{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5001406645619081,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001000281329123816,
      "grad_norm": 5.902005195617676,
      "learning_rate": 0.0001999999450590425,
      "loss": 3.1875,
      "step": 1
    },
    {
      "epoch": 0.002000562658247632,
      "grad_norm": 3.2577760219573975,
      "learning_rate": 0.00019999978023623033,
      "loss": 2.3666,
      "step": 2
    },
    {
      "epoch": 0.003000843987371448,
      "grad_norm": 5.3700995445251465,
      "learning_rate": 0.0001999995055317446,
      "loss": 2.8282,
      "step": 3
    },
    {
      "epoch": 0.004001125316495264,
      "grad_norm": 2.1445534229278564,
      "learning_rate": 0.00019999912094588717,
      "loss": 2.2322,
      "step": 4
    },
    {
      "epoch": 0.005001406645619081,
      "grad_norm": 1.5143821239471436,
      "learning_rate": 0.00019999862647908064,
      "loss": 2.1709,
      "step": 5
    },
    {
      "epoch": 0.006001687974742896,
      "grad_norm": 2.0491714477539062,
      "learning_rate": 0.00019999802213186834,
      "loss": 2.2863,
      "step": 6
    },
    {
      "epoch": 0.007001969303866712,
      "grad_norm": 1.2016857862472534,
      "learning_rate": 0.0001999973079049143,
      "loss": 1.5595,
      "step": 7
    },
    {
      "epoch": 0.008002250632990529,
      "grad_norm": 1.3860406875610352,
      "learning_rate": 0.00019999648379900338,
      "loss": 1.7264,
      "step": 8
    },
    {
      "epoch": 0.009002531962114344,
      "grad_norm": 1.0861930847167969,
      "learning_rate": 0.0001999955498150411,
      "loss": 2.0533,
      "step": 9
    },
    {
      "epoch": 0.010002813291238161,
      "grad_norm": 2.233243703842163,
      "learning_rate": 0.00019999450595405374,
      "loss": 1.9378,
      "step": 10
    },
    {
      "epoch": 0.011003094620361977,
      "grad_norm": 1.302808165550232,
      "learning_rate": 0.0001999933522171883,
      "loss": 1.9182,
      "step": 11
    },
    {
      "epoch": 0.012003375949485792,
      "grad_norm": 0.8285257816314697,
      "learning_rate": 0.00019999208860571255,
      "loss": 1.9146,
      "step": 12
    },
    {
      "epoch": 0.01300365727860961,
      "grad_norm": 1.2248319387435913,
      "learning_rate": 0.00019999071512101496,
      "loss": 1.7467,
      "step": 13
    },
    {
      "epoch": 0.014003938607733425,
      "grad_norm": 0.8307135105133057,
      "learning_rate": 0.00019998923176460474,
      "loss": 1.6896,
      "step": 14
    },
    {
      "epoch": 0.01500421993685724,
      "grad_norm": 1.1531301736831665,
      "learning_rate": 0.00019998763853811184,
      "loss": 1.7549,
      "step": 15
    },
    {
      "epoch": 0.016004501265981057,
      "grad_norm": 1.0071958303451538,
      "learning_rate": 0.00019998593544328692,
      "loss": 1.903,
      "step": 16
    },
    {
      "epoch": 0.017004782595104875,
      "grad_norm": 0.9111937284469604,
      "learning_rate": 0.00019998412248200138,
      "loss": 1.8372,
      "step": 17
    },
    {
      "epoch": 0.01800506392422869,
      "grad_norm": 0.9943836331367493,
      "learning_rate": 0.00019998219965624734,
      "loss": 1.7304,
      "step": 18
    },
    {
      "epoch": 0.019005345253352506,
      "grad_norm": 0.8139007687568665,
      "learning_rate": 0.0001999801669681376,
      "loss": 1.6932,
      "step": 19
    },
    {
      "epoch": 0.020005626582476323,
      "grad_norm": 0.7991273999214172,
      "learning_rate": 0.00019997802441990573,
      "loss": 1.9596,
      "step": 20
    },
    {
      "epoch": 0.021005907911600136,
      "grad_norm": 0.832266628742218,
      "learning_rate": 0.00019997577201390606,
      "loss": 1.7116,
      "step": 21
    },
    {
      "epoch": 0.022006189240723954,
      "grad_norm": 0.8465655446052551,
      "learning_rate": 0.00019997340975261353,
      "loss": 1.7711,
      "step": 22
    },
    {
      "epoch": 0.02300647056984777,
      "grad_norm": 1.032426118850708,
      "learning_rate": 0.00019997093763862383,
      "loss": 1.6746,
      "step": 23
    },
    {
      "epoch": 0.024006751898971584,
      "grad_norm": 1.0036743879318237,
      "learning_rate": 0.0001999683556746534,
      "loss": 1.7274,
      "step": 24
    },
    {
      "epoch": 0.0250070332280954,
      "grad_norm": 0.9491412043571472,
      "learning_rate": 0.0001999656638635393,
      "loss": 2.0302,
      "step": 25
    },
    {
      "epoch": 0.02600731455721922,
      "grad_norm": 0.9477822184562683,
      "learning_rate": 0.0001999628622082394,
      "loss": 1.6107,
      "step": 26
    },
    {
      "epoch": 0.027007595886343033,
      "grad_norm": 1.0687041282653809,
      "learning_rate": 0.0001999599507118322,
      "loss": 1.8225,
      "step": 27
    },
    {
      "epoch": 0.02800787721546685,
      "grad_norm": 1.6572712659835815,
      "learning_rate": 0.00019995692937751683,
      "loss": 1.896,
      "step": 28
    },
    {
      "epoch": 0.029008158544590667,
      "grad_norm": 1.013258695602417,
      "learning_rate": 0.0001999537982086133,
      "loss": 1.7847,
      "step": 29
    },
    {
      "epoch": 0.03000843987371448,
      "grad_norm": 0.7584932446479797,
      "learning_rate": 0.00019995055720856218,
      "loss": 1.5841,
      "step": 30
    },
    {
      "epoch": 0.031008721202838298,
      "grad_norm": 1.1543537378311157,
      "learning_rate": 0.00019994720638092468,
      "loss": 1.8362,
      "step": 31
    },
    {
      "epoch": 0.032009002531962115,
      "grad_norm": 0.8389608860015869,
      "learning_rate": 0.00019994374572938277,
      "loss": 1.7913,
      "step": 32
    },
    {
      "epoch": 0.03300928386108593,
      "grad_norm": 0.7582125663757324,
      "learning_rate": 0.00019994017525773913,
      "loss": 1.5406,
      "step": 33
    },
    {
      "epoch": 0.03400956519020975,
      "grad_norm": 0.7866935133934021,
      "learning_rate": 0.00019993649496991705,
      "loss": 1.5363,
      "step": 34
    },
    {
      "epoch": 0.03500984651933356,
      "grad_norm": 0.8007768988609314,
      "learning_rate": 0.00019993270486996046,
      "loss": 1.7597,
      "step": 35
    },
    {
      "epoch": 0.03601012784845738,
      "grad_norm": 0.8109031319618225,
      "learning_rate": 0.000199928804962034,
      "loss": 1.5554,
      "step": 36
    },
    {
      "epoch": 0.037010409177581194,
      "grad_norm": 0.7722628116607666,
      "learning_rate": 0.00019992479525042303,
      "loss": 1.6437,
      "step": 37
    },
    {
      "epoch": 0.03801069050670501,
      "grad_norm": 0.7336480021476746,
      "learning_rate": 0.00019992067573953342,
      "loss": 1.7276,
      "step": 38
    },
    {
      "epoch": 0.03901097183582883,
      "grad_norm": 0.6940280795097351,
      "learning_rate": 0.0001999164464338918,
      "loss": 1.846,
      "step": 39
    },
    {
      "epoch": 0.040011253164952645,
      "grad_norm": 0.7079702615737915,
      "learning_rate": 0.0001999121073381454,
      "loss": 1.7017,
      "step": 40
    },
    {
      "epoch": 0.041011534494076456,
      "grad_norm": 0.7438498139381409,
      "learning_rate": 0.0001999076584570621,
      "loss": 1.665,
      "step": 41
    },
    {
      "epoch": 0.04201181582320027,
      "grad_norm": 0.6951525211334229,
      "learning_rate": 0.00019990309979553045,
      "loss": 1.588,
      "step": 42
    },
    {
      "epoch": 0.04301209715232409,
      "grad_norm": 0.9398604035377502,
      "learning_rate": 0.00019989843135855958,
      "loss": 1.6513,
      "step": 43
    },
    {
      "epoch": 0.04401237848144791,
      "grad_norm": 0.7384347319602966,
      "learning_rate": 0.00019989365315127922,
      "loss": 1.5975,
      "step": 44
    },
    {
      "epoch": 0.045012659810571724,
      "grad_norm": 0.9856846332550049,
      "learning_rate": 0.0001998887651789398,
      "loss": 1.644,
      "step": 45
    },
    {
      "epoch": 0.04601294113969554,
      "grad_norm": 0.7322820425033569,
      "learning_rate": 0.0001998837674469123,
      "loss": 1.5207,
      "step": 46
    },
    {
      "epoch": 0.04701322246881936,
      "grad_norm": 0.8695257902145386,
      "learning_rate": 0.00019987865996068833,
      "loss": 1.5572,
      "step": 47
    },
    {
      "epoch": 0.04801350379794317,
      "grad_norm": 0.7231017351150513,
      "learning_rate": 0.00019987344272588006,
      "loss": 1.5841,
      "step": 48
    },
    {
      "epoch": 0.049013785127066986,
      "grad_norm": 0.7147384285926819,
      "learning_rate": 0.00019986811574822033,
      "loss": 1.8628,
      "step": 49
    },
    {
      "epoch": 0.0500140664561908,
      "grad_norm": 0.8631477355957031,
      "learning_rate": 0.00019986267903356254,
      "loss": 1.8487,
      "step": 50
    },
    {
      "epoch": 0.05101434778531462,
      "grad_norm": 0.7995486855506897,
      "learning_rate": 0.0001998571325878806,
      "loss": 1.6491,
      "step": 51
    },
    {
      "epoch": 0.05201462911443844,
      "grad_norm": 0.7828657031059265,
      "learning_rate": 0.0001998514764172691,
      "loss": 1.7496,
      "step": 52
    },
    {
      "epoch": 0.053014910443562255,
      "grad_norm": 0.7789833545684814,
      "learning_rate": 0.00019984571052794313,
      "loss": 1.6628,
      "step": 53
    },
    {
      "epoch": 0.054015191772686065,
      "grad_norm": 0.7077661752700806,
      "learning_rate": 0.00019983983492623833,
      "loss": 1.771,
      "step": 54
    },
    {
      "epoch": 0.05501547310180988,
      "grad_norm": 0.7939582467079163,
      "learning_rate": 0.00019983384961861096,
      "loss": 1.707,
      "step": 55
    },
    {
      "epoch": 0.0560157544309337,
      "grad_norm": 0.9438828229904175,
      "learning_rate": 0.0001998277546116378,
      "loss": 1.8334,
      "step": 56
    },
    {
      "epoch": 0.05701603576005752,
      "grad_norm": 0.8028286695480347,
      "learning_rate": 0.00019982154991201608,
      "loss": 1.9117,
      "step": 57
    },
    {
      "epoch": 0.058016317089181334,
      "grad_norm": 0.6563037037849426,
      "learning_rate": 0.00019981523552656377,
      "loss": 1.4767,
      "step": 58
    },
    {
      "epoch": 0.05901659841830515,
      "grad_norm": 0.6600964665412903,
      "learning_rate": 0.00019980881146221914,
      "loss": 1.6656,
      "step": 59
    },
    {
      "epoch": 0.06001687974742896,
      "grad_norm": 0.7966578602790833,
      "learning_rate": 0.00019980227772604112,
      "loss": 1.4844,
      "step": 60
    },
    {
      "epoch": 0.06101716107655278,
      "grad_norm": 0.8019976615905762,
      "learning_rate": 0.0001997956343252091,
      "loss": 1.5682,
      "step": 61
    },
    {
      "epoch": 0.062017442405676595,
      "grad_norm": 0.8935349583625793,
      "learning_rate": 0.00019978888126702296,
      "loss": 1.8131,
      "step": 62
    },
    {
      "epoch": 0.06301772373480041,
      "grad_norm": 0.8085179924964905,
      "learning_rate": 0.00019978201855890308,
      "loss": 1.5602,
      "step": 63
    },
    {
      "epoch": 0.06401800506392423,
      "grad_norm": 0.7631951570510864,
      "learning_rate": 0.00019977504620839035,
      "loss": 1.8008,
      "step": 64
    },
    {
      "epoch": 0.06501828639304805,
      "grad_norm": 0.7315165996551514,
      "learning_rate": 0.00019976796422314615,
      "loss": 1.5735,
      "step": 65
    },
    {
      "epoch": 0.06601856772217186,
      "grad_norm": 0.745726466178894,
      "learning_rate": 0.00019976077261095226,
      "loss": 1.5775,
      "step": 66
    },
    {
      "epoch": 0.06701884905129568,
      "grad_norm": 0.9082249999046326,
      "learning_rate": 0.00019975347137971098,
      "loss": 1.7427,
      "step": 67
    },
    {
      "epoch": 0.0680191303804195,
      "grad_norm": 0.6575669050216675,
      "learning_rate": 0.00019974606053744503,
      "loss": 1.5231,
      "step": 68
    },
    {
      "epoch": 0.06901941170954332,
      "grad_norm": 0.7749233245849609,
      "learning_rate": 0.00019973854009229763,
      "loss": 1.5703,
      "step": 69
    },
    {
      "epoch": 0.07001969303866712,
      "grad_norm": 0.7240824699401855,
      "learning_rate": 0.00019973091005253232,
      "loss": 1.5197,
      "step": 70
    },
    {
      "epoch": 0.07101997436779094,
      "grad_norm": 0.8683856725692749,
      "learning_rate": 0.0001997231704265332,
      "loss": 1.6183,
      "step": 71
    },
    {
      "epoch": 0.07202025569691475,
      "grad_norm": 0.6885640621185303,
      "learning_rate": 0.00019971532122280464,
      "loss": 1.6565,
      "step": 72
    },
    {
      "epoch": 0.07302053702603857,
      "grad_norm": 0.6648329496383667,
      "learning_rate": 0.0001997073624499716,
      "loss": 1.5943,
      "step": 73
    },
    {
      "epoch": 0.07402081835516239,
      "grad_norm": 0.8867416977882385,
      "learning_rate": 0.0001996992941167792,
      "loss": 1.7855,
      "step": 74
    },
    {
      "epoch": 0.0750210996842862,
      "grad_norm": 0.7790491580963135,
      "learning_rate": 0.00019969111623209323,
      "loss": 1.6723,
      "step": 75
    },
    {
      "epoch": 0.07602138101341002,
      "grad_norm": 0.7999201416969299,
      "learning_rate": 0.00019968282880489957,
      "loss": 1.5619,
      "step": 76
    },
    {
      "epoch": 0.07702166234253384,
      "grad_norm": 0.6316407322883606,
      "learning_rate": 0.00019967443184430467,
      "loss": 1.6377,
      "step": 77
    },
    {
      "epoch": 0.07802194367165766,
      "grad_norm": 0.7680445313453674,
      "learning_rate": 0.0001996659253595353,
      "loss": 1.5433,
      "step": 78
    },
    {
      "epoch": 0.07902222500078147,
      "grad_norm": 0.7158446907997131,
      "learning_rate": 0.0001996573093599385,
      "loss": 1.5436,
      "step": 79
    },
    {
      "epoch": 0.08002250632990529,
      "grad_norm": 0.7354825139045715,
      "learning_rate": 0.00019964858385498172,
      "loss": 1.6512,
      "step": 80
    },
    {
      "epoch": 0.08102278765902911,
      "grad_norm": 0.7031210660934448,
      "learning_rate": 0.00019963974885425266,
      "loss": 1.6411,
      "step": 81
    },
    {
      "epoch": 0.08202306898815291,
      "grad_norm": 0.8451045751571655,
      "learning_rate": 0.00019963080436745945,
      "loss": 1.6622,
      "step": 82
    },
    {
      "epoch": 0.08302335031727673,
      "grad_norm": 0.8329521417617798,
      "learning_rate": 0.00019962175040443044,
      "loss": 1.9269,
      "step": 83
    },
    {
      "epoch": 0.08402363164640055,
      "grad_norm": 0.6967645883560181,
      "learning_rate": 0.0001996125869751143,
      "loss": 1.7243,
      "step": 84
    },
    {
      "epoch": 0.08502391297552436,
      "grad_norm": 0.8699042797088623,
      "learning_rate": 0.00019960331408957997,
      "loss": 1.7211,
      "step": 85
    },
    {
      "epoch": 0.08602419430464818,
      "grad_norm": 0.6780512928962708,
      "learning_rate": 0.00019959393175801671,
      "loss": 1.6376,
      "step": 86
    },
    {
      "epoch": 0.087024475633772,
      "grad_norm": 0.7213720679283142,
      "learning_rate": 0.00019958443999073397,
      "loss": 1.6048,
      "step": 87
    },
    {
      "epoch": 0.08802475696289581,
      "grad_norm": 0.6077585816383362,
      "learning_rate": 0.00019957483879816151,
      "loss": 1.5231,
      "step": 88
    },
    {
      "epoch": 0.08902503829201963,
      "grad_norm": 0.6854611039161682,
      "learning_rate": 0.00019956512819084928,
      "loss": 1.4726,
      "step": 89
    },
    {
      "epoch": 0.09002531962114345,
      "grad_norm": 0.6969390511512756,
      "learning_rate": 0.00019955530817946748,
      "loss": 1.6435,
      "step": 90
    },
    {
      "epoch": 0.09102560095026727,
      "grad_norm": 0.7178792953491211,
      "learning_rate": 0.00019954537877480655,
      "loss": 1.6967,
      "step": 91
    },
    {
      "epoch": 0.09202588227939108,
      "grad_norm": 0.8248458504676819,
      "learning_rate": 0.00019953533998777706,
      "loss": 1.5884,
      "step": 92
    },
    {
      "epoch": 0.0930261636085149,
      "grad_norm": 0.6472075581550598,
      "learning_rate": 0.00019952519182940993,
      "loss": 1.5074,
      "step": 93
    },
    {
      "epoch": 0.09402644493763872,
      "grad_norm": 0.7548672556877136,
      "learning_rate": 0.00019951493431085603,
      "loss": 1.6934,
      "step": 94
    },
    {
      "epoch": 0.09502672626676252,
      "grad_norm": 0.6680666208267212,
      "learning_rate": 0.00019950456744338658,
      "loss": 1.4014,
      "step": 95
    },
    {
      "epoch": 0.09602700759588634,
      "grad_norm": 0.7270862460136414,
      "learning_rate": 0.00019949409123839288,
      "loss": 1.6991,
      "step": 96
    },
    {
      "epoch": 0.09702728892501016,
      "grad_norm": 0.682833731174469,
      "learning_rate": 0.00019948350570738642,
      "loss": 1.4926,
      "step": 97
    },
    {
      "epoch": 0.09802757025413397,
      "grad_norm": 0.6598315238952637,
      "learning_rate": 0.0001994728108619987,
      "loss": 1.6329,
      "step": 98
    },
    {
      "epoch": 0.09902785158325779,
      "grad_norm": 0.830845832824707,
      "learning_rate": 0.0001994620067139815,
      "loss": 1.8517,
      "step": 99
    },
    {
      "epoch": 0.1000281329123816,
      "grad_norm": 0.6964694857597351,
      "learning_rate": 0.00019945109327520658,
      "loss": 1.5459,
      "step": 100
    },
    {
      "epoch": 0.10102841424150542,
      "grad_norm": 0.8640177249908447,
      "learning_rate": 0.00019944007055766586,
      "loss": 1.6638,
      "step": 101
    },
    {
      "epoch": 0.10202869557062924,
      "grad_norm": 0.6485210657119751,
      "learning_rate": 0.00019942893857347128,
      "loss": 1.8025,
      "step": 102
    },
    {
      "epoch": 0.10302897689975306,
      "grad_norm": 0.6746248006820679,
      "learning_rate": 0.00019941769733485494,
      "loss": 1.6954,
      "step": 103
    },
    {
      "epoch": 0.10402925822887688,
      "grad_norm": 0.7386549115180969,
      "learning_rate": 0.00019940634685416888,
      "loss": 1.4547,
      "step": 104
    },
    {
      "epoch": 0.10502953955800069,
      "grad_norm": 0.7518633008003235,
      "learning_rate": 0.00019939488714388524,
      "loss": 1.5098,
      "step": 105
    },
    {
      "epoch": 0.10602982088712451,
      "grad_norm": 0.7350422739982605,
      "learning_rate": 0.00019938331821659614,
      "loss": 1.5452,
      "step": 106
    },
    {
      "epoch": 0.10703010221624833,
      "grad_norm": 0.6544668674468994,
      "learning_rate": 0.0001993716400850138,
      "loss": 1.5106,
      "step": 107
    },
    {
      "epoch": 0.10803038354537213,
      "grad_norm": 0.6108564138412476,
      "learning_rate": 0.0001993598527619703,
      "loss": 1.5818,
      "step": 108
    },
    {
      "epoch": 0.10903066487449595,
      "grad_norm": 0.731071949005127,
      "learning_rate": 0.00019934795626041783,
      "loss": 1.4819,
      "step": 109
    },
    {
      "epoch": 0.11003094620361976,
      "grad_norm": 0.5978986620903015,
      "learning_rate": 0.0001993359505934285,
      "loss": 1.5469,
      "step": 110
    },
    {
      "epoch": 0.11103122753274358,
      "grad_norm": 0.7249881029129028,
      "learning_rate": 0.00019932383577419432,
      "loss": 1.7466,
      "step": 111
    },
    {
      "epoch": 0.1120315088618674,
      "grad_norm": 0.6161806583404541,
      "learning_rate": 0.0001993116118160273,
      "loss": 1.3411,
      "step": 112
    },
    {
      "epoch": 0.11303179019099122,
      "grad_norm": 0.6745229363441467,
      "learning_rate": 0.00019929927873235938,
      "loss": 1.5615,
      "step": 113
    },
    {
      "epoch": 0.11403207152011503,
      "grad_norm": 0.6489872336387634,
      "learning_rate": 0.00019928683653674237,
      "loss": 1.6279,
      "step": 114
    },
    {
      "epoch": 0.11503235284923885,
      "grad_norm": 0.7769975662231445,
      "learning_rate": 0.00019927428524284805,
      "loss": 1.6155,
      "step": 115
    },
    {
      "epoch": 0.11603263417836267,
      "grad_norm": 0.734336793422699,
      "learning_rate": 0.00019926162486446792,
      "loss": 1.6837,
      "step": 116
    },
    {
      "epoch": 0.11703291550748648,
      "grad_norm": 0.6966903805732727,
      "learning_rate": 0.0001992488554155135,
      "loss": 1.6659,
      "step": 117
    },
    {
      "epoch": 0.1180331968366103,
      "grad_norm": 0.6714586615562439,
      "learning_rate": 0.00019923597691001615,
      "loss": 1.5161,
      "step": 118
    },
    {
      "epoch": 0.11903347816573412,
      "grad_norm": 0.6390894651412964,
      "learning_rate": 0.0001992229893621269,
      "loss": 1.4561,
      "step": 119
    },
    {
      "epoch": 0.12003375949485792,
      "grad_norm": 0.6481143832206726,
      "learning_rate": 0.00019920989278611687,
      "loss": 1.6331,
      "step": 120
    },
    {
      "epoch": 0.12103404082398174,
      "grad_norm": 0.6819384694099426,
      "learning_rate": 0.0001991966871963767,
      "loss": 1.6508,
      "step": 121
    },
    {
      "epoch": 0.12203432215310556,
      "grad_norm": 0.6839059591293335,
      "learning_rate": 0.000199183372607417,
      "loss": 1.6514,
      "step": 122
    },
    {
      "epoch": 0.12303460348222937,
      "grad_norm": 0.6401050090789795,
      "learning_rate": 0.0001991699490338681,
      "loss": 1.8065,
      "step": 123
    },
    {
      "epoch": 0.12403488481135319,
      "grad_norm": 0.6860588788986206,
      "learning_rate": 0.00019915641649048005,
      "loss": 1.7658,
      "step": 124
    },
    {
      "epoch": 0.12503516614047702,
      "grad_norm": 0.6286434531211853,
      "learning_rate": 0.0001991427749921227,
      "loss": 1.7678,
      "step": 125
    },
    {
      "epoch": 0.12603544746960083,
      "grad_norm": 0.6609922647476196,
      "learning_rate": 0.00019912902455378556,
      "loss": 1.4934,
      "step": 126
    },
    {
      "epoch": 0.12703572879872463,
      "grad_norm": 0.7058399319648743,
      "learning_rate": 0.00019911516519057788,
      "loss": 1.6058,
      "step": 127
    },
    {
      "epoch": 0.12803601012784846,
      "grad_norm": 0.6362051963806152,
      "learning_rate": 0.00019910119691772863,
      "loss": 1.502,
      "step": 128
    },
    {
      "epoch": 0.12903629145697226,
      "grad_norm": 0.7493100762367249,
      "learning_rate": 0.00019908711975058637,
      "loss": 1.5287,
      "step": 129
    },
    {
      "epoch": 0.1300365727860961,
      "grad_norm": 0.6492393612861633,
      "learning_rate": 0.0001990729337046194,
      "loss": 1.5716,
      "step": 130
    },
    {
      "epoch": 0.1310368541152199,
      "grad_norm": 0.70331871509552,
      "learning_rate": 0.0001990586387954156,
      "loss": 1.5882,
      "step": 131
    },
    {
      "epoch": 0.13203713544434373,
      "grad_norm": 0.7581572532653809,
      "learning_rate": 0.00019904423503868247,
      "loss": 1.7627,
      "step": 132
    },
    {
      "epoch": 0.13303741677346753,
      "grad_norm": 0.7087228894233704,
      "learning_rate": 0.00019902972245024715,
      "loss": 1.6257,
      "step": 133
    },
    {
      "epoch": 0.13403769810259136,
      "grad_norm": 0.7920627593994141,
      "learning_rate": 0.00019901510104605637,
      "loss": 1.572,
      "step": 134
    },
    {
      "epoch": 0.13503797943171517,
      "grad_norm": 0.6869202256202698,
      "learning_rate": 0.00019900037084217637,
      "loss": 1.5478,
      "step": 135
    },
    {
      "epoch": 0.136038260760839,
      "grad_norm": 0.6879409551620483,
      "learning_rate": 0.00019898553185479303,
      "loss": 1.3104,
      "step": 136
    },
    {
      "epoch": 0.1370385420899628,
      "grad_norm": 0.6574143767356873,
      "learning_rate": 0.00019897058410021167,
      "loss": 1.7041,
      "step": 137
    },
    {
      "epoch": 0.13803882341908663,
      "grad_norm": 0.7793259620666504,
      "learning_rate": 0.00019895552759485722,
      "loss": 1.5417,
      "step": 138
    },
    {
      "epoch": 0.13903910474821043,
      "grad_norm": 0.6310438513755798,
      "learning_rate": 0.00019894036235527395,
      "loss": 1.4978,
      "step": 139
    },
    {
      "epoch": 0.14003938607733424,
      "grad_norm": 0.6298012137413025,
      "learning_rate": 0.00019892508839812584,
      "loss": 1.5367,
      "step": 140
    },
    {
      "epoch": 0.14103966740645807,
      "grad_norm": 0.5647856593132019,
      "learning_rate": 0.00019890970574019617,
      "loss": 1.537,
      "step": 141
    },
    {
      "epoch": 0.14203994873558187,
      "grad_norm": 0.6491876244544983,
      "learning_rate": 0.00019889421439838763,
      "loss": 1.6992,
      "step": 142
    },
    {
      "epoch": 0.1430402300647057,
      "grad_norm": 0.6574720144271851,
      "learning_rate": 0.00019887861438972246,
      "loss": 1.3837,
      "step": 143
    },
    {
      "epoch": 0.1440405113938295,
      "grad_norm": 0.6267092227935791,
      "learning_rate": 0.00019886290573134228,
      "loss": 1.6307,
      "step": 144
    },
    {
      "epoch": 0.14504079272295334,
      "grad_norm": 0.6785029172897339,
      "learning_rate": 0.000198847088440508,
      "loss": 1.574,
      "step": 145
    },
    {
      "epoch": 0.14604107405207714,
      "grad_norm": 0.6218644380569458,
      "learning_rate": 0.0001988311625346,
      "loss": 1.4676,
      "step": 146
    },
    {
      "epoch": 0.14704135538120097,
      "grad_norm": 0.6047986745834351,
      "learning_rate": 0.00019881512803111796,
      "loss": 1.4316,
      "step": 147
    },
    {
      "epoch": 0.14804163671032478,
      "grad_norm": 0.7340937256813049,
      "learning_rate": 0.00019879898494768093,
      "loss": 1.5185,
      "step": 148
    },
    {
      "epoch": 0.1490419180394486,
      "grad_norm": 0.5874620676040649,
      "learning_rate": 0.00019878273330202717,
      "loss": 1.5031,
      "step": 149
    },
    {
      "epoch": 0.1500421993685724,
      "grad_norm": 0.6943556666374207,
      "learning_rate": 0.00019876637311201433,
      "loss": 1.7323,
      "step": 150
    },
    {
      "epoch": 0.15104248069769624,
      "grad_norm": 0.6345832347869873,
      "learning_rate": 0.00019874990439561934,
      "loss": 1.4691,
      "step": 151
    },
    {
      "epoch": 0.15204276202682004,
      "grad_norm": 0.7047753930091858,
      "learning_rate": 0.0001987333271709383,
      "loss": 1.5198,
      "step": 152
    },
    {
      "epoch": 0.15304304335594385,
      "grad_norm": 0.6043322086334229,
      "learning_rate": 0.00019871664145618657,
      "loss": 1.5488,
      "step": 153
    },
    {
      "epoch": 0.15404332468506768,
      "grad_norm": 0.5978446006774902,
      "learning_rate": 0.00019869984726969878,
      "loss": 1.4278,
      "step": 154
    },
    {
      "epoch": 0.15504360601419148,
      "grad_norm": 0.6796436905860901,
      "learning_rate": 0.00019868294462992866,
      "loss": 1.5845,
      "step": 155
    },
    {
      "epoch": 0.1560438873433153,
      "grad_norm": 0.7113372087478638,
      "learning_rate": 0.00019866593355544922,
      "loss": 1.7509,
      "step": 156
    },
    {
      "epoch": 0.15704416867243912,
      "grad_norm": 0.5908107757568359,
      "learning_rate": 0.00019864881406495246,
      "loss": 1.5693,
      "step": 157
    },
    {
      "epoch": 0.15804445000156295,
      "grad_norm": 0.7135252952575684,
      "learning_rate": 0.00019863158617724967,
      "loss": 1.6109,
      "step": 158
    },
    {
      "epoch": 0.15904473133068675,
      "grad_norm": 0.5621710419654846,
      "learning_rate": 0.00019861424991127115,
      "loss": 1.5368,
      "step": 159
    },
    {
      "epoch": 0.16004501265981058,
      "grad_norm": 0.6205443143844604,
      "learning_rate": 0.00019859680528606637,
      "loss": 1.5181,
      "step": 160
    },
    {
      "epoch": 0.16104529398893438,
      "grad_norm": 0.6933260560035706,
      "learning_rate": 0.00019857925232080373,
      "loss": 1.4508,
      "step": 161
    },
    {
      "epoch": 0.16204557531805822,
      "grad_norm": 0.6911661028862,
      "learning_rate": 0.00019856159103477086,
      "loss": 1.5423,
      "step": 162
    },
    {
      "epoch": 0.16304585664718202,
      "grad_norm": 0.7684744000434875,
      "learning_rate": 0.00019854382144737426,
      "loss": 1.4097,
      "step": 163
    },
    {
      "epoch": 0.16404613797630582,
      "grad_norm": 0.6657288074493408,
      "learning_rate": 0.00019852594357813952,
      "loss": 1.6145,
      "step": 164
    },
    {
      "epoch": 0.16504641930542965,
      "grad_norm": 0.7030160427093506,
      "learning_rate": 0.00019850795744671116,
      "loss": 1.6551,
      "step": 165
    },
    {
      "epoch": 0.16604670063455346,
      "grad_norm": 0.87894207239151,
      "learning_rate": 0.0001984898630728527,
      "loss": 1.6316,
      "step": 166
    },
    {
      "epoch": 0.1670469819636773,
      "grad_norm": 0.6282681226730347,
      "learning_rate": 0.0001984716604764466,
      "loss": 1.451,
      "step": 167
    },
    {
      "epoch": 0.1680472632928011,
      "grad_norm": 0.6729792952537537,
      "learning_rate": 0.0001984533496774942,
      "loss": 1.4381,
      "step": 168
    },
    {
      "epoch": 0.16904754462192492,
      "grad_norm": 0.7300116419792175,
      "learning_rate": 0.0001984349306961158,
      "loss": 1.4244,
      "step": 169
    },
    {
      "epoch": 0.17004782595104873,
      "grad_norm": 0.6853480935096741,
      "learning_rate": 0.00019841640355255043,
      "loss": 1.6174,
      "step": 170
    },
    {
      "epoch": 0.17104810728017256,
      "grad_norm": 0.735612690448761,
      "learning_rate": 0.00019839776826715614,
      "loss": 1.5085,
      "step": 171
    },
    {
      "epoch": 0.17204838860929636,
      "grad_norm": 0.6735563278198242,
      "learning_rate": 0.00019837902486040978,
      "loss": 1.507,
      "step": 172
    },
    {
      "epoch": 0.1730486699384202,
      "grad_norm": 0.6617917418479919,
      "learning_rate": 0.0001983601733529069,
      "loss": 1.6774,
      "step": 173
    },
    {
      "epoch": 0.174048951267544,
      "grad_norm": 0.7137823700904846,
      "learning_rate": 0.00019834121376536187,
      "loss": 1.4665,
      "step": 174
    },
    {
      "epoch": 0.17504923259666783,
      "grad_norm": 0.6372626423835754,
      "learning_rate": 0.00019832214611860793,
      "loss": 1.3597,
      "step": 175
    },
    {
      "epoch": 0.17604951392579163,
      "grad_norm": 0.7131632566452026,
      "learning_rate": 0.00019830297043359692,
      "loss": 1.4833,
      "step": 176
    },
    {
      "epoch": 0.17704979525491543,
      "grad_norm": 0.7538559436798096,
      "learning_rate": 0.00019828368673139947,
      "loss": 1.4714,
      "step": 177
    },
    {
      "epoch": 0.17805007658403926,
      "grad_norm": 0.5684806108474731,
      "learning_rate": 0.0001982642950332049,
      "loss": 1.5012,
      "step": 178
    },
    {
      "epoch": 0.17905035791316307,
      "grad_norm": 0.621658444404602,
      "learning_rate": 0.00019824479536032112,
      "loss": 1.9119,
      "step": 179
    },
    {
      "epoch": 0.1800506392422869,
      "grad_norm": 0.6564679741859436,
      "learning_rate": 0.0001982251877341748,
      "loss": 1.5131,
      "step": 180
    },
    {
      "epoch": 0.1810509205714107,
      "grad_norm": 0.6546526551246643,
      "learning_rate": 0.00019820547217631117,
      "loss": 1.4493,
      "step": 181
    },
    {
      "epoch": 0.18205120190053453,
      "grad_norm": 0.6504479050636292,
      "learning_rate": 0.00019818564870839405,
      "loss": 1.6131,
      "step": 182
    },
    {
      "epoch": 0.18305148322965833,
      "grad_norm": 0.6269803047180176,
      "learning_rate": 0.00019816571735220583,
      "loss": 1.5936,
      "step": 183
    },
    {
      "epoch": 0.18405176455878217,
      "grad_norm": 0.6303942799568176,
      "learning_rate": 0.00019814567812964748,
      "loss": 1.6948,
      "step": 184
    },
    {
      "epoch": 0.18505204588790597,
      "grad_norm": 0.6562885046005249,
      "learning_rate": 0.00019812553106273847,
      "loss": 1.5542,
      "step": 185
    },
    {
      "epoch": 0.1860523272170298,
      "grad_norm": 0.5844212174415588,
      "learning_rate": 0.00019810527617361681,
      "loss": 1.539,
      "step": 186
    },
    {
      "epoch": 0.1870526085461536,
      "grad_norm": 0.6402295231819153,
      "learning_rate": 0.00019808491348453894,
      "loss": 1.4748,
      "step": 187
    },
    {
      "epoch": 0.18805288987527743,
      "grad_norm": 0.6579477190971375,
      "learning_rate": 0.00019806444301787978,
      "loss": 1.5114,
      "step": 188
    },
    {
      "epoch": 0.18905317120440124,
      "grad_norm": 0.6511597037315369,
      "learning_rate": 0.0001980438647961327,
      "loss": 1.4678,
      "step": 189
    },
    {
      "epoch": 0.19005345253352504,
      "grad_norm": 0.6911427974700928,
      "learning_rate": 0.00019802317884190935,
      "loss": 1.6876,
      "step": 190
    },
    {
      "epoch": 0.19105373386264887,
      "grad_norm": 0.6146433353424072,
      "learning_rate": 0.00019800238517793996,
      "loss": 1.5986,
      "step": 191
    },
    {
      "epoch": 0.19205401519177268,
      "grad_norm": 0.6126302480697632,
      "learning_rate": 0.00019798148382707296,
      "loss": 1.571,
      "step": 192
    },
    {
      "epoch": 0.1930542965208965,
      "grad_norm": 0.5751072764396667,
      "learning_rate": 0.00019796047481227515,
      "loss": 1.4921,
      "step": 193
    },
    {
      "epoch": 0.1940545778500203,
      "grad_norm": 0.6484839916229248,
      "learning_rate": 0.00019793935815663163,
      "loss": 1.7495,
      "step": 194
    },
    {
      "epoch": 0.19505485917914414,
      "grad_norm": 0.6875973343849182,
      "learning_rate": 0.00019791813388334581,
      "loss": 1.5782,
      "step": 195
    },
    {
      "epoch": 0.19605514050826794,
      "grad_norm": 0.8130943179130554,
      "learning_rate": 0.00019789680201573933,
      "loss": 1.4964,
      "step": 196
    },
    {
      "epoch": 0.19705542183739178,
      "grad_norm": 0.6734403371810913,
      "learning_rate": 0.00019787536257725202,
      "loss": 1.4787,
      "step": 197
    },
    {
      "epoch": 0.19805570316651558,
      "grad_norm": 0.6480582356452942,
      "learning_rate": 0.00019785381559144196,
      "loss": 1.5629,
      "step": 198
    },
    {
      "epoch": 0.1990559844956394,
      "grad_norm": 0.6554624438285828,
      "learning_rate": 0.00019783216108198542,
      "loss": 1.5806,
      "step": 199
    },
    {
      "epoch": 0.2000562658247632,
      "grad_norm": 0.705443263053894,
      "learning_rate": 0.00019781039907267677,
      "loss": 1.8372,
      "step": 200
    },
    {
      "epoch": 0.20105654715388704,
      "grad_norm": 0.706923246383667,
      "learning_rate": 0.00019778852958742853,
      "loss": 1.6405,
      "step": 201
    },
    {
      "epoch": 0.20205682848301085,
      "grad_norm": 0.7062544822692871,
      "learning_rate": 0.00019776655265027127,
      "loss": 1.6,
      "step": 202
    },
    {
      "epoch": 0.20305710981213465,
      "grad_norm": 0.7227569222450256,
      "learning_rate": 0.00019774446828535371,
      "loss": 1.5172,
      "step": 203
    },
    {
      "epoch": 0.20405739114125848,
      "grad_norm": 0.6762563586235046,
      "learning_rate": 0.00019772227651694256,
      "loss": 1.6753,
      "step": 204
    },
    {
      "epoch": 0.20505767247038229,
      "grad_norm": 0.6048421859741211,
      "learning_rate": 0.00019769997736942258,
      "loss": 1.4827,
      "step": 205
    },
    {
      "epoch": 0.20605795379950612,
      "grad_norm": 0.6002956032752991,
      "learning_rate": 0.00019767757086729647,
      "loss": 1.5438,
      "step": 206
    },
    {
      "epoch": 0.20705823512862992,
      "grad_norm": 0.7948954701423645,
      "learning_rate": 0.00019765505703518496,
      "loss": 1.4988,
      "step": 207
    },
    {
      "epoch": 0.20805851645775375,
      "grad_norm": 0.6495680809020996,
      "learning_rate": 0.00019763243589782662,
      "loss": 1.5738,
      "step": 208
    },
    {
      "epoch": 0.20905879778687755,
      "grad_norm": 0.6413107514381409,
      "learning_rate": 0.00019760970748007803,
      "loss": 1.3794,
      "step": 209
    },
    {
      "epoch": 0.21005907911600138,
      "grad_norm": 0.5999665260314941,
      "learning_rate": 0.0001975868718069136,
      "loss": 1.4313,
      "step": 210
    },
    {
      "epoch": 0.2110593604451252,
      "grad_norm": 0.6355773210525513,
      "learning_rate": 0.00019756392890342563,
      "loss": 1.5107,
      "step": 211
    },
    {
      "epoch": 0.21205964177424902,
      "grad_norm": 0.6068251729011536,
      "learning_rate": 0.00019754087879482422,
      "loss": 1.536,
      "step": 212
    },
    {
      "epoch": 0.21305992310337282,
      "grad_norm": 0.5568909049034119,
      "learning_rate": 0.00019751772150643722,
      "loss": 1.5372,
      "step": 213
    },
    {
      "epoch": 0.21406020443249665,
      "grad_norm": 0.5771281719207764,
      "learning_rate": 0.00019749445706371038,
      "loss": 1.487,
      "step": 214
    },
    {
      "epoch": 0.21506048576162046,
      "grad_norm": 0.6146671772003174,
      "learning_rate": 0.00019747108549220702,
      "loss": 1.4585,
      "step": 215
    },
    {
      "epoch": 0.21606076709074426,
      "grad_norm": 0.5595754981040955,
      "learning_rate": 0.00019744760681760832,
      "loss": 1.4224,
      "step": 216
    },
    {
      "epoch": 0.2170610484198681,
      "grad_norm": 0.5873929858207703,
      "learning_rate": 0.00019742402106571314,
      "loss": 1.4581,
      "step": 217
    },
    {
      "epoch": 0.2180613297489919,
      "grad_norm": 0.5725668668746948,
      "learning_rate": 0.00019740032826243788,
      "loss": 1.4393,
      "step": 218
    },
    {
      "epoch": 0.21906161107811573,
      "grad_norm": 0.6452648043632507,
      "learning_rate": 0.0001973765284338167,
      "loss": 1.6048,
      "step": 219
    },
    {
      "epoch": 0.22006189240723953,
      "grad_norm": 0.6166092753410339,
      "learning_rate": 0.00019735262160600127,
      "loss": 1.4976,
      "step": 220
    },
    {
      "epoch": 0.22106217373636336,
      "grad_norm": 0.7053269147872925,
      "learning_rate": 0.00019732860780526088,
      "loss": 1.6882,
      "step": 221
    },
    {
      "epoch": 0.22206245506548716,
      "grad_norm": 0.7072796821594238,
      "learning_rate": 0.00019730448705798239,
      "loss": 1.5441,
      "step": 222
    },
    {
      "epoch": 0.223062736394611,
      "grad_norm": 0.6704496145248413,
      "learning_rate": 0.00019728025939067008,
      "loss": 1.3791,
      "step": 223
    },
    {
      "epoch": 0.2240630177237348,
      "grad_norm": 0.6141743659973145,
      "learning_rate": 0.00019725592482994583,
      "loss": 1.5831,
      "step": 224
    },
    {
      "epoch": 0.22506329905285863,
      "grad_norm": 0.6235673427581787,
      "learning_rate": 0.00019723148340254892,
      "loss": 1.6103,
      "step": 225
    },
    {
      "epoch": 0.22606358038198243,
      "grad_norm": 0.6383673548698425,
      "learning_rate": 0.00019720693513533598,
      "loss": 1.6284,
      "step": 226
    },
    {
      "epoch": 0.22706386171110624,
      "grad_norm": 0.7666104435920715,
      "learning_rate": 0.00019718228005528122,
      "loss": 1.702,
      "step": 227
    },
    {
      "epoch": 0.22806414304023007,
      "grad_norm": 0.6431383490562439,
      "learning_rate": 0.00019715751818947603,
      "loss": 1.4571,
      "step": 228
    },
    {
      "epoch": 0.22906442436935387,
      "grad_norm": 0.6177626252174377,
      "learning_rate": 0.0001971326495651293,
      "loss": 1.4326,
      "step": 229
    },
    {
      "epoch": 0.2300647056984777,
      "grad_norm": 0.7352898120880127,
      "learning_rate": 0.00019710767420956705,
      "loss": 1.7427,
      "step": 230
    },
    {
      "epoch": 0.2310649870276015,
      "grad_norm": 0.6259469389915466,
      "learning_rate": 0.0001970825921502328,
      "loss": 1.634,
      "step": 231
    },
    {
      "epoch": 0.23206526835672533,
      "grad_norm": 0.6699635982513428,
      "learning_rate": 0.0001970574034146871,
      "loss": 1.4705,
      "step": 232
    },
    {
      "epoch": 0.23306554968584914,
      "grad_norm": 0.5577033162117004,
      "learning_rate": 0.00019703210803060782,
      "loss": 1.5438,
      "step": 233
    },
    {
      "epoch": 0.23406583101497297,
      "grad_norm": 0.6063429117202759,
      "learning_rate": 0.00019700670602579008,
      "loss": 1.555,
      "step": 234
    },
    {
      "epoch": 0.23506611234409677,
      "grad_norm": 0.6069104671478271,
      "learning_rate": 0.00019698119742814606,
      "loss": 1.5036,
      "step": 235
    },
    {
      "epoch": 0.2360663936732206,
      "grad_norm": 0.6158379316329956,
      "learning_rate": 0.00019695558226570507,
      "loss": 1.3741,
      "step": 236
    },
    {
      "epoch": 0.2370666750023444,
      "grad_norm": 0.6366294622421265,
      "learning_rate": 0.00019692986056661356,
      "loss": 1.4467,
      "step": 237
    },
    {
      "epoch": 0.23806695633146824,
      "grad_norm": 0.6726595163345337,
      "learning_rate": 0.00019690403235913504,
      "loss": 1.3861,
      "step": 238
    },
    {
      "epoch": 0.23906723766059204,
      "grad_norm": 0.6546512842178345,
      "learning_rate": 0.00019687809767165,
      "loss": 1.6886,
      "step": 239
    },
    {
      "epoch": 0.24006751898971584,
      "grad_norm": 0.6623121500015259,
      "learning_rate": 0.000196852056532656,
      "loss": 1.5925,
      "step": 240
    },
    {
      "epoch": 0.24106780031883968,
      "grad_norm": 0.6577529311180115,
      "learning_rate": 0.00019682590897076752,
      "loss": 1.4509,
      "step": 241
    },
    {
      "epoch": 0.24206808164796348,
      "grad_norm": 0.5586327314376831,
      "learning_rate": 0.00019679965501471608,
      "loss": 1.6346,
      "step": 242
    },
    {
      "epoch": 0.2430683629770873,
      "grad_norm": 0.6459937691688538,
      "learning_rate": 0.0001967732946933499,
      "loss": 1.4129,
      "step": 243
    },
    {
      "epoch": 0.2440686443062111,
      "grad_norm": 0.778732180595398,
      "learning_rate": 0.00019674682803563428,
      "loss": 1.5129,
      "step": 244
    },
    {
      "epoch": 0.24506892563533494,
      "grad_norm": 0.7264451384544373,
      "learning_rate": 0.00019672025507065131,
      "loss": 1.4483,
      "step": 245
    },
    {
      "epoch": 0.24606920696445875,
      "grad_norm": 0.616084635257721,
      "learning_rate": 0.00019669357582759983,
      "loss": 1.5947,
      "step": 246
    },
    {
      "epoch": 0.24706948829358258,
      "grad_norm": 0.5911642909049988,
      "learning_rate": 0.00019666679033579552,
      "loss": 1.6407,
      "step": 247
    },
    {
      "epoch": 0.24806976962270638,
      "grad_norm": 0.6102796792984009,
      "learning_rate": 0.00019663989862467082,
      "loss": 1.5251,
      "step": 248
    },
    {
      "epoch": 0.2490700509518302,
      "grad_norm": 0.5973434448242188,
      "learning_rate": 0.00019661290072377482,
      "loss": 1.3969,
      "step": 249
    },
    {
      "epoch": 0.25007033228095404,
      "grad_norm": 0.8515523076057434,
      "learning_rate": 0.00019658579666277334,
      "loss": 1.5687,
      "step": 250
    },
    {
      "epoch": 0.2510706136100778,
      "grad_norm": 0.5003417134284973,
      "learning_rate": 0.0001965585864714488,
      "loss": 1.4102,
      "step": 251
    },
    {
      "epoch": 0.25207089493920165,
      "grad_norm": 0.5215190052986145,
      "learning_rate": 0.00019653127017970034,
      "loss": 1.2471,
      "step": 252
    },
    {
      "epoch": 0.2530711762683255,
      "grad_norm": 0.6491619348526001,
      "learning_rate": 0.0001965038478175436,
      "loss": 1.6969,
      "step": 253
    },
    {
      "epoch": 0.25407145759744926,
      "grad_norm": 0.6176133155822754,
      "learning_rate": 0.00019647631941511082,
      "loss": 1.5351,
      "step": 254
    },
    {
      "epoch": 0.2550717389265731,
      "grad_norm": 0.6913408041000366,
      "learning_rate": 0.0001964486850026507,
      "loss": 1.4309,
      "step": 255
    },
    {
      "epoch": 0.2560720202556969,
      "grad_norm": 0.5875718593597412,
      "learning_rate": 0.00019642094461052852,
      "loss": 1.4679,
      "step": 256
    },
    {
      "epoch": 0.25707230158482075,
      "grad_norm": 0.6682264804840088,
      "learning_rate": 0.00019639309826922585,
      "loss": 1.5393,
      "step": 257
    },
    {
      "epoch": 0.2580725829139445,
      "grad_norm": 0.7241432666778564,
      "learning_rate": 0.0001963651460093409,
      "loss": 1.4998,
      "step": 258
    },
    {
      "epoch": 0.25907286424306836,
      "grad_norm": 0.5210353136062622,
      "learning_rate": 0.00019633708786158806,
      "loss": 1.3837,
      "step": 259
    },
    {
      "epoch": 0.2600731455721922,
      "grad_norm": 0.584020733833313,
      "learning_rate": 0.00019630892385679818,
      "loss": 1.4961,
      "step": 260
    },
    {
      "epoch": 0.261073426901316,
      "grad_norm": 0.6708115935325623,
      "learning_rate": 0.00019628065402591845,
      "loss": 1.5277,
      "step": 261
    },
    {
      "epoch": 0.2620737082304398,
      "grad_norm": 0.5480003952980042,
      "learning_rate": 0.00019625227840001225,
      "loss": 1.556,
      "step": 262
    },
    {
      "epoch": 0.2630739895595636,
      "grad_norm": 0.595191478729248,
      "learning_rate": 0.0001962237970102593,
      "loss": 1.3514,
      "step": 263
    },
    {
      "epoch": 0.26407427088868746,
      "grad_norm": 0.7332099080085754,
      "learning_rate": 0.0001961952098879555,
      "loss": 1.5394,
      "step": 264
    },
    {
      "epoch": 0.26507455221781123,
      "grad_norm": 0.596319317817688,
      "learning_rate": 0.00019616651706451287,
      "loss": 1.3828,
      "step": 265
    },
    {
      "epoch": 0.26607483354693506,
      "grad_norm": 0.5998026132583618,
      "learning_rate": 0.0001961377185714597,
      "loss": 1.4479,
      "step": 266
    },
    {
      "epoch": 0.2670751148760589,
      "grad_norm": 0.6220220923423767,
      "learning_rate": 0.0001961088144404403,
      "loss": 1.5121,
      "step": 267
    },
    {
      "epoch": 0.2680753962051827,
      "grad_norm": 0.5865943431854248,
      "learning_rate": 0.00019607980470321505,
      "loss": 1.6747,
      "step": 268
    },
    {
      "epoch": 0.2690756775343065,
      "grad_norm": 0.5790852904319763,
      "learning_rate": 0.00019605068939166045,
      "loss": 1.3798,
      "step": 269
    },
    {
      "epoch": 0.27007595886343033,
      "grad_norm": 0.6157498955726624,
      "learning_rate": 0.00019602146853776894,
      "loss": 1.6799,
      "step": 270
    },
    {
      "epoch": 0.27107624019255416,
      "grad_norm": 0.6214422583580017,
      "learning_rate": 0.000195992142173649,
      "loss": 1.4782,
      "step": 271
    },
    {
      "epoch": 0.272076521521678,
      "grad_norm": 0.6460129618644714,
      "learning_rate": 0.0001959627103315249,
      "loss": 1.4874,
      "step": 272
    },
    {
      "epoch": 0.27307680285080177,
      "grad_norm": 0.5928930640220642,
      "learning_rate": 0.00019593317304373705,
      "loss": 1.4557,
      "step": 273
    },
    {
      "epoch": 0.2740770841799256,
      "grad_norm": 0.5123687982559204,
      "learning_rate": 0.00019590353034274144,
      "loss": 1.445,
      "step": 274
    },
    {
      "epoch": 0.27507736550904943,
      "grad_norm": 0.607455313205719,
      "learning_rate": 0.00019587378226111014,
      "loss": 1.4468,
      "step": 275
    },
    {
      "epoch": 0.27607764683817326,
      "grad_norm": 0.6108120083808899,
      "learning_rate": 0.00019584392883153088,
      "loss": 1.3834,
      "step": 276
    },
    {
      "epoch": 0.27707792816729704,
      "grad_norm": 0.680404543876648,
      "learning_rate": 0.00019581397008680717,
      "loss": 1.5094,
      "step": 277
    },
    {
      "epoch": 0.27807820949642087,
      "grad_norm": 0.6419563889503479,
      "learning_rate": 0.00019578390605985826,
      "loss": 1.6933,
      "step": 278
    },
    {
      "epoch": 0.2790784908255447,
      "grad_norm": 0.5788853764533997,
      "learning_rate": 0.00019575373678371909,
      "loss": 1.4754,
      "step": 279
    },
    {
      "epoch": 0.2800787721546685,
      "grad_norm": 0.5943770408630371,
      "learning_rate": 0.00019572346229154025,
      "loss": 1.2949,
      "step": 280
    },
    {
      "epoch": 0.2810790534837923,
      "grad_norm": 0.5997135043144226,
      "learning_rate": 0.00019569308261658787,
      "loss": 1.5365,
      "step": 281
    },
    {
      "epoch": 0.28207933481291614,
      "grad_norm": 0.692401647567749,
      "learning_rate": 0.00019566259779224378,
      "loss": 1.4946,
      "step": 282
    },
    {
      "epoch": 0.28307961614203997,
      "grad_norm": 0.5856708884239197,
      "learning_rate": 0.00019563200785200526,
      "loss": 1.426,
      "step": 283
    },
    {
      "epoch": 0.28407989747116374,
      "grad_norm": 1.2516822814941406,
      "learning_rate": 0.00019560131282948516,
      "loss": 1.5119,
      "step": 284
    },
    {
      "epoch": 0.2850801788002876,
      "grad_norm": 0.6360501050949097,
      "learning_rate": 0.0001955705127584117,
      "loss": 1.3916,
      "step": 285
    },
    {
      "epoch": 0.2860804601294114,
      "grad_norm": 0.6822036504745483,
      "learning_rate": 0.00019553960767262863,
      "loss": 1.5565,
      "step": 286
    },
    {
      "epoch": 0.28708074145853524,
      "grad_norm": 0.6973714828491211,
      "learning_rate": 0.00019550859760609503,
      "loss": 1.5559,
      "step": 287
    },
    {
      "epoch": 0.288081022787659,
      "grad_norm": 0.6595618724822998,
      "learning_rate": 0.00019547748259288536,
      "loss": 1.5824,
      "step": 288
    },
    {
      "epoch": 0.28908130411678284,
      "grad_norm": 0.5625808238983154,
      "learning_rate": 0.0001954462626671894,
      "loss": 1.2669,
      "step": 289
    },
    {
      "epoch": 0.2900815854459067,
      "grad_norm": 0.6318663358688354,
      "learning_rate": 0.0001954149378633122,
      "loss": 1.3896,
      "step": 290
    },
    {
      "epoch": 0.29108186677503045,
      "grad_norm": 0.6655906438827515,
      "learning_rate": 0.00019538350821567404,
      "loss": 1.3889,
      "step": 291
    },
    {
      "epoch": 0.2920821481041543,
      "grad_norm": 0.5947337746620178,
      "learning_rate": 0.00019535197375881045,
      "loss": 1.6112,
      "step": 292
    },
    {
      "epoch": 0.2930824294332781,
      "grad_norm": 0.6139295101165771,
      "learning_rate": 0.00019532033452737205,
      "loss": 1.5185,
      "step": 293
    },
    {
      "epoch": 0.29408271076240194,
      "grad_norm": 0.579953670501709,
      "learning_rate": 0.00019528859055612468,
      "loss": 1.3874,
      "step": 294
    },
    {
      "epoch": 0.2950829920915257,
      "grad_norm": 0.6101506352424622,
      "learning_rate": 0.0001952567418799492,
      "loss": 1.5965,
      "step": 295
    },
    {
      "epoch": 0.29608327342064955,
      "grad_norm": 0.6393965482711792,
      "learning_rate": 0.00019522478853384155,
      "loss": 1.4124,
      "step": 296
    },
    {
      "epoch": 0.2970835547497734,
      "grad_norm": 0.6147856712341309,
      "learning_rate": 0.00019519273055291266,
      "loss": 1.3776,
      "step": 297
    },
    {
      "epoch": 0.2980838360788972,
      "grad_norm": 0.6056416630744934,
      "learning_rate": 0.00019516056797238846,
      "loss": 1.4453,
      "step": 298
    },
    {
      "epoch": 0.299084117408021,
      "grad_norm": 0.6705831289291382,
      "learning_rate": 0.00019512830082760987,
      "loss": 1.3248,
      "step": 299
    },
    {
      "epoch": 0.3000843987371448,
      "grad_norm": 0.6664314866065979,
      "learning_rate": 0.00019509592915403255,
      "loss": 1.5865,
      "step": 300
    },
    {
      "epoch": 0.30108468006626865,
      "grad_norm": 0.5325604677200317,
      "learning_rate": 0.00019506345298722717,
      "loss": 1.0646,
      "step": 301
    },
    {
      "epoch": 0.3020849613953925,
      "grad_norm": 0.589242160320282,
      "learning_rate": 0.00019503087236287913,
      "loss": 1.2297,
      "step": 302
    },
    {
      "epoch": 0.30308524272451626,
      "grad_norm": 0.5677699446678162,
      "learning_rate": 0.00019499818731678873,
      "loss": 1.3961,
      "step": 303
    },
    {
      "epoch": 0.3040855240536401,
      "grad_norm": 0.5676394701004028,
      "learning_rate": 0.00019496539788487082,
      "loss": 1.3276,
      "step": 304
    },
    {
      "epoch": 0.3050858053827639,
      "grad_norm": 0.7280861139297485,
      "learning_rate": 0.0001949325041031551,
      "loss": 1.6731,
      "step": 305
    },
    {
      "epoch": 0.3060860867118877,
      "grad_norm": 0.690636396408081,
      "learning_rate": 0.0001948995060077859,
      "loss": 1.5443,
      "step": 306
    },
    {
      "epoch": 0.3070863680410115,
      "grad_norm": 0.611426055431366,
      "learning_rate": 0.0001948664036350221,
      "loss": 1.5827,
      "step": 307
    },
    {
      "epoch": 0.30808664937013536,
      "grad_norm": 0.7112497091293335,
      "learning_rate": 0.00019483319702123732,
      "loss": 1.5401,
      "step": 308
    },
    {
      "epoch": 0.3090869306992592,
      "grad_norm": 0.6598275303840637,
      "learning_rate": 0.00019479988620291956,
      "loss": 1.6432,
      "step": 309
    },
    {
      "epoch": 0.31008721202838296,
      "grad_norm": 0.5019932985305786,
      "learning_rate": 0.00019476647121667137,
      "loss": 1.2561,
      "step": 310
    },
    {
      "epoch": 0.3110874933575068,
      "grad_norm": 0.7777897715568542,
      "learning_rate": 0.00019473295209920983,
      "loss": 1.6118,
      "step": 311
    },
    {
      "epoch": 0.3120877746866306,
      "grad_norm": 0.6028640866279602,
      "learning_rate": 0.00019469932888736632,
      "loss": 1.4682,
      "step": 312
    },
    {
      "epoch": 0.31308805601575446,
      "grad_norm": 0.554381251335144,
      "learning_rate": 0.00019466560161808674,
      "loss": 1.4179,
      "step": 313
    },
    {
      "epoch": 0.31408833734487823,
      "grad_norm": 0.6212736368179321,
      "learning_rate": 0.00019463177032843124,
      "loss": 1.4327,
      "step": 314
    },
    {
      "epoch": 0.31508861867400206,
      "grad_norm": 0.6829814910888672,
      "learning_rate": 0.00019459783505557424,
      "loss": 1.4455,
      "step": 315
    },
    {
      "epoch": 0.3160889000031259,
      "grad_norm": 0.5808065533638,
      "learning_rate": 0.00019456379583680452,
      "loss": 1.3583,
      "step": 316
    },
    {
      "epoch": 0.31708918133224967,
      "grad_norm": 0.6354159712791443,
      "learning_rate": 0.000194529652709525,
      "loss": 1.6916,
      "step": 317
    },
    {
      "epoch": 0.3180894626613735,
      "grad_norm": 0.6299159526824951,
      "learning_rate": 0.00019449540571125286,
      "loss": 1.47,
      "step": 318
    },
    {
      "epoch": 0.31908974399049733,
      "grad_norm": 0.6222877502441406,
      "learning_rate": 0.00019446105487961926,
      "loss": 1.4137,
      "step": 319
    },
    {
      "epoch": 0.32009002531962116,
      "grad_norm": 0.5995916724205017,
      "learning_rate": 0.0001944266002523696,
      "loss": 1.3679,
      "step": 320
    },
    {
      "epoch": 0.32109030664874494,
      "grad_norm": 0.599814236164093,
      "learning_rate": 0.0001943920418673633,
      "loss": 1.4075,
      "step": 321
    },
    {
      "epoch": 0.32209058797786877,
      "grad_norm": 0.5409269332885742,
      "learning_rate": 0.00019435737976257377,
      "loss": 1.4289,
      "step": 322
    },
    {
      "epoch": 0.3230908693069926,
      "grad_norm": 0.5298951864242554,
      "learning_rate": 0.00019432261397608834,
      "loss": 1.2834,
      "step": 323
    },
    {
      "epoch": 0.32409115063611643,
      "grad_norm": 0.7196112871170044,
      "learning_rate": 0.00019428774454610843,
      "loss": 1.4845,
      "step": 324
    },
    {
      "epoch": 0.3250914319652402,
      "grad_norm": 0.5605450868606567,
      "learning_rate": 0.00019425277151094913,
      "loss": 1.4575,
      "step": 325
    },
    {
      "epoch": 0.32609171329436404,
      "grad_norm": 0.573080837726593,
      "learning_rate": 0.00019421769490903957,
      "loss": 1.5757,
      "step": 326
    },
    {
      "epoch": 0.32709199462348787,
      "grad_norm": 0.5017902851104736,
      "learning_rate": 0.0001941825147789225,
      "loss": 1.5794,
      "step": 327
    },
    {
      "epoch": 0.32809227595261165,
      "grad_norm": 0.643267810344696,
      "learning_rate": 0.00019414723115925456,
      "loss": 1.4903,
      "step": 328
    },
    {
      "epoch": 0.3290925572817355,
      "grad_norm": 0.6522070169448853,
      "learning_rate": 0.0001941118440888061,
      "loss": 1.5907,
      "step": 329
    },
    {
      "epoch": 0.3300928386108593,
      "grad_norm": 0.6496105790138245,
      "learning_rate": 0.0001940763536064611,
      "loss": 1.4225,
      "step": 330
    },
    {
      "epoch": 0.33109311993998314,
      "grad_norm": 0.6011468768119812,
      "learning_rate": 0.00019404075975121716,
      "loss": 1.5022,
      "step": 331
    },
    {
      "epoch": 0.3320934012691069,
      "grad_norm": 0.6327878832817078,
      "learning_rate": 0.0001940050625621855,
      "loss": 1.468,
      "step": 332
    },
    {
      "epoch": 0.33309368259823074,
      "grad_norm": 0.6187490820884705,
      "learning_rate": 0.00019396926207859084,
      "loss": 1.5183,
      "step": 333
    },
    {
      "epoch": 0.3340939639273546,
      "grad_norm": 0.7625093460083008,
      "learning_rate": 0.0001939333583397715,
      "loss": 1.4813,
      "step": 334
    },
    {
      "epoch": 0.3350942452564784,
      "grad_norm": 0.5286359190940857,
      "learning_rate": 0.00019389735138517915,
      "loss": 1.3674,
      "step": 335
    },
    {
      "epoch": 0.3360945265856022,
      "grad_norm": 0.5798503160476685,
      "learning_rate": 0.00019386124125437895,
      "loss": 1.3016,
      "step": 336
    },
    {
      "epoch": 0.337094807914726,
      "grad_norm": 0.48794126510620117,
      "learning_rate": 0.00019382502798704935,
      "loss": 1.3642,
      "step": 337
    },
    {
      "epoch": 0.33809508924384984,
      "grad_norm": 0.7394312620162964,
      "learning_rate": 0.00019378871162298227,
      "loss": 1.327,
      "step": 338
    },
    {
      "epoch": 0.3390953705729737,
      "grad_norm": 0.5598319172859192,
      "learning_rate": 0.00019375229220208276,
      "loss": 1.4247,
      "step": 339
    },
    {
      "epoch": 0.34009565190209745,
      "grad_norm": 0.6099628806114197,
      "learning_rate": 0.00019371576976436917,
      "loss": 1.4906,
      "step": 340
    },
    {
      "epoch": 0.3410959332312213,
      "grad_norm": 0.6749781370162964,
      "learning_rate": 0.00019367914434997312,
      "loss": 1.367,
      "step": 341
    },
    {
      "epoch": 0.3420962145603451,
      "grad_norm": 0.7721238136291504,
      "learning_rate": 0.00019364241599913924,
      "loss": 1.4464,
      "step": 342
    },
    {
      "epoch": 0.3430964958894689,
      "grad_norm": 0.5762369632720947,
      "learning_rate": 0.0001936055847522254,
      "loss": 1.409,
      "step": 343
    },
    {
      "epoch": 0.3440967772185927,
      "grad_norm": 0.6960498690605164,
      "learning_rate": 0.00019356865064970244,
      "loss": 1.3907,
      "step": 344
    },
    {
      "epoch": 0.34509705854771655,
      "grad_norm": 0.5805984735488892,
      "learning_rate": 0.0001935316137321543,
      "loss": 1.4539,
      "step": 345
    },
    {
      "epoch": 0.3460973398768404,
      "grad_norm": 0.5686045289039612,
      "learning_rate": 0.00019349447404027782,
      "loss": 1.4493,
      "step": 346
    },
    {
      "epoch": 0.34709762120596416,
      "grad_norm": 0.5448501706123352,
      "learning_rate": 0.00019345723161488283,
      "loss": 1.5633,
      "step": 347
    },
    {
      "epoch": 0.348097902535088,
      "grad_norm": 0.6388784050941467,
      "learning_rate": 0.000193419886496892,
      "loss": 1.7179,
      "step": 348
    },
    {
      "epoch": 0.3490981838642118,
      "grad_norm": 0.5240457653999329,
      "learning_rate": 0.00019338243872734086,
      "loss": 1.4411,
      "step": 349
    },
    {
      "epoch": 0.35009846519333565,
      "grad_norm": 0.5460641384124756,
      "learning_rate": 0.00019334488834737775,
      "loss": 1.361,
      "step": 350
    },
    {
      "epoch": 0.3510987465224594,
      "grad_norm": 0.5495695471763611,
      "learning_rate": 0.00019330723539826375,
      "loss": 1.5891,
      "step": 351
    },
    {
      "epoch": 0.35209902785158326,
      "grad_norm": 0.5618153214454651,
      "learning_rate": 0.00019326947992137262,
      "loss": 1.3084,
      "step": 352
    },
    {
      "epoch": 0.3530993091807071,
      "grad_norm": 0.5603707432746887,
      "learning_rate": 0.00019323162195819082,
      "loss": 1.5732,
      "step": 353
    },
    {
      "epoch": 0.35409959050983086,
      "grad_norm": 0.5732563138008118,
      "learning_rate": 0.0001931936615503174,
      "loss": 1.5045,
      "step": 354
    },
    {
      "epoch": 0.3550998718389547,
      "grad_norm": 0.5997583866119385,
      "learning_rate": 0.000193155598739464,
      "loss": 1.4175,
      "step": 355
    },
    {
      "epoch": 0.3561001531680785,
      "grad_norm": 0.5769765377044678,
      "learning_rate": 0.0001931174335674547,
      "loss": 1.4834,
      "step": 356
    },
    {
      "epoch": 0.35710043449720236,
      "grad_norm": 0.5902683138847351,
      "learning_rate": 0.0001930791660762262,
      "loss": 1.4664,
      "step": 357
    },
    {
      "epoch": 0.35810071582632613,
      "grad_norm": 0.6354758143424988,
      "learning_rate": 0.00019304079630782752,
      "loss": 1.3891,
      "step": 358
    },
    {
      "epoch": 0.35910099715544996,
      "grad_norm": 0.6018317341804504,
      "learning_rate": 0.0001930023243044201,
      "loss": 1.4514,
      "step": 359
    },
    {
      "epoch": 0.3601012784845738,
      "grad_norm": 0.5409123301506042,
      "learning_rate": 0.00019296375010827773,
      "loss": 1.4708,
      "step": 360
    },
    {
      "epoch": 0.3611015598136976,
      "grad_norm": 0.5457523465156555,
      "learning_rate": 0.00019292507376178643,
      "loss": 1.4988,
      "step": 361
    },
    {
      "epoch": 0.3621018411428214,
      "grad_norm": 0.626768946647644,
      "learning_rate": 0.00019288629530744454,
      "loss": 1.5722,
      "step": 362
    },
    {
      "epoch": 0.36310212247194523,
      "grad_norm": 0.566554069519043,
      "learning_rate": 0.0001928474147878626,
      "loss": 1.2135,
      "step": 363
    },
    {
      "epoch": 0.36410240380106906,
      "grad_norm": 0.7327786684036255,
      "learning_rate": 0.0001928084322457632,
      "loss": 1.5245,
      "step": 364
    },
    {
      "epoch": 0.3651026851301929,
      "grad_norm": 0.5205698609352112,
      "learning_rate": 0.00019276934772398114,
      "loss": 1.2068,
      "step": 365
    },
    {
      "epoch": 0.36610296645931667,
      "grad_norm": 1.0956753492355347,
      "learning_rate": 0.00019273016126546323,
      "loss": 1.5044,
      "step": 366
    },
    {
      "epoch": 0.3671032477884405,
      "grad_norm": 0.6484043598175049,
      "learning_rate": 0.00019269087291326833,
      "loss": 1.6369,
      "step": 367
    },
    {
      "epoch": 0.36810352911756433,
      "grad_norm": 0.6363429427146912,
      "learning_rate": 0.00019265148271056722,
      "loss": 1.4338,
      "step": 368
    },
    {
      "epoch": 0.3691038104466881,
      "grad_norm": 0.6295244693756104,
      "learning_rate": 0.0001926119907006426,
      "loss": 1.4701,
      "step": 369
    },
    {
      "epoch": 0.37010409177581194,
      "grad_norm": 0.6013259887695312,
      "learning_rate": 0.00019257239692688907,
      "loss": 1.7629,
      "step": 370
    },
    {
      "epoch": 0.37110437310493577,
      "grad_norm": 0.6949493885040283,
      "learning_rate": 0.00019253270143281296,
      "loss": 1.6713,
      "step": 371
    },
    {
      "epoch": 0.3721046544340596,
      "grad_norm": 0.6933801174163818,
      "learning_rate": 0.00019249290426203252,
      "loss": 1.6131,
      "step": 372
    },
    {
      "epoch": 0.3731049357631834,
      "grad_norm": 0.5847527384757996,
      "learning_rate": 0.0001924530054582776,
      "loss": 1.3968,
      "step": 373
    },
    {
      "epoch": 0.3741052170923072,
      "grad_norm": 0.6053057312965393,
      "learning_rate": 0.0001924130050653898,
      "loss": 1.3311,
      "step": 374
    },
    {
      "epoch": 0.37510549842143104,
      "grad_norm": 0.5513793230056763,
      "learning_rate": 0.00019237290312732226,
      "loss": 1.5063,
      "step": 375
    },
    {
      "epoch": 0.37610577975055487,
      "grad_norm": 0.5859197378158569,
      "learning_rate": 0.00019233269968813984,
      "loss": 1.3556,
      "step": 376
    },
    {
      "epoch": 0.37710606107967864,
      "grad_norm": 0.5623495578765869,
      "learning_rate": 0.00019229239479201876,
      "loss": 1.3859,
      "step": 377
    },
    {
      "epoch": 0.3781063424088025,
      "grad_norm": 0.602118968963623,
      "learning_rate": 0.0001922519884832469,
      "loss": 1.334,
      "step": 378
    },
    {
      "epoch": 0.3791066237379263,
      "grad_norm": 0.5212380886077881,
      "learning_rate": 0.0001922114808062234,
      "loss": 1.401,
      "step": 379
    },
    {
      "epoch": 0.3801069050670501,
      "grad_norm": 0.4969455599784851,
      "learning_rate": 0.00019217087180545893,
      "loss": 1.2292,
      "step": 380
    },
    {
      "epoch": 0.3811071863961739,
      "grad_norm": 0.578629732131958,
      "learning_rate": 0.0001921301615255754,
      "loss": 1.5015,
      "step": 381
    },
    {
      "epoch": 0.38210746772529774,
      "grad_norm": 0.593053936958313,
      "learning_rate": 0.0001920893500113061,
      "loss": 1.302,
      "step": 382
    },
    {
      "epoch": 0.3831077490544216,
      "grad_norm": 0.5832563638687134,
      "learning_rate": 0.00019204843730749547,
      "loss": 1.3695,
      "step": 383
    },
    {
      "epoch": 0.38410803038354535,
      "grad_norm": 0.5608510375022888,
      "learning_rate": 0.00019200742345909915,
      "loss": 1.3792,
      "step": 384
    },
    {
      "epoch": 0.3851083117126692,
      "grad_norm": 0.5337334275245667,
      "learning_rate": 0.00019196630851118398,
      "loss": 1.4163,
      "step": 385
    },
    {
      "epoch": 0.386108593041793,
      "grad_norm": 0.5460125803947449,
      "learning_rate": 0.0001919250925089278,
      "loss": 1.2439,
      "step": 386
    },
    {
      "epoch": 0.38710887437091684,
      "grad_norm": 0.6217851638793945,
      "learning_rate": 0.00019188377549761963,
      "loss": 1.6428,
      "step": 387
    },
    {
      "epoch": 0.3881091557000406,
      "grad_norm": 0.7154502868652344,
      "learning_rate": 0.00019184235752265928,
      "loss": 1.3468,
      "step": 388
    },
    {
      "epoch": 0.38910943702916445,
      "grad_norm": 0.5044635534286499,
      "learning_rate": 0.00019180083862955772,
      "loss": 1.1877,
      "step": 389
    },
    {
      "epoch": 0.3901097183582883,
      "grad_norm": 0.5755971074104309,
      "learning_rate": 0.00019175921886393666,
      "loss": 1.3475,
      "step": 390
    },
    {
      "epoch": 0.39110999968741206,
      "grad_norm": 0.6121137738227844,
      "learning_rate": 0.00019171749827152869,
      "loss": 1.4342,
      "step": 391
    },
    {
      "epoch": 0.3921102810165359,
      "grad_norm": 0.5615536570549011,
      "learning_rate": 0.0001916756768981772,
      "loss": 1.5471,
      "step": 392
    },
    {
      "epoch": 0.3931105623456597,
      "grad_norm": 0.6527026295661926,
      "learning_rate": 0.00019163375478983632,
      "loss": 1.6363,
      "step": 393
    },
    {
      "epoch": 0.39411084367478355,
      "grad_norm": 0.6465044617652893,
      "learning_rate": 0.00019159173199257085,
      "loss": 1.3823,
      "step": 394
    },
    {
      "epoch": 0.3951111250039073,
      "grad_norm": 0.5620000958442688,
      "learning_rate": 0.00019154960855255628,
      "loss": 1.5418,
      "step": 395
    },
    {
      "epoch": 0.39611140633303116,
      "grad_norm": 0.7090588808059692,
      "learning_rate": 0.0001915073845160786,
      "loss": 1.4593,
      "step": 396
    },
    {
      "epoch": 0.397111687662155,
      "grad_norm": 0.6644489169120789,
      "learning_rate": 0.00019146505992953446,
      "loss": 1.4236,
      "step": 397
    },
    {
      "epoch": 0.3981119689912788,
      "grad_norm": 0.6038135886192322,
      "learning_rate": 0.00019142263483943085,
      "loss": 1.1805,
      "step": 398
    },
    {
      "epoch": 0.3991122503204026,
      "grad_norm": 0.6746726036071777,
      "learning_rate": 0.00019138010929238534,
      "loss": 1.5264,
      "step": 399
    },
    {
      "epoch": 0.4001125316495264,
      "grad_norm": 0.5871374607086182,
      "learning_rate": 0.00019133748333512575,
      "loss": 1.3709,
      "step": 400
    },
    {
      "epoch": 0.40111281297865026,
      "grad_norm": 0.5743412375450134,
      "learning_rate": 0.00019129475701449035,
      "loss": 1.4677,
      "step": 401
    },
    {
      "epoch": 0.4021130943077741,
      "grad_norm": 0.6184396743774414,
      "learning_rate": 0.0001912519303774276,
      "loss": 1.4228,
      "step": 402
    },
    {
      "epoch": 0.40311337563689786,
      "grad_norm": 0.5872434973716736,
      "learning_rate": 0.0001912090034709963,
      "loss": 1.3495,
      "step": 403
    },
    {
      "epoch": 0.4041136569660217,
      "grad_norm": 0.6500155925750732,
      "learning_rate": 0.00019116597634236525,
      "loss": 1.4315,
      "step": 404
    },
    {
      "epoch": 0.4051139382951455,
      "grad_norm": 0.5240740180015564,
      "learning_rate": 0.0001911228490388136,
      "loss": 1.4954,
      "step": 405
    },
    {
      "epoch": 0.4061142196242693,
      "grad_norm": 0.5531806945800781,
      "learning_rate": 0.00019107962160773035,
      "loss": 1.3949,
      "step": 406
    },
    {
      "epoch": 0.40711450095339313,
      "grad_norm": 0.5266262888908386,
      "learning_rate": 0.0001910362940966147,
      "loss": 1.2859,
      "step": 407
    },
    {
      "epoch": 0.40811478228251696,
      "grad_norm": 0.5734869241714478,
      "learning_rate": 0.00019099286655307568,
      "loss": 1.2451,
      "step": 408
    },
    {
      "epoch": 0.4091150636116408,
      "grad_norm": 0.5922874212265015,
      "learning_rate": 0.0001909493390248324,
      "loss": 1.5429,
      "step": 409
    },
    {
      "epoch": 0.41011534494076457,
      "grad_norm": 0.542540431022644,
      "learning_rate": 0.00019090571155971366,
      "loss": 1.4138,
      "step": 410
    },
    {
      "epoch": 0.4111156262698884,
      "grad_norm": 0.57356196641922,
      "learning_rate": 0.00019086198420565823,
      "loss": 1.2592,
      "step": 411
    },
    {
      "epoch": 0.41211590759901223,
      "grad_norm": 0.6042733192443848,
      "learning_rate": 0.00019081815701071445,
      "loss": 1.5524,
      "step": 412
    },
    {
      "epoch": 0.41311618892813606,
      "grad_norm": 0.46550241112709045,
      "learning_rate": 0.0001907742300230406,
      "loss": 1.308,
      "step": 413
    },
    {
      "epoch": 0.41411647025725984,
      "grad_norm": 0.6283137798309326,
      "learning_rate": 0.00019073020329090444,
      "loss": 1.4753,
      "step": 414
    },
    {
      "epoch": 0.41511675158638367,
      "grad_norm": 0.5254876613616943,
      "learning_rate": 0.0001906860768626834,
      "loss": 1.2157,
      "step": 415
    },
    {
      "epoch": 0.4161170329155075,
      "grad_norm": 0.59089195728302,
      "learning_rate": 0.00019064185078686443,
      "loss": 1.2684,
      "step": 416
    },
    {
      "epoch": 0.4171173142446313,
      "grad_norm": 0.7129126787185669,
      "learning_rate": 0.000190597525112044,
      "loss": 1.3974,
      "step": 417
    },
    {
      "epoch": 0.4181175955737551,
      "grad_norm": 0.607305109500885,
      "learning_rate": 0.000190553099886928,
      "loss": 1.4312,
      "step": 418
    },
    {
      "epoch": 0.41911787690287894,
      "grad_norm": 0.49921515583992004,
      "learning_rate": 0.00019050857516033173,
      "loss": 1.3469,
      "step": 419
    },
    {
      "epoch": 0.42011815823200277,
      "grad_norm": 0.6167325377464294,
      "learning_rate": 0.00019046395098117983,
      "loss": 1.4723,
      "step": 420
    },
    {
      "epoch": 0.42111843956112655,
      "grad_norm": 0.6144593358039856,
      "learning_rate": 0.00019041922739850616,
      "loss": 1.5502,
      "step": 421
    },
    {
      "epoch": 0.4221187208902504,
      "grad_norm": 0.61333167552948,
      "learning_rate": 0.00019037440446145385,
      "loss": 1.3283,
      "step": 422
    },
    {
      "epoch": 0.4231190022193742,
      "grad_norm": 0.5881702303886414,
      "learning_rate": 0.00019032948221927524,
      "loss": 1.4206,
      "step": 423
    },
    {
      "epoch": 0.42411928354849804,
      "grad_norm": 0.5334322452545166,
      "learning_rate": 0.00019028446072133175,
      "loss": 1.4603,
      "step": 424
    },
    {
      "epoch": 0.4251195648776218,
      "grad_norm": 0.5730605721473694,
      "learning_rate": 0.00019023934001709383,
      "loss": 1.4375,
      "step": 425
    },
    {
      "epoch": 0.42611984620674564,
      "grad_norm": 0.6227820515632629,
      "learning_rate": 0.00019019412015614098,
      "loss": 1.4888,
      "step": 426
    },
    {
      "epoch": 0.4271201275358695,
      "grad_norm": 0.5811313390731812,
      "learning_rate": 0.00019014880118816164,
      "loss": 1.3492,
      "step": 427
    },
    {
      "epoch": 0.4281204088649933,
      "grad_norm": 0.5685800313949585,
      "learning_rate": 0.0001901033831629532,
      "loss": 1.5052,
      "step": 428
    },
    {
      "epoch": 0.4291206901941171,
      "grad_norm": 0.5961394309997559,
      "learning_rate": 0.00019005786613042185,
      "loss": 1.3324,
      "step": 429
    },
    {
      "epoch": 0.4301209715232409,
      "grad_norm": 0.5845314860343933,
      "learning_rate": 0.00019001225014058255,
      "loss": 1.5733,
      "step": 430
    },
    {
      "epoch": 0.43112125285236474,
      "grad_norm": 0.5400176048278809,
      "learning_rate": 0.00018996653524355902,
      "loss": 1.3973,
      "step": 431
    },
    {
      "epoch": 0.4321215341814885,
      "grad_norm": 0.5462201833724976,
      "learning_rate": 0.00018992072148958368,
      "loss": 1.2167,
      "step": 432
    },
    {
      "epoch": 0.43312181551061235,
      "grad_norm": 0.6200360059738159,
      "learning_rate": 0.00018987480892899758,
      "loss": 1.5596,
      "step": 433
    },
    {
      "epoch": 0.4341220968397362,
      "grad_norm": 0.5230718851089478,
      "learning_rate": 0.00018982879761225027,
      "loss": 1.3661,
      "step": 434
    },
    {
      "epoch": 0.43512237816886,
      "grad_norm": 0.5868643522262573,
      "learning_rate": 0.00018978268758989991,
      "loss": 1.4792,
      "step": 435
    },
    {
      "epoch": 0.4361226594979838,
      "grad_norm": 0.580892026424408,
      "learning_rate": 0.00018973647891261307,
      "loss": 1.3275,
      "step": 436
    },
    {
      "epoch": 0.4371229408271076,
      "grad_norm": 0.5903263688087463,
      "learning_rate": 0.00018969017163116472,
      "loss": 1.4721,
      "step": 437
    },
    {
      "epoch": 0.43812322215623145,
      "grad_norm": 0.5108968019485474,
      "learning_rate": 0.0001896437657964382,
      "loss": 1.3785,
      "step": 438
    },
    {
      "epoch": 0.4391235034853553,
      "grad_norm": 0.6707500219345093,
      "learning_rate": 0.00018959726145942508,
      "loss": 1.5033,
      "step": 439
    },
    {
      "epoch": 0.44012378481447906,
      "grad_norm": 0.5793184638023376,
      "learning_rate": 0.00018955065867122528,
      "loss": 1.3629,
      "step": 440
    },
    {
      "epoch": 0.4411240661436029,
      "grad_norm": 0.5549041628837585,
      "learning_rate": 0.00018950395748304678,
      "loss": 1.5557,
      "step": 441
    },
    {
      "epoch": 0.4421243474727267,
      "grad_norm": 0.5406919121742249,
      "learning_rate": 0.0001894571579462058,
      "loss": 1.4441,
      "step": 442
    },
    {
      "epoch": 0.4431246288018505,
      "grad_norm": 0.5131089091300964,
      "learning_rate": 0.00018941026011212654,
      "loss": 1.3051,
      "step": 443
    },
    {
      "epoch": 0.4441249101309743,
      "grad_norm": 0.601586639881134,
      "learning_rate": 0.00018936326403234125,
      "loss": 1.5297,
      "step": 444
    },
    {
      "epoch": 0.44512519146009816,
      "grad_norm": 0.5036457180976868,
      "learning_rate": 0.00018931616975849006,
      "loss": 1.357,
      "step": 445
    },
    {
      "epoch": 0.446125472789222,
      "grad_norm": 0.5471266508102417,
      "learning_rate": 0.00018926897734232115,
      "loss": 1.2176,
      "step": 446
    },
    {
      "epoch": 0.44712575411834576,
      "grad_norm": 0.6057867407798767,
      "learning_rate": 0.0001892216868356904,
      "loss": 1.4763,
      "step": 447
    },
    {
      "epoch": 0.4481260354474696,
      "grad_norm": 0.5384593605995178,
      "learning_rate": 0.0001891742982905615,
      "loss": 1.513,
      "step": 448
    },
    {
      "epoch": 0.4491263167765934,
      "grad_norm": 0.6144880056381226,
      "learning_rate": 0.00018912681175900598,
      "loss": 1.5782,
      "step": 449
    },
    {
      "epoch": 0.45012659810571726,
      "grad_norm": 0.4838174879550934,
      "learning_rate": 0.00018907922729320285,
      "loss": 1.4085,
      "step": 450
    },
    {
      "epoch": 0.45112687943484103,
      "grad_norm": 0.6852928400039673,
      "learning_rate": 0.00018903154494543889,
      "loss": 1.5989,
      "step": 451
    },
    {
      "epoch": 0.45212716076396486,
      "grad_norm": 0.47527411580085754,
      "learning_rate": 0.00018898376476810834,
      "loss": 1.3409,
      "step": 452
    },
    {
      "epoch": 0.4531274420930887,
      "grad_norm": 0.5665884613990784,
      "learning_rate": 0.00018893588681371303,
      "loss": 1.5395,
      "step": 453
    },
    {
      "epoch": 0.45412772342221247,
      "grad_norm": 0.5792158246040344,
      "learning_rate": 0.00018888791113486213,
      "loss": 1.516,
      "step": 454
    },
    {
      "epoch": 0.4551280047513363,
      "grad_norm": 0.5223523378372192,
      "learning_rate": 0.00018883983778427227,
      "loss": 1.3678,
      "step": 455
    },
    {
      "epoch": 0.45612828608046013,
      "grad_norm": 0.5927590131759644,
      "learning_rate": 0.0001887916668147673,
      "loss": 1.3617,
      "step": 456
    },
    {
      "epoch": 0.45712856740958396,
      "grad_norm": 0.7266496419906616,
      "learning_rate": 0.00018874339827927846,
      "loss": 1.3734,
      "step": 457
    },
    {
      "epoch": 0.45812884873870774,
      "grad_norm": 0.6495805978775024,
      "learning_rate": 0.00018869503223084414,
      "loss": 1.5282,
      "step": 458
    },
    {
      "epoch": 0.45912913006783157,
      "grad_norm": 0.6099816560745239,
      "learning_rate": 0.00018864656872260985,
      "loss": 1.4691,
      "step": 459
    },
    {
      "epoch": 0.4601294113969554,
      "grad_norm": 0.5208227038383484,
      "learning_rate": 0.00018859800780782828,
      "loss": 1.3949,
      "step": 460
    },
    {
      "epoch": 0.46112969272607923,
      "grad_norm": 0.5526600480079651,
      "learning_rate": 0.000188549349539859,
      "loss": 1.3557,
      "step": 461
    },
    {
      "epoch": 0.462129974055203,
      "grad_norm": 0.5537740588188171,
      "learning_rate": 0.00018850059397216876,
      "loss": 1.4703,
      "step": 462
    },
    {
      "epoch": 0.46313025538432684,
      "grad_norm": 0.5553976893424988,
      "learning_rate": 0.00018845174115833099,
      "loss": 1.4356,
      "step": 463
    },
    {
      "epoch": 0.46413053671345067,
      "grad_norm": 0.6027779579162598,
      "learning_rate": 0.0001884027911520262,
      "loss": 1.4763,
      "step": 464
    },
    {
      "epoch": 0.4651308180425745,
      "grad_norm": 0.5559154748916626,
      "learning_rate": 0.00018835374400704154,
      "loss": 1.4148,
      "step": 465
    },
    {
      "epoch": 0.4661310993716983,
      "grad_norm": 0.6124109029769897,
      "learning_rate": 0.00018830459977727096,
      "loss": 1.4468,
      "step": 466
    },
    {
      "epoch": 0.4671313807008221,
      "grad_norm": 0.4762580692768097,
      "learning_rate": 0.0001882553585167151,
      "loss": 1.3714,
      "step": 467
    },
    {
      "epoch": 0.46813166202994594,
      "grad_norm": 0.5793487429618835,
      "learning_rate": 0.00018820602027948114,
      "loss": 1.4828,
      "step": 468
    },
    {
      "epoch": 0.4691319433590697,
      "grad_norm": 0.55177241563797,
      "learning_rate": 0.00018815658511978298,
      "loss": 1.4157,
      "step": 469
    },
    {
      "epoch": 0.47013222468819355,
      "grad_norm": 0.5065292716026306,
      "learning_rate": 0.00018810705309194083,
      "loss": 1.4519,
      "step": 470
    },
    {
      "epoch": 0.4711325060173174,
      "grad_norm": 0.5401413440704346,
      "learning_rate": 0.00018805742425038145,
      "loss": 1.4344,
      "step": 471
    },
    {
      "epoch": 0.4721327873464412,
      "grad_norm": 0.7173880338668823,
      "learning_rate": 0.00018800769864963802,
      "loss": 1.7325,
      "step": 472
    },
    {
      "epoch": 0.473133068675565,
      "grad_norm": 0.507682204246521,
      "learning_rate": 0.00018795787634434994,
      "loss": 1.37,
      "step": 473
    },
    {
      "epoch": 0.4741333500046888,
      "grad_norm": 0.551888644695282,
      "learning_rate": 0.0001879079573892629,
      "loss": 1.3695,
      "step": 474
    },
    {
      "epoch": 0.47513363133381264,
      "grad_norm": 0.5109260082244873,
      "learning_rate": 0.00018785794183922883,
      "loss": 1.4001,
      "step": 475
    },
    {
      "epoch": 0.4761339126629365,
      "grad_norm": 0.4565551280975342,
      "learning_rate": 0.00018780782974920572,
      "loss": 1.1752,
      "step": 476
    },
    {
      "epoch": 0.47713419399206025,
      "grad_norm": 0.5651509761810303,
      "learning_rate": 0.00018775762117425777,
      "loss": 1.4291,
      "step": 477
    },
    {
      "epoch": 0.4781344753211841,
      "grad_norm": 0.5827792286872864,
      "learning_rate": 0.0001877073161695551,
      "loss": 1.3438,
      "step": 478
    },
    {
      "epoch": 0.4791347566503079,
      "grad_norm": 0.5719752907752991,
      "learning_rate": 0.00018765691479037376,
      "loss": 1.4683,
      "step": 479
    },
    {
      "epoch": 0.4801350379794317,
      "grad_norm": 0.5153111815452576,
      "learning_rate": 0.00018760641709209583,
      "loss": 1.4392,
      "step": 480
    },
    {
      "epoch": 0.4811353193085555,
      "grad_norm": 0.5455904603004456,
      "learning_rate": 0.0001875558231302091,
      "loss": 1.1603,
      "step": 481
    },
    {
      "epoch": 0.48213560063767935,
      "grad_norm": 0.5857074856758118,
      "learning_rate": 0.00018750513296030718,
      "loss": 1.3099,
      "step": 482
    },
    {
      "epoch": 0.4831358819668032,
      "grad_norm": 0.6051676273345947,
      "learning_rate": 0.00018745434663808942,
      "loss": 1.3587,
      "step": 483
    },
    {
      "epoch": 0.48413616329592696,
      "grad_norm": 0.588749885559082,
      "learning_rate": 0.0001874034642193608,
      "loss": 1.5277,
      "step": 484
    },
    {
      "epoch": 0.4851364446250508,
      "grad_norm": 0.5295410752296448,
      "learning_rate": 0.0001873524857600319,
      "loss": 1.2084,
      "step": 485
    },
    {
      "epoch": 0.4861367259541746,
      "grad_norm": 0.5313368439674377,
      "learning_rate": 0.00018730141131611882,
      "loss": 1.4002,
      "step": 486
    },
    {
      "epoch": 0.48713700728329845,
      "grad_norm": 0.5166353583335876,
      "learning_rate": 0.00018725024094374315,
      "loss": 1.208,
      "step": 487
    },
    {
      "epoch": 0.4881372886124222,
      "grad_norm": 0.5478363037109375,
      "learning_rate": 0.00018719897469913184,
      "loss": 1.3236,
      "step": 488
    },
    {
      "epoch": 0.48913756994154606,
      "grad_norm": 0.5531913042068481,
      "learning_rate": 0.00018714761263861728,
      "loss": 1.4938,
      "step": 489
    },
    {
      "epoch": 0.4901378512706699,
      "grad_norm": 0.5334530472755432,
      "learning_rate": 0.000187096154818637,
      "loss": 1.4172,
      "step": 490
    },
    {
      "epoch": 0.4911381325997937,
      "grad_norm": 0.5667001605033875,
      "learning_rate": 0.00018704460129573391,
      "loss": 1.3517,
      "step": 491
    },
    {
      "epoch": 0.4921384139289175,
      "grad_norm": 0.5568780303001404,
      "learning_rate": 0.00018699295212655596,
      "loss": 1.4287,
      "step": 492
    },
    {
      "epoch": 0.4931386952580413,
      "grad_norm": 0.6663610935211182,
      "learning_rate": 0.00018694120736785632,
      "loss": 1.5416,
      "step": 493
    },
    {
      "epoch": 0.49413897658716516,
      "grad_norm": 0.5753045082092285,
      "learning_rate": 0.00018688936707649304,
      "loss": 1.5552,
      "step": 494
    },
    {
      "epoch": 0.49513925791628893,
      "grad_norm": 0.5707410573959351,
      "learning_rate": 0.00018683743130942928,
      "loss": 1.5332,
      "step": 495
    },
    {
      "epoch": 0.49613953924541276,
      "grad_norm": 0.5847951173782349,
      "learning_rate": 0.00018678540012373302,
      "loss": 1.3488,
      "step": 496
    },
    {
      "epoch": 0.4971398205745366,
      "grad_norm": 0.60503751039505,
      "learning_rate": 0.00018673327357657715,
      "loss": 1.3924,
      "step": 497
    },
    {
      "epoch": 0.4981401019036604,
      "grad_norm": 0.635142982006073,
      "learning_rate": 0.0001866810517252393,
      "loss": 1.4392,
      "step": 498
    },
    {
      "epoch": 0.4991403832327842,
      "grad_norm": 0.5536782741546631,
      "learning_rate": 0.00018662873462710184,
      "loss": 1.286,
      "step": 499
    },
    {
      "epoch": 0.5001406645619081,
      "grad_norm": 0.5676659345626831,
      "learning_rate": 0.0001865763223396518,
      "loss": 1.3006,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 2997,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.234921192731443e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}