{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9991726094322524,
  "eval_steps": 200,
  "global_step": 3172,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006303928135219258,
      "grad_norm": 0.23900407552719116,
      "learning_rate": 1.40625e-06,
      "loss": 0.6566,
      "step": 1
    },
    {
      "epoch": 0.0012607856270438517,
      "grad_norm": 0.26570916175842285,
      "learning_rate": 2.8125e-06,
      "loss": 0.691,
      "step": 2
    },
    {
      "epoch": 0.0018911784405657775,
      "grad_norm": 0.2456347942352295,
      "learning_rate": 4.21875e-06,
      "loss": 0.7193,
      "step": 3
    },
    {
      "epoch": 0.0025215712540877034,
      "grad_norm": 0.2497250735759735,
      "learning_rate": 5.625e-06,
      "loss": 0.63,
      "step": 4
    },
    {
      "epoch": 0.0031519640676096294,
      "grad_norm": 0.2710248827934265,
      "learning_rate": 7.031250000000001e-06,
      "loss": 0.6894,
      "step": 5
    },
    {
      "epoch": 0.003782356881131555,
      "grad_norm": 0.24278788268566132,
      "learning_rate": 8.4375e-06,
      "loss": 0.6779,
      "step": 6
    },
    {
      "epoch": 0.004412749694653481,
      "grad_norm": 0.27044716477394104,
      "learning_rate": 9.84375e-06,
      "loss": 0.7366,
      "step": 7
    },
    {
      "epoch": 0.005043142508175407,
      "grad_norm": 0.22668607532978058,
      "learning_rate": 1.125e-05,
      "loss": 0.6066,
      "step": 8
    },
    {
      "epoch": 0.005673535321697332,
      "grad_norm": 0.2294527143239975,
      "learning_rate": 1.2656250000000001e-05,
      "loss": 0.7224,
      "step": 9
    },
    {
      "epoch": 0.006303928135219259,
      "grad_norm": 0.22373972833156586,
      "learning_rate": 1.4062500000000001e-05,
      "loss": 0.6607,
      "step": 10
    },
    {
      "epoch": 0.0069343209487411845,
      "grad_norm": 0.22508320212364197,
      "learning_rate": 1.546875e-05,
      "loss": 0.7065,
      "step": 11
    },
    {
      "epoch": 0.00756471376226311,
      "grad_norm": 0.22706764936447144,
      "learning_rate": 1.6875e-05,
      "loss": 0.648,
      "step": 12
    },
    {
      "epoch": 0.008195106575785037,
      "grad_norm": 0.21693125367164612,
      "learning_rate": 1.828125e-05,
      "loss": 0.6044,
      "step": 13
    },
    {
      "epoch": 0.008825499389306962,
      "grad_norm": 0.1876903772354126,
      "learning_rate": 1.96875e-05,
      "loss": 0.5912,
      "step": 14
    },
    {
      "epoch": 0.009455892202828888,
      "grad_norm": 0.18589751422405243,
      "learning_rate": 2.109375e-05,
      "loss": 0.6084,
      "step": 15
    },
    {
      "epoch": 0.010086285016350813,
      "grad_norm": 0.21171413362026215,
      "learning_rate": 2.25e-05,
      "loss": 0.5924,
      "step": 16
    },
    {
      "epoch": 0.010716677829872739,
      "grad_norm": 0.20840738713741302,
      "learning_rate": 2.3906250000000002e-05,
      "loss": 0.5527,
      "step": 17
    },
    {
      "epoch": 0.011347070643394665,
      "grad_norm": 0.20482507348060608,
      "learning_rate": 2.5312500000000002e-05,
      "loss": 0.6516,
      "step": 18
    },
    {
      "epoch": 0.01197746345691659,
      "grad_norm": 0.25472748279571533,
      "learning_rate": 2.6718750000000002e-05,
      "loss": 0.6341,
      "step": 19
    },
    {
      "epoch": 0.012607856270438518,
      "grad_norm": 0.23628972470760345,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.5931,
      "step": 20
    },
    {
      "epoch": 0.013238249083960443,
      "grad_norm": 0.19538824260234833,
      "learning_rate": 2.9531250000000003e-05,
      "loss": 0.5447,
      "step": 21
    },
    {
      "epoch": 0.013868641897482369,
      "grad_norm": 0.17556734383106232,
      "learning_rate": 3.09375e-05,
      "loss": 0.5111,
      "step": 22
    },
    {
      "epoch": 0.014499034711004295,
      "grad_norm": 0.19505266845226288,
      "learning_rate": 3.2343750000000004e-05,
      "loss": 0.5879,
      "step": 23
    },
    {
      "epoch": 0.01512942752452622,
      "grad_norm": 0.1797773689031601,
      "learning_rate": 3.375e-05,
      "loss": 0.5436,
      "step": 24
    },
    {
      "epoch": 0.015759820338048146,
      "grad_norm": 0.2432139664888382,
      "learning_rate": 3.5156250000000004e-05,
      "loss": 0.669,
      "step": 25
    },
    {
      "epoch": 0.016390213151570073,
      "grad_norm": 0.19737608730793,
      "learning_rate": 3.65625e-05,
      "loss": 0.5622,
      "step": 26
    },
    {
      "epoch": 0.017020605965091997,
      "grad_norm": 0.17237292230129242,
      "learning_rate": 3.7968750000000005e-05,
      "loss": 0.6074,
      "step": 27
    },
    {
      "epoch": 0.017650998778613924,
      "grad_norm": 0.17980431020259857,
      "learning_rate": 3.9375e-05,
      "loss": 0.4692,
      "step": 28
    },
    {
      "epoch": 0.01828139159213585,
      "grad_norm": 0.18721145391464233,
      "learning_rate": 4.0781250000000005e-05,
      "loss": 0.6626,
      "step": 29
    },
    {
      "epoch": 0.018911784405657776,
      "grad_norm": 0.18952064216136932,
      "learning_rate": 4.21875e-05,
      "loss": 0.5898,
      "step": 30
    },
    {
      "epoch": 0.019542177219179703,
      "grad_norm": 0.16960866749286652,
      "learning_rate": 4.3593750000000006e-05,
      "loss": 0.488,
      "step": 31
    },
    {
      "epoch": 0.020172570032701627,
      "grad_norm": 0.15774346888065338,
      "learning_rate": 4.5e-05,
      "loss": 0.538,
      "step": 32
    },
    {
      "epoch": 0.020802962846223554,
      "grad_norm": 0.1621467024087906,
      "learning_rate": 4.499998873858572e-05,
      "loss": 0.4839,
      "step": 33
    },
    {
      "epoch": 0.021433355659745478,
      "grad_norm": 0.17073126137256622,
      "learning_rate": 4.4999954954354146e-05,
      "loss": 0.5118,
      "step": 34
    },
    {
      "epoch": 0.022063748473267406,
      "grad_norm": 0.1898810863494873,
      "learning_rate": 4.4999898647339106e-05,
      "loss": 0.564,
      "step": 35
    },
    {
      "epoch": 0.02269414128678933,
      "grad_norm": 0.1633978635072708,
      "learning_rate": 4.4999819817596954e-05,
      "loss": 0.5113,
      "step": 36
    },
    {
      "epoch": 0.023324534100311257,
      "grad_norm": 0.17059823870658875,
      "learning_rate": 4.4999718465206607e-05,
      "loss": 0.5268,
      "step": 37
    },
    {
      "epoch": 0.02395492691383318,
      "grad_norm": 0.1455584019422531,
      "learning_rate": 4.499959459026952e-05,
      "loss": 0.4203,
      "step": 38
    },
    {
      "epoch": 0.024585319727355108,
      "grad_norm": 0.14489693939685822,
      "learning_rate": 4.499944819290968e-05,
      "loss": 0.4815,
      "step": 39
    },
    {
      "epoch": 0.025215712540877035,
      "grad_norm": 0.16840510070323944,
      "learning_rate": 4.4999279273273656e-05,
      "loss": 0.5266,
      "step": 40
    },
    {
      "epoch": 0.02584610535439896,
      "grad_norm": 0.13356979191303253,
      "learning_rate": 4.4999087831530515e-05,
      "loss": 0.5271,
      "step": 41
    },
    {
      "epoch": 0.026476498167920887,
      "grad_norm": 0.14093619585037231,
      "learning_rate": 4.4998873867871916e-05,
      "loss": 0.4549,
      "step": 42
    },
    {
      "epoch": 0.02710689098144281,
      "grad_norm": 0.16740302741527557,
      "learning_rate": 4.4998637382512016e-05,
      "loss": 0.5066,
      "step": 43
    },
    {
      "epoch": 0.027737283794964738,
      "grad_norm": 0.15250779688358307,
      "learning_rate": 4.4998378375687565e-05,
      "loss": 0.4683,
      "step": 44
    },
    {
      "epoch": 0.028367676608486662,
      "grad_norm": 0.1856726109981537,
      "learning_rate": 4.4998096847657815e-05,
      "loss": 0.5442,
      "step": 45
    },
    {
      "epoch": 0.02899806942200859,
      "grad_norm": 0.18478403985500336,
      "learning_rate": 4.499779279870459e-05,
      "loss": 0.6015,
      "step": 46
    },
    {
      "epoch": 0.029628462235530516,
      "grad_norm": 0.211555615067482,
      "learning_rate": 4.499746622913224e-05,
      "loss": 0.4835,
      "step": 47
    },
    {
      "epoch": 0.03025885504905244,
      "grad_norm": 0.16362585127353668,
      "learning_rate": 4.499711713926767e-05,
      "loss": 0.5241,
      "step": 48
    },
    {
      "epoch": 0.030889247862574368,
      "grad_norm": 0.16383248567581177,
      "learning_rate": 4.499674552946032e-05,
      "loss": 0.4803,
      "step": 49
    },
    {
      "epoch": 0.03151964067609629,
      "grad_norm": 0.17049849033355713,
      "learning_rate": 4.4996351400082177e-05,
      "loss": 0.5351,
      "step": 50
    },
    {
      "epoch": 0.032150033489618215,
      "grad_norm": 0.1521013379096985,
      "learning_rate": 4.4995934751527774e-05,
      "loss": 0.4387,
      "step": 51
    },
    {
      "epoch": 0.032780426303140146,
      "grad_norm": 0.1371651589870453,
      "learning_rate": 4.499549558421419e-05,
      "loss": 0.4201,
      "step": 52
    },
    {
      "epoch": 0.03341081911666207,
      "grad_norm": 0.18806736171245575,
      "learning_rate": 4.4995033898581017e-05,
      "loss": 0.4741,
      "step": 53
    },
    {
      "epoch": 0.034041211930183994,
      "grad_norm": 0.17703312635421753,
      "learning_rate": 4.499454969509043e-05,
      "loss": 0.5133,
      "step": 54
    },
    {
      "epoch": 0.034671604743705925,
      "grad_norm": 0.18577425181865692,
      "learning_rate": 4.499404297422711e-05,
      "loss": 0.4545,
      "step": 55
    },
    {
      "epoch": 0.03530199755722785,
      "grad_norm": 0.1458091139793396,
      "learning_rate": 4.4993513736498296e-05,
      "loss": 0.4347,
      "step": 56
    },
    {
      "epoch": 0.03593239037074977,
      "grad_norm": 0.1535828560590744,
      "learning_rate": 4.499296198243376e-05,
      "loss": 0.447,
      "step": 57
    },
    {
      "epoch": 0.0365627831842717,
      "grad_norm": 0.16728930175304413,
      "learning_rate": 4.4992387712585824e-05,
      "loss": 0.5249,
      "step": 58
    },
    {
      "epoch": 0.03719317599779363,
      "grad_norm": 0.17529086768627167,
      "learning_rate": 4.499179092752934e-05,
      "loss": 0.5676,
      "step": 59
    },
    {
      "epoch": 0.03782356881131555,
      "grad_norm": 0.155259370803833,
      "learning_rate": 4.4991171627861686e-05,
      "loss": 0.4816,
      "step": 60
    },
    {
      "epoch": 0.038453961624837475,
      "grad_norm": 0.17953960597515106,
      "learning_rate": 4.4990529814202806e-05,
      "loss": 0.4924,
      "step": 61
    },
    {
      "epoch": 0.039084354438359406,
      "grad_norm": 0.19025997817516327,
      "learning_rate": 4.4989865487195156e-05,
      "loss": 0.4962,
      "step": 62
    },
    {
      "epoch": 0.03971474725188133,
      "grad_norm": 0.22502540051937103,
      "learning_rate": 4.498917864750374e-05,
      "loss": 0.5378,
      "step": 63
    },
    {
      "epoch": 0.040345140065403254,
      "grad_norm": 0.15279769897460938,
      "learning_rate": 4.498846929581608e-05,
      "loss": 0.4582,
      "step": 64
    },
    {
      "epoch": 0.04097553287892518,
      "grad_norm": 0.15880808234214783,
      "learning_rate": 4.498773743284227e-05,
      "loss": 0.5224,
      "step": 65
    },
    {
      "epoch": 0.04160592569244711,
      "grad_norm": 0.16909122467041016,
      "learning_rate": 4.498698305931491e-05,
      "loss": 0.4602,
      "step": 66
    },
    {
      "epoch": 0.04223631850596903,
      "grad_norm": 0.16990695893764496,
      "learning_rate": 4.498620617598913e-05,
      "loss": 0.4741,
      "step": 67
    },
    {
      "epoch": 0.042866711319490956,
      "grad_norm": 0.16151531040668488,
      "learning_rate": 4.49854067836426e-05,
      "loss": 0.4053,
      "step": 68
    },
    {
      "epoch": 0.04349710413301288,
      "grad_norm": 0.18641473352909088,
      "learning_rate": 4.498458488307554e-05,
      "loss": 0.4993,
      "step": 69
    },
    {
      "epoch": 0.04412749694653481,
      "grad_norm": 0.1472584754228592,
      "learning_rate": 4.498374047511067e-05,
      "loss": 0.4444,
      "step": 70
    },
    {
      "epoch": 0.044757889760056735,
      "grad_norm": 0.13417768478393555,
      "learning_rate": 4.498287356059326e-05,
      "loss": 0.4316,
      "step": 71
    },
    {
      "epoch": 0.04538828257357866,
      "grad_norm": 0.1857917308807373,
      "learning_rate": 4.4981984140391097e-05,
      "loss": 0.463,
      "step": 72
    },
    {
      "epoch": 0.04601867538710059,
      "grad_norm": 0.16314299404621124,
      "learning_rate": 4.4981072215394514e-05,
      "loss": 0.5322,
      "step": 73
    },
    {
      "epoch": 0.046649068200622514,
      "grad_norm": 0.1854691356420517,
      "learning_rate": 4.498013778651636e-05,
      "loss": 0.4225,
      "step": 74
    },
    {
      "epoch": 0.04727946101414444,
      "grad_norm": 0.19411827623844147,
      "learning_rate": 4.4979180854692005e-05,
      "loss": 0.4597,
      "step": 75
    },
    {
      "epoch": 0.04790985382766636,
      "grad_norm": 0.14335975050926208,
      "learning_rate": 4.4978201420879354e-05,
      "loss": 0.4483,
      "step": 76
    },
    {
      "epoch": 0.04854024664118829,
      "grad_norm": 0.16315117478370667,
      "learning_rate": 4.4977199486058836e-05,
      "loss": 0.4745,
      "step": 77
    },
    {
      "epoch": 0.049170639454710216,
      "grad_norm": 0.15294815599918365,
      "learning_rate": 4.49761750512334e-05,
      "loss": 0.392,
      "step": 78
    },
    {
      "epoch": 0.04980103226823214,
      "grad_norm": 0.165781170129776,
      "learning_rate": 4.497512811742853e-05,
      "loss": 0.401,
      "step": 79
    },
    {
      "epoch": 0.05043142508175407,
      "grad_norm": 0.17177148163318634,
      "learning_rate": 4.497405868569221e-05,
      "loss": 0.49,
      "step": 80
    },
    {
      "epoch": 0.051061817895275995,
      "grad_norm": 0.15722817182540894,
      "learning_rate": 4.497296675709497e-05,
      "loss": 0.4317,
      "step": 81
    },
    {
      "epoch": 0.05169221070879792,
      "grad_norm": 0.14051659405231476,
      "learning_rate": 4.4971852332729826e-05,
      "loss": 0.4655,
      "step": 82
    },
    {
      "epoch": 0.05232260352231984,
      "grad_norm": 0.1831762194633484,
      "learning_rate": 4.4970715413712346e-05,
      "loss": 0.4518,
      "step": 83
    },
    {
      "epoch": 0.05295299633584177,
      "grad_norm": 0.225497305393219,
      "learning_rate": 4.4969556001180605e-05,
      "loss": 0.5071,
      "step": 84
    },
    {
      "epoch": 0.0535833891493637,
      "grad_norm": 0.18495135009288788,
      "learning_rate": 4.496837409629519e-05,
      "loss": 0.491,
      "step": 85
    },
    {
      "epoch": 0.05421378196288562,
      "grad_norm": 0.1926666647195816,
      "learning_rate": 4.4967169700239204e-05,
      "loss": 0.4815,
      "step": 86
    },
    {
      "epoch": 0.05484417477640755,
      "grad_norm": 0.2034398317337036,
      "learning_rate": 4.496594281421826e-05,
      "loss": 0.5421,
      "step": 87
    },
    {
      "epoch": 0.055474567589929476,
      "grad_norm": 0.18834415078163147,
      "learning_rate": 4.4964693439460494e-05,
      "loss": 0.5217,
      "step": 88
    },
    {
      "epoch": 0.0561049604034514,
      "grad_norm": 0.16215714812278748,
      "learning_rate": 4.4963421577216546e-05,
      "loss": 0.5103,
      "step": 89
    },
    {
      "epoch": 0.056735353216973324,
      "grad_norm": 0.15685607492923737,
      "learning_rate": 4.496212722875958e-05,
      "loss": 0.54,
      "step": 90
    },
    {
      "epoch": 0.057365746030495254,
      "grad_norm": 0.18496496975421906,
      "learning_rate": 4.496081039538524e-05,
      "loss": 0.4635,
      "step": 91
    },
    {
      "epoch": 0.05799613884401718,
      "grad_norm": 0.17332017421722412,
      "learning_rate": 4.4959471078411706e-05,
      "loss": 0.4878,
      "step": 92
    },
    {
      "epoch": 0.0586265316575391,
      "grad_norm": 0.16615472733974457,
      "learning_rate": 4.4958109279179644e-05,
      "loss": 0.45,
      "step": 93
    },
    {
      "epoch": 0.05925692447106103,
      "grad_norm": 0.17992667853832245,
      "learning_rate": 4.495672499905225e-05,
      "loss": 0.4838,
      "step": 94
    },
    {
      "epoch": 0.05988731728458296,
      "grad_norm": 0.15236912667751312,
      "learning_rate": 4.49553182394152e-05,
      "loss": 0.4665,
      "step": 95
    },
    {
      "epoch": 0.06051771009810488,
      "grad_norm": 0.17254213988780975,
      "learning_rate": 4.495388900167668e-05,
      "loss": 0.5133,
      "step": 96
    },
    {
      "epoch": 0.061148102911626805,
      "grad_norm": 0.15270383656024933,
      "learning_rate": 4.4952437287267376e-05,
      "loss": 0.4934,
      "step": 97
    },
    {
      "epoch": 0.061778495725148735,
      "grad_norm": 0.14834915101528168,
      "learning_rate": 4.4950963097640486e-05,
      "loss": 0.4598,
      "step": 98
    },
    {
      "epoch": 0.06240888853867066,
      "grad_norm": 0.14422562718391418,
      "learning_rate": 4.4949466434271685e-05,
      "loss": 0.4134,
      "step": 99
    },
    {
      "epoch": 0.06303928135219258,
      "grad_norm": 0.16636601090431213,
      "learning_rate": 4.494794729865916e-05,
      "loss": 0.4554,
      "step": 100
    },
    {
      "epoch": 0.06366967416571451,
      "grad_norm": 0.17716509103775024,
      "learning_rate": 4.4946405692323586e-05,
      "loss": 0.4314,
      "step": 101
    },
    {
      "epoch": 0.06430006697923643,
      "grad_norm": 0.17074547708034515,
      "learning_rate": 4.494484161680813e-05,
      "loss": 0.4192,
      "step": 102
    },
    {
      "epoch": 0.06493045979275837,
      "grad_norm": 0.14279553294181824,
      "learning_rate": 4.494325507367846e-05,
      "loss": 0.3965,
      "step": 103
    },
    {
      "epoch": 0.06556085260628029,
      "grad_norm": 0.17137177288532257,
      "learning_rate": 4.4941646064522724e-05,
      "loss": 0.4707,
      "step": 104
    },
    {
      "epoch": 0.06619124541980222,
      "grad_norm": 0.1495380699634552,
      "learning_rate": 4.4940014590951575e-05,
      "loss": 0.4655,
      "step": 105
    },
    {
      "epoch": 0.06682163823332414,
      "grad_norm": 0.15420159697532654,
      "learning_rate": 4.493836065459813e-05,
      "loss": 0.5068,
      "step": 106
    },
    {
      "epoch": 0.06745203104684606,
      "grad_norm": 0.1462813764810562,
      "learning_rate": 4.4936684257118e-05,
      "loss": 0.3858,
      "step": 107
    },
    {
      "epoch": 0.06808242386036799,
      "grad_norm": 0.18409891426563263,
      "learning_rate": 4.4934985400189296e-05,
      "loss": 0.4831,
      "step": 108
    },
    {
      "epoch": 0.06871281667388991,
      "grad_norm": 0.1710345447063446,
      "learning_rate": 4.493326408551259e-05,
      "loss": 0.4393,
      "step": 109
    },
    {
      "epoch": 0.06934320948741185,
      "grad_norm": 0.17826968431472778,
      "learning_rate": 4.493152031481095e-05,
      "loss": 0.4414,
      "step": 110
    },
    {
      "epoch": 0.06997360230093377,
      "grad_norm": 0.16518640518188477,
      "learning_rate": 4.492975408982991e-05,
      "loss": 0.4779,
      "step": 111
    },
    {
      "epoch": 0.0706039951144557,
      "grad_norm": 0.2125261276960373,
      "learning_rate": 4.492796541233749e-05,
      "loss": 0.4996,
      "step": 112
    },
    {
      "epoch": 0.07123438792797762,
      "grad_norm": 0.16218961775302887,
      "learning_rate": 4.4926154284124176e-05,
      "loss": 0.4495,
      "step": 113
    },
    {
      "epoch": 0.07186478074149955,
      "grad_norm": 0.16842223703861237,
      "learning_rate": 4.492432070700295e-05,
      "loss": 0.4078,
      "step": 114
    },
    {
      "epoch": 0.07249517355502147,
      "grad_norm": 0.16266340017318726,
      "learning_rate": 4.492246468280922e-05,
      "loss": 0.482,
      "step": 115
    },
    {
      "epoch": 0.0731255663685434,
      "grad_norm": 0.157426655292511,
      "learning_rate": 4.492058621340093e-05,
      "loss": 0.4512,
      "step": 116
    },
    {
      "epoch": 0.07375595918206533,
      "grad_norm": 0.15796181559562683,
      "learning_rate": 4.4918685300658424e-05,
      "loss": 0.4002,
      "step": 117
    },
    {
      "epoch": 0.07438635199558725,
      "grad_norm": 0.15874150395393372,
      "learning_rate": 4.4916761946484566e-05,
      "loss": 0.4356,
      "step": 118
    },
    {
      "epoch": 0.07501674480910918,
      "grad_norm": 0.1592918336391449,
      "learning_rate": 4.4914816152804645e-05,
      "loss": 0.4492,
      "step": 119
    },
    {
      "epoch": 0.0756471376226311,
      "grad_norm": 0.1482664942741394,
      "learning_rate": 4.491284792156644e-05,
      "loss": 0.4358,
      "step": 120
    },
    {
      "epoch": 0.07627753043615303,
      "grad_norm": 0.16430510580539703,
      "learning_rate": 4.491085725474018e-05,
      "loss": 0.4256,
      "step": 121
    },
    {
      "epoch": 0.07690792324967495,
      "grad_norm": 0.16677382588386536,
      "learning_rate": 4.4908844154318534e-05,
      "loss": 0.4289,
      "step": 122
    },
    {
      "epoch": 0.07753831606319687,
      "grad_norm": 0.213206946849823,
      "learning_rate": 4.490680862231667e-05,
      "loss": 0.5544,
      "step": 123
    },
    {
      "epoch": 0.07816870887671881,
      "grad_norm": 0.15621592104434967,
      "learning_rate": 4.4904750660772165e-05,
      "loss": 0.4191,
      "step": 124
    },
    {
      "epoch": 0.07879910169024074,
      "grad_norm": 0.19092267751693726,
      "learning_rate": 4.490267027174508e-05,
      "loss": 0.4882,
      "step": 125
    },
    {
      "epoch": 0.07942949450376266,
      "grad_norm": 0.22929497063159943,
      "learning_rate": 4.4900567457317904e-05,
      "loss": 0.5008,
      "step": 126
    },
    {
      "epoch": 0.08005988731728458,
      "grad_norm": 0.18289487063884735,
      "learning_rate": 4.48984422195956e-05,
      "loss": 0.4999,
      "step": 127
    },
    {
      "epoch": 0.08069028013080651,
      "grad_norm": 0.19063183665275574,
      "learning_rate": 4.489629456070555e-05,
      "loss": 0.539,
      "step": 128
    },
    {
      "epoch": 0.08132067294432843,
      "grad_norm": 0.15307193994522095,
      "learning_rate": 4.48941244827976e-05,
      "loss": 0.4312,
      "step": 129
    },
    {
      "epoch": 0.08195106575785036,
      "grad_norm": 0.1701776683330536,
      "learning_rate": 4.4891931988044025e-05,
      "loss": 0.4498,
      "step": 130
    },
    {
      "epoch": 0.0825814585713723,
      "grad_norm": 0.1710740327835083,
      "learning_rate": 4.488971707863954e-05,
      "loss": 0.4704,
      "step": 131
    },
    {
      "epoch": 0.08321185138489422,
      "grad_norm": 0.17142195999622345,
      "learning_rate": 4.4887479756801315e-05,
      "loss": 0.4366,
      "step": 132
    },
    {
      "epoch": 0.08384224419841614,
      "grad_norm": 0.1602955162525177,
      "learning_rate": 4.488522002476893e-05,
      "loss": 0.4615,
      "step": 133
    },
    {
      "epoch": 0.08447263701193806,
      "grad_norm": 0.19058693945407867,
      "learning_rate": 4.488293788480441e-05,
      "loss": 0.4678,
      "step": 134
    },
    {
      "epoch": 0.08510302982545999,
      "grad_norm": 0.1879880428314209,
      "learning_rate": 4.488063333919222e-05,
      "loss": 0.4998,
      "step": 135
    },
    {
      "epoch": 0.08573342263898191,
      "grad_norm": 0.15370860695838928,
      "learning_rate": 4.4878306390239236e-05,
      "loss": 0.4588,
      "step": 136
    },
    {
      "epoch": 0.08636381545250384,
      "grad_norm": 0.18422965705394745,
      "learning_rate": 4.487595704027476e-05,
      "loss": 0.4493,
      "step": 137
    },
    {
      "epoch": 0.08699420826602576,
      "grad_norm": 0.15197817981243134,
      "learning_rate": 4.487358529165054e-05,
      "loss": 0.4298,
      "step": 138
    },
    {
      "epoch": 0.0876246010795477,
      "grad_norm": 0.19497714936733246,
      "learning_rate": 4.4871191146740733e-05,
      "loss": 0.4643,
      "step": 139
    },
    {
      "epoch": 0.08825499389306962,
      "grad_norm": 0.16512389481067657,
      "learning_rate": 4.4868774607941894e-05,
      "loss": 0.4278,
      "step": 140
    },
    {
      "epoch": 0.08888538670659155,
      "grad_norm": 0.19072328507900238,
      "learning_rate": 4.4866335677673036e-05,
      "loss": 0.4674,
      "step": 141
    },
    {
      "epoch": 0.08951577952011347,
      "grad_norm": 0.18181082606315613,
      "learning_rate": 4.486387435837555e-05,
      "loss": 0.5149,
      "step": 142
    },
    {
      "epoch": 0.0901461723336354,
      "grad_norm": 0.15561337769031525,
      "learning_rate": 4.486139065251326e-05,
      "loss": 0.3933,
      "step": 143
    },
    {
      "epoch": 0.09077656514715732,
      "grad_norm": 0.16787169873714447,
      "learning_rate": 4.485888456257238e-05,
      "loss": 0.3935,
      "step": 144
    },
    {
      "epoch": 0.09140695796067924,
      "grad_norm": 0.19648166000843048,
      "learning_rate": 4.485635609106156e-05,
      "loss": 0.4691,
      "step": 145
    },
    {
      "epoch": 0.09203735077420118,
      "grad_norm": 0.16800902783870697,
      "learning_rate": 4.485380524051182e-05,
      "loss": 0.4609,
      "step": 146
    },
    {
      "epoch": 0.0926677435877231,
      "grad_norm": 0.1822555959224701,
      "learning_rate": 4.485123201347661e-05,
      "loss": 0.4519,
      "step": 147
    },
    {
      "epoch": 0.09329813640124503,
      "grad_norm": 0.18529456853866577,
      "learning_rate": 4.484863641253176e-05,
      "loss": 0.4701,
      "step": 148
    },
    {
      "epoch": 0.09392852921476695,
      "grad_norm": 0.17306600511074066,
      "learning_rate": 4.484601844027551e-05,
      "loss": 0.4405,
      "step": 149
    },
    {
      "epoch": 0.09455892202828887,
      "grad_norm": 0.16045962274074554,
      "learning_rate": 4.484337809932849e-05,
      "loss": 0.3381,
      "step": 150
    },
    {
      "epoch": 0.0951893148418108,
      "grad_norm": 0.20732329785823822,
      "learning_rate": 4.484071539233371e-05,
      "loss": 0.5217,
      "step": 151
    },
    {
      "epoch": 0.09581970765533272,
      "grad_norm": 0.17176291346549988,
      "learning_rate": 4.483803032195659e-05,
      "loss": 0.5006,
      "step": 152
    },
    {
      "epoch": 0.09645010046885466,
      "grad_norm": 0.18901783227920532,
      "learning_rate": 4.483532289088492e-05,
      "loss": 0.5125,
      "step": 153
    },
    {
      "epoch": 0.09708049328237658,
      "grad_norm": 0.16525357961654663,
      "learning_rate": 4.4832593101828874e-05,
      "loss": 0.4842,
      "step": 154
    },
    {
      "epoch": 0.09771088609589851,
      "grad_norm": 0.16560229659080505,
      "learning_rate": 4.4829840957521014e-05,
      "loss": 0.4463,
      "step": 155
    },
    {
      "epoch": 0.09834127890942043,
      "grad_norm": 0.18390001356601715,
      "learning_rate": 4.482706646071629e-05,
      "loss": 0.4678,
      "step": 156
    },
    {
      "epoch": 0.09897167172294236,
      "grad_norm": 0.17232733964920044,
      "learning_rate": 4.4824269614191985e-05,
      "loss": 0.4734,
      "step": 157
    },
    {
      "epoch": 0.09960206453646428,
      "grad_norm": 0.18861058354377747,
      "learning_rate": 4.482145042074781e-05,
      "loss": 0.4508,
      "step": 158
    },
    {
      "epoch": 0.1002324573499862,
      "grad_norm": 0.20135930180549622,
      "learning_rate": 4.481860888320581e-05,
      "loss": 0.5127,
      "step": 159
    },
    {
      "epoch": 0.10086285016350814,
      "grad_norm": 0.15137121081352234,
      "learning_rate": 4.4815745004410394e-05,
      "loss": 0.4318,
      "step": 160
    },
    {
      "epoch": 0.10149324297703007,
      "grad_norm": 0.16308103501796722,
      "learning_rate": 4.481285878722835e-05,
      "loss": 0.5055,
      "step": 161
    },
    {
      "epoch": 0.10212363579055199,
      "grad_norm": 0.14730596542358398,
      "learning_rate": 4.480995023454885e-05,
      "loss": 0.4132,
      "step": 162
    },
    {
      "epoch": 0.10275402860407391,
      "grad_norm": 0.1659645140171051,
      "learning_rate": 4.4807019349283365e-05,
      "loss": 0.4884,
      "step": 163
    },
    {
      "epoch": 0.10338442141759584,
      "grad_norm": 0.17068709433078766,
      "learning_rate": 4.4804066134365765e-05,
      "loss": 0.4674,
      "step": 164
    },
    {
      "epoch": 0.10401481423111776,
      "grad_norm": 0.1809260994195938,
      "learning_rate": 4.480109059275227e-05,
      "loss": 0.4232,
      "step": 165
    },
    {
      "epoch": 0.10464520704463968,
      "grad_norm": 0.17435210943222046,
      "learning_rate": 4.479809272742143e-05,
      "loss": 0.4616,
      "step": 166
    },
    {
      "epoch": 0.10527559985816162,
      "grad_norm": 0.15691553056240082,
      "learning_rate": 4.479507254137415e-05,
      "loss": 0.3953,
      "step": 167
    },
    {
      "epoch": 0.10590599267168355,
      "grad_norm": 0.1464596688747406,
      "learning_rate": 4.479203003763369e-05,
      "loss": 0.4254,
      "step": 168
    },
    {
      "epoch": 0.10653638548520547,
      "grad_norm": 0.16407829523086548,
      "learning_rate": 4.478896521924564e-05,
      "loss": 0.4266,
      "step": 169
    },
    {
      "epoch": 0.1071667782987274,
      "grad_norm": 0.15943580865859985,
      "learning_rate": 4.478587808927792e-05,
      "loss": 0.3904,
      "step": 170
    },
    {
      "epoch": 0.10779717111224932,
      "grad_norm": 0.18695223331451416,
      "learning_rate": 4.4782768650820794e-05,
      "loss": 0.5015,
      "step": 171
    },
    {
      "epoch": 0.10842756392577124,
      "grad_norm": 0.17117992043495178,
      "learning_rate": 4.477963690698687e-05,
      "loss": 0.3751,
      "step": 172
    },
    {
      "epoch": 0.10905795673929317,
      "grad_norm": 0.1756797730922699,
      "learning_rate": 4.477648286091104e-05,
      "loss": 0.4789,
      "step": 173
    },
    {
      "epoch": 0.1096883495528151,
      "grad_norm": 0.16904820501804352,
      "learning_rate": 4.477330651575057e-05,
      "loss": 0.5275,
      "step": 174
    },
    {
      "epoch": 0.11031874236633703,
      "grad_norm": 0.1578703373670578,
      "learning_rate": 4.4770107874685035e-05,
      "loss": 0.4494,
      "step": 175
    },
    {
      "epoch": 0.11094913517985895,
      "grad_norm": 0.16940182447433472,
      "learning_rate": 4.4766886940916305e-05,
      "loss": 0.4139,
      "step": 176
    },
    {
      "epoch": 0.11157952799338088,
      "grad_norm": 0.15721531212329865,
      "learning_rate": 4.476364371766859e-05,
      "loss": 0.3958,
      "step": 177
    },
    {
      "epoch": 0.1122099208069028,
      "grad_norm": 0.17659851908683777,
      "learning_rate": 4.47603782081884e-05,
      "loss": 0.4595,
      "step": 178
    },
    {
      "epoch": 0.11284031362042472,
      "grad_norm": 0.1890040636062622,
      "learning_rate": 4.475709041574456e-05,
      "loss": 0.4771,
      "step": 179
    },
    {
      "epoch": 0.11347070643394665,
      "grad_norm": 0.1727362871170044,
      "learning_rate": 4.47537803436282e-05,
      "loss": 0.499,
      "step": 180
    },
    {
      "epoch": 0.11410109924746858,
      "grad_norm": 0.18926569819450378,
      "learning_rate": 4.475044799515275e-05,
      "loss": 0.4292,
      "step": 181
    },
    {
      "epoch": 0.11473149206099051,
      "grad_norm": 0.17432500422000885,
      "learning_rate": 4.474709337365393e-05,
      "loss": 0.4015,
      "step": 182
    },
    {
      "epoch": 0.11536188487451243,
      "grad_norm": 0.1681092083454132,
      "learning_rate": 4.474371648248979e-05,
      "loss": 0.4705,
      "step": 183
    },
    {
      "epoch": 0.11599227768803436,
      "grad_norm": 0.14282703399658203,
      "learning_rate": 4.4740317325040615e-05,
      "loss": 0.3439,
      "step": 184
    },
    {
      "epoch": 0.11662267050155628,
      "grad_norm": 0.16238975524902344,
      "learning_rate": 4.473689590470903e-05,
      "loss": 0.3679,
      "step": 185
    },
    {
      "epoch": 0.1172530633150782,
      "grad_norm": 0.15502974390983582,
      "learning_rate": 4.4733452224919926e-05,
      "loss": 0.4495,
      "step": 186
    },
    {
      "epoch": 0.11788345612860013,
      "grad_norm": 0.18629074096679688,
      "learning_rate": 4.472998628912047e-05,
      "loss": 0.4752,
      "step": 187
    },
    {
      "epoch": 0.11851384894212207,
      "grad_norm": 0.19327090680599213,
      "learning_rate": 4.4726498100780116e-05,
      "loss": 0.4649,
      "step": 188
    },
    {
      "epoch": 0.11914424175564399,
      "grad_norm": 0.18306037783622742,
      "learning_rate": 4.47229876633906e-05,
      "loss": 0.4112,
      "step": 189
    },
    {
      "epoch": 0.11977463456916591,
      "grad_norm": 0.13204658031463623,
      "learning_rate": 4.471945498046591e-05,
      "loss": 0.3987,
      "step": 190
    },
    {
      "epoch": 0.12040502738268784,
      "grad_norm": 0.13157318532466888,
      "learning_rate": 4.471590005554233e-05,
      "loss": 0.3778,
      "step": 191
    },
    {
      "epoch": 0.12103542019620976,
      "grad_norm": 0.140708789229393,
      "learning_rate": 4.471232289217837e-05,
      "loss": 0.3759,
      "step": 192
    },
    {
      "epoch": 0.12166581300973169,
      "grad_norm": 0.15047132968902588,
      "learning_rate": 4.4708723493954836e-05,
      "loss": 0.3844,
      "step": 193
    },
    {
      "epoch": 0.12229620582325361,
      "grad_norm": 0.1584540754556656,
      "learning_rate": 4.4705101864474775e-05,
      "loss": 0.4779,
      "step": 194
    },
    {
      "epoch": 0.12292659863677555,
      "grad_norm": 0.16349102556705475,
      "learning_rate": 4.4701458007363485e-05,
      "loss": 0.3726,
      "step": 195
    },
    {
      "epoch": 0.12355699145029747,
      "grad_norm": 0.14389654994010925,
      "learning_rate": 4.4697791926268535e-05,
      "loss": 0.4332,
      "step": 196
    },
    {
      "epoch": 0.1241873842638194,
      "grad_norm": 0.15718668699264526,
      "learning_rate": 4.4694103624859706e-05,
      "loss": 0.3834,
      "step": 197
    },
    {
      "epoch": 0.12481777707734132,
      "grad_norm": 0.15873458981513977,
      "learning_rate": 4.4690393106829065e-05,
      "loss": 0.4657,
      "step": 198
    },
    {
      "epoch": 0.12544816989086324,
      "grad_norm": 0.17128777503967285,
      "learning_rate": 4.468666037589088e-05,
      "loss": 0.5291,
      "step": 199
    },
    {
      "epoch": 0.12607856270438517,
      "grad_norm": 0.17686426639556885,
      "learning_rate": 4.4682905435781666e-05,
      "loss": 0.3717,
      "step": 200
    },
    {
      "epoch": 0.12607856270438517,
      "eval_loss": 0.47015270590782166,
      "eval_runtime": 226.3613,
      "eval_samples_per_second": 4.418,
      "eval_steps_per_second": 4.418,
      "step": 200
    },
    {
      "epoch": 0.1267089555179071,
      "grad_norm": 0.18956759572029114,
      "learning_rate": 4.4679128290260173e-05,
      "loss": 0.4833,
      "step": 201
    },
    {
      "epoch": 0.12733934833142901,
      "grad_norm": 0.17402750253677368,
      "learning_rate": 4.4675328943107394e-05,
      "loss": 0.3957,
      "step": 202
    },
    {
      "epoch": 0.12796974114495094,
      "grad_norm": 0.1294248402118683,
      "learning_rate": 4.467150739812652e-05,
      "loss": 0.3687,
      "step": 203
    },
    {
      "epoch": 0.12860013395847286,
      "grad_norm": 0.15464654564857483,
      "learning_rate": 4.466766365914297e-05,
      "loss": 0.4203,
      "step": 204
    },
    {
      "epoch": 0.12923052677199479,
      "grad_norm": 0.1686786562204361,
      "learning_rate": 4.4663797730004384e-05,
      "loss": 0.4615,
      "step": 205
    },
    {
      "epoch": 0.12986091958551674,
      "grad_norm": 0.18524964153766632,
      "learning_rate": 4.4659909614580625e-05,
      "loss": 0.4875,
      "step": 206
    },
    {
      "epoch": 0.13049131239903866,
      "grad_norm": 0.14806324243545532,
      "learning_rate": 4.4655999316763744e-05,
      "loss": 0.4546,
      "step": 207
    },
    {
      "epoch": 0.13112170521256059,
      "grad_norm": 0.14976917207241058,
      "learning_rate": 4.465206684046801e-05,
      "loss": 0.4205,
      "step": 208
    },
    {
      "epoch": 0.1317520980260825,
      "grad_norm": 0.21427291631698608,
      "learning_rate": 4.4648112189629884e-05,
      "loss": 0.4247,
      "step": 209
    },
    {
      "epoch": 0.13238249083960443,
      "grad_norm": 0.1890469193458557,
      "learning_rate": 4.4644135368208025e-05,
      "loss": 0.4702,
      "step": 210
    },
    {
      "epoch": 0.13301288365312636,
      "grad_norm": 0.14907729625701904,
      "learning_rate": 4.4640136380183296e-05,
      "loss": 0.3715,
      "step": 211
    },
    {
      "epoch": 0.13364327646664828,
      "grad_norm": 0.1610517054796219,
      "learning_rate": 4.463611522955875e-05,
      "loss": 0.4277,
      "step": 212
    },
    {
      "epoch": 0.1342736692801702,
      "grad_norm": 0.16458484530448914,
      "learning_rate": 4.463207192035961e-05,
      "loss": 0.4274,
      "step": 213
    },
    {
      "epoch": 0.13490406209369213,
      "grad_norm": 0.1576109677553177,
      "learning_rate": 4.462800645663328e-05,
      "loss": 0.3847,
      "step": 214
    },
    {
      "epoch": 0.13553445490721405,
      "grad_norm": 0.1626528799533844,
      "learning_rate": 4.462391884244936e-05,
      "loss": 0.475,
      "step": 215
    },
    {
      "epoch": 0.13616484772073598,
      "grad_norm": 0.19211681187152863,
      "learning_rate": 4.461980908189961e-05,
      "loss": 0.4966,
      "step": 216
    },
    {
      "epoch": 0.1367952405342579,
      "grad_norm": 0.13622134923934937,
      "learning_rate": 4.461567717909796e-05,
      "loss": 0.4061,
      "step": 217
    },
    {
      "epoch": 0.13742563334777982,
      "grad_norm": 0.18433628976345062,
      "learning_rate": 4.46115231381805e-05,
      "loss": 0.4502,
      "step": 218
    },
    {
      "epoch": 0.13805602616130175,
      "grad_norm": 0.1514837145805359,
      "learning_rate": 4.460734696330548e-05,
      "loss": 0.4209,
      "step": 219
    },
    {
      "epoch": 0.1386864189748237,
      "grad_norm": 0.18339303135871887,
      "learning_rate": 4.4603148658653326e-05,
      "loss": 0.4564,
      "step": 220
    },
    {
      "epoch": 0.13931681178834562,
      "grad_norm": 0.18273302912712097,
      "learning_rate": 4.45989282284266e-05,
      "loss": 0.4833,
      "step": 221
    },
    {
      "epoch": 0.13994720460186755,
      "grad_norm": 0.15928077697753906,
      "learning_rate": 4.459468567685001e-05,
      "loss": 0.4816,
      "step": 222
    },
    {
      "epoch": 0.14057759741538947,
      "grad_norm": 0.16953788697719574,
      "learning_rate": 4.459042100817041e-05,
      "loss": 0.4381,
      "step": 223
    },
    {
      "epoch": 0.1412079902289114,
      "grad_norm": 0.19472193717956543,
      "learning_rate": 4.4586134226656804e-05,
      "loss": 0.4603,
      "step": 224
    },
    {
      "epoch": 0.14183838304243332,
      "grad_norm": 0.12984232604503632,
      "learning_rate": 4.458182533660031e-05,
      "loss": 0.3997,
      "step": 225
    },
    {
      "epoch": 0.14246877585595524,
      "grad_norm": 0.12314029783010483,
      "learning_rate": 4.4577494342314205e-05,
      "loss": 0.3807,
      "step": 226
    },
    {
      "epoch": 0.14309916866947717,
      "grad_norm": 0.16256219148635864,
      "learning_rate": 4.457314124813387e-05,
      "loss": 0.4091,
      "step": 227
    },
    {
      "epoch": 0.1437295614829991,
      "grad_norm": 0.1522122621536255,
      "learning_rate": 4.456876605841681e-05,
      "loss": 0.4658,
      "step": 228
    },
    {
      "epoch": 0.14435995429652101,
      "grad_norm": 0.15753582119941711,
      "learning_rate": 4.4564368777542665e-05,
      "loss": 0.4563,
      "step": 229
    },
    {
      "epoch": 0.14499034711004294,
      "grad_norm": 0.15977296233177185,
      "learning_rate": 4.4559949409913166e-05,
      "loss": 0.398,
      "step": 230
    },
    {
      "epoch": 0.14562073992356486,
      "grad_norm": 0.13135172426700592,
      "learning_rate": 4.455550795995218e-05,
      "loss": 0.4045,
      "step": 231
    },
    {
      "epoch": 0.1462511327370868,
      "grad_norm": 0.1841035783290863,
      "learning_rate": 4.4551044432105646e-05,
      "loss": 0.4978,
      "step": 232
    },
    {
      "epoch": 0.1468815255506087,
      "grad_norm": 0.13868924975395203,
      "learning_rate": 4.454655883084163e-05,
      "loss": 0.4585,
      "step": 233
    },
    {
      "epoch": 0.14751191836413066,
      "grad_norm": 0.17346960306167603,
      "learning_rate": 4.454205116065029e-05,
      "loss": 0.4473,
      "step": 234
    },
    {
      "epoch": 0.14814231117765259,
      "grad_norm": 0.1713954508304596,
      "learning_rate": 4.453752142604385e-05,
      "loss": 0.4346,
      "step": 235
    },
    {
      "epoch": 0.1487727039911745,
      "grad_norm": 0.17876474559307098,
      "learning_rate": 4.453296963155666e-05,
      "loss": 0.5037,
      "step": 236
    },
    {
      "epoch": 0.14940309680469643,
      "grad_norm": 0.14628243446350098,
      "learning_rate": 4.452839578174512e-05,
      "loss": 0.416,
      "step": 237
    },
    {
      "epoch": 0.15003348961821836,
      "grad_norm": 0.17380249500274658,
      "learning_rate": 4.452379988118774e-05,
      "loss": 0.4427,
      "step": 238
    },
    {
      "epoch": 0.15066388243174028,
      "grad_norm": 0.17569275200366974,
      "learning_rate": 4.4519181934485054e-05,
      "loss": 0.4676,
      "step": 239
    },
    {
      "epoch": 0.1512942752452622,
      "grad_norm": 0.16404937207698822,
      "learning_rate": 4.451454194625973e-05,
      "loss": 0.425,
      "step": 240
    },
    {
      "epoch": 0.15192466805878413,
      "grad_norm": 0.15239328145980835,
      "learning_rate": 4.450987992115642e-05,
      "loss": 0.3907,
      "step": 241
    },
    {
      "epoch": 0.15255506087230605,
      "grad_norm": 0.16638530790805817,
      "learning_rate": 4.450519586384192e-05,
      "loss": 0.3964,
      "step": 242
    },
    {
      "epoch": 0.15318545368582798,
      "grad_norm": 0.16477054357528687,
      "learning_rate": 4.4500489779005016e-05,
      "loss": 0.4793,
      "step": 243
    },
    {
      "epoch": 0.1538158464993499,
      "grad_norm": 0.19650088250637054,
      "learning_rate": 4.4495761671356574e-05,
      "loss": 0.5265,
      "step": 244
    },
    {
      "epoch": 0.15444623931287182,
      "grad_norm": 0.1595885008573532,
      "learning_rate": 4.4491011545629495e-05,
      "loss": 0.4893,
      "step": 245
    },
    {
      "epoch": 0.15507663212639375,
      "grad_norm": 0.17801198363304138,
      "learning_rate": 4.4486239406578736e-05,
      "loss": 0.5026,
      "step": 246
    },
    {
      "epoch": 0.15570702493991567,
      "grad_norm": 0.16517923772335052,
      "learning_rate": 4.4481445258981266e-05,
      "loss": 0.3776,
      "step": 247
    },
    {
      "epoch": 0.15633741775343762,
      "grad_norm": 0.1921938955783844,
      "learning_rate": 4.44766291076361e-05,
      "loss": 0.4077,
      "step": 248
    },
    {
      "epoch": 0.15696781056695955,
      "grad_norm": 0.15358708798885345,
      "learning_rate": 4.4471790957364284e-05,
      "loss": 0.4195,
      "step": 249
    },
    {
      "epoch": 0.15759820338048147,
      "grad_norm": 0.16735200583934784,
      "learning_rate": 4.446693081300887e-05,
      "loss": 0.4189,
      "step": 250
    },
    {
      "epoch": 0.1582285961940034,
      "grad_norm": 0.16738048195838928,
      "learning_rate": 4.446204867943493e-05,
      "loss": 0.4908,
      "step": 251
    },
    {
      "epoch": 0.15885898900752532,
      "grad_norm": 0.15153180062770844,
      "learning_rate": 4.445714456152955e-05,
      "loss": 0.5082,
      "step": 252
    },
    {
      "epoch": 0.15948938182104724,
      "grad_norm": 0.17264169454574585,
      "learning_rate": 4.445221846420183e-05,
      "loss": 0.4502,
      "step": 253
    },
    {
      "epoch": 0.16011977463456917,
      "grad_norm": 0.17027878761291504,
      "learning_rate": 4.4447270392382865e-05,
      "loss": 0.4815,
      "step": 254
    },
    {
      "epoch": 0.1607501674480911,
      "grad_norm": 0.17320281267166138,
      "learning_rate": 4.444230035102575e-05,
      "loss": 0.5007,
      "step": 255
    },
    {
      "epoch": 0.16138056026161302,
      "grad_norm": 0.1453806608915329,
      "learning_rate": 4.4437308345105554e-05,
      "loss": 0.4388,
      "step": 256
    },
    {
      "epoch": 0.16201095307513494,
      "grad_norm": 0.1596660614013672,
      "learning_rate": 4.443229437961937e-05,
      "loss": 0.5086,
      "step": 257
    },
    {
      "epoch": 0.16264134588865686,
      "grad_norm": 0.1486014723777771,
      "learning_rate": 4.442725845958623e-05,
      "loss": 0.4556,
      "step": 258
    },
    {
      "epoch": 0.1632717387021788,
      "grad_norm": 0.15127785503864288,
      "learning_rate": 4.442220059004717e-05,
      "loss": 0.467,
      "step": 259
    },
    {
      "epoch": 0.1639021315157007,
      "grad_norm": 0.1469348967075348,
      "learning_rate": 4.4417120776065207e-05,
      "loss": 0.4327,
      "step": 260
    },
    {
      "epoch": 0.16453252432922263,
      "grad_norm": 0.18044887483119965,
      "learning_rate": 4.441201902272529e-05,
      "loss": 0.4523,
      "step": 261
    },
    {
      "epoch": 0.1651629171427446,
      "grad_norm": 0.17343007028102875,
      "learning_rate": 4.440689533513435e-05,
      "loss": 0.4976,
      "step": 262
    },
    {
      "epoch": 0.1657933099562665,
      "grad_norm": 0.17072178423404694,
      "learning_rate": 4.440174971842128e-05,
      "loss": 0.4717,
      "step": 263
    },
    {
      "epoch": 0.16642370276978843,
      "grad_norm": 0.1547555923461914,
      "learning_rate": 4.439658217773691e-05,
      "loss": 0.3654,
      "step": 264
    },
    {
      "epoch": 0.16705409558331036,
      "grad_norm": 0.15340355038642883,
      "learning_rate": 4.4391392718254044e-05,
      "loss": 0.4425,
      "step": 265
    },
    {
      "epoch": 0.16768448839683228,
      "grad_norm": 0.1566683053970337,
      "learning_rate": 4.438618134516738e-05,
      "loss": 0.399,
      "step": 266
    },
    {
      "epoch": 0.1683148812103542,
      "grad_norm": 0.19167183339595795,
      "learning_rate": 4.438094806369359e-05,
      "loss": 0.4727,
      "step": 267
    },
    {
      "epoch": 0.16894527402387613,
      "grad_norm": 0.1682709902524948,
      "learning_rate": 4.4375692879071265e-05,
      "loss": 0.4369,
      "step": 268
    },
    {
      "epoch": 0.16957566683739805,
      "grad_norm": 0.19236905872821808,
      "learning_rate": 4.437041579656092e-05,
      "loss": 0.4765,
      "step": 269
    },
    {
      "epoch": 0.17020605965091998,
      "grad_norm": 0.20469754934310913,
      "learning_rate": 4.4365116821445e-05,
      "loss": 0.5317,
      "step": 270
    },
    {
      "epoch": 0.1708364524644419,
      "grad_norm": 0.15680178999900818,
      "learning_rate": 4.435979595902785e-05,
      "loss": 0.4359,
      "step": 271
    },
    {
      "epoch": 0.17146684527796383,
      "grad_norm": 0.1784251481294632,
      "learning_rate": 4.4354453214635725e-05,
      "loss": 0.5123,
      "step": 272
    },
    {
      "epoch": 0.17209723809148575,
      "grad_norm": 0.14888904988765717,
      "learning_rate": 4.4349088593616796e-05,
      "loss": 0.4192,
      "step": 273
    },
    {
      "epoch": 0.17272763090500767,
      "grad_norm": 0.18458880484104156,
      "learning_rate": 4.434370210134113e-05,
      "loss": 0.5188,
      "step": 274
    },
    {
      "epoch": 0.1733580237185296,
      "grad_norm": 0.15294192731380463,
      "learning_rate": 4.433829374320068e-05,
      "loss": 0.4913,
      "step": 275
    },
    {
      "epoch": 0.17398841653205152,
      "grad_norm": 0.15933866798877716,
      "learning_rate": 4.433286352460929e-05,
      "loss": 0.4916,
      "step": 276
    },
    {
      "epoch": 0.17461880934557347,
      "grad_norm": 0.21269841492176056,
      "learning_rate": 4.43274114510027e-05,
      "loss": 0.5095,
      "step": 277
    },
    {
      "epoch": 0.1752492021590954,
      "grad_norm": 0.1545397788286209,
      "learning_rate": 4.4321937527838496e-05,
      "loss": 0.4158,
      "step": 278
    },
    {
      "epoch": 0.17587959497261732,
      "grad_norm": 0.1311575025320053,
      "learning_rate": 4.431644176059617e-05,
      "loss": 0.3272,
      "step": 279
    },
    {
      "epoch": 0.17650998778613924,
      "grad_norm": 0.18615727126598358,
      "learning_rate": 4.431092415477706e-05,
      "loss": 0.4475,
      "step": 280
    },
    {
      "epoch": 0.17714038059966117,
      "grad_norm": 0.15645916759967804,
      "learning_rate": 4.430538471590437e-05,
      "loss": 0.4821,
      "step": 281
    },
    {
      "epoch": 0.1777707734131831,
      "grad_norm": 0.15182305872440338,
      "learning_rate": 4.429982344952316e-05,
      "loss": 0.3472,
      "step": 282
    },
    {
      "epoch": 0.17840116622670502,
      "grad_norm": 0.1452929824590683,
      "learning_rate": 4.429424036120034e-05,
      "loss": 0.3909,
      "step": 283
    },
    {
      "epoch": 0.17903155904022694,
      "grad_norm": 0.13374878466129303,
      "learning_rate": 4.4288635456524654e-05,
      "loss": 0.396,
      "step": 284
    },
    {
      "epoch": 0.17966195185374886,
      "grad_norm": 0.17100194096565247,
      "learning_rate": 4.428300874110671e-05,
      "loss": 0.5189,
      "step": 285
    },
    {
      "epoch": 0.1802923446672708,
      "grad_norm": 0.21815529465675354,
      "learning_rate": 4.427736022057892e-05,
      "loss": 0.4698,
      "step": 286
    },
    {
      "epoch": 0.1809227374807927,
      "grad_norm": 0.13070501387119293,
      "learning_rate": 4.427168990059554e-05,
      "loss": 0.4313,
      "step": 287
    },
    {
      "epoch": 0.18155313029431464,
      "grad_norm": 0.15077956020832062,
      "learning_rate": 4.426599778683263e-05,
      "loss": 0.4386,
      "step": 288
    },
    {
      "epoch": 0.18218352310783656,
      "grad_norm": 0.16122879087924957,
      "learning_rate": 4.426028388498811e-05,
      "loss": 0.4387,
      "step": 289
    },
    {
      "epoch": 0.18281391592135848,
      "grad_norm": 0.16982218623161316,
      "learning_rate": 4.4254548200781644e-05,
      "loss": 0.5156,
      "step": 290
    },
    {
      "epoch": 0.18344430873488043,
      "grad_norm": 0.16496285796165466,
      "learning_rate": 4.4248790739954764e-05,
      "loss": 0.4783,
      "step": 291
    },
    {
      "epoch": 0.18407470154840236,
      "grad_norm": 0.15385004878044128,
      "learning_rate": 4.424301150827075e-05,
      "loss": 0.4368,
      "step": 292
    },
    {
      "epoch": 0.18470509436192428,
      "grad_norm": 0.16213051974773407,
      "learning_rate": 4.423721051151472e-05,
      "loss": 0.4644,
      "step": 293
    },
    {
      "epoch": 0.1853354871754462,
      "grad_norm": 0.13787910342216492,
      "learning_rate": 4.423138775549353e-05,
      "loss": 0.4552,
      "step": 294
    },
    {
      "epoch": 0.18596587998896813,
      "grad_norm": 0.18587398529052734,
      "learning_rate": 4.422554324603586e-05,
      "loss": 0.5059,
      "step": 295
    },
    {
      "epoch": 0.18659627280249005,
      "grad_norm": 0.15263552963733673,
      "learning_rate": 4.421967698899215e-05,
      "loss": 0.4196,
      "step": 296
    },
    {
      "epoch": 0.18722666561601198,
      "grad_norm": 0.1526150405406952,
      "learning_rate": 4.421378899023461e-05,
      "loss": 0.403,
      "step": 297
    },
    {
      "epoch": 0.1878570584295339,
      "grad_norm": 0.15569239854812622,
      "learning_rate": 4.420787925565721e-05,
      "loss": 0.4181,
      "step": 298
    },
    {
      "epoch": 0.18848745124305583,
      "grad_norm": 0.15740294754505157,
      "learning_rate": 4.420194779117567e-05,
      "loss": 0.4911,
      "step": 299
    },
    {
      "epoch": 0.18911784405657775,
      "grad_norm": 0.1633409857749939,
      "learning_rate": 4.419599460272748e-05,
      "loss": 0.4817,
      "step": 300
    },
    {
      "epoch": 0.18974823687009967,
      "grad_norm": 0.14763297140598297,
      "learning_rate": 4.419001969627188e-05,
      "loss": 0.4,
      "step": 301
    },
    {
      "epoch": 0.1903786296836216,
      "grad_norm": 0.15027102828025818,
      "learning_rate": 4.4184023077789824e-05,
      "loss": 0.4388,
      "step": 302
    },
    {
      "epoch": 0.19100902249714352,
      "grad_norm": 0.1539442092180252,
      "learning_rate": 4.417800475328402e-05,
      "loss": 0.4624,
      "step": 303
    },
    {
      "epoch": 0.19163941531066545,
      "grad_norm": 0.17854462563991547,
      "learning_rate": 4.41719647287789e-05,
      "loss": 0.4952,
      "step": 304
    },
    {
      "epoch": 0.1922698081241874,
      "grad_norm": 0.1574363261461258,
      "learning_rate": 4.4165903010320614e-05,
      "loss": 0.4102,
      "step": 305
    },
    {
      "epoch": 0.19290020093770932,
      "grad_norm": 0.14707572758197784,
      "learning_rate": 4.4159819603977035e-05,
      "loss": 0.3848,
      "step": 306
    },
    {
      "epoch": 0.19353059375123124,
      "grad_norm": 0.1398138850927353,
      "learning_rate": 4.4153714515837734e-05,
      "loss": 0.3968,
      "step": 307
    },
    {
      "epoch": 0.19416098656475317,
      "grad_norm": 0.16176103055477142,
      "learning_rate": 4.4147587752014e-05,
      "loss": 0.4948,
      "step": 308
    },
    {
      "epoch": 0.1947913793782751,
      "grad_norm": 0.1668642908334732,
      "learning_rate": 4.414143931863881e-05,
      "loss": 0.4635,
      "step": 309
    },
    {
      "epoch": 0.19542177219179702,
      "grad_norm": 0.14107371866703033,
      "learning_rate": 4.413526922186683e-05,
      "loss": 0.3535,
      "step": 310
    },
    {
      "epoch": 0.19605216500531894,
      "grad_norm": 0.15369193255901337,
      "learning_rate": 4.412907746787444e-05,
      "loss": 0.3719,
      "step": 311
    },
    {
      "epoch": 0.19668255781884086,
      "grad_norm": 0.16373153030872345,
      "learning_rate": 4.412286406285964e-05,
      "loss": 0.4232,
      "step": 312
    },
    {
      "epoch": 0.1973129506323628,
      "grad_norm": 0.13650447130203247,
      "learning_rate": 4.411662901304217e-05,
      "loss": 0.4114,
      "step": 313
    },
    {
      "epoch": 0.1979433434458847,
      "grad_norm": 0.17574451863765717,
      "learning_rate": 4.411037232466339e-05,
      "loss": 0.5067,
      "step": 314
    },
    {
      "epoch": 0.19857373625940664,
      "grad_norm": 0.19951093196868896,
      "learning_rate": 4.410409400398634e-05,
      "loss": 0.5274,
      "step": 315
    },
    {
      "epoch": 0.19920412907292856,
      "grad_norm": 0.15199385583400726,
      "learning_rate": 4.409779405729572e-05,
      "loss": 0.4369,
      "step": 316
    },
    {
      "epoch": 0.19983452188645048,
      "grad_norm": 0.1495632529258728,
      "learning_rate": 4.409147249089786e-05,
      "loss": 0.4449,
      "step": 317
    },
    {
      "epoch": 0.2004649146999724,
      "grad_norm": 0.16323602199554443,
      "learning_rate": 4.408512931112073e-05,
      "loss": 0.42,
      "step": 318
    },
    {
      "epoch": 0.20109530751349436,
      "grad_norm": 0.16794022917747498,
      "learning_rate": 4.407876452431397e-05,
      "loss": 0.4548,
      "step": 319
    },
    {
      "epoch": 0.20172570032701628,
      "grad_norm": 0.1724035143852234,
      "learning_rate": 4.40723781368488e-05,
      "loss": 0.4969,
      "step": 320
    },
    {
      "epoch": 0.2023560931405382,
      "grad_norm": 0.16559875011444092,
      "learning_rate": 4.40659701551181e-05,
      "loss": 0.4681,
      "step": 321
    },
    {
      "epoch": 0.20298648595406013,
      "grad_norm": 0.1622600555419922,
      "learning_rate": 4.405954058553635e-05,
      "loss": 0.5081,
      "step": 322
    },
    {
      "epoch": 0.20361687876758205,
      "grad_norm": 0.1619727909564972,
      "learning_rate": 4.405308943453965e-05,
      "loss": 0.4478,
      "step": 323
    },
    {
      "epoch": 0.20424727158110398,
      "grad_norm": 0.12752674520015717,
      "learning_rate": 4.404661670858568e-05,
      "loss": 0.3779,
      "step": 324
    },
    {
      "epoch": 0.2048776643946259,
      "grad_norm": 0.163992777466774,
      "learning_rate": 4.404012241415375e-05,
      "loss": 0.3743,
      "step": 325
    },
    {
      "epoch": 0.20550805720814783,
      "grad_norm": 0.16608774662017822,
      "learning_rate": 4.403360655774474e-05,
      "loss": 0.4472,
      "step": 326
    },
    {
      "epoch": 0.20613845002166975,
      "grad_norm": 0.1689046025276184,
      "learning_rate": 4.4027069145881116e-05,
      "loss": 0.4166,
      "step": 327
    },
    {
      "epoch": 0.20676884283519167,
      "grad_norm": 0.180549755692482,
      "learning_rate": 4.402051018510691e-05,
      "loss": 0.4872,
      "step": 328
    },
    {
      "epoch": 0.2073992356487136,
      "grad_norm": 0.18701449036598206,
      "learning_rate": 4.4013929681987766e-05,
      "loss": 0.4188,
      "step": 329
    },
    {
      "epoch": 0.20802962846223552,
      "grad_norm": 0.161424919962883,
      "learning_rate": 4.400732764311084e-05,
      "loss": 0.4628,
      "step": 330
    },
    {
      "epoch": 0.20866002127575745,
      "grad_norm": 0.140961155295372,
      "learning_rate": 4.400070407508487e-05,
      "loss": 0.4086,
      "step": 331
    },
    {
      "epoch": 0.20929041408927937,
      "grad_norm": 0.12989573180675507,
      "learning_rate": 4.399405898454015e-05,
      "loss": 0.4245,
      "step": 332
    },
    {
      "epoch": 0.20992080690280132,
      "grad_norm": 0.1708124876022339,
      "learning_rate": 4.398739237812852e-05,
      "loss": 0.4754,
      "step": 333
    },
    {
      "epoch": 0.21055119971632325,
      "grad_norm": 0.1742548793554306,
      "learning_rate": 4.3980704262523335e-05,
      "loss": 0.4322,
      "step": 334
    },
    {
      "epoch": 0.21118159252984517,
      "grad_norm": 0.12426616251468658,
      "learning_rate": 4.3973994644419515e-05,
      "loss": 0.3293,
      "step": 335
    },
    {
      "epoch": 0.2118119853433671,
      "grad_norm": 0.12249540537595749,
      "learning_rate": 4.396726353053346e-05,
      "loss": 0.3001,
      "step": 336
    },
    {
      "epoch": 0.21244237815688902,
      "grad_norm": 0.19791898131370544,
      "learning_rate": 4.396051092760314e-05,
      "loss": 0.4886,
      "step": 337
    },
    {
      "epoch": 0.21307277097041094,
      "grad_norm": 0.18151599168777466,
      "learning_rate": 4.3953736842388006e-05,
      "loss": 0.4644,
      "step": 338
    },
    {
      "epoch": 0.21370316378393286,
      "grad_norm": 0.15587259829044342,
      "learning_rate": 4.3946941281669e-05,
      "loss": 0.4522,
      "step": 339
    },
    {
      "epoch": 0.2143335565974548,
      "grad_norm": 0.14205743372440338,
      "learning_rate": 4.394012425224858e-05,
      "loss": 0.4457,
      "step": 340
    },
    {
      "epoch": 0.2149639494109767,
      "grad_norm": 0.1723662167787552,
      "learning_rate": 4.39332857609507e-05,
      "loss": 0.4654,
      "step": 341
    },
    {
      "epoch": 0.21559434222449864,
      "grad_norm": 0.169973224401474,
      "learning_rate": 4.392642581462079e-05,
      "loss": 0.4172,
      "step": 342
    },
    {
      "epoch": 0.21622473503802056,
      "grad_norm": 0.20729491114616394,
      "learning_rate": 4.391954442012576e-05,
      "loss": 0.5264,
      "step": 343
    },
    {
      "epoch": 0.21685512785154248,
      "grad_norm": 0.14070910215377808,
      "learning_rate": 4.3912641584353965e-05,
      "loss": 0.3569,
      "step": 344
    },
    {
      "epoch": 0.2174855206650644,
      "grad_norm": 0.1460694670677185,
      "learning_rate": 4.390571731421527e-05,
      "loss": 0.3603,
      "step": 345
    },
    {
      "epoch": 0.21811591347858633,
      "grad_norm": 0.17333248257637024,
      "learning_rate": 4.389877161664096e-05,
      "loss": 0.4305,
      "step": 346
    },
    {
      "epoch": 0.21874630629210828,
      "grad_norm": 0.17414788901805878,
      "learning_rate": 4.3891804498583776e-05,
      "loss": 0.4788,
      "step": 347
    },
    {
      "epoch": 0.2193766991056302,
      "grad_norm": 0.13664431869983673,
      "learning_rate": 4.388481596701791e-05,
      "loss": 0.3736,
      "step": 348
    },
    {
      "epoch": 0.22000709191915213,
      "grad_norm": 0.17150871455669403,
      "learning_rate": 4.387780602893898e-05,
      "loss": 0.4915,
      "step": 349
    },
    {
      "epoch": 0.22063748473267406,
      "grad_norm": 0.15720312297344208,
      "learning_rate": 4.387077469136404e-05,
      "loss": 0.4359,
      "step": 350
    },
    {
      "epoch": 0.22126787754619598,
      "grad_norm": 0.15001578629016876,
      "learning_rate": 4.386372196133157e-05,
      "loss": 0.4145,
      "step": 351
    },
    {
      "epoch": 0.2218982703597179,
      "grad_norm": 0.18784606456756592,
      "learning_rate": 4.385664784590144e-05,
      "loss": 0.5409,
      "step": 352
    },
    {
      "epoch": 0.22252866317323983,
      "grad_norm": 0.18099607527256012,
      "learning_rate": 4.384955235215495e-05,
      "loss": 0.427,
      "step": 353
    },
    {
      "epoch": 0.22315905598676175,
      "grad_norm": 0.13089902698993683,
      "learning_rate": 4.3842435487194786e-05,
      "loss": 0.471,
      "step": 354
    },
    {
      "epoch": 0.22378944880028367,
      "grad_norm": 0.1351536512374878,
      "learning_rate": 4.383529725814504e-05,
      "loss": 0.4013,
      "step": 355
    },
    {
      "epoch": 0.2244198416138056,
      "grad_norm": 0.14495325088500977,
      "learning_rate": 4.3828137672151194e-05,
      "loss": 0.4244,
      "step": 356
    },
    {
      "epoch": 0.22505023442732752,
      "grad_norm": 0.1470225751399994,
      "learning_rate": 4.382095673638008e-05,
      "loss": 0.424,
      "step": 357
    },
    {
      "epoch": 0.22568062724084945,
      "grad_norm": 0.1563224196434021,
      "learning_rate": 4.3813754458019925e-05,
      "loss": 0.4162,
      "step": 358
    },
    {
      "epoch": 0.22631102005437137,
      "grad_norm": 0.13134637475013733,
      "learning_rate": 4.3806530844280324e-05,
      "loss": 0.3668,
      "step": 359
    },
    {
      "epoch": 0.2269414128678933,
      "grad_norm": 0.14569929242134094,
      "learning_rate": 4.379928590239221e-05,
      "loss": 0.3828,
      "step": 360
    },
    {
      "epoch": 0.22757180568141525,
      "grad_norm": 0.138423353433609,
      "learning_rate": 4.379201963960788e-05,
      "loss": 0.4409,
      "step": 361
    },
    {
      "epoch": 0.22820219849493717,
      "grad_norm": 0.16363699734210968,
      "learning_rate": 4.378473206320097e-05,
      "loss": 0.449,
      "step": 362
    },
    {
      "epoch": 0.2288325913084591,
      "grad_norm": 0.14801304042339325,
      "learning_rate": 4.3777423180466436e-05,
      "loss": 0.4623,
      "step": 363
    },
    {
      "epoch": 0.22946298412198102,
      "grad_norm": 0.20460020005702972,
      "learning_rate": 4.3770092998720605e-05,
      "loss": 0.5039,
      "step": 364
    },
    {
      "epoch": 0.23009337693550294,
      "grad_norm": 0.17865395545959473,
      "learning_rate": 4.376274152530107e-05,
      "loss": 0.4483,
      "step": 365
    },
    {
      "epoch": 0.23072376974902487,
      "grad_norm": 0.17483441531658173,
      "learning_rate": 4.3755368767566766e-05,
      "loss": 0.3916,
      "step": 366
    },
    {
      "epoch": 0.2313541625625468,
      "grad_norm": 0.13383352756500244,
      "learning_rate": 4.3747974732897945e-05,
      "loss": 0.379,
      "step": 367
    },
    {
      "epoch": 0.2319845553760687,
      "grad_norm": 0.1292165219783783,
      "learning_rate": 4.374055942869614e-05,
      "loss": 0.3773,
      "step": 368
    },
    {
      "epoch": 0.23261494818959064,
      "grad_norm": 0.1722308248281479,
      "learning_rate": 4.373312286238417e-05,
      "loss": 0.4104,
      "step": 369
    },
    {
      "epoch": 0.23324534100311256,
      "grad_norm": 0.13454218208789825,
      "learning_rate": 4.372566504140615e-05,
      "loss": 0.4483,
      "step": 370
    },
    {
      "epoch": 0.23387573381663448,
      "grad_norm": 0.16832982003688812,
      "learning_rate": 4.3718185973227477e-05,
      "loss": 0.3664,
      "step": 371
    },
    {
      "epoch": 0.2345061266301564,
      "grad_norm": 0.18032234907150269,
      "learning_rate": 4.371068566533479e-05,
      "loss": 0.4919,
      "step": 372
    },
    {
      "epoch": 0.23513651944367833,
      "grad_norm": 0.16197332739830017,
      "learning_rate": 4.3703164125236024e-05,
      "loss": 0.4078,
      "step": 373
    },
    {
      "epoch": 0.23576691225720026,
      "grad_norm": 0.14080944657325745,
      "learning_rate": 4.369562136046034e-05,
      "loss": 0.4235,
      "step": 374
    },
    {
      "epoch": 0.23639730507072218,
      "grad_norm": 0.13689732551574707,
      "learning_rate": 4.368805737855816e-05,
      "loss": 0.4384,
      "step": 375
    },
    {
      "epoch": 0.23702769788424413,
      "grad_norm": 0.13256917893886566,
      "learning_rate": 4.3680472187101135e-05,
      "loss": 0.3782,
      "step": 376
    },
    {
      "epoch": 0.23765809069776606,
      "grad_norm": 0.16238057613372803,
      "learning_rate": 4.367286579368216e-05,
      "loss": 0.4169,
      "step": 377
    },
    {
      "epoch": 0.23828848351128798,
      "grad_norm": 0.18068745732307434,
      "learning_rate": 4.366523820591535e-05,
      "loss": 0.4064,
      "step": 378
    },
    {
      "epoch": 0.2389188763248099,
      "grad_norm": 0.15773656964302063,
      "learning_rate": 4.365758943143602e-05,
      "loss": 0.4131,
      "step": 379
    },
    {
      "epoch": 0.23954926913833183,
      "grad_norm": 0.1459636241197586,
      "learning_rate": 4.36499194779007e-05,
      "loss": 0.4,
      "step": 380
    },
    {
      "epoch": 0.24017966195185375,
      "grad_norm": 0.15780167281627655,
      "learning_rate": 4.364222835298714e-05,
      "loss": 0.3767,
      "step": 381
    },
    {
      "epoch": 0.24081005476537568,
      "grad_norm": 0.19021441042423248,
      "learning_rate": 4.3634516064394264e-05,
      "loss": 0.4205,
      "step": 382
    },
    {
      "epoch": 0.2414404475788976,
      "grad_norm": 0.1581646203994751,
      "learning_rate": 4.362678261984218e-05,
      "loss": 0.4128,
      "step": 383
    },
    {
      "epoch": 0.24207084039241952,
      "grad_norm": 0.17075695097446442,
      "learning_rate": 4.361902802707219e-05,
      "loss": 0.4132,
      "step": 384
    },
    {
      "epoch": 0.24270123320594145,
      "grad_norm": 0.14225608110427856,
      "learning_rate": 4.361125229384674e-05,
      "loss": 0.3943,
      "step": 385
    },
    {
      "epoch": 0.24333162601946337,
      "grad_norm": 0.14167103171348572,
      "learning_rate": 4.3603455427949465e-05,
      "loss": 0.4041,
      "step": 386
    },
    {
      "epoch": 0.2439620188329853,
      "grad_norm": 0.16991916298866272,
      "learning_rate": 4.3595637437185133e-05,
      "loss": 0.4617,
      "step": 387
    },
    {
      "epoch": 0.24459241164650722,
      "grad_norm": 0.1786823570728302,
      "learning_rate": 4.358779832937968e-05,
      "loss": 0.5302,
      "step": 388
    },
    {
      "epoch": 0.24522280446002914,
      "grad_norm": 0.17845723032951355,
      "learning_rate": 4.3579938112380154e-05,
      "loss": 0.4843,
      "step": 389
    },
    {
      "epoch": 0.2458531972735511,
      "grad_norm": 0.14564692974090576,
      "learning_rate": 4.357205679405475e-05,
      "loss": 0.3841,
      "step": 390
    },
    {
      "epoch": 0.24648359008707302,
      "grad_norm": 0.1389242708683014,
      "learning_rate": 4.356415438229279e-05,
      "loss": 0.3282,
      "step": 391
    },
    {
      "epoch": 0.24711398290059494,
      "grad_norm": 0.15760691463947296,
      "learning_rate": 4.355623088500469e-05,
      "loss": 0.3952,
      "step": 392
    },
    {
      "epoch": 0.24774437571411687,
      "grad_norm": 0.24291901290416718,
      "learning_rate": 4.354828631012199e-05,
      "loss": 0.5629,
      "step": 393
    },
    {
      "epoch": 0.2483747685276388,
      "grad_norm": 0.1695348471403122,
      "learning_rate": 4.354032066559735e-05,
      "loss": 0.4149,
      "step": 394
    },
    {
      "epoch": 0.2490051613411607,
      "grad_norm": 0.16605013608932495,
      "learning_rate": 4.353233395940445e-05,
      "loss": 0.4163,
      "step": 395
    },
    {
      "epoch": 0.24963555415468264,
      "grad_norm": 0.1307298243045807,
      "learning_rate": 4.3524326199538144e-05,
      "loss": 0.3642,
      "step": 396
    },
    {
      "epoch": 0.2502659469682046,
      "grad_norm": 0.15121614933013916,
      "learning_rate": 4.351629739401429e-05,
      "loss": 0.3987,
      "step": 397
    },
    {
      "epoch": 0.2508963397817265,
      "grad_norm": 0.13745561242103577,
      "learning_rate": 4.3508247550869856e-05,
      "loss": 0.39,
      "step": 398
    },
    {
      "epoch": 0.25152673259524844,
      "grad_norm": 0.14871643483638763,
      "learning_rate": 4.350017667816283e-05,
      "loss": 0.3458,
      "step": 399
    },
    {
      "epoch": 0.25215712540877033,
      "grad_norm": 0.14765329658985138,
      "learning_rate": 4.34920847839723e-05,
      "loss": 0.3782,
      "step": 400
    },
    {
      "epoch": 0.25215712540877033,
      "eval_loss": 0.45531120896339417,
      "eval_runtime": 222.4652,
      "eval_samples_per_second": 4.495,
      "eval_steps_per_second": 4.495,
      "step": 400
    },
    {
      "epoch": 0.2527875182222923,
      "grad_norm": 0.18058902025222778,
      "learning_rate": 4.348397187639835e-05,
      "loss": 0.4872,
      "step": 401
    },
    {
      "epoch": 0.2534179110358142,
      "grad_norm": 0.1355857402086258,
      "learning_rate": 4.347583796356213e-05,
      "loss": 0.3838,
      "step": 402
    },
    {
      "epoch": 0.25404830384933613,
      "grad_norm": 0.14737853407859802,
      "learning_rate": 4.3467683053605806e-05,
      "loss": 0.4694,
      "step": 403
    },
    {
      "epoch": 0.25467869666285803,
      "grad_norm": 0.14057587087154388,
      "learning_rate": 4.3459507154692555e-05,
      "loss": 0.4103,
      "step": 404
    },
    {
      "epoch": 0.25530908947638,
      "grad_norm": 0.1664663404226303,
      "learning_rate": 4.3451310275006576e-05,
      "loss": 0.5004,
      "step": 405
    },
    {
      "epoch": 0.2559394822899019,
      "grad_norm": 0.12404580414295197,
      "learning_rate": 4.344309242275306e-05,
      "loss": 0.3744,
      "step": 406
    },
    {
      "epoch": 0.25656987510342383,
      "grad_norm": 0.13321265578269958,
      "learning_rate": 4.343485360615821e-05,
      "loss": 0.3566,
      "step": 407
    },
    {
      "epoch": 0.2572002679169457,
      "grad_norm": 0.17092850804328918,
      "learning_rate": 4.342659383346918e-05,
      "loss": 0.4143,
      "step": 408
    },
    {
      "epoch": 0.2578306607304677,
      "grad_norm": 0.14100387692451477,
      "learning_rate": 4.341831311295415e-05,
      "loss": 0.4269,
      "step": 409
    },
    {
      "epoch": 0.25846105354398957,
      "grad_norm": 0.1483694463968277,
      "learning_rate": 4.341001145290222e-05,
      "loss": 0.4244,
      "step": 410
    },
    {
      "epoch": 0.2590914463575115,
      "grad_norm": 0.1616392880678177,
      "learning_rate": 4.340168886162348e-05,
      "loss": 0.4426,
      "step": 411
    },
    {
      "epoch": 0.2597218391710335,
      "grad_norm": 0.1394892781972885,
      "learning_rate": 4.3393345347448964e-05,
      "loss": 0.4675,
      "step": 412
    },
    {
      "epoch": 0.26035223198455537,
      "grad_norm": 0.14571578800678253,
      "learning_rate": 4.338498091873066e-05,
      "loss": 0.3836,
      "step": 413
    },
    {
      "epoch": 0.2609826247980773,
      "grad_norm": 0.16842153668403625,
      "learning_rate": 4.3376595583841474e-05,
      "loss": 0.4068,
      "step": 414
    },
    {
      "epoch": 0.2616130176115992,
      "grad_norm": 0.16881997883319855,
      "learning_rate": 4.336818935117525e-05,
      "loss": 0.4047,
      "step": 415
    },
    {
      "epoch": 0.26224341042512117,
      "grad_norm": 0.15377576649188995,
      "learning_rate": 4.335976222914675e-05,
      "loss": 0.4123,
      "step": 416
    },
    {
      "epoch": 0.26287380323864307,
      "grad_norm": 0.1499597281217575,
      "learning_rate": 4.3351314226191656e-05,
      "loss": 0.4568,
      "step": 417
    },
    {
      "epoch": 0.263504196052165,
      "grad_norm": 0.13230498135089874,
      "learning_rate": 4.334284535076653e-05,
      "loss": 0.3797,
      "step": 418
    },
    {
      "epoch": 0.2641345888656869,
      "grad_norm": 0.14309710264205933,
      "learning_rate": 4.333435561134885e-05,
      "loss": 0.458,
      "step": 419
    },
    {
      "epoch": 0.26476498167920887,
      "grad_norm": 0.18320833146572113,
      "learning_rate": 4.332584501643695e-05,
      "loss": 0.4548,
      "step": 420
    },
    {
      "epoch": 0.26539537449273076,
      "grad_norm": 0.17113445699214935,
      "learning_rate": 4.331731357455009e-05,
      "loss": 0.4286,
      "step": 421
    },
    {
      "epoch": 0.2660257673062527,
      "grad_norm": 0.1291307806968689,
      "learning_rate": 4.3308761294228334e-05,
      "loss": 0.3992,
      "step": 422
    },
    {
      "epoch": 0.2666561601197746,
      "grad_norm": 0.13772858679294586,
      "learning_rate": 4.330018818403266e-05,
      "loss": 0.4343,
      "step": 423
    },
    {
      "epoch": 0.26728655293329656,
      "grad_norm": 0.16687078773975372,
      "learning_rate": 4.329159425254487e-05,
      "loss": 0.4587,
      "step": 424
    },
    {
      "epoch": 0.2679169457468185,
      "grad_norm": 0.13438096642494202,
      "learning_rate": 4.328297950836762e-05,
      "loss": 0.355,
      "step": 425
    },
    {
      "epoch": 0.2685473385603404,
      "grad_norm": 0.1611950397491455,
      "learning_rate": 4.3274343960124407e-05,
      "loss": 0.3935,
      "step": 426
    },
    {
      "epoch": 0.26917773137386236,
      "grad_norm": 0.143065944314003,
      "learning_rate": 4.3265687616459506e-05,
      "loss": 0.4134,
      "step": 427
    },
    {
      "epoch": 0.26980812418738426,
      "grad_norm": 0.14825253188610077,
      "learning_rate": 4.325701048603808e-05,
      "loss": 0.4583,
      "step": 428
    },
    {
      "epoch": 0.2704385170009062,
      "grad_norm": 0.17021827399730682,
      "learning_rate": 4.324831257754603e-05,
      "loss": 0.5149,
      "step": 429
    },
    {
      "epoch": 0.2710689098144281,
      "grad_norm": 0.1331031769514084,
      "learning_rate": 4.3239593899690125e-05,
      "loss": 0.3283,
      "step": 430
    },
    {
      "epoch": 0.27169930262795006,
      "grad_norm": 0.1685783565044403,
      "learning_rate": 4.323085446119785e-05,
      "loss": 0.52,
      "step": 431
    },
    {
      "epoch": 0.27232969544147195,
      "grad_norm": 0.15166236460208893,
      "learning_rate": 4.3222094270817546e-05,
      "loss": 0.4498,
      "step": 432
    },
    {
      "epoch": 0.2729600882549939,
      "grad_norm": 0.15902966260910034,
      "learning_rate": 4.321331333731828e-05,
      "loss": 0.4609,
      "step": 433
    },
    {
      "epoch": 0.2735904810685158,
      "grad_norm": 0.17017829418182373,
      "learning_rate": 4.3204511669489874e-05,
      "loss": 0.4214,
      "step": 434
    },
    {
      "epoch": 0.27422087388203775,
      "grad_norm": 0.163391575217247,
      "learning_rate": 4.319568927614295e-05,
      "loss": 0.4652,
      "step": 435
    },
    {
      "epoch": 0.27485126669555965,
      "grad_norm": 0.16209104657173157,
      "learning_rate": 4.318684616610885e-05,
      "loss": 0.4726,
      "step": 436
    },
    {
      "epoch": 0.2754816595090816,
      "grad_norm": 0.16812391579151154,
      "learning_rate": 4.3177982348239656e-05,
      "loss": 0.4635,
      "step": 437
    },
    {
      "epoch": 0.2761120523226035,
      "grad_norm": 0.16276055574417114,
      "learning_rate": 4.316909783140816e-05,
      "loss": 0.4833,
      "step": 438
    },
    {
      "epoch": 0.27674244513612545,
      "grad_norm": 0.1471432000398636,
      "learning_rate": 4.316019262450792e-05,
      "loss": 0.4585,
      "step": 439
    },
    {
      "epoch": 0.2773728379496474,
      "grad_norm": 0.1366051733493805,
      "learning_rate": 4.315126673645316e-05,
      "loss": 0.3718,
      "step": 440
    },
    {
      "epoch": 0.2780032307631693,
      "grad_norm": 0.14559337496757507,
      "learning_rate": 4.314232017617883e-05,
      "loss": 0.4245,
      "step": 441
    },
    {
      "epoch": 0.27863362357669125,
      "grad_norm": 0.14588779211044312,
      "learning_rate": 4.313335295264058e-05,
      "loss": 0.4753,
      "step": 442
    },
    {
      "epoch": 0.27926401639021314,
      "grad_norm": 0.16295523941516876,
      "learning_rate": 4.31243650748147e-05,
      "loss": 0.4217,
      "step": 443
    },
    {
      "epoch": 0.2798944092037351,
      "grad_norm": 0.1639915555715561,
      "learning_rate": 4.311535655169822e-05,
      "loss": 0.3766,
      "step": 444
    },
    {
      "epoch": 0.280524802017257,
      "grad_norm": 0.140822172164917,
      "learning_rate": 4.310632739230879e-05,
      "loss": 0.3774,
      "step": 445
    },
    {
      "epoch": 0.28115519483077894,
      "grad_norm": 0.14565636217594147,
      "learning_rate": 4.309727760568473e-05,
      "loss": 0.4208,
      "step": 446
    },
    {
      "epoch": 0.28178558764430084,
      "grad_norm": 0.14595231413841248,
      "learning_rate": 4.3088207200884995e-05,
      "loss": 0.4299,
      "step": 447
    },
    {
      "epoch": 0.2824159804578228,
      "grad_norm": 0.1536986231803894,
      "learning_rate": 4.307911618698922e-05,
      "loss": 0.4555,
      "step": 448
    },
    {
      "epoch": 0.2830463732713447,
      "grad_norm": 0.12186324596405029,
      "learning_rate": 4.307000457309763e-05,
      "loss": 0.4347,
      "step": 449
    },
    {
      "epoch": 0.28367676608486664,
      "grad_norm": 0.1660432517528534,
      "learning_rate": 4.3060872368331074e-05,
      "loss": 0.4679,
      "step": 450
    },
    {
      "epoch": 0.28430715889838853,
      "grad_norm": 0.15520548820495605,
      "learning_rate": 4.3051719581831036e-05,
      "loss": 0.4418,
      "step": 451
    },
    {
      "epoch": 0.2849375517119105,
      "grad_norm": 0.14485125243663788,
      "learning_rate": 4.304254622275958e-05,
      "loss": 0.4048,
      "step": 452
    },
    {
      "epoch": 0.28556794452543244,
      "grad_norm": 0.15754371881484985,
      "learning_rate": 4.3033352300299386e-05,
      "loss": 0.5319,
      "step": 453
    },
    {
      "epoch": 0.28619833733895433,
      "grad_norm": 0.17553724348545074,
      "learning_rate": 4.302413782365369e-05,
      "loss": 0.4475,
      "step": 454
    },
    {
      "epoch": 0.2868287301524763,
      "grad_norm": 0.1662897914648056,
      "learning_rate": 4.3014902802046315e-05,
      "loss": 0.4004,
      "step": 455
    },
    {
      "epoch": 0.2874591229659982,
      "grad_norm": 0.15161022543907166,
      "learning_rate": 4.300564724472167e-05,
      "loss": 0.4386,
      "step": 456
    },
    {
      "epoch": 0.28808951577952013,
      "grad_norm": 0.1543479859828949,
      "learning_rate": 4.2996371160944684e-05,
      "loss": 0.5125,
      "step": 457
    },
    {
      "epoch": 0.28871990859304203,
      "grad_norm": 0.14676740765571594,
      "learning_rate": 4.298707456000087e-05,
      "loss": 0.4285,
      "step": 458
    },
    {
      "epoch": 0.289350301406564,
      "grad_norm": 0.14895333349704742,
      "learning_rate": 4.297775745119625e-05,
      "loss": 0.4448,
      "step": 459
    },
    {
      "epoch": 0.2899806942200859,
      "grad_norm": 0.12872503697872162,
      "learning_rate": 4.296841984385739e-05,
      "loss": 0.3532,
      "step": 460
    },
    {
      "epoch": 0.29061108703360783,
      "grad_norm": 0.15547753870487213,
      "learning_rate": 4.295906174733137e-05,
      "loss": 0.4387,
      "step": 461
    },
    {
      "epoch": 0.2912414798471297,
      "grad_norm": 0.12598636746406555,
      "learning_rate": 4.294968317098578e-05,
      "loss": 0.3872,
      "step": 462
    },
    {
      "epoch": 0.2918718726606517,
      "grad_norm": 0.1377699226140976,
      "learning_rate": 4.2940284124208704e-05,
      "loss": 0.3701,
      "step": 463
    },
    {
      "epoch": 0.2925022654741736,
      "grad_norm": 0.1805099993944168,
      "learning_rate": 4.293086461640875e-05,
      "loss": 0.488,
      "step": 464
    },
    {
      "epoch": 0.2931326582876955,
      "grad_norm": 0.12898395955562592,
      "learning_rate": 4.2921424657014954e-05,
      "loss": 0.319,
      "step": 465
    },
    {
      "epoch": 0.2937630511012174,
      "grad_norm": 0.16145475208759308,
      "learning_rate": 4.291196425547687e-05,
      "loss": 0.4519,
      "step": 466
    },
    {
      "epoch": 0.29439344391473937,
      "grad_norm": 0.1678912341594696,
      "learning_rate": 4.290248342126451e-05,
      "loss": 0.4225,
      "step": 467
    },
    {
      "epoch": 0.2950238367282613,
      "grad_norm": 0.1473826915025711,
      "learning_rate": 4.289298216386829e-05,
      "loss": 0.3601,
      "step": 468
    },
    {
      "epoch": 0.2956542295417832,
      "grad_norm": 0.1619638204574585,
      "learning_rate": 4.288346049279914e-05,
      "loss": 0.4531,
      "step": 469
    },
    {
      "epoch": 0.29628462235530517,
      "grad_norm": 0.17025795578956604,
      "learning_rate": 4.287391841758838e-05,
      "loss": 0.4791,
      "step": 470
    },
    {
      "epoch": 0.29691501516882707,
      "grad_norm": 0.15568993985652924,
      "learning_rate": 4.286435594778776e-05,
      "loss": 0.4427,
      "step": 471
    },
    {
      "epoch": 0.297545407982349,
      "grad_norm": 0.1638522893190384,
      "learning_rate": 4.285477309296948e-05,
      "loss": 0.4029,
      "step": 472
    },
    {
      "epoch": 0.2981758007958709,
      "grad_norm": 0.1754094958305359,
      "learning_rate": 4.284516986272607e-05,
      "loss": 0.3822,
      "step": 473
    },
    {
      "epoch": 0.29880619360939287,
      "grad_norm": 0.15131361782550812,
      "learning_rate": 4.283554626667055e-05,
      "loss": 0.3986,
      "step": 474
    },
    {
      "epoch": 0.29943658642291476,
      "grad_norm": 0.1398685723543167,
      "learning_rate": 4.282590231443625e-05,
      "loss": 0.4285,
      "step": 475
    },
    {
      "epoch": 0.3000669792364367,
      "grad_norm": 0.15816210210323334,
      "learning_rate": 4.281623801567693e-05,
      "loss": 0.4436,
      "step": 476
    },
    {
      "epoch": 0.3006973720499586,
      "grad_norm": 0.16846077144145966,
      "learning_rate": 4.2806553380066674e-05,
      "loss": 0.3985,
      "step": 477
    },
    {
      "epoch": 0.30132776486348056,
      "grad_norm": 0.17154532670974731,
      "learning_rate": 4.279684841729996e-05,
      "loss": 0.4526,
      "step": 478
    },
    {
      "epoch": 0.30195815767700246,
      "grad_norm": 0.13398754596710205,
      "learning_rate": 4.2787123137091576e-05,
      "loss": 0.3518,
      "step": 479
    },
    {
      "epoch": 0.3025885504905244,
      "grad_norm": 0.14154496788978577,
      "learning_rate": 4.277737754917669e-05,
      "loss": 0.3814,
      "step": 480
    },
    {
      "epoch": 0.3032189433040463,
      "grad_norm": 0.1629718393087387,
      "learning_rate": 4.276761166331077e-05,
      "loss": 0.4,
      "step": 481
    },
    {
      "epoch": 0.30384933611756826,
      "grad_norm": 0.14869852364063263,
      "learning_rate": 4.2757825489269616e-05,
      "loss": 0.3429,
      "step": 482
    },
    {
      "epoch": 0.3044797289310902,
      "grad_norm": 0.15486106276512146,
      "learning_rate": 4.2748019036849325e-05,
      "loss": 0.3757,
      "step": 483
    },
    {
      "epoch": 0.3051101217446121,
      "grad_norm": 0.1632099747657776,
      "learning_rate": 4.273819231586629e-05,
      "loss": 0.3948,
      "step": 484
    },
    {
      "epoch": 0.30574051455813406,
      "grad_norm": 0.17350728809833527,
      "learning_rate": 4.272834533615723e-05,
      "loss": 0.4351,
      "step": 485
    },
    {
      "epoch": 0.30637090737165595,
      "grad_norm": 0.14235197007656097,
      "learning_rate": 4.271847810757908e-05,
      "loss": 0.3961,
      "step": 486
    },
    {
      "epoch": 0.3070013001851779,
      "grad_norm": 0.1834893822669983,
      "learning_rate": 4.270859064000911e-05,
      "loss": 0.4808,
      "step": 487
    },
    {
      "epoch": 0.3076316929986998,
      "grad_norm": 0.13989755511283875,
      "learning_rate": 4.2698682943344804e-05,
      "loss": 0.4272,
      "step": 488
    },
    {
      "epoch": 0.30826208581222175,
      "grad_norm": 0.1976575404405594,
      "learning_rate": 4.268875502750391e-05,
      "loss": 0.469,
      "step": 489
    },
    {
      "epoch": 0.30889247862574365,
      "grad_norm": 0.17637616395950317,
      "learning_rate": 4.267880690242443e-05,
      "loss": 0.4712,
      "step": 490
    },
    {
      "epoch": 0.3095228714392656,
      "grad_norm": 0.16208115220069885,
      "learning_rate": 4.266883857806456e-05,
      "loss": 0.4406,
      "step": 491
    },
    {
      "epoch": 0.3101532642527875,
      "grad_norm": 0.14250457286834717,
      "learning_rate": 4.2658850064402763e-05,
      "loss": 0.4061,
      "step": 492
    },
    {
      "epoch": 0.31078365706630945,
      "grad_norm": 0.15330681204795837,
      "learning_rate": 4.264884137143767e-05,
      "loss": 0.3706,
      "step": 493
    },
    {
      "epoch": 0.31141404987983135,
      "grad_norm": 0.13295991718769073,
      "learning_rate": 4.263881250918814e-05,
      "loss": 0.4272,
      "step": 494
    },
    {
      "epoch": 0.3120444426933533,
      "grad_norm": 0.143403097987175,
      "learning_rate": 4.2628763487693205e-05,
      "loss": 0.3909,
      "step": 495
    },
    {
      "epoch": 0.31267483550687525,
      "grad_norm": 0.17583061754703522,
      "learning_rate": 4.261869431701208e-05,
      "loss": 0.4029,
      "step": 496
    },
    {
      "epoch": 0.31330522832039714,
      "grad_norm": 0.134292870759964,
      "learning_rate": 4.260860500722415e-05,
      "loss": 0.3743,
      "step": 497
    },
    {
      "epoch": 0.3139356211339191,
      "grad_norm": 0.14064282178878784,
      "learning_rate": 4.259849556842898e-05,
      "loss": 0.3862,
      "step": 498
    },
    {
      "epoch": 0.314566013947441,
      "grad_norm": 0.2020834982395172,
      "learning_rate": 4.258836601074624e-05,
      "loss": 0.4301,
      "step": 499
    },
    {
      "epoch": 0.31519640676096294,
      "grad_norm": 0.17847569286823273,
      "learning_rate": 4.257821634431578e-05,
      "loss": 0.4587,
      "step": 500
    },
    {
      "epoch": 0.31582679957448484,
      "grad_norm": 0.16131030023097992,
      "learning_rate": 4.256804657929756e-05,
      "loss": 0.4235,
      "step": 501
    },
    {
      "epoch": 0.3164571923880068,
      "grad_norm": 0.11063003540039062,
      "learning_rate": 4.255785672587166e-05,
      "loss": 0.3409,
      "step": 502
    },
    {
      "epoch": 0.3170875852015287,
      "grad_norm": 0.16887235641479492,
      "learning_rate": 4.2547646794238276e-05,
      "loss": 0.4017,
      "step": 503
    },
    {
      "epoch": 0.31771797801505064,
      "grad_norm": 0.1624932438135147,
      "learning_rate": 4.253741679461771e-05,
      "loss": 0.4041,
      "step": 504
    },
    {
      "epoch": 0.31834837082857254,
      "grad_norm": 0.17934207618236542,
      "learning_rate": 4.2527166737250324e-05,
      "loss": 0.4966,
      "step": 505
    },
    {
      "epoch": 0.3189787636420945,
      "grad_norm": 0.15029865503311157,
      "learning_rate": 4.2516896632396576e-05,
      "loss": 0.4555,
      "step": 506
    },
    {
      "epoch": 0.3196091564556164,
      "grad_norm": 0.14951272308826447,
      "learning_rate": 4.2506606490337006e-05,
      "loss": 0.4224,
      "step": 507
    },
    {
      "epoch": 0.32023954926913833,
      "grad_norm": 0.15957953035831451,
      "learning_rate": 4.249629632137218e-05,
      "loss": 0.4615,
      "step": 508
    },
    {
      "epoch": 0.32086994208266023,
      "grad_norm": 0.18410411477088928,
      "learning_rate": 4.2485966135822736e-05,
      "loss": 0.522,
      "step": 509
    },
    {
      "epoch": 0.3215003348961822,
      "grad_norm": 0.13453824818134308,
      "learning_rate": 4.247561594402934e-05,
      "loss": 0.3849,
      "step": 510
    },
    {
      "epoch": 0.32213072770970413,
      "grad_norm": 0.1895299255847931,
      "learning_rate": 4.246524575635269e-05,
      "loss": 0.4298,
      "step": 511
    },
    {
      "epoch": 0.32276112052322603,
      "grad_norm": 0.17183034121990204,
      "learning_rate": 4.245485558317348e-05,
      "loss": 0.4133,
      "step": 512
    },
    {
      "epoch": 0.323391513336748,
      "grad_norm": 0.1502702534198761,
      "learning_rate": 4.244444543489244e-05,
      "loss": 0.4127,
      "step": 513
    },
    {
      "epoch": 0.3240219061502699,
      "grad_norm": 0.11526153981685638,
      "learning_rate": 4.243401532193027e-05,
      "loss": 0.3994,
      "step": 514
    },
    {
      "epoch": 0.32465229896379183,
      "grad_norm": 0.13428835570812225,
      "learning_rate": 4.242356525472767e-05,
      "loss": 0.4021,
      "step": 515
    },
    {
      "epoch": 0.3252826917773137,
      "grad_norm": 0.16996537148952484,
      "learning_rate": 4.241309524374531e-05,
      "loss": 0.4596,
      "step": 516
    },
    {
      "epoch": 0.3259130845908357,
      "grad_norm": 0.19075356423854828,
      "learning_rate": 4.240260529946383e-05,
      "loss": 0.5052,
      "step": 517
    },
    {
      "epoch": 0.3265434774043576,
      "grad_norm": 0.16824814677238464,
      "learning_rate": 4.239209543238381e-05,
      "loss": 0.4298,
      "step": 518
    },
    {
      "epoch": 0.3271738702178795,
      "grad_norm": 0.1488647311925888,
      "learning_rate": 4.238156565302578e-05,
      "loss": 0.3637,
      "step": 519
    },
    {
      "epoch": 0.3278042630314014,
      "grad_norm": 0.14669464528560638,
      "learning_rate": 4.2371015971930204e-05,
      "loss": 0.3611,
      "step": 520
    },
    {
      "epoch": 0.3284346558449234,
      "grad_norm": 0.14535008370876312,
      "learning_rate": 4.236044639965747e-05,
      "loss": 0.4349,
      "step": 521
    },
    {
      "epoch": 0.32906504865844527,
      "grad_norm": 0.14576545357704163,
      "learning_rate": 4.234985694678787e-05,
      "loss": 0.3882,
      "step": 522
    },
    {
      "epoch": 0.3296954414719672,
      "grad_norm": 0.16440042853355408,
      "learning_rate": 4.23392476239216e-05,
      "loss": 0.4597,
      "step": 523
    },
    {
      "epoch": 0.3303258342854892,
      "grad_norm": 0.15812115371227264,
      "learning_rate": 4.232861844167875e-05,
      "loss": 0.4408,
      "step": 524
    },
    {
      "epoch": 0.33095622709901107,
      "grad_norm": 0.15020525455474854,
      "learning_rate": 4.231796941069929e-05,
      "loss": 0.3864,
      "step": 525
    },
    {
      "epoch": 0.331586619912533,
      "grad_norm": 0.14310987293720245,
      "learning_rate": 4.230730054164305e-05,
      "loss": 0.4158,
      "step": 526
    },
    {
      "epoch": 0.3322170127260549,
      "grad_norm": 0.1488959640264511,
      "learning_rate": 4.229661184518973e-05,
      "loss": 0.3834,
      "step": 527
    },
    {
      "epoch": 0.33284740553957687,
      "grad_norm": 0.15532805025577545,
      "learning_rate": 4.228590333203886e-05,
      "loss": 0.383,
      "step": 528
    },
    {
      "epoch": 0.33347779835309876,
      "grad_norm": 0.16951856017112732,
      "learning_rate": 4.227517501290983e-05,
      "loss": 0.4878,
      "step": 529
    },
    {
      "epoch": 0.3341081911666207,
      "grad_norm": 0.1473962813615799,
      "learning_rate": 4.226442689854184e-05,
      "loss": 0.4814,
      "step": 530
    },
    {
      "epoch": 0.3347385839801426,
      "grad_norm": 0.15749426186084747,
      "learning_rate": 4.225365899969391e-05,
      "loss": 0.4194,
      "step": 531
    },
    {
      "epoch": 0.33536897679366456,
      "grad_norm": 0.16457794606685638,
      "learning_rate": 4.2242871327144855e-05,
      "loss": 0.4758,
      "step": 532
    },
    {
      "epoch": 0.33599936960718646,
      "grad_norm": 0.19138643145561218,
      "learning_rate": 4.2232063891693305e-05,
      "loss": 0.4677,
      "step": 533
    },
    {
      "epoch": 0.3366297624207084,
      "grad_norm": 0.15384329855442047,
      "learning_rate": 4.222123670415766e-05,
      "loss": 0.4107,
      "step": 534
    },
    {
      "epoch": 0.3372601552342303,
      "grad_norm": 0.1445811688899994,
      "learning_rate": 4.2210389775376076e-05,
      "loss": 0.3981,
      "step": 535
    },
    {
      "epoch": 0.33789054804775226,
      "grad_norm": 0.14559048414230347,
      "learning_rate": 4.219952311620651e-05,
      "loss": 0.4463,
      "step": 536
    },
    {
      "epoch": 0.33852094086127416,
      "grad_norm": 0.1466781497001648,
      "learning_rate": 4.2188636737526634e-05,
      "loss": 0.464,
      "step": 537
    },
    {
      "epoch": 0.3391513336747961,
      "grad_norm": 0.14743733406066895,
      "learning_rate": 4.217773065023388e-05,
      "loss": 0.4083,
      "step": 538
    },
    {
      "epoch": 0.33978172648831806,
      "grad_norm": 0.15406803786754608,
      "learning_rate": 4.2166804865245384e-05,
      "loss": 0.4374,
      "step": 539
    },
    {
      "epoch": 0.34041211930183995,
      "grad_norm": 0.1378432959318161,
      "learning_rate": 4.2155859393498034e-05,
      "loss": 0.3851,
      "step": 540
    },
    {
      "epoch": 0.3410425121153619,
      "grad_norm": 0.15630073845386505,
      "learning_rate": 4.2144894245948404e-05,
      "loss": 0.39,
      "step": 541
    },
    {
      "epoch": 0.3416729049288838,
      "grad_norm": 0.1572435051202774,
      "learning_rate": 4.213390943357275e-05,
      "loss": 0.4445,
      "step": 542
    },
    {
      "epoch": 0.34230329774240575,
      "grad_norm": 0.16819582879543304,
      "learning_rate": 4.2122904967367056e-05,
      "loss": 0.4271,
      "step": 543
    },
    {
      "epoch": 0.34293369055592765,
      "grad_norm": 0.16458523273468018,
      "learning_rate": 4.211188085834694e-05,
      "loss": 0.4699,
      "step": 544
    },
    {
      "epoch": 0.3435640833694496,
      "grad_norm": 0.1598554253578186,
      "learning_rate": 4.210083711754769e-05,
      "loss": 0.4468,
      "step": 545
    },
    {
      "epoch": 0.3441944761829715,
      "grad_norm": 0.18264490365982056,
      "learning_rate": 4.208977375602426e-05,
      "loss": 0.3785,
      "step": 546
    },
    {
      "epoch": 0.34482486899649345,
      "grad_norm": 0.1308012455701828,
      "learning_rate": 4.2078690784851236e-05,
      "loss": 0.3724,
      "step": 547
    },
    {
      "epoch": 0.34545526181001535,
      "grad_norm": 0.14296194911003113,
      "learning_rate": 4.206758821512283e-05,
      "loss": 0.389,
      "step": 548
    },
    {
      "epoch": 0.3460856546235373,
      "grad_norm": 0.14228825271129608,
      "learning_rate": 4.205646605795289e-05,
      "loss": 0.3776,
      "step": 549
    },
    {
      "epoch": 0.3467160474370592,
      "grad_norm": 0.1453482061624527,
      "learning_rate": 4.204532432447484e-05,
      "loss": 0.4125,
      "step": 550
    },
    {
      "epoch": 0.34734644025058115,
      "grad_norm": 0.1499478816986084,
      "learning_rate": 4.203416302584172e-05,
      "loss": 0.378,
      "step": 551
    },
    {
      "epoch": 0.34797683306410304,
      "grad_norm": 0.16379188001155853,
      "learning_rate": 4.202298217322616e-05,
      "loss": 0.4554,
      "step": 552
    },
    {
      "epoch": 0.348607225877625,
      "grad_norm": 0.16106747090816498,
      "learning_rate": 4.201178177782036e-05,
      "loss": 0.3583,
      "step": 553
    },
    {
      "epoch": 0.34923761869114694,
      "grad_norm": 0.13125036656856537,
      "learning_rate": 4.2000561850836056e-05,
      "loss": 0.403,
      "step": 554
    },
    {
      "epoch": 0.34986801150466884,
      "grad_norm": 0.15825971961021423,
      "learning_rate": 4.19893224035046e-05,
      "loss": 0.3882,
      "step": 555
    },
    {
      "epoch": 0.3504984043181908,
      "grad_norm": 0.12176766246557236,
      "learning_rate": 4.19780634470768e-05,
      "loss": 0.3691,
      "step": 556
    },
    {
      "epoch": 0.3511287971317127,
      "grad_norm": 0.14360551536083221,
      "learning_rate": 4.196678499282307e-05,
      "loss": 0.4186,
      "step": 557
    },
    {
      "epoch": 0.35175918994523464,
      "grad_norm": 0.15045621991157532,
      "learning_rate": 4.195548705203328e-05,
      "loss": 0.4297,
      "step": 558
    },
    {
      "epoch": 0.35238958275875654,
      "grad_norm": 0.15988411009311676,
      "learning_rate": 4.194416963601685e-05,
      "loss": 0.4631,
      "step": 559
    },
    {
      "epoch": 0.3530199755722785,
      "grad_norm": 0.15470652282238007,
      "learning_rate": 4.193283275610268e-05,
      "loss": 0.3725,
      "step": 560
    },
    {
      "epoch": 0.3536503683858004,
      "grad_norm": 0.15270668268203735,
      "learning_rate": 4.192147642363915e-05,
      "loss": 0.4065,
      "step": 561
    },
    {
      "epoch": 0.35428076119932234,
      "grad_norm": 0.14180682599544525,
      "learning_rate": 4.19101006499941e-05,
      "loss": 0.4172,
      "step": 562
    },
    {
      "epoch": 0.35491115401284423,
      "grad_norm": 0.1411629319190979,
      "learning_rate": 4.189870544655487e-05,
      "loss": 0.394,
      "step": 563
    },
    {
      "epoch": 0.3555415468263662,
      "grad_norm": 0.1397046446800232,
      "learning_rate": 4.1887290824728205e-05,
      "loss": 0.3998,
      "step": 564
    },
    {
      "epoch": 0.3561719396398881,
      "grad_norm": 0.11284170299768448,
      "learning_rate": 4.1875856795940316e-05,
      "loss": 0.3202,
      "step": 565
    },
    {
      "epoch": 0.35680233245341003,
      "grad_norm": 0.13387620449066162,
      "learning_rate": 4.186440337163684e-05,
      "loss": 0.4024,
      "step": 566
    },
    {
      "epoch": 0.357432725266932,
      "grad_norm": 0.12860415875911713,
      "learning_rate": 4.18529305632828e-05,
      "loss": 0.4669,
      "step": 567
    },
    {
      "epoch": 0.3580631180804539,
      "grad_norm": 0.1481974571943283,
      "learning_rate": 4.184143838236267e-05,
      "loss": 0.4428,
      "step": 568
    },
    {
      "epoch": 0.35869351089397583,
      "grad_norm": 0.15754854679107666,
      "learning_rate": 4.182992684038028e-05,
      "loss": 0.4612,
      "step": 569
    },
    {
      "epoch": 0.3593239037074977,
      "grad_norm": 0.17479227483272552,
      "learning_rate": 4.181839594885884e-05,
      "loss": 0.47,
      "step": 570
    },
    {
      "epoch": 0.3599542965210197,
      "grad_norm": 0.15785165131092072,
      "learning_rate": 4.180684571934097e-05,
      "loss": 0.4033,
      "step": 571
    },
    {
      "epoch": 0.3605846893345416,
      "grad_norm": 0.16270500421524048,
      "learning_rate": 4.179527616338859e-05,
      "loss": 0.4469,
      "step": 572
    },
    {
      "epoch": 0.3612150821480635,
      "grad_norm": 0.14534880220890045,
      "learning_rate": 4.1783687292583004e-05,
      "loss": 0.3873,
      "step": 573
    },
    {
      "epoch": 0.3618454749615854,
      "grad_norm": 0.12993444502353668,
      "learning_rate": 4.177207911852485e-05,
      "loss": 0.3789,
      "step": 574
    },
    {
      "epoch": 0.3624758677751074,
      "grad_norm": 0.12628543376922607,
      "learning_rate": 4.176045165283407e-05,
      "loss": 0.3762,
      "step": 575
    },
    {
      "epoch": 0.36310626058862927,
      "grad_norm": 0.15444335341453552,
      "learning_rate": 4.1748804907149925e-05,
      "loss": 0.3623,
      "step": 576
    },
    {
      "epoch": 0.3637366534021512,
      "grad_norm": 0.17257718741893768,
      "learning_rate": 4.173713889313099e-05,
      "loss": 0.4264,
      "step": 577
    },
    {
      "epoch": 0.3643670462156731,
      "grad_norm": 0.1389797180891037,
      "learning_rate": 4.1725453622455094e-05,
      "loss": 0.3736,
      "step": 578
    },
    {
      "epoch": 0.36499743902919507,
      "grad_norm": 0.14798113703727722,
      "learning_rate": 4.171374910681938e-05,
      "loss": 0.4604,
      "step": 579
    },
    {
      "epoch": 0.36562783184271697,
      "grad_norm": 0.16598446667194366,
      "learning_rate": 4.170202535794024e-05,
      "loss": 0.4466,
      "step": 580
    },
    {
      "epoch": 0.3662582246562389,
      "grad_norm": 0.14121782779693604,
      "learning_rate": 4.169028238755331e-05,
      "loss": 0.3377,
      "step": 581
    },
    {
      "epoch": 0.36688861746976087,
      "grad_norm": 0.13968297839164734,
      "learning_rate": 4.1678520207413485e-05,
      "loss": 0.4261,
      "step": 582
    },
    {
      "epoch": 0.36751901028328277,
      "grad_norm": 0.14780119061470032,
      "learning_rate": 4.166673882929487e-05,
      "loss": 0.4008,
      "step": 583
    },
    {
      "epoch": 0.3681494030968047,
      "grad_norm": 0.1448272168636322,
      "learning_rate": 4.1654938264990785e-05,
      "loss": 0.399,
      "step": 584
    },
    {
      "epoch": 0.3687797959103266,
      "grad_norm": 0.1570119857788086,
      "learning_rate": 4.16431185263138e-05,
      "loss": 0.4486,
      "step": 585
    },
    {
      "epoch": 0.36941018872384856,
      "grad_norm": 0.13365842401981354,
      "learning_rate": 4.1631279625095616e-05,
      "loss": 0.4559,
      "step": 586
    },
    {
      "epoch": 0.37004058153737046,
      "grad_norm": 0.142498180270195,
      "learning_rate": 4.161942157318716e-05,
      "loss": 0.3981,
      "step": 587
    },
    {
      "epoch": 0.3706709743508924,
      "grad_norm": 0.15279634296894073,
      "learning_rate": 4.160754438245851e-05,
      "loss": 0.3982,
      "step": 588
    },
    {
      "epoch": 0.3713013671644143,
      "grad_norm": 0.1295880526304245,
      "learning_rate": 4.159564806479891e-05,
      "loss": 0.3687,
      "step": 589
    },
    {
      "epoch": 0.37193175997793626,
      "grad_norm": 0.18762393295764923,
      "learning_rate": 4.1583732632116755e-05,
      "loss": 0.4631,
      "step": 590
    },
    {
      "epoch": 0.37256215279145816,
      "grad_norm": 0.12456995248794556,
      "learning_rate": 4.1571798096339544e-05,
      "loss": 0.4014,
      "step": 591
    },
    {
      "epoch": 0.3731925456049801,
      "grad_norm": 0.1376989632844925,
      "learning_rate": 4.1559844469413945e-05,
      "loss": 0.3587,
      "step": 592
    },
    {
      "epoch": 0.373822938418502,
      "grad_norm": 0.16048815846443176,
      "learning_rate": 4.154787176330571e-05,
      "loss": 0.4018,
      "step": 593
    },
    {
      "epoch": 0.37445333123202396,
      "grad_norm": 0.13161379098892212,
      "learning_rate": 4.153587998999969e-05,
      "loss": 0.4113,
      "step": 594
    },
    {
      "epoch": 0.3750837240455459,
      "grad_norm": 0.1387777328491211,
      "learning_rate": 4.152386916149981e-05,
      "loss": 0.3606,
      "step": 595
    },
    {
      "epoch": 0.3757141168590678,
      "grad_norm": 0.17377012968063354,
      "learning_rate": 4.1511839289829106e-05,
      "loss": 0.4833,
      "step": 596
    },
    {
      "epoch": 0.37634450967258976,
      "grad_norm": 0.1923179179430008,
      "learning_rate": 4.1499790387029646e-05,
      "loss": 0.4308,
      "step": 597
    },
    {
      "epoch": 0.37697490248611165,
      "grad_norm": 0.15764404833316803,
      "learning_rate": 4.1487722465162555e-05,
      "loss": 0.4458,
      "step": 598
    },
    {
      "epoch": 0.3776052952996336,
      "grad_norm": 0.1681838035583496,
      "learning_rate": 4.1475635536308e-05,
      "loss": 0.4405,
      "step": 599
    },
    {
      "epoch": 0.3782356881131555,
      "grad_norm": 0.154836967587471,
      "learning_rate": 4.146352961256518e-05,
      "loss": 0.4381,
      "step": 600
    },
    {
      "epoch": 0.3782356881131555,
      "eval_loss": 0.444912314414978,
      "eval_runtime": 222.7209,
      "eval_samples_per_second": 4.49,
      "eval_steps_per_second": 4.49,
      "step": 600
    },
    {
      "epoch": 0.37886608092667745,
      "grad_norm": 0.16694478690624237,
      "learning_rate": 4.14514047060523e-05,
      "loss": 0.4219,
      "step": 601
    },
    {
      "epoch": 0.37949647374019935,
      "grad_norm": 0.17993447184562683,
      "learning_rate": 4.1439260828906556e-05,
      "loss": 0.4688,
      "step": 602
    },
    {
      "epoch": 0.3801268665537213,
      "grad_norm": 0.14644283056259155,
      "learning_rate": 4.142709799328417e-05,
      "loss": 0.4383,
      "step": 603
    },
    {
      "epoch": 0.3807572593672432,
      "grad_norm": 0.1624111384153366,
      "learning_rate": 4.141491621136029e-05,
      "loss": 0.4815,
      "step": 604
    },
    {
      "epoch": 0.38138765218076515,
      "grad_norm": 0.14999568462371826,
      "learning_rate": 4.140271549532908e-05,
      "loss": 0.4006,
      "step": 605
    },
    {
      "epoch": 0.38201804499428704,
      "grad_norm": 0.1663047969341278,
      "learning_rate": 4.139049585740363e-05,
      "loss": 0.4411,
      "step": 606
    },
    {
      "epoch": 0.382648437807809,
      "grad_norm": 0.1374640315771103,
      "learning_rate": 4.137825730981597e-05,
      "loss": 0.3797,
      "step": 607
    },
    {
      "epoch": 0.3832788306213309,
      "grad_norm": 0.16739337146282196,
      "learning_rate": 4.136599986481706e-05,
      "loss": 0.3494,
      "step": 608
    },
    {
      "epoch": 0.38390922343485284,
      "grad_norm": 0.15691962838172913,
      "learning_rate": 4.1353723534676795e-05,
      "loss": 0.3653,
      "step": 609
    },
    {
      "epoch": 0.3845396162483748,
      "grad_norm": 0.14026010036468506,
      "learning_rate": 4.1341428331683954e-05,
      "loss": 0.425,
      "step": 610
    },
    {
      "epoch": 0.3851700090618967,
      "grad_norm": 0.13732515275478363,
      "learning_rate": 4.132911426814621e-05,
      "loss": 0.3926,
      "step": 611
    },
    {
      "epoch": 0.38580040187541864,
      "grad_norm": 0.21322603523731232,
      "learning_rate": 4.1316781356390134e-05,
      "loss": 0.5112,
      "step": 612
    },
    {
      "epoch": 0.38643079468894054,
      "grad_norm": 0.16545470058918,
      "learning_rate": 4.130442960876113e-05,
      "loss": 0.4436,
      "step": 613
    },
    {
      "epoch": 0.3870611875024625,
      "grad_norm": 0.1380622237920761,
      "learning_rate": 4.129205903762351e-05,
      "loss": 0.4221,
      "step": 614
    },
    {
      "epoch": 0.3876915803159844,
      "grad_norm": 0.14784802496433258,
      "learning_rate": 4.127966965536036e-05,
      "loss": 0.3305,
      "step": 615
    },
    {
      "epoch": 0.38832197312950634,
      "grad_norm": 0.1654745489358902,
      "learning_rate": 4.126726147437366e-05,
      "loss": 0.4587,
      "step": 616
    },
    {
      "epoch": 0.38895236594302823,
      "grad_norm": 0.16123948991298676,
      "learning_rate": 4.1254834507084165e-05,
      "loss": 0.4208,
      "step": 617
    },
    {
      "epoch": 0.3895827587565502,
      "grad_norm": 0.1417972445487976,
      "learning_rate": 4.124238876593145e-05,
      "loss": 0.349,
      "step": 618
    },
    {
      "epoch": 0.3902131515700721,
      "grad_norm": 0.13607749342918396,
      "learning_rate": 4.122992426337389e-05,
      "loss": 0.3971,
      "step": 619
    },
    {
      "epoch": 0.39084354438359403,
      "grad_norm": 0.16371870040893555,
      "learning_rate": 4.1217441011888644e-05,
      "loss": 0.4607,
      "step": 620
    },
    {
      "epoch": 0.39147393719711593,
      "grad_norm": 0.15990233421325684,
      "learning_rate": 4.120493902397161e-05,
      "loss": 0.4536,
      "step": 621
    },
    {
      "epoch": 0.3921043300106379,
      "grad_norm": 0.1312706023454666,
      "learning_rate": 4.119241831213747e-05,
      "loss": 0.3943,
      "step": 622
    },
    {
      "epoch": 0.39273472282415983,
      "grad_norm": 0.16973647475242615,
      "learning_rate": 4.117987888891963e-05,
      "loss": 0.4315,
      "step": 623
    },
    {
      "epoch": 0.39336511563768173,
      "grad_norm": 0.1553269475698471,
      "learning_rate": 4.116732076687026e-05,
      "loss": 0.451,
      "step": 624
    },
    {
      "epoch": 0.3939955084512037,
      "grad_norm": 0.13316068053245544,
      "learning_rate": 4.1154743958560195e-05,
      "loss": 0.3701,
      "step": 625
    },
    {
      "epoch": 0.3946259012647256,
      "grad_norm": 0.142917200922966,
      "learning_rate": 4.114214847657902e-05,
      "loss": 0.4027,
      "step": 626
    },
    {
      "epoch": 0.3952562940782475,
      "grad_norm": 0.16515448689460754,
      "learning_rate": 4.112953433353499e-05,
      "loss": 0.4192,
      "step": 627
    },
    {
      "epoch": 0.3958866868917694,
      "grad_norm": 0.1329452246427536,
      "learning_rate": 4.1116901542055044e-05,
      "loss": 0.3855,
      "step": 628
    },
    {
      "epoch": 0.3965170797052914,
      "grad_norm": 0.19352348148822784,
      "learning_rate": 4.1104250114784805e-05,
      "loss": 0.4795,
      "step": 629
    },
    {
      "epoch": 0.39714747251881327,
      "grad_norm": 0.1591757833957672,
      "learning_rate": 4.109158006438853e-05,
      "loss": 0.4383,
      "step": 630
    },
    {
      "epoch": 0.3977778653323352,
      "grad_norm": 0.15093408524990082,
      "learning_rate": 4.107889140354911e-05,
      "loss": 0.428,
      "step": 631
    },
    {
      "epoch": 0.3984082581458571,
      "grad_norm": 0.13541729748249054,
      "learning_rate": 4.1066184144968096e-05,
      "loss": 0.3743,
      "step": 632
    },
    {
      "epoch": 0.39903865095937907,
      "grad_norm": 0.15193648636341095,
      "learning_rate": 4.105345830136563e-05,
      "loss": 0.4469,
      "step": 633
    },
    {
      "epoch": 0.39966904377290097,
      "grad_norm": 0.1544993370771408,
      "learning_rate": 4.104071388548048e-05,
      "loss": 0.3963,
      "step": 634
    },
    {
      "epoch": 0.4002994365864229,
      "grad_norm": 0.17132483422756195,
      "learning_rate": 4.102795091006998e-05,
      "loss": 0.4785,
      "step": 635
    },
    {
      "epoch": 0.4009298293999448,
      "grad_norm": 0.18319329619407654,
      "learning_rate": 4.1015169387910056e-05,
      "loss": 0.398,
      "step": 636
    },
    {
      "epoch": 0.40156022221346677,
      "grad_norm": 0.1142900213599205,
      "learning_rate": 4.100236933179521e-05,
      "loss": 0.3647,
      "step": 637
    },
    {
      "epoch": 0.4021906150269887,
      "grad_norm": 0.1454845666885376,
      "learning_rate": 4.098955075453847e-05,
      "loss": 0.4765,
      "step": 638
    },
    {
      "epoch": 0.4028210078405106,
      "grad_norm": 0.15059219300746918,
      "learning_rate": 4.097671366897143e-05,
      "loss": 0.382,
      "step": 639
    },
    {
      "epoch": 0.40345140065403257,
      "grad_norm": 0.15379232168197632,
      "learning_rate": 4.0963858087944186e-05,
      "loss": 0.3915,
      "step": 640
    },
    {
      "epoch": 0.40408179346755446,
      "grad_norm": 0.14436917006969452,
      "learning_rate": 4.095098402432538e-05,
      "loss": 0.4359,
      "step": 641
    },
    {
      "epoch": 0.4047121862810764,
      "grad_norm": 0.16968920826911926,
      "learning_rate": 4.0938091491002125e-05,
      "loss": 0.3922,
      "step": 642
    },
    {
      "epoch": 0.4053425790945983,
      "grad_norm": 0.14354845881462097,
      "learning_rate": 4.092518050088005e-05,
      "loss": 0.4116,
      "step": 643
    },
    {
      "epoch": 0.40597297190812026,
      "grad_norm": 0.13624794781208038,
      "learning_rate": 4.091225106688322e-05,
      "loss": 0.4232,
      "step": 644
    },
    {
      "epoch": 0.40660336472164216,
      "grad_norm": 0.16531537473201752,
      "learning_rate": 4.089930320195421e-05,
      "loss": 0.4418,
      "step": 645
    },
    {
      "epoch": 0.4072337575351641,
      "grad_norm": 0.1489722579717636,
      "learning_rate": 4.088633691905402e-05,
      "loss": 0.41,
      "step": 646
    },
    {
      "epoch": 0.407864150348686,
      "grad_norm": 0.140624538064003,
      "learning_rate": 4.0873352231162064e-05,
      "loss": 0.4101,
      "step": 647
    },
    {
      "epoch": 0.40849454316220796,
      "grad_norm": 0.14307831227779388,
      "learning_rate": 4.0860349151276235e-05,
      "loss": 0.3795,
      "step": 648
    },
    {
      "epoch": 0.40912493597572985,
      "grad_norm": 0.1908441036939621,
      "learning_rate": 4.0847327692412787e-05,
      "loss": 0.4538,
      "step": 649
    },
    {
      "epoch": 0.4097553287892518,
      "grad_norm": 0.1263931393623352,
      "learning_rate": 4.08342878676064e-05,
      "loss": 0.3504,
      "step": 650
    },
    {
      "epoch": 0.4103857216027737,
      "grad_norm": 0.14092887938022614,
      "learning_rate": 4.0821229689910125e-05,
      "loss": 0.4195,
      "step": 651
    },
    {
      "epoch": 0.41101611441629565,
      "grad_norm": 0.1607782244682312,
      "learning_rate": 4.0808153172395394e-05,
      "loss": 0.3918,
      "step": 652
    },
    {
      "epoch": 0.4116465072298176,
      "grad_norm": 0.12833623588085175,
      "learning_rate": 4.079505832815198e-05,
      "loss": 0.3821,
      "step": 653
    },
    {
      "epoch": 0.4122769000433395,
      "grad_norm": 0.14452293515205383,
      "learning_rate": 4.078194517028802e-05,
      "loss": 0.4111,
      "step": 654
    },
    {
      "epoch": 0.41290729285686145,
      "grad_norm": 0.17976075410842896,
      "learning_rate": 4.076881371192998e-05,
      "loss": 0.473,
      "step": 655
    },
    {
      "epoch": 0.41353768567038335,
      "grad_norm": 0.1213824599981308,
      "learning_rate": 4.0755663966222645e-05,
      "loss": 0.3589,
      "step": 656
    },
    {
      "epoch": 0.4141680784839053,
      "grad_norm": 0.13618458807468414,
      "learning_rate": 4.0742495946329086e-05,
      "loss": 0.3873,
      "step": 657
    },
    {
      "epoch": 0.4147984712974272,
      "grad_norm": 0.12898501753807068,
      "learning_rate": 4.07293096654307e-05,
      "loss": 0.3933,
      "step": 658
    },
    {
      "epoch": 0.41542886411094915,
      "grad_norm": 0.1469496637582779,
      "learning_rate": 4.071610513672713e-05,
      "loss": 0.4204,
      "step": 659
    },
    {
      "epoch": 0.41605925692447104,
      "grad_norm": 0.14399708807468414,
      "learning_rate": 4.070288237343632e-05,
      "loss": 0.4844,
      "step": 660
    },
    {
      "epoch": 0.416689649737993,
      "grad_norm": 0.14005786180496216,
      "learning_rate": 4.068964138879444e-05,
      "loss": 0.4463,
      "step": 661
    },
    {
      "epoch": 0.4173200425515149,
      "grad_norm": 0.15872123837471008,
      "learning_rate": 4.0676382196055906e-05,
      "loss": 0.4447,
      "step": 662
    },
    {
      "epoch": 0.41795043536503684,
      "grad_norm": 0.1603812277317047,
      "learning_rate": 4.066310480849337e-05,
      "loss": 0.3953,
      "step": 663
    },
    {
      "epoch": 0.41858082817855874,
      "grad_norm": 0.1350301206111908,
      "learning_rate": 4.064980923939769e-05,
      "loss": 0.415,
      "step": 664
    },
    {
      "epoch": 0.4192112209920807,
      "grad_norm": 0.16543471813201904,
      "learning_rate": 4.063649550207792e-05,
      "loss": 0.4775,
      "step": 665
    },
    {
      "epoch": 0.41984161380560264,
      "grad_norm": 0.1276547908782959,
      "learning_rate": 4.0623163609861316e-05,
      "loss": 0.375,
      "step": 666
    },
    {
      "epoch": 0.42047200661912454,
      "grad_norm": 0.14537298679351807,
      "learning_rate": 4.0609813576093285e-05,
      "loss": 0.3697,
      "step": 667
    },
    {
      "epoch": 0.4211023994326465,
      "grad_norm": 0.14902625977993011,
      "learning_rate": 4.059644541413741e-05,
      "loss": 0.4501,
      "step": 668
    },
    {
      "epoch": 0.4217327922461684,
      "grad_norm": 0.1425814926624298,
      "learning_rate": 4.058305913737542e-05,
      "loss": 0.3701,
      "step": 669
    },
    {
      "epoch": 0.42236318505969034,
      "grad_norm": 0.1408214420080185,
      "learning_rate": 4.0569654759207164e-05,
      "loss": 0.357,
      "step": 670
    },
    {
      "epoch": 0.42299357787321223,
      "grad_norm": 0.15346527099609375,
      "learning_rate": 4.055623229305063e-05,
      "loss": 0.3679,
      "step": 671
    },
    {
      "epoch": 0.4236239706867342,
      "grad_norm": 0.14200004935264587,
      "learning_rate": 4.0542791752341904e-05,
      "loss": 0.374,
      "step": 672
    },
    {
      "epoch": 0.4242543635002561,
      "grad_norm": 0.13923034071922302,
      "learning_rate": 4.0529333150535145e-05,
      "loss": 0.4081,
      "step": 673
    },
    {
      "epoch": 0.42488475631377803,
      "grad_norm": 0.1957700550556183,
      "learning_rate": 4.051585650110264e-05,
      "loss": 0.5211,
      "step": 674
    },
    {
      "epoch": 0.42551514912729993,
      "grad_norm": 0.1250585913658142,
      "learning_rate": 4.0502361817534686e-05,
      "loss": 0.3656,
      "step": 675
    },
    {
      "epoch": 0.4261455419408219,
      "grad_norm": 0.14980201423168182,
      "learning_rate": 4.048884911333967e-05,
      "loss": 0.3976,
      "step": 676
    },
    {
      "epoch": 0.4267759347543438,
      "grad_norm": 0.14705874025821686,
      "learning_rate": 4.0475318402043996e-05,
      "loss": 0.4219,
      "step": 677
    },
    {
      "epoch": 0.42740632756786573,
      "grad_norm": 0.17089523375034332,
      "learning_rate": 4.046176969719212e-05,
      "loss": 0.5121,
      "step": 678
    },
    {
      "epoch": 0.4280367203813876,
      "grad_norm": 0.14580880105495453,
      "learning_rate": 4.0448203012346486e-05,
      "loss": 0.3927,
      "step": 679
    },
    {
      "epoch": 0.4286671131949096,
      "grad_norm": 0.14165879786014557,
      "learning_rate": 4.043461836108754e-05,
      "loss": 0.3613,
      "step": 680
    },
    {
      "epoch": 0.42929750600843153,
      "grad_norm": 0.17066994309425354,
      "learning_rate": 4.0421015757013714e-05,
      "loss": 0.4222,
      "step": 681
    },
    {
      "epoch": 0.4299278988219534,
      "grad_norm": 0.1570414900779724,
      "learning_rate": 4.040739521374143e-05,
      "loss": 0.42,
      "step": 682
    },
    {
      "epoch": 0.4305582916354754,
      "grad_norm": 0.16085675358772278,
      "learning_rate": 4.0393756744905025e-05,
      "loss": 0.3998,
      "step": 683
    },
    {
      "epoch": 0.4311886844489973,
      "grad_norm": 0.13752546906471252,
      "learning_rate": 4.038010036415683e-05,
      "loss": 0.4599,
      "step": 684
    },
    {
      "epoch": 0.4318190772625192,
      "grad_norm": 0.14098380506038666,
      "learning_rate": 4.036642608516707e-05,
      "loss": 0.4026,
      "step": 685
    },
    {
      "epoch": 0.4324494700760411,
      "grad_norm": 0.14366577565670013,
      "learning_rate": 4.035273392162389e-05,
      "loss": 0.469,
      "step": 686
    },
    {
      "epoch": 0.43307986288956307,
      "grad_norm": 0.14651672542095184,
      "learning_rate": 4.0339023887233366e-05,
      "loss": 0.4062,
      "step": 687
    },
    {
      "epoch": 0.43371025570308497,
      "grad_norm": 0.14978663623332977,
      "learning_rate": 4.032529599571942e-05,
      "loss": 0.3823,
      "step": 688
    },
    {
      "epoch": 0.4343406485166069,
      "grad_norm": 0.17241071164608002,
      "learning_rate": 4.031155026082389e-05,
      "loss": 0.418,
      "step": 689
    },
    {
      "epoch": 0.4349710413301288,
      "grad_norm": 0.18321415781974792,
      "learning_rate": 4.029778669630645e-05,
      "loss": 0.4246,
      "step": 690
    },
    {
      "epoch": 0.43560143414365077,
      "grad_norm": 0.12590643763542175,
      "learning_rate": 4.028400531594463e-05,
      "loss": 0.3959,
      "step": 691
    },
    {
      "epoch": 0.43623182695717266,
      "grad_norm": 0.1435738354921341,
      "learning_rate": 4.027020613353379e-05,
      "loss": 0.3956,
      "step": 692
    },
    {
      "epoch": 0.4368622197706946,
      "grad_norm": 0.15861615538597107,
      "learning_rate": 4.025638916288711e-05,
      "loss": 0.4415,
      "step": 693
    },
    {
      "epoch": 0.43749261258421657,
      "grad_norm": 0.14547286927700043,
      "learning_rate": 4.02425544178356e-05,
      "loss": 0.3686,
      "step": 694
    },
    {
      "epoch": 0.43812300539773846,
      "grad_norm": 0.13145437836647034,
      "learning_rate": 4.0228701912228015e-05,
      "loss": 0.3877,
      "step": 695
    },
    {
      "epoch": 0.4387533982112604,
      "grad_norm": 0.14630243182182312,
      "learning_rate": 4.021483165993093e-05,
      "loss": 0.3912,
      "step": 696
    },
    {
      "epoch": 0.4393837910247823,
      "grad_norm": 0.17564375698566437,
      "learning_rate": 4.020094367482867e-05,
      "loss": 0.4798,
      "step": 697
    },
    {
      "epoch": 0.44001418383830426,
      "grad_norm": 0.1785062700510025,
      "learning_rate": 4.0187037970823304e-05,
      "loss": 0.4223,
      "step": 698
    },
    {
      "epoch": 0.44064457665182616,
      "grad_norm": 0.14428603649139404,
      "learning_rate": 4.017311456183465e-05,
      "loss": 0.3492,
      "step": 699
    },
    {
      "epoch": 0.4412749694653481,
      "grad_norm": 0.15606176853179932,
      "learning_rate": 4.015917346180025e-05,
      "loss": 0.4348,
      "step": 700
    },
    {
      "epoch": 0.44190536227887,
      "grad_norm": 0.1602056622505188,
      "learning_rate": 4.014521468467533e-05,
      "loss": 0.4422,
      "step": 701
    },
    {
      "epoch": 0.44253575509239196,
      "grad_norm": 0.11997710913419724,
      "learning_rate": 4.013123824443284e-05,
      "loss": 0.4088,
      "step": 702
    },
    {
      "epoch": 0.44316614790591385,
      "grad_norm": 0.12170203775167465,
      "learning_rate": 4.011724415506341e-05,
      "loss": 0.3383,
      "step": 703
    },
    {
      "epoch": 0.4437965407194358,
      "grad_norm": 0.13026146590709686,
      "learning_rate": 4.0103232430575314e-05,
      "loss": 0.4095,
      "step": 704
    },
    {
      "epoch": 0.4444269335329577,
      "grad_norm": 0.1414802074432373,
      "learning_rate": 4.00892030849945e-05,
      "loss": 0.3865,
      "step": 705
    },
    {
      "epoch": 0.44505732634647965,
      "grad_norm": 0.1612880527973175,
      "learning_rate": 4.007515613236455e-05,
      "loss": 0.4371,
      "step": 706
    },
    {
      "epoch": 0.44568771916000155,
      "grad_norm": 0.13235218822956085,
      "learning_rate": 4.0061091586746665e-05,
      "loss": 0.4008,
      "step": 707
    },
    {
      "epoch": 0.4463181119735235,
      "grad_norm": 0.12737032771110535,
      "learning_rate": 4.0047009462219655e-05,
      "loss": 0.3763,
      "step": 708
    },
    {
      "epoch": 0.44694850478704545,
      "grad_norm": 0.13643917441368103,
      "learning_rate": 4.003290977287994e-05,
      "loss": 0.394,
      "step": 709
    },
    {
      "epoch": 0.44757889760056735,
      "grad_norm": 0.19716039299964905,
      "learning_rate": 4.001879253284152e-05,
      "loss": 0.5007,
      "step": 710
    },
    {
      "epoch": 0.4482092904140893,
      "grad_norm": 0.1554064154624939,
      "learning_rate": 4.0004657756235944e-05,
      "loss": 0.4375,
      "step": 711
    },
    {
      "epoch": 0.4488396832276112,
      "grad_norm": 0.17151057720184326,
      "learning_rate": 3.999050545721234e-05,
      "loss": 0.3899,
      "step": 712
    },
    {
      "epoch": 0.44947007604113315,
      "grad_norm": 0.1316985934972763,
      "learning_rate": 3.997633564993736e-05,
      "loss": 0.3585,
      "step": 713
    },
    {
      "epoch": 0.45010046885465504,
      "grad_norm": 0.16340312361717224,
      "learning_rate": 3.996214834859519e-05,
      "loss": 0.4085,
      "step": 714
    },
    {
      "epoch": 0.450730861668177,
      "grad_norm": 0.11996844410896301,
      "learning_rate": 3.994794356738753e-05,
      "loss": 0.361,
      "step": 715
    },
    {
      "epoch": 0.4513612544816989,
      "grad_norm": 0.14515595138072968,
      "learning_rate": 3.993372132053356e-05,
      "loss": 0.4109,
      "step": 716
    },
    {
      "epoch": 0.45199164729522084,
      "grad_norm": 0.14319062232971191,
      "learning_rate": 3.991948162226999e-05,
      "loss": 0.3465,
      "step": 717
    },
    {
      "epoch": 0.45262204010874274,
      "grad_norm": 0.16570813953876495,
      "learning_rate": 3.990522448685092e-05,
      "loss": 0.4548,
      "step": 718
    },
    {
      "epoch": 0.4532524329222647,
      "grad_norm": 0.17157788574695587,
      "learning_rate": 3.989094992854798e-05,
      "loss": 0.4332,
      "step": 719
    },
    {
      "epoch": 0.4538828257357866,
      "grad_norm": 0.15220706164836884,
      "learning_rate": 3.987665796165021e-05,
      "loss": 0.3821,
      "step": 720
    },
    {
      "epoch": 0.45451321854930854,
      "grad_norm": 0.13397742807865143,
      "learning_rate": 3.986234860046407e-05,
      "loss": 0.3754,
      "step": 721
    },
    {
      "epoch": 0.4551436113628305,
      "grad_norm": 0.14851902425289154,
      "learning_rate": 3.984802185931344e-05,
      "loss": 0.432,
      "step": 722
    },
    {
      "epoch": 0.4557740041763524,
      "grad_norm": 0.16764099895954132,
      "learning_rate": 3.98336777525396e-05,
      "loss": 0.4495,
      "step": 723
    },
    {
      "epoch": 0.45640439698987434,
      "grad_norm": 0.15480101108551025,
      "learning_rate": 3.9819316294501217e-05,
      "loss": 0.4505,
      "step": 724
    },
    {
      "epoch": 0.45703478980339624,
      "grad_norm": 0.15106666088104248,
      "learning_rate": 3.9804937499574296e-05,
      "loss": 0.4407,
      "step": 725
    },
    {
      "epoch": 0.4576651826169182,
      "grad_norm": 0.16314244270324707,
      "learning_rate": 3.979054138215225e-05,
      "loss": 0.3506,
      "step": 726
    },
    {
      "epoch": 0.4582955754304401,
      "grad_norm": 0.14463144540786743,
      "learning_rate": 3.977612795664579e-05,
      "loss": 0.4328,
      "step": 727
    },
    {
      "epoch": 0.45892596824396203,
      "grad_norm": 0.16658100485801697,
      "learning_rate": 3.976169723748296e-05,
      "loss": 0.4492,
      "step": 728
    },
    {
      "epoch": 0.45955636105748393,
      "grad_norm": 0.1776987612247467,
      "learning_rate": 3.9747249239109126e-05,
      "loss": 0.4352,
      "step": 729
    },
    {
      "epoch": 0.4601867538710059,
      "grad_norm": 0.1609821766614914,
      "learning_rate": 3.973278397598695e-05,
      "loss": 0.4014,
      "step": 730
    },
    {
      "epoch": 0.4608171466845278,
      "grad_norm": 0.16667957603931427,
      "learning_rate": 3.971830146259637e-05,
      "loss": 0.422,
      "step": 731
    },
    {
      "epoch": 0.46144753949804973,
      "grad_norm": 0.1358121931552887,
      "learning_rate": 3.970380171343458e-05,
      "loss": 0.3901,
      "step": 732
    },
    {
      "epoch": 0.4620779323115716,
      "grad_norm": 0.12884113192558289,
      "learning_rate": 3.968928474301606e-05,
      "loss": 0.3777,
      "step": 733
    },
    {
      "epoch": 0.4627083251250936,
      "grad_norm": 0.14654743671417236,
      "learning_rate": 3.9674750565872495e-05,
      "loss": 0.4344,
      "step": 734
    },
    {
      "epoch": 0.4633387179386155,
      "grad_norm": 0.15058912336826324,
      "learning_rate": 3.966019919655282e-05,
      "loss": 0.4133,
      "step": 735
    },
    {
      "epoch": 0.4639691107521374,
      "grad_norm": 0.13025817275047302,
      "learning_rate": 3.964563064962316e-05,
      "loss": 0.3639,
      "step": 736
    },
    {
      "epoch": 0.4645995035656594,
      "grad_norm": 0.146138533949852,
      "learning_rate": 3.9631044939666836e-05,
      "loss": 0.37,
      "step": 737
    },
    {
      "epoch": 0.4652298963791813,
      "grad_norm": 0.14348472654819489,
      "learning_rate": 3.961644208128438e-05,
      "loss": 0.4255,
      "step": 738
    },
    {
      "epoch": 0.4658602891927032,
      "grad_norm": 0.14459875226020813,
      "learning_rate": 3.9601822089093444e-05,
      "loss": 0.412,
      "step": 739
    },
    {
      "epoch": 0.4664906820062251,
      "grad_norm": 0.17654073238372803,
      "learning_rate": 3.9587184977728866e-05,
      "loss": 0.4538,
      "step": 740
    },
    {
      "epoch": 0.4671210748197471,
      "grad_norm": 0.13895660638809204,
      "learning_rate": 3.957253076184261e-05,
      "loss": 0.425,
      "step": 741
    },
    {
      "epoch": 0.46775146763326897,
      "grad_norm": 0.1470203697681427,
      "learning_rate": 3.9557859456103746e-05,
      "loss": 0.4476,
      "step": 742
    },
    {
      "epoch": 0.4683818604467909,
      "grad_norm": 0.14507576823234558,
      "learning_rate": 3.954317107519848e-05,
      "loss": 0.4101,
      "step": 743
    },
    {
      "epoch": 0.4690122532603128,
      "grad_norm": 0.129420667886734,
      "learning_rate": 3.9528465633830076e-05,
      "loss": 0.3631,
      "step": 744
    },
    {
      "epoch": 0.46964264607383477,
      "grad_norm": 0.12379541993141174,
      "learning_rate": 3.951374314671893e-05,
      "loss": 0.3204,
      "step": 745
    },
    {
      "epoch": 0.47027303888735666,
      "grad_norm": 0.14565017819404602,
      "learning_rate": 3.949900362860243e-05,
      "loss": 0.3734,
      "step": 746
    },
    {
      "epoch": 0.4709034317008786,
      "grad_norm": 0.1342315524816513,
      "learning_rate": 3.948424709423507e-05,
      "loss": 0.4007,
      "step": 747
    },
    {
      "epoch": 0.4715338245144005,
      "grad_norm": 0.13480636477470398,
      "learning_rate": 3.946947355838835e-05,
      "loss": 0.4612,
      "step": 748
    },
    {
      "epoch": 0.47216421732792246,
      "grad_norm": 0.13085278868675232,
      "learning_rate": 3.94546830358508e-05,
      "loss": 0.4104,
      "step": 749
    },
    {
      "epoch": 0.47279461014144436,
      "grad_norm": 0.16482636332511902,
      "learning_rate": 3.943987554142794e-05,
      "loss": 0.4542,
      "step": 750
    },
    {
      "epoch": 0.4734250029549663,
      "grad_norm": 0.15673352777957916,
      "learning_rate": 3.94250510899423e-05,
      "loss": 0.3808,
      "step": 751
    },
    {
      "epoch": 0.47405539576848826,
      "grad_norm": 0.15453855693340302,
      "learning_rate": 3.9410209696233356e-05,
      "loss": 0.3816,
      "step": 752
    },
    {
      "epoch": 0.47468578858201016,
      "grad_norm": 0.15016509592533112,
      "learning_rate": 3.939535137515758e-05,
      "loss": 0.39,
      "step": 753
    },
    {
      "epoch": 0.4753161813955321,
      "grad_norm": 0.13890962302684784,
      "learning_rate": 3.9380476141588355e-05,
      "loss": 0.412,
      "step": 754
    },
    {
      "epoch": 0.475946574209054,
      "grad_norm": 0.1330081820487976,
      "learning_rate": 3.9365584010416e-05,
      "loss": 0.3242,
      "step": 755
    },
    {
      "epoch": 0.47657696702257596,
      "grad_norm": 0.1311984807252884,
      "learning_rate": 3.935067499654777e-05,
      "loss": 0.3583,
      "step": 756
    },
    {
      "epoch": 0.47720735983609786,
      "grad_norm": 0.14548636972904205,
      "learning_rate": 3.9335749114907816e-05,
      "loss": 0.3684,
      "step": 757
    },
    {
      "epoch": 0.4778377526496198,
      "grad_norm": 0.1494605988264084,
      "learning_rate": 3.932080638043714e-05,
      "loss": 0.4363,
      "step": 758
    },
    {
      "epoch": 0.4784681454631417,
      "grad_norm": 0.126301571726799,
      "learning_rate": 3.930584680809364e-05,
      "loss": 0.3807,
      "step": 759
    },
    {
      "epoch": 0.47909853827666365,
      "grad_norm": 0.1573512852191925,
      "learning_rate": 3.9290870412852076e-05,
      "loss": 0.4454,
      "step": 760
    },
    {
      "epoch": 0.47972893109018555,
      "grad_norm": 0.16862505674362183,
      "learning_rate": 3.927587720970404e-05,
      "loss": 0.4264,
      "step": 761
    },
    {
      "epoch": 0.4803593239037075,
      "grad_norm": 0.16232459247112274,
      "learning_rate": 3.9260867213657934e-05,
      "loss": 0.3882,
      "step": 762
    },
    {
      "epoch": 0.4809897167172294,
      "grad_norm": 0.12893806397914886,
      "learning_rate": 3.9245840439739e-05,
      "loss": 0.3486,
      "step": 763
    },
    {
      "epoch": 0.48162010953075135,
      "grad_norm": 0.2132662981748581,
      "learning_rate": 3.923079690298924e-05,
      "loss": 0.5119,
      "step": 764
    },
    {
      "epoch": 0.4822505023442733,
      "grad_norm": 0.15508213639259338,
      "learning_rate": 3.921573661846747e-05,
      "loss": 0.3704,
      "step": 765
    },
    {
      "epoch": 0.4828808951577952,
      "grad_norm": 0.12923885881900787,
      "learning_rate": 3.920065960124925e-05,
      "loss": 0.4451,
      "step": 766
    },
    {
      "epoch": 0.48351128797131715,
      "grad_norm": 0.12894199788570404,
      "learning_rate": 3.918556586642689e-05,
      "loss": 0.3758,
      "step": 767
    },
    {
      "epoch": 0.48414168078483905,
      "grad_norm": 0.16268831491470337,
      "learning_rate": 3.917045542910944e-05,
      "loss": 0.4023,
      "step": 768
    },
    {
      "epoch": 0.484772073598361,
      "grad_norm": 0.14713963866233826,
      "learning_rate": 3.915532830442267e-05,
      "loss": 0.4138,
      "step": 769
    },
    {
      "epoch": 0.4854024664118829,
      "grad_norm": 0.14321781694889069,
      "learning_rate": 3.914018450750905e-05,
      "loss": 0.3415,
      "step": 770
    },
    {
      "epoch": 0.48603285922540485,
      "grad_norm": 0.15645352005958557,
      "learning_rate": 3.912502405352776e-05,
      "loss": 0.4201,
      "step": 771
    },
    {
      "epoch": 0.48666325203892674,
      "grad_norm": 0.15770842134952545,
      "learning_rate": 3.9109846957654605e-05,
      "loss": 0.3694,
      "step": 772
    },
    {
      "epoch": 0.4872936448524487,
      "grad_norm": 0.1686742752790451,
      "learning_rate": 3.9094653235082104e-05,
      "loss": 0.384,
      "step": 773
    },
    {
      "epoch": 0.4879240376659706,
      "grad_norm": 0.1598564088344574,
      "learning_rate": 3.9079442901019384e-05,
      "loss": 0.4244,
      "step": 774
    },
    {
      "epoch": 0.48855443047949254,
      "grad_norm": 0.14372321963310242,
      "learning_rate": 3.90642159706922e-05,
      "loss": 0.4358,
      "step": 775
    },
    {
      "epoch": 0.48918482329301444,
      "grad_norm": 0.13739392161369324,
      "learning_rate": 3.904897245934297e-05,
      "loss": 0.4267,
      "step": 776
    },
    {
      "epoch": 0.4898152161065364,
      "grad_norm": 0.16683433949947357,
      "learning_rate": 3.903371238223062e-05,
      "loss": 0.3607,
      "step": 777
    },
    {
      "epoch": 0.4904456089200583,
      "grad_norm": 0.16543376445770264,
      "learning_rate": 3.901843575463076e-05,
      "loss": 0.4113,
      "step": 778
    },
    {
      "epoch": 0.49107600173358024,
      "grad_norm": 0.1504923403263092,
      "learning_rate": 3.900314259183548e-05,
      "loss": 0.4765,
      "step": 779
    },
    {
      "epoch": 0.4917063945471022,
      "grad_norm": 0.16944220662117004,
      "learning_rate": 3.8987832909153484e-05,
      "loss": 0.4301,
      "step": 780
    },
    {
      "epoch": 0.4923367873606241,
      "grad_norm": 0.13818518817424774,
      "learning_rate": 3.8972506721909975e-05,
      "loss": 0.487,
      "step": 781
    },
    {
      "epoch": 0.49296718017414604,
      "grad_norm": 0.15189217031002045,
      "learning_rate": 3.895716404544669e-05,
      "loss": 0.3897,
      "step": 782
    },
    {
      "epoch": 0.49359757298766793,
      "grad_norm": 0.1592871993780136,
      "learning_rate": 3.894180489512187e-05,
      "loss": 0.3851,
      "step": 783
    },
    {
      "epoch": 0.4942279658011899,
      "grad_norm": 0.15947246551513672,
      "learning_rate": 3.8926429286310255e-05,
      "loss": 0.3861,
      "step": 784
    },
    {
      "epoch": 0.4948583586147118,
      "grad_norm": 0.1814534068107605,
      "learning_rate": 3.891103723440306e-05,
      "loss": 0.3973,
      "step": 785
    },
    {
      "epoch": 0.49548875142823373,
      "grad_norm": 0.12373247742652893,
      "learning_rate": 3.8895628754807936e-05,
      "loss": 0.3791,
      "step": 786
    },
    {
      "epoch": 0.4961191442417556,
      "grad_norm": 0.18603551387786865,
      "learning_rate": 3.888020386294902e-05,
      "loss": 0.3958,
      "step": 787
    },
    {
      "epoch": 0.4967495370552776,
      "grad_norm": 0.16354039311408997,
      "learning_rate": 3.886476257426683e-05,
      "loss": 0.3913,
      "step": 788
    },
    {
      "epoch": 0.4973799298687995,
      "grad_norm": 0.1754021793603897,
      "learning_rate": 3.8849304904218335e-05,
      "loss": 0.4388,
      "step": 789
    },
    {
      "epoch": 0.4980103226823214,
      "grad_norm": 0.16621164977550507,
      "learning_rate": 3.883383086827688e-05,
      "loss": 0.4342,
      "step": 790
    },
    {
      "epoch": 0.4986407154958433,
      "grad_norm": 0.18488451838493347,
      "learning_rate": 3.8818340481932206e-05,
      "loss": 0.4792,
      "step": 791
    },
    {
      "epoch": 0.4992711083093653,
      "grad_norm": 0.14420366287231445,
      "learning_rate": 3.8802833760690424e-05,
      "loss": 0.361,
      "step": 792
    },
    {
      "epoch": 0.4999015011228872,
      "grad_norm": 0.16407188773155212,
      "learning_rate": 3.878731072007398e-05,
      "loss": 0.3657,
      "step": 793
    },
    {
      "epoch": 0.5005318939364092,
      "grad_norm": 0.13707882165908813,
      "learning_rate": 3.877177137562166e-05,
      "loss": 0.4247,
      "step": 794
    },
    {
      "epoch": 0.501162286749931,
      "grad_norm": 0.13158880174160004,
      "learning_rate": 3.875621574288858e-05,
      "loss": 0.4068,
      "step": 795
    },
    {
      "epoch": 0.501792679563453,
      "grad_norm": 0.16040393710136414,
      "learning_rate": 3.874064383744616e-05,
      "loss": 0.4172,
      "step": 796
    },
    {
      "epoch": 0.5024230723769749,
      "grad_norm": 0.15780168771743774,
      "learning_rate": 3.8725055674882096e-05,
      "loss": 0.4144,
      "step": 797
    },
    {
      "epoch": 0.5030534651904969,
      "grad_norm": 0.14710697531700134,
      "learning_rate": 3.870945127080037e-05,
      "loss": 0.3704,
      "step": 798
    },
    {
      "epoch": 0.5036838580040187,
      "grad_norm": 0.18354450166225433,
      "learning_rate": 3.869383064082122e-05,
      "loss": 0.4544,
      "step": 799
    },
    {
      "epoch": 0.5043142508175407,
      "grad_norm": 0.14307357370853424,
      "learning_rate": 3.867819380058112e-05,
      "loss": 0.3872,
      "step": 800
    },
    {
      "epoch": 0.5043142508175407,
      "eval_loss": 0.4390900135040283,
      "eval_runtime": 222.7922,
      "eval_samples_per_second": 4.488,
      "eval_steps_per_second": 4.488,
      "step": 800
    },
    {
      "epoch": 0.5049446436310626,
      "grad_norm": 0.1444072276353836,
      "learning_rate": 3.866254076573278e-05,
      "loss": 0.4046,
      "step": 801
    },
    {
      "epoch": 0.5055750364445846,
      "grad_norm": 0.13857287168502808,
      "learning_rate": 3.864687155194512e-05,
      "loss": 0.4448,
      "step": 802
    },
    {
      "epoch": 0.5062054292581064,
      "grad_norm": 0.16345997154712677,
      "learning_rate": 3.8631186174903236e-05,
      "loss": 0.3848,
      "step": 803
    },
    {
      "epoch": 0.5068358220716284,
      "grad_norm": 0.13920150697231293,
      "learning_rate": 3.861548465030845e-05,
      "loss": 0.3894,
      "step": 804
    },
    {
      "epoch": 0.5074662148851503,
      "grad_norm": 0.15711119771003723,
      "learning_rate": 3.8599766993878173e-05,
      "loss": 0.433,
      "step": 805
    },
    {
      "epoch": 0.5080966076986723,
      "grad_norm": 0.16589997708797455,
      "learning_rate": 3.858403322134605e-05,
      "loss": 0.4011,
      "step": 806
    },
    {
      "epoch": 0.5087270005121942,
      "grad_norm": 0.15303505957126617,
      "learning_rate": 3.8568283348461794e-05,
      "loss": 0.4127,
      "step": 807
    },
    {
      "epoch": 0.5093573933257161,
      "grad_norm": 0.13526608049869537,
      "learning_rate": 3.855251739099127e-05,
      "loss": 0.3545,
      "step": 808
    },
    {
      "epoch": 0.509987786139238,
      "grad_norm": 0.14198428392410278,
      "learning_rate": 3.8536735364716425e-05,
      "loss": 0.4364,
      "step": 809
    },
    {
      "epoch": 0.51061817895276,
      "grad_norm": 0.14397798478603363,
      "learning_rate": 3.852093728543529e-05,
      "loss": 0.4014,
      "step": 810
    },
    {
      "epoch": 0.5112485717662819,
      "grad_norm": 0.13944432139396667,
      "learning_rate": 3.8505123168961987e-05,
      "loss": 0.4326,
      "step": 811
    },
    {
      "epoch": 0.5118789645798038,
      "grad_norm": 0.157311350107193,
      "learning_rate": 3.848929303112668e-05,
      "loss": 0.453,
      "step": 812
    },
    {
      "epoch": 0.5125093573933257,
      "grad_norm": 0.14742758870124817,
      "learning_rate": 3.8473446887775546e-05,
      "loss": 0.5064,
      "step": 813
    },
    {
      "epoch": 0.5131397502068477,
      "grad_norm": 0.1551235318183899,
      "learning_rate": 3.845758475477083e-05,
      "loss": 0.4191,
      "step": 814
    },
    {
      "epoch": 0.5137701430203696,
      "grad_norm": 0.13968229293823242,
      "learning_rate": 3.844170664799074e-05,
      "loss": 0.3774,
      "step": 815
    },
    {
      "epoch": 0.5144005358338914,
      "grad_norm": 0.14213190972805023,
      "learning_rate": 3.8425812583329514e-05,
      "loss": 0.411,
      "step": 816
    },
    {
      "epoch": 0.5150309286474134,
      "grad_norm": 0.1836906224489212,
      "learning_rate": 3.840990257669732e-05,
      "loss": 0.4508,
      "step": 817
    },
    {
      "epoch": 0.5156613214609354,
      "grad_norm": 0.13046447932720184,
      "learning_rate": 3.8393976644020325e-05,
      "loss": 0.3677,
      "step": 818
    },
    {
      "epoch": 0.5162917142744573,
      "grad_norm": 0.1387125700712204,
      "learning_rate": 3.837803480124061e-05,
      "loss": 0.3913,
      "step": 819
    },
    {
      "epoch": 0.5169221070879791,
      "grad_norm": 0.1600324660539627,
      "learning_rate": 3.8362077064316195e-05,
      "loss": 0.3776,
      "step": 820
    },
    {
      "epoch": 0.5175524999015011,
      "grad_norm": 0.12740956246852875,
      "learning_rate": 3.834610344922101e-05,
      "loss": 0.4008,
      "step": 821
    },
    {
      "epoch": 0.518182892715023,
      "grad_norm": 0.13033899664878845,
      "learning_rate": 3.833011397194487e-05,
      "loss": 0.4032,
      "step": 822
    },
    {
      "epoch": 0.518813285528545,
      "grad_norm": 0.15081040561199188,
      "learning_rate": 3.831410864849348e-05,
      "loss": 0.3461,
      "step": 823
    },
    {
      "epoch": 0.519443678342067,
      "grad_norm": 0.13452240824699402,
      "learning_rate": 3.829808749488841e-05,
      "loss": 0.3746,
      "step": 824
    },
    {
      "epoch": 0.5200740711555888,
      "grad_norm": 0.19129051268100739,
      "learning_rate": 3.8282050527167056e-05,
      "loss": 0.483,
      "step": 825
    },
    {
      "epoch": 0.5207044639691107,
      "grad_norm": 0.1633075773715973,
      "learning_rate": 3.826599776138266e-05,
      "loss": 0.4815,
      "step": 826
    },
    {
      "epoch": 0.5213348567826327,
      "grad_norm": 0.1542254090309143,
      "learning_rate": 3.824992921360428e-05,
      "loss": 0.4134,
      "step": 827
    },
    {
      "epoch": 0.5219652495961546,
      "grad_norm": 0.1658627986907959,
      "learning_rate": 3.8233844899916765e-05,
      "loss": 0.4263,
      "step": 828
    },
    {
      "epoch": 0.5225956424096765,
      "grad_norm": 0.15097402036190033,
      "learning_rate": 3.821774483642075e-05,
      "loss": 0.4252,
      "step": 829
    },
    {
      "epoch": 0.5232260352231984,
      "grad_norm": 0.14764727652072906,
      "learning_rate": 3.8201629039232624e-05,
      "loss": 0.4535,
      "step": 830
    },
    {
      "epoch": 0.5238564280367204,
      "grad_norm": 0.1661219298839569,
      "learning_rate": 3.818549752448454e-05,
      "loss": 0.4581,
      "step": 831
    },
    {
      "epoch": 0.5244868208502423,
      "grad_norm": 0.16364407539367676,
      "learning_rate": 3.816935030832439e-05,
      "loss": 0.415,
      "step": 832
    },
    {
      "epoch": 0.5251172136637642,
      "grad_norm": 0.14157086610794067,
      "learning_rate": 3.815318740691576e-05,
      "loss": 0.3201,
      "step": 833
    },
    {
      "epoch": 0.5257476064772861,
      "grad_norm": 0.1508474200963974,
      "learning_rate": 3.8137008836437954e-05,
      "loss": 0.4632,
      "step": 834
    },
    {
      "epoch": 0.5263779992908081,
      "grad_norm": 0.16357192397117615,
      "learning_rate": 3.8120814613085965e-05,
      "loss": 0.4291,
      "step": 835
    },
    {
      "epoch": 0.52700839210433,
      "grad_norm": 0.1774810403585434,
      "learning_rate": 3.810460475307044e-05,
      "loss": 0.4183,
      "step": 836
    },
    {
      "epoch": 0.527638784917852,
      "grad_norm": 0.14707089960575104,
      "learning_rate": 3.808837927261768e-05,
      "loss": 0.4873,
      "step": 837
    },
    {
      "epoch": 0.5282691777313738,
      "grad_norm": 0.17293687164783478,
      "learning_rate": 3.8072138187969637e-05,
      "loss": 0.3572,
      "step": 838
    },
    {
      "epoch": 0.5288995705448958,
      "grad_norm": 0.15435566008090973,
      "learning_rate": 3.805588151538387e-05,
      "loss": 0.4386,
      "step": 839
    },
    {
      "epoch": 0.5295299633584177,
      "grad_norm": 0.13135646283626556,
      "learning_rate": 3.803960927113355e-05,
      "loss": 0.3141,
      "step": 840
    },
    {
      "epoch": 0.5301603561719397,
      "grad_norm": 0.1465950310230255,
      "learning_rate": 3.802332147150742e-05,
      "loss": 0.4001,
      "step": 841
    },
    {
      "epoch": 0.5307907489854615,
      "grad_norm": 0.15999934077262878,
      "learning_rate": 3.800701813280982e-05,
      "loss": 0.4008,
      "step": 842
    },
    {
      "epoch": 0.5314211417989835,
      "grad_norm": 0.20061102509498596,
      "learning_rate": 3.7990699271360624e-05,
      "loss": 0.496,
      "step": 843
    },
    {
      "epoch": 0.5320515346125054,
      "grad_norm": 0.15036503970623016,
      "learning_rate": 3.7974364903495244e-05,
      "loss": 0.4079,
      "step": 844
    },
    {
      "epoch": 0.5326819274260274,
      "grad_norm": 0.14602142572402954,
      "learning_rate": 3.795801504556464e-05,
      "loss": 0.3936,
      "step": 845
    },
    {
      "epoch": 0.5333123202395492,
      "grad_norm": 0.15145988762378693,
      "learning_rate": 3.7941649713935235e-05,
      "loss": 0.4181,
      "step": 846
    },
    {
      "epoch": 0.5339427130530712,
      "grad_norm": 0.155500590801239,
      "learning_rate": 3.792526892498897e-05,
      "loss": 0.4588,
      "step": 847
    },
    {
      "epoch": 0.5345731058665931,
      "grad_norm": 0.15288515388965607,
      "learning_rate": 3.790887269512326e-05,
      "loss": 0.3514,
      "step": 848
    },
    {
      "epoch": 0.5352034986801151,
      "grad_norm": 0.13540484011173248,
      "learning_rate": 3.7892461040750975e-05,
      "loss": 0.41,
      "step": 849
    },
    {
      "epoch": 0.535833891493637,
      "grad_norm": 0.1624019742012024,
      "learning_rate": 3.787603397830042e-05,
      "loss": 0.4208,
      "step": 850
    },
    {
      "epoch": 0.5364642843071589,
      "grad_norm": 0.14963330328464508,
      "learning_rate": 3.785959152421531e-05,
      "loss": 0.4226,
      "step": 851
    },
    {
      "epoch": 0.5370946771206808,
      "grad_norm": 0.1439056098461151,
      "learning_rate": 3.784313369495479e-05,
      "loss": 0.3824,
      "step": 852
    },
    {
      "epoch": 0.5377250699342028,
      "grad_norm": 0.13627079129219055,
      "learning_rate": 3.78266605069934e-05,
      "loss": 0.3943,
      "step": 853
    },
    {
      "epoch": 0.5383554627477247,
      "grad_norm": 0.12575384974479675,
      "learning_rate": 3.7810171976821014e-05,
      "loss": 0.4188,
      "step": 854
    },
    {
      "epoch": 0.5389858555612466,
      "grad_norm": 0.11646067351102829,
      "learning_rate": 3.779366812094292e-05,
      "loss": 0.391,
      "step": 855
    },
    {
      "epoch": 0.5396162483747685,
      "grad_norm": 0.19494909048080444,
      "learning_rate": 3.777714895587971e-05,
      "loss": 0.4874,
      "step": 856
    },
    {
      "epoch": 0.5402466411882905,
      "grad_norm": 0.1547946184873581,
      "learning_rate": 3.7760614498167294e-05,
      "loss": 0.4467,
      "step": 857
    },
    {
      "epoch": 0.5408770340018124,
      "grad_norm": 0.12604691088199615,
      "learning_rate": 3.7744064764356926e-05,
      "loss": 0.3483,
      "step": 858
    },
    {
      "epoch": 0.5415074268153343,
      "grad_norm": 0.18869976699352264,
      "learning_rate": 3.772749977101512e-05,
      "loss": 0.4451,
      "step": 859
    },
    {
      "epoch": 0.5421378196288562,
      "grad_norm": 0.16798147559165955,
      "learning_rate": 3.771091953472368e-05,
      "loss": 0.4239,
      "step": 860
    },
    {
      "epoch": 0.5427682124423782,
      "grad_norm": 0.136243999004364,
      "learning_rate": 3.769432407207966e-05,
      "loss": 0.3578,
      "step": 861
    },
    {
      "epoch": 0.5433986052559001,
      "grad_norm": 0.1705186516046524,
      "learning_rate": 3.767771339969537e-05,
      "loss": 0.4837,
      "step": 862
    },
    {
      "epoch": 0.544028998069422,
      "grad_norm": 0.1542588621377945,
      "learning_rate": 3.766108753419832e-05,
      "loss": 0.3601,
      "step": 863
    },
    {
      "epoch": 0.5446593908829439,
      "grad_norm": 0.15121257305145264,
      "learning_rate": 3.764444649223126e-05,
      "loss": 0.3634,
      "step": 864
    },
    {
      "epoch": 0.5452897836964659,
      "grad_norm": 0.1671563982963562,
      "learning_rate": 3.762779029045211e-05,
      "loss": 0.3991,
      "step": 865
    },
    {
      "epoch": 0.5459201765099878,
      "grad_norm": 0.16092930734157562,
      "learning_rate": 3.7611118945533974e-05,
      "loss": 0.4546,
      "step": 866
    },
    {
      "epoch": 0.5465505693235098,
      "grad_norm": 0.1620696634054184,
      "learning_rate": 3.759443247416511e-05,
      "loss": 0.417,
      "step": 867
    },
    {
      "epoch": 0.5471809621370316,
      "grad_norm": 0.1499662548303604,
      "learning_rate": 3.7577730893048906e-05,
      "loss": 0.4409,
      "step": 868
    },
    {
      "epoch": 0.5478113549505536,
      "grad_norm": 0.14857828617095947,
      "learning_rate": 3.7561014218903915e-05,
      "loss": 0.392,
      "step": 869
    },
    {
      "epoch": 0.5484417477640755,
      "grad_norm": 0.1388327032327652,
      "learning_rate": 3.754428246846375e-05,
      "loss": 0.4249,
      "step": 870
    },
    {
      "epoch": 0.5490721405775975,
      "grad_norm": 0.12996181845664978,
      "learning_rate": 3.752753565847715e-05,
      "loss": 0.4354,
      "step": 871
    },
    {
      "epoch": 0.5497025333911193,
      "grad_norm": 0.14674219489097595,
      "learning_rate": 3.75107738057079e-05,
      "loss": 0.3574,
      "step": 872
    },
    {
      "epoch": 0.5503329262046412,
      "grad_norm": 0.14478866755962372,
      "learning_rate": 3.7493996926934886e-05,
      "loss": 0.4462,
      "step": 873
    },
    {
      "epoch": 0.5509633190181632,
      "grad_norm": 0.17569395899772644,
      "learning_rate": 3.7477205038952e-05,
      "loss": 0.4337,
      "step": 874
    },
    {
      "epoch": 0.5515937118316852,
      "grad_norm": 0.19228047132492065,
      "learning_rate": 3.7460398158568145e-05,
      "loss": 0.4272,
      "step": 875
    },
    {
      "epoch": 0.552224104645207,
      "grad_norm": 0.16651074588298798,
      "learning_rate": 3.7443576302607273e-05,
      "loss": 0.4306,
      "step": 876
    },
    {
      "epoch": 0.5528544974587289,
      "grad_norm": 0.15477438271045685,
      "learning_rate": 3.74267394879083e-05,
      "loss": 0.4356,
      "step": 877
    },
    {
      "epoch": 0.5534848902722509,
      "grad_norm": 0.13447880744934082,
      "learning_rate": 3.740988773132513e-05,
      "loss": 0.3399,
      "step": 878
    },
    {
      "epoch": 0.5541152830857728,
      "grad_norm": 0.1480575054883957,
      "learning_rate": 3.739302104972662e-05,
      "loss": 0.3658,
      "step": 879
    },
    {
      "epoch": 0.5547456758992948,
      "grad_norm": 0.1701166033744812,
      "learning_rate": 3.737613945999656e-05,
      "loss": 0.3752,
      "step": 880
    },
    {
      "epoch": 0.5553760687128166,
      "grad_norm": 0.16578589379787445,
      "learning_rate": 3.7359242979033655e-05,
      "loss": 0.378,
      "step": 881
    },
    {
      "epoch": 0.5560064615263386,
      "grad_norm": 0.1562201976776123,
      "learning_rate": 3.7342331623751554e-05,
      "loss": 0.3944,
      "step": 882
    },
    {
      "epoch": 0.5566368543398605,
      "grad_norm": 0.18078267574310303,
      "learning_rate": 3.732540541107876e-05,
      "loss": 0.4377,
      "step": 883
    },
    {
      "epoch": 0.5572672471533825,
      "grad_norm": 0.13611632585525513,
      "learning_rate": 3.730846435795864e-05,
      "loss": 0.3743,
      "step": 884
    },
    {
      "epoch": 0.5578976399669043,
      "grad_norm": 0.1585019826889038,
      "learning_rate": 3.729150848134947e-05,
      "loss": 0.4132,
      "step": 885
    },
    {
      "epoch": 0.5585280327804263,
      "grad_norm": 0.14006251096725464,
      "learning_rate": 3.727453779822431e-05,
      "loss": 0.369,
      "step": 886
    },
    {
      "epoch": 0.5591584255939482,
      "grad_norm": 0.1613488793373108,
      "learning_rate": 3.7257552325571055e-05,
      "loss": 0.4728,
      "step": 887
    },
    {
      "epoch": 0.5597888184074702,
      "grad_norm": 0.1657625436782837,
      "learning_rate": 3.7240552080392425e-05,
      "loss": 0.4398,
      "step": 888
    },
    {
      "epoch": 0.560419211220992,
      "grad_norm": 0.15457575023174286,
      "learning_rate": 3.722353707970591e-05,
      "loss": 0.4068,
      "step": 889
    },
    {
      "epoch": 0.561049604034514,
      "grad_norm": 0.13711117208003998,
      "learning_rate": 3.720650734054378e-05,
      "loss": 0.3746,
      "step": 890
    },
    {
      "epoch": 0.5616799968480359,
      "grad_norm": 0.1615670621395111,
      "learning_rate": 3.7189462879953035e-05,
      "loss": 0.4717,
      "step": 891
    },
    {
      "epoch": 0.5623103896615579,
      "grad_norm": 0.17041678726673126,
      "learning_rate": 3.717240371499545e-05,
      "loss": 0.4522,
      "step": 892
    },
    {
      "epoch": 0.5629407824750798,
      "grad_norm": 0.14367422461509705,
      "learning_rate": 3.715532986274748e-05,
      "loss": 0.3958,
      "step": 893
    },
    {
      "epoch": 0.5635711752886017,
      "grad_norm": 0.1326402723789215,
      "learning_rate": 3.713824134030031e-05,
      "loss": 0.3662,
      "step": 894
    },
    {
      "epoch": 0.5642015681021236,
      "grad_norm": 0.15629824995994568,
      "learning_rate": 3.7121138164759804e-05,
      "loss": 0.4241,
      "step": 895
    },
    {
      "epoch": 0.5648319609156456,
      "grad_norm": 0.14097222685813904,
      "learning_rate": 3.710402035324648e-05,
      "loss": 0.3501,
      "step": 896
    },
    {
      "epoch": 0.5654623537291675,
      "grad_norm": 0.1376744508743286,
      "learning_rate": 3.7086887922895525e-05,
      "loss": 0.3405,
      "step": 897
    },
    {
      "epoch": 0.5660927465426894,
      "grad_norm": 0.1320733278989792,
      "learning_rate": 3.706974089085675e-05,
      "loss": 0.3882,
      "step": 898
    },
    {
      "epoch": 0.5667231393562113,
      "grad_norm": 0.15197254717350006,
      "learning_rate": 3.705257927429458e-05,
      "loss": 0.3226,
      "step": 899
    },
    {
      "epoch": 0.5673535321697333,
      "grad_norm": 0.16653065383434296,
      "learning_rate": 3.703540309038805e-05,
      "loss": 0.4761,
      "step": 900
    },
    {
      "epoch": 0.5679839249832552,
      "grad_norm": 0.14119040966033936,
      "learning_rate": 3.7018212356330766e-05,
      "loss": 0.4173,
      "step": 901
    },
    {
      "epoch": 0.5686143177967771,
      "grad_norm": 0.12635064125061035,
      "learning_rate": 3.700100708933091e-05,
      "loss": 0.3189,
      "step": 902
    },
    {
      "epoch": 0.569244710610299,
      "grad_norm": 0.15768444538116455,
      "learning_rate": 3.698378730661121e-05,
      "loss": 0.4362,
      "step": 903
    },
    {
      "epoch": 0.569875103423821,
      "grad_norm": 0.1234264075756073,
      "learning_rate": 3.6966553025408906e-05,
      "loss": 0.3721,
      "step": 904
    },
    {
      "epoch": 0.5705054962373429,
      "grad_norm": 0.2133094221353531,
      "learning_rate": 3.694930426297577e-05,
      "loss": 0.5438,
      "step": 905
    },
    {
      "epoch": 0.5711358890508649,
      "grad_norm": 0.15444080531597137,
      "learning_rate": 3.693204103657807e-05,
      "loss": 0.4884,
      "step": 906
    },
    {
      "epoch": 0.5717662818643867,
      "grad_norm": 0.14413847029209137,
      "learning_rate": 3.691476336349654e-05,
      "loss": 0.3625,
      "step": 907
    },
    {
      "epoch": 0.5723966746779087,
      "grad_norm": 0.14719709753990173,
      "learning_rate": 3.689747126102639e-05,
      "loss": 0.4258,
      "step": 908
    },
    {
      "epoch": 0.5730270674914306,
      "grad_norm": 0.16312751173973083,
      "learning_rate": 3.6880164746477256e-05,
      "loss": 0.3715,
      "step": 909
    },
    {
      "epoch": 0.5736574603049526,
      "grad_norm": 0.15884438157081604,
      "learning_rate": 3.686284383717323e-05,
      "loss": 0.3683,
      "step": 910
    },
    {
      "epoch": 0.5742878531184744,
      "grad_norm": 0.1744864583015442,
      "learning_rate": 3.6845508550452784e-05,
      "loss": 0.4181,
      "step": 911
    },
    {
      "epoch": 0.5749182459319964,
      "grad_norm": 0.16491571068763733,
      "learning_rate": 3.682815890366879e-05,
      "loss": 0.4146,
      "step": 912
    },
    {
      "epoch": 0.5755486387455183,
      "grad_norm": 0.13504120707511902,
      "learning_rate": 3.6810794914188505e-05,
      "loss": 0.3924,
      "step": 913
    },
    {
      "epoch": 0.5761790315590403,
      "grad_norm": 0.16097284853458405,
      "learning_rate": 3.679341659939353e-05,
      "loss": 0.3781,
      "step": 914
    },
    {
      "epoch": 0.5768094243725621,
      "grad_norm": 0.1496012806892395,
      "learning_rate": 3.677602397667982e-05,
      "loss": 0.3979,
      "step": 915
    },
    {
      "epoch": 0.5774398171860841,
      "grad_norm": 0.14697562158107758,
      "learning_rate": 3.675861706345763e-05,
      "loss": 0.4282,
      "step": 916
    },
    {
      "epoch": 0.578070209999606,
      "grad_norm": 0.15116776525974274,
      "learning_rate": 3.674119587715156e-05,
      "loss": 0.3885,
      "step": 917
    },
    {
      "epoch": 0.578700602813128,
      "grad_norm": 0.20044632256031036,
      "learning_rate": 3.6723760435200446e-05,
      "loss": 0.5018,
      "step": 918
    },
    {
      "epoch": 0.5793309956266498,
      "grad_norm": 0.1666855663061142,
      "learning_rate": 3.670631075505743e-05,
      "loss": 0.4903,
      "step": 919
    },
    {
      "epoch": 0.5799613884401718,
      "grad_norm": 0.14125750958919525,
      "learning_rate": 3.66888468541899e-05,
      "loss": 0.4394,
      "step": 920
    },
    {
      "epoch": 0.5805917812536937,
      "grad_norm": 0.14239953458309174,
      "learning_rate": 3.6671368750079464e-05,
      "loss": 0.4216,
      "step": 921
    },
    {
      "epoch": 0.5812221740672157,
      "grad_norm": 0.15734338760375977,
      "learning_rate": 3.6653876460221975e-05,
      "loss": 0.4036,
      "step": 922
    },
    {
      "epoch": 0.5818525668807376,
      "grad_norm": 0.15128806233406067,
      "learning_rate": 3.6636370002127454e-05,
      "loss": 0.4421,
      "step": 923
    },
    {
      "epoch": 0.5824829596942595,
      "grad_norm": 0.1520044207572937,
      "learning_rate": 3.661884939332014e-05,
      "loss": 0.4424,
      "step": 924
    },
    {
      "epoch": 0.5831133525077814,
      "grad_norm": 0.1292216032743454,
      "learning_rate": 3.66013146513384e-05,
      "loss": 0.3393,
      "step": 925
    },
    {
      "epoch": 0.5837437453213034,
      "grad_norm": 0.13135753571987152,
      "learning_rate": 3.658376579373478e-05,
      "loss": 0.4032,
      "step": 926
    },
    {
      "epoch": 0.5843741381348253,
      "grad_norm": 0.15352147817611694,
      "learning_rate": 3.656620283807594e-05,
      "loss": 0.4581,
      "step": 927
    },
    {
      "epoch": 0.5850045309483471,
      "grad_norm": 0.131231427192688,
      "learning_rate": 3.654862580194265e-05,
      "loss": 0.3887,
      "step": 928
    },
    {
      "epoch": 0.5856349237618691,
      "grad_norm": 0.14569136500358582,
      "learning_rate": 3.6531034702929786e-05,
      "loss": 0.3301,
      "step": 929
    },
    {
      "epoch": 0.586265316575391,
      "grad_norm": 0.12439321726560593,
      "learning_rate": 3.651342955864629e-05,
      "loss": 0.3648,
      "step": 930
    },
    {
      "epoch": 0.586895709388913,
      "grad_norm": 0.16859029233455658,
      "learning_rate": 3.6495810386715173e-05,
      "loss": 0.3427,
      "step": 931
    },
    {
      "epoch": 0.5875261022024348,
      "grad_norm": 0.15363626182079315,
      "learning_rate": 3.647817720477348e-05,
      "loss": 0.4368,
      "step": 932
    },
    {
      "epoch": 0.5881564950159568,
      "grad_norm": 0.15837733447551727,
      "learning_rate": 3.6460530030472284e-05,
      "loss": 0.4757,
      "step": 933
    },
    {
      "epoch": 0.5887868878294787,
      "grad_norm": 0.13571250438690186,
      "learning_rate": 3.644286888147666e-05,
      "loss": 0.4153,
      "step": 934
    },
    {
      "epoch": 0.5894172806430007,
      "grad_norm": 0.15629878640174866,
      "learning_rate": 3.642519377546568e-05,
      "loss": 0.3564,
      "step": 935
    },
    {
      "epoch": 0.5900476734565226,
      "grad_norm": 0.18650545179843903,
      "learning_rate": 3.640750473013239e-05,
      "loss": 0.453,
      "step": 936
    },
    {
      "epoch": 0.5906780662700445,
      "grad_norm": 0.12352810800075531,
      "learning_rate": 3.638980176318378e-05,
      "loss": 0.3529,
      "step": 937
    },
    {
      "epoch": 0.5913084590835664,
      "grad_norm": 0.15796920657157898,
      "learning_rate": 3.637208489234077e-05,
      "loss": 0.4279,
      "step": 938
    },
    {
      "epoch": 0.5919388518970884,
      "grad_norm": 0.12148329615592957,
      "learning_rate": 3.635435413533821e-05,
      "loss": 0.329,
      "step": 939
    },
    {
      "epoch": 0.5925692447106103,
      "grad_norm": 0.13628233969211578,
      "learning_rate": 3.633660950992485e-05,
      "loss": 0.3973,
      "step": 940
    },
    {
      "epoch": 0.5931996375241322,
      "grad_norm": 0.16014662384986877,
      "learning_rate": 3.631885103386331e-05,
      "loss": 0.4577,
      "step": 941
    },
    {
      "epoch": 0.5938300303376541,
      "grad_norm": 0.15572448074817657,
      "learning_rate": 3.63010787249301e-05,
      "loss": 0.3086,
      "step": 942
    },
    {
      "epoch": 0.5944604231511761,
      "grad_norm": 0.13830454647541046,
      "learning_rate": 3.6283292600915554e-05,
      "loss": 0.3489,
      "step": 943
    },
    {
      "epoch": 0.595090815964698,
      "grad_norm": 0.19943471252918243,
      "learning_rate": 3.626549267962383e-05,
      "loss": 0.4587,
      "step": 944
    },
    {
      "epoch": 0.5957212087782199,
      "grad_norm": 0.17432811856269836,
      "learning_rate": 3.624767897887293e-05,
      "loss": 0.4835,
      "step": 945
    },
    {
      "epoch": 0.5963516015917418,
      "grad_norm": 0.16056682169437408,
      "learning_rate": 3.622985151649461e-05,
      "loss": 0.4116,
      "step": 946
    },
    {
      "epoch": 0.5969819944052638,
      "grad_norm": 0.1521116942167282,
      "learning_rate": 3.621201031033443e-05,
      "loss": 0.3751,
      "step": 947
    },
    {
      "epoch": 0.5976123872187857,
      "grad_norm": 0.15403449535369873,
      "learning_rate": 3.6194155378251694e-05,
      "loss": 0.4144,
      "step": 948
    },
    {
      "epoch": 0.5982427800323077,
      "grad_norm": 0.15799763798713684,
      "learning_rate": 3.617628673811945e-05,
      "loss": 0.3858,
      "step": 949
    },
    {
      "epoch": 0.5988731728458295,
      "grad_norm": 0.1562921553850174,
      "learning_rate": 3.615840440782447e-05,
      "loss": 0.3842,
      "step": 950
    },
    {
      "epoch": 0.5995035656593515,
      "grad_norm": 0.1535552740097046,
      "learning_rate": 3.6140508405267236e-05,
      "loss": 0.4072,
      "step": 951
    },
    {
      "epoch": 0.6001339584728734,
      "grad_norm": 0.14467324316501617,
      "learning_rate": 3.612259874836188e-05,
      "loss": 0.3785,
      "step": 952
    },
    {
      "epoch": 0.6007643512863954,
      "grad_norm": 0.17959435284137726,
      "learning_rate": 3.610467545503627e-05,
      "loss": 0.4085,
      "step": 953
    },
    {
      "epoch": 0.6013947440999172,
      "grad_norm": 0.1359519064426422,
      "learning_rate": 3.608673854323185e-05,
      "loss": 0.3582,
      "step": 954
    },
    {
      "epoch": 0.6020251369134392,
      "grad_norm": 0.1733047515153885,
      "learning_rate": 3.606878803090375e-05,
      "loss": 0.403,
      "step": 955
    },
    {
      "epoch": 0.6026555297269611,
      "grad_norm": 0.1633727103471756,
      "learning_rate": 3.6050823936020684e-05,
      "loss": 0.3801,
      "step": 956
    },
    {
      "epoch": 0.6032859225404831,
      "grad_norm": 0.1312204748392105,
      "learning_rate": 3.603284627656498e-05,
      "loss": 0.35,
      "step": 957
    },
    {
      "epoch": 0.6039163153540049,
      "grad_norm": 0.15054136514663696,
      "learning_rate": 3.601485507053254e-05,
      "loss": 0.3937,
      "step": 958
    },
    {
      "epoch": 0.6045467081675269,
      "grad_norm": 0.14886727929115295,
      "learning_rate": 3.5996850335932803e-05,
      "loss": 0.3875,
      "step": 959
    },
    {
      "epoch": 0.6051771009810488,
      "grad_norm": 0.15318524837493896,
      "learning_rate": 3.59788320907888e-05,
      "loss": 0.4236,
      "step": 960
    },
    {
      "epoch": 0.6058074937945708,
      "grad_norm": 0.12440776824951172,
      "learning_rate": 3.5960800353137044e-05,
      "loss": 0.4105,
      "step": 961
    },
    {
      "epoch": 0.6064378866080926,
      "grad_norm": 0.13105151057243347,
      "learning_rate": 3.594275514102757e-05,
      "loss": 0.3824,
      "step": 962
    },
    {
      "epoch": 0.6070682794216146,
      "grad_norm": 0.14685268700122833,
      "learning_rate": 3.59246964725239e-05,
      "loss": 0.4375,
      "step": 963
    },
    {
      "epoch": 0.6076986722351365,
      "grad_norm": 0.13163559138774872,
      "learning_rate": 3.5906624365703024e-05,
      "loss": 0.3601,
      "step": 964
    },
    {
      "epoch": 0.6083290650486585,
      "grad_norm": 0.17752410471439362,
      "learning_rate": 3.588853883865538e-05,
      "loss": 0.4102,
      "step": 965
    },
    {
      "epoch": 0.6089594578621804,
      "grad_norm": 0.13089518249034882,
      "learning_rate": 3.587043990948485e-05,
      "loss": 0.4087,
      "step": 966
    },
    {
      "epoch": 0.6095898506757023,
      "grad_norm": 0.13238897919654846,
      "learning_rate": 3.5852327596308724e-05,
      "loss": 0.4105,
      "step": 967
    },
    {
      "epoch": 0.6102202434892242,
      "grad_norm": 0.17239437997341156,
      "learning_rate": 3.583420191725769e-05,
      "loss": 0.4815,
      "step": 968
    },
    {
      "epoch": 0.6108506363027462,
      "grad_norm": 0.1444268673658371,
      "learning_rate": 3.5816062890475824e-05,
      "loss": 0.4018,
      "step": 969
    },
    {
      "epoch": 0.6114810291162681,
      "grad_norm": 0.14324289560317993,
      "learning_rate": 3.579791053412056e-05,
      "loss": 0.3432,
      "step": 970
    },
    {
      "epoch": 0.61211142192979,
      "grad_norm": 0.17291143536567688,
      "learning_rate": 3.5779744866362645e-05,
      "loss": 0.4084,
      "step": 971
    },
    {
      "epoch": 0.6127418147433119,
      "grad_norm": 0.1407811939716339,
      "learning_rate": 3.576156590538621e-05,
      "loss": 0.3445,
      "step": 972
    },
    {
      "epoch": 0.6133722075568339,
      "grad_norm": 0.158295676112175,
      "learning_rate": 3.5743373669388644e-05,
      "loss": 0.3933,
      "step": 973
    },
    {
      "epoch": 0.6140026003703558,
      "grad_norm": 0.1508656144142151,
      "learning_rate": 3.572516817658065e-05,
      "loss": 0.4026,
      "step": 974
    },
    {
      "epoch": 0.6146329931838777,
      "grad_norm": 0.12717203795909882,
      "learning_rate": 3.570694944518618e-05,
      "loss": 0.395,
      "step": 975
    },
    {
      "epoch": 0.6152633859973996,
      "grad_norm": 0.14242561161518097,
      "learning_rate": 3.568871749344246e-05,
      "loss": 0.347,
      "step": 976
    },
    {
      "epoch": 0.6158937788109216,
      "grad_norm": 0.15714222192764282,
      "learning_rate": 3.5670472339599944e-05,
      "loss": 0.4108,
      "step": 977
    },
    {
      "epoch": 0.6165241716244435,
      "grad_norm": 0.18858179450035095,
      "learning_rate": 3.565221400192227e-05,
      "loss": 0.4192,
      "step": 978
    },
    {
      "epoch": 0.6171545644379655,
      "grad_norm": 0.12744100391864777,
      "learning_rate": 3.5633942498686344e-05,
      "loss": 0.4141,
      "step": 979
    },
    {
      "epoch": 0.6177849572514873,
      "grad_norm": 0.12514697015285492,
      "learning_rate": 3.5615657848182175e-05,
      "loss": 0.3693,
      "step": 980
    },
    {
      "epoch": 0.6184153500650093,
      "grad_norm": 0.11576073616743088,
      "learning_rate": 3.5597360068712976e-05,
      "loss": 0.3417,
      "step": 981
    },
    {
      "epoch": 0.6190457428785312,
      "grad_norm": 0.14353668689727783,
      "learning_rate": 3.557904917859508e-05,
      "loss": 0.368,
      "step": 982
    },
    {
      "epoch": 0.6196761356920532,
      "grad_norm": 0.15207786858081818,
      "learning_rate": 3.556072519615797e-05,
      "loss": 0.3417,
      "step": 983
    },
    {
      "epoch": 0.620306528505575,
      "grad_norm": 0.22169116139411926,
      "learning_rate": 3.554238813974423e-05,
      "loss": 0.5027,
      "step": 984
    },
    {
      "epoch": 0.620936921319097,
      "grad_norm": 0.15463010966777802,
      "learning_rate": 3.5524038027709495e-05,
      "loss": 0.3696,
      "step": 985
    },
    {
      "epoch": 0.6215673141326189,
      "grad_norm": 0.19080068171024323,
      "learning_rate": 3.550567487842251e-05,
      "loss": 0.4981,
      "step": 986
    },
    {
      "epoch": 0.6221977069461408,
      "grad_norm": 0.14745832979679108,
      "learning_rate": 3.548729871026505e-05,
      "loss": 0.3288,
      "step": 987
    },
    {
      "epoch": 0.6228280997596627,
      "grad_norm": 0.16360951960086823,
      "learning_rate": 3.5468909541631925e-05,
      "loss": 0.4344,
      "step": 988
    },
    {
      "epoch": 0.6234584925731846,
      "grad_norm": 0.14530690014362335,
      "learning_rate": 3.545050739093097e-05,
      "loss": 0.3505,
      "step": 989
    },
    {
      "epoch": 0.6240888853867066,
      "grad_norm": 0.1620369851589203,
      "learning_rate": 3.5432092276583004e-05,
      "loss": 0.3794,
      "step": 990
    },
    {
      "epoch": 0.6247192782002285,
      "grad_norm": 0.15294252336025238,
      "learning_rate": 3.541366421702183e-05,
      "loss": 0.4205,
      "step": 991
    },
    {
      "epoch": 0.6253496710137505,
      "grad_norm": 0.1411382108926773,
      "learning_rate": 3.5395223230694186e-05,
      "loss": 0.4292,
      "step": 992
    },
    {
      "epoch": 0.6259800638272723,
      "grad_norm": 0.15150560438632965,
      "learning_rate": 3.537676933605979e-05,
      "loss": 0.3929,
      "step": 993
    },
    {
      "epoch": 0.6266104566407943,
      "grad_norm": 0.15619826316833496,
      "learning_rate": 3.535830255159125e-05,
      "loss": 0.3956,
      "step": 994
    },
    {
      "epoch": 0.6272408494543162,
      "grad_norm": 0.1517878621816635,
      "learning_rate": 3.533982289577409e-05,
      "loss": 0.3902,
      "step": 995
    },
    {
      "epoch": 0.6278712422678382,
      "grad_norm": 0.13279902935028076,
      "learning_rate": 3.5321330387106714e-05,
      "loss": 0.4051,
      "step": 996
    },
    {
      "epoch": 0.62850163508136,
      "grad_norm": 0.15745455026626587,
      "learning_rate": 3.5302825044100396e-05,
      "loss": 0.4214,
      "step": 997
    },
    {
      "epoch": 0.629132027894882,
      "grad_norm": 0.13961806893348694,
      "learning_rate": 3.528430688527925e-05,
      "loss": 0.3664,
      "step": 998
    },
    {
      "epoch": 0.6297624207084039,
      "grad_norm": 0.18050052225589752,
      "learning_rate": 3.5265775929180224e-05,
      "loss": 0.505,
      "step": 999
    },
    {
      "epoch": 0.6303928135219259,
      "grad_norm": 0.13805602490901947,
      "learning_rate": 3.5247232194353076e-05,
      "loss": 0.3696,
      "step": 1000
    },
    {
      "epoch": 0.6303928135219259,
      "eval_loss": 0.4341304302215576,
      "eval_runtime": 222.67,
      "eval_samples_per_second": 4.491,
      "eval_steps_per_second": 4.491,
      "step": 1000
    },
    {
      "epoch": 0.6310232063354477,
      "grad_norm": 0.1473533809185028,
      "learning_rate": 3.522867569936036e-05,
      "loss": 0.3384,
      "step": 1001
    },
    {
      "epoch": 0.6316535991489697,
      "grad_norm": 0.17155957221984863,
      "learning_rate": 3.5210106462777386e-05,
      "loss": 0.4029,
      "step": 1002
    },
    {
      "epoch": 0.6322839919624916,
      "grad_norm": 0.1242462620139122,
      "learning_rate": 3.519152450319225e-05,
      "loss": 0.3844,
      "step": 1003
    },
    {
      "epoch": 0.6329143847760136,
      "grad_norm": 0.15569186210632324,
      "learning_rate": 3.517292983920575e-05,
      "loss": 0.3944,
      "step": 1004
    },
    {
      "epoch": 0.6335447775895354,
      "grad_norm": 0.14550328254699707,
      "learning_rate": 3.5154322489431415e-05,
      "loss": 0.3772,
      "step": 1005
    },
    {
      "epoch": 0.6341751704030574,
      "grad_norm": 0.13598184287548065,
      "learning_rate": 3.5135702472495485e-05,
      "loss": 0.3739,
      "step": 1006
    },
    {
      "epoch": 0.6348055632165793,
      "grad_norm": 0.17206841707229614,
      "learning_rate": 3.511706980703686e-05,
      "loss": 0.4588,
      "step": 1007
    },
    {
      "epoch": 0.6354359560301013,
      "grad_norm": 0.1455783098936081,
      "learning_rate": 3.509842451170712e-05,
      "loss": 0.4551,
      "step": 1008
    },
    {
      "epoch": 0.6360663488436232,
      "grad_norm": 0.15823356807231903,
      "learning_rate": 3.507976660517047e-05,
      "loss": 0.3975,
      "step": 1009
    },
    {
      "epoch": 0.6366967416571451,
      "grad_norm": 0.1431056261062622,
      "learning_rate": 3.506109610610374e-05,
      "loss": 0.4013,
      "step": 1010
    },
    {
      "epoch": 0.637327134470667,
      "grad_norm": 0.14573250710964203,
      "learning_rate": 3.504241303319639e-05,
      "loss": 0.4071,
      "step": 1011
    },
    {
      "epoch": 0.637957527284189,
      "grad_norm": 0.15625494718551636,
      "learning_rate": 3.5023717405150436e-05,
      "loss": 0.4716,
      "step": 1012
    },
    {
      "epoch": 0.6385879200977109,
      "grad_norm": 0.17605744302272797,
      "learning_rate": 3.500500924068048e-05,
      "loss": 0.4658,
      "step": 1013
    },
    {
      "epoch": 0.6392183129112328,
      "grad_norm": 0.1206241101026535,
      "learning_rate": 3.4986288558513665e-05,
      "loss": 0.3339,
      "step": 1014
    },
    {
      "epoch": 0.6398487057247547,
      "grad_norm": 0.15223607420921326,
      "learning_rate": 3.4967555377389665e-05,
      "loss": 0.4369,
      "step": 1015
    },
    {
      "epoch": 0.6404790985382767,
      "grad_norm": 0.14725369215011597,
      "learning_rate": 3.494880971606068e-05,
      "loss": 0.4466,
      "step": 1016
    },
    {
      "epoch": 0.6411094913517986,
      "grad_norm": 0.14819376170635223,
      "learning_rate": 3.493005159329137e-05,
      "loss": 0.3763,
      "step": 1017
    },
    {
      "epoch": 0.6417398841653205,
      "grad_norm": 0.13168439269065857,
      "learning_rate": 3.491128102785891e-05,
      "loss": 0.3794,
      "step": 1018
    },
    {
      "epoch": 0.6423702769788424,
      "grad_norm": 0.17571082711219788,
      "learning_rate": 3.48924980385529e-05,
      "loss": 0.4371,
      "step": 1019
    },
    {
      "epoch": 0.6430006697923644,
      "grad_norm": 0.154144749045372,
      "learning_rate": 3.487370264417539e-05,
      "loss": 0.4427,
      "step": 1020
    },
    {
      "epoch": 0.6436310626058863,
      "grad_norm": 0.2012409120798111,
      "learning_rate": 3.4854894863540834e-05,
      "loss": 0.5475,
      "step": 1021
    },
    {
      "epoch": 0.6442614554194083,
      "grad_norm": 0.14221590757369995,
      "learning_rate": 3.4836074715476105e-05,
      "loss": 0.475,
      "step": 1022
    },
    {
      "epoch": 0.6448918482329301,
      "grad_norm": 0.12871815264225006,
      "learning_rate": 3.481724221882045e-05,
      "loss": 0.3711,
      "step": 1023
    },
    {
      "epoch": 0.6455222410464521,
      "grad_norm": 0.16108739376068115,
      "learning_rate": 3.479839739242547e-05,
      "loss": 0.4116,
      "step": 1024
    },
    {
      "epoch": 0.646152633859974,
      "grad_norm": 0.16816940903663635,
      "learning_rate": 3.47795402551551e-05,
      "loss": 0.4328,
      "step": 1025
    },
    {
      "epoch": 0.646783026673496,
      "grad_norm": 0.156079962849617,
      "learning_rate": 3.476067082588562e-05,
      "loss": 0.388,
      "step": 1026
    },
    {
      "epoch": 0.6474134194870178,
      "grad_norm": 0.1362450271844864,
      "learning_rate": 3.474178912350561e-05,
      "loss": 0.4717,
      "step": 1027
    },
    {
      "epoch": 0.6480438123005398,
      "grad_norm": 0.14149360358715057,
      "learning_rate": 3.4722895166915925e-05,
      "loss": 0.3397,
      "step": 1028
    },
    {
      "epoch": 0.6486742051140617,
      "grad_norm": 0.141607403755188,
      "learning_rate": 3.470398897502968e-05,
      "loss": 0.4298,
      "step": 1029
    },
    {
      "epoch": 0.6493045979275837,
      "grad_norm": 0.1447194516658783,
      "learning_rate": 3.4685070566772265e-05,
      "loss": 0.4161,
      "step": 1030
    },
    {
      "epoch": 0.6499349907411055,
      "grad_norm": 0.16443675756454468,
      "learning_rate": 3.4666139961081275e-05,
      "loss": 0.4223,
      "step": 1031
    },
    {
      "epoch": 0.6505653835546275,
      "grad_norm": 0.1416953057050705,
      "learning_rate": 3.4647197176906525e-05,
      "loss": 0.4227,
      "step": 1032
    },
    {
      "epoch": 0.6511957763681494,
      "grad_norm": 0.14385829865932465,
      "learning_rate": 3.462824223321002e-05,
      "loss": 0.3944,
      "step": 1033
    },
    {
      "epoch": 0.6518261691816714,
      "grad_norm": 0.13178785145282745,
      "learning_rate": 3.460927514896593e-05,
      "loss": 0.3748,
      "step": 1034
    },
    {
      "epoch": 0.6524565619951933,
      "grad_norm": 0.17496028542518616,
      "learning_rate": 3.459029594316058e-05,
      "loss": 0.4521,
      "step": 1035
    },
    {
      "epoch": 0.6530869548087151,
      "grad_norm": 0.13025426864624023,
      "learning_rate": 3.457130463479245e-05,
      "loss": 0.3935,
      "step": 1036
    },
    {
      "epoch": 0.6537173476222371,
      "grad_norm": 0.17129521071910858,
      "learning_rate": 3.4552301242872095e-05,
      "loss": 0.4211,
      "step": 1037
    },
    {
      "epoch": 0.654347740435759,
      "grad_norm": 0.14599891006946564,
      "learning_rate": 3.4533285786422206e-05,
      "loss": 0.3644,
      "step": 1038
    },
    {
      "epoch": 0.654978133249281,
      "grad_norm": 0.14229898154735565,
      "learning_rate": 3.451425828447752e-05,
      "loss": 0.3399,
      "step": 1039
    },
    {
      "epoch": 0.6556085260628028,
      "grad_norm": 0.13468730449676514,
      "learning_rate": 3.4495218756084855e-05,
      "loss": 0.4602,
      "step": 1040
    },
    {
      "epoch": 0.6562389188763248,
      "grad_norm": 0.14791233837604523,
      "learning_rate": 3.447616722030305e-05,
      "loss": 0.4587,
      "step": 1041
    },
    {
      "epoch": 0.6568693116898467,
      "grad_norm": 0.1709253489971161,
      "learning_rate": 3.445710369620297e-05,
      "loss": 0.4111,
      "step": 1042
    },
    {
      "epoch": 0.6574997045033687,
      "grad_norm": 0.1572764813899994,
      "learning_rate": 3.443802820286748e-05,
      "loss": 0.4076,
      "step": 1043
    },
    {
      "epoch": 0.6581300973168905,
      "grad_norm": 0.1494017392396927,
      "learning_rate": 3.441894075939144e-05,
      "loss": 0.3271,
      "step": 1044
    },
    {
      "epoch": 0.6587604901304125,
      "grad_norm": 0.1699901521205902,
      "learning_rate": 3.439984138488164e-05,
      "loss": 0.3869,
      "step": 1045
    },
    {
      "epoch": 0.6593908829439344,
      "grad_norm": 0.15090258419513702,
      "learning_rate": 3.438073009845684e-05,
      "loss": 0.3876,
      "step": 1046
    },
    {
      "epoch": 0.6600212757574564,
      "grad_norm": 0.14121317863464355,
      "learning_rate": 3.4361606919247733e-05,
      "loss": 0.3792,
      "step": 1047
    },
    {
      "epoch": 0.6606516685709783,
      "grad_norm": 0.19512726366519928,
      "learning_rate": 3.434247186639688e-05,
      "loss": 0.4897,
      "step": 1048
    },
    {
      "epoch": 0.6612820613845002,
      "grad_norm": 0.15658080577850342,
      "learning_rate": 3.432332495905875e-05,
      "loss": 0.4556,
      "step": 1049
    },
    {
      "epoch": 0.6619124541980221,
      "grad_norm": 0.13722775876522064,
      "learning_rate": 3.4304166216399696e-05,
      "loss": 0.4073,
      "step": 1050
    },
    {
      "epoch": 0.6625428470115441,
      "grad_norm": 0.1241544634103775,
      "learning_rate": 3.428499565759787e-05,
      "loss": 0.3478,
      "step": 1051
    },
    {
      "epoch": 0.663173239825066,
      "grad_norm": 0.1600281447172165,
      "learning_rate": 3.426581330184331e-05,
      "loss": 0.4351,
      "step": 1052
    },
    {
      "epoch": 0.6638036326385879,
      "grad_norm": 0.14616727828979492,
      "learning_rate": 3.4246619168337814e-05,
      "loss": 0.4346,
      "step": 1053
    },
    {
      "epoch": 0.6644340254521098,
      "grad_norm": 0.1428317278623581,
      "learning_rate": 3.4227413276295e-05,
      "loss": 0.4193,
      "step": 1054
    },
    {
      "epoch": 0.6650644182656318,
      "grad_norm": 0.15601269900798798,
      "learning_rate": 3.4208195644940246e-05,
      "loss": 0.3534,
      "step": 1055
    },
    {
      "epoch": 0.6656948110791537,
      "grad_norm": 0.12326395511627197,
      "learning_rate": 3.418896629351067e-05,
      "loss": 0.3143,
      "step": 1056
    },
    {
      "epoch": 0.6663252038926756,
      "grad_norm": 0.15271161496639252,
      "learning_rate": 3.416972524125515e-05,
      "loss": 0.4322,
      "step": 1057
    },
    {
      "epoch": 0.6669555967061975,
      "grad_norm": 0.15423282980918884,
      "learning_rate": 3.4150472507434245e-05,
      "loss": 0.3973,
      "step": 1058
    },
    {
      "epoch": 0.6675859895197195,
      "grad_norm": 0.14533115923404694,
      "learning_rate": 3.413120811132024e-05,
      "loss": 0.3989,
      "step": 1059
    },
    {
      "epoch": 0.6682163823332414,
      "grad_norm": 0.13678240776062012,
      "learning_rate": 3.411193207219705e-05,
      "loss": 0.4486,
      "step": 1060
    },
    {
      "epoch": 0.6688467751467633,
      "grad_norm": 0.17480261623859406,
      "learning_rate": 3.409264440936031e-05,
      "loss": 0.3908,
      "step": 1061
    },
    {
      "epoch": 0.6694771679602852,
      "grad_norm": 0.15733547508716583,
      "learning_rate": 3.407334514211722e-05,
      "loss": 0.3599,
      "step": 1062
    },
    {
      "epoch": 0.6701075607738072,
      "grad_norm": 0.14672769606113434,
      "learning_rate": 3.4054034289786643e-05,
      "loss": 0.3635,
      "step": 1063
    },
    {
      "epoch": 0.6707379535873291,
      "grad_norm": 0.1510699838399887,
      "learning_rate": 3.4034711871699026e-05,
      "loss": 0.4569,
      "step": 1064
    },
    {
      "epoch": 0.6713683464008511,
      "grad_norm": 0.13803254067897797,
      "learning_rate": 3.401537790719637e-05,
      "loss": 0.3643,
      "step": 1065
    },
    {
      "epoch": 0.6719987392143729,
      "grad_norm": 0.16187086701393127,
      "learning_rate": 3.3996032415632296e-05,
      "loss": 0.436,
      "step": 1066
    },
    {
      "epoch": 0.6726291320278949,
      "grad_norm": 0.16465528309345245,
      "learning_rate": 3.3976675416371874e-05,
      "loss": 0.413,
      "step": 1067
    },
    {
      "epoch": 0.6732595248414168,
      "grad_norm": 0.15643639862537384,
      "learning_rate": 3.395730692879178e-05,
      "loss": 0.469,
      "step": 1068
    },
    {
      "epoch": 0.6738899176549388,
      "grad_norm": 0.18642200529575348,
      "learning_rate": 3.393792697228014e-05,
      "loss": 0.4506,
      "step": 1069
    },
    {
      "epoch": 0.6745203104684606,
      "grad_norm": 0.1551331728696823,
      "learning_rate": 3.3918535566236585e-05,
      "loss": 0.451,
      "step": 1070
    },
    {
      "epoch": 0.6751507032819826,
      "grad_norm": 0.1358264684677124,
      "learning_rate": 3.3899132730072174e-05,
      "loss": 0.4065,
      "step": 1071
    },
    {
      "epoch": 0.6757810960955045,
      "grad_norm": 0.13260263204574585,
      "learning_rate": 3.387971848320944e-05,
      "loss": 0.3795,
      "step": 1072
    },
    {
      "epoch": 0.6764114889090265,
      "grad_norm": 0.13498187065124512,
      "learning_rate": 3.3860292845082335e-05,
      "loss": 0.4265,
      "step": 1073
    },
    {
      "epoch": 0.6770418817225483,
      "grad_norm": 0.15217430889606476,
      "learning_rate": 3.38408558351362e-05,
      "loss": 0.433,
      "step": 1074
    },
    {
      "epoch": 0.6776722745360703,
      "grad_norm": 0.14895272254943848,
      "learning_rate": 3.3821407472827764e-05,
      "loss": 0.3977,
      "step": 1075
    },
    {
      "epoch": 0.6783026673495922,
      "grad_norm": 0.14306458830833435,
      "learning_rate": 3.380194777762513e-05,
      "loss": 0.3816,
      "step": 1076
    },
    {
      "epoch": 0.6789330601631142,
      "grad_norm": 0.15742669999599457,
      "learning_rate": 3.378247676900772e-05,
      "loss": 0.4769,
      "step": 1077
    },
    {
      "epoch": 0.6795634529766361,
      "grad_norm": 0.1568499207496643,
      "learning_rate": 3.376299446646631e-05,
      "loss": 0.438,
      "step": 1078
    },
    {
      "epoch": 0.680193845790158,
      "grad_norm": 0.1393679827451706,
      "learning_rate": 3.374350088950297e-05,
      "loss": 0.4329,
      "step": 1079
    },
    {
      "epoch": 0.6808242386036799,
      "grad_norm": 0.14324618875980377,
      "learning_rate": 3.372399605763106e-05,
      "loss": 0.347,
      "step": 1080
    },
    {
      "epoch": 0.6814546314172019,
      "grad_norm": 0.14731600880622864,
      "learning_rate": 3.370447999037518e-05,
      "loss": 0.3856,
      "step": 1081
    },
    {
      "epoch": 0.6820850242307238,
      "grad_norm": 0.13627250492572784,
      "learning_rate": 3.3684952707271226e-05,
      "loss": 0.3602,
      "step": 1082
    },
    {
      "epoch": 0.6827154170442457,
      "grad_norm": 0.15852227807044983,
      "learning_rate": 3.366541422786627e-05,
      "loss": 0.3785,
      "step": 1083
    },
    {
      "epoch": 0.6833458098577676,
      "grad_norm": 0.16282619535923004,
      "learning_rate": 3.3645864571718634e-05,
      "loss": 0.3979,
      "step": 1084
    },
    {
      "epoch": 0.6839762026712896,
      "grad_norm": 0.14918841421604156,
      "learning_rate": 3.3626303758397795e-05,
      "loss": 0.4108,
      "step": 1085
    },
    {
      "epoch": 0.6846065954848115,
      "grad_norm": 0.16575288772583008,
      "learning_rate": 3.3606731807484426e-05,
      "loss": 0.4184,
      "step": 1086
    },
    {
      "epoch": 0.6852369882983333,
      "grad_norm": 0.1275065392255783,
      "learning_rate": 3.358714873857031e-05,
      "loss": 0.4038,
      "step": 1087
    },
    {
      "epoch": 0.6858673811118553,
      "grad_norm": 0.14004617929458618,
      "learning_rate": 3.356755457125843e-05,
      "loss": 0.4864,
      "step": 1088
    },
    {
      "epoch": 0.6864977739253773,
      "grad_norm": 0.1653287708759308,
      "learning_rate": 3.354794932516279e-05,
      "loss": 0.4619,
      "step": 1089
    },
    {
      "epoch": 0.6871281667388992,
      "grad_norm": 0.15715204179286957,
      "learning_rate": 3.3528333019908546e-05,
      "loss": 0.4256,
      "step": 1090
    },
    {
      "epoch": 0.6877585595524212,
      "grad_norm": 0.1482582837343216,
      "learning_rate": 3.350870567513191e-05,
      "loss": 0.3792,
      "step": 1091
    },
    {
      "epoch": 0.688388952365943,
      "grad_norm": 0.14185462892055511,
      "learning_rate": 3.3489067310480124e-05,
      "loss": 0.3564,
      "step": 1092
    },
    {
      "epoch": 0.689019345179465,
      "grad_norm": 0.14399947226047516,
      "learning_rate": 3.34694179456115e-05,
      "loss": 0.4025,
      "step": 1093
    },
    {
      "epoch": 0.6896497379929869,
      "grad_norm": 0.17147956788539886,
      "learning_rate": 3.3449757600195324e-05,
      "loss": 0.4628,
      "step": 1094
    },
    {
      "epoch": 0.6902801308065089,
      "grad_norm": 0.18218888342380524,
      "learning_rate": 3.343008629391189e-05,
      "loss": 0.4818,
      "step": 1095
    },
    {
      "epoch": 0.6909105236200307,
      "grad_norm": 0.1327589452266693,
      "learning_rate": 3.341040404645247e-05,
      "loss": 0.4052,
      "step": 1096
    },
    {
      "epoch": 0.6915409164335526,
      "grad_norm": 0.1555880606174469,
      "learning_rate": 3.339071087751928e-05,
      "loss": 0.4009,
      "step": 1097
    },
    {
      "epoch": 0.6921713092470746,
      "grad_norm": 0.14132007956504822,
      "learning_rate": 3.337100680682546e-05,
      "loss": 0.3736,
      "step": 1098
    },
    {
      "epoch": 0.6928017020605965,
      "grad_norm": 0.12532183527946472,
      "learning_rate": 3.335129185409509e-05,
      "loss": 0.3825,
      "step": 1099
    },
    {
      "epoch": 0.6934320948741184,
      "grad_norm": 0.1209065243601799,
      "learning_rate": 3.3331566039063104e-05,
      "loss": 0.3834,
      "step": 1100
    },
    {
      "epoch": 0.6940624876876403,
      "grad_norm": 0.15299884974956512,
      "learning_rate": 3.331182938147534e-05,
      "loss": 0.3554,
      "step": 1101
    },
    {
      "epoch": 0.6946928805011623,
      "grad_norm": 0.15046365559101105,
      "learning_rate": 3.329208190108849e-05,
      "loss": 0.358,
      "step": 1102
    },
    {
      "epoch": 0.6953232733146842,
      "grad_norm": 0.1376636028289795,
      "learning_rate": 3.327232361767006e-05,
      "loss": 0.374,
      "step": 1103
    },
    {
      "epoch": 0.6959536661282061,
      "grad_norm": 0.17495515942573547,
      "learning_rate": 3.3252554550998376e-05,
      "loss": 0.4924,
      "step": 1104
    },
    {
      "epoch": 0.696584058941728,
      "grad_norm": 0.1590355485677719,
      "learning_rate": 3.323277472086258e-05,
      "loss": 0.4235,
      "step": 1105
    },
    {
      "epoch": 0.69721445175525,
      "grad_norm": 0.13074256479740143,
      "learning_rate": 3.3212984147062556e-05,
      "loss": 0.3432,
      "step": 1106
    },
    {
      "epoch": 0.6978448445687719,
      "grad_norm": 0.1273060142993927,
      "learning_rate": 3.319318284940896e-05,
      "loss": 0.3498,
      "step": 1107
    },
    {
      "epoch": 0.6984752373822939,
      "grad_norm": 0.142560213804245,
      "learning_rate": 3.3173370847723185e-05,
      "loss": 0.4031,
      "step": 1108
    },
    {
      "epoch": 0.6991056301958157,
      "grad_norm": 0.16981838643550873,
      "learning_rate": 3.315354816183732e-05,
      "loss": 0.4806,
      "step": 1109
    },
    {
      "epoch": 0.6997360230093377,
      "grad_norm": 0.12655171751976013,
      "learning_rate": 3.313371481159418e-05,
      "loss": 0.3589,
      "step": 1110
    },
    {
      "epoch": 0.7003664158228596,
      "grad_norm": 0.15416717529296875,
      "learning_rate": 3.3113870816847234e-05,
      "loss": 0.4265,
      "step": 1111
    },
    {
      "epoch": 0.7009968086363816,
      "grad_norm": 0.1576196402311325,
      "learning_rate": 3.309401619746061e-05,
      "loss": 0.3929,
      "step": 1112
    },
    {
      "epoch": 0.7016272014499034,
      "grad_norm": 0.142515167593956,
      "learning_rate": 3.307415097330906e-05,
      "loss": 0.3744,
      "step": 1113
    },
    {
      "epoch": 0.7022575942634254,
      "grad_norm": 0.12994253635406494,
      "learning_rate": 3.305427516427798e-05,
      "loss": 0.3346,
      "step": 1114
    },
    {
      "epoch": 0.7028879870769473,
      "grad_norm": 0.1487005203962326,
      "learning_rate": 3.303438879026334e-05,
      "loss": 0.3755,
      "step": 1115
    },
    {
      "epoch": 0.7035183798904693,
      "grad_norm": 0.13746176660060883,
      "learning_rate": 3.301449187117168e-05,
      "loss": 0.3748,
      "step": 1116
    },
    {
      "epoch": 0.7041487727039911,
      "grad_norm": 0.10952261835336685,
      "learning_rate": 3.299458442692014e-05,
      "loss": 0.3291,
      "step": 1117
    },
    {
      "epoch": 0.7047791655175131,
      "grad_norm": 0.14707626402378082,
      "learning_rate": 3.297466647743631e-05,
      "loss": 0.376,
      "step": 1118
    },
    {
      "epoch": 0.705409558331035,
      "grad_norm": 0.15899558365345,
      "learning_rate": 3.295473804265839e-05,
      "loss": 0.4085,
      "step": 1119
    },
    {
      "epoch": 0.706039951144557,
      "grad_norm": 0.1395518183708191,
      "learning_rate": 3.293479914253502e-05,
      "loss": 0.3351,
      "step": 1120
    },
    {
      "epoch": 0.7066703439580789,
      "grad_norm": 0.13215899467468262,
      "learning_rate": 3.291484979702533e-05,
      "loss": 0.3844,
      "step": 1121
    },
    {
      "epoch": 0.7073007367716008,
      "grad_norm": 0.12787476181983948,
      "learning_rate": 3.28948900260989e-05,
      "loss": 0.4299,
      "step": 1122
    },
    {
      "epoch": 0.7079311295851227,
      "grad_norm": 0.1378302276134491,
      "learning_rate": 3.2874919849735766e-05,
      "loss": 0.3924,
      "step": 1123
    },
    {
      "epoch": 0.7085615223986447,
      "grad_norm": 0.18550285696983337,
      "learning_rate": 3.285493928792636e-05,
      "loss": 0.5406,
      "step": 1124
    },
    {
      "epoch": 0.7091919152121666,
      "grad_norm": 0.15087875723838806,
      "learning_rate": 3.283494836067152e-05,
      "loss": 0.4411,
      "step": 1125
    },
    {
      "epoch": 0.7098223080256885,
      "grad_norm": 0.14296133816242218,
      "learning_rate": 3.281494708798245e-05,
      "loss": 0.42,
      "step": 1126
    },
    {
      "epoch": 0.7104527008392104,
      "grad_norm": 0.16558092832565308,
      "learning_rate": 3.2794935489880715e-05,
      "loss": 0.4505,
      "step": 1127
    },
    {
      "epoch": 0.7110830936527324,
      "grad_norm": 0.14892105758190155,
      "learning_rate": 3.277491358639822e-05,
      "loss": 0.4671,
      "step": 1128
    },
    {
      "epoch": 0.7117134864662543,
      "grad_norm": 0.14008575677871704,
      "learning_rate": 3.275488139757718e-05,
      "loss": 0.3752,
      "step": 1129
    },
    {
      "epoch": 0.7123438792797762,
      "grad_norm": 0.12162633240222931,
      "learning_rate": 3.2734838943470116e-05,
      "loss": 0.3555,
      "step": 1130
    },
    {
      "epoch": 0.7129742720932981,
      "grad_norm": 0.17720535397529602,
      "learning_rate": 3.271478624413981e-05,
      "loss": 0.4747,
      "step": 1131
    },
    {
      "epoch": 0.7136046649068201,
      "grad_norm": 0.13535228371620178,
      "learning_rate": 3.2694723319659304e-05,
      "loss": 0.4067,
      "step": 1132
    },
    {
      "epoch": 0.714235057720342,
      "grad_norm": 0.15453457832336426,
      "learning_rate": 3.267465019011189e-05,
      "loss": 0.4482,
      "step": 1133
    },
    {
      "epoch": 0.714865450533864,
      "grad_norm": 0.13449645042419434,
      "learning_rate": 3.2654566875591054e-05,
      "loss": 0.3486,
      "step": 1134
    },
    {
      "epoch": 0.7154958433473858,
      "grad_norm": 0.14120006561279297,
      "learning_rate": 3.2634473396200495e-05,
      "loss": 0.4511,
      "step": 1135
    },
    {
      "epoch": 0.7161262361609078,
      "grad_norm": 0.13961990177631378,
      "learning_rate": 3.261436977205407e-05,
      "loss": 0.3933,
      "step": 1136
    },
    {
      "epoch": 0.7167566289744297,
      "grad_norm": 0.1289156675338745,
      "learning_rate": 3.259425602327581e-05,
      "loss": 0.3766,
      "step": 1137
    },
    {
      "epoch": 0.7173870217879517,
      "grad_norm": 0.15146243572235107,
      "learning_rate": 3.2574132169999856e-05,
      "loss": 0.4708,
      "step": 1138
    },
    {
      "epoch": 0.7180174146014735,
      "grad_norm": 0.1951146125793457,
      "learning_rate": 3.25539982323705e-05,
      "loss": 0.4631,
      "step": 1139
    },
    {
      "epoch": 0.7186478074149955,
      "grad_norm": 0.17794398963451385,
      "learning_rate": 3.253385423054209e-05,
      "loss": 0.4093,
      "step": 1140
    },
    {
      "epoch": 0.7192782002285174,
      "grad_norm": 0.13891717791557312,
      "learning_rate": 3.251370018467908e-05,
      "loss": 0.3927,
      "step": 1141
    },
    {
      "epoch": 0.7199085930420394,
      "grad_norm": 0.15158259868621826,
      "learning_rate": 3.249353611495595e-05,
      "loss": 0.4271,
      "step": 1142
    },
    {
      "epoch": 0.7205389858555612,
      "grad_norm": 0.14888045191764832,
      "learning_rate": 3.247336204155724e-05,
      "loss": 0.3671,
      "step": 1143
    },
    {
      "epoch": 0.7211693786690831,
      "grad_norm": 0.15289460122585297,
      "learning_rate": 3.245317798467748e-05,
      "loss": 0.4726,
      "step": 1144
    },
    {
      "epoch": 0.7217997714826051,
      "grad_norm": 0.15524503588676453,
      "learning_rate": 3.243298396452122e-05,
      "loss": 0.4251,
      "step": 1145
    },
    {
      "epoch": 0.722430164296127,
      "grad_norm": 0.1572655588388443,
      "learning_rate": 3.241278000130297e-05,
      "loss": 0.3803,
      "step": 1146
    },
    {
      "epoch": 0.723060557109649,
      "grad_norm": 0.15959100425243378,
      "learning_rate": 3.2392566115247166e-05,
      "loss": 0.3879,
      "step": 1147
    },
    {
      "epoch": 0.7236909499231708,
      "grad_norm": 0.1515599489212036,
      "learning_rate": 3.2372342326588234e-05,
      "loss": 0.3734,
      "step": 1148
    },
    {
      "epoch": 0.7243213427366928,
      "grad_norm": 0.15384040772914886,
      "learning_rate": 3.235210865557047e-05,
      "loss": 0.4155,
      "step": 1149
    },
    {
      "epoch": 0.7249517355502147,
      "grad_norm": 0.13184258341789246,
      "learning_rate": 3.233186512244807e-05,
      "loss": 0.3707,
      "step": 1150
    },
    {
      "epoch": 0.7255821283637367,
      "grad_norm": 0.1614120900630951,
      "learning_rate": 3.23116117474851e-05,
      "loss": 0.3841,
      "step": 1151
    },
    {
      "epoch": 0.7262125211772585,
      "grad_norm": 0.16377252340316772,
      "learning_rate": 3.2291348550955506e-05,
      "loss": 0.4358,
      "step": 1152
    },
    {
      "epoch": 0.7268429139907805,
      "grad_norm": 0.17715105414390564,
      "learning_rate": 3.227107555314303e-05,
      "loss": 0.4311,
      "step": 1153
    },
    {
      "epoch": 0.7274733068043024,
      "grad_norm": 0.12353774160146713,
      "learning_rate": 3.2250792774341235e-05,
      "loss": 0.3998,
      "step": 1154
    },
    {
      "epoch": 0.7281036996178244,
      "grad_norm": 0.18568025529384613,
      "learning_rate": 3.223050023485349e-05,
      "loss": 0.3944,
      "step": 1155
    },
    {
      "epoch": 0.7287340924313462,
      "grad_norm": 0.14664076268672943,
      "learning_rate": 3.221019795499291e-05,
      "loss": 0.4321,
      "step": 1156
    },
    {
      "epoch": 0.7293644852448682,
      "grad_norm": 0.14460389316082,
      "learning_rate": 3.218988595508239e-05,
      "loss": 0.4517,
      "step": 1157
    },
    {
      "epoch": 0.7299948780583901,
      "grad_norm": 0.13419942557811737,
      "learning_rate": 3.2169564255454525e-05,
      "loss": 0.3954,
      "step": 1158
    },
    {
      "epoch": 0.7306252708719121,
      "grad_norm": 0.12872156500816345,
      "learning_rate": 3.2149232876451636e-05,
      "loss": 0.3859,
      "step": 1159
    },
    {
      "epoch": 0.7312556636854339,
      "grad_norm": 0.17439115047454834,
      "learning_rate": 3.212889183842573e-05,
      "loss": 0.4542,
      "step": 1160
    },
    {
      "epoch": 0.7318860564989559,
      "grad_norm": 0.1659138798713684,
      "learning_rate": 3.210854116173849e-05,
      "loss": 0.4502,
      "step": 1161
    },
    {
      "epoch": 0.7325164493124778,
      "grad_norm": 0.1282382756471634,
      "learning_rate": 3.208818086676123e-05,
      "loss": 0.345,
      "step": 1162
    },
    {
      "epoch": 0.7331468421259998,
      "grad_norm": 0.14103926718235016,
      "learning_rate": 3.206781097387492e-05,
      "loss": 0.3867,
      "step": 1163
    },
    {
      "epoch": 0.7337772349395217,
      "grad_norm": 0.13906313478946686,
      "learning_rate": 3.2047431503470094e-05,
      "loss": 0.423,
      "step": 1164
    },
    {
      "epoch": 0.7344076277530436,
      "grad_norm": 0.18526320159435272,
      "learning_rate": 3.2027042475946924e-05,
      "loss": 0.4396,
      "step": 1165
    },
    {
      "epoch": 0.7350380205665655,
      "grad_norm": 0.12118617445230484,
      "learning_rate": 3.200664391171511e-05,
      "loss": 0.3873,
      "step": 1166
    },
    {
      "epoch": 0.7356684133800875,
      "grad_norm": 0.1466676890850067,
      "learning_rate": 3.198623583119392e-05,
      "loss": 0.3754,
      "step": 1167
    },
    {
      "epoch": 0.7362988061936094,
      "grad_norm": 0.1386253833770752,
      "learning_rate": 3.196581825481213e-05,
      "loss": 0.4402,
      "step": 1168
    },
    {
      "epoch": 0.7369291990071313,
      "grad_norm": 0.14989392459392548,
      "learning_rate": 3.194539120300804e-05,
      "loss": 0.3425,
      "step": 1169
    },
    {
      "epoch": 0.7375595918206532,
      "grad_norm": 0.14861762523651123,
      "learning_rate": 3.192495469622943e-05,
      "loss": 0.3832,
      "step": 1170
    },
    {
      "epoch": 0.7381899846341752,
      "grad_norm": 0.14428891241550446,
      "learning_rate": 3.190450875493354e-05,
      "loss": 0.4072,
      "step": 1171
    },
    {
      "epoch": 0.7388203774476971,
      "grad_norm": 0.12648852169513702,
      "learning_rate": 3.1884053399587054e-05,
      "loss": 0.3947,
      "step": 1172
    },
    {
      "epoch": 0.739450770261219,
      "grad_norm": 0.1434190273284912,
      "learning_rate": 3.186358865066608e-05,
      "loss": 0.3826,
      "step": 1173
    },
    {
      "epoch": 0.7400811630747409,
      "grad_norm": 0.1685863435268402,
      "learning_rate": 3.184311452865613e-05,
      "loss": 0.4157,
      "step": 1174
    },
    {
      "epoch": 0.7407115558882629,
      "grad_norm": 0.15299765765666962,
      "learning_rate": 3.182263105405211e-05,
      "loss": 0.3789,
      "step": 1175
    },
    {
      "epoch": 0.7413419487017848,
      "grad_norm": 0.11780162155628204,
      "learning_rate": 3.1802138247358255e-05,
      "loss": 0.3864,
      "step": 1176
    },
    {
      "epoch": 0.7419723415153068,
      "grad_norm": 0.1341855227947235,
      "learning_rate": 3.1781636129088186e-05,
      "loss": 0.3601,
      "step": 1177
    },
    {
      "epoch": 0.7426027343288286,
      "grad_norm": 0.14695020020008087,
      "learning_rate": 3.176112471976481e-05,
      "loss": 0.4271,
      "step": 1178
    },
    {
      "epoch": 0.7432331271423506,
      "grad_norm": 0.15160372853279114,
      "learning_rate": 3.1740604039920354e-05,
      "loss": 0.4123,
      "step": 1179
    },
    {
      "epoch": 0.7438635199558725,
      "grad_norm": 0.14544130861759186,
      "learning_rate": 3.172007411009631e-05,
      "loss": 0.4237,
      "step": 1180
    },
    {
      "epoch": 0.7444939127693945,
      "grad_norm": 0.1619604080915451,
      "learning_rate": 3.169953495084345e-05,
      "loss": 0.4467,
      "step": 1181
    },
    {
      "epoch": 0.7451243055829163,
      "grad_norm": 0.1389506459236145,
      "learning_rate": 3.167898658272176e-05,
      "loss": 0.3846,
      "step": 1182
    },
    {
      "epoch": 0.7457546983964383,
      "grad_norm": 0.14674296975135803,
      "learning_rate": 3.165842902630047e-05,
      "loss": 0.3836,
      "step": 1183
    },
    {
      "epoch": 0.7463850912099602,
      "grad_norm": 0.1308245211839676,
      "learning_rate": 3.163786230215798e-05,
      "loss": 0.3929,
      "step": 1184
    },
    {
      "epoch": 0.7470154840234822,
      "grad_norm": 0.14882703125476837,
      "learning_rate": 3.161728643088189e-05,
      "loss": 0.4624,
      "step": 1185
    },
    {
      "epoch": 0.747645876837004,
      "grad_norm": 0.14521747827529907,
      "learning_rate": 3.159670143306893e-05,
      "loss": 0.3438,
      "step": 1186
    },
    {
      "epoch": 0.748276269650526,
      "grad_norm": 0.14668028056621552,
      "learning_rate": 3.157610732932502e-05,
      "loss": 0.3262,
      "step": 1187
    },
    {
      "epoch": 0.7489066624640479,
      "grad_norm": 0.13926194608211517,
      "learning_rate": 3.155550414026512e-05,
      "loss": 0.3905,
      "step": 1188
    },
    {
      "epoch": 0.7495370552775699,
      "grad_norm": 0.19401229918003082,
      "learning_rate": 3.153489188651335e-05,
      "loss": 0.5161,
      "step": 1189
    },
    {
      "epoch": 0.7501674480910918,
      "grad_norm": 0.19589434564113617,
      "learning_rate": 3.151427058870287e-05,
      "loss": 0.557,
      "step": 1190
    },
    {
      "epoch": 0.7507978409046137,
      "grad_norm": 0.12772412598133087,
      "learning_rate": 3.1493640267475883e-05,
      "loss": 0.393,
      "step": 1191
    },
    {
      "epoch": 0.7514282337181356,
      "grad_norm": 0.15101544559001923,
      "learning_rate": 3.147300094348366e-05,
      "loss": 0.3938,
      "step": 1192
    },
    {
      "epoch": 0.7520586265316576,
      "grad_norm": 0.12661443650722504,
      "learning_rate": 3.1452352637386465e-05,
      "loss": 0.3683,
      "step": 1193
    },
    {
      "epoch": 0.7526890193451795,
      "grad_norm": 0.14246030151844025,
      "learning_rate": 3.1431695369853554e-05,
      "loss": 0.4059,
      "step": 1194
    },
    {
      "epoch": 0.7533194121587014,
      "grad_norm": 0.14124950766563416,
      "learning_rate": 3.1411029161563146e-05,
      "loss": 0.3998,
      "step": 1195
    },
    {
      "epoch": 0.7539498049722233,
      "grad_norm": 0.14489232003688812,
      "learning_rate": 3.139035403320242e-05,
      "loss": 0.3704,
      "step": 1196
    },
    {
      "epoch": 0.7545801977857453,
      "grad_norm": 0.16528914868831635,
      "learning_rate": 3.1369670005467486e-05,
      "loss": 0.3777,
      "step": 1197
    },
    {
      "epoch": 0.7552105905992672,
      "grad_norm": 0.1324455440044403,
      "learning_rate": 3.1348977099063354e-05,
      "loss": 0.3648,
      "step": 1198
    },
    {
      "epoch": 0.755840983412789,
      "grad_norm": 0.14387087523937225,
      "learning_rate": 3.132827533470393e-05,
      "loss": 0.4182,
      "step": 1199
    },
    {
      "epoch": 0.756471376226311,
      "grad_norm": 0.17798291146755219,
      "learning_rate": 3.130756473311197e-05,
      "loss": 0.4266,
      "step": 1200
    },
    {
      "epoch": 0.756471376226311,
      "eval_loss": 0.42969560623168945,
      "eval_runtime": 222.3502,
      "eval_samples_per_second": 4.497,
      "eval_steps_per_second": 4.497,
      "step": 1200
    },
    {
      "epoch": 0.757101769039833,
      "grad_norm": 0.13976815342903137,
      "learning_rate": 3.12868453150191e-05,
      "loss": 0.4242,
      "step": 1201
    },
    {
      "epoch": 0.7577321618533549,
      "grad_norm": 0.16970422863960266,
      "learning_rate": 3.1266117101165755e-05,
      "loss": 0.3967,
      "step": 1202
    },
    {
      "epoch": 0.7583625546668767,
      "grad_norm": 0.14190421998500824,
      "learning_rate": 3.124538011230119e-05,
      "loss": 0.4174,
      "step": 1203
    },
    {
      "epoch": 0.7589929474803987,
      "grad_norm": 0.18532924354076385,
      "learning_rate": 3.1224634369183415e-05,
      "loss": 0.4296,
      "step": 1204
    },
    {
      "epoch": 0.7596233402939206,
      "grad_norm": 0.14926187694072723,
      "learning_rate": 3.120387989257924e-05,
      "loss": 0.3885,
      "step": 1205
    },
    {
      "epoch": 0.7602537331074426,
      "grad_norm": 0.15586240589618683,
      "learning_rate": 3.118311670326418e-05,
      "loss": 0.4022,
      "step": 1206
    },
    {
      "epoch": 0.7608841259209645,
      "grad_norm": 0.15296033024787903,
      "learning_rate": 3.116234482202252e-05,
      "loss": 0.384,
      "step": 1207
    },
    {
      "epoch": 0.7615145187344864,
      "grad_norm": 0.13187125325202942,
      "learning_rate": 3.114156426964719e-05,
      "loss": 0.399,
      "step": 1208
    },
    {
      "epoch": 0.7621449115480083,
      "grad_norm": 0.15693902969360352,
      "learning_rate": 3.1120775066939835e-05,
      "loss": 0.418,
      "step": 1209
    },
    {
      "epoch": 0.7627753043615303,
      "grad_norm": 0.13276666402816772,
      "learning_rate": 3.1099977234710743e-05,
      "loss": 0.4256,
      "step": 1210
    },
    {
      "epoch": 0.7634056971750522,
      "grad_norm": 0.16410839557647705,
      "learning_rate": 3.107917079377886e-05,
      "loss": 0.4195,
      "step": 1211
    },
    {
      "epoch": 0.7640360899885741,
      "grad_norm": 0.1418546438217163,
      "learning_rate": 3.105835576497174e-05,
      "loss": 0.3913,
      "step": 1212
    },
    {
      "epoch": 0.764666482802096,
      "grad_norm": 0.1336127668619156,
      "learning_rate": 3.103753216912552e-05,
      "loss": 0.4342,
      "step": 1213
    },
    {
      "epoch": 0.765296875615618,
      "grad_norm": 0.14342859387397766,
      "learning_rate": 3.101670002708492e-05,
      "loss": 0.3532,
      "step": 1214
    },
    {
      "epoch": 0.7659272684291399,
      "grad_norm": 0.13100619614124298,
      "learning_rate": 3.0995859359703234e-05,
      "loss": 0.3356,
      "step": 1215
    },
    {
      "epoch": 0.7665576612426618,
      "grad_norm": 0.18905891478061676,
      "learning_rate": 3.0975010187842256e-05,
      "loss": 0.4619,
      "step": 1216
    },
    {
      "epoch": 0.7671880540561837,
      "grad_norm": 0.16561256349086761,
      "learning_rate": 3.0954152532372335e-05,
      "loss": 0.4726,
      "step": 1217
    },
    {
      "epoch": 0.7678184468697057,
      "grad_norm": 0.14073455333709717,
      "learning_rate": 3.093328641417226e-05,
      "loss": 0.3954,
      "step": 1218
    },
    {
      "epoch": 0.7684488396832276,
      "grad_norm": 0.15636882185935974,
      "learning_rate": 3.0912411854129344e-05,
      "loss": 0.4532,
      "step": 1219
    },
    {
      "epoch": 0.7690792324967496,
      "grad_norm": 0.14515434205532074,
      "learning_rate": 3.089152887313931e-05,
      "loss": 0.3979,
      "step": 1220
    },
    {
      "epoch": 0.7697096253102714,
      "grad_norm": 0.15273843705654144,
      "learning_rate": 3.0870637492106346e-05,
      "loss": 0.4146,
      "step": 1221
    },
    {
      "epoch": 0.7703400181237934,
      "grad_norm": 0.14053373038768768,
      "learning_rate": 3.084973773194301e-05,
      "loss": 0.4118,
      "step": 1222
    },
    {
      "epoch": 0.7709704109373153,
      "grad_norm": 0.16905413568019867,
      "learning_rate": 3.0828829613570275e-05,
      "loss": 0.4402,
      "step": 1223
    },
    {
      "epoch": 0.7716008037508373,
      "grad_norm": 0.1700507253408432,
      "learning_rate": 3.080791315791747e-05,
      "loss": 0.3915,
      "step": 1224
    },
    {
      "epoch": 0.7722311965643591,
      "grad_norm": 0.1725083589553833,
      "learning_rate": 3.0786988385922274e-05,
      "loss": 0.4382,
      "step": 1225
    },
    {
      "epoch": 0.7728615893778811,
      "grad_norm": 0.12850132584571838,
      "learning_rate": 3.07660553185307e-05,
      "loss": 0.4087,
      "step": 1226
    },
    {
      "epoch": 0.773491982191403,
      "grad_norm": 0.14398561418056488,
      "learning_rate": 3.0745113976697036e-05,
      "loss": 0.4638,
      "step": 1227
    },
    {
      "epoch": 0.774122375004925,
      "grad_norm": 0.1383698284626007,
      "learning_rate": 3.0724164381383884e-05,
      "loss": 0.3122,
      "step": 1228
    },
    {
      "epoch": 0.7747527678184468,
      "grad_norm": 0.14155955612659454,
      "learning_rate": 3.070320655356209e-05,
      "loss": 0.3587,
      "step": 1229
    },
    {
      "epoch": 0.7753831606319688,
      "grad_norm": 0.12928208708763123,
      "learning_rate": 3.068224051421074e-05,
      "loss": 0.3619,
      "step": 1230
    },
    {
      "epoch": 0.7760135534454907,
      "grad_norm": 0.16862641274929047,
      "learning_rate": 3.0661266284317164e-05,
      "loss": 0.4479,
      "step": 1231
    },
    {
      "epoch": 0.7766439462590127,
      "grad_norm": 0.14298008382320404,
      "learning_rate": 3.064028388487685e-05,
      "loss": 0.349,
      "step": 1232
    },
    {
      "epoch": 0.7772743390725346,
      "grad_norm": 0.14893986284732819,
      "learning_rate": 3.06192933368935e-05,
      "loss": 0.3044,
      "step": 1233
    },
    {
      "epoch": 0.7779047318860565,
      "grad_norm": 0.16283908486366272,
      "learning_rate": 3.059829466137895e-05,
      "loss": 0.506,
      "step": 1234
    },
    {
      "epoch": 0.7785351246995784,
      "grad_norm": 0.14608779549598694,
      "learning_rate": 3.057728787935319e-05,
      "loss": 0.4079,
      "step": 1235
    },
    {
      "epoch": 0.7791655175131004,
      "grad_norm": 0.14887581765651703,
      "learning_rate": 3.055627301184431e-05,
      "loss": 0.3728,
      "step": 1236
    },
    {
      "epoch": 0.7797959103266223,
      "grad_norm": 0.15198326110839844,
      "learning_rate": 3.053525007988851e-05,
      "loss": 0.4171,
      "step": 1237
    },
    {
      "epoch": 0.7804263031401442,
      "grad_norm": 0.14626415073871613,
      "learning_rate": 3.0514219104530034e-05,
      "loss": 0.3743,
      "step": 1238
    },
    {
      "epoch": 0.7810566959536661,
      "grad_norm": 0.15413986146450043,
      "learning_rate": 3.049318010682121e-05,
      "loss": 0.4569,
      "step": 1239
    },
    {
      "epoch": 0.7816870887671881,
      "grad_norm": 0.16661998629570007,
      "learning_rate": 3.047213310782237e-05,
      "loss": 0.3922,
      "step": 1240
    },
    {
      "epoch": 0.78231748158071,
      "grad_norm": 0.14461380243301392,
      "learning_rate": 3.0451078128601875e-05,
      "loss": 0.3893,
      "step": 1241
    },
    {
      "epoch": 0.7829478743942319,
      "grad_norm": 0.15647661685943604,
      "learning_rate": 3.0430015190236064e-05,
      "loss": 0.4276,
      "step": 1242
    },
    {
      "epoch": 0.7835782672077538,
      "grad_norm": 0.1723053753376007,
      "learning_rate": 3.0408944313809254e-05,
      "loss": 0.4251,
      "step": 1243
    },
    {
      "epoch": 0.7842086600212758,
      "grad_norm": 0.1330244243144989,
      "learning_rate": 3.0387865520413694e-05,
      "loss": 0.3697,
      "step": 1244
    },
    {
      "epoch": 0.7848390528347977,
      "grad_norm": 0.1832405924797058,
      "learning_rate": 3.0366778831149553e-05,
      "loss": 0.5055,
      "step": 1245
    },
    {
      "epoch": 0.7854694456483197,
      "grad_norm": 0.1423068344593048,
      "learning_rate": 3.0345684267124928e-05,
      "loss": 0.3877,
      "step": 1246
    },
    {
      "epoch": 0.7860998384618415,
      "grad_norm": 0.13884073495864868,
      "learning_rate": 3.0324581849455786e-05,
      "loss": 0.3952,
      "step": 1247
    },
    {
      "epoch": 0.7867302312753635,
      "grad_norm": 0.14646825194358826,
      "learning_rate": 3.030347159926595e-05,
      "loss": 0.3548,
      "step": 1248
    },
    {
      "epoch": 0.7873606240888854,
      "grad_norm": 0.13451075553894043,
      "learning_rate": 3.028235353768709e-05,
      "loss": 0.4441,
      "step": 1249
    },
    {
      "epoch": 0.7879910169024074,
      "grad_norm": 0.13031382858753204,
      "learning_rate": 3.0261227685858693e-05,
      "loss": 0.415,
      "step": 1250
    },
    {
      "epoch": 0.7886214097159292,
      "grad_norm": 0.16893799602985382,
      "learning_rate": 3.024009406492805e-05,
      "loss": 0.4956,
      "step": 1251
    },
    {
      "epoch": 0.7892518025294512,
      "grad_norm": 0.14744479954242706,
      "learning_rate": 3.0218952696050224e-05,
      "loss": 0.386,
      "step": 1252
    },
    {
      "epoch": 0.7898821953429731,
      "grad_norm": 0.12958194315433502,
      "learning_rate": 3.019780360038802e-05,
      "loss": 0.374,
      "step": 1253
    },
    {
      "epoch": 0.790512588156495,
      "grad_norm": 0.14195840060710907,
      "learning_rate": 3.0176646799112015e-05,
      "loss": 0.3455,
      "step": 1254
    },
    {
      "epoch": 0.7911429809700169,
      "grad_norm": 0.1527252197265625,
      "learning_rate": 3.015548231340046e-05,
      "loss": 0.4011,
      "step": 1255
    },
    {
      "epoch": 0.7917733737835388,
      "grad_norm": 0.14155863225460052,
      "learning_rate": 3.0134310164439315e-05,
      "loss": 0.3637,
      "step": 1256
    },
    {
      "epoch": 0.7924037665970608,
      "grad_norm": 0.16283807158470154,
      "learning_rate": 3.0113130373422213e-05,
      "loss": 0.4651,
      "step": 1257
    },
    {
      "epoch": 0.7930341594105828,
      "grad_norm": 0.14141714572906494,
      "learning_rate": 3.0091942961550433e-05,
      "loss": 0.3961,
      "step": 1258
    },
    {
      "epoch": 0.7936645522241046,
      "grad_norm": 0.19669808447360992,
      "learning_rate": 3.0070747950032894e-05,
      "loss": 0.3923,
      "step": 1259
    },
    {
      "epoch": 0.7942949450376265,
      "grad_norm": 0.16972172260284424,
      "learning_rate": 3.0049545360086087e-05,
      "loss": 0.4162,
      "step": 1260
    },
    {
      "epoch": 0.7949253378511485,
      "grad_norm": 0.1589929312467575,
      "learning_rate": 3.0028335212934143e-05,
      "loss": 0.4726,
      "step": 1261
    },
    {
      "epoch": 0.7955557306646704,
      "grad_norm": 0.154354989528656,
      "learning_rate": 3.0007117529808713e-05,
      "loss": 0.3982,
      "step": 1262
    },
    {
      "epoch": 0.7961861234781924,
      "grad_norm": 0.13349230587482452,
      "learning_rate": 2.998589233194901e-05,
      "loss": 0.3245,
      "step": 1263
    },
    {
      "epoch": 0.7968165162917142,
      "grad_norm": 0.12602095305919647,
      "learning_rate": 2.996465964060177e-05,
      "loss": 0.3663,
      "step": 1264
    },
    {
      "epoch": 0.7974469091052362,
      "grad_norm": 0.15197496116161346,
      "learning_rate": 2.994341947702122e-05,
      "loss": 0.4175,
      "step": 1265
    },
    {
      "epoch": 0.7980773019187581,
      "grad_norm": 0.14406231045722961,
      "learning_rate": 2.992217186246908e-05,
      "loss": 0.3619,
      "step": 1266
    },
    {
      "epoch": 0.7987076947322801,
      "grad_norm": 0.16416853666305542,
      "learning_rate": 2.9900916818214522e-05,
      "loss": 0.4152,
      "step": 1267
    },
    {
      "epoch": 0.7993380875458019,
      "grad_norm": 0.1644122153520584,
      "learning_rate": 2.9879654365534157e-05,
      "loss": 0.4807,
      "step": 1268
    },
    {
      "epoch": 0.7999684803593239,
      "grad_norm": 0.13329042494297028,
      "learning_rate": 2.9858384525712007e-05,
      "loss": 0.3947,
      "step": 1269
    },
    {
      "epoch": 0.8005988731728458,
      "grad_norm": 0.17329329252243042,
      "learning_rate": 2.983710732003949e-05,
      "loss": 0.5129,
      "step": 1270
    },
    {
      "epoch": 0.8012292659863678,
      "grad_norm": 0.20010189712047577,
      "learning_rate": 2.9815822769815405e-05,
      "loss": 0.4258,
      "step": 1271
    },
    {
      "epoch": 0.8018596587998896,
      "grad_norm": 0.1383606642484665,
      "learning_rate": 2.97945308963459e-05,
      "loss": 0.4292,
      "step": 1272
    },
    {
      "epoch": 0.8024900516134116,
      "grad_norm": 0.18742382526397705,
      "learning_rate": 2.977323172094444e-05,
      "loss": 0.4427,
      "step": 1273
    },
    {
      "epoch": 0.8031204444269335,
      "grad_norm": 0.12892279028892517,
      "learning_rate": 2.9751925264931825e-05,
      "loss": 0.3814,
      "step": 1274
    },
    {
      "epoch": 0.8037508372404555,
      "grad_norm": 0.13529056310653687,
      "learning_rate": 2.973061154963612e-05,
      "loss": 0.3902,
      "step": 1275
    },
    {
      "epoch": 0.8043812300539774,
      "grad_norm": 0.1281760334968567,
      "learning_rate": 2.9709290596392665e-05,
      "loss": 0.3686,
      "step": 1276
    },
    {
      "epoch": 0.8050116228674993,
      "grad_norm": 0.12750031054019928,
      "learning_rate": 2.9687962426544054e-05,
      "loss": 0.3661,
      "step": 1277
    },
    {
      "epoch": 0.8056420156810212,
      "grad_norm": 0.19774696230888367,
      "learning_rate": 2.966662706144008e-05,
      "loss": 0.4035,
      "step": 1278
    },
    {
      "epoch": 0.8062724084945432,
      "grad_norm": 0.1890394538640976,
      "learning_rate": 2.9645284522437767e-05,
      "loss": 0.4657,
      "step": 1279
    },
    {
      "epoch": 0.8069028013080651,
      "grad_norm": 0.1572059839963913,
      "learning_rate": 2.9623934830901302e-05,
      "loss": 0.3899,
      "step": 1280
    },
    {
      "epoch": 0.807533194121587,
      "grad_norm": 0.1299111694097519,
      "learning_rate": 2.960257800820205e-05,
      "loss": 0.3763,
      "step": 1281
    },
    {
      "epoch": 0.8081635869351089,
      "grad_norm": 0.17201797664165497,
      "learning_rate": 2.958121407571849e-05,
      "loss": 0.4203,
      "step": 1282
    },
    {
      "epoch": 0.8087939797486309,
      "grad_norm": 0.1822853982448578,
      "learning_rate": 2.9559843054836234e-05,
      "loss": 0.4675,
      "step": 1283
    },
    {
      "epoch": 0.8094243725621528,
      "grad_norm": 0.14285458624362946,
      "learning_rate": 2.953846496694799e-05,
      "loss": 0.381,
      "step": 1284
    },
    {
      "epoch": 0.8100547653756747,
      "grad_norm": 0.13120800256729126,
      "learning_rate": 2.951707983345353e-05,
      "loss": 0.351,
      "step": 1285
    },
    {
      "epoch": 0.8106851581891966,
      "grad_norm": 0.15498490631580353,
      "learning_rate": 2.949568767575969e-05,
      "loss": 0.4421,
      "step": 1286
    },
    {
      "epoch": 0.8113155510027186,
      "grad_norm": 0.13642612099647522,
      "learning_rate": 2.9474288515280327e-05,
      "loss": 0.4148,
      "step": 1287
    },
    {
      "epoch": 0.8119459438162405,
      "grad_norm": 0.14929735660552979,
      "learning_rate": 2.945288237343632e-05,
      "loss": 0.4098,
      "step": 1288
    },
    {
      "epoch": 0.8125763366297625,
      "grad_norm": 0.15047243237495422,
      "learning_rate": 2.9431469271655517e-05,
      "loss": 0.4056,
      "step": 1289
    },
    {
      "epoch": 0.8132067294432843,
      "grad_norm": 0.17344744503498077,
      "learning_rate": 2.9410049231372764e-05,
      "loss": 0.4745,
      "step": 1290
    },
    {
      "epoch": 0.8138371222568063,
      "grad_norm": 0.13470345735549927,
      "learning_rate": 2.9388622274029818e-05,
      "loss": 0.3921,
      "step": 1291
    },
    {
      "epoch": 0.8144675150703282,
      "grad_norm": 0.18539802730083466,
      "learning_rate": 2.936718842107539e-05,
      "loss": 0.4469,
      "step": 1292
    },
    {
      "epoch": 0.8150979078838502,
      "grad_norm": 0.1868763566017151,
      "learning_rate": 2.9345747693965062e-05,
      "loss": 0.4339,
      "step": 1293
    },
    {
      "epoch": 0.815728300697372,
      "grad_norm": 0.13025294244289398,
      "learning_rate": 2.9324300114161335e-05,
      "loss": 0.3505,
      "step": 1294
    },
    {
      "epoch": 0.816358693510894,
      "grad_norm": 0.14217989146709442,
      "learning_rate": 2.9302845703133535e-05,
      "loss": 0.482,
      "step": 1295
    },
    {
      "epoch": 0.8169890863244159,
      "grad_norm": 0.1494067758321762,
      "learning_rate": 2.9281384482357847e-05,
      "loss": 0.4544,
      "step": 1296
    },
    {
      "epoch": 0.8176194791379379,
      "grad_norm": 0.139066681265831,
      "learning_rate": 2.9259916473317262e-05,
      "loss": 0.3703,
      "step": 1297
    },
    {
      "epoch": 0.8182498719514597,
      "grad_norm": 0.1567106544971466,
      "learning_rate": 2.9238441697501573e-05,
      "loss": 0.4444,
      "step": 1298
    },
    {
      "epoch": 0.8188802647649817,
      "grad_norm": 0.12469413131475449,
      "learning_rate": 2.9216960176407353e-05,
      "loss": 0.3793,
      "step": 1299
    },
    {
      "epoch": 0.8195106575785036,
      "grad_norm": 0.15910683572292328,
      "learning_rate": 2.91954719315379e-05,
      "loss": 0.3861,
      "step": 1300
    },
    {
      "epoch": 0.8201410503920256,
      "grad_norm": 0.12822704017162323,
      "learning_rate": 2.917397698440328e-05,
      "loss": 0.337,
      "step": 1301
    },
    {
      "epoch": 0.8207714432055474,
      "grad_norm": 0.1806858628988266,
      "learning_rate": 2.915247535652023e-05,
      "loss": 0.4272,
      "step": 1302
    },
    {
      "epoch": 0.8214018360190694,
      "grad_norm": 0.1507682353258133,
      "learning_rate": 2.913096706941221e-05,
      "loss": 0.4612,
      "step": 1303
    },
    {
      "epoch": 0.8220322288325913,
      "grad_norm": 0.1620669811964035,
      "learning_rate": 2.9109452144609315e-05,
      "loss": 0.4811,
      "step": 1304
    },
    {
      "epoch": 0.8226626216461133,
      "grad_norm": 0.1458032876253128,
      "learning_rate": 2.908793060364832e-05,
      "loss": 0.4287,
      "step": 1305
    },
    {
      "epoch": 0.8232930144596352,
      "grad_norm": 0.14256137609481812,
      "learning_rate": 2.906640246807258e-05,
      "loss": 0.3712,
      "step": 1306
    },
    {
      "epoch": 0.823923407273157,
      "grad_norm": 0.1815001517534256,
      "learning_rate": 2.904486775943208e-05,
      "loss": 0.4335,
      "step": 1307
    },
    {
      "epoch": 0.824553800086679,
      "grad_norm": 0.18930688500404358,
      "learning_rate": 2.902332649928339e-05,
      "loss": 0.4208,
      "step": 1308
    },
    {
      "epoch": 0.825184192900201,
      "grad_norm": 0.15047843754291534,
      "learning_rate": 2.90017787091896e-05,
      "loss": 0.4442,
      "step": 1309
    },
    {
      "epoch": 0.8258145857137229,
      "grad_norm": 0.1436319202184677,
      "learning_rate": 2.8980224410720387e-05,
      "loss": 0.3839,
      "step": 1310
    },
    {
      "epoch": 0.8264449785272447,
      "grad_norm": 0.16568589210510254,
      "learning_rate": 2.895866362545191e-05,
      "loss": 0.4122,
      "step": 1311
    },
    {
      "epoch": 0.8270753713407667,
      "grad_norm": 0.13160791993141174,
      "learning_rate": 2.893709637496683e-05,
      "loss": 0.383,
      "step": 1312
    },
    {
      "epoch": 0.8277057641542886,
      "grad_norm": 0.12996850907802582,
      "learning_rate": 2.891552268085427e-05,
      "loss": 0.3561,
      "step": 1313
    },
    {
      "epoch": 0.8283361569678106,
      "grad_norm": 0.12997615337371826,
      "learning_rate": 2.8893942564709828e-05,
      "loss": 0.367,
      "step": 1314
    },
    {
      "epoch": 0.8289665497813324,
      "grad_norm": 0.1627953201532364,
      "learning_rate": 2.8872356048135515e-05,
      "loss": 0.4475,
      "step": 1315
    },
    {
      "epoch": 0.8295969425948544,
      "grad_norm": 0.15016132593154907,
      "learning_rate": 2.8850763152739736e-05,
      "loss": 0.3783,
      "step": 1316
    },
    {
      "epoch": 0.8302273354083763,
      "grad_norm": 0.14278893172740936,
      "learning_rate": 2.8829163900137306e-05,
      "loss": 0.3903,
      "step": 1317
    },
    {
      "epoch": 0.8308577282218983,
      "grad_norm": 0.13539397716522217,
      "learning_rate": 2.8807558311949387e-05,
      "loss": 0.3694,
      "step": 1318
    },
    {
      "epoch": 0.8314881210354202,
      "grad_norm": 0.15693554282188416,
      "learning_rate": 2.8785946409803485e-05,
      "loss": 0.4324,
      "step": 1319
    },
    {
      "epoch": 0.8321185138489421,
      "grad_norm": 0.13900567591190338,
      "learning_rate": 2.876432821533344e-05,
      "loss": 0.3951,
      "step": 1320
    },
    {
      "epoch": 0.832748906662464,
      "grad_norm": 0.15492331981658936,
      "learning_rate": 2.8742703750179375e-05,
      "loss": 0.4227,
      "step": 1321
    },
    {
      "epoch": 0.833379299475986,
      "grad_norm": 0.18184980750083923,
      "learning_rate": 2.8721073035987694e-05,
      "loss": 0.416,
      "step": 1322
    },
    {
      "epoch": 0.8340096922895079,
      "grad_norm": 0.16962742805480957,
      "learning_rate": 2.8699436094411058e-05,
      "loss": 0.4497,
      "step": 1323
    },
    {
      "epoch": 0.8346400851030298,
      "grad_norm": 0.1357308179140091,
      "learning_rate": 2.8677792947108362e-05,
      "loss": 0.3872,
      "step": 1324
    },
    {
      "epoch": 0.8352704779165517,
      "grad_norm": 0.13641200959682465,
      "learning_rate": 2.8656143615744708e-05,
      "loss": 0.3701,
      "step": 1325
    },
    {
      "epoch": 0.8359008707300737,
      "grad_norm": 0.15859255194664001,
      "learning_rate": 2.8634488121991412e-05,
      "loss": 0.3139,
      "step": 1326
    },
    {
      "epoch": 0.8365312635435956,
      "grad_norm": 0.1266074925661087,
      "learning_rate": 2.8612826487525913e-05,
      "loss": 0.3832,
      "step": 1327
    },
    {
      "epoch": 0.8371616563571175,
      "grad_norm": 0.16178956627845764,
      "learning_rate": 2.859115873403184e-05,
      "loss": 0.4407,
      "step": 1328
    },
    {
      "epoch": 0.8377920491706394,
      "grad_norm": 0.14203162491321564,
      "learning_rate": 2.8569484883198924e-05,
      "loss": 0.4151,
      "step": 1329
    },
    {
      "epoch": 0.8384224419841614,
      "grad_norm": 0.16555646061897278,
      "learning_rate": 2.854780495672301e-05,
      "loss": 0.3897,
      "step": 1330
    },
    {
      "epoch": 0.8390528347976833,
      "grad_norm": 0.11057502776384354,
      "learning_rate": 2.852611897630602e-05,
      "loss": 0.3404,
      "step": 1331
    },
    {
      "epoch": 0.8396832276112053,
      "grad_norm": 0.12985214591026306,
      "learning_rate": 2.850442696365594e-05,
      "loss": 0.3201,
      "step": 1332
    },
    {
      "epoch": 0.8403136204247271,
      "grad_norm": 0.15618446469306946,
      "learning_rate": 2.8482728940486774e-05,
      "loss": 0.4648,
      "step": 1333
    },
    {
      "epoch": 0.8409440132382491,
      "grad_norm": 0.1684415489435196,
      "learning_rate": 2.846102492851859e-05,
      "loss": 0.4042,
      "step": 1334
    },
    {
      "epoch": 0.841574406051771,
      "grad_norm": 0.16451717913150787,
      "learning_rate": 2.8439314949477393e-05,
      "loss": 0.4144,
      "step": 1335
    },
    {
      "epoch": 0.842204798865293,
      "grad_norm": 0.1372554451227188,
      "learning_rate": 2.841759902509521e-05,
      "loss": 0.3653,
      "step": 1336
    },
    {
      "epoch": 0.8428351916788148,
      "grad_norm": 0.14489951729774475,
      "learning_rate": 2.8395877177109986e-05,
      "loss": 0.3859,
      "step": 1337
    },
    {
      "epoch": 0.8434655844923368,
      "grad_norm": 0.13698627054691315,
      "learning_rate": 2.837414942726561e-05,
      "loss": 0.4311,
      "step": 1338
    },
    {
      "epoch": 0.8440959773058587,
      "grad_norm": 0.12621767818927765,
      "learning_rate": 2.835241579731188e-05,
      "loss": 0.3912,
      "step": 1339
    },
    {
      "epoch": 0.8447263701193807,
      "grad_norm": 0.18279710412025452,
      "learning_rate": 2.8330676309004474e-05,
      "loss": 0.4154,
      "step": 1340
    },
    {
      "epoch": 0.8453567629329025,
      "grad_norm": 0.1633724719285965,
      "learning_rate": 2.830893098410494e-05,
      "loss": 0.3936,
      "step": 1341
    },
    {
      "epoch": 0.8459871557464245,
      "grad_norm": 0.1703907698392868,
      "learning_rate": 2.8287179844380662e-05,
      "loss": 0.4769,
      "step": 1342
    },
    {
      "epoch": 0.8466175485599464,
      "grad_norm": 0.14323486387729645,
      "learning_rate": 2.826542291160485e-05,
      "loss": 0.3686,
      "step": 1343
    },
    {
      "epoch": 0.8472479413734684,
      "grad_norm": 0.1713230162858963,
      "learning_rate": 2.8243660207556516e-05,
      "loss": 0.3874,
      "step": 1344
    },
    {
      "epoch": 0.8478783341869903,
      "grad_norm": 0.1804395169019699,
      "learning_rate": 2.822189175402044e-05,
      "loss": 0.4208,
      "step": 1345
    },
    {
      "epoch": 0.8485087270005122,
      "grad_norm": 0.12683406472206116,
      "learning_rate": 2.820011757278716e-05,
      "loss": 0.34,
      "step": 1346
    },
    {
      "epoch": 0.8491391198140341,
      "grad_norm": 0.119718998670578,
      "learning_rate": 2.8178337685652956e-05,
      "loss": 0.4118,
      "step": 1347
    },
    {
      "epoch": 0.8497695126275561,
      "grad_norm": 0.13840489089488983,
      "learning_rate": 2.815655211441981e-05,
      "loss": 0.4252,
      "step": 1348
    },
    {
      "epoch": 0.850399905441078,
      "grad_norm": 0.13725785911083221,
      "learning_rate": 2.8134760880895384e-05,
      "loss": 0.4144,
      "step": 1349
    },
    {
      "epoch": 0.8510302982545999,
      "grad_norm": 0.147429421544075,
      "learning_rate": 2.8112964006893044e-05,
      "loss": 0.4391,
      "step": 1350
    },
    {
      "epoch": 0.8516606910681218,
      "grad_norm": 0.14347843825817108,
      "learning_rate": 2.8091161514231764e-05,
      "loss": 0.3921,
      "step": 1351
    },
    {
      "epoch": 0.8522910838816438,
      "grad_norm": 0.1464390754699707,
      "learning_rate": 2.806935342473616e-05,
      "loss": 0.3983,
      "step": 1352
    },
    {
      "epoch": 0.8529214766951657,
      "grad_norm": 0.1333981603384018,
      "learning_rate": 2.8047539760236457e-05,
      "loss": 0.3999,
      "step": 1353
    },
    {
      "epoch": 0.8535518695086876,
      "grad_norm": 0.142788827419281,
      "learning_rate": 2.8025720542568438e-05,
      "loss": 0.3837,
      "step": 1354
    },
    {
      "epoch": 0.8541822623222095,
      "grad_norm": 0.1281784325838089,
      "learning_rate": 2.8003895793573467e-05,
      "loss": 0.3303,
      "step": 1355
    },
    {
      "epoch": 0.8548126551357315,
      "grad_norm": 0.14220544695854187,
      "learning_rate": 2.7982065535098436e-05,
      "loss": 0.3639,
      "step": 1356
    },
    {
      "epoch": 0.8554430479492534,
      "grad_norm": 0.13342589139938354,
      "learning_rate": 2.796022978899575e-05,
      "loss": 0.357,
      "step": 1357
    },
    {
      "epoch": 0.8560734407627753,
      "grad_norm": 0.15548692643642426,
      "learning_rate": 2.7938388577123307e-05,
      "loss": 0.4021,
      "step": 1358
    },
    {
      "epoch": 0.8567038335762972,
      "grad_norm": 0.12782569229602814,
      "learning_rate": 2.791654192134449e-05,
      "loss": 0.3451,
      "step": 1359
    },
    {
      "epoch": 0.8573342263898192,
      "grad_norm": 0.15871256589889526,
      "learning_rate": 2.7894689843528103e-05,
      "loss": 0.3999,
      "step": 1360
    },
    {
      "epoch": 0.8579646192033411,
      "grad_norm": 0.14577703177928925,
      "learning_rate": 2.7872832365548415e-05,
      "loss": 0.4041,
      "step": 1361
    },
    {
      "epoch": 0.8585950120168631,
      "grad_norm": 0.1306508630514145,
      "learning_rate": 2.785096950928506e-05,
      "loss": 0.4234,
      "step": 1362
    },
    {
      "epoch": 0.8592254048303849,
      "grad_norm": 0.18583562970161438,
      "learning_rate": 2.7829101296623095e-05,
      "loss": 0.4297,
      "step": 1363
    },
    {
      "epoch": 0.8598557976439068,
      "grad_norm": 0.14241017401218414,
      "learning_rate": 2.780722774945291e-05,
      "loss": 0.3435,
      "step": 1364
    },
    {
      "epoch": 0.8604861904574288,
      "grad_norm": 0.18705232441425323,
      "learning_rate": 2.7785348889670246e-05,
      "loss": 0.4676,
      "step": 1365
    },
    {
      "epoch": 0.8611165832709508,
      "grad_norm": 0.1302051544189453,
      "learning_rate": 2.7763464739176167e-05,
      "loss": 0.3381,
      "step": 1366
    },
    {
      "epoch": 0.8617469760844726,
      "grad_norm": 0.1736343502998352,
      "learning_rate": 2.7741575319877023e-05,
      "loss": 0.4425,
      "step": 1367
    },
    {
      "epoch": 0.8623773688979945,
      "grad_norm": 0.14309068024158478,
      "learning_rate": 2.771968065368444e-05,
      "loss": 0.4022,
      "step": 1368
    },
    {
      "epoch": 0.8630077617115165,
      "grad_norm": 0.1872050166130066,
      "learning_rate": 2.7697780762515295e-05,
      "loss": 0.4522,
      "step": 1369
    },
    {
      "epoch": 0.8636381545250384,
      "grad_norm": 0.1302749216556549,
      "learning_rate": 2.7675875668291715e-05,
      "loss": 0.3819,
      "step": 1370
    },
    {
      "epoch": 0.8642685473385603,
      "grad_norm": 0.14372815191745758,
      "learning_rate": 2.7653965392940994e-05,
      "loss": 0.3799,
      "step": 1371
    },
    {
      "epoch": 0.8648989401520822,
      "grad_norm": 0.13147087395191193,
      "learning_rate": 2.763204995839567e-05,
      "loss": 0.3821,
      "step": 1372
    },
    {
      "epoch": 0.8655293329656042,
      "grad_norm": 0.16376769542694092,
      "learning_rate": 2.761012938659338e-05,
      "loss": 0.4577,
      "step": 1373
    },
    {
      "epoch": 0.8661597257791261,
      "grad_norm": 0.13191644847393036,
      "learning_rate": 2.758820369947696e-05,
      "loss": 0.3956,
      "step": 1374
    },
    {
      "epoch": 0.8667901185926481,
      "grad_norm": 0.14816711843013763,
      "learning_rate": 2.756627291899433e-05,
      "loss": 0.3705,
      "step": 1375
    },
    {
      "epoch": 0.8674205114061699,
      "grad_norm": 0.16423572599887848,
      "learning_rate": 2.7544337067098536e-05,
      "loss": 0.4717,
      "step": 1376
    },
    {
      "epoch": 0.8680509042196919,
      "grad_norm": 0.1640395224094391,
      "learning_rate": 2.752239616574767e-05,
      "loss": 0.4146,
      "step": 1377
    },
    {
      "epoch": 0.8686812970332138,
      "grad_norm": 0.15758387744426727,
      "learning_rate": 2.75004502369049e-05,
      "loss": 0.4276,
      "step": 1378
    },
    {
      "epoch": 0.8693116898467358,
      "grad_norm": 0.16378600895404816,
      "learning_rate": 2.7478499302538426e-05,
      "loss": 0.3942,
      "step": 1379
    },
    {
      "epoch": 0.8699420826602576,
      "grad_norm": 0.15028297901153564,
      "learning_rate": 2.7456543384621444e-05,
      "loss": 0.383,
      "step": 1380
    },
    {
      "epoch": 0.8705724754737796,
      "grad_norm": 0.16802988946437836,
      "learning_rate": 2.743458250513216e-05,
      "loss": 0.4308,
      "step": 1381
    },
    {
      "epoch": 0.8712028682873015,
      "grad_norm": 0.16376203298568726,
      "learning_rate": 2.7412616686053725e-05,
      "loss": 0.3934,
      "step": 1382
    },
    {
      "epoch": 0.8718332611008235,
      "grad_norm": 0.14017964899539948,
      "learning_rate": 2.739064594937425e-05,
      "loss": 0.3852,
      "step": 1383
    },
    {
      "epoch": 0.8724636539143453,
      "grad_norm": 0.1617506444454193,
      "learning_rate": 2.7368670317086766e-05,
      "loss": 0.4662,
      "step": 1384
    },
    {
      "epoch": 0.8730940467278673,
      "grad_norm": 0.14962440729141235,
      "learning_rate": 2.73466898111892e-05,
      "loss": 0.3929,
      "step": 1385
    },
    {
      "epoch": 0.8737244395413892,
      "grad_norm": 0.16706913709640503,
      "learning_rate": 2.7324704453684352e-05,
      "loss": 0.4208,
      "step": 1386
    },
    {
      "epoch": 0.8743548323549112,
      "grad_norm": 0.16016672551631927,
      "learning_rate": 2.7302714266579897e-05,
      "loss": 0.3574,
      "step": 1387
    },
    {
      "epoch": 0.8749852251684331,
      "grad_norm": 0.13596835732460022,
      "learning_rate": 2.7280719271888326e-05,
      "loss": 0.4341,
      "step": 1388
    },
    {
      "epoch": 0.875615617981955,
      "grad_norm": 0.11969844996929169,
      "learning_rate": 2.725871949162695e-05,
      "loss": 0.378,
      "step": 1389
    },
    {
      "epoch": 0.8762460107954769,
      "grad_norm": 0.13828162848949432,
      "learning_rate": 2.7236714947817875e-05,
      "loss": 0.3713,
      "step": 1390
    },
    {
      "epoch": 0.8768764036089989,
      "grad_norm": 0.15005554258823395,
      "learning_rate": 2.721470566248796e-05,
      "loss": 0.4491,
      "step": 1391
    },
    {
      "epoch": 0.8775067964225208,
      "grad_norm": 0.1482902467250824,
      "learning_rate": 2.7192691657668842e-05,
      "loss": 0.4512,
      "step": 1392
    },
    {
      "epoch": 0.8781371892360427,
      "grad_norm": 0.12331166118383408,
      "learning_rate": 2.7170672955396837e-05,
      "loss": 0.3675,
      "step": 1393
    },
    {
      "epoch": 0.8787675820495646,
      "grad_norm": 0.13660334050655365,
      "learning_rate": 2.7148649577713005e-05,
      "loss": 0.3699,
      "step": 1394
    },
    {
      "epoch": 0.8793979748630866,
      "grad_norm": 0.137289896607399,
      "learning_rate": 2.712662154666306e-05,
      "loss": 0.3468,
      "step": 1395
    },
    {
      "epoch": 0.8800283676766085,
      "grad_norm": 0.16336098313331604,
      "learning_rate": 2.7104588884297388e-05,
      "loss": 0.4172,
      "step": 1396
    },
    {
      "epoch": 0.8806587604901304,
      "grad_norm": 0.16273409128189087,
      "learning_rate": 2.7082551612671002e-05,
      "loss": 0.4059,
      "step": 1397
    },
    {
      "epoch": 0.8812891533036523,
      "grad_norm": 0.16388723254203796,
      "learning_rate": 2.7060509753843532e-05,
      "loss": 0.4598,
      "step": 1398
    },
    {
      "epoch": 0.8819195461171743,
      "grad_norm": 0.1399746984243393,
      "learning_rate": 2.7038463329879205e-05,
      "loss": 0.4212,
      "step": 1399
    },
    {
      "epoch": 0.8825499389306962,
      "grad_norm": 0.14388558268547058,
      "learning_rate": 2.7016412362846804e-05,
      "loss": 0.4263,
      "step": 1400
    },
    {
      "epoch": 0.8825499389306962,
      "eval_loss": 0.42592132091522217,
      "eval_runtime": 222.6108,
      "eval_samples_per_second": 4.492,
      "eval_steps_per_second": 4.492,
      "step": 1400
    },
    {
      "epoch": 0.8831803317442181,
      "grad_norm": 0.14138349890708923,
      "learning_rate": 2.6994356874819683e-05,
      "loss": 0.4132,
      "step": 1401
    },
    {
      "epoch": 0.88381072455774,
      "grad_norm": 0.1343000829219818,
      "learning_rate": 2.6972296887875695e-05,
      "loss": 0.3542,
      "step": 1402
    },
    {
      "epoch": 0.884441117371262,
      "grad_norm": 0.15190397202968597,
      "learning_rate": 2.6950232424097215e-05,
      "loss": 0.4265,
      "step": 1403
    },
    {
      "epoch": 0.8850715101847839,
      "grad_norm": 0.1640225201845169,
      "learning_rate": 2.6928163505571093e-05,
      "loss": 0.3829,
      "step": 1404
    },
    {
      "epoch": 0.8857019029983059,
      "grad_norm": 0.14996233582496643,
      "learning_rate": 2.690609015438864e-05,
      "loss": 0.3758,
      "step": 1405
    },
    {
      "epoch": 0.8863322958118277,
      "grad_norm": 0.1581767052412033,
      "learning_rate": 2.68840123926456e-05,
      "loss": 0.4537,
      "step": 1406
    },
    {
      "epoch": 0.8869626886253497,
      "grad_norm": 0.16030016541481018,
      "learning_rate": 2.6861930242442137e-05,
      "loss": 0.3675,
      "step": 1407
    },
    {
      "epoch": 0.8875930814388716,
      "grad_norm": 0.1415213793516159,
      "learning_rate": 2.6839843725882805e-05,
      "loss": 0.3931,
      "step": 1408
    },
    {
      "epoch": 0.8882234742523936,
      "grad_norm": 0.1286267340183258,
      "learning_rate": 2.6817752865076526e-05,
      "loss": 0.3673,
      "step": 1409
    },
    {
      "epoch": 0.8888538670659154,
      "grad_norm": 0.1294373869895935,
      "learning_rate": 2.6795657682136582e-05,
      "loss": 0.3691,
      "step": 1410
    },
    {
      "epoch": 0.8894842598794374,
      "grad_norm": 0.13593226671218872,
      "learning_rate": 2.6773558199180566e-05,
      "loss": 0.32,
      "step": 1411
    },
    {
      "epoch": 0.8901146526929593,
      "grad_norm": 0.1556108444929123,
      "learning_rate": 2.675145443833039e-05,
      "loss": 0.4227,
      "step": 1412
    },
    {
      "epoch": 0.8907450455064813,
      "grad_norm": 0.13537348806858063,
      "learning_rate": 2.6729346421712237e-05,
      "loss": 0.3302,
      "step": 1413
    },
    {
      "epoch": 0.8913754383200031,
      "grad_norm": 0.14231304824352264,
      "learning_rate": 2.6707234171456555e-05,
      "loss": 0.4624,
      "step": 1414
    },
    {
      "epoch": 0.892005831133525,
      "grad_norm": 0.16324934363365173,
      "learning_rate": 2.668511770969803e-05,
      "loss": 0.4407,
      "step": 1415
    },
    {
      "epoch": 0.892636223947047,
      "grad_norm": 0.146931454539299,
      "learning_rate": 2.666299705857556e-05,
      "loss": 0.4546,
      "step": 1416
    },
    {
      "epoch": 0.893266616760569,
      "grad_norm": 0.1308489888906479,
      "learning_rate": 2.6640872240232247e-05,
      "loss": 0.3906,
      "step": 1417
    },
    {
      "epoch": 0.8938970095740909,
      "grad_norm": 0.1489260494709015,
      "learning_rate": 2.661874327681536e-05,
      "loss": 0.3559,
      "step": 1418
    },
    {
      "epoch": 0.8945274023876127,
      "grad_norm": 0.14801530539989471,
      "learning_rate": 2.65966101904763e-05,
      "loss": 0.4261,
      "step": 1419
    },
    {
      "epoch": 0.8951577952011347,
      "grad_norm": 0.138169065117836,
      "learning_rate": 2.6574473003370612e-05,
      "loss": 0.3814,
      "step": 1420
    },
    {
      "epoch": 0.8957881880146567,
      "grad_norm": 0.15858194231987,
      "learning_rate": 2.6552331737657956e-05,
      "loss": 0.4008,
      "step": 1421
    },
    {
      "epoch": 0.8964185808281786,
      "grad_norm": 0.1418287456035614,
      "learning_rate": 2.6530186415502055e-05,
      "loss": 0.3684,
      "step": 1422
    },
    {
      "epoch": 0.8970489736417004,
      "grad_norm": 0.1921471655368805,
      "learning_rate": 2.6508037059070698e-05,
      "loss": 0.4454,
      "step": 1423
    },
    {
      "epoch": 0.8976793664552224,
      "grad_norm": 0.16956478357315063,
      "learning_rate": 2.6485883690535716e-05,
      "loss": 0.4696,
      "step": 1424
    },
    {
      "epoch": 0.8983097592687443,
      "grad_norm": 0.15770865976810455,
      "learning_rate": 2.6463726332072958e-05,
      "loss": 0.3732,
      "step": 1425
    },
    {
      "epoch": 0.8989401520822663,
      "grad_norm": 0.15551583468914032,
      "learning_rate": 2.6441565005862253e-05,
      "loss": 0.3584,
      "step": 1426
    },
    {
      "epoch": 0.8995705448957881,
      "grad_norm": 0.16392450034618378,
      "learning_rate": 2.6419399734087426e-05,
      "loss": 0.3833,
      "step": 1427
    },
    {
      "epoch": 0.9002009377093101,
      "grad_norm": 0.17206740379333496,
      "learning_rate": 2.6397230538936233e-05,
      "loss": 0.4071,
      "step": 1428
    },
    {
      "epoch": 0.900831330522832,
      "grad_norm": 0.13827435672283173,
      "learning_rate": 2.6375057442600353e-05,
      "loss": 0.3501,
      "step": 1429
    },
    {
      "epoch": 0.901461723336354,
      "grad_norm": 0.20176725089550018,
      "learning_rate": 2.635288046727539e-05,
      "loss": 0.4255,
      "step": 1430
    },
    {
      "epoch": 0.9020921161498759,
      "grad_norm": 0.16234435141086578,
      "learning_rate": 2.6330699635160814e-05,
      "loss": 0.4142,
      "step": 1431
    },
    {
      "epoch": 0.9027225089633978,
      "grad_norm": 0.1656573861837387,
      "learning_rate": 2.6308514968459968e-05,
      "loss": 0.4694,
      "step": 1432
    },
    {
      "epoch": 0.9033529017769197,
      "grad_norm": 0.16659680008888245,
      "learning_rate": 2.628632648938002e-05,
      "loss": 0.4834,
      "step": 1433
    },
    {
      "epoch": 0.9039832945904417,
      "grad_norm": 0.14273500442504883,
      "learning_rate": 2.626413422013197e-05,
      "loss": 0.3792,
      "step": 1434
    },
    {
      "epoch": 0.9046136874039636,
      "grad_norm": 0.15806609392166138,
      "learning_rate": 2.6241938182930593e-05,
      "loss": 0.3034,
      "step": 1435
    },
    {
      "epoch": 0.9052440802174855,
      "grad_norm": 0.14422351121902466,
      "learning_rate": 2.6219738399994453e-05,
      "loss": 0.4349,
      "step": 1436
    },
    {
      "epoch": 0.9058744730310074,
      "grad_norm": 0.1485350877046585,
      "learning_rate": 2.6197534893545858e-05,
      "loss": 0.3553,
      "step": 1437
    },
    {
      "epoch": 0.9065048658445294,
      "grad_norm": 0.14150355756282806,
      "learning_rate": 2.617532768581084e-05,
      "loss": 0.3638,
      "step": 1438
    },
    {
      "epoch": 0.9071352586580513,
      "grad_norm": 0.13242359459400177,
      "learning_rate": 2.6153116799019138e-05,
      "loss": 0.3722,
      "step": 1439
    },
    {
      "epoch": 0.9077656514715732,
      "grad_norm": 0.15447986125946045,
      "learning_rate": 2.6130902255404177e-05,
      "loss": 0.3917,
      "step": 1440
    },
    {
      "epoch": 0.9083960442850951,
      "grad_norm": 0.18113017082214355,
      "learning_rate": 2.6108684077203037e-05,
      "loss": 0.3853,
      "step": 1441
    },
    {
      "epoch": 0.9090264370986171,
      "grad_norm": 0.19897674024105072,
      "learning_rate": 2.6086462286656432e-05,
      "loss": 0.4622,
      "step": 1442
    },
    {
      "epoch": 0.909656829912139,
      "grad_norm": 0.1397695243358612,
      "learning_rate": 2.606423690600871e-05,
      "loss": 0.3387,
      "step": 1443
    },
    {
      "epoch": 0.910287222725661,
      "grad_norm": 0.1728425920009613,
      "learning_rate": 2.6042007957507797e-05,
      "loss": 0.4727,
      "step": 1444
    },
    {
      "epoch": 0.9109176155391828,
      "grad_norm": 0.1335446983575821,
      "learning_rate": 2.60197754634052e-05,
      "loss": 0.3824,
      "step": 1445
    },
    {
      "epoch": 0.9115480083527048,
      "grad_norm": 0.16681966185569763,
      "learning_rate": 2.5997539445955955e-05,
      "loss": 0.4989,
      "step": 1446
    },
    {
      "epoch": 0.9121784011662267,
      "grad_norm": 0.14989428222179413,
      "learning_rate": 2.597529992741866e-05,
      "loss": 0.3926,
      "step": 1447
    },
    {
      "epoch": 0.9128087939797487,
      "grad_norm": 0.14452800154685974,
      "learning_rate": 2.595305693005538e-05,
      "loss": 0.3706,
      "step": 1448
    },
    {
      "epoch": 0.9134391867932705,
      "grad_norm": 0.12137465924024582,
      "learning_rate": 2.5930810476131697e-05,
      "loss": 0.3181,
      "step": 1449
    },
    {
      "epoch": 0.9140695796067925,
      "grad_norm": 0.1557312160730362,
      "learning_rate": 2.590856058791663e-05,
      "loss": 0.4069,
      "step": 1450
    },
    {
      "epoch": 0.9146999724203144,
      "grad_norm": 0.14804257452487946,
      "learning_rate": 2.588630728768263e-05,
      "loss": 0.3671,
      "step": 1451
    },
    {
      "epoch": 0.9153303652338364,
      "grad_norm": 0.17212040722370148,
      "learning_rate": 2.5864050597705593e-05,
      "loss": 0.4591,
      "step": 1452
    },
    {
      "epoch": 0.9159607580473582,
      "grad_norm": 0.119450144469738,
      "learning_rate": 2.5841790540264776e-05,
      "loss": 0.3433,
      "step": 1453
    },
    {
      "epoch": 0.9165911508608802,
      "grad_norm": 0.14319778978824615,
      "learning_rate": 2.581952713764284e-05,
      "loss": 0.4054,
      "step": 1454
    },
    {
      "epoch": 0.9172215436744021,
      "grad_norm": 0.12508279085159302,
      "learning_rate": 2.5797260412125756e-05,
      "loss": 0.3926,
      "step": 1455
    },
    {
      "epoch": 0.9178519364879241,
      "grad_norm": 0.146225243806839,
      "learning_rate": 2.5774990386002857e-05,
      "loss": 0.3393,
      "step": 1456
    },
    {
      "epoch": 0.9184823293014459,
      "grad_norm": 0.14512450993061066,
      "learning_rate": 2.575271708156675e-05,
      "loss": 0.363,
      "step": 1457
    },
    {
      "epoch": 0.9191127221149679,
      "grad_norm": 0.12677498161792755,
      "learning_rate": 2.5730440521113347e-05,
      "loss": 0.4141,
      "step": 1458
    },
    {
      "epoch": 0.9197431149284898,
      "grad_norm": 0.16313442587852478,
      "learning_rate": 2.570816072694181e-05,
      "loss": 0.387,
      "step": 1459
    },
    {
      "epoch": 0.9203735077420118,
      "grad_norm": 0.12096891552209854,
      "learning_rate": 2.5685877721354537e-05,
      "loss": 0.3407,
      "step": 1460
    },
    {
      "epoch": 0.9210039005555337,
      "grad_norm": 0.1445506513118744,
      "learning_rate": 2.5663591526657146e-05,
      "loss": 0.429,
      "step": 1461
    },
    {
      "epoch": 0.9216342933690556,
      "grad_norm": 0.1301969736814499,
      "learning_rate": 2.5641302165158435e-05,
      "loss": 0.3671,
      "step": 1462
    },
    {
      "epoch": 0.9222646861825775,
      "grad_norm": 0.14147797226905823,
      "learning_rate": 2.5619009659170385e-05,
      "loss": 0.4343,
      "step": 1463
    },
    {
      "epoch": 0.9228950789960995,
      "grad_norm": 0.15561676025390625,
      "learning_rate": 2.559671403100812e-05,
      "loss": 0.4057,
      "step": 1464
    },
    {
      "epoch": 0.9235254718096214,
      "grad_norm": 0.16878941655158997,
      "learning_rate": 2.5574415302989894e-05,
      "loss": 0.4392,
      "step": 1465
    },
    {
      "epoch": 0.9241558646231433,
      "grad_norm": 0.13875456154346466,
      "learning_rate": 2.5552113497437047e-05,
      "loss": 0.3901,
      "step": 1466
    },
    {
      "epoch": 0.9247862574366652,
      "grad_norm": 0.1460646241903305,
      "learning_rate": 2.5529808636674022e-05,
      "loss": 0.3717,
      "step": 1467
    },
    {
      "epoch": 0.9254166502501872,
      "grad_norm": 0.15384356677532196,
      "learning_rate": 2.550750074302831e-05,
      "loss": 0.3628,
      "step": 1468
    },
    {
      "epoch": 0.9260470430637091,
      "grad_norm": 0.16636264324188232,
      "learning_rate": 2.5485189838830437e-05,
      "loss": 0.3677,
      "step": 1469
    },
    {
      "epoch": 0.926677435877231,
      "grad_norm": 0.17758719623088837,
      "learning_rate": 2.5462875946413945e-05,
      "loss": 0.4612,
      "step": 1470
    },
    {
      "epoch": 0.9273078286907529,
      "grad_norm": 0.12851199507713318,
      "learning_rate": 2.544055908811536e-05,
      "loss": 0.393,
      "step": 1471
    },
    {
      "epoch": 0.9279382215042749,
      "grad_norm": 0.20510762929916382,
      "learning_rate": 2.541823928627419e-05,
      "loss": 0.4241,
      "step": 1472
    },
    {
      "epoch": 0.9285686143177968,
      "grad_norm": 0.14449749886989594,
      "learning_rate": 2.5395916563232873e-05,
      "loss": 0.3731,
      "step": 1473
    },
    {
      "epoch": 0.9291990071313188,
      "grad_norm": 0.13960114121437073,
      "learning_rate": 2.5373590941336798e-05,
      "loss": 0.3782,
      "step": 1474
    },
    {
      "epoch": 0.9298293999448406,
      "grad_norm": 0.15464681386947632,
      "learning_rate": 2.5351262442934225e-05,
      "loss": 0.3528,
      "step": 1475
    },
    {
      "epoch": 0.9304597927583625,
      "grad_norm": 0.16516605019569397,
      "learning_rate": 2.5328931090376307e-05,
      "loss": 0.4447,
      "step": 1476
    },
    {
      "epoch": 0.9310901855718845,
      "grad_norm": 0.16184210777282715,
      "learning_rate": 2.5306596906017057e-05,
      "loss": 0.4308,
      "step": 1477
    },
    {
      "epoch": 0.9317205783854065,
      "grad_norm": 0.12153437733650208,
      "learning_rate": 2.5284259912213328e-05,
      "loss": 0.3333,
      "step": 1478
    },
    {
      "epoch": 0.9323509711989283,
      "grad_norm": 0.14005841314792633,
      "learning_rate": 2.526192013132476e-05,
      "loss": 0.3686,
      "step": 1479
    },
    {
      "epoch": 0.9329813640124502,
      "grad_norm": 0.14325931668281555,
      "learning_rate": 2.523957758571381e-05,
      "loss": 0.3911,
      "step": 1480
    },
    {
      "epoch": 0.9336117568259722,
      "grad_norm": 0.14206531643867493,
      "learning_rate": 2.5217232297745695e-05,
      "loss": 0.339,
      "step": 1481
    },
    {
      "epoch": 0.9342421496394941,
      "grad_norm": 0.12405907362699509,
      "learning_rate": 2.519488428978837e-05,
      "loss": 0.3635,
      "step": 1482
    },
    {
      "epoch": 0.934872542453016,
      "grad_norm": 0.14162199199199677,
      "learning_rate": 2.5172533584212515e-05,
      "loss": 0.3568,
      "step": 1483
    },
    {
      "epoch": 0.9355029352665379,
      "grad_norm": 0.13539136946201324,
      "learning_rate": 2.5150180203391514e-05,
      "loss": 0.3735,
      "step": 1484
    },
    {
      "epoch": 0.9361333280800599,
      "grad_norm": 0.13682807981967926,
      "learning_rate": 2.5127824169701437e-05,
      "loss": 0.3539,
      "step": 1485
    },
    {
      "epoch": 0.9367637208935818,
      "grad_norm": 0.15052659809589386,
      "learning_rate": 2.5105465505520986e-05,
      "loss": 0.4234,
      "step": 1486
    },
    {
      "epoch": 0.9373941137071038,
      "grad_norm": 0.15359781682491302,
      "learning_rate": 2.508310423323152e-05,
      "loss": 0.4297,
      "step": 1487
    },
    {
      "epoch": 0.9380245065206256,
      "grad_norm": 0.15407221019268036,
      "learning_rate": 2.506074037521699e-05,
      "loss": 0.4242,
      "step": 1488
    },
    {
      "epoch": 0.9386548993341476,
      "grad_norm": 0.13012506067752838,
      "learning_rate": 2.503837395386396e-05,
      "loss": 0.4042,
      "step": 1489
    },
    {
      "epoch": 0.9392852921476695,
      "grad_norm": 0.1508699506521225,
      "learning_rate": 2.5016004991561532e-05,
      "loss": 0.4303,
      "step": 1490
    },
    {
      "epoch": 0.9399156849611915,
      "grad_norm": 0.18919318914413452,
      "learning_rate": 2.4993633510701366e-05,
      "loss": 0.4513,
      "step": 1491
    },
    {
      "epoch": 0.9405460777747133,
      "grad_norm": 0.17468729615211487,
      "learning_rate": 2.4971259533677642e-05,
      "loss": 0.4648,
      "step": 1492
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.13577426970005035,
      "learning_rate": 2.4948883082887034e-05,
      "loss": 0.3959,
      "step": 1493
    },
    {
      "epoch": 0.9418068634017572,
      "grad_norm": 0.15910327434539795,
      "learning_rate": 2.4926504180728703e-05,
      "loss": 0.4816,
      "step": 1494
    },
    {
      "epoch": 0.9424372562152792,
      "grad_norm": 0.1459847092628479,
      "learning_rate": 2.490412284960425e-05,
      "loss": 0.3895,
      "step": 1495
    },
    {
      "epoch": 0.943067649028801,
      "grad_norm": 0.15411172807216644,
      "learning_rate": 2.488173911191772e-05,
      "loss": 0.4221,
      "step": 1496
    },
    {
      "epoch": 0.943698041842323,
      "grad_norm": 0.13575363159179688,
      "learning_rate": 2.4859352990075558e-05,
      "loss": 0.3708,
      "step": 1497
    },
    {
      "epoch": 0.9443284346558449,
      "grad_norm": 0.16765348613262177,
      "learning_rate": 2.48369645064866e-05,
      "loss": 0.4337,
      "step": 1498
    },
    {
      "epoch": 0.9449588274693669,
      "grad_norm": 0.1297028809785843,
      "learning_rate": 2.4814573683562044e-05,
      "loss": 0.3579,
      "step": 1499
    },
    {
      "epoch": 0.9455892202828887,
      "grad_norm": 0.1517636775970459,
      "learning_rate": 2.4792180543715438e-05,
      "loss": 0.4615,
      "step": 1500
    },
    {
      "epoch": 0.9462196130964107,
      "grad_norm": 0.17348432540893555,
      "learning_rate": 2.476978510936263e-05,
      "loss": 0.4367,
      "step": 1501
    },
    {
      "epoch": 0.9468500059099326,
      "grad_norm": 0.15146148204803467,
      "learning_rate": 2.4747387402921776e-05,
      "loss": 0.336,
      "step": 1502
    },
    {
      "epoch": 0.9474803987234546,
      "grad_norm": 0.13965347409248352,
      "learning_rate": 2.4724987446813324e-05,
      "loss": 0.331,
      "step": 1503
    },
    {
      "epoch": 0.9481107915369765,
      "grad_norm": 0.1368507295846939,
      "learning_rate": 2.470258526345994e-05,
      "loss": 0.4138,
      "step": 1504
    },
    {
      "epoch": 0.9487411843504984,
      "grad_norm": 0.13618862628936768,
      "learning_rate": 2.4680180875286548e-05,
      "loss": 0.354,
      "step": 1505
    },
    {
      "epoch": 0.9493715771640203,
      "grad_norm": 0.17908114194869995,
      "learning_rate": 2.465777430472026e-05,
      "loss": 0.4809,
      "step": 1506
    },
    {
      "epoch": 0.9500019699775423,
      "grad_norm": 0.13375219702720642,
      "learning_rate": 2.4635365574190383e-05,
      "loss": 0.3861,
      "step": 1507
    },
    {
      "epoch": 0.9506323627910642,
      "grad_norm": 0.1537475287914276,
      "learning_rate": 2.4612954706128387e-05,
      "loss": 0.4372,
      "step": 1508
    },
    {
      "epoch": 0.9512627556045861,
      "grad_norm": 0.15866872668266296,
      "learning_rate": 2.4590541722967872e-05,
      "loss": 0.3914,
      "step": 1509
    },
    {
      "epoch": 0.951893148418108,
      "grad_norm": 0.15313230454921722,
      "learning_rate": 2.4568126647144573e-05,
      "loss": 0.3593,
      "step": 1510
    },
    {
      "epoch": 0.95252354123163,
      "grad_norm": 0.14494845271110535,
      "learning_rate": 2.4545709501096288e-05,
      "loss": 0.3509,
      "step": 1511
    },
    {
      "epoch": 0.9531539340451519,
      "grad_norm": 0.1457335203886032,
      "learning_rate": 2.4523290307262925e-05,
      "loss": 0.4383,
      "step": 1512
    },
    {
      "epoch": 0.9537843268586738,
      "grad_norm": 0.15442660450935364,
      "learning_rate": 2.4500869088086415e-05,
      "loss": 0.369,
      "step": 1513
    },
    {
      "epoch": 0.9544147196721957,
      "grad_norm": 0.12074555456638336,
      "learning_rate": 2.4478445866010736e-05,
      "loss": 0.2969,
      "step": 1514
    },
    {
      "epoch": 0.9550451124857177,
      "grad_norm": 0.1371067464351654,
      "learning_rate": 2.4456020663481845e-05,
      "loss": 0.356,
      "step": 1515
    },
    {
      "epoch": 0.9556755052992396,
      "grad_norm": 0.19766010344028473,
      "learning_rate": 2.4433593502947717e-05,
      "loss": 0.4266,
      "step": 1516
    },
    {
      "epoch": 0.9563058981127616,
      "grad_norm": 0.11943485587835312,
      "learning_rate": 2.441116440685825e-05,
      "loss": 0.3104,
      "step": 1517
    },
    {
      "epoch": 0.9569362909262834,
      "grad_norm": 0.16624711453914642,
      "learning_rate": 2.438873339766531e-05,
      "loss": 0.4282,
      "step": 1518
    },
    {
      "epoch": 0.9575666837398054,
      "grad_norm": 0.12300334125757217,
      "learning_rate": 2.4366300497822646e-05,
      "loss": 0.3538,
      "step": 1519
    },
    {
      "epoch": 0.9581970765533273,
      "grad_norm": 0.17065255343914032,
      "learning_rate": 2.4343865729785945e-05,
      "loss": 0.4407,
      "step": 1520
    },
    {
      "epoch": 0.9588274693668493,
      "grad_norm": 0.16122186183929443,
      "learning_rate": 2.4321429116012716e-05,
      "loss": 0.4291,
      "step": 1521
    },
    {
      "epoch": 0.9594578621803711,
      "grad_norm": 0.12593595683574677,
      "learning_rate": 2.429899067896234e-05,
      "loss": 0.3771,
      "step": 1522
    },
    {
      "epoch": 0.960088254993893,
      "grad_norm": 0.1505429744720459,
      "learning_rate": 2.427655044109603e-05,
      "loss": 0.3829,
      "step": 1523
    },
    {
      "epoch": 0.960718647807415,
      "grad_norm": 0.1560903638601303,
      "learning_rate": 2.425410842487678e-05,
      "loss": 0.3833,
      "step": 1524
    },
    {
      "epoch": 0.961349040620937,
      "grad_norm": 0.14203353226184845,
      "learning_rate": 2.423166465276939e-05,
      "loss": 0.3857,
      "step": 1525
    },
    {
      "epoch": 0.9619794334344588,
      "grad_norm": 0.1508108377456665,
      "learning_rate": 2.4209219147240385e-05,
      "loss": 0.4051,
      "step": 1526
    },
    {
      "epoch": 0.9626098262479807,
      "grad_norm": 0.12329906225204468,
      "learning_rate": 2.4186771930758066e-05,
      "loss": 0.3441,
      "step": 1527
    },
    {
      "epoch": 0.9632402190615027,
      "grad_norm": 0.13662351667881012,
      "learning_rate": 2.4164323025792402e-05,
      "loss": 0.3676,
      "step": 1528
    },
    {
      "epoch": 0.9638706118750247,
      "grad_norm": 0.15648755431175232,
      "learning_rate": 2.4141872454815093e-05,
      "loss": 0.4035,
      "step": 1529
    },
    {
      "epoch": 0.9645010046885466,
      "grad_norm": 0.15462183952331543,
      "learning_rate": 2.4119420240299476e-05,
      "loss": 0.3807,
      "step": 1530
    },
    {
      "epoch": 0.9651313975020684,
      "grad_norm": 0.14754685759544373,
      "learning_rate": 2.4096966404720555e-05,
      "loss": 0.3369,
      "step": 1531
    },
    {
      "epoch": 0.9657617903155904,
      "grad_norm": 0.13474981486797333,
      "learning_rate": 2.407451097055494e-05,
      "loss": 0.3715,
      "step": 1532
    },
    {
      "epoch": 0.9663921831291123,
      "grad_norm": 0.19845913350582123,
      "learning_rate": 2.4052053960280847e-05,
      "loss": 0.4587,
      "step": 1533
    },
    {
      "epoch": 0.9670225759426343,
      "grad_norm": 0.1516077220439911,
      "learning_rate": 2.402959539637808e-05,
      "loss": 0.387,
      "step": 1534
    },
    {
      "epoch": 0.9676529687561561,
      "grad_norm": 0.160874143242836,
      "learning_rate": 2.4007135301327985e-05,
      "loss": 0.3907,
      "step": 1535
    },
    {
      "epoch": 0.9682833615696781,
      "grad_norm": 0.16115260124206543,
      "learning_rate": 2.3984673697613446e-05,
      "loss": 0.4068,
      "step": 1536
    },
    {
      "epoch": 0.9689137543832,
      "grad_norm": 0.11696796864271164,
      "learning_rate": 2.396221060771885e-05,
      "loss": 0.3385,
      "step": 1537
    },
    {
      "epoch": 0.969544147196722,
      "grad_norm": 0.19254839420318604,
      "learning_rate": 2.393974605413009e-05,
      "loss": 0.4503,
      "step": 1538
    },
    {
      "epoch": 0.9701745400102438,
      "grad_norm": 0.13219331204891205,
      "learning_rate": 2.39172800593345e-05,
      "loss": 0.3776,
      "step": 1539
    },
    {
      "epoch": 0.9708049328237658,
      "grad_norm": 0.138895183801651,
      "learning_rate": 2.3894812645820877e-05,
      "loss": 0.4329,
      "step": 1540
    },
    {
      "epoch": 0.9714353256372877,
      "grad_norm": 0.13477686047554016,
      "learning_rate": 2.387234383607943e-05,
      "loss": 0.3652,
      "step": 1541
    },
    {
      "epoch": 0.9720657184508097,
      "grad_norm": 0.15185682475566864,
      "learning_rate": 2.384987365260176e-05,
      "loss": 0.4455,
      "step": 1542
    },
    {
      "epoch": 0.9726961112643316,
      "grad_norm": 0.14270374178886414,
      "learning_rate": 2.3827402117880854e-05,
      "loss": 0.4097,
      "step": 1543
    },
    {
      "epoch": 0.9733265040778535,
      "grad_norm": 0.14587655663490295,
      "learning_rate": 2.380492925441104e-05,
      "loss": 0.3952,
      "step": 1544
    },
    {
      "epoch": 0.9739568968913754,
      "grad_norm": 0.13921095430850983,
      "learning_rate": 2.378245508468799e-05,
      "loss": 0.3444,
      "step": 1545
    },
    {
      "epoch": 0.9745872897048974,
      "grad_norm": 0.16872672736644745,
      "learning_rate": 2.375997963120867e-05,
      "loss": 0.4061,
      "step": 1546
    },
    {
      "epoch": 0.9752176825184193,
      "grad_norm": 0.14401352405548096,
      "learning_rate": 2.3737502916471346e-05,
      "loss": 0.4147,
      "step": 1547
    },
    {
      "epoch": 0.9758480753319412,
      "grad_norm": 0.16273754835128784,
      "learning_rate": 2.371502496297552e-05,
      "loss": 0.3681,
      "step": 1548
    },
    {
      "epoch": 0.9764784681454631,
      "grad_norm": 0.1970919370651245,
      "learning_rate": 2.3692545793221974e-05,
      "loss": 0.4568,
      "step": 1549
    },
    {
      "epoch": 0.9771088609589851,
      "grad_norm": 0.12945570051670074,
      "learning_rate": 2.3670065429712665e-05,
      "loss": 0.3762,
      "step": 1550
    },
    {
      "epoch": 0.977739253772507,
      "grad_norm": 0.1520601063966751,
      "learning_rate": 2.3647583894950787e-05,
      "loss": 0.3706,
      "step": 1551
    },
    {
      "epoch": 0.9783696465860289,
      "grad_norm": 0.14032632112503052,
      "learning_rate": 2.3625101211440665e-05,
      "loss": 0.3885,
      "step": 1552
    },
    {
      "epoch": 0.9790000393995508,
      "grad_norm": 0.1562827080488205,
      "learning_rate": 2.36026174016878e-05,
      "loss": 0.4199,
      "step": 1553
    },
    {
      "epoch": 0.9796304322130728,
      "grad_norm": 0.18005892634391785,
      "learning_rate": 2.358013248819882e-05,
      "loss": 0.4708,
      "step": 1554
    },
    {
      "epoch": 0.9802608250265947,
      "grad_norm": 0.17608360946178436,
      "learning_rate": 2.355764649348144e-05,
      "loss": 0.4308,
      "step": 1555
    },
    {
      "epoch": 0.9808912178401166,
      "grad_norm": 0.19211320579051971,
      "learning_rate": 2.353515944004448e-05,
      "loss": 0.5204,
      "step": 1556
    },
    {
      "epoch": 0.9815216106536385,
      "grad_norm": 0.12346751987934113,
      "learning_rate": 2.35126713503978e-05,
      "loss": 0.3678,
      "step": 1557
    },
    {
      "epoch": 0.9821520034671605,
      "grad_norm": 0.1775476336479187,
      "learning_rate": 2.3490182247052312e-05,
      "loss": 0.4032,
      "step": 1558
    },
    {
      "epoch": 0.9827823962806824,
      "grad_norm": 0.14146657288074493,
      "learning_rate": 2.3467692152519934e-05,
      "loss": 0.3351,
      "step": 1559
    },
    {
      "epoch": 0.9834127890942044,
      "grad_norm": 0.16345363855361938,
      "learning_rate": 2.3445201089313583e-05,
      "loss": 0.3726,
      "step": 1560
    },
    {
      "epoch": 0.9840431819077262,
      "grad_norm": 0.14180757105350494,
      "learning_rate": 2.342270907994714e-05,
      "loss": 0.346,
      "step": 1561
    },
    {
      "epoch": 0.9846735747212482,
      "grad_norm": 0.15017127990722656,
      "learning_rate": 2.3400216146935433e-05,
      "loss": 0.4195,
      "step": 1562
    },
    {
      "epoch": 0.9853039675347701,
      "grad_norm": 0.1315138190984726,
      "learning_rate": 2.337772231279422e-05,
      "loss": 0.3626,
      "step": 1563
    },
    {
      "epoch": 0.9859343603482921,
      "grad_norm": 0.1329270601272583,
      "learning_rate": 2.3355227600040146e-05,
      "loss": 0.3395,
      "step": 1564
    },
    {
      "epoch": 0.9865647531618139,
      "grad_norm": 0.14351126551628113,
      "learning_rate": 2.333273203119076e-05,
      "loss": 0.378,
      "step": 1565
    },
    {
      "epoch": 0.9871951459753359,
      "grad_norm": 0.14194701611995697,
      "learning_rate": 2.331023562876445e-05,
      "loss": 0.3862,
      "step": 1566
    },
    {
      "epoch": 0.9878255387888578,
      "grad_norm": 0.12102647870779037,
      "learning_rate": 2.328773841528045e-05,
      "loss": 0.3812,
      "step": 1567
    },
    {
      "epoch": 0.9884559316023798,
      "grad_norm": 0.18585304915905,
      "learning_rate": 2.3265240413258784e-05,
      "loss": 0.4972,
      "step": 1568
    },
    {
      "epoch": 0.9890863244159016,
      "grad_norm": 0.1484466940164566,
      "learning_rate": 2.3242741645220293e-05,
      "loss": 0.3705,
      "step": 1569
    },
    {
      "epoch": 0.9897167172294236,
      "grad_norm": 0.1592293083667755,
      "learning_rate": 2.322024213368657e-05,
      "loss": 0.4305,
      "step": 1570
    },
    {
      "epoch": 0.9903471100429455,
      "grad_norm": 0.15290561318397522,
      "learning_rate": 2.3197741901179957e-05,
      "loss": 0.4293,
      "step": 1571
    },
    {
      "epoch": 0.9909775028564675,
      "grad_norm": 0.14812232553958893,
      "learning_rate": 2.3175240970223506e-05,
      "loss": 0.421,
      "step": 1572
    },
    {
      "epoch": 0.9916078956699894,
      "grad_norm": 0.1472492218017578,
      "learning_rate": 2.315273936334099e-05,
      "loss": 0.4024,
      "step": 1573
    },
    {
      "epoch": 0.9922382884835113,
      "grad_norm": 0.12985913455486298,
      "learning_rate": 2.3130237103056834e-05,
      "loss": 0.3311,
      "step": 1574
    },
    {
      "epoch": 0.9928686812970332,
      "grad_norm": 0.16290757060050964,
      "learning_rate": 2.3107734211896132e-05,
      "loss": 0.4139,
      "step": 1575
    },
    {
      "epoch": 0.9934990741105552,
      "grad_norm": 0.14641249179840088,
      "learning_rate": 2.3085230712384618e-05,
      "loss": 0.4256,
      "step": 1576
    },
    {
      "epoch": 0.9941294669240771,
      "grad_norm": 0.1735769808292389,
      "learning_rate": 2.3062726627048607e-05,
      "loss": 0.4302,
      "step": 1577
    },
    {
      "epoch": 0.994759859737599,
      "grad_norm": 0.12449291348457336,
      "learning_rate": 2.3040221978415033e-05,
      "loss": 0.3884,
      "step": 1578
    },
    {
      "epoch": 0.9953902525511209,
      "grad_norm": 0.15413033962249756,
      "learning_rate": 2.301771678901136e-05,
      "loss": 0.4274,
      "step": 1579
    },
    {
      "epoch": 0.9960206453646429,
      "grad_norm": 0.14631548523902893,
      "learning_rate": 2.299521108136563e-05,
      "loss": 0.372,
      "step": 1580
    },
    {
      "epoch": 0.9966510381781648,
      "grad_norm": 0.17966602742671967,
      "learning_rate": 2.2972704878006366e-05,
      "loss": 0.4,
      "step": 1581
    },
    {
      "epoch": 0.9972814309916866,
      "grad_norm": 0.13735203444957733,
      "learning_rate": 2.2950198201462623e-05,
      "loss": 0.4069,
      "step": 1582
    },
    {
      "epoch": 0.9979118238052086,
      "grad_norm": 0.15251851081848145,
      "learning_rate": 2.292769107426391e-05,
      "loss": 0.4148,
      "step": 1583
    },
    {
      "epoch": 0.9985422166187305,
      "grad_norm": 0.14987708628177643,
      "learning_rate": 2.2905183518940175e-05,
      "loss": 0.4326,
      "step": 1584
    },
    {
      "epoch": 0.9991726094322525,
      "grad_norm": 0.14043983817100525,
      "learning_rate": 2.2882675558021835e-05,
      "loss": 0.3963,
      "step": 1585
    },
    {
      "epoch": 0.9998030022457745,
      "grad_norm": 0.1419934183359146,
      "learning_rate": 2.286016721403966e-05,
      "loss": 0.3756,
      "step": 1586
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2177281379699707,
      "learning_rate": 2.2837658509524855e-05,
      "loss": 0.4348,
      "step": 1587
    },
    {
      "epoch": 1.0006303928135218,
      "grad_norm": 0.13241924345493317,
      "learning_rate": 2.281514946700894e-05,
      "loss": 0.3234,
      "step": 1588
    },
    {
      "epoch": 1.001260785627044,
      "grad_norm": 0.14077149331569672,
      "learning_rate": 2.279264010902382e-05,
      "loss": 0.3567,
      "step": 1589
    },
    {
      "epoch": 1.0018911784405657,
      "grad_norm": 0.1576465517282486,
      "learning_rate": 2.2770130458101662e-05,
      "loss": 0.3527,
      "step": 1590
    },
    {
      "epoch": 1.0025215712540878,
      "grad_norm": 0.15964816510677338,
      "learning_rate": 2.2747620536774984e-05,
      "loss": 0.4073,
      "step": 1591
    },
    {
      "epoch": 1.0031519640676096,
      "grad_norm": 0.15599475800991058,
      "learning_rate": 2.2725110367576528e-05,
      "loss": 0.343,
      "step": 1592
    },
    {
      "epoch": 1.0037823568811315,
      "grad_norm": 0.13730469346046448,
      "learning_rate": 2.2702599973039306e-05,
      "loss": 0.3627,
      "step": 1593
    },
    {
      "epoch": 1.0044127496946536,
      "grad_norm": 0.1274457722902298,
      "learning_rate": 2.2680089375696554e-05,
      "loss": 0.3106,
      "step": 1594
    },
    {
      "epoch": 1.0050431425081754,
      "grad_norm": 0.1535639464855194,
      "learning_rate": 2.26575785980817e-05,
      "loss": 0.3128,
      "step": 1595
    },
    {
      "epoch": 1.0056735353216972,
      "grad_norm": 0.13755956292152405,
      "learning_rate": 2.2635067662728382e-05,
      "loss": 0.3614,
      "step": 1596
    },
    {
      "epoch": 1.0063039281352193,
      "grad_norm": 0.14729587733745575,
      "learning_rate": 2.261255659217035e-05,
      "loss": 0.3973,
      "step": 1597
    },
    {
      "epoch": 1.0069343209487411,
      "grad_norm": 0.13851574063301086,
      "learning_rate": 2.259004540894153e-05,
      "loss": 0.3385,
      "step": 1598
    },
    {
      "epoch": 1.0075647137622632,
      "grad_norm": 0.19435077905654907,
      "learning_rate": 2.256753413557594e-05,
      "loss": 0.428,
      "step": 1599
    },
    {
      "epoch": 1.008195106575785,
      "grad_norm": 0.1472771167755127,
      "learning_rate": 2.254502279460769e-05,
      "loss": 0.351,
      "step": 1600
    },
    {
      "epoch": 1.008195106575785,
      "eval_loss": 0.4261360764503479,
      "eval_runtime": 222.4194,
      "eval_samples_per_second": 4.496,
      "eval_steps_per_second": 4.496,
      "step": 1600
    },
    {
      "epoch": 1.0088254993893069,
      "grad_norm": 0.15507906675338745,
      "learning_rate": 2.252251140857097e-05,
      "loss": 0.298,
      "step": 1601
    },
    {
      "epoch": 1.009455892202829,
      "grad_norm": 0.15746253728866577,
      "learning_rate": 2.25e-05,
      "loss": 0.3894,
      "step": 1602
    },
    {
      "epoch": 1.0100862850163508,
      "grad_norm": 0.14388109743595123,
      "learning_rate": 2.2477488591429033e-05,
      "loss": 0.3008,
      "step": 1603
    },
    {
      "epoch": 1.0107166778298728,
      "grad_norm": 0.1770758479833603,
      "learning_rate": 2.245497720539231e-05,
      "loss": 0.3707,
      "step": 1604
    },
    {
      "epoch": 1.0113470706433947,
      "grad_norm": 0.15947437286376953,
      "learning_rate": 2.2432465864424062e-05,
      "loss": 0.3468,
      "step": 1605
    },
    {
      "epoch": 1.0119774634569165,
      "grad_norm": 0.14842569828033447,
      "learning_rate": 2.2409954591058474e-05,
      "loss": 0.2939,
      "step": 1606
    },
    {
      "epoch": 1.0126078562704386,
      "grad_norm": 0.1525873839855194,
      "learning_rate": 2.238744340782965e-05,
      "loss": 0.3332,
      "step": 1607
    },
    {
      "epoch": 1.0132382490839604,
      "grad_norm": 0.14072124660015106,
      "learning_rate": 2.2364932337271627e-05,
      "loss": 0.3691,
      "step": 1608
    },
    {
      "epoch": 1.0138686418974823,
      "grad_norm": 0.14068391919136047,
      "learning_rate": 2.2342421401918298e-05,
      "loss": 0.3615,
      "step": 1609
    },
    {
      "epoch": 1.0144990347110043,
      "grad_norm": 0.15972381830215454,
      "learning_rate": 2.2319910624303452e-05,
      "loss": 0.2993,
      "step": 1610
    },
    {
      "epoch": 1.0151294275245262,
      "grad_norm": 0.16424645483493805,
      "learning_rate": 2.2297400026960697e-05,
      "loss": 0.3269,
      "step": 1611
    },
    {
      "epoch": 1.0157598203380482,
      "grad_norm": 0.14915169775485992,
      "learning_rate": 2.2274889632423475e-05,
      "loss": 0.385,
      "step": 1612
    },
    {
      "epoch": 1.01639021315157,
      "grad_norm": 0.1401686817407608,
      "learning_rate": 2.225237946322502e-05,
      "loss": 0.3614,
      "step": 1613
    },
    {
      "epoch": 1.017020605965092,
      "grad_norm": 0.16717317700386047,
      "learning_rate": 2.222986954189834e-05,
      "loss": 0.3794,
      "step": 1614
    },
    {
      "epoch": 1.017650998778614,
      "grad_norm": 0.16320808231830597,
      "learning_rate": 2.2207359890976184e-05,
      "loss": 0.4097,
      "step": 1615
    },
    {
      "epoch": 1.0182813915921358,
      "grad_norm": 0.1880963146686554,
      "learning_rate": 2.2184850532991065e-05,
      "loss": 0.3861,
      "step": 1616
    },
    {
      "epoch": 1.0189117844056579,
      "grad_norm": 0.16419583559036255,
      "learning_rate": 2.2162341490475148e-05,
      "loss": 0.3844,
      "step": 1617
    },
    {
      "epoch": 1.0195421772191797,
      "grad_norm": 0.1421871930360794,
      "learning_rate": 2.2139832785960342e-05,
      "loss": 0.354,
      "step": 1618
    },
    {
      "epoch": 1.0201725700327016,
      "grad_norm": 0.1516357809305191,
      "learning_rate": 2.2117324441978175e-05,
      "loss": 0.3302,
      "step": 1619
    },
    {
      "epoch": 1.0208029628462236,
      "grad_norm": 0.1635531634092331,
      "learning_rate": 2.2094816481059827e-05,
      "loss": 0.3938,
      "step": 1620
    },
    {
      "epoch": 1.0214333556597455,
      "grad_norm": 0.1413624882698059,
      "learning_rate": 2.20723089257361e-05,
      "loss": 0.3953,
      "step": 1621
    },
    {
      "epoch": 1.0220637484732673,
      "grad_norm": 0.18992073833942413,
      "learning_rate": 2.204980179853738e-05,
      "loss": 0.3696,
      "step": 1622
    },
    {
      "epoch": 1.0226941412867894,
      "grad_norm": 0.1756129264831543,
      "learning_rate": 2.2027295121993637e-05,
      "loss": 0.3644,
      "step": 1623
    },
    {
      "epoch": 1.0233245341003112,
      "grad_norm": 0.18942801654338837,
      "learning_rate": 2.200478891863438e-05,
      "loss": 0.3855,
      "step": 1624
    },
    {
      "epoch": 1.0239549269138333,
      "grad_norm": 0.15456733107566833,
      "learning_rate": 2.198228321098864e-05,
      "loss": 0.3558,
      "step": 1625
    },
    {
      "epoch": 1.0245853197273551,
      "grad_norm": 0.16767869889736176,
      "learning_rate": 2.1959778021584977e-05,
      "loss": 0.3341,
      "step": 1626
    },
    {
      "epoch": 1.025215712540877,
      "grad_norm": 0.1457701325416565,
      "learning_rate": 2.1937273372951392e-05,
      "loss": 0.3264,
      "step": 1627
    },
    {
      "epoch": 1.025846105354399,
      "grad_norm": 0.2074301540851593,
      "learning_rate": 2.1914769287615388e-05,
      "loss": 0.3717,
      "step": 1628
    },
    {
      "epoch": 1.0264764981679209,
      "grad_norm": 0.17067231237888336,
      "learning_rate": 2.1892265788103867e-05,
      "loss": 0.3604,
      "step": 1629
    },
    {
      "epoch": 1.0271068909814427,
      "grad_norm": 0.14882232248783112,
      "learning_rate": 2.1869762896943172e-05,
      "loss": 0.3606,
      "step": 1630
    },
    {
      "epoch": 1.0277372837949648,
      "grad_norm": 0.20086149871349335,
      "learning_rate": 2.1847260636659014e-05,
      "loss": 0.4098,
      "step": 1631
    },
    {
      "epoch": 1.0283676766084866,
      "grad_norm": 0.16884002089500427,
      "learning_rate": 2.1824759029776497e-05,
      "loss": 0.3098,
      "step": 1632
    },
    {
      "epoch": 1.0289980694220087,
      "grad_norm": 0.13446485996246338,
      "learning_rate": 2.1802258098820045e-05,
      "loss": 0.2947,
      "step": 1633
    },
    {
      "epoch": 1.0296284622355305,
      "grad_norm": 0.18780672550201416,
      "learning_rate": 2.1779757866313433e-05,
      "loss": 0.3743,
      "step": 1634
    },
    {
      "epoch": 1.0302588550490523,
      "grad_norm": 0.16584482789039612,
      "learning_rate": 2.1757258354779704e-05,
      "loss": 0.3698,
      "step": 1635
    },
    {
      "epoch": 1.0308892478625744,
      "grad_norm": 0.1275024265050888,
      "learning_rate": 2.1734759586741222e-05,
      "loss": 0.3158,
      "step": 1636
    },
    {
      "epoch": 1.0315196406760962,
      "grad_norm": 0.15358369052410126,
      "learning_rate": 2.171226158471955e-05,
      "loss": 0.3502,
      "step": 1637
    },
    {
      "epoch": 1.0321500334896183,
      "grad_norm": 0.13834697008132935,
      "learning_rate": 2.1689764371235556e-05,
      "loss": 0.3342,
      "step": 1638
    },
    {
      "epoch": 1.0327804263031402,
      "grad_norm": 0.16860130429267883,
      "learning_rate": 2.1667267968809242e-05,
      "loss": 0.3234,
      "step": 1639
    },
    {
      "epoch": 1.033410819116662,
      "grad_norm": 0.17650333046913147,
      "learning_rate": 2.164477239995986e-05,
      "loss": 0.3425,
      "step": 1640
    },
    {
      "epoch": 1.034041211930184,
      "grad_norm": 0.1395651251077652,
      "learning_rate": 2.162227768720579e-05,
      "loss": 0.303,
      "step": 1641
    },
    {
      "epoch": 1.034671604743706,
      "grad_norm": 0.1621188074350357,
      "learning_rate": 2.1599783853064574e-05,
      "loss": 0.3546,
      "step": 1642
    },
    {
      "epoch": 1.035301997557228,
      "grad_norm": 0.15481868386268616,
      "learning_rate": 2.1577290920052863e-05,
      "loss": 0.3367,
      "step": 1643
    },
    {
      "epoch": 1.0359323903707498,
      "grad_norm": 0.1645256131887436,
      "learning_rate": 2.155479891068642e-05,
      "loss": 0.3282,
      "step": 1644
    },
    {
      "epoch": 1.0365627831842716,
      "grad_norm": 0.1496310532093048,
      "learning_rate": 2.153230784748007e-05,
      "loss": 0.284,
      "step": 1645
    },
    {
      "epoch": 1.0371931759977937,
      "grad_norm": 0.21159502863883972,
      "learning_rate": 2.150981775294769e-05,
      "loss": 0.3727,
      "step": 1646
    },
    {
      "epoch": 1.0378235688113155,
      "grad_norm": 0.16791081428527832,
      "learning_rate": 2.14873286496022e-05,
      "loss": 0.3644,
      "step": 1647
    },
    {
      "epoch": 1.0384539616248374,
      "grad_norm": 0.19133862853050232,
      "learning_rate": 2.1464840559955525e-05,
      "loss": 0.3774,
      "step": 1648
    },
    {
      "epoch": 1.0390843544383594,
      "grad_norm": 0.1551189422607422,
      "learning_rate": 2.144235350651856e-05,
      "loss": 0.4179,
      "step": 1649
    },
    {
      "epoch": 1.0397147472518813,
      "grad_norm": 0.14688484370708466,
      "learning_rate": 2.1419867511801187e-05,
      "loss": 0.3788,
      "step": 1650
    },
    {
      "epoch": 1.0403451400654034,
      "grad_norm": 0.16884516179561615,
      "learning_rate": 2.13973825983122e-05,
      "loss": 0.362,
      "step": 1651
    },
    {
      "epoch": 1.0409755328789252,
      "grad_norm": 0.15721917152404785,
      "learning_rate": 2.1374898788559338e-05,
      "loss": 0.3686,
      "step": 1652
    },
    {
      "epoch": 1.041605925692447,
      "grad_norm": 0.18979185819625854,
      "learning_rate": 2.1352416105049215e-05,
      "loss": 0.3916,
      "step": 1653
    },
    {
      "epoch": 1.042236318505969,
      "grad_norm": 0.13208013772964478,
      "learning_rate": 2.1329934570287338e-05,
      "loss": 0.3572,
      "step": 1654
    },
    {
      "epoch": 1.042866711319491,
      "grad_norm": 0.12008725106716156,
      "learning_rate": 2.130745420677803e-05,
      "loss": 0.3206,
      "step": 1655
    },
    {
      "epoch": 1.0434971041330128,
      "grad_norm": 0.1520049124956131,
      "learning_rate": 2.1284975037024486e-05,
      "loss": 0.3464,
      "step": 1656
    },
    {
      "epoch": 1.0441274969465348,
      "grad_norm": 0.18334481120109558,
      "learning_rate": 2.1262497083528657e-05,
      "loss": 0.2915,
      "step": 1657
    },
    {
      "epoch": 1.0447578897600567,
      "grad_norm": 0.14017915725708008,
      "learning_rate": 2.1240020368791336e-05,
      "loss": 0.316,
      "step": 1658
    },
    {
      "epoch": 1.0453882825735787,
      "grad_norm": 0.16284134984016418,
      "learning_rate": 2.1217544915312014e-05,
      "loss": 0.3782,
      "step": 1659
    },
    {
      "epoch": 1.0460186753871006,
      "grad_norm": 0.13631150126457214,
      "learning_rate": 2.1195070745588966e-05,
      "loss": 0.3128,
      "step": 1660
    },
    {
      "epoch": 1.0466490682006224,
      "grad_norm": 0.17620840668678284,
      "learning_rate": 2.1172597882119152e-05,
      "loss": 0.3419,
      "step": 1661
    },
    {
      "epoch": 1.0472794610141445,
      "grad_norm": 0.15279735624790192,
      "learning_rate": 2.1150126347398248e-05,
      "loss": 0.3849,
      "step": 1662
    },
    {
      "epoch": 1.0479098538276663,
      "grad_norm": 0.1570388674736023,
      "learning_rate": 2.1127656163920575e-05,
      "loss": 0.2943,
      "step": 1663
    },
    {
      "epoch": 1.0485402466411884,
      "grad_norm": 0.16196565330028534,
      "learning_rate": 2.1105187354179126e-05,
      "loss": 0.3545,
      "step": 1664
    },
    {
      "epoch": 1.0491706394547102,
      "grad_norm": 0.16797305643558502,
      "learning_rate": 2.10827199406655e-05,
      "loss": 0.3147,
      "step": 1665
    },
    {
      "epoch": 1.049801032268232,
      "grad_norm": 0.16055622696876526,
      "learning_rate": 2.1060253945869915e-05,
      "loss": 0.3912,
      "step": 1666
    },
    {
      "epoch": 1.0504314250817541,
      "grad_norm": 0.14581353962421417,
      "learning_rate": 2.103778939228115e-05,
      "loss": 0.3271,
      "step": 1667
    },
    {
      "epoch": 1.051061817895276,
      "grad_norm": 0.16159775853157043,
      "learning_rate": 2.101532630238656e-05,
      "loss": 0.3505,
      "step": 1668
    },
    {
      "epoch": 1.0516922107087978,
      "grad_norm": 0.15286363661289215,
      "learning_rate": 2.0992864698672018e-05,
      "loss": 0.3272,
      "step": 1669
    },
    {
      "epoch": 1.0523226035223199,
      "grad_norm": 0.14952370524406433,
      "learning_rate": 2.0970404603621925e-05,
      "loss": 0.3497,
      "step": 1670
    },
    {
      "epoch": 1.0529529963358417,
      "grad_norm": 0.15633340179920197,
      "learning_rate": 2.0947946039719153e-05,
      "loss": 0.3394,
      "step": 1671
    },
    {
      "epoch": 1.0535833891493638,
      "grad_norm": 0.17084918916225433,
      "learning_rate": 2.0925489029445066e-05,
      "loss": 0.3367,
      "step": 1672
    },
    {
      "epoch": 1.0542137819628856,
      "grad_norm": 0.15072505176067352,
      "learning_rate": 2.0903033595279448e-05,
      "loss": 0.3312,
      "step": 1673
    },
    {
      "epoch": 1.0548441747764075,
      "grad_norm": 0.18470540642738342,
      "learning_rate": 2.0880579759700534e-05,
      "loss": 0.361,
      "step": 1674
    },
    {
      "epoch": 1.0554745675899295,
      "grad_norm": 0.1253766268491745,
      "learning_rate": 2.085812754518491e-05,
      "loss": 0.2552,
      "step": 1675
    },
    {
      "epoch": 1.0561049604034514,
      "grad_norm": 0.15152914822101593,
      "learning_rate": 2.0835676974207604e-05,
      "loss": 0.3833,
      "step": 1676
    },
    {
      "epoch": 1.0567353532169734,
      "grad_norm": 0.15731677412986755,
      "learning_rate": 2.0813228069241937e-05,
      "loss": 0.331,
      "step": 1677
    },
    {
      "epoch": 1.0573657460304953,
      "grad_norm": 0.17662854492664337,
      "learning_rate": 2.0790780852759618e-05,
      "loss": 0.4095,
      "step": 1678
    },
    {
      "epoch": 1.057996138844017,
      "grad_norm": 0.142434760928154,
      "learning_rate": 2.076833534723061e-05,
      "loss": 0.3207,
      "step": 1679
    },
    {
      "epoch": 1.0586265316575392,
      "grad_norm": 0.16224971413612366,
      "learning_rate": 2.0745891575123224e-05,
      "loss": 0.3565,
      "step": 1680
    },
    {
      "epoch": 1.059256924471061,
      "grad_norm": 0.16012756526470184,
      "learning_rate": 2.0723449558903976e-05,
      "loss": 0.3237,
      "step": 1681
    },
    {
      "epoch": 1.0598873172845829,
      "grad_norm": 0.15715405344963074,
      "learning_rate": 2.0701009321037666e-05,
      "loss": 0.3296,
      "step": 1682
    },
    {
      "epoch": 1.060517710098105,
      "grad_norm": 0.17331728339195251,
      "learning_rate": 2.067857088398729e-05,
      "loss": 0.3466,
      "step": 1683
    },
    {
      "epoch": 1.0611481029116268,
      "grad_norm": 0.14372803270816803,
      "learning_rate": 2.0656134270214064e-05,
      "loss": 0.3718,
      "step": 1684
    },
    {
      "epoch": 1.0617784957251488,
      "grad_norm": 0.18391430377960205,
      "learning_rate": 2.0633699502177353e-05,
      "loss": 0.4057,
      "step": 1685
    },
    {
      "epoch": 1.0624088885386707,
      "grad_norm": 0.15161548554897308,
      "learning_rate": 2.0611266602334698e-05,
      "loss": 0.3942,
      "step": 1686
    },
    {
      "epoch": 1.0630392813521925,
      "grad_norm": 0.21158887445926666,
      "learning_rate": 2.058883559314175e-05,
      "loss": 0.4186,
      "step": 1687
    },
    {
      "epoch": 1.0636696741657146,
      "grad_norm": 0.14235830307006836,
      "learning_rate": 2.0566406497052286e-05,
      "loss": 0.3225,
      "step": 1688
    },
    {
      "epoch": 1.0643000669792364,
      "grad_norm": 0.14076681435108185,
      "learning_rate": 2.054397933651815e-05,
      "loss": 0.3632,
      "step": 1689
    },
    {
      "epoch": 1.0649304597927585,
      "grad_norm": 0.1497487723827362,
      "learning_rate": 2.052155413398927e-05,
      "loss": 0.3167,
      "step": 1690
    },
    {
      "epoch": 1.0655608526062803,
      "grad_norm": 0.16239683330059052,
      "learning_rate": 2.0499130911913584e-05,
      "loss": 0.3964,
      "step": 1691
    },
    {
      "epoch": 1.0661912454198021,
      "grad_norm": 0.1456744372844696,
      "learning_rate": 2.0476709692737078e-05,
      "loss": 0.3236,
      "step": 1692
    },
    {
      "epoch": 1.0668216382333242,
      "grad_norm": 0.1592414826154709,
      "learning_rate": 2.0454290498903715e-05,
      "loss": 0.3432,
      "step": 1693
    },
    {
      "epoch": 1.067452031046846,
      "grad_norm": 0.13664771616458893,
      "learning_rate": 2.0431873352855436e-05,
      "loss": 0.357,
      "step": 1694
    },
    {
      "epoch": 1.068082423860368,
      "grad_norm": 0.14796030521392822,
      "learning_rate": 2.0409458277032127e-05,
      "loss": 0.3883,
      "step": 1695
    },
    {
      "epoch": 1.06871281667389,
      "grad_norm": 0.14164042472839355,
      "learning_rate": 2.038704529387162e-05,
      "loss": 0.3135,
      "step": 1696
    },
    {
      "epoch": 1.0693432094874118,
      "grad_norm": 0.15346303582191467,
      "learning_rate": 2.0364634425809613e-05,
      "loss": 0.3435,
      "step": 1697
    },
    {
      "epoch": 1.0699736023009339,
      "grad_norm": 0.1760052591562271,
      "learning_rate": 2.034222569527975e-05,
      "loss": 0.4371,
      "step": 1698
    },
    {
      "epoch": 1.0706039951144557,
      "grad_norm": 0.1785290688276291,
      "learning_rate": 2.0319819124713452e-05,
      "loss": 0.4007,
      "step": 1699
    },
    {
      "epoch": 1.0712343879279775,
      "grad_norm": 0.13086946308612823,
      "learning_rate": 2.0297414736540066e-05,
      "loss": 0.288,
      "step": 1700
    },
    {
      "epoch": 1.0718647807414996,
      "grad_norm": 0.18402013182640076,
      "learning_rate": 2.0275012553186682e-05,
      "loss": 0.3993,
      "step": 1701
    },
    {
      "epoch": 1.0724951735550214,
      "grad_norm": 0.14493998885154724,
      "learning_rate": 2.0252612597078227e-05,
      "loss": 0.3638,
      "step": 1702
    },
    {
      "epoch": 1.0731255663685433,
      "grad_norm": 0.17171445488929749,
      "learning_rate": 2.0230214890637377e-05,
      "loss": 0.4127,
      "step": 1703
    },
    {
      "epoch": 1.0737559591820653,
      "grad_norm": 0.16459907591342926,
      "learning_rate": 2.0207819456284572e-05,
      "loss": 0.2843,
      "step": 1704
    },
    {
      "epoch": 1.0743863519955872,
      "grad_norm": 0.16528962552547455,
      "learning_rate": 2.0185426316437956e-05,
      "loss": 0.3822,
      "step": 1705
    },
    {
      "epoch": 1.0750167448091092,
      "grad_norm": 0.14452378451824188,
      "learning_rate": 2.0163035493513405e-05,
      "loss": 0.301,
      "step": 1706
    },
    {
      "epoch": 1.075647137622631,
      "grad_norm": 0.1770389974117279,
      "learning_rate": 2.0140647009924445e-05,
      "loss": 0.3514,
      "step": 1707
    },
    {
      "epoch": 1.076277530436153,
      "grad_norm": 0.15465545654296875,
      "learning_rate": 2.0118260888082286e-05,
      "loss": 0.327,
      "step": 1708
    },
    {
      "epoch": 1.076907923249675,
      "grad_norm": 0.12285317480564117,
      "learning_rate": 2.0095877150395754e-05,
      "loss": 0.2968,
      "step": 1709
    },
    {
      "epoch": 1.0775383160631968,
      "grad_norm": 0.13278360664844513,
      "learning_rate": 2.0073495819271303e-05,
      "loss": 0.3318,
      "step": 1710
    },
    {
      "epoch": 1.078168708876719,
      "grad_norm": 0.16636794805526733,
      "learning_rate": 2.005111691711297e-05,
      "loss": 0.4132,
      "step": 1711
    },
    {
      "epoch": 1.0787991016902407,
      "grad_norm": 0.17786742746829987,
      "learning_rate": 2.0028740466322367e-05,
      "loss": 0.3533,
      "step": 1712
    },
    {
      "epoch": 1.0794294945037626,
      "grad_norm": 0.17079973220825195,
      "learning_rate": 2.0006366489298637e-05,
      "loss": 0.3396,
      "step": 1713
    },
    {
      "epoch": 1.0800598873172846,
      "grad_norm": 0.17479632794857025,
      "learning_rate": 1.9983995008438474e-05,
      "loss": 0.3884,
      "step": 1714
    },
    {
      "epoch": 1.0806902801308065,
      "grad_norm": 0.1269015371799469,
      "learning_rate": 1.996162604613604e-05,
      "loss": 0.3791,
      "step": 1715
    },
    {
      "epoch": 1.0813206729443285,
      "grad_norm": 0.15591496229171753,
      "learning_rate": 1.9939259624783013e-05,
      "loss": 0.3828,
      "step": 1716
    },
    {
      "epoch": 1.0819510657578504,
      "grad_norm": 0.13570600748062134,
      "learning_rate": 1.991689576676848e-05,
      "loss": 0.3429,
      "step": 1717
    },
    {
      "epoch": 1.0825814585713722,
      "grad_norm": 0.1412353813648224,
      "learning_rate": 1.989453449447902e-05,
      "loss": 0.3769,
      "step": 1718
    },
    {
      "epoch": 1.0832118513848943,
      "grad_norm": 0.1421315222978592,
      "learning_rate": 1.9872175830298565e-05,
      "loss": 0.3805,
      "step": 1719
    },
    {
      "epoch": 1.0838422441984161,
      "grad_norm": 0.14278772473335266,
      "learning_rate": 1.984981979660849e-05,
      "loss": 0.3063,
      "step": 1720
    },
    {
      "epoch": 1.084472637011938,
      "grad_norm": 0.16554735600948334,
      "learning_rate": 1.982746641578749e-05,
      "loss": 0.3635,
      "step": 1721
    },
    {
      "epoch": 1.08510302982546,
      "grad_norm": 0.14530974626541138,
      "learning_rate": 1.980511571021164e-05,
      "loss": 0.3411,
      "step": 1722
    },
    {
      "epoch": 1.0857334226389819,
      "grad_norm": 0.12940828502178192,
      "learning_rate": 1.978276770225431e-05,
      "loss": 0.3391,
      "step": 1723
    },
    {
      "epoch": 1.086363815452504,
      "grad_norm": 0.16847068071365356,
      "learning_rate": 1.9760422414286198e-05,
      "loss": 0.3779,
      "step": 1724
    },
    {
      "epoch": 1.0869942082660258,
      "grad_norm": 0.15065135061740875,
      "learning_rate": 1.9738079868675245e-05,
      "loss": 0.3144,
      "step": 1725
    },
    {
      "epoch": 1.0876246010795476,
      "grad_norm": 0.16651591658592224,
      "learning_rate": 1.9715740087786682e-05,
      "loss": 0.3138,
      "step": 1726
    },
    {
      "epoch": 1.0882549938930697,
      "grad_norm": 0.16407909989356995,
      "learning_rate": 1.9693403093982942e-05,
      "loss": 0.3587,
      "step": 1727
    },
    {
      "epoch": 1.0888853867065915,
      "grad_norm": 0.18480424582958221,
      "learning_rate": 1.96710689096237e-05,
      "loss": 0.3492,
      "step": 1728
    },
    {
      "epoch": 1.0895157795201134,
      "grad_norm": 0.12064468115568161,
      "learning_rate": 1.9648737557065778e-05,
      "loss": 0.2683,
      "step": 1729
    },
    {
      "epoch": 1.0901461723336354,
      "grad_norm": 0.1776457130908966,
      "learning_rate": 1.962640905866321e-05,
      "loss": 0.3334,
      "step": 1730
    },
    {
      "epoch": 1.0907765651471573,
      "grad_norm": 0.18231508135795593,
      "learning_rate": 1.9604083436767123e-05,
      "loss": 0.3286,
      "step": 1731
    },
    {
      "epoch": 1.0914069579606793,
      "grad_norm": 0.17567265033721924,
      "learning_rate": 1.9581760713725814e-05,
      "loss": 0.3887,
      "step": 1732
    },
    {
      "epoch": 1.0920373507742012,
      "grad_norm": 0.14002569019794464,
      "learning_rate": 1.9559440911884643e-05,
      "loss": 0.3509,
      "step": 1733
    },
    {
      "epoch": 1.092667743587723,
      "grad_norm": 0.16191992163658142,
      "learning_rate": 1.953712405358606e-05,
      "loss": 0.3311,
      "step": 1734
    },
    {
      "epoch": 1.093298136401245,
      "grad_norm": 0.16020630300045013,
      "learning_rate": 1.9514810161169562e-05,
      "loss": 0.3539,
      "step": 1735
    },
    {
      "epoch": 1.093928529214767,
      "grad_norm": 0.16078157722949982,
      "learning_rate": 1.9492499256971698e-05,
      "loss": 0.3259,
      "step": 1736
    },
    {
      "epoch": 1.094558922028289,
      "grad_norm": 0.1885467916727066,
      "learning_rate": 1.9470191363325974e-05,
      "loss": 0.419,
      "step": 1737
    },
    {
      "epoch": 1.0951893148418108,
      "grad_norm": 0.14935602247714996,
      "learning_rate": 1.944788650256296e-05,
      "loss": 0.2964,
      "step": 1738
    },
    {
      "epoch": 1.0958197076553327,
      "grad_norm": 0.15341497957706451,
      "learning_rate": 1.942558469701011e-05,
      "loss": 0.303,
      "step": 1739
    },
    {
      "epoch": 1.0964501004688547,
      "grad_norm": 0.14188827574253082,
      "learning_rate": 1.940328596899189e-05,
      "loss": 0.302,
      "step": 1740
    },
    {
      "epoch": 1.0970804932823766,
      "grad_norm": 0.14294935762882233,
      "learning_rate": 1.9380990340829614e-05,
      "loss": 0.3174,
      "step": 1741
    },
    {
      "epoch": 1.0977108860958986,
      "grad_norm": 0.17663875222206116,
      "learning_rate": 1.935869783484157e-05,
      "loss": 0.3676,
      "step": 1742
    },
    {
      "epoch": 1.0983412789094205,
      "grad_norm": 0.15765130519866943,
      "learning_rate": 1.933640847334286e-05,
      "loss": 0.34,
      "step": 1743
    },
    {
      "epoch": 1.0989716717229423,
      "grad_norm": 0.15313151478767395,
      "learning_rate": 1.931412227864547e-05,
      "loss": 0.3366,
      "step": 1744
    },
    {
      "epoch": 1.0996020645364644,
      "grad_norm": 0.19496582448482513,
      "learning_rate": 1.929183927305819e-05,
      "loss": 0.4336,
      "step": 1745
    },
    {
      "epoch": 1.1002324573499862,
      "grad_norm": 0.17884379625320435,
      "learning_rate": 1.926955947888666e-05,
      "loss": 0.3065,
      "step": 1746
    },
    {
      "epoch": 1.100862850163508,
      "grad_norm": 0.19162435829639435,
      "learning_rate": 1.9247282918433253e-05,
      "loss": 0.3758,
      "step": 1747
    },
    {
      "epoch": 1.10149324297703,
      "grad_norm": 0.217264324426651,
      "learning_rate": 1.9225009613997152e-05,
      "loss": 0.3557,
      "step": 1748
    },
    {
      "epoch": 1.102123635790552,
      "grad_norm": 0.15280327200889587,
      "learning_rate": 1.9202739587874247e-05,
      "loss": 0.3406,
      "step": 1749
    },
    {
      "epoch": 1.102754028604074,
      "grad_norm": 0.16503563523292542,
      "learning_rate": 1.9180472862357166e-05,
      "loss": 0.3315,
      "step": 1750
    },
    {
      "epoch": 1.1033844214175959,
      "grad_norm": 0.180845245718956,
      "learning_rate": 1.915820945973522e-05,
      "loss": 0.3854,
      "step": 1751
    },
    {
      "epoch": 1.1040148142311177,
      "grad_norm": 0.21107171475887299,
      "learning_rate": 1.913594940229441e-05,
      "loss": 0.406,
      "step": 1752
    },
    {
      "epoch": 1.1046452070446398,
      "grad_norm": 0.1476421058177948,
      "learning_rate": 1.911369271231737e-05,
      "loss": 0.3955,
      "step": 1753
    },
    {
      "epoch": 1.1052755998581616,
      "grad_norm": 0.1788562536239624,
      "learning_rate": 1.9091439412083374e-05,
      "loss": 0.3686,
      "step": 1754
    },
    {
      "epoch": 1.1059059926716834,
      "grad_norm": 0.11860582232475281,
      "learning_rate": 1.90691895238683e-05,
      "loss": 0.2905,
      "step": 1755
    },
    {
      "epoch": 1.1065363854852055,
      "grad_norm": 0.1615850329399109,
      "learning_rate": 1.9046943069944623e-05,
      "loss": 0.365,
      "step": 1756
    },
    {
      "epoch": 1.1071667782987273,
      "grad_norm": 0.1617215871810913,
      "learning_rate": 1.9024700072581343e-05,
      "loss": 0.3549,
      "step": 1757
    },
    {
      "epoch": 1.1077971711122494,
      "grad_norm": 0.1754276603460312,
      "learning_rate": 1.9002460554044048e-05,
      "loss": 0.3082,
      "step": 1758
    },
    {
      "epoch": 1.1084275639257712,
      "grad_norm": 0.17688316106796265,
      "learning_rate": 1.8980224536594804e-05,
      "loss": 0.3494,
      "step": 1759
    },
    {
      "epoch": 1.109057956739293,
      "grad_norm": 0.14567002654075623,
      "learning_rate": 1.895799204249221e-05,
      "loss": 0.2989,
      "step": 1760
    },
    {
      "epoch": 1.1096883495528151,
      "grad_norm": 0.1490413397550583,
      "learning_rate": 1.893576309399129e-05,
      "loss": 0.3265,
      "step": 1761
    },
    {
      "epoch": 1.110318742366337,
      "grad_norm": 0.17052440345287323,
      "learning_rate": 1.8913537713343575e-05,
      "loss": 0.3928,
      "step": 1762
    },
    {
      "epoch": 1.110949135179859,
      "grad_norm": 0.17583999037742615,
      "learning_rate": 1.8891315922796972e-05,
      "loss": 0.3459,
      "step": 1763
    },
    {
      "epoch": 1.111579527993381,
      "grad_norm": 0.19256591796875,
      "learning_rate": 1.8869097744595832e-05,
      "loss": 0.3684,
      "step": 1764
    },
    {
      "epoch": 1.1122099208069027,
      "grad_norm": 0.13533738255500793,
      "learning_rate": 1.8846883200980865e-05,
      "loss": 0.3343,
      "step": 1765
    },
    {
      "epoch": 1.1128403136204248,
      "grad_norm": 0.13766324520111084,
      "learning_rate": 1.8824672314189165e-05,
      "loss": 0.324,
      "step": 1766
    },
    {
      "epoch": 1.1134707064339466,
      "grad_norm": 0.1536354422569275,
      "learning_rate": 1.8802465106454145e-05,
      "loss": 0.326,
      "step": 1767
    },
    {
      "epoch": 1.1141010992474687,
      "grad_norm": 0.17327311635017395,
      "learning_rate": 1.878026160000555e-05,
      "loss": 0.3406,
      "step": 1768
    },
    {
      "epoch": 1.1147314920609905,
      "grad_norm": 0.1594127118587494,
      "learning_rate": 1.875806181706941e-05,
      "loss": 0.3698,
      "step": 1769
    },
    {
      "epoch": 1.1153618848745124,
      "grad_norm": 0.15584924817085266,
      "learning_rate": 1.8735865779868038e-05,
      "loss": 0.3275,
      "step": 1770
    },
    {
      "epoch": 1.1159922776880344,
      "grad_norm": 0.15024584531784058,
      "learning_rate": 1.871367351061998e-05,
      "loss": 0.3456,
      "step": 1771
    },
    {
      "epoch": 1.1166226705015563,
      "grad_norm": 0.14913159608840942,
      "learning_rate": 1.8691485031540035e-05,
      "loss": 0.3551,
      "step": 1772
    },
    {
      "epoch": 1.1172530633150781,
      "grad_norm": 0.18090124428272247,
      "learning_rate": 1.8669300364839185e-05,
      "loss": 0.3763,
      "step": 1773
    },
    {
      "epoch": 1.1178834561286002,
      "grad_norm": 0.14488834142684937,
      "learning_rate": 1.8647119532724616e-05,
      "loss": 0.354,
      "step": 1774
    },
    {
      "epoch": 1.118513848942122,
      "grad_norm": 0.1709679216146469,
      "learning_rate": 1.862494255739965e-05,
      "loss": 0.3878,
      "step": 1775
    },
    {
      "epoch": 1.119144241755644,
      "grad_norm": 0.12193349003791809,
      "learning_rate": 1.8602769461063773e-05,
      "loss": 0.3088,
      "step": 1776
    },
    {
      "epoch": 1.119774634569166,
      "grad_norm": 0.15026722848415375,
      "learning_rate": 1.8580600265912573e-05,
      "loss": 0.39,
      "step": 1777
    },
    {
      "epoch": 1.1204050273826878,
      "grad_norm": 0.14931859076023102,
      "learning_rate": 1.855843499413775e-05,
      "loss": 0.3017,
      "step": 1778
    },
    {
      "epoch": 1.1210354201962098,
      "grad_norm": 0.1615418940782547,
      "learning_rate": 1.8536273667927045e-05,
      "loss": 0.3286,
      "step": 1779
    },
    {
      "epoch": 1.1216658130097317,
      "grad_norm": 0.18689070641994476,
      "learning_rate": 1.8514116309464294e-05,
      "loss": 0.3529,
      "step": 1780
    },
    {
      "epoch": 1.1222962058232535,
      "grad_norm": 0.1535871922969818,
      "learning_rate": 1.8491962940929302e-05,
      "loss": 0.3603,
      "step": 1781
    },
    {
      "epoch": 1.1229265986367756,
      "grad_norm": 0.12987935543060303,
      "learning_rate": 1.8469813584497954e-05,
      "loss": 0.303,
      "step": 1782
    },
    {
      "epoch": 1.1235569914502974,
      "grad_norm": 0.1690494865179062,
      "learning_rate": 1.8447668262342047e-05,
      "loss": 0.3499,
      "step": 1783
    },
    {
      "epoch": 1.1241873842638195,
      "grad_norm": 0.1679484099149704,
      "learning_rate": 1.8425526996629394e-05,
      "loss": 0.3893,
      "step": 1784
    },
    {
      "epoch": 1.1248177770773413,
      "grad_norm": 0.19813470542430878,
      "learning_rate": 1.8403389809523707e-05,
      "loss": 0.3479,
      "step": 1785
    },
    {
      "epoch": 1.1254481698908632,
      "grad_norm": 0.17155030369758606,
      "learning_rate": 1.838125672318465e-05,
      "loss": 0.3111,
      "step": 1786
    },
    {
      "epoch": 1.1260785627043852,
      "grad_norm": 0.16126155853271484,
      "learning_rate": 1.8359127759767753e-05,
      "loss": 0.3233,
      "step": 1787
    },
    {
      "epoch": 1.126708955517907,
      "grad_norm": 0.18906235694885254,
      "learning_rate": 1.8337002941424442e-05,
      "loss": 0.4023,
      "step": 1788
    },
    {
      "epoch": 1.1273393483314291,
      "grad_norm": 0.1417372077703476,
      "learning_rate": 1.8314882290301973e-05,
      "loss": 0.2812,
      "step": 1789
    },
    {
      "epoch": 1.127969741144951,
      "grad_norm": 0.2051158845424652,
      "learning_rate": 1.829276582854345e-05,
      "loss": 0.3346,
      "step": 1790
    },
    {
      "epoch": 1.1286001339584728,
      "grad_norm": 0.1490676999092102,
      "learning_rate": 1.8270653578287766e-05,
      "loss": 0.3457,
      "step": 1791
    },
    {
      "epoch": 1.1292305267719949,
      "grad_norm": 0.1721116602420807,
      "learning_rate": 1.8248545561669614e-05,
      "loss": 0.3525,
      "step": 1792
    },
    {
      "epoch": 1.1298609195855167,
      "grad_norm": 0.15207171440124512,
      "learning_rate": 1.8226441800819434e-05,
      "loss": 0.3986,
      "step": 1793
    },
    {
      "epoch": 1.1304913123990388,
      "grad_norm": 0.15704981982707977,
      "learning_rate": 1.820434231786342e-05,
      "loss": 0.3752,
      "step": 1794
    },
    {
      "epoch": 1.1311217052125606,
      "grad_norm": 0.17935959994792938,
      "learning_rate": 1.8182247134923474e-05,
      "loss": 0.4284,
      "step": 1795
    },
    {
      "epoch": 1.1317520980260825,
      "grad_norm": 0.15869015455245972,
      "learning_rate": 1.81601562741172e-05,
      "loss": 0.3213,
      "step": 1796
    },
    {
      "epoch": 1.1323824908396045,
      "grad_norm": 0.1865074187517166,
      "learning_rate": 1.8138069757557866e-05,
      "loss": 0.3612,
      "step": 1797
    },
    {
      "epoch": 1.1330128836531264,
      "grad_norm": 0.153265580534935,
      "learning_rate": 1.811598760735441e-05,
      "loss": 0.3984,
      "step": 1798
    },
    {
      "epoch": 1.1336432764666482,
      "grad_norm": 0.1499812752008438,
      "learning_rate": 1.809390984561136e-05,
      "loss": 0.3472,
      "step": 1799
    },
    {
      "epoch": 1.1342736692801703,
      "grad_norm": 0.15162378549575806,
      "learning_rate": 1.807183649442891e-05,
      "loss": 0.384,
      "step": 1800
    },
    {
      "epoch": 1.1342736692801703,
      "eval_loss": 0.42443564534187317,
      "eval_runtime": 222.4133,
      "eval_samples_per_second": 4.496,
      "eval_steps_per_second": 4.496,
      "step": 1800
    },
    {
      "epoch": 1.134904062093692,
      "grad_norm": 0.2255977839231491,
      "learning_rate": 1.8049767575902784e-05,
      "loss": 0.4607,
      "step": 1801
    },
    {
      "epoch": 1.135534454907214,
      "grad_norm": 0.15398207306861877,
      "learning_rate": 1.802770311212431e-05,
      "loss": 0.3314,
      "step": 1802
    },
    {
      "epoch": 1.136164847720736,
      "grad_norm": 0.1576896458864212,
      "learning_rate": 1.8005643125180323e-05,
      "loss": 0.3493,
      "step": 1803
    },
    {
      "epoch": 1.1367952405342578,
      "grad_norm": 0.17448103427886963,
      "learning_rate": 1.7983587637153202e-05,
      "loss": 0.3861,
      "step": 1804
    },
    {
      "epoch": 1.13742563334778,
      "grad_norm": 0.19642406702041626,
      "learning_rate": 1.79615366701208e-05,
      "loss": 0.3913,
      "step": 1805
    },
    {
      "epoch": 1.1380560261613017,
      "grad_norm": 0.1716988980770111,
      "learning_rate": 1.7939490246156474e-05,
      "loss": 0.3679,
      "step": 1806
    },
    {
      "epoch": 1.1386864189748236,
      "grad_norm": 0.1577298492193222,
      "learning_rate": 1.7917448387329e-05,
      "loss": 0.3465,
      "step": 1807
    },
    {
      "epoch": 1.1393168117883457,
      "grad_norm": 0.16155940294265747,
      "learning_rate": 1.789541111570262e-05,
      "loss": 0.3487,
      "step": 1808
    },
    {
      "epoch": 1.1399472046018675,
      "grad_norm": 0.16611319780349731,
      "learning_rate": 1.787337845333694e-05,
      "loss": 0.3402,
      "step": 1809
    },
    {
      "epoch": 1.1405775974153896,
      "grad_norm": 0.17726682126522064,
      "learning_rate": 1.7851350422286998e-05,
      "loss": 0.3009,
      "step": 1810
    },
    {
      "epoch": 1.1412079902289114,
      "grad_norm": 0.17919518053531647,
      "learning_rate": 1.7829327044603162e-05,
      "loss": 0.3766,
      "step": 1811
    },
    {
      "epoch": 1.1418383830424332,
      "grad_norm": 0.15726324915885925,
      "learning_rate": 1.7807308342331164e-05,
      "loss": 0.336,
      "step": 1812
    },
    {
      "epoch": 1.1424687758559553,
      "grad_norm": 0.18717758357524872,
      "learning_rate": 1.7785294337512036e-05,
      "loss": 0.3744,
      "step": 1813
    },
    {
      "epoch": 1.1430991686694771,
      "grad_norm": 0.1685945689678192,
      "learning_rate": 1.776328505218213e-05,
      "loss": 0.3486,
      "step": 1814
    },
    {
      "epoch": 1.1437295614829992,
      "grad_norm": 0.17545783519744873,
      "learning_rate": 1.7741280508373052e-05,
      "loss": 0.3274,
      "step": 1815
    },
    {
      "epoch": 1.144359954296521,
      "grad_norm": 0.1535610854625702,
      "learning_rate": 1.771928072811168e-05,
      "loss": 0.3411,
      "step": 1816
    },
    {
      "epoch": 1.1449903471100429,
      "grad_norm": 0.15853679180145264,
      "learning_rate": 1.7697285733420106e-05,
      "loss": 0.3439,
      "step": 1817
    },
    {
      "epoch": 1.145620739923565,
      "grad_norm": 0.199609637260437,
      "learning_rate": 1.7675295546315654e-05,
      "loss": 0.4275,
      "step": 1818
    },
    {
      "epoch": 1.1462511327370868,
      "grad_norm": 0.1530158966779709,
      "learning_rate": 1.7653310188810803e-05,
      "loss": 0.3901,
      "step": 1819
    },
    {
      "epoch": 1.1468815255506086,
      "grad_norm": 0.1571042537689209,
      "learning_rate": 1.763132968291324e-05,
      "loss": 0.3225,
      "step": 1820
    },
    {
      "epoch": 1.1475119183641307,
      "grad_norm": 0.17838561534881592,
      "learning_rate": 1.7609354050625745e-05,
      "loss": 0.3634,
      "step": 1821
    },
    {
      "epoch": 1.1481423111776525,
      "grad_norm": 0.1688927263021469,
      "learning_rate": 1.758738331394628e-05,
      "loss": 0.3285,
      "step": 1822
    },
    {
      "epoch": 1.1487727039911746,
      "grad_norm": 0.1764656901359558,
      "learning_rate": 1.756541749486784e-05,
      "loss": 0.4225,
      "step": 1823
    },
    {
      "epoch": 1.1494030968046964,
      "grad_norm": 0.172949880361557,
      "learning_rate": 1.7543456615378562e-05,
      "loss": 0.3206,
      "step": 1824
    },
    {
      "epoch": 1.1500334896182183,
      "grad_norm": 0.15606196224689484,
      "learning_rate": 1.752150069746158e-05,
      "loss": 0.3027,
      "step": 1825
    },
    {
      "epoch": 1.1506638824317403,
      "grad_norm": 0.14963605999946594,
      "learning_rate": 1.7499549763095104e-05,
      "loss": 0.3614,
      "step": 1826
    },
    {
      "epoch": 1.1512942752452622,
      "grad_norm": 0.1628952920436859,
      "learning_rate": 1.7477603834252337e-05,
      "loss": 0.3728,
      "step": 1827
    },
    {
      "epoch": 1.151924668058784,
      "grad_norm": 0.15831774473190308,
      "learning_rate": 1.745566293290147e-05,
      "loss": 0.3207,
      "step": 1828
    },
    {
      "epoch": 1.152555060872306,
      "grad_norm": 0.16237981617450714,
      "learning_rate": 1.7433727081005667e-05,
      "loss": 0.3094,
      "step": 1829
    },
    {
      "epoch": 1.153185453685828,
      "grad_norm": 0.16303876042366028,
      "learning_rate": 1.7411796300523045e-05,
      "loss": 0.3491,
      "step": 1830
    },
    {
      "epoch": 1.15381584649935,
      "grad_norm": 0.17776191234588623,
      "learning_rate": 1.738987061340662e-05,
      "loss": 0.3718,
      "step": 1831
    },
    {
      "epoch": 1.1544462393128718,
      "grad_norm": 0.16025635600090027,
      "learning_rate": 1.7367950041604338e-05,
      "loss": 0.3671,
      "step": 1832
    },
    {
      "epoch": 1.1550766321263937,
      "grad_norm": 0.1274735927581787,
      "learning_rate": 1.7346034607059002e-05,
      "loss": 0.3221,
      "step": 1833
    },
    {
      "epoch": 1.1557070249399157,
      "grad_norm": 0.15472093224525452,
      "learning_rate": 1.7324124331708295e-05,
      "loss": 0.3289,
      "step": 1834
    },
    {
      "epoch": 1.1563374177534376,
      "grad_norm": 0.17002683877944946,
      "learning_rate": 1.7302219237484705e-05,
      "loss": 0.3171,
      "step": 1835
    },
    {
      "epoch": 1.1569678105669596,
      "grad_norm": 0.14287510514259338,
      "learning_rate": 1.7280319346315566e-05,
      "loss": 0.3065,
      "step": 1836
    },
    {
      "epoch": 1.1575982033804815,
      "grad_norm": 0.18563435971736908,
      "learning_rate": 1.725842468012298e-05,
      "loss": 0.379,
      "step": 1837
    },
    {
      "epoch": 1.1582285961940033,
      "grad_norm": 0.16103070974349976,
      "learning_rate": 1.7236535260823843e-05,
      "loss": 0.3817,
      "step": 1838
    },
    {
      "epoch": 1.1588589890075254,
      "grad_norm": 0.16440634429454803,
      "learning_rate": 1.721465111032975e-05,
      "loss": 0.3782,
      "step": 1839
    },
    {
      "epoch": 1.1594893818210472,
      "grad_norm": 0.1373133808374405,
      "learning_rate": 1.7192772250547097e-05,
      "loss": 0.301,
      "step": 1840
    },
    {
      "epoch": 1.1601197746345693,
      "grad_norm": 0.17497479915618896,
      "learning_rate": 1.7170898703376905e-05,
      "loss": 0.3573,
      "step": 1841
    },
    {
      "epoch": 1.1607501674480911,
      "grad_norm": 0.13609153032302856,
      "learning_rate": 1.7149030490714945e-05,
      "loss": 0.3322,
      "step": 1842
    },
    {
      "epoch": 1.161380560261613,
      "grad_norm": 0.22188684344291687,
      "learning_rate": 1.7127167634451588e-05,
      "loss": 0.4429,
      "step": 1843
    },
    {
      "epoch": 1.162010953075135,
      "grad_norm": 0.2214953899383545,
      "learning_rate": 1.7105310156471903e-05,
      "loss": 0.419,
      "step": 1844
    },
    {
      "epoch": 1.1626413458886569,
      "grad_norm": 0.15619449317455292,
      "learning_rate": 1.708345807865552e-05,
      "loss": 0.4114,
      "step": 1845
    },
    {
      "epoch": 1.1632717387021787,
      "grad_norm": 0.15597301721572876,
      "learning_rate": 1.70616114228767e-05,
      "loss": 0.3476,
      "step": 1846
    },
    {
      "epoch": 1.1639021315157008,
      "grad_norm": 0.1534939855337143,
      "learning_rate": 1.7039770211004256e-05,
      "loss": 0.3128,
      "step": 1847
    },
    {
      "epoch": 1.1645325243292226,
      "grad_norm": 0.1397959440946579,
      "learning_rate": 1.701793446490157e-05,
      "loss": 0.3239,
      "step": 1848
    },
    {
      "epoch": 1.1651629171427447,
      "grad_norm": 0.1773579716682434,
      "learning_rate": 1.6996104206426536e-05,
      "loss": 0.3735,
      "step": 1849
    },
    {
      "epoch": 1.1657933099562665,
      "grad_norm": 0.14740358293056488,
      "learning_rate": 1.6974279457431565e-05,
      "loss": 0.3029,
      "step": 1850
    },
    {
      "epoch": 1.1664237027697884,
      "grad_norm": 0.15667730569839478,
      "learning_rate": 1.6952460239763546e-05,
      "loss": 0.3595,
      "step": 1851
    },
    {
      "epoch": 1.1670540955833104,
      "grad_norm": 0.21429184079170227,
      "learning_rate": 1.693064657526384e-05,
      "loss": 0.4161,
      "step": 1852
    },
    {
      "epoch": 1.1676844883968323,
      "grad_norm": 0.13290826976299286,
      "learning_rate": 1.690883848576824e-05,
      "loss": 0.3327,
      "step": 1853
    },
    {
      "epoch": 1.168314881210354,
      "grad_norm": 0.1636621654033661,
      "learning_rate": 1.688703599310696e-05,
      "loss": 0.3268,
      "step": 1854
    },
    {
      "epoch": 1.1689452740238762,
      "grad_norm": 0.13847965002059937,
      "learning_rate": 1.686523911910462e-05,
      "loss": 0.3375,
      "step": 1855
    },
    {
      "epoch": 1.169575666837398,
      "grad_norm": 0.1506374478340149,
      "learning_rate": 1.68434478855802e-05,
      "loss": 0.3697,
      "step": 1856
    },
    {
      "epoch": 1.17020605965092,
      "grad_norm": 0.15315689146518707,
      "learning_rate": 1.6821662314347047e-05,
      "loss": 0.3167,
      "step": 1857
    },
    {
      "epoch": 1.170836452464442,
      "grad_norm": 0.16829730570316315,
      "learning_rate": 1.6799882427212843e-05,
      "loss": 0.3697,
      "step": 1858
    },
    {
      "epoch": 1.1714668452779637,
      "grad_norm": 0.18139970302581787,
      "learning_rate": 1.6778108245979564e-05,
      "loss": 0.3428,
      "step": 1859
    },
    {
      "epoch": 1.1720972380914858,
      "grad_norm": 0.17139574885368347,
      "learning_rate": 1.6756339792443494e-05,
      "loss": 0.3537,
      "step": 1860
    },
    {
      "epoch": 1.1727276309050076,
      "grad_norm": 0.17136725783348083,
      "learning_rate": 1.6734577088395147e-05,
      "loss": 0.3734,
      "step": 1861
    },
    {
      "epoch": 1.1733580237185297,
      "grad_norm": 0.1732012778520584,
      "learning_rate": 1.6712820155619348e-05,
      "loss": 0.4209,
      "step": 1862
    },
    {
      "epoch": 1.1739884165320515,
      "grad_norm": 0.17362134158611298,
      "learning_rate": 1.669106901589506e-05,
      "loss": 0.3422,
      "step": 1863
    },
    {
      "epoch": 1.1746188093455734,
      "grad_norm": 0.1846878081560135,
      "learning_rate": 1.6669323690995532e-05,
      "loss": 0.4042,
      "step": 1864
    },
    {
      "epoch": 1.1752492021590955,
      "grad_norm": 0.1782120168209076,
      "learning_rate": 1.6647584202688126e-05,
      "loss": 0.3604,
      "step": 1865
    },
    {
      "epoch": 1.1758795949726173,
      "grad_norm": 0.15381821990013123,
      "learning_rate": 1.6625850572734398e-05,
      "loss": 0.3528,
      "step": 1866
    },
    {
      "epoch": 1.1765099877861394,
      "grad_norm": 0.20381847023963928,
      "learning_rate": 1.6604122822890017e-05,
      "loss": 0.3198,
      "step": 1867
    },
    {
      "epoch": 1.1771403805996612,
      "grad_norm": 0.16701090335845947,
      "learning_rate": 1.6582400974904797e-05,
      "loss": 0.3492,
      "step": 1868
    },
    {
      "epoch": 1.177770773413183,
      "grad_norm": 0.14173492789268494,
      "learning_rate": 1.656068505052261e-05,
      "loss": 0.2858,
      "step": 1869
    },
    {
      "epoch": 1.178401166226705,
      "grad_norm": 0.16685733199119568,
      "learning_rate": 1.653897507148142e-05,
      "loss": 0.3855,
      "step": 1870
    },
    {
      "epoch": 1.179031559040227,
      "grad_norm": 0.15488892793655396,
      "learning_rate": 1.6517271059513225e-05,
      "loss": 0.3503,
      "step": 1871
    },
    {
      "epoch": 1.1796619518537488,
      "grad_norm": 0.15705163776874542,
      "learning_rate": 1.649557303634407e-05,
      "loss": 0.3356,
      "step": 1872
    },
    {
      "epoch": 1.1802923446672708,
      "grad_norm": 0.17007462680339813,
      "learning_rate": 1.6473881023693984e-05,
      "loss": 0.3568,
      "step": 1873
    },
    {
      "epoch": 1.1809227374807927,
      "grad_norm": 0.15087947249412537,
      "learning_rate": 1.6452195043276995e-05,
      "loss": 0.3495,
      "step": 1874
    },
    {
      "epoch": 1.1815531302943145,
      "grad_norm": 0.16111180186271667,
      "learning_rate": 1.643051511680108e-05,
      "loss": 0.3755,
      "step": 1875
    },
    {
      "epoch": 1.1821835231078366,
      "grad_norm": 0.16729417443275452,
      "learning_rate": 1.6408841265968166e-05,
      "loss": 0.3515,
      "step": 1876
    },
    {
      "epoch": 1.1828139159213584,
      "grad_norm": 0.20641054213047028,
      "learning_rate": 1.638717351247409e-05,
      "loss": 0.4274,
      "step": 1877
    },
    {
      "epoch": 1.1834443087348805,
      "grad_norm": 0.15898416936397552,
      "learning_rate": 1.6365511878008594e-05,
      "loss": 0.3482,
      "step": 1878
    },
    {
      "epoch": 1.1840747015484023,
      "grad_norm": 0.15798181295394897,
      "learning_rate": 1.6343856384255288e-05,
      "loss": 0.3218,
      "step": 1879
    },
    {
      "epoch": 1.1847050943619242,
      "grad_norm": 0.1792026311159134,
      "learning_rate": 1.6322207052891648e-05,
      "loss": 0.3921,
      "step": 1880
    },
    {
      "epoch": 1.1853354871754462,
      "grad_norm": 0.1666623055934906,
      "learning_rate": 1.6300563905588945e-05,
      "loss": 0.3244,
      "step": 1881
    },
    {
      "epoch": 1.185965879988968,
      "grad_norm": 0.1739160269498825,
      "learning_rate": 1.6278926964012316e-05,
      "loss": 0.3323,
      "step": 1882
    },
    {
      "epoch": 1.1865962728024901,
      "grad_norm": 0.15621982514858246,
      "learning_rate": 1.6257296249820628e-05,
      "loss": 0.3212,
      "step": 1883
    },
    {
      "epoch": 1.187226665616012,
      "grad_norm": 0.13840575516223907,
      "learning_rate": 1.6235671784666564e-05,
      "loss": 0.3351,
      "step": 1884
    },
    {
      "epoch": 1.1878570584295338,
      "grad_norm": 0.14696210622787476,
      "learning_rate": 1.6214053590196517e-05,
      "loss": 0.3669,
      "step": 1885
    },
    {
      "epoch": 1.1884874512430559,
      "grad_norm": 0.17982372641563416,
      "learning_rate": 1.6192441688050623e-05,
      "loss": 0.3206,
      "step": 1886
    },
    {
      "epoch": 1.1891178440565777,
      "grad_norm": 0.13978564739227295,
      "learning_rate": 1.6170836099862697e-05,
      "loss": 0.339,
      "step": 1887
    },
    {
      "epoch": 1.1897482368700998,
      "grad_norm": 0.12686683237552643,
      "learning_rate": 1.614923684726027e-05,
      "loss": 0.2845,
      "step": 1888
    },
    {
      "epoch": 1.1903786296836216,
      "grad_norm": 0.17271322011947632,
      "learning_rate": 1.6127643951864488e-05,
      "loss": 0.3649,
      "step": 1889
    },
    {
      "epoch": 1.1910090224971435,
      "grad_norm": 0.15250203013420105,
      "learning_rate": 1.6106057435290175e-05,
      "loss": 0.3493,
      "step": 1890
    },
    {
      "epoch": 1.1916394153106655,
      "grad_norm": 0.21040686964988708,
      "learning_rate": 1.608447731914573e-05,
      "loss": 0.4127,
      "step": 1891
    },
    {
      "epoch": 1.1922698081241874,
      "grad_norm": 0.16836979985237122,
      "learning_rate": 1.6062903625033177e-05,
      "loss": 0.378,
      "step": 1892
    },
    {
      "epoch": 1.1929002009377094,
      "grad_norm": 0.16101545095443726,
      "learning_rate": 1.6041336374548093e-05,
      "loss": 0.3219,
      "step": 1893
    },
    {
      "epoch": 1.1935305937512313,
      "grad_norm": 0.1391444206237793,
      "learning_rate": 1.6019775589279616e-05,
      "loss": 0.3327,
      "step": 1894
    },
    {
      "epoch": 1.1941609865647531,
      "grad_norm": 0.16221654415130615,
      "learning_rate": 1.5998221290810398e-05,
      "loss": 0.3633,
      "step": 1895
    },
    {
      "epoch": 1.1947913793782752,
      "grad_norm": 0.15801361203193665,
      "learning_rate": 1.597667350071662e-05,
      "loss": 0.3559,
      "step": 1896
    },
    {
      "epoch": 1.195421772191797,
      "grad_norm": 0.17644530534744263,
      "learning_rate": 1.5955132240567918e-05,
      "loss": 0.3376,
      "step": 1897
    },
    {
      "epoch": 1.1960521650053189,
      "grad_norm": 0.17779093980789185,
      "learning_rate": 1.5933597531927422e-05,
      "loss": 0.3468,
      "step": 1898
    },
    {
      "epoch": 1.196682557818841,
      "grad_norm": 0.17327743768692017,
      "learning_rate": 1.591206939635168e-05,
      "loss": 0.3524,
      "step": 1899
    },
    {
      "epoch": 1.1973129506323628,
      "grad_norm": 0.18291594088077545,
      "learning_rate": 1.5890547855390687e-05,
      "loss": 0.3162,
      "step": 1900
    },
    {
      "epoch": 1.1979433434458846,
      "grad_norm": 0.18050126731395721,
      "learning_rate": 1.5869032930587788e-05,
      "loss": 0.3384,
      "step": 1901
    },
    {
      "epoch": 1.1985737362594067,
      "grad_norm": 0.1861979067325592,
      "learning_rate": 1.5847524643479776e-05,
      "loss": 0.3859,
      "step": 1902
    },
    {
      "epoch": 1.1992041290729285,
      "grad_norm": 0.15310519933700562,
      "learning_rate": 1.5826023015596723e-05,
      "loss": 0.2921,
      "step": 1903
    },
    {
      "epoch": 1.1998345218864506,
      "grad_norm": 0.18371252715587616,
      "learning_rate": 1.5804528068462103e-05,
      "loss": 0.3805,
      "step": 1904
    },
    {
      "epoch": 1.2004649146999724,
      "grad_norm": 0.19243298470973969,
      "learning_rate": 1.578303982359265e-05,
      "loss": 0.4009,
      "step": 1905
    },
    {
      "epoch": 1.2010953075134942,
      "grad_norm": 0.17407071590423584,
      "learning_rate": 1.576155830249843e-05,
      "loss": 0.3416,
      "step": 1906
    },
    {
      "epoch": 1.2017257003270163,
      "grad_norm": 0.12393369525671005,
      "learning_rate": 1.574008352668274e-05,
      "loss": 0.2966,
      "step": 1907
    },
    {
      "epoch": 1.2023560931405382,
      "grad_norm": 0.20292018353939056,
      "learning_rate": 1.571861551764216e-05,
      "loss": 0.3988,
      "step": 1908
    },
    {
      "epoch": 1.2029864859540602,
      "grad_norm": 0.15854638814926147,
      "learning_rate": 1.5697154296866468e-05,
      "loss": 0.3335,
      "step": 1909
    },
    {
      "epoch": 1.203616878767582,
      "grad_norm": 0.1645781546831131,
      "learning_rate": 1.567569988583867e-05,
      "loss": 0.3353,
      "step": 1910
    },
    {
      "epoch": 1.204247271581104,
      "grad_norm": 0.1488713175058365,
      "learning_rate": 1.5654252306034934e-05,
      "loss": 0.3427,
      "step": 1911
    },
    {
      "epoch": 1.204877664394626,
      "grad_norm": 0.14664226770401,
      "learning_rate": 1.5632811578924614e-05,
      "loss": 0.3492,
      "step": 1912
    },
    {
      "epoch": 1.2055080572081478,
      "grad_norm": 0.18963125348091125,
      "learning_rate": 1.5611377725970182e-05,
      "loss": 0.3855,
      "step": 1913
    },
    {
      "epoch": 1.2061384500216699,
      "grad_norm": 0.14855507016181946,
      "learning_rate": 1.558995076862724e-05,
      "loss": 0.3235,
      "step": 1914
    },
    {
      "epoch": 1.2067688428351917,
      "grad_norm": 0.14204628765583038,
      "learning_rate": 1.556853072834448e-05,
      "loss": 0.3609,
      "step": 1915
    },
    {
      "epoch": 1.2073992356487135,
      "grad_norm": 0.13496729731559753,
      "learning_rate": 1.5547117626563687e-05,
      "loss": 0.2936,
      "step": 1916
    },
    {
      "epoch": 1.2080296284622356,
      "grad_norm": 0.16182400286197662,
      "learning_rate": 1.5525711484719676e-05,
      "loss": 0.2771,
      "step": 1917
    },
    {
      "epoch": 1.2086600212757574,
      "grad_norm": 0.14185205101966858,
      "learning_rate": 1.5504312324240312e-05,
      "loss": 0.2653,
      "step": 1918
    },
    {
      "epoch": 1.2092904140892793,
      "grad_norm": 0.13705161213874817,
      "learning_rate": 1.5482920166546473e-05,
      "loss": 0.3319,
      "step": 1919
    },
    {
      "epoch": 1.2099208069028013,
      "grad_norm": 0.15866032242774963,
      "learning_rate": 1.546153503305202e-05,
      "loss": 0.3551,
      "step": 1920
    },
    {
      "epoch": 1.2105511997163232,
      "grad_norm": 0.15233084559440613,
      "learning_rate": 1.5440156945163762e-05,
      "loss": 0.3683,
      "step": 1921
    },
    {
      "epoch": 1.2111815925298453,
      "grad_norm": 0.17361125349998474,
      "learning_rate": 1.5418785924281516e-05,
      "loss": 0.3799,
      "step": 1922
    },
    {
      "epoch": 1.211811985343367,
      "grad_norm": 0.15449875593185425,
      "learning_rate": 1.5397421991797952e-05,
      "loss": 0.3449,
      "step": 1923
    },
    {
      "epoch": 1.212442378156889,
      "grad_norm": 0.1920737475156784,
      "learning_rate": 1.53760651690987e-05,
      "loss": 0.3539,
      "step": 1924
    },
    {
      "epoch": 1.213072770970411,
      "grad_norm": 0.1675095111131668,
      "learning_rate": 1.5354715477562236e-05,
      "loss": 0.3222,
      "step": 1925
    },
    {
      "epoch": 1.2137031637839328,
      "grad_norm": 0.15904296934604645,
      "learning_rate": 1.533337293855993e-05,
      "loss": 0.3203,
      "step": 1926
    },
    {
      "epoch": 1.2143335565974547,
      "grad_norm": 0.1559150665998459,
      "learning_rate": 1.5312037573455955e-05,
      "loss": 0.3502,
      "step": 1927
    },
    {
      "epoch": 1.2149639494109767,
      "grad_norm": 0.14907754957675934,
      "learning_rate": 1.529070940360734e-05,
      "loss": 0.3525,
      "step": 1928
    },
    {
      "epoch": 1.2155943422244986,
      "grad_norm": 0.19221338629722595,
      "learning_rate": 1.526938845036388e-05,
      "loss": 0.3653,
      "step": 1929
    },
    {
      "epoch": 1.2162247350380206,
      "grad_norm": 0.14959098398685455,
      "learning_rate": 1.5248074735068178e-05,
      "loss": 0.3614,
      "step": 1930
    },
    {
      "epoch": 1.2168551278515425,
      "grad_norm": 0.15030638873577118,
      "learning_rate": 1.5226768279055559e-05,
      "loss": 0.3303,
      "step": 1931
    },
    {
      "epoch": 1.2174855206650643,
      "grad_norm": 0.1545165628194809,
      "learning_rate": 1.5205469103654104e-05,
      "loss": 0.3569,
      "step": 1932
    },
    {
      "epoch": 1.2181159134785864,
      "grad_norm": 0.16187258064746857,
      "learning_rate": 1.5184177230184595e-05,
      "loss": 0.3707,
      "step": 1933
    },
    {
      "epoch": 1.2187463062921082,
      "grad_norm": 0.1676671952009201,
      "learning_rate": 1.5162892679960513e-05,
      "loss": 0.3382,
      "step": 1934
    },
    {
      "epoch": 1.2193766991056303,
      "grad_norm": 0.1787586361169815,
      "learning_rate": 1.5141615474287996e-05,
      "loss": 0.33,
      "step": 1935
    },
    {
      "epoch": 1.2200070919191521,
      "grad_norm": 0.18215793371200562,
      "learning_rate": 1.5120345634465848e-05,
      "loss": 0.3825,
      "step": 1936
    },
    {
      "epoch": 1.220637484732674,
      "grad_norm": 0.17851123213768005,
      "learning_rate": 1.5099083181785479e-05,
      "loss": 0.4282,
      "step": 1937
    },
    {
      "epoch": 1.221267877546196,
      "grad_norm": 0.17377333343029022,
      "learning_rate": 1.5077828137530923e-05,
      "loss": 0.3273,
      "step": 1938
    },
    {
      "epoch": 1.2218982703597179,
      "grad_norm": 0.16321879625320435,
      "learning_rate": 1.5056580522978783e-05,
      "loss": 0.3081,
      "step": 1939
    },
    {
      "epoch": 1.22252866317324,
      "grad_norm": 0.1564747542142868,
      "learning_rate": 1.5035340359398237e-05,
      "loss": 0.3613,
      "step": 1940
    },
    {
      "epoch": 1.2231590559867618,
      "grad_norm": 0.1720079630613327,
      "learning_rate": 1.5014107668050992e-05,
      "loss": 0.3934,
      "step": 1941
    },
    {
      "epoch": 1.2237894488002836,
      "grad_norm": 0.15148408710956573,
      "learning_rate": 1.4992882470191297e-05,
      "loss": 0.2705,
      "step": 1942
    },
    {
      "epoch": 1.2244198416138057,
      "grad_norm": 0.1540118306875229,
      "learning_rate": 1.4971664787065858e-05,
      "loss": 0.309,
      "step": 1943
    },
    {
      "epoch": 1.2250502344273275,
      "grad_norm": 0.14092116057872772,
      "learning_rate": 1.4950454639913916e-05,
      "loss": 0.3714,
      "step": 1944
    },
    {
      "epoch": 1.2256806272408494,
      "grad_norm": 0.1863245815038681,
      "learning_rate": 1.492925204996711e-05,
      "loss": 0.4278,
      "step": 1945
    },
    {
      "epoch": 1.2263110200543714,
      "grad_norm": 0.14283518493175507,
      "learning_rate": 1.4908057038449573e-05,
      "loss": 0.2725,
      "step": 1946
    },
    {
      "epoch": 1.2269414128678933,
      "grad_norm": 0.1478254646062851,
      "learning_rate": 1.4886869626577791e-05,
      "loss": 0.339,
      "step": 1947
    },
    {
      "epoch": 1.2275718056814153,
      "grad_norm": 0.1424347460269928,
      "learning_rate": 1.4865689835560691e-05,
      "loss": 0.2964,
      "step": 1948
    },
    {
      "epoch": 1.2282021984949372,
      "grad_norm": 0.1579347401857376,
      "learning_rate": 1.4844517686599545e-05,
      "loss": 0.3859,
      "step": 1949
    },
    {
      "epoch": 1.228832591308459,
      "grad_norm": 0.1495109647512436,
      "learning_rate": 1.482335320088799e-05,
      "loss": 0.3267,
      "step": 1950
    },
    {
      "epoch": 1.229462984121981,
      "grad_norm": 0.14779378473758698,
      "learning_rate": 1.4802196399611978e-05,
      "loss": 0.3386,
      "step": 1951
    },
    {
      "epoch": 1.230093376935503,
      "grad_norm": 0.13087581098079681,
      "learning_rate": 1.4781047303949784e-05,
      "loss": 0.3595,
      "step": 1952
    },
    {
      "epoch": 1.2307237697490248,
      "grad_norm": 0.18340842425823212,
      "learning_rate": 1.4759905935071952e-05,
      "loss": 0.3052,
      "step": 1953
    },
    {
      "epoch": 1.2313541625625468,
      "grad_norm": 0.1894017606973648,
      "learning_rate": 1.473877231414131e-05,
      "loss": 0.3803,
      "step": 1954
    },
    {
      "epoch": 1.2319845553760687,
      "grad_norm": 0.17660388350486755,
      "learning_rate": 1.4717646462312912e-05,
      "loss": 0.3708,
      "step": 1955
    },
    {
      "epoch": 1.2326149481895907,
      "grad_norm": 0.19020117819309235,
      "learning_rate": 1.4696528400734054e-05,
      "loss": 0.2913,
      "step": 1956
    },
    {
      "epoch": 1.2332453410031126,
      "grad_norm": 0.15870848298072815,
      "learning_rate": 1.4675418150544217e-05,
      "loss": 0.38,
      "step": 1957
    },
    {
      "epoch": 1.2338757338166344,
      "grad_norm": 0.1692182570695877,
      "learning_rate": 1.4654315732875073e-05,
      "loss": 0.3718,
      "step": 1958
    },
    {
      "epoch": 1.2345061266301565,
      "grad_norm": 0.19379208981990814,
      "learning_rate": 1.463322116885045e-05,
      "loss": 0.4182,
      "step": 1959
    },
    {
      "epoch": 1.2351365194436783,
      "grad_norm": 0.14317487180233002,
      "learning_rate": 1.4612134479586316e-05,
      "loss": 0.338,
      "step": 1960
    },
    {
      "epoch": 1.2357669122572004,
      "grad_norm": 0.14631275832653046,
      "learning_rate": 1.4591055686190749e-05,
      "loss": 0.2957,
      "step": 1961
    },
    {
      "epoch": 1.2363973050707222,
      "grad_norm": 0.1501506268978119,
      "learning_rate": 1.4569984809763939e-05,
      "loss": 0.3617,
      "step": 1962
    },
    {
      "epoch": 1.237027697884244,
      "grad_norm": 0.14559954404830933,
      "learning_rate": 1.4548921871398122e-05,
      "loss": 0.3309,
      "step": 1963
    },
    {
      "epoch": 1.2376580906977661,
      "grad_norm": 0.16393865644931793,
      "learning_rate": 1.4527866892177639e-05,
      "loss": 0.3721,
      "step": 1964
    },
    {
      "epoch": 1.238288483511288,
      "grad_norm": 0.15368694067001343,
      "learning_rate": 1.4506819893178792e-05,
      "loss": 0.3812,
      "step": 1965
    },
    {
      "epoch": 1.23891887632481,
      "grad_norm": 0.17005257308483124,
      "learning_rate": 1.4485780895469974e-05,
      "loss": 0.3725,
      "step": 1966
    },
    {
      "epoch": 1.2395492691383319,
      "grad_norm": 0.17467091977596283,
      "learning_rate": 1.4464749920111495e-05,
      "loss": 0.337,
      "step": 1967
    },
    {
      "epoch": 1.2401796619518537,
      "grad_norm": 0.14071719348430634,
      "learning_rate": 1.444372698815569e-05,
      "loss": 0.3416,
      "step": 1968
    },
    {
      "epoch": 1.2408100547653758,
      "grad_norm": 0.13080553710460663,
      "learning_rate": 1.4422712120646809e-05,
      "loss": 0.288,
      "step": 1969
    },
    {
      "epoch": 1.2414404475788976,
      "grad_norm": 0.1427469402551651,
      "learning_rate": 1.4401705338621053e-05,
      "loss": 0.3098,
      "step": 1970
    },
    {
      "epoch": 1.2420708403924194,
      "grad_norm": 0.15654031932353973,
      "learning_rate": 1.4380706663106503e-05,
      "loss": 0.345,
      "step": 1971
    },
    {
      "epoch": 1.2427012332059415,
      "grad_norm": 0.18579089641571045,
      "learning_rate": 1.4359716115123156e-05,
      "loss": 0.3677,
      "step": 1972
    },
    {
      "epoch": 1.2433316260194633,
      "grad_norm": 0.15906280279159546,
      "learning_rate": 1.4338733715682842e-05,
      "loss": 0.356,
      "step": 1973
    },
    {
      "epoch": 1.2439620188329852,
      "grad_norm": 0.16996033489704132,
      "learning_rate": 1.4317759485789262e-05,
      "loss": 0.3887,
      "step": 1974
    },
    {
      "epoch": 1.2445924116465072,
      "grad_norm": 0.16264496743679047,
      "learning_rate": 1.4296793446437915e-05,
      "loss": 0.3404,
      "step": 1975
    },
    {
      "epoch": 1.245222804460029,
      "grad_norm": 0.18813878297805786,
      "learning_rate": 1.427583561861612e-05,
      "loss": 0.3911,
      "step": 1976
    },
    {
      "epoch": 1.2458531972735511,
      "grad_norm": 0.17374739050865173,
      "learning_rate": 1.4254886023302962e-05,
      "loss": 0.3744,
      "step": 1977
    },
    {
      "epoch": 1.246483590087073,
      "grad_norm": 0.16600753366947174,
      "learning_rate": 1.4233944681469304e-05,
      "loss": 0.3308,
      "step": 1978
    },
    {
      "epoch": 1.2471139829005948,
      "grad_norm": 0.16969601809978485,
      "learning_rate": 1.4213011614077722e-05,
      "loss": 0.3617,
      "step": 1979
    },
    {
      "epoch": 1.247744375714117,
      "grad_norm": 0.14293773472309113,
      "learning_rate": 1.4192086842082532e-05,
      "loss": 0.3456,
      "step": 1980
    },
    {
      "epoch": 1.2483747685276387,
      "grad_norm": 0.1606130748987198,
      "learning_rate": 1.4171170386429728e-05,
      "loss": 0.3757,
      "step": 1981
    },
    {
      "epoch": 1.2490051613411608,
      "grad_norm": 0.15878310799598694,
      "learning_rate": 1.4150262268057002e-05,
      "loss": 0.3353,
      "step": 1982
    },
    {
      "epoch": 1.2496355541546826,
      "grad_norm": 0.17191563546657562,
      "learning_rate": 1.4129362507893657e-05,
      "loss": 0.354,
      "step": 1983
    },
    {
      "epoch": 1.2502659469682045,
      "grad_norm": 0.16638456284999847,
      "learning_rate": 1.4108471126860693e-05,
      "loss": 0.378,
      "step": 1984
    },
    {
      "epoch": 1.2508963397817265,
      "grad_norm": 0.16778291761875153,
      "learning_rate": 1.4087588145870657e-05,
      "loss": 0.3866,
      "step": 1985
    },
    {
      "epoch": 1.2515267325952484,
      "grad_norm": 0.1632990837097168,
      "learning_rate": 1.4066713585827747e-05,
      "loss": 0.3486,
      "step": 1986
    },
    {
      "epoch": 1.2521571254087704,
      "grad_norm": 0.15390026569366455,
      "learning_rate": 1.404584746762767e-05,
      "loss": 0.3778,
      "step": 1987
    },
    {
      "epoch": 1.2527875182222923,
      "grad_norm": 0.15472039580345154,
      "learning_rate": 1.4024989812157745e-05,
      "loss": 0.3031,
      "step": 1988
    },
    {
      "epoch": 1.2534179110358141,
      "grad_norm": 0.14273808896541595,
      "learning_rate": 1.400414064029677e-05,
      "loss": 0.3105,
      "step": 1989
    },
    {
      "epoch": 1.2540483038493362,
      "grad_norm": 0.13901756703853607,
      "learning_rate": 1.3983299972915085e-05,
      "loss": 0.3639,
      "step": 1990
    },
    {
      "epoch": 1.254678696662858,
      "grad_norm": 0.15769080817699432,
      "learning_rate": 1.3962467830874486e-05,
      "loss": 0.3572,
      "step": 1991
    },
    {
      "epoch": 1.25530908947638,
      "grad_norm": 0.18810272216796875,
      "learning_rate": 1.3941644235028269e-05,
      "loss": 0.3733,
      "step": 1992
    },
    {
      "epoch": 1.255939482289902,
      "grad_norm": 0.13771487772464752,
      "learning_rate": 1.3920829206221138e-05,
      "loss": 0.3542,
      "step": 1993
    },
    {
      "epoch": 1.2565698751034238,
      "grad_norm": 0.18687589466571808,
      "learning_rate": 1.3900022765289261e-05,
      "loss": 0.3899,
      "step": 1994
    },
    {
      "epoch": 1.2572002679169456,
      "grad_norm": 0.15968801081180573,
      "learning_rate": 1.3879224933060176e-05,
      "loss": 0.4121,
      "step": 1995
    },
    {
      "epoch": 1.2578306607304677,
      "grad_norm": 0.1799834817647934,
      "learning_rate": 1.3858435730352819e-05,
      "loss": 0.3319,
      "step": 1996
    },
    {
      "epoch": 1.2584610535439895,
      "grad_norm": 0.20274387300014496,
      "learning_rate": 1.3837655177977489e-05,
      "loss": 0.3676,
      "step": 1997
    },
    {
      "epoch": 1.2590914463575116,
      "grad_norm": 0.19322608411312103,
      "learning_rate": 1.3816883296735821e-05,
      "loss": 0.4142,
      "step": 1998
    },
    {
      "epoch": 1.2597218391710334,
      "grad_norm": 0.18503984808921814,
      "learning_rate": 1.3796120107420767e-05,
      "loss": 0.3196,
      "step": 1999
    },
    {
      "epoch": 1.2603522319845553,
      "grad_norm": 0.17512302100658417,
      "learning_rate": 1.3775365630816591e-05,
      "loss": 0.4094,
      "step": 2000
    },
    {
      "epoch": 1.2603522319845553,
      "eval_loss": 0.4224383533000946,
      "eval_runtime": 222.5349,
      "eval_samples_per_second": 4.494,
      "eval_steps_per_second": 4.494,
      "step": 2000
    },
    {
      "epoch": 1.2609826247980773,
      "grad_norm": 0.17680183053016663,
      "learning_rate": 1.375461988769882e-05,
      "loss": 0.3257,
      "step": 2001
    },
    {
      "epoch": 1.2616130176115992,
      "grad_norm": 0.16907788813114166,
      "learning_rate": 1.3733882898834255e-05,
      "loss": 0.3264,
      "step": 2002
    },
    {
      "epoch": 1.2622434104251212,
      "grad_norm": 0.19145682454109192,
      "learning_rate": 1.3713154684980903e-05,
      "loss": 0.3352,
      "step": 2003
    },
    {
      "epoch": 1.262873803238643,
      "grad_norm": 0.15243153274059296,
      "learning_rate": 1.3692435266888044e-05,
      "loss": 0.3908,
      "step": 2004
    },
    {
      "epoch": 1.263504196052165,
      "grad_norm": 0.17341181635856628,
      "learning_rate": 1.367172466529608e-05,
      "loss": 0.3609,
      "step": 2005
    },
    {
      "epoch": 1.264134588865687,
      "grad_norm": 0.207267627120018,
      "learning_rate": 1.3651022900936656e-05,
      "loss": 0.3579,
      "step": 2006
    },
    {
      "epoch": 1.2647649816792088,
      "grad_norm": 0.15846067667007446,
      "learning_rate": 1.3630329994532515e-05,
      "loss": 0.3012,
      "step": 2007
    },
    {
      "epoch": 1.2653953744927309,
      "grad_norm": 0.15553094446659088,
      "learning_rate": 1.3609645966797588e-05,
      "loss": 0.3624,
      "step": 2008
    },
    {
      "epoch": 1.2660257673062527,
      "grad_norm": 0.16129595041275024,
      "learning_rate": 1.3588970838436862e-05,
      "loss": 0.3522,
      "step": 2009
    },
    {
      "epoch": 1.2666561601197746,
      "grad_norm": 0.16139130294322968,
      "learning_rate": 1.3568304630146457e-05,
      "loss": 0.3419,
      "step": 2010
    },
    {
      "epoch": 1.2672865529332966,
      "grad_norm": 0.1251128762960434,
      "learning_rate": 1.3547647362613538e-05,
      "loss": 0.299,
      "step": 2011
    },
    {
      "epoch": 1.2679169457468185,
      "grad_norm": 0.17684251070022583,
      "learning_rate": 1.3526999056516346e-05,
      "loss": 0.3454,
      "step": 2012
    },
    {
      "epoch": 1.2685473385603405,
      "grad_norm": 0.16017553210258484,
      "learning_rate": 1.3506359732524123e-05,
      "loss": 0.3684,
      "step": 2013
    },
    {
      "epoch": 1.2691777313738624,
      "grad_norm": 0.14525946974754333,
      "learning_rate": 1.3485729411297147e-05,
      "loss": 0.3145,
      "step": 2014
    },
    {
      "epoch": 1.2698081241873842,
      "grad_norm": 0.1517396867275238,
      "learning_rate": 1.3465108113486656e-05,
      "loss": 0.3048,
      "step": 2015
    },
    {
      "epoch": 1.2704385170009063,
      "grad_norm": 0.12699000537395477,
      "learning_rate": 1.3444495859734885e-05,
      "loss": 0.2737,
      "step": 2016
    },
    {
      "epoch": 1.271068909814428,
      "grad_norm": 0.14624783396720886,
      "learning_rate": 1.3423892670674989e-05,
      "loss": 0.3573,
      "step": 2017
    },
    {
      "epoch": 1.2716993026279502,
      "grad_norm": 0.16006223857402802,
      "learning_rate": 1.340329856693107e-05,
      "loss": 0.3779,
      "step": 2018
    },
    {
      "epoch": 1.272329695441472,
      "grad_norm": 0.17546787858009338,
      "learning_rate": 1.3382713569118122e-05,
      "loss": 0.3166,
      "step": 2019
    },
    {
      "epoch": 1.2729600882549938,
      "grad_norm": 0.1470988541841507,
      "learning_rate": 1.3362137697842032e-05,
      "loss": 0.3314,
      "step": 2020
    },
    {
      "epoch": 1.2735904810685157,
      "grad_norm": 0.16196255385875702,
      "learning_rate": 1.3341570973699536e-05,
      "loss": 0.367,
      "step": 2021
    },
    {
      "epoch": 1.2742208738820378,
      "grad_norm": 0.15870201587677002,
      "learning_rate": 1.3321013417278243e-05,
      "loss": 0.3421,
      "step": 2022
    },
    {
      "epoch": 1.2748512666955596,
      "grad_norm": 0.14421947300434113,
      "learning_rate": 1.3300465049156553e-05,
      "loss": 0.2975,
      "step": 2023
    },
    {
      "epoch": 1.2754816595090817,
      "grad_norm": 0.16223183274269104,
      "learning_rate": 1.3279925889903695e-05,
      "loss": 0.3614,
      "step": 2024
    },
    {
      "epoch": 1.2761120523226035,
      "grad_norm": 0.17637160420417786,
      "learning_rate": 1.325939596007965e-05,
      "loss": 0.4007,
      "step": 2025
    },
    {
      "epoch": 1.2767424451361253,
      "grad_norm": 0.17480182647705078,
      "learning_rate": 1.32388752802352e-05,
      "loss": 0.3837,
      "step": 2026
    },
    {
      "epoch": 1.2773728379496474,
      "grad_norm": 0.15375201404094696,
      "learning_rate": 1.3218363870911818e-05,
      "loss": 0.253,
      "step": 2027
    },
    {
      "epoch": 1.2780032307631692,
      "grad_norm": 0.1700395941734314,
      "learning_rate": 1.3197861752641752e-05,
      "loss": 0.3247,
      "step": 2028
    },
    {
      "epoch": 1.2786336235766913,
      "grad_norm": 0.17732472717761993,
      "learning_rate": 1.3177368945947902e-05,
      "loss": 0.3686,
      "step": 2029
    },
    {
      "epoch": 1.2792640163902131,
      "grad_norm": 0.15792888402938843,
      "learning_rate": 1.3156885471343877e-05,
      "loss": 0.3475,
      "step": 2030
    },
    {
      "epoch": 1.279894409203735,
      "grad_norm": 0.16673409938812256,
      "learning_rate": 1.3136411349333921e-05,
      "loss": 0.3013,
      "step": 2031
    },
    {
      "epoch": 1.280524802017257,
      "grad_norm": 0.14620770514011383,
      "learning_rate": 1.3115946600412955e-05,
      "loss": 0.3683,
      "step": 2032
    },
    {
      "epoch": 1.2811551948307789,
      "grad_norm": 0.17873959243297577,
      "learning_rate": 1.3095491245066465e-05,
      "loss": 0.3408,
      "step": 2033
    },
    {
      "epoch": 1.281785587644301,
      "grad_norm": 0.1617429405450821,
      "learning_rate": 1.3075045303770581e-05,
      "loss": 0.359,
      "step": 2034
    },
    {
      "epoch": 1.2824159804578228,
      "grad_norm": 0.187462717294693,
      "learning_rate": 1.3054608796991957e-05,
      "loss": 0.3613,
      "step": 2035
    },
    {
      "epoch": 1.2830463732713446,
      "grad_norm": 0.14734844863414764,
      "learning_rate": 1.3034181745187877e-05,
      "loss": 0.3299,
      "step": 2036
    },
    {
      "epoch": 1.2836767660848667,
      "grad_norm": 0.18680807948112488,
      "learning_rate": 1.3013764168806091e-05,
      "loss": 0.3326,
      "step": 2037
    },
    {
      "epoch": 1.2843071588983885,
      "grad_norm": 0.1547849178314209,
      "learning_rate": 1.2993356088284902e-05,
      "loss": 0.3732,
      "step": 2038
    },
    {
      "epoch": 1.2849375517119106,
      "grad_norm": 0.15319564938545227,
      "learning_rate": 1.2972957524053088e-05,
      "loss": 0.3105,
      "step": 2039
    },
    {
      "epoch": 1.2855679445254324,
      "grad_norm": 0.16970223188400269,
      "learning_rate": 1.2952568496529912e-05,
      "loss": 0.3467,
      "step": 2040
    },
    {
      "epoch": 1.2861983373389543,
      "grad_norm": 0.18748734891414642,
      "learning_rate": 1.2932189026125086e-05,
      "loss": 0.3412,
      "step": 2041
    },
    {
      "epoch": 1.2868287301524763,
      "grad_norm": 0.15902206301689148,
      "learning_rate": 1.2911819133238775e-05,
      "loss": 0.3001,
      "step": 2042
    },
    {
      "epoch": 1.2874591229659982,
      "grad_norm": 0.20623432099819183,
      "learning_rate": 1.2891458838261515e-05,
      "loss": 0.3626,
      "step": 2043
    },
    {
      "epoch": 1.2880895157795202,
      "grad_norm": 0.18049001693725586,
      "learning_rate": 1.2871108161574274e-05,
      "loss": 0.3765,
      "step": 2044
    },
    {
      "epoch": 1.288719908593042,
      "grad_norm": 0.17701010406017303,
      "learning_rate": 1.2850767123548368e-05,
      "loss": 0.3259,
      "step": 2045
    },
    {
      "epoch": 1.289350301406564,
      "grad_norm": 0.20592482388019562,
      "learning_rate": 1.2830435744545487e-05,
      "loss": 0.3574,
      "step": 2046
    },
    {
      "epoch": 1.2899806942200858,
      "grad_norm": 0.1588132679462433,
      "learning_rate": 1.2810114044917618e-05,
      "loss": 0.3072,
      "step": 2047
    },
    {
      "epoch": 1.2906110870336078,
      "grad_norm": 0.16019609570503235,
      "learning_rate": 1.2789802045007092e-05,
      "loss": 0.402,
      "step": 2048
    },
    {
      "epoch": 1.2912414798471297,
      "grad_norm": 0.14784568548202515,
      "learning_rate": 1.2769499765146514e-05,
      "loss": 0.3119,
      "step": 2049
    },
    {
      "epoch": 1.2918718726606517,
      "grad_norm": 0.20480933785438538,
      "learning_rate": 1.2749207225658772e-05,
      "loss": 0.4032,
      "step": 2050
    },
    {
      "epoch": 1.2925022654741736,
      "grad_norm": 0.19020628929138184,
      "learning_rate": 1.2728924446856978e-05,
      "loss": 0.36,
      "step": 2051
    },
    {
      "epoch": 1.2931326582876954,
      "grad_norm": 0.16703538596630096,
      "learning_rate": 1.2708651449044498e-05,
      "loss": 0.3897,
      "step": 2052
    },
    {
      "epoch": 1.2937630511012175,
      "grad_norm": 0.15481066703796387,
      "learning_rate": 1.2688388252514897e-05,
      "loss": 0.3668,
      "step": 2053
    },
    {
      "epoch": 1.2943934439147393,
      "grad_norm": 0.17388780415058136,
      "learning_rate": 1.2668134877551945e-05,
      "loss": 0.3633,
      "step": 2054
    },
    {
      "epoch": 1.2950238367282614,
      "grad_norm": 0.15646809339523315,
      "learning_rate": 1.2647891344429541e-05,
      "loss": 0.3962,
      "step": 2055
    },
    {
      "epoch": 1.2956542295417832,
      "grad_norm": 0.17562836408615112,
      "learning_rate": 1.262765767341178e-05,
      "loss": 0.3227,
      "step": 2056
    },
    {
      "epoch": 1.296284622355305,
      "grad_norm": 0.1547730267047882,
      "learning_rate": 1.2607433884752831e-05,
      "loss": 0.3233,
      "step": 2057
    },
    {
      "epoch": 1.2969150151688271,
      "grad_norm": 0.17791743576526642,
      "learning_rate": 1.2587219998697041e-05,
      "loss": 0.4285,
      "step": 2058
    },
    {
      "epoch": 1.297545407982349,
      "grad_norm": 0.16143293678760529,
      "learning_rate": 1.256701603547878e-05,
      "loss": 0.3373,
      "step": 2059
    },
    {
      "epoch": 1.298175800795871,
      "grad_norm": 0.16266663372516632,
      "learning_rate": 1.2546822015322526e-05,
      "loss": 0.2953,
      "step": 2060
    },
    {
      "epoch": 1.2988061936093929,
      "grad_norm": 0.17324328422546387,
      "learning_rate": 1.252663795844277e-05,
      "loss": 0.3864,
      "step": 2061
    },
    {
      "epoch": 1.2994365864229147,
      "grad_norm": 0.14631371200084686,
      "learning_rate": 1.2506463885044056e-05,
      "loss": 0.3076,
      "step": 2062
    },
    {
      "epoch": 1.3000669792364368,
      "grad_norm": 0.17927135527133942,
      "learning_rate": 1.2486299815320925e-05,
      "loss": 0.3168,
      "step": 2063
    },
    {
      "epoch": 1.3006973720499586,
      "grad_norm": 0.14138492941856384,
      "learning_rate": 1.2466145769457918e-05,
      "loss": 0.2993,
      "step": 2064
    },
    {
      "epoch": 1.3013277648634807,
      "grad_norm": 0.14290207624435425,
      "learning_rate": 1.244600176762951e-05,
      "loss": 0.343,
      "step": 2065
    },
    {
      "epoch": 1.3019581576770025,
      "grad_norm": 0.1918410360813141,
      "learning_rate": 1.2425867830000147e-05,
      "loss": 0.3881,
      "step": 2066
    },
    {
      "epoch": 1.3025885504905244,
      "grad_norm": 0.14091849327087402,
      "learning_rate": 1.2405743976724194e-05,
      "loss": 0.3455,
      "step": 2067
    },
    {
      "epoch": 1.3032189433040462,
      "grad_norm": 0.1309073269367218,
      "learning_rate": 1.238563022794594e-05,
      "loss": 0.2491,
      "step": 2068
    },
    {
      "epoch": 1.3038493361175683,
      "grad_norm": 0.14994269609451294,
      "learning_rate": 1.2365526603799513e-05,
      "loss": 0.3525,
      "step": 2069
    },
    {
      "epoch": 1.3044797289310903,
      "grad_norm": 0.18416768312454224,
      "learning_rate": 1.2345433124408949e-05,
      "loss": 0.3717,
      "step": 2070
    },
    {
      "epoch": 1.3051101217446122,
      "grad_norm": 0.15417777001857758,
      "learning_rate": 1.2325349809888112e-05,
      "loss": 0.3617,
      "step": 2071
    },
    {
      "epoch": 1.305740514558134,
      "grad_norm": 0.17371056973934174,
      "learning_rate": 1.2305276680340704e-05,
      "loss": 0.3664,
      "step": 2072
    },
    {
      "epoch": 1.3063709073716558,
      "grad_norm": 0.19204451143741608,
      "learning_rate": 1.2285213755860199e-05,
      "loss": 0.3769,
      "step": 2073
    },
    {
      "epoch": 1.307001300185178,
      "grad_norm": 0.16981463134288788,
      "learning_rate": 1.2265161056529899e-05,
      "loss": 0.3194,
      "step": 2074
    },
    {
      "epoch": 1.3076316929986997,
      "grad_norm": 0.16115683317184448,
      "learning_rate": 1.2245118602422822e-05,
      "loss": 0.2995,
      "step": 2075
    },
    {
      "epoch": 1.3082620858122218,
      "grad_norm": 0.16867254674434662,
      "learning_rate": 1.222508641360179e-05,
      "loss": 0.362,
      "step": 2076
    },
    {
      "epoch": 1.3088924786257436,
      "grad_norm": 0.20875564217567444,
      "learning_rate": 1.2205064510119293e-05,
      "loss": 0.42,
      "step": 2077
    },
    {
      "epoch": 1.3095228714392655,
      "grad_norm": 0.14017994701862335,
      "learning_rate": 1.2185052912017564e-05,
      "loss": 0.3294,
      "step": 2078
    },
    {
      "epoch": 1.3101532642527876,
      "grad_norm": 0.17594243586063385,
      "learning_rate": 1.216505163932848e-05,
      "loss": 0.3656,
      "step": 2079
    },
    {
      "epoch": 1.3107836570663094,
      "grad_norm": 0.18787644803524017,
      "learning_rate": 1.2145060712073642e-05,
      "loss": 0.3789,
      "step": 2080
    },
    {
      "epoch": 1.3114140498798315,
      "grad_norm": 0.14373567700386047,
      "learning_rate": 1.2125080150264233e-05,
      "loss": 0.3008,
      "step": 2081
    },
    {
      "epoch": 1.3120444426933533,
      "grad_norm": 0.16203191876411438,
      "learning_rate": 1.2105109973901107e-05,
      "loss": 0.3153,
      "step": 2082
    },
    {
      "epoch": 1.3126748355068751,
      "grad_norm": 0.1497282087802887,
      "learning_rate": 1.208515020297468e-05,
      "loss": 0.3249,
      "step": 2083
    },
    {
      "epoch": 1.3133052283203972,
      "grad_norm": 0.14576971530914307,
      "learning_rate": 1.2065200857464989e-05,
      "loss": 0.3079,
      "step": 2084
    },
    {
      "epoch": 1.313935621133919,
      "grad_norm": 0.19296002388000488,
      "learning_rate": 1.2045261957341617e-05,
      "loss": 0.3555,
      "step": 2085
    },
    {
      "epoch": 1.314566013947441,
      "grad_norm": 0.1784832924604416,
      "learning_rate": 1.2025333522563699e-05,
      "loss": 0.3714,
      "step": 2086
    },
    {
      "epoch": 1.315196406760963,
      "grad_norm": 0.21400387585163116,
      "learning_rate": 1.200541557307988e-05,
      "loss": 0.434,
      "step": 2087
    },
    {
      "epoch": 1.3158267995744848,
      "grad_norm": 0.18918611109256744,
      "learning_rate": 1.1985508128828318e-05,
      "loss": 0.3567,
      "step": 2088
    },
    {
      "epoch": 1.3164571923880068,
      "grad_norm": 0.16789653897285461,
      "learning_rate": 1.1965611209736663e-05,
      "loss": 0.3048,
      "step": 2089
    },
    {
      "epoch": 1.3170875852015287,
      "grad_norm": 0.20766571164131165,
      "learning_rate": 1.1945724835722029e-05,
      "loss": 0.3841,
      "step": 2090
    },
    {
      "epoch": 1.3177179780150508,
      "grad_norm": 0.15403051674365997,
      "learning_rate": 1.1925849026690944e-05,
      "loss": 0.3633,
      "step": 2091
    },
    {
      "epoch": 1.3183483708285726,
      "grad_norm": 0.17515239119529724,
      "learning_rate": 1.1905983802539399e-05,
      "loss": 0.4071,
      "step": 2092
    },
    {
      "epoch": 1.3189787636420944,
      "grad_norm": 0.18618489801883698,
      "learning_rate": 1.1886129183152768e-05,
      "loss": 0.3248,
      "step": 2093
    },
    {
      "epoch": 1.3196091564556163,
      "grad_norm": 0.15004846453666687,
      "learning_rate": 1.1866285188405826e-05,
      "loss": 0.3313,
      "step": 2094
    },
    {
      "epoch": 1.3202395492691383,
      "grad_norm": 0.1696581244468689,
      "learning_rate": 1.1846451838162683e-05,
      "loss": 0.3451,
      "step": 2095
    },
    {
      "epoch": 1.3208699420826602,
      "grad_norm": 0.13555723428726196,
      "learning_rate": 1.1826629152276831e-05,
      "loss": 0.3332,
      "step": 2096
    },
    {
      "epoch": 1.3215003348961822,
      "grad_norm": 0.14103135466575623,
      "learning_rate": 1.1806817150591042e-05,
      "loss": 0.3281,
      "step": 2097
    },
    {
      "epoch": 1.322130727709704,
      "grad_norm": 0.16602687537670135,
      "learning_rate": 1.1787015852937451e-05,
      "loss": 0.3702,
      "step": 2098
    },
    {
      "epoch": 1.322761120523226,
      "grad_norm": 0.18209081888198853,
      "learning_rate": 1.1767225279137426e-05,
      "loss": 0.371,
      "step": 2099
    },
    {
      "epoch": 1.323391513336748,
      "grad_norm": 0.15361028909683228,
      "learning_rate": 1.1747445449001628e-05,
      "loss": 0.3664,
      "step": 2100
    },
    {
      "epoch": 1.3240219061502698,
      "grad_norm": 0.1743720918893814,
      "learning_rate": 1.1727676382329951e-05,
      "loss": 0.3428,
      "step": 2101
    },
    {
      "epoch": 1.3246522989637919,
      "grad_norm": 0.15461906790733337,
      "learning_rate": 1.1707918098911518e-05,
      "loss": 0.3322,
      "step": 2102
    },
    {
      "epoch": 1.3252826917773137,
      "grad_norm": 0.17772087454795837,
      "learning_rate": 1.1688170618524662e-05,
      "loss": 0.3166,
      "step": 2103
    },
    {
      "epoch": 1.3259130845908356,
      "grad_norm": 0.17029738426208496,
      "learning_rate": 1.1668433960936909e-05,
      "loss": 0.3881,
      "step": 2104
    },
    {
      "epoch": 1.3265434774043576,
      "grad_norm": 0.1838579922914505,
      "learning_rate": 1.1648708145904926e-05,
      "loss": 0.3728,
      "step": 2105
    },
    {
      "epoch": 1.3271738702178795,
      "grad_norm": 0.16194473206996918,
      "learning_rate": 1.1628993193174543e-05,
      "loss": 0.3705,
      "step": 2106
    },
    {
      "epoch": 1.3278042630314015,
      "grad_norm": 0.16256876289844513,
      "learning_rate": 1.1609289122480726e-05,
      "loss": 0.3453,
      "step": 2107
    },
    {
      "epoch": 1.3284346558449234,
      "grad_norm": 0.13366344571113586,
      "learning_rate": 1.158959595354754e-05,
      "loss": 0.2767,
      "step": 2108
    },
    {
      "epoch": 1.3290650486584452,
      "grad_norm": 0.17386309802532196,
      "learning_rate": 1.1569913706088115e-05,
      "loss": 0.4169,
      "step": 2109
    },
    {
      "epoch": 1.3296954414719673,
      "grad_norm": 0.16652558743953705,
      "learning_rate": 1.155024239980468e-05,
      "loss": 0.3581,
      "step": 2110
    },
    {
      "epoch": 1.3303258342854891,
      "grad_norm": 0.14673203229904175,
      "learning_rate": 1.1530582054388504e-05,
      "loss": 0.3549,
      "step": 2111
    },
    {
      "epoch": 1.3309562270990112,
      "grad_norm": 0.15078188478946686,
      "learning_rate": 1.1510932689519882e-05,
      "loss": 0.348,
      "step": 2112
    },
    {
      "epoch": 1.331586619912533,
      "grad_norm": 0.19112563133239746,
      "learning_rate": 1.1491294324868103e-05,
      "loss": 0.3475,
      "step": 2113
    },
    {
      "epoch": 1.3322170127260549,
      "grad_norm": 0.1545761227607727,
      "learning_rate": 1.1471666980091462e-05,
      "loss": 0.3901,
      "step": 2114
    },
    {
      "epoch": 1.332847405539577,
      "grad_norm": 0.16541506350040436,
      "learning_rate": 1.1452050674837217e-05,
      "loss": 0.4023,
      "step": 2115
    },
    {
      "epoch": 1.3334777983530988,
      "grad_norm": 0.15039975941181183,
      "learning_rate": 1.1432445428741585e-05,
      "loss": 0.378,
      "step": 2116
    },
    {
      "epoch": 1.3341081911666208,
      "grad_norm": 0.15127238631248474,
      "learning_rate": 1.1412851261429689e-05,
      "loss": 0.3087,
      "step": 2117
    },
    {
      "epoch": 1.3347385839801427,
      "grad_norm": 0.1384199857711792,
      "learning_rate": 1.1393268192515592e-05,
      "loss": 0.2421,
      "step": 2118
    },
    {
      "epoch": 1.3353689767936645,
      "grad_norm": 0.15712042152881622,
      "learning_rate": 1.1373696241602206e-05,
      "loss": 0.3087,
      "step": 2119
    },
    {
      "epoch": 1.3359993696071863,
      "grad_norm": 0.1710873246192932,
      "learning_rate": 1.1354135428281377e-05,
      "loss": 0.405,
      "step": 2120
    },
    {
      "epoch": 1.3366297624207084,
      "grad_norm": 0.16128316521644592,
      "learning_rate": 1.1334585772133736e-05,
      "loss": 0.3004,
      "step": 2121
    },
    {
      "epoch": 1.3372601552342303,
      "grad_norm": 0.1520003229379654,
      "learning_rate": 1.131504729272879e-05,
      "loss": 0.3234,
      "step": 2122
    },
    {
      "epoch": 1.3378905480477523,
      "grad_norm": 0.17195023596286774,
      "learning_rate": 1.1295520009624825e-05,
      "loss": 0.3663,
      "step": 2123
    },
    {
      "epoch": 1.3385209408612742,
      "grad_norm": 0.15014538168907166,
      "learning_rate": 1.1276003942368952e-05,
      "loss": 0.3301,
      "step": 2124
    },
    {
      "epoch": 1.339151333674796,
      "grad_norm": 0.1437128782272339,
      "learning_rate": 1.1256499110497032e-05,
      "loss": 0.3344,
      "step": 2125
    },
    {
      "epoch": 1.339781726488318,
      "grad_norm": 0.15259809792041779,
      "learning_rate": 1.12370055335337e-05,
      "loss": 0.3522,
      "step": 2126
    },
    {
      "epoch": 1.34041211930184,
      "grad_norm": 0.17295295000076294,
      "learning_rate": 1.1217523230992287e-05,
      "loss": 0.3374,
      "step": 2127
    },
    {
      "epoch": 1.341042512115362,
      "grad_norm": 0.12949199974536896,
      "learning_rate": 1.1198052222374879e-05,
      "loss": 0.3374,
      "step": 2128
    },
    {
      "epoch": 1.3416729049288838,
      "grad_norm": 0.17765846848487854,
      "learning_rate": 1.1178592527172235e-05,
      "loss": 0.382,
      "step": 2129
    },
    {
      "epoch": 1.3423032977424056,
      "grad_norm": 0.18448440730571747,
      "learning_rate": 1.1159144164863806e-05,
      "loss": 0.3577,
      "step": 2130
    },
    {
      "epoch": 1.3429336905559277,
      "grad_norm": 0.18258066475391388,
      "learning_rate": 1.1139707154917666e-05,
      "loss": 0.3923,
      "step": 2131
    },
    {
      "epoch": 1.3435640833694495,
      "grad_norm": 0.16213437914848328,
      "learning_rate": 1.1120281516790557e-05,
      "loss": 0.4219,
      "step": 2132
    },
    {
      "epoch": 1.3441944761829716,
      "grad_norm": 0.17956312000751495,
      "learning_rate": 1.1100867269927825e-05,
      "loss": 0.3848,
      "step": 2133
    },
    {
      "epoch": 1.3448248689964934,
      "grad_norm": 0.14712972939014435,
      "learning_rate": 1.1081464433763425e-05,
      "loss": 0.3267,
      "step": 2134
    },
    {
      "epoch": 1.3454552618100153,
      "grad_norm": 0.16251666843891144,
      "learning_rate": 1.106207302771986e-05,
      "loss": 0.3602,
      "step": 2135
    },
    {
      "epoch": 1.3460856546235374,
      "grad_norm": 0.17616884410381317,
      "learning_rate": 1.1042693071208228e-05,
      "loss": 0.352,
      "step": 2136
    },
    {
      "epoch": 1.3467160474370592,
      "grad_norm": 0.13910022377967834,
      "learning_rate": 1.1023324583628122e-05,
      "loss": 0.2441,
      "step": 2137
    },
    {
      "epoch": 1.3473464402505813,
      "grad_norm": 0.1513376086950302,
      "learning_rate": 1.100396758436772e-05,
      "loss": 0.3176,
      "step": 2138
    },
    {
      "epoch": 1.347976833064103,
      "grad_norm": 0.17463596165180206,
      "learning_rate": 1.098462209280363e-05,
      "loss": 0.365,
      "step": 2139
    },
    {
      "epoch": 1.348607225877625,
      "grad_norm": 0.17284253239631653,
      "learning_rate": 1.096528812830099e-05,
      "loss": 0.3961,
      "step": 2140
    },
    {
      "epoch": 1.349237618691147,
      "grad_norm": 0.14869636297225952,
      "learning_rate": 1.0945965710213364e-05,
      "loss": 0.3043,
      "step": 2141
    },
    {
      "epoch": 1.3498680115046688,
      "grad_norm": 0.17893320322036743,
      "learning_rate": 1.0926654857882784e-05,
      "loss": 0.3753,
      "step": 2142
    },
    {
      "epoch": 1.350498404318191,
      "grad_norm": 0.20987479388713837,
      "learning_rate": 1.0907355590639695e-05,
      "loss": 0.3507,
      "step": 2143
    },
    {
      "epoch": 1.3511287971317127,
      "grad_norm": 0.14228267967700958,
      "learning_rate": 1.0888067927802953e-05,
      "loss": 0.2932,
      "step": 2144
    },
    {
      "epoch": 1.3517591899452346,
      "grad_norm": 0.15405458211898804,
      "learning_rate": 1.0868791888679774e-05,
      "loss": 0.2966,
      "step": 2145
    },
    {
      "epoch": 1.3523895827587564,
      "grad_norm": 0.14483889937400818,
      "learning_rate": 1.0849527492565762e-05,
      "loss": 0.3889,
      "step": 2146
    },
    {
      "epoch": 1.3530199755722785,
      "grad_norm": 0.18476107716560364,
      "learning_rate": 1.0830274758744857e-05,
      "loss": 0.3637,
      "step": 2147
    },
    {
      "epoch": 1.3536503683858003,
      "grad_norm": 0.20455743372440338,
      "learning_rate": 1.0811033706489342e-05,
      "loss": 0.4167,
      "step": 2148
    },
    {
      "epoch": 1.3542807611993224,
      "grad_norm": 0.1842804104089737,
      "learning_rate": 1.0791804355059767e-05,
      "loss": 0.3541,
      "step": 2149
    },
    {
      "epoch": 1.3549111540128442,
      "grad_norm": 0.20898349583148956,
      "learning_rate": 1.0772586723705007e-05,
      "loss": 0.4065,
      "step": 2150
    },
    {
      "epoch": 1.355541546826366,
      "grad_norm": 0.16361692547798157,
      "learning_rate": 1.0753380831662187e-05,
      "loss": 0.4172,
      "step": 2151
    },
    {
      "epoch": 1.3561719396398881,
      "grad_norm": 0.19685503840446472,
      "learning_rate": 1.07341866981567e-05,
      "loss": 0.3342,
      "step": 2152
    },
    {
      "epoch": 1.35680233245341,
      "grad_norm": 0.16768209636211395,
      "learning_rate": 1.0715004342402133e-05,
      "loss": 0.3338,
      "step": 2153
    },
    {
      "epoch": 1.357432725266932,
      "grad_norm": 0.16943420469760895,
      "learning_rate": 1.0695833783600312e-05,
      "loss": 0.328,
      "step": 2154
    },
    {
      "epoch": 1.3580631180804539,
      "grad_norm": 0.1994352638721466,
      "learning_rate": 1.0676675040941247e-05,
      "loss": 0.376,
      "step": 2155
    },
    {
      "epoch": 1.3586935108939757,
      "grad_norm": 0.1642264872789383,
      "learning_rate": 1.0657528133603128e-05,
      "loss": 0.2913,
      "step": 2156
    },
    {
      "epoch": 1.3593239037074978,
      "grad_norm": 0.17089161276817322,
      "learning_rate": 1.0638393080752273e-05,
      "loss": 0.3303,
      "step": 2157
    },
    {
      "epoch": 1.3599542965210196,
      "grad_norm": 0.1726408749818802,
      "learning_rate": 1.0619269901543167e-05,
      "loss": 0.3242,
      "step": 2158
    },
    {
      "epoch": 1.3605846893345417,
      "grad_norm": 0.16919533908367157,
      "learning_rate": 1.0600158615118361e-05,
      "loss": 0.4038,
      "step": 2159
    },
    {
      "epoch": 1.3612150821480635,
      "grad_norm": 0.19589287042617798,
      "learning_rate": 1.0581059240608571e-05,
      "loss": 0.3371,
      "step": 2160
    },
    {
      "epoch": 1.3618454749615854,
      "grad_norm": 0.1313522905111313,
      "learning_rate": 1.0561971797132524e-05,
      "loss": 0.303,
      "step": 2161
    },
    {
      "epoch": 1.3624758677751074,
      "grad_norm": 0.144204244017601,
      "learning_rate": 1.0542896303797043e-05,
      "loss": 0.3262,
      "step": 2162
    },
    {
      "epoch": 1.3631062605886293,
      "grad_norm": 0.18240010738372803,
      "learning_rate": 1.0523832779696962e-05,
      "loss": 0.3823,
      "step": 2163
    },
    {
      "epoch": 1.3637366534021513,
      "grad_norm": 0.12797421216964722,
      "learning_rate": 1.0504781243915153e-05,
      "loss": 0.3045,
      "step": 2164
    },
    {
      "epoch": 1.3643670462156732,
      "grad_norm": 0.1489725559949875,
      "learning_rate": 1.0485741715522484e-05,
      "loss": 0.3592,
      "step": 2165
    },
    {
      "epoch": 1.364997439029195,
      "grad_norm": 0.17021243274211884,
      "learning_rate": 1.0466714213577807e-05,
      "loss": 0.3543,
      "step": 2166
    },
    {
      "epoch": 1.3656278318427169,
      "grad_norm": 0.14448320865631104,
      "learning_rate": 1.0447698757127913e-05,
      "loss": 0.3197,
      "step": 2167
    },
    {
      "epoch": 1.366258224656239,
      "grad_norm": 0.17550738155841827,
      "learning_rate": 1.0428695365207561e-05,
      "loss": 0.3569,
      "step": 2168
    },
    {
      "epoch": 1.366888617469761,
      "grad_norm": 0.16940660774707794,
      "learning_rate": 1.0409704056839423e-05,
      "loss": 0.3336,
      "step": 2169
    },
    {
      "epoch": 1.3675190102832828,
      "grad_norm": 0.1698468029499054,
      "learning_rate": 1.0390724851034088e-05,
      "loss": 0.3461,
      "step": 2170
    },
    {
      "epoch": 1.3681494030968047,
      "grad_norm": 0.21772147715091705,
      "learning_rate": 1.0371757766789994e-05,
      "loss": 0.3958,
      "step": 2171
    },
    {
      "epoch": 1.3687797959103265,
      "grad_norm": 0.1430359184741974,
      "learning_rate": 1.0352802823093483e-05,
      "loss": 0.2761,
      "step": 2172
    },
    {
      "epoch": 1.3694101887238486,
      "grad_norm": 0.1858154535293579,
      "learning_rate": 1.0333860038918728e-05,
      "loss": 0.2856,
      "step": 2173
    },
    {
      "epoch": 1.3700405815373704,
      "grad_norm": 0.15812067687511444,
      "learning_rate": 1.0314929433227745e-05,
      "loss": 0.3216,
      "step": 2174
    },
    {
      "epoch": 1.3706709743508925,
      "grad_norm": 0.19749779999256134,
      "learning_rate": 1.0296011024970326e-05,
      "loss": 0.4061,
      "step": 2175
    },
    {
      "epoch": 1.3713013671644143,
      "grad_norm": 0.1705770045518875,
      "learning_rate": 1.0277104833084083e-05,
      "loss": 0.3822,
      "step": 2176
    },
    {
      "epoch": 1.3719317599779361,
      "grad_norm": 0.18999949097633362,
      "learning_rate": 1.025821087649439e-05,
      "loss": 0.4095,
      "step": 2177
    },
    {
      "epoch": 1.3725621527914582,
      "grad_norm": 0.16586129367351532,
      "learning_rate": 1.0239329174114383e-05,
      "loss": 0.3592,
      "step": 2178
    },
    {
      "epoch": 1.37319254560498,
      "grad_norm": 0.15350428223609924,
      "learning_rate": 1.0220459744844905e-05,
      "loss": 0.3844,
      "step": 2179
    },
    {
      "epoch": 1.3738229384185021,
      "grad_norm": 0.16602548956871033,
      "learning_rate": 1.0201602607574546e-05,
      "loss": 0.4005,
      "step": 2180
    },
    {
      "epoch": 1.374453331232024,
      "grad_norm": 0.16380132734775543,
      "learning_rate": 1.0182757781179552e-05,
      "loss": 0.3572,
      "step": 2181
    },
    {
      "epoch": 1.3750837240455458,
      "grad_norm": 0.20154309272766113,
      "learning_rate": 1.01639252845239e-05,
      "loss": 0.3688,
      "step": 2182
    },
    {
      "epoch": 1.3757141168590679,
      "grad_norm": 0.17154088616371155,
      "learning_rate": 1.0145105136459172e-05,
      "loss": 0.3749,
      "step": 2183
    },
    {
      "epoch": 1.3763445096725897,
      "grad_norm": 0.13803677260875702,
      "learning_rate": 1.0126297355824628e-05,
      "loss": 0.3064,
      "step": 2184
    },
    {
      "epoch": 1.3769749024861118,
      "grad_norm": 0.15523208677768707,
      "learning_rate": 1.0107501961447112e-05,
      "loss": 0.3564,
      "step": 2185
    },
    {
      "epoch": 1.3776052952996336,
      "grad_norm": 0.18922999501228333,
      "learning_rate": 1.0088718972141097e-05,
      "loss": 0.3339,
      "step": 2186
    },
    {
      "epoch": 1.3782356881131554,
      "grad_norm": 0.17507292330265045,
      "learning_rate": 1.0069948406708632e-05,
      "loss": 0.3967,
      "step": 2187
    },
    {
      "epoch": 1.3788660809266775,
      "grad_norm": 0.2225351333618164,
      "learning_rate": 1.0051190283939332e-05,
      "loss": 0.4534,
      "step": 2188
    },
    {
      "epoch": 1.3794964737401993,
      "grad_norm": 0.15008077025413513,
      "learning_rate": 1.0032444622610339e-05,
      "loss": 0.2755,
      "step": 2189
    },
    {
      "epoch": 1.3801268665537214,
      "grad_norm": 0.1385604292154312,
      "learning_rate": 1.0013711441486339e-05,
      "loss": 0.2841,
      "step": 2190
    },
    {
      "epoch": 1.3807572593672433,
      "grad_norm": 0.2007773369550705,
      "learning_rate": 9.994990759319522e-06,
      "loss": 0.3863,
      "step": 2191
    },
    {
      "epoch": 1.381387652180765,
      "grad_norm": 0.1429567039012909,
      "learning_rate": 9.976282594849569e-06,
      "loss": 0.3438,
      "step": 2192
    },
    {
      "epoch": 1.382018044994287,
      "grad_norm": 0.14305874705314636,
      "learning_rate": 9.957586966803615e-06,
      "loss": 0.2912,
      "step": 2193
    },
    {
      "epoch": 1.382648437807809,
      "grad_norm": 0.14862282574176788,
      "learning_rate": 9.93890389389626e-06,
      "loss": 0.3041,
      "step": 2194
    },
    {
      "epoch": 1.3832788306213308,
      "grad_norm": 0.16503016650676727,
      "learning_rate": 9.920233394829533e-06,
      "loss": 0.2807,
      "step": 2195
    },
    {
      "epoch": 1.383909223434853,
      "grad_norm": 0.16887208819389343,
      "learning_rate": 9.901575488292886e-06,
      "loss": 0.3668,
      "step": 2196
    },
    {
      "epoch": 1.3845396162483747,
      "grad_norm": 0.1434229612350464,
      "learning_rate": 9.882930192963142e-06,
      "loss": 0.2897,
      "step": 2197
    },
    {
      "epoch": 1.3851700090618966,
      "grad_norm": 0.21803705394268036,
      "learning_rate": 9.864297527504525e-06,
      "loss": 0.4214,
      "step": 2198
    },
    {
      "epoch": 1.3858004018754186,
      "grad_norm": 0.1603592485189438,
      "learning_rate": 9.845677510568584e-06,
      "loss": 0.3886,
      "step": 2199
    },
    {
      "epoch": 1.3864307946889405,
      "grad_norm": 0.161991149187088,
      "learning_rate": 9.82707016079426e-06,
      "loss": 0.2961,
      "step": 2200
    },
    {
      "epoch": 1.3864307946889405,
      "eval_loss": 0.4210342466831207,
      "eval_runtime": 222.2931,
      "eval_samples_per_second": 4.499,
      "eval_steps_per_second": 4.499,
      "step": 2200
    },
    {
      "epoch": 1.3870611875024625,
      "grad_norm": 0.15873432159423828,
      "learning_rate": 9.808475496807757e-06,
      "loss": 0.3394,
      "step": 2201
    },
    {
      "epoch": 1.3876915803159844,
      "grad_norm": 0.15210144221782684,
      "learning_rate": 9.789893537222622e-06,
      "loss": 0.3294,
      "step": 2202
    },
    {
      "epoch": 1.3883219731295062,
      "grad_norm": 0.15166395902633667,
      "learning_rate": 9.771324300639652e-06,
      "loss": 0.4074,
      "step": 2203
    },
    {
      "epoch": 1.3889523659430283,
      "grad_norm": 0.18236121535301208,
      "learning_rate": 9.752767805646932e-06,
      "loss": 0.3576,
      "step": 2204
    },
    {
      "epoch": 1.3895827587565501,
      "grad_norm": 0.15060827136039734,
      "learning_rate": 9.734224070819779e-06,
      "loss": 0.3644,
      "step": 2205
    },
    {
      "epoch": 1.3902131515700722,
      "grad_norm": 0.15586620569229126,
      "learning_rate": 9.71569311472076e-06,
      "loss": 0.2944,
      "step": 2206
    },
    {
      "epoch": 1.390843544383594,
      "grad_norm": 0.1764363795518875,
      "learning_rate": 9.697174955899612e-06,
      "loss": 0.3629,
      "step": 2207
    },
    {
      "epoch": 1.3914739371971159,
      "grad_norm": 0.13980361819267273,
      "learning_rate": 9.678669612893287e-06,
      "loss": 0.3418,
      "step": 2208
    },
    {
      "epoch": 1.392104330010638,
      "grad_norm": 0.19479088485240936,
      "learning_rate": 9.660177104225908e-06,
      "loss": 0.4051,
      "step": 2209
    },
    {
      "epoch": 1.3927347228241598,
      "grad_norm": 0.16188308596611023,
      "learning_rate": 9.641697448408756e-06,
      "loss": 0.3823,
      "step": 2210
    },
    {
      "epoch": 1.3933651156376818,
      "grad_norm": 0.16864702105522156,
      "learning_rate": 9.623230663940215e-06,
      "loss": 0.3365,
      "step": 2211
    },
    {
      "epoch": 1.3939955084512037,
      "grad_norm": 0.1589280664920807,
      "learning_rate": 9.604776769305815e-06,
      "loss": 0.3006,
      "step": 2212
    },
    {
      "epoch": 1.3946259012647255,
      "grad_norm": 0.16485220193862915,
      "learning_rate": 9.586335782978175e-06,
      "loss": 0.2999,
      "step": 2213
    },
    {
      "epoch": 1.3952562940782476,
      "grad_norm": 0.16778141260147095,
      "learning_rate": 9.567907723417002e-06,
      "loss": 0.3433,
      "step": 2214
    },
    {
      "epoch": 1.3958866868917694,
      "grad_norm": 0.17085351049900055,
      "learning_rate": 9.549492609069036e-06,
      "loss": 0.3674,
      "step": 2215
    },
    {
      "epoch": 1.3965170797052915,
      "grad_norm": 0.1922229677438736,
      "learning_rate": 9.531090458368078e-06,
      "loss": 0.3561,
      "step": 2216
    },
    {
      "epoch": 1.3971474725188133,
      "grad_norm": 0.1994408369064331,
      "learning_rate": 9.512701289734957e-06,
      "loss": 0.3808,
      "step": 2217
    },
    {
      "epoch": 1.3977778653323352,
      "grad_norm": 0.14683650434017181,
      "learning_rate": 9.494325121577504e-06,
      "loss": 0.3088,
      "step": 2218
    },
    {
      "epoch": 1.398408258145857,
      "grad_norm": 0.18013975024223328,
      "learning_rate": 9.475961972290513e-06,
      "loss": 0.3011,
      "step": 2219
    },
    {
      "epoch": 1.399038650959379,
      "grad_norm": 0.17953065037727356,
      "learning_rate": 9.457611860255786e-06,
      "loss": 0.4288,
      "step": 2220
    },
    {
      "epoch": 1.399669043772901,
      "grad_norm": 0.13922902941703796,
      "learning_rate": 9.439274803842024e-06,
      "loss": 0.3322,
      "step": 2221
    },
    {
      "epoch": 1.400299436586423,
      "grad_norm": 0.1882917881011963,
      "learning_rate": 9.420950821404922e-06,
      "loss": 0.3984,
      "step": 2222
    },
    {
      "epoch": 1.4009298293999448,
      "grad_norm": 0.13556498289108276,
      "learning_rate": 9.402639931287032e-06,
      "loss": 0.3062,
      "step": 2223
    },
    {
      "epoch": 1.4015602222134667,
      "grad_norm": 0.16906505823135376,
      "learning_rate": 9.38434215181784e-06,
      "loss": 0.3651,
      "step": 2224
    },
    {
      "epoch": 1.4021906150269887,
      "grad_norm": 0.16540083289146423,
      "learning_rate": 9.366057501313665e-06,
      "loss": 0.3759,
      "step": 2225
    },
    {
      "epoch": 1.4028210078405106,
      "grad_norm": 0.1791892945766449,
      "learning_rate": 9.347785998077727e-06,
      "loss": 0.3819,
      "step": 2226
    },
    {
      "epoch": 1.4034514006540326,
      "grad_norm": 0.15491949021816254,
      "learning_rate": 9.329527660400064e-06,
      "loss": 0.3316,
      "step": 2227
    },
    {
      "epoch": 1.4040817934675545,
      "grad_norm": 0.17092517018318176,
      "learning_rate": 9.311282506557547e-06,
      "loss": 0.368,
      "step": 2228
    },
    {
      "epoch": 1.4047121862810763,
      "grad_norm": 0.15667341649532318,
      "learning_rate": 9.293050554813825e-06,
      "loss": 0.2982,
      "step": 2229
    },
    {
      "epoch": 1.4053425790945984,
      "grad_norm": 0.14160802960395813,
      "learning_rate": 9.274831823419357e-06,
      "loss": 0.3122,
      "step": 2230
    },
    {
      "epoch": 1.4059729719081202,
      "grad_norm": 0.19099406898021698,
      "learning_rate": 9.256626330611357e-06,
      "loss": 0.377,
      "step": 2231
    },
    {
      "epoch": 1.4066033647216423,
      "grad_norm": 0.1867620050907135,
      "learning_rate": 9.2384340946138e-06,
      "loss": 0.3396,
      "step": 2232
    },
    {
      "epoch": 1.407233757535164,
      "grad_norm": 0.17459844052791595,
      "learning_rate": 9.22025513363736e-06,
      "loss": 0.3364,
      "step": 2233
    },
    {
      "epoch": 1.407864150348686,
      "grad_norm": 0.16290386021137238,
      "learning_rate": 9.202089465879452e-06,
      "loss": 0.3307,
      "step": 2234
    },
    {
      "epoch": 1.408494543162208,
      "grad_norm": 0.16637465357780457,
      "learning_rate": 9.183937109524177e-06,
      "loss": 0.3042,
      "step": 2235
    },
    {
      "epoch": 1.4091249359757299,
      "grad_norm": 0.19044050574302673,
      "learning_rate": 9.165798082742316e-06,
      "loss": 0.3813,
      "step": 2236
    },
    {
      "epoch": 1.409755328789252,
      "grad_norm": 0.19611325860023499,
      "learning_rate": 9.147672403691284e-06,
      "loss": 0.344,
      "step": 2237
    },
    {
      "epoch": 1.4103857216027738,
      "grad_norm": 0.16075120866298676,
      "learning_rate": 9.129560090515161e-06,
      "loss": 0.3024,
      "step": 2238
    },
    {
      "epoch": 1.4110161144162956,
      "grad_norm": 0.15303272008895874,
      "learning_rate": 9.11146116134462e-06,
      "loss": 0.332,
      "step": 2239
    },
    {
      "epoch": 1.4116465072298177,
      "grad_norm": 0.1527310013771057,
      "learning_rate": 9.093375634296984e-06,
      "loss": 0.3317,
      "step": 2240
    },
    {
      "epoch": 1.4122769000433395,
      "grad_norm": 0.20200815796852112,
      "learning_rate": 9.075303527476101e-06,
      "loss": 0.4097,
      "step": 2241
    },
    {
      "epoch": 1.4129072928568616,
      "grad_norm": 0.17556019127368927,
      "learning_rate": 9.057244858972435e-06,
      "loss": 0.3812,
      "step": 2242
    },
    {
      "epoch": 1.4135376856703834,
      "grad_norm": 0.17567972838878632,
      "learning_rate": 9.039199646862952e-06,
      "loss": 0.4171,
      "step": 2243
    },
    {
      "epoch": 1.4141680784839052,
      "grad_norm": 0.19437840580940247,
      "learning_rate": 9.021167909211204e-06,
      "loss": 0.3437,
      "step": 2244
    },
    {
      "epoch": 1.414798471297427,
      "grad_norm": 0.1861252784729004,
      "learning_rate": 9.003149664067198e-06,
      "loss": 0.412,
      "step": 2245
    },
    {
      "epoch": 1.4154288641109491,
      "grad_norm": 0.18544334173202515,
      "learning_rate": 8.985144929467478e-06,
      "loss": 0.3603,
      "step": 2246
    },
    {
      "epoch": 1.416059256924471,
      "grad_norm": 0.16569247841835022,
      "learning_rate": 8.96715372343503e-06,
      "loss": 0.3239,
      "step": 2247
    },
    {
      "epoch": 1.416689649737993,
      "grad_norm": 0.20697587728500366,
      "learning_rate": 8.949176063979324e-06,
      "loss": 0.4214,
      "step": 2248
    },
    {
      "epoch": 1.417320042551515,
      "grad_norm": 0.17573077976703644,
      "learning_rate": 8.931211969096257e-06,
      "loss": 0.2999,
      "step": 2249
    },
    {
      "epoch": 1.4179504353650367,
      "grad_norm": 0.16955791413784027,
      "learning_rate": 8.91326145676816e-06,
      "loss": 0.342,
      "step": 2250
    },
    {
      "epoch": 1.4185808281785588,
      "grad_norm": 0.21412505209445953,
      "learning_rate": 8.89532454496374e-06,
      "loss": 0.3414,
      "step": 2251
    },
    {
      "epoch": 1.4192112209920806,
      "grad_norm": 0.17389221489429474,
      "learning_rate": 8.877401251638119e-06,
      "loss": 0.3694,
      "step": 2252
    },
    {
      "epoch": 1.4198416138056027,
      "grad_norm": 0.1807471066713333,
      "learning_rate": 8.85949159473277e-06,
      "loss": 0.4271,
      "step": 2253
    },
    {
      "epoch": 1.4204720066191245,
      "grad_norm": 0.15700294077396393,
      "learning_rate": 8.841595592175534e-06,
      "loss": 0.2964,
      "step": 2254
    },
    {
      "epoch": 1.4211023994326464,
      "grad_norm": 0.19284594058990479,
      "learning_rate": 8.823713261880553e-06,
      "loss": 0.4243,
      "step": 2255
    },
    {
      "epoch": 1.4217327922461684,
      "grad_norm": 0.2265140563249588,
      "learning_rate": 8.80584462174831e-06,
      "loss": 0.4123,
      "step": 2256
    },
    {
      "epoch": 1.4223631850596903,
      "grad_norm": 0.16101014614105225,
      "learning_rate": 8.787989689665575e-06,
      "loss": 0.3469,
      "step": 2257
    },
    {
      "epoch": 1.4229935778732123,
      "grad_norm": 0.14722800254821777,
      "learning_rate": 8.770148483505399e-06,
      "loss": 0.3058,
      "step": 2258
    },
    {
      "epoch": 1.4236239706867342,
      "grad_norm": 0.21112282574176788,
      "learning_rate": 8.752321021127079e-06,
      "loss": 0.3872,
      "step": 2259
    },
    {
      "epoch": 1.424254363500256,
      "grad_norm": 0.23365963995456696,
      "learning_rate": 8.734507320376179e-06,
      "loss": 0.4056,
      "step": 2260
    },
    {
      "epoch": 1.424884756313778,
      "grad_norm": 0.17326068878173828,
      "learning_rate": 8.716707399084452e-06,
      "loss": 0.3536,
      "step": 2261
    },
    {
      "epoch": 1.4255151491273,
      "grad_norm": 0.14267979562282562,
      "learning_rate": 8.698921275069906e-06,
      "loss": 0.3063,
      "step": 2262
    },
    {
      "epoch": 1.426145541940822,
      "grad_norm": 0.22588102519512177,
      "learning_rate": 8.681148966136691e-06,
      "loss": 0.3752,
      "step": 2263
    },
    {
      "epoch": 1.4267759347543438,
      "grad_norm": 0.20228253304958344,
      "learning_rate": 8.663390490075164e-06,
      "loss": 0.3556,
      "step": 2264
    },
    {
      "epoch": 1.4274063275678657,
      "grad_norm": 0.18117617070674896,
      "learning_rate": 8.6456458646618e-06,
      "loss": 0.3498,
      "step": 2265
    },
    {
      "epoch": 1.4280367203813875,
      "grad_norm": 0.14785583317279816,
      "learning_rate": 8.62791510765924e-06,
      "loss": 0.2724,
      "step": 2266
    },
    {
      "epoch": 1.4286671131949096,
      "grad_norm": 0.17025531828403473,
      "learning_rate": 8.610198236816226e-06,
      "loss": 0.3058,
      "step": 2267
    },
    {
      "epoch": 1.4292975060084316,
      "grad_norm": 0.1619190275669098,
      "learning_rate": 8.592495269867618e-06,
      "loss": 0.3293,
      "step": 2268
    },
    {
      "epoch": 1.4299278988219535,
      "grad_norm": 0.14388024806976318,
      "learning_rate": 8.574806224534324e-06,
      "loss": 0.3411,
      "step": 2269
    },
    {
      "epoch": 1.4305582916354753,
      "grad_norm": 0.16051793098449707,
      "learning_rate": 8.557131118523345e-06,
      "loss": 0.4002,
      "step": 2270
    },
    {
      "epoch": 1.4311886844489972,
      "grad_norm": 0.17801295220851898,
      "learning_rate": 8.539469969527723e-06,
      "loss": 0.3817,
      "step": 2271
    },
    {
      "epoch": 1.4318190772625192,
      "grad_norm": 0.18220989406108856,
      "learning_rate": 8.521822795226532e-06,
      "loss": 0.333,
      "step": 2272
    },
    {
      "epoch": 1.432449470076041,
      "grad_norm": 0.16216981410980225,
      "learning_rate": 8.504189613284834e-06,
      "loss": 0.3632,
      "step": 2273
    },
    {
      "epoch": 1.4330798628895631,
      "grad_norm": 0.20460912585258484,
      "learning_rate": 8.486570441353714e-06,
      "loss": 0.4964,
      "step": 2274
    },
    {
      "epoch": 1.433710255703085,
      "grad_norm": 0.20223969221115112,
      "learning_rate": 8.468965297070215e-06,
      "loss": 0.3536,
      "step": 2275
    },
    {
      "epoch": 1.4343406485166068,
      "grad_norm": 0.21065102517604828,
      "learning_rate": 8.451374198057354e-06,
      "loss": 0.3481,
      "step": 2276
    },
    {
      "epoch": 1.4349710413301289,
      "grad_norm": 0.12952826917171478,
      "learning_rate": 8.433797161924065e-06,
      "loss": 0.2528,
      "step": 2277
    },
    {
      "epoch": 1.4356014341436507,
      "grad_norm": 0.15633657574653625,
      "learning_rate": 8.416234206265222e-06,
      "loss": 0.3528,
      "step": 2278
    },
    {
      "epoch": 1.4362318269571728,
      "grad_norm": 0.13650497794151306,
      "learning_rate": 8.3986853486616e-06,
      "loss": 0.3002,
      "step": 2279
    },
    {
      "epoch": 1.4368622197706946,
      "grad_norm": 0.21313124895095825,
      "learning_rate": 8.381150606679868e-06,
      "loss": 0.4149,
      "step": 2280
    },
    {
      "epoch": 1.4374926125842165,
      "grad_norm": 0.16620555520057678,
      "learning_rate": 8.363629997872549e-06,
      "loss": 0.3494,
      "step": 2281
    },
    {
      "epoch": 1.4381230053977385,
      "grad_norm": 0.1616056263446808,
      "learning_rate": 8.346123539778038e-06,
      "loss": 0.3469,
      "step": 2282
    },
    {
      "epoch": 1.4387533982112604,
      "grad_norm": 0.15768451988697052,
      "learning_rate": 8.328631249920537e-06,
      "loss": 0.365,
      "step": 2283
    },
    {
      "epoch": 1.4393837910247824,
      "grad_norm": 0.13965880870819092,
      "learning_rate": 8.311153145810112e-06,
      "loss": 0.3409,
      "step": 2284
    },
    {
      "epoch": 1.4400141838383043,
      "grad_norm": 0.13279767334461212,
      "learning_rate": 8.293689244942575e-06,
      "loss": 0.3028,
      "step": 2285
    },
    {
      "epoch": 1.440644576651826,
      "grad_norm": 0.2043708711862564,
      "learning_rate": 8.276239564799564e-06,
      "loss": 0.3897,
      "step": 2286
    },
    {
      "epoch": 1.4412749694653482,
      "grad_norm": 0.16391275823116302,
      "learning_rate": 8.258804122848448e-06,
      "loss": 0.314,
      "step": 2287
    },
    {
      "epoch": 1.44190536227887,
      "grad_norm": 0.16918890178203583,
      "learning_rate": 8.241382936542367e-06,
      "loss": 0.3634,
      "step": 2288
    },
    {
      "epoch": 1.442535755092392,
      "grad_norm": 0.19429834187030792,
      "learning_rate": 8.223976023320185e-06,
      "loss": 0.3305,
      "step": 2289
    },
    {
      "epoch": 1.443166147905914,
      "grad_norm": 0.15197838842868805,
      "learning_rate": 8.206583400606478e-06,
      "loss": 0.2995,
      "step": 2290
    },
    {
      "epoch": 1.4437965407194358,
      "grad_norm": 0.1665520817041397,
      "learning_rate": 8.189205085811502e-06,
      "loss": 0.3707,
      "step": 2291
    },
    {
      "epoch": 1.4444269335329576,
      "grad_norm": 0.16098156571388245,
      "learning_rate": 8.171841096331216e-06,
      "loss": 0.3067,
      "step": 2292
    },
    {
      "epoch": 1.4450573263464797,
      "grad_norm": 0.1439000517129898,
      "learning_rate": 8.154491449547222e-06,
      "loss": 0.3317,
      "step": 2293
    },
    {
      "epoch": 1.4456877191600015,
      "grad_norm": 0.175605908036232,
      "learning_rate": 8.13715616282678e-06,
      "loss": 0.4274,
      "step": 2294
    },
    {
      "epoch": 1.4463181119735236,
      "grad_norm": 0.17509125173091888,
      "learning_rate": 8.119835253522749e-06,
      "loss": 0.4218,
      "step": 2295
    },
    {
      "epoch": 1.4469485047870454,
      "grad_norm": 0.1781781017780304,
      "learning_rate": 8.102528738973618e-06,
      "loss": 0.3294,
      "step": 2296
    },
    {
      "epoch": 1.4475788976005672,
      "grad_norm": 0.19787555932998657,
      "learning_rate": 8.085236636503464e-06,
      "loss": 0.4119,
      "step": 2297
    },
    {
      "epoch": 1.4482092904140893,
      "grad_norm": 0.1802581250667572,
      "learning_rate": 8.067958963421942e-06,
      "loss": 0.4097,
      "step": 2298
    },
    {
      "epoch": 1.4488396832276111,
      "grad_norm": 0.19002848863601685,
      "learning_rate": 8.050695737024237e-06,
      "loss": 0.3125,
      "step": 2299
    },
    {
      "epoch": 1.4494700760411332,
      "grad_norm": 0.19523286819458008,
      "learning_rate": 8.033446974591107e-06,
      "loss": 0.3657,
      "step": 2300
    },
    {
      "epoch": 1.450100468854655,
      "grad_norm": 0.15178027749061584,
      "learning_rate": 8.016212693388793e-06,
      "loss": 0.3377,
      "step": 2301
    },
    {
      "epoch": 1.4507308616681769,
      "grad_norm": 0.1866513192653656,
      "learning_rate": 7.99899291066909e-06,
      "loss": 0.4018,
      "step": 2302
    },
    {
      "epoch": 1.451361254481699,
      "grad_norm": 0.1409262716770172,
      "learning_rate": 7.981787643669234e-06,
      "loss": 0.3302,
      "step": 2303
    },
    {
      "epoch": 1.4519916472952208,
      "grad_norm": 0.14522545039653778,
      "learning_rate": 7.964596909611959e-06,
      "loss": 0.3362,
      "step": 2304
    },
    {
      "epoch": 1.4526220401087429,
      "grad_norm": 0.14141473174095154,
      "learning_rate": 7.94742072570543e-06,
      "loss": 0.3365,
      "step": 2305
    },
    {
      "epoch": 1.4532524329222647,
      "grad_norm": 0.1505013406276703,
      "learning_rate": 7.930259109143259e-06,
      "loss": 0.3251,
      "step": 2306
    },
    {
      "epoch": 1.4538828257357865,
      "grad_norm": 0.14216575026512146,
      "learning_rate": 7.91311207710448e-06,
      "loss": 0.3438,
      "step": 2307
    },
    {
      "epoch": 1.4545132185493086,
      "grad_norm": 0.148826003074646,
      "learning_rate": 7.895979646753533e-06,
      "loss": 0.3608,
      "step": 2308
    },
    {
      "epoch": 1.4551436113628304,
      "grad_norm": 0.19999346137046814,
      "learning_rate": 7.878861835240207e-06,
      "loss": 0.3857,
      "step": 2309
    },
    {
      "epoch": 1.4557740041763525,
      "grad_norm": 0.19859881699085236,
      "learning_rate": 7.861758659699695e-06,
      "loss": 0.3852,
      "step": 2310
    },
    {
      "epoch": 1.4564043969898743,
      "grad_norm": 0.1679404079914093,
      "learning_rate": 7.844670137252523e-06,
      "loss": 0.3958,
      "step": 2311
    },
    {
      "epoch": 1.4570347898033962,
      "grad_norm": 0.19567959010601044,
      "learning_rate": 7.827596285004562e-06,
      "loss": 0.3816,
      "step": 2312
    },
    {
      "epoch": 1.4576651826169182,
      "grad_norm": 0.16922706365585327,
      "learning_rate": 7.810537120046968e-06,
      "loss": 0.3706,
      "step": 2313
    },
    {
      "epoch": 1.45829557543044,
      "grad_norm": 0.15642282366752625,
      "learning_rate": 7.793492659456228e-06,
      "loss": 0.3059,
      "step": 2314
    },
    {
      "epoch": 1.4589259682439621,
      "grad_norm": 0.15599018335342407,
      "learning_rate": 7.77646292029409e-06,
      "loss": 0.3399,
      "step": 2315
    },
    {
      "epoch": 1.459556361057484,
      "grad_norm": 0.19208070635795593,
      "learning_rate": 7.75944791960758e-06,
      "loss": 0.378,
      "step": 2316
    },
    {
      "epoch": 1.4601867538710058,
      "grad_norm": 0.17060023546218872,
      "learning_rate": 7.742447674428948e-06,
      "loss": 0.3805,
      "step": 2317
    },
    {
      "epoch": 1.4608171466845277,
      "grad_norm": 0.18306058645248413,
      "learning_rate": 7.725462201775696e-06,
      "loss": 0.3457,
      "step": 2318
    },
    {
      "epoch": 1.4614475394980497,
      "grad_norm": 0.17282027006149292,
      "learning_rate": 7.708491518650531e-06,
      "loss": 0.3346,
      "step": 2319
    },
    {
      "epoch": 1.4620779323115716,
      "grad_norm": 0.22159677743911743,
      "learning_rate": 7.69153564204136e-06,
      "loss": 0.4076,
      "step": 2320
    },
    {
      "epoch": 1.4627083251250936,
      "grad_norm": 0.1395348459482193,
      "learning_rate": 7.674594588921249e-06,
      "loss": 0.34,
      "step": 2321
    },
    {
      "epoch": 1.4633387179386155,
      "grad_norm": 0.1594412624835968,
      "learning_rate": 7.657668376248456e-06,
      "loss": 0.3198,
      "step": 2322
    },
    {
      "epoch": 1.4639691107521373,
      "grad_norm": 0.18476234376430511,
      "learning_rate": 7.64075702096634e-06,
      "loss": 0.3493,
      "step": 2323
    },
    {
      "epoch": 1.4645995035656594,
      "grad_norm": 0.14744731783866882,
      "learning_rate": 7.623860540003451e-06,
      "loss": 0.4012,
      "step": 2324
    },
    {
      "epoch": 1.4652298963791812,
      "grad_norm": 0.15878044068813324,
      "learning_rate": 7.6069789502733835e-06,
      "loss": 0.3396,
      "step": 2325
    },
    {
      "epoch": 1.4658602891927033,
      "grad_norm": 0.16711211204528809,
      "learning_rate": 7.590112268674876e-06,
      "loss": 0.3348,
      "step": 2326
    },
    {
      "epoch": 1.4664906820062251,
      "grad_norm": 0.14225564897060394,
      "learning_rate": 7.5732605120917064e-06,
      "loss": 0.3227,
      "step": 2327
    },
    {
      "epoch": 1.467121074819747,
      "grad_norm": 0.14761362969875336,
      "learning_rate": 7.556423697392737e-06,
      "loss": 0.372,
      "step": 2328
    },
    {
      "epoch": 1.467751467633269,
      "grad_norm": 0.1955820918083191,
      "learning_rate": 7.539601841431863e-06,
      "loss": 0.3537,
      "step": 2329
    },
    {
      "epoch": 1.4683818604467909,
      "grad_norm": 0.14523617923259735,
      "learning_rate": 7.5227949610480175e-06,
      "loss": 0.2896,
      "step": 2330
    },
    {
      "epoch": 1.469012253260313,
      "grad_norm": 0.13613255321979523,
      "learning_rate": 7.506003073065118e-06,
      "loss": 0.2918,
      "step": 2331
    },
    {
      "epoch": 1.4696426460738348,
      "grad_norm": 0.1531195491552353,
      "learning_rate": 7.489226194292096e-06,
      "loss": 0.3214,
      "step": 2332
    },
    {
      "epoch": 1.4702730388873566,
      "grad_norm": 0.12154468894004822,
      "learning_rate": 7.472464341522855e-06,
      "loss": 0.3022,
      "step": 2333
    },
    {
      "epoch": 1.4709034317008787,
      "grad_norm": 0.16331082582473755,
      "learning_rate": 7.455717531536259e-06,
      "loss": 0.3501,
      "step": 2334
    },
    {
      "epoch": 1.4715338245144005,
      "grad_norm": 0.15346533060073853,
      "learning_rate": 7.438985781096092e-06,
      "loss": 0.2929,
      "step": 2335
    },
    {
      "epoch": 1.4721642173279226,
      "grad_norm": 0.15735021233558655,
      "learning_rate": 7.422269106951096e-06,
      "loss": 0.3296,
      "step": 2336
    },
    {
      "epoch": 1.4727946101414444,
      "grad_norm": 0.150700643658638,
      "learning_rate": 7.405567525834898e-06,
      "loss": 0.323,
      "step": 2337
    },
    {
      "epoch": 1.4734250029549663,
      "grad_norm": 0.21111014485359192,
      "learning_rate": 7.3888810544660365e-06,
      "loss": 0.3881,
      "step": 2338
    },
    {
      "epoch": 1.4740553957684883,
      "grad_norm": 0.1844063103199005,
      "learning_rate": 7.372209709547894e-06,
      "loss": 0.3513,
      "step": 2339
    },
    {
      "epoch": 1.4746857885820102,
      "grad_norm": 0.1757941097021103,
      "learning_rate": 7.355553507768742e-06,
      "loss": 0.3798,
      "step": 2340
    },
    {
      "epoch": 1.4753161813955322,
      "grad_norm": 0.15453548729419708,
      "learning_rate": 7.338912465801681e-06,
      "loss": 0.3721,
      "step": 2341
    },
    {
      "epoch": 1.475946574209054,
      "grad_norm": 0.1763160228729248,
      "learning_rate": 7.322286600304643e-06,
      "loss": 0.4228,
      "step": 2342
    },
    {
      "epoch": 1.476576967022576,
      "grad_norm": 0.15534138679504395,
      "learning_rate": 7.305675927920348e-06,
      "loss": 0.411,
      "step": 2343
    },
    {
      "epoch": 1.4772073598360977,
      "grad_norm": 0.17668470740318298,
      "learning_rate": 7.289080465276335e-06,
      "loss": 0.3669,
      "step": 2344
    },
    {
      "epoch": 1.4778377526496198,
      "grad_norm": 0.17460039258003235,
      "learning_rate": 7.272500228984886e-06,
      "loss": 0.441,
      "step": 2345
    },
    {
      "epoch": 1.4784681454631416,
      "grad_norm": 0.20688629150390625,
      "learning_rate": 7.255935235643084e-06,
      "loss": 0.424,
      "step": 2346
    },
    {
      "epoch": 1.4790985382766637,
      "grad_norm": 0.15839360654354095,
      "learning_rate": 7.239385501832711e-06,
      "loss": 0.3268,
      "step": 2347
    },
    {
      "epoch": 1.4797289310901856,
      "grad_norm": 0.1791486293077469,
      "learning_rate": 7.222851044120307e-06,
      "loss": 0.3273,
      "step": 2348
    },
    {
      "epoch": 1.4803593239037074,
      "grad_norm": 0.1676637977361679,
      "learning_rate": 7.2063318790570875e-06,
      "loss": 0.3515,
      "step": 2349
    },
    {
      "epoch": 1.4809897167172295,
      "grad_norm": 0.1364106982946396,
      "learning_rate": 7.189828023178987e-06,
      "loss": 0.2846,
      "step": 2350
    },
    {
      "epoch": 1.4816201095307513,
      "grad_norm": 0.15322741866111755,
      "learning_rate": 7.173339493006609e-06,
      "loss": 0.3028,
      "step": 2351
    },
    {
      "epoch": 1.4822505023442734,
      "grad_norm": 0.18440741300582886,
      "learning_rate": 7.156866305045218e-06,
      "loss": 0.3662,
      "step": 2352
    },
    {
      "epoch": 1.4828808951577952,
      "grad_norm": 0.17829926311969757,
      "learning_rate": 7.1404084757847004e-06,
      "loss": 0.371,
      "step": 2353
    },
    {
      "epoch": 1.483511287971317,
      "grad_norm": 0.14179544150829315,
      "learning_rate": 7.12396602169959e-06,
      "loss": 0.3318,
      "step": 2354
    },
    {
      "epoch": 1.484141680784839,
      "grad_norm": 0.16233308613300323,
      "learning_rate": 7.107538959249027e-06,
      "loss": 0.3349,
      "step": 2355
    },
    {
      "epoch": 1.484772073598361,
      "grad_norm": 0.15612120926380157,
      "learning_rate": 7.091127304876745e-06,
      "loss": 0.2996,
      "step": 2356
    },
    {
      "epoch": 1.485402466411883,
      "grad_norm": 0.22952015697956085,
      "learning_rate": 7.074731075011037e-06,
      "loss": 0.3868,
      "step": 2357
    },
    {
      "epoch": 1.4860328592254048,
      "grad_norm": 0.1616676300764084,
      "learning_rate": 7.058350286064775e-06,
      "loss": 0.3905,
      "step": 2358
    },
    {
      "epoch": 1.4866632520389267,
      "grad_norm": 0.15389202535152435,
      "learning_rate": 7.041984954435369e-06,
      "loss": 0.3495,
      "step": 2359
    },
    {
      "epoch": 1.4872936448524487,
      "grad_norm": 0.1607460081577301,
      "learning_rate": 7.025635096504761e-06,
      "loss": 0.2944,
      "step": 2360
    },
    {
      "epoch": 1.4879240376659706,
      "grad_norm": 0.18009735643863678,
      "learning_rate": 7.00930072863938e-06,
      "loss": 0.3959,
      "step": 2361
    },
    {
      "epoch": 1.4885544304794927,
      "grad_norm": 0.189145028591156,
      "learning_rate": 6.992981867190189e-06,
      "loss": 0.3319,
      "step": 2362
    },
    {
      "epoch": 1.4891848232930145,
      "grad_norm": 0.16663429141044617,
      "learning_rate": 6.976678528492578e-06,
      "loss": 0.375,
      "step": 2363
    },
    {
      "epoch": 1.4898152161065363,
      "grad_norm": 0.1959405094385147,
      "learning_rate": 6.960390728866458e-06,
      "loss": 0.344,
      "step": 2364
    },
    {
      "epoch": 1.4904456089200582,
      "grad_norm": 0.12720222771167755,
      "learning_rate": 6.944118484616134e-06,
      "loss": 0.3095,
      "step": 2365
    },
    {
      "epoch": 1.4910760017335802,
      "grad_norm": 0.15662875771522522,
      "learning_rate": 6.927861812030374e-06,
      "loss": 0.3688,
      "step": 2366
    },
    {
      "epoch": 1.4917063945471023,
      "grad_norm": 0.17900823056697845,
      "learning_rate": 6.911620727382329e-06,
      "loss": 0.3871,
      "step": 2367
    },
    {
      "epoch": 1.4923367873606241,
      "grad_norm": 0.17844952642917633,
      "learning_rate": 6.89539524692957e-06,
      "loss": 0.3272,
      "step": 2368
    },
    {
      "epoch": 1.492967180174146,
      "grad_norm": 0.17715677618980408,
      "learning_rate": 6.8791853869140395e-06,
      "loss": 0.3838,
      "step": 2369
    },
    {
      "epoch": 1.4935975729876678,
      "grad_norm": 0.18059316277503967,
      "learning_rate": 6.862991163562051e-06,
      "loss": 0.3856,
      "step": 2370
    },
    {
      "epoch": 1.4942279658011899,
      "grad_norm": 0.15726926922798157,
      "learning_rate": 6.846812593084246e-06,
      "loss": 0.4097,
      "step": 2371
    },
    {
      "epoch": 1.4948583586147117,
      "grad_norm": 0.15653415024280548,
      "learning_rate": 6.830649691675614e-06,
      "loss": 0.3041,
      "step": 2372
    },
    {
      "epoch": 1.4954887514282338,
      "grad_norm": 0.22607919573783875,
      "learning_rate": 6.81450247551546e-06,
      "loss": 0.3754,
      "step": 2373
    },
    {
      "epoch": 1.4961191442417556,
      "grad_norm": 0.1733357161283493,
      "learning_rate": 6.798370960767385e-06,
      "loss": 0.3802,
      "step": 2374
    },
    {
      "epoch": 1.4967495370552775,
      "grad_norm": 0.1510259062051773,
      "learning_rate": 6.78225516357926e-06,
      "loss": 0.3057,
      "step": 2375
    },
    {
      "epoch": 1.4973799298687995,
      "grad_norm": 0.17537620663642883,
      "learning_rate": 6.766155100083238e-06,
      "loss": 0.3372,
      "step": 2376
    },
    {
      "epoch": 1.4980103226823214,
      "grad_norm": 0.19684992730617523,
      "learning_rate": 6.750070786395722e-06,
      "loss": 0.4168,
      "step": 2377
    },
    {
      "epoch": 1.4986407154958434,
      "grad_norm": 0.17185229063034058,
      "learning_rate": 6.734002238617346e-06,
      "loss": 0.3734,
      "step": 2378
    },
    {
      "epoch": 1.4992711083093653,
      "grad_norm": 0.18762269616127014,
      "learning_rate": 6.7179494728329514e-06,
      "loss": 0.3612,
      "step": 2379
    },
    {
      "epoch": 1.4999015011228871,
      "grad_norm": 0.17824314534664154,
      "learning_rate": 6.701912505111595e-06,
      "loss": 0.3269,
      "step": 2380
    },
    {
      "epoch": 1.5005318939364092,
      "grad_norm": 0.17873038351535797,
      "learning_rate": 6.68589135150652e-06,
      "loss": 0.3581,
      "step": 2381
    },
    {
      "epoch": 1.501162286749931,
      "grad_norm": 0.15424737334251404,
      "learning_rate": 6.669886028055138e-06,
      "loss": 0.3541,
      "step": 2382
    },
    {
      "epoch": 1.501792679563453,
      "grad_norm": 0.205153688788414,
      "learning_rate": 6.653896550778998e-06,
      "loss": 0.3629,
      "step": 2383
    },
    {
      "epoch": 1.502423072376975,
      "grad_norm": 0.17122958600521088,
      "learning_rate": 6.637922935683816e-06,
      "loss": 0.4008,
      "step": 2384
    },
    {
      "epoch": 1.5030534651904968,
      "grad_norm": 0.19354306161403656,
      "learning_rate": 6.621965198759393e-06,
      "loss": 0.3817,
      "step": 2385
    },
    {
      "epoch": 1.5036838580040186,
      "grad_norm": 0.18343815207481384,
      "learning_rate": 6.606023355979683e-06,
      "loss": 0.4202,
      "step": 2386
    },
    {
      "epoch": 1.5043142508175407,
      "grad_norm": 0.18962843716144562,
      "learning_rate": 6.5900974233026824e-06,
      "loss": 0.3502,
      "step": 2387
    },
    {
      "epoch": 1.5049446436310627,
      "grad_norm": 0.16619205474853516,
      "learning_rate": 6.574187416670491e-06,
      "loss": 0.3029,
      "step": 2388
    },
    {
      "epoch": 1.5055750364445846,
      "grad_norm": 0.17878541350364685,
      "learning_rate": 6.558293352009264e-06,
      "loss": 0.3589,
      "step": 2389
    },
    {
      "epoch": 1.5062054292581064,
      "grad_norm": 0.15167735517024994,
      "learning_rate": 6.542415245229172e-06,
      "loss": 0.3249,
      "step": 2390
    },
    {
      "epoch": 1.5068358220716283,
      "grad_norm": 0.2101573944091797,
      "learning_rate": 6.526553112224457e-06,
      "loss": 0.3233,
      "step": 2391
    },
    {
      "epoch": 1.5074662148851503,
      "grad_norm": 0.17878898978233337,
      "learning_rate": 6.510706968873328e-06,
      "loss": 0.3383,
      "step": 2392
    },
    {
      "epoch": 1.5080966076986724,
      "grad_norm": 0.1626843959093094,
      "learning_rate": 6.49487683103802e-06,
      "loss": 0.3771,
      "step": 2393
    },
    {
      "epoch": 1.5087270005121942,
      "grad_norm": 0.19067011773586273,
      "learning_rate": 6.4790627145647164e-06,
      "loss": 0.3521,
      "step": 2394
    },
    {
      "epoch": 1.509357393325716,
      "grad_norm": 0.17487511038780212,
      "learning_rate": 6.463264635283584e-06,
      "loss": 0.3363,
      "step": 2395
    },
    {
      "epoch": 1.509987786139238,
      "grad_norm": 0.1691151261329651,
      "learning_rate": 6.447482609008733e-06,
      "loss": 0.3659,
      "step": 2396
    },
    {
      "epoch": 1.51061817895276,
      "grad_norm": 0.1458681970834732,
      "learning_rate": 6.4317166515382125e-06,
      "loss": 0.3196,
      "step": 2397
    },
    {
      "epoch": 1.511248571766282,
      "grad_norm": 0.14862768352031708,
      "learning_rate": 6.415966778653955e-06,
      "loss": 0.3243,
      "step": 2398
    },
    {
      "epoch": 1.5118789645798039,
      "grad_norm": 0.18596191704273224,
      "learning_rate": 6.400233006121826e-06,
      "loss": 0.3664,
      "step": 2399
    },
    {
      "epoch": 1.5125093573933257,
      "grad_norm": 0.15921002626419067,
      "learning_rate": 6.38451534969156e-06,
      "loss": 0.3127,
      "step": 2400
    },
    {
      "epoch": 1.5125093573933257,
      "eval_loss": 0.4196794927120209,
      "eval_runtime": 222.4679,
      "eval_samples_per_second": 4.495,
      "eval_steps_per_second": 4.495,
      "step": 2400
    },
    {
      "epoch": 1.5131397502068475,
      "grad_norm": 0.1784636527299881,
      "learning_rate": 6.368813825096767e-06,
      "loss": 0.3868,
      "step": 2401
    },
    {
      "epoch": 1.5137701430203696,
      "grad_norm": 0.1569874882698059,
      "learning_rate": 6.353128448054886e-06,
      "loss": 0.3194,
      "step": 2402
    },
    {
      "epoch": 1.5144005358338914,
      "grad_norm": 0.1486312448978424,
      "learning_rate": 6.33745923426722e-06,
      "loss": 0.3188,
      "step": 2403
    },
    {
      "epoch": 1.5150309286474135,
      "grad_norm": 0.18690063059329987,
      "learning_rate": 6.321806199418879e-06,
      "loss": 0.4137,
      "step": 2404
    },
    {
      "epoch": 1.5156613214609354,
      "grad_norm": 0.17441874742507935,
      "learning_rate": 6.306169359178786e-06,
      "loss": 0.3199,
      "step": 2405
    },
    {
      "epoch": 1.5162917142744572,
      "grad_norm": 0.16934332251548767,
      "learning_rate": 6.290548729199634e-06,
      "loss": 0.3482,
      "step": 2406
    },
    {
      "epoch": 1.516922107087979,
      "grad_norm": 0.174238383769989,
      "learning_rate": 6.274944325117907e-06,
      "loss": 0.2894,
      "step": 2407
    },
    {
      "epoch": 1.517552499901501,
      "grad_norm": 0.17666368186473846,
      "learning_rate": 6.2593561625538416e-06,
      "loss": 0.393,
      "step": 2408
    },
    {
      "epoch": 1.5181828927150232,
      "grad_norm": 0.16942931711673737,
      "learning_rate": 6.243784257111422e-06,
      "loss": 0.3233,
      "step": 2409
    },
    {
      "epoch": 1.518813285528545,
      "grad_norm": 0.15660184621810913,
      "learning_rate": 6.228228624378343e-06,
      "loss": 0.3481,
      "step": 2410
    },
    {
      "epoch": 1.5194436783420668,
      "grad_norm": 0.18719768524169922,
      "learning_rate": 6.212689279926028e-06,
      "loss": 0.3512,
      "step": 2411
    },
    {
      "epoch": 1.5200740711555887,
      "grad_norm": 0.15997354686260223,
      "learning_rate": 6.197166239309573e-06,
      "loss": 0.371,
      "step": 2412
    },
    {
      "epoch": 1.5207044639691107,
      "grad_norm": 0.15153126418590546,
      "learning_rate": 6.181659518067792e-06,
      "loss": 0.3215,
      "step": 2413
    },
    {
      "epoch": 1.5213348567826328,
      "grad_norm": 0.17869645357131958,
      "learning_rate": 6.166169131723122e-06,
      "loss": 0.3497,
      "step": 2414
    },
    {
      "epoch": 1.5219652495961546,
      "grad_norm": 0.16944412887096405,
      "learning_rate": 6.150695095781676e-06,
      "loss": 0.3523,
      "step": 2415
    },
    {
      "epoch": 1.5225956424096765,
      "grad_norm": 0.186680868268013,
      "learning_rate": 6.135237425733178e-06,
      "loss": 0.3314,
      "step": 2416
    },
    {
      "epoch": 1.5232260352231983,
      "grad_norm": 0.23459716141223907,
      "learning_rate": 6.119796137050989e-06,
      "loss": 0.4228,
      "step": 2417
    },
    {
      "epoch": 1.5238564280367204,
      "grad_norm": 0.14739306271076202,
      "learning_rate": 6.1043712451920615e-06,
      "loss": 0.3521,
      "step": 2418
    },
    {
      "epoch": 1.5244868208502425,
      "grad_norm": 0.16764120757579803,
      "learning_rate": 6.088962765596949e-06,
      "loss": 0.383,
      "step": 2419
    },
    {
      "epoch": 1.5251172136637643,
      "grad_norm": 0.18274492025375366,
      "learning_rate": 6.073570713689747e-06,
      "loss": 0.3351,
      "step": 2420
    },
    {
      "epoch": 1.5257476064772861,
      "grad_norm": 0.17499703168869019,
      "learning_rate": 6.0581951048781325e-06,
      "loss": 0.3788,
      "step": 2421
    },
    {
      "epoch": 1.526377999290808,
      "grad_norm": 0.20380593836307526,
      "learning_rate": 6.042835954553315e-06,
      "loss": 0.3575,
      "step": 2422
    },
    {
      "epoch": 1.52700839210433,
      "grad_norm": 0.2026897519826889,
      "learning_rate": 6.027493278090036e-06,
      "loss": 0.3352,
      "step": 2423
    },
    {
      "epoch": 1.527638784917852,
      "grad_norm": 0.15653148293495178,
      "learning_rate": 6.012167090846522e-06,
      "loss": 0.3743,
      "step": 2424
    },
    {
      "epoch": 1.528269177731374,
      "grad_norm": 0.19078733026981354,
      "learning_rate": 5.996857408164517e-06,
      "loss": 0.3306,
      "step": 2425
    },
    {
      "epoch": 1.5288995705448958,
      "grad_norm": 0.18270157277584076,
      "learning_rate": 5.98156424536924e-06,
      "loss": 0.3914,
      "step": 2426
    },
    {
      "epoch": 1.5295299633584176,
      "grad_norm": 0.17126402258872986,
      "learning_rate": 5.966287617769377e-06,
      "loss": 0.3557,
      "step": 2427
    },
    {
      "epoch": 1.5301603561719397,
      "grad_norm": 0.17806404829025269,
      "learning_rate": 5.95102754065704e-06,
      "loss": 0.3603,
      "step": 2428
    },
    {
      "epoch": 1.5307907489854615,
      "grad_norm": 0.15630225837230682,
      "learning_rate": 5.935784029307802e-06,
      "loss": 0.3428,
      "step": 2429
    },
    {
      "epoch": 1.5314211417989836,
      "grad_norm": 0.1432841718196869,
      "learning_rate": 5.920557098980622e-06,
      "loss": 0.3472,
      "step": 2430
    },
    {
      "epoch": 1.5320515346125054,
      "grad_norm": 0.1667039692401886,
      "learning_rate": 5.905346764917904e-06,
      "loss": 0.3284,
      "step": 2431
    },
    {
      "epoch": 1.5326819274260273,
      "grad_norm": 0.15188711881637573,
      "learning_rate": 5.890153042345397e-06,
      "loss": 0.3359,
      "step": 2432
    },
    {
      "epoch": 1.533312320239549,
      "grad_norm": 0.18007594347000122,
      "learning_rate": 5.8749759464722525e-06,
      "loss": 0.3996,
      "step": 2433
    },
    {
      "epoch": 1.5339427130530712,
      "grad_norm": 0.17693552374839783,
      "learning_rate": 5.85981549249095e-06,
      "loss": 0.3524,
      "step": 2434
    },
    {
      "epoch": 1.5345731058665932,
      "grad_norm": 0.18030700087547302,
      "learning_rate": 5.844671695577333e-06,
      "loss": 0.3913,
      "step": 2435
    },
    {
      "epoch": 1.535203498680115,
      "grad_norm": 0.1581149399280548,
      "learning_rate": 5.829544570890562e-06,
      "loss": 0.3069,
      "step": 2436
    },
    {
      "epoch": 1.535833891493637,
      "grad_norm": 0.18180450797080994,
      "learning_rate": 5.8144341335731175e-06,
      "loss": 0.3135,
      "step": 2437
    },
    {
      "epoch": 1.5364642843071588,
      "grad_norm": 0.15132294595241547,
      "learning_rate": 5.799340398750756e-06,
      "loss": 0.2986,
      "step": 2438
    },
    {
      "epoch": 1.5370946771206808,
      "grad_norm": 0.16063569486141205,
      "learning_rate": 5.784263381532531e-06,
      "loss": 0.3967,
      "step": 2439
    },
    {
      "epoch": 1.5377250699342029,
      "grad_norm": 0.16005367040634155,
      "learning_rate": 5.7692030970107585e-06,
      "loss": 0.2788,
      "step": 2440
    },
    {
      "epoch": 1.5383554627477247,
      "grad_norm": 0.16369113326072693,
      "learning_rate": 5.7541595602610096e-06,
      "loss": 0.3814,
      "step": 2441
    },
    {
      "epoch": 1.5389858555612466,
      "grad_norm": 0.18679864704608917,
      "learning_rate": 5.73913278634207e-06,
      "loss": 0.3903,
      "step": 2442
    },
    {
      "epoch": 1.5396162483747684,
      "grad_norm": 0.18288299441337585,
      "learning_rate": 5.724122790295966e-06,
      "loss": 0.3872,
      "step": 2443
    },
    {
      "epoch": 1.5402466411882905,
      "grad_norm": 0.15036456286907196,
      "learning_rate": 5.709129587147923e-06,
      "loss": 0.3277,
      "step": 2444
    },
    {
      "epoch": 1.5408770340018125,
      "grad_norm": 0.18015111982822418,
      "learning_rate": 5.694153191906366e-06,
      "loss": 0.3685,
      "step": 2445
    },
    {
      "epoch": 1.5415074268153344,
      "grad_norm": 0.1756429523229599,
      "learning_rate": 5.67919361956287e-06,
      "loss": 0.3749,
      "step": 2446
    },
    {
      "epoch": 1.5421378196288562,
      "grad_norm": 0.20075921714305878,
      "learning_rate": 5.66425088509219e-06,
      "loss": 0.401,
      "step": 2447
    },
    {
      "epoch": 1.542768212442378,
      "grad_norm": 0.19082263112068176,
      "learning_rate": 5.649325003452224e-06,
      "loss": 0.3243,
      "step": 2448
    },
    {
      "epoch": 1.5433986052559001,
      "grad_norm": 0.19560390710830688,
      "learning_rate": 5.634415989584001e-06,
      "loss": 0.3659,
      "step": 2449
    },
    {
      "epoch": 1.544028998069422,
      "grad_norm": 0.15874971449375153,
      "learning_rate": 5.619523858411653e-06,
      "loss": 0.3534,
      "step": 2450
    },
    {
      "epoch": 1.544659390882944,
      "grad_norm": 0.18899878859519958,
      "learning_rate": 5.604648624842429e-06,
      "loss": 0.4562,
      "step": 2451
    },
    {
      "epoch": 1.5452897836964659,
      "grad_norm": 0.151338592171669,
      "learning_rate": 5.5897903037666405e-06,
      "loss": 0.3257,
      "step": 2452
    },
    {
      "epoch": 1.5459201765099877,
      "grad_norm": 0.1475290209054947,
      "learning_rate": 5.574948910057708e-06,
      "loss": 0.3249,
      "step": 2453
    },
    {
      "epoch": 1.5465505693235098,
      "grad_norm": 0.1572064757347107,
      "learning_rate": 5.56012445857206e-06,
      "loss": 0.3617,
      "step": 2454
    },
    {
      "epoch": 1.5471809621370316,
      "grad_norm": 0.13381150364875793,
      "learning_rate": 5.5453169641492085e-06,
      "loss": 0.3332,
      "step": 2455
    },
    {
      "epoch": 1.5478113549505537,
      "grad_norm": 0.2090974599123001,
      "learning_rate": 5.530526441611654e-06,
      "loss": 0.3652,
      "step": 2456
    },
    {
      "epoch": 1.5484417477640755,
      "grad_norm": 0.16606074571609497,
      "learning_rate": 5.515752905764933e-06,
      "loss": 0.3139,
      "step": 2457
    },
    {
      "epoch": 1.5490721405775973,
      "grad_norm": 0.20185938477516174,
      "learning_rate": 5.500996371397571e-06,
      "loss": 0.3398,
      "step": 2458
    },
    {
      "epoch": 1.5497025333911192,
      "grad_norm": 0.1493961215019226,
      "learning_rate": 5.486256853281081e-06,
      "loss": 0.3541,
      "step": 2459
    },
    {
      "epoch": 1.5503329262046412,
      "grad_norm": 0.20136871933937073,
      "learning_rate": 5.471534366169921e-06,
      "loss": 0.3318,
      "step": 2460
    },
    {
      "epoch": 1.5509633190181633,
      "grad_norm": 0.1520240306854248,
      "learning_rate": 5.456828924801526e-06,
      "loss": 0.3053,
      "step": 2461
    },
    {
      "epoch": 1.5515937118316852,
      "grad_norm": 0.14991526305675507,
      "learning_rate": 5.442140543896254e-06,
      "loss": 0.375,
      "step": 2462
    },
    {
      "epoch": 1.552224104645207,
      "grad_norm": 0.16552947461605072,
      "learning_rate": 5.427469238157397e-06,
      "loss": 0.3324,
      "step": 2463
    },
    {
      "epoch": 1.5528544974587288,
      "grad_norm": 0.15674886107444763,
      "learning_rate": 5.412815022271133e-06,
      "loss": 0.3353,
      "step": 2464
    },
    {
      "epoch": 1.553484890272251,
      "grad_norm": 0.19823868572711945,
      "learning_rate": 5.398177910906554e-06,
      "loss": 0.4067,
      "step": 2465
    },
    {
      "epoch": 1.554115283085773,
      "grad_norm": 0.16715441644191742,
      "learning_rate": 5.383557918715618e-06,
      "loss": 0.3385,
      "step": 2466
    },
    {
      "epoch": 1.5547456758992948,
      "grad_norm": 0.1568594127893448,
      "learning_rate": 5.3689550603331616e-06,
      "loss": 0.3172,
      "step": 2467
    },
    {
      "epoch": 1.5553760687128166,
      "grad_norm": 0.13004900515079498,
      "learning_rate": 5.354369350376844e-06,
      "loss": 0.3248,
      "step": 2468
    },
    {
      "epoch": 1.5560064615263385,
      "grad_norm": 0.16289322078227997,
      "learning_rate": 5.3398008034471815e-06,
      "loss": 0.3251,
      "step": 2469
    },
    {
      "epoch": 1.5566368543398605,
      "grad_norm": 0.17474089562892914,
      "learning_rate": 5.325249434127501e-06,
      "loss": 0.318,
      "step": 2470
    },
    {
      "epoch": 1.5572672471533826,
      "grad_norm": 0.15705354511737823,
      "learning_rate": 5.310715256983943e-06,
      "loss": 0.3237,
      "step": 2471
    },
    {
      "epoch": 1.5578976399669044,
      "grad_norm": 0.18768437206745148,
      "learning_rate": 5.29619828656542e-06,
      "loss": 0.3563,
      "step": 2472
    },
    {
      "epoch": 1.5585280327804263,
      "grad_norm": 0.14832916855812073,
      "learning_rate": 5.281698537403641e-06,
      "loss": 0.3264,
      "step": 2473
    },
    {
      "epoch": 1.5591584255939481,
      "grad_norm": 0.19431886076927185,
      "learning_rate": 5.267216024013048e-06,
      "loss": 0.3663,
      "step": 2474
    },
    {
      "epoch": 1.5597888184074702,
      "grad_norm": 0.18100936710834503,
      "learning_rate": 5.252750760890875e-06,
      "loss": 0.374,
      "step": 2475
    },
    {
      "epoch": 1.560419211220992,
      "grad_norm": 0.18944475054740906,
      "learning_rate": 5.23830276251704e-06,
      "loss": 0.3784,
      "step": 2476
    },
    {
      "epoch": 1.561049604034514,
      "grad_norm": 0.20163822174072266,
      "learning_rate": 5.2238720433542194e-06,
      "loss": 0.3453,
      "step": 2477
    },
    {
      "epoch": 1.561679996848036,
      "grad_norm": 0.13501304388046265,
      "learning_rate": 5.209458617847754e-06,
      "loss": 0.3022,
      "step": 2478
    },
    {
      "epoch": 1.5623103896615578,
      "grad_norm": 0.18465079367160797,
      "learning_rate": 5.195062500425703e-06,
      "loss": 0.3521,
      "step": 2479
    },
    {
      "epoch": 1.5629407824750798,
      "grad_norm": 0.17045170068740845,
      "learning_rate": 5.180683705498789e-06,
      "loss": 0.3355,
      "step": 2480
    },
    {
      "epoch": 1.5635711752886017,
      "grad_norm": 0.1863599270582199,
      "learning_rate": 5.166322247460405e-06,
      "loss": 0.3979,
      "step": 2481
    },
    {
      "epoch": 1.5642015681021237,
      "grad_norm": 0.19586798548698425,
      "learning_rate": 5.151978140686565e-06,
      "loss": 0.3579,
      "step": 2482
    },
    {
      "epoch": 1.5648319609156456,
      "grad_norm": 0.14061041176319122,
      "learning_rate": 5.137651399535935e-06,
      "loss": 0.3152,
      "step": 2483
    },
    {
      "epoch": 1.5654623537291674,
      "grad_norm": 0.17973418533802032,
      "learning_rate": 5.123342038349792e-06,
      "loss": 0.3428,
      "step": 2484
    },
    {
      "epoch": 1.5660927465426893,
      "grad_norm": 0.13237828016281128,
      "learning_rate": 5.1090500714520214e-06,
      "loss": 0.2866,
      "step": 2485
    },
    {
      "epoch": 1.5667231393562113,
      "grad_norm": 0.1529563069343567,
      "learning_rate": 5.094775513149082e-06,
      "loss": 0.3204,
      "step": 2486
    },
    {
      "epoch": 1.5673535321697334,
      "grad_norm": 0.18000224232673645,
      "learning_rate": 5.08051837773002e-06,
      "loss": 0.3938,
      "step": 2487
    },
    {
      "epoch": 1.5679839249832552,
      "grad_norm": 0.1832749843597412,
      "learning_rate": 5.0662786794664335e-06,
      "loss": 0.383,
      "step": 2488
    },
    {
      "epoch": 1.568614317796777,
      "grad_norm": 0.14123855531215668,
      "learning_rate": 5.052056432612477e-06,
      "loss": 0.3813,
      "step": 2489
    },
    {
      "epoch": 1.569244710610299,
      "grad_norm": 0.22280830144882202,
      "learning_rate": 5.037851651404814e-06,
      "loss": 0.3823,
      "step": 2490
    },
    {
      "epoch": 1.569875103423821,
      "grad_norm": 0.16046370565891266,
      "learning_rate": 5.023664350062651e-06,
      "loss": 0.4011,
      "step": 2491
    },
    {
      "epoch": 1.570505496237343,
      "grad_norm": 0.15080615878105164,
      "learning_rate": 5.009494542787663e-06,
      "loss": 0.3499,
      "step": 2492
    },
    {
      "epoch": 1.5711358890508649,
      "grad_norm": 0.1666601300239563,
      "learning_rate": 4.995342243764061e-06,
      "loss": 0.3163,
      "step": 2493
    },
    {
      "epoch": 1.5717662818643867,
      "grad_norm": 0.18129931390285492,
      "learning_rate": 4.981207467158486e-06,
      "loss": 0.3544,
      "step": 2494
    },
    {
      "epoch": 1.5723966746779086,
      "grad_norm": 0.1465536504983902,
      "learning_rate": 4.967090227120064e-06,
      "loss": 0.2443,
      "step": 2495
    },
    {
      "epoch": 1.5730270674914306,
      "grad_norm": 0.17997393012046814,
      "learning_rate": 4.952990537780347e-06,
      "loss": 0.367,
      "step": 2496
    },
    {
      "epoch": 1.5736574603049527,
      "grad_norm": 0.18684880435466766,
      "learning_rate": 4.938908413253338e-06,
      "loss": 0.4089,
      "step": 2497
    },
    {
      "epoch": 1.5742878531184745,
      "grad_norm": 0.1652742475271225,
      "learning_rate": 4.924843867635449e-06,
      "loss": 0.357,
      "step": 2498
    },
    {
      "epoch": 1.5749182459319964,
      "grad_norm": 0.1724511831998825,
      "learning_rate": 4.910796915005503e-06,
      "loss": 0.3158,
      "step": 2499
    },
    {
      "epoch": 1.5755486387455182,
      "grad_norm": 0.16976529359817505,
      "learning_rate": 4.896767569424689e-06,
      "loss": 0.2956,
      "step": 2500
    },
    {
      "epoch": 1.5761790315590403,
      "grad_norm": 0.14916852116584778,
      "learning_rate": 4.8827558449365935e-06,
      "loss": 0.3085,
      "step": 2501
    },
    {
      "epoch": 1.576809424372562,
      "grad_norm": 0.16239811480045319,
      "learning_rate": 4.868761755567159e-06,
      "loss": 0.353,
      "step": 2502
    },
    {
      "epoch": 1.5774398171860842,
      "grad_norm": 0.17747247219085693,
      "learning_rate": 4.854785315324677e-06,
      "loss": 0.3268,
      "step": 2503
    },
    {
      "epoch": 1.578070209999606,
      "grad_norm": 0.16685880720615387,
      "learning_rate": 4.8408265381997615e-06,
      "loss": 0.341,
      "step": 2504
    },
    {
      "epoch": 1.5787006028131279,
      "grad_norm": 0.14966802299022675,
      "learning_rate": 4.826885438165352e-06,
      "loss": 0.3725,
      "step": 2505
    },
    {
      "epoch": 1.5793309956266497,
      "grad_norm": 0.18024073541164398,
      "learning_rate": 4.812962029176696e-06,
      "loss": 0.3488,
      "step": 2506
    },
    {
      "epoch": 1.5799613884401718,
      "grad_norm": 0.17087945342063904,
      "learning_rate": 4.799056325171336e-06,
      "loss": 0.3603,
      "step": 2507
    },
    {
      "epoch": 1.5805917812536938,
      "grad_norm": 0.16011197865009308,
      "learning_rate": 4.7851683400690724e-06,
      "loss": 0.3461,
      "step": 2508
    },
    {
      "epoch": 1.5812221740672157,
      "grad_norm": 0.1560867875814438,
      "learning_rate": 4.771298087771987e-06,
      "loss": 0.3271,
      "step": 2509
    },
    {
      "epoch": 1.5818525668807375,
      "grad_norm": 0.14889803528785706,
      "learning_rate": 4.757445582164402e-06,
      "loss": 0.3311,
      "step": 2510
    },
    {
      "epoch": 1.5824829596942593,
      "grad_norm": 0.19574882090091705,
      "learning_rate": 4.743610837112888e-06,
      "loss": 0.3039,
      "step": 2511
    },
    {
      "epoch": 1.5831133525077814,
      "grad_norm": 0.15672369301319122,
      "learning_rate": 4.7297938664662124e-06,
      "loss": 0.35,
      "step": 2512
    },
    {
      "epoch": 1.5837437453213035,
      "grad_norm": 0.17589780688285828,
      "learning_rate": 4.715994684055378e-06,
      "loss": 0.3492,
      "step": 2513
    },
    {
      "epoch": 1.5843741381348253,
      "grad_norm": 0.17307446897029877,
      "learning_rate": 4.702213303693549e-06,
      "loss": 0.3479,
      "step": 2514
    },
    {
      "epoch": 1.5850045309483471,
      "grad_norm": 0.136874258518219,
      "learning_rate": 4.6884497391761125e-06,
      "loss": 0.3115,
      "step": 2515
    },
    {
      "epoch": 1.585634923761869,
      "grad_norm": 0.17769919335842133,
      "learning_rate": 4.67470400428058e-06,
      "loss": 0.3022,
      "step": 2516
    },
    {
      "epoch": 1.586265316575391,
      "grad_norm": 0.14690563082695007,
      "learning_rate": 4.660976112766643e-06,
      "loss": 0.3269,
      "step": 2517
    },
    {
      "epoch": 1.5868957093889131,
      "grad_norm": 0.17333537340164185,
      "learning_rate": 4.647266078376113e-06,
      "loss": 0.3409,
      "step": 2518
    },
    {
      "epoch": 1.587526102202435,
      "grad_norm": 0.1596616804599762,
      "learning_rate": 4.633573914832937e-06,
      "loss": 0.3709,
      "step": 2519
    },
    {
      "epoch": 1.5881564950159568,
      "grad_norm": 0.1438072770833969,
      "learning_rate": 4.619899635843173e-06,
      "loss": 0.291,
      "step": 2520
    },
    {
      "epoch": 1.5887868878294786,
      "grad_norm": 0.1735801249742508,
      "learning_rate": 4.606243255094979e-06,
      "loss": 0.3477,
      "step": 2521
    },
    {
      "epoch": 1.5894172806430007,
      "grad_norm": 0.1408032774925232,
      "learning_rate": 4.592604786258578e-06,
      "loss": 0.3318,
      "step": 2522
    },
    {
      "epoch": 1.5900476734565228,
      "grad_norm": 0.16737176477909088,
      "learning_rate": 4.578984242986286e-06,
      "loss": 0.3742,
      "step": 2523
    },
    {
      "epoch": 1.5906780662700446,
      "grad_norm": 0.16467176377773285,
      "learning_rate": 4.565381638912463e-06,
      "loss": 0.3153,
      "step": 2524
    },
    {
      "epoch": 1.5913084590835664,
      "grad_norm": 0.1506030112504959,
      "learning_rate": 4.551796987653522e-06,
      "loss": 0.2828,
      "step": 2525
    },
    {
      "epoch": 1.5919388518970883,
      "grad_norm": 0.16982591152191162,
      "learning_rate": 4.538230302807883e-06,
      "loss": 0.3579,
      "step": 2526
    },
    {
      "epoch": 1.5925692447106103,
      "grad_norm": 0.1681368499994278,
      "learning_rate": 4.524681597956003e-06,
      "loss": 0.3045,
      "step": 2527
    },
    {
      "epoch": 1.5931996375241322,
      "grad_norm": 0.15242716670036316,
      "learning_rate": 4.511150886660335e-06,
      "loss": 0.3063,
      "step": 2528
    },
    {
      "epoch": 1.5938300303376542,
      "grad_norm": 0.1426815241575241,
      "learning_rate": 4.497638182465323e-06,
      "loss": 0.3325,
      "step": 2529
    },
    {
      "epoch": 1.594460423151176,
      "grad_norm": 0.15260830521583557,
      "learning_rate": 4.484143498897367e-06,
      "loss": 0.2992,
      "step": 2530
    },
    {
      "epoch": 1.595090815964698,
      "grad_norm": 0.16513516008853912,
      "learning_rate": 4.470666849464857e-06,
      "loss": 0.3382,
      "step": 2531
    },
    {
      "epoch": 1.5957212087782198,
      "grad_norm": 0.16228486597537994,
      "learning_rate": 4.4572082476581e-06,
      "loss": 0.3705,
      "step": 2532
    },
    {
      "epoch": 1.5963516015917418,
      "grad_norm": 0.15340639650821686,
      "learning_rate": 4.443767706949372e-06,
      "loss": 0.3295,
      "step": 2533
    },
    {
      "epoch": 1.596981994405264,
      "grad_norm": 0.1555001437664032,
      "learning_rate": 4.430345240792837e-06,
      "loss": 0.349,
      "step": 2534
    },
    {
      "epoch": 1.5976123872187857,
      "grad_norm": 0.16534945368766785,
      "learning_rate": 4.416940862624589e-06,
      "loss": 0.3621,
      "step": 2535
    },
    {
      "epoch": 1.5982427800323076,
      "grad_norm": 0.18221977353096008,
      "learning_rate": 4.403554585862589e-06,
      "loss": 0.3357,
      "step": 2536
    },
    {
      "epoch": 1.5988731728458294,
      "grad_norm": 0.14937010407447815,
      "learning_rate": 4.390186423906717e-06,
      "loss": 0.3208,
      "step": 2537
    },
    {
      "epoch": 1.5995035656593515,
      "grad_norm": 0.24392864108085632,
      "learning_rate": 4.376836390138687e-06,
      "loss": 0.3775,
      "step": 2538
    },
    {
      "epoch": 1.6001339584728735,
      "grad_norm": 0.14190733432769775,
      "learning_rate": 4.363504497922081e-06,
      "loss": 0.3168,
      "step": 2539
    },
    {
      "epoch": 1.6007643512863954,
      "grad_norm": 0.19658371806144714,
      "learning_rate": 4.350190760602313e-06,
      "loss": 0.3723,
      "step": 2540
    },
    {
      "epoch": 1.6013947440999172,
      "grad_norm": 0.18854087591171265,
      "learning_rate": 4.336895191506629e-06,
      "loss": 0.3595,
      "step": 2541
    },
    {
      "epoch": 1.602025136913439,
      "grad_norm": 0.16696226596832275,
      "learning_rate": 4.323617803944094e-06,
      "loss": 0.3631,
      "step": 2542
    },
    {
      "epoch": 1.6026555297269611,
      "grad_norm": 0.13967862725257874,
      "learning_rate": 4.310358611205567e-06,
      "loss": 0.2913,
      "step": 2543
    },
    {
      "epoch": 1.6032859225404832,
      "grad_norm": 0.17204873263835907,
      "learning_rate": 4.297117626563685e-06,
      "loss": 0.3681,
      "step": 2544
    },
    {
      "epoch": 1.603916315354005,
      "grad_norm": 0.1861438751220703,
      "learning_rate": 4.283894863272871e-06,
      "loss": 0.3924,
      "step": 2545
    },
    {
      "epoch": 1.6045467081675269,
      "grad_norm": 0.1662445217370987,
      "learning_rate": 4.270690334569305e-06,
      "loss": 0.3553,
      "step": 2546
    },
    {
      "epoch": 1.6051771009810487,
      "grad_norm": 0.16762596368789673,
      "learning_rate": 4.257504053670921e-06,
      "loss": 0.3248,
      "step": 2547
    },
    {
      "epoch": 1.6058074937945708,
      "grad_norm": 0.16183894872665405,
      "learning_rate": 4.244336033777362e-06,
      "loss": 0.3225,
      "step": 2548
    },
    {
      "epoch": 1.6064378866080926,
      "grad_norm": 0.1885928511619568,
      "learning_rate": 4.2311862880700205e-06,
      "loss": 0.4148,
      "step": 2549
    },
    {
      "epoch": 1.6070682794216147,
      "grad_norm": 0.20267227292060852,
      "learning_rate": 4.2180548297119785e-06,
      "loss": 0.3956,
      "step": 2550
    },
    {
      "epoch": 1.6076986722351365,
      "grad_norm": 0.1709766834974289,
      "learning_rate": 4.204941671848024e-06,
      "loss": 0.3204,
      "step": 2551
    },
    {
      "epoch": 1.6083290650486584,
      "grad_norm": 0.1575269252061844,
      "learning_rate": 4.19184682760461e-06,
      "loss": 0.3385,
      "step": 2552
    },
    {
      "epoch": 1.6089594578621804,
      "grad_norm": 0.1714513599872589,
      "learning_rate": 4.17877031008988e-06,
      "loss": 0.2944,
      "step": 2553
    },
    {
      "epoch": 1.6095898506757023,
      "grad_norm": 0.16309994459152222,
      "learning_rate": 4.165712132393597e-06,
      "loss": 0.3186,
      "step": 2554
    },
    {
      "epoch": 1.6102202434892243,
      "grad_norm": 0.13920311629772186,
      "learning_rate": 4.1526723075872146e-06,
      "loss": 0.2782,
      "step": 2555
    },
    {
      "epoch": 1.6108506363027462,
      "grad_norm": 0.17000916600227356,
      "learning_rate": 4.1396508487237676e-06,
      "loss": 0.3481,
      "step": 2556
    },
    {
      "epoch": 1.611481029116268,
      "grad_norm": 0.1467917263507843,
      "learning_rate": 4.1266477688379384e-06,
      "loss": 0.3042,
      "step": 2557
    },
    {
      "epoch": 1.6121114219297898,
      "grad_norm": 0.18692618608474731,
      "learning_rate": 4.113663080945993e-06,
      "loss": 0.3545,
      "step": 2558
    },
    {
      "epoch": 1.612741814743312,
      "grad_norm": 0.1667277216911316,
      "learning_rate": 4.100696798045792e-06,
      "loss": 0.3248,
      "step": 2559
    },
    {
      "epoch": 1.613372207556834,
      "grad_norm": 0.20179854333400726,
      "learning_rate": 4.08774893311678e-06,
      "loss": 0.3872,
      "step": 2560
    },
    {
      "epoch": 1.6140026003703558,
      "grad_norm": 0.17441004514694214,
      "learning_rate": 4.07481949911996e-06,
      "loss": 0.3422,
      "step": 2561
    },
    {
      "epoch": 1.6146329931838777,
      "grad_norm": 0.16981568932533264,
      "learning_rate": 4.061908508997876e-06,
      "loss": 0.3244,
      "step": 2562
    },
    {
      "epoch": 1.6152633859973995,
      "grad_norm": 0.1381535828113556,
      "learning_rate": 4.049015975674623e-06,
      "loss": 0.2821,
      "step": 2563
    },
    {
      "epoch": 1.6158937788109216,
      "grad_norm": 0.22367283701896667,
      "learning_rate": 4.036141912055813e-06,
      "loss": 0.4001,
      "step": 2564
    },
    {
      "epoch": 1.6165241716244436,
      "grad_norm": 0.1576678603887558,
      "learning_rate": 4.0232863310285795e-06,
      "loss": 0.3758,
      "step": 2565
    },
    {
      "epoch": 1.6171545644379655,
      "grad_norm": 0.17275619506835938,
      "learning_rate": 4.010449245461535e-06,
      "loss": 0.3641,
      "step": 2566
    },
    {
      "epoch": 1.6177849572514873,
      "grad_norm": 0.15438169240951538,
      "learning_rate": 3.997630668204796e-06,
      "loss": 0.3748,
      "step": 2567
    },
    {
      "epoch": 1.6184153500650091,
      "grad_norm": 0.1662559062242508,
      "learning_rate": 3.984830612089941e-06,
      "loss": 0.2801,
      "step": 2568
    },
    {
      "epoch": 1.6190457428785312,
      "grad_norm": 0.1477411836385727,
      "learning_rate": 3.972049089930026e-06,
      "loss": 0.324,
      "step": 2569
    },
    {
      "epoch": 1.6196761356920533,
      "grad_norm": 0.17186160385608673,
      "learning_rate": 3.959286114519524e-06,
      "loss": 0.3551,
      "step": 2570
    },
    {
      "epoch": 1.620306528505575,
      "grad_norm": 0.1736433207988739,
      "learning_rate": 3.946541698634368e-06,
      "loss": 0.324,
      "step": 2571
    },
    {
      "epoch": 1.620936921319097,
      "grad_norm": 0.20666708052158356,
      "learning_rate": 3.933815855031907e-06,
      "loss": 0.3898,
      "step": 2572
    },
    {
      "epoch": 1.6215673141326188,
      "grad_norm": 0.16306789219379425,
      "learning_rate": 3.921108596450895e-06,
      "loss": 0.3686,
      "step": 2573
    },
    {
      "epoch": 1.6221977069461408,
      "grad_norm": 0.17801955342292786,
      "learning_rate": 3.9084199356114796e-06,
      "loss": 0.3579,
      "step": 2574
    },
    {
      "epoch": 1.6228280997596627,
      "grad_norm": 0.15669108927249908,
      "learning_rate": 3.895749885215201e-06,
      "loss": 0.3413,
      "step": 2575
    },
    {
      "epoch": 1.6234584925731848,
      "grad_norm": 0.1938735842704773,
      "learning_rate": 3.883098457944951e-06,
      "loss": 0.3592,
      "step": 2576
    },
    {
      "epoch": 1.6240888853867066,
      "grad_norm": 0.15771815180778503,
      "learning_rate": 3.8704656664650145e-06,
      "loss": 0.3474,
      "step": 2577
    },
    {
      "epoch": 1.6247192782002284,
      "grad_norm": 0.14929990470409393,
      "learning_rate": 3.857851523420986e-06,
      "loss": 0.2952,
      "step": 2578
    },
    {
      "epoch": 1.6253496710137505,
      "grad_norm": 0.20388369262218475,
      "learning_rate": 3.845256041439812e-06,
      "loss": 0.3406,
      "step": 2579
    },
    {
      "epoch": 1.6259800638272723,
      "grad_norm": 0.1572895348072052,
      "learning_rate": 3.832679233129748e-06,
      "loss": 0.3672,
      "step": 2580
    },
    {
      "epoch": 1.6266104566407944,
      "grad_norm": 0.1904786080121994,
      "learning_rate": 3.820121111080368e-06,
      "loss": 0.3725,
      "step": 2581
    },
    {
      "epoch": 1.6272408494543162,
      "grad_norm": 0.1755596548318863,
      "learning_rate": 3.807581687862533e-06,
      "loss": 0.3525,
      "step": 2582
    },
    {
      "epoch": 1.627871242267838,
      "grad_norm": 0.17767442762851715,
      "learning_rate": 3.7950609760283983e-06,
      "loss": 0.3402,
      "step": 2583
    },
    {
      "epoch": 1.62850163508136,
      "grad_norm": 0.1524641066789627,
      "learning_rate": 3.782558988111363e-06,
      "loss": 0.3572,
      "step": 2584
    },
    {
      "epoch": 1.629132027894882,
      "grad_norm": 0.15821711719036102,
      "learning_rate": 3.7700757366261087e-06,
      "loss": 0.3378,
      "step": 2585
    },
    {
      "epoch": 1.629762420708404,
      "grad_norm": 0.12943841516971588,
      "learning_rate": 3.7576112340685497e-06,
      "loss": 0.2988,
      "step": 2586
    },
    {
      "epoch": 1.6303928135219259,
      "grad_norm": 0.18882150948047638,
      "learning_rate": 3.745165492915843e-06,
      "loss": 0.3968,
      "step": 2587
    },
    {
      "epoch": 1.6310232063354477,
      "grad_norm": 0.1451679915189743,
      "learning_rate": 3.7327385256263476e-06,
      "loss": 0.3529,
      "step": 2588
    },
    {
      "epoch": 1.6316535991489696,
      "grad_norm": 0.16361628472805023,
      "learning_rate": 3.720330344639641e-06,
      "loss": 0.3535,
      "step": 2589
    },
    {
      "epoch": 1.6322839919624916,
      "grad_norm": 0.13375185430049896,
      "learning_rate": 3.707940962376494e-06,
      "loss": 0.3064,
      "step": 2590
    },
    {
      "epoch": 1.6329143847760137,
      "grad_norm": 0.15740998089313507,
      "learning_rate": 3.695570391238867e-06,
      "loss": 0.3922,
      "step": 2591
    },
    {
      "epoch": 1.6335447775895355,
      "grad_norm": 0.21207581460475922,
      "learning_rate": 3.6832186436098723e-06,
      "loss": 0.3297,
      "step": 2592
    },
    {
      "epoch": 1.6341751704030574,
      "grad_norm": 0.15415050089359283,
      "learning_rate": 3.6708857318537964e-06,
      "loss": 0.3319,
      "step": 2593
    },
    {
      "epoch": 1.6348055632165792,
      "grad_norm": 0.17594383656978607,
      "learning_rate": 3.658571668316049e-06,
      "loss": 0.2924,
      "step": 2594
    },
    {
      "epoch": 1.6354359560301013,
      "grad_norm": 0.1760425716638565,
      "learning_rate": 3.646276465323212e-06,
      "loss": 0.386,
      "step": 2595
    },
    {
      "epoch": 1.6360663488436233,
      "grad_norm": 0.16595324873924255,
      "learning_rate": 3.634000135182944e-06,
      "loss": 0.2909,
      "step": 2596
    },
    {
      "epoch": 1.6366967416571452,
      "grad_norm": 0.20991137623786926,
      "learning_rate": 3.6217426901840434e-06,
      "loss": 0.4425,
      "step": 2597
    },
    {
      "epoch": 1.637327134470667,
      "grad_norm": 0.1507931649684906,
      "learning_rate": 3.6095041425963793e-06,
      "loss": 0.3358,
      "step": 2598
    },
    {
      "epoch": 1.6379575272841889,
      "grad_norm": 0.17338827252388,
      "learning_rate": 3.5972845046709208e-06,
      "loss": 0.298,
      "step": 2599
    },
    {
      "epoch": 1.638587920097711,
      "grad_norm": 0.14650113880634308,
      "learning_rate": 3.585083788639709e-06,
      "loss": 0.3326,
      "step": 2600
    },
    {
      "epoch": 1.638587920097711,
      "eval_loss": 0.4189898371696472,
      "eval_runtime": 222.1834,
      "eval_samples_per_second": 4.501,
      "eval_steps_per_second": 4.501,
      "step": 2600
    },
    {
      "epoch": 1.6392183129112328,
      "grad_norm": 0.18736357986927032,
      "learning_rate": 3.5729020067158394e-06,
      "loss": 0.3513,
      "step": 2601
    },
    {
      "epoch": 1.6398487057247548,
      "grad_norm": 0.17417657375335693,
      "learning_rate": 3.5607391710934446e-06,
      "loss": 0.2972,
      "step": 2602
    },
    {
      "epoch": 1.6404790985382767,
      "grad_norm": 0.16662700474262238,
      "learning_rate": 3.548595293947706e-06,
      "loss": 0.3203,
      "step": 2603
    },
    {
      "epoch": 1.6411094913517985,
      "grad_norm": 0.15887589752674103,
      "learning_rate": 3.536470387434822e-06,
      "loss": 0.338,
      "step": 2604
    },
    {
      "epoch": 1.6417398841653204,
      "grad_norm": 0.21726834774017334,
      "learning_rate": 3.524364463692004e-06,
      "loss": 0.4007,
      "step": 2605
    },
    {
      "epoch": 1.6423702769788424,
      "grad_norm": 0.20520001649856567,
      "learning_rate": 3.5122775348374533e-06,
      "loss": 0.3685,
      "step": 2606
    },
    {
      "epoch": 1.6430006697923645,
      "grad_norm": 0.1584928184747696,
      "learning_rate": 3.5002096129703615e-06,
      "loss": 0.3241,
      "step": 2607
    },
    {
      "epoch": 1.6436310626058863,
      "grad_norm": 0.17537303268909454,
      "learning_rate": 3.4881607101708977e-06,
      "loss": 0.3486,
      "step": 2608
    },
    {
      "epoch": 1.6442614554194082,
      "grad_norm": 0.1952475756406784,
      "learning_rate": 3.476130838500197e-06,
      "loss": 0.3844,
      "step": 2609
    },
    {
      "epoch": 1.64489184823293,
      "grad_norm": 0.1399722397327423,
      "learning_rate": 3.4641200100003214e-06,
      "loss": 0.3307,
      "step": 2610
    },
    {
      "epoch": 1.645522241046452,
      "grad_norm": 0.17013977468013763,
      "learning_rate": 3.4521282366942916e-06,
      "loss": 0.3501,
      "step": 2611
    },
    {
      "epoch": 1.6461526338599741,
      "grad_norm": 0.17964443564414978,
      "learning_rate": 3.440155530586052e-06,
      "loss": 0.3822,
      "step": 2612
    },
    {
      "epoch": 1.646783026673496,
      "grad_norm": 0.17054222524166107,
      "learning_rate": 3.4282019036604564e-06,
      "loss": 0.3256,
      "step": 2613
    },
    {
      "epoch": 1.6474134194870178,
      "grad_norm": 0.1672014743089676,
      "learning_rate": 3.4162673678832526e-06,
      "loss": 0.3639,
      "step": 2614
    },
    {
      "epoch": 1.6480438123005396,
      "grad_norm": 0.1616799235343933,
      "learning_rate": 3.404351935201095e-06,
      "loss": 0.2757,
      "step": 2615
    },
    {
      "epoch": 1.6486742051140617,
      "grad_norm": 0.16986674070358276,
      "learning_rate": 3.3924556175414895e-06,
      "loss": 0.3926,
      "step": 2616
    },
    {
      "epoch": 1.6493045979275838,
      "grad_norm": 0.20208294689655304,
      "learning_rate": 3.3805784268128456e-06,
      "loss": 0.3595,
      "step": 2617
    },
    {
      "epoch": 1.6499349907411056,
      "grad_norm": 0.15416187047958374,
      "learning_rate": 3.368720374904388e-06,
      "loss": 0.2523,
      "step": 2618
    },
    {
      "epoch": 1.6505653835546275,
      "grad_norm": 0.148421049118042,
      "learning_rate": 3.3568814736862115e-06,
      "loss": 0.3518,
      "step": 2619
    },
    {
      "epoch": 1.6511957763681493,
      "grad_norm": 0.16845634579658508,
      "learning_rate": 3.3450617350092163e-06,
      "loss": 0.3474,
      "step": 2620
    },
    {
      "epoch": 1.6518261691816714,
      "grad_norm": 0.20402076840400696,
      "learning_rate": 3.3332611707051404e-06,
      "loss": 0.3908,
      "step": 2621
    },
    {
      "epoch": 1.6524565619951934,
      "grad_norm": 0.16622604429721832,
      "learning_rate": 3.3214797925865198e-06,
      "loss": 0.3393,
      "step": 2622
    },
    {
      "epoch": 1.6530869548087153,
      "grad_norm": 0.18637624382972717,
      "learning_rate": 3.3097176124466936e-06,
      "loss": 0.3607,
      "step": 2623
    },
    {
      "epoch": 1.653717347622237,
      "grad_norm": 0.18721283972263336,
      "learning_rate": 3.2979746420597634e-06,
      "loss": 0.3442,
      "step": 2624
    },
    {
      "epoch": 1.654347740435759,
      "grad_norm": 0.18155747652053833,
      "learning_rate": 3.2862508931806207e-06,
      "loss": 0.3225,
      "step": 2625
    },
    {
      "epoch": 1.654978133249281,
      "grad_norm": 0.16414645314216614,
      "learning_rate": 3.2745463775449087e-06,
      "loss": 0.3125,
      "step": 2626
    },
    {
      "epoch": 1.6556085260628028,
      "grad_norm": 0.1629631072282791,
      "learning_rate": 3.262861106869022e-06,
      "loss": 0.3336,
      "step": 2627
    },
    {
      "epoch": 1.656238918876325,
      "grad_norm": 0.20036570727825165,
      "learning_rate": 3.25119509285008e-06,
      "loss": 0.4085,
      "step": 2628
    },
    {
      "epoch": 1.6568693116898467,
      "grad_norm": 0.16448689997196198,
      "learning_rate": 3.2395483471659357e-06,
      "loss": 0.294,
      "step": 2629
    },
    {
      "epoch": 1.6574997045033686,
      "grad_norm": 0.18873000144958496,
      "learning_rate": 3.2279208814751527e-06,
      "loss": 0.3816,
      "step": 2630
    },
    {
      "epoch": 1.6581300973168904,
      "grad_norm": 0.18073073029518127,
      "learning_rate": 3.216312707416999e-06,
      "loss": 0.3841,
      "step": 2631
    },
    {
      "epoch": 1.6587604901304125,
      "grad_norm": 0.19359445571899414,
      "learning_rate": 3.204723836611414e-06,
      "loss": 0.3783,
      "step": 2632
    },
    {
      "epoch": 1.6593908829439346,
      "grad_norm": 0.19595137238502502,
      "learning_rate": 3.1931542806590356e-06,
      "loss": 0.4084,
      "step": 2633
    },
    {
      "epoch": 1.6600212757574564,
      "grad_norm": 0.19038718938827515,
      "learning_rate": 3.1816040511411552e-06,
      "loss": 0.429,
      "step": 2634
    },
    {
      "epoch": 1.6606516685709782,
      "grad_norm": 0.18803559243679047,
      "learning_rate": 3.1700731596197288e-06,
      "loss": 0.3572,
      "step": 2635
    },
    {
      "epoch": 1.6612820613845,
      "grad_norm": 0.1585325002670288,
      "learning_rate": 3.1585616176373336e-06,
      "loss": 0.3065,
      "step": 2636
    },
    {
      "epoch": 1.6619124541980221,
      "grad_norm": 0.16231995820999146,
      "learning_rate": 3.1470694367172044e-06,
      "loss": 0.3307,
      "step": 2637
    },
    {
      "epoch": 1.6625428470115442,
      "grad_norm": 0.1627531200647354,
      "learning_rate": 3.1355966283631664e-06,
      "loss": 0.3506,
      "step": 2638
    },
    {
      "epoch": 1.663173239825066,
      "grad_norm": 0.14652608335018158,
      "learning_rate": 3.124143204059688e-06,
      "loss": 0.3366,
      "step": 2639
    },
    {
      "epoch": 1.6638036326385879,
      "grad_norm": 0.13706839084625244,
      "learning_rate": 3.1127091752717983e-06,
      "loss": 0.3214,
      "step": 2640
    },
    {
      "epoch": 1.6644340254521097,
      "grad_norm": 0.15212666988372803,
      "learning_rate": 3.101294553445139e-06,
      "loss": 0.3175,
      "step": 2641
    },
    {
      "epoch": 1.6650644182656318,
      "grad_norm": 0.16928543150424957,
      "learning_rate": 3.0898993500059028e-06,
      "loss": 0.3812,
      "step": 2642
    },
    {
      "epoch": 1.6656948110791538,
      "grad_norm": 0.15869659185409546,
      "learning_rate": 3.078523576360858e-06,
      "loss": 0.3082,
      "step": 2643
    },
    {
      "epoch": 1.6663252038926757,
      "grad_norm": 0.16265471279621124,
      "learning_rate": 3.06716724389732e-06,
      "loss": 0.2794,
      "step": 2644
    },
    {
      "epoch": 1.6669555967061975,
      "grad_norm": 0.18983694911003113,
      "learning_rate": 3.055830363983153e-06,
      "loss": 0.3445,
      "step": 2645
    },
    {
      "epoch": 1.6675859895197194,
      "grad_norm": 0.18641945719718933,
      "learning_rate": 3.0445129479667224e-06,
      "loss": 0.3768,
      "step": 2646
    },
    {
      "epoch": 1.6682163823332414,
      "grad_norm": 0.1502719521522522,
      "learning_rate": 3.0332150071769383e-06,
      "loss": 0.2855,
      "step": 2647
    },
    {
      "epoch": 1.6688467751467633,
      "grad_norm": 0.1414182037115097,
      "learning_rate": 3.021936552923199e-06,
      "loss": 0.3237,
      "step": 2648
    },
    {
      "epoch": 1.6694771679602853,
      "grad_norm": 0.21098460257053375,
      "learning_rate": 3.010677596495411e-06,
      "loss": 0.426,
      "step": 2649
    },
    {
      "epoch": 1.6701075607738072,
      "grad_norm": 0.19079872965812683,
      "learning_rate": 2.9994381491639435e-06,
      "loss": 0.349,
      "step": 2650
    },
    {
      "epoch": 1.670737953587329,
      "grad_norm": 0.16061921417713165,
      "learning_rate": 2.98821822217965e-06,
      "loss": 0.3617,
      "step": 2651
    },
    {
      "epoch": 1.671368346400851,
      "grad_norm": 0.20180317759513855,
      "learning_rate": 2.9770178267738435e-06,
      "loss": 0.3442,
      "step": 2652
    },
    {
      "epoch": 1.671998739214373,
      "grad_norm": 0.15677542984485626,
      "learning_rate": 2.9658369741582877e-06,
      "loss": 0.3209,
      "step": 2653
    },
    {
      "epoch": 1.672629132027895,
      "grad_norm": 0.14801929891109467,
      "learning_rate": 2.9546756755251683e-06,
      "loss": 0.2879,
      "step": 2654
    },
    {
      "epoch": 1.6732595248414168,
      "grad_norm": 0.16312134265899658,
      "learning_rate": 2.9435339420471195e-06,
      "loss": 0.3165,
      "step": 2655
    },
    {
      "epoch": 1.6738899176549387,
      "grad_norm": 0.16553540527820587,
      "learning_rate": 2.9324117848771653e-06,
      "loss": 0.2867,
      "step": 2656
    },
    {
      "epoch": 1.6745203104684605,
      "grad_norm": 0.16059130430221558,
      "learning_rate": 2.921309215148765e-06,
      "loss": 0.3347,
      "step": 2657
    },
    {
      "epoch": 1.6751507032819826,
      "grad_norm": 0.19318628311157227,
      "learning_rate": 2.910226243975741e-06,
      "loss": 0.3434,
      "step": 2658
    },
    {
      "epoch": 1.6757810960955046,
      "grad_norm": 0.17356213927268982,
      "learning_rate": 2.899162882452313e-06,
      "loss": 0.3595,
      "step": 2659
    },
    {
      "epoch": 1.6764114889090265,
      "grad_norm": 0.13051897287368774,
      "learning_rate": 2.888119141653066e-06,
      "loss": 0.3069,
      "step": 2660
    },
    {
      "epoch": 1.6770418817225483,
      "grad_norm": 0.203187957406044,
      "learning_rate": 2.877095032632946e-06,
      "loss": 0.3804,
      "step": 2661
    },
    {
      "epoch": 1.6776722745360702,
      "grad_norm": 0.1899317353963852,
      "learning_rate": 2.8660905664272462e-06,
      "loss": 0.3632,
      "step": 2662
    },
    {
      "epoch": 1.6783026673495922,
      "grad_norm": 0.1583564132452011,
      "learning_rate": 2.8551057540516045e-06,
      "loss": 0.3131,
      "step": 2663
    },
    {
      "epoch": 1.6789330601631143,
      "grad_norm": 0.1547822654247284,
      "learning_rate": 2.8441406065019684e-06,
      "loss": 0.2862,
      "step": 2664
    },
    {
      "epoch": 1.6795634529766361,
      "grad_norm": 0.1812221109867096,
      "learning_rate": 2.8331951347546167e-06,
      "loss": 0.3665,
      "step": 2665
    },
    {
      "epoch": 1.680193845790158,
      "grad_norm": 0.18095368146896362,
      "learning_rate": 2.822269349766126e-06,
      "loss": 0.3622,
      "step": 2666
    },
    {
      "epoch": 1.6808242386036798,
      "grad_norm": 0.15226151049137115,
      "learning_rate": 2.8113632624733697e-06,
      "loss": 0.348,
      "step": 2667
    },
    {
      "epoch": 1.6814546314172019,
      "grad_norm": 0.1840267926454544,
      "learning_rate": 2.8004768837934926e-06,
      "loss": 0.3193,
      "step": 2668
    },
    {
      "epoch": 1.682085024230724,
      "grad_norm": 0.18752466142177582,
      "learning_rate": 2.789610224623925e-06,
      "loss": 0.4,
      "step": 2669
    },
    {
      "epoch": 1.6827154170442458,
      "grad_norm": 0.22188887000083923,
      "learning_rate": 2.778763295842346e-06,
      "loss": 0.3561,
      "step": 2670
    },
    {
      "epoch": 1.6833458098577676,
      "grad_norm": 0.14395049214363098,
      "learning_rate": 2.7679361083067e-06,
      "loss": 0.3108,
      "step": 2671
    },
    {
      "epoch": 1.6839762026712894,
      "grad_norm": 0.16563533246517181,
      "learning_rate": 2.7571286728551497e-06,
      "loss": 0.3403,
      "step": 2672
    },
    {
      "epoch": 1.6846065954848115,
      "grad_norm": 0.17470870912075043,
      "learning_rate": 2.7463410003060965e-06,
      "loss": 0.3393,
      "step": 2673
    },
    {
      "epoch": 1.6852369882983333,
      "grad_norm": 0.1701815128326416,
      "learning_rate": 2.735573101458162e-06,
      "loss": 0.3538,
      "step": 2674
    },
    {
      "epoch": 1.6858673811118554,
      "grad_norm": 0.2105788141489029,
      "learning_rate": 2.724824987090172e-06,
      "loss": 0.4069,
      "step": 2675
    },
    {
      "epoch": 1.6864977739253773,
      "grad_norm": 0.1775672733783722,
      "learning_rate": 2.71409666796114e-06,
      "loss": 0.3541,
      "step": 2676
    },
    {
      "epoch": 1.687128166738899,
      "grad_norm": 0.16868631541728973,
      "learning_rate": 2.7033881548102783e-06,
      "loss": 0.3065,
      "step": 2677
    },
    {
      "epoch": 1.6877585595524212,
      "grad_norm": 0.16328738629817963,
      "learning_rate": 2.6926994583569494e-06,
      "loss": 0.3419,
      "step": 2678
    },
    {
      "epoch": 1.688388952365943,
      "grad_norm": 0.16287265717983246,
      "learning_rate": 2.6820305893007143e-06,
      "loss": 0.3393,
      "step": 2679
    },
    {
      "epoch": 1.689019345179465,
      "grad_norm": 0.19886572659015656,
      "learning_rate": 2.671381558321251e-06,
      "loss": 0.4079,
      "step": 2680
    },
    {
      "epoch": 1.689649737992987,
      "grad_norm": 0.19469165802001953,
      "learning_rate": 2.660752376078407e-06,
      "loss": 0.3149,
      "step": 2681
    },
    {
      "epoch": 1.6902801308065087,
      "grad_norm": 0.15739621222019196,
      "learning_rate": 2.650143053212138e-06,
      "loss": 0.3475,
      "step": 2682
    },
    {
      "epoch": 1.6909105236200306,
      "grad_norm": 0.13259677588939667,
      "learning_rate": 2.639553600342538e-06,
      "loss": 0.3312,
      "step": 2683
    },
    {
      "epoch": 1.6915409164335526,
      "grad_norm": 0.16224607825279236,
      "learning_rate": 2.6289840280698013e-06,
      "loss": 0.3378,
      "step": 2684
    },
    {
      "epoch": 1.6921713092470747,
      "grad_norm": 0.15360943973064423,
      "learning_rate": 2.61843434697423e-06,
      "loss": 0.2946,
      "step": 2685
    },
    {
      "epoch": 1.6928017020605965,
      "grad_norm": 0.1765449494123459,
      "learning_rate": 2.6079045676161963e-06,
      "loss": 0.3056,
      "step": 2686
    },
    {
      "epoch": 1.6934320948741184,
      "grad_norm": 0.17461608350276947,
      "learning_rate": 2.5973947005361746e-06,
      "loss": 0.3048,
      "step": 2687
    },
    {
      "epoch": 1.6940624876876402,
      "grad_norm": 0.16787251830101013,
      "learning_rate": 2.5869047562546885e-06,
      "loss": 0.3737,
      "step": 2688
    },
    {
      "epoch": 1.6946928805011623,
      "grad_norm": 0.1929512768983841,
      "learning_rate": 2.5764347452723365e-06,
      "loss": 0.3367,
      "step": 2689
    },
    {
      "epoch": 1.6953232733146844,
      "grad_norm": 0.16267262399196625,
      "learning_rate": 2.565984678069737e-06,
      "loss": 0.3386,
      "step": 2690
    },
    {
      "epoch": 1.6959536661282062,
      "grad_norm": 0.19082540273666382,
      "learning_rate": 2.5555545651075677e-06,
      "loss": 0.3568,
      "step": 2691
    },
    {
      "epoch": 1.696584058941728,
      "grad_norm": 0.17961396276950836,
      "learning_rate": 2.545144416826524e-06,
      "loss": 0.3649,
      "step": 2692
    },
    {
      "epoch": 1.6972144517552499,
      "grad_norm": 0.17428696155548096,
      "learning_rate": 2.5347542436473204e-06,
      "loss": 0.3468,
      "step": 2693
    },
    {
      "epoch": 1.697844844568772,
      "grad_norm": 0.17337742447853088,
      "learning_rate": 2.524384055970662e-06,
      "loss": 0.3284,
      "step": 2694
    },
    {
      "epoch": 1.698475237382294,
      "grad_norm": 0.20851732790470123,
      "learning_rate": 2.5140338641772705e-06,
      "loss": 0.3598,
      "step": 2695
    },
    {
      "epoch": 1.6991056301958158,
      "grad_norm": 0.16625341773033142,
      "learning_rate": 2.5037036786278202e-06,
      "loss": 0.3261,
      "step": 2696
    },
    {
      "epoch": 1.6997360230093377,
      "grad_norm": 0.22402916848659515,
      "learning_rate": 2.493393509663e-06,
      "loss": 0.4392,
      "step": 2697
    },
    {
      "epoch": 1.7003664158228595,
      "grad_norm": 0.16897734999656677,
      "learning_rate": 2.483103367603424e-06,
      "loss": 0.3401,
      "step": 2698
    },
    {
      "epoch": 1.7009968086363816,
      "grad_norm": 0.1630307286977768,
      "learning_rate": 2.472833262749685e-06,
      "loss": 0.3425,
      "step": 2699
    },
    {
      "epoch": 1.7016272014499034,
      "grad_norm": 0.19418928027153015,
      "learning_rate": 2.462583205382292e-06,
      "loss": 0.3906,
      "step": 2700
    },
    {
      "epoch": 1.7022575942634255,
      "grad_norm": 0.16761890053749084,
      "learning_rate": 2.452353205761725e-06,
      "loss": 0.3793,
      "step": 2701
    },
    {
      "epoch": 1.7028879870769473,
      "grad_norm": 0.18117816746234894,
      "learning_rate": 2.442143274128345e-06,
      "loss": 0.3681,
      "step": 2702
    },
    {
      "epoch": 1.7035183798904692,
      "grad_norm": 0.14692558348178864,
      "learning_rate": 2.4319534207024514e-06,
      "loss": 0.3868,
      "step": 2703
    },
    {
      "epoch": 1.704148772703991,
      "grad_norm": 0.19921517372131348,
      "learning_rate": 2.42178365568423e-06,
      "loss": 0.3649,
      "step": 2704
    },
    {
      "epoch": 1.704779165517513,
      "grad_norm": 0.16821995377540588,
      "learning_rate": 2.4116339892537684e-06,
      "loss": 0.3404,
      "step": 2705
    },
    {
      "epoch": 1.7054095583310351,
      "grad_norm": 0.19933214783668518,
      "learning_rate": 2.401504431571028e-06,
      "loss": 0.3387,
      "step": 2706
    },
    {
      "epoch": 1.706039951144557,
      "grad_norm": 0.18399402499198914,
      "learning_rate": 2.3913949927758514e-06,
      "loss": 0.3846,
      "step": 2707
    },
    {
      "epoch": 1.7066703439580788,
      "grad_norm": 0.15911422669887543,
      "learning_rate": 2.381305682987926e-06,
      "loss": 0.3916,
      "step": 2708
    },
    {
      "epoch": 1.7073007367716007,
      "grad_norm": 0.1464301496744156,
      "learning_rate": 2.3712365123067997e-06,
      "loss": 0.3244,
      "step": 2709
    },
    {
      "epoch": 1.7079311295851227,
      "grad_norm": 0.20365366339683533,
      "learning_rate": 2.361187490811861e-06,
      "loss": 0.3821,
      "step": 2710
    },
    {
      "epoch": 1.7085615223986448,
      "grad_norm": 0.16726025938987732,
      "learning_rate": 2.3511586285623324e-06,
      "loss": 0.3617,
      "step": 2711
    },
    {
      "epoch": 1.7091919152121666,
      "grad_norm": 0.17762507498264313,
      "learning_rate": 2.341149935597239e-06,
      "loss": 0.3814,
      "step": 2712
    },
    {
      "epoch": 1.7098223080256885,
      "grad_norm": 0.13304100930690765,
      "learning_rate": 2.331161421935439e-06,
      "loss": 0.2826,
      "step": 2713
    },
    {
      "epoch": 1.7104527008392103,
      "grad_norm": 0.1622227430343628,
      "learning_rate": 2.321193097575577e-06,
      "loss": 0.3359,
      "step": 2714
    },
    {
      "epoch": 1.7110830936527324,
      "grad_norm": 0.1425076723098755,
      "learning_rate": 2.311244972496093e-06,
      "loss": 0.3394,
      "step": 2715
    },
    {
      "epoch": 1.7117134864662544,
      "grad_norm": 0.16510890424251556,
      "learning_rate": 2.301317056655202e-06,
      "loss": 0.2648,
      "step": 2716
    },
    {
      "epoch": 1.7123438792797763,
      "grad_norm": 0.1962651163339615,
      "learning_rate": 2.2914093599908985e-06,
      "loss": 0.3774,
      "step": 2717
    },
    {
      "epoch": 1.7129742720932981,
      "grad_norm": 0.18967847526073456,
      "learning_rate": 2.2815218924209198e-06,
      "loss": 0.3603,
      "step": 2718
    },
    {
      "epoch": 1.71360466490682,
      "grad_norm": 0.16082076728343964,
      "learning_rate": 2.271654663842781e-06,
      "loss": 0.3075,
      "step": 2719
    },
    {
      "epoch": 1.714235057720342,
      "grad_norm": 0.15744170546531677,
      "learning_rate": 2.261807684133709e-06,
      "loss": 0.3584,
      "step": 2720
    },
    {
      "epoch": 1.714865450533864,
      "grad_norm": 0.215196430683136,
      "learning_rate": 2.2519809631506858e-06,
      "loss": 0.3892,
      "step": 2721
    },
    {
      "epoch": 1.715495843347386,
      "grad_norm": 0.16610518097877502,
      "learning_rate": 2.242174510730391e-06,
      "loss": 0.3303,
      "step": 2722
    },
    {
      "epoch": 1.7161262361609078,
      "grad_norm": 0.1572638601064682,
      "learning_rate": 2.232388336689232e-06,
      "loss": 0.362,
      "step": 2723
    },
    {
      "epoch": 1.7167566289744296,
      "grad_norm": 0.17560486495494843,
      "learning_rate": 2.2226224508233105e-06,
      "loss": 0.3483,
      "step": 2724
    },
    {
      "epoch": 1.7173870217879517,
      "grad_norm": 0.18823911249637604,
      "learning_rate": 2.2128768629084286e-06,
      "loss": 0.3773,
      "step": 2725
    },
    {
      "epoch": 1.7180174146014735,
      "grad_norm": 0.12789592146873474,
      "learning_rate": 2.203151582700051e-06,
      "loss": 0.3035,
      "step": 2726
    },
    {
      "epoch": 1.7186478074149956,
      "grad_norm": 0.19134822487831116,
      "learning_rate": 2.1934466199333298e-06,
      "loss": 0.391,
      "step": 2727
    },
    {
      "epoch": 1.7192782002285174,
      "grad_norm": 0.14222867786884308,
      "learning_rate": 2.1837619843230744e-06,
      "loss": 0.2925,
      "step": 2728
    },
    {
      "epoch": 1.7199085930420392,
      "grad_norm": 0.1417618989944458,
      "learning_rate": 2.1740976855637517e-06,
      "loss": 0.3337,
      "step": 2729
    },
    {
      "epoch": 1.720538985855561,
      "grad_norm": 0.18344461917877197,
      "learning_rate": 2.1644537333294543e-06,
      "loss": 0.3417,
      "step": 2730
    },
    {
      "epoch": 1.7211693786690831,
      "grad_norm": 0.16136455535888672,
      "learning_rate": 2.1548301372739292e-06,
      "loss": 0.3542,
      "step": 2731
    },
    {
      "epoch": 1.7217997714826052,
      "grad_norm": 0.18665267527103424,
      "learning_rate": 2.145226907030531e-06,
      "loss": 0.4172,
      "step": 2732
    },
    {
      "epoch": 1.722430164296127,
      "grad_norm": 0.18608105182647705,
      "learning_rate": 2.1356440522122403e-06,
      "loss": 0.3145,
      "step": 2733
    },
    {
      "epoch": 1.723060557109649,
      "grad_norm": 0.1521705538034439,
      "learning_rate": 2.1260815824116256e-06,
      "loss": 0.2793,
      "step": 2734
    },
    {
      "epoch": 1.7236909499231707,
      "grad_norm": 0.21205557882785797,
      "learning_rate": 2.1165395072008627e-06,
      "loss": 0.3642,
      "step": 2735
    },
    {
      "epoch": 1.7243213427366928,
      "grad_norm": 0.1696559190750122,
      "learning_rate": 2.1070178361317113e-06,
      "loss": 0.3955,
      "step": 2736
    },
    {
      "epoch": 1.7249517355502149,
      "grad_norm": 0.1602529138326645,
      "learning_rate": 2.0975165787355016e-06,
      "loss": 0.3284,
      "step": 2737
    },
    {
      "epoch": 1.7255821283637367,
      "grad_norm": 0.19696998596191406,
      "learning_rate": 2.0880357445231265e-06,
      "loss": 0.4423,
      "step": 2738
    },
    {
      "epoch": 1.7262125211772585,
      "grad_norm": 0.14066126942634583,
      "learning_rate": 2.0785753429850485e-06,
      "loss": 0.3609,
      "step": 2739
    },
    {
      "epoch": 1.7268429139907804,
      "grad_norm": 0.18645574152469635,
      "learning_rate": 2.069135383591253e-06,
      "loss": 0.3536,
      "step": 2740
    },
    {
      "epoch": 1.7274733068043024,
      "grad_norm": 0.1929475963115692,
      "learning_rate": 2.059715875791296e-06,
      "loss": 0.4631,
      "step": 2741
    },
    {
      "epoch": 1.7281036996178245,
      "grad_norm": 0.16178543865680695,
      "learning_rate": 2.0503168290142277e-06,
      "loss": 0.3382,
      "step": 2742
    },
    {
      "epoch": 1.7287340924313463,
      "grad_norm": 0.1649865061044693,
      "learning_rate": 2.040938252668639e-06,
      "loss": 0.3416,
      "step": 2743
    },
    {
      "epoch": 1.7293644852448682,
      "grad_norm": 0.1723279356956482,
      "learning_rate": 2.031580156142618e-06,
      "loss": 0.3325,
      "step": 2744
    },
    {
      "epoch": 1.72999487805839,
      "grad_norm": 0.20794516801834106,
      "learning_rate": 2.022242548803754e-06,
      "loss": 0.414,
      "step": 2745
    },
    {
      "epoch": 1.730625270871912,
      "grad_norm": 0.17778415977954865,
      "learning_rate": 2.0129254399991324e-06,
      "loss": 0.3701,
      "step": 2746
    },
    {
      "epoch": 1.731255663685434,
      "grad_norm": 0.14724238216876984,
      "learning_rate": 2.003628839055318e-06,
      "loss": 0.2701,
      "step": 2747
    },
    {
      "epoch": 1.731886056498956,
      "grad_norm": 0.2008165717124939,
      "learning_rate": 1.9943527552783365e-06,
      "loss": 0.3588,
      "step": 2748
    },
    {
      "epoch": 1.7325164493124778,
      "grad_norm": 0.15005536377429962,
      "learning_rate": 1.985097197953685e-06,
      "loss": 0.397,
      "step": 2749
    },
    {
      "epoch": 1.7331468421259997,
      "grad_norm": 0.1658148318529129,
      "learning_rate": 1.975862176346316e-06,
      "loss": 0.3324,
      "step": 2750
    },
    {
      "epoch": 1.7337772349395217,
      "grad_norm": 0.16382184624671936,
      "learning_rate": 1.966647699700621e-06,
      "loss": 0.3103,
      "step": 2751
    },
    {
      "epoch": 1.7344076277530436,
      "grad_norm": 0.1729460507631302,
      "learning_rate": 1.9574537772404195e-06,
      "loss": 0.3479,
      "step": 2752
    },
    {
      "epoch": 1.7350380205665656,
      "grad_norm": 0.17231756448745728,
      "learning_rate": 1.9482804181689674e-06,
      "loss": 0.3712,
      "step": 2753
    },
    {
      "epoch": 1.7356684133800875,
      "grad_norm": 0.16201011836528778,
      "learning_rate": 1.9391276316689267e-06,
      "loss": 0.3607,
      "step": 2754
    },
    {
      "epoch": 1.7362988061936093,
      "grad_norm": 0.15200699865818024,
      "learning_rate": 1.9299954269023784e-06,
      "loss": 0.3093,
      "step": 2755
    },
    {
      "epoch": 1.7369291990071312,
      "grad_norm": 0.1864054799079895,
      "learning_rate": 1.920883813010783e-06,
      "loss": 0.3405,
      "step": 2756
    },
    {
      "epoch": 1.7375595918206532,
      "grad_norm": 0.17108328640460968,
      "learning_rate": 1.9117927991150066e-06,
      "loss": 0.345,
      "step": 2757
    },
    {
      "epoch": 1.7381899846341753,
      "grad_norm": 0.1786588728427887,
      "learning_rate": 1.9027223943152772e-06,
      "loss": 0.3149,
      "step": 2758
    },
    {
      "epoch": 1.7388203774476971,
      "grad_norm": 0.17140208184719086,
      "learning_rate": 1.8936726076912172e-06,
      "loss": 0.308,
      "step": 2759
    },
    {
      "epoch": 1.739450770261219,
      "grad_norm": 0.18755091726779938,
      "learning_rate": 1.8846434483017828e-06,
      "loss": 0.3674,
      "step": 2760
    },
    {
      "epoch": 1.7400811630747408,
      "grad_norm": 0.1827147752046585,
      "learning_rate": 1.8756349251853012e-06,
      "loss": 0.3668,
      "step": 2761
    },
    {
      "epoch": 1.7407115558882629,
      "grad_norm": 0.1812572330236435,
      "learning_rate": 1.8666470473594316e-06,
      "loss": 0.3738,
      "step": 2762
    },
    {
      "epoch": 1.741341948701785,
      "grad_norm": 0.18684326112270355,
      "learning_rate": 1.8576798238211698e-06,
      "loss": 0.3566,
      "step": 2763
    },
    {
      "epoch": 1.7419723415153068,
      "grad_norm": 0.1563086211681366,
      "learning_rate": 1.848733263546843e-06,
      "loss": 0.3371,
      "step": 2764
    },
    {
      "epoch": 1.7426027343288286,
      "grad_norm": 0.23187503218650818,
      "learning_rate": 1.8398073754920873e-06,
      "loss": 0.3441,
      "step": 2765
    },
    {
      "epoch": 1.7432331271423505,
      "grad_norm": 0.16062569618225098,
      "learning_rate": 1.8309021685918428e-06,
      "loss": 0.4071,
      "step": 2766
    },
    {
      "epoch": 1.7438635199558725,
      "grad_norm": 0.16154485940933228,
      "learning_rate": 1.8220176517603519e-06,
      "loss": 0.3936,
      "step": 2767
    },
    {
      "epoch": 1.7444939127693946,
      "grad_norm": 0.17705531418323517,
      "learning_rate": 1.81315383389115e-06,
      "loss": 0.3389,
      "step": 2768
    },
    {
      "epoch": 1.7451243055829164,
      "grad_norm": 0.16158095002174377,
      "learning_rate": 1.8043107238570498e-06,
      "loss": 0.3602,
      "step": 2769
    },
    {
      "epoch": 1.7457546983964383,
      "grad_norm": 0.18866123259067535,
      "learning_rate": 1.7954883305101277e-06,
      "loss": 0.3823,
      "step": 2770
    },
    {
      "epoch": 1.74638509120996,
      "grad_norm": 0.14326314628124237,
      "learning_rate": 1.7866866626817297e-06,
      "loss": 0.3356,
      "step": 2771
    },
    {
      "epoch": 1.7470154840234822,
      "grad_norm": 0.1846877783536911,
      "learning_rate": 1.7779057291824527e-06,
      "loss": 0.3421,
      "step": 2772
    },
    {
      "epoch": 1.747645876837004,
      "grad_norm": 0.15310382843017578,
      "learning_rate": 1.7691455388021484e-06,
      "loss": 0.2835,
      "step": 2773
    },
    {
      "epoch": 1.748276269650526,
      "grad_norm": 0.14761358499526978,
      "learning_rate": 1.7604061003098817e-06,
      "loss": 0.3675,
      "step": 2774
    },
    {
      "epoch": 1.748906662464048,
      "grad_norm": 0.19517937302589417,
      "learning_rate": 1.751687422453968e-06,
      "loss": 0.3041,
      "step": 2775
    },
    {
      "epoch": 1.7495370552775698,
      "grad_norm": 0.1950213611125946,
      "learning_rate": 1.7429895139619252e-06,
      "loss": 0.3819,
      "step": 2776
    },
    {
      "epoch": 1.7501674480910918,
      "grad_norm": 0.15891462564468384,
      "learning_rate": 1.734312383540494e-06,
      "loss": 0.3552,
      "step": 2777
    },
    {
      "epoch": 1.7507978409046137,
      "grad_norm": 0.16390115022659302,
      "learning_rate": 1.7256560398756014e-06,
      "loss": 0.4012,
      "step": 2778
    },
    {
      "epoch": 1.7514282337181357,
      "grad_norm": 0.1448567807674408,
      "learning_rate": 1.7170204916323784e-06,
      "loss": 0.3193,
      "step": 2779
    },
    {
      "epoch": 1.7520586265316576,
      "grad_norm": 0.1577872782945633,
      "learning_rate": 1.7084057474551254e-06,
      "loss": 0.3678,
      "step": 2780
    },
    {
      "epoch": 1.7526890193451794,
      "grad_norm": 0.161997988820076,
      "learning_rate": 1.6998118159673442e-06,
      "loss": 0.3477,
      "step": 2781
    },
    {
      "epoch": 1.7533194121587012,
      "grad_norm": 0.176879420876503,
      "learning_rate": 1.691238705771671e-06,
      "loss": 0.331,
      "step": 2782
    },
    {
      "epoch": 1.7539498049722233,
      "grad_norm": 0.16836316883563995,
      "learning_rate": 1.6826864254499236e-06,
      "loss": 0.3691,
      "step": 2783
    },
    {
      "epoch": 1.7545801977857454,
      "grad_norm": 0.20967160165309906,
      "learning_rate": 1.6741549835630538e-06,
      "loss": 0.3648,
      "step": 2784
    },
    {
      "epoch": 1.7552105905992672,
      "grad_norm": 0.13090410828590393,
      "learning_rate": 1.6656443886511577e-06,
      "loss": 0.2312,
      "step": 2785
    },
    {
      "epoch": 1.755840983412789,
      "grad_norm": 0.15169045329093933,
      "learning_rate": 1.6571546492334708e-06,
      "loss": 0.2898,
      "step": 2786
    },
    {
      "epoch": 1.7564713762263109,
      "grad_norm": 0.1762060821056366,
      "learning_rate": 1.6486857738083474e-06,
      "loss": 0.3035,
      "step": 2787
    },
    {
      "epoch": 1.757101769039833,
      "grad_norm": 0.17813783884048462,
      "learning_rate": 1.6402377708532466e-06,
      "loss": 0.3459,
      "step": 2788
    },
    {
      "epoch": 1.757732161853355,
      "grad_norm": 0.12713901698589325,
      "learning_rate": 1.631810648824751e-06,
      "loss": 0.2806,
      "step": 2789
    },
    {
      "epoch": 1.7583625546668769,
      "grad_norm": 0.16314174234867096,
      "learning_rate": 1.6234044161585276e-06,
      "loss": 0.3242,
      "step": 2790
    },
    {
      "epoch": 1.7589929474803987,
      "grad_norm": 0.18997551500797272,
      "learning_rate": 1.6150190812693458e-06,
      "loss": 0.4186,
      "step": 2791
    },
    {
      "epoch": 1.7596233402939205,
      "grad_norm": 0.14270219206809998,
      "learning_rate": 1.6066546525510384e-06,
      "loss": 0.2792,
      "step": 2792
    },
    {
      "epoch": 1.7602537331074426,
      "grad_norm": 0.18702076375484467,
      "learning_rate": 1.5983111383765254e-06,
      "loss": 0.3878,
      "step": 2793
    },
    {
      "epoch": 1.7608841259209647,
      "grad_norm": 0.17163795232772827,
      "learning_rate": 1.5899885470977858e-06,
      "loss": 0.3613,
      "step": 2794
    },
    {
      "epoch": 1.7615145187344865,
      "grad_norm": 0.14220699667930603,
      "learning_rate": 1.581686887045858e-06,
      "loss": 0.3274,
      "step": 2795
    },
    {
      "epoch": 1.7621449115480083,
      "grad_norm": 0.1297381967306137,
      "learning_rate": 1.5734061665308194e-06,
      "loss": 0.3298,
      "step": 2796
    },
    {
      "epoch": 1.7627753043615302,
      "grad_norm": 0.14467160403728485,
      "learning_rate": 1.5651463938417948e-06,
      "loss": 0.3203,
      "step": 2797
    },
    {
      "epoch": 1.7634056971750522,
      "grad_norm": 0.1549520492553711,
      "learning_rate": 1.5569075772469378e-06,
      "loss": 0.3047,
      "step": 2798
    },
    {
      "epoch": 1.764036089988574,
      "grad_norm": 0.1654859334230423,
      "learning_rate": 1.5486897249934285e-06,
      "loss": 0.3584,
      "step": 2799
    },
    {
      "epoch": 1.7646664828020961,
      "grad_norm": 0.18609672784805298,
      "learning_rate": 1.5404928453074468e-06,
      "loss": 0.3791,
      "step": 2800
    },
    {
      "epoch": 1.7646664828020961,
      "eval_loss": 0.41850462555885315,
      "eval_runtime": 222.3838,
      "eval_samples_per_second": 4.497,
      "eval_steps_per_second": 4.497,
      "step": 2800
    },
    {
      "epoch": 1.765296875615618,
      "grad_norm": 0.2150953710079193,
      "learning_rate": 1.5323169463941994e-06,
      "loss": 0.4229,
      "step": 2801
    },
    {
      "epoch": 1.7659272684291398,
      "grad_norm": 0.18508629500865936,
      "learning_rate": 1.5241620364378695e-06,
      "loss": 0.3557,
      "step": 2802
    },
    {
      "epoch": 1.7665576612426617,
      "grad_norm": 0.15115678310394287,
      "learning_rate": 1.5160281236016522e-06,
      "loss": 0.3079,
      "step": 2803
    },
    {
      "epoch": 1.7671880540561837,
      "grad_norm": 0.2158491164445877,
      "learning_rate": 1.5079152160277045e-06,
      "loss": 0.3583,
      "step": 2804
    },
    {
      "epoch": 1.7678184468697058,
      "grad_norm": 0.17843793332576752,
      "learning_rate": 1.4998233218371724e-06,
      "loss": 0.4217,
      "step": 2805
    },
    {
      "epoch": 1.7684488396832276,
      "grad_norm": 0.1688946932554245,
      "learning_rate": 1.4917524491301543e-06,
      "loss": 0.3485,
      "step": 2806
    },
    {
      "epoch": 1.7690792324967495,
      "grad_norm": 0.1464250385761261,
      "learning_rate": 1.4837026059857127e-06,
      "loss": 0.3557,
      "step": 2807
    },
    {
      "epoch": 1.7697096253102713,
      "grad_norm": 0.16816261410713196,
      "learning_rate": 1.4756738004618614e-06,
      "loss": 0.2934,
      "step": 2808
    },
    {
      "epoch": 1.7703400181237934,
      "grad_norm": 0.15366758406162262,
      "learning_rate": 1.467666040595552e-06,
      "loss": 0.3111,
      "step": 2809
    },
    {
      "epoch": 1.7709704109373154,
      "grad_norm": 0.15366709232330322,
      "learning_rate": 1.459679334402662e-06,
      "loss": 0.336,
      "step": 2810
    },
    {
      "epoch": 1.7716008037508373,
      "grad_norm": 0.18162818253040314,
      "learning_rate": 1.4517136898780068e-06,
      "loss": 0.3466,
      "step": 2811
    },
    {
      "epoch": 1.7722311965643591,
      "grad_norm": 0.16830721497535706,
      "learning_rate": 1.4437691149953124e-06,
      "loss": 0.313,
      "step": 2812
    },
    {
      "epoch": 1.772861589377881,
      "grad_norm": 0.20500795543193817,
      "learning_rate": 1.435845617707218e-06,
      "loss": 0.4075,
      "step": 2813
    },
    {
      "epoch": 1.773491982191403,
      "grad_norm": 0.16557179391384125,
      "learning_rate": 1.4279432059452511e-06,
      "loss": 0.3837,
      "step": 2814
    },
    {
      "epoch": 1.774122375004925,
      "grad_norm": 0.20282112061977386,
      "learning_rate": 1.4200618876198465e-06,
      "loss": 0.3548,
      "step": 2815
    },
    {
      "epoch": 1.774752767818447,
      "grad_norm": 0.17897149920463562,
      "learning_rate": 1.4122016706203188e-06,
      "loss": 0.4087,
      "step": 2816
    },
    {
      "epoch": 1.7753831606319688,
      "grad_norm": 0.16336628794670105,
      "learning_rate": 1.4043625628148635e-06,
      "loss": 0.3451,
      "step": 2817
    },
    {
      "epoch": 1.7760135534454906,
      "grad_norm": 0.1643591672182083,
      "learning_rate": 1.3965445720505365e-06,
      "loss": 0.3211,
      "step": 2818
    },
    {
      "epoch": 1.7766439462590127,
      "grad_norm": 0.17228056490421295,
      "learning_rate": 1.3887477061532628e-06,
      "loss": 0.2829,
      "step": 2819
    },
    {
      "epoch": 1.7772743390725347,
      "grad_norm": 0.14726178348064423,
      "learning_rate": 1.380971972927814e-06,
      "loss": 0.3419,
      "step": 2820
    },
    {
      "epoch": 1.7779047318860566,
      "grad_norm": 0.16053885221481323,
      "learning_rate": 1.3732173801578213e-06,
      "loss": 0.3403,
      "step": 2821
    },
    {
      "epoch": 1.7785351246995784,
      "grad_norm": 0.15379776060581207,
      "learning_rate": 1.3654839356057405e-06,
      "loss": 0.3309,
      "step": 2822
    },
    {
      "epoch": 1.7791655175131003,
      "grad_norm": 0.14749456942081451,
      "learning_rate": 1.3577716470128668e-06,
      "loss": 0.3119,
      "step": 2823
    },
    {
      "epoch": 1.7797959103266223,
      "grad_norm": 0.17150136828422546,
      "learning_rate": 1.3500805220993045e-06,
      "loss": 0.3999,
      "step": 2824
    },
    {
      "epoch": 1.7804263031401442,
      "grad_norm": 0.20718194544315338,
      "learning_rate": 1.3424105685639904e-06,
      "loss": 0.3495,
      "step": 2825
    },
    {
      "epoch": 1.7810566959536662,
      "grad_norm": 0.18642765283584595,
      "learning_rate": 1.3347617940846552e-06,
      "loss": 0.3682,
      "step": 2826
    },
    {
      "epoch": 1.781687088767188,
      "grad_norm": 0.18365077674388885,
      "learning_rate": 1.3271342063178393e-06,
      "loss": 0.3234,
      "step": 2827
    },
    {
      "epoch": 1.78231748158071,
      "grad_norm": 0.15567843616008759,
      "learning_rate": 1.3195278128988653e-06,
      "loss": 0.3824,
      "step": 2828
    },
    {
      "epoch": 1.7829478743942317,
      "grad_norm": 0.18231748044490814,
      "learning_rate": 1.3119426214418423e-06,
      "loss": 0.3105,
      "step": 2829
    },
    {
      "epoch": 1.7835782672077538,
      "grad_norm": 0.21656112372875214,
      "learning_rate": 1.3043786395396617e-06,
      "loss": 0.4615,
      "step": 2830
    },
    {
      "epoch": 1.7842086600212759,
      "grad_norm": 0.16485925018787384,
      "learning_rate": 1.2968358747639794e-06,
      "loss": 0.2991,
      "step": 2831
    },
    {
      "epoch": 1.7848390528347977,
      "grad_norm": 0.17655867338180542,
      "learning_rate": 1.2893143346652105e-06,
      "loss": 0.3318,
      "step": 2832
    },
    {
      "epoch": 1.7854694456483196,
      "grad_norm": 0.14479409158229828,
      "learning_rate": 1.2818140267725275e-06,
      "loss": 0.3241,
      "step": 2833
    },
    {
      "epoch": 1.7860998384618414,
      "grad_norm": 0.18551374971866608,
      "learning_rate": 1.2743349585938497e-06,
      "loss": 0.3531,
      "step": 2834
    },
    {
      "epoch": 1.7867302312753635,
      "grad_norm": 0.15372885763645172,
      "learning_rate": 1.2668771376158356e-06,
      "loss": 0.3643,
      "step": 2835
    },
    {
      "epoch": 1.7873606240888855,
      "grad_norm": 0.17137745022773743,
      "learning_rate": 1.2594405713038661e-06,
      "loss": 0.3494,
      "step": 2836
    },
    {
      "epoch": 1.7879910169024074,
      "grad_norm": 0.17164146900177002,
      "learning_rate": 1.2520252671020568e-06,
      "loss": 0.2951,
      "step": 2837
    },
    {
      "epoch": 1.7886214097159292,
      "grad_norm": 0.17525000870227814,
      "learning_rate": 1.2446312324332348e-06,
      "loss": 0.3379,
      "step": 2838
    },
    {
      "epoch": 1.789251802529451,
      "grad_norm": 0.17154620587825775,
      "learning_rate": 1.2372584746989397e-06,
      "loss": 0.3236,
      "step": 2839
    },
    {
      "epoch": 1.789882195342973,
      "grad_norm": 0.17524504661560059,
      "learning_rate": 1.229907001279405e-06,
      "loss": 0.386,
      "step": 2840
    },
    {
      "epoch": 1.7905125881564952,
      "grad_norm": 0.17549632489681244,
      "learning_rate": 1.222576819533567e-06,
      "loss": 0.381,
      "step": 2841
    },
    {
      "epoch": 1.791142980970017,
      "grad_norm": 0.1916823834180832,
      "learning_rate": 1.2152679367990364e-06,
      "loss": 0.3531,
      "step": 2842
    },
    {
      "epoch": 1.7917733737835388,
      "grad_norm": 0.21655133366584778,
      "learning_rate": 1.2079803603921253e-06,
      "loss": 0.3728,
      "step": 2843
    },
    {
      "epoch": 1.7924037665970607,
      "grad_norm": 0.16216030716896057,
      "learning_rate": 1.200714097607791e-06,
      "loss": 0.3782,
      "step": 2844
    },
    {
      "epoch": 1.7930341594105828,
      "grad_norm": 0.17528338730335236,
      "learning_rate": 1.1934691557196804e-06,
      "loss": 0.3318,
      "step": 2845
    },
    {
      "epoch": 1.7936645522241046,
      "grad_norm": 0.20596998929977417,
      "learning_rate": 1.1862455419800744e-06,
      "loss": 0.3781,
      "step": 2846
    },
    {
      "epoch": 1.7942949450376267,
      "grad_norm": 0.17023244500160217,
      "learning_rate": 1.1790432636199241e-06,
      "loss": 0.4052,
      "step": 2847
    },
    {
      "epoch": 1.7949253378511485,
      "grad_norm": 0.14573776721954346,
      "learning_rate": 1.17186232784881e-06,
      "loss": 0.2995,
      "step": 2848
    },
    {
      "epoch": 1.7955557306646703,
      "grad_norm": 0.178813636302948,
      "learning_rate": 1.1647027418549594e-06,
      "loss": 0.3702,
      "step": 2849
    },
    {
      "epoch": 1.7961861234781924,
      "grad_norm": 0.17194363474845886,
      "learning_rate": 1.1575645128052195e-06,
      "loss": 0.3203,
      "step": 2850
    },
    {
      "epoch": 1.7968165162917142,
      "grad_norm": 0.16584280133247375,
      "learning_rate": 1.1504476478450595e-06,
      "loss": 0.321,
      "step": 2851
    },
    {
      "epoch": 1.7974469091052363,
      "grad_norm": 0.1823454350233078,
      "learning_rate": 1.1433521540985658e-06,
      "loss": 0.3391,
      "step": 2852
    },
    {
      "epoch": 1.7980773019187581,
      "grad_norm": 0.17875845730304718,
      "learning_rate": 1.1362780386684369e-06,
      "loss": 0.3315,
      "step": 2853
    },
    {
      "epoch": 1.79870769473228,
      "grad_norm": 0.18415741622447968,
      "learning_rate": 1.1292253086359606e-06,
      "loss": 0.3427,
      "step": 2854
    },
    {
      "epoch": 1.7993380875458018,
      "grad_norm": 0.14331680536270142,
      "learning_rate": 1.1221939710610197e-06,
      "loss": 0.3506,
      "step": 2855
    },
    {
      "epoch": 1.7999684803593239,
      "grad_norm": 0.16605164110660553,
      "learning_rate": 1.1151840329820915e-06,
      "loss": 0.3184,
      "step": 2856
    },
    {
      "epoch": 1.800598873172846,
      "grad_norm": 0.16677996516227722,
      "learning_rate": 1.1081955014162279e-06,
      "loss": 0.287,
      "step": 2857
    },
    {
      "epoch": 1.8012292659863678,
      "grad_norm": 0.17730239033699036,
      "learning_rate": 1.1012283833590457e-06,
      "loss": 0.3395,
      "step": 2858
    },
    {
      "epoch": 1.8018596587998896,
      "grad_norm": 0.20990370213985443,
      "learning_rate": 1.0942826857847358e-06,
      "loss": 0.3315,
      "step": 2859
    },
    {
      "epoch": 1.8024900516134115,
      "grad_norm": 0.19944962859153748,
      "learning_rate": 1.0873584156460343e-06,
      "loss": 0.3541,
      "step": 2860
    },
    {
      "epoch": 1.8031204444269335,
      "grad_norm": 0.19780515134334564,
      "learning_rate": 1.0804555798742492e-06,
      "loss": 0.432,
      "step": 2861
    },
    {
      "epoch": 1.8037508372404556,
      "grad_norm": 0.1663893759250641,
      "learning_rate": 1.0735741853792105e-06,
      "loss": 0.3963,
      "step": 2862
    },
    {
      "epoch": 1.8043812300539774,
      "grad_norm": 0.15331469476222992,
      "learning_rate": 1.0667142390493034e-06,
      "loss": 0.3588,
      "step": 2863
    },
    {
      "epoch": 1.8050116228674993,
      "grad_norm": 0.16489370167255402,
      "learning_rate": 1.0598757477514224e-06,
      "loss": 0.3506,
      "step": 2864
    },
    {
      "epoch": 1.8056420156810211,
      "grad_norm": 0.2095303237438202,
      "learning_rate": 1.053058718331009e-06,
      "loss": 0.3441,
      "step": 2865
    },
    {
      "epoch": 1.8062724084945432,
      "grad_norm": 0.15023407340049744,
      "learning_rate": 1.0462631576119996e-06,
      "loss": 0.3304,
      "step": 2866
    },
    {
      "epoch": 1.8069028013080652,
      "grad_norm": 0.15889757871627808,
      "learning_rate": 1.0394890723968607e-06,
      "loss": 0.3887,
      "step": 2867
    },
    {
      "epoch": 1.807533194121587,
      "grad_norm": 0.16765765845775604,
      "learning_rate": 1.0327364694665376e-06,
      "loss": 0.3276,
      "step": 2868
    },
    {
      "epoch": 1.808163586935109,
      "grad_norm": 0.14473390579223633,
      "learning_rate": 1.026005355580491e-06,
      "loss": 0.3083,
      "step": 2869
    },
    {
      "epoch": 1.8087939797486308,
      "grad_norm": 0.20213527977466583,
      "learning_rate": 1.0192957374766663e-06,
      "loss": 0.3985,
      "step": 2870
    },
    {
      "epoch": 1.8094243725621528,
      "grad_norm": 0.14654994010925293,
      "learning_rate": 1.0126076218714881e-06,
      "loss": 0.3177,
      "step": 2871
    },
    {
      "epoch": 1.8100547653756747,
      "grad_norm": 0.17342987656593323,
      "learning_rate": 1.0059410154598538e-06,
      "loss": 0.3344,
      "step": 2872
    },
    {
      "epoch": 1.8106851581891967,
      "grad_norm": 0.1641921103000641,
      "learning_rate": 9.992959249151373e-07,
      "loss": 0.3203,
      "step": 2873
    },
    {
      "epoch": 1.8113155510027186,
      "grad_norm": 0.14185187220573425,
      "learning_rate": 9.926723568891683e-07,
      "loss": 0.3284,
      "step": 2874
    },
    {
      "epoch": 1.8119459438162404,
      "grad_norm": 0.1535300761461258,
      "learning_rate": 9.860703180122427e-07,
      "loss": 0.3436,
      "step": 2875
    },
    {
      "epoch": 1.8125763366297625,
      "grad_norm": 0.16254208981990814,
      "learning_rate": 9.794898148930893e-07,
      "loss": 0.3308,
      "step": 2876
    },
    {
      "epoch": 1.8132067294432843,
      "grad_norm": 0.17688316106796265,
      "learning_rate": 9.729308541188886e-07,
      "loss": 0.3402,
      "step": 2877
    },
    {
      "epoch": 1.8138371222568064,
      "grad_norm": 0.1664654016494751,
      "learning_rate": 9.663934422552615e-07,
      "loss": 0.3684,
      "step": 2878
    },
    {
      "epoch": 1.8144675150703282,
      "grad_norm": 0.17551520466804504,
      "learning_rate": 9.598775858462505e-07,
      "loss": 0.3491,
      "step": 2879
    },
    {
      "epoch": 1.81509790788385,
      "grad_norm": 0.19198037683963776,
      "learning_rate": 9.53383291414321e-07,
      "loss": 0.392,
      "step": 2880
    },
    {
      "epoch": 1.815728300697372,
      "grad_norm": 0.1434807926416397,
      "learning_rate": 9.4691056546036e-07,
      "loss": 0.33,
      "step": 2881
    },
    {
      "epoch": 1.816358693510894,
      "grad_norm": 0.1824355274438858,
      "learning_rate": 9.404594144636523e-07,
      "loss": 0.434,
      "step": 2882
    },
    {
      "epoch": 1.816989086324416,
      "grad_norm": 0.18050681054592133,
      "learning_rate": 9.340298448819046e-07,
      "loss": 0.3432,
      "step": 2883
    },
    {
      "epoch": 1.8176194791379379,
      "grad_norm": 0.16399838030338287,
      "learning_rate": 9.276218631512063e-07,
      "loss": 0.3352,
      "step": 2884
    },
    {
      "epoch": 1.8182498719514597,
      "grad_norm": 0.1734437495470047,
      "learning_rate": 9.212354756860411e-07,
      "loss": 0.3327,
      "step": 2885
    },
    {
      "epoch": 1.8188802647649815,
      "grad_norm": 0.14438305795192719,
      "learning_rate": 9.14870688879273e-07,
      "loss": 0.3377,
      "step": 2886
    },
    {
      "epoch": 1.8195106575785036,
      "grad_norm": 0.14935597777366638,
      "learning_rate": 9.085275091021478e-07,
      "loss": 0.3107,
      "step": 2887
    },
    {
      "epoch": 1.8201410503920257,
      "grad_norm": 0.18067923188209534,
      "learning_rate": 9.022059427042814e-07,
      "loss": 0.333,
      "step": 2888
    },
    {
      "epoch": 1.8207714432055475,
      "grad_norm": 0.1463068276643753,
      "learning_rate": 8.959059960136596e-07,
      "loss": 0.3535,
      "step": 2889
    },
    {
      "epoch": 1.8214018360190694,
      "grad_norm": 0.18298909068107605,
      "learning_rate": 8.896276753366126e-07,
      "loss": 0.3211,
      "step": 2890
    },
    {
      "epoch": 1.8220322288325912,
      "grad_norm": 0.184007465839386,
      "learning_rate": 8.833709869578333e-07,
      "loss": 0.3909,
      "step": 2891
    },
    {
      "epoch": 1.8226626216461133,
      "grad_norm": 0.14271940290927887,
      "learning_rate": 8.771359371403592e-07,
      "loss": 0.3538,
      "step": 2892
    },
    {
      "epoch": 1.8232930144596353,
      "grad_norm": 0.18465162813663483,
      "learning_rate": 8.709225321255704e-07,
      "loss": 0.3873,
      "step": 2893
    },
    {
      "epoch": 1.8239234072731572,
      "grad_norm": 0.14754216372966766,
      "learning_rate": 8.64730778133169e-07,
      "loss": 0.3445,
      "step": 2894
    },
    {
      "epoch": 1.824553800086679,
      "grad_norm": 0.15001064538955688,
      "learning_rate": 8.585606813611947e-07,
      "loss": 0.3217,
      "step": 2895
    },
    {
      "epoch": 1.8251841929002008,
      "grad_norm": 0.19190865755081177,
      "learning_rate": 8.524122479860042e-07,
      "loss": 0.365,
      "step": 2896
    },
    {
      "epoch": 1.825814585713723,
      "grad_norm": 0.1557355672121048,
      "learning_rate": 8.462854841622719e-07,
      "loss": 0.3154,
      "step": 2897
    },
    {
      "epoch": 1.8264449785272447,
      "grad_norm": 0.16147379577159882,
      "learning_rate": 8.401803960229696e-07,
      "loss": 0.3415,
      "step": 2898
    },
    {
      "epoch": 1.8270753713407668,
      "grad_norm": 0.1741449236869812,
      "learning_rate": 8.340969896793885e-07,
      "loss": 0.3075,
      "step": 2899
    },
    {
      "epoch": 1.8277057641542886,
      "grad_norm": 0.22572091221809387,
      "learning_rate": 8.280352712211e-07,
      "loss": 0.3748,
      "step": 2900
    },
    {
      "epoch": 1.8283361569678105,
      "grad_norm": 0.18792511522769928,
      "learning_rate": 8.21995246715983e-07,
      "loss": 0.3281,
      "step": 2901
    },
    {
      "epoch": 1.8289665497813323,
      "grad_norm": 0.19930507242679596,
      "learning_rate": 8.159769222101785e-07,
      "loss": 0.3866,
      "step": 2902
    },
    {
      "epoch": 1.8295969425948544,
      "grad_norm": 0.15603318810462952,
      "learning_rate": 8.099803037281248e-07,
      "loss": 0.3551,
      "step": 2903
    },
    {
      "epoch": 1.8302273354083765,
      "grad_norm": 0.17856952548027039,
      "learning_rate": 8.040053972725179e-07,
      "loss": 0.343,
      "step": 2904
    },
    {
      "epoch": 1.8308577282218983,
      "grad_norm": 0.18663626909255981,
      "learning_rate": 7.980522088243358e-07,
      "loss": 0.3805,
      "step": 2905
    },
    {
      "epoch": 1.8314881210354201,
      "grad_norm": 0.14066052436828613,
      "learning_rate": 7.921207443427994e-07,
      "loss": 0.3002,
      "step": 2906
    },
    {
      "epoch": 1.832118513848942,
      "grad_norm": 0.1478758603334427,
      "learning_rate": 7.862110097653966e-07,
      "loss": 0.274,
      "step": 2907
    },
    {
      "epoch": 1.832748906662464,
      "grad_norm": 0.1529366672039032,
      "learning_rate": 7.803230110078502e-07,
      "loss": 0.3466,
      "step": 2908
    },
    {
      "epoch": 1.833379299475986,
      "grad_norm": 0.16557784378528595,
      "learning_rate": 7.744567539641406e-07,
      "loss": 0.3323,
      "step": 2909
    },
    {
      "epoch": 1.834009692289508,
      "grad_norm": 0.2191508263349533,
      "learning_rate": 7.686122445064731e-07,
      "loss": 0.4284,
      "step": 2910
    },
    {
      "epoch": 1.8346400851030298,
      "grad_norm": 0.21287937462329865,
      "learning_rate": 7.627894884852902e-07,
      "loss": 0.3492,
      "step": 2911
    },
    {
      "epoch": 1.8352704779165516,
      "grad_norm": 0.1814602017402649,
      "learning_rate": 7.569884917292519e-07,
      "loss": 0.3924,
      "step": 2912
    },
    {
      "epoch": 1.8359008707300737,
      "grad_norm": 0.18028287589550018,
      "learning_rate": 7.512092600452408e-07,
      "loss": 0.386,
      "step": 2913
    },
    {
      "epoch": 1.8365312635435957,
      "grad_norm": 0.15933865308761597,
      "learning_rate": 7.454517992183569e-07,
      "loss": 0.3295,
      "step": 2914
    },
    {
      "epoch": 1.8371616563571176,
      "grad_norm": 0.14482073485851288,
      "learning_rate": 7.397161150118998e-07,
      "loss": 0.3374,
      "step": 2915
    },
    {
      "epoch": 1.8377920491706394,
      "grad_norm": 0.17769312858581543,
      "learning_rate": 7.340022131673693e-07,
      "loss": 0.2856,
      "step": 2916
    },
    {
      "epoch": 1.8384224419841613,
      "grad_norm": 0.18236099183559418,
      "learning_rate": 7.283100994044678e-07,
      "loss": 0.3551,
      "step": 2917
    },
    {
      "epoch": 1.8390528347976833,
      "grad_norm": 0.138231098651886,
      "learning_rate": 7.226397794210848e-07,
      "loss": 0.3117,
      "step": 2918
    },
    {
      "epoch": 1.8396832276112054,
      "grad_norm": 0.18965476751327515,
      "learning_rate": 7.169912588932976e-07,
      "loss": 0.3806,
      "step": 2919
    },
    {
      "epoch": 1.8403136204247272,
      "grad_norm": 0.17229288816452026,
      "learning_rate": 7.11364543475348e-07,
      "loss": 0.2963,
      "step": 2920
    },
    {
      "epoch": 1.840944013238249,
      "grad_norm": 0.1469162553548813,
      "learning_rate": 7.057596387996707e-07,
      "loss": 0.2906,
      "step": 2921
    },
    {
      "epoch": 1.841574406051771,
      "grad_norm": 0.16973532736301422,
      "learning_rate": 7.001765504768451e-07,
      "loss": 0.3742,
      "step": 2922
    },
    {
      "epoch": 1.842204798865293,
      "grad_norm": 0.1712886393070221,
      "learning_rate": 6.946152840956358e-07,
      "loss": 0.397,
      "step": 2923
    },
    {
      "epoch": 1.8428351916788148,
      "grad_norm": 0.18453745543956757,
      "learning_rate": 6.890758452229471e-07,
      "loss": 0.3905,
      "step": 2924
    },
    {
      "epoch": 1.8434655844923369,
      "grad_norm": 0.16405226290225983,
      "learning_rate": 6.835582394038359e-07,
      "loss": 0.2815,
      "step": 2925
    },
    {
      "epoch": 1.8440959773058587,
      "grad_norm": 0.16712190210819244,
      "learning_rate": 6.780624721615092e-07,
      "loss": 0.3236,
      "step": 2926
    },
    {
      "epoch": 1.8447263701193806,
      "grad_norm": 0.19651249051094055,
      "learning_rate": 6.725885489973089e-07,
      "loss": 0.4105,
      "step": 2927
    },
    {
      "epoch": 1.8453567629329024,
      "grad_norm": 0.18198879063129425,
      "learning_rate": 6.671364753907116e-07,
      "loss": 0.3217,
      "step": 2928
    },
    {
      "epoch": 1.8459871557464245,
      "grad_norm": 0.1818927526473999,
      "learning_rate": 6.617062567993245e-07,
      "loss": 0.309,
      "step": 2929
    },
    {
      "epoch": 1.8466175485599465,
      "grad_norm": 0.17822176218032837,
      "learning_rate": 6.562978986588744e-07,
      "loss": 0.3661,
      "step": 2930
    },
    {
      "epoch": 1.8472479413734684,
      "grad_norm": 0.1938237100839615,
      "learning_rate": 6.50911406383208e-07,
      "loss": 0.3514,
      "step": 2931
    },
    {
      "epoch": 1.8478783341869902,
      "grad_norm": 0.1587604582309723,
      "learning_rate": 6.4554678536428e-07,
      "loss": 0.3102,
      "step": 2932
    },
    {
      "epoch": 1.848508727000512,
      "grad_norm": 0.17696255445480347,
      "learning_rate": 6.402040409721568e-07,
      "loss": 0.3909,
      "step": 2933
    },
    {
      "epoch": 1.8491391198140341,
      "grad_norm": 0.13390642404556274,
      "learning_rate": 6.348831785550033e-07,
      "loss": 0.2746,
      "step": 2934
    },
    {
      "epoch": 1.8497695126275562,
      "grad_norm": 0.13030406832695007,
      "learning_rate": 6.295842034390786e-07,
      "loss": 0.3389,
      "step": 2935
    },
    {
      "epoch": 1.850399905441078,
      "grad_norm": 0.12831594049930573,
      "learning_rate": 6.243071209287374e-07,
      "loss": 0.2808,
      "step": 2936
    },
    {
      "epoch": 1.8510302982545999,
      "grad_norm": 0.19339363276958466,
      "learning_rate": 6.19051936306414e-07,
      "loss": 0.4201,
      "step": 2937
    },
    {
      "epoch": 1.8516606910681217,
      "grad_norm": 0.1678164303302765,
      "learning_rate": 6.138186548326255e-07,
      "loss": 0.369,
      "step": 2938
    },
    {
      "epoch": 1.8522910838816438,
      "grad_norm": 0.17880061268806458,
      "learning_rate": 6.086072817459615e-07,
      "loss": 0.3824,
      "step": 2939
    },
    {
      "epoch": 1.8529214766951658,
      "grad_norm": 0.22390973567962646,
      "learning_rate": 6.034178222630865e-07,
      "loss": 0.4395,
      "step": 2940
    },
    {
      "epoch": 1.8535518695086877,
      "grad_norm": 0.15214017033576965,
      "learning_rate": 5.982502815787228e-07,
      "loss": 0.3766,
      "step": 2941
    },
    {
      "epoch": 1.8541822623222095,
      "grad_norm": 0.17880332469940186,
      "learning_rate": 5.931046648656524e-07,
      "loss": 0.4008,
      "step": 2942
    },
    {
      "epoch": 1.8548126551357313,
      "grad_norm": 0.1607338786125183,
      "learning_rate": 5.879809772747177e-07,
      "loss": 0.3282,
      "step": 2943
    },
    {
      "epoch": 1.8554430479492534,
      "grad_norm": 0.15719808638095856,
      "learning_rate": 5.82879223934796e-07,
      "loss": 0.355,
      "step": 2944
    },
    {
      "epoch": 1.8560734407627753,
      "grad_norm": 0.15147340297698975,
      "learning_rate": 5.777994099528272e-07,
      "loss": 0.3112,
      "step": 2945
    },
    {
      "epoch": 1.8567038335762973,
      "grad_norm": 0.15864863991737366,
      "learning_rate": 5.727415404137737e-07,
      "loss": 0.3386,
      "step": 2946
    },
    {
      "epoch": 1.8573342263898192,
      "grad_norm": 0.1567879170179367,
      "learning_rate": 5.67705620380638e-07,
      "loss": 0.3387,
      "step": 2947
    },
    {
      "epoch": 1.857964619203341,
      "grad_norm": 0.14126014709472656,
      "learning_rate": 5.626916548944478e-07,
      "loss": 0.3969,
      "step": 2948
    },
    {
      "epoch": 1.858595012016863,
      "grad_norm": 0.1492016315460205,
      "learning_rate": 5.576996489742581e-07,
      "loss": 0.3638,
      "step": 2949
    },
    {
      "epoch": 1.859225404830385,
      "grad_norm": 0.21211956441402435,
      "learning_rate": 5.527296076171368e-07,
      "loss": 0.382,
      "step": 2950
    },
    {
      "epoch": 1.859855797643907,
      "grad_norm": 0.17853829264640808,
      "learning_rate": 5.477815357981742e-07,
      "loss": 0.3108,
      "step": 2951
    },
    {
      "epoch": 1.8604861904574288,
      "grad_norm": 0.18469850718975067,
      "learning_rate": 5.428554384704557e-07,
      "loss": 0.365,
      "step": 2952
    },
    {
      "epoch": 1.8611165832709506,
      "grad_norm": 0.15979118645191193,
      "learning_rate": 5.379513205650768e-07,
      "loss": 0.2674,
      "step": 2953
    },
    {
      "epoch": 1.8617469760844725,
      "grad_norm": 0.21204063296318054,
      "learning_rate": 5.330691869911359e-07,
      "loss": 0.3816,
      "step": 2954
    },
    {
      "epoch": 1.8623773688979945,
      "grad_norm": 0.19262298941612244,
      "learning_rate": 5.282090426357208e-07,
      "loss": 0.3409,
      "step": 2955
    },
    {
      "epoch": 1.8630077617115166,
      "grad_norm": 0.16887032985687256,
      "learning_rate": 5.233708923638978e-07,
      "loss": 0.3895,
      "step": 2956
    },
    {
      "epoch": 1.8636381545250384,
      "grad_norm": 0.15567231178283691,
      "learning_rate": 5.185547410187349e-07,
      "loss": 0.3607,
      "step": 2957
    },
    {
      "epoch": 1.8642685473385603,
      "grad_norm": 0.17275381088256836,
      "learning_rate": 5.137605934212661e-07,
      "loss": 0.2954,
      "step": 2958
    },
    {
      "epoch": 1.8648989401520821,
      "grad_norm": 0.18041570484638214,
      "learning_rate": 5.089884543705073e-07,
      "loss": 0.4095,
      "step": 2959
    },
    {
      "epoch": 1.8655293329656042,
      "grad_norm": 0.16471639275550842,
      "learning_rate": 5.042383286434324e-07,
      "loss": 0.3647,
      "step": 2960
    },
    {
      "epoch": 1.8661597257791263,
      "grad_norm": 0.15917225182056427,
      "learning_rate": 4.995102209949904e-07,
      "loss": 0.3822,
      "step": 2961
    },
    {
      "epoch": 1.866790118592648,
      "grad_norm": 0.174025759100914,
      "learning_rate": 4.948041361580852e-07,
      "loss": 0.3648,
      "step": 2962
    },
    {
      "epoch": 1.86742051140617,
      "grad_norm": 0.151805579662323,
      "learning_rate": 4.901200788435811e-07,
      "loss": 0.3146,
      "step": 2963
    },
    {
      "epoch": 1.8680509042196918,
      "grad_norm": 0.19342181086540222,
      "learning_rate": 4.8545805374028e-07,
      "loss": 0.3677,
      "step": 2964
    },
    {
      "epoch": 1.8686812970332138,
      "grad_norm": 0.14720351994037628,
      "learning_rate": 4.808180655149438e-07,
      "loss": 0.3263,
      "step": 2965
    },
    {
      "epoch": 1.869311689846736,
      "grad_norm": 0.1777946650981903,
      "learning_rate": 4.7620011881226224e-07,
      "loss": 0.3635,
      "step": 2966
    },
    {
      "epoch": 1.8699420826602577,
      "grad_norm": 0.16213665902614594,
      "learning_rate": 4.7160421825487493e-07,
      "loss": 0.3436,
      "step": 2967
    },
    {
      "epoch": 1.8705724754737796,
      "grad_norm": 0.18879404664039612,
      "learning_rate": 4.670303684433394e-07,
      "loss": 0.3891,
      "step": 2968
    },
    {
      "epoch": 1.8712028682873014,
      "grad_norm": 0.13427011668682098,
      "learning_rate": 4.6247857395615076e-07,
      "loss": 0.3509,
      "step": 2969
    },
    {
      "epoch": 1.8718332611008235,
      "grad_norm": 0.18270568549633026,
      "learning_rate": 4.579488393497167e-07,
      "loss": 0.3303,
      "step": 2970
    },
    {
      "epoch": 1.8724636539143453,
      "grad_norm": 0.16502298414707184,
      "learning_rate": 4.534411691583701e-07,
      "loss": 0.3537,
      "step": 2971
    },
    {
      "epoch": 1.8730940467278674,
      "grad_norm": 0.2042229175567627,
      "learning_rate": 4.489555678943541e-07,
      "loss": 0.386,
      "step": 2972
    },
    {
      "epoch": 1.8737244395413892,
      "grad_norm": 0.1804782897233963,
      "learning_rate": 4.4449204004782445e-07,
      "loss": 0.3777,
      "step": 2973
    },
    {
      "epoch": 1.874354832354911,
      "grad_norm": 0.200051411986351,
      "learning_rate": 4.400505900868321e-07,
      "loss": 0.34,
      "step": 2974
    },
    {
      "epoch": 1.8749852251684331,
      "grad_norm": 0.1387588083744049,
      "learning_rate": 4.3563122245733805e-07,
      "loss": 0.2821,
      "step": 2975
    },
    {
      "epoch": 1.875615617981955,
      "grad_norm": 0.16911667585372925,
      "learning_rate": 4.312339415831912e-07,
      "loss": 0.3272,
      "step": 2976
    },
    {
      "epoch": 1.876246010795477,
      "grad_norm": 0.18176445364952087,
      "learning_rate": 4.2685875186613547e-07,
      "loss": 0.3403,
      "step": 2977
    },
    {
      "epoch": 1.8768764036089989,
      "grad_norm": 0.15195323526859283,
      "learning_rate": 4.225056576857975e-07,
      "loss": 0.337,
      "step": 2978
    },
    {
      "epoch": 1.8775067964225207,
      "grad_norm": 0.17995168268680573,
      "learning_rate": 4.1817466339968914e-07,
      "loss": 0.3082,
      "step": 2979
    },
    {
      "epoch": 1.8781371892360426,
      "grad_norm": 0.18412546813488007,
      "learning_rate": 4.1386577334319997e-07,
      "loss": 0.3054,
      "step": 2980
    },
    {
      "epoch": 1.8787675820495646,
      "grad_norm": 0.18003132939338684,
      "learning_rate": 4.0957899182959226e-07,
      "loss": 0.3801,
      "step": 2981
    },
    {
      "epoch": 1.8793979748630867,
      "grad_norm": 0.17081038653850555,
      "learning_rate": 4.053143231499934e-07,
      "loss": 0.3192,
      "step": 2982
    },
    {
      "epoch": 1.8800283676766085,
      "grad_norm": 0.1643000692129135,
      "learning_rate": 4.010717715734036e-07,
      "loss": 0.3418,
      "step": 2983
    },
    {
      "epoch": 1.8806587604901304,
      "grad_norm": 0.12995314598083496,
      "learning_rate": 3.9685134134667317e-07,
      "loss": 0.2888,
      "step": 2984
    },
    {
      "epoch": 1.8812891533036522,
      "grad_norm": 0.1708272397518158,
      "learning_rate": 3.9265303669452277e-07,
      "loss": 0.3322,
      "step": 2985
    },
    {
      "epoch": 1.8819195461171743,
      "grad_norm": 0.16013602912425995,
      "learning_rate": 3.884768618195106e-07,
      "loss": 0.3303,
      "step": 2986
    },
    {
      "epoch": 1.8825499389306963,
      "grad_norm": 0.1538516879081726,
      "learning_rate": 3.8432282090205023e-07,
      "loss": 0.3555,
      "step": 2987
    },
    {
      "epoch": 1.8831803317442182,
      "grad_norm": 0.19950860738754272,
      "learning_rate": 3.8019091810039544e-07,
      "loss": 0.4115,
      "step": 2988
    },
    {
      "epoch": 1.88381072455774,
      "grad_norm": 0.20162293314933777,
      "learning_rate": 3.760811575506426e-07,
      "loss": 0.3634,
      "step": 2989
    },
    {
      "epoch": 1.8844411173712619,
      "grad_norm": 0.17130547761917114,
      "learning_rate": 3.71993543366721e-07,
      "loss": 0.3337,
      "step": 2990
    },
    {
      "epoch": 1.885071510184784,
      "grad_norm": 0.14211808145046234,
      "learning_rate": 3.679280796403975e-07,
      "loss": 0.3429,
      "step": 2991
    },
    {
      "epoch": 1.885701902998306,
      "grad_norm": 0.16880738735198975,
      "learning_rate": 3.638847704412543e-07,
      "loss": 0.3424,
      "step": 2992
    },
    {
      "epoch": 1.8863322958118278,
      "grad_norm": 0.18482467532157898,
      "learning_rate": 3.598636198167038e-07,
      "loss": 0.3143,
      "step": 2993
    },
    {
      "epoch": 1.8869626886253497,
      "grad_norm": 0.15687884390354156,
      "learning_rate": 3.5586463179197864e-07,
      "loss": 0.2924,
      "step": 2994
    },
    {
      "epoch": 1.8875930814388715,
      "grad_norm": 0.19636908173561096,
      "learning_rate": 3.518878103701268e-07,
      "loss": 0.3539,
      "step": 2995
    },
    {
      "epoch": 1.8882234742523936,
      "grad_norm": 0.15740831196308136,
      "learning_rate": 3.4793315953199624e-07,
      "loss": 0.3046,
      "step": 2996
    },
    {
      "epoch": 1.8888538670659154,
      "grad_norm": 0.13536562025547028,
      "learning_rate": 3.4400068323625803e-07,
      "loss": 0.3097,
      "step": 2997
    },
    {
      "epoch": 1.8894842598794375,
      "grad_norm": 0.1597716510295868,
      "learning_rate": 3.4009038541937574e-07,
      "loss": 0.3146,
      "step": 2998
    },
    {
      "epoch": 1.8901146526929593,
      "grad_norm": 0.1682119220495224,
      "learning_rate": 3.3620226999561325e-07,
      "loss": 0.4298,
      "step": 2999
    },
    {
      "epoch": 1.8907450455064811,
      "grad_norm": 0.20387521386146545,
      "learning_rate": 3.3233634085703236e-07,
      "loss": 0.4097,
      "step": 3000
    },
    {
      "epoch": 1.8907450455064811,
      "eval_loss": 0.41817376017570496,
      "eval_runtime": 222.7094,
      "eval_samples_per_second": 4.49,
      "eval_steps_per_second": 4.49,
      "step": 3000
    },
    {
      "epoch": 1.891375438320003,
      "grad_norm": 0.19256849586963654,
      "learning_rate": 3.284926018734849e-07,
      "loss": 0.309,
      "step": 3001
    },
    {
      "epoch": 1.892005831133525,
      "grad_norm": 0.20908156037330627,
      "learning_rate": 3.246710568926081e-07,
      "loss": 0.396,
      "step": 3002
    },
    {
      "epoch": 1.8926362239470471,
      "grad_norm": 0.14932669699192047,
      "learning_rate": 3.20871709739827e-07,
      "loss": 0.3048,
      "step": 3003
    },
    {
      "epoch": 1.893266616760569,
      "grad_norm": 0.1737649291753769,
      "learning_rate": 3.170945642183393e-07,
      "loss": 0.3218,
      "step": 3004
    },
    {
      "epoch": 1.8938970095740908,
      "grad_norm": 0.1602880358695984,
      "learning_rate": 3.133396241091305e-07,
      "loss": 0.3153,
      "step": 3005
    },
    {
      "epoch": 1.8945274023876126,
      "grad_norm": 0.15219104290008545,
      "learning_rate": 3.0960689317093896e-07,
      "loss": 0.3298,
      "step": 3006
    },
    {
      "epoch": 1.8951577952011347,
      "grad_norm": 0.16922716796398163,
      "learning_rate": 3.0589637514029326e-07,
      "loss": 0.3161,
      "step": 3007
    },
    {
      "epoch": 1.8957881880146568,
      "grad_norm": 0.18126600980758667,
      "learning_rate": 3.0220807373146976e-07,
      "loss": 0.3512,
      "step": 3008
    },
    {
      "epoch": 1.8964185808281786,
      "grad_norm": 0.17149627208709717,
      "learning_rate": 2.985419926365177e-07,
      "loss": 0.3605,
      "step": 3009
    },
    {
      "epoch": 1.8970489736417004,
      "grad_norm": 0.18697969615459442,
      "learning_rate": 2.9489813552523146e-07,
      "loss": 0.3286,
      "step": 3010
    },
    {
      "epoch": 1.8976793664552223,
      "grad_norm": 0.17346805334091187,
      "learning_rate": 2.912765060451683e-07,
      "loss": 0.3158,
      "step": 3011
    },
    {
      "epoch": 1.8983097592687443,
      "grad_norm": 0.16184809803962708,
      "learning_rate": 2.876771078216334e-07,
      "loss": 0.2843,
      "step": 3012
    },
    {
      "epoch": 1.8989401520822664,
      "grad_norm": 0.16137875616550446,
      "learning_rate": 2.8409994445767696e-07,
      "loss": 0.3306,
      "step": 3013
    },
    {
      "epoch": 1.8995705448957882,
      "grad_norm": 0.17828133702278137,
      "learning_rate": 2.8054501953408725e-07,
      "loss": 0.3278,
      "step": 3014
    },
    {
      "epoch": 1.90020093770931,
      "grad_norm": 0.17460158467292786,
      "learning_rate": 2.770123366094027e-07,
      "loss": 0.3224,
      "step": 3015
    },
    {
      "epoch": 1.900831330522832,
      "grad_norm": 0.18349096179008484,
      "learning_rate": 2.735018992198846e-07,
      "loss": 0.3714,
      "step": 3016
    },
    {
      "epoch": 1.901461723336354,
      "grad_norm": 0.17420694231987,
      "learning_rate": 2.7001371087953697e-07,
      "loss": 0.3074,
      "step": 3017
    },
    {
      "epoch": 1.902092116149876,
      "grad_norm": 0.17939402163028717,
      "learning_rate": 2.665477750800818e-07,
      "loss": 0.3582,
      "step": 3018
    },
    {
      "epoch": 1.902722508963398,
      "grad_norm": 0.18356801569461823,
      "learning_rate": 2.6310409529097636e-07,
      "loss": 0.3741,
      "step": 3019
    },
    {
      "epoch": 1.9033529017769197,
      "grad_norm": 0.16717529296875,
      "learning_rate": 2.5968267495939056e-07,
      "loss": 0.3853,
      "step": 3020
    },
    {
      "epoch": 1.9039832945904416,
      "grad_norm": 0.15661168098449707,
      "learning_rate": 2.562835175102199e-07,
      "loss": 0.3774,
      "step": 3021
    },
    {
      "epoch": 1.9046136874039636,
      "grad_norm": 0.15401984751224518,
      "learning_rate": 2.5290662634606997e-07,
      "loss": 0.3381,
      "step": 3022
    },
    {
      "epoch": 1.9052440802174855,
      "grad_norm": 0.20636636018753052,
      "learning_rate": 2.495520048472569e-07,
      "loss": 0.3507,
      "step": 3023
    },
    {
      "epoch": 1.9058744730310075,
      "grad_norm": 0.1823091357946396,
      "learning_rate": 2.4621965637180195e-07,
      "loss": 0.283,
      "step": 3024
    },
    {
      "epoch": 1.9065048658445294,
      "grad_norm": 0.14718377590179443,
      "learning_rate": 2.4290958425544187e-07,
      "loss": 0.4077,
      "step": 3025
    },
    {
      "epoch": 1.9071352586580512,
      "grad_norm": 0.15838302671909332,
      "learning_rate": 2.396217918116037e-07,
      "loss": 0.3339,
      "step": 3026
    },
    {
      "epoch": 1.907765651471573,
      "grad_norm": 0.18296922743320465,
      "learning_rate": 2.363562823314147e-07,
      "loss": 0.3178,
      "step": 3027
    },
    {
      "epoch": 1.9083960442850951,
      "grad_norm": 0.19694828987121582,
      "learning_rate": 2.3311305908369774e-07,
      "loss": 0.3308,
      "step": 3028
    },
    {
      "epoch": 1.9090264370986172,
      "grad_norm": 0.166620671749115,
      "learning_rate": 2.298921253149708e-07,
      "loss": 0.2882,
      "step": 3029
    },
    {
      "epoch": 1.909656829912139,
      "grad_norm": 0.18683145940303802,
      "learning_rate": 2.2669348424942977e-07,
      "loss": 0.3599,
      "step": 3030
    },
    {
      "epoch": 1.9102872227256609,
      "grad_norm": 0.18327577412128448,
      "learning_rate": 2.2351713908896594e-07,
      "loss": 0.2976,
      "step": 3031
    },
    {
      "epoch": 1.9109176155391827,
      "grad_norm": 0.1980234980583191,
      "learning_rate": 2.203630930131434e-07,
      "loss": 0.3111,
      "step": 3032
    },
    {
      "epoch": 1.9115480083527048,
      "grad_norm": 0.18122927844524384,
      "learning_rate": 2.1723134917920913e-07,
      "loss": 0.4303,
      "step": 3033
    },
    {
      "epoch": 1.9121784011662268,
      "grad_norm": 0.17239895462989807,
      "learning_rate": 2.1412191072208543e-07,
      "loss": 0.3429,
      "step": 3034
    },
    {
      "epoch": 1.9128087939797487,
      "grad_norm": 0.15282370150089264,
      "learning_rate": 2.110347807543675e-07,
      "loss": 0.3253,
      "step": 3035
    },
    {
      "epoch": 1.9134391867932705,
      "grad_norm": 0.18070636689662933,
      "learning_rate": 2.079699623663134e-07,
      "loss": 0.3788,
      "step": 3036
    },
    {
      "epoch": 1.9140695796067924,
      "grad_norm": 0.20107102394104004,
      "learning_rate": 2.049274586258515e-07,
      "loss": 0.3869,
      "step": 3037
    },
    {
      "epoch": 1.9146999724203144,
      "grad_norm": 0.18471233546733856,
      "learning_rate": 2.019072725785756e-07,
      "loss": 0.4217,
      "step": 3038
    },
    {
      "epoch": 1.9153303652338365,
      "grad_norm": 0.15934079885482788,
      "learning_rate": 1.9890940724773738e-07,
      "loss": 0.3242,
      "step": 3039
    },
    {
      "epoch": 1.9159607580473583,
      "grad_norm": 0.16995728015899658,
      "learning_rate": 1.9593386563423882e-07,
      "loss": 0.3333,
      "step": 3040
    },
    {
      "epoch": 1.9165911508608802,
      "grad_norm": 0.2071707844734192,
      "learning_rate": 1.9298065071663985e-07,
      "loss": 0.3952,
      "step": 3041
    },
    {
      "epoch": 1.917221543674402,
      "grad_norm": 0.17585161328315735,
      "learning_rate": 1.900497654511582e-07,
      "loss": 0.3582,
      "step": 3042
    },
    {
      "epoch": 1.917851936487924,
      "grad_norm": 0.18772096931934357,
      "learning_rate": 1.8714121277164958e-07,
      "loss": 0.326,
      "step": 3043
    },
    {
      "epoch": 1.918482329301446,
      "grad_norm": 0.15703828632831573,
      "learning_rate": 1.8425499558961502e-07,
      "loss": 0.345,
      "step": 3044
    },
    {
      "epoch": 1.919112722114968,
      "grad_norm": 0.1644994169473648,
      "learning_rate": 1.813911167942034e-07,
      "loss": 0.3261,
      "step": 3045
    },
    {
      "epoch": 1.9197431149284898,
      "grad_norm": 0.15160581469535828,
      "learning_rate": 1.7854957925219657e-07,
      "loss": 0.3345,
      "step": 3046
    },
    {
      "epoch": 1.9203735077420117,
      "grad_norm": 0.16602809727191925,
      "learning_rate": 1.757303858080167e-07,
      "loss": 0.328,
      "step": 3047
    },
    {
      "epoch": 1.9210039005555337,
      "grad_norm": 0.16758157312870026,
      "learning_rate": 1.7293353928371893e-07,
      "loss": 0.3538,
      "step": 3048
    },
    {
      "epoch": 1.9216342933690556,
      "grad_norm": 0.18970537185668945,
      "learning_rate": 1.7015904247898618e-07,
      "loss": 0.3269,
      "step": 3049
    },
    {
      "epoch": 1.9222646861825776,
      "grad_norm": 0.15211471915245056,
      "learning_rate": 1.6740689817112936e-07,
      "loss": 0.35,
      "step": 3050
    },
    {
      "epoch": 1.9228950789960995,
      "grad_norm": 0.14648528397083282,
      "learning_rate": 1.6467710911508476e-07,
      "loss": 0.2858,
      "step": 3051
    },
    {
      "epoch": 1.9235254718096213,
      "grad_norm": 0.22460030019283295,
      "learning_rate": 1.6196967804341408e-07,
      "loss": 0.4544,
      "step": 3052
    },
    {
      "epoch": 1.9241558646231431,
      "grad_norm": 0.14347076416015625,
      "learning_rate": 1.5928460766629432e-07,
      "loss": 0.3336,
      "step": 3053
    },
    {
      "epoch": 1.9247862574366652,
      "grad_norm": 0.2079204022884369,
      "learning_rate": 1.5662190067151805e-07,
      "loss": 0.3914,
      "step": 3054
    },
    {
      "epoch": 1.9254166502501873,
      "grad_norm": 0.1357291042804718,
      "learning_rate": 1.5398155972449314e-07,
      "loss": 0.2695,
      "step": 3055
    },
    {
      "epoch": 1.926047043063709,
      "grad_norm": 0.1850150227546692,
      "learning_rate": 1.5136358746824287e-07,
      "loss": 0.3429,
      "step": 3056
    },
    {
      "epoch": 1.926677435877231,
      "grad_norm": 0.15783745050430298,
      "learning_rate": 1.4876798652339595e-07,
      "loss": 0.39,
      "step": 3057
    },
    {
      "epoch": 1.9273078286907528,
      "grad_norm": 0.17066796123981476,
      "learning_rate": 1.4619475948818407e-07,
      "loss": 0.3225,
      "step": 3058
    },
    {
      "epoch": 1.9279382215042749,
      "grad_norm": 0.15452978014945984,
      "learning_rate": 1.436439089384467e-07,
      "loss": 0.3813,
      "step": 3059
    },
    {
      "epoch": 1.928568614317797,
      "grad_norm": 0.1883036345243454,
      "learning_rate": 1.4111543742762135e-07,
      "loss": 0.3995,
      "step": 3060
    },
    {
      "epoch": 1.9291990071313188,
      "grad_norm": 0.14374081790447235,
      "learning_rate": 1.3860934748674588e-07,
      "loss": 0.3013,
      "step": 3061
    },
    {
      "epoch": 1.9298293999448406,
      "grad_norm": 0.21677044034004211,
      "learning_rate": 1.3612564162445106e-07,
      "loss": 0.4131,
      "step": 3062
    },
    {
      "epoch": 1.9304597927583624,
      "grad_norm": 0.16398075222969055,
      "learning_rate": 1.3366432232696568e-07,
      "loss": 0.3285,
      "step": 3063
    },
    {
      "epoch": 1.9310901855718845,
      "grad_norm": 0.15119622647762299,
      "learning_rate": 1.3122539205810387e-07,
      "loss": 0.2927,
      "step": 3064
    },
    {
      "epoch": 1.9317205783854066,
      "grad_norm": 0.1611369550228119,
      "learning_rate": 1.2880885325927268e-07,
      "loss": 0.3363,
      "step": 3065
    },
    {
      "epoch": 1.9323509711989284,
      "grad_norm": 0.1843705177307129,
      "learning_rate": 1.2641470834945963e-07,
      "loss": 0.3361,
      "step": 3066
    },
    {
      "epoch": 1.9329813640124502,
      "grad_norm": 0.21802350878715515,
      "learning_rate": 1.2404295972524266e-07,
      "loss": 0.4132,
      "step": 3067
    },
    {
      "epoch": 1.933611756825972,
      "grad_norm": 0.1921495646238327,
      "learning_rate": 1.2169360976077266e-07,
      "loss": 0.3443,
      "step": 3068
    },
    {
      "epoch": 1.9342421496394941,
      "grad_norm": 0.1394999772310257,
      "learning_rate": 1.193666608077859e-07,
      "loss": 0.2871,
      "step": 3069
    },
    {
      "epoch": 1.934872542453016,
      "grad_norm": 0.1905948966741562,
      "learning_rate": 1.1706211519559169e-07,
      "loss": 0.4448,
      "step": 3070
    },
    {
      "epoch": 1.935502935266538,
      "grad_norm": 0.18064671754837036,
      "learning_rate": 1.1477997523107464e-07,
      "loss": 0.3654,
      "step": 3071
    },
    {
      "epoch": 1.93613332808006,
      "grad_norm": 0.16649015247821808,
      "learning_rate": 1.1252024319868992e-07,
      "loss": 0.3733,
      "step": 3072
    },
    {
      "epoch": 1.9367637208935817,
      "grad_norm": 0.19247853755950928,
      "learning_rate": 1.1028292136046059e-07,
      "loss": 0.4663,
      "step": 3073
    },
    {
      "epoch": 1.9373941137071038,
      "grad_norm": 0.18122366070747375,
      "learning_rate": 1.0806801195597766e-07,
      "loss": 0.3317,
      "step": 3074
    },
    {
      "epoch": 1.9380245065206256,
      "grad_norm": 0.15887488424777985,
      "learning_rate": 1.0587551720240258e-07,
      "loss": 0.3566,
      "step": 3075
    },
    {
      "epoch": 1.9386548993341477,
      "grad_norm": 0.19291207194328308,
      "learning_rate": 1.0370543929444975e-07,
      "loss": 0.3351,
      "step": 3076
    },
    {
      "epoch": 1.9392852921476695,
      "grad_norm": 0.1809178739786148,
      "learning_rate": 1.0155778040440156e-07,
      "loss": 0.3887,
      "step": 3077
    },
    {
      "epoch": 1.9399156849611914,
      "grad_norm": 0.1738986223936081,
      "learning_rate": 9.943254268209579e-08,
      "loss": 0.3293,
      "step": 3078
    },
    {
      "epoch": 1.9405460777747132,
      "grad_norm": 0.16989794373512268,
      "learning_rate": 9.73297282549257e-08,
      "loss": 0.3517,
      "step": 3079
    },
    {
      "epoch": 1.9411764705882353,
      "grad_norm": 0.17106318473815918,
      "learning_rate": 9.524933922784002e-08,
      "loss": 0.3602,
      "step": 3080
    },
    {
      "epoch": 1.9418068634017573,
      "grad_norm": 0.18975426256656647,
      "learning_rate": 9.319137768333541e-08,
      "loss": 0.3415,
      "step": 3081
    },
    {
      "epoch": 1.9424372562152792,
      "grad_norm": 0.17912082374095917,
      "learning_rate": 9.115584568146651e-08,
      "loss": 0.3994,
      "step": 3082
    },
    {
      "epoch": 1.943067649028801,
      "grad_norm": 0.16902948915958405,
      "learning_rate": 8.914274525982841e-08,
      "loss": 0.3515,
      "step": 3083
    },
    {
      "epoch": 1.9436980418423229,
      "grad_norm": 0.17020246386528015,
      "learning_rate": 8.715207843356165e-08,
      "loss": 0.2807,
      "step": 3084
    },
    {
      "epoch": 1.944328434655845,
      "grad_norm": 0.1424531638622284,
      "learning_rate": 8.518384719535477e-08,
      "loss": 0.3058,
      "step": 3085
    },
    {
      "epoch": 1.944958827469367,
      "grad_norm": 0.17307209968566895,
      "learning_rate": 8.323805351543673e-08,
      "loss": 0.3989,
      "step": 3086
    },
    {
      "epoch": 1.9455892202828888,
      "grad_norm": 0.197873055934906,
      "learning_rate": 8.131469934157448e-08,
      "loss": 0.365,
      "step": 3087
    },
    {
      "epoch": 1.9462196130964107,
      "grad_norm": 0.14806973934173584,
      "learning_rate": 7.941378659907295e-08,
      "loss": 0.3045,
      "step": 3088
    },
    {
      "epoch": 1.9468500059099325,
      "grad_norm": 0.14717578887939453,
      "learning_rate": 7.75353171907775e-08,
      "loss": 0.3438,
      "step": 3089
    },
    {
      "epoch": 1.9474803987234546,
      "grad_norm": 0.15752723813056946,
      "learning_rate": 7.567929299705651e-08,
      "loss": 0.3682,
      "step": 3090
    },
    {
      "epoch": 1.9481107915369766,
      "grad_norm": 0.1642657220363617,
      "learning_rate": 7.384571587582378e-08,
      "loss": 0.3269,
      "step": 3091
    },
    {
      "epoch": 1.9487411843504985,
      "grad_norm": 0.17168262600898743,
      "learning_rate": 7.203458766251364e-08,
      "loss": 0.3386,
      "step": 3092
    },
    {
      "epoch": 1.9493715771640203,
      "grad_norm": 0.18933595716953278,
      "learning_rate": 7.024591017009086e-08,
      "loss": 0.4002,
      "step": 3093
    },
    {
      "epoch": 1.9500019699775422,
      "grad_norm": 0.13463959097862244,
      "learning_rate": 6.847968518905068e-08,
      "loss": 0.2955,
      "step": 3094
    },
    {
      "epoch": 1.9506323627910642,
      "grad_norm": 0.19657647609710693,
      "learning_rate": 6.673591448740885e-08,
      "loss": 0.3551,
      "step": 3095
    },
    {
      "epoch": 1.951262755604586,
      "grad_norm": 0.21734939515590668,
      "learning_rate": 6.501459981070407e-08,
      "loss": 0.4166,
      "step": 3096
    },
    {
      "epoch": 1.9518931484181081,
      "grad_norm": 0.16016525030136108,
      "learning_rate": 6.331574288200054e-08,
      "loss": 0.3323,
      "step": 3097
    },
    {
      "epoch": 1.95252354123163,
      "grad_norm": 0.16205845773220062,
      "learning_rate": 6.163934540187544e-08,
      "loss": 0.351,
      "step": 3098
    },
    {
      "epoch": 1.9531539340451518,
      "grad_norm": 0.1572217047214508,
      "learning_rate": 5.998540904842643e-08,
      "loss": 0.3103,
      "step": 3099
    },
    {
      "epoch": 1.9537843268586736,
      "grad_norm": 0.17913149297237396,
      "learning_rate": 5.835393547727414e-08,
      "loss": 0.3093,
      "step": 3100
    },
    {
      "epoch": 1.9544147196721957,
      "grad_norm": 0.16652511060237885,
      "learning_rate": 5.6744926321539715e-08,
      "loss": 0.3165,
      "step": 3101
    },
    {
      "epoch": 1.9550451124857178,
      "grad_norm": 0.1753828376531601,
      "learning_rate": 5.5158383191872275e-08,
      "loss": 0.3602,
      "step": 3102
    },
    {
      "epoch": 1.9556755052992396,
      "grad_norm": 0.1615012139081955,
      "learning_rate": 5.3594307676416436e-08,
      "loss": 0.2906,
      "step": 3103
    },
    {
      "epoch": 1.9563058981127615,
      "grad_norm": 0.16111844778060913,
      "learning_rate": 5.205270134084228e-08,
      "loss": 0.3132,
      "step": 3104
    },
    {
      "epoch": 1.9569362909262833,
      "grad_norm": 0.21132858097553253,
      "learning_rate": 5.0533565728315415e-08,
      "loss": 0.3911,
      "step": 3105
    },
    {
      "epoch": 1.9575666837398054,
      "grad_norm": 0.16164764761924744,
      "learning_rate": 4.903690235951441e-08,
      "loss": 0.2846,
      "step": 3106
    },
    {
      "epoch": 1.9581970765533274,
      "grad_norm": 0.14004522562026978,
      "learning_rate": 4.756271273262336e-08,
      "loss": 0.3344,
      "step": 3107
    },
    {
      "epoch": 1.9588274693668493,
      "grad_norm": 0.13915061950683594,
      "learning_rate": 4.611099832332183e-08,
      "loss": 0.3273,
      "step": 3108
    },
    {
      "epoch": 1.959457862180371,
      "grad_norm": 0.16548193991184235,
      "learning_rate": 4.4681760584804887e-08,
      "loss": 0.3241,
      "step": 3109
    },
    {
      "epoch": 1.960088254993893,
      "grad_norm": 0.16104203462600708,
      "learning_rate": 4.327500094775311e-08,
      "loss": 0.3391,
      "step": 3110
    },
    {
      "epoch": 1.960718647807415,
      "grad_norm": 0.15194769203662872,
      "learning_rate": 4.1890720820357555e-08,
      "loss": 0.2724,
      "step": 3111
    },
    {
      "epoch": 1.961349040620937,
      "grad_norm": 0.18014349043369293,
      "learning_rate": 4.0528921588299795e-08,
      "loss": 0.3043,
      "step": 3112
    },
    {
      "epoch": 1.961979433434459,
      "grad_norm": 0.14656884968280792,
      "learning_rate": 3.91896046147644e-08,
      "loss": 0.3392,
      "step": 3113
    },
    {
      "epoch": 1.9626098262479807,
      "grad_norm": 0.15012793242931366,
      "learning_rate": 3.787277124042643e-08,
      "loss": 0.3175,
      "step": 3114
    },
    {
      "epoch": 1.9632402190615026,
      "grad_norm": 0.14461544156074524,
      "learning_rate": 3.657842278345397e-08,
      "loss": 0.2879,
      "step": 3115
    },
    {
      "epoch": 1.9638706118750247,
      "grad_norm": 0.13914228975772858,
      "learning_rate": 3.530656053950809e-08,
      "loss": 0.321,
      "step": 3116
    },
    {
      "epoch": 1.9645010046885467,
      "grad_norm": 0.18909741938114166,
      "learning_rate": 3.405718578174289e-08,
      "loss": 0.3458,
      "step": 3117
    },
    {
      "epoch": 1.9651313975020686,
      "grad_norm": 0.1531994789838791,
      "learning_rate": 3.283029976080043e-08,
      "loss": 0.337,
      "step": 3118
    },
    {
      "epoch": 1.9657617903155904,
      "grad_norm": 0.1318083256483078,
      "learning_rate": 3.162590370481333e-08,
      "loss": 0.277,
      "step": 3119
    },
    {
      "epoch": 1.9663921831291122,
      "grad_norm": 0.18918323516845703,
      "learning_rate": 3.044399881939469e-08,
      "loss": 0.3567,
      "step": 3120
    },
    {
      "epoch": 1.9670225759426343,
      "grad_norm": 0.1714440882205963,
      "learning_rate": 2.9284586287655626e-08,
      "loss": 0.3354,
      "step": 3121
    },
    {
      "epoch": 1.9676529687561561,
      "grad_norm": 0.17648588120937347,
      "learning_rate": 2.814766727017776e-08,
      "loss": 0.3654,
      "step": 3122
    },
    {
      "epoch": 1.9682833615696782,
      "grad_norm": 0.16460299491882324,
      "learning_rate": 2.7033242905038226e-08,
      "loss": 0.3516,
      "step": 3123
    },
    {
      "epoch": 1.9689137543832,
      "grad_norm": 0.18309004604816437,
      "learning_rate": 2.594131430778968e-08,
      "loss": 0.3671,
      "step": 3124
    },
    {
      "epoch": 1.9695441471967219,
      "grad_norm": 0.16385690867900848,
      "learning_rate": 2.4871882571472772e-08,
      "loss": 0.3237,
      "step": 3125
    },
    {
      "epoch": 1.9701745400102437,
      "grad_norm": 0.15146617591381073,
      "learning_rate": 2.382494876659869e-08,
      "loss": 0.3169,
      "step": 3126
    },
    {
      "epoch": 1.9708049328237658,
      "grad_norm": 0.14356575906276703,
      "learning_rate": 2.2800513941166614e-08,
      "loss": 0.3171,
      "step": 3127
    },
    {
      "epoch": 1.9714353256372878,
      "grad_norm": 0.1531408727169037,
      "learning_rate": 2.1798579120648756e-08,
      "loss": 0.351,
      "step": 3128
    },
    {
      "epoch": 1.9720657184508097,
      "grad_norm": 0.16242480278015137,
      "learning_rate": 2.0819145308000333e-08,
      "loss": 0.34,
      "step": 3129
    },
    {
      "epoch": 1.9726961112643315,
      "grad_norm": 0.2029055804014206,
      "learning_rate": 1.9862213483644588e-08,
      "loss": 0.3804,
      "step": 3130
    },
    {
      "epoch": 1.9733265040778534,
      "grad_norm": 0.1916704624891281,
      "learning_rate": 1.8927784605487774e-08,
      "loss": 0.3823,
      "step": 3131
    },
    {
      "epoch": 1.9739568968913754,
      "grad_norm": 0.2057744860649109,
      "learning_rate": 1.8015859608904175e-08,
      "loss": 0.3805,
      "step": 3132
    },
    {
      "epoch": 1.9745872897048975,
      "grad_norm": 0.18788807094097137,
      "learning_rate": 1.7126439406746077e-08,
      "loss": 0.3535,
      "step": 3133
    },
    {
      "epoch": 1.9752176825184193,
      "grad_norm": 0.19672442972660065,
      "learning_rate": 1.62595248893338e-08,
      "loss": 0.3592,
      "step": 3134
    },
    {
      "epoch": 1.9758480753319412,
      "grad_norm": 0.20926621556282043,
      "learning_rate": 1.541511692446318e-08,
      "loss": 0.4156,
      "step": 3135
    },
    {
      "epoch": 1.976478468145463,
      "grad_norm": 0.15115287899971008,
      "learning_rate": 1.4593216357400574e-08,
      "loss": 0.2962,
      "step": 3136
    },
    {
      "epoch": 1.977108860958985,
      "grad_norm": 0.182939812541008,
      "learning_rate": 1.3793824010875366e-08,
      "loss": 0.3647,
      "step": 3137
    },
    {
      "epoch": 1.9777392537725071,
      "grad_norm": 0.16283048689365387,
      "learning_rate": 1.3016940685094959e-08,
      "loss": 0.3349,
      "step": 3138
    },
    {
      "epoch": 1.978369646586029,
      "grad_norm": 0.15308210253715515,
      "learning_rate": 1.2262567157729777e-08,
      "loss": 0.3077,
      "step": 3139
    },
    {
      "epoch": 1.9790000393995508,
      "grad_norm": 0.1855187863111496,
      "learning_rate": 1.1530704183918273e-08,
      "loss": 0.3904,
      "step": 3140
    },
    {
      "epoch": 1.9796304322130727,
      "grad_norm": 0.2238065004348755,
      "learning_rate": 1.0821352496269422e-08,
      "loss": 0.3726,
      "step": 3141
    },
    {
      "epoch": 1.9802608250265947,
      "grad_norm": 0.15885622799396515,
      "learning_rate": 1.013451280484773e-08,
      "loss": 0.3599,
      "step": 3142
    },
    {
      "epoch": 1.9808912178401166,
      "grad_norm": 0.1552300751209259,
      "learning_rate": 9.470185797195718e-09,
      "loss": 0.3267,
      "step": 3143
    },
    {
      "epoch": 1.9815216106536386,
      "grad_norm": 0.14502142369747162,
      "learning_rate": 8.828372138313945e-09,
      "loss": 0.3253,
      "step": 3144
    },
    {
      "epoch": 1.9821520034671605,
      "grad_norm": 0.15471288561820984,
      "learning_rate": 8.209072470663493e-09,
      "loss": 0.3686,
      "step": 3145
    },
    {
      "epoch": 1.9827823962806823,
      "grad_norm": 0.20301209390163422,
      "learning_rate": 7.612287414175968e-09,
      "loss": 0.3457,
      "step": 3146
    },
    {
      "epoch": 1.9834127890942044,
      "grad_norm": 0.18437117338180542,
      "learning_rate": 7.038017566238509e-09,
      "loss": 0.3005,
      "step": 3147
    },
    {
      "epoch": 1.9840431819077262,
      "grad_norm": 0.16809889674186707,
      "learning_rate": 6.486263501708778e-09,
      "loss": 0.3795,
      "step": 3148
    },
    {
      "epoch": 1.9846735747212483,
      "grad_norm": 0.17451220750808716,
      "learning_rate": 5.9570257728949754e-09,
      "loss": 0.398,
      "step": 3149
    },
    {
      "epoch": 1.9853039675347701,
      "grad_norm": 0.17353513836860657,
      "learning_rate": 5.450304909575821e-09,
      "loss": 0.3307,
      "step": 3150
    },
    {
      "epoch": 1.985934360348292,
      "grad_norm": 0.2020755410194397,
      "learning_rate": 4.966101418985569e-09,
      "loss": 0.3615,
      "step": 3151
    },
    {
      "epoch": 1.9865647531618138,
      "grad_norm": 0.18863660097122192,
      "learning_rate": 4.504415785816507e-09,
      "loss": 0.3231,
      "step": 3152
    },
    {
      "epoch": 1.9871951459753359,
      "grad_norm": 0.20244933664798737,
      "learning_rate": 4.0652484722264466e-09,
      "loss": 0.3875,
      "step": 3153
    },
    {
      "epoch": 1.987825538788858,
      "grad_norm": 0.14496862888336182,
      "learning_rate": 3.6485999178237363e-09,
      "loss": 0.2825,
      "step": 3154
    },
    {
      "epoch": 1.9884559316023798,
      "grad_norm": 0.17724725604057312,
      "learning_rate": 3.254470539684751e-09,
      "loss": 0.3615,
      "step": 3155
    },
    {
      "epoch": 1.9890863244159016,
      "grad_norm": 0.16778138279914856,
      "learning_rate": 2.882860732336401e-09,
      "loss": 0.3507,
      "step": 3156
    },
    {
      "epoch": 1.9897167172294234,
      "grad_norm": 0.18544743955135345,
      "learning_rate": 2.5337708677636295e-09,
      "loss": 0.3382,
      "step": 3157
    },
    {
      "epoch": 1.9903471100429455,
      "grad_norm": 0.15907777845859528,
      "learning_rate": 2.207201295411909e-09,
      "loss": 0.3352,
      "step": 3158
    },
    {
      "epoch": 1.9909775028564676,
      "grad_norm": 0.1705615222454071,
      "learning_rate": 1.903152342184744e-09,
      "loss": 0.3351,
      "step": 3159
    },
    {
      "epoch": 1.9916078956699894,
      "grad_norm": 0.15151086449623108,
      "learning_rate": 1.6216243124361764e-09,
      "loss": 0.3626,
      "step": 3160
    },
    {
      "epoch": 1.9922382884835113,
      "grad_norm": 0.15104210376739502,
      "learning_rate": 1.3626174879807775e-09,
      "loss": 0.3037,
      "step": 3161
    },
    {
      "epoch": 1.992868681297033,
      "grad_norm": 0.17009414732456207,
      "learning_rate": 1.1261321280886527e-09,
      "loss": 0.3687,
      "step": 3162
    },
    {
      "epoch": 1.9934990741105552,
      "grad_norm": 0.18878397345542908,
      "learning_rate": 9.121684694854415e-10,
      "loss": 0.4226,
      "step": 3163
    },
    {
      "epoch": 1.9941294669240772,
      "grad_norm": 0.1798999160528183,
      "learning_rate": 7.207267263498185e-10,
      "loss": 0.3507,
      "step": 3164
    },
    {
      "epoch": 1.994759859737599,
      "grad_norm": 0.15835590660572052,
      "learning_rate": 5.518070903209882e-10,
      "loss": 0.3359,
      "step": 3165
    },
    {
      "epoch": 1.995390252551121,
      "grad_norm": 0.1508297324180603,
      "learning_rate": 4.054097304886928e-10,
      "loss": 0.3145,
      "step": 3166
    },
    {
      "epoch": 1.9960206453646427,
      "grad_norm": 0.17179661989212036,
      "learning_rate": 2.815347933957102e-10,
      "loss": 0.3899,
      "step": 3167
    },
    {
      "epoch": 1.9966510381781648,
      "grad_norm": 0.16289018094539642,
      "learning_rate": 1.8018240304784606e-10,
      "loss": 0.3098,
      "step": 3168
    },
    {
      "epoch": 1.9972814309916866,
      "grad_norm": 0.1442747563123703,
      "learning_rate": 1.0135266089644768e-10,
      "loss": 0.3264,
      "step": 3169
    },
    {
      "epoch": 1.9979118238052087,
      "grad_norm": 0.17480942606925964,
      "learning_rate": 4.504564585339211e-11,
      "loss": 0.3562,
      "step": 3170
    },
    {
      "epoch": 1.9985422166187305,
      "grad_norm": 0.1754692643880844,
      "learning_rate": 1.1261414281094062e-11,
      "loss": 0.3207,
      "step": 3171
    },
    {
      "epoch": 1.9991726094322524,
      "grad_norm": 0.13669772446155548,
      "learning_rate": 0.0,
      "loss": 0.2699,
      "step": 3172
    }
  ],
  "logging_steps": 1,
  "max_steps": 3172,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.1449881977460634e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}