314735336194564,
      "grad_norm": 1.0096951723098755,
      "learning_rate": 0.0001547364191598579,
      "loss": 0.8862,
      "step": 1585
    },
    {
      "epoch": 3.6337625178826896,
      "grad_norm": 1.0112119913101196,
      "learning_rate": 0.00015468371930375834,
      "loss": 0.9107,
      "step": 1586
    },
    {
      "epoch": 3.6360515021459228,
      "grad_norm": 1.109299898147583,
      "learning_rate": 0.00015463099777277963,
      "loss": 0.7745,
      "step": 1587
    },
    {
      "epoch": 3.638340486409156,
      "grad_norm": 1.1044436693191528,
      "learning_rate": 0.00015457825458781887,
      "loss": 0.8227,
      "step": 1588
    },
    {
      "epoch": 3.640629470672389,
      "grad_norm": 0.9999521970748901,
      "learning_rate": 0.00015452548976978178,
      "loss": 0.6667,
      "step": 1589
    },
    {
      "epoch": 3.6429184549356224,
      "grad_norm": 1.1406440734863281,
      "learning_rate": 0.00015447270333958265,
      "loss": 0.6793,
      "step": 1590
    },
    {
      "epoch": 3.6452074391988556,
      "grad_norm": 1.1465872526168823,
      "learning_rate": 0.00015441989531814437,
      "loss": 0.5937,
      "step": 1591
    },
    {
      "epoch": 3.647496423462089,
      "grad_norm": 1.1943423748016357,
      "learning_rate": 0.00015436706572639826,
      "loss": 0.6536,
      "step": 1592
    },
    {
      "epoch": 3.649785407725322,
      "grad_norm": 1.4073413610458374,
      "learning_rate": 0.00015431421458528437,
      "loss": 0.6297,
      "step": 1593
    },
    {
      "epoch": 3.652074391988555,
      "grad_norm": 1.248165488243103,
      "learning_rate": 0.00015426134191575116,
      "loss": 0.5336,
      "step": 1594
    },
    {
      "epoch": 3.6543633762517884,
      "grad_norm": 1.0524934530258179,
      "learning_rate": 0.00015420844773875568,
      "loss": 0.4659,
      "step": 1595
    },
    {
      "epoch": 3.6566523605150216,
      "grad_norm": 1.3649994134902954,
      "learning_rate": 0.0001541555320752635,
      "loss": 0.4773,
      "step": 1596
    },
    {
      "epoch": 3.6589413447782544,
      "grad_norm": 1.3285311460494995,
      "learning_rate": 0.0001541025949462487,
      "loss": 0.4507,
      "step": 1597
    },
    {
      "epoch": 3.661230329041488,
      "grad_norm": 1.3772430419921875,
      "learning_rate": 0.00015404963637269381,
      "loss": 0.3966,
      "step": 1598
    },
    {
      "epoch": 3.663519313304721,
      "grad_norm": 1.4577609300613403,
      "learning_rate": 0.00015399665637559,
      "loss": 0.3689,
      "step": 1599
    },
    {
      "epoch": 3.6658082975679545,
      "grad_norm": 1.5130454301834106,
      "learning_rate": 0.00015394365497593682,
      "loss": 0.3042,
      "step": 1600
    },
    {
      "epoch": 3.668097281831187,
      "grad_norm": 1.4147976636886597,
      "learning_rate": 0.00015389063219474228,
      "loss": 0.3654,
      "step": 1601
    },
    {
      "epoch": 3.670386266094421,
      "grad_norm": 2.0109431743621826,
      "learning_rate": 0.000153837588053023,
      "loss": 0.3362,
      "step": 1602
    },
    {
      "epoch": 3.6726752503576536,
      "grad_norm": 2.0059592723846436,
      "learning_rate": 0.00015378452257180388,
      "loss": 0.279,
      "step": 1603
    },
    {
      "epoch": 3.674964234620887,
      "grad_norm": 2.262427806854248,
      "learning_rate": 0.0001537314357721185,
      "loss": 0.3572,
      "step": 1604
    },
    {
      "epoch": 3.67725321888412,
      "grad_norm": 1.9626609086990356,
      "learning_rate": 0.0001536783276750087,
      "loss": 0.2642,
      "step": 1605
    },
    {
      "epoch": 3.6795422031473533,
      "grad_norm": 2.9041426181793213,
      "learning_rate": 0.00015362519830152484,
      "loss": 0.3735,
      "step": 1606
    },
    {
      "epoch": 3.6818311874105865,
      "grad_norm": 2.211362838745117,
      "learning_rate": 0.00015357204767272572,
      "loss": 0.2951,
      "step": 1607
    },
    {
      "epoch": 3.6841201716738197,
      "grad_norm": 3.4044175148010254,
      "learning_rate": 0.00015351887580967855,
      "loss": 0.3566,
      "step": 1608
    },
    {
      "epoch": 3.686409155937053,
      "grad_norm": 3.792001485824585,
      "learning_rate": 0.00015346568273345897,
      "loss": 0.3662,
      "step": 1609
    },
    {
      "epoch": 3.688698140200286,
      "grad_norm": 1.4554511308670044,
      "learning_rate": 0.00015341246846515096,
      "loss": 1.6486,
      "step": 1610
    },
    {
      "epoch": 3.6909871244635193,
      "grad_norm": 0.5519815683364868,
      "learning_rate": 0.000153359233025847,
      "loss": 1.8248,
      "step": 1611
    },
    {
      "epoch": 3.6932761087267525,
      "grad_norm": 0.5921805500984192,
      "learning_rate": 0.0001533059764366479,
      "loss": 1.7423,
      "step": 1612
    },
    {
      "epoch": 3.6955650929899857,
      "grad_norm": 0.6531163454055786,
      "learning_rate": 0.00015325269871866291,
      "loss": 1.5875,
      "step": 1613
    },
    {
      "epoch": 3.697854077253219,
      "grad_norm": 0.6997845768928528,
      "learning_rate": 0.00015319939989300954,
      "loss": 1.59,
      "step": 1614
    },
    {
      "epoch": 3.700143061516452,
      "grad_norm": 0.7298904657363892,
      "learning_rate": 0.00015314607998081386,
      "loss": 1.6081,
      "step": 1615
    },
    {
      "epoch": 3.7024320457796853,
      "grad_norm": 0.7905436158180237,
      "learning_rate": 0.00015309273900321007,
      "loss": 1.5784,
      "step": 1616
    },
    {
      "epoch": 3.7047210300429185,
      "grad_norm": 0.7912278771400452,
      "learning_rate": 0.00015303937698134083,
      "loss": 1.4748,
      "step": 1617
    },
    {
      "epoch": 3.7070100143061517,
      "grad_norm": 0.8271948099136353,
      "learning_rate": 0.00015298599393635722,
      "loss": 1.489,
      "step": 1618
    },
    {
      "epoch": 3.709298998569385,
      "grad_norm": 0.7761908173561096,
      "learning_rate": 0.00015293258988941855,
      "loss": 1.41,
      "step": 1619
    },
    {
      "epoch": 3.711587982832618,
      "grad_norm": 0.8067419528961182,
      "learning_rate": 0.0001528791648616924,
      "loss": 1.4425,
      "step": 1620
    },
    {
      "epoch": 3.7138769670958514,
      "grad_norm": 0.8288164734840393,
      "learning_rate": 0.00015282571887435483,
      "loss": 1.4353,
      "step": 1621
    },
    {
      "epoch": 3.7161659513590846,
      "grad_norm": 0.7974148988723755,
      "learning_rate": 0.00015277225194859008,
      "loss": 1.3482,
      "step": 1622
    },
    {
      "epoch": 3.7184549356223178,
      "grad_norm": 0.7817927002906799,
      "learning_rate": 0.0001527187641055908,
      "loss": 1.3171,
      "step": 1623
    },
    {
      "epoch": 3.7207439198855505,
      "grad_norm": 0.8328514099121094,
      "learning_rate": 0.00015266525536655775,
      "loss": 1.4593,
      "step": 1624
    },
    {
      "epoch": 3.723032904148784,
      "grad_norm": 0.7940589189529419,
      "learning_rate": 0.00015261172575270016,
      "loss": 1.2146,
      "step": 1625
    },
    {
      "epoch": 3.725321888412017,
      "grad_norm": 0.8301095962524414,
      "learning_rate": 0.00015255817528523545,
      "loss": 1.2037,
      "step": 1626
    },
    {
      "epoch": 3.7276108726752506,
      "grad_norm": 0.8458488583564758,
      "learning_rate": 0.0001525046039853893,
      "loss": 1.2401,
      "step": 1627
    },
    {
      "epoch": 3.7298998569384834,
      "grad_norm": 0.85792076587677,
      "learning_rate": 0.00015245101187439563,
      "loss": 1.1158,
      "step": 1628
    },
    {
      "epoch": 3.732188841201717,
      "grad_norm": 0.8571141362190247,
      "learning_rate": 0.0001523973989734967,
      "loss": 1.0806,
      "step": 1629
    },
    {
      "epoch": 3.7344778254649498,
      "grad_norm": 0.8554336428642273,
      "learning_rate": 0.00015234376530394295,
      "loss": 1.1863,
      "step": 1630
    },
    {
      "epoch": 3.736766809728183,
      "grad_norm": 0.8635918498039246,
      "learning_rate": 0.00015229011088699302,
      "loss": 0.9766,
      "step": 1631
    },
    {
      "epoch": 3.739055793991416,
      "grad_norm": 0.9634860157966614,
      "learning_rate": 0.0001522364357439138,
      "loss": 1.1977,
      "step": 1632
    },
    {
      "epoch": 3.7413447782546494,
      "grad_norm": 0.9780722856521606,
      "learning_rate": 0.00015218273989598045,
      "loss": 1.0433,
      "step": 1633
    },
    {
      "epoch": 3.7436337625178826,
      "grad_norm": 0.9208489656448364,
      "learning_rate": 0.00015212902336447627,
      "loss": 1.0092,
      "step": 1634
    },
    {
      "epoch": 3.745922746781116,
      "grad_norm": 1.0279505252838135,
      "learning_rate": 0.00015207528617069272,
      "loss": 0.9366,
      "step": 1635
    },
    {
      "epoch": 3.748211731044349,
      "grad_norm": 1.0359374284744263,
      "learning_rate": 0.00015202152833592954,
      "loss": 0.8968,
      "step": 1636
    },
    {
      "epoch": 3.7505007153075822,
      "grad_norm": 1.081196665763855,
      "learning_rate": 0.00015196774988149466,
      "loss": 0.7949,
      "step": 1637
    },
    {
      "epoch": 3.7527896995708154,
      "grad_norm": 1.128497838973999,
      "learning_rate": 0.00015191395082870412,
      "loss": 0.8224,
      "step": 1638
    },
    {
      "epoch": 3.7550786838340486,
      "grad_norm": 1.035473346710205,
      "learning_rate": 0.00015186013119888213,
      "loss": 0.7625,
      "step": 1639
    },
    {
      "epoch": 3.757367668097282,
      "grad_norm": 1.1496644020080566,
      "learning_rate": 0.00015180629101336108,
      "loss": 0.8328,
      "step": 1640
    },
    {
      "epoch": 3.759656652360515,
      "grad_norm": 1.1070492267608643,
      "learning_rate": 0.0001517524302934815,
      "loss": 0.7106,
      "step": 1641
    },
    {
      "epoch": 3.7619456366237483,
      "grad_norm": 1.2384364604949951,
      "learning_rate": 0.000151698549060592,
      "loss": 0.6606,
      "step": 1642
    },
    {
      "epoch": 3.7642346208869815,
      "grad_norm": 1.1313549280166626,
      "learning_rate": 0.0001516446473360495,
      "loss": 0.5574,
      "step": 1643
    },
    {
      "epoch": 3.7665236051502147,
      "grad_norm": 1.3806557655334473,
      "learning_rate": 0.00015159072514121883,
      "loss": 0.6239,
      "step": 1644
    },
    {
      "epoch": 3.768812589413448,
      "grad_norm": 1.3479688167572021,
      "learning_rate": 0.00015153678249747306,
      "loss": 0.5784,
      "step": 1645
    },
    {
      "epoch": 3.771101573676681,
      "grad_norm": 1.3832741975784302,
      "learning_rate": 0.00015148281942619333,
      "loss": 0.4887,
      "step": 1646
    },
    {
      "epoch": 3.7733905579399143,
      "grad_norm": 1.5570096969604492,
      "learning_rate": 0.00015142883594876887,
      "loss": 0.6388,
      "step": 1647
    },
    {
      "epoch": 3.7756795422031475,
      "grad_norm": 1.3564070463180542,
      "learning_rate": 0.000151374832086597,
      "loss": 0.4629,
      "step": 1648
    },
    {
      "epoch": 3.7779685264663807,
      "grad_norm": 1.5236693620681763,
      "learning_rate": 0.00015132080786108313,
      "loss": 0.3627,
      "step": 1649
    },
    {
      "epoch": 3.780257510729614,
      "grad_norm": 1.6069036722183228,
      "learning_rate": 0.00015126676329364072,
      "loss": 0.3655,
      "step": 1650
    },
    {
      "epoch": 3.7825464949928467,
      "grad_norm": 1.5798829793930054,
      "learning_rate": 0.00015121269840569134,
      "loss": 0.367,
      "step": 1651
    },
    {
      "epoch": 3.7848354792560803,
      "grad_norm": 1.8589974641799927,
      "learning_rate": 0.00015115861321866455,
      "loss": 0.3061,
      "step": 1652
    },
    {
      "epoch": 3.787124463519313,
      "grad_norm": 1.395572543144226,
      "learning_rate": 0.00015110450775399805,
      "loss": 0.2572,
      "step": 1653
    },
    {
      "epoch": 3.7894134477825467,
      "grad_norm": 2.529019832611084,
      "learning_rate": 0.00015105038203313748,
      "loss": 0.4464,
      "step": 1654
    },
    {
      "epoch": 3.7917024320457795,
      "grad_norm": 1.760545253753662,
      "learning_rate": 0.0001509962360775365,
      "loss": 0.2562,
      "step": 1655
    },
    {
      "epoch": 3.7939914163090127,
      "grad_norm": 1.9058586359024048,
      "learning_rate": 0.00015094206990865693,
      "loss": 0.25,
      "step": 1656
    },
    {
      "epoch": 3.796280400572246,
      "grad_norm": 1.9067879915237427,
      "learning_rate": 0.00015088788354796843,
      "loss": 0.1715,
      "step": 1657
    },
    {
      "epoch": 3.798569384835479,
      "grad_norm": 2.5658154487609863,
      "learning_rate": 0.0001508336770169488,
      "loss": 0.2417,
      "step": 1658
    },
    {
      "epoch": 3.8008583690987123,
      "grad_norm": 5.0240912437438965,
      "learning_rate": 0.0001507794503370837,
      "loss": 0.2909,
      "step": 1659
    },
    {
      "epoch": 3.8031473533619455,
      "grad_norm": 1.011162519454956,
      "learning_rate": 0.00015072520352986697,
      "loss": 1.6048,
      "step": 1660
    },
    {
      "epoch": 3.8054363376251787,
      "grad_norm": 0.5463991165161133,
      "learning_rate": 0.0001506709366168002,
      "loss": 1.8731,
      "step": 1661
    },
    {
      "epoch": 3.807725321888412,
      "grad_norm": 0.6056814789772034,
      "learning_rate": 0.00015061664961939312,
      "loss": 1.7895,
      "step": 1662
    },
    {
      "epoch": 3.810014306151645,
      "grad_norm": 0.6620305776596069,
      "learning_rate": 0.00015056234255916335,
      "loss": 1.7562,
      "step": 1663
    },
    {
      "epoch": 3.8123032904148784,
      "grad_norm": 0.7236174941062927,
      "learning_rate": 0.00015050801545763646,
      "loss": 1.642,
      "step": 1664
    },
    {
      "epoch": 3.8145922746781116,
      "grad_norm": 0.7159740924835205,
      "learning_rate": 0.000150453668336346,
      "loss": 1.6126,
      "step": 1665
    },
    {
      "epoch": 3.816881258941345,
      "grad_norm": 0.7541245222091675,
      "learning_rate": 0.0001503993012168334,
      "loss": 1.4509,
      "step": 1666
    },
    {
      "epoch": 3.819170243204578,
      "grad_norm": 0.7407465577125549,
      "learning_rate": 0.00015034491412064806,
      "loss": 1.4913,
      "step": 1667
    },
    {
      "epoch": 3.821459227467811,
      "grad_norm": 0.8195914626121521,
      "learning_rate": 0.00015029050706934737,
      "loss": 1.6013,
      "step": 1668
    },
    {
      "epoch": 3.8237482117310444,
      "grad_norm": 0.8221433758735657,
      "learning_rate": 0.00015023608008449642,
      "loss": 1.4439,
      "step": 1669
    },
    {
      "epoch": 3.8260371959942776,
      "grad_norm": 0.8168205618858337,
      "learning_rate": 0.00015018163318766833,
      "loss": 1.3029,
      "step": 1670
    },
    {
      "epoch": 3.828326180257511,
      "grad_norm": 0.8368527889251709,
      "learning_rate": 0.0001501271664004442,
      "loss": 1.4415,
      "step": 1671
    },
    {
      "epoch": 3.830615164520744,
      "grad_norm": 0.8870606422424316,
      "learning_rate": 0.00015007267974441292,
      "loss": 1.4138,
      "step": 1672
    },
    {
      "epoch": 3.8329041487839772,
      "grad_norm": 0.8419132828712463,
      "learning_rate": 0.00015001817324117122,
      "loss": 1.3305,
      "step": 1673
    },
    {
      "epoch": 3.8351931330472104,
      "grad_norm": 0.8348716497421265,
      "learning_rate": 0.00014996364691232373,
      "loss": 1.2744,
      "step": 1674
    },
    {
      "epoch": 3.8374821173104436,
      "grad_norm": 0.8420340418815613,
      "learning_rate": 0.00014990910077948292,
      "loss": 1.3047,
      "step": 1675
    },
    {
      "epoch": 3.8397711015736764,
      "grad_norm": 0.8508443236351013,
      "learning_rate": 0.0001498545348642692,
      "loss": 1.285,
      "step": 1676
    },
    {
      "epoch": 3.84206008583691,
      "grad_norm": 0.8771832585334778,
      "learning_rate": 0.00014979994918831073,
      "loss": 1.3076,
      "step": 1677
    },
    {
      "epoch": 3.844349070100143,
      "grad_norm": 0.8925603628158569,
      "learning_rate": 0.00014974534377324355,
      "loss": 1.2541,
      "step": 1678
    },
    {
      "epoch": 3.8466380543633765,
      "grad_norm": 0.9128324389457703,
      "learning_rate": 0.00014969071864071143,
      "loss": 1.0793,
      "step": 1679
    },
    {
      "epoch": 3.8489270386266092,
      "grad_norm": 0.9458082914352417,
      "learning_rate": 0.00014963607381236608,
      "loss": 1.0655,
      "step": 1680
    },
    {
      "epoch": 3.851216022889843,
      "grad_norm": 0.9138215780258179,
      "learning_rate": 0.00014958140930986698,
      "loss": 1.0918,
      "step": 1681
    },
    {
      "epoch": 3.8535050071530756,
      "grad_norm": 0.9036693572998047,
      "learning_rate": 0.00014952672515488132,
      "loss": 0.8557,
      "step": 1682
    },
    {
      "epoch": 3.855793991416309,
      "grad_norm": 0.8931397795677185,
      "learning_rate": 0.00014947202136908426,
      "loss": 0.9188,
      "step": 1683
    },
    {
      "epoch": 3.858082975679542,
      "grad_norm": 0.9750208258628845,
      "learning_rate": 0.00014941729797415853,
      "loss": 0.9551,
      "step": 1684
    },
    {
      "epoch": 3.8603719599427753,
      "grad_norm": 1.0227867364883423,
      "learning_rate": 0.0001493625549917948,
      "loss": 0.8957,
      "step": 1685
    },
    {
      "epoch": 3.8626609442060085,
      "grad_norm": 1.04948890209198,
      "learning_rate": 0.00014930779244369142,
      "loss": 0.8852,
      "step": 1686
    },
    {
      "epoch": 3.8649499284692417,
      "grad_norm": 0.9962487816810608,
      "learning_rate": 0.0001492530103515545,
      "loss": 0.8416,
      "step": 1687
    },
    {
      "epoch": 3.867238912732475,
      "grad_norm": 1.0342847108840942,
      "learning_rate": 0.00014919820873709796,
      "loss": 0.9805,
      "step": 1688
    },
    {
      "epoch": 3.869527896995708,
      "grad_norm": 1.0398114919662476,
      "learning_rate": 0.00014914338762204331,
      "loss": 0.8916,
      "step": 1689
    },
    {
      "epoch": 3.8718168812589413,
      "grad_norm": 1.1402392387390137,
      "learning_rate": 0.00014908854702811998,
      "loss": 0.769,
      "step": 1690
    },
    {
      "epoch": 3.8741058655221745,
      "grad_norm": 1.217068076133728,
      "learning_rate": 0.00014903368697706502,
      "loss": 0.8346,
      "step": 1691
    },
    {
      "epoch": 3.8763948497854077,
      "grad_norm": 1.293504238128662,
      "learning_rate": 0.00014897880749062316,
      "loss": 0.8068,
      "step": 1692
    },
    {
      "epoch": 3.878683834048641,
      "grad_norm": 1.333113193511963,
      "learning_rate": 0.0001489239085905469,
      "loss": 0.6991,
      "step": 1693
    },
    {
      "epoch": 3.880972818311874,
      "grad_norm": 1.1802215576171875,
      "learning_rate": 0.00014886899029859643,
      "loss": 0.6316,
      "step": 1694
    },
    {
      "epoch": 3.8832618025751073,
      "grad_norm": 1.3808867931365967,
      "learning_rate": 0.00014881405263653954,
      "loss": 0.5928,
      "step": 1695
    },
    {
      "epoch": 3.8855507868383405,
      "grad_norm": 1.3361176252365112,
      "learning_rate": 0.00014875909562615184,
      "loss": 0.6402,
      "step": 1696
    },
    {
      "epoch": 3.8878397711015737,
      "grad_norm": 1.2890623807907104,
      "learning_rate": 0.0001487041192892165,
      "loss": 0.5675,
      "step": 1697
    },
    {
      "epoch": 3.890128755364807,
      "grad_norm": 1.430479884147644,
      "learning_rate": 0.00014864912364752434,
      "loss": 0.5139,
      "step": 1698
    },
    {
      "epoch": 3.89241773962804,
      "grad_norm": 1.2463384866714478,
      "learning_rate": 0.000148594108722874,
      "loss": 0.3208,
      "step": 1699
    },
    {
      "epoch": 3.8947067238912734,
      "grad_norm": 1.4024913311004639,
      "learning_rate": 0.00014853907453707147,
      "loss": 0.3998,
      "step": 1700
    },
    {
      "epoch": 3.8969957081545066,
      "grad_norm": 1.6548993587493896,
      "learning_rate": 0.00014848402111193067,
      "loss": 0.2924,
      "step": 1701
    },
    {
      "epoch": 3.89928469241774,
      "grad_norm": 1.315961480140686,
      "learning_rate": 0.000148428948469273,
      "loss": 0.2436,
      "step": 1702
    },
    {
      "epoch": 3.9015736766809725,
      "grad_norm": 2.049527168273926,
      "learning_rate": 0.00014837385663092744,
      "loss": 0.3896,
      "step": 1703
    },
    {
      "epoch": 3.903862660944206,
      "grad_norm": 1.8041912317276,
      "learning_rate": 0.0001483187456187307,
      "loss": 0.3867,
      "step": 1704
    },
    {
      "epoch": 3.906151645207439,
      "grad_norm": 1.984589695930481,
      "learning_rate": 0.00014826361545452696,
      "loss": 0.3802,
      "step": 1705
    },
    {
      "epoch": 3.9084406294706726,
      "grad_norm": 2.0554628372192383,
      "learning_rate": 0.0001482084661601681,
      "loss": 0.3724,
      "step": 1706
    },
    {
      "epoch": 3.9107296137339054,
      "grad_norm": 2.1720497608184814,
      "learning_rate": 0.00014815329775751357,
      "loss": 0.2343,
      "step": 1707
    },
    {
      "epoch": 3.913018597997139,
      "grad_norm": 3.0064074993133545,
      "learning_rate": 0.00014809811026843026,
      "loss": 0.2229,
      "step": 1708
    },
    {
      "epoch": 3.915307582260372,
      "grad_norm": 3.7488739490509033,
      "learning_rate": 0.0001480429037147928,
      "loss": 0.2322,
      "step": 1709
    },
    {
      "epoch": 3.917596566523605,
      "grad_norm": 0.7786121964454651,
      "learning_rate": 0.0001479876781184833,
      "loss": 1.6537,
      "step": 1710
    },
    {
      "epoch": 3.919885550786838,
      "grad_norm": 0.5747677087783813,
      "learning_rate": 0.00014793243350139142,
      "loss": 1.9256,
      "step": 1711
    },
    {
      "epoch": 3.9221745350500714,
      "grad_norm": 0.6486791372299194,
      "learning_rate": 0.00014787716988541436,
      "loss": 1.6479,
      "step": 1712
    },
    {
      "epoch": 3.9244635193133046,
      "grad_norm": 0.7661999464035034,
      "learning_rate": 0.00014782188729245684,
      "loss": 1.5968,
      "step": 1713
    },
    {
      "epoch": 3.926752503576538,
      "grad_norm": 0.7707553505897522,
      "learning_rate": 0.0001477665857444311,
      "loss": 1.5102,
      "step": 1714
    },
    {
      "epoch": 3.929041487839771,
      "grad_norm": 0.8604703545570374,
      "learning_rate": 0.00014771126526325694,
      "loss": 1.5109,
      "step": 1715
    },
    {
      "epoch": 3.9313304721030042,
      "grad_norm": 0.8465156555175781,
      "learning_rate": 0.00014765592587086162,
      "loss": 1.4792,
      "step": 1716
    },
    {
      "epoch": 3.9336194563662374,
      "grad_norm": 0.9079059362411499,
      "learning_rate": 0.0001476005675891799,
      "loss": 1.5047,
      "step": 1717
    },
    {
      "epoch": 3.9359084406294707,
      "grad_norm": 0.8793495297431946,
      "learning_rate": 0.00014754519044015405,
      "loss": 1.3054,
      "step": 1718
    },
    {
      "epoch": 3.938197424892704,
      "grad_norm": 0.9664410948753357,
      "learning_rate": 0.00014748979444573378,
      "loss": 1.2806,
      "step": 1719
    },
    {
      "epoch": 3.940486409155937,
      "grad_norm": 0.9473050236701965,
      "learning_rate": 0.0001474343796278763,
      "loss": 1.406,
      "step": 1720
    },
    {
      "epoch": 3.9427753934191703,
      "grad_norm": 0.9971415996551514,
      "learning_rate": 0.00014737894600854628,
      "loss": 1.3588,
      "step": 1721
    },
    {
      "epoch": 3.9450643776824035,
      "grad_norm": 1.0359668731689453,
      "learning_rate": 0.00014732349360971586,
      "loss": 1.226,
      "step": 1722
    },
    {
      "epoch": 3.9473533619456367,
      "grad_norm": 1.0823783874511719,
      "learning_rate": 0.0001472680224533645,
      "loss": 1.2084,
      "step": 1723
    },
    {
      "epoch": 3.94964234620887,
      "grad_norm": 0.9512320756912231,
      "learning_rate": 0.00014721253256147934,
      "loss": 1.0636,
      "step": 1724
    },
    {
      "epoch": 3.951931330472103,
      "grad_norm": 1.0435305833816528,
      "learning_rate": 0.0001471570239560547,
      "loss": 1.0408,
      "step": 1725
    },
    {
      "epoch": 3.9542203147353363,
      "grad_norm": 0.9986626505851746,
      "learning_rate": 0.00014710149665909242,
      "loss": 0.9198,
      "step": 1726
    },
    {
      "epoch": 3.9565092989985695,
      "grad_norm": 1.0072948932647705,
      "learning_rate": 0.00014704595069260182,
      "loss": 0.8214,
      "step": 1727
    },
    {
      "epoch": 3.9587982832618027,
      "grad_norm": 1.0602082014083862,
      "learning_rate": 0.00014699038607859946,
      "loss": 0.8308,
      "step": 1728
    },
    {
      "epoch": 3.961087267525036,
      "grad_norm": 1.0133410692214966,
      "learning_rate": 0.00014693480283910941,
      "loss": 0.8048,
      "step": 1729
    },
    {
      "epoch": 3.9633762517882687,
      "grad_norm": 1.1373006105422974,
      "learning_rate": 0.00014687920099616311,
      "loss": 0.984,
      "step": 1730
    },
    {
      "epoch": 3.9656652360515023,
      "grad_norm": 1.101943016052246,
      "learning_rate": 0.00014682358057179932,
      "loss": 0.8476,
      "step": 1731
    },
    {
      "epoch": 3.967954220314735,
      "grad_norm": 1.243362307548523,
      "learning_rate": 0.00014676794158806423,
      "loss": 0.8338,
      "step": 1732
    },
    {
      "epoch": 3.9702432045779688,
      "grad_norm": 1.2640669345855713,
      "learning_rate": 0.00014671228406701133,
      "loss": 0.8639,
      "step": 1733
    },
    {
      "epoch": 3.9725321888412015,
      "grad_norm": 1.2642742395401,
      "learning_rate": 0.00014665660803070153,
      "loss": 0.7593,
      "step": 1734
    },
    {
      "epoch": 3.974821173104435,
      "grad_norm": 1.2393301725387573,
      "learning_rate": 0.00014660091350120297,
      "loss": 0.532,
      "step": 1735
    },
    {
      "epoch": 3.977110157367668,
      "grad_norm": 1.4510412216186523,
      "learning_rate": 0.00014654520050059121,
      "loss": 0.6041,
      "step": 1736
    },
    {
      "epoch": 3.979399141630901,
      "grad_norm": 1.30900239944458,
      "learning_rate": 0.00014648946905094912,
      "loss": 0.3425,
      "step": 1737
    },
    {
      "epoch": 3.9816881258941343,
      "grad_norm": 1.4096959829330444,
      "learning_rate": 0.00014643371917436683,
      "loss": 0.4687,
      "step": 1738
    },
    {
      "epoch": 3.9839771101573676,
      "grad_norm": 1.5130640268325806,
      "learning_rate": 0.00014637795089294186,
      "loss": 0.4595,
      "step": 1739
    },
    {
      "epoch": 3.9862660944206008,
      "grad_norm": 1.5498608350753784,
      "learning_rate": 0.0001463221642287789,
      "loss": 0.4417,
      "step": 1740
    },
    {
      "epoch": 3.988555078683834,
      "grad_norm": 1.3768302202224731,
      "learning_rate": 0.00014626635920399008,
      "loss": 0.3324,
      "step": 1741
    },
    {
      "epoch": 3.990844062947067,
      "grad_norm": 1.9338836669921875,
      "learning_rate": 0.00014621053584069468,
      "loss": 0.3389,
      "step": 1742
    },
    {
      "epoch": 3.9931330472103004,
      "grad_norm": 1.5647732019424438,
      "learning_rate": 0.0001461546941610193,
      "loss": 0.2612,
      "step": 1743
    },
    {
      "epoch": 3.9954220314735336,
      "grad_norm": 2.102107524871826,
      "learning_rate": 0.00014609883418709784,
      "loss": 0.3016,
      "step": 1744
    },
    {
      "epoch": 3.997711015736767,
      "grad_norm": 1.862336277961731,
      "learning_rate": 0.00014604295594107137,
      "loss": 0.1563,
      "step": 1745
    },
    {
      "epoch": 4.001144492131616,
      "grad_norm": 5.682592391967773,
      "learning_rate": 0.00014598705944508824,
      "loss": 1.076,
      "step": 1746
    },
    {
      "epoch": 4.00343347639485,
      "grad_norm": 0.44566667079925537,
      "learning_rate": 0.00014593114472130406,
      "loss": 1.6827,
      "step": 1747
    },
    {
      "epoch": 4.005722460658083,
      "grad_norm": 0.4679805040359497,
      "learning_rate": 0.0001458752117918816,
      "loss": 1.851,
      "step": 1748
    },
    {
      "epoch": 4.0080114449213164,
      "grad_norm": 0.5022954344749451,
      "learning_rate": 0.0001458192606789909,
      "loss": 1.567,
      "step": 1749
    },
    {
      "epoch": 4.010300429184549,
      "grad_norm": 0.5355679392814636,
      "learning_rate": 0.00014576329140480925,
      "loss": 1.2447,
      "step": 1750
    },
    {
      "epoch": 4.012589413447783,
      "grad_norm": 0.6063524484634399,
      "learning_rate": 0.000145707303991521,
      "loss": 1.5429,
      "step": 1751
    },
    {
      "epoch": 4.014878397711016,
      "grad_norm": 0.6033669710159302,
      "learning_rate": 0.00014565129846131784,
      "loss": 1.2655,
      "step": 1752
    },
    {
      "epoch": 4.017167381974249,
      "grad_norm": 0.6651263236999512,
      "learning_rate": 0.0001455952748363985,
      "loss": 1.3036,
      "step": 1753
    },
    {
      "epoch": 4.019456366237482,
      "grad_norm": 0.6865150332450867,
      "learning_rate": 0.00014553923313896902,
      "loss": 1.2273,
      "step": 1754
    },
    {
      "epoch": 4.021745350500716,
      "grad_norm": 0.8032099604606628,
      "learning_rate": 0.00014548317339124252,
      "loss": 1.3473,
      "step": 1755
    },
    {
      "epoch": 4.0240343347639485,
      "grad_norm": 0.7824806571006775,
      "learning_rate": 0.0001454270956154393,
      "loss": 1.1447,
      "step": 1756
    },
    {
      "epoch": 4.026323319027182,
      "grad_norm": 0.816189169883728,
      "learning_rate": 0.00014537099983378675,
      "loss": 1.1298,
      "step": 1757
    },
    {
      "epoch": 4.028612303290415,
      "grad_norm": 0.8846980333328247,
      "learning_rate": 0.0001453148860685195,
      "loss": 1.1162,
      "step": 1758
    },
    {
      "epoch": 4.030901287553648,
      "grad_norm": 0.9195663332939148,
      "learning_rate": 0.00014525875434187924,
      "loss": 1.1653,
      "step": 1759
    },
    {
      "epoch": 4.033190271816881,
      "grad_norm": 0.9265051484107971,
      "learning_rate": 0.0001452026046761148,
      "loss": 1.0801,
      "step": 1760
    },
    {
      "epoch": 4.035479256080114,
      "grad_norm": 1.041785717010498,
      "learning_rate": 0.0001451464370934821,
      "loss": 1.0668,
      "step": 1761
    },
    {
      "epoch": 4.037768240343348,
      "grad_norm": 0.9964609742164612,
      "learning_rate": 0.00014509025161624412,
      "loss": 0.9903,
      "step": 1762
    },
    {
      "epoch": 4.0400572246065805,
      "grad_norm": 1.0553271770477295,
      "learning_rate": 0.0001450340482666711,
      "loss": 0.8965,
      "step": 1763
    },
    {
      "epoch": 4.042346208869814,
      "grad_norm": 1.1435829401016235,
      "learning_rate": 0.00014497782706704017,
      "loss": 0.8686,
      "step": 1764
    },
    {
      "epoch": 4.044635193133047,
      "grad_norm": 1.0699126720428467,
      "learning_rate": 0.00014492158803963568,
      "loss": 0.7944,
      "step": 1765
    },
    {
      "epoch": 4.0469241773962805,
      "grad_norm": 1.1480063199996948,
      "learning_rate": 0.00014486533120674892,
      "loss": 0.8461,
      "step": 1766
    },
    {
      "epoch": 4.049213161659513,
      "grad_norm": 1.246055006980896,
      "learning_rate": 0.00014480905659067833,
      "loss": 0.776,
      "step": 1767
    },
    {
      "epoch": 4.051502145922747,
      "grad_norm": 1.2290637493133545,
      "learning_rate": 0.00014475276421372935,
      "loss": 0.7211,
      "step": 1768
    },
    {
      "epoch": 4.05379113018598,
      "grad_norm": 1.1152641773223877,
      "learning_rate": 0.00014469645409821455,
      "loss": 0.5442,
      "step": 1769
    },
    {
      "epoch": 4.056080114449213,
      "grad_norm": 1.152429461479187,
      "learning_rate": 0.00014464012626645335,
      "loss": 0.5681,
      "step": 1770
    },
    {
      "epoch": 4.058369098712446,
      "grad_norm": 1.2199641466140747,
      "learning_rate": 0.00014458378074077243,
      "loss": 0.5907,
      "step": 1771
    },
    {
      "epoch": 4.06065808297568,
      "grad_norm": 1.0936607122421265,
      "learning_rate": 0.00014452741754350526,
      "loss": 0.4856,
      "step": 1772
    },
    {
      "epoch": 4.0629470672389125,
      "grad_norm": 1.371535301208496,
      "learning_rate": 0.00014447103669699242,
      "loss": 0.544,
      "step": 1773
    },
    {
      "epoch": 4.065236051502146,
      "grad_norm": 1.1755516529083252,
      "learning_rate": 0.0001444146382235815,
      "loss": 0.4203,
      "step": 1774
    },
    {
      "epoch": 4.067525035765379,
      "grad_norm": 1.295109748840332,
      "learning_rate": 0.00014435822214562704,
      "loss": 0.4168,
      "step": 1775
    },
    {
      "epoch": 4.069814020028613,
      "grad_norm": 1.1494113206863403,
      "learning_rate": 0.0001443017884854906,
      "loss": 0.3503,
      "step": 1776
    },
    {
      "epoch": 4.072103004291845,
      "grad_norm": 1.2138291597366333,
      "learning_rate": 0.00014424533726554065,
      "loss": 0.3473,
      "step": 1777
    },
    {
      "epoch": 4.074391988555079,
      "grad_norm": 1.2661527395248413,
      "learning_rate": 0.00014418886850815267,
      "loss": 0.3394,
      "step": 1778
    },
    {
      "epoch": 4.076680972818312,
      "grad_norm": 1.399715781211853,
      "learning_rate": 0.000144132382235709,
      "loss": 0.3488,
      "step": 1779
    },
    {
      "epoch": 4.078969957081545,
      "grad_norm": 1.4522619247436523,
      "learning_rate": 0.00014407587847059913,
      "loss": 0.2779,
      "step": 1780
    },
    {
      "epoch": 4.081258941344778,
      "grad_norm": 1.2140947580337524,
      "learning_rate": 0.00014401935723521928,
      "loss": 0.2534,
      "step": 1781
    },
    {
      "epoch": 4.083547925608012,
      "grad_norm": 1.366463541984558,
      "learning_rate": 0.0001439628185519726,
      "loss": 0.2425,
      "step": 1782
    },
    {
      "epoch": 4.085836909871245,
      "grad_norm": 1.3253388404846191,
      "learning_rate": 0.00014390626244326925,
      "loss": 0.2125,
      "step": 1783
    },
    {
      "epoch": 4.088125894134478,
      "grad_norm": 1.2972759008407593,
      "learning_rate": 0.00014384968893152634,
      "loss": 0.1956,
      "step": 1784
    },
    {
      "epoch": 4.090414878397711,
      "grad_norm": 1.161211609840393,
      "learning_rate": 0.0001437930980391677,
      "loss": 0.1329,
      "step": 1785
    },
    {
      "epoch": 4.092703862660944,
      "grad_norm": 1.3612005710601807,
      "learning_rate": 0.00014373648978862426,
      "loss": 0.1395,
      "step": 1786
    },
    {
      "epoch": 4.094992846924177,
      "grad_norm": 1.3977785110473633,
      "learning_rate": 0.0001436798642023336,
      "loss": 0.1925,
      "step": 1787
    },
    {
      "epoch": 4.09728183118741,
      "grad_norm": 1.3967665433883667,
      "learning_rate": 0.0001436232213027404,
      "loss": 0.1068,
      "step": 1788
    },
    {
      "epoch": 4.099570815450644,
      "grad_norm": 2.064885139465332,
      "learning_rate": 0.00014356656111229603,
      "loss": 0.1559,
      "step": 1789
    },
    {
      "epoch": 4.101859799713877,
      "grad_norm": 1.9638558626174927,
      "learning_rate": 0.00014350988365345878,
      "loss": 0.1913,
      "step": 1790
    },
    {
      "epoch": 4.10414878397711,
      "grad_norm": 2.244676113128662,
      "learning_rate": 0.00014345318894869383,
      "loss": 0.1793,
      "step": 1791
    },
    {
      "epoch": 4.106437768240343,
      "grad_norm": 2.107057809829712,
      "learning_rate": 0.0001433964770204731,
      "loss": 0.1607,
      "step": 1792
    },
    {
      "epoch": 4.108726752503577,
      "grad_norm": 2.0315301418304443,
      "learning_rate": 0.00014333974789127542,
      "loss": 0.0995,
      "step": 1793
    },
    {
      "epoch": 4.111015736766809,
      "grad_norm": 4.222057819366455,
      "learning_rate": 0.0001432830015835864,
      "loss": 0.2596,
      "step": 1794
    },
    {
      "epoch": 4.113304721030043,
      "grad_norm": 2.639320135116577,
      "learning_rate": 0.00014322623811989843,
      "loss": 0.2111,
      "step": 1795
    },
    {
      "epoch": 4.115593705293276,
      "grad_norm": 0.8525104522705078,
      "learning_rate": 0.0001431694575227108,
      "loss": 0.8591,
      "step": 1796
    },
    {
      "epoch": 4.1178826895565095,
      "grad_norm": 0.6069846153259277,
      "learning_rate": 0.00014311265981452946,
      "loss": 1.7719,
      "step": 1797
    },
    {
      "epoch": 4.120171673819742,
      "grad_norm": 0.6406371593475342,
      "learning_rate": 0.0001430558450178672,
      "loss": 1.6625,
      "step": 1798
    },
    {
      "epoch": 4.122460658082976,
      "grad_norm": 0.7228275537490845,
      "learning_rate": 0.00014299901315524368,
      "loss": 1.5061,
      "step": 1799
    },
    {
      "epoch": 4.124749642346209,
      "grad_norm": 0.7601426243782043,
      "learning_rate": 0.00014294216424918515,
      "loss": 1.4428,
      "step": 1800
    },
    {
      "epoch": 4.127038626609442,
      "grad_norm": 0.7522489428520203,
      "learning_rate": 0.00014288529832222474,
      "loss": 1.3279,
      "step": 1801
    },
    {
      "epoch": 4.129327610872675,
      "grad_norm": 0.7654016613960266,
      "learning_rate": 0.00014282841539690221,
      "loss": 1.2668,
      "step": 1802
    },
    {
      "epoch": 4.131616595135909,
      "grad_norm": 0.7934605479240417,
      "learning_rate": 0.00014277151549576427,
      "loss": 1.331,
      "step": 1803
    },
    {
      "epoch": 4.1339055793991415,
      "grad_norm": 0.8227596879005432,
      "learning_rate": 0.0001427145986413641,
      "loss": 1.2089,
      "step": 1804
    },
    {
      "epoch": 4.136194563662375,
      "grad_norm": 0.7974002361297607,
      "learning_rate": 0.0001426576648562618,
      "loss": 1.2156,
      "step": 1805
    },
    {
      "epoch": 4.138483547925608,
      "grad_norm": 0.8188871741294861,
      "learning_rate": 0.00014260071416302407,
      "loss": 1.1281,
      "step": 1806
    },
    {
      "epoch": 4.140772532188842,
      "grad_norm": 0.8775198459625244,
      "learning_rate": 0.00014254374658422434,
      "loss": 1.1846,
      "step": 1807
    },
    {
      "epoch": 4.143061516452074,
      "grad_norm": 0.8636072278022766,
      "learning_rate": 0.00014248676214244273,
      "loss": 0.969,
      "step": 1808
    },
    {
      "epoch": 4.145350500715308,
      "grad_norm": 0.9492164850234985,
      "learning_rate": 0.0001424297608602661,
      "loss": 1.0989,
      "step": 1809
    },
    {
      "epoch": 4.147639484978541,
      "grad_norm": 0.9339204430580139,
      "learning_rate": 0.0001423727427602879,
      "loss": 1.0414,
      "step": 1810
    },
    {
      "epoch": 4.149928469241774,
      "grad_norm": 0.9688881039619446,
      "learning_rate": 0.0001423157078651083,
      "loss": 1.0163,
      "step": 1811
    },
    {
      "epoch": 4.152217453505007,
      "grad_norm": 0.9714307188987732,
      "learning_rate": 0.0001422586561973341,
      "loss": 0.9926,
      "step": 1812
    },
    {
      "epoch": 4.154506437768241,
      "grad_norm": 1.0081877708435059,
      "learning_rate": 0.00014220158777957876,
      "loss": 0.8907,
      "step": 1813
    },
    {
      "epoch": 4.156795422031474,
      "grad_norm": 1.0111851692199707,
      "learning_rate": 0.00014214450263446234,
      "loss": 0.8518,
      "step": 1814
    },
    {
      "epoch": 4.159084406294706,
      "grad_norm": 1.0751018524169922,
      "learning_rate": 0.00014208740078461166,
      "loss": 0.8703,
      "step": 1815
    },
    {
      "epoch": 4.16137339055794,
      "grad_norm": 1.1573535203933716,
      "learning_rate": 0.00014203028225265999,
      "loss": 0.8156,
      "step": 1816
    },
    {
      "epoch": 4.163662374821173,
      "grad_norm": 1.0764294862747192,
      "learning_rate": 0.00014197314706124729,
      "loss": 0.8363,
      "step": 1817
    },
    {
      "epoch": 4.165951359084406,
      "grad_norm": 1.1808897256851196,
      "learning_rate": 0.00014191599523302018,
      "loss": 0.7065,
      "step": 1818
    },
    {
      "epoch": 4.168240343347639,
      "grad_norm": 1.208436369895935,
      "learning_rate": 0.00014185882679063175,
      "loss": 0.6565,
      "step": 1819
    },
    {
      "epoch": 4.170529327610873,
      "grad_norm": 1.145291805267334,
      "learning_rate": 0.00014180164175674184,
      "loss": 0.6326,
      "step": 1820
    },
    {
      "epoch": 4.172818311874106,
      "grad_norm": 1.1896421909332275,
      "learning_rate": 0.00014174444015401666,
      "loss": 0.6041,
      "step": 1821
    },
    {
      "epoch": 4.175107296137339,
      "grad_norm": 1.1539980173110962,
      "learning_rate": 0.00014168722200512917,
      "loss": 0.5126,
      "step": 1822
    },
    {
      "epoch": 4.177396280400572,
      "grad_norm": 1.2064073085784912,
      "learning_rate": 0.00014162998733275876,
      "loss": 0.547,
      "step": 1823
    },
    {
      "epoch": 4.179685264663806,
      "grad_norm": 1.160438895225525,
      "learning_rate": 0.0001415727361595915,
      "loss": 0.4952,
      "step": 1824
    },
    {
      "epoch": 4.181974248927038,
      "grad_norm": 1.2637494802474976,
      "learning_rate": 0.00014151546850831988,
      "loss": 0.5077,
      "step": 1825
    },
    {
      "epoch": 4.184263233190272,
      "grad_norm": 1.291296362876892,
      "learning_rate": 0.00014145818440164298,
      "loss": 0.4651,
      "step": 1826
    },
    {
      "epoch": 4.186552217453505,
      "grad_norm": 1.1155532598495483,
      "learning_rate": 0.00014140088386226638,
      "loss": 0.3349,
      "step": 1827
    },
    {
      "epoch": 4.1888412017167385,
      "grad_norm": 1.3315963745117188,
      "learning_rate": 0.00014134356691290217,
      "loss": 0.3803,
      "step": 1828
    },
    {
      "epoch": 4.191130185979971,
      "grad_norm": 1.1778143644332886,
      "learning_rate": 0.00014128623357626897,
      "loss": 0.3056,
      "step": 1829
    },
    {
      "epoch": 4.193419170243205,
      "grad_norm": 1.4100370407104492,
      "learning_rate": 0.00014122888387509193,
      "loss": 0.2322,
      "step": 1830
    },
    {
      "epoch": 4.195708154506438,
      "grad_norm": 1.3123832941055298,
      "learning_rate": 0.00014117151783210255,
      "loss": 0.2406,
      "step": 1831
    },
    {
      "epoch": 4.197997138769671,
      "grad_norm": 1.4539953470230103,
      "learning_rate": 0.00014111413547003896,
      "loss": 0.2775,
      "step": 1832
    },
    {
      "epoch": 4.200286123032904,
      "grad_norm": 1.5130772590637207,
      "learning_rate": 0.00014105673681164569,
      "loss": 0.2328,
      "step": 1833
    },
    {
      "epoch": 4.202575107296138,
      "grad_norm": 1.4674046039581299,
      "learning_rate": 0.00014099932187967368,
      "loss": 0.1935,
      "step": 1834
    },
    {
      "epoch": 4.2048640915593705,
      "grad_norm": 1.2991887331008911,
      "learning_rate": 0.00014094189069688046,
      "loss": 0.1563,
      "step": 1835
    },
    {
      "epoch": 4.207153075822604,
      "grad_norm": 1.2331081628799438,
      "learning_rate": 0.00014088444328602985,
      "loss": 0.1224,
      "step": 1836
    },
    {
      "epoch": 4.209442060085837,
      "grad_norm": 1.1798337697982788,
      "learning_rate": 0.00014082697966989223,
      "loss": 0.126,
      "step": 1837
    },
    {
      "epoch": 4.2117310443490705,
      "grad_norm": 1.1834417581558228,
      "learning_rate": 0.00014076949987124423,
      "loss": 0.1049,
      "step": 1838
    },
    {
      "epoch": 4.214020028612303,
      "grad_norm": 1.417945146560669,
      "learning_rate": 0.00014071200391286914,
      "loss": 0.1426,
      "step": 1839
    },
    {
      "epoch": 4.216309012875536,
      "grad_norm": 1.5263046026229858,
      "learning_rate": 0.00014065449181755642,
      "loss": 0.1735,
      "step": 1840
    },
    {
      "epoch": 4.21859799713877,
      "grad_norm": 2.0970356464385986,
      "learning_rate": 0.00014059696360810208,
      "loss": 0.1309,
      "step": 1841
    },
    {
      "epoch": 4.2208869814020025,
      "grad_norm": 2.2149500846862793,
      "learning_rate": 0.00014053941930730846,
      "loss": 0.1119,
      "step": 1842
    },
    {
      "epoch": 4.223175965665236,
      "grad_norm": 2.0250422954559326,
      "learning_rate": 0.00014048185893798424,
      "loss": 0.122,
      "step": 1843
    },
    {
      "epoch": 4.225464949928469,
      "grad_norm": 3.0951108932495117,
      "learning_rate": 0.00014042428252294457,
      "loss": 0.1005,
      "step": 1844
    },
    {
      "epoch": 4.2277539341917025,
      "grad_norm": 2.9070775508880615,
      "learning_rate": 0.0001403666900850109,
      "loss": 0.2827,
      "step": 1845
    },
    {
      "epoch": 4.230042918454935,
      "grad_norm": 0.9913147687911987,
      "learning_rate": 0.00014030908164701098,
      "loss": 0.9888,
      "step": 1846
    },
    {
      "epoch": 4.232331902718169,
      "grad_norm": 0.5928571224212646,
      "learning_rate": 0.00014025145723177895,
      "loss": 1.6881,
      "step": 1847
    },
    {
      "epoch": 4.234620886981402,
      "grad_norm": 0.6608178019523621,
      "learning_rate": 0.00014019381686215535,
      "loss": 1.7316,
      "step": 1848
    },
    {
      "epoch": 4.236909871244635,
      "grad_norm": 0.6736155152320862,
      "learning_rate": 0.00014013616056098694,
      "loss": 1.452,
      "step": 1849
    },
    {
      "epoch": 4.239198855507868,
      "grad_norm": 0.7292165160179138,
      "learning_rate": 0.00014007848835112687,
      "loss": 1.583,
      "step": 1850
    },
    {
      "epoch": 4.241487839771102,
      "grad_norm": 0.7763010263442993,
      "learning_rate": 0.00014002080025543451,
      "loss": 1.4831,
      "step": 1851
    },
    {
      "epoch": 4.2437768240343345,
      "grad_norm": 0.827728271484375,
      "learning_rate": 0.00013996309629677562,
      "loss": 1.52,
      "step": 1852
    },
    {
      "epoch": 4.246065808297568,
      "grad_norm": 0.8173840641975403,
      "learning_rate": 0.00013990537649802217,
      "loss": 1.31,
      "step": 1853
    },
    {
      "epoch": 4.248354792560801,
      "grad_norm": 0.8371086120605469,
      "learning_rate": 0.00013984764088205247,
      "loss": 1.259,
      "step": 1854
    },
    {
      "epoch": 4.250643776824035,
      "grad_norm": 0.8675817251205444,
      "learning_rate": 0.00013978988947175108,
      "loss": 1.2215,
      "step": 1855
    },
    {
      "epoch": 4.252932761087267,
      "grad_norm": 0.9176662564277649,
      "learning_rate": 0.0001397321222900088,
      "loss": 1.3494,
      "step": 1856
    },
    {
      "epoch": 4.255221745350501,
      "grad_norm": 0.8640986084938049,
      "learning_rate": 0.00013967433935972267,
      "loss": 0.9971,
      "step": 1857
    },
    {
      "epoch": 4.257510729613734,
      "grad_norm": 0.9197400212287903,
      "learning_rate": 0.00013961654070379604,
      "loss": 1.3061,
      "step": 1858
    },
    {
      "epoch": 4.259799713876967,
      "grad_norm": 0.9120035767555237,
      "learning_rate": 0.00013955872634513844,
      "loss": 0.9622,
      "step": 1859
    },
    {
      "epoch": 4.2620886981402,
      "grad_norm": 0.977082371711731,
      "learning_rate": 0.00013950089630666564,
      "loss": 1.1379,
      "step": 1860
    },
    {
      "epoch": 4.264377682403434,
      "grad_norm": 1.0064245462417603,
      "learning_rate": 0.00013944305061129957,
      "loss": 1.1149,
      "step": 1861
    },
    {
      "epoch": 4.266666666666667,
      "grad_norm": 1.0343989133834839,
      "learning_rate": 0.0001393851892819685,
      "loss": 1.0638,
      "step": 1862
    },
    {
      "epoch": 4.2689556509299,
      "grad_norm": 0.9780799746513367,
      "learning_rate": 0.00013932731234160675,
      "loss": 0.8629,
      "step": 1863
    },
    {
      "epoch": 4.271244635193133,
      "grad_norm": 1.014587163925171,
      "learning_rate": 0.00013926941981315486,
      "loss": 0.9844,
      "step": 1864
    },
    {
      "epoch": 4.273533619456366,
      "grad_norm": 1.0625803470611572,
      "learning_rate": 0.0001392115117195597,
      "loss": 0.8582,
      "step": 1865
    },
    {
      "epoch": 4.275822603719599,
      "grad_norm": 1.1477220058441162,
      "learning_rate": 0.00013915358808377405,
      "loss": 0.8825,
      "step": 1866
    },
    {
      "epoch": 4.278111587982833,
      "grad_norm": 1.1152499914169312,
      "learning_rate": 0.0001390956489287571,
      "loss": 0.6892,
      "step": 1867
    },
    {
      "epoch": 4.280400572246066,
      "grad_norm": 1.157660961151123,
      "learning_rate": 0.000139037694277474,
      "loss": 0.7486,
      "step": 1868
    },
    {
      "epoch": 4.282689556509299,
      "grad_norm": 1.1074392795562744,
      "learning_rate": 0.00013897972415289616,
      "loss": 0.6215,
      "step": 1869
    },
    {
      "epoch": 4.284978540772532,
      "grad_norm": 1.1611192226409912,
      "learning_rate": 0.00013892173857800108,
      "loss": 0.6177,
      "step": 1870
    },
    {
      "epoch": 4.287267525035765,
      "grad_norm": 1.1261109113693237,
      "learning_rate": 0.00013886373757577238,
      "loss": 0.5579,
      "step": 1871
    },
    {
      "epoch": 4.289556509298999,
      "grad_norm": 1.1944984197616577,
      "learning_rate": 0.00013880572116919983,
      "loss": 0.5132,
      "step": 1872
    },
    {
      "epoch": 4.291845493562231,
      "grad_norm": 1.183237075805664,
      "learning_rate": 0.00013874768938127924,
      "loss": 0.474,
      "step": 1873
    },
    {
      "epoch": 4.294134477825465,
      "grad_norm": 1.2248847484588623,
      "learning_rate": 0.00013868964223501262,
      "loss": 0.437,
      "step": 1874
    },
    {
      "epoch": 4.296423462088698,
      "grad_norm": 1.3775705099105835,
      "learning_rate": 0.00013863157975340791,
      "loss": 0.4138,
      "step": 1875
    },
    {
      "epoch": 4.2987124463519315,
      "grad_norm": 1.3187086582183838,
      "learning_rate": 0.00013857350195947932,
      "loss": 0.3899,
      "step": 1876
    },
    {
      "epoch": 4.301001430615164,
      "grad_norm": 1.2714444398880005,
      "learning_rate": 0.00013851540887624693,
      "loss": 0.3745,
      "step": 1877
    },
    {
      "epoch": 4.303290414878398,
      "grad_norm": 1.565042495727539,
      "learning_rate": 0.0001384573005267371,
      "loss": 0.3452,
      "step": 1878
    },
    {
      "epoch": 4.305579399141631,
      "grad_norm": 1.5059938430786133,
      "learning_rate": 0.00013839917693398202,
      "loss": 0.3695,
      "step": 1879
    },
    {
      "epoch": 4.307868383404864,
      "grad_norm": 1.5211122035980225,
      "learning_rate": 0.00013834103812102012,
      "loss": 0.3363,
      "step": 1880
    },
    {
      "epoch": 4.310157367668097,
      "grad_norm": 1.726873755455017,
      "learning_rate": 0.00013828288411089565,
      "loss": 0.2977,
      "step": 1881
    },
    {
      "epoch": 4.312446351931331,
      "grad_norm": 1.762004017829895,
      "learning_rate": 0.0001382247149266591,
      "loss": 0.289,
      "step": 1882
    },
    {
      "epoch": 4.3147353361945635,
      "grad_norm": 1.7233407497406006,
      "learning_rate": 0.00013816653059136683,
      "loss": 0.2772,
      "step": 1883
    },
    {
      "epoch": 4.317024320457797,
      "grad_norm": 1.2020918130874634,
      "learning_rate": 0.0001381083311280813,
      "loss": 0.1716,
      "step": 1884
    },
    {
      "epoch": 4.31931330472103,
      "grad_norm": 1.3072203397750854,
      "learning_rate": 0.00013805011655987083,
      "loss": 0.1881,
      "step": 1885
    },
    {
      "epoch": 4.321602288984264,
      "grad_norm": 1.6461601257324219,
      "learning_rate": 0.00013799188690980988,
      "loss": 0.2338,
      "step": 1886
    },
    {
      "epoch": 4.323891273247496,
      "grad_norm": 1.901731252670288,
      "learning_rate": 0.00013793364220097883,
      "loss": 0.2204,
      "step": 1887
    },
    {
      "epoch": 4.32618025751073,
      "grad_norm": 1.323927402496338,
      "learning_rate": 0.000137875382456464,
      "loss": 0.1716,
      "step": 1888
    },
    {
      "epoch": 4.328469241773963,
      "grad_norm": 1.318861722946167,
      "learning_rate": 0.00013781710769935772,
      "loss": 0.1424,
      "step": 1889
    },
    {
      "epoch": 4.330758226037196,
      "grad_norm": 1.673195719718933,
      "learning_rate": 0.00013775881795275816,
      "loss": 0.1442,
      "step": 1890
    },
    {
      "epoch": 4.333047210300429,
      "grad_norm": 1.337529182434082,
      "learning_rate": 0.0001377005132397696,
      "loss": 0.1543,
      "step": 1891
    },
    {
      "epoch": 4.335336194563663,
      "grad_norm": 2.0561656951904297,
      "learning_rate": 0.00013764219358350211,
      "loss": 0.1759,
      "step": 1892
    },
    {
      "epoch": 4.337625178826896,
      "grad_norm": 1.5654786825180054,
      "learning_rate": 0.0001375838590070718,
      "loss": 0.0971,
      "step": 1893
    },
    {
      "epoch": 4.339914163090128,
      "grad_norm": 3.0831820964813232,
      "learning_rate": 0.00013752550953360056,
      "loss": 0.0735,
      "step": 1894
    },
    {
      "epoch": 4.342203147353362,
      "grad_norm": 2.8486666679382324,
      "learning_rate": 0.0001374671451862163,
      "loss": 0.1149,
      "step": 1895
    },
    {
      "epoch": 4.344492131616595,
      "grad_norm": 1.0494810342788696,
      "learning_rate": 0.00013740876598805277,
      "loss": 0.9351,
      "step": 1896
    },
    {
      "epoch": 4.346781115879828,
      "grad_norm": 0.6020961999893188,
      "learning_rate": 0.00013735037196224963,
      "loss": 1.7943,
      "step": 1897
    },
    {
      "epoch": 4.349070100143061,
      "grad_norm": 0.6643849015235901,
      "learning_rate": 0.0001372919631319524,
      "loss": 1.7295,
      "step": 1898
    },
    {
      "epoch": 4.351359084406295,
      "grad_norm": 0.7122296094894409,
      "learning_rate": 0.00013723353952031246,
      "loss": 1.4286,
      "step": 1899
    },
    {
      "epoch": 4.353648068669528,
      "grad_norm": 0.7596356868743896,
      "learning_rate": 0.0001371751011504871,
      "loss": 1.4306,
      "step": 1900
    },
    {
      "epoch": 4.355937052932761,
      "grad_norm": 0.7785231471061707,
      "learning_rate": 0.00013711664804563935,
      "loss": 1.4756,
      "step": 1901
    },
    {
      "epoch": 4.358226037195994,
      "grad_norm": 0.7578595280647278,
      "learning_rate": 0.00013705818022893826,
      "loss": 1.2556,
      "step": 1902
    },
    {
      "epoch": 4.360515021459228,
      "grad_norm": 0.8627766966819763,
      "learning_rate": 0.00013699969772355852,
      "loss": 1.4254,
      "step": 1903
    },
    {
      "epoch": 4.36280400572246,
      "grad_norm": 0.8474234938621521,
      "learning_rate": 0.00013694120055268076,
      "loss": 1.2509,
      "step": 1904
    },
    {
      "epoch": 4.365092989985694,
      "grad_norm": 0.8625220656394958,
      "learning_rate": 0.00013688268873949134,
      "loss": 1.0913,
      "step": 1905
    },
    {
      "epoch": 4.367381974248927,
      "grad_norm": 0.9123669862747192,
      "learning_rate": 0.00013682416230718258,
      "loss": 1.085,
      "step": 1906
    },
    {
      "epoch": 4.3696709585121605,
      "grad_norm": 0.8739493489265442,
      "learning_rate": 0.00013676562127895237,
      "loss": 1.0619,
      "step": 1907
    },
    {
      "epoch": 4.371959942775393,
      "grad_norm": 0.9685913920402527,
      "learning_rate": 0.0001367070656780046,
      "loss": 1.2055,
      "step": 1908
    },
    {
      "epoch": 4.374248927038627,
      "grad_norm": 0.9431620836257935,
      "learning_rate": 0.00013664849552754877,
      "loss": 1.1056,
      "step": 1909
    },
    {
      "epoch": 4.37653791130186,
      "grad_norm": 0.9369290471076965,
      "learning_rate": 0.00013658991085080025,
      "loss": 1.0644,
      "step": 1910
    },
    {
      "epoch": 4.378826895565093,
      "grad_norm": 0.9952932000160217,
      "learning_rate": 0.00013653131167098013,
      "loss": 0.9128,
      "step": 1911
    },
    {
      "epoch": 4.381115879828326,
      "grad_norm": 1.0159112215042114,
      "learning_rate": 0.00013647269801131523,
      "loss": 0.9172,
      "step": 1912
    },
    {
      "epoch": 4.38340486409156,
      "grad_norm": 1.0917634963989258,
      "learning_rate": 0.00013641406989503815,
      "loss": 0.9761,
      "step": 1913
    },
    {
      "epoch": 4.3856938483547925,
      "grad_norm": 1.0909875631332397,
      "learning_rate": 0.00013635542734538725,
      "loss": 1.0508,
      "step": 1914
    },
    {
      "epoch": 4.387982832618026,
      "grad_norm": 1.0644196271896362,
      "learning_rate": 0.0001362967703856065,
      "loss": 0.8636,
      "step": 1915
    },
    {
      "epoch": 4.390271816881259,
      "grad_norm": 1.0008410215377808,
      "learning_rate": 0.00013623809903894562,
      "loss": 0.7,
      "step": 1916
    },
    {
      "epoch": 4.3925608011444925,
      "grad_norm": 1.0553101301193237,
      "learning_rate": 0.00013617941332866011,
      "loss": 0.903,
      "step": 1917
    },
    {
      "epoch": 4.394849785407725,
      "grad_norm": 1.1645140647888184,
      "learning_rate": 0.00013612071327801114,
      "loss": 0.8311,
      "step": 1918
    },
    {
      "epoch": 4.397138769670958,
      "grad_norm": 1.2693705558776855,
      "learning_rate": 0.00013606199891026544,
      "loss": 0.8169,
      "step": 1919
    },
    {
      "epoch": 4.399427753934192,
      "grad_norm": 1.233024001121521,
      "learning_rate": 0.0001360032702486956,
      "loss": 0.6266,
      "step": 1920
    },
    {
      "epoch": 4.4017167381974245,
      "grad_norm": 1.209771752357483,
      "learning_rate": 0.0001359445273165797,
      "loss": 0.5964,
      "step": 1921
    },
    {
      "epoch": 4.404005722460658,
      "grad_norm": 1.2486642599105835,
      "learning_rate": 0.00013588577013720166,
      "loss": 0.5071,
      "step": 1922
    },
    {
      "epoch": 4.406294706723891,
      "grad_norm": 1.131710171699524,
      "learning_rate": 0.0001358269987338509,
      "loss": 0.4773,
      "step": 1923
    },
    {
      "epoch": 4.4085836909871245,
      "grad_norm": 1.1338353157043457,
      "learning_rate": 0.0001357682131298225,
      "loss": 0.4969,
      "step": 1924
    },
    {
      "epoch": 4.410872675250357,
      "grad_norm": 1.1553444862365723,
      "learning_rate": 0.00013570941334841723,
      "loss": 0.5181,
      "step": 1925
    },
    {
      "epoch": 4.413161659513591,
      "grad_norm": 1.2361962795257568,
      "learning_rate": 0.00013565059941294142,
      "loss": 0.4904,
      "step": 1926
    },
    {
      "epoch": 4.415450643776824,
      "grad_norm": 1.3251562118530273,
      "learning_rate": 0.00013559177134670706,
      "loss": 0.4725,
      "step": 1927
    },
    {
      "epoch": 4.417739628040057,
      "grad_norm": 1.3303446769714355,
      "learning_rate": 0.0001355329291730317,
      "loss": 0.363,
      "step": 1928
    },
    {
      "epoch": 4.42002861230329,
      "grad_norm": 1.414873719215393,
      "learning_rate": 0.00013547407291523853,
      "loss": 0.3727,
      "step": 1929
    },
    {
      "epoch": 4.422317596566524,
      "grad_norm": 1.3383206129074097,
      "learning_rate": 0.0001354152025966562,
      "loss": 0.3201,
      "step": 1930
    },
    {
      "epoch": 4.4246065808297566,
      "grad_norm": 1.3916726112365723,
      "learning_rate": 0.00013535631824061913,
      "loss": 0.2309,
      "step": 1931
    },
    {
      "epoch": 4.42689556509299,
      "grad_norm": 1.3709685802459717,
      "learning_rate": 0.00013529741987046716,
      "loss": 0.2759,
      "step": 1932
    },
    {
      "epoch": 4.429184549356223,
      "grad_norm": 1.3704681396484375,
      "learning_rate": 0.0001352385075095457,
      "loss": 0.2658,
      "step": 1933
    },
    {
      "epoch": 4.431473533619457,
      "grad_norm": 1.756007432937622,
      "learning_rate": 0.00013517958118120576,
      "loss": 0.2528,
      "step": 1934
    },
    {
      "epoch": 4.433762517882689,
      "grad_norm": 1.0347654819488525,
      "learning_rate": 0.0001351206409088038,
      "loss": 0.1138,
      "step": 1935
    },
    {
      "epoch": 4.436051502145923,
      "grad_norm": 1.4181766510009766,
      "learning_rate": 0.00013506168671570192,
      "loss": 0.1408,
      "step": 1936
    },
    {
      "epoch": 4.438340486409156,
      "grad_norm": 1.846653699874878,
      "learning_rate": 0.00013500271862526764,
      "loss": 0.2026,
      "step": 1937
    },
    {
      "epoch": 4.4406294706723894,
      "grad_norm": 1.4231027364730835,
      "learning_rate": 0.0001349437366608741,
      "loss": 0.1554,
      "step": 1938
    },
    {
      "epoch": 4.442918454935622,
      "grad_norm": 1.0651813745498657,
      "learning_rate": 0.00013488474084589976,
      "loss": 0.114,
      "step": 1939
    },
    {
      "epoch": 4.445207439198856,
      "grad_norm": 1.6582481861114502,
      "learning_rate": 0.00013482573120372873,
      "loss": 0.1301,
      "step": 1940
    },
    {
      "epoch": 4.447496423462089,
      "grad_norm": 1.7687350511550903,
      "learning_rate": 0.00013476670775775052,
      "loss": 0.1559,
      "step": 1941
    },
    {
      "epoch": 4.449785407725322,
      "grad_norm": 1.3473821878433228,
      "learning_rate": 0.0001347076705313602,
      "loss": 0.0956,
      "step": 1942
    },
    {
      "epoch": 4.452074391988555,
      "grad_norm": 1.5209802389144897,
      "learning_rate": 0.0001346486195479582,
      "loss": 0.0777,
      "step": 1943
    },
    {
      "epoch": 4.454363376251788,
      "grad_norm": 2.565901279449463,
      "learning_rate": 0.00013458955483095042,
      "loss": 0.0962,
      "step": 1944
    },
    {
      "epoch": 4.4566523605150214,
      "grad_norm": 1.7184423208236694,
      "learning_rate": 0.00013453047640374827,
      "loss": 0.1249,
      "step": 1945
    },
    {
      "epoch": 4.458941344778255,
      "grad_norm": 1.5781924724578857,
      "learning_rate": 0.00013447138428976854,
      "loss": 1.0313,
      "step": 1946
    },
    {
      "epoch": 4.461230329041488,
      "grad_norm": 0.6325271129608154,
      "learning_rate": 0.0001344122785124334,
      "loss": 1.7381,
      "step": 1947
    },
    {
      "epoch": 4.463519313304721,
      "grad_norm": 0.6897770762443542,
      "learning_rate": 0.00013435315909517058,
      "loss": 1.7595,
      "step": 1948
    },
    {
      "epoch": 4.465808297567954,
      "grad_norm": 0.7247394323348999,
      "learning_rate": 0.0001342940260614131,
      "loss": 1.5707,
      "step": 1949
    },
    {
      "epoch": 4.468097281831187,
      "grad_norm": 0.7909899353981018,
      "learning_rate": 0.00013423487943459934,
      "loss": 1.4698,
      "step": 1950
    },
    {
      "epoch": 4.470386266094421,
      "grad_norm": 0.8112832903862,
      "learning_rate": 0.00013417571923817322,
      "loss": 1.4059,
      "step": 1951
    },
    {
      "epoch": 4.4726752503576535,
      "grad_norm": 0.8589115142822266,
      "learning_rate": 0.0001341165454955839,
      "loss": 1.5253,
      "step": 1952
    },
    {
      "epoch": 4.474964234620887,
      "grad_norm": 0.8326034545898438,
      "learning_rate": 0.00013405735823028602,
      "loss": 1.2412,
      "step": 1953
    },
    {
      "epoch": 4.47725321888412,
      "grad_norm": 0.8572644591331482,
      "learning_rate": 0.00013399815746573945,
      "loss": 1.3859,
      "step": 1954
    },
    {
      "epoch": 4.4795422031473535,
      "grad_norm": 0.899202287197113,
      "learning_rate": 0.0001339389432254095,
      "loss": 1.3033,
      "step": 1955
    },
    {
      "epoch": 4.481831187410586,
      "grad_norm": 0.9051918983459473,
      "learning_rate": 0.00013387971553276682,
      "loss": 1.2522,
      "step": 1956
    },
    {
      "epoch": 4.48412017167382,
      "grad_norm": 0.8995559811592102,
      "learning_rate": 0.0001338204744112874,
      "loss": 1.1044,
      "step": 1957
    },
    {
      "epoch": 4.486409155937053,
      "grad_norm": 0.9207310676574707,
      "learning_rate": 0.00013376121988445248,
      "loss": 1.0933,
      "step": 1958
    },
    {
      "epoch": 4.488698140200286,
      "grad_norm": 0.9572367072105408,
      "learning_rate": 0.00013370195197574865,
      "loss": 1.0125,
      "step": 1959
    },
    {
      "epoch": 4.490987124463519,
      "grad_norm": 0.9588605761528015,
      "learning_rate": 0.00013364267070866786,
      "loss": 1.0196,
      "step": 1960
    },
    {
      "epoch": 4.493276108726753,
      "grad_norm": 1.030645489692688,
      "learning_rate": 0.0001335833761067073,
      "loss": 1.1733,
      "step": 1961
    },
    {
      "epoch": 4.4955650929899855,
      "grad_norm": 1.0392740964889526,
      "learning_rate": 0.00013352406819336945,
      "loss": 1.0174,
      "step": 1962
    },
    {
      "epoch": 4.497854077253219,
      "grad_norm": 1.0212891101837158,
      "learning_rate": 0.00013346474699216208,
      "loss": 0.9613,
      "step": 1963
    },
    {
      "epoch": 4.500143061516452,
      "grad_norm": 1.131136417388916,
      "learning_rate": 0.00013340541252659818,
      "loss": 1.0567,
      "step": 1964
    },
    {
      "epoch": 4.502432045779686,
      "grad_norm": 1.0588047504425049,
      "learning_rate": 0.00013334606482019607,
      "loss": 0.7574,
      "step": 1965
    },
    {
      "epoch": 4.504721030042918,
      "grad_norm": 1.5936955213546753,
      "learning_rate": 0.0001332867038964793,
      "loss": 0.7497,
      "step": 1966
    },
    {
      "epoch": 4.507010014306152,
      "grad_norm": 1.227865219116211,
      "learning_rate": 0.00013322732977897662,
      "loss": 0.946,
      "step": 1967
    },
    {
      "epoch": 4.509298998569385,
      "grad_norm": 1.226654291152954,
      "learning_rate": 0.00013316794249122206,
      "loss": 0.7259,
      "step": 1968
    },
    {
      "epoch": 4.5115879828326175,
      "grad_norm": 1.1799983978271484,
      "learning_rate": 0.0001331085420567548,
      "loss": 0.6394,
      "step": 1969
    },
    {
      "epoch": 4.513876967095851,
      "grad_norm": 1.2049132585525513,
      "learning_rate": 0.00013304912849911927,
      "loss": 0.6035,
      "step": 1970
    },
    {
      "epoch": 4.516165951359085,
      "grad_norm": 1.347949743270874,
      "learning_rate": 0.00013298970184186514,
      "loss": 0.6276,
      "step": 1971
    },
    {
      "epoch": 4.518454935622318,
      "grad_norm": 1.2323335409164429,
      "learning_rate": 0.00013293026210854726,
      "loss": 0.5482,
      "step": 1972
    },
    {
      "epoch": 4.52074391988555,
      "grad_norm": 1.2629224061965942,
      "learning_rate": 0.0001328708093227256,
      "loss": 0.587,
      "step": 1973
    },
    {
      "epoch": 4.523032904148784,
      "grad_norm": 1.4115405082702637,
      "learning_rate": 0.00013281134350796539,
      "loss": 0.5042,
      "step": 1974
    },
    {
      "epoch": 4.525321888412018,
      "grad_norm": 1.3867058753967285,
      "learning_rate": 0.00013275186468783687,
      "loss": 0.5705,
      "step": 1975
    },
    {
      "epoch": 4.52761087267525,
      "grad_norm": 1.5213968753814697,
      "learning_rate": 0.00013269237288591565,
      "loss": 0.5916,
      "step": 1976
    },
    {
      "epoch": 4.529899856938483,
      "grad_norm": 1.5112426280975342,
      "learning_rate": 0.00013263286812578235,
      "loss": 0.4245,
      "step": 1977
    },
    {
      "epoch": 4.532188841201717,
      "grad_norm": 1.2434872388839722,
      "learning_rate": 0.00013257335043102277,
      "loss": 0.3977,
      "step": 1978
    },
    {
      "epoch": 4.53447782546495,
      "grad_norm": 1.2847955226898193,
      "learning_rate": 0.0001325138198252278,
      "loss": 0.3967,
      "step": 1979
    },
    {
      "epoch": 4.536766809728183,
      "grad_norm": 1.3733184337615967,
      "learning_rate": 0.00013245427633199345,
      "loss": 0.3502,
      "step": 1980
    },
    {
      "epoch": 4.539055793991416,
      "grad_norm": 1.3960249423980713,
      "learning_rate": 0.00013239471997492092,
      "loss": 0.3305,
      "step": 1981
    },
    {
      "epoch": 4.54134477825465,
      "grad_norm": 1.3462896347045898,
      "learning_rate": 0.0001323351507776164,
      "loss": 0.2457,
      "step": 1982
    },
    {
      "epoch": 4.543633762517882,
      "grad_norm": 1.4513142108917236,
      "learning_rate": 0.00013227556876369123,
      "loss": 0.2284,
      "step": 1983
    },
    {
      "epoch": 4.545922746781116,
      "grad_norm": 1.4674392938613892,
      "learning_rate": 0.0001322159739567618,
      "loss": 0.2691,
      "step": 1984
    },
    {
      "epoch": 4.548211731044349,
      "grad_norm": 1.4074273109436035,
      "learning_rate": 0.0001321563663804496,
      "loss": 0.1861,
      "step": 1985
    },
    {
      "epoch": 4.5505007153075825,
      "grad_norm": 1.080726146697998,
      "learning_rate": 0.00013209674605838117,
      "loss": 0.0974,
      "step": 1986
    },
    {
      "epoch": 4.552789699570815,
      "grad_norm": 1.5537816286087036,
      "learning_rate": 0.00013203711301418815,
      "loss": 0.1991,
      "step": 1987
    },
    {
      "epoch": 4.555078683834049,
      "grad_norm": 1.4420602321624756,
      "learning_rate": 0.00013197746727150708,
      "loss": 0.1775,
      "step": 1988
    },
    {
      "epoch": 4.557367668097282,
      "grad_norm": 1.2284140586853027,
      "learning_rate": 0.0001319178088539797,
      "loss": 0.1238,
      "step": 1989
    },
    {
      "epoch": 4.559656652360515,
      "grad_norm": 1.7686671018600464,
      "learning_rate": 0.00013185813778525263,
      "loss": 0.1987,
      "step": 1990
    },
    {
      "epoch": 4.561945636623748,
      "grad_norm": 2.3643531799316406,
      "learning_rate": 0.00013179845408897766,
      "loss": 0.1214,
      "step": 1991
    },
    {
      "epoch": 4.564234620886982,
      "grad_norm": 1.6342504024505615,
      "learning_rate": 0.00013173875778881147,
      "loss": 0.0951,
      "step": 1992
    },
    {
      "epoch": 4.5665236051502145,
      "grad_norm": 1.2456626892089844,
      "learning_rate": 0.0001316790489084157,
      "loss": 0.0993,
      "step": 1993
    },
    {
      "epoch": 4.568812589413448,
      "grad_norm": 3.6577072143554688,
      "learning_rate": 0.00013161932747145712,
      "loss": 0.1036,
      "step": 1994
    },
    {
      "epoch": 4.571101573676681,
      "grad_norm": 2.6007823944091797,
      "learning_rate": 0.00013155959350160738,
      "loss": 0.1888,
      "step": 1995
    },
    {
      "epoch": 4.573390557939915,
      "grad_norm": 4.163236618041992,
      "learning_rate": 0.00013149984702254312,
      "loss": 0.9818,
      "step": 1996
    },
    {
      "epoch": 4.575679542203147,
      "grad_norm": 0.6608015298843384,
      "learning_rate": 0.00013144008805794592,
      "loss": 1.9279,
      "step": 1997
    },
    {
      "epoch": 4.57796852646638,
      "grad_norm": 0.6357399225234985,
      "learning_rate": 0.00013138031663150235,
      "loss": 1.601,
      "step": 1998
    },
    {
      "epoch": 4.580257510729614,
      "grad_norm": 0.7199594974517822,
      "learning_rate": 0.00013132053276690385,
      "loss": 1.8058,
      "step": 1999
    },
    {
      "epoch": 4.582546494992847,
      "grad_norm": 0.774132490158081,
      "learning_rate": 0.00013126073648784688,
      "loss": 1.5383,
      "step": 2000
    },
    {
      "epoch": 4.582546494992847,
      "eval_loss": 2.1012461185455322,
      "eval_runtime": 24.8502,
      "eval_samples_per_second": 29.617,
      "eval_steps_per_second": 14.809,
      "step": 2000
    }
  ],
  "logging_steps": 1,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 12,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.4327953570953953e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}