{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9987760097919217,
  "eval_steps": 500,
  "global_step": 272,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0036719706242350062,
      "grad_norm": 0.03811819851398468,
      "learning_rate": 3.5714285714285714e-06,
      "loss": 0.4568,
      "step": 1
    },
    {
      "epoch": 0.0073439412484700125,
      "grad_norm": 0.037763386964797974,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.7457,
      "step": 2
    },
    {
      "epoch": 0.011015911872705019,
      "grad_norm": 0.03825872391462326,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 0.3669,
      "step": 3
    },
    {
      "epoch": 0.014687882496940025,
      "grad_norm": 0.03575674816966057,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.4611,
      "step": 4
    },
    {
      "epoch": 0.01835985312117503,
      "grad_norm": 0.048165928572416306,
      "learning_rate": 1.785714285714286e-05,
      "loss": 0.3727,
      "step": 5
    },
    {
      "epoch": 0.022031823745410038,
      "grad_norm": 0.042961642146110535,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.3609,
      "step": 6
    },
    {
      "epoch": 0.025703794369645042,
      "grad_norm": 0.10260031372308731,
      "learning_rate": 2.5e-05,
      "loss": 0.8697,
      "step": 7
    },
    {
      "epoch": 0.02937576499388005,
      "grad_norm": 0.06274840980768204,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.521,
      "step": 8
    },
    {
      "epoch": 0.033047735618115054,
      "grad_norm": 0.10686090588569641,
      "learning_rate": 3.2142857142857144e-05,
      "loss": 0.676,
      "step": 9
    },
    {
      "epoch": 0.03671970624235006,
      "grad_norm": 0.05962346866726875,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.306,
      "step": 10
    },
    {
      "epoch": 0.04039167686658507,
      "grad_norm": 0.07848496735095978,
      "learning_rate": 3.928571428571429e-05,
      "loss": 0.5941,
      "step": 11
    },
    {
      "epoch": 0.044063647490820076,
      "grad_norm": 0.042250536382198334,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.412,
      "step": 12
    },
    {
      "epoch": 0.04773561811505508,
      "grad_norm": 0.07003479450941086,
      "learning_rate": 4.642857142857143e-05,
      "loss": 0.4988,
      "step": 13
    },
    {
      "epoch": 0.051407588739290085,
      "grad_norm": 0.05183990299701691,
      "learning_rate": 5e-05,
      "loss": 0.3337,
      "step": 14
    },
    {
      "epoch": 0.05507955936352509,
      "grad_norm": 0.04867109656333923,
      "learning_rate": 5.3571428571428575e-05,
      "loss": 0.424,
      "step": 15
    },
    {
      "epoch": 0.0587515299877601,
      "grad_norm": 0.04919494688510895,
      "learning_rate": 5.714285714285714e-05,
      "loss": 0.3674,
      "step": 16
    },
    {
      "epoch": 0.06242350061199511,
      "grad_norm": 0.07788144797086716,
      "learning_rate": 6.0714285714285715e-05,
      "loss": 0.4244,
      "step": 17
    },
    {
      "epoch": 0.06609547123623011,
      "grad_norm": 0.08719762414693832,
      "learning_rate": 6.428571428571429e-05,
      "loss": 0.3832,
      "step": 18
    },
    {
      "epoch": 0.06976744186046512,
      "grad_norm": 0.16575194895267487,
      "learning_rate": 6.785714285714286e-05,
      "loss": 0.7971,
      "step": 19
    },
    {
      "epoch": 0.07343941248470012,
      "grad_norm": 0.08812964707612991,
      "learning_rate": 7.142857142857143e-05,
      "loss": 0.267,
      "step": 20
    },
    {
      "epoch": 0.07711138310893513,
      "grad_norm": 0.20091792941093445,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.4959,
      "step": 21
    },
    {
      "epoch": 0.08078335373317014,
      "grad_norm": 0.30549925565719604,
      "learning_rate": 7.857142857142858e-05,
      "loss": 0.6555,
      "step": 22
    },
    {
      "epoch": 0.08445532435740515,
      "grad_norm": 0.05145212262868881,
      "learning_rate": 8.214285714285714e-05,
      "loss": 0.365,
      "step": 23
    },
    {
      "epoch": 0.08812729498164015,
      "grad_norm": 0.04592301696538925,
      "learning_rate": 8.571428571428571e-05,
      "loss": 0.3245,
      "step": 24
    },
    {
      "epoch": 0.09179926560587515,
      "grad_norm": 0.13516275584697723,
      "learning_rate": 8.92857142857143e-05,
      "loss": 0.7651,
      "step": 25
    },
    {
      "epoch": 0.09547123623011015,
      "grad_norm": 0.10254071652889252,
      "learning_rate": 9.285714285714286e-05,
      "loss": 0.4515,
      "step": 26
    },
    {
      "epoch": 0.09914320685434516,
      "grad_norm": 0.10743008553981781,
      "learning_rate": 9.642857142857143e-05,
      "loss": 0.853,
      "step": 27
    },
    {
      "epoch": 0.10281517747858017,
      "grad_norm": 0.06928564608097076,
      "learning_rate": 0.0001,
      "loss": 0.5184,
      "step": 28
    },
    {
      "epoch": 0.10648714810281518,
      "grad_norm": 0.11129824072122574,
      "learning_rate": 9.999585567217439e-05,
      "loss": 0.4656,
      "step": 29
    },
    {
      "epoch": 0.11015911872705018,
      "grad_norm": 0.0913616195321083,
      "learning_rate": 9.998342337571565e-05,
      "loss": 1.1901,
      "step": 30
    },
    {
      "epoch": 0.11383108935128519,
      "grad_norm": 0.08568678051233292,
      "learning_rate": 9.99627051715643e-05,
      "loss": 0.3506,
      "step": 31
    },
    {
      "epoch": 0.1175030599755202,
      "grad_norm": 0.07405232638120651,
      "learning_rate": 9.993370449424153e-05,
      "loss": 0.3324,
      "step": 32
    },
    {
      "epoch": 0.1211750305997552,
      "grad_norm": 0.07847411185503006,
      "learning_rate": 9.98964261512799e-05,
      "loss": 0.2745,
      "step": 33
    },
    {
      "epoch": 0.12484700122399021,
      "grad_norm": 0.07299336791038513,
      "learning_rate": 9.985087632242633e-05,
      "loss": 0.5225,
      "step": 34
    },
    {
      "epoch": 0.12851897184822522,
      "grad_norm": 0.0863344594836235,
      "learning_rate": 9.97970625586178e-05,
      "loss": 0.3213,
      "step": 35
    },
    {
      "epoch": 0.13219094247246022,
      "grad_norm": 0.05981218069791794,
      "learning_rate": 9.973499378072945e-05,
      "loss": 0.3551,
      "step": 36
    },
    {
      "epoch": 0.13586291309669524,
      "grad_norm": 0.07763096690177917,
      "learning_rate": 9.966468027809582e-05,
      "loss": 0.3264,
      "step": 37
    },
    {
      "epoch": 0.13953488372093023,
      "grad_norm": 0.0758497565984726,
      "learning_rate": 9.958613370680507e-05,
      "loss": 0.397,
      "step": 38
    },
    {
      "epoch": 0.14320685434516525,
      "grad_norm": 0.3360985517501831,
      "learning_rate": 9.949936708776691e-05,
      "loss": 0.9732,
      "step": 39
    },
    {
      "epoch": 0.14687882496940025,
      "grad_norm": 0.07008073478937149,
      "learning_rate": 9.940439480455386e-05,
      "loss": 0.3739,
      "step": 40
    },
    {
      "epoch": 0.15055079559363524,
      "grad_norm": 0.07435714453458786,
      "learning_rate": 9.930123260101696e-05,
      "loss": 0.3762,
      "step": 41
    },
    {
      "epoch": 0.15422276621787026,
      "grad_norm": 0.3299938142299652,
      "learning_rate": 9.918989757867583e-05,
      "loss": 0.4278,
      "step": 42
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 0.27187252044677734,
      "learning_rate": 9.907040819388371e-05,
      "loss": 0.478,
      "step": 43
    },
    {
      "epoch": 0.16156670746634028,
      "grad_norm": 0.059857387095689774,
      "learning_rate": 9.89427842547679e-05,
      "loss": 0.3133,
      "step": 44
    },
    {
      "epoch": 0.16523867809057527,
      "grad_norm": 0.07612061500549316,
      "learning_rate": 9.880704691794608e-05,
      "loss": 0.4275,
      "step": 45
    },
    {
      "epoch": 0.1689106487148103,
      "grad_norm": 0.19199877977371216,
      "learning_rate": 9.866321868501914e-05,
      "loss": 0.7791,
      "step": 46
    },
    {
      "epoch": 0.17258261933904528,
      "grad_norm": 0.1403833031654358,
      "learning_rate": 9.851132339884096e-05,
      "loss": 0.6251,
      "step": 47
    },
    {
      "epoch": 0.1762545899632803,
      "grad_norm": 0.09072674065828323,
      "learning_rate": 9.835138623956603e-05,
      "loss": 0.4439,
      "step": 48
    },
    {
      "epoch": 0.1799265605875153,
      "grad_norm": 0.08053342252969742,
      "learning_rate": 9.81834337204751e-05,
      "loss": 0.2853,
      "step": 49
    },
    {
      "epoch": 0.1835985312117503,
      "grad_norm": 0.15429392457008362,
      "learning_rate": 9.800749368358009e-05,
      "loss": 0.5243,
      "step": 50
    },
    {
      "epoch": 0.18727050183598531,
      "grad_norm": 0.10327895730733871,
      "learning_rate": 9.782359529500867e-05,
      "loss": 0.6096,
      "step": 51
    },
    {
      "epoch": 0.1909424724602203,
      "grad_norm": 0.07972899079322815,
      "learning_rate": 9.763176904016913e-05,
      "loss": 0.4799,
      "step": 52
    },
    {
      "epoch": 0.19461444308445533,
      "grad_norm": 0.16347306966781616,
      "learning_rate": 9.743204671869694e-05,
      "loss": 0.6377,
      "step": 53
    },
    {
      "epoch": 0.19828641370869032,
      "grad_norm": 0.10270920395851135,
      "learning_rate": 9.722446143918306e-05,
      "loss": 0.5388,
      "step": 54
    },
    {
      "epoch": 0.20195838433292534,
      "grad_norm": 0.08964154124259949,
      "learning_rate": 9.70090476136855e-05,
      "loss": 0.5875,
      "step": 55
    },
    {
      "epoch": 0.20563035495716034,
      "grad_norm": 0.18301688134670258,
      "learning_rate": 9.678584095202468e-05,
      "loss": 0.3805,
      "step": 56
    },
    {
      "epoch": 0.20930232558139536,
      "grad_norm": 0.11658929288387299,
      "learning_rate": 9.655487845586377e-05,
      "loss": 0.5034,
      "step": 57
    },
    {
      "epoch": 0.21297429620563035,
      "grad_norm": 0.10647784918546677,
      "learning_rate": 9.631619841257476e-05,
      "loss": 0.4465,
      "step": 58
    },
    {
      "epoch": 0.21664626682986537,
      "grad_norm": 0.08420727401971817,
      "learning_rate": 9.60698403888914e-05,
      "loss": 0.3717,
      "step": 59
    },
    {
      "epoch": 0.22031823745410037,
      "grad_norm": 0.14984101057052612,
      "learning_rate": 9.581584522435024e-05,
      "loss": 0.5679,
      "step": 60
    },
    {
      "epoch": 0.22399020807833536,
      "grad_norm": 0.07012059539556503,
      "learning_rate": 9.555425502452038e-05,
      "loss": 0.3083,
      "step": 61
    },
    {
      "epoch": 0.22766217870257038,
      "grad_norm": 0.14937996864318848,
      "learning_rate": 9.528511315402358e-05,
      "loss": 0.575,
      "step": 62
    },
    {
      "epoch": 0.23133414932680538,
      "grad_norm": 0.08581522852182388,
      "learning_rate": 9.500846422934557e-05,
      "loss": 0.5254,
      "step": 63
    },
    {
      "epoch": 0.2350061199510404,
      "grad_norm": 0.21604809165000916,
      "learning_rate": 9.472435411143978e-05,
      "loss": 0.7045,
      "step": 64
    },
    {
      "epoch": 0.2386780905752754,
      "grad_norm": 0.1742328703403473,
      "learning_rate": 9.443282989812495e-05,
      "loss": 0.6068,
      "step": 65
    },
    {
      "epoch": 0.2423500611995104,
      "grad_norm": 0.24800752103328705,
      "learning_rate": 9.413393991627737e-05,
      "loss": 0.8401,
      "step": 66
    },
    {
      "epoch": 0.2460220318237454,
      "grad_norm": 0.10994048416614532,
      "learning_rate": 9.382773371381985e-05,
      "loss": 0.425,
      "step": 67
    },
    {
      "epoch": 0.24969400244798043,
      "grad_norm": 0.15951792895793915,
      "learning_rate": 9.351426205150777e-05,
      "loss": 0.5221,
      "step": 68
    },
    {
      "epoch": 0.2533659730722154,
      "grad_norm": 0.20339590311050415,
      "learning_rate": 9.319357689451443e-05,
      "loss": 0.5302,
      "step": 69
    },
    {
      "epoch": 0.25703794369645044,
      "grad_norm": 0.09229890257120132,
      "learning_rate": 9.286573140381662e-05,
      "loss": 0.3575,
      "step": 70
    },
    {
      "epoch": 0.2607099143206854,
      "grad_norm": 0.2451619654893875,
      "learning_rate": 9.253077992738192e-05,
      "loss": 0.7181,
      "step": 71
    },
    {
      "epoch": 0.26438188494492043,
      "grad_norm": 0.12739384174346924,
      "learning_rate": 9.218877799115928e-05,
      "loss": 0.5156,
      "step": 72
    },
    {
      "epoch": 0.26805385556915545,
      "grad_norm": 0.08134493976831436,
      "learning_rate": 9.183978228987436e-05,
      "loss": 0.3,
      "step": 73
    },
    {
      "epoch": 0.2717258261933905,
      "grad_norm": 0.11099664121866226,
      "learning_rate": 9.148385067763095e-05,
      "loss": 0.4198,
      "step": 74
    },
    {
      "epoch": 0.27539779681762544,
      "grad_norm": 0.07188478857278824,
      "learning_rate": 9.112104215832047e-05,
      "loss": 0.3915,
      "step": 75
    },
    {
      "epoch": 0.27906976744186046,
      "grad_norm": 0.06322924792766571,
      "learning_rate": 9.075141687584057e-05,
      "loss": 0.2867,
      "step": 76
    },
    {
      "epoch": 0.2827417380660955,
      "grad_norm": 0.07251914590597153,
      "learning_rate": 9.037503610412501e-05,
      "loss": 0.4824,
      "step": 77
    },
    {
      "epoch": 0.2864137086903305,
      "grad_norm": 0.2195172756910324,
      "learning_rate": 8.999196223698599e-05,
      "loss": 0.3936,
      "step": 78
    },
    {
      "epoch": 0.29008567931456547,
      "grad_norm": 0.21877703070640564,
      "learning_rate": 8.960225877777095e-05,
      "loss": 0.4454,
      "step": 79
    },
    {
      "epoch": 0.2937576499388005,
      "grad_norm": 0.08352579921483994,
      "learning_rate": 8.920599032883554e-05,
      "loss": 0.2931,
      "step": 80
    },
    {
      "epoch": 0.2974296205630355,
      "grad_norm": 0.09364911168813705,
      "learning_rate": 8.880322258083408e-05,
      "loss": 0.4871,
      "step": 81
    },
    {
      "epoch": 0.3011015911872705,
      "grad_norm": 0.07338505238294601,
      "learning_rate": 8.839402230183e-05,
      "loss": 0.3048,
      "step": 82
    },
    {
      "epoch": 0.3047735618115055,
      "grad_norm": 0.1507229208946228,
      "learning_rate": 8.797845732622743e-05,
      "loss": 0.4948,
      "step": 83
    },
    {
      "epoch": 0.3084455324357405,
      "grad_norm": 0.19068707525730133,
      "learning_rate": 8.755659654352599e-05,
      "loss": 0.6406,
      "step": 84
    },
    {
      "epoch": 0.31211750305997554,
      "grad_norm": 0.1288161277770996,
      "learning_rate": 8.712850988690093e-05,
      "loss": 0.362,
      "step": 85
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.07310109585523605,
      "learning_rate": 8.669426832160996e-05,
      "loss": 0.2735,
      "step": 86
    },
    {
      "epoch": 0.31946144430844553,
      "grad_norm": 0.2170391082763672,
      "learning_rate": 8.625394383322914e-05,
      "loss": 0.6501,
      "step": 87
    },
    {
      "epoch": 0.32313341493268055,
      "grad_norm": 0.13086941838264465,
      "learning_rate": 8.580760941571967e-05,
      "loss": 0.4452,
      "step": 88
    },
    {
      "epoch": 0.3268053855569155,
      "grad_norm": 0.08146587759256363,
      "learning_rate": 8.535533905932738e-05,
      "loss": 0.3644,
      "step": 89
    },
    {
      "epoch": 0.33047735618115054,
      "grad_norm": 0.1446744054555893,
      "learning_rate": 8.489720773831717e-05,
      "loss": 0.632,
      "step": 90
    },
    {
      "epoch": 0.33414932680538556,
      "grad_norm": 0.08141674101352692,
      "learning_rate": 8.443329139854433e-05,
      "loss": 0.5625,
      "step": 91
    },
    {
      "epoch": 0.3378212974296206,
      "grad_norm": 0.07421129941940308,
      "learning_rate": 8.396366694486466e-05,
      "loss": 0.3207,
      "step": 92
    },
    {
      "epoch": 0.34149326805385555,
      "grad_norm": 0.345011830329895,
      "learning_rate": 8.34884122283858e-05,
      "loss": 0.5898,
      "step": 93
    },
    {
      "epoch": 0.34516523867809057,
      "grad_norm": 0.12236194312572479,
      "learning_rate": 8.300760603356159e-05,
      "loss": 0.4658,
      "step": 94
    },
    {
      "epoch": 0.3488372093023256,
      "grad_norm": 0.10392044484615326,
      "learning_rate": 8.25213280651317e-05,
      "loss": 0.4421,
      "step": 95
    },
    {
      "epoch": 0.3525091799265606,
      "grad_norm": 0.0597870871424675,
      "learning_rate": 8.202965893490878e-05,
      "loss": 0.2782,
      "step": 96
    },
    {
      "epoch": 0.3561811505507956,
      "grad_norm": 0.2009837031364441,
      "learning_rate": 8.153268014841506e-05,
      "loss": 0.4888,
      "step": 97
    },
    {
      "epoch": 0.3598531211750306,
      "grad_norm": 0.24240681529045105,
      "learning_rate": 8.103047409137114e-05,
      "loss": 0.7807,
      "step": 98
    },
    {
      "epoch": 0.3635250917992656,
      "grad_norm": 0.0845351368188858,
      "learning_rate": 8.052312401603847e-05,
      "loss": 0.6298,
      "step": 99
    },
    {
      "epoch": 0.3671970624235006,
      "grad_norm": 0.11193636804819107,
      "learning_rate": 8.001071402741842e-05,
      "loss": 0.8064,
      "step": 100
    },
    {
      "epoch": 0.3708690330477356,
      "grad_norm": 0.0703149363398552,
      "learning_rate": 7.949332906930995e-05,
      "loss": 0.3406,
      "step": 101
    },
    {
      "epoch": 0.37454100367197063,
      "grad_norm": 0.1298230141401291,
      "learning_rate": 7.897105491022818e-05,
      "loss": 0.579,
      "step": 102
    },
    {
      "epoch": 0.37821297429620565,
      "grad_norm": 0.07925601303577423,
      "learning_rate": 7.844397812918635e-05,
      "loss": 0.3489,
      "step": 103
    },
    {
      "epoch": 0.3818849449204406,
      "grad_norm": 0.12708668410778046,
      "learning_rate": 7.791218610134323e-05,
      "loss": 0.4151,
      "step": 104
    },
    {
      "epoch": 0.38555691554467564,
      "grad_norm": 0.2936255931854248,
      "learning_rate": 7.737576698351878e-05,
      "loss": 1.1401,
      "step": 105
    },
    {
      "epoch": 0.38922888616891066,
      "grad_norm": 0.07473833113908768,
      "learning_rate": 7.683480969958004e-05,
      "loss": 0.3091,
      "step": 106
    },
    {
      "epoch": 0.3929008567931457,
      "grad_norm": 0.1253412663936615,
      "learning_rate": 7.628940392569994e-05,
      "loss": 0.4951,
      "step": 107
    },
    {
      "epoch": 0.39657282741738065,
      "grad_norm": 0.08265762031078339,
      "learning_rate": 7.573964007549155e-05,
      "loss": 0.4025,
      "step": 108
    },
    {
      "epoch": 0.40024479804161567,
      "grad_norm": 0.11429966986179352,
      "learning_rate": 7.518560928501969e-05,
      "loss": 0.4147,
      "step": 109
    },
    {
      "epoch": 0.4039167686658507,
      "grad_norm": 0.25090694427490234,
      "learning_rate": 7.462740339769324e-05,
      "loss": 0.5981,
      "step": 110
    },
    {
      "epoch": 0.40758873929008566,
      "grad_norm": 0.17391008138656616,
      "learning_rate": 7.406511494903982e-05,
      "loss": 0.4219,
      "step": 111
    },
    {
      "epoch": 0.4112607099143207,
      "grad_norm": 0.06932790577411652,
      "learning_rate": 7.3498837151366e-05,
      "loss": 0.4379,
      "step": 112
    },
    {
      "epoch": 0.4149326805385557,
      "grad_norm": 0.08115310966968536,
      "learning_rate": 7.292866387830514e-05,
      "loss": 0.3631,
      "step": 113
    },
    {
      "epoch": 0.4186046511627907,
      "grad_norm": 0.11411315202713013,
      "learning_rate": 7.23546896492557e-05,
      "loss": 0.6605,
      "step": 114
    },
    {
      "epoch": 0.4222766217870257,
      "grad_norm": 0.12723813951015472,
      "learning_rate": 7.177700961371239e-05,
      "loss": 0.5364,
      "step": 115
    },
    {
      "epoch": 0.4259485924112607,
      "grad_norm": 0.08049023896455765,
      "learning_rate": 7.119571953549305e-05,
      "loss": 0.3387,
      "step": 116
    },
    {
      "epoch": 0.42962056303549573,
      "grad_norm": 0.07781126350164413,
      "learning_rate": 7.06109157768635e-05,
      "loss": 0.5689,
      "step": 117
    },
    {
      "epoch": 0.43329253365973075,
      "grad_norm": 0.14163357019424438,
      "learning_rate": 7.002269528256334e-05,
      "loss": 0.4366,
      "step": 118
    },
    {
      "epoch": 0.4369645042839657,
      "grad_norm": 0.07881614565849304,
      "learning_rate": 6.943115556373502e-05,
      "loss": 0.3186,
      "step": 119
    },
    {
      "epoch": 0.44063647490820074,
      "grad_norm": 0.08862645924091339,
      "learning_rate": 6.883639468175927e-05,
      "loss": 0.3786,
      "step": 120
    },
    {
      "epoch": 0.44430844553243576,
      "grad_norm": 0.08052200824022293,
      "learning_rate": 6.823851123199893e-05,
      "loss": 0.3527,
      "step": 121
    },
    {
      "epoch": 0.4479804161566707,
      "grad_norm": 0.11888729780912399,
      "learning_rate": 6.763760432745475e-05,
      "loss": 0.4314,
      "step": 122
    },
    {
      "epoch": 0.45165238678090575,
      "grad_norm": 0.06493515521287918,
      "learning_rate": 6.703377358233489e-05,
      "loss": 0.2977,
      "step": 123
    },
    {
      "epoch": 0.45532435740514077,
      "grad_norm": 0.10048846155405045,
      "learning_rate": 6.642711909554174e-05,
      "loss": 0.3244,
      "step": 124
    },
    {
      "epoch": 0.4589963280293758,
      "grad_norm": 0.10203630477190018,
      "learning_rate": 6.58177414340781e-05,
      "loss": 0.4996,
      "step": 125
    },
    {
      "epoch": 0.46266829865361075,
      "grad_norm": 0.2413017451763153,
      "learning_rate": 6.52057416163759e-05,
      "loss": 0.555,
      "step": 126
    },
    {
      "epoch": 0.4663402692778458,
      "grad_norm": 0.08983640372753143,
      "learning_rate": 6.45912210955501e-05,
      "loss": 0.2582,
      "step": 127
    },
    {
      "epoch": 0.4700122399020808,
      "grad_norm": 0.21270190179347992,
      "learning_rate": 6.397428174258047e-05,
      "loss": 0.5315,
      "step": 128
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 0.12013928592205048,
      "learning_rate": 6.335502582942408e-05,
      "loss": 0.489,
      "step": 129
    },
    {
      "epoch": 0.4773561811505508,
      "grad_norm": 0.1020718440413475,
      "learning_rate": 6.273355601206144e-05,
      "loss": 0.456,
      "step": 130
    },
    {
      "epoch": 0.4810281517747858,
      "grad_norm": 0.08006691932678223,
      "learning_rate": 6.210997531347878e-05,
      "loss": 0.3459,
      "step": 131
    },
    {
      "epoch": 0.4847001223990208,
      "grad_norm": 0.10623283684253693,
      "learning_rate": 6.14843871065898e-05,
      "loss": 0.5545,
      "step": 132
    },
    {
      "epoch": 0.4883720930232558,
      "grad_norm": 0.0730510801076889,
      "learning_rate": 6.0856895097098934e-05,
      "loss": 0.3225,
      "step": 133
    },
    {
      "epoch": 0.4920440636474908,
      "grad_norm": 0.09953512251377106,
      "learning_rate": 6.022760330631005e-05,
      "loss": 0.4317,
      "step": 134
    },
    {
      "epoch": 0.49571603427172584,
      "grad_norm": 0.0781313106417656,
      "learning_rate": 5.9596616053882295e-05,
      "loss": 0.3251,
      "step": 135
    },
    {
      "epoch": 0.49938800489596086,
      "grad_norm": 0.11223123967647552,
      "learning_rate": 5.896403794053679e-05,
      "loss": 0.63,
      "step": 136
    },
    {
      "epoch": 0.5030599755201959,
      "grad_norm": 0.10512508451938629,
      "learning_rate": 5.83299738307166e-05,
      "loss": 0.481,
      "step": 137
    },
    {
      "epoch": 0.5067319461444308,
      "grad_norm": 0.08379629254341125,
      "learning_rate": 5.76945288352031e-05,
      "loss": 0.4922,
      "step": 138
    },
    {
      "epoch": 0.5104039167686658,
      "grad_norm": 0.08486858010292053,
      "learning_rate": 5.70578082936913e-05,
      "loss": 0.381,
      "step": 139
    },
    {
      "epoch": 0.5140758873929009,
      "grad_norm": 0.14607948064804077,
      "learning_rate": 5.6419917757327555e-05,
      "loss": 0.5921,
      "step": 140
    },
    {
      "epoch": 0.5177478580171359,
      "grad_norm": 0.1181630790233612,
      "learning_rate": 5.5780962971211795e-05,
      "loss": 0.6962,
      "step": 141
    },
    {
      "epoch": 0.5214198286413708,
      "grad_norm": 0.17947016656398773,
      "learning_rate": 5.514104985686802e-05,
      "loss": 0.6109,
      "step": 142
    },
    {
      "epoch": 0.5250917992656059,
      "grad_norm": 0.06438343226909637,
      "learning_rate": 5.450028449468527e-05,
      "loss": 0.2542,
      "step": 143
    },
    {
      "epoch": 0.5287637698898409,
      "grad_norm": 0.09820745885372162,
      "learning_rate": 5.385877310633233e-05,
      "loss": 0.6507,
      "step": 144
    },
    {
      "epoch": 0.5324357405140759,
      "grad_norm": 0.09040407836437225,
      "learning_rate": 5.321662203714909e-05,
      "loss": 0.399,
      "step": 145
    },
    {
      "epoch": 0.5361077111383109,
      "grad_norm": 0.2683495283126831,
      "learning_rate": 5.257393773851733e-05,
      "loss": 0.4017,
      "step": 146
    },
    {
      "epoch": 0.5397796817625459,
      "grad_norm": 0.1613844335079193,
      "learning_rate": 5.193082675021392e-05,
      "loss": 0.3849,
      "step": 147
    },
    {
      "epoch": 0.543451652386781,
      "grad_norm": 0.12207731604576111,
      "learning_rate": 5.128739568274944e-05,
      "loss": 0.5905,
      "step": 148
    },
    {
      "epoch": 0.5471236230110159,
      "grad_norm": 0.0870044156908989,
      "learning_rate": 5.064375119969491e-05,
      "loss": 0.2895,
      "step": 149
    },
    {
      "epoch": 0.5507955936352509,
      "grad_norm": 0.08491414785385132,
      "learning_rate": 5e-05,
      "loss": 0.3632,
      "step": 150
    },
    {
      "epoch": 0.554467564259486,
      "grad_norm": 0.11790735274553299,
      "learning_rate": 4.9356248800305106e-05,
      "loss": 0.7068,
      "step": 151
    },
    {
      "epoch": 0.5581395348837209,
      "grad_norm": 0.19284167885780334,
      "learning_rate": 4.8712604317250576e-05,
      "loss": 0.3605,
      "step": 152
    },
    {
      "epoch": 0.5618115055079559,
      "grad_norm": 0.09215154498815536,
      "learning_rate": 4.806917324978608e-05,
      "loss": 0.2476,
      "step": 153
    },
    {
      "epoch": 0.565483476132191,
      "grad_norm": 0.12108391523361206,
      "learning_rate": 4.7426062261482675e-05,
      "loss": 0.5039,
      "step": 154
    },
    {
      "epoch": 0.5691554467564259,
      "grad_norm": 0.14685741066932678,
      "learning_rate": 4.678337796285092e-05,
      "loss": 0.5615,
      "step": 155
    },
    {
      "epoch": 0.572827417380661,
      "grad_norm": 0.09982460737228394,
      "learning_rate": 4.6141226893667684e-05,
      "loss": 0.6,
      "step": 156
    },
    {
      "epoch": 0.576499388004896,
      "grad_norm": 0.10319037735462189,
      "learning_rate": 4.5499715505314745e-05,
      "loss": 0.4833,
      "step": 157
    },
    {
      "epoch": 0.5801713586291309,
      "grad_norm": 0.0910174548625946,
      "learning_rate": 4.4858950143131975e-05,
      "loss": 0.5044,
      "step": 158
    },
    {
      "epoch": 0.583843329253366,
      "grad_norm": 0.08110320568084717,
      "learning_rate": 4.4219037028788216e-05,
      "loss": 0.6077,
      "step": 159
    },
    {
      "epoch": 0.587515299877601,
      "grad_norm": 0.0765228420495987,
      "learning_rate": 4.358008224267245e-05,
      "loss": 0.2877,
      "step": 160
    },
    {
      "epoch": 0.591187270501836,
      "grad_norm": 0.16392959654331207,
      "learning_rate": 4.29421917063087e-05,
      "loss": 0.4462,
      "step": 161
    },
    {
      "epoch": 0.594859241126071,
      "grad_norm": 0.09697830677032471,
      "learning_rate": 4.2305471164796905e-05,
      "loss": 0.3803,
      "step": 162
    },
    {
      "epoch": 0.598531211750306,
      "grad_norm": 0.09458833932876587,
      "learning_rate": 4.167002616928341e-05,
      "loss": 0.4625,
      "step": 163
    },
    {
      "epoch": 0.602203182374541,
      "grad_norm": 0.08443703502416611,
      "learning_rate": 4.103596205946323e-05,
      "loss": 0.615,
      "step": 164
    },
    {
      "epoch": 0.605875152998776,
      "grad_norm": 0.10602038353681564,
      "learning_rate": 4.0403383946117717e-05,
      "loss": 0.6253,
      "step": 165
    },
    {
      "epoch": 0.609547123623011,
      "grad_norm": 0.18194644153118134,
      "learning_rate": 3.977239669368997e-05,
      "loss": 0.4052,
      "step": 166
    },
    {
      "epoch": 0.6132190942472461,
      "grad_norm": 0.1692446619272232,
      "learning_rate": 3.9143104902901085e-05,
      "loss": 0.9142,
      "step": 167
    },
    {
      "epoch": 0.616891064871481,
      "grad_norm": 0.19827115535736084,
      "learning_rate": 3.851561289341023e-05,
      "loss": 0.4136,
      "step": 168
    },
    {
      "epoch": 0.620563035495716,
      "grad_norm": 0.2976871132850647,
      "learning_rate": 3.789002468652121e-05,
      "loss": 0.5657,
      "step": 169
    },
    {
      "epoch": 0.6242350061199511,
      "grad_norm": 0.07852277904748917,
      "learning_rate": 3.726644398793857e-05,
      "loss": 0.3279,
      "step": 170
    },
    {
      "epoch": 0.627906976744186,
      "grad_norm": 0.12604489922523499,
      "learning_rate": 3.664497417057591e-05,
      "loss": 0.62,
      "step": 171
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.22574089467525482,
      "learning_rate": 3.602571825741953e-05,
      "loss": 0.8179,
      "step": 172
    },
    {
      "epoch": 0.6352509179926561,
      "grad_norm": 0.13447195291519165,
      "learning_rate": 3.540877890444989e-05,
      "loss": 0.5163,
      "step": 173
    },
    {
      "epoch": 0.6389228886168911,
      "grad_norm": 0.11539029330015182,
      "learning_rate": 3.4794258383624114e-05,
      "loss": 0.4173,
      "step": 174
    },
    {
      "epoch": 0.642594859241126,
      "grad_norm": 0.11384222656488419,
      "learning_rate": 3.4182258565921934e-05,
      "loss": 0.5455,
      "step": 175
    },
    {
      "epoch": 0.6462668298653611,
      "grad_norm": 0.12927398085594177,
      "learning_rate": 3.357288090445827e-05,
      "loss": 0.5327,
      "step": 176
    },
    {
      "epoch": 0.6499388004895961,
      "grad_norm": 0.16683682799339294,
      "learning_rate": 3.296622641766512e-05,
      "loss": 0.5541,
      "step": 177
    },
    {
      "epoch": 0.653610771113831,
      "grad_norm": 0.09209699183702469,
      "learning_rate": 3.236239567254526e-05,
      "loss": 0.4312,
      "step": 178
    },
    {
      "epoch": 0.6572827417380661,
      "grad_norm": 0.13782797753810883,
      "learning_rate": 3.1761488768001085e-05,
      "loss": 0.8246,
      "step": 179
    },
    {
      "epoch": 0.6609547123623011,
      "grad_norm": 0.07543042302131653,
      "learning_rate": 3.116360531824074e-05,
      "loss": 0.5633,
      "step": 180
    },
    {
      "epoch": 0.6646266829865362,
      "grad_norm": 0.14241543412208557,
      "learning_rate": 3.056884443626499e-05,
      "loss": 0.6657,
      "step": 181
    },
    {
      "epoch": 0.6682986536107711,
      "grad_norm": 0.11854031682014465,
      "learning_rate": 2.9977304717436672e-05,
      "loss": 0.6332,
      "step": 182
    },
    {
      "epoch": 0.6719706242350061,
      "grad_norm": 0.09301911294460297,
      "learning_rate": 2.938908422313652e-05,
      "loss": 0.4556,
      "step": 183
    },
    {
      "epoch": 0.6756425948592412,
      "grad_norm": 0.08479209989309311,
      "learning_rate": 2.8804280464506973e-05,
      "loss": 0.3345,
      "step": 184
    },
    {
      "epoch": 0.6793145654834761,
      "grad_norm": 0.08416282385587692,
      "learning_rate": 2.8222990386287617e-05,
      "loss": 0.4235,
      "step": 185
    },
    {
      "epoch": 0.6829865361077111,
      "grad_norm": 0.29337114095687866,
      "learning_rate": 2.7645310350744297e-05,
      "loss": 0.485,
      "step": 186
    },
    {
      "epoch": 0.6866585067319462,
      "grad_norm": 0.106132872402668,
      "learning_rate": 2.7071336121694858e-05,
      "loss": 0.3384,
      "step": 187
    },
    {
      "epoch": 0.6903304773561811,
      "grad_norm": 0.08644799143075943,
      "learning_rate": 2.650116284863402e-05,
      "loss": 0.9103,
      "step": 188
    },
    {
      "epoch": 0.6940024479804161,
      "grad_norm": 0.13192370533943176,
      "learning_rate": 2.5934885050960182e-05,
      "loss": 0.5258,
      "step": 189
    },
    {
      "epoch": 0.6976744186046512,
      "grad_norm": 0.10062063485383987,
      "learning_rate": 2.5372596602306786e-05,
      "loss": 0.4817,
      "step": 190
    },
    {
      "epoch": 0.7013463892288861,
      "grad_norm": 0.09146949648857117,
      "learning_rate": 2.4814390714980325e-05,
      "loss": 0.4289,
      "step": 191
    },
    {
      "epoch": 0.7050183598531212,
      "grad_norm": 0.17095047235488892,
      "learning_rate": 2.426035992450848e-05,
      "loss": 0.4383,
      "step": 192
    },
    {
      "epoch": 0.7086903304773562,
      "grad_norm": 0.2827148735523224,
      "learning_rate": 2.3710596074300045e-05,
      "loss": 0.7566,
      "step": 193
    },
    {
      "epoch": 0.7123623011015912,
      "grad_norm": 0.09297563880681992,
      "learning_rate": 2.316519030041998e-05,
      "loss": 0.4648,
      "step": 194
    },
    {
      "epoch": 0.7160342717258262,
      "grad_norm": 0.11713358759880066,
      "learning_rate": 2.262423301648122e-05,
      "loss": 0.668,
      "step": 195
    },
    {
      "epoch": 0.7197062423500612,
      "grad_norm": 0.10133064538240433,
      "learning_rate": 2.2087813898656774e-05,
      "loss": 0.4397,
      "step": 196
    },
    {
      "epoch": 0.7233782129742962,
      "grad_norm": 0.23745763301849365,
      "learning_rate": 2.1556021870813654e-05,
      "loss": 0.685,
      "step": 197
    },
    {
      "epoch": 0.7270501835985312,
      "grad_norm": 0.11296797543764114,
      "learning_rate": 2.102894508977182e-05,
      "loss": 0.2627,
      "step": 198
    },
    {
      "epoch": 0.7307221542227662,
      "grad_norm": 0.09247058629989624,
      "learning_rate": 2.0506670930690074e-05,
      "loss": 0.3164,
      "step": 199
    },
    {
      "epoch": 0.7343941248470012,
      "grad_norm": 0.08100568503141403,
      "learning_rate": 1.9989285972581595e-05,
      "loss": 0.4444,
      "step": 200
    },
    {
      "epoch": 0.7380660954712362,
      "grad_norm": 0.08765120804309845,
      "learning_rate": 1.947687598396154e-05,
      "loss": 0.8168,
      "step": 201
    },
    {
      "epoch": 0.7417380660954712,
      "grad_norm": 0.11285313963890076,
      "learning_rate": 1.896952590862886e-05,
      "loss": 0.3696,
      "step": 202
    },
    {
      "epoch": 0.7454100367197063,
      "grad_norm": 0.1029384583234787,
      "learning_rate": 1.8467319851584954e-05,
      "loss": 0.3903,
      "step": 203
    },
    {
      "epoch": 0.7490820073439413,
      "grad_norm": 0.14774741232395172,
      "learning_rate": 1.7970341065091245e-05,
      "loss": 0.5486,
      "step": 204
    },
    {
      "epoch": 0.7527539779681762,
      "grad_norm": 0.20584475994110107,
      "learning_rate": 1.7478671934868302e-05,
      "loss": 0.4969,
      "step": 205
    },
    {
      "epoch": 0.7564259485924113,
      "grad_norm": 0.10497330129146576,
      "learning_rate": 1.6992393966438407e-05,
      "loss": 0.6211,
      "step": 206
    },
    {
      "epoch": 0.7600979192166463,
      "grad_norm": 0.10213766247034073,
      "learning_rate": 1.6511587771614205e-05,
      "loss": 0.3785,
      "step": 207
    },
    {
      "epoch": 0.7637698898408812,
      "grad_norm": 0.08263826370239258,
      "learning_rate": 1.6036333055135344e-05,
      "loss": 0.4173,
      "step": 208
    },
    {
      "epoch": 0.7674418604651163,
      "grad_norm": 0.15048938989639282,
      "learning_rate": 1.556670860145567e-05,
      "loss": 0.7029,
      "step": 209
    },
    {
      "epoch": 0.7711138310893513,
      "grad_norm": 0.09334007650613785,
      "learning_rate": 1.5102792261682813e-05,
      "loss": 0.4614,
      "step": 210
    },
    {
      "epoch": 0.7747858017135862,
      "grad_norm": 0.15974034368991852,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 0.8333,
      "step": 211
    },
    {
      "epoch": 0.7784577723378213,
      "grad_norm": 0.08053223788738251,
      "learning_rate": 1.4192390584280346e-05,
      "loss": 0.3017,
      "step": 212
    },
    {
      "epoch": 0.7821297429620563,
      "grad_norm": 0.0834859237074852,
      "learning_rate": 1.374605616677087e-05,
      "loss": 0.304,
      "step": 213
    },
    {
      "epoch": 0.7858017135862914,
      "grad_norm": 0.0708678588271141,
      "learning_rate": 1.3305731678390048e-05,
      "loss": 0.4155,
      "step": 214
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 0.13682302832603455,
      "learning_rate": 1.2871490113099066e-05,
      "loss": 0.7763,
      "step": 215
    },
    {
      "epoch": 0.7931456548347613,
      "grad_norm": 0.12170913815498352,
      "learning_rate": 1.2443403456474017e-05,
      "loss": 0.5421,
      "step": 216
    },
    {
      "epoch": 0.7968176254589964,
      "grad_norm": 0.07957993447780609,
      "learning_rate": 1.2021542673772585e-05,
      "loss": 0.3638,
      "step": 217
    },
    {
      "epoch": 0.8004895960832313,
      "grad_norm": 0.14521080255508423,
      "learning_rate": 1.1605977698170001e-05,
      "loss": 0.5641,
      "step": 218
    },
    {
      "epoch": 0.8041615667074663,
      "grad_norm": 0.0813819020986557,
      "learning_rate": 1.1196777419165927e-05,
      "loss": 0.407,
      "step": 219
    },
    {
      "epoch": 0.8078335373317014,
      "grad_norm": 0.12135583162307739,
      "learning_rate": 1.0794009671164484e-05,
      "loss": 0.8802,
      "step": 220
    },
    {
      "epoch": 0.8115055079559363,
      "grad_norm": 0.1219213679432869,
      "learning_rate": 1.0397741222229057e-05,
      "loss": 0.5714,
      "step": 221
    },
    {
      "epoch": 0.8151774785801713,
      "grad_norm": 0.08637195825576782,
      "learning_rate": 1.0008037763014032e-05,
      "loss": 0.4146,
      "step": 222
    },
    {
      "epoch": 0.8188494492044064,
      "grad_norm": 0.22487489879131317,
      "learning_rate": 9.624963895874994e-06,
      "loss": 0.4979,
      "step": 223
    },
    {
      "epoch": 0.8225214198286414,
      "grad_norm": 0.28123998641967773,
      "learning_rate": 9.248583124159438e-06,
      "loss": 0.7919,
      "step": 224
    },
    {
      "epoch": 0.8261933904528764,
      "grad_norm": 0.14954394102096558,
      "learning_rate": 8.878957841679541e-06,
      "loss": 0.4572,
      "step": 225
    },
    {
      "epoch": 0.8298653610771114,
      "grad_norm": 0.08152212202548981,
      "learning_rate": 8.516149322369054e-06,
      "loss": 0.3024,
      "step": 226
    },
    {
      "epoch": 0.8335373317013464,
      "grad_norm": 0.08769659698009491,
      "learning_rate": 8.160217710125662e-06,
      "loss": 0.5087,
      "step": 227
    },
    {
      "epoch": 0.8372093023255814,
      "grad_norm": 0.14432811737060547,
      "learning_rate": 7.81122200884072e-06,
      "loss": 0.4048,
      "step": 228
    },
    {
      "epoch": 0.8408812729498164,
      "grad_norm": 0.2427009642124176,
      "learning_rate": 7.469220072618094e-06,
      "loss": 0.4902,
      "step": 229
    },
    {
      "epoch": 0.8445532435740514,
      "grad_norm": 0.09131976217031479,
      "learning_rate": 7.13426859618338e-06,
      "loss": 0.2719,
      "step": 230
    },
    {
      "epoch": 0.8482252141982864,
      "grad_norm": 0.0897597223520279,
      "learning_rate": 6.806423105485577e-06,
      "loss": 0.3147,
      "step": 231
    },
    {
      "epoch": 0.8518971848225214,
      "grad_norm": 0.11949042230844498,
      "learning_rate": 6.4857379484922375e-06,
      "loss": 0.5688,
      "step": 232
    },
    {
      "epoch": 0.8555691554467564,
      "grad_norm": 0.08959188312292099,
      "learning_rate": 6.1722662861801614e-06,
      "loss": 0.3724,
      "step": 233
    },
    {
      "epoch": 0.8592411260709915,
      "grad_norm": 0.10583740472793579,
      "learning_rate": 5.866060083722624e-06,
      "loss": 0.6448,
      "step": 234
    },
    {
      "epoch": 0.8629130966952264,
      "grad_norm": 0.1368107944726944,
      "learning_rate": 5.5671701018750745e-06,
      "loss": 0.7383,
      "step": 235
    },
    {
      "epoch": 0.8665850673194615,
      "grad_norm": 0.07287576794624329,
      "learning_rate": 5.275645888560232e-06,
      "loss": 0.3452,
      "step": 236
    },
    {
      "epoch": 0.8702570379436965,
      "grad_norm": 0.07863111793994904,
      "learning_rate": 4.9915357706544485e-06,
      "loss": 0.3893,
      "step": 237
    },
    {
      "epoch": 0.8739290085679314,
      "grad_norm": 0.08062013983726501,
      "learning_rate": 4.714886845976429e-06,
      "loss": 0.2986,
      "step": 238
    },
    {
      "epoch": 0.8776009791921665,
      "grad_norm": 0.07851500064134598,
      "learning_rate": 4.445744975479626e-06,
      "loss": 0.4395,
      "step": 239
    },
    {
      "epoch": 0.8812729498164015,
      "grad_norm": 0.07426175475120544,
      "learning_rate": 4.184154775649768e-06,
      "loss": 0.2981,
      "step": 240
    },
    {
      "epoch": 0.8849449204406364,
      "grad_norm": 0.10778014361858368,
      "learning_rate": 3.9301596111086025e-06,
      "loss": 0.5546,
      "step": 241
    },
    {
      "epoch": 0.8886168910648715,
      "grad_norm": 0.23396329581737518,
      "learning_rate": 3.683801587425251e-06,
      "loss": 0.4488,
      "step": 242
    },
    {
      "epoch": 0.8922888616891065,
      "grad_norm": 0.24511796236038208,
      "learning_rate": 3.4451215441362263e-06,
      "loss": 0.5273,
      "step": 243
    },
    {
      "epoch": 0.8959608323133414,
      "grad_norm": 0.11425651609897614,
      "learning_rate": 3.2141590479753236e-06,
      "loss": 0.6317,
      "step": 244
    },
    {
      "epoch": 0.8996328029375765,
      "grad_norm": 0.2402946650981903,
      "learning_rate": 2.990952386314505e-06,
      "loss": 0.5092,
      "step": 245
    },
    {
      "epoch": 0.9033047735618115,
      "grad_norm": 0.12040656059980392,
      "learning_rate": 2.775538560816937e-06,
      "loss": 0.5734,
      "step": 246
    },
    {
      "epoch": 0.9069767441860465,
      "grad_norm": 0.08450046181678772,
      "learning_rate": 2.5679532813030596e-06,
      "loss": 0.535,
      "step": 247
    },
    {
      "epoch": 0.9106487148102815,
      "grad_norm": 0.08185907453298569,
      "learning_rate": 2.3682309598308747e-06,
      "loss": 0.5474,
      "step": 248
    },
    {
      "epoch": 0.9143206854345165,
      "grad_norm": 0.08737199753522873,
      "learning_rate": 2.1764047049913527e-06,
      "loss": 0.3862,
      "step": 249
    },
    {
      "epoch": 0.9179926560587516,
      "grad_norm": 0.0846424475312233,
      "learning_rate": 1.992506316419912e-06,
      "loss": 0.3631,
      "step": 250
    },
    {
      "epoch": 0.9216646266829865,
      "grad_norm": 0.0750948116183281,
      "learning_rate": 1.8165662795249172e-06,
      "loss": 0.2861,
      "step": 251
    },
    {
      "epoch": 0.9253365973072215,
      "grad_norm": 0.09620046615600586,
      "learning_rate": 1.6486137604339813e-06,
      "loss": 0.6115,
      "step": 252
    },
    {
      "epoch": 0.9290085679314566,
      "grad_norm": 0.13249874114990234,
      "learning_rate": 1.4886766011590448e-06,
      "loss": 0.6009,
      "step": 253
    },
    {
      "epoch": 0.9326805385556916,
      "grad_norm": 0.07768455892801285,
      "learning_rate": 1.3367813149808729e-06,
      "loss": 0.5147,
      "step": 254
    },
    {
      "epoch": 0.9363525091799265,
      "grad_norm": 0.09131491929292679,
      "learning_rate": 1.1929530820539269e-06,
      "loss": 0.3699,
      "step": 255
    },
    {
      "epoch": 0.9400244798041616,
      "grad_norm": 0.0973149985074997,
      "learning_rate": 1.0572157452321097e-06,
      "loss": 0.4612,
      "step": 256
    },
    {
      "epoch": 0.9436964504283966,
      "grad_norm": 0.08338738232851028,
      "learning_rate": 9.295918061163033e-07,
      "loss": 0.3989,
      "step": 257
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 0.12276645749807358,
      "learning_rate": 8.101024213241826e-07,
      "loss": 0.7361,
      "step": 258
    },
    {
      "epoch": 0.9510403916768666,
      "grad_norm": 0.07636597752571106,
      "learning_rate": 6.987673989830523e-07,
      "loss": 0.356,
      "step": 259
    },
    {
      "epoch": 0.9547123623011016,
      "grad_norm": 0.0995926633477211,
      "learning_rate": 5.956051954461472e-07,
      "loss": 0.5988,
      "step": 260
    },
    {
      "epoch": 0.9583843329253366,
      "grad_norm": 0.12855686247348785,
      "learning_rate": 5.006329122330899e-07,
      "loss": 0.6091,
      "step": 261
    },
    {
      "epoch": 0.9620563035495716,
      "grad_norm": 0.10323388129472733,
      "learning_rate": 4.1386629319492556e-07,
      "loss": 0.5726,
      "step": 262
    },
    {
      "epoch": 0.9657282741738066,
      "grad_norm": 0.07876481115818024,
      "learning_rate": 3.3531972190419815e-07,
      "loss": 0.2667,
      "step": 263
    },
    {
      "epoch": 0.9694002447980417,
      "grad_norm": 0.08382689952850342,
      "learning_rate": 2.6500621927054715e-07,
      "loss": 0.2876,
      "step": 264
    },
    {
      "epoch": 0.9730722154222766,
      "grad_norm": 0.08943753689527512,
      "learning_rate": 2.0293744138219495e-07,
      "loss": 0.3169,
      "step": 265
    },
    {
      "epoch": 0.9767441860465116,
      "grad_norm": 0.07517411559820175,
      "learning_rate": 1.4912367757366487e-07,
      "loss": 0.3369,
      "step": 266
    },
    {
      "epoch": 0.9804161566707467,
      "grad_norm": 0.08059985190629959,
      "learning_rate": 1.0357384872011766e-07,
      "loss": 0.2544,
      "step": 267
    },
    {
      "epoch": 0.9840881272949816,
      "grad_norm": 0.15402479469776154,
      "learning_rate": 6.629550575847354e-08,
      "loss": 0.9444,
      "step": 268
    },
    {
      "epoch": 0.9877600979192166,
      "grad_norm": 0.07759124785661697,
      "learning_rate": 3.729482843569665e-08,
      "loss": 0.5641,
      "step": 269
    },
    {
      "epoch": 0.9914320685434517,
      "grad_norm": 0.08798715472221375,
      "learning_rate": 1.6576624284347918e-08,
      "loss": 0.3192,
      "step": 270
    },
    {
      "epoch": 0.9951040391676866,
      "grad_norm": 0.09718140214681625,
      "learning_rate": 4.1443278256170226e-09,
      "loss": 0.6975,
      "step": 271
    },
    {
      "epoch": 0.9987760097919217,
      "grad_norm": 0.09622414410114288,
      "learning_rate": 0.0,
      "loss": 0.4445,
      "step": 272
    },
    {
      "epoch": 0.9987760097919217,
      "step": 272,
      "total_flos": 1.7184161279184077e+17,
      "train_loss": 0.49135432271834684,
      "train_runtime": 6097.9323,
      "train_samples_per_second": 0.134,
      "train_steps_per_second": 0.045
    }
  ],
  "logging_steps": 1,
  "max_steps": 272,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7184161279184077e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}