{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999820939352159,
  "eval_steps": 500,
  "global_step": 27923,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.581212956828478e-05,
      "grad_norm": 10.962271690368652,
      "learning_rate": 2.386634844868735e-07,
      "loss": 3.4992,
      "step": 1
    },
    {
      "epoch": 7.162425913656956e-05,
      "grad_norm": 27.797842025756836,
      "learning_rate": 4.77326968973747e-07,
      "loss": 3.1642,
      "step": 2
    },
    {
      "epoch": 0.00010743638870485433,
      "grad_norm": 9.58411979675293,
      "learning_rate": 7.159904534606206e-07,
      "loss": 3.0653,
      "step": 3
    },
    {
      "epoch": 0.00014324851827313912,
      "grad_norm": 20.217010498046875,
      "learning_rate": 9.54653937947494e-07,
      "loss": 2.3841,
      "step": 4
    },
    {
      "epoch": 0.00017906064784142388,
      "grad_norm": 7.880495071411133,
      "learning_rate": 1.1933174224343676e-06,
      "loss": 2.9288,
      "step": 5
    },
    {
      "epoch": 0.00021487277740970867,
      "grad_norm": 51.29137420654297,
      "learning_rate": 1.4319809069212413e-06,
      "loss": 3.7625,
      "step": 6
    },
    {
      "epoch": 0.00025068490697799346,
      "grad_norm": 8.99178409576416,
      "learning_rate": 1.6706443914081146e-06,
      "loss": 3.156,
      "step": 7
    },
    {
      "epoch": 0.00028649703654627824,
      "grad_norm": 7.409815788269043,
      "learning_rate": 1.909307875894988e-06,
      "loss": 3.0058,
      "step": 8
    },
    {
      "epoch": 0.000322309166114563,
      "grad_norm": 26.714468002319336,
      "learning_rate": 2.1479713603818614e-06,
      "loss": 3.0568,
      "step": 9
    },
    {
      "epoch": 0.00035812129568284776,
      "grad_norm": 9.890267372131348,
      "learning_rate": 2.386634844868735e-06,
      "loss": 2.9411,
      "step": 10
    },
    {
      "epoch": 0.00039393342525113255,
      "grad_norm": 6.088392734527588,
      "learning_rate": 2.625298329355609e-06,
      "loss": 2.6308,
      "step": 11
    },
    {
      "epoch": 0.00042974555481941734,
      "grad_norm": 5.2657365798950195,
      "learning_rate": 2.8639618138424826e-06,
      "loss": 2.3566,
      "step": 12
    },
    {
      "epoch": 0.0004655576843877021,
      "grad_norm": 14.518162727355957,
      "learning_rate": 3.1026252983293554e-06,
      "loss": 3.2695,
      "step": 13
    },
    {
      "epoch": 0.0005013698139559869,
      "grad_norm": 42.87174987792969,
      "learning_rate": 3.341288782816229e-06,
      "loss": 3.3732,
      "step": 14
    },
    {
      "epoch": 0.0005371819435242717,
      "grad_norm": 5.752267837524414,
      "learning_rate": 3.579952267303103e-06,
      "loss": 2.6332,
      "step": 15
    },
    {
      "epoch": 0.0005729940730925565,
      "grad_norm": 5.9656572341918945,
      "learning_rate": 3.818615751789976e-06,
      "loss": 2.5045,
      "step": 16
    },
    {
      "epoch": 0.0006088062026608413,
      "grad_norm": 15.938884735107422,
      "learning_rate": 4.05727923627685e-06,
      "loss": 3.2342,
      "step": 17
    },
    {
      "epoch": 0.000644618332229126,
      "grad_norm": 12.065786361694336,
      "learning_rate": 4.295942720763723e-06,
      "loss": 2.4276,
      "step": 18
    },
    {
      "epoch": 0.0006804304617974107,
      "grad_norm": 9.022994995117188,
      "learning_rate": 4.5346062052505965e-06,
      "loss": 3.5128,
      "step": 19
    },
    {
      "epoch": 0.0007162425913656955,
      "grad_norm": 8.21558952331543,
      "learning_rate": 4.77326968973747e-06,
      "loss": 2.6403,
      "step": 20
    },
    {
      "epoch": 0.0007520547209339803,
      "grad_norm": 6.674815654754639,
      "learning_rate": 5.011933174224344e-06,
      "loss": 2.9362,
      "step": 21
    },
    {
      "epoch": 0.0007878668505022651,
      "grad_norm": 8.615503311157227,
      "learning_rate": 5.250596658711218e-06,
      "loss": 3.4083,
      "step": 22
    },
    {
      "epoch": 0.0008236789800705499,
      "grad_norm": 5.087034225463867,
      "learning_rate": 5.489260143198091e-06,
      "loss": 2.6269,
      "step": 23
    },
    {
      "epoch": 0.0008594911096388347,
      "grad_norm": 5.468980312347412,
      "learning_rate": 5.727923627684965e-06,
      "loss": 2.7802,
      "step": 24
    },
    {
      "epoch": 0.0008953032392071195,
      "grad_norm": 4.205332279205322,
      "learning_rate": 5.966587112171838e-06,
      "loss": 2.6768,
      "step": 25
    },
    {
      "epoch": 0.0009311153687754042,
      "grad_norm": 2.8609237670898438,
      "learning_rate": 6.205250596658711e-06,
      "loss": 2.3405,
      "step": 26
    },
    {
      "epoch": 0.000966927498343689,
      "grad_norm": 4.893277168273926,
      "learning_rate": 6.4439140811455855e-06,
      "loss": 2.9049,
      "step": 27
    },
    {
      "epoch": 0.0010027396279119738,
      "grad_norm": 2.7744317054748535,
      "learning_rate": 6.682577565632458e-06,
      "loss": 2.3621,
      "step": 28
    },
    {
      "epoch": 0.0010385517574802586,
      "grad_norm": 4.931545257568359,
      "learning_rate": 6.921241050119331e-06,
      "loss": 2.8006,
      "step": 29
    },
    {
      "epoch": 0.0010743638870485434,
      "grad_norm": 2.949612855911255,
      "learning_rate": 7.159904534606206e-06,
      "loss": 2.7418,
      "step": 30
    },
    {
      "epoch": 0.0011101760166168282,
      "grad_norm": 6.12894868850708,
      "learning_rate": 7.398568019093079e-06,
      "loss": 3.279,
      "step": 31
    },
    {
      "epoch": 0.001145988146185113,
      "grad_norm": 4.933887004852295,
      "learning_rate": 7.637231503579952e-06,
      "loss": 3.0113,
      "step": 32
    },
    {
      "epoch": 0.0011818002757533978,
      "grad_norm": 3.533989906311035,
      "learning_rate": 7.875894988066825e-06,
      "loss": 2.4149,
      "step": 33
    },
    {
      "epoch": 0.0012176124053216825,
      "grad_norm": 5.942819118499756,
      "learning_rate": 8.1145584725537e-06,
      "loss": 2.7419,
      "step": 34
    },
    {
      "epoch": 0.0012534245348899673,
      "grad_norm": 5.094061374664307,
      "learning_rate": 8.353221957040573e-06,
      "loss": 3.1046,
      "step": 35
    },
    {
      "epoch": 0.001289236664458252,
      "grad_norm": 4.425570487976074,
      "learning_rate": 8.591885441527446e-06,
      "loss": 2.9092,
      "step": 36
    },
    {
      "epoch": 0.0013250487940265367,
      "grad_norm": 3.693305253982544,
      "learning_rate": 8.83054892601432e-06,
      "loss": 2.7984,
      "step": 37
    },
    {
      "epoch": 0.0013608609235948215,
      "grad_norm": 4.119627475738525,
      "learning_rate": 9.069212410501193e-06,
      "loss": 2.4779,
      "step": 38
    },
    {
      "epoch": 0.0013966730531631063,
      "grad_norm": 4.064864158630371,
      "learning_rate": 9.307875894988068e-06,
      "loss": 2.5798,
      "step": 39
    },
    {
      "epoch": 0.001432485182731391,
      "grad_norm": 6.357635974884033,
      "learning_rate": 9.54653937947494e-06,
      "loss": 2.9759,
      "step": 40
    },
    {
      "epoch": 0.0014682973122996758,
      "grad_norm": 4.885557651519775,
      "learning_rate": 9.785202863961815e-06,
      "loss": 2.6862,
      "step": 41
    },
    {
      "epoch": 0.0015041094418679606,
      "grad_norm": 4.694367408752441,
      "learning_rate": 1.0023866348448688e-05,
      "loss": 2.8069,
      "step": 42
    },
    {
      "epoch": 0.0015399215714362454,
      "grad_norm": 6.279577732086182,
      "learning_rate": 1.026252983293556e-05,
      "loss": 2.5923,
      "step": 43
    },
    {
      "epoch": 0.0015757337010045302,
      "grad_norm": 3.011857032775879,
      "learning_rate": 1.0501193317422435e-05,
      "loss": 2.4332,
      "step": 44
    },
    {
      "epoch": 0.001611545830572815,
      "grad_norm": 5.952685832977295,
      "learning_rate": 1.0739856801909308e-05,
      "loss": 3.1646,
      "step": 45
    },
    {
      "epoch": 0.0016473579601410998,
      "grad_norm": 3.988813877105713,
      "learning_rate": 1.0978520286396183e-05,
      "loss": 2.7031,
      "step": 46
    },
    {
      "epoch": 0.0016831700897093846,
      "grad_norm": 4.060389041900635,
      "learning_rate": 1.1217183770883056e-05,
      "loss": 2.7739,
      "step": 47
    },
    {
      "epoch": 0.0017189822192776693,
      "grad_norm": 3.857228994369507,
      "learning_rate": 1.145584725536993e-05,
      "loss": 2.5465,
      "step": 48
    },
    {
      "epoch": 0.0017547943488459541,
      "grad_norm": 2.791111469268799,
      "learning_rate": 1.1694510739856803e-05,
      "loss": 2.465,
      "step": 49
    },
    {
      "epoch": 0.001790606478414239,
      "grad_norm": 3.5531413555145264,
      "learning_rate": 1.1933174224343676e-05,
      "loss": 2.6838,
      "step": 50
    },
    {
      "epoch": 0.0018264186079825237,
      "grad_norm": 3.4963018894195557,
      "learning_rate": 1.2171837708830549e-05,
      "loss": 2.4643,
      "step": 51
    },
    {
      "epoch": 0.0018622307375508085,
      "grad_norm": 2.362746238708496,
      "learning_rate": 1.2410501193317422e-05,
      "loss": 2.2958,
      "step": 52
    },
    {
      "epoch": 0.0018980428671190933,
      "grad_norm": 2.81494140625,
      "learning_rate": 1.2649164677804295e-05,
      "loss": 2.6014,
      "step": 53
    },
    {
      "epoch": 0.001933854996687378,
      "grad_norm": 2.551485061645508,
      "learning_rate": 1.2887828162291171e-05,
      "loss": 2.5691,
      "step": 54
    },
    {
      "epoch": 0.0019696671262556626,
      "grad_norm": 7.383233547210693,
      "learning_rate": 1.3126491646778044e-05,
      "loss": 3.3642,
      "step": 55
    },
    {
      "epoch": 0.0020054792558239476,
      "grad_norm": 6.4138360023498535,
      "learning_rate": 1.3365155131264917e-05,
      "loss": 3.1544,
      "step": 56
    },
    {
      "epoch": 0.002041291385392232,
      "grad_norm": 3.4359352588653564,
      "learning_rate": 1.360381861575179e-05,
      "loss": 2.7802,
      "step": 57
    },
    {
      "epoch": 0.0020771035149605172,
      "grad_norm": 1.671030879020691,
      "learning_rate": 1.3842482100238662e-05,
      "loss": 2.0855,
      "step": 58
    },
    {
      "epoch": 0.002112915644528802,
      "grad_norm": 2.736416816711426,
      "learning_rate": 1.4081145584725539e-05,
      "loss": 2.5001,
      "step": 59
    },
    {
      "epoch": 0.002148727774097087,
      "grad_norm": 2.362579584121704,
      "learning_rate": 1.4319809069212412e-05,
      "loss": 2.5147,
      "step": 60
    },
    {
      "epoch": 0.0021845399036653714,
      "grad_norm": 4.88362455368042,
      "learning_rate": 1.4558472553699284e-05,
      "loss": 2.7335,
      "step": 61
    },
    {
      "epoch": 0.0022203520332336564,
      "grad_norm": 3.396545171737671,
      "learning_rate": 1.4797136038186157e-05,
      "loss": 2.2491,
      "step": 62
    },
    {
      "epoch": 0.002256164162801941,
      "grad_norm": 2.045161008834839,
      "learning_rate": 1.5035799522673034e-05,
      "loss": 2.1031,
      "step": 63
    },
    {
      "epoch": 0.002291976292370226,
      "grad_norm": 2.4404568672180176,
      "learning_rate": 1.5274463007159905e-05,
      "loss": 2.4767,
      "step": 64
    },
    {
      "epoch": 0.0023277884219385105,
      "grad_norm": 1.991403341293335,
      "learning_rate": 1.551312649164678e-05,
      "loss": 2.2476,
      "step": 65
    },
    {
      "epoch": 0.0023636005515067955,
      "grad_norm": 2.6608493328094482,
      "learning_rate": 1.575178997613365e-05,
      "loss": 2.4446,
      "step": 66
    },
    {
      "epoch": 0.00239941268107508,
      "grad_norm": 4.269506454467773,
      "learning_rate": 1.5990453460620525e-05,
      "loss": 2.6558,
      "step": 67
    },
    {
      "epoch": 0.002435224810643365,
      "grad_norm": 2.757368564605713,
      "learning_rate": 1.62291169451074e-05,
      "loss": 2.4896,
      "step": 68
    },
    {
      "epoch": 0.0024710369402116497,
      "grad_norm": 3.7774527072906494,
      "learning_rate": 1.6467780429594274e-05,
      "loss": 2.6252,
      "step": 69
    },
    {
      "epoch": 0.0025068490697799347,
      "grad_norm": 2.459183692932129,
      "learning_rate": 1.6706443914081145e-05,
      "loss": 2.3044,
      "step": 70
    },
    {
      "epoch": 0.0025426611993482192,
      "grad_norm": 1.71919584274292,
      "learning_rate": 1.694510739856802e-05,
      "loss": 2.2353,
      "step": 71
    },
    {
      "epoch": 0.002578473328916504,
      "grad_norm": 2.3338592052459717,
      "learning_rate": 1.718377088305489e-05,
      "loss": 2.3211,
      "step": 72
    },
    {
      "epoch": 0.002614285458484789,
      "grad_norm": 4.527637958526611,
      "learning_rate": 1.742243436754177e-05,
      "loss": 2.5248,
      "step": 73
    },
    {
      "epoch": 0.0026500975880530734,
      "grad_norm": 4.273586273193359,
      "learning_rate": 1.766109785202864e-05,
      "loss": 2.7464,
      "step": 74
    },
    {
      "epoch": 0.0026859097176213584,
      "grad_norm": 2.748335838317871,
      "learning_rate": 1.7899761336515515e-05,
      "loss": 2.4126,
      "step": 75
    },
    {
      "epoch": 0.002721721847189643,
      "grad_norm": 4.273103713989258,
      "learning_rate": 1.8138424821002386e-05,
      "loss": 2.5641,
      "step": 76
    },
    {
      "epoch": 0.002757533976757928,
      "grad_norm": 2.175398826599121,
      "learning_rate": 1.837708830548926e-05,
      "loss": 2.335,
      "step": 77
    },
    {
      "epoch": 0.0027933461063262125,
      "grad_norm": 1.991686463356018,
      "learning_rate": 1.8615751789976135e-05,
      "loss": 2.1709,
      "step": 78
    },
    {
      "epoch": 0.0028291582358944975,
      "grad_norm": 2.3759572505950928,
      "learning_rate": 1.885441527446301e-05,
      "loss": 2.2788,
      "step": 79
    },
    {
      "epoch": 0.002864970365462782,
      "grad_norm": 2.7693066596984863,
      "learning_rate": 1.909307875894988e-05,
      "loss": 2.2775,
      "step": 80
    },
    {
      "epoch": 0.002900782495031067,
      "grad_norm": 3.2281436920166016,
      "learning_rate": 1.9331742243436756e-05,
      "loss": 2.453,
      "step": 81
    },
    {
      "epoch": 0.0029365946245993517,
      "grad_norm": 2.9001846313476562,
      "learning_rate": 1.957040572792363e-05,
      "loss": 2.4301,
      "step": 82
    },
    {
      "epoch": 0.0029724067541676367,
      "grad_norm": 4.701358795166016,
      "learning_rate": 1.98090692124105e-05,
      "loss": 2.3049,
      "step": 83
    },
    {
      "epoch": 0.0030082188837359213,
      "grad_norm": 3.314978837966919,
      "learning_rate": 2.0047732696897376e-05,
      "loss": 2.5192,
      "step": 84
    },
    {
      "epoch": 0.0030440310133042063,
      "grad_norm": 3.295654535293579,
      "learning_rate": 2.0286396181384247e-05,
      "loss": 2.3816,
      "step": 85
    },
    {
      "epoch": 0.003079843142872491,
      "grad_norm": 4.113006591796875,
      "learning_rate": 2.052505966587112e-05,
      "loss": 2.3823,
      "step": 86
    },
    {
      "epoch": 0.003115655272440776,
      "grad_norm": 6.134260654449463,
      "learning_rate": 2.0763723150357996e-05,
      "loss": 2.9283,
      "step": 87
    },
    {
      "epoch": 0.0031514674020090604,
      "grad_norm": 3.5906167030334473,
      "learning_rate": 2.100238663484487e-05,
      "loss": 2.7102,
      "step": 88
    },
    {
      "epoch": 0.0031872795315773454,
      "grad_norm": 2.333726167678833,
      "learning_rate": 2.1241050119331742e-05,
      "loss": 2.0401,
      "step": 89
    },
    {
      "epoch": 0.00322309166114563,
      "grad_norm": 2.3526418209075928,
      "learning_rate": 2.1479713603818617e-05,
      "loss": 2.0289,
      "step": 90
    },
    {
      "epoch": 0.003258903790713915,
      "grad_norm": 4.23220157623291,
      "learning_rate": 2.171837708830549e-05,
      "loss": 2.6101,
      "step": 91
    },
    {
      "epoch": 0.0032947159202821995,
      "grad_norm": 2.601986885070801,
      "learning_rate": 2.1957040572792366e-05,
      "loss": 2.1812,
      "step": 92
    },
    {
      "epoch": 0.0033305280498504846,
      "grad_norm": 3.328922748565674,
      "learning_rate": 2.2195704057279237e-05,
      "loss": 2.4358,
      "step": 93
    },
    {
      "epoch": 0.003366340179418769,
      "grad_norm": 3.1595191955566406,
      "learning_rate": 2.243436754176611e-05,
      "loss": 2.243,
      "step": 94
    },
    {
      "epoch": 0.003402152308987054,
      "grad_norm": 3.6129677295684814,
      "learning_rate": 2.2673031026252983e-05,
      "loss": 1.9055,
      "step": 95
    },
    {
      "epoch": 0.0034379644385553387,
      "grad_norm": 2.5094430446624756,
      "learning_rate": 2.291169451073986e-05,
      "loss": 1.9977,
      "step": 96
    },
    {
      "epoch": 0.0034737765681236233,
      "grad_norm": 5.377303600311279,
      "learning_rate": 2.3150357995226732e-05,
      "loss": 2.5395,
      "step": 97
    },
    {
      "epoch": 0.0035095886976919083,
      "grad_norm": 5.151118755340576,
      "learning_rate": 2.3389021479713606e-05,
      "loss": 2.4771,
      "step": 98
    },
    {
      "epoch": 0.003545400827260193,
      "grad_norm": 3.074028968811035,
      "learning_rate": 2.3627684964200477e-05,
      "loss": 2.1024,
      "step": 99
    },
    {
      "epoch": 0.003581212956828478,
      "grad_norm": 2.7826621532440186,
      "learning_rate": 2.3866348448687352e-05,
      "loss": 2.4725,
      "step": 100
    },
    {
      "epoch": 0.0036170250863967624,
      "grad_norm": 4.709033012390137,
      "learning_rate": 2.4105011933174227e-05,
      "loss": 2.6447,
      "step": 101
    },
    {
      "epoch": 0.0036528372159650474,
      "grad_norm": 5.2452311515808105,
      "learning_rate": 2.4343675417661098e-05,
      "loss": 2.373,
      "step": 102
    },
    {
      "epoch": 0.003688649345533332,
      "grad_norm": 2.4772093296051025,
      "learning_rate": 2.4582338902147972e-05,
      "loss": 2.3534,
      "step": 103
    },
    {
      "epoch": 0.003724461475101617,
      "grad_norm": 4.724703788757324,
      "learning_rate": 2.4821002386634844e-05,
      "loss": 2.6285,
      "step": 104
    },
    {
      "epoch": 0.0037602736046699016,
      "grad_norm": 2.5867512226104736,
      "learning_rate": 2.5059665871121718e-05,
      "loss": 2.2477,
      "step": 105
    },
    {
      "epoch": 0.0037960857342381866,
      "grad_norm": 4.422124862670898,
      "learning_rate": 2.529832935560859e-05,
      "loss": 2.3933,
      "step": 106
    },
    {
      "epoch": 0.003831897863806471,
      "grad_norm": 3.257288932800293,
      "learning_rate": 2.5536992840095464e-05,
      "loss": 1.8866,
      "step": 107
    },
    {
      "epoch": 0.003867709993374756,
      "grad_norm": 3.285341501235962,
      "learning_rate": 2.5775656324582342e-05,
      "loss": 2.3934,
      "step": 108
    },
    {
      "epoch": 0.0039035221229430407,
      "grad_norm": 3.1677615642547607,
      "learning_rate": 2.6014319809069216e-05,
      "loss": 2.4064,
      "step": 109
    },
    {
      "epoch": 0.003939334252511325,
      "grad_norm": 5.5628461837768555,
      "learning_rate": 2.6252983293556088e-05,
      "loss": 2.5679,
      "step": 110
    },
    {
      "epoch": 0.003975146382079611,
      "grad_norm": 3.9429681301116943,
      "learning_rate": 2.6491646778042962e-05,
      "loss": 2.5226,
      "step": 111
    },
    {
      "epoch": 0.004010958511647895,
      "grad_norm": 3.8189384937286377,
      "learning_rate": 2.6730310262529833e-05,
      "loss": 2.3188,
      "step": 112
    },
    {
      "epoch": 0.00404677064121618,
      "grad_norm": 3.60325026512146,
      "learning_rate": 2.6968973747016708e-05,
      "loss": 2.3814,
      "step": 113
    },
    {
      "epoch": 0.004082582770784464,
      "grad_norm": 4.04654598236084,
      "learning_rate": 2.720763723150358e-05,
      "loss": 2.0409,
      "step": 114
    },
    {
      "epoch": 0.00411839490035275,
      "grad_norm": 2.3440001010894775,
      "learning_rate": 2.7446300715990454e-05,
      "loss": 2.1214,
      "step": 115
    },
    {
      "epoch": 0.0041542070299210344,
      "grad_norm": 2.3456714153289795,
      "learning_rate": 2.7684964200477325e-05,
      "loss": 1.7589,
      "step": 116
    },
    {
      "epoch": 0.004190019159489319,
      "grad_norm": 3.4626646041870117,
      "learning_rate": 2.7923627684964203e-05,
      "loss": 2.4336,
      "step": 117
    },
    {
      "epoch": 0.004225831289057604,
      "grad_norm": 2.838393211364746,
      "learning_rate": 2.8162291169451077e-05,
      "loss": 1.9198,
      "step": 118
    },
    {
      "epoch": 0.004261643418625889,
      "grad_norm": 3.0637154579162598,
      "learning_rate": 2.840095465393795e-05,
      "loss": 2.0948,
      "step": 119
    },
    {
      "epoch": 0.004297455548194174,
      "grad_norm": 3.7256317138671875,
      "learning_rate": 2.8639618138424823e-05,
      "loss": 2.4361,
      "step": 120
    },
    {
      "epoch": 0.004333267677762458,
      "grad_norm": 3.2016754150390625,
      "learning_rate": 2.8878281622911694e-05,
      "loss": 2.1261,
      "step": 121
    },
    {
      "epoch": 0.004369079807330743,
      "grad_norm": 4.712606906890869,
      "learning_rate": 2.911694510739857e-05,
      "loss": 2.0613,
      "step": 122
    },
    {
      "epoch": 0.004404891936899027,
      "grad_norm": 3.5016684532165527,
      "learning_rate": 2.935560859188544e-05,
      "loss": 2.1061,
      "step": 123
    },
    {
      "epoch": 0.004440704066467313,
      "grad_norm": 2.8278017044067383,
      "learning_rate": 2.9594272076372315e-05,
      "loss": 2.5463,
      "step": 124
    },
    {
      "epoch": 0.004476516196035597,
      "grad_norm": 3.4511027336120605,
      "learning_rate": 2.983293556085919e-05,
      "loss": 2.4129,
      "step": 125
    },
    {
      "epoch": 0.004512328325603882,
      "grad_norm": 2.2344186305999756,
      "learning_rate": 3.0071599045346067e-05,
      "loss": 2.1861,
      "step": 126
    },
    {
      "epoch": 0.0045481404551721664,
      "grad_norm": 2.4487826824188232,
      "learning_rate": 3.031026252983294e-05,
      "loss": 1.985,
      "step": 127
    },
    {
      "epoch": 0.004583952584740452,
      "grad_norm": 3.2509713172912598,
      "learning_rate": 3.054892601431981e-05,
      "loss": 2.2934,
      "step": 128
    },
    {
      "epoch": 0.0046197647143087365,
      "grad_norm": 1.7187750339508057,
      "learning_rate": 3.0787589498806684e-05,
      "loss": 1.9708,
      "step": 129
    },
    {
      "epoch": 0.004655576843877021,
      "grad_norm": 5.076966285705566,
      "learning_rate": 3.102625298329356e-05,
      "loss": 2.6119,
      "step": 130
    },
    {
      "epoch": 0.004691388973445306,
      "grad_norm": 3.493896245956421,
      "learning_rate": 3.126491646778043e-05,
      "loss": 2.5397,
      "step": 131
    },
    {
      "epoch": 0.004727201103013591,
      "grad_norm": 2.5739636421203613,
      "learning_rate": 3.15035799522673e-05,
      "loss": 2.1216,
      "step": 132
    },
    {
      "epoch": 0.004763013232581876,
      "grad_norm": 2.577495574951172,
      "learning_rate": 3.1742243436754176e-05,
      "loss": 2.1758,
      "step": 133
    },
    {
      "epoch": 0.00479882536215016,
      "grad_norm": 2.5767970085144043,
      "learning_rate": 3.198090692124105e-05,
      "loss": 2.2066,
      "step": 134
    },
    {
      "epoch": 0.004834637491718445,
      "grad_norm": 3.6485743522644043,
      "learning_rate": 3.2219570405727925e-05,
      "loss": 2.4,
      "step": 135
    },
    {
      "epoch": 0.00487044962128673,
      "grad_norm": 3.2032318115234375,
      "learning_rate": 3.24582338902148e-05,
      "loss": 2.323,
      "step": 136
    },
    {
      "epoch": 0.004906261750855015,
      "grad_norm": 3.5218751430511475,
      "learning_rate": 3.2696897374701674e-05,
      "loss": 2.1281,
      "step": 137
    },
    {
      "epoch": 0.004942073880423299,
      "grad_norm": 3.35833477973938,
      "learning_rate": 3.293556085918855e-05,
      "loss": 2.3306,
      "step": 138
    },
    {
      "epoch": 0.004977886009991584,
      "grad_norm": 2.5601823329925537,
      "learning_rate": 3.3174224343675416e-05,
      "loss": 2.0496,
      "step": 139
    },
    {
      "epoch": 0.005013698139559869,
      "grad_norm": 5.0960373878479,
      "learning_rate": 3.341288782816229e-05,
      "loss": 2.6784,
      "step": 140
    },
    {
      "epoch": 0.005049510269128154,
      "grad_norm": 3.276700973510742,
      "learning_rate": 3.3651551312649165e-05,
      "loss": 2.0709,
      "step": 141
    },
    {
      "epoch": 0.0050853223986964385,
      "grad_norm": 3.034759283065796,
      "learning_rate": 3.389021479713604e-05,
      "loss": 2.1616,
      "step": 142
    },
    {
      "epoch": 0.005121134528264723,
      "grad_norm": 3.5838255882263184,
      "learning_rate": 3.4128878281622915e-05,
      "loss": 2.5447,
      "step": 143
    },
    {
      "epoch": 0.005156946657833008,
      "grad_norm": 3.239966630935669,
      "learning_rate": 3.436754176610978e-05,
      "loss": 2.4755,
      "step": 144
    },
    {
      "epoch": 0.005192758787401293,
      "grad_norm": 3.4092092514038086,
      "learning_rate": 3.4606205250596664e-05,
      "loss": 2.2712,
      "step": 145
    },
    {
      "epoch": 0.005228570916969578,
      "grad_norm": 2.1897785663604736,
      "learning_rate": 3.484486873508354e-05,
      "loss": 2.1974,
      "step": 146
    },
    {
      "epoch": 0.005264383046537862,
      "grad_norm": 3.396059513092041,
      "learning_rate": 3.5083532219570406e-05,
      "loss": 2.4299,
      "step": 147
    },
    {
      "epoch": 0.005300195176106147,
      "grad_norm": 4.3964104652404785,
      "learning_rate": 3.532219570405728e-05,
      "loss": 2.2913,
      "step": 148
    },
    {
      "epoch": 0.005336007305674432,
      "grad_norm": 2.8789312839508057,
      "learning_rate": 3.5560859188544155e-05,
      "loss": 1.6942,
      "step": 149
    },
    {
      "epoch": 0.005371819435242717,
      "grad_norm": 1.7784701585769653,
      "learning_rate": 3.579952267303103e-05,
      "loss": 1.9099,
      "step": 150
    },
    {
      "epoch": 0.005407631564811001,
      "grad_norm": 3.5115861892700195,
      "learning_rate": 3.60381861575179e-05,
      "loss": 2.4334,
      "step": 151
    },
    {
      "epoch": 0.005443443694379286,
      "grad_norm": 2.5266120433807373,
      "learning_rate": 3.627684964200477e-05,
      "loss": 2.1036,
      "step": 152
    },
    {
      "epoch": 0.005479255823947571,
      "grad_norm": 2.6542105674743652,
      "learning_rate": 3.651551312649165e-05,
      "loss": 2.1469,
      "step": 153
    },
    {
      "epoch": 0.005515067953515856,
      "grad_norm": 6.56311559677124,
      "learning_rate": 3.675417661097852e-05,
      "loss": 2.9395,
      "step": 154
    },
    {
      "epoch": 0.0055508800830841405,
      "grad_norm": 2.516366720199585,
      "learning_rate": 3.6992840095465396e-05,
      "loss": 1.9398,
      "step": 155
    },
    {
      "epoch": 0.005586692212652425,
      "grad_norm": 2.376039981842041,
      "learning_rate": 3.723150357995227e-05,
      "loss": 2.1975,
      "step": 156
    },
    {
      "epoch": 0.0056225043422207105,
      "grad_norm": 3.064851760864258,
      "learning_rate": 3.7470167064439145e-05,
      "loss": 2.2792,
      "step": 157
    },
    {
      "epoch": 0.005658316471788995,
      "grad_norm": 3.0628392696380615,
      "learning_rate": 3.770883054892602e-05,
      "loss": 2.1432,
      "step": 158
    },
    {
      "epoch": 0.00569412860135728,
      "grad_norm": 4.342270851135254,
      "learning_rate": 3.794749403341289e-05,
      "loss": 2.2726,
      "step": 159
    },
    {
      "epoch": 0.005729940730925564,
      "grad_norm": 3.9450583457946777,
      "learning_rate": 3.818615751789976e-05,
      "loss": 2.2563,
      "step": 160
    },
    {
      "epoch": 0.00576575286049385,
      "grad_norm": 2.65295672416687,
      "learning_rate": 3.8424821002386637e-05,
      "loss": 2.279,
      "step": 161
    },
    {
      "epoch": 0.005801564990062134,
      "grad_norm": 3.084883451461792,
      "learning_rate": 3.866348448687351e-05,
      "loss": 2.347,
      "step": 162
    },
    {
      "epoch": 0.005837377119630419,
      "grad_norm": 2.0659403800964355,
      "learning_rate": 3.8902147971360386e-05,
      "loss": 1.982,
      "step": 163
    },
    {
      "epoch": 0.005873189249198703,
      "grad_norm": 2.8707988262176514,
      "learning_rate": 3.914081145584726e-05,
      "loss": 2.2617,
      "step": 164
    },
    {
      "epoch": 0.005909001378766989,
      "grad_norm": 2.436180353164673,
      "learning_rate": 3.9379474940334135e-05,
      "loss": 2.0672,
      "step": 165
    },
    {
      "epoch": 0.005944813508335273,
      "grad_norm": 3.461872100830078,
      "learning_rate": 3.9618138424821e-05,
      "loss": 2.024,
      "step": 166
    },
    {
      "epoch": 0.005980625637903558,
      "grad_norm": 2.2728688716888428,
      "learning_rate": 3.985680190930788e-05,
      "loss": 1.7771,
      "step": 167
    },
    {
      "epoch": 0.0060164377674718425,
      "grad_norm": 2.9046618938446045,
      "learning_rate": 4.009546539379475e-05,
      "loss": 2.1021,
      "step": 168
    },
    {
      "epoch": 0.006052249897040127,
      "grad_norm": 3.4231221675872803,
      "learning_rate": 4.0334128878281626e-05,
      "loss": 2.4282,
      "step": 169
    },
    {
      "epoch": 0.0060880620266084125,
      "grad_norm": 2.6314404010772705,
      "learning_rate": 4.0572792362768494e-05,
      "loss": 2.5756,
      "step": 170
    },
    {
      "epoch": 0.006123874156176697,
      "grad_norm": 1.9490439891815186,
      "learning_rate": 4.081145584725537e-05,
      "loss": 1.9626,
      "step": 171
    },
    {
      "epoch": 0.006159686285744982,
      "grad_norm": 2.614915370941162,
      "learning_rate": 4.105011933174224e-05,
      "loss": 2.0636,
      "step": 172
    },
    {
      "epoch": 0.006195498415313266,
      "grad_norm": 2.4821505546569824,
      "learning_rate": 4.1288782816229125e-05,
      "loss": 2.2238,
      "step": 173
    },
    {
      "epoch": 0.006231310544881552,
      "grad_norm": 3.056910991668701,
      "learning_rate": 4.152744630071599e-05,
      "loss": 1.9857,
      "step": 174
    },
    {
      "epoch": 0.006267122674449836,
      "grad_norm": 3.037914514541626,
      "learning_rate": 4.176610978520287e-05,
      "loss": 2.0045,
      "step": 175
    },
    {
      "epoch": 0.006302934804018121,
      "grad_norm": 2.768630266189575,
      "learning_rate": 4.200477326968974e-05,
      "loss": 2.2729,
      "step": 176
    },
    {
      "epoch": 0.006338746933586405,
      "grad_norm": 6.603919506072998,
      "learning_rate": 4.2243436754176616e-05,
      "loss": 2.4431,
      "step": 177
    },
    {
      "epoch": 0.006374559063154691,
      "grad_norm": 2.241738796234131,
      "learning_rate": 4.2482100238663484e-05,
      "loss": 2.1826,
      "step": 178
    },
    {
      "epoch": 0.006410371192722975,
      "grad_norm": 2.1560134887695312,
      "learning_rate": 4.272076372315036e-05,
      "loss": 2.041,
      "step": 179
    },
    {
      "epoch": 0.00644618332229126,
      "grad_norm": 2.806037187576294,
      "learning_rate": 4.295942720763723e-05,
      "loss": 2.1842,
      "step": 180
    },
    {
      "epoch": 0.0064819954518595445,
      "grad_norm": 3.3710579872131348,
      "learning_rate": 4.319809069212411e-05,
      "loss": 1.8961,
      "step": 181
    },
    {
      "epoch": 0.00651780758142783,
      "grad_norm": 2.774545192718506,
      "learning_rate": 4.343675417661098e-05,
      "loss": 2.1378,
      "step": 182
    },
    {
      "epoch": 0.0065536197109961145,
      "grad_norm": 4.045608043670654,
      "learning_rate": 4.367541766109786e-05,
      "loss": 2.3613,
      "step": 183
    },
    {
      "epoch": 0.006589431840564399,
      "grad_norm": 4.307875633239746,
      "learning_rate": 4.391408114558473e-05,
      "loss": 2.5546,
      "step": 184
    },
    {
      "epoch": 0.006625243970132684,
      "grad_norm": 2.564769744873047,
      "learning_rate": 4.41527446300716e-05,
      "loss": 2.0365,
      "step": 185
    },
    {
      "epoch": 0.006661056099700969,
      "grad_norm": 3.592845916748047,
      "learning_rate": 4.4391408114558474e-05,
      "loss": 2.0478,
      "step": 186
    },
    {
      "epoch": 0.006696868229269254,
      "grad_norm": 2.4689202308654785,
      "learning_rate": 4.463007159904535e-05,
      "loss": 2.099,
      "step": 187
    },
    {
      "epoch": 0.006732680358837538,
      "grad_norm": 2.8406245708465576,
      "learning_rate": 4.486873508353222e-05,
      "loss": 1.972,
      "step": 188
    },
    {
      "epoch": 0.006768492488405823,
      "grad_norm": 3.0729517936706543,
      "learning_rate": 4.510739856801909e-05,
      "loss": 2.3021,
      "step": 189
    },
    {
      "epoch": 0.006804304617974108,
      "grad_norm": 2.427234649658203,
      "learning_rate": 4.5346062052505965e-05,
      "loss": 2.1842,
      "step": 190
    },
    {
      "epoch": 0.006840116747542393,
      "grad_norm": 2.7807438373565674,
      "learning_rate": 4.5584725536992847e-05,
      "loss": 2.2948,
      "step": 191
    },
    {
      "epoch": 0.006875928877110677,
      "grad_norm": 3.589139699935913,
      "learning_rate": 4.582338902147972e-05,
      "loss": 2.3455,
      "step": 192
    },
    {
      "epoch": 0.006911741006678962,
      "grad_norm": 3.561444044113159,
      "learning_rate": 4.606205250596659e-05,
      "loss": 1.9343,
      "step": 193
    },
    {
      "epoch": 0.0069475531362472465,
      "grad_norm": 2.3971855640411377,
      "learning_rate": 4.6300715990453463e-05,
      "loss": 2.144,
      "step": 194
    },
    {
      "epoch": 0.006983365265815532,
      "grad_norm": 1.9098275899887085,
      "learning_rate": 4.653937947494034e-05,
      "loss": 1.8509,
      "step": 195
    },
    {
      "epoch": 0.0070191773953838165,
      "grad_norm": 2.244335412979126,
      "learning_rate": 4.677804295942721e-05,
      "loss": 1.922,
      "step": 196
    },
    {
      "epoch": 0.007054989524952101,
      "grad_norm": 2.8799850940704346,
      "learning_rate": 4.701670644391408e-05,
      "loss": 1.9339,
      "step": 197
    },
    {
      "epoch": 0.007090801654520386,
      "grad_norm": 2.483138084411621,
      "learning_rate": 4.7255369928400955e-05,
      "loss": 2.0267,
      "step": 198
    },
    {
      "epoch": 0.007126613784088671,
      "grad_norm": 4.479922294616699,
      "learning_rate": 4.749403341288783e-05,
      "loss": 2.345,
      "step": 199
    },
    {
      "epoch": 0.007162425913656956,
      "grad_norm": 2.4808027744293213,
      "learning_rate": 4.7732696897374704e-05,
      "loss": 2.1023,
      "step": 200
    },
    {
      "epoch": 0.00719823804322524,
      "grad_norm": 3.2008464336395264,
      "learning_rate": 4.797136038186158e-05,
      "loss": 2.1962,
      "step": 201
    },
    {
      "epoch": 0.007234050172793525,
      "grad_norm": 3.3301191329956055,
      "learning_rate": 4.821002386634845e-05,
      "loss": 2.0843,
      "step": 202
    },
    {
      "epoch": 0.00726986230236181,
      "grad_norm": 2.2607581615448,
      "learning_rate": 4.844868735083533e-05,
      "loss": 1.812,
      "step": 203
    },
    {
      "epoch": 0.007305674431930095,
      "grad_norm": 2.308560371398926,
      "learning_rate": 4.8687350835322196e-05,
      "loss": 2.1753,
      "step": 204
    },
    {
      "epoch": 0.007341486561498379,
      "grad_norm": 1.8823864459991455,
      "learning_rate": 4.892601431980907e-05,
      "loss": 2.0771,
      "step": 205
    },
    {
      "epoch": 0.007377298691066664,
      "grad_norm": 2.2979161739349365,
      "learning_rate": 4.9164677804295945e-05,
      "loss": 2.1303,
      "step": 206
    },
    {
      "epoch": 0.007413110820634949,
      "grad_norm": 2.1987791061401367,
      "learning_rate": 4.940334128878282e-05,
      "loss": 2.1973,
      "step": 207
    },
    {
      "epoch": 0.007448922950203234,
      "grad_norm": 2.7911312580108643,
      "learning_rate": 4.964200477326969e-05,
      "loss": 2.2152,
      "step": 208
    },
    {
      "epoch": 0.0074847350797715186,
      "grad_norm": 4.45490837097168,
      "learning_rate": 4.988066825775656e-05,
      "loss": 2.0745,
      "step": 209
    },
    {
      "epoch": 0.007520547209339803,
      "grad_norm": 3.342639684677124,
      "learning_rate": 5.0119331742243436e-05,
      "loss": 2.3235,
      "step": 210
    },
    {
      "epoch": 0.0075563593389080886,
      "grad_norm": 2.095116376876831,
      "learning_rate": 5.035799522673032e-05,
      "loss": 1.9779,
      "step": 211
    },
    {
      "epoch": 0.007592171468476373,
      "grad_norm": 2.7536773681640625,
      "learning_rate": 5.059665871121718e-05,
      "loss": 2.11,
      "step": 212
    },
    {
      "epoch": 0.007627983598044658,
      "grad_norm": 2.3608434200286865,
      "learning_rate": 5.083532219570406e-05,
      "loss": 2.0265,
      "step": 213
    },
    {
      "epoch": 0.007663795727612942,
      "grad_norm": 2.917818307876587,
      "learning_rate": 5.107398568019093e-05,
      "loss": 1.9071,
      "step": 214
    },
    {
      "epoch": 0.007699607857181228,
      "grad_norm": 3.007936716079712,
      "learning_rate": 5.131264916467781e-05,
      "loss": 2.0646,
      "step": 215
    },
    {
      "epoch": 0.007735419986749512,
      "grad_norm": 2.02176833152771,
      "learning_rate": 5.1551312649164684e-05,
      "loss": 1.9931,
      "step": 216
    },
    {
      "epoch": 0.007771232116317797,
      "grad_norm": 2.787811040878296,
      "learning_rate": 5.178997613365155e-05,
      "loss": 2.1347,
      "step": 217
    },
    {
      "epoch": 0.007807044245886081,
      "grad_norm": 3.3416104316711426,
      "learning_rate": 5.202863961813843e-05,
      "loss": 2.1722,
      "step": 218
    },
    {
      "epoch": 0.007842856375454366,
      "grad_norm": 3.348334312438965,
      "learning_rate": 5.22673031026253e-05,
      "loss": 2.2807,
      "step": 219
    },
    {
      "epoch": 0.00787866850502265,
      "grad_norm": 5.854157447814941,
      "learning_rate": 5.2505966587112175e-05,
      "loss": 2.7781,
      "step": 220
    },
    {
      "epoch": 0.007914480634590935,
      "grad_norm": 2.1232118606567383,
      "learning_rate": 5.274463007159904e-05,
      "loss": 2.0395,
      "step": 221
    },
    {
      "epoch": 0.007950292764159221,
      "grad_norm": 3.457007646560669,
      "learning_rate": 5.2983293556085924e-05,
      "loss": 2.2375,
      "step": 222
    },
    {
      "epoch": 0.007986104893727506,
      "grad_norm": 3.5690152645111084,
      "learning_rate": 5.322195704057279e-05,
      "loss": 2.5559,
      "step": 223
    },
    {
      "epoch": 0.00802191702329579,
      "grad_norm": 2.673306941986084,
      "learning_rate": 5.346062052505967e-05,
      "loss": 2.2094,
      "step": 224
    },
    {
      "epoch": 0.008057729152864075,
      "grad_norm": 2.7809441089630127,
      "learning_rate": 5.369928400954655e-05,
      "loss": 1.9878,
      "step": 225
    },
    {
      "epoch": 0.00809354128243236,
      "grad_norm": 3.0256967544555664,
      "learning_rate": 5.3937947494033416e-05,
      "loss": 1.8963,
      "step": 226
    },
    {
      "epoch": 0.008129353412000644,
      "grad_norm": 3.9648005962371826,
      "learning_rate": 5.417661097852029e-05,
      "loss": 2.0862,
      "step": 227
    },
    {
      "epoch": 0.008165165541568929,
      "grad_norm": 2.245849609375,
      "learning_rate": 5.441527446300716e-05,
      "loss": 1.7293,
      "step": 228
    },
    {
      "epoch": 0.008200977671137213,
      "grad_norm": 2.413544178009033,
      "learning_rate": 5.465393794749404e-05,
      "loss": 2.0762,
      "step": 229
    },
    {
      "epoch": 0.0082367898007055,
      "grad_norm": 3.0067107677459717,
      "learning_rate": 5.489260143198091e-05,
      "loss": 2.1954,
      "step": 230
    },
    {
      "epoch": 0.008272601930273784,
      "grad_norm": 3.096357583999634,
      "learning_rate": 5.513126491646778e-05,
      "loss": 2.2491,
      "step": 231
    },
    {
      "epoch": 0.008308414059842069,
      "grad_norm": 2.373617172241211,
      "learning_rate": 5.536992840095465e-05,
      "loss": 2.1228,
      "step": 232
    },
    {
      "epoch": 0.008344226189410353,
      "grad_norm": 2.055159091949463,
      "learning_rate": 5.560859188544153e-05,
      "loss": 1.9121,
      "step": 233
    },
    {
      "epoch": 0.008380038318978638,
      "grad_norm": 3.7004029750823975,
      "learning_rate": 5.5847255369928406e-05,
      "loss": 2.3244,
      "step": 234
    },
    {
      "epoch": 0.008415850448546923,
      "grad_norm": 2.1520016193389893,
      "learning_rate": 5.6085918854415273e-05,
      "loss": 2.1335,
      "step": 235
    },
    {
      "epoch": 0.008451662578115207,
      "grad_norm": 4.440451622009277,
      "learning_rate": 5.6324582338902155e-05,
      "loss": 2.6448,
      "step": 236
    },
    {
      "epoch": 0.008487474707683492,
      "grad_norm": 1.7631467580795288,
      "learning_rate": 5.656324582338902e-05,
      "loss": 2.0462,
      "step": 237
    },
    {
      "epoch": 0.008523286837251778,
      "grad_norm": 3.681589365005493,
      "learning_rate": 5.68019093078759e-05,
      "loss": 2.3295,
      "step": 238
    },
    {
      "epoch": 0.008559098966820063,
      "grad_norm": 2.199773073196411,
      "learning_rate": 5.7040572792362765e-05,
      "loss": 1.9294,
      "step": 239
    },
    {
      "epoch": 0.008594911096388347,
      "grad_norm": 2.9512124061584473,
      "learning_rate": 5.7279236276849646e-05,
      "loss": 1.9352,
      "step": 240
    },
    {
      "epoch": 0.008630723225956632,
      "grad_norm": 2.6109206676483154,
      "learning_rate": 5.7517899761336514e-05,
      "loss": 2.1316,
      "step": 241
    },
    {
      "epoch": 0.008666535355524916,
      "grad_norm": 2.3685696125030518,
      "learning_rate": 5.775656324582339e-05,
      "loss": 2.1921,
      "step": 242
    },
    {
      "epoch": 0.008702347485093201,
      "grad_norm": 3.183547019958496,
      "learning_rate": 5.799522673031027e-05,
      "loss": 2.0476,
      "step": 243
    },
    {
      "epoch": 0.008738159614661485,
      "grad_norm": 2.3643414974212646,
      "learning_rate": 5.823389021479714e-05,
      "loss": 2.3041,
      "step": 244
    },
    {
      "epoch": 0.00877397174422977,
      "grad_norm": 2.646777868270874,
      "learning_rate": 5.847255369928402e-05,
      "loss": 1.8089,
      "step": 245
    },
    {
      "epoch": 0.008809783873798055,
      "grad_norm": 1.8658462762832642,
      "learning_rate": 5.871121718377088e-05,
      "loss": 2.2097,
      "step": 246
    },
    {
      "epoch": 0.008845596003366341,
      "grad_norm": 3.2187111377716064,
      "learning_rate": 5.894988066825776e-05,
      "loss": 2.3646,
      "step": 247
    },
    {
      "epoch": 0.008881408132934625,
      "grad_norm": 2.9397201538085938,
      "learning_rate": 5.918854415274463e-05,
      "loss": 1.8528,
      "step": 248
    },
    {
      "epoch": 0.00891722026250291,
      "grad_norm": 2.3005692958831787,
      "learning_rate": 5.942720763723151e-05,
      "loss": 1.6524,
      "step": 249
    },
    {
      "epoch": 0.008953032392071195,
      "grad_norm": 2.3031492233276367,
      "learning_rate": 5.966587112171838e-05,
      "loss": 1.9533,
      "step": 250
    },
    {
      "epoch": 0.00898884452163948,
      "grad_norm": 1.9477695226669312,
      "learning_rate": 5.990453460620525e-05,
      "loss": 1.8504,
      "step": 251
    },
    {
      "epoch": 0.009024656651207764,
      "grad_norm": 2.6072261333465576,
      "learning_rate": 6.0143198090692134e-05,
      "loss": 2.2766,
      "step": 252
    },
    {
      "epoch": 0.009060468780776048,
      "grad_norm": 1.8509987592697144,
      "learning_rate": 6.0381861575179e-05,
      "loss": 1.9699,
      "step": 253
    },
    {
      "epoch": 0.009096280910344333,
      "grad_norm": 2.4382662773132324,
      "learning_rate": 6.062052505966588e-05,
      "loss": 2.0657,
      "step": 254
    },
    {
      "epoch": 0.00913209303991262,
      "grad_norm": 2.3414926528930664,
      "learning_rate": 6.0859188544152745e-05,
      "loss": 2.0522,
      "step": 255
    },
    {
      "epoch": 0.009167905169480904,
      "grad_norm": 2.583178997039795,
      "learning_rate": 6.109785202863962e-05,
      "loss": 2.0446,
      "step": 256
    },
    {
      "epoch": 0.009203717299049188,
      "grad_norm": 4.623299598693848,
      "learning_rate": 6.133651551312649e-05,
      "loss": 2.2111,
      "step": 257
    },
    {
      "epoch": 0.009239529428617473,
      "grad_norm": 2.9717419147491455,
      "learning_rate": 6.157517899761337e-05,
      "loss": 2.0214,
      "step": 258
    },
    {
      "epoch": 0.009275341558185757,
      "grad_norm": 2.3703761100769043,
      "learning_rate": 6.181384248210024e-05,
      "loss": 2.0451,
      "step": 259
    },
    {
      "epoch": 0.009311153687754042,
      "grad_norm": 2.0671181678771973,
      "learning_rate": 6.205250596658712e-05,
      "loss": 1.9399,
      "step": 260
    },
    {
      "epoch": 0.009346965817322327,
      "grad_norm": 2.3600852489471436,
      "learning_rate": 6.2291169451074e-05,
      "loss": 1.9703,
      "step": 261
    },
    {
      "epoch": 0.009382777946890611,
      "grad_norm": 1.8281196355819702,
      "learning_rate": 6.252983293556087e-05,
      "loss": 1.9562,
      "step": 262
    },
    {
      "epoch": 0.009418590076458897,
      "grad_norm": 3.4217169284820557,
      "learning_rate": 6.276849642004773e-05,
      "loss": 2.1296,
      "step": 263
    },
    {
      "epoch": 0.009454402206027182,
      "grad_norm": 5.549176216125488,
      "learning_rate": 6.30071599045346e-05,
      "loss": 2.3743,
      "step": 264
    },
    {
      "epoch": 0.009490214335595467,
      "grad_norm": 1.707136869430542,
      "learning_rate": 6.324582338902148e-05,
      "loss": 1.9212,
      "step": 265
    },
    {
      "epoch": 0.009526026465163751,
      "grad_norm": 2.4918057918548584,
      "learning_rate": 6.348448687350835e-05,
      "loss": 2.0175,
      "step": 266
    },
    {
      "epoch": 0.009561838594732036,
      "grad_norm": 2.368617057800293,
      "learning_rate": 6.372315035799523e-05,
      "loss": 1.8226,
      "step": 267
    },
    {
      "epoch": 0.00959765072430032,
      "grad_norm": 2.9346976280212402,
      "learning_rate": 6.39618138424821e-05,
      "loss": 2.3414,
      "step": 268
    },
    {
      "epoch": 0.009633462853868605,
      "grad_norm": 2.7401106357574463,
      "learning_rate": 6.420047732696898e-05,
      "loss": 1.8663,
      "step": 269
    },
    {
      "epoch": 0.00966927498343689,
      "grad_norm": 2.8707168102264404,
      "learning_rate": 6.443914081145585e-05,
      "loss": 2.2932,
      "step": 270
    },
    {
      "epoch": 0.009705087113005174,
      "grad_norm": 2.1876916885375977,
      "learning_rate": 6.467780429594272e-05,
      "loss": 1.986,
      "step": 271
    },
    {
      "epoch": 0.00974089924257346,
      "grad_norm": 2.1263930797576904,
      "learning_rate": 6.49164677804296e-05,
      "loss": 1.8861,
      "step": 272
    },
    {
      "epoch": 0.009776711372141745,
      "grad_norm": 1.8206520080566406,
      "learning_rate": 6.515513126491647e-05,
      "loss": 1.8851,
      "step": 273
    },
    {
      "epoch": 0.00981252350171003,
      "grad_norm": 2.8688876628875732,
      "learning_rate": 6.539379474940335e-05,
      "loss": 1.8439,
      "step": 274
    },
    {
      "epoch": 0.009848335631278314,
      "grad_norm": 2.3246121406555176,
      "learning_rate": 6.563245823389022e-05,
      "loss": 2.1826,
      "step": 275
    },
    {
      "epoch": 0.009884147760846599,
      "grad_norm": 2.6065704822540283,
      "learning_rate": 6.58711217183771e-05,
      "loss": 2.229,
      "step": 276
    },
    {
      "epoch": 0.009919959890414883,
      "grad_norm": 2.4256670475006104,
      "learning_rate": 6.610978520286396e-05,
      "loss": 2.0945,
      "step": 277
    },
    {
      "epoch": 0.009955772019983168,
      "grad_norm": 1.7812882661819458,
      "learning_rate": 6.634844868735083e-05,
      "loss": 2.255,
      "step": 278
    },
    {
      "epoch": 0.009991584149551452,
      "grad_norm": 1.8207054138183594,
      "learning_rate": 6.65871121718377e-05,
      "loss": 1.6961,
      "step": 279
    },
    {
      "epoch": 0.010027396279119739,
      "grad_norm": 1.6386990547180176,
      "learning_rate": 6.682577565632458e-05,
      "loss": 1.856,
      "step": 280
    },
    {
      "epoch": 0.010063208408688023,
      "grad_norm": 3.287531614303589,
      "learning_rate": 6.706443914081146e-05,
      "loss": 2.3542,
      "step": 281
    },
    {
      "epoch": 0.010099020538256308,
      "grad_norm": 2.7156126499176025,
      "learning_rate": 6.730310262529833e-05,
      "loss": 1.8437,
      "step": 282
    },
    {
      "epoch": 0.010134832667824592,
      "grad_norm": 2.3014869689941406,
      "learning_rate": 6.754176610978521e-05,
      "loss": 2.447,
      "step": 283
    },
    {
      "epoch": 0.010170644797392877,
      "grad_norm": 2.7459521293640137,
      "learning_rate": 6.778042959427208e-05,
      "loss": 2.2216,
      "step": 284
    },
    {
      "epoch": 0.010206456926961162,
      "grad_norm": 2.640103816986084,
      "learning_rate": 6.801909307875896e-05,
      "loss": 2.0094,
      "step": 285
    },
    {
      "epoch": 0.010242269056529446,
      "grad_norm": 3.3041248321533203,
      "learning_rate": 6.825775656324583e-05,
      "loss": 2.6193,
      "step": 286
    },
    {
      "epoch": 0.01027808118609773,
      "grad_norm": 2.390202522277832,
      "learning_rate": 6.84964200477327e-05,
      "loss": 2.4139,
      "step": 287
    },
    {
      "epoch": 0.010313893315666015,
      "grad_norm": 2.879079580307007,
      "learning_rate": 6.873508353221956e-05,
      "loss": 2.1419,
      "step": 288
    },
    {
      "epoch": 0.010349705445234302,
      "grad_norm": 5.001767635345459,
      "learning_rate": 6.897374701670645e-05,
      "loss": 2.108,
      "step": 289
    },
    {
      "epoch": 0.010385517574802586,
      "grad_norm": 4.854882717132568,
      "learning_rate": 6.921241050119333e-05,
      "loss": 2.4277,
      "step": 290
    },
    {
      "epoch": 0.01042132970437087,
      "grad_norm": 2.883726119995117,
      "learning_rate": 6.94510739856802e-05,
      "loss": 2.0774,
      "step": 291
    },
    {
      "epoch": 0.010457141833939155,
      "grad_norm": 2.1024317741394043,
      "learning_rate": 6.968973747016708e-05,
      "loss": 2.1162,
      "step": 292
    },
    {
      "epoch": 0.01049295396350744,
      "grad_norm": 2.173642873764038,
      "learning_rate": 6.992840095465394e-05,
      "loss": 1.9923,
      "step": 293
    },
    {
      "epoch": 0.010528766093075724,
      "grad_norm": 2.249166250228882,
      "learning_rate": 7.016706443914081e-05,
      "loss": 2.0456,
      "step": 294
    },
    {
      "epoch": 0.010564578222644009,
      "grad_norm": 2.782399892807007,
      "learning_rate": 7.040572792362768e-05,
      "loss": 1.6617,
      "step": 295
    },
    {
      "epoch": 0.010600390352212294,
      "grad_norm": 2.9986255168914795,
      "learning_rate": 7.064439140811456e-05,
      "loss": 2.0264,
      "step": 296
    },
    {
      "epoch": 0.01063620248178058,
      "grad_norm": 1.7165770530700684,
      "learning_rate": 7.088305489260143e-05,
      "loss": 1.9433,
      "step": 297
    },
    {
      "epoch": 0.010672014611348864,
      "grad_norm": 2.9197936058044434,
      "learning_rate": 7.112171837708831e-05,
      "loss": 1.9866,
      "step": 298
    },
    {
      "epoch": 0.010707826740917149,
      "grad_norm": 3.9878487586975098,
      "learning_rate": 7.136038186157519e-05,
      "loss": 2.2649,
      "step": 299
    },
    {
      "epoch": 0.010743638870485434,
      "grad_norm": 2.6345372200012207,
      "learning_rate": 7.159904534606206e-05,
      "loss": 1.8249,
      "step": 300
    },
    {
      "epoch": 0.010779451000053718,
      "grad_norm": 2.4214723110198975,
      "learning_rate": 7.183770883054893e-05,
      "loss": 2.0411,
      "step": 301
    },
    {
      "epoch": 0.010815263129622003,
      "grad_norm": 3.2533586025238037,
      "learning_rate": 7.20763723150358e-05,
      "loss": 2.0109,
      "step": 302
    },
    {
      "epoch": 0.010851075259190287,
      "grad_norm": 2.417447566986084,
      "learning_rate": 7.231503579952268e-05,
      "loss": 1.996,
      "step": 303
    },
    {
      "epoch": 0.010886887388758572,
      "grad_norm": 3.2113840579986572,
      "learning_rate": 7.255369928400954e-05,
      "loss": 2.1155,
      "step": 304
    },
    {
      "epoch": 0.010922699518326858,
      "grad_norm": 2.6941497325897217,
      "learning_rate": 7.279236276849643e-05,
      "loss": 2.1457,
      "step": 305
    },
    {
      "epoch": 0.010958511647895143,
      "grad_norm": 1.8207515478134155,
      "learning_rate": 7.30310262529833e-05,
      "loss": 1.9003,
      "step": 306
    },
    {
      "epoch": 0.010994323777463427,
      "grad_norm": 3.3273072242736816,
      "learning_rate": 7.326968973747017e-05,
      "loss": 2.2891,
      "step": 307
    },
    {
      "epoch": 0.011030135907031712,
      "grad_norm": 1.6675465106964111,
      "learning_rate": 7.350835322195704e-05,
      "loss": 1.9006,
      "step": 308
    },
    {
      "epoch": 0.011065948036599996,
      "grad_norm": 2.2108771800994873,
      "learning_rate": 7.374701670644391e-05,
      "loss": 2.0193,
      "step": 309
    },
    {
      "epoch": 0.011101760166168281,
      "grad_norm": 1.6703202724456787,
      "learning_rate": 7.398568019093079e-05,
      "loss": 1.7437,
      "step": 310
    },
    {
      "epoch": 0.011137572295736566,
      "grad_norm": 2.188809394836426,
      "learning_rate": 7.422434367541766e-05,
      "loss": 1.97,
      "step": 311
    },
    {
      "epoch": 0.01117338442530485,
      "grad_norm": 2.2568600177764893,
      "learning_rate": 7.446300715990454e-05,
      "loss": 2.1866,
      "step": 312
    },
    {
      "epoch": 0.011209196554873135,
      "grad_norm": 2.189673662185669,
      "learning_rate": 7.470167064439141e-05,
      "loss": 1.8975,
      "step": 313
    },
    {
      "epoch": 0.011245008684441421,
      "grad_norm": 1.8501036167144775,
      "learning_rate": 7.494033412887829e-05,
      "loss": 2.0104,
      "step": 314
    },
    {
      "epoch": 0.011280820814009706,
      "grad_norm": 3.4657652378082275,
      "learning_rate": 7.517899761336516e-05,
      "loss": 2.1677,
      "step": 315
    },
    {
      "epoch": 0.01131663294357799,
      "grad_norm": 1.7842155694961548,
      "learning_rate": 7.541766109785204e-05,
      "loss": 2.162,
      "step": 316
    },
    {
      "epoch": 0.011352445073146275,
      "grad_norm": 2.6685571670532227,
      "learning_rate": 7.565632458233891e-05,
      "loss": 2.2024,
      "step": 317
    },
    {
      "epoch": 0.01138825720271456,
      "grad_norm": 2.2792065143585205,
      "learning_rate": 7.589498806682577e-05,
      "loss": 2.0813,
      "step": 318
    },
    {
      "epoch": 0.011424069332282844,
      "grad_norm": 1.8221955299377441,
      "learning_rate": 7.613365155131266e-05,
      "loss": 1.9769,
      "step": 319
    },
    {
      "epoch": 0.011459881461851128,
      "grad_norm": 4.3000922203063965,
      "learning_rate": 7.637231503579952e-05,
      "loss": 2.301,
      "step": 320
    },
    {
      "epoch": 0.011495693591419413,
      "grad_norm": 1.8403431177139282,
      "learning_rate": 7.66109785202864e-05,
      "loss": 1.9934,
      "step": 321
    },
    {
      "epoch": 0.0115315057209877,
      "grad_norm": 2.3018696308135986,
      "learning_rate": 7.684964200477327e-05,
      "loss": 1.9003,
      "step": 322
    },
    {
      "epoch": 0.011567317850555984,
      "grad_norm": 2.887930393218994,
      "learning_rate": 7.708830548926015e-05,
      "loss": 2.5743,
      "step": 323
    },
    {
      "epoch": 0.011603129980124268,
      "grad_norm": 1.5166923999786377,
      "learning_rate": 7.732696897374702e-05,
      "loss": 1.8831,
      "step": 324
    },
    {
      "epoch": 0.011638942109692553,
      "grad_norm": 2.4193575382232666,
      "learning_rate": 7.756563245823389e-05,
      "loss": 2.0361,
      "step": 325
    },
    {
      "epoch": 0.011674754239260838,
      "grad_norm": 1.8523463010787964,
      "learning_rate": 7.780429594272077e-05,
      "loss": 1.9684,
      "step": 326
    },
    {
      "epoch": 0.011710566368829122,
      "grad_norm": 2.535493850708008,
      "learning_rate": 7.804295942720764e-05,
      "loss": 2.1888,
      "step": 327
    },
    {
      "epoch": 0.011746378498397407,
      "grad_norm": 2.134873390197754,
      "learning_rate": 7.828162291169452e-05,
      "loss": 1.826,
      "step": 328
    },
    {
      "epoch": 0.011782190627965691,
      "grad_norm": 2.6702308654785156,
      "learning_rate": 7.852028639618139e-05,
      "loss": 1.9467,
      "step": 329
    },
    {
      "epoch": 0.011818002757533978,
      "grad_norm": 1.7101609706878662,
      "learning_rate": 7.875894988066827e-05,
      "loss": 2.0146,
      "step": 330
    },
    {
      "epoch": 0.011853814887102262,
      "grad_norm": 1.9963518381118774,
      "learning_rate": 7.899761336515514e-05,
      "loss": 1.8542,
      "step": 331
    },
    {
      "epoch": 0.011889627016670547,
      "grad_norm": 2.23158597946167,
      "learning_rate": 7.9236276849642e-05,
      "loss": 1.895,
      "step": 332
    },
    {
      "epoch": 0.011925439146238831,
      "grad_norm": 2.7184247970581055,
      "learning_rate": 7.947494033412887e-05,
      "loss": 2.2446,
      "step": 333
    },
    {
      "epoch": 0.011961251275807116,
      "grad_norm": 2.315206527709961,
      "learning_rate": 7.971360381861575e-05,
      "loss": 2.2946,
      "step": 334
    },
    {
      "epoch": 0.0119970634053754,
      "grad_norm": 2.8590247631073,
      "learning_rate": 7.995226730310262e-05,
      "loss": 2.0968,
      "step": 335
    },
    {
      "epoch": 0.012032875534943685,
      "grad_norm": 2.333500623703003,
      "learning_rate": 8.01909307875895e-05,
      "loss": 2.2555,
      "step": 336
    },
    {
      "epoch": 0.01206868766451197,
      "grad_norm": 1.6746714115142822,
      "learning_rate": 8.042959427207638e-05,
      "loss": 1.9062,
      "step": 337
    },
    {
      "epoch": 0.012104499794080254,
      "grad_norm": 2.6714026927948,
      "learning_rate": 8.066825775656325e-05,
      "loss": 2.5086,
      "step": 338
    },
    {
      "epoch": 0.01214031192364854,
      "grad_norm": 4.634089469909668,
      "learning_rate": 8.090692124105012e-05,
      "loss": 2.0091,
      "step": 339
    },
    {
      "epoch": 0.012176124053216825,
      "grad_norm": 2.3018248081207275,
      "learning_rate": 8.114558472553699e-05,
      "loss": 2.1707,
      "step": 340
    },
    {
      "epoch": 0.01221193618278511,
      "grad_norm": 2.3117361068725586,
      "learning_rate": 8.138424821002387e-05,
      "loss": 2.0692,
      "step": 341
    },
    {
      "epoch": 0.012247748312353394,
      "grad_norm": 2.316091299057007,
      "learning_rate": 8.162291169451074e-05,
      "loss": 2.1788,
      "step": 342
    },
    {
      "epoch": 0.012283560441921679,
      "grad_norm": 1.7986754179000854,
      "learning_rate": 8.186157517899762e-05,
      "loss": 2.0328,
      "step": 343
    },
    {
      "epoch": 0.012319372571489963,
      "grad_norm": 2.5290677547454834,
      "learning_rate": 8.210023866348449e-05,
      "loss": 2.2345,
      "step": 344
    },
    {
      "epoch": 0.012355184701058248,
      "grad_norm": 1.8213863372802734,
      "learning_rate": 8.233890214797137e-05,
      "loss": 1.833,
      "step": 345
    },
    {
      "epoch": 0.012390996830626532,
      "grad_norm": 2.757683753967285,
      "learning_rate": 8.257756563245825e-05,
      "loss": 2.2434,
      "step": 346
    },
    {
      "epoch": 0.012426808960194819,
      "grad_norm": 2.7653613090515137,
      "learning_rate": 8.28162291169451e-05,
      "loss": 1.9301,
      "step": 347
    },
    {
      "epoch": 0.012462621089763103,
      "grad_norm": 2.544404983520508,
      "learning_rate": 8.305489260143198e-05,
      "loss": 2.1534,
      "step": 348
    },
    {
      "epoch": 0.012498433219331388,
      "grad_norm": 2.3338091373443604,
      "learning_rate": 8.329355608591885e-05,
      "loss": 2.1738,
      "step": 349
    },
    {
      "epoch": 0.012534245348899672,
      "grad_norm": 2.3454766273498535,
      "learning_rate": 8.353221957040573e-05,
      "loss": 2.1207,
      "step": 350
    },
    {
      "epoch": 0.012570057478467957,
      "grad_norm": 2.4842939376831055,
      "learning_rate": 8.37708830548926e-05,
      "loss": 1.9311,
      "step": 351
    },
    {
      "epoch": 0.012605869608036242,
      "grad_norm": 3.1244163513183594,
      "learning_rate": 8.400954653937948e-05,
      "loss": 1.8968,
      "step": 352
    },
    {
      "epoch": 0.012641681737604526,
      "grad_norm": 1.9500224590301514,
      "learning_rate": 8.424821002386635e-05,
      "loss": 1.9746,
      "step": 353
    },
    {
      "epoch": 0.01267749386717281,
      "grad_norm": 2.537527561187744,
      "learning_rate": 8.448687350835323e-05,
      "loss": 1.8227,
      "step": 354
    },
    {
      "epoch": 0.012713305996741097,
      "grad_norm": 1.9497244358062744,
      "learning_rate": 8.47255369928401e-05,
      "loss": 1.9673,
      "step": 355
    },
    {
      "epoch": 0.012749118126309382,
      "grad_norm": 2.9225804805755615,
      "learning_rate": 8.496420047732697e-05,
      "loss": 2.253,
      "step": 356
    },
    {
      "epoch": 0.012784930255877666,
      "grad_norm": 2.1374170780181885,
      "learning_rate": 8.520286396181385e-05,
      "loss": 1.8416,
      "step": 357
    },
    {
      "epoch": 0.01282074238544595,
      "grad_norm": 2.318568468093872,
      "learning_rate": 8.544152744630072e-05,
      "loss": 2.1213,
      "step": 358
    },
    {
      "epoch": 0.012856554515014235,
      "grad_norm": 1.6215802431106567,
      "learning_rate": 8.56801909307876e-05,
      "loss": 1.9469,
      "step": 359
    },
    {
      "epoch": 0.01289236664458252,
      "grad_norm": 2.4686646461486816,
      "learning_rate": 8.591885441527447e-05,
      "loss": 2.2494,
      "step": 360
    },
    {
      "epoch": 0.012928178774150804,
      "grad_norm": 1.9479748010635376,
      "learning_rate": 8.615751789976135e-05,
      "loss": 2.0661,
      "step": 361
    },
    {
      "epoch": 0.012963990903719089,
      "grad_norm": 2.651142120361328,
      "learning_rate": 8.639618138424822e-05,
      "loss": 2.2081,
      "step": 362
    },
    {
      "epoch": 0.012999803033287374,
      "grad_norm": 2.298008441925049,
      "learning_rate": 8.663484486873508e-05,
      "loss": 2.0353,
      "step": 363
    },
    {
      "epoch": 0.01303561516285566,
      "grad_norm": 2.419844388961792,
      "learning_rate": 8.687350835322196e-05,
      "loss": 1.8743,
      "step": 364
    },
    {
      "epoch": 0.013071427292423944,
      "grad_norm": 1.65255868434906,
      "learning_rate": 8.711217183770883e-05,
      "loss": 1.9959,
      "step": 365
    },
    {
      "epoch": 0.013107239421992229,
      "grad_norm": 2.8441193103790283,
      "learning_rate": 8.735083532219571e-05,
      "loss": 1.9528,
      "step": 366
    },
    {
      "epoch": 0.013143051551560514,
      "grad_norm": 2.9021151065826416,
      "learning_rate": 8.758949880668258e-05,
      "loss": 1.7156,
      "step": 367
    },
    {
      "epoch": 0.013178863681128798,
      "grad_norm": 2.2067577838897705,
      "learning_rate": 8.782816229116946e-05,
      "loss": 1.9428,
      "step": 368
    },
    {
      "epoch": 0.013214675810697083,
      "grad_norm": 1.5661609172821045,
      "learning_rate": 8.806682577565633e-05,
      "loss": 1.7973,
      "step": 369
    },
    {
      "epoch": 0.013250487940265367,
      "grad_norm": 2.014951467514038,
      "learning_rate": 8.83054892601432e-05,
      "loss": 2.0513,
      "step": 370
    },
    {
      "epoch": 0.013286300069833652,
      "grad_norm": 2.001088857650757,
      "learning_rate": 8.854415274463007e-05,
      "loss": 2.0834,
      "step": 371
    },
    {
      "epoch": 0.013322112199401938,
      "grad_norm": 3.38238787651062,
      "learning_rate": 8.878281622911695e-05,
      "loss": 1.9486,
      "step": 372
    },
    {
      "epoch": 0.013357924328970223,
      "grad_norm": 2.687730550765991,
      "learning_rate": 8.902147971360383e-05,
      "loss": 2.0036,
      "step": 373
    },
    {
      "epoch": 0.013393736458538507,
      "grad_norm": 2.138909101486206,
      "learning_rate": 8.92601431980907e-05,
      "loss": 1.8376,
      "step": 374
    },
    {
      "epoch": 0.013429548588106792,
      "grad_norm": 2.748356580734253,
      "learning_rate": 8.949880668257758e-05,
      "loss": 2.4684,
      "step": 375
    },
    {
      "epoch": 0.013465360717675076,
      "grad_norm": 1.895658016204834,
      "learning_rate": 8.973747016706445e-05,
      "loss": 2.1803,
      "step": 376
    },
    {
      "epoch": 0.013501172847243361,
      "grad_norm": 2.7984509468078613,
      "learning_rate": 8.997613365155131e-05,
      "loss": 2.166,
      "step": 377
    },
    {
      "epoch": 0.013536984976811646,
      "grad_norm": 4.1571173667907715,
      "learning_rate": 9.021479713603818e-05,
      "loss": 2.1634,
      "step": 378
    },
    {
      "epoch": 0.01357279710637993,
      "grad_norm": 2.012040853500366,
      "learning_rate": 9.045346062052506e-05,
      "loss": 1.6297,
      "step": 379
    },
    {
      "epoch": 0.013608609235948217,
      "grad_norm": 1.9994226694107056,
      "learning_rate": 9.069212410501193e-05,
      "loss": 1.7911,
      "step": 380
    },
    {
      "epoch": 0.013644421365516501,
      "grad_norm": 2.6208250522613525,
      "learning_rate": 9.093078758949881e-05,
      "loss": 1.94,
      "step": 381
    },
    {
      "epoch": 0.013680233495084786,
      "grad_norm": 3.200495719909668,
      "learning_rate": 9.116945107398569e-05,
      "loss": 1.9532,
      "step": 382
    },
    {
      "epoch": 0.01371604562465307,
      "grad_norm": 1.4798295497894287,
      "learning_rate": 9.140811455847256e-05,
      "loss": 2.0838,
      "step": 383
    },
    {
      "epoch": 0.013751857754221355,
      "grad_norm": 2.466656446456909,
      "learning_rate": 9.164677804295944e-05,
      "loss": 2.0837,
      "step": 384
    },
    {
      "epoch": 0.01378766988378964,
      "grad_norm": 2.8002874851226807,
      "learning_rate": 9.18854415274463e-05,
      "loss": 2.3469,
      "step": 385
    },
    {
      "epoch": 0.013823482013357924,
      "grad_norm": 2.267125368118286,
      "learning_rate": 9.212410501193318e-05,
      "loss": 2.1056,
      "step": 386
    },
    {
      "epoch": 0.013859294142926208,
      "grad_norm": 2.0536201000213623,
      "learning_rate": 9.236276849642005e-05,
      "loss": 2.2548,
      "step": 387
    },
    {
      "epoch": 0.013895106272494493,
      "grad_norm": 1.349810242652893,
      "learning_rate": 9.260143198090693e-05,
      "loss": 2.0713,
      "step": 388
    },
    {
      "epoch": 0.01393091840206278,
      "grad_norm": 2.174856424331665,
      "learning_rate": 9.28400954653938e-05,
      "loss": 1.9262,
      "step": 389
    },
    {
      "epoch": 0.013966730531631064,
      "grad_norm": 1.979978322982788,
      "learning_rate": 9.307875894988068e-05,
      "loss": 2.2743,
      "step": 390
    },
    {
      "epoch": 0.014002542661199349,
      "grad_norm": 1.4002437591552734,
      "learning_rate": 9.331742243436754e-05,
      "loss": 1.8055,
      "step": 391
    },
    {
      "epoch": 0.014038354790767633,
      "grad_norm": 1.8472005128860474,
      "learning_rate": 9.355608591885443e-05,
      "loss": 2.1736,
      "step": 392
    },
    {
      "epoch": 0.014074166920335918,
      "grad_norm": 1.950952172279358,
      "learning_rate": 9.379474940334129e-05,
      "loss": 2.1521,
      "step": 393
    },
    {
      "epoch": 0.014109979049904202,
      "grad_norm": 1.622833251953125,
      "learning_rate": 9.403341288782816e-05,
      "loss": 1.8347,
      "step": 394
    },
    {
      "epoch": 0.014145791179472487,
      "grad_norm": 2.020577907562256,
      "learning_rate": 9.427207637231504e-05,
      "loss": 2.0858,
      "step": 395
    },
    {
      "epoch": 0.014181603309040771,
      "grad_norm": 1.5022003650665283,
      "learning_rate": 9.451073985680191e-05,
      "loss": 1.8131,
      "step": 396
    },
    {
      "epoch": 0.014217415438609058,
      "grad_norm": 1.293748140335083,
      "learning_rate": 9.474940334128879e-05,
      "loss": 1.7116,
      "step": 397
    },
    {
      "epoch": 0.014253227568177342,
      "grad_norm": 3.8732552528381348,
      "learning_rate": 9.498806682577566e-05,
      "loss": 2.359,
      "step": 398
    },
    {
      "epoch": 0.014289039697745627,
      "grad_norm": 2.218346118927002,
      "learning_rate": 9.522673031026254e-05,
      "loss": 2.3042,
      "step": 399
    },
    {
      "epoch": 0.014324851827313911,
      "grad_norm": 1.8491095304489136,
      "learning_rate": 9.546539379474941e-05,
      "loss": 1.8915,
      "step": 400
    },
    {
      "epoch": 0.014360663956882196,
      "grad_norm": 2.9908032417297363,
      "learning_rate": 9.570405727923628e-05,
      "loss": 2.0991,
      "step": 401
    },
    {
      "epoch": 0.01439647608645048,
      "grad_norm": 6.480159282684326,
      "learning_rate": 9.594272076372316e-05,
      "loss": 2.5134,
      "step": 402
    },
    {
      "epoch": 0.014432288216018765,
      "grad_norm": 1.737426996231079,
      "learning_rate": 9.618138424821003e-05,
      "loss": 1.8967,
      "step": 403
    },
    {
      "epoch": 0.01446810034558705,
      "grad_norm": 1.8148224353790283,
      "learning_rate": 9.64200477326969e-05,
      "loss": 1.7998,
      "step": 404
    },
    {
      "epoch": 0.014503912475155336,
      "grad_norm": 2.05126690864563,
      "learning_rate": 9.665871121718377e-05,
      "loss": 1.8313,
      "step": 405
    },
    {
      "epoch": 0.01453972460472362,
      "grad_norm": 3.2550830841064453,
      "learning_rate": 9.689737470167066e-05,
      "loss": 2.2962,
      "step": 406
    },
    {
      "epoch": 0.014575536734291905,
      "grad_norm": 2.001216411590576,
      "learning_rate": 9.713603818615752e-05,
      "loss": 2.031,
      "step": 407
    },
    {
      "epoch": 0.01461134886386019,
      "grad_norm": 1.8144034147262573,
      "learning_rate": 9.737470167064439e-05,
      "loss": 1.675,
      "step": 408
    },
    {
      "epoch": 0.014647160993428474,
      "grad_norm": 2.0882346630096436,
      "learning_rate": 9.761336515513126e-05,
      "loss": 2.0904,
      "step": 409
    },
    {
      "epoch": 0.014682973122996759,
      "grad_norm": 2.1910388469696045,
      "learning_rate": 9.785202863961814e-05,
      "loss": 1.8463,
      "step": 410
    },
    {
      "epoch": 0.014718785252565043,
      "grad_norm": 1.9278587102890015,
      "learning_rate": 9.809069212410502e-05,
      "loss": 2.0166,
      "step": 411
    },
    {
      "epoch": 0.014754597382133328,
      "grad_norm": 2.418215751647949,
      "learning_rate": 9.832935560859189e-05,
      "loss": 2.3441,
      "step": 412
    },
    {
      "epoch": 0.014790409511701613,
      "grad_norm": 1.1429497003555298,
      "learning_rate": 9.856801909307877e-05,
      "loss": 1.628,
      "step": 413
    },
    {
      "epoch": 0.014826221641269899,
      "grad_norm": 2.355159044265747,
      "learning_rate": 9.880668257756564e-05,
      "loss": 2.1252,
      "step": 414
    },
    {
      "epoch": 0.014862033770838183,
      "grad_norm": 1.5233746767044067,
      "learning_rate": 9.90453460620525e-05,
      "loss": 1.9219,
      "step": 415
    },
    {
      "epoch": 0.014897845900406468,
      "grad_norm": 1.9501924514770508,
      "learning_rate": 9.928400954653937e-05,
      "loss": 1.6977,
      "step": 416
    },
    {
      "epoch": 0.014933658029974753,
      "grad_norm": 2.0013651847839355,
      "learning_rate": 9.952267303102626e-05,
      "loss": 2.1085,
      "step": 417
    },
    {
      "epoch": 0.014969470159543037,
      "grad_norm": 2.1112027168273926,
      "learning_rate": 9.976133651551312e-05,
      "loss": 2.0397,
      "step": 418
    },
    {
      "epoch": 0.015005282289111322,
      "grad_norm": 1.9288190603256226,
      "learning_rate": 0.0001,
      "loss": 2.0879,
      "step": 419
    },
    {
      "epoch": 0.015041094418679606,
      "grad_norm": 2.2626709938049316,
      "learning_rate": 0.00010023866348448687,
      "loss": 2.2272,
      "step": 420
    },
    {
      "epoch": 0.01507690654824789,
      "grad_norm": 1.7174164056777954,
      "learning_rate": 0.00010047732696897377,
      "loss": 1.9436,
      "step": 421
    },
    {
      "epoch": 0.015112718677816177,
      "grad_norm": 2.5441973209381104,
      "learning_rate": 0.00010071599045346064,
      "loss": 2.2915,
      "step": 422
    },
    {
      "epoch": 0.015148530807384462,
      "grad_norm": 1.7958893775939941,
      "learning_rate": 0.0001009546539379475,
      "loss": 1.9233,
      "step": 423
    },
    {
      "epoch": 0.015184342936952746,
      "grad_norm": 2.6941654682159424,
      "learning_rate": 0.00010119331742243436,
      "loss": 2.0398,
      "step": 424
    },
    {
      "epoch": 0.01522015506652103,
      "grad_norm": 2.385948657989502,
      "learning_rate": 0.00010143198090692125,
      "loss": 2.2552,
      "step": 425
    },
    {
      "epoch": 0.015255967196089315,
      "grad_norm": 2.532027244567871,
      "learning_rate": 0.00010167064439140812,
      "loss": 2.3462,
      "step": 426
    },
    {
      "epoch": 0.0152917793256576,
      "grad_norm": 1.7785552740097046,
      "learning_rate": 0.00010190930787589499,
      "loss": 2.1202,
      "step": 427
    },
    {
      "epoch": 0.015327591455225885,
      "grad_norm": 4.266390323638916,
      "learning_rate": 0.00010214797136038186,
      "loss": 1.7885,
      "step": 428
    },
    {
      "epoch": 0.015363403584794169,
      "grad_norm": 1.9129207134246826,
      "learning_rate": 0.00010238663484486875,
      "loss": 1.9795,
      "step": 429
    },
    {
      "epoch": 0.015399215714362455,
      "grad_norm": 2.3568286895751953,
      "learning_rate": 0.00010262529832935562,
      "loss": 2.1897,
      "step": 430
    },
    {
      "epoch": 0.01543502784393074,
      "grad_norm": 2.463308811187744,
      "learning_rate": 0.00010286396181384249,
      "loss": 2.191,
      "step": 431
    },
    {
      "epoch": 0.015470839973499025,
      "grad_norm": 2.999436855316162,
      "learning_rate": 0.00010310262529832937,
      "loss": 2.1384,
      "step": 432
    },
    {
      "epoch": 0.01550665210306731,
      "grad_norm": 2.8959248065948486,
      "learning_rate": 0.00010334128878281624,
      "loss": 2.5383,
      "step": 433
    },
    {
      "epoch": 0.015542464232635594,
      "grad_norm": 1.6721136569976807,
      "learning_rate": 0.0001035799522673031,
      "loss": 2.1312,
      "step": 434
    },
    {
      "epoch": 0.015578276362203878,
      "grad_norm": 2.227095365524292,
      "learning_rate": 0.00010381861575178997,
      "loss": 1.9225,
      "step": 435
    },
    {
      "epoch": 0.015614088491772163,
      "grad_norm": 2.0040481090545654,
      "learning_rate": 0.00010405727923627687,
      "loss": 1.9404,
      "step": 436
    },
    {
      "epoch": 0.01564990062134045,
      "grad_norm": 2.118105173110962,
      "learning_rate": 0.00010429594272076373,
      "loss": 2.0456,
      "step": 437
    },
    {
      "epoch": 0.015685712750908732,
      "grad_norm": 1.629913568496704,
      "learning_rate": 0.0001045346062052506,
      "loss": 1.9724,
      "step": 438
    },
    {
      "epoch": 0.01572152488047702,
      "grad_norm": 1.544683575630188,
      "learning_rate": 0.00010477326968973748,
      "loss": 2.3597,
      "step": 439
    },
    {
      "epoch": 0.0157573370100453,
      "grad_norm": 2.007746934890747,
      "learning_rate": 0.00010501193317422435,
      "loss": 2.1682,
      "step": 440
    },
    {
      "epoch": 0.015793149139613587,
      "grad_norm": 2.2837698459625244,
      "learning_rate": 0.00010525059665871122,
      "loss": 2.0777,
      "step": 441
    },
    {
      "epoch": 0.01582896126918187,
      "grad_norm": 2.146202802658081,
      "learning_rate": 0.00010548926014319809,
      "loss": 2.0073,
      "step": 442
    },
    {
      "epoch": 0.015864773398750157,
      "grad_norm": 2.1583595275878906,
      "learning_rate": 0.00010572792362768498,
      "loss": 1.7767,
      "step": 443
    },
    {
      "epoch": 0.015900585528318443,
      "grad_norm": 2.5789666175842285,
      "learning_rate": 0.00010596658711217185,
      "loss": 2.2289,
      "step": 444
    },
    {
      "epoch": 0.015936397657886726,
      "grad_norm": 2.8515987396240234,
      "learning_rate": 0.00010620525059665872,
      "loss": 2.1816,
      "step": 445
    },
    {
      "epoch": 0.015972209787455012,
      "grad_norm": 2.2290050983428955,
      "learning_rate": 0.00010644391408114558,
      "loss": 2.1232,
      "step": 446
    },
    {
      "epoch": 0.016008021917023295,
      "grad_norm": 1.601799726486206,
      "learning_rate": 0.00010668257756563247,
      "loss": 2.1464,
      "step": 447
    },
    {
      "epoch": 0.01604383404659158,
      "grad_norm": 1.905332088470459,
      "learning_rate": 0.00010692124105011933,
      "loss": 1.8095,
      "step": 448
    },
    {
      "epoch": 0.016079646176159864,
      "grad_norm": 2.2088615894317627,
      "learning_rate": 0.0001071599045346062,
      "loss": 2.4517,
      "step": 449
    },
    {
      "epoch": 0.01611545830572815,
      "grad_norm": 1.835677146911621,
      "learning_rate": 0.0001073985680190931,
      "loss": 2.1423,
      "step": 450
    },
    {
      "epoch": 0.016151270435296437,
      "grad_norm": 1.556067705154419,
      "learning_rate": 0.00010763723150357996,
      "loss": 1.8783,
      "step": 451
    },
    {
      "epoch": 0.01618708256486472,
      "grad_norm": 2.0234811305999756,
      "learning_rate": 0.00010787589498806683,
      "loss": 2.0476,
      "step": 452
    },
    {
      "epoch": 0.016222894694433006,
      "grad_norm": 2.6546096801757812,
      "learning_rate": 0.0001081145584725537,
      "loss": 1.9325,
      "step": 453
    },
    {
      "epoch": 0.01625870682400129,
      "grad_norm": 1.8908900022506714,
      "learning_rate": 0.00010835322195704058,
      "loss": 1.9481,
      "step": 454
    },
    {
      "epoch": 0.016294518953569575,
      "grad_norm": 1.8198904991149902,
      "learning_rate": 0.00010859188544152745,
      "loss": 1.918,
      "step": 455
    },
    {
      "epoch": 0.016330331083137858,
      "grad_norm": 1.573096752166748,
      "learning_rate": 0.00010883054892601432,
      "loss": 2.0613,
      "step": 456
    },
    {
      "epoch": 0.016366143212706144,
      "grad_norm": 1.9454529285430908,
      "learning_rate": 0.00010906921241050121,
      "loss": 1.9583,
      "step": 457
    },
    {
      "epoch": 0.016401955342274427,
      "grad_norm": 1.7274153232574463,
      "learning_rate": 0.00010930787589498808,
      "loss": 1.7381,
      "step": 458
    },
    {
      "epoch": 0.016437767471842713,
      "grad_norm": 1.4126781225204468,
      "learning_rate": 0.00010954653937947495,
      "loss": 1.7259,
      "step": 459
    },
    {
      "epoch": 0.016473579601411,
      "grad_norm": 2.398538589477539,
      "learning_rate": 0.00010978520286396181,
      "loss": 2.2239,
      "step": 460
    },
    {
      "epoch": 0.016509391730979282,
      "grad_norm": 1.7049897909164429,
      "learning_rate": 0.0001100238663484487,
      "loss": 1.7488,
      "step": 461
    },
    {
      "epoch": 0.01654520386054757,
      "grad_norm": 1.9099923372268677,
      "learning_rate": 0.00011026252983293556,
      "loss": 1.8568,
      "step": 462
    },
    {
      "epoch": 0.01658101599011585,
      "grad_norm": 1.5508618354797363,
      "learning_rate": 0.00011050119331742243,
      "loss": 1.9427,
      "step": 463
    },
    {
      "epoch": 0.016616828119684138,
      "grad_norm": 3.0816149711608887,
      "learning_rate": 0.0001107398568019093,
      "loss": 2.3965,
      "step": 464
    },
    {
      "epoch": 0.01665264024925242,
      "grad_norm": 1.8753119707107544,
      "learning_rate": 0.0001109785202863962,
      "loss": 1.9468,
      "step": 465
    },
    {
      "epoch": 0.016688452378820707,
      "grad_norm": 1.9726189374923706,
      "learning_rate": 0.00011121718377088306,
      "loss": 1.7688,
      "step": 466
    },
    {
      "epoch": 0.01672426450838899,
      "grad_norm": 1.5842738151550293,
      "learning_rate": 0.00011145584725536993,
      "loss": 1.8443,
      "step": 467
    },
    {
      "epoch": 0.016760076637957276,
      "grad_norm": 2.3097848892211914,
      "learning_rate": 0.00011169451073985681,
      "loss": 2.4292,
      "step": 468
    },
    {
      "epoch": 0.016795888767525562,
      "grad_norm": 1.7131446599960327,
      "learning_rate": 0.00011193317422434368,
      "loss": 1.9989,
      "step": 469
    },
    {
      "epoch": 0.016831700897093845,
      "grad_norm": 1.544679045677185,
      "learning_rate": 0.00011217183770883055,
      "loss": 1.9196,
      "step": 470
    },
    {
      "epoch": 0.01686751302666213,
      "grad_norm": 2.726846218109131,
      "learning_rate": 0.00011241050119331741,
      "loss": 2.1967,
      "step": 471
    },
    {
      "epoch": 0.016903325156230414,
      "grad_norm": 3.2844207286834717,
      "learning_rate": 0.00011264916467780431,
      "loss": 1.8285,
      "step": 472
    },
    {
      "epoch": 0.0169391372857987,
      "grad_norm": 1.295161247253418,
      "learning_rate": 0.00011288782816229118,
      "loss": 1.9555,
      "step": 473
    },
    {
      "epoch": 0.016974949415366983,
      "grad_norm": 2.582324266433716,
      "learning_rate": 0.00011312649164677805,
      "loss": 2.1132,
      "step": 474
    },
    {
      "epoch": 0.01701076154493527,
      "grad_norm": 1.7749077081680298,
      "learning_rate": 0.00011336515513126493,
      "loss": 2.0865,
      "step": 475
    },
    {
      "epoch": 0.017046573674503556,
      "grad_norm": 2.0523269176483154,
      "learning_rate": 0.0001136038186157518,
      "loss": 2.2231,
      "step": 476
    },
    {
      "epoch": 0.01708238580407184,
      "grad_norm": 3.3681750297546387,
      "learning_rate": 0.00011384248210023866,
      "loss": 2.6565,
      "step": 477
    },
    {
      "epoch": 0.017118197933640125,
      "grad_norm": 2.334472417831421,
      "learning_rate": 0.00011408114558472553,
      "loss": 2.1217,
      "step": 478
    },
    {
      "epoch": 0.017154010063208408,
      "grad_norm": 2.677645444869995,
      "learning_rate": 0.00011431980906921242,
      "loss": 2.2404,
      "step": 479
    },
    {
      "epoch": 0.017189822192776694,
      "grad_norm": 2.0976386070251465,
      "learning_rate": 0.00011455847255369929,
      "loss": 2.2931,
      "step": 480
    },
    {
      "epoch": 0.017225634322344977,
      "grad_norm": 1.753631591796875,
      "learning_rate": 0.00011479713603818616,
      "loss": 1.9063,
      "step": 481
    },
    {
      "epoch": 0.017261446451913263,
      "grad_norm": 1.7365729808807373,
      "learning_rate": 0.00011503579952267303,
      "loss": 1.9847,
      "step": 482
    },
    {
      "epoch": 0.017297258581481546,
      "grad_norm": 2.4236369132995605,
      "learning_rate": 0.00011527446300715991,
      "loss": 1.9288,
      "step": 483
    },
    {
      "epoch": 0.017333070711049833,
      "grad_norm": 1.8561084270477295,
      "learning_rate": 0.00011551312649164678,
      "loss": 1.743,
      "step": 484
    },
    {
      "epoch": 0.01736888284061812,
      "grad_norm": 1.5186161994934082,
      "learning_rate": 0.00011575178997613365,
      "loss": 2.0968,
      "step": 485
    },
    {
      "epoch": 0.017404694970186402,
      "grad_norm": 2.043816089630127,
      "learning_rate": 0.00011599045346062054,
      "loss": 2.2149,
      "step": 486
    },
    {
      "epoch": 0.017440507099754688,
      "grad_norm": 2.2540132999420166,
      "learning_rate": 0.00011622911694510741,
      "loss": 2.089,
      "step": 487
    },
    {
      "epoch": 0.01747631922932297,
      "grad_norm": 1.8196076154708862,
      "learning_rate": 0.00011646778042959428,
      "loss": 2.051,
      "step": 488
    },
    {
      "epoch": 0.017512131358891257,
      "grad_norm": 1.6149146556854248,
      "learning_rate": 0.00011670644391408114,
      "loss": 1.991,
      "step": 489
    },
    {
      "epoch": 0.01754794348845954,
      "grad_norm": 1.9776016473770142,
      "learning_rate": 0.00011694510739856804,
      "loss": 2.2675,
      "step": 490
    },
    {
      "epoch": 0.017583755618027826,
      "grad_norm": 2.040417432785034,
      "learning_rate": 0.0001171837708830549,
      "loss": 2.0448,
      "step": 491
    },
    {
      "epoch": 0.01761956774759611,
      "grad_norm": 1.90510094165802,
      "learning_rate": 0.00011742243436754176,
      "loss": 2.0016,
      "step": 492
    },
    {
      "epoch": 0.017655379877164395,
      "grad_norm": 1.4178807735443115,
      "learning_rate": 0.00011766109785202863,
      "loss": 2.1013,
      "step": 493
    },
    {
      "epoch": 0.017691192006732682,
      "grad_norm": 3.005915641784668,
      "learning_rate": 0.00011789976133651552,
      "loss": 2.3571,
      "step": 494
    },
    {
      "epoch": 0.017727004136300965,
      "grad_norm": 1.7287548780441284,
      "learning_rate": 0.00011813842482100239,
      "loss": 1.8362,
      "step": 495
    },
    {
      "epoch": 0.01776281626586925,
      "grad_norm": 2.2629318237304688,
      "learning_rate": 0.00011837708830548926,
      "loss": 2.0679,
      "step": 496
    },
    {
      "epoch": 0.017798628395437534,
      "grad_norm": 2.3710145950317383,
      "learning_rate": 0.00011861575178997615,
      "loss": 2.2409,
      "step": 497
    },
    {
      "epoch": 0.01783444052500582,
      "grad_norm": 2.7783234119415283,
      "learning_rate": 0.00011885441527446302,
      "loss": 2.2683,
      "step": 498
    },
    {
      "epoch": 0.017870252654574103,
      "grad_norm": 2.343177556991577,
      "learning_rate": 0.00011909307875894989,
      "loss": 1.9737,
      "step": 499
    },
    {
      "epoch": 0.01790606478414239,
      "grad_norm": 1.3416262865066528,
      "learning_rate": 0.00011933174224343676,
      "loss": 1.7674,
      "step": 500
    },
    {
      "epoch": 0.017941876913710676,
      "grad_norm": 2.2444570064544678,
      "learning_rate": 0.00011957040572792364,
      "loss": 2.3294,
      "step": 501
    },
    {
      "epoch": 0.01797768904327896,
      "grad_norm": 2.3115077018737793,
      "learning_rate": 0.0001198090692124105,
      "loss": 2.3566,
      "step": 502
    },
    {
      "epoch": 0.018013501172847245,
      "grad_norm": 1.9268572330474854,
      "learning_rate": 0.00012004773269689737,
      "loss": 1.8323,
      "step": 503
    },
    {
      "epoch": 0.018049313302415528,
      "grad_norm": 3.032489061355591,
      "learning_rate": 0.00012028639618138427,
      "loss": 1.9294,
      "step": 504
    },
    {
      "epoch": 0.018085125431983814,
      "grad_norm": 1.1588308811187744,
      "learning_rate": 0.00012052505966587114,
      "loss": 1.9001,
      "step": 505
    },
    {
      "epoch": 0.018120937561552097,
      "grad_norm": 1.5301387310028076,
      "learning_rate": 0.000120763723150358,
      "loss": 1.9317,
      "step": 506
    },
    {
      "epoch": 0.018156749691120383,
      "grad_norm": 2.175445079803467,
      "learning_rate": 0.00012100238663484487,
      "loss": 1.7914,
      "step": 507
    },
    {
      "epoch": 0.018192561820688666,
      "grad_norm": 1.8067626953125,
      "learning_rate": 0.00012124105011933175,
      "loss": 1.9488,
      "step": 508
    },
    {
      "epoch": 0.018228373950256952,
      "grad_norm": 2.2004895210266113,
      "learning_rate": 0.00012147971360381862,
      "loss": 1.9453,
      "step": 509
    },
    {
      "epoch": 0.01826418607982524,
      "grad_norm": 1.2897497415542603,
      "learning_rate": 0.00012171837708830549,
      "loss": 1.8825,
      "step": 510
    },
    {
      "epoch": 0.01829999820939352,
      "grad_norm": 1.6512136459350586,
      "learning_rate": 0.00012195704057279236,
      "loss": 2.1278,
      "step": 511
    },
    {
      "epoch": 0.018335810338961808,
      "grad_norm": 1.7471429109573364,
      "learning_rate": 0.00012219570405727924,
      "loss": 1.994,
      "step": 512
    },
    {
      "epoch": 0.01837162246853009,
      "grad_norm": 1.75509774684906,
      "learning_rate": 0.0001224343675417661,
      "loss": 1.9485,
      "step": 513
    },
    {
      "epoch": 0.018407434598098377,
      "grad_norm": 1.647789478302002,
      "learning_rate": 0.00012267303102625297,
      "loss": 1.8526,
      "step": 514
    },
    {
      "epoch": 0.01844324672766666,
      "grad_norm": 1.4581859111785889,
      "learning_rate": 0.00012291169451073987,
      "loss": 1.9781,
      "step": 515
    },
    {
      "epoch": 0.018479058857234946,
      "grad_norm": 1.680022120475769,
      "learning_rate": 0.00012315035799522674,
      "loss": 1.9252,
      "step": 516
    },
    {
      "epoch": 0.01851487098680323,
      "grad_norm": 1.8121978044509888,
      "learning_rate": 0.0001233890214797136,
      "loss": 1.5773,
      "step": 517
    },
    {
      "epoch": 0.018550683116371515,
      "grad_norm": 2.650664806365967,
      "learning_rate": 0.00012362768496420047,
      "loss": 2.358,
      "step": 518
    },
    {
      "epoch": 0.0185864952459398,
      "grad_norm": 1.9430338144302368,
      "learning_rate": 0.00012386634844868737,
      "loss": 2.1443,
      "step": 519
    },
    {
      "epoch": 0.018622307375508084,
      "grad_norm": 1.728421688079834,
      "learning_rate": 0.00012410501193317423,
      "loss": 1.8808,
      "step": 520
    },
    {
      "epoch": 0.01865811950507637,
      "grad_norm": 2.1015219688415527,
      "learning_rate": 0.0001243436754176611,
      "loss": 1.8069,
      "step": 521
    },
    {
      "epoch": 0.018693931634644653,
      "grad_norm": 3.219109058380127,
      "learning_rate": 0.000124582338902148,
      "loss": 1.9532,
      "step": 522
    },
    {
      "epoch": 0.01872974376421294,
      "grad_norm": 1.7970722913742065,
      "learning_rate": 0.00012482100238663487,
      "loss": 2.1112,
      "step": 523
    },
    {
      "epoch": 0.018765555893781222,
      "grad_norm": 2.274960517883301,
      "learning_rate": 0.00012505966587112173,
      "loss": 1.6838,
      "step": 524
    },
    {
      "epoch": 0.01880136802334951,
      "grad_norm": 1.3818376064300537,
      "learning_rate": 0.0001252983293556086,
      "loss": 1.9166,
      "step": 525
    },
    {
      "epoch": 0.018837180152917795,
      "grad_norm": 1.6130000352859497,
      "learning_rate": 0.00012553699284009547,
      "loss": 2.2296,
      "step": 526
    },
    {
      "epoch": 0.018872992282486078,
      "grad_norm": 1.9274317026138306,
      "learning_rate": 0.00012577565632458234,
      "loss": 2.098,
      "step": 527
    },
    {
      "epoch": 0.018908804412054364,
      "grad_norm": 1.789971947669983,
      "learning_rate": 0.0001260143198090692,
      "loss": 1.7194,
      "step": 528
    },
    {
      "epoch": 0.018944616541622647,
      "grad_norm": 1.8698501586914062,
      "learning_rate": 0.00012625298329355607,
      "loss": 2.3349,
      "step": 529
    },
    {
      "epoch": 0.018980428671190933,
      "grad_norm": 1.8852440118789673,
      "learning_rate": 0.00012649164677804297,
      "loss": 1.918,
      "step": 530
    },
    {
      "epoch": 0.019016240800759216,
      "grad_norm": 2.258939743041992,
      "learning_rate": 0.00012673031026252983,
      "loss": 2.038,
      "step": 531
    },
    {
      "epoch": 0.019052052930327502,
      "grad_norm": 1.3770029544830322,
      "learning_rate": 0.0001269689737470167,
      "loss": 1.8361,
      "step": 532
    },
    {
      "epoch": 0.019087865059895785,
      "grad_norm": 1.3910750150680542,
      "learning_rate": 0.0001272076372315036,
      "loss": 2.0795,
      "step": 533
    },
    {
      "epoch": 0.01912367718946407,
      "grad_norm": 1.963090181350708,
      "learning_rate": 0.00012744630071599047,
      "loss": 2.2053,
      "step": 534
    },
    {
      "epoch": 0.019159489319032358,
      "grad_norm": 2.1430723667144775,
      "learning_rate": 0.00012768496420047733,
      "loss": 1.7194,
      "step": 535
    },
    {
      "epoch": 0.01919530144860064,
      "grad_norm": 1.6133919954299927,
      "learning_rate": 0.0001279236276849642,
      "loss": 1.821,
      "step": 536
    },
    {
      "epoch": 0.019231113578168927,
      "grad_norm": 1.79860258102417,
      "learning_rate": 0.0001281622911694511,
      "loss": 1.9787,
      "step": 537
    },
    {
      "epoch": 0.01926692570773721,
      "grad_norm": 1.847287893295288,
      "learning_rate": 0.00012840095465393796,
      "loss": 2.0806,
      "step": 538
    },
    {
      "epoch": 0.019302737837305496,
      "grad_norm": 1.458694338798523,
      "learning_rate": 0.00012863961813842483,
      "loss": 2.1716,
      "step": 539
    },
    {
      "epoch": 0.01933854996687378,
      "grad_norm": 2.063096761703491,
      "learning_rate": 0.0001288782816229117,
      "loss": 1.9751,
      "step": 540
    },
    {
      "epoch": 0.019374362096442065,
      "grad_norm": 1.8983663320541382,
      "learning_rate": 0.00012911694510739857,
      "loss": 1.9129,
      "step": 541
    },
    {
      "epoch": 0.019410174226010348,
      "grad_norm": 2.726100444793701,
      "learning_rate": 0.00012935560859188543,
      "loss": 2.3521,
      "step": 542
    },
    {
      "epoch": 0.019445986355578634,
      "grad_norm": 1.60848069190979,
      "learning_rate": 0.0001295942720763723,
      "loss": 1.673,
      "step": 543
    },
    {
      "epoch": 0.01948179848514692,
      "grad_norm": 1.7081117630004883,
      "learning_rate": 0.0001298329355608592,
      "loss": 2.2052,
      "step": 544
    },
    {
      "epoch": 0.019517610614715204,
      "grad_norm": 1.2822455167770386,
      "learning_rate": 0.00013007159904534607,
      "loss": 1.8536,
      "step": 545
    },
    {
      "epoch": 0.01955342274428349,
      "grad_norm": 1.9416059255599976,
      "learning_rate": 0.00013031026252983293,
      "loss": 2.3582,
      "step": 546
    },
    {
      "epoch": 0.019589234873851773,
      "grad_norm": 1.0728063583374023,
      "learning_rate": 0.0001305489260143198,
      "loss": 1.6316,
      "step": 547
    },
    {
      "epoch": 0.01962504700342006,
      "grad_norm": 1.4513641595840454,
      "learning_rate": 0.0001307875894988067,
      "loss": 1.9188,
      "step": 548
    },
    {
      "epoch": 0.019660859132988342,
      "grad_norm": 1.5579917430877686,
      "learning_rate": 0.00013102625298329356,
      "loss": 1.758,
      "step": 549
    },
    {
      "epoch": 0.019696671262556628,
      "grad_norm": 1.304106593132019,
      "learning_rate": 0.00013126491646778043,
      "loss": 1.9442,
      "step": 550
    },
    {
      "epoch": 0.01973248339212491,
      "grad_norm": 1.6284905672073364,
      "learning_rate": 0.00013150357995226733,
      "loss": 2.0928,
      "step": 551
    },
    {
      "epoch": 0.019768295521693197,
      "grad_norm": 2.598228931427002,
      "learning_rate": 0.0001317422434367542,
      "loss": 1.9553,
      "step": 552
    },
    {
      "epoch": 0.019804107651261484,
      "grad_norm": 2.1815645694732666,
      "learning_rate": 0.00013198090692124106,
      "loss": 2.2695,
      "step": 553
    },
    {
      "epoch": 0.019839919780829766,
      "grad_norm": 1.7056655883789062,
      "learning_rate": 0.00013221957040572793,
      "loss": 1.9625,
      "step": 554
    },
    {
      "epoch": 0.019875731910398053,
      "grad_norm": 1.9924041032791138,
      "learning_rate": 0.00013245823389021482,
      "loss": 2.1729,
      "step": 555
    },
    {
      "epoch": 0.019911544039966336,
      "grad_norm": 1.7806648015975952,
      "learning_rate": 0.00013269689737470167,
      "loss": 1.845,
      "step": 556
    },
    {
      "epoch": 0.019947356169534622,
      "grad_norm": 1.9855705499649048,
      "learning_rate": 0.00013293556085918853,
      "loss": 1.8858,
      "step": 557
    },
    {
      "epoch": 0.019983168299102905,
      "grad_norm": 2.107879161834717,
      "learning_rate": 0.0001331742243436754,
      "loss": 2.1954,
      "step": 558
    },
    {
      "epoch": 0.02001898042867119,
      "grad_norm": 1.5310916900634766,
      "learning_rate": 0.0001334128878281623,
      "loss": 2.0236,
      "step": 559
    },
    {
      "epoch": 0.020054792558239477,
      "grad_norm": 1.5715898275375366,
      "learning_rate": 0.00013365155131264916,
      "loss": 2.3037,
      "step": 560
    },
    {
      "epoch": 0.02009060468780776,
      "grad_norm": 1.846575379371643,
      "learning_rate": 0.00013389021479713603,
      "loss": 1.9472,
      "step": 561
    },
    {
      "epoch": 0.020126416817376046,
      "grad_norm": 1.5027644634246826,
      "learning_rate": 0.00013412887828162293,
      "loss": 1.8978,
      "step": 562
    },
    {
      "epoch": 0.02016222894694433,
      "grad_norm": 2.3320515155792236,
      "learning_rate": 0.0001343675417661098,
      "loss": 2.0998,
      "step": 563
    },
    {
      "epoch": 0.020198041076512616,
      "grad_norm": 1.4879544973373413,
      "learning_rate": 0.00013460620525059666,
      "loss": 2.0195,
      "step": 564
    },
    {
      "epoch": 0.0202338532060809,
      "grad_norm": 1.6790188550949097,
      "learning_rate": 0.00013484486873508353,
      "loss": 2.0034,
      "step": 565
    },
    {
      "epoch": 0.020269665335649185,
      "grad_norm": 1.3368083238601685,
      "learning_rate": 0.00013508353221957042,
      "loss": 1.853,
      "step": 566
    },
    {
      "epoch": 0.020305477465217468,
      "grad_norm": 2.40515398979187,
      "learning_rate": 0.0001353221957040573,
      "loss": 1.6864,
      "step": 567
    },
    {
      "epoch": 0.020341289594785754,
      "grad_norm": 3.6255276203155518,
      "learning_rate": 0.00013556085918854416,
      "loss": 2.0015,
      "step": 568
    },
    {
      "epoch": 0.02037710172435404,
      "grad_norm": 1.3990453481674194,
      "learning_rate": 0.00013579952267303105,
      "loss": 1.9736,
      "step": 569
    },
    {
      "epoch": 0.020412913853922323,
      "grad_norm": 1.815877079963684,
      "learning_rate": 0.00013603818615751792,
      "loss": 2.0062,
      "step": 570
    },
    {
      "epoch": 0.02044872598349061,
      "grad_norm": 2.521155595779419,
      "learning_rate": 0.0001362768496420048,
      "loss": 2.2658,
      "step": 571
    },
    {
      "epoch": 0.020484538113058892,
      "grad_norm": 1.6638094186782837,
      "learning_rate": 0.00013651551312649166,
      "loss": 2.0926,
      "step": 572
    },
    {
      "epoch": 0.02052035024262718,
      "grad_norm": 1.9844567775726318,
      "learning_rate": 0.00013675417661097853,
      "loss": 2.2214,
      "step": 573
    },
    {
      "epoch": 0.02055616237219546,
      "grad_norm": 1.8711490631103516,
      "learning_rate": 0.0001369928400954654,
      "loss": 1.9504,
      "step": 574
    },
    {
      "epoch": 0.020591974501763748,
      "grad_norm": 2.035768985748291,
      "learning_rate": 0.00013723150357995226,
      "loss": 1.8715,
      "step": 575
    },
    {
      "epoch": 0.02062778663133203,
      "grad_norm": 1.6506410837173462,
      "learning_rate": 0.00013747016706443913,
      "loss": 2.1253,
      "step": 576
    },
    {
      "epoch": 0.020663598760900317,
      "grad_norm": 1.5059458017349243,
      "learning_rate": 0.00013770883054892602,
      "loss": 1.9463,
      "step": 577
    },
    {
      "epoch": 0.020699410890468603,
      "grad_norm": 2.002347230911255,
      "learning_rate": 0.0001379474940334129,
      "loss": 2.3318,
      "step": 578
    },
    {
      "epoch": 0.020735223020036886,
      "grad_norm": 2.4913315773010254,
      "learning_rate": 0.00013818615751789976,
      "loss": 2.0054,
      "step": 579
    },
    {
      "epoch": 0.020771035149605172,
      "grad_norm": 1.9572597742080688,
      "learning_rate": 0.00013842482100238665,
      "loss": 2.24,
      "step": 580
    },
    {
      "epoch": 0.020806847279173455,
      "grad_norm": 2.6330034732818604,
      "learning_rate": 0.00013866348448687352,
      "loss": 2.0444,
      "step": 581
    },
    {
      "epoch": 0.02084265940874174,
      "grad_norm": 2.075014591217041,
      "learning_rate": 0.0001389021479713604,
      "loss": 2.1214,
      "step": 582
    },
    {
      "epoch": 0.020878471538310024,
      "grad_norm": 1.4259493350982666,
      "learning_rate": 0.00013914081145584726,
      "loss": 1.6096,
      "step": 583
    },
    {
      "epoch": 0.02091428366787831,
      "grad_norm": 2.2150115966796875,
      "learning_rate": 0.00013937947494033415,
      "loss": 2.0084,
      "step": 584
    },
    {
      "epoch": 0.020950095797446597,
      "grad_norm": 2.1223316192626953,
      "learning_rate": 0.00013961813842482102,
      "loss": 2.2784,
      "step": 585
    },
    {
      "epoch": 0.02098590792701488,
      "grad_norm": 1.5994501113891602,
      "learning_rate": 0.0001398568019093079,
      "loss": 2.2577,
      "step": 586
    },
    {
      "epoch": 0.021021720056583166,
      "grad_norm": 1.6532303094863892,
      "learning_rate": 0.00014009546539379476,
      "loss": 1.965,
      "step": 587
    },
    {
      "epoch": 0.02105753218615145,
      "grad_norm": 1.9446479082107544,
      "learning_rate": 0.00014033412887828162,
      "loss": 2.207,
      "step": 588
    },
    {
      "epoch": 0.021093344315719735,
      "grad_norm": 1.3874430656433105,
      "learning_rate": 0.0001405727923627685,
      "loss": 2.0884,
      "step": 589
    },
    {
      "epoch": 0.021129156445288018,
      "grad_norm": 1.4150493144989014,
      "learning_rate": 0.00014081145584725536,
      "loss": 1.8492,
      "step": 590
    },
    {
      "epoch": 0.021164968574856304,
      "grad_norm": 2.022547483444214,
      "learning_rate": 0.00014105011933174225,
      "loss": 1.9806,
      "step": 591
    },
    {
      "epoch": 0.021200780704424587,
      "grad_norm": 2.3493235111236572,
      "learning_rate": 0.00014128878281622912,
      "loss": 2.4232,
      "step": 592
    },
    {
      "epoch": 0.021236592833992873,
      "grad_norm": 1.6374825239181519,
      "learning_rate": 0.000141527446300716,
      "loss": 1.9931,
      "step": 593
    },
    {
      "epoch": 0.02127240496356116,
      "grad_norm": 1.9927897453308105,
      "learning_rate": 0.00014176610978520286,
      "loss": 1.9837,
      "step": 594
    },
    {
      "epoch": 0.021308217093129442,
      "grad_norm": 2.0843703746795654,
      "learning_rate": 0.00014200477326968975,
      "loss": 2.2173,
      "step": 595
    },
    {
      "epoch": 0.02134402922269773,
      "grad_norm": 2.041806936264038,
      "learning_rate": 0.00014224343675417662,
      "loss": 2.1302,
      "step": 596
    },
    {
      "epoch": 0.02137984135226601,
      "grad_norm": 2.7673728466033936,
      "learning_rate": 0.0001424821002386635,
      "loss": 2.6909,
      "step": 597
    },
    {
      "epoch": 0.021415653481834298,
      "grad_norm": 1.646714210510254,
      "learning_rate": 0.00014272076372315038,
      "loss": 2.1574,
      "step": 598
    },
    {
      "epoch": 0.02145146561140258,
      "grad_norm": 1.549854040145874,
      "learning_rate": 0.00014295942720763725,
      "loss": 2.1436,
      "step": 599
    },
    {
      "epoch": 0.021487277740970867,
      "grad_norm": 2.0806381702423096,
      "learning_rate": 0.00014319809069212412,
      "loss": 1.6527,
      "step": 600
    },
    {
      "epoch": 0.02152308987053915,
      "grad_norm": 1.1766951084136963,
      "learning_rate": 0.000143436754176611,
      "loss": 1.5967,
      "step": 601
    },
    {
      "epoch": 0.021558902000107436,
      "grad_norm": 1.7814152240753174,
      "learning_rate": 0.00014367541766109785,
      "loss": 1.8676,
      "step": 602
    },
    {
      "epoch": 0.021594714129675723,
      "grad_norm": 1.4170676469802856,
      "learning_rate": 0.00014391408114558472,
      "loss": 1.9861,
      "step": 603
    },
    {
      "epoch": 0.021630526259244005,
      "grad_norm": 1.3304917812347412,
      "learning_rate": 0.0001441527446300716,
      "loss": 1.7847,
      "step": 604
    },
    {
      "epoch": 0.02166633838881229,
      "grad_norm": 1.353492259979248,
      "learning_rate": 0.00014439140811455846,
      "loss": 1.9458,
      "step": 605
    },
    {
      "epoch": 0.021702150518380574,
      "grad_norm": 1.9017729759216309,
      "learning_rate": 0.00014463007159904535,
      "loss": 2.2398,
      "step": 606
    },
    {
      "epoch": 0.02173796264794886,
      "grad_norm": 1.3059676885604858,
      "learning_rate": 0.00014486873508353222,
      "loss": 1.7462,
      "step": 607
    },
    {
      "epoch": 0.021773774777517144,
      "grad_norm": 1.7438218593597412,
      "learning_rate": 0.0001451073985680191,
      "loss": 2.2967,
      "step": 608
    },
    {
      "epoch": 0.02180958690708543,
      "grad_norm": 1.7737010717391968,
      "learning_rate": 0.00014534606205250598,
      "loss": 1.9835,
      "step": 609
    },
    {
      "epoch": 0.021845399036653716,
      "grad_norm": 1.6507363319396973,
      "learning_rate": 0.00014558472553699285,
      "loss": 1.7111,
      "step": 610
    },
    {
      "epoch": 0.021881211166222,
      "grad_norm": 1.657170295715332,
      "learning_rate": 0.00014582338902147972,
      "loss": 2.1349,
      "step": 611
    },
    {
      "epoch": 0.021917023295790285,
      "grad_norm": 2.128286838531494,
      "learning_rate": 0.0001460620525059666,
      "loss": 2.0744,
      "step": 612
    },
    {
      "epoch": 0.021952835425358568,
      "grad_norm": 1.6936157941818237,
      "learning_rate": 0.00014630071599045348,
      "loss": 2.2095,
      "step": 613
    },
    {
      "epoch": 0.021988647554926855,
      "grad_norm": 1.3525550365447998,
      "learning_rate": 0.00014653937947494035,
      "loss": 1.9486,
      "step": 614
    },
    {
      "epoch": 0.022024459684495137,
      "grad_norm": 1.4725382328033447,
      "learning_rate": 0.00014677804295942722,
      "loss": 2.0803,
      "step": 615
    },
    {
      "epoch": 0.022060271814063424,
      "grad_norm": 1.4054583311080933,
      "learning_rate": 0.00014701670644391409,
      "loss": 1.6478,
      "step": 616
    },
    {
      "epoch": 0.022096083943631706,
      "grad_norm": 1.6811931133270264,
      "learning_rate": 0.00014725536992840095,
      "loss": 2.2387,
      "step": 617
    },
    {
      "epoch": 0.022131896073199993,
      "grad_norm": 2.0554606914520264,
      "learning_rate": 0.00014749403341288782,
      "loss": 1.9838,
      "step": 618
    },
    {
      "epoch": 0.02216770820276828,
      "grad_norm": 1.9794045686721802,
      "learning_rate": 0.0001477326968973747,
      "loss": 2.1976,
      "step": 619
    },
    {
      "epoch": 0.022203520332336562,
      "grad_norm": 1.6070706844329834,
      "learning_rate": 0.00014797136038186158,
      "loss": 2.0375,
      "step": 620
    },
    {
      "epoch": 0.022239332461904848,
      "grad_norm": 1.8594540357589722,
      "learning_rate": 0.00014821002386634845,
      "loss": 2.2466,
      "step": 621
    },
    {
      "epoch": 0.02227514459147313,
      "grad_norm": 2.2988474369049072,
      "learning_rate": 0.00014844868735083532,
      "loss": 2.0812,
      "step": 622
    },
    {
      "epoch": 0.022310956721041417,
      "grad_norm": 1.9472583532333374,
      "learning_rate": 0.0001486873508353222,
      "loss": 1.7122,
      "step": 623
    },
    {
      "epoch": 0.0223467688506097,
      "grad_norm": 2.4133808612823486,
      "learning_rate": 0.00014892601431980908,
      "loss": 2.0218,
      "step": 624
    },
    {
      "epoch": 0.022382580980177987,
      "grad_norm": 1.960550308227539,
      "learning_rate": 0.00014916467780429595,
      "loss": 1.6735,
      "step": 625
    },
    {
      "epoch": 0.02241839310974627,
      "grad_norm": 1.429086446762085,
      "learning_rate": 0.00014940334128878282,
      "loss": 1.8953,
      "step": 626
    },
    {
      "epoch": 0.022454205239314556,
      "grad_norm": 2.099316120147705,
      "learning_rate": 0.0001496420047732697,
      "loss": 2.0225,
      "step": 627
    },
    {
      "epoch": 0.022490017368882842,
      "grad_norm": 1.6147634983062744,
      "learning_rate": 0.00014988066825775658,
      "loss": 1.9627,
      "step": 628
    },
    {
      "epoch": 0.022525829498451125,
      "grad_norm": 3.438903570175171,
      "learning_rate": 0.00015011933174224345,
      "loss": 2.2949,
      "step": 629
    },
    {
      "epoch": 0.02256164162801941,
      "grad_norm": 2.2410268783569336,
      "learning_rate": 0.00015035799522673032,
      "loss": 2.1214,
      "step": 630
    },
    {
      "epoch": 0.022597453757587694,
      "grad_norm": 1.8363879919052124,
      "learning_rate": 0.0001505966587112172,
      "loss": 1.6187,
      "step": 631
    },
    {
      "epoch": 0.02263326588715598,
      "grad_norm": 1.890332579612732,
      "learning_rate": 0.00015083532219570408,
      "loss": 2.244,
      "step": 632
    },
    {
      "epoch": 0.022669078016724263,
      "grad_norm": 2.458477258682251,
      "learning_rate": 0.00015107398568019092,
      "loss": 2.4482,
      "step": 633
    },
    {
      "epoch": 0.02270489014629255,
      "grad_norm": 1.5786428451538086,
      "learning_rate": 0.00015131264916467781,
      "loss": 1.9677,
      "step": 634
    },
    {
      "epoch": 0.022740702275860836,
      "grad_norm": 1.7664666175842285,
      "learning_rate": 0.00015155131264916468,
      "loss": 2.1842,
      "step": 635
    },
    {
      "epoch": 0.02277651440542912,
      "grad_norm": 1.6327040195465088,
      "learning_rate": 0.00015178997613365155,
      "loss": 1.7608,
      "step": 636
    },
    {
      "epoch": 0.022812326534997405,
      "grad_norm": 1.7589260339736938,
      "learning_rate": 0.00015202863961813842,
      "loss": 1.9472,
      "step": 637
    },
    {
      "epoch": 0.022848138664565688,
      "grad_norm": 3.3443808555603027,
      "learning_rate": 0.0001522673031026253,
      "loss": 1.894,
      "step": 638
    },
    {
      "epoch": 0.022883950794133974,
      "grad_norm": 1.7203983068466187,
      "learning_rate": 0.00015250596658711218,
      "loss": 1.616,
      "step": 639
    },
    {
      "epoch": 0.022919762923702257,
      "grad_norm": 2.1859230995178223,
      "learning_rate": 0.00015274463007159905,
      "loss": 1.9724,
      "step": 640
    },
    {
      "epoch": 0.022955575053270543,
      "grad_norm": 1.5211198329925537,
      "learning_rate": 0.00015298329355608592,
      "loss": 2.0265,
      "step": 641
    },
    {
      "epoch": 0.022991387182838826,
      "grad_norm": 1.601146936416626,
      "learning_rate": 0.0001532219570405728,
      "loss": 1.9278,
      "step": 642
    },
    {
      "epoch": 0.023027199312407112,
      "grad_norm": 1.6272515058517456,
      "learning_rate": 0.00015346062052505968,
      "loss": 1.8003,
      "step": 643
    },
    {
      "epoch": 0.0230630114419754,
      "grad_norm": 2.8821325302124023,
      "learning_rate": 0.00015369928400954655,
      "loss": 2.173,
      "step": 644
    },
    {
      "epoch": 0.02309882357154368,
      "grad_norm": 1.6864391565322876,
      "learning_rate": 0.00015393794749403344,
      "loss": 2.0027,
      "step": 645
    },
    {
      "epoch": 0.023134635701111968,
      "grad_norm": 1.6988509893417358,
      "learning_rate": 0.0001541766109785203,
      "loss": 2.1044,
      "step": 646
    },
    {
      "epoch": 0.02317044783068025,
      "grad_norm": 1.5111042261123657,
      "learning_rate": 0.00015441527446300718,
      "loss": 1.9445,
      "step": 647
    },
    {
      "epoch": 0.023206259960248537,
      "grad_norm": 1.5460760593414307,
      "learning_rate": 0.00015465393794749404,
      "loss": 1.9057,
      "step": 648
    },
    {
      "epoch": 0.02324207208981682,
      "grad_norm": 2.1058406829833984,
      "learning_rate": 0.0001548926014319809,
      "loss": 2.2557,
      "step": 649
    },
    {
      "epoch": 0.023277884219385106,
      "grad_norm": 2.062448263168335,
      "learning_rate": 0.00015513126491646778,
      "loss": 2.3574,
      "step": 650
    },
    {
      "epoch": 0.02331369634895339,
      "grad_norm": 1.2844550609588623,
      "learning_rate": 0.00015536992840095465,
      "loss": 2.0645,
      "step": 651
    },
    {
      "epoch": 0.023349508478521675,
      "grad_norm": 1.7018535137176514,
      "learning_rate": 0.00015560859188544154,
      "loss": 2.0612,
      "step": 652
    },
    {
      "epoch": 0.02338532060808996,
      "grad_norm": 2.8740715980529785,
      "learning_rate": 0.0001558472553699284,
      "loss": 2.0078,
      "step": 653
    },
    {
      "epoch": 0.023421132737658244,
      "grad_norm": 2.4455902576446533,
      "learning_rate": 0.00015608591885441528,
      "loss": 1.9655,
      "step": 654
    },
    {
      "epoch": 0.02345694486722653,
      "grad_norm": 1.9727590084075928,
      "learning_rate": 0.00015632458233890215,
      "loss": 2.1078,
      "step": 655
    },
    {
      "epoch": 0.023492756996794813,
      "grad_norm": 1.6687122583389282,
      "learning_rate": 0.00015656324582338904,
      "loss": 2.0421,
      "step": 656
    },
    {
      "epoch": 0.0235285691263631,
      "grad_norm": 2.6429245471954346,
      "learning_rate": 0.0001568019093078759,
      "loss": 2.3127,
      "step": 657
    },
    {
      "epoch": 0.023564381255931383,
      "grad_norm": 2.1367905139923096,
      "learning_rate": 0.00015704057279236278,
      "loss": 2.2663,
      "step": 658
    },
    {
      "epoch": 0.02360019338549967,
      "grad_norm": 1.8748291730880737,
      "learning_rate": 0.00015727923627684964,
      "loss": 1.6464,
      "step": 659
    },
    {
      "epoch": 0.023636005515067955,
      "grad_norm": 1.3424737453460693,
      "learning_rate": 0.00015751789976133654,
      "loss": 1.9569,
      "step": 660
    },
    {
      "epoch": 0.023671817644636238,
      "grad_norm": 1.6000856161117554,
      "learning_rate": 0.0001577565632458234,
      "loss": 1.9856,
      "step": 661
    },
    {
      "epoch": 0.023707629774204524,
      "grad_norm": 1.6837635040283203,
      "learning_rate": 0.00015799522673031027,
      "loss": 1.8739,
      "step": 662
    },
    {
      "epoch": 0.023743441903772807,
      "grad_norm": 1.6797798871994019,
      "learning_rate": 0.00015823389021479714,
      "loss": 1.9404,
      "step": 663
    },
    {
      "epoch": 0.023779254033341093,
      "grad_norm": 1.217158317565918,
      "learning_rate": 0.000158472553699284,
      "loss": 2.0562,
      "step": 664
    },
    {
      "epoch": 0.023815066162909376,
      "grad_norm": 1.4763151407241821,
      "learning_rate": 0.00015871121718377088,
      "loss": 1.8707,
      "step": 665
    },
    {
      "epoch": 0.023850878292477663,
      "grad_norm": 1.5236214399337769,
      "learning_rate": 0.00015894988066825775,
      "loss": 1.7224,
      "step": 666
    },
    {
      "epoch": 0.023886690422045945,
      "grad_norm": 1.8331769704818726,
      "learning_rate": 0.00015918854415274464,
      "loss": 2.1492,
      "step": 667
    },
    {
      "epoch": 0.02392250255161423,
      "grad_norm": 1.3603184223175049,
      "learning_rate": 0.0001594272076372315,
      "loss": 2.0236,
      "step": 668
    },
    {
      "epoch": 0.023958314681182518,
      "grad_norm": 1.4922575950622559,
      "learning_rate": 0.00015966587112171838,
      "loss": 1.7288,
      "step": 669
    },
    {
      "epoch": 0.0239941268107508,
      "grad_norm": 1.5984807014465332,
      "learning_rate": 0.00015990453460620524,
      "loss": 2.249,
      "step": 670
    },
    {
      "epoch": 0.024029938940319087,
      "grad_norm": 1.735472321510315,
      "learning_rate": 0.00016014319809069214,
      "loss": 1.7396,
      "step": 671
    },
    {
      "epoch": 0.02406575106988737,
      "grad_norm": 1.040790319442749,
      "learning_rate": 0.000160381861575179,
      "loss": 1.7578,
      "step": 672
    },
    {
      "epoch": 0.024101563199455656,
      "grad_norm": 1.53122878074646,
      "learning_rate": 0.00016062052505966587,
      "loss": 1.7265,
      "step": 673
    },
    {
      "epoch": 0.02413737532902394,
      "grad_norm": 3.3196752071380615,
      "learning_rate": 0.00016085918854415277,
      "loss": 2.035,
      "step": 674
    },
    {
      "epoch": 0.024173187458592225,
      "grad_norm": 1.471759557723999,
      "learning_rate": 0.00016109785202863964,
      "loss": 1.9174,
      "step": 675
    },
    {
      "epoch": 0.02420899958816051,
      "grad_norm": 1.670170545578003,
      "learning_rate": 0.0001613365155131265,
      "loss": 2.0755,
      "step": 676
    },
    {
      "epoch": 0.024244811717728795,
      "grad_norm": 1.6233552694320679,
      "learning_rate": 0.00016157517899761337,
      "loss": 1.8244,
      "step": 677
    },
    {
      "epoch": 0.02428062384729708,
      "grad_norm": 3.0857901573181152,
      "learning_rate": 0.00016181384248210024,
      "loss": 2.8201,
      "step": 678
    },
    {
      "epoch": 0.024316435976865364,
      "grad_norm": 1.895977258682251,
      "learning_rate": 0.0001620525059665871,
      "loss": 1.9378,
      "step": 679
    },
    {
      "epoch": 0.02435224810643365,
      "grad_norm": 1.9350051879882812,
      "learning_rate": 0.00016229116945107398,
      "loss": 2.0559,
      "step": 680
    },
    {
      "epoch": 0.024388060236001933,
      "grad_norm": 2.0227410793304443,
      "learning_rate": 0.00016252983293556087,
      "loss": 2.2404,
      "step": 681
    },
    {
      "epoch": 0.02442387236557022,
      "grad_norm": 1.1764206886291504,
      "learning_rate": 0.00016276849642004774,
      "loss": 1.8332,
      "step": 682
    },
    {
      "epoch": 0.024459684495138502,
      "grad_norm": 2.8057363033294678,
      "learning_rate": 0.0001630071599045346,
      "loss": 2.0951,
      "step": 683
    },
    {
      "epoch": 0.02449549662470679,
      "grad_norm": 1.5808459520339966,
      "learning_rate": 0.00016324582338902147,
      "loss": 2.0856,
      "step": 684
    },
    {
      "epoch": 0.024531308754275075,
      "grad_norm": 2.126241683959961,
      "learning_rate": 0.00016348448687350837,
      "loss": 1.9048,
      "step": 685
    },
    {
      "epoch": 0.024567120883843357,
      "grad_norm": 1.388526201248169,
      "learning_rate": 0.00016372315035799524,
      "loss": 1.8886,
      "step": 686
    },
    {
      "epoch": 0.024602933013411644,
      "grad_norm": 1.420893907546997,
      "learning_rate": 0.0001639618138424821,
      "loss": 1.9233,
      "step": 687
    },
    {
      "epoch": 0.024638745142979927,
      "grad_norm": 1.5350810289382935,
      "learning_rate": 0.00016420047732696897,
      "loss": 2.1028,
      "step": 688
    },
    {
      "epoch": 0.024674557272548213,
      "grad_norm": 1.220397710800171,
      "learning_rate": 0.00016443914081145587,
      "loss": 1.9763,
      "step": 689
    },
    {
      "epoch": 0.024710369402116496,
      "grad_norm": 1.369965672492981,
      "learning_rate": 0.00016467780429594274,
      "loss": 2.074,
      "step": 690
    },
    {
      "epoch": 0.024746181531684782,
      "grad_norm": 2.692288398742676,
      "learning_rate": 0.0001649164677804296,
      "loss": 2.4718,
      "step": 691
    },
    {
      "epoch": 0.024781993661253065,
      "grad_norm": 1.9050400257110596,
      "learning_rate": 0.0001651551312649165,
      "loss": 2.2467,
      "step": 692
    },
    {
      "epoch": 0.02481780579082135,
      "grad_norm": 1.706419825553894,
      "learning_rate": 0.00016539379474940334,
      "loss": 1.7131,
      "step": 693
    },
    {
      "epoch": 0.024853617920389637,
      "grad_norm": 1.5882086753845215,
      "learning_rate": 0.0001656324582338902,
      "loss": 2.0457,
      "step": 694
    },
    {
      "epoch": 0.02488943004995792,
      "grad_norm": 1.5576844215393066,
      "learning_rate": 0.00016587112171837707,
      "loss": 1.7335,
      "step": 695
    },
    {
      "epoch": 0.024925242179526207,
      "grad_norm": 1.3042941093444824,
      "learning_rate": 0.00016610978520286397,
      "loss": 1.7833,
      "step": 696
    },
    {
      "epoch": 0.02496105430909449,
      "grad_norm": 1.5248041152954102,
      "learning_rate": 0.00016634844868735084,
      "loss": 1.8416,
      "step": 697
    },
    {
      "epoch": 0.024996866438662776,
      "grad_norm": 3.5359699726104736,
      "learning_rate": 0.0001665871121718377,
      "loss": 2.8096,
      "step": 698
    },
    {
      "epoch": 0.02503267856823106,
      "grad_norm": 1.6402137279510498,
      "learning_rate": 0.0001668257756563246,
      "loss": 2.0446,
      "step": 699
    },
    {
      "epoch": 0.025068490697799345,
      "grad_norm": 1.531969428062439,
      "learning_rate": 0.00016706443914081147,
      "loss": 1.9463,
      "step": 700
    },
    {
      "epoch": 0.025104302827367628,
      "grad_norm": 2.4962432384490967,
      "learning_rate": 0.00016730310262529834,
      "loss": 2.3453,
      "step": 701
    },
    {
      "epoch": 0.025140114956935914,
      "grad_norm": 1.790243148803711,
      "learning_rate": 0.0001675417661097852,
      "loss": 2.0987,
      "step": 702
    },
    {
      "epoch": 0.0251759270865042,
      "grad_norm": 1.7493826150894165,
      "learning_rate": 0.0001677804295942721,
      "loss": 1.7627,
      "step": 703
    },
    {
      "epoch": 0.025211739216072483,
      "grad_norm": 1.5792869329452515,
      "learning_rate": 0.00016801909307875897,
      "loss": 1.7321,
      "step": 704
    },
    {
      "epoch": 0.02524755134564077,
      "grad_norm": 1.3130288124084473,
      "learning_rate": 0.00016825775656324583,
      "loss": 2.0553,
      "step": 705
    },
    {
      "epoch": 0.025283363475209052,
      "grad_norm": 1.769005537033081,
      "learning_rate": 0.0001684964200477327,
      "loss": 2.0413,
      "step": 706
    },
    {
      "epoch": 0.02531917560477734,
      "grad_norm": 1.4005001783370972,
      "learning_rate": 0.0001687350835322196,
      "loss": 1.8354,
      "step": 707
    },
    {
      "epoch": 0.02535498773434562,
      "grad_norm": 1.722219705581665,
      "learning_rate": 0.00016897374701670646,
      "loss": 2.1277,
      "step": 708
    },
    {
      "epoch": 0.025390799863913908,
      "grad_norm": 1.9481533765792847,
      "learning_rate": 0.00016921241050119333,
      "loss": 2.1439,
      "step": 709
    },
    {
      "epoch": 0.025426611993482194,
      "grad_norm": 1.281445026397705,
      "learning_rate": 0.0001694510739856802,
      "loss": 2.1655,
      "step": 710
    },
    {
      "epoch": 0.025462424123050477,
      "grad_norm": 1.7158055305480957,
      "learning_rate": 0.00016968973747016707,
      "loss": 2.3323,
      "step": 711
    },
    {
      "epoch": 0.025498236252618763,
      "grad_norm": 1.4600533246994019,
      "learning_rate": 0.00016992840095465394,
      "loss": 1.9099,
      "step": 712
    },
    {
      "epoch": 0.025534048382187046,
      "grad_norm": 1.8186371326446533,
      "learning_rate": 0.0001701670644391408,
      "loss": 1.8916,
      "step": 713
    },
    {
      "epoch": 0.025569860511755332,
      "grad_norm": 1.6856729984283447,
      "learning_rate": 0.0001704057279236277,
      "loss": 1.8507,
      "step": 714
    },
    {
      "epoch": 0.025605672641323615,
      "grad_norm": 1.6164995431900024,
      "learning_rate": 0.00017064439140811457,
      "loss": 1.7116,
      "step": 715
    },
    {
      "epoch": 0.0256414847708919,
      "grad_norm": 1.3906581401824951,
      "learning_rate": 0.00017088305489260143,
      "loss": 2.0011,
      "step": 716
    },
    {
      "epoch": 0.025677296900460184,
      "grad_norm": 1.490162968635559,
      "learning_rate": 0.00017112171837708833,
      "loss": 2.2157,
      "step": 717
    },
    {
      "epoch": 0.02571310903002847,
      "grad_norm": 1.9008619785308838,
      "learning_rate": 0.0001713603818615752,
      "loss": 2.0239,
      "step": 718
    },
    {
      "epoch": 0.025748921159596757,
      "grad_norm": 1.612532377243042,
      "learning_rate": 0.00017159904534606206,
      "loss": 2.0756,
      "step": 719
    },
    {
      "epoch": 0.02578473328916504,
      "grad_norm": 1.4195812940597534,
      "learning_rate": 0.00017183770883054893,
      "loss": 2.0478,
      "step": 720
    },
    {
      "epoch": 0.025820545418733326,
      "grad_norm": 1.613661289215088,
      "learning_rate": 0.00017207637231503583,
      "loss": 2.3087,
      "step": 721
    },
    {
      "epoch": 0.02585635754830161,
      "grad_norm": 1.243811845779419,
      "learning_rate": 0.0001723150357995227,
      "loss": 1.9042,
      "step": 722
    },
    {
      "epoch": 0.025892169677869895,
      "grad_norm": 1.657544493675232,
      "learning_rate": 0.00017255369928400956,
      "loss": 1.998,
      "step": 723
    },
    {
      "epoch": 0.025927981807438178,
      "grad_norm": 2.2602198123931885,
      "learning_rate": 0.00017279236276849643,
      "loss": 2.2622,
      "step": 724
    },
    {
      "epoch": 0.025963793937006464,
      "grad_norm": 1.8787742853164673,
      "learning_rate": 0.0001730310262529833,
      "loss": 2.1541,
      "step": 725
    },
    {
      "epoch": 0.025999606066574747,
      "grad_norm": 1.5164096355438232,
      "learning_rate": 0.00017326968973747017,
      "loss": 1.8409,
      "step": 726
    },
    {
      "epoch": 0.026035418196143034,
      "grad_norm": 1.5882408618927002,
      "learning_rate": 0.00017350835322195703,
      "loss": 2.0037,
      "step": 727
    },
    {
      "epoch": 0.02607123032571132,
      "grad_norm": 2.327857732772827,
      "learning_rate": 0.00017374701670644393,
      "loss": 2.1629,
      "step": 728
    },
    {
      "epoch": 0.026107042455279603,
      "grad_norm": 2.476983070373535,
      "learning_rate": 0.0001739856801909308,
      "loss": 2.3573,
      "step": 729
    },
    {
      "epoch": 0.02614285458484789,
      "grad_norm": 2.473822593688965,
      "learning_rate": 0.00017422434367541766,
      "loss": 2.3806,
      "step": 730
    },
    {
      "epoch": 0.026178666714416172,
      "grad_norm": 2.2253236770629883,
      "learning_rate": 0.00017446300715990453,
      "loss": 2.1016,
      "step": 731
    },
    {
      "epoch": 0.026214478843984458,
      "grad_norm": 2.6160786151885986,
      "learning_rate": 0.00017470167064439143,
      "loss": 1.7466,
      "step": 732
    },
    {
      "epoch": 0.02625029097355274,
      "grad_norm": 1.6734675168991089,
      "learning_rate": 0.0001749403341288783,
      "loss": 1.9496,
      "step": 733
    },
    {
      "epoch": 0.026286103103121027,
      "grad_norm": 2.585047721862793,
      "learning_rate": 0.00017517899761336516,
      "loss": 2.2801,
      "step": 734
    },
    {
      "epoch": 0.026321915232689314,
      "grad_norm": 2.7207157611846924,
      "learning_rate": 0.00017541766109785203,
      "loss": 1.9879,
      "step": 735
    },
    {
      "epoch": 0.026357727362257596,
      "grad_norm": 1.3990424871444702,
      "learning_rate": 0.00017565632458233893,
      "loss": 2.206,
      "step": 736
    },
    {
      "epoch": 0.026393539491825883,
      "grad_norm": 1.7303647994995117,
      "learning_rate": 0.0001758949880668258,
      "loss": 2.1254,
      "step": 737
    },
    {
      "epoch": 0.026429351621394166,
      "grad_norm": 1.4347045421600342,
      "learning_rate": 0.00017613365155131266,
      "loss": 2.0242,
      "step": 738
    },
    {
      "epoch": 0.026465163750962452,
      "grad_norm": 2.1285247802734375,
      "learning_rate": 0.00017637231503579953,
      "loss": 1.7194,
      "step": 739
    },
    {
      "epoch": 0.026500975880530735,
      "grad_norm": 1.445741891860962,
      "learning_rate": 0.0001766109785202864,
      "loss": 1.951,
      "step": 740
    },
    {
      "epoch": 0.02653678801009902,
      "grad_norm": 1.6613547801971436,
      "learning_rate": 0.00017684964200477326,
      "loss": 1.9932,
      "step": 741
    },
    {
      "epoch": 0.026572600139667304,
      "grad_norm": 1.2473163604736328,
      "learning_rate": 0.00017708830548926013,
      "loss": 2.0447,
      "step": 742
    },
    {
      "epoch": 0.02660841226923559,
      "grad_norm": 1.4733924865722656,
      "learning_rate": 0.00017732696897374703,
      "loss": 1.9246,
      "step": 743
    },
    {
      "epoch": 0.026644224398803876,
      "grad_norm": 1.6123089790344238,
      "learning_rate": 0.0001775656324582339,
      "loss": 1.7595,
      "step": 744
    },
    {
      "epoch": 0.02668003652837216,
      "grad_norm": 1.358934998512268,
      "learning_rate": 0.00017780429594272076,
      "loss": 1.9185,
      "step": 745
    },
    {
      "epoch": 0.026715848657940446,
      "grad_norm": 1.765520691871643,
      "learning_rate": 0.00017804295942720766,
      "loss": 1.8516,
      "step": 746
    },
    {
      "epoch": 0.02675166078750873,
      "grad_norm": 2.589219331741333,
      "learning_rate": 0.00017828162291169453,
      "loss": 1.9495,
      "step": 747
    },
    {
      "epoch": 0.026787472917077015,
      "grad_norm": 1.5191903114318848,
      "learning_rate": 0.0001785202863961814,
      "loss": 2.121,
      "step": 748
    },
    {
      "epoch": 0.026823285046645298,
      "grad_norm": 2.5188381671905518,
      "learning_rate": 0.00017875894988066826,
      "loss": 2.154,
      "step": 749
    },
    {
      "epoch": 0.026859097176213584,
      "grad_norm": 1.1737215518951416,
      "learning_rate": 0.00017899761336515516,
      "loss": 1.6,
      "step": 750
    },
    {
      "epoch": 0.026894909305781867,
      "grad_norm": 1.4904850721359253,
      "learning_rate": 0.00017923627684964202,
      "loss": 2.1385,
      "step": 751
    },
    {
      "epoch": 0.026930721435350153,
      "grad_norm": 1.7092859745025635,
      "learning_rate": 0.0001794749403341289,
      "loss": 2.4254,
      "step": 752
    },
    {
      "epoch": 0.02696653356491844,
      "grad_norm": 1.8370356559753418,
      "learning_rate": 0.00017971360381861576,
      "loss": 2.0809,
      "step": 753
    },
    {
      "epoch": 0.027002345694486722,
      "grad_norm": 1.6396968364715576,
      "learning_rate": 0.00017995226730310263,
      "loss": 2.0178,
      "step": 754
    },
    {
      "epoch": 0.02703815782405501,
      "grad_norm": 1.7405651807785034,
      "learning_rate": 0.0001801909307875895,
      "loss": 2.1105,
      "step": 755
    },
    {
      "epoch": 0.02707396995362329,
      "grad_norm": 1.99338698387146,
      "learning_rate": 0.00018042959427207636,
      "loss": 2.1981,
      "step": 756
    },
    {
      "epoch": 0.027109782083191578,
      "grad_norm": 1.7673587799072266,
      "learning_rate": 0.00018066825775656326,
      "loss": 2.343,
      "step": 757
    },
    {
      "epoch": 0.02714559421275986,
      "grad_norm": 2.105565309524536,
      "learning_rate": 0.00018090692124105013,
      "loss": 2.0746,
      "step": 758
    },
    {
      "epoch": 0.027181406342328147,
      "grad_norm": 1.3338245153427124,
      "learning_rate": 0.000181145584725537,
      "loss": 1.9739,
      "step": 759
    },
    {
      "epoch": 0.027217218471896433,
      "grad_norm": 1.8781588077545166,
      "learning_rate": 0.00018138424821002386,
      "loss": 2.0575,
      "step": 760
    },
    {
      "epoch": 0.027253030601464716,
      "grad_norm": 1.1376395225524902,
      "learning_rate": 0.00018162291169451076,
      "loss": 1.8376,
      "step": 761
    },
    {
      "epoch": 0.027288842731033002,
      "grad_norm": 1.6982131004333496,
      "learning_rate": 0.00018186157517899762,
      "loss": 2.3264,
      "step": 762
    },
    {
      "epoch": 0.027324654860601285,
      "grad_norm": 1.391781210899353,
      "learning_rate": 0.0001821002386634845,
      "loss": 2.0107,
      "step": 763
    },
    {
      "epoch": 0.02736046699016957,
      "grad_norm": 1.4247759580612183,
      "learning_rate": 0.00018233890214797139,
      "loss": 1.8508,
      "step": 764
    },
    {
      "epoch": 0.027396279119737854,
      "grad_norm": 1.8408862352371216,
      "learning_rate": 0.00018257756563245825,
      "loss": 1.9833,
      "step": 765
    },
    {
      "epoch": 0.02743209124930614,
      "grad_norm": 2.116222381591797,
      "learning_rate": 0.00018281622911694512,
      "loss": 2.0208,
      "step": 766
    },
    {
      "epoch": 0.027467903378874423,
      "grad_norm": 1.4811813831329346,
      "learning_rate": 0.000183054892601432,
      "loss": 1.9656,
      "step": 767
    },
    {
      "epoch": 0.02750371550844271,
      "grad_norm": 1.1970711946487427,
      "learning_rate": 0.00018329355608591888,
      "loss": 2.0476,
      "step": 768
    },
    {
      "epoch": 0.027539527638010996,
      "grad_norm": 1.6708041429519653,
      "learning_rate": 0.00018353221957040575,
      "loss": 2.0781,
      "step": 769
    },
    {
      "epoch": 0.02757533976757928,
      "grad_norm": 1.3129223585128784,
      "learning_rate": 0.0001837708830548926,
      "loss": 1.7613,
      "step": 770
    },
    {
      "epoch": 0.027611151897147565,
      "grad_norm": 1.6428635120391846,
      "learning_rate": 0.00018400954653937946,
      "loss": 1.9753,
      "step": 771
    },
    {
      "epoch": 0.027646964026715848,
      "grad_norm": 1.3336197137832642,
      "learning_rate": 0.00018424821002386636,
      "loss": 1.8005,
      "step": 772
    },
    {
      "epoch": 0.027682776156284134,
      "grad_norm": 1.9145218133926392,
      "learning_rate": 0.00018448687350835322,
      "loss": 1.8981,
      "step": 773
    },
    {
      "epoch": 0.027718588285852417,
      "grad_norm": 1.456026315689087,
      "learning_rate": 0.0001847255369928401,
      "loss": 2.0075,
      "step": 774
    },
    {
      "epoch": 0.027754400415420703,
      "grad_norm": 1.5138424634933472,
      "learning_rate": 0.00018496420047732699,
      "loss": 2.2897,
      "step": 775
    },
    {
      "epoch": 0.027790212544988986,
      "grad_norm": 1.3635684251785278,
      "learning_rate": 0.00018520286396181385,
      "loss": 1.8127,
      "step": 776
    },
    {
      "epoch": 0.027826024674557272,
      "grad_norm": 2.9247641563415527,
      "learning_rate": 0.00018544152744630072,
      "loss": 2.0023,
      "step": 777
    },
    {
      "epoch": 0.02786183680412556,
      "grad_norm": 2.099508285522461,
      "learning_rate": 0.0001856801909307876,
      "loss": 1.8379,
      "step": 778
    },
    {
      "epoch": 0.02789764893369384,
      "grad_norm": 1.8196276426315308,
      "learning_rate": 0.00018591885441527448,
      "loss": 2.3346,
      "step": 779
    },
    {
      "epoch": 0.027933461063262128,
      "grad_norm": 1.5537874698638916,
      "learning_rate": 0.00018615751789976135,
      "loss": 2.0353,
      "step": 780
    },
    {
      "epoch": 0.02796927319283041,
      "grad_norm": 1.42340087890625,
      "learning_rate": 0.00018639618138424822,
      "loss": 1.8038,
      "step": 781
    },
    {
      "epoch": 0.028005085322398697,
      "grad_norm": 2.276153802871704,
      "learning_rate": 0.0001866348448687351,
      "loss": 1.9849,
      "step": 782
    },
    {
      "epoch": 0.02804089745196698,
      "grad_norm": 1.2486404180526733,
      "learning_rate": 0.00018687350835322198,
      "loss": 1.8228,
      "step": 783
    },
    {
      "epoch": 0.028076709581535266,
      "grad_norm": 2.1660590171813965,
      "learning_rate": 0.00018711217183770885,
      "loss": 1.8634,
      "step": 784
    },
    {
      "epoch": 0.028112521711103552,
      "grad_norm": 1.5528640747070312,
      "learning_rate": 0.00018735083532219572,
      "loss": 2.2286,
      "step": 785
    },
    {
      "epoch": 0.028148333840671835,
      "grad_norm": 1.3248339891433716,
      "learning_rate": 0.00018758949880668259,
      "loss": 2.0061,
      "step": 786
    },
    {
      "epoch": 0.02818414597024012,
      "grad_norm": 1.7929283380508423,
      "learning_rate": 0.00018782816229116945,
      "loss": 2.3811,
      "step": 787
    },
    {
      "epoch": 0.028219958099808404,
      "grad_norm": 1.729906678199768,
      "learning_rate": 0.00018806682577565632,
      "loss": 1.8049,
      "step": 788
    },
    {
      "epoch": 0.02825577022937669,
      "grad_norm": 1.7999067306518555,
      "learning_rate": 0.0001883054892601432,
      "loss": 1.9822,
      "step": 789
    },
    {
      "epoch": 0.028291582358944974,
      "grad_norm": 1.5044533014297485,
      "learning_rate": 0.00018854415274463008,
      "loss": 1.8759,
      "step": 790
    },
    {
      "epoch": 0.02832739448851326,
      "grad_norm": 1.5658330917358398,
      "learning_rate": 0.00018878281622911695,
      "loss": 2.1563,
      "step": 791
    },
    {
      "epoch": 0.028363206618081543,
      "grad_norm": 1.8020349740982056,
      "learning_rate": 0.00018902147971360382,
      "loss": 2.1909,
      "step": 792
    },
    {
      "epoch": 0.02839901874764983,
      "grad_norm": 1.3612864017486572,
      "learning_rate": 0.00018926014319809071,
      "loss": 1.7796,
      "step": 793
    },
    {
      "epoch": 0.028434830877218115,
      "grad_norm": 1.2705191373825073,
      "learning_rate": 0.00018949880668257758,
      "loss": 2.071,
      "step": 794
    },
    {
      "epoch": 0.028470643006786398,
      "grad_norm": 1.3243223428726196,
      "learning_rate": 0.00018973747016706445,
      "loss": 2.1743,
      "step": 795
    },
    {
      "epoch": 0.028506455136354684,
      "grad_norm": 1.805530309677124,
      "learning_rate": 0.00018997613365155132,
      "loss": 1.9458,
      "step": 796
    },
    {
      "epoch": 0.028542267265922967,
      "grad_norm": 1.6725308895111084,
      "learning_rate": 0.0001902147971360382,
      "loss": 1.9733,
      "step": 797
    },
    {
      "epoch": 0.028578079395491254,
      "grad_norm": 2.9171323776245117,
      "learning_rate": 0.00019045346062052508,
      "loss": 1.8562,
      "step": 798
    },
    {
      "epoch": 0.028613891525059536,
      "grad_norm": 1.2938868999481201,
      "learning_rate": 0.00019069212410501195,
      "loss": 2.2139,
      "step": 799
    },
    {
      "epoch": 0.028649703654627823,
      "grad_norm": 1.9000681638717651,
      "learning_rate": 0.00019093078758949882,
      "loss": 1.9803,
      "step": 800
    },
    {
      "epoch": 0.028685515784196106,
      "grad_norm": 1.895714282989502,
      "learning_rate": 0.00019116945107398568,
      "loss": 1.8723,
      "step": 801
    },
    {
      "epoch": 0.028721327913764392,
      "grad_norm": 1.1832196712493896,
      "learning_rate": 0.00019140811455847255,
      "loss": 1.7096,
      "step": 802
    },
    {
      "epoch": 0.028757140043332678,
      "grad_norm": 1.382149577140808,
      "learning_rate": 0.00019164677804295942,
      "loss": 1.7541,
      "step": 803
    },
    {
      "epoch": 0.02879295217290096,
      "grad_norm": 1.5034754276275635,
      "learning_rate": 0.00019188544152744631,
      "loss": 2.3756,
      "step": 804
    },
    {
      "epoch": 0.028828764302469247,
      "grad_norm": 1.5388857126235962,
      "learning_rate": 0.00019212410501193318,
      "loss": 1.9153,
      "step": 805
    },
    {
      "epoch": 0.02886457643203753,
      "grad_norm": 2.118048667907715,
      "learning_rate": 0.00019236276849642005,
      "loss": 2.2704,
      "step": 806
    },
    {
      "epoch": 0.028900388561605816,
      "grad_norm": 1.6065908670425415,
      "learning_rate": 0.00019260143198090692,
      "loss": 2.0055,
      "step": 807
    },
    {
      "epoch": 0.0289362006911741,
      "grad_norm": 2.0480384826660156,
      "learning_rate": 0.0001928400954653938,
      "loss": 2.033,
      "step": 808
    },
    {
      "epoch": 0.028972012820742386,
      "grad_norm": 2.066574811935425,
      "learning_rate": 0.00019307875894988068,
      "loss": 2.2037,
      "step": 809
    },
    {
      "epoch": 0.029007824950310672,
      "grad_norm": 1.3903077840805054,
      "learning_rate": 0.00019331742243436755,
      "loss": 1.7733,
      "step": 810
    },
    {
      "epoch": 0.029043637079878955,
      "grad_norm": 2.8067617416381836,
      "learning_rate": 0.00019355608591885444,
      "loss": 1.7627,
      "step": 811
    },
    {
      "epoch": 0.02907944920944724,
      "grad_norm": 1.9488036632537842,
      "learning_rate": 0.0001937947494033413,
      "loss": 1.9536,
      "step": 812
    },
    {
      "epoch": 0.029115261339015524,
      "grad_norm": 1.4101473093032837,
      "learning_rate": 0.00019403341288782818,
      "loss": 2.1673,
      "step": 813
    },
    {
      "epoch": 0.02915107346858381,
      "grad_norm": 1.2836867570877075,
      "learning_rate": 0.00019427207637231505,
      "loss": 1.7867,
      "step": 814
    },
    {
      "epoch": 0.029186885598152093,
      "grad_norm": 1.4783426523208618,
      "learning_rate": 0.00019451073985680191,
      "loss": 2.1039,
      "step": 815
    },
    {
      "epoch": 0.02922269772772038,
      "grad_norm": 1.3614662885665894,
      "learning_rate": 0.00019474940334128878,
      "loss": 2.0084,
      "step": 816
    },
    {
      "epoch": 0.029258509857288662,
      "grad_norm": 1.2852891683578491,
      "learning_rate": 0.00019498806682577565,
      "loss": 1.8364,
      "step": 817
    },
    {
      "epoch": 0.02929432198685695,
      "grad_norm": 1.6663340330123901,
      "learning_rate": 0.00019522673031026252,
      "loss": 2.1946,
      "step": 818
    },
    {
      "epoch": 0.029330134116425235,
      "grad_norm": 2.088148832321167,
      "learning_rate": 0.0001954653937947494,
      "loss": 2.1451,
      "step": 819
    },
    {
      "epoch": 0.029365946245993518,
      "grad_norm": 1.2781169414520264,
      "learning_rate": 0.00019570405727923628,
      "loss": 1.8716,
      "step": 820
    },
    {
      "epoch": 0.029401758375561804,
      "grad_norm": 1.7055004835128784,
      "learning_rate": 0.00019594272076372315,
      "loss": 2.1419,
      "step": 821
    },
    {
      "epoch": 0.029437570505130087,
      "grad_norm": 2.0636518001556396,
      "learning_rate": 0.00019618138424821004,
      "loss": 2.2445,
      "step": 822
    },
    {
      "epoch": 0.029473382634698373,
      "grad_norm": 1.557982325553894,
      "learning_rate": 0.0001964200477326969,
      "loss": 1.8256,
      "step": 823
    },
    {
      "epoch": 0.029509194764266656,
      "grad_norm": 1.3213343620300293,
      "learning_rate": 0.00019665871121718378,
      "loss": 1.6728,
      "step": 824
    },
    {
      "epoch": 0.029545006893834942,
      "grad_norm": 2.1603245735168457,
      "learning_rate": 0.00019689737470167065,
      "loss": 2.0918,
      "step": 825
    },
    {
      "epoch": 0.029580819023403225,
      "grad_norm": 1.5868972539901733,
      "learning_rate": 0.00019713603818615754,
      "loss": 2.0132,
      "step": 826
    },
    {
      "epoch": 0.02961663115297151,
      "grad_norm": 2.6218371391296387,
      "learning_rate": 0.0001973747016706444,
      "loss": 1.97,
      "step": 827
    },
    {
      "epoch": 0.029652443282539798,
      "grad_norm": 1.8237935304641724,
      "learning_rate": 0.00019761336515513128,
      "loss": 1.8931,
      "step": 828
    },
    {
      "epoch": 0.02968825541210808,
      "grad_norm": 1.9808688163757324,
      "learning_rate": 0.00019785202863961817,
      "loss": 1.9166,
      "step": 829
    },
    {
      "epoch": 0.029724067541676367,
      "grad_norm": 1.4820518493652344,
      "learning_rate": 0.000198090692124105,
      "loss": 1.9317,
      "step": 830
    },
    {
      "epoch": 0.02975987967124465,
      "grad_norm": 2.2204062938690186,
      "learning_rate": 0.00019832935560859188,
      "loss": 2.0906,
      "step": 831
    },
    {
      "epoch": 0.029795691800812936,
      "grad_norm": 1.4470603466033936,
      "learning_rate": 0.00019856801909307875,
      "loss": 1.9674,
      "step": 832
    },
    {
      "epoch": 0.02983150393038122,
      "grad_norm": 1.444513201713562,
      "learning_rate": 0.00019880668257756564,
      "loss": 1.9251,
      "step": 833
    },
    {
      "epoch": 0.029867316059949505,
      "grad_norm": 1.6284515857696533,
      "learning_rate": 0.0001990453460620525,
      "loss": 1.8075,
      "step": 834
    },
    {
      "epoch": 0.02990312818951779,
      "grad_norm": 1.885413646697998,
      "learning_rate": 0.00019928400954653938,
      "loss": 2.0856,
      "step": 835
    },
    {
      "epoch": 0.029938940319086074,
      "grad_norm": 1.7370575666427612,
      "learning_rate": 0.00019952267303102625,
      "loss": 1.7621,
      "step": 836
    },
    {
      "epoch": 0.02997475244865436,
      "grad_norm": 2.0905561447143555,
      "learning_rate": 0.00019976133651551314,
      "loss": 2.5441,
      "step": 837
    },
    {
      "epoch": 0.030010564578222643,
      "grad_norm": 1.5622786283493042,
      "learning_rate": 0.0002,
      "loss": 1.9866,
      "step": 838
    },
    {
      "epoch": 0.03004637670779093,
      "grad_norm": 1.1882922649383545,
      "learning_rate": 0.0001999999993273145,
      "loss": 1.807,
      "step": 839
    },
    {
      "epoch": 0.030082188837359213,
      "grad_norm": 1.3145676851272583,
      "learning_rate": 0.000199999997309258,
      "loss": 1.9652,
      "step": 840
    },
    {
      "epoch": 0.0301180009669275,
      "grad_norm": 1.4804152250289917,
      "learning_rate": 0.00019999999394583053,
      "loss": 2.1154,
      "step": 841
    },
    {
      "epoch": 0.03015381309649578,
      "grad_norm": 1.6515103578567505,
      "learning_rate": 0.00019999998923703213,
      "loss": 1.7606,
      "step": 842
    },
    {
      "epoch": 0.030189625226064068,
      "grad_norm": 2.046954393386841,
      "learning_rate": 0.00019999998318286286,
      "loss": 1.9643,
      "step": 843
    },
    {
      "epoch": 0.030225437355632354,
      "grad_norm": 2.260840654373169,
      "learning_rate": 0.0001999999757833228,
      "loss": 2.0192,
      "step": 844
    },
    {
      "epoch": 0.030261249485200637,
      "grad_norm": 2.859248161315918,
      "learning_rate": 0.00019999996703841207,
      "loss": 1.9989,
      "step": 845
    },
    {
      "epoch": 0.030297061614768923,
      "grad_norm": 1.7310001850128174,
      "learning_rate": 0.00019999995694813073,
      "loss": 2.0006,
      "step": 846
    },
    {
      "epoch": 0.030332873744337206,
      "grad_norm": 1.423134207725525,
      "learning_rate": 0.00019999994551247901,
      "loss": 1.7585,
      "step": 847
    },
    {
      "epoch": 0.030368685873905493,
      "grad_norm": 1.5323225259780884,
      "learning_rate": 0.000199999932731457,
      "loss": 2.1375,
      "step": 848
    },
    {
      "epoch": 0.030404498003473775,
      "grad_norm": 1.53788423538208,
      "learning_rate": 0.00019999991860506492,
      "loss": 1.9595,
      "step": 849
    },
    {
      "epoch": 0.03044031013304206,
      "grad_norm": 1.2339884042739868,
      "learning_rate": 0.00019999990313330286,
      "loss": 2.0193,
      "step": 850
    },
    {
      "epoch": 0.030476122262610345,
      "grad_norm": 1.9879727363586426,
      "learning_rate": 0.00019999988631617114,
      "loss": 1.5824,
      "step": 851
    },
    {
      "epoch": 0.03051193439217863,
      "grad_norm": 1.245160460472107,
      "learning_rate": 0.00019999986815366993,
      "loss": 1.7495,
      "step": 852
    },
    {
      "epoch": 0.030547746521746917,
      "grad_norm": 1.278892159461975,
      "learning_rate": 0.0001999998486457995,
      "loss": 2.207,
      "step": 853
    },
    {
      "epoch": 0.0305835586513152,
      "grad_norm": 1.3570001125335693,
      "learning_rate": 0.00019999982779256005,
      "loss": 1.7336,
      "step": 854
    },
    {
      "epoch": 0.030619370780883486,
      "grad_norm": 1.5163108110427856,
      "learning_rate": 0.00019999980559395195,
      "loss": 2.1173,
      "step": 855
    },
    {
      "epoch": 0.03065518291045177,
      "grad_norm": 1.1349409818649292,
      "learning_rate": 0.00019999978204997545,
      "loss": 1.8362,
      "step": 856
    },
    {
      "epoch": 0.030690995040020055,
      "grad_norm": 1.9292572736740112,
      "learning_rate": 0.00019999975716063087,
      "loss": 2.0006,
      "step": 857
    },
    {
      "epoch": 0.030726807169588338,
      "grad_norm": 1.8030412197113037,
      "learning_rate": 0.0001999997309259185,
      "loss": 2.0975,
      "step": 858
    },
    {
      "epoch": 0.030762619299156625,
      "grad_norm": 1.3017148971557617,
      "learning_rate": 0.0001999997033458388,
      "loss": 1.9676,
      "step": 859
    },
    {
      "epoch": 0.03079843142872491,
      "grad_norm": 1.4081376791000366,
      "learning_rate": 0.00019999967442039206,
      "loss": 1.985,
      "step": 860
    },
    {
      "epoch": 0.030834243558293194,
      "grad_norm": 1.465991735458374,
      "learning_rate": 0.0001999996441495787,
      "loss": 1.7478,
      "step": 861
    },
    {
      "epoch": 0.03087005568786148,
      "grad_norm": 1.6048805713653564,
      "learning_rate": 0.0001999996125333991,
      "loss": 2.0733,
      "step": 862
    },
    {
      "epoch": 0.030905867817429763,
      "grad_norm": 1.3127105236053467,
      "learning_rate": 0.00019999957957185375,
      "loss": 1.9589,
      "step": 863
    },
    {
      "epoch": 0.03094167994699805,
      "grad_norm": 1.815430760383606,
      "learning_rate": 0.000199999545264943,
      "loss": 2.246,
      "step": 864
    },
    {
      "epoch": 0.030977492076566332,
      "grad_norm": 1.334842324256897,
      "learning_rate": 0.00019999950961266738,
      "loss": 1.8453,
      "step": 865
    },
    {
      "epoch": 0.03101330420613462,
      "grad_norm": 1.6176645755767822,
      "learning_rate": 0.00019999947261502735,
      "loss": 2.0127,
      "step": 866
    },
    {
      "epoch": 0.0310491163357029,
      "grad_norm": 1.444711446762085,
      "learning_rate": 0.0001999994342720234,
      "loss": 1.7625,
      "step": 867
    },
    {
      "epoch": 0.031084928465271187,
      "grad_norm": 1.4336127042770386,
      "learning_rate": 0.00019999939458365605,
      "loss": 1.799,
      "step": 868
    },
    {
      "epoch": 0.031120740594839474,
      "grad_norm": 1.5195722579956055,
      "learning_rate": 0.00019999935354992582,
      "loss": 1.9003,
      "step": 869
    },
    {
      "epoch": 0.031156552724407757,
      "grad_norm": 2.2031455039978027,
      "learning_rate": 0.0001999993111708333,
      "loss": 2.1488,
      "step": 870
    },
    {
      "epoch": 0.031192364853976043,
      "grad_norm": 1.6597728729248047,
      "learning_rate": 0.00019999926744637903,
      "loss": 1.9993,
      "step": 871
    },
    {
      "epoch": 0.031228176983544326,
      "grad_norm": 1.707897424697876,
      "learning_rate": 0.0001999992223765636,
      "loss": 2.0831,
      "step": 872
    },
    {
      "epoch": 0.03126398911311261,
      "grad_norm": 1.9093265533447266,
      "learning_rate": 0.00019999917596138765,
      "loss": 2.2365,
      "step": 873
    },
    {
      "epoch": 0.0312998012426809,
      "grad_norm": 2.5550174713134766,
      "learning_rate": 0.00019999912820085176,
      "loss": 2.1307,
      "step": 874
    },
    {
      "epoch": 0.03133561337224918,
      "grad_norm": 1.5057379007339478,
      "learning_rate": 0.0001999990790949566,
      "loss": 2.0101,
      "step": 875
    },
    {
      "epoch": 0.031371425501817464,
      "grad_norm": 1.5937858819961548,
      "learning_rate": 0.0001999990286437028,
      "loss": 2.0092,
      "step": 876
    },
    {
      "epoch": 0.031407237631385754,
      "grad_norm": 1.1650123596191406,
      "learning_rate": 0.00019999897684709104,
      "loss": 1.8871,
      "step": 877
    },
    {
      "epoch": 0.03144304976095404,
      "grad_norm": 2.418663501739502,
      "learning_rate": 0.00019999892370512208,
      "loss": 2.106,
      "step": 878
    },
    {
      "epoch": 0.03147886189052232,
      "grad_norm": 1.407043218612671,
      "learning_rate": 0.00019999886921779657,
      "loss": 1.9746,
      "step": 879
    },
    {
      "epoch": 0.0315146740200906,
      "grad_norm": 1.4505114555358887,
      "learning_rate": 0.00019999881338511526,
      "loss": 1.9175,
      "step": 880
    },
    {
      "epoch": 0.03155048614965889,
      "grad_norm": 1.4700239896774292,
      "learning_rate": 0.0001999987562070789,
      "loss": 1.9464,
      "step": 881
    },
    {
      "epoch": 0.031586298279227175,
      "grad_norm": 1.1745959520339966,
      "learning_rate": 0.00019999869768368828,
      "loss": 1.7727,
      "step": 882
    },
    {
      "epoch": 0.03162211040879546,
      "grad_norm": 1.303771734237671,
      "learning_rate": 0.0001999986378149442,
      "loss": 2.056,
      "step": 883
    },
    {
      "epoch": 0.03165792253836374,
      "grad_norm": 1.8219273090362549,
      "learning_rate": 0.00019999857660084737,
      "loss": 1.768,
      "step": 884
    },
    {
      "epoch": 0.03169373466793203,
      "grad_norm": 1.5007424354553223,
      "learning_rate": 0.00019999851404139873,
      "loss": 1.9576,
      "step": 885
    },
    {
      "epoch": 0.03172954679750031,
      "grad_norm": 1.8747928142547607,
      "learning_rate": 0.00019999845013659906,
      "loss": 1.8605,
      "step": 886
    },
    {
      "epoch": 0.031765358927068596,
      "grad_norm": 1.2296186685562134,
      "learning_rate": 0.00019999838488644924,
      "loss": 2.0188,
      "step": 887
    },
    {
      "epoch": 0.031801171056636886,
      "grad_norm": 1.2861896753311157,
      "learning_rate": 0.00019999831829095013,
      "loss": 2.0889,
      "step": 888
    },
    {
      "epoch": 0.03183698318620517,
      "grad_norm": 1.5891250371932983,
      "learning_rate": 0.00019999825035010263,
      "loss": 1.9346,
      "step": 889
    },
    {
      "epoch": 0.03187279531577345,
      "grad_norm": 1.4180065393447876,
      "learning_rate": 0.00019999818106390766,
      "loss": 1.7997,
      "step": 890
    },
    {
      "epoch": 0.031908607445341734,
      "grad_norm": 1.7038114070892334,
      "learning_rate": 0.0001999981104323662,
      "loss": 1.7128,
      "step": 891
    },
    {
      "epoch": 0.031944419574910024,
      "grad_norm": 1.518619179725647,
      "learning_rate": 0.00019999803845547907,
      "loss": 1.802,
      "step": 892
    },
    {
      "epoch": 0.03198023170447831,
      "grad_norm": 1.3319896459579468,
      "learning_rate": 0.00019999796513324735,
      "loss": 1.8977,
      "step": 893
    },
    {
      "epoch": 0.03201604383404659,
      "grad_norm": 1.7978893518447876,
      "learning_rate": 0.00019999789046567203,
      "loss": 1.8772,
      "step": 894
    },
    {
      "epoch": 0.03205185596361488,
      "grad_norm": 2.5152993202209473,
      "learning_rate": 0.00019999781445275406,
      "loss": 1.9364,
      "step": 895
    },
    {
      "epoch": 0.03208766809318316,
      "grad_norm": 1.4267518520355225,
      "learning_rate": 0.0001999977370944945,
      "loss": 2.3957,
      "step": 896
    },
    {
      "epoch": 0.032123480222751445,
      "grad_norm": 1.907285451889038,
      "learning_rate": 0.00019999765839089434,
      "loss": 1.6652,
      "step": 897
    },
    {
      "epoch": 0.03215929235231973,
      "grad_norm": 1.9558316469192505,
      "learning_rate": 0.00019999757834195472,
      "loss": 1.9541,
      "step": 898
    },
    {
      "epoch": 0.03219510448188802,
      "grad_norm": 1.6896051168441772,
      "learning_rate": 0.00019999749694767666,
      "loss": 1.722,
      "step": 899
    },
    {
      "epoch": 0.0322309166114563,
      "grad_norm": 2.4404208660125732,
      "learning_rate": 0.0001999974142080612,
      "loss": 2.4644,
      "step": 900
    },
    {
      "epoch": 0.03226672874102458,
      "grad_norm": 1.4918023347854614,
      "learning_rate": 0.00019999733012310958,
      "loss": 1.9803,
      "step": 901
    },
    {
      "epoch": 0.03230254087059287,
      "grad_norm": 1.9694902896881104,
      "learning_rate": 0.00019999724469282288,
      "loss": 1.9535,
      "step": 902
    },
    {
      "epoch": 0.032338353000161156,
      "grad_norm": 1.778144359588623,
      "learning_rate": 0.00019999715791720223,
      "loss": 1.8126,
      "step": 903
    },
    {
      "epoch": 0.03237416512972944,
      "grad_norm": 1.5544754266738892,
      "learning_rate": 0.00019999706979624877,
      "loss": 2.0249,
      "step": 904
    },
    {
      "epoch": 0.03240997725929772,
      "grad_norm": 2.4764208793640137,
      "learning_rate": 0.00019999698032996377,
      "loss": 2.1015,
      "step": 905
    },
    {
      "epoch": 0.03244578938886601,
      "grad_norm": 1.926424503326416,
      "learning_rate": 0.00019999688951834836,
      "loss": 1.8125,
      "step": 906
    },
    {
      "epoch": 0.032481601518434294,
      "grad_norm": 2.0330772399902344,
      "learning_rate": 0.0001999967973614038,
      "loss": 2.0531,
      "step": 907
    },
    {
      "epoch": 0.03251741364800258,
      "grad_norm": 2.0009052753448486,
      "learning_rate": 0.00019999670385913133,
      "loss": 2.2191,
      "step": 908
    },
    {
      "epoch": 0.03255322577757086,
      "grad_norm": 1.3479782342910767,
      "learning_rate": 0.00019999660901153218,
      "loss": 1.7542,
      "step": 909
    },
    {
      "epoch": 0.03258903790713915,
      "grad_norm": 1.28345787525177,
      "learning_rate": 0.00019999651281860762,
      "loss": 1.9944,
      "step": 910
    },
    {
      "epoch": 0.03262485003670743,
      "grad_norm": 1.4357455968856812,
      "learning_rate": 0.00019999641528035898,
      "loss": 2.0591,
      "step": 911
    },
    {
      "epoch": 0.032660662166275715,
      "grad_norm": 1.6938787698745728,
      "learning_rate": 0.0001999963163967876,
      "loss": 2.2751,
      "step": 912
    },
    {
      "epoch": 0.032696474295844005,
      "grad_norm": 1.267437219619751,
      "learning_rate": 0.00019999621616789473,
      "loss": 1.8928,
      "step": 913
    },
    {
      "epoch": 0.03273228642541229,
      "grad_norm": 1.2574923038482666,
      "learning_rate": 0.00019999611459368174,
      "loss": 1.793,
      "step": 914
    },
    {
      "epoch": 0.03276809855498057,
      "grad_norm": 1.0232667922973633,
      "learning_rate": 0.00019999601167415006,
      "loss": 1.8391,
      "step": 915
    },
    {
      "epoch": 0.032803910684548854,
      "grad_norm": 2.3581461906433105,
      "learning_rate": 0.000199995907409301,
      "loss": 2.4764,
      "step": 916
    },
    {
      "epoch": 0.032839722814117144,
      "grad_norm": 1.5463016033172607,
      "learning_rate": 0.000199995801799136,
      "loss": 1.9629,
      "step": 917
    },
    {
      "epoch": 0.032875534943685426,
      "grad_norm": 2.3027689456939697,
      "learning_rate": 0.00019999569484365645,
      "loss": 1.9298,
      "step": 918
    },
    {
      "epoch": 0.03291134707325371,
      "grad_norm": 1.575325846672058,
      "learning_rate": 0.00019999558654286385,
      "loss": 1.8611,
      "step": 919
    },
    {
      "epoch": 0.032947159202822,
      "grad_norm": 2.7732012271881104,
      "learning_rate": 0.0001999954768967596,
      "loss": 2.2494,
      "step": 920
    },
    {
      "epoch": 0.03298297133239028,
      "grad_norm": 2.0338802337646484,
      "learning_rate": 0.0001999953659053452,
      "loss": 1.8992,
      "step": 921
    },
    {
      "epoch": 0.033018783461958565,
      "grad_norm": 1.5755022764205933,
      "learning_rate": 0.0001999952535686221,
      "loss": 2.1845,
      "step": 922
    },
    {
      "epoch": 0.03305459559152685,
      "grad_norm": 1.2108244895935059,
      "learning_rate": 0.00019999513988659188,
      "loss": 1.6432,
      "step": 923
    },
    {
      "epoch": 0.03309040772109514,
      "grad_norm": 2.210641860961914,
      "learning_rate": 0.00019999502485925605,
      "loss": 2.4049,
      "step": 924
    },
    {
      "epoch": 0.03312621985066342,
      "grad_norm": 1.6159656047821045,
      "learning_rate": 0.00019999490848661612,
      "loss": 1.6542,
      "step": 925
    },
    {
      "epoch": 0.0331620319802317,
      "grad_norm": 1.2654778957366943,
      "learning_rate": 0.00019999479076867368,
      "loss": 1.991,
      "step": 926
    },
    {
      "epoch": 0.03319784410979999,
      "grad_norm": 1.8192824125289917,
      "learning_rate": 0.00019999467170543031,
      "loss": 1.9816,
      "step": 927
    },
    {
      "epoch": 0.033233656239368276,
      "grad_norm": 1.5657546520233154,
      "learning_rate": 0.00019999455129688764,
      "loss": 1.8487,
      "step": 928
    },
    {
      "epoch": 0.03326946836893656,
      "grad_norm": 1.4876744747161865,
      "learning_rate": 0.00019999442954304729,
      "loss": 1.7911,
      "step": 929
    },
    {
      "epoch": 0.03330528049850484,
      "grad_norm": 1.8739250898361206,
      "learning_rate": 0.00019999430644391082,
      "loss": 1.9183,
      "step": 930
    },
    {
      "epoch": 0.03334109262807313,
      "grad_norm": 1.3608872890472412,
      "learning_rate": 0.00019999418199947994,
      "loss": 1.8478,
      "step": 931
    },
    {
      "epoch": 0.033376904757641414,
      "grad_norm": 1.5598634481430054,
      "learning_rate": 0.00019999405620975636,
      "loss": 2.0591,
      "step": 932
    },
    {
      "epoch": 0.0334127168872097,
      "grad_norm": 2.9095773696899414,
      "learning_rate": 0.00019999392907474174,
      "loss": 2.0628,
      "step": 933
    },
    {
      "epoch": 0.03344852901677798,
      "grad_norm": 1.6652559041976929,
      "learning_rate": 0.00019999380059443773,
      "loss": 1.9522,
      "step": 934
    },
    {
      "epoch": 0.03348434114634627,
      "grad_norm": 1.8389030694961548,
      "learning_rate": 0.00019999367076884616,
      "loss": 1.8165,
      "step": 935
    },
    {
      "epoch": 0.03352015327591455,
      "grad_norm": 1.5243996381759644,
      "learning_rate": 0.00019999353959796872,
      "loss": 1.9362,
      "step": 936
    },
    {
      "epoch": 0.033555965405482835,
      "grad_norm": 1.587996006011963,
      "learning_rate": 0.0001999934070818072,
      "loss": 1.9498,
      "step": 937
    },
    {
      "epoch": 0.033591777535051125,
      "grad_norm": 1.3211830854415894,
      "learning_rate": 0.00019999327322036336,
      "loss": 1.8008,
      "step": 938
    },
    {
      "epoch": 0.03362758966461941,
      "grad_norm": 1.775557518005371,
      "learning_rate": 0.00019999313801363902,
      "loss": 2.0888,
      "step": 939
    },
    {
      "epoch": 0.03366340179418769,
      "grad_norm": 1.5863924026489258,
      "learning_rate": 0.00019999300146163597,
      "loss": 1.7688,
      "step": 940
    },
    {
      "epoch": 0.03369921392375597,
      "grad_norm": 2.072439193725586,
      "learning_rate": 0.00019999286356435608,
      "loss": 1.9447,
      "step": 941
    },
    {
      "epoch": 0.03373502605332426,
      "grad_norm": 1.8205657005310059,
      "learning_rate": 0.0001999927243218012,
      "loss": 1.852,
      "step": 942
    },
    {
      "epoch": 0.033770838182892546,
      "grad_norm": 2.22773814201355,
      "learning_rate": 0.0001999925837339732,
      "loss": 1.898,
      "step": 943
    },
    {
      "epoch": 0.03380665031246083,
      "grad_norm": 1.3842288255691528,
      "learning_rate": 0.00019999244180087395,
      "loss": 2.0167,
      "step": 944
    },
    {
      "epoch": 0.03384246244202912,
      "grad_norm": 1.9286949634552002,
      "learning_rate": 0.00019999229852250537,
      "loss": 1.9268,
      "step": 945
    },
    {
      "epoch": 0.0338782745715974,
      "grad_norm": 2.1182754039764404,
      "learning_rate": 0.00019999215389886942,
      "loss": 2.0501,
      "step": 946
    },
    {
      "epoch": 0.033914086701165684,
      "grad_norm": 1.6029773950576782,
      "learning_rate": 0.000199992007929968,
      "loss": 1.9263,
      "step": 947
    },
    {
      "epoch": 0.03394989883073397,
      "grad_norm": 1.2110930681228638,
      "learning_rate": 0.0001999918606158031,
      "loss": 1.9129,
      "step": 948
    },
    {
      "epoch": 0.03398571096030226,
      "grad_norm": 1.507839322090149,
      "learning_rate": 0.0001999917119563767,
      "loss": 1.9622,
      "step": 949
    },
    {
      "epoch": 0.03402152308987054,
      "grad_norm": 1.6795588731765747,
      "learning_rate": 0.00019999156195169078,
      "loss": 1.864,
      "step": 950
    },
    {
      "epoch": 0.03405733521943882,
      "grad_norm": 1.9231281280517578,
      "learning_rate": 0.0001999914106017474,
      "loss": 2.1288,
      "step": 951
    },
    {
      "epoch": 0.03409314734900711,
      "grad_norm": 1.7240833044052124,
      "learning_rate": 0.00019999125790654855,
      "loss": 1.7942,
      "step": 952
    },
    {
      "epoch": 0.034128959478575395,
      "grad_norm": 1.4035639762878418,
      "learning_rate": 0.0001999911038660963,
      "loss": 1.9428,
      "step": 953
    },
    {
      "epoch": 0.03416477160814368,
      "grad_norm": 1.2519588470458984,
      "learning_rate": 0.00019999094848039274,
      "loss": 1.7459,
      "step": 954
    },
    {
      "epoch": 0.03420058373771196,
      "grad_norm": 1.201668620109558,
      "learning_rate": 0.00019999079174943995,
      "loss": 1.6607,
      "step": 955
    },
    {
      "epoch": 0.03423639586728025,
      "grad_norm": 1.6537754535675049,
      "learning_rate": 0.00019999063367324003,
      "loss": 2.0089,
      "step": 956
    },
    {
      "epoch": 0.03427220799684853,
      "grad_norm": 2.3439764976501465,
      "learning_rate": 0.0001999904742517951,
      "loss": 2.1381,
      "step": 957
    },
    {
      "epoch": 0.034308020126416816,
      "grad_norm": 1.219859004020691,
      "learning_rate": 0.00019999031348510733,
      "loss": 2.0814,
      "step": 958
    },
    {
      "epoch": 0.0343438322559851,
      "grad_norm": 1.969250202178955,
      "learning_rate": 0.00019999015137317887,
      "loss": 2.0864,
      "step": 959
    },
    {
      "epoch": 0.03437964438555339,
      "grad_norm": 2.401399850845337,
      "learning_rate": 0.0001999899879160119,
      "loss": 1.9438,
      "step": 960
    },
    {
      "epoch": 0.03441545651512167,
      "grad_norm": 1.4327278137207031,
      "learning_rate": 0.00019998982311360863,
      "loss": 1.9214,
      "step": 961
    },
    {
      "epoch": 0.034451268644689954,
      "grad_norm": 2.319035768508911,
      "learning_rate": 0.00019998965696597126,
      "loss": 2.3354,
      "step": 962
    },
    {
      "epoch": 0.034487080774258244,
      "grad_norm": 1.1964077949523926,
      "learning_rate": 0.00019998948947310202,
      "loss": 1.9042,
      "step": 963
    },
    {
      "epoch": 0.03452289290382653,
      "grad_norm": 1.8281431198120117,
      "learning_rate": 0.0001999893206350032,
      "loss": 1.9747,
      "step": 964
    },
    {
      "epoch": 0.03455870503339481,
      "grad_norm": 1.1979892253875732,
      "learning_rate": 0.00019998915045167702,
      "loss": 1.6347,
      "step": 965
    },
    {
      "epoch": 0.03459451716296309,
      "grad_norm": 1.382490873336792,
      "learning_rate": 0.0001999889789231258,
      "loss": 2.0574,
      "step": 966
    },
    {
      "epoch": 0.03463032929253138,
      "grad_norm": 1.8725450038909912,
      "learning_rate": 0.00019998880604935187,
      "loss": 1.9832,
      "step": 967
    },
    {
      "epoch": 0.034666141422099665,
      "grad_norm": 1.4053820371627808,
      "learning_rate": 0.00019998863183035752,
      "loss": 1.5536,
      "step": 968
    },
    {
      "epoch": 0.03470195355166795,
      "grad_norm": 2.361921787261963,
      "learning_rate": 0.0001999884562661451,
      "loss": 2.347,
      "step": 969
    },
    {
      "epoch": 0.03473776568123624,
      "grad_norm": 1.5455695390701294,
      "learning_rate": 0.00019998827935671697,
      "loss": 1.5136,
      "step": 970
    },
    {
      "epoch": 0.03477357781080452,
      "grad_norm": 1.388515830039978,
      "learning_rate": 0.00019998810110207553,
      "loss": 1.7991,
      "step": 971
    },
    {
      "epoch": 0.034809389940372804,
      "grad_norm": 1.5065860748291016,
      "learning_rate": 0.00019998792150222316,
      "loss": 2.0703,
      "step": 972
    },
    {
      "epoch": 0.034845202069941086,
      "grad_norm": 2.07222580909729,
      "learning_rate": 0.0001999877405571623,
      "loss": 1.9729,
      "step": 973
    },
    {
      "epoch": 0.034881014199509376,
      "grad_norm": 1.977397084236145,
      "learning_rate": 0.00019998755826689535,
      "loss": 2.313,
      "step": 974
    },
    {
      "epoch": 0.03491682632907766,
      "grad_norm": 1.2814414501190186,
      "learning_rate": 0.00019998737463142478,
      "loss": 1.7694,
      "step": 975
    },
    {
      "epoch": 0.03495263845864594,
      "grad_norm": 2.0901002883911133,
      "learning_rate": 0.00019998718965075305,
      "loss": 2.186,
      "step": 976
    },
    {
      "epoch": 0.03498845058821423,
      "grad_norm": 1.2248455286026,
      "learning_rate": 0.00019998700332488265,
      "loss": 1.7233,
      "step": 977
    },
    {
      "epoch": 0.035024262717782514,
      "grad_norm": 1.0053720474243164,
      "learning_rate": 0.00019998681565381611,
      "loss": 1.5713,
      "step": 978
    },
    {
      "epoch": 0.0350600748473508,
      "grad_norm": 1.413627028465271,
      "learning_rate": 0.00019998662663755595,
      "loss": 1.908,
      "step": 979
    },
    {
      "epoch": 0.03509588697691908,
      "grad_norm": 1.2800226211547852,
      "learning_rate": 0.00019998643627610466,
      "loss": 1.83,
      "step": 980
    },
    {
      "epoch": 0.03513169910648737,
      "grad_norm": 1.8141751289367676,
      "learning_rate": 0.00019998624456946492,
      "loss": 1.6135,
      "step": 981
    },
    {
      "epoch": 0.03516751123605565,
      "grad_norm": 1.8859864473342896,
      "learning_rate": 0.00019998605151763917,
      "loss": 2.0094,
      "step": 982
    },
    {
      "epoch": 0.035203323365623936,
      "grad_norm": 1.6842153072357178,
      "learning_rate": 0.00019998585712063008,
      "loss": 1.8373,
      "step": 983
    },
    {
      "epoch": 0.03523913549519222,
      "grad_norm": 1.4130045175552368,
      "learning_rate": 0.00019998566137844026,
      "loss": 2.2282,
      "step": 984
    },
    {
      "epoch": 0.03527494762476051,
      "grad_norm": 2.0987722873687744,
      "learning_rate": 0.00019998546429107235,
      "loss": 2.1985,
      "step": 985
    },
    {
      "epoch": 0.03531075975432879,
      "grad_norm": 1.7609018087387085,
      "learning_rate": 0.00019998526585852898,
      "loss": 2.0455,
      "step": 986
    },
    {
      "epoch": 0.035346571883897074,
      "grad_norm": 1.6820363998413086,
      "learning_rate": 0.00019998506608081282,
      "loss": 2.0726,
      "step": 987
    },
    {
      "epoch": 0.035382384013465364,
      "grad_norm": 2.3095192909240723,
      "learning_rate": 0.00019998486495792657,
      "loss": 1.9518,
      "step": 988
    },
    {
      "epoch": 0.035418196143033646,
      "grad_norm": 1.5085362195968628,
      "learning_rate": 0.00019998466248987294,
      "loss": 2.0056,
      "step": 989
    },
    {
      "epoch": 0.03545400827260193,
      "grad_norm": 1.2365782260894775,
      "learning_rate": 0.00019998445867665463,
      "loss": 2.0304,
      "step": 990
    },
    {
      "epoch": 0.03548982040217021,
      "grad_norm": 1.5272630453109741,
      "learning_rate": 0.0001999842535182744,
      "loss": 1.9765,
      "step": 991
    },
    {
      "epoch": 0.0355256325317385,
      "grad_norm": 1.3069766759872437,
      "learning_rate": 0.00019998404701473504,
      "loss": 2.0625,
      "step": 992
    },
    {
      "epoch": 0.035561444661306785,
      "grad_norm": 1.3852381706237793,
      "learning_rate": 0.00019998383916603927,
      "loss": 2.0252,
      "step": 993
    },
    {
      "epoch": 0.03559725679087507,
      "grad_norm": 2.0161495208740234,
      "learning_rate": 0.00019998362997218993,
      "loss": 2.197,
      "step": 994
    },
    {
      "epoch": 0.03563306892044336,
      "grad_norm": 1.4747849702835083,
      "learning_rate": 0.0001999834194331898,
      "loss": 2.02,
      "step": 995
    },
    {
      "epoch": 0.03566888105001164,
      "grad_norm": 2.3067164421081543,
      "learning_rate": 0.00019998320754904177,
      "loss": 2.6484,
      "step": 996
    },
    {
      "epoch": 0.03570469317957992,
      "grad_norm": 1.9584856033325195,
      "learning_rate": 0.0001999829943197486,
      "loss": 1.8812,
      "step": 997
    },
    {
      "epoch": 0.035740505309148206,
      "grad_norm": 1.4480103254318237,
      "learning_rate": 0.00019998277974531326,
      "loss": 1.9695,
      "step": 998
    },
    {
      "epoch": 0.035776317438716496,
      "grad_norm": 1.4160569906234741,
      "learning_rate": 0.00019998256382573856,
      "loss": 2.1651,
      "step": 999
    },
    {
      "epoch": 0.03581212956828478,
      "grad_norm": 1.8961275815963745,
      "learning_rate": 0.0001999823465610274,
      "loss": 2.2543,
      "step": 1000
    },
    {
      "epoch": 0.03584794169785306,
      "grad_norm": 1.6168714761734009,
      "learning_rate": 0.0001999821279511828,
      "loss": 1.8404,
      "step": 1001
    },
    {
      "epoch": 0.03588375382742135,
      "grad_norm": 1.4820830821990967,
      "learning_rate": 0.0001999819079962076,
      "loss": 1.7631,
      "step": 1002
    },
    {
      "epoch": 0.035919565956989634,
      "grad_norm": 1.6105800867080688,
      "learning_rate": 0.0001999816866961048,
      "loss": 2.187,
      "step": 1003
    },
    {
      "epoch": 0.03595537808655792,
      "grad_norm": 1.5409481525421143,
      "learning_rate": 0.00019998146405087738,
      "loss": 1.8704,
      "step": 1004
    },
    {
      "epoch": 0.0359911902161262,
      "grad_norm": 1.9356262683868408,
      "learning_rate": 0.00019998124006052832,
      "loss": 1.9801,
      "step": 1005
    },
    {
      "epoch": 0.03602700234569449,
      "grad_norm": 1.5342742204666138,
      "learning_rate": 0.00019998101472506064,
      "loss": 2.0569,
      "step": 1006
    },
    {
      "epoch": 0.03606281447526277,
      "grad_norm": 1.746179223060608,
      "learning_rate": 0.00019998078804447738,
      "loss": 2.0516,
      "step": 1007
    },
    {
      "epoch": 0.036098626604831055,
      "grad_norm": 1.6185777187347412,
      "learning_rate": 0.00019998056001878158,
      "loss": 1.9516,
      "step": 1008
    },
    {
      "epoch": 0.03613443873439934,
      "grad_norm": 1.815090537071228,
      "learning_rate": 0.0001999803306479763,
      "loss": 1.9385,
      "step": 1009
    },
    {
      "epoch": 0.03617025086396763,
      "grad_norm": 1.338623285293579,
      "learning_rate": 0.00019998009993206462,
      "loss": 1.8261,
      "step": 1010
    },
    {
      "epoch": 0.03620606299353591,
      "grad_norm": 1.5827144384384155,
      "learning_rate": 0.0001999798678710497,
      "loss": 1.7827,
      "step": 1011
    },
    {
      "epoch": 0.03624187512310419,
      "grad_norm": 1.5187708139419556,
      "learning_rate": 0.00019997963446493461,
      "loss": 2.2097,
      "step": 1012
    },
    {
      "epoch": 0.03627768725267248,
      "grad_norm": 1.234924554824829,
      "learning_rate": 0.00019997939971372252,
      "loss": 1.7321,
      "step": 1013
    },
    {
      "epoch": 0.036313499382240766,
      "grad_norm": 1.9228838682174683,
      "learning_rate": 0.00019997916361741655,
      "loss": 2.0538,
      "step": 1014
    },
    {
      "epoch": 0.03634931151180905,
      "grad_norm": 1.6233713626861572,
      "learning_rate": 0.0001999789261760199,
      "loss": 2.216,
      "step": 1015
    },
    {
      "epoch": 0.03638512364137733,
      "grad_norm": 1.3126624822616577,
      "learning_rate": 0.00019997868738953577,
      "loss": 1.7939,
      "step": 1016
    },
    {
      "epoch": 0.03642093577094562,
      "grad_norm": 1.9218897819519043,
      "learning_rate": 0.00019997844725796733,
      "loss": 2.1702,
      "step": 1017
    },
    {
      "epoch": 0.036456747900513904,
      "grad_norm": 1.6152637004852295,
      "learning_rate": 0.0001999782057813179,
      "loss": 1.9599,
      "step": 1018
    },
    {
      "epoch": 0.03649256003008219,
      "grad_norm": 1.2749489545822144,
      "learning_rate": 0.00019997796295959065,
      "loss": 1.9781,
      "step": 1019
    },
    {
      "epoch": 0.03652837215965048,
      "grad_norm": 1.346900224685669,
      "learning_rate": 0.00019997771879278883,
      "loss": 1.8683,
      "step": 1020
    },
    {
      "epoch": 0.03656418428921876,
      "grad_norm": 1.3836091756820679,
      "learning_rate": 0.00019997747328091584,
      "loss": 1.7748,
      "step": 1021
    },
    {
      "epoch": 0.03659999641878704,
      "grad_norm": 1.3248271942138672,
      "learning_rate": 0.00019997722642397484,
      "loss": 2.0045,
      "step": 1022
    },
    {
      "epoch": 0.036635808548355325,
      "grad_norm": 1.6735011339187622,
      "learning_rate": 0.00019997697822196926,
      "loss": 1.9793,
      "step": 1023
    },
    {
      "epoch": 0.036671620677923615,
      "grad_norm": 2.2710859775543213,
      "learning_rate": 0.00019997672867490238,
      "loss": 1.8652,
      "step": 1024
    },
    {
      "epoch": 0.0367074328074919,
      "grad_norm": 1.8319227695465088,
      "learning_rate": 0.0001999764777827776,
      "loss": 2.1901,
      "step": 1025
    },
    {
      "epoch": 0.03674324493706018,
      "grad_norm": 1.5663321018218994,
      "learning_rate": 0.00019997622554559824,
      "loss": 1.9499,
      "step": 1026
    },
    {
      "epoch": 0.03677905706662847,
      "grad_norm": 1.2792712450027466,
      "learning_rate": 0.00019997597196336775,
      "loss": 2.1186,
      "step": 1027
    },
    {
      "epoch": 0.03681486919619675,
      "grad_norm": 1.6958818435668945,
      "learning_rate": 0.00019997571703608952,
      "loss": 1.9744,
      "step": 1028
    },
    {
      "epoch": 0.036850681325765036,
      "grad_norm": 1.8845629692077637,
      "learning_rate": 0.00019997546076376695,
      "loss": 2.0421,
      "step": 1029
    },
    {
      "epoch": 0.03688649345533332,
      "grad_norm": 2.2948620319366455,
      "learning_rate": 0.00019997520314640356,
      "loss": 2.1033,
      "step": 1030
    },
    {
      "epoch": 0.03692230558490161,
      "grad_norm": 1.8859891891479492,
      "learning_rate": 0.00019997494418400272,
      "loss": 1.7958,
      "step": 1031
    },
    {
      "epoch": 0.03695811771446989,
      "grad_norm": 1.6800627708435059,
      "learning_rate": 0.00019997468387656796,
      "loss": 2.0049,
      "step": 1032
    },
    {
      "epoch": 0.036993929844038174,
      "grad_norm": 2.595055341720581,
      "learning_rate": 0.00019997442222410283,
      "loss": 2.2109,
      "step": 1033
    },
    {
      "epoch": 0.03702974197360646,
      "grad_norm": 1.660823941230774,
      "learning_rate": 0.0001999741592266108,
      "loss": 1.9758,
      "step": 1034
    },
    {
      "epoch": 0.03706555410317475,
      "grad_norm": 2.998129367828369,
      "learning_rate": 0.0001999738948840954,
      "loss": 1.9876,
      "step": 1035
    },
    {
      "epoch": 0.03710136623274303,
      "grad_norm": 1.9035344123840332,
      "learning_rate": 0.0001999736291965602,
      "loss": 2.2105,
      "step": 1036
    },
    {
      "epoch": 0.03713717836231131,
      "grad_norm": 1.6378567218780518,
      "learning_rate": 0.00019997336216400876,
      "loss": 1.985,
      "step": 1037
    },
    {
      "epoch": 0.0371729904918796,
      "grad_norm": 2.2616209983825684,
      "learning_rate": 0.00019997309378644472,
      "loss": 2.1327,
      "step": 1038
    },
    {
      "epoch": 0.037208802621447885,
      "grad_norm": 1.207955241203308,
      "learning_rate": 0.00019997282406387167,
      "loss": 1.7478,
      "step": 1039
    },
    {
      "epoch": 0.03724461475101617,
      "grad_norm": 1.3184400796890259,
      "learning_rate": 0.00019997255299629318,
      "loss": 1.6028,
      "step": 1040
    },
    {
      "epoch": 0.03728042688058445,
      "grad_norm": 1.752772569656372,
      "learning_rate": 0.00019997228058371298,
      "loss": 1.9437,
      "step": 1041
    },
    {
      "epoch": 0.03731623901015274,
      "grad_norm": 1.965004563331604,
      "learning_rate": 0.00019997200682613468,
      "loss": 2.4803,
      "step": 1042
    },
    {
      "epoch": 0.037352051139721024,
      "grad_norm": 1.6652560234069824,
      "learning_rate": 0.00019997173172356202,
      "loss": 2.1181,
      "step": 1043
    },
    {
      "epoch": 0.037387863269289306,
      "grad_norm": 1.3832405805587769,
      "learning_rate": 0.00019997145527599864,
      "loss": 2.0115,
      "step": 1044
    },
    {
      "epoch": 0.037423675398857596,
      "grad_norm": 1.9811400175094604,
      "learning_rate": 0.00019997117748344825,
      "loss": 2.0448,
      "step": 1045
    },
    {
      "epoch": 0.03745948752842588,
      "grad_norm": 1.222140908241272,
      "learning_rate": 0.00019997089834591466,
      "loss": 1.8386,
      "step": 1046
    },
    {
      "epoch": 0.03749529965799416,
      "grad_norm": 1.2641700506210327,
      "learning_rate": 0.00019997061786340158,
      "loss": 1.7789,
      "step": 1047
    },
    {
      "epoch": 0.037531111787562445,
      "grad_norm": 2.9230399131774902,
      "learning_rate": 0.00019997033603591277,
      "loss": 2.0249,
      "step": 1048
    },
    {
      "epoch": 0.037566923917130735,
      "grad_norm": 1.4298175573349,
      "learning_rate": 0.00019997005286345208,
      "loss": 1.6235,
      "step": 1049
    },
    {
      "epoch": 0.03760273604669902,
      "grad_norm": 2.0398781299591064,
      "learning_rate": 0.00019996976834602324,
      "loss": 2.2335,
      "step": 1050
    },
    {
      "epoch": 0.0376385481762673,
      "grad_norm": 2.0259313583374023,
      "learning_rate": 0.00019996948248363015,
      "loss": 2.1829,
      "step": 1051
    },
    {
      "epoch": 0.03767436030583559,
      "grad_norm": 1.503429651260376,
      "learning_rate": 0.0001999691952762766,
      "loss": 1.8359,
      "step": 1052
    },
    {
      "epoch": 0.03771017243540387,
      "grad_norm": 1.3695441484451294,
      "learning_rate": 0.00019996890672396652,
      "loss": 1.8916,
      "step": 1053
    },
    {
      "epoch": 0.037745984564972156,
      "grad_norm": 2.213430881500244,
      "learning_rate": 0.0001999686168267037,
      "loss": 2.0709,
      "step": 1054
    },
    {
      "epoch": 0.03778179669454044,
      "grad_norm": 1.7021877765655518,
      "learning_rate": 0.0001999683255844921,
      "loss": 1.8707,
      "step": 1055
    },
    {
      "epoch": 0.03781760882410873,
      "grad_norm": 1.7013872861862183,
      "learning_rate": 0.00019996803299733565,
      "loss": 1.8043,
      "step": 1056
    },
    {
      "epoch": 0.03785342095367701,
      "grad_norm": 2.1002440452575684,
      "learning_rate": 0.00019996773906523827,
      "loss": 1.7684,
      "step": 1057
    },
    {
      "epoch": 0.037889233083245294,
      "grad_norm": 1.6813231706619263,
      "learning_rate": 0.0001999674437882039,
      "loss": 1.9193,
      "step": 1058
    },
    {
      "epoch": 0.03792504521281358,
      "grad_norm": 1.3570369482040405,
      "learning_rate": 0.0001999671471662365,
      "loss": 1.7127,
      "step": 1059
    },
    {
      "epoch": 0.03796085734238187,
      "grad_norm": 3.0057036876678467,
      "learning_rate": 0.0001999668491993401,
      "loss": 2.5179,
      "step": 1060
    },
    {
      "epoch": 0.03799666947195015,
      "grad_norm": 1.5875788927078247,
      "learning_rate": 0.00019996654988751867,
      "loss": 1.9084,
      "step": 1061
    },
    {
      "epoch": 0.03803248160151843,
      "grad_norm": 4.081822395324707,
      "learning_rate": 0.0001999662492307763,
      "loss": 2.2503,
      "step": 1062
    },
    {
      "epoch": 0.03806829373108672,
      "grad_norm": 1.6430948972702026,
      "learning_rate": 0.000199965947229117,
      "loss": 1.6658,
      "step": 1063
    },
    {
      "epoch": 0.038104105860655005,
      "grad_norm": 1.858124017715454,
      "learning_rate": 0.0001999656438825448,
      "loss": 1.8926,
      "step": 1064
    },
    {
      "epoch": 0.03813991799022329,
      "grad_norm": 1.7443957328796387,
      "learning_rate": 0.0001999653391910638,
      "loss": 1.8012,
      "step": 1065
    },
    {
      "epoch": 0.03817573011979157,
      "grad_norm": 1.912501335144043,
      "learning_rate": 0.00019996503315467811,
      "loss": 1.8669,
      "step": 1066
    },
    {
      "epoch": 0.03821154224935986,
      "grad_norm": 1.8698126077651978,
      "learning_rate": 0.00019996472577339186,
      "loss": 2.3256,
      "step": 1067
    },
    {
      "epoch": 0.03824735437892814,
      "grad_norm": 1.444370985031128,
      "learning_rate": 0.00019996441704720917,
      "loss": 1.724,
      "step": 1068
    },
    {
      "epoch": 0.038283166508496426,
      "grad_norm": 2.0056838989257812,
      "learning_rate": 0.00019996410697613418,
      "loss": 1.8959,
      "step": 1069
    },
    {
      "epoch": 0.038318978638064716,
      "grad_norm": 1.4529732465744019,
      "learning_rate": 0.0001999637955601711,
      "loss": 2.0943,
      "step": 1070
    },
    {
      "epoch": 0.038354790767633,
      "grad_norm": 1.5077389478683472,
      "learning_rate": 0.00019996348279932406,
      "loss": 1.6368,
      "step": 1071
    },
    {
      "epoch": 0.03839060289720128,
      "grad_norm": 1.458603858947754,
      "learning_rate": 0.0001999631686935973,
      "loss": 2.1051,
      "step": 1072
    },
    {
      "epoch": 0.038426415026769564,
      "grad_norm": 1.7620093822479248,
      "learning_rate": 0.0001999628532429951,
      "loss": 2.2006,
      "step": 1073
    },
    {
      "epoch": 0.038462227156337854,
      "grad_norm": 1.673052430152893,
      "learning_rate": 0.00019996253644752158,
      "loss": 1.9259,
      "step": 1074
    },
    {
      "epoch": 0.03849803928590614,
      "grad_norm": 1.3855998516082764,
      "learning_rate": 0.00019996221830718115,
      "loss": 1.9441,
      "step": 1075
    },
    {
      "epoch": 0.03853385141547442,
      "grad_norm": 1.4624788761138916,
      "learning_rate": 0.00019996189882197797,
      "loss": 1.9249,
      "step": 1076
    },
    {
      "epoch": 0.0385696635450427,
      "grad_norm": 2.3567802906036377,
      "learning_rate": 0.0001999615779919164,
      "loss": 1.8877,
      "step": 1077
    },
    {
      "epoch": 0.03860547567461099,
      "grad_norm": 1.6427150964736938,
      "learning_rate": 0.0001999612558170007,
      "loss": 2.2207,
      "step": 1078
    },
    {
      "epoch": 0.038641287804179275,
      "grad_norm": 1.4462939500808716,
      "learning_rate": 0.0001999609322972353,
      "loss": 1.9105,
      "step": 1079
    },
    {
      "epoch": 0.03867709993374756,
      "grad_norm": 1.3780089616775513,
      "learning_rate": 0.00019996060743262447,
      "loss": 1.9752,
      "step": 1080
    },
    {
      "epoch": 0.03871291206331585,
      "grad_norm": 1.4172141551971436,
      "learning_rate": 0.00019996028122317257,
      "loss": 1.9202,
      "step": 1081
    },
    {
      "epoch": 0.03874872419288413,
      "grad_norm": 1.6056876182556152,
      "learning_rate": 0.00019995995366888408,
      "loss": 2.0305,
      "step": 1082
    },
    {
      "epoch": 0.03878453632245241,
      "grad_norm": 1.9504435062408447,
      "learning_rate": 0.00019995962476976336,
      "loss": 1.7623,
      "step": 1083
    },
    {
      "epoch": 0.038820348452020696,
      "grad_norm": 1.192026138305664,
      "learning_rate": 0.00019995929452581478,
      "loss": 2.051,
      "step": 1084
    },
    {
      "epoch": 0.038856160581588986,
      "grad_norm": 2.1449334621429443,
      "learning_rate": 0.00019995896293704285,
      "loss": 1.9216,
      "step": 1085
    },
    {
      "epoch": 0.03889197271115727,
      "grad_norm": 1.6051068305969238,
      "learning_rate": 0.00019995863000345202,
      "loss": 1.7794,
      "step": 1086
    },
    {
      "epoch": 0.03892778484072555,
      "grad_norm": 1.9584460258483887,
      "learning_rate": 0.00019995829572504677,
      "loss": 1.84,
      "step": 1087
    },
    {
      "epoch": 0.03896359697029384,
      "grad_norm": 1.3237011432647705,
      "learning_rate": 0.00019995796010183157,
      "loss": 1.9595,
      "step": 1088
    },
    {
      "epoch": 0.038999409099862124,
      "grad_norm": 1.275923728942871,
      "learning_rate": 0.00019995762313381095,
      "loss": 1.8077,
      "step": 1089
    },
    {
      "epoch": 0.03903522122943041,
      "grad_norm": 1.2776323556900024,
      "learning_rate": 0.00019995728482098945,
      "loss": 1.798,
      "step": 1090
    },
    {
      "epoch": 0.03907103335899869,
      "grad_norm": 1.3595540523529053,
      "learning_rate": 0.00019995694516337164,
      "loss": 2.0992,
      "step": 1091
    },
    {
      "epoch": 0.03910684548856698,
      "grad_norm": 1.3693207502365112,
      "learning_rate": 0.00019995660416096206,
      "loss": 1.9173,
      "step": 1092
    },
    {
      "epoch": 0.03914265761813526,
      "grad_norm": 1.6067368984222412,
      "learning_rate": 0.00019995626181376527,
      "loss": 1.9124,
      "step": 1093
    },
    {
      "epoch": 0.039178469747703545,
      "grad_norm": 1.5119096040725708,
      "learning_rate": 0.00019995591812178596,
      "loss": 2.2585,
      "step": 1094
    },
    {
      "epoch": 0.039214281877271835,
      "grad_norm": 1.692460060119629,
      "learning_rate": 0.00019995557308502866,
      "loss": 1.9322,
      "step": 1095
    },
    {
      "epoch": 0.03925009400684012,
      "grad_norm": 2.6956653594970703,
      "learning_rate": 0.00019995522670349808,
      "loss": 2.2339,
      "step": 1096
    },
    {
      "epoch": 0.0392859061364084,
      "grad_norm": 1.5072599649429321,
      "learning_rate": 0.00019995487897719888,
      "loss": 2.0304,
      "step": 1097
    },
    {
      "epoch": 0.039321718265976684,
      "grad_norm": 1.522534728050232,
      "learning_rate": 0.00019995452990613567,
      "loss": 1.7841,
      "step": 1098
    },
    {
      "epoch": 0.03935753039554497,
      "grad_norm": 3.1163241863250732,
      "learning_rate": 0.00019995417949031323,
      "loss": 2.0245,
      "step": 1099
    },
    {
      "epoch": 0.039393342525113256,
      "grad_norm": 2.7022454738616943,
      "learning_rate": 0.00019995382772973623,
      "loss": 2.064,
      "step": 1100
    },
    {
      "epoch": 0.03942915465468154,
      "grad_norm": 1.7741551399230957,
      "learning_rate": 0.00019995347462440938,
      "loss": 1.8704,
      "step": 1101
    },
    {
      "epoch": 0.03946496678424982,
      "grad_norm": 2.4735567569732666,
      "learning_rate": 0.0001999531201743375,
      "loss": 2.2503,
      "step": 1102
    },
    {
      "epoch": 0.03950077891381811,
      "grad_norm": 2.0130088329315186,
      "learning_rate": 0.0001999527643795253,
      "loss": 2.0751,
      "step": 1103
    },
    {
      "epoch": 0.039536591043386395,
      "grad_norm": 1.6272753477096558,
      "learning_rate": 0.00019995240723997757,
      "loss": 2.1966,
      "step": 1104
    },
    {
      "epoch": 0.03957240317295468,
      "grad_norm": 1.5757356882095337,
      "learning_rate": 0.00019995204875569914,
      "loss": 1.7593,
      "step": 1105
    },
    {
      "epoch": 0.03960821530252297,
      "grad_norm": 1.7916936874389648,
      "learning_rate": 0.00019995168892669485,
      "loss": 1.9154,
      "step": 1106
    },
    {
      "epoch": 0.03964402743209125,
      "grad_norm": 1.6669237613677979,
      "learning_rate": 0.00019995132775296948,
      "loss": 2.0772,
      "step": 1107
    },
    {
      "epoch": 0.03967983956165953,
      "grad_norm": 1.3182016611099243,
      "learning_rate": 0.00019995096523452795,
      "loss": 1.6875,
      "step": 1108
    },
    {
      "epoch": 0.039715651691227816,
      "grad_norm": 1.801761269569397,
      "learning_rate": 0.0001999506013713751,
      "loss": 1.8703,
      "step": 1109
    },
    {
      "epoch": 0.039751463820796105,
      "grad_norm": 1.1991573572158813,
      "learning_rate": 0.0001999502361635158,
      "loss": 2.0746,
      "step": 1110
    },
    {
      "epoch": 0.03978727595036439,
      "grad_norm": 1.6609413623809814,
      "learning_rate": 0.00019994986961095504,
      "loss": 2.0265,
      "step": 1111
    },
    {
      "epoch": 0.03982308807993267,
      "grad_norm": 1.425610065460205,
      "learning_rate": 0.0001999495017136977,
      "loss": 1.757,
      "step": 1112
    },
    {
      "epoch": 0.03985890020950096,
      "grad_norm": 1.956095576286316,
      "learning_rate": 0.00019994913247174876,
      "loss": 2.0188,
      "step": 1113
    },
    {
      "epoch": 0.039894712339069244,
      "grad_norm": 1.771166205406189,
      "learning_rate": 0.00019994876188511314,
      "loss": 2.2367,
      "step": 1114
    },
    {
      "epoch": 0.03993052446863753,
      "grad_norm": 1.6214959621429443,
      "learning_rate": 0.00019994838995379585,
      "loss": 1.8904,
      "step": 1115
    },
    {
      "epoch": 0.03996633659820581,
      "grad_norm": 1.3988220691680908,
      "learning_rate": 0.0001999480166778019,
      "loss": 1.7266,
      "step": 1116
    },
    {
      "epoch": 0.0400021487277741,
      "grad_norm": 1.1129361391067505,
      "learning_rate": 0.00019994764205713631,
      "loss": 1.9944,
      "step": 1117
    },
    {
      "epoch": 0.04003796085734238,
      "grad_norm": 1.3052695989608765,
      "learning_rate": 0.00019994726609180415,
      "loss": 1.8591,
      "step": 1118
    },
    {
      "epoch": 0.040073772986910665,
      "grad_norm": 1.9215950965881348,
      "learning_rate": 0.00019994688878181044,
      "loss": 2.0272,
      "step": 1119
    },
    {
      "epoch": 0.040109585116478955,
      "grad_norm": 1.3812592029571533,
      "learning_rate": 0.0001999465101271602,
      "loss": 1.8704,
      "step": 1120
    },
    {
      "epoch": 0.04014539724604724,
      "grad_norm": 1.5294747352600098,
      "learning_rate": 0.00019994613012785868,
      "loss": 1.8968,
      "step": 1121
    },
    {
      "epoch": 0.04018120937561552,
      "grad_norm": 1.8965163230895996,
      "learning_rate": 0.00019994574878391084,
      "loss": 2.0483,
      "step": 1122
    },
    {
      "epoch": 0.0402170215051838,
      "grad_norm": 1.571008324623108,
      "learning_rate": 0.00019994536609532187,
      "loss": 1.5518,
      "step": 1123
    },
    {
      "epoch": 0.04025283363475209,
      "grad_norm": 2.02764630317688,
      "learning_rate": 0.00019994498206209695,
      "loss": 1.5406,
      "step": 1124
    },
    {
      "epoch": 0.040288645764320376,
      "grad_norm": 1.47233247756958,
      "learning_rate": 0.0001999445966842412,
      "loss": 1.9526,
      "step": 1125
    },
    {
      "epoch": 0.04032445789388866,
      "grad_norm": 1.4829063415527344,
      "learning_rate": 0.00019994420996175983,
      "loss": 1.8804,
      "step": 1126
    },
    {
      "epoch": 0.04036027002345694,
      "grad_norm": 1.8773620128631592,
      "learning_rate": 0.00019994382189465802,
      "loss": 2.0499,
      "step": 1127
    },
    {
      "epoch": 0.04039608215302523,
      "grad_norm": 2.0403313636779785,
      "learning_rate": 0.000199943432482941,
      "loss": 1.7639,
      "step": 1128
    },
    {
      "epoch": 0.040431894282593514,
      "grad_norm": 2.01694917678833,
      "learning_rate": 0.00019994304172661403,
      "loss": 2.5179,
      "step": 1129
    },
    {
      "epoch": 0.0404677064121618,
      "grad_norm": 1.556822419166565,
      "learning_rate": 0.00019994264962568234,
      "loss": 2.0218,
      "step": 1130
    },
    {
      "epoch": 0.04050351854173009,
      "grad_norm": 2.926945924758911,
      "learning_rate": 0.00019994225618015125,
      "loss": 2.1212,
      "step": 1131
    },
    {
      "epoch": 0.04053933067129837,
      "grad_norm": 2.010427713394165,
      "learning_rate": 0.000199941861390026,
      "loss": 2.0122,
      "step": 1132
    },
    {
      "epoch": 0.04057514280086665,
      "grad_norm": 1.425561547279358,
      "learning_rate": 0.0001999414652553119,
      "loss": 1.9191,
      "step": 1133
    },
    {
      "epoch": 0.040610954930434935,
      "grad_norm": 1.2319929599761963,
      "learning_rate": 0.00019994106777601432,
      "loss": 1.7172,
      "step": 1134
    },
    {
      "epoch": 0.040646767060003225,
      "grad_norm": 1.7433674335479736,
      "learning_rate": 0.00019994066895213857,
      "loss": 1.8323,
      "step": 1135
    },
    {
      "epoch": 0.04068257918957151,
      "grad_norm": 1.2644333839416504,
      "learning_rate": 0.00019994026878369003,
      "loss": 1.8999,
      "step": 1136
    },
    {
      "epoch": 0.04071839131913979,
      "grad_norm": 1.5659446716308594,
      "learning_rate": 0.00019993986727067414,
      "loss": 1.7909,
      "step": 1137
    },
    {
      "epoch": 0.04075420344870808,
      "grad_norm": 2.269122362136841,
      "learning_rate": 0.0001999394644130962,
      "loss": 2.0758,
      "step": 1138
    },
    {
      "epoch": 0.04079001557827636,
      "grad_norm": 1.7199034690856934,
      "learning_rate": 0.00019993906021096168,
      "loss": 1.8833,
      "step": 1139
    },
    {
      "epoch": 0.040825827707844646,
      "grad_norm": 1.6728589534759521,
      "learning_rate": 0.00019993865466427603,
      "loss": 1.8298,
      "step": 1140
    },
    {
      "epoch": 0.04086163983741293,
      "grad_norm": 2.495798349380493,
      "learning_rate": 0.00019993824777304469,
      "loss": 1.639,
      "step": 1141
    },
    {
      "epoch": 0.04089745196698122,
      "grad_norm": 1.8760560750961304,
      "learning_rate": 0.0001999378395372731,
      "loss": 1.5388,
      "step": 1142
    },
    {
      "epoch": 0.0409332640965495,
      "grad_norm": 1.9759643077850342,
      "learning_rate": 0.00019993742995696686,
      "loss": 2.1794,
      "step": 1143
    },
    {
      "epoch": 0.040969076226117784,
      "grad_norm": 1.5241389274597168,
      "learning_rate": 0.0001999370190321314,
      "loss": 1.9345,
      "step": 1144
    },
    {
      "epoch": 0.041004888355686074,
      "grad_norm": 1.6700853109359741,
      "learning_rate": 0.0001999366067627722,
      "loss": 1.6506,
      "step": 1145
    },
    {
      "epoch": 0.04104070048525436,
      "grad_norm": 1.6093263626098633,
      "learning_rate": 0.0001999361931488949,
      "loss": 1.9827,
      "step": 1146
    },
    {
      "epoch": 0.04107651261482264,
      "grad_norm": 1.726548194885254,
      "learning_rate": 0.00019993577819050505,
      "loss": 1.9425,
      "step": 1147
    },
    {
      "epoch": 0.04111232474439092,
      "grad_norm": 1.4879534244537354,
      "learning_rate": 0.00019993536188760817,
      "loss": 1.8571,
      "step": 1148
    },
    {
      "epoch": 0.04114813687395921,
      "grad_norm": 2.0006051063537598,
      "learning_rate": 0.00019993494424020992,
      "loss": 1.7725,
      "step": 1149
    },
    {
      "epoch": 0.041183949003527495,
      "grad_norm": 1.6221108436584473,
      "learning_rate": 0.00019993452524831592,
      "loss": 1.8815,
      "step": 1150
    },
    {
      "epoch": 0.04121976113309578,
      "grad_norm": 1.5773656368255615,
      "learning_rate": 0.0001999341049119318,
      "loss": 1.7668,
      "step": 1151
    },
    {
      "epoch": 0.04125557326266406,
      "grad_norm": 1.7638592720031738,
      "learning_rate": 0.00019993368323106315,
      "loss": 1.9433,
      "step": 1152
    },
    {
      "epoch": 0.04129138539223235,
      "grad_norm": 1.4809341430664062,
      "learning_rate": 0.0001999332602057157,
      "loss": 1.6965,
      "step": 1153
    },
    {
      "epoch": 0.041327197521800633,
      "grad_norm": 1.7813769578933716,
      "learning_rate": 0.0001999328358358952,
      "loss": 1.8184,
      "step": 1154
    },
    {
      "epoch": 0.041363009651368916,
      "grad_norm": 1.9684855937957764,
      "learning_rate": 0.00019993241012160727,
      "loss": 1.9786,
      "step": 1155
    },
    {
      "epoch": 0.041398821780937206,
      "grad_norm": 1.7000457048416138,
      "learning_rate": 0.00019993198306285766,
      "loss": 1.742,
      "step": 1156
    },
    {
      "epoch": 0.04143463391050549,
      "grad_norm": 1.2915273904800415,
      "learning_rate": 0.0001999315546596521,
      "loss": 1.774,
      "step": 1157
    },
    {
      "epoch": 0.04147044604007377,
      "grad_norm": 1.3124654293060303,
      "learning_rate": 0.0001999311249119964,
      "loss": 1.9188,
      "step": 1158
    },
    {
      "epoch": 0.041506258169642055,
      "grad_norm": 1.339920997619629,
      "learning_rate": 0.0001999306938198963,
      "loss": 1.9007,
      "step": 1159
    },
    {
      "epoch": 0.041542070299210344,
      "grad_norm": 1.8238441944122314,
      "learning_rate": 0.00019993026138335763,
      "loss": 2.3424,
      "step": 1160
    },
    {
      "epoch": 0.04157788242877863,
      "grad_norm": 1.7708998918533325,
      "learning_rate": 0.0001999298276023862,
      "loss": 2.0156,
      "step": 1161
    },
    {
      "epoch": 0.04161369455834691,
      "grad_norm": 1.3194271326065063,
      "learning_rate": 0.00019992939247698784,
      "loss": 1.7528,
      "step": 1162
    },
    {
      "epoch": 0.0416495066879152,
      "grad_norm": 1.9072855710983276,
      "learning_rate": 0.00019992895600716838,
      "loss": 1.944,
      "step": 1163
    },
    {
      "epoch": 0.04168531881748348,
      "grad_norm": 1.2703431844711304,
      "learning_rate": 0.00019992851819293373,
      "loss": 1.8563,
      "step": 1164
    },
    {
      "epoch": 0.041721130947051766,
      "grad_norm": 1.5858049392700195,
      "learning_rate": 0.00019992807903428976,
      "loss": 2.0275,
      "step": 1165
    },
    {
      "epoch": 0.04175694307662005,
      "grad_norm": 1.2060877084732056,
      "learning_rate": 0.0001999276385312424,
      "loss": 1.811,
      "step": 1166
    },
    {
      "epoch": 0.04179275520618834,
      "grad_norm": 1.198927640914917,
      "learning_rate": 0.00019992719668379753,
      "loss": 1.9143,
      "step": 1167
    },
    {
      "epoch": 0.04182856733575662,
      "grad_norm": 2.0849688053131104,
      "learning_rate": 0.00019992675349196114,
      "loss": 2.4221,
      "step": 1168
    },
    {
      "epoch": 0.041864379465324904,
      "grad_norm": 1.9343287944793701,
      "learning_rate": 0.0001999263089557392,
      "loss": 2.047,
      "step": 1169
    },
    {
      "epoch": 0.041900191594893194,
      "grad_norm": 1.5256718397140503,
      "learning_rate": 0.00019992586307513767,
      "loss": 2.1052,
      "step": 1170
    },
    {
      "epoch": 0.041936003724461476,
      "grad_norm": 1.3801934719085693,
      "learning_rate": 0.00019992541585016254,
      "loss": 1.6852,
      "step": 1171
    },
    {
      "epoch": 0.04197181585402976,
      "grad_norm": 1.4266293048858643,
      "learning_rate": 0.0001999249672808198,
      "loss": 1.9,
      "step": 1172
    },
    {
      "epoch": 0.04200762798359804,
      "grad_norm": 1.2615375518798828,
      "learning_rate": 0.00019992451736711554,
      "loss": 2.0813,
      "step": 1173
    },
    {
      "epoch": 0.04204344011316633,
      "grad_norm": 1.4383363723754883,
      "learning_rate": 0.00019992406610905582,
      "loss": 1.6228,
      "step": 1174
    },
    {
      "epoch": 0.042079252242734615,
      "grad_norm": 1.2527996301651,
      "learning_rate": 0.00019992361350664663,
      "loss": 1.8077,
      "step": 1175
    },
    {
      "epoch": 0.0421150643723029,
      "grad_norm": 1.6648797988891602,
      "learning_rate": 0.00019992315955989415,
      "loss": 2.051,
      "step": 1176
    },
    {
      "epoch": 0.04215087650187118,
      "grad_norm": 1.7546112537384033,
      "learning_rate": 0.00019992270426880446,
      "loss": 1.871,
      "step": 1177
    },
    {
      "epoch": 0.04218668863143947,
      "grad_norm": 1.7627754211425781,
      "learning_rate": 0.00019992224763338366,
      "loss": 1.7231,
      "step": 1178
    },
    {
      "epoch": 0.04222250076100775,
      "grad_norm": 1.8482389450073242,
      "learning_rate": 0.00019992178965363787,
      "loss": 1.698,
      "step": 1179
    },
    {
      "epoch": 0.042258312890576036,
      "grad_norm": 1.2556909322738647,
      "learning_rate": 0.00019992133032957336,
      "loss": 1.9038,
      "step": 1180
    },
    {
      "epoch": 0.042294125020144326,
      "grad_norm": 1.856414794921875,
      "learning_rate": 0.0001999208696611962,
      "loss": 1.7274,
      "step": 1181
    },
    {
      "epoch": 0.04232993714971261,
      "grad_norm": 1.5887939929962158,
      "learning_rate": 0.00019992040764851263,
      "loss": 1.7081,
      "step": 1182
    },
    {
      "epoch": 0.04236574927928089,
      "grad_norm": 1.9165586233139038,
      "learning_rate": 0.00019991994429152888,
      "loss": 1.8554,
      "step": 1183
    },
    {
      "epoch": 0.042401561408849174,
      "grad_norm": 2.281977653503418,
      "learning_rate": 0.00019991947959025112,
      "loss": 2.5105,
      "step": 1184
    },
    {
      "epoch": 0.042437373538417464,
      "grad_norm": 1.645222544670105,
      "learning_rate": 0.0001999190135446857,
      "loss": 2.0212,
      "step": 1185
    },
    {
      "epoch": 0.04247318566798575,
      "grad_norm": 1.8959003686904907,
      "learning_rate": 0.00019991854615483882,
      "loss": 2.2363,
      "step": 1186
    },
    {
      "epoch": 0.04250899779755403,
      "grad_norm": 1.5478854179382324,
      "learning_rate": 0.00019991807742071678,
      "loss": 1.9947,
      "step": 1187
    },
    {
      "epoch": 0.04254480992712232,
      "grad_norm": 1.8765732049942017,
      "learning_rate": 0.0001999176073423259,
      "loss": 1.7691,
      "step": 1188
    },
    {
      "epoch": 0.0425806220566906,
      "grad_norm": 1.8564085960388184,
      "learning_rate": 0.00019991713591967252,
      "loss": 1.7588,
      "step": 1189
    },
    {
      "epoch": 0.042616434186258885,
      "grad_norm": 1.8056801557540894,
      "learning_rate": 0.00019991666315276292,
      "loss": 2.027,
      "step": 1190
    },
    {
      "epoch": 0.04265224631582717,
      "grad_norm": 1.8482861518859863,
      "learning_rate": 0.0001999161890416035,
      "loss": 1.8143,
      "step": 1191
    },
    {
      "epoch": 0.04268805844539546,
      "grad_norm": 1.6742010116577148,
      "learning_rate": 0.00019991571358620068,
      "loss": 2.162,
      "step": 1192
    },
    {
      "epoch": 0.04272387057496374,
      "grad_norm": 1.4236180782318115,
      "learning_rate": 0.0001999152367865608,
      "loss": 1.8787,
      "step": 1193
    },
    {
      "epoch": 0.04275968270453202,
      "grad_norm": 1.2990878820419312,
      "learning_rate": 0.0001999147586426903,
      "loss": 1.8382,
      "step": 1194
    },
    {
      "epoch": 0.04279549483410031,
      "grad_norm": 1.8911601305007935,
      "learning_rate": 0.00019991427915459558,
      "loss": 1.8486,
      "step": 1195
    },
    {
      "epoch": 0.042831306963668596,
      "grad_norm": 1.2889195680618286,
      "learning_rate": 0.0001999137983222831,
      "loss": 1.9353,
      "step": 1196
    },
    {
      "epoch": 0.04286711909323688,
      "grad_norm": 1.7363015413284302,
      "learning_rate": 0.0001999133161457594,
      "loss": 1.7446,
      "step": 1197
    },
    {
      "epoch": 0.04290293122280516,
      "grad_norm": 1.3156371116638184,
      "learning_rate": 0.00019991283262503083,
      "loss": 1.9038,
      "step": 1198
    },
    {
      "epoch": 0.04293874335237345,
      "grad_norm": 1.2407479286193848,
      "learning_rate": 0.00019991234776010406,
      "loss": 1.8502,
      "step": 1199
    },
    {
      "epoch": 0.042974555481941734,
      "grad_norm": 1.3936583995819092,
      "learning_rate": 0.0001999118615509855,
      "loss": 1.5118,
      "step": 1200
    },
    {
      "epoch": 0.04301036761151002,
      "grad_norm": 1.2559468746185303,
      "learning_rate": 0.00019991137399768166,
      "loss": 1.9056,
      "step": 1201
    },
    {
      "epoch": 0.0430461797410783,
      "grad_norm": 2.0051515102386475,
      "learning_rate": 0.00019991088510019924,
      "loss": 1.8493,
      "step": 1202
    },
    {
      "epoch": 0.04308199187064659,
      "grad_norm": 1.2575451135635376,
      "learning_rate": 0.0001999103948585447,
      "loss": 1.4424,
      "step": 1203
    },
    {
      "epoch": 0.04311780400021487,
      "grad_norm": 1.596622109413147,
      "learning_rate": 0.00019990990327272467,
      "loss": 2.141,
      "step": 1204
    },
    {
      "epoch": 0.043153616129783155,
      "grad_norm": 1.6816015243530273,
      "learning_rate": 0.00019990941034274577,
      "loss": 1.962,
      "step": 1205
    },
    {
      "epoch": 0.043189428259351445,
      "grad_norm": 1.785096287727356,
      "learning_rate": 0.00019990891606861463,
      "loss": 1.5008,
      "step": 1206
    },
    {
      "epoch": 0.04322524038891973,
      "grad_norm": 1.742447853088379,
      "learning_rate": 0.0001999084204503379,
      "loss": 1.8909,
      "step": 1207
    },
    {
      "epoch": 0.04326105251848801,
      "grad_norm": 1.6396584510803223,
      "learning_rate": 0.00019990792348792224,
      "loss": 1.8476,
      "step": 1208
    },
    {
      "epoch": 0.043296864648056294,
      "grad_norm": 1.9370383024215698,
      "learning_rate": 0.00019990742518137436,
      "loss": 2.1081,
      "step": 1209
    },
    {
      "epoch": 0.04333267677762458,
      "grad_norm": 2.301330089569092,
      "learning_rate": 0.00019990692553070093,
      "loss": 2.0177,
      "step": 1210
    },
    {
      "epoch": 0.043368488907192866,
      "grad_norm": 1.2723262310028076,
      "learning_rate": 0.0001999064245359087,
      "loss": 1.6058,
      "step": 1211
    },
    {
      "epoch": 0.04340430103676115,
      "grad_norm": 1.3464289903640747,
      "learning_rate": 0.00019990592219700437,
      "loss": 1.7071,
      "step": 1212
    },
    {
      "epoch": 0.04344011316632944,
      "grad_norm": 1.548187017440796,
      "learning_rate": 0.00019990541851399476,
      "loss": 1.9317,
      "step": 1213
    },
    {
      "epoch": 0.04347592529589772,
      "grad_norm": 1.9910813570022583,
      "learning_rate": 0.00019990491348688657,
      "loss": 2.1553,
      "step": 1214
    },
    {
      "epoch": 0.043511737425466004,
      "grad_norm": 1.2897824048995972,
      "learning_rate": 0.00019990440711568666,
      "loss": 1.831,
      "step": 1215
    },
    {
      "epoch": 0.04354754955503429,
      "grad_norm": 1.9216833114624023,
      "learning_rate": 0.00019990389940040184,
      "loss": 1.9973,
      "step": 1216
    },
    {
      "epoch": 0.04358336168460258,
      "grad_norm": 2.3305580615997314,
      "learning_rate": 0.0001999033903410389,
      "loss": 1.934,
      "step": 1217
    },
    {
      "epoch": 0.04361917381417086,
      "grad_norm": 1.5975700616836548,
      "learning_rate": 0.00019990287993760473,
      "loss": 1.829,
      "step": 1218
    },
    {
      "epoch": 0.04365498594373914,
      "grad_norm": 1.2177056074142456,
      "learning_rate": 0.00019990236819010615,
      "loss": 1.7145,
      "step": 1219
    },
    {
      "epoch": 0.04369079807330743,
      "grad_norm": 1.501842737197876,
      "learning_rate": 0.0001999018550985501,
      "loss": 1.9816,
      "step": 1220
    },
    {
      "epoch": 0.043726610202875715,
      "grad_norm": 2.3533644676208496,
      "learning_rate": 0.00019990134066294338,
      "loss": 2.2937,
      "step": 1221
    },
    {
      "epoch": 0.043762422332444,
      "grad_norm": 1.9165318012237549,
      "learning_rate": 0.00019990082488329308,
      "loss": 2.0315,
      "step": 1222
    },
    {
      "epoch": 0.04379823446201228,
      "grad_norm": 1.4523035287857056,
      "learning_rate": 0.000199900307759606,
      "loss": 2.2864,
      "step": 1223
    },
    {
      "epoch": 0.04383404659158057,
      "grad_norm": 1.336516261100769,
      "learning_rate": 0.00019989978929188914,
      "loss": 1.9243,
      "step": 1224
    },
    {
      "epoch": 0.043869858721148854,
      "grad_norm": 1.738339900970459,
      "learning_rate": 0.00019989926948014945,
      "loss": 1.7763,
      "step": 1225
    },
    {
      "epoch": 0.043905670850717136,
      "grad_norm": 2.4466097354888916,
      "learning_rate": 0.000199898748324394,
      "loss": 2.1759,
      "step": 1226
    },
    {
      "epoch": 0.04394148298028542,
      "grad_norm": 1.6366623640060425,
      "learning_rate": 0.00019989822582462972,
      "loss": 1.8826,
      "step": 1227
    },
    {
      "epoch": 0.04397729510985371,
      "grad_norm": 1.563489556312561,
      "learning_rate": 0.00019989770198086367,
      "loss": 1.6678,
      "step": 1228
    },
    {
      "epoch": 0.04401310723942199,
      "grad_norm": 1.8374176025390625,
      "learning_rate": 0.0001998971767931029,
      "loss": 1.7684,
      "step": 1229
    },
    {
      "epoch": 0.044048919368990275,
      "grad_norm": 1.9586941003799438,
      "learning_rate": 0.0001998966502613545,
      "loss": 2.132,
      "step": 1230
    },
    {
      "epoch": 0.044084731498558564,
      "grad_norm": 1.2699816226959229,
      "learning_rate": 0.0001998961223856255,
      "loss": 1.9222,
      "step": 1231
    },
    {
      "epoch": 0.04412054362812685,
      "grad_norm": 1.6662757396697998,
      "learning_rate": 0.00019989559316592305,
      "loss": 1.6608,
      "step": 1232
    },
    {
      "epoch": 0.04415635575769513,
      "grad_norm": 1.3291850090026855,
      "learning_rate": 0.00019989506260225426,
      "loss": 2.0618,
      "step": 1233
    },
    {
      "epoch": 0.04419216788726341,
      "grad_norm": 2.277132272720337,
      "learning_rate": 0.00019989453069462623,
      "loss": 1.7717,
      "step": 1234
    },
    {
      "epoch": 0.0442279800168317,
      "grad_norm": 1.6302454471588135,
      "learning_rate": 0.00019989399744304616,
      "loss": 2.0715,
      "step": 1235
    },
    {
      "epoch": 0.044263792146399986,
      "grad_norm": 1.285477876663208,
      "learning_rate": 0.0001998934628475212,
      "loss": 1.5437,
      "step": 1236
    },
    {
      "epoch": 0.04429960427596827,
      "grad_norm": 2.632361888885498,
      "learning_rate": 0.00019989292690805854,
      "loss": 1.8132,
      "step": 1237
    },
    {
      "epoch": 0.04433541640553656,
      "grad_norm": 1.5045816898345947,
      "learning_rate": 0.00019989238962466542,
      "loss": 1.8698,
      "step": 1238
    },
    {
      "epoch": 0.04437122853510484,
      "grad_norm": 1.6610292196273804,
      "learning_rate": 0.00019989185099734903,
      "loss": 1.8337,
      "step": 1239
    },
    {
      "epoch": 0.044407040664673124,
      "grad_norm": 1.562566876411438,
      "learning_rate": 0.00019989131102611667,
      "loss": 1.9222,
      "step": 1240
    },
    {
      "epoch": 0.04444285279424141,
      "grad_norm": 2.671149969100952,
      "learning_rate": 0.00019989076971097555,
      "loss": 1.8991,
      "step": 1241
    },
    {
      "epoch": 0.044478664923809696,
      "grad_norm": 2.1588051319122314,
      "learning_rate": 0.00019989022705193299,
      "loss": 1.9221,
      "step": 1242
    },
    {
      "epoch": 0.04451447705337798,
      "grad_norm": 1.949323296546936,
      "learning_rate": 0.00019988968304899624,
      "loss": 2.0363,
      "step": 1243
    },
    {
      "epoch": 0.04455028918294626,
      "grad_norm": 1.7885140180587769,
      "learning_rate": 0.00019988913770217269,
      "loss": 1.8394,
      "step": 1244
    },
    {
      "epoch": 0.04458610131251455,
      "grad_norm": 1.935902714729309,
      "learning_rate": 0.00019988859101146962,
      "loss": 2.0142,
      "step": 1245
    },
    {
      "epoch": 0.044621913442082835,
      "grad_norm": 1.9690736532211304,
      "learning_rate": 0.00019988804297689438,
      "loss": 1.8359,
      "step": 1246
    },
    {
      "epoch": 0.04465772557165112,
      "grad_norm": 1.5136144161224365,
      "learning_rate": 0.0001998874935984544,
      "loss": 1.9559,
      "step": 1247
    },
    {
      "epoch": 0.0446935377012194,
      "grad_norm": 1.742443561553955,
      "learning_rate": 0.00019988694287615704,
      "loss": 1.9451,
      "step": 1248
    },
    {
      "epoch": 0.04472934983078769,
      "grad_norm": 1.5419561862945557,
      "learning_rate": 0.0001998863908100097,
      "loss": 1.7377,
      "step": 1249
    },
    {
      "epoch": 0.04476516196035597,
      "grad_norm": 1.9197380542755127,
      "learning_rate": 0.00019988583740001984,
      "loss": 1.7331,
      "step": 1250
    },
    {
      "epoch": 0.044800974089924256,
      "grad_norm": 2.089766502380371,
      "learning_rate": 0.00019988528264619485,
      "loss": 1.8769,
      "step": 1251
    },
    {
      "epoch": 0.04483678621949254,
      "grad_norm": 1.1897839307785034,
      "learning_rate": 0.00019988472654854222,
      "loss": 1.9805,
      "step": 1252
    },
    {
      "epoch": 0.04487259834906083,
      "grad_norm": 2.2890515327453613,
      "learning_rate": 0.00019988416910706947,
      "loss": 2.2311,
      "step": 1253
    },
    {
      "epoch": 0.04490841047862911,
      "grad_norm": 1.2708755731582642,
      "learning_rate": 0.00019988361032178403,
      "loss": 1.7924,
      "step": 1254
    },
    {
      "epoch": 0.044944222608197394,
      "grad_norm": 1.8061339855194092,
      "learning_rate": 0.00019988305019269346,
      "loss": 1.7318,
      "step": 1255
    },
    {
      "epoch": 0.044980034737765684,
      "grad_norm": 1.5179500579833984,
      "learning_rate": 0.00019988248871980532,
      "loss": 1.6443,
      "step": 1256
    },
    {
      "epoch": 0.04501584686733397,
      "grad_norm": 1.4282045364379883,
      "learning_rate": 0.0001998819259031271,
      "loss": 1.8233,
      "step": 1257
    },
    {
      "epoch": 0.04505165899690225,
      "grad_norm": 1.1604241132736206,
      "learning_rate": 0.00019988136174266643,
      "loss": 1.8747,
      "step": 1258
    },
    {
      "epoch": 0.04508747112647053,
      "grad_norm": 2.2839112281799316,
      "learning_rate": 0.00019988079623843087,
      "loss": 2.0667,
      "step": 1259
    },
    {
      "epoch": 0.04512328325603882,
      "grad_norm": 2.5396506786346436,
      "learning_rate": 0.000199880229390428,
      "loss": 1.9539,
      "step": 1260
    },
    {
      "epoch": 0.045159095385607105,
      "grad_norm": 1.8665293455123901,
      "learning_rate": 0.00019987966119866554,
      "loss": 1.7051,
      "step": 1261
    },
    {
      "epoch": 0.04519490751517539,
      "grad_norm": 2.028208017349243,
      "learning_rate": 0.00019987909166315103,
      "loss": 1.7659,
      "step": 1262
    },
    {
      "epoch": 0.04523071964474368,
      "grad_norm": 2.1394529342651367,
      "learning_rate": 0.00019987852078389218,
      "loss": 1.6273,
      "step": 1263
    },
    {
      "epoch": 0.04526653177431196,
      "grad_norm": 2.826866626739502,
      "learning_rate": 0.00019987794856089668,
      "loss": 2.0664,
      "step": 1264
    },
    {
      "epoch": 0.04530234390388024,
      "grad_norm": 1.5801968574523926,
      "learning_rate": 0.0001998773749941722,
      "loss": 2.1104,
      "step": 1265
    },
    {
      "epoch": 0.045338156033448526,
      "grad_norm": 1.9282563924789429,
      "learning_rate": 0.00019987680008372647,
      "loss": 1.9696,
      "step": 1266
    },
    {
      "epoch": 0.045373968163016816,
      "grad_norm": 2.0032389163970947,
      "learning_rate": 0.00019987622382956722,
      "loss": 1.8998,
      "step": 1267
    },
    {
      "epoch": 0.0454097802925851,
      "grad_norm": 1.5961376428604126,
      "learning_rate": 0.00019987564623170226,
      "loss": 1.9436,
      "step": 1268
    },
    {
      "epoch": 0.04544559242215338,
      "grad_norm": 1.54341721534729,
      "learning_rate": 0.00019987506729013927,
      "loss": 1.8716,
      "step": 1269
    },
    {
      "epoch": 0.04548140455172167,
      "grad_norm": 2.1734094619750977,
      "learning_rate": 0.0001998744870048861,
      "loss": 1.7878,
      "step": 1270
    },
    {
      "epoch": 0.045517216681289954,
      "grad_norm": 1.2922240495681763,
      "learning_rate": 0.0001998739053759505,
      "loss": 1.7814,
      "step": 1271
    },
    {
      "epoch": 0.04555302881085824,
      "grad_norm": 1.7018311023712158,
      "learning_rate": 0.00019987332240334037,
      "loss": 1.9387,
      "step": 1272
    },
    {
      "epoch": 0.04558884094042652,
      "grad_norm": 1.7782323360443115,
      "learning_rate": 0.00019987273808706347,
      "loss": 1.7283,
      "step": 1273
    },
    {
      "epoch": 0.04562465306999481,
      "grad_norm": 1.5118334293365479,
      "learning_rate": 0.00019987215242712775,
      "loss": 1.7824,
      "step": 1274
    },
    {
      "epoch": 0.04566046519956309,
      "grad_norm": 1.525036096572876,
      "learning_rate": 0.00019987156542354103,
      "loss": 1.7238,
      "step": 1275
    },
    {
      "epoch": 0.045696277329131375,
      "grad_norm": 1.968327522277832,
      "learning_rate": 0.00019987097707631124,
      "loss": 2.0203,
      "step": 1276
    },
    {
      "epoch": 0.04573208945869966,
      "grad_norm": 1.6078931093215942,
      "learning_rate": 0.00019987038738544625,
      "loss": 1.4283,
      "step": 1277
    },
    {
      "epoch": 0.04576790158826795,
      "grad_norm": 1.5513797998428345,
      "learning_rate": 0.00019986979635095402,
      "loss": 1.811,
      "step": 1278
    },
    {
      "epoch": 0.04580371371783623,
      "grad_norm": 1.5902775526046753,
      "learning_rate": 0.00019986920397284253,
      "loss": 1.8791,
      "step": 1279
    },
    {
      "epoch": 0.045839525847404514,
      "grad_norm": 1.7717511653900146,
      "learning_rate": 0.0001998686102511197,
      "loss": 2.1192,
      "step": 1280
    },
    {
      "epoch": 0.0458753379769728,
      "grad_norm": 1.5827869176864624,
      "learning_rate": 0.00019986801518579353,
      "loss": 1.8253,
      "step": 1281
    },
    {
      "epoch": 0.045911150106541086,
      "grad_norm": 2.1853511333465576,
      "learning_rate": 0.00019986741877687207,
      "loss": 1.8936,
      "step": 1282
    },
    {
      "epoch": 0.04594696223610937,
      "grad_norm": 2.5177838802337646,
      "learning_rate": 0.00019986682102436328,
      "loss": 1.8655,
      "step": 1283
    },
    {
      "epoch": 0.04598277436567765,
      "grad_norm": 1.5945955514907837,
      "learning_rate": 0.00019986622192827525,
      "loss": 2.0518,
      "step": 1284
    },
    {
      "epoch": 0.04601858649524594,
      "grad_norm": 1.603678584098816,
      "learning_rate": 0.000199865621488616,
      "loss": 2.0015,
      "step": 1285
    },
    {
      "epoch": 0.046054398624814225,
      "grad_norm": 1.593038558959961,
      "learning_rate": 0.00019986501970539367,
      "loss": 1.9955,
      "step": 1286
    },
    {
      "epoch": 0.04609021075438251,
      "grad_norm": 1.295050024986267,
      "learning_rate": 0.0001998644165786163,
      "loss": 1.8378,
      "step": 1287
    },
    {
      "epoch": 0.0461260228839508,
      "grad_norm": 1.8528512716293335,
      "learning_rate": 0.00019986381210829199,
      "loss": 2.0828,
      "step": 1288
    },
    {
      "epoch": 0.04616183501351908,
      "grad_norm": 1.5784971714019775,
      "learning_rate": 0.00019986320629442893,
      "loss": 1.728,
      "step": 1289
    },
    {
      "epoch": 0.04619764714308736,
      "grad_norm": 1.469783902168274,
      "learning_rate": 0.00019986259913703526,
      "loss": 1.6864,
      "step": 1290
    },
    {
      "epoch": 0.046233459272655646,
      "grad_norm": 1.7400978803634644,
      "learning_rate": 0.00019986199063611913,
      "loss": 1.7075,
      "step": 1291
    },
    {
      "epoch": 0.046269271402223935,
      "grad_norm": 1.7644522190093994,
      "learning_rate": 0.0001998613807916887,
      "loss": 2.161,
      "step": 1292
    },
    {
      "epoch": 0.04630508353179222,
      "grad_norm": 2.0017952919006348,
      "learning_rate": 0.00019986076960375223,
      "loss": 1.9667,
      "step": 1293
    },
    {
      "epoch": 0.0463408956613605,
      "grad_norm": 1.2968863248825073,
      "learning_rate": 0.00019986015707231788,
      "loss": 1.8902,
      "step": 1294
    },
    {
      "epoch": 0.04637670779092879,
      "grad_norm": 1.572568416595459,
      "learning_rate": 0.00019985954319739392,
      "loss": 1.9558,
      "step": 1295
    },
    {
      "epoch": 0.046412519920497074,
      "grad_norm": 1.356263518333435,
      "learning_rate": 0.00019985892797898865,
      "loss": 1.8009,
      "step": 1296
    },
    {
      "epoch": 0.04644833205006536,
      "grad_norm": 1.2009116411209106,
      "learning_rate": 0.00019985831141711033,
      "loss": 1.8879,
      "step": 1297
    },
    {
      "epoch": 0.04648414417963364,
      "grad_norm": 1.1283270120620728,
      "learning_rate": 0.00019985769351176723,
      "loss": 1.7523,
      "step": 1298
    },
    {
      "epoch": 0.04651995630920193,
      "grad_norm": 1.6041669845581055,
      "learning_rate": 0.00019985707426296764,
      "loss": 1.823,
      "step": 1299
    },
    {
      "epoch": 0.04655576843877021,
      "grad_norm": 1.4035066366195679,
      "learning_rate": 0.00019985645367071993,
      "loss": 1.6812,
      "step": 1300
    },
    {
      "epoch": 0.046591580568338495,
      "grad_norm": 1.4297456741333008,
      "learning_rate": 0.00019985583173503244,
      "loss": 1.7875,
      "step": 1301
    },
    {
      "epoch": 0.04662739269790678,
      "grad_norm": 1.2370003461837769,
      "learning_rate": 0.00019985520845591356,
      "loss": 1.7563,
      "step": 1302
    },
    {
      "epoch": 0.04666320482747507,
      "grad_norm": 1.1710233688354492,
      "learning_rate": 0.00019985458383337164,
      "loss": 1.467,
      "step": 1303
    },
    {
      "epoch": 0.04669901695704335,
      "grad_norm": 1.7941819429397583,
      "learning_rate": 0.0001998539578674151,
      "loss": 1.7955,
      "step": 1304
    },
    {
      "epoch": 0.04673482908661163,
      "grad_norm": 1.6461565494537354,
      "learning_rate": 0.00019985333055805236,
      "loss": 1.6701,
      "step": 1305
    },
    {
      "epoch": 0.04677064121617992,
      "grad_norm": 1.215015172958374,
      "learning_rate": 0.00019985270190529187,
      "loss": 1.6302,
      "step": 1306
    },
    {
      "epoch": 0.046806453345748206,
      "grad_norm": 3.887476921081543,
      "learning_rate": 0.00019985207190914206,
      "loss": 1.687,
      "step": 1307
    },
    {
      "epoch": 0.04684226547531649,
      "grad_norm": 1.9764825105667114,
      "learning_rate": 0.00019985144056961141,
      "loss": 2.1641,
      "step": 1308
    },
    {
      "epoch": 0.04687807760488477,
      "grad_norm": 1.4992644786834717,
      "learning_rate": 0.00019985080788670847,
      "loss": 2.0364,
      "step": 1309
    },
    {
      "epoch": 0.04691388973445306,
      "grad_norm": 2.266002655029297,
      "learning_rate": 0.00019985017386044167,
      "loss": 1.8181,
      "step": 1310
    },
    {
      "epoch": 0.046949701864021344,
      "grad_norm": 2.006157636642456,
      "learning_rate": 0.00019984953849081958,
      "loss": 1.9546,
      "step": 1311
    },
    {
      "epoch": 0.04698551399358963,
      "grad_norm": 1.8313133716583252,
      "learning_rate": 0.00019984890177785077,
      "loss": 1.9454,
      "step": 1312
    },
    {
      "epoch": 0.04702132612315792,
      "grad_norm": 2.395103931427002,
      "learning_rate": 0.00019984826372154374,
      "loss": 1.9007,
      "step": 1313
    },
    {
      "epoch": 0.0470571382527262,
      "grad_norm": 1.6291462182998657,
      "learning_rate": 0.00019984762432190717,
      "loss": 1.7289,
      "step": 1314
    },
    {
      "epoch": 0.04709295038229448,
      "grad_norm": 1.8135216236114502,
      "learning_rate": 0.00019984698357894957,
      "loss": 2.0032,
      "step": 1315
    },
    {
      "epoch": 0.047128762511862765,
      "grad_norm": 1.8101180791854858,
      "learning_rate": 0.00019984634149267962,
      "loss": 1.8555,
      "step": 1316
    },
    {
      "epoch": 0.047164574641431055,
      "grad_norm": 1.441382646560669,
      "learning_rate": 0.00019984569806310592,
      "loss": 1.596,
      "step": 1317
    },
    {
      "epoch": 0.04720038677099934,
      "grad_norm": 1.289527416229248,
      "learning_rate": 0.00019984505329023717,
      "loss": 1.6556,
      "step": 1318
    },
    {
      "epoch": 0.04723619890056762,
      "grad_norm": 2.152682304382324,
      "learning_rate": 0.000199844407174082,
      "loss": 2.0037,
      "step": 1319
    },
    {
      "epoch": 0.04727201103013591,
      "grad_norm": 1.3137688636779785,
      "learning_rate": 0.00019984375971464913,
      "loss": 1.4988,
      "step": 1320
    },
    {
      "epoch": 0.04730782315970419,
      "grad_norm": 2.3812808990478516,
      "learning_rate": 0.00019984311091194725,
      "loss": 2.0996,
      "step": 1321
    },
    {
      "epoch": 0.047343635289272476,
      "grad_norm": 1.7241475582122803,
      "learning_rate": 0.0001998424607659851,
      "loss": 1.7122,
      "step": 1322
    },
    {
      "epoch": 0.04737944741884076,
      "grad_norm": 1.3988949060440063,
      "learning_rate": 0.00019984180927677146,
      "loss": 1.6472,
      "step": 1323
    },
    {
      "epoch": 0.04741525954840905,
      "grad_norm": 1.1787669658660889,
      "learning_rate": 0.00019984115644431502,
      "loss": 1.76,
      "step": 1324
    },
    {
      "epoch": 0.04745107167797733,
      "grad_norm": 2.1474668979644775,
      "learning_rate": 0.00019984050226862462,
      "loss": 1.9005,
      "step": 1325
    },
    {
      "epoch": 0.047486883807545614,
      "grad_norm": 1.143454670906067,
      "learning_rate": 0.00019983984674970905,
      "loss": 1.629,
      "step": 1326
    },
    {
      "epoch": 0.0475226959371139,
      "grad_norm": 1.7645509243011475,
      "learning_rate": 0.00019983918988757715,
      "loss": 2.044,
      "step": 1327
    },
    {
      "epoch": 0.04755850806668219,
      "grad_norm": 1.4978320598602295,
      "learning_rate": 0.0001998385316822377,
      "loss": 1.9889,
      "step": 1328
    },
    {
      "epoch": 0.04759432019625047,
      "grad_norm": 1.7313258647918701,
      "learning_rate": 0.0001998378721336996,
      "loss": 2.0302,
      "step": 1329
    },
    {
      "epoch": 0.04763013232581875,
      "grad_norm": 2.19752836227417,
      "learning_rate": 0.0001998372112419717,
      "loss": 2.1565,
      "step": 1330
    },
    {
      "epoch": 0.04766594445538704,
      "grad_norm": 1.5281023979187012,
      "learning_rate": 0.00019983654900706293,
      "loss": 1.7766,
      "step": 1331
    },
    {
      "epoch": 0.047701756584955325,
      "grad_norm": 1.273146152496338,
      "learning_rate": 0.00019983588542898218,
      "loss": 1.3964,
      "step": 1332
    },
    {
      "epoch": 0.04773756871452361,
      "grad_norm": 1.923797607421875,
      "learning_rate": 0.00019983522050773833,
      "loss": 2.0575,
      "step": 1333
    },
    {
      "epoch": 0.04777338084409189,
      "grad_norm": 2.0104284286499023,
      "learning_rate": 0.00019983455424334038,
      "loss": 2.059,
      "step": 1334
    },
    {
      "epoch": 0.04780919297366018,
      "grad_norm": 1.3276079893112183,
      "learning_rate": 0.0001998338866357973,
      "loss": 1.7016,
      "step": 1335
    },
    {
      "epoch": 0.04784500510322846,
      "grad_norm": 1.1934891939163208,
      "learning_rate": 0.00019983321768511801,
      "loss": 1.7879,
      "step": 1336
    },
    {
      "epoch": 0.047880817232796746,
      "grad_norm": 1.8240382671356201,
      "learning_rate": 0.00019983254739131158,
      "loss": 1.9411,
      "step": 1337
    },
    {
      "epoch": 0.047916629362365036,
      "grad_norm": 1.425972819328308,
      "learning_rate": 0.000199831875754387,
      "loss": 1.7801,
      "step": 1338
    },
    {
      "epoch": 0.04795244149193332,
      "grad_norm": 1.7534042596817017,
      "learning_rate": 0.00019983120277435333,
      "loss": 1.6276,
      "step": 1339
    },
    {
      "epoch": 0.0479882536215016,
      "grad_norm": 1.249691128730774,
      "learning_rate": 0.00019983052845121954,
      "loss": 2.0146,
      "step": 1340
    },
    {
      "epoch": 0.048024065751069885,
      "grad_norm": 1.7718329429626465,
      "learning_rate": 0.00019982985278499483,
      "loss": 1.8151,
      "step": 1341
    },
    {
      "epoch": 0.048059877880638174,
      "grad_norm": 2.18198561668396,
      "learning_rate": 0.0001998291757756882,
      "loss": 2.0883,
      "step": 1342
    },
    {
      "epoch": 0.04809569001020646,
      "grad_norm": 1.7033090591430664,
      "learning_rate": 0.00019982849742330875,
      "loss": 1.8902,
      "step": 1343
    },
    {
      "epoch": 0.04813150213977474,
      "grad_norm": 1.6161448955535889,
      "learning_rate": 0.00019982781772786564,
      "loss": 1.9358,
      "step": 1344
    },
    {
      "epoch": 0.04816731426934303,
      "grad_norm": 2.100522041320801,
      "learning_rate": 0.00019982713668936805,
      "loss": 1.798,
      "step": 1345
    },
    {
      "epoch": 0.04820312639891131,
      "grad_norm": 2.4044883251190186,
      "learning_rate": 0.00019982645430782506,
      "loss": 2.3565,
      "step": 1346
    },
    {
      "epoch": 0.048238938528479595,
      "grad_norm": 1.768221378326416,
      "learning_rate": 0.00019982577058324589,
      "loss": 1.9625,
      "step": 1347
    },
    {
      "epoch": 0.04827475065804788,
      "grad_norm": 1.4462714195251465,
      "learning_rate": 0.00019982508551563978,
      "loss": 1.7243,
      "step": 1348
    },
    {
      "epoch": 0.04831056278761617,
      "grad_norm": 2.0050547122955322,
      "learning_rate": 0.00019982439910501588,
      "loss": 1.57,
      "step": 1349
    },
    {
      "epoch": 0.04834637491718445,
      "grad_norm": 1.5143630504608154,
      "learning_rate": 0.0001998237113513835,
      "loss": 1.708,
      "step": 1350
    },
    {
      "epoch": 0.048382187046752734,
      "grad_norm": 1.6981946229934692,
      "learning_rate": 0.00019982302225475182,
      "loss": 2.1118,
      "step": 1351
    },
    {
      "epoch": 0.04841799917632102,
      "grad_norm": 1.5785638093948364,
      "learning_rate": 0.0001998223318151301,
      "loss": 1.6616,
      "step": 1352
    },
    {
      "epoch": 0.048453811305889306,
      "grad_norm": 1.7291991710662842,
      "learning_rate": 0.00019982164003252772,
      "loss": 1.8341,
      "step": 1353
    },
    {
      "epoch": 0.04848962343545759,
      "grad_norm": 2.4716439247131348,
      "learning_rate": 0.0001998209469069539,
      "loss": 2.0056,
      "step": 1354
    },
    {
      "epoch": 0.04852543556502587,
      "grad_norm": 2.480513334274292,
      "learning_rate": 0.00019982025243841804,
      "loss": 2.1853,
      "step": 1355
    },
    {
      "epoch": 0.04856124769459416,
      "grad_norm": 2.5003104209899902,
      "learning_rate": 0.00019981955662692942,
      "loss": 2.2649,
      "step": 1356
    },
    {
      "epoch": 0.048597059824162445,
      "grad_norm": 1.948498010635376,
      "learning_rate": 0.00019981885947249742,
      "loss": 1.8958,
      "step": 1357
    },
    {
      "epoch": 0.04863287195373073,
      "grad_norm": 1.2755658626556396,
      "learning_rate": 0.0001998181609751314,
      "loss": 1.4372,
      "step": 1358
    },
    {
      "epoch": 0.04866868408329901,
      "grad_norm": 1.3627718687057495,
      "learning_rate": 0.00019981746113484082,
      "loss": 1.8786,
      "step": 1359
    },
    {
      "epoch": 0.0487044962128673,
      "grad_norm": 1.4305596351623535,
      "learning_rate": 0.00019981675995163505,
      "loss": 1.7181,
      "step": 1360
    },
    {
      "epoch": 0.04874030834243558,
      "grad_norm": 1.3786169290542603,
      "learning_rate": 0.00019981605742552352,
      "loss": 1.6552,
      "step": 1361
    },
    {
      "epoch": 0.048776120472003866,
      "grad_norm": 2.100572109222412,
      "learning_rate": 0.00019981535355651569,
      "loss": 1.9594,
      "step": 1362
    },
    {
      "epoch": 0.048811932601572156,
      "grad_norm": 2.110886573791504,
      "learning_rate": 0.00019981464834462103,
      "loss": 2.0575,
      "step": 1363
    },
    {
      "epoch": 0.04884774473114044,
      "grad_norm": 1.4668046236038208,
      "learning_rate": 0.00019981394178984903,
      "loss": 1.7259,
      "step": 1364
    },
    {
      "epoch": 0.04888355686070872,
      "grad_norm": 1.8313469886779785,
      "learning_rate": 0.0001998132338922092,
      "loss": 1.9957,
      "step": 1365
    },
    {
      "epoch": 0.048919368990277004,
      "grad_norm": 1.4437497854232788,
      "learning_rate": 0.00019981252465171102,
      "loss": 1.8628,
      "step": 1366
    },
    {
      "epoch": 0.048955181119845294,
      "grad_norm": 1.769682765007019,
      "learning_rate": 0.0001998118140683641,
      "loss": 1.8354,
      "step": 1367
    },
    {
      "epoch": 0.04899099324941358,
      "grad_norm": 1.4478873014450073,
      "learning_rate": 0.00019981110214217798,
      "loss": 1.9215,
      "step": 1368
    },
    {
      "epoch": 0.04902680537898186,
      "grad_norm": 1.3279414176940918,
      "learning_rate": 0.00019981038887316221,
      "loss": 1.6335,
      "step": 1369
    },
    {
      "epoch": 0.04906261750855015,
      "grad_norm": 1.3693251609802246,
      "learning_rate": 0.00019980967426132642,
      "loss": 1.9246,
      "step": 1370
    },
    {
      "epoch": 0.04909842963811843,
      "grad_norm": 1.9764519929885864,
      "learning_rate": 0.0001998089583066802,
      "loss": 1.881,
      "step": 1371
    },
    {
      "epoch": 0.049134241767686715,
      "grad_norm": 2.154409646987915,
      "learning_rate": 0.00019980824100923318,
      "loss": 1.8377,
      "step": 1372
    },
    {
      "epoch": 0.049170053897255,
      "grad_norm": 1.876036524772644,
      "learning_rate": 0.00019980752236899502,
      "loss": 1.976,
      "step": 1373
    },
    {
      "epoch": 0.04920586602682329,
      "grad_norm": 1.0766030550003052,
      "learning_rate": 0.00019980680238597542,
      "loss": 1.7499,
      "step": 1374
    },
    {
      "epoch": 0.04924167815639157,
      "grad_norm": 1.5491044521331787,
      "learning_rate": 0.000199806081060184,
      "loss": 1.9504,
      "step": 1375
    },
    {
      "epoch": 0.04927749028595985,
      "grad_norm": 1.6977920532226562,
      "learning_rate": 0.00019980535839163053,
      "loss": 1.6755,
      "step": 1376
    },
    {
      "epoch": 0.049313302415528136,
      "grad_norm": 2.227975606918335,
      "learning_rate": 0.00019980463438032468,
      "loss": 2.0132,
      "step": 1377
    },
    {
      "epoch": 0.049349114545096426,
      "grad_norm": 1.4667011499404907,
      "learning_rate": 0.0001998039090262762,
      "loss": 1.8883,
      "step": 1378
    },
    {
      "epoch": 0.04938492667466471,
      "grad_norm": 1.5283125638961792,
      "learning_rate": 0.0001998031823294949,
      "loss": 1.8843,
      "step": 1379
    },
    {
      "epoch": 0.04942073880423299,
      "grad_norm": 1.5257043838500977,
      "learning_rate": 0.0001998024542899905,
      "loss": 1.8381,
      "step": 1380
    },
    {
      "epoch": 0.04945655093380128,
      "grad_norm": 1.9778072834014893,
      "learning_rate": 0.00019980172490777283,
      "loss": 1.9028,
      "step": 1381
    },
    {
      "epoch": 0.049492363063369564,
      "grad_norm": 1.3122044801712036,
      "learning_rate": 0.00019980099418285166,
      "loss": 1.6511,
      "step": 1382
    },
    {
      "epoch": 0.04952817519293785,
      "grad_norm": 1.8901945352554321,
      "learning_rate": 0.00019980026211523686,
      "loss": 1.6198,
      "step": 1383
    },
    {
      "epoch": 0.04956398732250613,
      "grad_norm": 1.79074227809906,
      "learning_rate": 0.00019979952870493824,
      "loss": 1.8086,
      "step": 1384
    },
    {
      "epoch": 0.04959979945207442,
      "grad_norm": 1.6266603469848633,
      "learning_rate": 0.00019979879395196575,
      "loss": 1.6447,
      "step": 1385
    },
    {
      "epoch": 0.0496356115816427,
      "grad_norm": 1.4488394260406494,
      "learning_rate": 0.00019979805785632916,
      "loss": 2.0551,
      "step": 1386
    },
    {
      "epoch": 0.049671423711210985,
      "grad_norm": 3.32766056060791,
      "learning_rate": 0.00019979732041803847,
      "loss": 2.1713,
      "step": 1387
    },
    {
      "epoch": 0.049707235840779275,
      "grad_norm": 1.405636191368103,
      "learning_rate": 0.00019979658163710355,
      "loss": 1.6463,
      "step": 1388
    },
    {
      "epoch": 0.04974304797034756,
      "grad_norm": 1.9788293838500977,
      "learning_rate": 0.00019979584151353437,
      "loss": 1.9736,
      "step": 1389
    },
    {
      "epoch": 0.04977886009991584,
      "grad_norm": 2.140531539916992,
      "learning_rate": 0.00019979510004734083,
      "loss": 1.9015,
      "step": 1390
    },
    {
      "epoch": 0.049814672229484123,
      "grad_norm": 2.0656092166900635,
      "learning_rate": 0.00019979435723853296,
      "loss": 1.8466,
      "step": 1391
    },
    {
      "epoch": 0.04985048435905241,
      "grad_norm": 1.4488940238952637,
      "learning_rate": 0.00019979361308712073,
      "loss": 1.7686,
      "step": 1392
    },
    {
      "epoch": 0.049886296488620696,
      "grad_norm": 2.0686659812927246,
      "learning_rate": 0.00019979286759311423,
      "loss": 1.8533,
      "step": 1393
    },
    {
      "epoch": 0.04992210861818898,
      "grad_norm": 2.168595314025879,
      "learning_rate": 0.00019979212075652334,
      "loss": 2.1408,
      "step": 1394
    },
    {
      "epoch": 0.04995792074775727,
      "grad_norm": 1.4077308177947998,
      "learning_rate": 0.00019979137257735823,
      "loss": 1.8534,
      "step": 1395
    },
    {
      "epoch": 0.04999373287732555,
      "grad_norm": 1.5167765617370605,
      "learning_rate": 0.0001997906230556289,
      "loss": 1.7809,
      "step": 1396
    },
    {
      "epoch": 0.050029545006893834,
      "grad_norm": 2.342071771621704,
      "learning_rate": 0.00019978987219134545,
      "loss": 1.8212,
      "step": 1397
    },
    {
      "epoch": 0.05006535713646212,
      "grad_norm": 1.417253851890564,
      "learning_rate": 0.000199789119984518,
      "loss": 2.1048,
      "step": 1398
    },
    {
      "epoch": 0.05010116926603041,
      "grad_norm": 1.8908305168151855,
      "learning_rate": 0.0001997883664351567,
      "loss": 1.951,
      "step": 1399
    },
    {
      "epoch": 0.05013698139559869,
      "grad_norm": 2.8274893760681152,
      "learning_rate": 0.00019978761154327158,
      "loss": 2.3188,
      "step": 1400
    },
    {
      "epoch": 0.05017279352516697,
      "grad_norm": 1.62851083278656,
      "learning_rate": 0.0001997868553088729,
      "loss": 1.9725,
      "step": 1401
    },
    {
      "epoch": 0.050208605654735255,
      "grad_norm": 2.5111052989959717,
      "learning_rate": 0.00019978609773197082,
      "loss": 2.2518,
      "step": 1402
    },
    {
      "epoch": 0.050244417784303545,
      "grad_norm": 1.185157299041748,
      "learning_rate": 0.00019978533881257547,
      "loss": 1.675,
      "step": 1403
    },
    {
      "epoch": 0.05028022991387183,
      "grad_norm": 1.112566590309143,
      "learning_rate": 0.0001997845785506971,
      "loss": 1.7595,
      "step": 1404
    },
    {
      "epoch": 0.05031604204344011,
      "grad_norm": 1.807395339012146,
      "learning_rate": 0.00019978381694634595,
      "loss": 1.9387,
      "step": 1405
    },
    {
      "epoch": 0.0503518541730084,
      "grad_norm": 1.576179027557373,
      "learning_rate": 0.00019978305399953228,
      "loss": 1.6791,
      "step": 1406
    },
    {
      "epoch": 0.050387666302576684,
      "grad_norm": 1.4238735437393188,
      "learning_rate": 0.0001997822897102663,
      "loss": 1.8154,
      "step": 1407
    },
    {
      "epoch": 0.050423478432144966,
      "grad_norm": 1.8287601470947266,
      "learning_rate": 0.00019978152407855833,
      "loss": 1.8966,
      "step": 1408
    },
    {
      "epoch": 0.05045929056171325,
      "grad_norm": 1.644075632095337,
      "learning_rate": 0.00019978075710441867,
      "loss": 1.8428,
      "step": 1409
    },
    {
      "epoch": 0.05049510269128154,
      "grad_norm": 1.835912823677063,
      "learning_rate": 0.0001997799887878576,
      "loss": 1.8366,
      "step": 1410
    },
    {
      "epoch": 0.05053091482084982,
      "grad_norm": 1.9325989484786987,
      "learning_rate": 0.0001997792191288855,
      "loss": 1.534,
      "step": 1411
    },
    {
      "epoch": 0.050566726950418105,
      "grad_norm": 1.7003639936447144,
      "learning_rate": 0.00019977844812751273,
      "loss": 1.8233,
      "step": 1412
    },
    {
      "epoch": 0.050602539079986394,
      "grad_norm": 1.4617531299591064,
      "learning_rate": 0.00019977767578374965,
      "loss": 1.8938,
      "step": 1413
    },
    {
      "epoch": 0.05063835120955468,
      "grad_norm": 1.3252395391464233,
      "learning_rate": 0.0001997769020976066,
      "loss": 1.9042,
      "step": 1414
    },
    {
      "epoch": 0.05067416333912296,
      "grad_norm": 1.881447672843933,
      "learning_rate": 0.0001997761270690941,
      "loss": 1.8318,
      "step": 1415
    },
    {
      "epoch": 0.05070997546869124,
      "grad_norm": 2.197857618331909,
      "learning_rate": 0.00019977535069822246,
      "loss": 1.7475,
      "step": 1416
    },
    {
      "epoch": 0.05074578759825953,
      "grad_norm": 1.581891655921936,
      "learning_rate": 0.0001997745729850022,
      "loss": 1.7239,
      "step": 1417
    },
    {
      "epoch": 0.050781599727827816,
      "grad_norm": 1.7551162242889404,
      "learning_rate": 0.00019977379392944377,
      "loss": 1.8272,
      "step": 1418
    },
    {
      "epoch": 0.0508174118573961,
      "grad_norm": 1.9000366926193237,
      "learning_rate": 0.00019977301353155764,
      "loss": 2.1745,
      "step": 1419
    },
    {
      "epoch": 0.05085322398696439,
      "grad_norm": 1.7322187423706055,
      "learning_rate": 0.00019977223179135428,
      "loss": 1.9419,
      "step": 1420
    },
    {
      "epoch": 0.05088903611653267,
      "grad_norm": 1.9449477195739746,
      "learning_rate": 0.0001997714487088443,
      "loss": 1.7893,
      "step": 1421
    },
    {
      "epoch": 0.050924848246100954,
      "grad_norm": 1.3871630430221558,
      "learning_rate": 0.0001997706642840381,
      "loss": 1.9459,
      "step": 1422
    },
    {
      "epoch": 0.05096066037566924,
      "grad_norm": 1.2880792617797852,
      "learning_rate": 0.00019976987851694634,
      "loss": 1.7042,
      "step": 1423
    },
    {
      "epoch": 0.050996472505237526,
      "grad_norm": 2.2730605602264404,
      "learning_rate": 0.00019976909140757956,
      "loss": 1.744,
      "step": 1424
    },
    {
      "epoch": 0.05103228463480581,
      "grad_norm": 2.3292646408081055,
      "learning_rate": 0.00019976830295594832,
      "loss": 2.1704,
      "step": 1425
    },
    {
      "epoch": 0.05106809676437409,
      "grad_norm": 2.2148680686950684,
      "learning_rate": 0.0001997675131620633,
      "loss": 1.7412,
      "step": 1426
    },
    {
      "epoch": 0.051103908893942375,
      "grad_norm": 2.116534471511841,
      "learning_rate": 0.00019976672202593506,
      "loss": 1.7296,
      "step": 1427
    },
    {
      "epoch": 0.051139721023510665,
      "grad_norm": 1.278947114944458,
      "learning_rate": 0.00019976592954757427,
      "loss": 1.4632,
      "step": 1428
    },
    {
      "epoch": 0.05117553315307895,
      "grad_norm": 1.4900505542755127,
      "learning_rate": 0.00019976513572699157,
      "loss": 1.8935,
      "step": 1429
    },
    {
      "epoch": 0.05121134528264723,
      "grad_norm": 1.7054023742675781,
      "learning_rate": 0.00019976434056419767,
      "loss": 1.681,
      "step": 1430
    },
    {
      "epoch": 0.05124715741221552,
      "grad_norm": 2.0350306034088135,
      "learning_rate": 0.00019976354405920328,
      "loss": 1.8399,
      "step": 1431
    },
    {
      "epoch": 0.0512829695417838,
      "grad_norm": 1.9636918306350708,
      "learning_rate": 0.00019976274621201907,
      "loss": 2.0888,
      "step": 1432
    },
    {
      "epoch": 0.051318781671352086,
      "grad_norm": 1.9971915483474731,
      "learning_rate": 0.00019976194702265578,
      "loss": 1.7744,
      "step": 1433
    },
    {
      "epoch": 0.05135459380092037,
      "grad_norm": 2.1631476879119873,
      "learning_rate": 0.00019976114649112418,
      "loss": 2.0092,
      "step": 1434
    },
    {
      "epoch": 0.05139040593048866,
      "grad_norm": 3.263909339904785,
      "learning_rate": 0.00019976034461743504,
      "loss": 2.3351,
      "step": 1435
    },
    {
      "epoch": 0.05142621806005694,
      "grad_norm": 1.6741386651992798,
      "learning_rate": 0.00019975954140159915,
      "loss": 1.7882,
      "step": 1436
    },
    {
      "epoch": 0.051462030189625224,
      "grad_norm": 1.8211408853530884,
      "learning_rate": 0.0001997587368436273,
      "loss": 1.6995,
      "step": 1437
    },
    {
      "epoch": 0.051497842319193514,
      "grad_norm": 1.880556344985962,
      "learning_rate": 0.00019975793094353036,
      "loss": 2.0151,
      "step": 1438
    },
    {
      "epoch": 0.0515336544487618,
      "grad_norm": 1.701261281967163,
      "learning_rate": 0.0001997571237013191,
      "loss": 1.6038,
      "step": 1439
    },
    {
      "epoch": 0.05156946657833008,
      "grad_norm": 1.4975484609603882,
      "learning_rate": 0.00019975631511700442,
      "loss": 1.5615,
      "step": 1440
    },
    {
      "epoch": 0.05160527870789836,
      "grad_norm": 1.6058564186096191,
      "learning_rate": 0.00019975550519059723,
      "loss": 2.0746,
      "step": 1441
    },
    {
      "epoch": 0.05164109083746665,
      "grad_norm": 1.5407977104187012,
      "learning_rate": 0.00019975469392210834,
      "loss": 1.893,
      "step": 1442
    },
    {
      "epoch": 0.051676902967034935,
      "grad_norm": 2.4655606746673584,
      "learning_rate": 0.00019975388131154875,
      "loss": 2.4228,
      "step": 1443
    },
    {
      "epoch": 0.05171271509660322,
      "grad_norm": 2.1224660873413086,
      "learning_rate": 0.00019975306735892936,
      "loss": 1.84,
      "step": 1444
    },
    {
      "epoch": 0.05174852722617151,
      "grad_norm": 1.920393943786621,
      "learning_rate": 0.00019975225206426113,
      "loss": 1.8256,
      "step": 1445
    },
    {
      "epoch": 0.05178433935573979,
      "grad_norm": 1.4037801027297974,
      "learning_rate": 0.000199751435427555,
      "loss": 1.6021,
      "step": 1446
    },
    {
      "epoch": 0.05182015148530807,
      "grad_norm": 2.118389844894409,
      "learning_rate": 0.000199750617448822,
      "loss": 1.7848,
      "step": 1447
    },
    {
      "epoch": 0.051855963614876356,
      "grad_norm": 1.7762823104858398,
      "learning_rate": 0.0001997497981280731,
      "loss": 1.8298,
      "step": 1448
    },
    {
      "epoch": 0.051891775744444646,
      "grad_norm": 2.352095127105713,
      "learning_rate": 0.0001997489774653193,
      "loss": 1.9828,
      "step": 1449
    },
    {
      "epoch": 0.05192758787401293,
      "grad_norm": 3.4863343238830566,
      "learning_rate": 0.00019974815546057172,
      "loss": 2.2459,
      "step": 1450
    },
    {
      "epoch": 0.05196340000358121,
      "grad_norm": 2.1996467113494873,
      "learning_rate": 0.00019974733211384135,
      "loss": 1.7053,
      "step": 1451
    },
    {
      "epoch": 0.051999212133149494,
      "grad_norm": 1.548524022102356,
      "learning_rate": 0.0001997465074251393,
      "loss": 1.6912,
      "step": 1452
    },
    {
      "epoch": 0.052035024262717784,
      "grad_norm": 1.4116077423095703,
      "learning_rate": 0.00019974568139447666,
      "loss": 1.6227,
      "step": 1453
    },
    {
      "epoch": 0.05207083639228607,
      "grad_norm": 1.6894891262054443,
      "learning_rate": 0.00019974485402186453,
      "loss": 1.599,
      "step": 1454
    },
    {
      "epoch": 0.05210664852185435,
      "grad_norm": 2.850682497024536,
      "learning_rate": 0.00019974402530731407,
      "loss": 1.929,
      "step": 1455
    },
    {
      "epoch": 0.05214246065142264,
      "grad_norm": 2.964691400527954,
      "learning_rate": 0.0001997431952508364,
      "loss": 1.4999,
      "step": 1456
    },
    {
      "epoch": 0.05217827278099092,
      "grad_norm": 2.3899643421173096,
      "learning_rate": 0.00019974236385244268,
      "loss": 1.8135,
      "step": 1457
    },
    {
      "epoch": 0.052214084910559205,
      "grad_norm": 3.404496431350708,
      "learning_rate": 0.00019974153111214414,
      "loss": 2.2774,
      "step": 1458
    },
    {
      "epoch": 0.05224989704012749,
      "grad_norm": 1.3085155487060547,
      "learning_rate": 0.00019974069702995194,
      "loss": 1.7506,
      "step": 1459
    },
    {
      "epoch": 0.05228570916969578,
      "grad_norm": 1.9195306301116943,
      "learning_rate": 0.00019973986160587732,
      "loss": 2.0551,
      "step": 1460
    },
    {
      "epoch": 0.05232152129926406,
      "grad_norm": 1.2769794464111328,
      "learning_rate": 0.0001997390248399315,
      "loss": 1.6659,
      "step": 1461
    },
    {
      "epoch": 0.052357333428832344,
      "grad_norm": 1.5254219770431519,
      "learning_rate": 0.00019973818673212578,
      "loss": 1.3987,
      "step": 1462
    },
    {
      "epoch": 0.05239314555840063,
      "grad_norm": 1.952865481376648,
      "learning_rate": 0.00019973734728247143,
      "loss": 1.6659,
      "step": 1463
    },
    {
      "epoch": 0.052428957687968916,
      "grad_norm": 1.9222220182418823,
      "learning_rate": 0.0001997365064909797,
      "loss": 1.6806,
      "step": 1464
    },
    {
      "epoch": 0.0524647698175372,
      "grad_norm": 1.1028681993484497,
      "learning_rate": 0.0001997356643576619,
      "loss": 1.8794,
      "step": 1465
    },
    {
      "epoch": 0.05250058194710548,
      "grad_norm": 1.5954684019088745,
      "learning_rate": 0.00019973482088252943,
      "loss": 1.6585,
      "step": 1466
    },
    {
      "epoch": 0.05253639407667377,
      "grad_norm": 2.3468151092529297,
      "learning_rate": 0.00019973397606559354,
      "loss": 1.9015,
      "step": 1467
    },
    {
      "epoch": 0.052572206206242054,
      "grad_norm": 1.6654571294784546,
      "learning_rate": 0.0001997331299068657,
      "loss": 1.9589,
      "step": 1468
    },
    {
      "epoch": 0.05260801833581034,
      "grad_norm": 1.9264411926269531,
      "learning_rate": 0.00019973228240635722,
      "loss": 1.9634,
      "step": 1469
    },
    {
      "epoch": 0.05264383046537863,
      "grad_norm": 1.5382975339889526,
      "learning_rate": 0.00019973143356407952,
      "loss": 1.7071,
      "step": 1470
    },
    {
      "epoch": 0.05267964259494691,
      "grad_norm": 1.8750600814819336,
      "learning_rate": 0.00019973058338004407,
      "loss": 1.8874,
      "step": 1471
    },
    {
      "epoch": 0.05271545472451519,
      "grad_norm": 2.4676332473754883,
      "learning_rate": 0.00019972973185426222,
      "loss": 2.0747,
      "step": 1472
    },
    {
      "epoch": 0.052751266854083476,
      "grad_norm": 1.5208827257156372,
      "learning_rate": 0.0001997288789867455,
      "loss": 1.8408,
      "step": 1473
    },
    {
      "epoch": 0.052787078983651765,
      "grad_norm": 2.3117616176605225,
      "learning_rate": 0.0001997280247775053,
      "loss": 1.6432,
      "step": 1474
    },
    {
      "epoch": 0.05282289111322005,
      "grad_norm": 1.715316653251648,
      "learning_rate": 0.0001997271692265532,
      "loss": 1.494,
      "step": 1475
    },
    {
      "epoch": 0.05285870324278833,
      "grad_norm": 1.7642710208892822,
      "learning_rate": 0.0001997263123339007,
      "loss": 1.6121,
      "step": 1476
    },
    {
      "epoch": 0.052894515372356614,
      "grad_norm": 1.6220943927764893,
      "learning_rate": 0.00019972545409955927,
      "loss": 1.4739,
      "step": 1477
    },
    {
      "epoch": 0.052930327501924904,
      "grad_norm": 1.6621453762054443,
      "learning_rate": 0.0001997245945235405,
      "loss": 2.0465,
      "step": 1478
    },
    {
      "epoch": 0.052966139631493186,
      "grad_norm": 1.6994413137435913,
      "learning_rate": 0.00019972373360585598,
      "loss": 1.8201,
      "step": 1479
    },
    {
      "epoch": 0.05300195176106147,
      "grad_norm": 1.6507011651992798,
      "learning_rate": 0.0001997228713465172,
      "loss": 1.7818,
      "step": 1480
    },
    {
      "epoch": 0.05303776389062976,
      "grad_norm": 2.683297872543335,
      "learning_rate": 0.00019972200774553587,
      "loss": 2.1151,
      "step": 1481
    },
    {
      "epoch": 0.05307357602019804,
      "grad_norm": 1.821467399597168,
      "learning_rate": 0.00019972114280292355,
      "loss": 1.759,
      "step": 1482
    },
    {
      "epoch": 0.053109388149766325,
      "grad_norm": 1.6540615558624268,
      "learning_rate": 0.00019972027651869186,
      "loss": 1.6425,
      "step": 1483
    },
    {
      "epoch": 0.05314520027933461,
      "grad_norm": 1.185801386833191,
      "learning_rate": 0.0001997194088928525,
      "loss": 1.598,
      "step": 1484
    },
    {
      "epoch": 0.0531810124089029,
      "grad_norm": 1.5758860111236572,
      "learning_rate": 0.0001997185399254171,
      "loss": 1.8597,
      "step": 1485
    },
    {
      "epoch": 0.05321682453847118,
      "grad_norm": 1.5715807676315308,
      "learning_rate": 0.00019971766961639738,
      "loss": 1.6007,
      "step": 1486
    },
    {
      "epoch": 0.05325263666803946,
      "grad_norm": 2.0418407917022705,
      "learning_rate": 0.00019971679796580504,
      "loss": 2.1591,
      "step": 1487
    },
    {
      "epoch": 0.05328844879760775,
      "grad_norm": 1.2468957901000977,
      "learning_rate": 0.00019971592497365184,
      "loss": 1.6722,
      "step": 1488
    },
    {
      "epoch": 0.053324260927176036,
      "grad_norm": 1.5520515441894531,
      "learning_rate": 0.00019971505063994948,
      "loss": 1.7877,
      "step": 1489
    },
    {
      "epoch": 0.05336007305674432,
      "grad_norm": 1.990512728691101,
      "learning_rate": 0.0001997141749647097,
      "loss": 1.8976,
      "step": 1490
    },
    {
      "epoch": 0.0533958851863126,
      "grad_norm": 1.7781422138214111,
      "learning_rate": 0.00019971329794794436,
      "loss": 2.0116,
      "step": 1491
    },
    {
      "epoch": 0.05343169731588089,
      "grad_norm": 1.6702563762664795,
      "learning_rate": 0.0001997124195896652,
      "loss": 1.7436,
      "step": 1492
    },
    {
      "epoch": 0.053467509445449174,
      "grad_norm": 2.4951159954071045,
      "learning_rate": 0.00019971153988988406,
      "loss": 1.7853,
      "step": 1493
    },
    {
      "epoch": 0.05350332157501746,
      "grad_norm": 1.4511957168579102,
      "learning_rate": 0.00019971065884861276,
      "loss": 1.6299,
      "step": 1494
    },
    {
      "epoch": 0.05353913370458575,
      "grad_norm": 1.593092679977417,
      "learning_rate": 0.00019970977646586319,
      "loss": 2.1281,
      "step": 1495
    },
    {
      "epoch": 0.05357494583415403,
      "grad_norm": 2.2723426818847656,
      "learning_rate": 0.00019970889274164715,
      "loss": 1.4845,
      "step": 1496
    },
    {
      "epoch": 0.05361075796372231,
      "grad_norm": 2.1010289192199707,
      "learning_rate": 0.00019970800767597663,
      "loss": 1.995,
      "step": 1497
    },
    {
      "epoch": 0.053646570093290595,
      "grad_norm": 1.3042644262313843,
      "learning_rate": 0.00019970712126886342,
      "loss": 2.1141,
      "step": 1498
    },
    {
      "epoch": 0.053682382222858885,
      "grad_norm": 1.9947757720947266,
      "learning_rate": 0.00019970623352031952,
      "loss": 1.8331,
      "step": 1499
    },
    {
      "epoch": 0.05371819435242717,
      "grad_norm": 2.3290586471557617,
      "learning_rate": 0.00019970534443035688,
      "loss": 1.6654,
      "step": 1500
    },
    {
      "epoch": 0.05375400648199545,
      "grad_norm": 3.0495433807373047,
      "learning_rate": 0.00019970445399898745,
      "loss": 1.8413,
      "step": 1501
    },
    {
      "epoch": 0.05378981861156373,
      "grad_norm": 2.522512435913086,
      "learning_rate": 0.0001997035622262232,
      "loss": 1.8241,
      "step": 1502
    },
    {
      "epoch": 0.05382563074113202,
      "grad_norm": 2.1961100101470947,
      "learning_rate": 0.00019970266911207608,
      "loss": 1.6946,
      "step": 1503
    },
    {
      "epoch": 0.053861442870700306,
      "grad_norm": 1.8775655031204224,
      "learning_rate": 0.00019970177465655818,
      "loss": 1.8042,
      "step": 1504
    },
    {
      "epoch": 0.05389725500026859,
      "grad_norm": 3.8398947715759277,
      "learning_rate": 0.00019970087885968154,
      "loss": 2.0227,
      "step": 1505
    },
    {
      "epoch": 0.05393306712983688,
      "grad_norm": 1.3387746810913086,
      "learning_rate": 0.00019969998172145815,
      "loss": 1.6747,
      "step": 1506
    },
    {
      "epoch": 0.05396887925940516,
      "grad_norm": 1.517892599105835,
      "learning_rate": 0.00019969908324190012,
      "loss": 1.8804,
      "step": 1507
    },
    {
      "epoch": 0.054004691388973444,
      "grad_norm": 1.1590577363967896,
      "learning_rate": 0.0001996981834210195,
      "loss": 1.6391,
      "step": 1508
    },
    {
      "epoch": 0.05404050351854173,
      "grad_norm": 2.3728671073913574,
      "learning_rate": 0.00019969728225882846,
      "loss": 1.6555,
      "step": 1509
    },
    {
      "epoch": 0.05407631564811002,
      "grad_norm": 1.6440011262893677,
      "learning_rate": 0.0001996963797553391,
      "loss": 1.7857,
      "step": 1510
    },
    {
      "epoch": 0.0541121277776783,
      "grad_norm": 1.4567910432815552,
      "learning_rate": 0.0001996954759105635,
      "loss": 1.9391,
      "step": 1511
    },
    {
      "epoch": 0.05414793990724658,
      "grad_norm": 1.946500539779663,
      "learning_rate": 0.00019969457072451392,
      "loss": 1.7353,
      "step": 1512
    },
    {
      "epoch": 0.05418375203681487,
      "grad_norm": 2.7675740718841553,
      "learning_rate": 0.00019969366419720245,
      "loss": 2.133,
      "step": 1513
    },
    {
      "epoch": 0.054219564166383155,
      "grad_norm": 2.4415030479431152,
      "learning_rate": 0.00019969275632864133,
      "loss": 1.9024,
      "step": 1514
    },
    {
      "epoch": 0.05425537629595144,
      "grad_norm": 1.4757903814315796,
      "learning_rate": 0.0001996918471188428,
      "loss": 1.7258,
      "step": 1515
    },
    {
      "epoch": 0.05429118842551972,
      "grad_norm": 1.7123427391052246,
      "learning_rate": 0.00019969093656781902,
      "loss": 1.7771,
      "step": 1516
    },
    {
      "epoch": 0.05432700055508801,
      "grad_norm": 1.3325016498565674,
      "learning_rate": 0.00019969002467558228,
      "loss": 1.7978,
      "step": 1517
    },
    {
      "epoch": 0.05436281268465629,
      "grad_norm": 1.9106487035751343,
      "learning_rate": 0.00019968911144214486,
      "loss": 1.8826,
      "step": 1518
    },
    {
      "epoch": 0.054398624814224576,
      "grad_norm": 2.213944435119629,
      "learning_rate": 0.00019968819686751906,
      "loss": 1.9959,
      "step": 1519
    },
    {
      "epoch": 0.054434436943792866,
      "grad_norm": 2.123725175857544,
      "learning_rate": 0.00019968728095171715,
      "loss": 2.1413,
      "step": 1520
    },
    {
      "epoch": 0.05447024907336115,
      "grad_norm": 1.2367082834243774,
      "learning_rate": 0.00019968636369475142,
      "loss": 1.6893,
      "step": 1521
    },
    {
      "epoch": 0.05450606120292943,
      "grad_norm": 1.6155388355255127,
      "learning_rate": 0.00019968544509663428,
      "loss": 1.9238,
      "step": 1522
    },
    {
      "epoch": 0.054541873332497715,
      "grad_norm": 2.1532177925109863,
      "learning_rate": 0.00019968452515737805,
      "loss": 1.8312,
      "step": 1523
    },
    {
      "epoch": 0.054577685462066004,
      "grad_norm": 2.00732421875,
      "learning_rate": 0.00019968360387699513,
      "loss": 1.8616,
      "step": 1524
    },
    {
      "epoch": 0.05461349759163429,
      "grad_norm": 2.6617205142974854,
      "learning_rate": 0.00019968268125549794,
      "loss": 1.9007,
      "step": 1525
    },
    {
      "epoch": 0.05464930972120257,
      "grad_norm": 1.5837981700897217,
      "learning_rate": 0.0001996817572928988,
      "loss": 1.8432,
      "step": 1526
    },
    {
      "epoch": 0.05468512185077085,
      "grad_norm": 1.387168526649475,
      "learning_rate": 0.0001996808319892102,
      "loss": 1.7933,
      "step": 1527
    },
    {
      "epoch": 0.05472093398033914,
      "grad_norm": 1.664420247077942,
      "learning_rate": 0.00019967990534444462,
      "loss": 1.8014,
      "step": 1528
    },
    {
      "epoch": 0.054756746109907425,
      "grad_norm": 1.184078574180603,
      "learning_rate": 0.00019967897735861446,
      "loss": 1.451,
      "step": 1529
    },
    {
      "epoch": 0.05479255823947571,
      "grad_norm": 2.2369894981384277,
      "learning_rate": 0.00019967804803173227,
      "loss": 1.5848,
      "step": 1530
    },
    {
      "epoch": 0.054828370369044,
      "grad_norm": 1.4040497541427612,
      "learning_rate": 0.00019967711736381048,
      "loss": 1.8061,
      "step": 1531
    },
    {
      "epoch": 0.05486418249861228,
      "grad_norm": 2.491568088531494,
      "learning_rate": 0.00019967618535486164,
      "loss": 1.9725,
      "step": 1532
    },
    {
      "epoch": 0.054899994628180564,
      "grad_norm": 1.9477348327636719,
      "learning_rate": 0.00019967525200489833,
      "loss": 1.6782,
      "step": 1533
    },
    {
      "epoch": 0.054935806757748847,
      "grad_norm": 1.5153226852416992,
      "learning_rate": 0.00019967431731393308,
      "loss": 1.8764,
      "step": 1534
    },
    {
      "epoch": 0.054971618887317136,
      "grad_norm": 2.029557704925537,
      "learning_rate": 0.00019967338128197847,
      "loss": 1.8136,
      "step": 1535
    },
    {
      "epoch": 0.05500743101688542,
      "grad_norm": 2.0749216079711914,
      "learning_rate": 0.00019967244390904708,
      "loss": 1.9191,
      "step": 1536
    },
    {
      "epoch": 0.0550432431464537,
      "grad_norm": 1.8724786043167114,
      "learning_rate": 0.0001996715051951515,
      "loss": 1.7789,
      "step": 1537
    },
    {
      "epoch": 0.05507905527602199,
      "grad_norm": 1.8553539514541626,
      "learning_rate": 0.0001996705651403044,
      "loss": 1.9742,
      "step": 1538
    },
    {
      "epoch": 0.055114867405590275,
      "grad_norm": 1.4529833793640137,
      "learning_rate": 0.0001996696237445184,
      "loss": 1.5149,
      "step": 1539
    },
    {
      "epoch": 0.05515067953515856,
      "grad_norm": 1.6312896013259888,
      "learning_rate": 0.0001996686810078062,
      "loss": 1.8811,
      "step": 1540
    },
    {
      "epoch": 0.05518649166472684,
      "grad_norm": 2.4548962116241455,
      "learning_rate": 0.00019966773693018045,
      "loss": 1.9498,
      "step": 1541
    },
    {
      "epoch": 0.05522230379429513,
      "grad_norm": 1.8022762537002563,
      "learning_rate": 0.00019966679151165384,
      "loss": 1.9692,
      "step": 1542
    },
    {
      "epoch": 0.05525811592386341,
      "grad_norm": 2.094322919845581,
      "learning_rate": 0.00019966584475223913,
      "loss": 1.7657,
      "step": 1543
    },
    {
      "epoch": 0.055293928053431696,
      "grad_norm": 1.8352880477905273,
      "learning_rate": 0.00019966489665194904,
      "loss": 1.7019,
      "step": 1544
    },
    {
      "epoch": 0.055329740182999985,
      "grad_norm": 3.184751033782959,
      "learning_rate": 0.0001996639472107963,
      "loss": 2.2141,
      "step": 1545
    },
    {
      "epoch": 0.05536555231256827,
      "grad_norm": 1.6981626749038696,
      "learning_rate": 0.00019966299642879375,
      "loss": 1.7203,
      "step": 1546
    },
    {
      "epoch": 0.05540136444213655,
      "grad_norm": 1.5179368257522583,
      "learning_rate": 0.00019966204430595412,
      "loss": 1.907,
      "step": 1547
    },
    {
      "epoch": 0.055437176571704834,
      "grad_norm": 3.7298059463500977,
      "learning_rate": 0.00019966109084229024,
      "loss": 1.9523,
      "step": 1548
    },
    {
      "epoch": 0.055472988701273124,
      "grad_norm": 2.229259729385376,
      "learning_rate": 0.00019966013603781493,
      "loss": 1.9006,
      "step": 1549
    },
    {
      "epoch": 0.05550880083084141,
      "grad_norm": 1.454759120941162,
      "learning_rate": 0.00019965917989254103,
      "loss": 1.7634,
      "step": 1550
    },
    {
      "epoch": 0.05554461296040969,
      "grad_norm": 1.9848576784133911,
      "learning_rate": 0.00019965822240648143,
      "loss": 2.0198,
      "step": 1551
    },
    {
      "epoch": 0.05558042508997797,
      "grad_norm": 1.571907877922058,
      "learning_rate": 0.00019965726357964902,
      "loss": 1.6818,
      "step": 1552
    },
    {
      "epoch": 0.05561623721954626,
      "grad_norm": 2.924750328063965,
      "learning_rate": 0.00019965630341205664,
      "loss": 1.8103,
      "step": 1553
    },
    {
      "epoch": 0.055652049349114545,
      "grad_norm": 1.6184571981430054,
      "learning_rate": 0.00019965534190371725,
      "loss": 1.8719,
      "step": 1554
    },
    {
      "epoch": 0.05568786147868283,
      "grad_norm": 1.6031067371368408,
      "learning_rate": 0.0001996543790546438,
      "loss": 2.0676,
      "step": 1555
    },
    {
      "epoch": 0.05572367360825112,
      "grad_norm": 2.796915054321289,
      "learning_rate": 0.00019965341486484923,
      "loss": 1.7383,
      "step": 1556
    },
    {
      "epoch": 0.0557594857378194,
      "grad_norm": 3.0277481079101562,
      "learning_rate": 0.00019965244933434648,
      "loss": 1.8378,
      "step": 1557
    },
    {
      "epoch": 0.05579529786738768,
      "grad_norm": 1.5920354127883911,
      "learning_rate": 0.00019965148246314858,
      "loss": 1.9057,
      "step": 1558
    },
    {
      "epoch": 0.055831109996955966,
      "grad_norm": 2.056974172592163,
      "learning_rate": 0.00019965051425126852,
      "loss": 1.6761,
      "step": 1559
    },
    {
      "epoch": 0.055866922126524256,
      "grad_norm": 1.310811161994934,
      "learning_rate": 0.00019964954469871936,
      "loss": 1.5442,
      "step": 1560
    },
    {
      "epoch": 0.05590273425609254,
      "grad_norm": 2.0601959228515625,
      "learning_rate": 0.0001996485738055141,
      "loss": 1.8756,
      "step": 1561
    },
    {
      "epoch": 0.05593854638566082,
      "grad_norm": 1.5039955377578735,
      "learning_rate": 0.00019964760157166578,
      "loss": 1.9241,
      "step": 1562
    },
    {
      "epoch": 0.05597435851522911,
      "grad_norm": 2.9783098697662354,
      "learning_rate": 0.00019964662799718753,
      "loss": 1.995,
      "step": 1563
    },
    {
      "epoch": 0.056010170644797394,
      "grad_norm": 2.375992774963379,
      "learning_rate": 0.00019964565308209248,
      "loss": 2.1893,
      "step": 1564
    },
    {
      "epoch": 0.05604598277436568,
      "grad_norm": 1.5130183696746826,
      "learning_rate": 0.00019964467682639364,
      "loss": 1.5716,
      "step": 1565
    },
    {
      "epoch": 0.05608179490393396,
      "grad_norm": 1.9226343631744385,
      "learning_rate": 0.00019964369923010424,
      "loss": 1.9059,
      "step": 1566
    },
    {
      "epoch": 0.05611760703350225,
      "grad_norm": 3.1932523250579834,
      "learning_rate": 0.00019964272029323742,
      "loss": 2.0947,
      "step": 1567
    },
    {
      "epoch": 0.05615341916307053,
      "grad_norm": 1.9467480182647705,
      "learning_rate": 0.00019964174001580628,
      "loss": 2.0824,
      "step": 1568
    },
    {
      "epoch": 0.056189231292638815,
      "grad_norm": 1.4615256786346436,
      "learning_rate": 0.00019964075839782407,
      "loss": 1.7063,
      "step": 1569
    },
    {
      "epoch": 0.056225043422207105,
      "grad_norm": 2.083310604095459,
      "learning_rate": 0.000199639775439304,
      "loss": 2.1004,
      "step": 1570
    },
    {
      "epoch": 0.05626085555177539,
      "grad_norm": 2.23530650138855,
      "learning_rate": 0.00019963879114025926,
      "loss": 1.7032,
      "step": 1571
    },
    {
      "epoch": 0.05629666768134367,
      "grad_norm": 1.425076961517334,
      "learning_rate": 0.00019963780550070315,
      "loss": 1.7797,
      "step": 1572
    },
    {
      "epoch": 0.05633247981091195,
      "grad_norm": 2.6845715045928955,
      "learning_rate": 0.00019963681852064883,
      "loss": 1.8535,
      "step": 1573
    },
    {
      "epoch": 0.05636829194048024,
      "grad_norm": 2.078648567199707,
      "learning_rate": 0.0001996358302001097,
      "loss": 1.6292,
      "step": 1574
    },
    {
      "epoch": 0.056404104070048526,
      "grad_norm": 2.747649908065796,
      "learning_rate": 0.00019963484053909896,
      "loss": 1.877,
      "step": 1575
    },
    {
      "epoch": 0.05643991619961681,
      "grad_norm": 2.103194236755371,
      "learning_rate": 0.00019963384953762995,
      "loss": 1.8571,
      "step": 1576
    },
    {
      "epoch": 0.05647572832918509,
      "grad_norm": 1.530626654624939,
      "learning_rate": 0.00019963285719571604,
      "loss": 1.5274,
      "step": 1577
    },
    {
      "epoch": 0.05651154045875338,
      "grad_norm": 1.7835010290145874,
      "learning_rate": 0.00019963186351337054,
      "loss": 1.678,
      "step": 1578
    },
    {
      "epoch": 0.056547352588321664,
      "grad_norm": 1.6494067907333374,
      "learning_rate": 0.00019963086849060684,
      "loss": 1.6927,
      "step": 1579
    },
    {
      "epoch": 0.05658316471788995,
      "grad_norm": 2.0887563228607178,
      "learning_rate": 0.0001996298721274383,
      "loss": 1.8169,
      "step": 1580
    },
    {
      "epoch": 0.05661897684745824,
      "grad_norm": 1.8033913373947144,
      "learning_rate": 0.00019962887442387834,
      "loss": 1.7722,
      "step": 1581
    },
    {
      "epoch": 0.05665478897702652,
      "grad_norm": 2.5214874744415283,
      "learning_rate": 0.0001996278753799404,
      "loss": 1.7613,
      "step": 1582
    },
    {
      "epoch": 0.0566906011065948,
      "grad_norm": 1.961786150932312,
      "learning_rate": 0.00019962687499563793,
      "loss": 1.8367,
      "step": 1583
    },
    {
      "epoch": 0.056726413236163085,
      "grad_norm": 2.0807454586029053,
      "learning_rate": 0.00019962587327098435,
      "loss": 1.7686,
      "step": 1584
    },
    {
      "epoch": 0.056762225365731375,
      "grad_norm": 1.7898211479187012,
      "learning_rate": 0.00019962487020599315,
      "loss": 1.2548,
      "step": 1585
    },
    {
      "epoch": 0.05679803749529966,
      "grad_norm": 1.7042467594146729,
      "learning_rate": 0.00019962386580067782,
      "loss": 1.6725,
      "step": 1586
    },
    {
      "epoch": 0.05683384962486794,
      "grad_norm": 1.3166635036468506,
      "learning_rate": 0.00019962286005505188,
      "loss": 1.9565,
      "step": 1587
    },
    {
      "epoch": 0.05686966175443623,
      "grad_norm": 1.8134676218032837,
      "learning_rate": 0.00019962185296912887,
      "loss": 1.7871,
      "step": 1588
    },
    {
      "epoch": 0.056905473884004514,
      "grad_norm": 1.3248748779296875,
      "learning_rate": 0.00019962084454292235,
      "loss": 1.7441,
      "step": 1589
    },
    {
      "epoch": 0.056941286013572796,
      "grad_norm": 2.7689707279205322,
      "learning_rate": 0.00019961983477644583,
      "loss": 1.7567,
      "step": 1590
    },
    {
      "epoch": 0.05697709814314108,
      "grad_norm": 1.4286848306655884,
      "learning_rate": 0.00019961882366971296,
      "loss": 1.5907,
      "step": 1591
    },
    {
      "epoch": 0.05701291027270937,
      "grad_norm": 1.531070590019226,
      "learning_rate": 0.00019961781122273734,
      "loss": 1.8205,
      "step": 1592
    },
    {
      "epoch": 0.05704872240227765,
      "grad_norm": 3.7116222381591797,
      "learning_rate": 0.00019961679743553252,
      "loss": 1.7679,
      "step": 1593
    },
    {
      "epoch": 0.057084534531845935,
      "grad_norm": 1.6121838092803955,
      "learning_rate": 0.0001996157823081122,
      "loss": 1.728,
      "step": 1594
    },
    {
      "epoch": 0.057120346661414224,
      "grad_norm": 1.6923259496688843,
      "learning_rate": 0.00019961476584049004,
      "loss": 1.5902,
      "step": 1595
    },
    {
      "epoch": 0.05715615879098251,
      "grad_norm": 1.7286845445632935,
      "learning_rate": 0.00019961374803267968,
      "loss": 1.8698,
      "step": 1596
    },
    {
      "epoch": 0.05719197092055079,
      "grad_norm": 1.6534048318862915,
      "learning_rate": 0.00019961272888469484,
      "loss": 1.8303,
      "step": 1597
    },
    {
      "epoch": 0.05722778305011907,
      "grad_norm": 2.573026180267334,
      "learning_rate": 0.00019961170839654922,
      "loss": 1.8427,
      "step": 1598
    },
    {
      "epoch": 0.05726359517968736,
      "grad_norm": 1.91075599193573,
      "learning_rate": 0.00019961068656825656,
      "loss": 1.7374,
      "step": 1599
    },
    {
      "epoch": 0.057299407309255646,
      "grad_norm": 1.761756181716919,
      "learning_rate": 0.0001996096633998306,
      "loss": 1.9732,
      "step": 1600
    },
    {
      "epoch": 0.05733521943882393,
      "grad_norm": 1.8080767393112183,
      "learning_rate": 0.0001996086388912851,
      "loss": 1.5933,
      "step": 1601
    },
    {
      "epoch": 0.05737103156839221,
      "grad_norm": 2.0051701068878174,
      "learning_rate": 0.00019960761304263386,
      "loss": 1.8956,
      "step": 1602
    },
    {
      "epoch": 0.0574068436979605,
      "grad_norm": 2.3873651027679443,
      "learning_rate": 0.0001996065858538907,
      "loss": 1.7061,
      "step": 1603
    },
    {
      "epoch": 0.057442655827528784,
      "grad_norm": 1.7301465272903442,
      "learning_rate": 0.00019960555732506937,
      "loss": 1.5289,
      "step": 1604
    },
    {
      "epoch": 0.05747846795709707,
      "grad_norm": 1.9653252363204956,
      "learning_rate": 0.00019960452745618375,
      "loss": 1.6136,
      "step": 1605
    },
    {
      "epoch": 0.057514280086665356,
      "grad_norm": 1.3157404661178589,
      "learning_rate": 0.0001996034962472477,
      "loss": 1.833,
      "step": 1606
    },
    {
      "epoch": 0.05755009221623364,
      "grad_norm": 1.559603214263916,
      "learning_rate": 0.0001996024636982751,
      "loss": 1.6462,
      "step": 1607
    },
    {
      "epoch": 0.05758590434580192,
      "grad_norm": 1.8252371549606323,
      "learning_rate": 0.0001996014298092798,
      "loss": 1.6637,
      "step": 1608
    },
    {
      "epoch": 0.057621716475370205,
      "grad_norm": 1.6523933410644531,
      "learning_rate": 0.00019960039458027576,
      "loss": 1.6047,
      "step": 1609
    },
    {
      "epoch": 0.057657528604938495,
      "grad_norm": 1.5484936237335205,
      "learning_rate": 0.00019959935801127686,
      "loss": 1.7067,
      "step": 1610
    },
    {
      "epoch": 0.05769334073450678,
      "grad_norm": 2.451167583465576,
      "learning_rate": 0.00019959832010229712,
      "loss": 1.419,
      "step": 1611
    },
    {
      "epoch": 0.05772915286407506,
      "grad_norm": 1.6821362972259521,
      "learning_rate": 0.0001995972808533504,
      "loss": 2.039,
      "step": 1612
    },
    {
      "epoch": 0.05776496499364335,
      "grad_norm": 1.6939760446548462,
      "learning_rate": 0.00019959624026445077,
      "loss": 1.6131,
      "step": 1613
    },
    {
      "epoch": 0.05780077712321163,
      "grad_norm": 1.5620489120483398,
      "learning_rate": 0.0001995951983356122,
      "loss": 1.6838,
      "step": 1614
    },
    {
      "epoch": 0.057836589252779916,
      "grad_norm": 2.6445093154907227,
      "learning_rate": 0.0001995941550668487,
      "loss": 1.8336,
      "step": 1615
    },
    {
      "epoch": 0.0578724013823482,
      "grad_norm": 1.760237693786621,
      "learning_rate": 0.00019959311045817432,
      "loss": 1.6198,
      "step": 1616
    },
    {
      "epoch": 0.05790821351191649,
      "grad_norm": 1.3533624410629272,
      "learning_rate": 0.00019959206450960307,
      "loss": 1.7068,
      "step": 1617
    },
    {
      "epoch": 0.05794402564148477,
      "grad_norm": 2.5859081745147705,
      "learning_rate": 0.0001995910172211491,
      "loss": 1.9612,
      "step": 1618
    },
    {
      "epoch": 0.057979837771053054,
      "grad_norm": 1.4971524477005005,
      "learning_rate": 0.0001995899685928264,
      "loss": 1.7951,
      "step": 1619
    },
    {
      "epoch": 0.058015649900621344,
      "grad_norm": 1.8958418369293213,
      "learning_rate": 0.0001995889186246492,
      "loss": 1.5899,
      "step": 1620
    },
    {
      "epoch": 0.05805146203018963,
      "grad_norm": 2.1714656352996826,
      "learning_rate": 0.0001995878673166315,
      "loss": 1.5648,
      "step": 1621
    },
    {
      "epoch": 0.05808727415975791,
      "grad_norm": 2.7349958419799805,
      "learning_rate": 0.00019958681466878756,
      "loss": 1.9454,
      "step": 1622
    },
    {
      "epoch": 0.05812308628932619,
      "grad_norm": 2.072175979614258,
      "learning_rate": 0.00019958576068113145,
      "loss": 1.5197,
      "step": 1623
    },
    {
      "epoch": 0.05815889841889448,
      "grad_norm": 2.4490714073181152,
      "learning_rate": 0.00019958470535367742,
      "loss": 1.5733,
      "step": 1624
    },
    {
      "epoch": 0.058194710548462765,
      "grad_norm": 1.668999195098877,
      "learning_rate": 0.00019958364868643958,
      "loss": 1.5685,
      "step": 1625
    },
    {
      "epoch": 0.05823052267803105,
      "grad_norm": 1.6361305713653564,
      "learning_rate": 0.00019958259067943225,
      "loss": 1.9354,
      "step": 1626
    },
    {
      "epoch": 0.05826633480759933,
      "grad_norm": 1.7708817720413208,
      "learning_rate": 0.0001995815313326696,
      "loss": 1.6394,
      "step": 1627
    },
    {
      "epoch": 0.05830214693716762,
      "grad_norm": 1.748575210571289,
      "learning_rate": 0.0001995804706461659,
      "loss": 1.8433,
      "step": 1628
    },
    {
      "epoch": 0.0583379590667359,
      "grad_norm": 1.4690053462982178,
      "learning_rate": 0.0001995794086199354,
      "loss": 1.5964,
      "step": 1629
    },
    {
      "epoch": 0.058373771196304186,
      "grad_norm": 1.3967971801757812,
      "learning_rate": 0.00019957834525399242,
      "loss": 1.657,
      "step": 1630
    },
    {
      "epoch": 0.058409583325872476,
      "grad_norm": 1.4848518371582031,
      "learning_rate": 0.00019957728054835125,
      "loss": 1.8209,
      "step": 1631
    },
    {
      "epoch": 0.05844539545544076,
      "grad_norm": 2.1335513591766357,
      "learning_rate": 0.0001995762145030262,
      "loss": 1.7809,
      "step": 1632
    },
    {
      "epoch": 0.05848120758500904,
      "grad_norm": 3.323810577392578,
      "learning_rate": 0.00019957514711803164,
      "loss": 1.4154,
      "step": 1633
    },
    {
      "epoch": 0.058517019714577324,
      "grad_norm": 1.775437593460083,
      "learning_rate": 0.00019957407839338193,
      "loss": 1.6845,
      "step": 1634
    },
    {
      "epoch": 0.058552831844145614,
      "grad_norm": 2.64595365524292,
      "learning_rate": 0.00019957300832909144,
      "loss": 1.5673,
      "step": 1635
    },
    {
      "epoch": 0.0585886439737139,
      "grad_norm": 1.9590623378753662,
      "learning_rate": 0.00019957193692517455,
      "loss": 1.7878,
      "step": 1636
    },
    {
      "epoch": 0.05862445610328218,
      "grad_norm": 1.5874866247177124,
      "learning_rate": 0.00019957086418164567,
      "loss": 1.5769,
      "step": 1637
    },
    {
      "epoch": 0.05866026823285047,
      "grad_norm": 2.0571467876434326,
      "learning_rate": 0.00019956979009851927,
      "loss": 1.5945,
      "step": 1638
    },
    {
      "epoch": 0.05869608036241875,
      "grad_norm": 1.961224913597107,
      "learning_rate": 0.00019956871467580978,
      "loss": 1.8875,
      "step": 1639
    },
    {
      "epoch": 0.058731892491987035,
      "grad_norm": 1.3154363632202148,
      "learning_rate": 0.00019956763791353165,
      "loss": 1.4418,
      "step": 1640
    },
    {
      "epoch": 0.05876770462155532,
      "grad_norm": 1.3892945051193237,
      "learning_rate": 0.00019956655981169942,
      "loss": 1.7232,
      "step": 1641
    },
    {
      "epoch": 0.05880351675112361,
      "grad_norm": 1.9630827903747559,
      "learning_rate": 0.00019956548037032752,
      "loss": 1.8138,
      "step": 1642
    },
    {
      "epoch": 0.05883932888069189,
      "grad_norm": 1.364547848701477,
      "learning_rate": 0.00019956439958943053,
      "loss": 1.8376,
      "step": 1643
    },
    {
      "epoch": 0.058875141010260174,
      "grad_norm": 1.8685306310653687,
      "learning_rate": 0.00019956331746902298,
      "loss": 1.8356,
      "step": 1644
    },
    {
      "epoch": 0.05891095313982846,
      "grad_norm": 2.003082036972046,
      "learning_rate": 0.00019956223400911943,
      "loss": 1.7245,
      "step": 1645
    },
    {
      "epoch": 0.058946765269396746,
      "grad_norm": 2.334925651550293,
      "learning_rate": 0.00019956114920973442,
      "loss": 1.9533,
      "step": 1646
    },
    {
      "epoch": 0.05898257739896503,
      "grad_norm": 2.2185535430908203,
      "learning_rate": 0.00019956006307088258,
      "loss": 1.6938,
      "step": 1647
    },
    {
      "epoch": 0.05901838952853331,
      "grad_norm": 1.9847118854522705,
      "learning_rate": 0.00019955897559257853,
      "loss": 1.7853,
      "step": 1648
    },
    {
      "epoch": 0.0590542016581016,
      "grad_norm": 2.876863956451416,
      "learning_rate": 0.00019955788677483686,
      "loss": 1.6786,
      "step": 1649
    },
    {
      "epoch": 0.059090013787669884,
      "grad_norm": 1.6888865232467651,
      "learning_rate": 0.00019955679661767226,
      "loss": 1.4412,
      "step": 1650
    },
    {
      "epoch": 0.05912582591723817,
      "grad_norm": 2.452845335006714,
      "learning_rate": 0.0001995557051210994,
      "loss": 1.7187,
      "step": 1651
    },
    {
      "epoch": 0.05916163804680645,
      "grad_norm": 1.8312182426452637,
      "learning_rate": 0.0001995546122851329,
      "loss": 1.9031,
      "step": 1652
    },
    {
      "epoch": 0.05919745017637474,
      "grad_norm": 1.6904535293579102,
      "learning_rate": 0.00019955351810978754,
      "loss": 1.9099,
      "step": 1653
    },
    {
      "epoch": 0.05923326230594302,
      "grad_norm": 1.750017762184143,
      "learning_rate": 0.000199552422595078,
      "loss": 1.8949,
      "step": 1654
    },
    {
      "epoch": 0.059269074435511306,
      "grad_norm": 1.7873555421829224,
      "learning_rate": 0.000199551325741019,
      "loss": 1.6408,
      "step": 1655
    },
    {
      "epoch": 0.059304886565079595,
      "grad_norm": 2.2488410472869873,
      "learning_rate": 0.00019955022754762535,
      "loss": 2.0474,
      "step": 1656
    },
    {
      "epoch": 0.05934069869464788,
      "grad_norm": 1.8136587142944336,
      "learning_rate": 0.0001995491280149118,
      "loss": 1.9466,
      "step": 1657
    },
    {
      "epoch": 0.05937651082421616,
      "grad_norm": 1.3177629709243774,
      "learning_rate": 0.00019954802714289315,
      "loss": 1.7,
      "step": 1658
    },
    {
      "epoch": 0.059412322953784444,
      "grad_norm": 1.3142410516738892,
      "learning_rate": 0.0001995469249315842,
      "loss": 1.6965,
      "step": 1659
    },
    {
      "epoch": 0.059448135083352734,
      "grad_norm": 1.6144740581512451,
      "learning_rate": 0.00019954582138099978,
      "loss": 1.8903,
      "step": 1660
    },
    {
      "epoch": 0.059483947212921016,
      "grad_norm": 1.29384446144104,
      "learning_rate": 0.00019954471649115475,
      "loss": 1.614,
      "step": 1661
    },
    {
      "epoch": 0.0595197593424893,
      "grad_norm": 1.294191598892212,
      "learning_rate": 0.00019954361026206394,
      "loss": 1.5788,
      "step": 1662
    },
    {
      "epoch": 0.05955557147205759,
      "grad_norm": 1.5210322141647339,
      "learning_rate": 0.00019954250269374227,
      "loss": 1.5188,
      "step": 1663
    },
    {
      "epoch": 0.05959138360162587,
      "grad_norm": 2.4071707725524902,
      "learning_rate": 0.0001995413937862046,
      "loss": 1.7177,
      "step": 1664
    },
    {
      "epoch": 0.059627195731194155,
      "grad_norm": 1.5951025485992432,
      "learning_rate": 0.0001995402835394659,
      "loss": 1.5914,
      "step": 1665
    },
    {
      "epoch": 0.05966300786076244,
      "grad_norm": 2.301365852355957,
      "learning_rate": 0.00019953917195354105,
      "loss": 1.6783,
      "step": 1666
    },
    {
      "epoch": 0.05969881999033073,
      "grad_norm": 1.3707059621810913,
      "learning_rate": 0.00019953805902844508,
      "loss": 1.9201,
      "step": 1667
    },
    {
      "epoch": 0.05973463211989901,
      "grad_norm": 1.7182759046554565,
      "learning_rate": 0.00019953694476419293,
      "loss": 1.8426,
      "step": 1668
    },
    {
      "epoch": 0.05977044424946729,
      "grad_norm": 1.4055579900741577,
      "learning_rate": 0.00019953582916079957,
      "loss": 1.4242,
      "step": 1669
    },
    {
      "epoch": 0.05980625637903558,
      "grad_norm": 1.643748164176941,
      "learning_rate": 0.00019953471221827998,
      "loss": 1.7121,
      "step": 1670
    },
    {
      "epoch": 0.059842068508603866,
      "grad_norm": 1.5763031244277954,
      "learning_rate": 0.00019953359393664927,
      "loss": 1.564,
      "step": 1671
    },
    {
      "epoch": 0.05987788063817215,
      "grad_norm": 1.790825605392456,
      "learning_rate": 0.0001995324743159224,
      "loss": 1.5434,
      "step": 1672
    },
    {
      "epoch": 0.05991369276774043,
      "grad_norm": 2.168774127960205,
      "learning_rate": 0.00019953135335611452,
      "loss": 1.6468,
      "step": 1673
    },
    {
      "epoch": 0.05994950489730872,
      "grad_norm": 1.6681972742080688,
      "learning_rate": 0.00019953023105724068,
      "loss": 1.9108,
      "step": 1674
    },
    {
      "epoch": 0.059985317026877004,
      "grad_norm": 1.9940334558486938,
      "learning_rate": 0.00019952910741931592,
      "loss": 1.5585,
      "step": 1675
    },
    {
      "epoch": 0.06002112915644529,
      "grad_norm": 2.587232828140259,
      "learning_rate": 0.0001995279824423554,
      "loss": 1.8064,
      "step": 1676
    },
    {
      "epoch": 0.06005694128601357,
      "grad_norm": 1.9780170917510986,
      "learning_rate": 0.0001995268561263743,
      "loss": 1.8043,
      "step": 1677
    },
    {
      "epoch": 0.06009275341558186,
      "grad_norm": 1.620010256767273,
      "learning_rate": 0.00019952572847138772,
      "loss": 1.865,
      "step": 1678
    },
    {
      "epoch": 0.06012856554515014,
      "grad_norm": 2.4111506938934326,
      "learning_rate": 0.00019952459947741082,
      "loss": 1.9095,
      "step": 1679
    },
    {
      "epoch": 0.060164377674718425,
      "grad_norm": 2.624497652053833,
      "learning_rate": 0.00019952346914445883,
      "loss": 1.902,
      "step": 1680
    },
    {
      "epoch": 0.060200189804286715,
      "grad_norm": 1.8209869861602783,
      "learning_rate": 0.00019952233747254691,
      "loss": 1.7628,
      "step": 1681
    },
    {
      "epoch": 0.060236001933855,
      "grad_norm": 1.5332977771759033,
      "learning_rate": 0.00019952120446169037,
      "loss": 1.6418,
      "step": 1682
    },
    {
      "epoch": 0.06027181406342328,
      "grad_norm": 1.8602702617645264,
      "learning_rate": 0.00019952007011190433,
      "loss": 1.918,
      "step": 1683
    },
    {
      "epoch": 0.06030762619299156,
      "grad_norm": 1.2004653215408325,
      "learning_rate": 0.00019951893442320416,
      "loss": 1.5955,
      "step": 1684
    },
    {
      "epoch": 0.06034343832255985,
      "grad_norm": 1.2076934576034546,
      "learning_rate": 0.0001995177973956051,
      "loss": 1.7053,
      "step": 1685
    },
    {
      "epoch": 0.060379250452128136,
      "grad_norm": 2.4891135692596436,
      "learning_rate": 0.00019951665902912243,
      "loss": 1.902,
      "step": 1686
    },
    {
      "epoch": 0.06041506258169642,
      "grad_norm": 1.7342336177825928,
      "learning_rate": 0.00019951551932377148,
      "loss": 1.7217,
      "step": 1687
    },
    {
      "epoch": 0.06045087471126471,
      "grad_norm": 2.313361406326294,
      "learning_rate": 0.00019951437827956758,
      "loss": 1.5658,
      "step": 1688
    },
    {
      "epoch": 0.06048668684083299,
      "grad_norm": 1.3781776428222656,
      "learning_rate": 0.0001995132358965261,
      "loss": 1.5952,
      "step": 1689
    },
    {
      "epoch": 0.060522498970401274,
      "grad_norm": 2.3602588176727295,
      "learning_rate": 0.00019951209217466238,
      "loss": 1.8622,
      "step": 1690
    },
    {
      "epoch": 0.06055831109996956,
      "grad_norm": 2.0375075340270996,
      "learning_rate": 0.0001995109471139918,
      "loss": 1.6982,
      "step": 1691
    },
    {
      "epoch": 0.06059412322953785,
      "grad_norm": 1.8000264167785645,
      "learning_rate": 0.0001995098007145298,
      "loss": 1.8062,
      "step": 1692
    },
    {
      "epoch": 0.06062993535910613,
      "grad_norm": 1.7053439617156982,
      "learning_rate": 0.00019950865297629184,
      "loss": 1.789,
      "step": 1693
    },
    {
      "epoch": 0.06066574748867441,
      "grad_norm": 3.238233804702759,
      "learning_rate": 0.00019950750389929328,
      "loss": 2.2441,
      "step": 1694
    },
    {
      "epoch": 0.0607015596182427,
      "grad_norm": 1.5956525802612305,
      "learning_rate": 0.0001995063534835496,
      "loss": 1.8782,
      "step": 1695
    },
    {
      "epoch": 0.060737371747810985,
      "grad_norm": 2.0592153072357178,
      "learning_rate": 0.0001995052017290763,
      "loss": 1.846,
      "step": 1696
    },
    {
      "epoch": 0.06077318387737927,
      "grad_norm": 1.7917789220809937,
      "learning_rate": 0.00019950404863588883,
      "loss": 1.7443,
      "step": 1697
    },
    {
      "epoch": 0.06080899600694755,
      "grad_norm": 1.8193100690841675,
      "learning_rate": 0.00019950289420400278,
      "loss": 2.2193,
      "step": 1698
    },
    {
      "epoch": 0.06084480813651584,
      "grad_norm": 1.6950918436050415,
      "learning_rate": 0.00019950173843343364,
      "loss": 1.9775,
      "step": 1699
    },
    {
      "epoch": 0.06088062026608412,
      "grad_norm": 1.3480393886566162,
      "learning_rate": 0.00019950058132419692,
      "loss": 1.5773,
      "step": 1700
    },
    {
      "epoch": 0.060916432395652406,
      "grad_norm": 2.6421098709106445,
      "learning_rate": 0.00019949942287630825,
      "loss": 1.8098,
      "step": 1701
    },
    {
      "epoch": 0.06095224452522069,
      "grad_norm": 1.7833307981491089,
      "learning_rate": 0.00019949826308978316,
      "loss": 1.758,
      "step": 1702
    },
    {
      "epoch": 0.06098805665478898,
      "grad_norm": 2.0496833324432373,
      "learning_rate": 0.0001994971019646373,
      "loss": 1.7918,
      "step": 1703
    },
    {
      "epoch": 0.06102386878435726,
      "grad_norm": 2.668501377105713,
      "learning_rate": 0.0001994959395008863,
      "loss": 1.8106,
      "step": 1704
    },
    {
      "epoch": 0.061059680913925544,
      "grad_norm": 1.7229629755020142,
      "learning_rate": 0.00019949477569854575,
      "loss": 1.7763,
      "step": 1705
    },
    {
      "epoch": 0.061095493043493834,
      "grad_norm": 2.5283937454223633,
      "learning_rate": 0.00019949361055763133,
      "loss": 1.9133,
      "step": 1706
    },
    {
      "epoch": 0.06113130517306212,
      "grad_norm": 2.3446457386016846,
      "learning_rate": 0.00019949244407815875,
      "loss": 1.5309,
      "step": 1707
    },
    {
      "epoch": 0.0611671173026304,
      "grad_norm": 1.1809579133987427,
      "learning_rate": 0.00019949127626014363,
      "loss": 1.4884,
      "step": 1708
    },
    {
      "epoch": 0.06120292943219868,
      "grad_norm": 1.2034236192703247,
      "learning_rate": 0.00019949010710360173,
      "loss": 1.5259,
      "step": 1709
    },
    {
      "epoch": 0.06123874156176697,
      "grad_norm": 2.4860334396362305,
      "learning_rate": 0.0001994889366085488,
      "loss": 1.9134,
      "step": 1710
    },
    {
      "epoch": 0.061274553691335255,
      "grad_norm": 1.7052736282348633,
      "learning_rate": 0.00019948776477500053,
      "loss": 1.8427,
      "step": 1711
    },
    {
      "epoch": 0.06131036582090354,
      "grad_norm": 1.8961480855941772,
      "learning_rate": 0.0001994865916029727,
      "loss": 1.6836,
      "step": 1712
    },
    {
      "epoch": 0.06134617795047183,
      "grad_norm": 1.532853603363037,
      "learning_rate": 0.00019948541709248116,
      "loss": 1.7492,
      "step": 1713
    },
    {
      "epoch": 0.06138199008004011,
      "grad_norm": 1.632897138595581,
      "learning_rate": 0.00019948424124354163,
      "loss": 1.7434,
      "step": 1714
    },
    {
      "epoch": 0.061417802209608394,
      "grad_norm": 2.043008804321289,
      "learning_rate": 0.00019948306405616996,
      "loss": 1.8477,
      "step": 1715
    },
    {
      "epoch": 0.061453614339176676,
      "grad_norm": 1.3530820608139038,
      "learning_rate": 0.00019948188553038198,
      "loss": 1.9704,
      "step": 1716
    },
    {
      "epoch": 0.061489426468744966,
      "grad_norm": 1.3419357538223267,
      "learning_rate": 0.0001994807056661936,
      "loss": 1.6517,
      "step": 1717
    },
    {
      "epoch": 0.06152523859831325,
      "grad_norm": 1.2969164848327637,
      "learning_rate": 0.00019947952446362058,
      "loss": 1.8418,
      "step": 1718
    },
    {
      "epoch": 0.06156105072788153,
      "grad_norm": 2.865709066390991,
      "learning_rate": 0.00019947834192267892,
      "loss": 1.862,
      "step": 1719
    },
    {
      "epoch": 0.06159686285744982,
      "grad_norm": 1.7418867349624634,
      "learning_rate": 0.00019947715804338447,
      "loss": 1.8999,
      "step": 1720
    },
    {
      "epoch": 0.061632674987018105,
      "grad_norm": 2.2759385108947754,
      "learning_rate": 0.00019947597282575318,
      "loss": 1.6186,
      "step": 1721
    },
    {
      "epoch": 0.06166848711658639,
      "grad_norm": 1.5044995546340942,
      "learning_rate": 0.00019947478626980097,
      "loss": 1.7092,
      "step": 1722
    },
    {
      "epoch": 0.06170429924615467,
      "grad_norm": 1.620406150817871,
      "learning_rate": 0.00019947359837554384,
      "loss": 1.6904,
      "step": 1723
    },
    {
      "epoch": 0.06174011137572296,
      "grad_norm": 1.9609277248382568,
      "learning_rate": 0.00019947240914299776,
      "loss": 1.6958,
      "step": 1724
    },
    {
      "epoch": 0.06177592350529124,
      "grad_norm": 2.1059086322784424,
      "learning_rate": 0.00019947121857217875,
      "loss": 1.7419,
      "step": 1725
    },
    {
      "epoch": 0.061811735634859526,
      "grad_norm": 2.0715277194976807,
      "learning_rate": 0.00019947002666310276,
      "loss": 1.6952,
      "step": 1726
    },
    {
      "epoch": 0.06184754776442781,
      "grad_norm": 1.6598073244094849,
      "learning_rate": 0.00019946883341578588,
      "loss": 1.6857,
      "step": 1727
    },
    {
      "epoch": 0.0618833598939961,
      "grad_norm": 1.5604004859924316,
      "learning_rate": 0.00019946763883024415,
      "loss": 1.6462,
      "step": 1728
    },
    {
      "epoch": 0.06191917202356438,
      "grad_norm": 2.105992555618286,
      "learning_rate": 0.00019946644290649368,
      "loss": 1.77,
      "step": 1729
    },
    {
      "epoch": 0.061954984153132664,
      "grad_norm": 2.2045366764068604,
      "learning_rate": 0.00019946524564455048,
      "loss": 2.0856,
      "step": 1730
    },
    {
      "epoch": 0.061990796282700954,
      "grad_norm": 1.7695341110229492,
      "learning_rate": 0.00019946404704443072,
      "loss": 1.7301,
      "step": 1731
    },
    {
      "epoch": 0.06202660841226924,
      "grad_norm": 1.6013516187667847,
      "learning_rate": 0.00019946284710615052,
      "loss": 1.5589,
      "step": 1732
    },
    {
      "epoch": 0.06206242054183752,
      "grad_norm": 1.9727685451507568,
      "learning_rate": 0.00019946164582972594,
      "loss": 1.8941,
      "step": 1733
    },
    {
      "epoch": 0.0620982326714058,
      "grad_norm": 1.8168443441390991,
      "learning_rate": 0.0001994604432151733,
      "loss": 1.5806,
      "step": 1734
    },
    {
      "epoch": 0.06213404480097409,
      "grad_norm": 1.5395933389663696,
      "learning_rate": 0.00019945923926250865,
      "loss": 1.88,
      "step": 1735
    },
    {
      "epoch": 0.062169856930542375,
      "grad_norm": 1.3251415491104126,
      "learning_rate": 0.00019945803397174823,
      "loss": 1.6235,
      "step": 1736
    },
    {
      "epoch": 0.06220566906011066,
      "grad_norm": 1.7758703231811523,
      "learning_rate": 0.00019945682734290825,
      "loss": 1.7609,
      "step": 1737
    },
    {
      "epoch": 0.06224148118967895,
      "grad_norm": 1.366156816482544,
      "learning_rate": 0.00019945561937600496,
      "loss": 1.629,
      "step": 1738
    },
    {
      "epoch": 0.06227729331924723,
      "grad_norm": 2.5820953845977783,
      "learning_rate": 0.0001994544100710546,
      "loss": 1.6179,
      "step": 1739
    },
    {
      "epoch": 0.06231310544881551,
      "grad_norm": 1.8584883213043213,
      "learning_rate": 0.0001994531994280734,
      "loss": 1.8395,
      "step": 1740
    },
    {
      "epoch": 0.062348917578383796,
      "grad_norm": 1.4980369806289673,
      "learning_rate": 0.00019945198744707776,
      "loss": 1.8487,
      "step": 1741
    },
    {
      "epoch": 0.062384729707952086,
      "grad_norm": 2.2591607570648193,
      "learning_rate": 0.00019945077412808387,
      "loss": 1.8087,
      "step": 1742
    },
    {
      "epoch": 0.06242054183752037,
      "grad_norm": 4.082409381866455,
      "learning_rate": 0.0001994495594711081,
      "loss": 1.7609,
      "step": 1743
    },
    {
      "epoch": 0.06245635396708865,
      "grad_norm": 1.8433887958526611,
      "learning_rate": 0.0001994483434761668,
      "loss": 1.6445,
      "step": 1744
    },
    {
      "epoch": 0.06249216609665694,
      "grad_norm": 1.5228683948516846,
      "learning_rate": 0.0001994471261432763,
      "loss": 1.8867,
      "step": 1745
    },
    {
      "epoch": 0.06252797822622522,
      "grad_norm": 1.2005826234817505,
      "learning_rate": 0.00019944590747245298,
      "loss": 1.7023,
      "step": 1746
    },
    {
      "epoch": 0.06256379035579351,
      "grad_norm": 1.6530516147613525,
      "learning_rate": 0.0001994446874637133,
      "loss": 1.9049,
      "step": 1747
    },
    {
      "epoch": 0.0625996024853618,
      "grad_norm": 2.352060556411743,
      "learning_rate": 0.00019944346611707356,
      "loss": 1.6247,
      "step": 1748
    },
    {
      "epoch": 0.06263541461493008,
      "grad_norm": 2.1047065258026123,
      "learning_rate": 0.0001994422434325503,
      "loss": 1.6373,
      "step": 1749
    },
    {
      "epoch": 0.06267122674449836,
      "grad_norm": 1.7999935150146484,
      "learning_rate": 0.00019944101941015994,
      "loss": 1.7002,
      "step": 1750
    },
    {
      "epoch": 0.06270703887406665,
      "grad_norm": 1.833964467048645,
      "learning_rate": 0.0001994397940499189,
      "loss": 1.6692,
      "step": 1751
    },
    {
      "epoch": 0.06274285100363493,
      "grad_norm": 1.7198271751403809,
      "learning_rate": 0.00019943856735184373,
      "loss": 1.9069,
      "step": 1752
    },
    {
      "epoch": 0.06277866313320321,
      "grad_norm": 2.55334734916687,
      "learning_rate": 0.00019943733931595086,
      "loss": 1.789,
      "step": 1753
    },
    {
      "epoch": 0.06281447526277151,
      "grad_norm": 1.8098565340042114,
      "learning_rate": 0.0001994361099422569,
      "loss": 1.7267,
      "step": 1754
    },
    {
      "epoch": 0.06285028739233979,
      "grad_norm": 1.1642190217971802,
      "learning_rate": 0.00019943487923077831,
      "loss": 1.4918,
      "step": 1755
    },
    {
      "epoch": 0.06288609952190807,
      "grad_norm": 2.4731671810150146,
      "learning_rate": 0.0001994336471815317,
      "loss": 1.7026,
      "step": 1756
    },
    {
      "epoch": 0.06292191165147636,
      "grad_norm": 1.744199514389038,
      "learning_rate": 0.00019943241379453364,
      "loss": 1.6319,
      "step": 1757
    },
    {
      "epoch": 0.06295772378104464,
      "grad_norm": 2.089003562927246,
      "learning_rate": 0.00019943117906980068,
      "loss": 1.5497,
      "step": 1758
    },
    {
      "epoch": 0.06299353591061292,
      "grad_norm": 1.1575398445129395,
      "learning_rate": 0.00019942994300734947,
      "loss": 1.7987,
      "step": 1759
    },
    {
      "epoch": 0.0630293480401812,
      "grad_norm": 1.753517746925354,
      "learning_rate": 0.00019942870560719664,
      "loss": 1.7975,
      "step": 1760
    },
    {
      "epoch": 0.06306516016974949,
      "grad_norm": 2.7385034561157227,
      "learning_rate": 0.00019942746686935883,
      "loss": 1.641,
      "step": 1761
    },
    {
      "epoch": 0.06310097229931778,
      "grad_norm": 3.074314594268799,
      "learning_rate": 0.0001994262267938527,
      "loss": 1.6846,
      "step": 1762
    },
    {
      "epoch": 0.06313678442888607,
      "grad_norm": 1.2961831092834473,
      "learning_rate": 0.00019942498538069495,
      "loss": 1.7411,
      "step": 1763
    },
    {
      "epoch": 0.06317259655845435,
      "grad_norm": 2.032168388366699,
      "learning_rate": 0.00019942374262990224,
      "loss": 1.8094,
      "step": 1764
    },
    {
      "epoch": 0.06320840868802263,
      "grad_norm": 1.753873586654663,
      "learning_rate": 0.00019942249854149136,
      "loss": 1.7005,
      "step": 1765
    },
    {
      "epoch": 0.06324422081759092,
      "grad_norm": 2.4005773067474365,
      "learning_rate": 0.000199421253115479,
      "loss": 1.8196,
      "step": 1766
    },
    {
      "epoch": 0.0632800329471592,
      "grad_norm": 1.66982102394104,
      "learning_rate": 0.0001994200063518819,
      "loss": 1.7387,
      "step": 1767
    },
    {
      "epoch": 0.06331584507672748,
      "grad_norm": 1.7754584550857544,
      "learning_rate": 0.0001994187582507169,
      "loss": 1.7563,
      "step": 1768
    },
    {
      "epoch": 0.06335165720629578,
      "grad_norm": 2.168553590774536,
      "learning_rate": 0.0001994175088120007,
      "loss": 1.5715,
      "step": 1769
    },
    {
      "epoch": 0.06338746933586406,
      "grad_norm": 2.892993450164795,
      "learning_rate": 0.00019941625803575019,
      "loss": 1.6514,
      "step": 1770
    },
    {
      "epoch": 0.06342328146543234,
      "grad_norm": 1.5328547954559326,
      "learning_rate": 0.00019941500592198216,
      "loss": 1.7424,
      "step": 1771
    },
    {
      "epoch": 0.06345909359500063,
      "grad_norm": 1.4240583181381226,
      "learning_rate": 0.00019941375247071346,
      "loss": 1.7308,
      "step": 1772
    },
    {
      "epoch": 0.06349490572456891,
      "grad_norm": 1.379332184791565,
      "learning_rate": 0.00019941249768196093,
      "loss": 1.6878,
      "step": 1773
    },
    {
      "epoch": 0.06353071785413719,
      "grad_norm": 1.5497807264328003,
      "learning_rate": 0.0001994112415557415,
      "loss": 1.6895,
      "step": 1774
    },
    {
      "epoch": 0.06356652998370547,
      "grad_norm": 2.0320279598236084,
      "learning_rate": 0.00019940998409207205,
      "loss": 1.9394,
      "step": 1775
    },
    {
      "epoch": 0.06360234211327377,
      "grad_norm": 1.5920579433441162,
      "learning_rate": 0.00019940872529096947,
      "loss": 1.4275,
      "step": 1776
    },
    {
      "epoch": 0.06363815424284205,
      "grad_norm": 1.7592054605484009,
      "learning_rate": 0.00019940746515245073,
      "loss": 1.8923,
      "step": 1777
    },
    {
      "epoch": 0.06367396637241034,
      "grad_norm": 1.5628163814544678,
      "learning_rate": 0.0001994062036765328,
      "loss": 1.9191,
      "step": 1778
    },
    {
      "epoch": 0.06370977850197862,
      "grad_norm": 2.930593729019165,
      "learning_rate": 0.0001994049408632326,
      "loss": 1.7287,
      "step": 1779
    },
    {
      "epoch": 0.0637455906315469,
      "grad_norm": 2.2003262042999268,
      "learning_rate": 0.00019940367671256716,
      "loss": 1.8437,
      "step": 1780
    },
    {
      "epoch": 0.06378140276111519,
      "grad_norm": 1.2991431951522827,
      "learning_rate": 0.00019940241122455346,
      "loss": 1.8275,
      "step": 1781
    },
    {
      "epoch": 0.06381721489068347,
      "grad_norm": 1.6640808582305908,
      "learning_rate": 0.00019940114439920853,
      "loss": 1.7935,
      "step": 1782
    },
    {
      "epoch": 0.06385302702025177,
      "grad_norm": 1.9864023923873901,
      "learning_rate": 0.00019939987623654944,
      "loss": 1.7738,
      "step": 1783
    },
    {
      "epoch": 0.06388883914982005,
      "grad_norm": 1.4357529878616333,
      "learning_rate": 0.0001993986067365932,
      "loss": 1.7587,
      "step": 1784
    },
    {
      "epoch": 0.06392465127938833,
      "grad_norm": 1.5999445915222168,
      "learning_rate": 0.00019939733589935694,
      "loss": 1.7795,
      "step": 1785
    },
    {
      "epoch": 0.06396046340895661,
      "grad_norm": 2.447521448135376,
      "learning_rate": 0.00019939606372485776,
      "loss": 1.9686,
      "step": 1786
    },
    {
      "epoch": 0.0639962755385249,
      "grad_norm": 1.655842900276184,
      "learning_rate": 0.00019939479021311273,
      "loss": 2.0018,
      "step": 1787
    },
    {
      "epoch": 0.06403208766809318,
      "grad_norm": 1.4144165515899658,
      "learning_rate": 0.000199393515364139,
      "loss": 1.5992,
      "step": 1788
    },
    {
      "epoch": 0.06406789979766146,
      "grad_norm": 2.399679183959961,
      "learning_rate": 0.00019939223917795373,
      "loss": 1.7488,
      "step": 1789
    },
    {
      "epoch": 0.06410371192722976,
      "grad_norm": 2.1720006465911865,
      "learning_rate": 0.00019939096165457411,
      "loss": 1.8954,
      "step": 1790
    },
    {
      "epoch": 0.06413952405679804,
      "grad_norm": 1.701594352722168,
      "learning_rate": 0.00019938968279401727,
      "loss": 1.9236,
      "step": 1791
    },
    {
      "epoch": 0.06417533618636632,
      "grad_norm": 1.7471704483032227,
      "learning_rate": 0.0001993884025963005,
      "loss": 1.9107,
      "step": 1792
    },
    {
      "epoch": 0.06421114831593461,
      "grad_norm": 1.7068006992340088,
      "learning_rate": 0.00019938712106144096,
      "loss": 1.9236,
      "step": 1793
    },
    {
      "epoch": 0.06424696044550289,
      "grad_norm": 1.5164170265197754,
      "learning_rate": 0.0001993858381894559,
      "loss": 2.014,
      "step": 1794
    },
    {
      "epoch": 0.06428277257507117,
      "grad_norm": 1.6961904764175415,
      "learning_rate": 0.00019938455398036257,
      "loss": 2.0187,
      "step": 1795
    },
    {
      "epoch": 0.06431858470463946,
      "grad_norm": 1.8207290172576904,
      "learning_rate": 0.00019938326843417826,
      "loss": 1.533,
      "step": 1796
    },
    {
      "epoch": 0.06435439683420775,
      "grad_norm": 2.8234851360321045,
      "learning_rate": 0.0001993819815509203,
      "loss": 1.7141,
      "step": 1797
    },
    {
      "epoch": 0.06439020896377604,
      "grad_norm": 2.3623080253601074,
      "learning_rate": 0.00019938069333060593,
      "loss": 2.0943,
      "step": 1798
    },
    {
      "epoch": 0.06442602109334432,
      "grad_norm": 1.3733361959457397,
      "learning_rate": 0.00019937940377325256,
      "loss": 1.6068,
      "step": 1799
    },
    {
      "epoch": 0.0644618332229126,
      "grad_norm": 1.828696370124817,
      "learning_rate": 0.0001993781128788775,
      "loss": 1.6581,
      "step": 1800
    },
    {
      "epoch": 0.06449764535248088,
      "grad_norm": 1.8768473863601685,
      "learning_rate": 0.00019937682064749811,
      "loss": 1.8659,
      "step": 1801
    },
    {
      "epoch": 0.06453345748204917,
      "grad_norm": 1.796538233757019,
      "learning_rate": 0.0001993755270791318,
      "loss": 1.6696,
      "step": 1802
    },
    {
      "epoch": 0.06456926961161745,
      "grad_norm": 2.022562026977539,
      "learning_rate": 0.00019937423217379594,
      "loss": 1.5342,
      "step": 1803
    },
    {
      "epoch": 0.06460508174118575,
      "grad_norm": 1.4857107400894165,
      "learning_rate": 0.00019937293593150796,
      "loss": 1.3543,
      "step": 1804
    },
    {
      "epoch": 0.06464089387075403,
      "grad_norm": 1.9994350671768188,
      "learning_rate": 0.00019937163835228534,
      "loss": 1.6907,
      "step": 1805
    },
    {
      "epoch": 0.06467670600032231,
      "grad_norm": 1.9123289585113525,
      "learning_rate": 0.0001993703394361455,
      "loss": 1.5374,
      "step": 1806
    },
    {
      "epoch": 0.0647125181298906,
      "grad_norm": 1.8468797206878662,
      "learning_rate": 0.0001993690391831059,
      "loss": 1.4841,
      "step": 1807
    },
    {
      "epoch": 0.06474833025945888,
      "grad_norm": 2.5288283824920654,
      "learning_rate": 0.00019936773759318408,
      "loss": 1.5264,
      "step": 1808
    },
    {
      "epoch": 0.06478414238902716,
      "grad_norm": 1.552699089050293,
      "learning_rate": 0.0001993664346663975,
      "loss": 1.6451,
      "step": 1809
    },
    {
      "epoch": 0.06481995451859544,
      "grad_norm": 1.2891765832901,
      "learning_rate": 0.00019936513040276371,
      "loss": 1.4697,
      "step": 1810
    },
    {
      "epoch": 0.06485576664816373,
      "grad_norm": 2.329200267791748,
      "learning_rate": 0.00019936382480230028,
      "loss": 1.55,
      "step": 1811
    },
    {
      "epoch": 0.06489157877773202,
      "grad_norm": 1.7297874689102173,
      "learning_rate": 0.00019936251786502478,
      "loss": 1.4764,
      "step": 1812
    },
    {
      "epoch": 0.0649273909073003,
      "grad_norm": 1.8127317428588867,
      "learning_rate": 0.00019936120959095473,
      "loss": 1.5498,
      "step": 1813
    },
    {
      "epoch": 0.06496320303686859,
      "grad_norm": 1.7599560022354126,
      "learning_rate": 0.0001993598999801078,
      "loss": 1.7308,
      "step": 1814
    },
    {
      "epoch": 0.06499901516643687,
      "grad_norm": 2.6168103218078613,
      "learning_rate": 0.00019935858903250155,
      "loss": 1.8455,
      "step": 1815
    },
    {
      "epoch": 0.06503482729600515,
      "grad_norm": 2.163510799407959,
      "learning_rate": 0.00019935727674815369,
      "loss": 1.6476,
      "step": 1816
    },
    {
      "epoch": 0.06507063942557344,
      "grad_norm": 1.5857428312301636,
      "learning_rate": 0.0001993559631270818,
      "loss": 1.771,
      "step": 1817
    },
    {
      "epoch": 0.06510645155514172,
      "grad_norm": 2.8085968494415283,
      "learning_rate": 0.0001993546481693036,
      "loss": 1.7182,
      "step": 1818
    },
    {
      "epoch": 0.06514226368471002,
      "grad_norm": 1.4190795421600342,
      "learning_rate": 0.00019935333187483676,
      "loss": 1.6694,
      "step": 1819
    },
    {
      "epoch": 0.0651780758142783,
      "grad_norm": 1.375416874885559,
      "learning_rate": 0.000199352014243699,
      "loss": 1.7377,
      "step": 1820
    },
    {
      "epoch": 0.06521388794384658,
      "grad_norm": 2.3879919052124023,
      "learning_rate": 0.00019935069527590805,
      "loss": 1.703,
      "step": 1821
    },
    {
      "epoch": 0.06524970007341487,
      "grad_norm": 1.3292384147644043,
      "learning_rate": 0.00019934937497148163,
      "loss": 1.657,
      "step": 1822
    },
    {
      "epoch": 0.06528551220298315,
      "grad_norm": 1.2860294580459595,
      "learning_rate": 0.00019934805333043752,
      "loss": 1.6298,
      "step": 1823
    },
    {
      "epoch": 0.06532132433255143,
      "grad_norm": 1.6525280475616455,
      "learning_rate": 0.00019934673035279353,
      "loss": 1.8054,
      "step": 1824
    },
    {
      "epoch": 0.06535713646211971,
      "grad_norm": 1.3846570253372192,
      "learning_rate": 0.00019934540603856743,
      "loss": 1.6464,
      "step": 1825
    },
    {
      "epoch": 0.06539294859168801,
      "grad_norm": 1.7888760566711426,
      "learning_rate": 0.000199344080387777,
      "loss": 1.9671,
      "step": 1826
    },
    {
      "epoch": 0.0654287607212563,
      "grad_norm": 1.9460965394973755,
      "learning_rate": 0.00019934275340044013,
      "loss": 1.7569,
      "step": 1827
    },
    {
      "epoch": 0.06546457285082458,
      "grad_norm": 2.1091198921203613,
      "learning_rate": 0.0001993414250765747,
      "loss": 1.6684,
      "step": 1828
    },
    {
      "epoch": 0.06550038498039286,
      "grad_norm": 1.813179612159729,
      "learning_rate": 0.0001993400954161985,
      "loss": 1.6355,
      "step": 1829
    },
    {
      "epoch": 0.06553619710996114,
      "grad_norm": 1.5191233158111572,
      "learning_rate": 0.00019933876441932943,
      "loss": 1.7589,
      "step": 1830
    },
    {
      "epoch": 0.06557200923952942,
      "grad_norm": 2.3240489959716797,
      "learning_rate": 0.00019933743208598546,
      "loss": 1.6427,
      "step": 1831
    },
    {
      "epoch": 0.06560782136909771,
      "grad_norm": 1.6643846035003662,
      "learning_rate": 0.00019933609841618445,
      "loss": 1.8053,
      "step": 1832
    },
    {
      "epoch": 0.065643633498666,
      "grad_norm": 2.014047145843506,
      "learning_rate": 0.0001993347634099444,
      "loss": 1.3374,
      "step": 1833
    },
    {
      "epoch": 0.06567944562823429,
      "grad_norm": 1.8650319576263428,
      "learning_rate": 0.00019933342706728323,
      "loss": 1.9769,
      "step": 1834
    },
    {
      "epoch": 0.06571525775780257,
      "grad_norm": 1.386111855506897,
      "learning_rate": 0.0001993320893882189,
      "loss": 1.2967,
      "step": 1835
    },
    {
      "epoch": 0.06575106988737085,
      "grad_norm": 1.4114490747451782,
      "learning_rate": 0.00019933075037276949,
      "loss": 1.5003,
      "step": 1836
    },
    {
      "epoch": 0.06578688201693914,
      "grad_norm": 1.2881723642349243,
      "learning_rate": 0.00019932941002095294,
      "loss": 2.0214,
      "step": 1837
    },
    {
      "epoch": 0.06582269414650742,
      "grad_norm": 1.827712059020996,
      "learning_rate": 0.00019932806833278726,
      "loss": 1.4548,
      "step": 1838
    },
    {
      "epoch": 0.0658585062760757,
      "grad_norm": 2.9244232177734375,
      "learning_rate": 0.0001993267253082906,
      "loss": 1.6894,
      "step": 1839
    },
    {
      "epoch": 0.065894318405644,
      "grad_norm": 1.268809199333191,
      "learning_rate": 0.00019932538094748098,
      "loss": 1.5501,
      "step": 1840
    },
    {
      "epoch": 0.06593013053521228,
      "grad_norm": 1.1067763566970825,
      "learning_rate": 0.00019932403525037642,
      "loss": 1.6284,
      "step": 1841
    },
    {
      "epoch": 0.06596594266478056,
      "grad_norm": 2.2531087398529053,
      "learning_rate": 0.00019932268821699513,
      "loss": 1.7967,
      "step": 1842
    },
    {
      "epoch": 0.06600175479434885,
      "grad_norm": 2.0509607791900635,
      "learning_rate": 0.0001993213398473552,
      "loss": 1.2527,
      "step": 1843
    },
    {
      "epoch": 0.06603756692391713,
      "grad_norm": 2.42293381690979,
      "learning_rate": 0.00019931999014147472,
      "loss": 1.8235,
      "step": 1844
    },
    {
      "epoch": 0.06607337905348541,
      "grad_norm": 1.7757221460342407,
      "learning_rate": 0.0001993186390993719,
      "loss": 2.0193,
      "step": 1845
    },
    {
      "epoch": 0.0661091911830537,
      "grad_norm": 1.6632944345474243,
      "learning_rate": 0.0001993172867210649,
      "loss": 1.8347,
      "step": 1846
    },
    {
      "epoch": 0.06614500331262199,
      "grad_norm": 2.6417300701141357,
      "learning_rate": 0.00019931593300657192,
      "loss": 2.097,
      "step": 1847
    },
    {
      "epoch": 0.06618081544219027,
      "grad_norm": 2.0074949264526367,
      "learning_rate": 0.00019931457795591118,
      "loss": 1.5325,
      "step": 1848
    },
    {
      "epoch": 0.06621662757175856,
      "grad_norm": 1.9241271018981934,
      "learning_rate": 0.00019931322156910088,
      "loss": 1.7955,
      "step": 1849
    },
    {
      "epoch": 0.06625243970132684,
      "grad_norm": 1.4078401327133179,
      "learning_rate": 0.0001993118638461593,
      "loss": 1.6546,
      "step": 1850
    },
    {
      "epoch": 0.06628825183089512,
      "grad_norm": 1.9178889989852905,
      "learning_rate": 0.00019931050478710468,
      "loss": 1.7818,
      "step": 1851
    },
    {
      "epoch": 0.0663240639604634,
      "grad_norm": 2.1118886470794678,
      "learning_rate": 0.00019930914439195534,
      "loss": 1.7646,
      "step": 1852
    },
    {
      "epoch": 0.06635987609003169,
      "grad_norm": 2.849180221557617,
      "learning_rate": 0.00019930778266072957,
      "loss": 1.6227,
      "step": 1853
    },
    {
      "epoch": 0.06639568821959999,
      "grad_norm": 2.1016101837158203,
      "learning_rate": 0.00019930641959344566,
      "loss": 1.263,
      "step": 1854
    },
    {
      "epoch": 0.06643150034916827,
      "grad_norm": 1.8419221639633179,
      "learning_rate": 0.00019930505519012197,
      "loss": 1.5502,
      "step": 1855
    },
    {
      "epoch": 0.06646731247873655,
      "grad_norm": 2.3808183670043945,
      "learning_rate": 0.0001993036894507769,
      "loss": 1.8669,
      "step": 1856
    },
    {
      "epoch": 0.06650312460830483,
      "grad_norm": 1.957215666770935,
      "learning_rate": 0.00019930232237542873,
      "loss": 1.7474,
      "step": 1857
    },
    {
      "epoch": 0.06653893673787312,
      "grad_norm": 1.7244082689285278,
      "learning_rate": 0.0001993009539640959,
      "loss": 1.6863,
      "step": 1858
    },
    {
      "epoch": 0.0665747488674414,
      "grad_norm": 1.3676176071166992,
      "learning_rate": 0.00019929958421679685,
      "loss": 1.6681,
      "step": 1859
    },
    {
      "epoch": 0.06661056099700968,
      "grad_norm": 1.637736439704895,
      "learning_rate": 0.00019929821313354997,
      "loss": 1.8369,
      "step": 1860
    },
    {
      "epoch": 0.06664637312657797,
      "grad_norm": 1.4250011444091797,
      "learning_rate": 0.00019929684071437373,
      "loss": 1.7336,
      "step": 1861
    },
    {
      "epoch": 0.06668218525614626,
      "grad_norm": 1.5548502206802368,
      "learning_rate": 0.00019929546695928658,
      "loss": 1.8462,
      "step": 1862
    },
    {
      "epoch": 0.06671799738571454,
      "grad_norm": 1.8258202075958252,
      "learning_rate": 0.000199294091868307,
      "loss": 1.6896,
      "step": 1863
    },
    {
      "epoch": 0.06675380951528283,
      "grad_norm": 1.2990753650665283,
      "learning_rate": 0.0001992927154414535,
      "loss": 1.6857,
      "step": 1864
    },
    {
      "epoch": 0.06678962164485111,
      "grad_norm": 1.2354774475097656,
      "learning_rate": 0.00019929133767874454,
      "loss": 1.5531,
      "step": 1865
    },
    {
      "epoch": 0.0668254337744194,
      "grad_norm": 1.4240732192993164,
      "learning_rate": 0.0001992899585801988,
      "loss": 1.8465,
      "step": 1866
    },
    {
      "epoch": 0.06686124590398768,
      "grad_norm": 1.7457743883132935,
      "learning_rate": 0.0001992885781458347,
      "loss": 1.7631,
      "step": 1867
    },
    {
      "epoch": 0.06689705803355596,
      "grad_norm": 1.6174548864364624,
      "learning_rate": 0.0001992871963756708,
      "loss": 1.5175,
      "step": 1868
    },
    {
      "epoch": 0.06693287016312426,
      "grad_norm": 1.0970457792282104,
      "learning_rate": 0.00019928581326972582,
      "loss": 1.6494,
      "step": 1869
    },
    {
      "epoch": 0.06696868229269254,
      "grad_norm": 1.3978979587554932,
      "learning_rate": 0.00019928442882801825,
      "loss": 1.5092,
      "step": 1870
    },
    {
      "epoch": 0.06700449442226082,
      "grad_norm": 1.5778833627700806,
      "learning_rate": 0.00019928304305056677,
      "loss": 1.7631,
      "step": 1871
    },
    {
      "epoch": 0.0670403065518291,
      "grad_norm": 1.6992747783660889,
      "learning_rate": 0.00019928165593739,
      "loss": 1.8914,
      "step": 1872
    },
    {
      "epoch": 0.06707611868139739,
      "grad_norm": 1.8649605512619019,
      "learning_rate": 0.00019928026748850663,
      "loss": 1.8694,
      "step": 1873
    },
    {
      "epoch": 0.06711193081096567,
      "grad_norm": 1.2663445472717285,
      "learning_rate": 0.00019927887770393533,
      "loss": 1.757,
      "step": 1874
    },
    {
      "epoch": 0.06714774294053395,
      "grad_norm": 1.696965217590332,
      "learning_rate": 0.0001992774865836948,
      "loss": 1.7698,
      "step": 1875
    },
    {
      "epoch": 0.06718355507010225,
      "grad_norm": 2.0695016384124756,
      "learning_rate": 0.0001992760941278037,
      "loss": 1.6721,
      "step": 1876
    },
    {
      "epoch": 0.06721936719967053,
      "grad_norm": 1.6472669839859009,
      "learning_rate": 0.00019927470033628087,
      "loss": 1.5199,
      "step": 1877
    },
    {
      "epoch": 0.06725517932923882,
      "grad_norm": 3.179795742034912,
      "learning_rate": 0.00019927330520914496,
      "loss": 1.4682,
      "step": 1878
    },
    {
      "epoch": 0.0672909914588071,
      "grad_norm": 1.6882466077804565,
      "learning_rate": 0.00019927190874641478,
      "loss": 1.8529,
      "step": 1879
    },
    {
      "epoch": 0.06732680358837538,
      "grad_norm": 1.6953445672988892,
      "learning_rate": 0.00019927051094810913,
      "loss": 1.9795,
      "step": 1880
    },
    {
      "epoch": 0.06736261571794366,
      "grad_norm": 1.3161673545837402,
      "learning_rate": 0.00019926911181424682,
      "loss": 1.7497,
      "step": 1881
    },
    {
      "epoch": 0.06739842784751195,
      "grad_norm": 1.8265464305877686,
      "learning_rate": 0.00019926771134484662,
      "loss": 1.5825,
      "step": 1882
    },
    {
      "epoch": 0.06743423997708024,
      "grad_norm": 2.6041676998138428,
      "learning_rate": 0.00019926630953992746,
      "loss": 1.9153,
      "step": 1883
    },
    {
      "epoch": 0.06747005210664853,
      "grad_norm": 2.21744441986084,
      "learning_rate": 0.00019926490639950812,
      "loss": 1.8852,
      "step": 1884
    },
    {
      "epoch": 0.06750586423621681,
      "grad_norm": 2.1859962940216064,
      "learning_rate": 0.00019926350192360753,
      "loss": 1.441,
      "step": 1885
    },
    {
      "epoch": 0.06754167636578509,
      "grad_norm": 1.6695815324783325,
      "learning_rate": 0.00019926209611224454,
      "loss": 1.5176,
      "step": 1886
    },
    {
      "epoch": 0.06757748849535337,
      "grad_norm": 1.3804771900177002,
      "learning_rate": 0.00019926068896543807,
      "loss": 1.5485,
      "step": 1887
    },
    {
      "epoch": 0.06761330062492166,
      "grad_norm": 1.9427920579910278,
      "learning_rate": 0.0001992592804832071,
      "loss": 1.6327,
      "step": 1888
    },
    {
      "epoch": 0.06764911275448994,
      "grad_norm": 1.2968367338180542,
      "learning_rate": 0.00019925787066557053,
      "loss": 1.7162,
      "step": 1889
    },
    {
      "epoch": 0.06768492488405824,
      "grad_norm": 1.447363257408142,
      "learning_rate": 0.00019925645951254735,
      "loss": 1.6678,
      "step": 1890
    },
    {
      "epoch": 0.06772073701362652,
      "grad_norm": 1.4006659984588623,
      "learning_rate": 0.00019925504702415653,
      "loss": 1.7161,
      "step": 1891
    },
    {
      "epoch": 0.0677565491431948,
      "grad_norm": 1.27907395362854,
      "learning_rate": 0.00019925363320041708,
      "loss": 1.5937,
      "step": 1892
    },
    {
      "epoch": 0.06779236127276309,
      "grad_norm": 1.2862017154693604,
      "learning_rate": 0.00019925221804134805,
      "loss": 1.6961,
      "step": 1893
    },
    {
      "epoch": 0.06782817340233137,
      "grad_norm": 2.0907084941864014,
      "learning_rate": 0.0001992508015469684,
      "loss": 1.5839,
      "step": 1894
    },
    {
      "epoch": 0.06786398553189965,
      "grad_norm": 2.4519996643066406,
      "learning_rate": 0.00019924938371729728,
      "loss": 1.772,
      "step": 1895
    },
    {
      "epoch": 0.06789979766146793,
      "grad_norm": 2.106823205947876,
      "learning_rate": 0.00019924796455235373,
      "loss": 1.9488,
      "step": 1896
    },
    {
      "epoch": 0.06793560979103623,
      "grad_norm": 1.8093961477279663,
      "learning_rate": 0.00019924654405215682,
      "loss": 1.5178,
      "step": 1897
    },
    {
      "epoch": 0.06797142192060451,
      "grad_norm": 2.3577721118927,
      "learning_rate": 0.00019924512221672572,
      "loss": 1.6168,
      "step": 1898
    },
    {
      "epoch": 0.0680072340501728,
      "grad_norm": 1.4707239866256714,
      "learning_rate": 0.00019924369904607945,
      "loss": 1.6463,
      "step": 1899
    },
    {
      "epoch": 0.06804304617974108,
      "grad_norm": 1.9914259910583496,
      "learning_rate": 0.00019924227454023728,
      "loss": 1.6292,
      "step": 1900
    },
    {
      "epoch": 0.06807885830930936,
      "grad_norm": 1.2515363693237305,
      "learning_rate": 0.0001992408486992183,
      "loss": 1.754,
      "step": 1901
    },
    {
      "epoch": 0.06811467043887764,
      "grad_norm": 2.06872296333313,
      "learning_rate": 0.00019923942152304169,
      "loss": 1.3609,
      "step": 1902
    },
    {
      "epoch": 0.06815048256844593,
      "grad_norm": 1.7905656099319458,
      "learning_rate": 0.0001992379930117267,
      "loss": 1.7038,
      "step": 1903
    },
    {
      "epoch": 0.06818629469801422,
      "grad_norm": 1.7369552850723267,
      "learning_rate": 0.00019923656316529252,
      "loss": 1.5613,
      "step": 1904
    },
    {
      "epoch": 0.06822210682758251,
      "grad_norm": 1.4253979921340942,
      "learning_rate": 0.00019923513198375837,
      "loss": 1.53,
      "step": 1905
    },
    {
      "epoch": 0.06825791895715079,
      "grad_norm": 1.3576210737228394,
      "learning_rate": 0.00019923369946714354,
      "loss": 1.9039,
      "step": 1906
    },
    {
      "epoch": 0.06829373108671907,
      "grad_norm": 2.030214548110962,
      "learning_rate": 0.00019923226561546726,
      "loss": 1.5617,
      "step": 1907
    },
    {
      "epoch": 0.06832954321628736,
      "grad_norm": 2.2546305656433105,
      "learning_rate": 0.00019923083042874885,
      "loss": 2.2339,
      "step": 1908
    },
    {
      "epoch": 0.06836535534585564,
      "grad_norm": 2.327280044555664,
      "learning_rate": 0.00019922939390700767,
      "loss": 1.5703,
      "step": 1909
    },
    {
      "epoch": 0.06840116747542392,
      "grad_norm": 2.571589708328247,
      "learning_rate": 0.00019922795605026295,
      "loss": 1.6462,
      "step": 1910
    },
    {
      "epoch": 0.0684369796049922,
      "grad_norm": 1.4403351545333862,
      "learning_rate": 0.00019922651685853407,
      "loss": 1.6363,
      "step": 1911
    },
    {
      "epoch": 0.0684727917345605,
      "grad_norm": 2.2027952671051025,
      "learning_rate": 0.0001992250763318404,
      "loss": 1.6944,
      "step": 1912
    },
    {
      "epoch": 0.06850860386412878,
      "grad_norm": 2.0483181476593018,
      "learning_rate": 0.00019922363447020134,
      "loss": 2.0797,
      "step": 1913
    },
    {
      "epoch": 0.06854441599369707,
      "grad_norm": 1.3529871702194214,
      "learning_rate": 0.00019922219127363624,
      "loss": 1.6436,
      "step": 1914
    },
    {
      "epoch": 0.06858022812326535,
      "grad_norm": 1.9239258766174316,
      "learning_rate": 0.00019922074674216456,
      "loss": 1.5636,
      "step": 1915
    },
    {
      "epoch": 0.06861604025283363,
      "grad_norm": 2.6495797634124756,
      "learning_rate": 0.00019921930087580573,
      "loss": 1.9851,
      "step": 1916
    },
    {
      "epoch": 0.06865185238240192,
      "grad_norm": 1.5434764623641968,
      "learning_rate": 0.00019921785367457917,
      "loss": 1.8326,
      "step": 1917
    },
    {
      "epoch": 0.0686876645119702,
      "grad_norm": 1.602738857269287,
      "learning_rate": 0.00019921640513850437,
      "loss": 1.5884,
      "step": 1918
    },
    {
      "epoch": 0.0687234766415385,
      "grad_norm": 1.9453281164169312,
      "learning_rate": 0.00019921495526760083,
      "loss": 1.6429,
      "step": 1919
    },
    {
      "epoch": 0.06875928877110678,
      "grad_norm": 1.6963716745376587,
      "learning_rate": 0.00019921350406188805,
      "loss": 1.642,
      "step": 1920
    },
    {
      "epoch": 0.06879510090067506,
      "grad_norm": 1.7032421827316284,
      "learning_rate": 0.00019921205152138556,
      "loss": 1.5783,
      "step": 1921
    },
    {
      "epoch": 0.06883091303024334,
      "grad_norm": 1.9125220775604248,
      "learning_rate": 0.00019921059764611284,
      "loss": 1.9889,
      "step": 1922
    },
    {
      "epoch": 0.06886672515981163,
      "grad_norm": 1.6646088361740112,
      "learning_rate": 0.00019920914243608956,
      "loss": 1.7274,
      "step": 1923
    },
    {
      "epoch": 0.06890253728937991,
      "grad_norm": 1.3573722839355469,
      "learning_rate": 0.0001992076858913352,
      "loss": 1.3907,
      "step": 1924
    },
    {
      "epoch": 0.06893834941894819,
      "grad_norm": 1.8801019191741943,
      "learning_rate": 0.0001992062280118694,
      "loss": 1.8737,
      "step": 1925
    },
    {
      "epoch": 0.06897416154851649,
      "grad_norm": 1.404797911643982,
      "learning_rate": 0.0001992047687977118,
      "loss": 1.5543,
      "step": 1926
    },
    {
      "epoch": 0.06900997367808477,
      "grad_norm": 2.4929728507995605,
      "learning_rate": 0.00019920330824888197,
      "loss": 1.8148,
      "step": 1927
    },
    {
      "epoch": 0.06904578580765305,
      "grad_norm": 1.5967156887054443,
      "learning_rate": 0.0001992018463653996,
      "loss": 1.7098,
      "step": 1928
    },
    {
      "epoch": 0.06908159793722134,
      "grad_norm": 1.629167079925537,
      "learning_rate": 0.00019920038314728434,
      "loss": 1.6494,
      "step": 1929
    },
    {
      "epoch": 0.06911741006678962,
      "grad_norm": 1.807207703590393,
      "learning_rate": 0.00019919891859455588,
      "loss": 1.8693,
      "step": 1930
    },
    {
      "epoch": 0.0691532221963579,
      "grad_norm": 1.9468473196029663,
      "learning_rate": 0.00019919745270723395,
      "loss": 1.7264,
      "step": 1931
    },
    {
      "epoch": 0.06918903432592619,
      "grad_norm": 1.7176305055618286,
      "learning_rate": 0.00019919598548533824,
      "loss": 1.8964,
      "step": 1932
    },
    {
      "epoch": 0.06922484645549448,
      "grad_norm": 2.3279569149017334,
      "learning_rate": 0.00019919451692888848,
      "loss": 1.7493,
      "step": 1933
    },
    {
      "epoch": 0.06926065858506276,
      "grad_norm": 2.4404523372650146,
      "learning_rate": 0.00019919304703790446,
      "loss": 1.8917,
      "step": 1934
    },
    {
      "epoch": 0.06929647071463105,
      "grad_norm": 1.455718755722046,
      "learning_rate": 0.00019919157581240596,
      "loss": 1.6557,
      "step": 1935
    },
    {
      "epoch": 0.06933228284419933,
      "grad_norm": 2.637948751449585,
      "learning_rate": 0.00019919010325241275,
      "loss": 1.4312,
      "step": 1936
    },
    {
      "epoch": 0.06936809497376761,
      "grad_norm": 1.4223843812942505,
      "learning_rate": 0.00019918862935794463,
      "loss": 2.0284,
      "step": 1937
    },
    {
      "epoch": 0.0694039071033359,
      "grad_norm": 2.231555461883545,
      "learning_rate": 0.00019918715412902142,
      "loss": 1.9927,
      "step": 1938
    },
    {
      "epoch": 0.06943971923290418,
      "grad_norm": 1.6516687870025635,
      "learning_rate": 0.00019918567756566305,
      "loss": 1.5935,
      "step": 1939
    },
    {
      "epoch": 0.06947553136247248,
      "grad_norm": 1.9006787538528442,
      "learning_rate": 0.0001991841996678893,
      "loss": 1.2902,
      "step": 1940
    },
    {
      "epoch": 0.06951134349204076,
      "grad_norm": 2.0806474685668945,
      "learning_rate": 0.0001991827204357201,
      "loss": 1.3603,
      "step": 1941
    },
    {
      "epoch": 0.06954715562160904,
      "grad_norm": 1.810377836227417,
      "learning_rate": 0.0001991812398691753,
      "loss": 1.6762,
      "step": 1942
    },
    {
      "epoch": 0.06958296775117732,
      "grad_norm": 1.6597081422805786,
      "learning_rate": 0.00019917975796827488,
      "loss": 1.6292,
      "step": 1943
    },
    {
      "epoch": 0.06961877988074561,
      "grad_norm": 1.8197176456451416,
      "learning_rate": 0.00019917827473303875,
      "loss": 1.5692,
      "step": 1944
    },
    {
      "epoch": 0.06965459201031389,
      "grad_norm": 1.2852716445922852,
      "learning_rate": 0.00019917679016348685,
      "loss": 1.6318,
      "step": 1945
    },
    {
      "epoch": 0.06969040413988217,
      "grad_norm": 1.7549954652786255,
      "learning_rate": 0.00019917530425963916,
      "loss": 1.8909,
      "step": 1946
    },
    {
      "epoch": 0.06972621626945047,
      "grad_norm": 1.5758832693099976,
      "learning_rate": 0.0001991738170215157,
      "loss": 1.5813,
      "step": 1947
    },
    {
      "epoch": 0.06976202839901875,
      "grad_norm": 2.270517587661743,
      "learning_rate": 0.00019917232844913644,
      "loss": 1.8209,
      "step": 1948
    },
    {
      "epoch": 0.06979784052858704,
      "grad_norm": 2.2871134281158447,
      "learning_rate": 0.00019917083854252142,
      "loss": 1.7456,
      "step": 1949
    },
    {
      "epoch": 0.06983365265815532,
      "grad_norm": 1.7921063899993896,
      "learning_rate": 0.00019916934730169073,
      "loss": 1.5367,
      "step": 1950
    },
    {
      "epoch": 0.0698694647877236,
      "grad_norm": 1.8753576278686523,
      "learning_rate": 0.00019916785472666435,
      "loss": 1.6617,
      "step": 1951
    },
    {
      "epoch": 0.06990527691729188,
      "grad_norm": 2.1969332695007324,
      "learning_rate": 0.0001991663608174624,
      "loss": 1.4074,
      "step": 1952
    },
    {
      "epoch": 0.06994108904686017,
      "grad_norm": 1.6860895156860352,
      "learning_rate": 0.000199164865574105,
      "loss": 1.9061,
      "step": 1953
    },
    {
      "epoch": 0.06997690117642846,
      "grad_norm": 1.7256300449371338,
      "learning_rate": 0.00019916336899661224,
      "loss": 1.7184,
      "step": 1954
    },
    {
      "epoch": 0.07001271330599675,
      "grad_norm": 1.6683882474899292,
      "learning_rate": 0.00019916187108500428,
      "loss": 1.6855,
      "step": 1955
    },
    {
      "epoch": 0.07004852543556503,
      "grad_norm": 1.431535243988037,
      "learning_rate": 0.00019916037183930122,
      "loss": 1.8098,
      "step": 1956
    },
    {
      "epoch": 0.07008433756513331,
      "grad_norm": 2.2073168754577637,
      "learning_rate": 0.00019915887125952327,
      "loss": 1.5942,
      "step": 1957
    },
    {
      "epoch": 0.0701201496947016,
      "grad_norm": 1.7813090085983276,
      "learning_rate": 0.00019915736934569066,
      "loss": 1.7848,
      "step": 1958
    },
    {
      "epoch": 0.07015596182426988,
      "grad_norm": 1.7625131607055664,
      "learning_rate": 0.0001991558660978235,
      "loss": 1.6697,
      "step": 1959
    },
    {
      "epoch": 0.07019177395383816,
      "grad_norm": 1.319684386253357,
      "learning_rate": 0.0001991543615159421,
      "loss": 1.8004,
      "step": 1960
    },
    {
      "epoch": 0.07022758608340644,
      "grad_norm": 1.4817261695861816,
      "learning_rate": 0.00019915285560006662,
      "loss": 1.7313,
      "step": 1961
    },
    {
      "epoch": 0.07026339821297474,
      "grad_norm": 1.8547471761703491,
      "learning_rate": 0.00019915134835021738,
      "loss": 1.8569,
      "step": 1962
    },
    {
      "epoch": 0.07029921034254302,
      "grad_norm": 1.6080464124679565,
      "learning_rate": 0.00019914983976641466,
      "loss": 1.4447,
      "step": 1963
    },
    {
      "epoch": 0.0703350224721113,
      "grad_norm": 1.452889323234558,
      "learning_rate": 0.00019914832984867874,
      "loss": 1.6103,
      "step": 1964
    },
    {
      "epoch": 0.07037083460167959,
      "grad_norm": 1.8950872421264648,
      "learning_rate": 0.0001991468185970299,
      "loss": 1.7781,
      "step": 1965
    },
    {
      "epoch": 0.07040664673124787,
      "grad_norm": 1.769492745399475,
      "learning_rate": 0.00019914530601148855,
      "loss": 1.708,
      "step": 1966
    },
    {
      "epoch": 0.07044245886081615,
      "grad_norm": 1.1686387062072754,
      "learning_rate": 0.000199143792092075,
      "loss": 1.6729,
      "step": 1967
    },
    {
      "epoch": 0.07047827099038444,
      "grad_norm": 1.3533289432525635,
      "learning_rate": 0.00019914227683880958,
      "loss": 1.6236,
      "step": 1968
    },
    {
      "epoch": 0.07051408311995273,
      "grad_norm": 1.8892587423324585,
      "learning_rate": 0.0001991407602517127,
      "loss": 1.7556,
      "step": 1969
    },
    {
      "epoch": 0.07054989524952102,
      "grad_norm": 1.8320790529251099,
      "learning_rate": 0.00019913924233080482,
      "loss": 1.4439,
      "step": 1970
    },
    {
      "epoch": 0.0705857073790893,
      "grad_norm": 2.168354034423828,
      "learning_rate": 0.0001991377230761063,
      "loss": 1.8053,
      "step": 1971
    },
    {
      "epoch": 0.07062151950865758,
      "grad_norm": 1.4819917678833008,
      "learning_rate": 0.00019913620248763756,
      "loss": 1.6767,
      "step": 1972
    },
    {
      "epoch": 0.07065733163822586,
      "grad_norm": 1.7416330575942993,
      "learning_rate": 0.0001991346805654191,
      "loss": 1.7843,
      "step": 1973
    },
    {
      "epoch": 0.07069314376779415,
      "grad_norm": 2.728843927383423,
      "learning_rate": 0.00019913315730947143,
      "loss": 1.7568,
      "step": 1974
    },
    {
      "epoch": 0.07072895589736243,
      "grad_norm": 1.4552961587905884,
      "learning_rate": 0.00019913163271981495,
      "loss": 1.3882,
      "step": 1975
    },
    {
      "epoch": 0.07076476802693073,
      "grad_norm": 1.9386544227600098,
      "learning_rate": 0.00019913010679647027,
      "loss": 1.5637,
      "step": 1976
    },
    {
      "epoch": 0.07080058015649901,
      "grad_norm": 3.1994082927703857,
      "learning_rate": 0.00019912857953945784,
      "loss": 1.7099,
      "step": 1977
    },
    {
      "epoch": 0.07083639228606729,
      "grad_norm": 1.4596532583236694,
      "learning_rate": 0.00019912705094879827,
      "loss": 1.5022,
      "step": 1978
    },
    {
      "epoch": 0.07087220441563558,
      "grad_norm": 1.96254563331604,
      "learning_rate": 0.00019912552102451206,
      "loss": 1.6754,
      "step": 1979
    },
    {
      "epoch": 0.07090801654520386,
      "grad_norm": 1.8212988376617432,
      "learning_rate": 0.00019912398976661984,
      "loss": 1.6848,
      "step": 1980
    },
    {
      "epoch": 0.07094382867477214,
      "grad_norm": 1.8282296657562256,
      "learning_rate": 0.0001991224571751422,
      "loss": 1.6715,
      "step": 1981
    },
    {
      "epoch": 0.07097964080434042,
      "grad_norm": 2.6742913722991943,
      "learning_rate": 0.00019912092325009975,
      "loss": 1.4798,
      "step": 1982
    },
    {
      "epoch": 0.07101545293390872,
      "grad_norm": 2.115812063217163,
      "learning_rate": 0.00019911938799151315,
      "loss": 1.8264,
      "step": 1983
    },
    {
      "epoch": 0.071051265063477,
      "grad_norm": 1.2611589431762695,
      "learning_rate": 0.00019911785139940303,
      "loss": 1.6957,
      "step": 1984
    },
    {
      "epoch": 0.07108707719304529,
      "grad_norm": 1.0900263786315918,
      "learning_rate": 0.00019911631347379008,
      "loss": 1.563,
      "step": 1985
    },
    {
      "epoch": 0.07112288932261357,
      "grad_norm": 1.8478407859802246,
      "learning_rate": 0.00019911477421469495,
      "loss": 1.6451,
      "step": 1986
    },
    {
      "epoch": 0.07115870145218185,
      "grad_norm": 2.0498433113098145,
      "learning_rate": 0.0001991132336221384,
      "loss": 1.4951,
      "step": 1987
    },
    {
      "epoch": 0.07119451358175014,
      "grad_norm": 1.6745867729187012,
      "learning_rate": 0.00019911169169614117,
      "loss": 1.7767,
      "step": 1988
    },
    {
      "epoch": 0.07123032571131842,
      "grad_norm": 1.856895923614502,
      "learning_rate": 0.00019911014843672394,
      "loss": 1.5433,
      "step": 1989
    },
    {
      "epoch": 0.07126613784088671,
      "grad_norm": 1.378844976425171,
      "learning_rate": 0.0001991086038439075,
      "loss": 1.7412,
      "step": 1990
    },
    {
      "epoch": 0.071301949970455,
      "grad_norm": 1.3555171489715576,
      "learning_rate": 0.00019910705791771263,
      "loss": 1.5704,
      "step": 1991
    },
    {
      "epoch": 0.07133776210002328,
      "grad_norm": 1.2136765718460083,
      "learning_rate": 0.00019910551065816017,
      "loss": 1.6872,
      "step": 1992
    },
    {
      "epoch": 0.07137357422959156,
      "grad_norm": 2.4193828105926514,
      "learning_rate": 0.0001991039620652709,
      "loss": 1.693,
      "step": 1993
    },
    {
      "epoch": 0.07140938635915985,
      "grad_norm": 1.5262309312820435,
      "learning_rate": 0.00019910241213906565,
      "loss": 1.6757,
      "step": 1994
    },
    {
      "epoch": 0.07144519848872813,
      "grad_norm": 1.5123639106750488,
      "learning_rate": 0.00019910086087956527,
      "loss": 1.556,
      "step": 1995
    },
    {
      "epoch": 0.07148101061829641,
      "grad_norm": 1.8455959558486938,
      "learning_rate": 0.00019909930828679063,
      "loss": 1.3509,
      "step": 1996
    },
    {
      "epoch": 0.07151682274786471,
      "grad_norm": 1.946921706199646,
      "learning_rate": 0.0001990977543607626,
      "loss": 1.7793,
      "step": 1997
    },
    {
      "epoch": 0.07155263487743299,
      "grad_norm": 2.3060195446014404,
      "learning_rate": 0.00019909619910150216,
      "loss": 1.8093,
      "step": 1998
    },
    {
      "epoch": 0.07158844700700127,
      "grad_norm": 1.4833565950393677,
      "learning_rate": 0.0001990946425090302,
      "loss": 1.5529,
      "step": 1999
    },
    {
      "epoch": 0.07162425913656956,
      "grad_norm": 1.3298144340515137,
      "learning_rate": 0.00019909308458336759,
      "loss": 1.7373,
      "step": 2000
    },
    {
      "epoch": 0.07166007126613784,
      "grad_norm": 1.6872481107711792,
      "learning_rate": 0.0001990915253245354,
      "loss": 1.7242,
      "step": 2001
    },
    {
      "epoch": 0.07169588339570612,
      "grad_norm": 1.7926667928695679,
      "learning_rate": 0.0001990899647325545,
      "loss": 1.6452,
      "step": 2002
    },
    {
      "epoch": 0.0717316955252744,
      "grad_norm": 1.7860251665115356,
      "learning_rate": 0.000199088402807446,
      "loss": 1.6085,
      "step": 2003
    },
    {
      "epoch": 0.0717675076548427,
      "grad_norm": 2.7218711376190186,
      "learning_rate": 0.00019908683954923082,
      "loss": 1.9528,
      "step": 2004
    },
    {
      "epoch": 0.07180331978441098,
      "grad_norm": 2.7485177516937256,
      "learning_rate": 0.00019908527495793004,
      "loss": 1.5438,
      "step": 2005
    },
    {
      "epoch": 0.07183913191397927,
      "grad_norm": 2.6853740215301514,
      "learning_rate": 0.00019908370903356468,
      "loss": 1.7663,
      "step": 2006
    },
    {
      "epoch": 0.07187494404354755,
      "grad_norm": 1.6789944171905518,
      "learning_rate": 0.00019908214177615584,
      "loss": 1.919,
      "step": 2007
    },
    {
      "epoch": 0.07191075617311583,
      "grad_norm": 2.4104981422424316,
      "learning_rate": 0.00019908057318572458,
      "loss": 1.3113,
      "step": 2008
    },
    {
      "epoch": 0.07194656830268412,
      "grad_norm": 1.2793664932250977,
      "learning_rate": 0.000199079003262292,
      "loss": 1.4809,
      "step": 2009
    },
    {
      "epoch": 0.0719823804322524,
      "grad_norm": 1.5413146018981934,
      "learning_rate": 0.00019907743200587926,
      "loss": 1.5148,
      "step": 2010
    },
    {
      "epoch": 0.07201819256182068,
      "grad_norm": 1.1401458978652954,
      "learning_rate": 0.00019907585941650747,
      "loss": 1.733,
      "step": 2011
    },
    {
      "epoch": 0.07205400469138898,
      "grad_norm": 1.4969557523727417,
      "learning_rate": 0.00019907428549419777,
      "loss": 1.7206,
      "step": 2012
    },
    {
      "epoch": 0.07208981682095726,
      "grad_norm": 1.5214531421661377,
      "learning_rate": 0.00019907271023897138,
      "loss": 1.8043,
      "step": 2013
    },
    {
      "epoch": 0.07212562895052554,
      "grad_norm": 1.6244412660598755,
      "learning_rate": 0.00019907113365084947,
      "loss": 1.7689,
      "step": 2014
    },
    {
      "epoch": 0.07216144108009383,
      "grad_norm": 1.6461654901504517,
      "learning_rate": 0.0001990695557298532,
      "loss": 1.5748,
      "step": 2015
    },
    {
      "epoch": 0.07219725320966211,
      "grad_norm": 1.6046760082244873,
      "learning_rate": 0.0001990679764760039,
      "loss": 1.8966,
      "step": 2016
    },
    {
      "epoch": 0.07223306533923039,
      "grad_norm": 1.2768783569335938,
      "learning_rate": 0.00019906639588932276,
      "loss": 1.6619,
      "step": 2017
    },
    {
      "epoch": 0.07226887746879868,
      "grad_norm": 1.3331384658813477,
      "learning_rate": 0.00019906481396983103,
      "loss": 1.6712,
      "step": 2018
    },
    {
      "epoch": 0.07230468959836697,
      "grad_norm": 1.4289017915725708,
      "learning_rate": 0.00019906323071755005,
      "loss": 1.7709,
      "step": 2019
    },
    {
      "epoch": 0.07234050172793526,
      "grad_norm": 1.8819347620010376,
      "learning_rate": 0.00019906164613250104,
      "loss": 1.9157,
      "step": 2020
    },
    {
      "epoch": 0.07237631385750354,
      "grad_norm": 1.380541443824768,
      "learning_rate": 0.00019906006021470538,
      "loss": 1.5728,
      "step": 2021
    },
    {
      "epoch": 0.07241212598707182,
      "grad_norm": 1.3870627880096436,
      "learning_rate": 0.00019905847296418437,
      "loss": 1.6477,
      "step": 2022
    },
    {
      "epoch": 0.0724479381166401,
      "grad_norm": 1.8019845485687256,
      "learning_rate": 0.0001990568843809594,
      "loss": 1.4675,
      "step": 2023
    },
    {
      "epoch": 0.07248375024620839,
      "grad_norm": 1.3413184881210327,
      "learning_rate": 0.00019905529446505183,
      "loss": 1.588,
      "step": 2024
    },
    {
      "epoch": 0.07251956237577667,
      "grad_norm": 1.5961971282958984,
      "learning_rate": 0.00019905370321648302,
      "loss": 1.6738,
      "step": 2025
    },
    {
      "epoch": 0.07255537450534497,
      "grad_norm": 2.0965704917907715,
      "learning_rate": 0.00019905211063527442,
      "loss": 1.5145,
      "step": 2026
    },
    {
      "epoch": 0.07259118663491325,
      "grad_norm": 1.2881782054901123,
      "learning_rate": 0.00019905051672144746,
      "loss": 1.6991,
      "step": 2027
    },
    {
      "epoch": 0.07262699876448153,
      "grad_norm": 1.2879416942596436,
      "learning_rate": 0.00019904892147502352,
      "loss": 1.795,
      "step": 2028
    },
    {
      "epoch": 0.07266281089404981,
      "grad_norm": 1.2281581163406372,
      "learning_rate": 0.00019904732489602417,
      "loss": 1.5426,
      "step": 2029
    },
    {
      "epoch": 0.0726986230236181,
      "grad_norm": 1.7174360752105713,
      "learning_rate": 0.00019904572698447077,
      "loss": 1.6864,
      "step": 2030
    },
    {
      "epoch": 0.07273443515318638,
      "grad_norm": 1.7847421169281006,
      "learning_rate": 0.00019904412774038487,
      "loss": 1.6732,
      "step": 2031
    },
    {
      "epoch": 0.07277024728275466,
      "grad_norm": 1.5283077955245972,
      "learning_rate": 0.000199042527163788,
      "loss": 1.7374,
      "step": 2032
    },
    {
      "epoch": 0.07280605941232296,
      "grad_norm": 1.7698839902877808,
      "learning_rate": 0.0001990409252547017,
      "loss": 1.8254,
      "step": 2033
    },
    {
      "epoch": 0.07284187154189124,
      "grad_norm": 1.3195661306381226,
      "learning_rate": 0.0001990393220131475,
      "loss": 1.8756,
      "step": 2034
    },
    {
      "epoch": 0.07287768367145953,
      "grad_norm": 1.8096301555633545,
      "learning_rate": 0.00019903771743914696,
      "loss": 1.4076,
      "step": 2035
    },
    {
      "epoch": 0.07291349580102781,
      "grad_norm": 1.5130648612976074,
      "learning_rate": 0.00019903611153272168,
      "loss": 1.2862,
      "step": 2036
    },
    {
      "epoch": 0.07294930793059609,
      "grad_norm": 2.1466357707977295,
      "learning_rate": 0.0001990345042938933,
      "loss": 1.6471,
      "step": 2037
    },
    {
      "epoch": 0.07298512006016437,
      "grad_norm": 1.3187168836593628,
      "learning_rate": 0.00019903289572268336,
      "loss": 1.6238,
      "step": 2038
    },
    {
      "epoch": 0.07302093218973266,
      "grad_norm": 1.909706473350525,
      "learning_rate": 0.0001990312858191136,
      "loss": 1.6599,
      "step": 2039
    },
    {
      "epoch": 0.07305674431930095,
      "grad_norm": 1.4692286252975464,
      "learning_rate": 0.0001990296745832056,
      "loss": 1.7824,
      "step": 2040
    },
    {
      "epoch": 0.07309255644886924,
      "grad_norm": 1.2459222078323364,
      "learning_rate": 0.00019902806201498106,
      "loss": 1.7793,
      "step": 2041
    },
    {
      "epoch": 0.07312836857843752,
      "grad_norm": 1.560295581817627,
      "learning_rate": 0.0001990264481144617,
      "loss": 1.7668,
      "step": 2042
    },
    {
      "epoch": 0.0731641807080058,
      "grad_norm": 1.6913843154907227,
      "learning_rate": 0.00019902483288166922,
      "loss": 1.6637,
      "step": 2043
    },
    {
      "epoch": 0.07319999283757408,
      "grad_norm": 1.3807941675186157,
      "learning_rate": 0.00019902321631662533,
      "loss": 1.5641,
      "step": 2044
    },
    {
      "epoch": 0.07323580496714237,
      "grad_norm": 2.5262813568115234,
      "learning_rate": 0.0001990215984193518,
      "loss": 1.4331,
      "step": 2045
    },
    {
      "epoch": 0.07327161709671065,
      "grad_norm": 1.294885277748108,
      "learning_rate": 0.00019901997918987042,
      "loss": 1.7085,
      "step": 2046
    },
    {
      "epoch": 0.07330742922627895,
      "grad_norm": 1.254361867904663,
      "learning_rate": 0.0001990183586282029,
      "loss": 1.5558,
      "step": 2047
    },
    {
      "epoch": 0.07334324135584723,
      "grad_norm": 1.6477164030075073,
      "learning_rate": 0.00019901673673437112,
      "loss": 1.6353,
      "step": 2048
    },
    {
      "epoch": 0.07337905348541551,
      "grad_norm": 2.1215851306915283,
      "learning_rate": 0.00019901511350839686,
      "loss": 2.1082,
      "step": 2049
    },
    {
      "epoch": 0.0734148656149838,
      "grad_norm": 1.359099268913269,
      "learning_rate": 0.00019901348895030196,
      "loss": 1.8811,
      "step": 2050
    },
    {
      "epoch": 0.07345067774455208,
      "grad_norm": 1.2427027225494385,
      "learning_rate": 0.0001990118630601083,
      "loss": 1.7966,
      "step": 2051
    },
    {
      "epoch": 0.07348648987412036,
      "grad_norm": 2.6582934856414795,
      "learning_rate": 0.00019901023583783776,
      "loss": 1.7436,
      "step": 2052
    },
    {
      "epoch": 0.07352230200368864,
      "grad_norm": 2.8324716091156006,
      "learning_rate": 0.00019900860728351216,
      "loss": 1.4201,
      "step": 2053
    },
    {
      "epoch": 0.07355811413325694,
      "grad_norm": 1.3952288627624512,
      "learning_rate": 0.00019900697739715347,
      "loss": 1.7557,
      "step": 2054
    },
    {
      "epoch": 0.07359392626282522,
      "grad_norm": 1.437296748161316,
      "learning_rate": 0.00019900534617878365,
      "loss": 1.7427,
      "step": 2055
    },
    {
      "epoch": 0.0736297383923935,
      "grad_norm": 2.397494077682495,
      "learning_rate": 0.0001990037136284246,
      "loss": 1.6845,
      "step": 2056
    },
    {
      "epoch": 0.07366555052196179,
      "grad_norm": 1.7484171390533447,
      "learning_rate": 0.00019900207974609822,
      "loss": 1.5456,
      "step": 2057
    },
    {
      "epoch": 0.07370136265153007,
      "grad_norm": 1.8446656465530396,
      "learning_rate": 0.00019900044453182662,
      "loss": 1.7223,
      "step": 2058
    },
    {
      "epoch": 0.07373717478109836,
      "grad_norm": 1.7288898229599,
      "learning_rate": 0.00019899880798563172,
      "loss": 1.6006,
      "step": 2059
    },
    {
      "epoch": 0.07377298691066664,
      "grad_norm": 2.0306966304779053,
      "learning_rate": 0.00019899717010753558,
      "loss": 1.6236,
      "step": 2060
    },
    {
      "epoch": 0.07380879904023492,
      "grad_norm": 1.7749613523483276,
      "learning_rate": 0.0001989955308975602,
      "loss": 1.7825,
      "step": 2061
    },
    {
      "epoch": 0.07384461116980322,
      "grad_norm": 1.8691487312316895,
      "learning_rate": 0.00019899389035572763,
      "loss": 1.8369,
      "step": 2062
    },
    {
      "epoch": 0.0738804232993715,
      "grad_norm": 1.8923448324203491,
      "learning_rate": 0.00019899224848205998,
      "loss": 1.7227,
      "step": 2063
    },
    {
      "epoch": 0.07391623542893978,
      "grad_norm": 1.278397798538208,
      "learning_rate": 0.0001989906052765793,
      "loss": 1.5775,
      "step": 2064
    },
    {
      "epoch": 0.07395204755850807,
      "grad_norm": 1.4953033924102783,
      "learning_rate": 0.00019898896073930776,
      "loss": 1.7734,
      "step": 2065
    },
    {
      "epoch": 0.07398785968807635,
      "grad_norm": 1.5151599645614624,
      "learning_rate": 0.00019898731487026742,
      "loss": 1.6955,
      "step": 2066
    },
    {
      "epoch": 0.07402367181764463,
      "grad_norm": 1.816370964050293,
      "learning_rate": 0.00019898566766948038,
      "loss": 1.5642,
      "step": 2067
    },
    {
      "epoch": 0.07405948394721291,
      "grad_norm": 1.5039783716201782,
      "learning_rate": 0.00019898401913696892,
      "loss": 1.492,
      "step": 2068
    },
    {
      "epoch": 0.07409529607678121,
      "grad_norm": 1.5567251443862915,
      "learning_rate": 0.00019898236927275517,
      "loss": 1.8026,
      "step": 2069
    },
    {
      "epoch": 0.0741311082063495,
      "grad_norm": 2.611471176147461,
      "learning_rate": 0.0001989807180768613,
      "loss": 1.8471,
      "step": 2070
    },
    {
      "epoch": 0.07416692033591778,
      "grad_norm": 1.6151888370513916,
      "learning_rate": 0.00019897906554930956,
      "loss": 1.6592,
      "step": 2071
    },
    {
      "epoch": 0.07420273246548606,
      "grad_norm": 2.0822031497955322,
      "learning_rate": 0.00019897741169012213,
      "loss": 1.8254,
      "step": 2072
    },
    {
      "epoch": 0.07423854459505434,
      "grad_norm": 1.3963028192520142,
      "learning_rate": 0.00019897575649932135,
      "loss": 1.6788,
      "step": 2073
    },
    {
      "epoch": 0.07427435672462263,
      "grad_norm": 1.506298542022705,
      "learning_rate": 0.0001989740999769294,
      "loss": 1.6061,
      "step": 2074
    },
    {
      "epoch": 0.07431016885419091,
      "grad_norm": 1.5757018327713013,
      "learning_rate": 0.0001989724421229686,
      "loss": 1.8792,
      "step": 2075
    },
    {
      "epoch": 0.0743459809837592,
      "grad_norm": 1.4515763521194458,
      "learning_rate": 0.00019897078293746128,
      "loss": 1.7142,
      "step": 2076
    },
    {
      "epoch": 0.07438179311332749,
      "grad_norm": 2.8991751670837402,
      "learning_rate": 0.0001989691224204297,
      "loss": 1.757,
      "step": 2077
    },
    {
      "epoch": 0.07441760524289577,
      "grad_norm": 1.243881344795227,
      "learning_rate": 0.0001989674605718963,
      "loss": 1.6967,
      "step": 2078
    },
    {
      "epoch": 0.07445341737246405,
      "grad_norm": 2.1475493907928467,
      "learning_rate": 0.00019896579739188335,
      "loss": 1.7954,
      "step": 2079
    },
    {
      "epoch": 0.07448922950203234,
      "grad_norm": 1.5451580286026,
      "learning_rate": 0.00019896413288041323,
      "loss": 1.6241,
      "step": 2080
    },
    {
      "epoch": 0.07452504163160062,
      "grad_norm": 1.7768465280532837,
      "learning_rate": 0.00019896246703750837,
      "loss": 1.6177,
      "step": 2081
    },
    {
      "epoch": 0.0745608537611689,
      "grad_norm": 1.5759029388427734,
      "learning_rate": 0.00019896079986319118,
      "loss": 1.8915,
      "step": 2082
    },
    {
      "epoch": 0.0745966658907372,
      "grad_norm": 1.9887797832489014,
      "learning_rate": 0.00019895913135748407,
      "loss": 1.6981,
      "step": 2083
    },
    {
      "epoch": 0.07463247802030548,
      "grad_norm": 1.390030860900879,
      "learning_rate": 0.0001989574615204095,
      "loss": 1.5878,
      "step": 2084
    },
    {
      "epoch": 0.07466829014987376,
      "grad_norm": 1.6922458410263062,
      "learning_rate": 0.0001989557903519899,
      "loss": 1.4414,
      "step": 2085
    },
    {
      "epoch": 0.07470410227944205,
      "grad_norm": 2.327239990234375,
      "learning_rate": 0.0001989541178522478,
      "loss": 1.753,
      "step": 2086
    },
    {
      "epoch": 0.07473991440901033,
      "grad_norm": 1.5192580223083496,
      "learning_rate": 0.0001989524440212057,
      "loss": 1.5827,
      "step": 2087
    },
    {
      "epoch": 0.07477572653857861,
      "grad_norm": 1.4516956806182861,
      "learning_rate": 0.00019895076885888613,
      "loss": 1.4345,
      "step": 2088
    },
    {
      "epoch": 0.0748115386681469,
      "grad_norm": 1.7795008420944214,
      "learning_rate": 0.00019894909236531158,
      "loss": 1.3601,
      "step": 2089
    },
    {
      "epoch": 0.07484735079771519,
      "grad_norm": 1.4862284660339355,
      "learning_rate": 0.0001989474145405046,
      "loss": 1.6291,
      "step": 2090
    },
    {
      "epoch": 0.07488316292728348,
      "grad_norm": 1.9060381650924683,
      "learning_rate": 0.00019894573538448783,
      "loss": 1.9217,
      "step": 2091
    },
    {
      "epoch": 0.07491897505685176,
      "grad_norm": 1.4660017490386963,
      "learning_rate": 0.0001989440548972838,
      "loss": 1.5406,
      "step": 2092
    },
    {
      "epoch": 0.07495478718642004,
      "grad_norm": 1.3798494338989258,
      "learning_rate": 0.0001989423730789151,
      "loss": 1.4285,
      "step": 2093
    },
    {
      "epoch": 0.07499059931598832,
      "grad_norm": 1.282052755355835,
      "learning_rate": 0.00019894068992940448,
      "loss": 1.7026,
      "step": 2094
    },
    {
      "epoch": 0.0750264114455566,
      "grad_norm": 3.1386232376098633,
      "learning_rate": 0.00019893900544877443,
      "loss": 1.7159,
      "step": 2095
    },
    {
      "epoch": 0.07506222357512489,
      "grad_norm": 1.8487104177474976,
      "learning_rate": 0.00019893731963704773,
      "loss": 1.4776,
      "step": 2096
    },
    {
      "epoch": 0.07509803570469319,
      "grad_norm": 1.7013107538223267,
      "learning_rate": 0.000198935632494247,
      "loss": 1.9181,
      "step": 2097
    },
    {
      "epoch": 0.07513384783426147,
      "grad_norm": 1.594666600227356,
      "learning_rate": 0.00019893394402039496,
      "loss": 1.7788,
      "step": 2098
    },
    {
      "epoch": 0.07516965996382975,
      "grad_norm": 1.4459965229034424,
      "learning_rate": 0.00019893225421551428,
      "loss": 1.6597,
      "step": 2099
    },
    {
      "epoch": 0.07520547209339803,
      "grad_norm": 2.2312240600585938,
      "learning_rate": 0.0001989305630796278,
      "loss": 1.5934,
      "step": 2100
    },
    {
      "epoch": 0.07524128422296632,
      "grad_norm": 1.4842166900634766,
      "learning_rate": 0.00019892887061275815,
      "loss": 1.7209,
      "step": 2101
    },
    {
      "epoch": 0.0752770963525346,
      "grad_norm": 2.3655948638916016,
      "learning_rate": 0.00019892717681492815,
      "loss": 1.5201,
      "step": 2102
    },
    {
      "epoch": 0.07531290848210288,
      "grad_norm": 1.5361484289169312,
      "learning_rate": 0.00019892548168616063,
      "loss": 1.6127,
      "step": 2103
    },
    {
      "epoch": 0.07534872061167118,
      "grad_norm": 2.2546679973602295,
      "learning_rate": 0.00019892378522647834,
      "loss": 1.3852,
      "step": 2104
    },
    {
      "epoch": 0.07538453274123946,
      "grad_norm": 1.4017812013626099,
      "learning_rate": 0.00019892208743590412,
      "loss": 1.4618,
      "step": 2105
    },
    {
      "epoch": 0.07542034487080775,
      "grad_norm": 1.399861216545105,
      "learning_rate": 0.00019892038831446085,
      "loss": 1.2668,
      "step": 2106
    },
    {
      "epoch": 0.07545615700037603,
      "grad_norm": 1.8196576833724976,
      "learning_rate": 0.0001989186878621713,
      "loss": 1.7222,
      "step": 2107
    },
    {
      "epoch": 0.07549196912994431,
      "grad_norm": 1.6640617847442627,
      "learning_rate": 0.00019891698607905843,
      "loss": 1.5111,
      "step": 2108
    },
    {
      "epoch": 0.0755277812595126,
      "grad_norm": 1.4460179805755615,
      "learning_rate": 0.0001989152829651451,
      "loss": 1.7697,
      "step": 2109
    },
    {
      "epoch": 0.07556359338908088,
      "grad_norm": 2.077554941177368,
      "learning_rate": 0.00019891357852045422,
      "loss": 1.5551,
      "step": 2110
    },
    {
      "epoch": 0.07559940551864916,
      "grad_norm": 2.658271312713623,
      "learning_rate": 0.00019891187274500874,
      "loss": 1.9367,
      "step": 2111
    },
    {
      "epoch": 0.07563521764821746,
      "grad_norm": 1.6757729053497314,
      "learning_rate": 0.0001989101656388316,
      "loss": 1.5789,
      "step": 2112
    },
    {
      "epoch": 0.07567102977778574,
      "grad_norm": 1.1914464235305786,
      "learning_rate": 0.00019890845720194576,
      "loss": 1.6118,
      "step": 2113
    },
    {
      "epoch": 0.07570684190735402,
      "grad_norm": 1.547780990600586,
      "learning_rate": 0.00019890674743437424,
      "loss": 1.8614,
      "step": 2114
    },
    {
      "epoch": 0.0757426540369223,
      "grad_norm": 2.2313425540924072,
      "learning_rate": 0.00019890503633614,
      "loss": 1.5434,
      "step": 2115
    },
    {
      "epoch": 0.07577846616649059,
      "grad_norm": 1.3298251628875732,
      "learning_rate": 0.00019890332390726606,
      "loss": 1.6958,
      "step": 2116
    },
    {
      "epoch": 0.07581427829605887,
      "grad_norm": 1.4307975769042969,
      "learning_rate": 0.00019890161014777546,
      "loss": 1.4825,
      "step": 2117
    },
    {
      "epoch": 0.07585009042562715,
      "grad_norm": 3.3594889640808105,
      "learning_rate": 0.0001988998950576913,
      "loss": 1.695,
      "step": 2118
    },
    {
      "epoch": 0.07588590255519545,
      "grad_norm": 1.8578377962112427,
      "learning_rate": 0.00019889817863703662,
      "loss": 1.7816,
      "step": 2119
    },
    {
      "epoch": 0.07592171468476373,
      "grad_norm": 1.5581252574920654,
      "learning_rate": 0.0001988964608858345,
      "loss": 1.889,
      "step": 2120
    },
    {
      "epoch": 0.07595752681433202,
      "grad_norm": 1.5691393613815308,
      "learning_rate": 0.00019889474180410805,
      "loss": 1.5541,
      "step": 2121
    },
    {
      "epoch": 0.0759933389439003,
      "grad_norm": 1.5680208206176758,
      "learning_rate": 0.00019889302139188044,
      "loss": 1.5807,
      "step": 2122
    },
    {
      "epoch": 0.07602915107346858,
      "grad_norm": 1.4914289712905884,
      "learning_rate": 0.00019889129964917478,
      "loss": 1.8856,
      "step": 2123
    },
    {
      "epoch": 0.07606496320303686,
      "grad_norm": 1.8300025463104248,
      "learning_rate": 0.00019888957657601425,
      "loss": 1.6327,
      "step": 2124
    },
    {
      "epoch": 0.07610077533260515,
      "grad_norm": 1.6856653690338135,
      "learning_rate": 0.00019888785217242206,
      "loss": 1.5229,
      "step": 2125
    },
    {
      "epoch": 0.07613658746217344,
      "grad_norm": 1.4205312728881836,
      "learning_rate": 0.00019888612643842132,
      "loss": 1.5932,
      "step": 2126
    },
    {
      "epoch": 0.07617239959174173,
      "grad_norm": 1.4903684854507446,
      "learning_rate": 0.00019888439937403534,
      "loss": 1.7948,
      "step": 2127
    },
    {
      "epoch": 0.07620821172131001,
      "grad_norm": 1.770424485206604,
      "learning_rate": 0.0001988826709792873,
      "loss": 1.8025,
      "step": 2128
    },
    {
      "epoch": 0.07624402385087829,
      "grad_norm": 1.9164286851882935,
      "learning_rate": 0.00019888094125420044,
      "loss": 1.7622,
      "step": 2129
    },
    {
      "epoch": 0.07627983598044658,
      "grad_norm": 2.1902430057525635,
      "learning_rate": 0.00019887921019879812,
      "loss": 1.94,
      "step": 2130
    },
    {
      "epoch": 0.07631564811001486,
      "grad_norm": 1.942354679107666,
      "learning_rate": 0.00019887747781310356,
      "loss": 1.6829,
      "step": 2131
    },
    {
      "epoch": 0.07635146023958314,
      "grad_norm": 2.437299966812134,
      "learning_rate": 0.00019887574409714005,
      "loss": 1.8157,
      "step": 2132
    },
    {
      "epoch": 0.07638727236915144,
      "grad_norm": 1.7092030048370361,
      "learning_rate": 0.00019887400905093096,
      "loss": 1.8297,
      "step": 2133
    },
    {
      "epoch": 0.07642308449871972,
      "grad_norm": 1.8825984001159668,
      "learning_rate": 0.00019887227267449963,
      "loss": 1.739,
      "step": 2134
    },
    {
      "epoch": 0.076458896628288,
      "grad_norm": 1.9578901529312134,
      "learning_rate": 0.00019887053496786937,
      "loss": 1.772,
      "step": 2135
    },
    {
      "epoch": 0.07649470875785629,
      "grad_norm": 0.9991260170936584,
      "learning_rate": 0.00019886879593106365,
      "loss": 1.3756,
      "step": 2136
    },
    {
      "epoch": 0.07653052088742457,
      "grad_norm": 1.4741476774215698,
      "learning_rate": 0.00019886705556410576,
      "loss": 1.4068,
      "step": 2137
    },
    {
      "epoch": 0.07656633301699285,
      "grad_norm": 1.7536135911941528,
      "learning_rate": 0.0001988653138670192,
      "loss": 1.7613,
      "step": 2138
    },
    {
      "epoch": 0.07660214514656113,
      "grad_norm": 1.7891243696212769,
      "learning_rate": 0.00019886357083982734,
      "loss": 1.5153,
      "step": 2139
    },
    {
      "epoch": 0.07663795727612943,
      "grad_norm": 1.2617411613464355,
      "learning_rate": 0.0001988618264825537,
      "loss": 1.597,
      "step": 2140
    },
    {
      "epoch": 0.07667376940569771,
      "grad_norm": 1.4487390518188477,
      "learning_rate": 0.00019886008079522167,
      "loss": 1.6301,
      "step": 2141
    },
    {
      "epoch": 0.076709581535266,
      "grad_norm": 1.4209492206573486,
      "learning_rate": 0.0001988583337778548,
      "loss": 1.6636,
      "step": 2142
    },
    {
      "epoch": 0.07674539366483428,
      "grad_norm": 1.6687378883361816,
      "learning_rate": 0.00019885658543047655,
      "loss": 1.7059,
      "step": 2143
    },
    {
      "epoch": 0.07678120579440256,
      "grad_norm": 1.7952252626419067,
      "learning_rate": 0.00019885483575311045,
      "loss": 2.0421,
      "step": 2144
    },
    {
      "epoch": 0.07681701792397085,
      "grad_norm": 1.4075217247009277,
      "learning_rate": 0.00019885308474578008,
      "loss": 1.5278,
      "step": 2145
    },
    {
      "epoch": 0.07685283005353913,
      "grad_norm": 1.9520444869995117,
      "learning_rate": 0.00019885133240850892,
      "loss": 1.9848,
      "step": 2146
    },
    {
      "epoch": 0.07688864218310743,
      "grad_norm": 2.3137176036834717,
      "learning_rate": 0.00019884957874132065,
      "loss": 2.0316,
      "step": 2147
    },
    {
      "epoch": 0.07692445431267571,
      "grad_norm": 1.3930728435516357,
      "learning_rate": 0.00019884782374423877,
      "loss": 1.6199,
      "step": 2148
    },
    {
      "epoch": 0.07696026644224399,
      "grad_norm": 1.7720856666564941,
      "learning_rate": 0.00019884606741728692,
      "loss": 1.4512,
      "step": 2149
    },
    {
      "epoch": 0.07699607857181227,
      "grad_norm": 1.3794426918029785,
      "learning_rate": 0.00019884430976048877,
      "loss": 1.8775,
      "step": 2150
    },
    {
      "epoch": 0.07703189070138056,
      "grad_norm": 1.3538085222244263,
      "learning_rate": 0.00019884255077386788,
      "loss": 1.4677,
      "step": 2151
    },
    {
      "epoch": 0.07706770283094884,
      "grad_norm": 1.767635703086853,
      "learning_rate": 0.000198840790457448,
      "loss": 1.4348,
      "step": 2152
    },
    {
      "epoch": 0.07710351496051712,
      "grad_norm": 1.4579626321792603,
      "learning_rate": 0.00019883902881125278,
      "loss": 1.756,
      "step": 2153
    },
    {
      "epoch": 0.0771393270900854,
      "grad_norm": 1.3255012035369873,
      "learning_rate": 0.00019883726583530594,
      "loss": 1.6512,
      "step": 2154
    },
    {
      "epoch": 0.0771751392196537,
      "grad_norm": 1.5879746675491333,
      "learning_rate": 0.00019883550152963113,
      "loss": 1.6947,
      "step": 2155
    },
    {
      "epoch": 0.07721095134922198,
      "grad_norm": 1.8822317123413086,
      "learning_rate": 0.00019883373589425215,
      "loss": 1.702,
      "step": 2156
    },
    {
      "epoch": 0.07724676347879027,
      "grad_norm": 2.246804714202881,
      "learning_rate": 0.00019883196892919275,
      "loss": 1.6123,
      "step": 2157
    },
    {
      "epoch": 0.07728257560835855,
      "grad_norm": 1.7975350618362427,
      "learning_rate": 0.00019883020063447672,
      "loss": 1.6843,
      "step": 2158
    },
    {
      "epoch": 0.07731838773792683,
      "grad_norm": 1.6588022708892822,
      "learning_rate": 0.00019882843101012778,
      "loss": 1.7215,
      "step": 2159
    },
    {
      "epoch": 0.07735419986749512,
      "grad_norm": 1.990808367729187,
      "learning_rate": 0.00019882666005616978,
      "loss": 1.729,
      "step": 2160
    },
    {
      "epoch": 0.0773900119970634,
      "grad_norm": 2.0240907669067383,
      "learning_rate": 0.00019882488777262655,
      "loss": 1.299,
      "step": 2161
    },
    {
      "epoch": 0.0774258241266317,
      "grad_norm": 1.273319959640503,
      "learning_rate": 0.00019882311415952194,
      "loss": 1.2768,
      "step": 2162
    },
    {
      "epoch": 0.07746163625619998,
      "grad_norm": 3.0127789974212646,
      "learning_rate": 0.00019882133921687983,
      "loss": 1.7046,
      "step": 2163
    },
    {
      "epoch": 0.07749744838576826,
      "grad_norm": 1.886521577835083,
      "learning_rate": 0.00019881956294472405,
      "loss": 1.62,
      "step": 2164
    },
    {
      "epoch": 0.07753326051533654,
      "grad_norm": 1.6154439449310303,
      "learning_rate": 0.00019881778534307852,
      "loss": 1.5143,
      "step": 2165
    },
    {
      "epoch": 0.07756907264490483,
      "grad_norm": 2.255967617034912,
      "learning_rate": 0.0001988160064119671,
      "loss": 1.6454,
      "step": 2166
    },
    {
      "epoch": 0.07760488477447311,
      "grad_norm": 2.1320412158966064,
      "learning_rate": 0.00019881422615141385,
      "loss": 1.6159,
      "step": 2167
    },
    {
      "epoch": 0.07764069690404139,
      "grad_norm": 1.293569803237915,
      "learning_rate": 0.00019881244456144262,
      "loss": 1.7265,
      "step": 2168
    },
    {
      "epoch": 0.07767650903360969,
      "grad_norm": 1.4700331687927246,
      "learning_rate": 0.00019881066164207742,
      "loss": 1.6755,
      "step": 2169
    },
    {
      "epoch": 0.07771232116317797,
      "grad_norm": 2.1993863582611084,
      "learning_rate": 0.0001988088773933422,
      "loss": 1.7127,
      "step": 2170
    },
    {
      "epoch": 0.07774813329274625,
      "grad_norm": 2.8095591068267822,
      "learning_rate": 0.000198807091815261,
      "loss": 1.6606,
      "step": 2171
    },
    {
      "epoch": 0.07778394542231454,
      "grad_norm": 1.5221842527389526,
      "learning_rate": 0.00019880530490785784,
      "loss": 1.543,
      "step": 2172
    },
    {
      "epoch": 0.07781975755188282,
      "grad_norm": 1.5565546751022339,
      "learning_rate": 0.00019880351667115673,
      "loss": 1.3438,
      "step": 2173
    },
    {
      "epoch": 0.0778555696814511,
      "grad_norm": 1.7649588584899902,
      "learning_rate": 0.00019880172710518178,
      "loss": 1.7558,
      "step": 2174
    },
    {
      "epoch": 0.07789138181101939,
      "grad_norm": 2.050035238265991,
      "learning_rate": 0.00019879993620995702,
      "loss": 1.6594,
      "step": 2175
    },
    {
      "epoch": 0.07792719394058768,
      "grad_norm": 1.595670461654663,
      "learning_rate": 0.00019879814398550657,
      "loss": 2.0138,
      "step": 2176
    },
    {
      "epoch": 0.07796300607015597,
      "grad_norm": 1.8371297121047974,
      "learning_rate": 0.00019879635043185454,
      "loss": 1.6345,
      "step": 2177
    },
    {
      "epoch": 0.07799881819972425,
      "grad_norm": 1.5137767791748047,
      "learning_rate": 0.00019879455554902502,
      "loss": 1.8678,
      "step": 2178
    },
    {
      "epoch": 0.07803463032929253,
      "grad_norm": 1.4583414793014526,
      "learning_rate": 0.00019879275933704224,
      "loss": 1.7769,
      "step": 2179
    },
    {
      "epoch": 0.07807044245886081,
      "grad_norm": 1.8222814798355103,
      "learning_rate": 0.00019879096179593027,
      "loss": 1.7373,
      "step": 2180
    },
    {
      "epoch": 0.0781062545884291,
      "grad_norm": 1.8903967142105103,
      "learning_rate": 0.00019878916292571334,
      "loss": 2.003,
      "step": 2181
    },
    {
      "epoch": 0.07814206671799738,
      "grad_norm": 2.2535934448242188,
      "learning_rate": 0.00019878736272641568,
      "loss": 1.661,
      "step": 2182
    },
    {
      "epoch": 0.07817787884756568,
      "grad_norm": 1.6508339643478394,
      "learning_rate": 0.00019878556119806148,
      "loss": 1.3687,
      "step": 2183
    },
    {
      "epoch": 0.07821369097713396,
      "grad_norm": 1.6056736707687378,
      "learning_rate": 0.00019878375834067496,
      "loss": 1.3639,
      "step": 2184
    },
    {
      "epoch": 0.07824950310670224,
      "grad_norm": 1.481017827987671,
      "learning_rate": 0.0001987819541542804,
      "loss": 1.9444,
      "step": 2185
    },
    {
      "epoch": 0.07828531523627053,
      "grad_norm": 1.7288168668746948,
      "learning_rate": 0.0001987801486389021,
      "loss": 1.9123,
      "step": 2186
    },
    {
      "epoch": 0.07832112736583881,
      "grad_norm": 2.189270257949829,
      "learning_rate": 0.00019877834179456424,
      "loss": 1.5186,
      "step": 2187
    },
    {
      "epoch": 0.07835693949540709,
      "grad_norm": 1.4086220264434814,
      "learning_rate": 0.00019877653362129126,
      "loss": 1.5263,
      "step": 2188
    },
    {
      "epoch": 0.07839275162497537,
      "grad_norm": 1.423419713973999,
      "learning_rate": 0.00019877472411910745,
      "loss": 1.6879,
      "step": 2189
    },
    {
      "epoch": 0.07842856375454367,
      "grad_norm": 1.8120468854904175,
      "learning_rate": 0.0001987729132880371,
      "loss": 1.725,
      "step": 2190
    },
    {
      "epoch": 0.07846437588411195,
      "grad_norm": 1.2922435998916626,
      "learning_rate": 0.00019877110112810463,
      "loss": 1.5935,
      "step": 2191
    },
    {
      "epoch": 0.07850018801368024,
      "grad_norm": 1.484522819519043,
      "learning_rate": 0.00019876928763933437,
      "loss": 1.5979,
      "step": 2192
    },
    {
      "epoch": 0.07853600014324852,
      "grad_norm": 1.5501383543014526,
      "learning_rate": 0.00019876747282175078,
      "loss": 1.6291,
      "step": 2193
    },
    {
      "epoch": 0.0785718122728168,
      "grad_norm": 1.7386257648468018,
      "learning_rate": 0.00019876565667537824,
      "loss": 1.5312,
      "step": 2194
    },
    {
      "epoch": 0.07860762440238508,
      "grad_norm": 1.6912481784820557,
      "learning_rate": 0.00019876383920024117,
      "loss": 1.5414,
      "step": 2195
    },
    {
      "epoch": 0.07864343653195337,
      "grad_norm": 1.5818910598754883,
      "learning_rate": 0.00019876202039636405,
      "loss": 1.8032,
      "step": 2196
    },
    {
      "epoch": 0.07867924866152166,
      "grad_norm": 1.9245262145996094,
      "learning_rate": 0.00019876020026377136,
      "loss": 1.3893,
      "step": 2197
    },
    {
      "epoch": 0.07871506079108995,
      "grad_norm": 1.422369360923767,
      "learning_rate": 0.00019875837880248756,
      "loss": 1.6805,
      "step": 2198
    },
    {
      "epoch": 0.07875087292065823,
      "grad_norm": 1.8367788791656494,
      "learning_rate": 0.00019875655601253714,
      "loss": 1.7323,
      "step": 2199
    },
    {
      "epoch": 0.07878668505022651,
      "grad_norm": 1.3831416368484497,
      "learning_rate": 0.00019875473189394463,
      "loss": 1.5551,
      "step": 2200
    },
    {
      "epoch": 0.0788224971797948,
      "grad_norm": 1.5825157165527344,
      "learning_rate": 0.00019875290644673463,
      "loss": 1.7588,
      "step": 2201
    },
    {
      "epoch": 0.07885830930936308,
      "grad_norm": 1.0356022119522095,
      "learning_rate": 0.00019875107967093163,
      "loss": 1.7062,
      "step": 2202
    },
    {
      "epoch": 0.07889412143893136,
      "grad_norm": 1.5128309726715088,
      "learning_rate": 0.00019874925156656024,
      "loss": 1.6296,
      "step": 2203
    },
    {
      "epoch": 0.07892993356849964,
      "grad_norm": 2.1393330097198486,
      "learning_rate": 0.00019874742213364506,
      "loss": 1.8722,
      "step": 2204
    },
    {
      "epoch": 0.07896574569806794,
      "grad_norm": 1.531294345855713,
      "learning_rate": 0.00019874559137221068,
      "loss": 1.5297,
      "step": 2205
    },
    {
      "epoch": 0.07900155782763622,
      "grad_norm": 1.653439998626709,
      "learning_rate": 0.00019874375928228175,
      "loss": 1.5084,
      "step": 2206
    },
    {
      "epoch": 0.0790373699572045,
      "grad_norm": 1.6008787155151367,
      "learning_rate": 0.00019874192586388288,
      "loss": 1.4493,
      "step": 2207
    },
    {
      "epoch": 0.07907318208677279,
      "grad_norm": 1.8827781677246094,
      "learning_rate": 0.00019874009111703878,
      "loss": 1.6156,
      "step": 2208
    },
    {
      "epoch": 0.07910899421634107,
      "grad_norm": 1.7337536811828613,
      "learning_rate": 0.00019873825504177414,
      "loss": 1.7076,
      "step": 2209
    },
    {
      "epoch": 0.07914480634590935,
      "grad_norm": 1.7455438375473022,
      "learning_rate": 0.0001987364176381136,
      "loss": 1.4984,
      "step": 2210
    },
    {
      "epoch": 0.07918061847547764,
      "grad_norm": 1.7682832479476929,
      "learning_rate": 0.00019873457890608198,
      "loss": 1.7267,
      "step": 2211
    },
    {
      "epoch": 0.07921643060504593,
      "grad_norm": 1.8477669954299927,
      "learning_rate": 0.0001987327388457039,
      "loss": 1.8183,
      "step": 2212
    },
    {
      "epoch": 0.07925224273461422,
      "grad_norm": 2.0341742038726807,
      "learning_rate": 0.0001987308974570042,
      "loss": 1.5857,
      "step": 2213
    },
    {
      "epoch": 0.0792880548641825,
      "grad_norm": 1.6047203540802002,
      "learning_rate": 0.0001987290547400076,
      "loss": 1.4304,
      "step": 2214
    },
    {
      "epoch": 0.07932386699375078,
      "grad_norm": 1.6200206279754639,
      "learning_rate": 0.000198727210694739,
      "loss": 1.3838,
      "step": 2215
    },
    {
      "epoch": 0.07935967912331907,
      "grad_norm": 1.4143503904342651,
      "learning_rate": 0.00019872536532122305,
      "loss": 1.4999,
      "step": 2216
    },
    {
      "epoch": 0.07939549125288735,
      "grad_norm": 1.7010365724563599,
      "learning_rate": 0.0001987235186194847,
      "loss": 1.9153,
      "step": 2217
    },
    {
      "epoch": 0.07943130338245563,
      "grad_norm": 1.9490126371383667,
      "learning_rate": 0.00019872167058954874,
      "loss": 1.5034,
      "step": 2218
    },
    {
      "epoch": 0.07946711551202393,
      "grad_norm": 1.496788740158081,
      "learning_rate": 0.00019871982123144004,
      "loss": 1.7765,
      "step": 2219
    },
    {
      "epoch": 0.07950292764159221,
      "grad_norm": 1.3413259983062744,
      "learning_rate": 0.00019871797054518347,
      "loss": 1.6507,
      "step": 2220
    },
    {
      "epoch": 0.0795387397711605,
      "grad_norm": 2.239360809326172,
      "learning_rate": 0.00019871611853080397,
      "loss": 1.8348,
      "step": 2221
    },
    {
      "epoch": 0.07957455190072878,
      "grad_norm": 1.3430578708648682,
      "learning_rate": 0.00019871426518832644,
      "loss": 1.7134,
      "step": 2222
    },
    {
      "epoch": 0.07961036403029706,
      "grad_norm": 1.7762681245803833,
      "learning_rate": 0.00019871241051777576,
      "loss": 1.8258,
      "step": 2223
    },
    {
      "epoch": 0.07964617615986534,
      "grad_norm": 3.437670946121216,
      "learning_rate": 0.00019871055451917694,
      "loss": 1.598,
      "step": 2224
    },
    {
      "epoch": 0.07968198828943363,
      "grad_norm": 1.7100651264190674,
      "learning_rate": 0.00019870869719255496,
      "loss": 1.5688,
      "step": 2225
    },
    {
      "epoch": 0.07971780041900192,
      "grad_norm": 1.7335412502288818,
      "learning_rate": 0.00019870683853793474,
      "loss": 1.4256,
      "step": 2226
    },
    {
      "epoch": 0.0797536125485702,
      "grad_norm": 1.4160618782043457,
      "learning_rate": 0.00019870497855534137,
      "loss": 1.6183,
      "step": 2227
    },
    {
      "epoch": 0.07978942467813849,
      "grad_norm": 1.4244014024734497,
      "learning_rate": 0.00019870311724479983,
      "loss": 1.7856,
      "step": 2228
    },
    {
      "epoch": 0.07982523680770677,
      "grad_norm": 1.4283264875411987,
      "learning_rate": 0.00019870125460633514,
      "loss": 1.4689,
      "step": 2229
    },
    {
      "epoch": 0.07986104893727505,
      "grad_norm": 1.4369860887527466,
      "learning_rate": 0.00019869939063997243,
      "loss": 1.6801,
      "step": 2230
    },
    {
      "epoch": 0.07989686106684334,
      "grad_norm": 1.8066489696502686,
      "learning_rate": 0.00019869752534573668,
      "loss": 1.4535,
      "step": 2231
    },
    {
      "epoch": 0.07993267319641162,
      "grad_norm": 2.8576557636260986,
      "learning_rate": 0.00019869565872365308,
      "loss": 1.7293,
      "step": 2232
    },
    {
      "epoch": 0.07996848532597992,
      "grad_norm": 1.3145085573196411,
      "learning_rate": 0.00019869379077374667,
      "loss": 1.7171,
      "step": 2233
    },
    {
      "epoch": 0.0800042974555482,
      "grad_norm": 2.2998170852661133,
      "learning_rate": 0.00019869192149604264,
      "loss": 1.6707,
      "step": 2234
    },
    {
      "epoch": 0.08004010958511648,
      "grad_norm": 1.755245566368103,
      "learning_rate": 0.0001986900508905661,
      "loss": 1.5189,
      "step": 2235
    },
    {
      "epoch": 0.08007592171468476,
      "grad_norm": 1.8190783262252808,
      "learning_rate": 0.00019868817895734222,
      "loss": 1.7181,
      "step": 2236
    },
    {
      "epoch": 0.08011173384425305,
      "grad_norm": 1.505210280418396,
      "learning_rate": 0.00019868630569639618,
      "loss": 1.5393,
      "step": 2237
    },
    {
      "epoch": 0.08014754597382133,
      "grad_norm": 2.068084478378296,
      "learning_rate": 0.0001986844311077532,
      "loss": 1.5744,
      "step": 2238
    },
    {
      "epoch": 0.08018335810338961,
      "grad_norm": 1.9969322681427002,
      "learning_rate": 0.0001986825551914385,
      "loss": 1.507,
      "step": 2239
    },
    {
      "epoch": 0.08021917023295791,
      "grad_norm": 3.0546181201934814,
      "learning_rate": 0.00019868067794747728,
      "loss": 1.8212,
      "step": 2240
    },
    {
      "epoch": 0.08025498236252619,
      "grad_norm": 1.3269965648651123,
      "learning_rate": 0.00019867879937589486,
      "loss": 1.6769,
      "step": 2241
    },
    {
      "epoch": 0.08029079449209447,
      "grad_norm": 1.6832436323165894,
      "learning_rate": 0.0001986769194767165,
      "loss": 1.7213,
      "step": 2242
    },
    {
      "epoch": 0.08032660662166276,
      "grad_norm": 1.4924203157424927,
      "learning_rate": 0.00019867503824996745,
      "loss": 1.6567,
      "step": 2243
    },
    {
      "epoch": 0.08036241875123104,
      "grad_norm": 1.7548980712890625,
      "learning_rate": 0.00019867315569567303,
      "loss": 1.571,
      "step": 2244
    },
    {
      "epoch": 0.08039823088079932,
      "grad_norm": 2.1904239654541016,
      "learning_rate": 0.0001986712718138586,
      "loss": 1.6707,
      "step": 2245
    },
    {
      "epoch": 0.0804340430103676,
      "grad_norm": 1.3834635019302368,
      "learning_rate": 0.00019866938660454949,
      "loss": 1.5114,
      "step": 2246
    },
    {
      "epoch": 0.0804698551399359,
      "grad_norm": 1.6187025308609009,
      "learning_rate": 0.00019866750006777102,
      "loss": 1.6336,
      "step": 2247
    },
    {
      "epoch": 0.08050566726950419,
      "grad_norm": 1.5296133756637573,
      "learning_rate": 0.00019866561220354862,
      "loss": 1.6381,
      "step": 2248
    },
    {
      "epoch": 0.08054147939907247,
      "grad_norm": 1.5018010139465332,
      "learning_rate": 0.0001986637230119077,
      "loss": 1.5588,
      "step": 2249
    },
    {
      "epoch": 0.08057729152864075,
      "grad_norm": 1.3861911296844482,
      "learning_rate": 0.00019866183249287364,
      "loss": 1.5929,
      "step": 2250
    },
    {
      "epoch": 0.08061310365820903,
      "grad_norm": 1.2317131757736206,
      "learning_rate": 0.00019865994064647188,
      "loss": 1.4665,
      "step": 2251
    },
    {
      "epoch": 0.08064891578777732,
      "grad_norm": 1.849687933921814,
      "learning_rate": 0.0001986580474727279,
      "loss": 1.5341,
      "step": 2252
    },
    {
      "epoch": 0.0806847279173456,
      "grad_norm": 1.3923594951629639,
      "learning_rate": 0.00019865615297166714,
      "loss": 1.7859,
      "step": 2253
    },
    {
      "epoch": 0.08072054004691388,
      "grad_norm": 1.4301036596298218,
      "learning_rate": 0.0001986542571433151,
      "loss": 1.8605,
      "step": 2254
    },
    {
      "epoch": 0.08075635217648218,
      "grad_norm": 1.6380246877670288,
      "learning_rate": 0.00019865235998769727,
      "loss": 1.7913,
      "step": 2255
    },
    {
      "epoch": 0.08079216430605046,
      "grad_norm": 2.1478631496429443,
      "learning_rate": 0.0001986504615048392,
      "loss": 1.5993,
      "step": 2256
    },
    {
      "epoch": 0.08082797643561875,
      "grad_norm": 1.3344511985778809,
      "learning_rate": 0.0001986485616947664,
      "loss": 1.718,
      "step": 2257
    },
    {
      "epoch": 0.08086378856518703,
      "grad_norm": 1.8746141195297241,
      "learning_rate": 0.00019864666055750452,
      "loss": 1.6871,
      "step": 2258
    },
    {
      "epoch": 0.08089960069475531,
      "grad_norm": 1.6850662231445312,
      "learning_rate": 0.000198644758093079,
      "loss": 1.8947,
      "step": 2259
    },
    {
      "epoch": 0.0809354128243236,
      "grad_norm": 1.7207536697387695,
      "learning_rate": 0.00019864285430151553,
      "loss": 1.6498,
      "step": 2260
    },
    {
      "epoch": 0.08097122495389188,
      "grad_norm": 1.387789011001587,
      "learning_rate": 0.00019864094918283968,
      "loss": 1.5868,
      "step": 2261
    },
    {
      "epoch": 0.08100703708346017,
      "grad_norm": 1.4519275426864624,
      "learning_rate": 0.0001986390427370771,
      "loss": 1.4775,
      "step": 2262
    },
    {
      "epoch": 0.08104284921302846,
      "grad_norm": 1.7079511880874634,
      "learning_rate": 0.00019863713496425347,
      "loss": 1.4262,
      "step": 2263
    },
    {
      "epoch": 0.08107866134259674,
      "grad_norm": 2.308537483215332,
      "learning_rate": 0.0001986352258643944,
      "loss": 1.5502,
      "step": 2264
    },
    {
      "epoch": 0.08111447347216502,
      "grad_norm": 1.9541075229644775,
      "learning_rate": 0.00019863331543752558,
      "loss": 1.6206,
      "step": 2265
    },
    {
      "epoch": 0.0811502856017333,
      "grad_norm": 1.6608632802963257,
      "learning_rate": 0.00019863140368367273,
      "loss": 1.5943,
      "step": 2266
    },
    {
      "epoch": 0.08118609773130159,
      "grad_norm": 1.6765403747558594,
      "learning_rate": 0.00019862949060286158,
      "loss": 1.852,
      "step": 2267
    },
    {
      "epoch": 0.08122190986086987,
      "grad_norm": 1.4967211484909058,
      "learning_rate": 0.00019862757619511784,
      "loss": 1.4903,
      "step": 2268
    },
    {
      "epoch": 0.08125772199043817,
      "grad_norm": 1.465468168258667,
      "learning_rate": 0.0001986256604604673,
      "loss": 1.5695,
      "step": 2269
    },
    {
      "epoch": 0.08129353412000645,
      "grad_norm": 1.5892409086227417,
      "learning_rate": 0.0001986237433989357,
      "loss": 1.6337,
      "step": 2270
    },
    {
      "epoch": 0.08132934624957473,
      "grad_norm": 1.3970975875854492,
      "learning_rate": 0.0001986218250105489,
      "loss": 1.6544,
      "step": 2271
    },
    {
      "epoch": 0.08136515837914302,
      "grad_norm": 2.048312187194824,
      "learning_rate": 0.0001986199052953326,
      "loss": 1.4003,
      "step": 2272
    },
    {
      "epoch": 0.0814009705087113,
      "grad_norm": 1.864230751991272,
      "learning_rate": 0.0001986179842533127,
      "loss": 1.4919,
      "step": 2273
    },
    {
      "epoch": 0.08143678263827958,
      "grad_norm": 1.4341579675674438,
      "learning_rate": 0.00019861606188451502,
      "loss": 1.5127,
      "step": 2274
    },
    {
      "epoch": 0.08147259476784786,
      "grad_norm": 1.5005865097045898,
      "learning_rate": 0.00019861413818896546,
      "loss": 1.6661,
      "step": 2275
    },
    {
      "epoch": 0.08150840689741616,
      "grad_norm": 1.1538370847702026,
      "learning_rate": 0.00019861221316668984,
      "loss": 1.4567,
      "step": 2276
    },
    {
      "epoch": 0.08154421902698444,
      "grad_norm": 1.613469123840332,
      "learning_rate": 0.0001986102868177141,
      "loss": 1.7542,
      "step": 2277
    },
    {
      "epoch": 0.08158003115655273,
      "grad_norm": 1.6482155323028564,
      "learning_rate": 0.0001986083591420642,
      "loss": 1.8616,
      "step": 2278
    },
    {
      "epoch": 0.08161584328612101,
      "grad_norm": 1.7545753717422485,
      "learning_rate": 0.00019860643013976597,
      "loss": 1.4438,
      "step": 2279
    },
    {
      "epoch": 0.08165165541568929,
      "grad_norm": 2.3131508827209473,
      "learning_rate": 0.00019860449981084545,
      "loss": 1.6012,
      "step": 2280
    },
    {
      "epoch": 0.08168746754525757,
      "grad_norm": 1.8130494356155396,
      "learning_rate": 0.00019860256815532854,
      "loss": 1.2831,
      "step": 2281
    },
    {
      "epoch": 0.08172327967482586,
      "grad_norm": 1.7892794609069824,
      "learning_rate": 0.0001986006351732413,
      "loss": 1.5343,
      "step": 2282
    },
    {
      "epoch": 0.08175909180439415,
      "grad_norm": 1.544555902481079,
      "learning_rate": 0.00019859870086460965,
      "loss": 1.708,
      "step": 2283
    },
    {
      "epoch": 0.08179490393396244,
      "grad_norm": 2.1215953826904297,
      "learning_rate": 0.0001985967652294597,
      "loss": 1.7509,
      "step": 2284
    },
    {
      "epoch": 0.08183071606353072,
      "grad_norm": 1.4176714420318604,
      "learning_rate": 0.00019859482826781744,
      "loss": 1.5661,
      "step": 2285
    },
    {
      "epoch": 0.081866528193099,
      "grad_norm": 1.7044744491577148,
      "learning_rate": 0.00019859288997970895,
      "loss": 1.7327,
      "step": 2286
    },
    {
      "epoch": 0.08190234032266729,
      "grad_norm": 1.9832544326782227,
      "learning_rate": 0.0001985909503651603,
      "loss": 1.4708,
      "step": 2287
    },
    {
      "epoch": 0.08193815245223557,
      "grad_norm": 1.8185745477676392,
      "learning_rate": 0.0001985890094241976,
      "loss": 1.7814,
      "step": 2288
    },
    {
      "epoch": 0.08197396458180385,
      "grad_norm": 1.557418942451477,
      "learning_rate": 0.0001985870671568469,
      "loss": 1.7562,
      "step": 2289
    },
    {
      "epoch": 0.08200977671137215,
      "grad_norm": 1.7640334367752075,
      "learning_rate": 0.00019858512356313445,
      "loss": 1.4121,
      "step": 2290
    },
    {
      "epoch": 0.08204558884094043,
      "grad_norm": 1.6936959028244019,
      "learning_rate": 0.00019858317864308628,
      "loss": 1.6866,
      "step": 2291
    },
    {
      "epoch": 0.08208140097050871,
      "grad_norm": 1.587742805480957,
      "learning_rate": 0.0001985812323967286,
      "loss": 1.6117,
      "step": 2292
    },
    {
      "epoch": 0.082117213100077,
      "grad_norm": 1.4579888582229614,
      "learning_rate": 0.00019857928482408763,
      "loss": 1.7537,
      "step": 2293
    },
    {
      "epoch": 0.08215302522964528,
      "grad_norm": 1.2377456426620483,
      "learning_rate": 0.00019857733592518954,
      "loss": 1.2892,
      "step": 2294
    },
    {
      "epoch": 0.08218883735921356,
      "grad_norm": 2.1315197944641113,
      "learning_rate": 0.00019857538570006053,
      "loss": 1.7453,
      "step": 2295
    },
    {
      "epoch": 0.08222464948878185,
      "grad_norm": 2.444917678833008,
      "learning_rate": 0.00019857343414872685,
      "loss": 1.8851,
      "step": 2296
    },
    {
      "epoch": 0.08226046161835014,
      "grad_norm": 2.752296209335327,
      "learning_rate": 0.0001985714812712148,
      "loss": 1.7393,
      "step": 2297
    },
    {
      "epoch": 0.08229627374791842,
      "grad_norm": 1.5612289905548096,
      "learning_rate": 0.0001985695270675506,
      "loss": 1.5019,
      "step": 2298
    },
    {
      "epoch": 0.08233208587748671,
      "grad_norm": 1.6323474645614624,
      "learning_rate": 0.00019856757153776058,
      "loss": 1.6256,
      "step": 2299
    },
    {
      "epoch": 0.08236789800705499,
      "grad_norm": 2.15275239944458,
      "learning_rate": 0.000198565614681871,
      "loss": 1.4934,
      "step": 2300
    },
    {
      "epoch": 0.08240371013662327,
      "grad_norm": 2.2830469608306885,
      "learning_rate": 0.0001985636564999082,
      "loss": 1.8445,
      "step": 2301
    },
    {
      "epoch": 0.08243952226619156,
      "grad_norm": 1.283078670501709,
      "learning_rate": 0.00019856169699189856,
      "loss": 1.6053,
      "step": 2302
    },
    {
      "epoch": 0.08247533439575984,
      "grad_norm": 2.0136337280273438,
      "learning_rate": 0.00019855973615786842,
      "loss": 1.7864,
      "step": 2303
    },
    {
      "epoch": 0.08251114652532812,
      "grad_norm": 1.7052009105682373,
      "learning_rate": 0.0001985577739978442,
      "loss": 1.2617,
      "step": 2304
    },
    {
      "epoch": 0.08254695865489642,
      "grad_norm": 1.6163235902786255,
      "learning_rate": 0.0001985558105118522,
      "loss": 1.575,
      "step": 2305
    },
    {
      "epoch": 0.0825827707844647,
      "grad_norm": 1.2028448581695557,
      "learning_rate": 0.00019855384569991892,
      "loss": 1.575,
      "step": 2306
    },
    {
      "epoch": 0.08261858291403298,
      "grad_norm": 1.1882579326629639,
      "learning_rate": 0.0001985518795620708,
      "loss": 1.7627,
      "step": 2307
    },
    {
      "epoch": 0.08265439504360127,
      "grad_norm": 1.8147140741348267,
      "learning_rate": 0.0001985499120983342,
      "loss": 1.8266,
      "step": 2308
    },
    {
      "epoch": 0.08269020717316955,
      "grad_norm": 1.386354684829712,
      "learning_rate": 0.00019854794330873568,
      "loss": 1.6571,
      "step": 2309
    },
    {
      "epoch": 0.08272601930273783,
      "grad_norm": 2.043935775756836,
      "learning_rate": 0.00019854597319330175,
      "loss": 1.6551,
      "step": 2310
    },
    {
      "epoch": 0.08276183143230612,
      "grad_norm": 1.749604344367981,
      "learning_rate": 0.00019854400175205883,
      "loss": 1.7481,
      "step": 2311
    },
    {
      "epoch": 0.08279764356187441,
      "grad_norm": 1.7542747259140015,
      "learning_rate": 0.00019854202898503346,
      "loss": 1.4622,
      "step": 2312
    },
    {
      "epoch": 0.0828334556914427,
      "grad_norm": 1.2417621612548828,
      "learning_rate": 0.00019854005489225224,
      "loss": 1.7303,
      "step": 2313
    },
    {
      "epoch": 0.08286926782101098,
      "grad_norm": 1.5347504615783691,
      "learning_rate": 0.00019853807947374166,
      "loss": 1.866,
      "step": 2314
    },
    {
      "epoch": 0.08290507995057926,
      "grad_norm": 1.5232512950897217,
      "learning_rate": 0.0001985361027295283,
      "loss": 1.5288,
      "step": 2315
    },
    {
      "epoch": 0.08294089208014754,
      "grad_norm": 1.9294697046279907,
      "learning_rate": 0.00019853412465963883,
      "loss": 1.5636,
      "step": 2316
    },
    {
      "epoch": 0.08297670420971583,
      "grad_norm": 1.68217933177948,
      "learning_rate": 0.0001985321452640998,
      "loss": 1.547,
      "step": 2317
    },
    {
      "epoch": 0.08301251633928411,
      "grad_norm": 2.1691644191741943,
      "learning_rate": 0.00019853016454293785,
      "loss": 1.6431,
      "step": 2318
    },
    {
      "epoch": 0.0830483284688524,
      "grad_norm": 1.4858300685882568,
      "learning_rate": 0.00019852818249617963,
      "loss": 1.7581,
      "step": 2319
    },
    {
      "epoch": 0.08308414059842069,
      "grad_norm": 1.6669294834136963,
      "learning_rate": 0.0001985261991238518,
      "loss": 1.4139,
      "step": 2320
    },
    {
      "epoch": 0.08311995272798897,
      "grad_norm": 1.9712706804275513,
      "learning_rate": 0.00019852421442598107,
      "loss": 1.7628,
      "step": 2321
    },
    {
      "epoch": 0.08315576485755725,
      "grad_norm": 2.3515987396240234,
      "learning_rate": 0.0001985222284025941,
      "loss": 1.9166,
      "step": 2322
    },
    {
      "epoch": 0.08319157698712554,
      "grad_norm": 1.5739303827285767,
      "learning_rate": 0.00019852024105371764,
      "loss": 1.7737,
      "step": 2323
    },
    {
      "epoch": 0.08322738911669382,
      "grad_norm": 1.4964985847473145,
      "learning_rate": 0.0001985182523793784,
      "loss": 1.3749,
      "step": 2324
    },
    {
      "epoch": 0.0832632012462621,
      "grad_norm": 1.9431986808776855,
      "learning_rate": 0.00019851626237960316,
      "loss": 1.7314,
      "step": 2325
    },
    {
      "epoch": 0.0832990133758304,
      "grad_norm": 1.4260855913162231,
      "learning_rate": 0.00019851427105441874,
      "loss": 1.6923,
      "step": 2326
    },
    {
      "epoch": 0.08333482550539868,
      "grad_norm": 2.3210411071777344,
      "learning_rate": 0.00019851227840385184,
      "loss": 1.6412,
      "step": 2327
    },
    {
      "epoch": 0.08337063763496697,
      "grad_norm": 1.6222409009933472,
      "learning_rate": 0.00019851028442792928,
      "loss": 1.7586,
      "step": 2328
    },
    {
      "epoch": 0.08340644976453525,
      "grad_norm": 1.9236634969711304,
      "learning_rate": 0.00019850828912667794,
      "loss": 1.8013,
      "step": 2329
    },
    {
      "epoch": 0.08344226189410353,
      "grad_norm": 1.6394051313400269,
      "learning_rate": 0.0001985062925001246,
      "loss": 1.5605,
      "step": 2330
    },
    {
      "epoch": 0.08347807402367181,
      "grad_norm": 1.769962191581726,
      "learning_rate": 0.0001985042945482962,
      "loss": 1.6226,
      "step": 2331
    },
    {
      "epoch": 0.0835138861532401,
      "grad_norm": 1.7239434719085693,
      "learning_rate": 0.00019850229527121956,
      "loss": 1.5937,
      "step": 2332
    },
    {
      "epoch": 0.0835496982828084,
      "grad_norm": 1.6355654001235962,
      "learning_rate": 0.00019850029466892161,
      "loss": 1.7279,
      "step": 2333
    },
    {
      "epoch": 0.08358551041237668,
      "grad_norm": 2.545929193496704,
      "learning_rate": 0.00019849829274142924,
      "loss": 1.472,
      "step": 2334
    },
    {
      "epoch": 0.08362132254194496,
      "grad_norm": 2.478335380554199,
      "learning_rate": 0.00019849628948876943,
      "loss": 2.0368,
      "step": 2335
    },
    {
      "epoch": 0.08365713467151324,
      "grad_norm": 1.5632565021514893,
      "learning_rate": 0.00019849428491096904,
      "loss": 1.5616,
      "step": 2336
    },
    {
      "epoch": 0.08369294680108152,
      "grad_norm": 1.9258129596710205,
      "learning_rate": 0.0001984922790080551,
      "loss": 1.7881,
      "step": 2337
    },
    {
      "epoch": 0.08372875893064981,
      "grad_norm": 1.5572803020477295,
      "learning_rate": 0.0001984902717800546,
      "loss": 1.4918,
      "step": 2338
    },
    {
      "epoch": 0.08376457106021809,
      "grad_norm": 1.6713584661483765,
      "learning_rate": 0.00019848826322699456,
      "loss": 1.5777,
      "step": 2339
    },
    {
      "epoch": 0.08380038318978639,
      "grad_norm": 1.7208898067474365,
      "learning_rate": 0.000198486253348902,
      "loss": 1.4879,
      "step": 2340
    },
    {
      "epoch": 0.08383619531935467,
      "grad_norm": 3.210822343826294,
      "learning_rate": 0.0001984842421458039,
      "loss": 1.8145,
      "step": 2341
    },
    {
      "epoch": 0.08387200744892295,
      "grad_norm": 1.435042142868042,
      "learning_rate": 0.00019848222961772733,
      "loss": 1.7716,
      "step": 2342
    },
    {
      "epoch": 0.08390781957849124,
      "grad_norm": 1.7752784490585327,
      "learning_rate": 0.00019848021576469944,
      "loss": 1.6036,
      "step": 2343
    },
    {
      "epoch": 0.08394363170805952,
      "grad_norm": 1.3135857582092285,
      "learning_rate": 0.00019847820058674728,
      "loss": 1.6017,
      "step": 2344
    },
    {
      "epoch": 0.0839794438376278,
      "grad_norm": 2.173564910888672,
      "learning_rate": 0.00019847618408389792,
      "loss": 1.7796,
      "step": 2345
    },
    {
      "epoch": 0.08401525596719608,
      "grad_norm": 2.061150312423706,
      "learning_rate": 0.00019847416625617855,
      "loss": 1.7466,
      "step": 2346
    },
    {
      "epoch": 0.08405106809676438,
      "grad_norm": 1.2300703525543213,
      "learning_rate": 0.0001984721471036163,
      "loss": 1.6624,
      "step": 2347
    },
    {
      "epoch": 0.08408688022633266,
      "grad_norm": 2.010895252227783,
      "learning_rate": 0.00019847012662623832,
      "loss": 1.5831,
      "step": 2348
    },
    {
      "epoch": 0.08412269235590095,
      "grad_norm": 1.3102781772613525,
      "learning_rate": 0.00019846810482407182,
      "loss": 1.7545,
      "step": 2349
    },
    {
      "epoch": 0.08415850448546923,
      "grad_norm": 1.5911515951156616,
      "learning_rate": 0.00019846608169714398,
      "loss": 1.3566,
      "step": 2350
    },
    {
      "epoch": 0.08419431661503751,
      "grad_norm": 2.5576908588409424,
      "learning_rate": 0.00019846405724548204,
      "loss": 1.5922,
      "step": 2351
    },
    {
      "epoch": 0.0842301287446058,
      "grad_norm": 1.8098796606063843,
      "learning_rate": 0.00019846203146911318,
      "loss": 1.8392,
      "step": 2352
    },
    {
      "epoch": 0.08426594087417408,
      "grad_norm": 1.828391432762146,
      "learning_rate": 0.00019846000436806471,
      "loss": 1.9675,
      "step": 2353
    },
    {
      "epoch": 0.08430175300374236,
      "grad_norm": 1.516196846961975,
      "learning_rate": 0.00019845797594236387,
      "loss": 1.596,
      "step": 2354
    },
    {
      "epoch": 0.08433756513331066,
      "grad_norm": 1.5106902122497559,
      "learning_rate": 0.00019845594619203797,
      "loss": 1.6667,
      "step": 2355
    },
    {
      "epoch": 0.08437337726287894,
      "grad_norm": 1.6493037939071655,
      "learning_rate": 0.00019845391511711435,
      "loss": 1.5757,
      "step": 2356
    },
    {
      "epoch": 0.08440918939244722,
      "grad_norm": 1.7102857828140259,
      "learning_rate": 0.00019845188271762029,
      "loss": 1.6952,
      "step": 2357
    },
    {
      "epoch": 0.0844450015220155,
      "grad_norm": 1.6236246824264526,
      "learning_rate": 0.0001984498489935831,
      "loss": 1.542,
      "step": 2358
    },
    {
      "epoch": 0.08448081365158379,
      "grad_norm": 1.3867934942245483,
      "learning_rate": 0.00019844781394503022,
      "loss": 1.5048,
      "step": 2359
    },
    {
      "epoch": 0.08451662578115207,
      "grad_norm": 2.9754772186279297,
      "learning_rate": 0.00019844577757198898,
      "loss": 1.6699,
      "step": 2360
    },
    {
      "epoch": 0.08455243791072035,
      "grad_norm": 1.5601086616516113,
      "learning_rate": 0.00019844373987448676,
      "loss": 1.5336,
      "step": 2361
    },
    {
      "epoch": 0.08458825004028865,
      "grad_norm": 1.989410638809204,
      "learning_rate": 0.00019844170085255104,
      "loss": 1.5408,
      "step": 2362
    },
    {
      "epoch": 0.08462406216985693,
      "grad_norm": 1.0998789072036743,
      "learning_rate": 0.0001984396605062092,
      "loss": 1.5359,
      "step": 2363
    },
    {
      "epoch": 0.08465987429942522,
      "grad_norm": 2.2412939071655273,
      "learning_rate": 0.00019843761883548872,
      "loss": 1.448,
      "step": 2364
    },
    {
      "epoch": 0.0846956864289935,
      "grad_norm": 1.2157700061798096,
      "learning_rate": 0.00019843557584041705,
      "loss": 1.4621,
      "step": 2365
    },
    {
      "epoch": 0.08473149855856178,
      "grad_norm": 1.5647035837173462,
      "learning_rate": 0.0001984335315210217,
      "loss": 1.8045,
      "step": 2366
    },
    {
      "epoch": 0.08476731068813007,
      "grad_norm": 1.3524651527404785,
      "learning_rate": 0.00019843148587733012,
      "loss": 1.778,
      "step": 2367
    },
    {
      "epoch": 0.08480312281769835,
      "grad_norm": 1.6187690496444702,
      "learning_rate": 0.00019842943890936986,
      "loss": 1.6222,
      "step": 2368
    },
    {
      "epoch": 0.08483893494726664,
      "grad_norm": 3.2864222526550293,
      "learning_rate": 0.00019842739061716848,
      "loss": 1.8601,
      "step": 2369
    },
    {
      "epoch": 0.08487474707683493,
      "grad_norm": 1.786603569984436,
      "learning_rate": 0.00019842534100075355,
      "loss": 1.9872,
      "step": 2370
    },
    {
      "epoch": 0.08491055920640321,
      "grad_norm": 1.6398714780807495,
      "learning_rate": 0.00019842329006015255,
      "loss": 1.5213,
      "step": 2371
    },
    {
      "epoch": 0.0849463713359715,
      "grad_norm": 1.576285719871521,
      "learning_rate": 0.0001984212377953932,
      "loss": 1.9445,
      "step": 2372
    },
    {
      "epoch": 0.08498218346553978,
      "grad_norm": 1.1794153451919556,
      "learning_rate": 0.00019841918420650302,
      "loss": 1.6204,
      "step": 2373
    },
    {
      "epoch": 0.08501799559510806,
      "grad_norm": 1.7451518774032593,
      "learning_rate": 0.00019841712929350965,
      "loss": 1.7095,
      "step": 2374
    },
    {
      "epoch": 0.08505380772467634,
      "grad_norm": 1.3258010149002075,
      "learning_rate": 0.0001984150730564408,
      "loss": 1.8024,
      "step": 2375
    },
    {
      "epoch": 0.08508961985424464,
      "grad_norm": 1.6353540420532227,
      "learning_rate": 0.00019841301549532409,
      "loss": 1.7716,
      "step": 2376
    },
    {
      "epoch": 0.08512543198381292,
      "grad_norm": 1.306315302848816,
      "learning_rate": 0.00019841095661018716,
      "loss": 1.6493,
      "step": 2377
    },
    {
      "epoch": 0.0851612441133812,
      "grad_norm": 1.7025679349899292,
      "learning_rate": 0.00019840889640105775,
      "loss": 1.8987,
      "step": 2378
    },
    {
      "epoch": 0.08519705624294949,
      "grad_norm": 2.471372127532959,
      "learning_rate": 0.00019840683486796362,
      "loss": 1.5562,
      "step": 2379
    },
    {
      "epoch": 0.08523286837251777,
      "grad_norm": 1.2974969148635864,
      "learning_rate": 0.00019840477201093243,
      "loss": 1.6721,
      "step": 2380
    },
    {
      "epoch": 0.08526868050208605,
      "grad_norm": 1.7710561752319336,
      "learning_rate": 0.00019840270782999197,
      "loss": 1.6326,
      "step": 2381
    },
    {
      "epoch": 0.08530449263165434,
      "grad_norm": 1.3455687761306763,
      "learning_rate": 0.00019840064232517,
      "loss": 1.7118,
      "step": 2382
    },
    {
      "epoch": 0.08534030476122263,
      "grad_norm": 1.3794631958007812,
      "learning_rate": 0.0001983985754964943,
      "loss": 1.5416,
      "step": 2383
    },
    {
      "epoch": 0.08537611689079092,
      "grad_norm": 1.446144938468933,
      "learning_rate": 0.00019839650734399276,
      "loss": 1.7194,
      "step": 2384
    },
    {
      "epoch": 0.0854119290203592,
      "grad_norm": 1.3965164422988892,
      "learning_rate": 0.0001983944378676931,
      "loss": 1.5514,
      "step": 2385
    },
    {
      "epoch": 0.08544774114992748,
      "grad_norm": 1.8332926034927368,
      "learning_rate": 0.00019839236706762318,
      "loss": 1.9941,
      "step": 2386
    },
    {
      "epoch": 0.08548355327949576,
      "grad_norm": 2.237839460372925,
      "learning_rate": 0.00019839029494381086,
      "loss": 1.5284,
      "step": 2387
    },
    {
      "epoch": 0.08551936540906405,
      "grad_norm": 1.652097463607788,
      "learning_rate": 0.0001983882214962841,
      "loss": 1.8933,
      "step": 2388
    },
    {
      "epoch": 0.08555517753863233,
      "grad_norm": 1.1951125860214233,
      "learning_rate": 0.00019838614672507067,
      "loss": 1.5353,
      "step": 2389
    },
    {
      "epoch": 0.08559098966820063,
      "grad_norm": 1.575891375541687,
      "learning_rate": 0.00019838407063019857,
      "loss": 1.9357,
      "step": 2390
    },
    {
      "epoch": 0.08562680179776891,
      "grad_norm": 1.398511290550232,
      "learning_rate": 0.0001983819932116957,
      "loss": 1.6936,
      "step": 2391
    },
    {
      "epoch": 0.08566261392733719,
      "grad_norm": 1.8677663803100586,
      "learning_rate": 0.00019837991446959005,
      "loss": 1.8502,
      "step": 2392
    },
    {
      "epoch": 0.08569842605690547,
      "grad_norm": 1.2182732820510864,
      "learning_rate": 0.0001983778344039095,
      "loss": 1.4945,
      "step": 2393
    },
    {
      "epoch": 0.08573423818647376,
      "grad_norm": 1.4245761632919312,
      "learning_rate": 0.00019837575301468211,
      "loss": 1.8349,
      "step": 2394
    },
    {
      "epoch": 0.08577005031604204,
      "grad_norm": 1.4133869409561157,
      "learning_rate": 0.00019837367030193587,
      "loss": 1.7366,
      "step": 2395
    },
    {
      "epoch": 0.08580586244561032,
      "grad_norm": 2.530789852142334,
      "learning_rate": 0.00019837158626569878,
      "loss": 2.013,
      "step": 2396
    },
    {
      "epoch": 0.08584167457517862,
      "grad_norm": 1.37733793258667,
      "learning_rate": 0.0001983695009059989,
      "loss": 1.7408,
      "step": 2397
    },
    {
      "epoch": 0.0858774867047469,
      "grad_norm": 1.6104776859283447,
      "learning_rate": 0.00019836741422286425,
      "loss": 1.5966,
      "step": 2398
    },
    {
      "epoch": 0.08591329883431519,
      "grad_norm": 1.435455322265625,
      "learning_rate": 0.00019836532621632293,
      "loss": 1.5554,
      "step": 2399
    },
    {
      "epoch": 0.08594911096388347,
      "grad_norm": 1.2741247415542603,
      "learning_rate": 0.000198363236886403,
      "loss": 1.3838,
      "step": 2400
    },
    {
      "epoch": 0.08598492309345175,
      "grad_norm": 1.459208369255066,
      "learning_rate": 0.00019836114623313265,
      "loss": 1.6547,
      "step": 2401
    },
    {
      "epoch": 0.08602073522302003,
      "grad_norm": 2.1880502700805664,
      "learning_rate": 0.00019835905425653994,
      "loss": 1.3995,
      "step": 2402
    },
    {
      "epoch": 0.08605654735258832,
      "grad_norm": 2.445836305618286,
      "learning_rate": 0.00019835696095665302,
      "loss": 1.9346,
      "step": 2403
    },
    {
      "epoch": 0.0860923594821566,
      "grad_norm": 2.707475185394287,
      "learning_rate": 0.00019835486633350006,
      "loss": 1.8816,
      "step": 2404
    },
    {
      "epoch": 0.0861281716117249,
      "grad_norm": 1.7417428493499756,
      "learning_rate": 0.00019835277038710928,
      "loss": 1.5025,
      "step": 2405
    },
    {
      "epoch": 0.08616398374129318,
      "grad_norm": 1.2903729677200317,
      "learning_rate": 0.00019835067311750878,
      "loss": 1.5742,
      "step": 2406
    },
    {
      "epoch": 0.08619979587086146,
      "grad_norm": 1.8366937637329102,
      "learning_rate": 0.00019834857452472686,
      "loss": 1.5567,
      "step": 2407
    },
    {
      "epoch": 0.08623560800042974,
      "grad_norm": 1.5683704614639282,
      "learning_rate": 0.00019834647460879174,
      "loss": 1.3516,
      "step": 2408
    },
    {
      "epoch": 0.08627142012999803,
      "grad_norm": 1.3883376121520996,
      "learning_rate": 0.00019834437336973165,
      "loss": 1.6229,
      "step": 2409
    },
    {
      "epoch": 0.08630723225956631,
      "grad_norm": 2.0175423622131348,
      "learning_rate": 0.00019834227080757488,
      "loss": 1.6116,
      "step": 2410
    },
    {
      "epoch": 0.0863430443891346,
      "grad_norm": 1.6487847566604614,
      "learning_rate": 0.0001983401669223497,
      "loss": 1.674,
      "step": 2411
    },
    {
      "epoch": 0.08637885651870289,
      "grad_norm": 1.7350643873214722,
      "learning_rate": 0.00019833806171408442,
      "loss": 1.7811,
      "step": 2412
    },
    {
      "epoch": 0.08641466864827117,
      "grad_norm": 1.6598446369171143,
      "learning_rate": 0.0001983359551828074,
      "loss": 1.6918,
      "step": 2413
    },
    {
      "epoch": 0.08645048077783946,
      "grad_norm": 5.592057228088379,
      "learning_rate": 0.0001983338473285469,
      "loss": 1.6704,
      "step": 2414
    },
    {
      "epoch": 0.08648629290740774,
      "grad_norm": 1.5842877626419067,
      "learning_rate": 0.00019833173815133134,
      "loss": 1.4705,
      "step": 2415
    },
    {
      "epoch": 0.08652210503697602,
      "grad_norm": 1.3319464921951294,
      "learning_rate": 0.0001983296276511891,
      "loss": 1.6647,
      "step": 2416
    },
    {
      "epoch": 0.0865579171665443,
      "grad_norm": 1.8691999912261963,
      "learning_rate": 0.00019832751582814855,
      "loss": 1.83,
      "step": 2417
    },
    {
      "epoch": 0.08659372929611259,
      "grad_norm": 2.428990125656128,
      "learning_rate": 0.0001983254026822381,
      "loss": 1.6859,
      "step": 2418
    },
    {
      "epoch": 0.08662954142568088,
      "grad_norm": 2.839390516281128,
      "learning_rate": 0.0001983232882134862,
      "loss": 1.4562,
      "step": 2419
    },
    {
      "epoch": 0.08666535355524917,
      "grad_norm": 1.6174662113189697,
      "learning_rate": 0.00019832117242192128,
      "loss": 1.4465,
      "step": 2420
    },
    {
      "epoch": 0.08670116568481745,
      "grad_norm": 1.4822794198989868,
      "learning_rate": 0.0001983190553075718,
      "loss": 1.4847,
      "step": 2421
    },
    {
      "epoch": 0.08673697781438573,
      "grad_norm": 1.9179503917694092,
      "learning_rate": 0.00019831693687046627,
      "loss": 1.8159,
      "step": 2422
    },
    {
      "epoch": 0.08677278994395402,
      "grad_norm": 3.1191704273223877,
      "learning_rate": 0.00019831481711063314,
      "loss": 1.8706,
      "step": 2423
    },
    {
      "epoch": 0.0868086020735223,
      "grad_norm": 1.2060275077819824,
      "learning_rate": 0.000198312696028101,
      "loss": 1.5291,
      "step": 2424
    },
    {
      "epoch": 0.08684441420309058,
      "grad_norm": 1.2168149948120117,
      "learning_rate": 0.00019831057362289833,
      "loss": 1.5218,
      "step": 2425
    },
    {
      "epoch": 0.08688022633265888,
      "grad_norm": 2.720611095428467,
      "learning_rate": 0.00019830844989505373,
      "loss": 1.8275,
      "step": 2426
    },
    {
      "epoch": 0.08691603846222716,
      "grad_norm": 1.599690318107605,
      "learning_rate": 0.00019830632484459573,
      "loss": 1.4722,
      "step": 2427
    },
    {
      "epoch": 0.08695185059179544,
      "grad_norm": 1.5266869068145752,
      "learning_rate": 0.00019830419847155292,
      "loss": 1.5618,
      "step": 2428
    },
    {
      "epoch": 0.08698766272136373,
      "grad_norm": 2.5052943229675293,
      "learning_rate": 0.00019830207077595392,
      "loss": 1.9813,
      "step": 2429
    },
    {
      "epoch": 0.08702347485093201,
      "grad_norm": 2.039900779724121,
      "learning_rate": 0.00019829994175782738,
      "loss": 1.7817,
      "step": 2430
    },
    {
      "epoch": 0.08705928698050029,
      "grad_norm": 1.7723355293273926,
      "learning_rate": 0.0001982978114172019,
      "loss": 1.8192,
      "step": 2431
    },
    {
      "epoch": 0.08709509911006857,
      "grad_norm": 1.4781324863433838,
      "learning_rate": 0.0001982956797541062,
      "loss": 1.5802,
      "step": 2432
    },
    {
      "epoch": 0.08713091123963687,
      "grad_norm": 1.499817967414856,
      "learning_rate": 0.0001982935467685689,
      "loss": 1.7145,
      "step": 2433
    },
    {
      "epoch": 0.08716672336920515,
      "grad_norm": 1.2955687046051025,
      "learning_rate": 0.0001982914124606187,
      "loss": 1.4076,
      "step": 2434
    },
    {
      "epoch": 0.08720253549877344,
      "grad_norm": 1.1812690496444702,
      "learning_rate": 0.00019828927683028435,
      "loss": 1.6311,
      "step": 2435
    },
    {
      "epoch": 0.08723834762834172,
      "grad_norm": 1.2968649864196777,
      "learning_rate": 0.00019828713987759454,
      "loss": 1.377,
      "step": 2436
    },
    {
      "epoch": 0.08727415975791,
      "grad_norm": 1.2757093906402588,
      "learning_rate": 0.00019828500160257807,
      "loss": 1.7077,
      "step": 2437
    },
    {
      "epoch": 0.08730997188747829,
      "grad_norm": 1.4437847137451172,
      "learning_rate": 0.0001982828620052637,
      "loss": 1.4253,
      "step": 2438
    },
    {
      "epoch": 0.08734578401704657,
      "grad_norm": 1.79239821434021,
      "learning_rate": 0.00019828072108568016,
      "loss": 1.8537,
      "step": 2439
    },
    {
      "epoch": 0.08738159614661486,
      "grad_norm": 1.204079031944275,
      "learning_rate": 0.0001982785788438563,
      "loss": 1.5567,
      "step": 2440
    },
    {
      "epoch": 0.08741740827618315,
      "grad_norm": 2.0811383724212646,
      "learning_rate": 0.00019827643527982095,
      "loss": 1.806,
      "step": 2441
    },
    {
      "epoch": 0.08745322040575143,
      "grad_norm": 1.5582301616668701,
      "learning_rate": 0.00019827429039360293,
      "loss": 1.7261,
      "step": 2442
    },
    {
      "epoch": 0.08748903253531971,
      "grad_norm": 1.1033960580825806,
      "learning_rate": 0.00019827214418523107,
      "loss": 1.7938,
      "step": 2443
    },
    {
      "epoch": 0.087524844664888,
      "grad_norm": 1.3593032360076904,
      "learning_rate": 0.0001982699966547343,
      "loss": 1.5597,
      "step": 2444
    },
    {
      "epoch": 0.08756065679445628,
      "grad_norm": 1.574497938156128,
      "learning_rate": 0.00019826784780214147,
      "loss": 1.3391,
      "step": 2445
    },
    {
      "epoch": 0.08759646892402456,
      "grad_norm": 2.002777338027954,
      "learning_rate": 0.0001982656976274815,
      "loss": 1.3655,
      "step": 2446
    },
    {
      "epoch": 0.08763228105359286,
      "grad_norm": 2.2492942810058594,
      "learning_rate": 0.00019826354613078332,
      "loss": 1.948,
      "step": 2447
    },
    {
      "epoch": 0.08766809318316114,
      "grad_norm": 1.3055859804153442,
      "learning_rate": 0.0001982613933120759,
      "loss": 1.6317,
      "step": 2448
    },
    {
      "epoch": 0.08770390531272942,
      "grad_norm": 2.0596539974212646,
      "learning_rate": 0.00019825923917138818,
      "loss": 1.8837,
      "step": 2449
    },
    {
      "epoch": 0.08773971744229771,
      "grad_norm": 1.592313289642334,
      "learning_rate": 0.0001982570837087491,
      "loss": 1.5169,
      "step": 2450
    },
    {
      "epoch": 0.08777552957186599,
      "grad_norm": 1.5785194635391235,
      "learning_rate": 0.00019825492692418774,
      "loss": 1.5085,
      "step": 2451
    },
    {
      "epoch": 0.08781134170143427,
      "grad_norm": 1.0770231485366821,
      "learning_rate": 0.00019825276881773308,
      "loss": 1.5045,
      "step": 2452
    },
    {
      "epoch": 0.08784715383100256,
      "grad_norm": 1.7435587644577026,
      "learning_rate": 0.00019825060938941414,
      "loss": 1.7879,
      "step": 2453
    },
    {
      "epoch": 0.08788296596057084,
      "grad_norm": 1.482975959777832,
      "learning_rate": 0.00019824844863925998,
      "loss": 1.722,
      "step": 2454
    },
    {
      "epoch": 0.08791877809013914,
      "grad_norm": 1.9586156606674194,
      "learning_rate": 0.0001982462865672997,
      "loss": 1.6066,
      "step": 2455
    },
    {
      "epoch": 0.08795459021970742,
      "grad_norm": 1.9015264511108398,
      "learning_rate": 0.00019824412317356234,
      "loss": 1.4479,
      "step": 2456
    },
    {
      "epoch": 0.0879904023492757,
      "grad_norm": 1.943438172340393,
      "learning_rate": 0.00019824195845807703,
      "loss": 1.5558,
      "step": 2457
    },
    {
      "epoch": 0.08802621447884398,
      "grad_norm": 1.8411654233932495,
      "learning_rate": 0.00019823979242087288,
      "loss": 1.3188,
      "step": 2458
    },
    {
      "epoch": 0.08806202660841227,
      "grad_norm": 1.542543888092041,
      "learning_rate": 0.00019823762506197907,
      "loss": 1.7211,
      "step": 2459
    },
    {
      "epoch": 0.08809783873798055,
      "grad_norm": 1.1874290704727173,
      "learning_rate": 0.0001982354563814247,
      "loss": 1.4066,
      "step": 2460
    },
    {
      "epoch": 0.08813365086754883,
      "grad_norm": 2.709726095199585,
      "learning_rate": 0.000198233286379239,
      "loss": 1.8428,
      "step": 2461
    },
    {
      "epoch": 0.08816946299711713,
      "grad_norm": 1.5549380779266357,
      "learning_rate": 0.00019823111505545114,
      "loss": 1.537,
      "step": 2462
    },
    {
      "epoch": 0.08820527512668541,
      "grad_norm": 2.0994925498962402,
      "learning_rate": 0.00019822894241009037,
      "loss": 1.4763,
      "step": 2463
    },
    {
      "epoch": 0.0882410872562537,
      "grad_norm": 1.4595978260040283,
      "learning_rate": 0.00019822676844318582,
      "loss": 1.6066,
      "step": 2464
    },
    {
      "epoch": 0.08827689938582198,
      "grad_norm": 1.3861949443817139,
      "learning_rate": 0.00019822459315476686,
      "loss": 1.5408,
      "step": 2465
    },
    {
      "epoch": 0.08831271151539026,
      "grad_norm": 1.820631980895996,
      "learning_rate": 0.00019822241654486266,
      "loss": 1.9246,
      "step": 2466
    },
    {
      "epoch": 0.08834852364495854,
      "grad_norm": 2.3096110820770264,
      "learning_rate": 0.00019822023861350256,
      "loss": 1.8691,
      "step": 2467
    },
    {
      "epoch": 0.08838433577452683,
      "grad_norm": 1.4352091550827026,
      "learning_rate": 0.00019821805936071584,
      "loss": 1.6238,
      "step": 2468
    },
    {
      "epoch": 0.08842014790409512,
      "grad_norm": 1.4374312162399292,
      "learning_rate": 0.00019821587878653184,
      "loss": 1.6068,
      "step": 2469
    },
    {
      "epoch": 0.0884559600336634,
      "grad_norm": 1.8493610620498657,
      "learning_rate": 0.00019821369689097988,
      "loss": 1.6721,
      "step": 2470
    },
    {
      "epoch": 0.08849177216323169,
      "grad_norm": 1.8211166858673096,
      "learning_rate": 0.00019821151367408927,
      "loss": 1.3961,
      "step": 2471
    },
    {
      "epoch": 0.08852758429279997,
      "grad_norm": 2.59659743309021,
      "learning_rate": 0.00019820932913588947,
      "loss": 1.8369,
      "step": 2472
    },
    {
      "epoch": 0.08856339642236825,
      "grad_norm": 1.3814586400985718,
      "learning_rate": 0.00019820714327640983,
      "loss": 1.2698,
      "step": 2473
    },
    {
      "epoch": 0.08859920855193654,
      "grad_norm": 1.4654902219772339,
      "learning_rate": 0.00019820495609567976,
      "loss": 1.6292,
      "step": 2474
    },
    {
      "epoch": 0.08863502068150482,
      "grad_norm": 1.361952543258667,
      "learning_rate": 0.00019820276759372867,
      "loss": 1.9109,
      "step": 2475
    },
    {
      "epoch": 0.08867083281107312,
      "grad_norm": 1.8597922325134277,
      "learning_rate": 0.00019820057777058598,
      "loss": 1.5002,
      "step": 2476
    },
    {
      "epoch": 0.0887066449406414,
      "grad_norm": 1.6137717962265015,
      "learning_rate": 0.00019819838662628122,
      "loss": 1.6735,
      "step": 2477
    },
    {
      "epoch": 0.08874245707020968,
      "grad_norm": 1.753365397453308,
      "learning_rate": 0.00019819619416084385,
      "loss": 1.8299,
      "step": 2478
    },
    {
      "epoch": 0.08877826919977796,
      "grad_norm": 1.4413976669311523,
      "learning_rate": 0.00019819400037430332,
      "loss": 1.5304,
      "step": 2479
    },
    {
      "epoch": 0.08881408132934625,
      "grad_norm": 1.604013442993164,
      "learning_rate": 0.0001981918052666892,
      "loss": 1.5374,
      "step": 2480
    },
    {
      "epoch": 0.08884989345891453,
      "grad_norm": 2.0614285469055176,
      "learning_rate": 0.00019818960883803097,
      "loss": 1.5122,
      "step": 2481
    },
    {
      "epoch": 0.08888570558848281,
      "grad_norm": 1.5279723405838013,
      "learning_rate": 0.00019818741108835824,
      "loss": 1.3099,
      "step": 2482
    },
    {
      "epoch": 0.08892151771805111,
      "grad_norm": 2.621316432952881,
      "learning_rate": 0.00019818521201770052,
      "loss": 1.8033,
      "step": 2483
    },
    {
      "epoch": 0.08895732984761939,
      "grad_norm": 1.9032632112503052,
      "learning_rate": 0.00019818301162608743,
      "loss": 1.7269,
      "step": 2484
    },
    {
      "epoch": 0.08899314197718768,
      "grad_norm": 1.6183884143829346,
      "learning_rate": 0.00019818080991354858,
      "loss": 1.4296,
      "step": 2485
    },
    {
      "epoch": 0.08902895410675596,
      "grad_norm": 1.493016004562378,
      "learning_rate": 0.00019817860688011357,
      "loss": 1.558,
      "step": 2486
    },
    {
      "epoch": 0.08906476623632424,
      "grad_norm": 1.3361989259719849,
      "learning_rate": 0.00019817640252581202,
      "loss": 1.3134,
      "step": 2487
    },
    {
      "epoch": 0.08910057836589252,
      "grad_norm": 2.0151519775390625,
      "learning_rate": 0.00019817419685067364,
      "loss": 2.0462,
      "step": 2488
    },
    {
      "epoch": 0.08913639049546081,
      "grad_norm": 1.6625710725784302,
      "learning_rate": 0.00019817198985472807,
      "loss": 1.5406,
      "step": 2489
    },
    {
      "epoch": 0.0891722026250291,
      "grad_norm": 1.9225102663040161,
      "learning_rate": 0.00019816978153800504,
      "loss": 1.7719,
      "step": 2490
    },
    {
      "epoch": 0.08920801475459739,
      "grad_norm": 1.7934019565582275,
      "learning_rate": 0.00019816757190053416,
      "loss": 1.6073,
      "step": 2491
    },
    {
      "epoch": 0.08924382688416567,
      "grad_norm": 1.8220930099487305,
      "learning_rate": 0.00019816536094234528,
      "loss": 1.7796,
      "step": 2492
    },
    {
      "epoch": 0.08927963901373395,
      "grad_norm": 1.3911093473434448,
      "learning_rate": 0.00019816314866346807,
      "loss": 1.4431,
      "step": 2493
    },
    {
      "epoch": 0.08931545114330224,
      "grad_norm": 1.3377423286437988,
      "learning_rate": 0.00019816093506393233,
      "loss": 1.6441,
      "step": 2494
    },
    {
      "epoch": 0.08935126327287052,
      "grad_norm": 2.121514320373535,
      "learning_rate": 0.00019815872014376784,
      "loss": 1.6375,
      "step": 2495
    },
    {
      "epoch": 0.0893870754024388,
      "grad_norm": 2.1512534618377686,
      "learning_rate": 0.00019815650390300434,
      "loss": 1.8018,
      "step": 2496
    },
    {
      "epoch": 0.0894228875320071,
      "grad_norm": 1.6939420700073242,
      "learning_rate": 0.00019815428634167176,
      "loss": 1.2262,
      "step": 2497
    },
    {
      "epoch": 0.08945869966157538,
      "grad_norm": 1.2576205730438232,
      "learning_rate": 0.00019815206745979981,
      "loss": 1.6021,
      "step": 2498
    },
    {
      "epoch": 0.08949451179114366,
      "grad_norm": 1.6956746578216553,
      "learning_rate": 0.00019814984725741842,
      "loss": 1.5378,
      "step": 2499
    },
    {
      "epoch": 0.08953032392071195,
      "grad_norm": 1.830729365348816,
      "learning_rate": 0.00019814762573455743,
      "loss": 1.9099,
      "step": 2500
    },
    {
      "epoch": 0.08956613605028023,
      "grad_norm": 1.476012110710144,
      "learning_rate": 0.00019814540289124675,
      "loss": 1.3992,
      "step": 2501
    },
    {
      "epoch": 0.08960194817984851,
      "grad_norm": 2.3555774688720703,
      "learning_rate": 0.00019814317872751626,
      "loss": 1.9159,
      "step": 2502
    },
    {
      "epoch": 0.0896377603094168,
      "grad_norm": 1.2655014991760254,
      "learning_rate": 0.0001981409532433959,
      "loss": 1.7051,
      "step": 2503
    },
    {
      "epoch": 0.08967357243898508,
      "grad_norm": 1.7695574760437012,
      "learning_rate": 0.00019813872643891563,
      "loss": 1.7181,
      "step": 2504
    },
    {
      "epoch": 0.08970938456855337,
      "grad_norm": 1.8145886659622192,
      "learning_rate": 0.00019813649831410535,
      "loss": 1.5635,
      "step": 2505
    },
    {
      "epoch": 0.08974519669812166,
      "grad_norm": 1.3770238161087036,
      "learning_rate": 0.00019813426886899509,
      "loss": 1.2597,
      "step": 2506
    },
    {
      "epoch": 0.08978100882768994,
      "grad_norm": 2.0309836864471436,
      "learning_rate": 0.00019813203810361483,
      "loss": 1.4215,
      "step": 2507
    },
    {
      "epoch": 0.08981682095725822,
      "grad_norm": 1.656128168106079,
      "learning_rate": 0.00019812980601799458,
      "loss": 1.5579,
      "step": 2508
    },
    {
      "epoch": 0.0898526330868265,
      "grad_norm": 1.2524288892745972,
      "learning_rate": 0.00019812757261216435,
      "loss": 1.7402,
      "step": 2509
    },
    {
      "epoch": 0.08988844521639479,
      "grad_norm": 1.3505594730377197,
      "learning_rate": 0.0001981253378861542,
      "loss": 1.6665,
      "step": 2510
    },
    {
      "epoch": 0.08992425734596307,
      "grad_norm": 2.9887189865112305,
      "learning_rate": 0.00019812310183999423,
      "loss": 1.9483,
      "step": 2511
    },
    {
      "epoch": 0.08996006947553137,
      "grad_norm": 1.1449756622314453,
      "learning_rate": 0.00019812086447371446,
      "loss": 1.6492,
      "step": 2512
    },
    {
      "epoch": 0.08999588160509965,
      "grad_norm": 1.5508500337600708,
      "learning_rate": 0.00019811862578734507,
      "loss": 1.9411,
      "step": 2513
    },
    {
      "epoch": 0.09003169373466793,
      "grad_norm": 1.6970720291137695,
      "learning_rate": 0.0001981163857809161,
      "loss": 1.7011,
      "step": 2514
    },
    {
      "epoch": 0.09006750586423622,
      "grad_norm": 1.9046391248703003,
      "learning_rate": 0.00019811414445445772,
      "loss": 1.4795,
      "step": 2515
    },
    {
      "epoch": 0.0901033179938045,
      "grad_norm": 1.9710471630096436,
      "learning_rate": 0.00019811190180800013,
      "loss": 1.8442,
      "step": 2516
    },
    {
      "epoch": 0.09013913012337278,
      "grad_norm": 1.5773948431015015,
      "learning_rate": 0.0001981096578415734,
      "loss": 1.569,
      "step": 2517
    },
    {
      "epoch": 0.09017494225294106,
      "grad_norm": 1.4161051511764526,
      "learning_rate": 0.00019810741255520782,
      "loss": 1.4553,
      "step": 2518
    },
    {
      "epoch": 0.09021075438250936,
      "grad_norm": 1.7593234777450562,
      "learning_rate": 0.0001981051659489335,
      "loss": 1.7854,
      "step": 2519
    },
    {
      "epoch": 0.09024656651207764,
      "grad_norm": 1.5166726112365723,
      "learning_rate": 0.00019810291802278078,
      "loss": 1.403,
      "step": 2520
    },
    {
      "epoch": 0.09028237864164593,
      "grad_norm": 1.500707983970642,
      "learning_rate": 0.00019810066877677982,
      "loss": 2.0719,
      "step": 2521
    },
    {
      "epoch": 0.09031819077121421,
      "grad_norm": 2.0746140480041504,
      "learning_rate": 0.00019809841821096086,
      "loss": 1.3798,
      "step": 2522
    },
    {
      "epoch": 0.09035400290078249,
      "grad_norm": 1.631017804145813,
      "learning_rate": 0.00019809616632535427,
      "loss": 1.6161,
      "step": 2523
    },
    {
      "epoch": 0.09038981503035078,
      "grad_norm": 1.3964747190475464,
      "learning_rate": 0.00019809391311999028,
      "loss": 1.7918,
      "step": 2524
    },
    {
      "epoch": 0.09042562715991906,
      "grad_norm": 1.8602650165557861,
      "learning_rate": 0.00019809165859489922,
      "loss": 1.7201,
      "step": 2525
    },
    {
      "epoch": 0.09046143928948736,
      "grad_norm": 1.7261444330215454,
      "learning_rate": 0.00019808940275011145,
      "loss": 1.7615,
      "step": 2526
    },
    {
      "epoch": 0.09049725141905564,
      "grad_norm": 1.6224415302276611,
      "learning_rate": 0.00019808714558565727,
      "loss": 1.1881,
      "step": 2527
    },
    {
      "epoch": 0.09053306354862392,
      "grad_norm": 1.8537174463272095,
      "learning_rate": 0.00019808488710156707,
      "loss": 1.7823,
      "step": 2528
    },
    {
      "epoch": 0.0905688756781922,
      "grad_norm": 2.799915313720703,
      "learning_rate": 0.0001980826272978712,
      "loss": 1.5794,
      "step": 2529
    },
    {
      "epoch": 0.09060468780776049,
      "grad_norm": 1.8718516826629639,
      "learning_rate": 0.00019808036617460016,
      "loss": 1.6623,
      "step": 2530
    },
    {
      "epoch": 0.09064049993732877,
      "grad_norm": 1.234557867050171,
      "learning_rate": 0.00019807810373178425,
      "loss": 1.5993,
      "step": 2531
    },
    {
      "epoch": 0.09067631206689705,
      "grad_norm": 1.8687853813171387,
      "learning_rate": 0.000198075839969454,
      "loss": 1.6352,
      "step": 2532
    },
    {
      "epoch": 0.09071212419646535,
      "grad_norm": 1.8785006999969482,
      "learning_rate": 0.00019807357488763985,
      "loss": 1.7187,
      "step": 2533
    },
    {
      "epoch": 0.09074793632603363,
      "grad_norm": 1.3195416927337646,
      "learning_rate": 0.00019807130848637224,
      "loss": 1.7064,
      "step": 2534
    },
    {
      "epoch": 0.09078374845560191,
      "grad_norm": 2.186288833618164,
      "learning_rate": 0.00019806904076568165,
      "loss": 1.5541,
      "step": 2535
    },
    {
      "epoch": 0.0908195605851702,
      "grad_norm": 1.6952531337738037,
      "learning_rate": 0.00019806677172559865,
      "loss": 1.3094,
      "step": 2536
    },
    {
      "epoch": 0.09085537271473848,
      "grad_norm": 4.69820499420166,
      "learning_rate": 0.00019806450136615372,
      "loss": 1.5628,
      "step": 2537
    },
    {
      "epoch": 0.09089118484430676,
      "grad_norm": 1.3726061582565308,
      "learning_rate": 0.0001980622296873774,
      "loss": 1.7239,
      "step": 2538
    },
    {
      "epoch": 0.09092699697387505,
      "grad_norm": 1.5278609991073608,
      "learning_rate": 0.0001980599566893003,
      "loss": 2.0344,
      "step": 2539
    },
    {
      "epoch": 0.09096280910344334,
      "grad_norm": 1.4546926021575928,
      "learning_rate": 0.00019805768237195296,
      "loss": 1.671,
      "step": 2540
    },
    {
      "epoch": 0.09099862123301163,
      "grad_norm": 2.053382396697998,
      "learning_rate": 0.00019805540673536597,
      "loss": 1.7412,
      "step": 2541
    },
    {
      "epoch": 0.09103443336257991,
      "grad_norm": 2.1475985050201416,
      "learning_rate": 0.00019805312977956997,
      "loss": 1.3434,
      "step": 2542
    },
    {
      "epoch": 0.09107024549214819,
      "grad_norm": 2.8895585536956787,
      "learning_rate": 0.0001980508515045956,
      "loss": 1.545,
      "step": 2543
    },
    {
      "epoch": 0.09110605762171647,
      "grad_norm": 1.509041428565979,
      "learning_rate": 0.00019804857191047353,
      "loss": 1.936,
      "step": 2544
    },
    {
      "epoch": 0.09114186975128476,
      "grad_norm": 2.280407190322876,
      "learning_rate": 0.00019804629099723435,
      "loss": 1.9202,
      "step": 2545
    },
    {
      "epoch": 0.09117768188085304,
      "grad_norm": 2.5649526119232178,
      "learning_rate": 0.00019804400876490883,
      "loss": 1.7697,
      "step": 2546
    },
    {
      "epoch": 0.09121349401042134,
      "grad_norm": 2.6087114810943604,
      "learning_rate": 0.00019804172521352761,
      "loss": 1.6824,
      "step": 2547
    },
    {
      "epoch": 0.09124930613998962,
      "grad_norm": 1.8538936376571655,
      "learning_rate": 0.00019803944034312148,
      "loss": 1.6512,
      "step": 2548
    },
    {
      "epoch": 0.0912851182695579,
      "grad_norm": 2.424487352371216,
      "learning_rate": 0.0001980371541537211,
      "loss": 1.4249,
      "step": 2549
    },
    {
      "epoch": 0.09132093039912619,
      "grad_norm": 1.9333043098449707,
      "learning_rate": 0.0001980348666453573,
      "loss": 1.8535,
      "step": 2550
    },
    {
      "epoch": 0.09135674252869447,
      "grad_norm": 1.7312296628952026,
      "learning_rate": 0.00019803257781806082,
      "loss": 1.7266,
      "step": 2551
    },
    {
      "epoch": 0.09139255465826275,
      "grad_norm": 1.5011173486709595,
      "learning_rate": 0.00019803028767186246,
      "loss": 1.7465,
      "step": 2552
    },
    {
      "epoch": 0.09142836678783103,
      "grad_norm": 2.4707727432250977,
      "learning_rate": 0.000198027996206793,
      "loss": 1.6072,
      "step": 2553
    },
    {
      "epoch": 0.09146417891739932,
      "grad_norm": 3.124415159225464,
      "learning_rate": 0.0001980257034228833,
      "loss": 1.8117,
      "step": 2554
    },
    {
      "epoch": 0.09149999104696761,
      "grad_norm": 1.7809544801712036,
      "learning_rate": 0.00019802340932016424,
      "loss": 1.7285,
      "step": 2555
    },
    {
      "epoch": 0.0915358031765359,
      "grad_norm": 1.9475129842758179,
      "learning_rate": 0.00019802111389866664,
      "loss": 1.6601,
      "step": 2556
    },
    {
      "epoch": 0.09157161530610418,
      "grad_norm": 1.3887070417404175,
      "learning_rate": 0.00019801881715842136,
      "loss": 1.5281,
      "step": 2557
    },
    {
      "epoch": 0.09160742743567246,
      "grad_norm": 1.341725468635559,
      "learning_rate": 0.00019801651909945935,
      "loss": 1.5264,
      "step": 2558
    },
    {
      "epoch": 0.09164323956524074,
      "grad_norm": 1.8760310411453247,
      "learning_rate": 0.0001980142197218115,
      "loss": 1.6517,
      "step": 2559
    },
    {
      "epoch": 0.09167905169480903,
      "grad_norm": 2.533712863922119,
      "learning_rate": 0.0001980119190255088,
      "loss": 1.6518,
      "step": 2560
    },
    {
      "epoch": 0.09171486382437731,
      "grad_norm": 2.026277780532837,
      "learning_rate": 0.0001980096170105821,
      "loss": 1.5654,
      "step": 2561
    },
    {
      "epoch": 0.0917506759539456,
      "grad_norm": 2.1434755325317383,
      "learning_rate": 0.00019800731367706248,
      "loss": 1.7728,
      "step": 2562
    },
    {
      "epoch": 0.09178648808351389,
      "grad_norm": 1.6237499713897705,
      "learning_rate": 0.0001980050090249808,
      "loss": 1.651,
      "step": 2563
    },
    {
      "epoch": 0.09182230021308217,
      "grad_norm": 1.460422396659851,
      "learning_rate": 0.0001980027030543682,
      "loss": 1.5255,
      "step": 2564
    },
    {
      "epoch": 0.09185811234265046,
      "grad_norm": 1.3004599809646606,
      "learning_rate": 0.00019800039576525562,
      "loss": 1.7805,
      "step": 2565
    },
    {
      "epoch": 0.09189392447221874,
      "grad_norm": 1.7502729892730713,
      "learning_rate": 0.00019799808715767413,
      "loss": 1.8473,
      "step": 2566
    },
    {
      "epoch": 0.09192973660178702,
      "grad_norm": 1.482386827468872,
      "learning_rate": 0.00019799577723165479,
      "loss": 1.647,
      "step": 2567
    },
    {
      "epoch": 0.0919655487313553,
      "grad_norm": 2.2665460109710693,
      "learning_rate": 0.0001979934659872287,
      "loss": 1.8273,
      "step": 2568
    },
    {
      "epoch": 0.0920013608609236,
      "grad_norm": 1.5462538003921509,
      "learning_rate": 0.00019799115342442687,
      "loss": 1.6563,
      "step": 2569
    },
    {
      "epoch": 0.09203717299049188,
      "grad_norm": 1.4237371683120728,
      "learning_rate": 0.0001979888395432805,
      "loss": 1.7318,
      "step": 2570
    },
    {
      "epoch": 0.09207298512006017,
      "grad_norm": 1.7229256629943848,
      "learning_rate": 0.00019798652434382068,
      "loss": 1.7655,
      "step": 2571
    },
    {
      "epoch": 0.09210879724962845,
      "grad_norm": 1.699660062789917,
      "learning_rate": 0.0001979842078260786,
      "loss": 1.4854,
      "step": 2572
    },
    {
      "epoch": 0.09214460937919673,
      "grad_norm": 1.2170562744140625,
      "learning_rate": 0.00019798188999008536,
      "loss": 1.2979,
      "step": 2573
    },
    {
      "epoch": 0.09218042150876501,
      "grad_norm": 1.5595293045043945,
      "learning_rate": 0.00019797957083587218,
      "loss": 1.4223,
      "step": 2574
    },
    {
      "epoch": 0.0922162336383333,
      "grad_norm": 1.6665608882904053,
      "learning_rate": 0.00019797725036347025,
      "loss": 1.5744,
      "step": 2575
    },
    {
      "epoch": 0.0922520457679016,
      "grad_norm": 1.5953999757766724,
      "learning_rate": 0.00019797492857291085,
      "loss": 1.4263,
      "step": 2576
    },
    {
      "epoch": 0.09228785789746988,
      "grad_norm": 1.8051010370254517,
      "learning_rate": 0.00019797260546422512,
      "loss": 2.0402,
      "step": 2577
    },
    {
      "epoch": 0.09232367002703816,
      "grad_norm": 2.281867027282715,
      "learning_rate": 0.00019797028103744438,
      "loss": 1.7244,
      "step": 2578
    },
    {
      "epoch": 0.09235948215660644,
      "grad_norm": 1.6062850952148438,
      "learning_rate": 0.00019796795529259986,
      "loss": 1.6111,
      "step": 2579
    },
    {
      "epoch": 0.09239529428617473,
      "grad_norm": 1.4826256036758423,
      "learning_rate": 0.0001979656282297229,
      "loss": 1.442,
      "step": 2580
    },
    {
      "epoch": 0.09243110641574301,
      "grad_norm": 2.0289320945739746,
      "learning_rate": 0.00019796329984884473,
      "loss": 1.986,
      "step": 2581
    },
    {
      "epoch": 0.09246691854531129,
      "grad_norm": 1.5660099983215332,
      "learning_rate": 0.00019796097014999678,
      "loss": 1.7942,
      "step": 2582
    },
    {
      "epoch": 0.09250273067487959,
      "grad_norm": 1.9263335466384888,
      "learning_rate": 0.0001979586391332103,
      "loss": 1.7872,
      "step": 2583
    },
    {
      "epoch": 0.09253854280444787,
      "grad_norm": 1.0887078046798706,
      "learning_rate": 0.0001979563067985167,
      "loss": 1.591,
      "step": 2584
    },
    {
      "epoch": 0.09257435493401615,
      "grad_norm": 1.3899720907211304,
      "learning_rate": 0.00019795397314594735,
      "loss": 1.4923,
      "step": 2585
    },
    {
      "epoch": 0.09261016706358444,
      "grad_norm": 1.539726972579956,
      "learning_rate": 0.00019795163817553363,
      "loss": 1.5147,
      "step": 2586
    },
    {
      "epoch": 0.09264597919315272,
      "grad_norm": 2.2429847717285156,
      "learning_rate": 0.000197949301887307,
      "loss": 1.6219,
      "step": 2587
    },
    {
      "epoch": 0.092681791322721,
      "grad_norm": 1.6642656326293945,
      "learning_rate": 0.00019794696428129883,
      "loss": 1.3772,
      "step": 2588
    },
    {
      "epoch": 0.09271760345228929,
      "grad_norm": 1.2310926914215088,
      "learning_rate": 0.0001979446253575406,
      "loss": 1.5615,
      "step": 2589
    },
    {
      "epoch": 0.09275341558185758,
      "grad_norm": 2.3761885166168213,
      "learning_rate": 0.00019794228511606376,
      "loss": 1.5446,
      "step": 2590
    },
    {
      "epoch": 0.09278922771142586,
      "grad_norm": 1.7739567756652832,
      "learning_rate": 0.00019793994355689985,
      "loss": 1.5243,
      "step": 2591
    },
    {
      "epoch": 0.09282503984099415,
      "grad_norm": 1.9059218168258667,
      "learning_rate": 0.0001979376006800803,
      "loss": 1.5244,
      "step": 2592
    },
    {
      "epoch": 0.09286085197056243,
      "grad_norm": 2.342510938644409,
      "learning_rate": 0.00019793525648563668,
      "loss": 1.6978,
      "step": 2593
    },
    {
      "epoch": 0.09289666410013071,
      "grad_norm": 1.4292327165603638,
      "learning_rate": 0.0001979329109736005,
      "loss": 1.78,
      "step": 2594
    },
    {
      "epoch": 0.092932476229699,
      "grad_norm": 2.344578504562378,
      "learning_rate": 0.00019793056414400332,
      "loss": 1.8671,
      "step": 2595
    },
    {
      "epoch": 0.09296828835926728,
      "grad_norm": 1.9726289510726929,
      "learning_rate": 0.00019792821599687676,
      "loss": 1.5574,
      "step": 2596
    },
    {
      "epoch": 0.09300410048883558,
      "grad_norm": 1.629486322402954,
      "learning_rate": 0.00019792586653225237,
      "loss": 1.5039,
      "step": 2597
    },
    {
      "epoch": 0.09303991261840386,
      "grad_norm": 1.6008448600769043,
      "learning_rate": 0.00019792351575016173,
      "loss": 1.5486,
      "step": 2598
    },
    {
      "epoch": 0.09307572474797214,
      "grad_norm": 1.9387474060058594,
      "learning_rate": 0.0001979211636506365,
      "loss": 1.5127,
      "step": 2599
    },
    {
      "epoch": 0.09311153687754042,
      "grad_norm": 2.0888354778289795,
      "learning_rate": 0.0001979188102337083,
      "loss": 1.6695,
      "step": 2600
    },
    {
      "epoch": 0.0931473490071087,
      "grad_norm": 1.6669467687606812,
      "learning_rate": 0.00019791645549940886,
      "loss": 1.6549,
      "step": 2601
    },
    {
      "epoch": 0.09318316113667699,
      "grad_norm": 1.3782527446746826,
      "learning_rate": 0.0001979140994477698,
      "loss": 1.6938,
      "step": 2602
    },
    {
      "epoch": 0.09321897326624527,
      "grad_norm": 1.7212045192718506,
      "learning_rate": 0.00019791174207882284,
      "loss": 1.3562,
      "step": 2603
    },
    {
      "epoch": 0.09325478539581356,
      "grad_norm": 1.2293366193771362,
      "learning_rate": 0.00019790938339259967,
      "loss": 1.65,
      "step": 2604
    },
    {
      "epoch": 0.09329059752538185,
      "grad_norm": 1.813463568687439,
      "learning_rate": 0.00019790702338913204,
      "loss": 1.5209,
      "step": 2605
    },
    {
      "epoch": 0.09332640965495013,
      "grad_norm": 1.8608582019805908,
      "learning_rate": 0.0001979046620684517,
      "loss": 1.5321,
      "step": 2606
    },
    {
      "epoch": 0.09336222178451842,
      "grad_norm": 2.023027181625366,
      "learning_rate": 0.00019790229943059045,
      "loss": 1.6548,
      "step": 2607
    },
    {
      "epoch": 0.0933980339140867,
      "grad_norm": 1.4660706520080566,
      "learning_rate": 0.00019789993547558,
      "loss": 1.7817,
      "step": 2608
    },
    {
      "epoch": 0.09343384604365498,
      "grad_norm": 1.455670714378357,
      "learning_rate": 0.00019789757020345224,
      "loss": 1.6645,
      "step": 2609
    },
    {
      "epoch": 0.09346965817322327,
      "grad_norm": 1.8113867044448853,
      "learning_rate": 0.00019789520361423893,
      "loss": 1.7678,
      "step": 2610
    },
    {
      "epoch": 0.09350547030279155,
      "grad_norm": 1.1966438293457031,
      "learning_rate": 0.00019789283570797192,
      "loss": 1.7918,
      "step": 2611
    },
    {
      "epoch": 0.09354128243235985,
      "grad_norm": 1.6652956008911133,
      "learning_rate": 0.0001978904664846831,
      "loss": 1.5339,
      "step": 2612
    },
    {
      "epoch": 0.09357709456192813,
      "grad_norm": 1.756519079208374,
      "learning_rate": 0.00019788809594440432,
      "loss": 1.5227,
      "step": 2613
    },
    {
      "epoch": 0.09361290669149641,
      "grad_norm": 1.711736798286438,
      "learning_rate": 0.00019788572408716747,
      "loss": 1.8176,
      "step": 2614
    },
    {
      "epoch": 0.0936487188210647,
      "grad_norm": 2.1952521800994873,
      "learning_rate": 0.00019788335091300448,
      "loss": 1.7203,
      "step": 2615
    },
    {
      "epoch": 0.09368453095063298,
      "grad_norm": 1.3425790071487427,
      "learning_rate": 0.00019788097642194725,
      "loss": 1.7342,
      "step": 2616
    },
    {
      "epoch": 0.09372034308020126,
      "grad_norm": 1.7646204233169556,
      "learning_rate": 0.00019787860061402774,
      "loss": 1.5705,
      "step": 2617
    },
    {
      "epoch": 0.09375615520976954,
      "grad_norm": 2.6315975189208984,
      "learning_rate": 0.00019787622348927793,
      "loss": 1.5904,
      "step": 2618
    },
    {
      "epoch": 0.09379196733933784,
      "grad_norm": 2.2786176204681396,
      "learning_rate": 0.00019787384504772976,
      "loss": 1.4719,
      "step": 2619
    },
    {
      "epoch": 0.09382777946890612,
      "grad_norm": 3.4464972019195557,
      "learning_rate": 0.00019787146528941528,
      "loss": 2.1013,
      "step": 2620
    },
    {
      "epoch": 0.0938635915984744,
      "grad_norm": 1.4851138591766357,
      "learning_rate": 0.00019786908421436645,
      "loss": 1.6989,
      "step": 2621
    },
    {
      "epoch": 0.09389940372804269,
      "grad_norm": 1.547202229499817,
      "learning_rate": 0.00019786670182261534,
      "loss": 1.8292,
      "step": 2622
    },
    {
      "epoch": 0.09393521585761097,
      "grad_norm": 1.8076845407485962,
      "learning_rate": 0.00019786431811419402,
      "loss": 1.5035,
      "step": 2623
    },
    {
      "epoch": 0.09397102798717925,
      "grad_norm": 2.3341751098632812,
      "learning_rate": 0.0001978619330891345,
      "loss": 2.0307,
      "step": 2624
    },
    {
      "epoch": 0.09400684011674754,
      "grad_norm": 2.7499022483825684,
      "learning_rate": 0.0001978595467474689,
      "loss": 1.51,
      "step": 2625
    },
    {
      "epoch": 0.09404265224631583,
      "grad_norm": 1.925464153289795,
      "learning_rate": 0.00019785715908922938,
      "loss": 1.4552,
      "step": 2626
    },
    {
      "epoch": 0.09407846437588412,
      "grad_norm": 2.460425853729248,
      "learning_rate": 0.00019785477011444798,
      "loss": 1.5344,
      "step": 2627
    },
    {
      "epoch": 0.0941142765054524,
      "grad_norm": 1.5089377164840698,
      "learning_rate": 0.00019785237982315686,
      "loss": 1.6439,
      "step": 2628
    },
    {
      "epoch": 0.09415008863502068,
      "grad_norm": 1.501597285270691,
      "learning_rate": 0.0001978499882153882,
      "loss": 1.5609,
      "step": 2629
    },
    {
      "epoch": 0.09418590076458896,
      "grad_norm": 1.284104824066162,
      "learning_rate": 0.00019784759529117415,
      "loss": 1.4742,
      "step": 2630
    },
    {
      "epoch": 0.09422171289415725,
      "grad_norm": 2.3818297386169434,
      "learning_rate": 0.000197845201050547,
      "loss": 1.5001,
      "step": 2631
    },
    {
      "epoch": 0.09425752502372553,
      "grad_norm": 1.6480205059051514,
      "learning_rate": 0.0001978428054935388,
      "loss": 1.5732,
      "step": 2632
    },
    {
      "epoch": 0.09429333715329383,
      "grad_norm": 2.133923053741455,
      "learning_rate": 0.00019784040862018184,
      "loss": 1.5003,
      "step": 2633
    },
    {
      "epoch": 0.09432914928286211,
      "grad_norm": 1.5461385250091553,
      "learning_rate": 0.00019783801043050844,
      "loss": 1.4475,
      "step": 2634
    },
    {
      "epoch": 0.09436496141243039,
      "grad_norm": 1.5656708478927612,
      "learning_rate": 0.0001978356109245508,
      "loss": 1.5606,
      "step": 2635
    },
    {
      "epoch": 0.09440077354199868,
      "grad_norm": 1.454704999923706,
      "learning_rate": 0.00019783321010234122,
      "loss": 1.643,
      "step": 2636
    },
    {
      "epoch": 0.09443658567156696,
      "grad_norm": 1.9194536209106445,
      "learning_rate": 0.000197830807963912,
      "loss": 1.4815,
      "step": 2637
    },
    {
      "epoch": 0.09447239780113524,
      "grad_norm": 1.1359739303588867,
      "learning_rate": 0.00019782840450929543,
      "loss": 1.6183,
      "step": 2638
    },
    {
      "epoch": 0.09450820993070352,
      "grad_norm": 1.4093936681747437,
      "learning_rate": 0.00019782599973852387,
      "loss": 1.7041,
      "step": 2639
    },
    {
      "epoch": 0.09454402206027182,
      "grad_norm": 1.9672014713287354,
      "learning_rate": 0.0001978235936516297,
      "loss": 1.4104,
      "step": 2640
    },
    {
      "epoch": 0.0945798341898401,
      "grad_norm": 1.2967243194580078,
      "learning_rate": 0.0001978211862486452,
      "loss": 1.5651,
      "step": 2641
    },
    {
      "epoch": 0.09461564631940839,
      "grad_norm": 1.4249886274337769,
      "learning_rate": 0.00019781877752960285,
      "loss": 1.6584,
      "step": 2642
    },
    {
      "epoch": 0.09465145844897667,
      "grad_norm": 1.4482895135879517,
      "learning_rate": 0.00019781636749453504,
      "loss": 1.6222,
      "step": 2643
    },
    {
      "epoch": 0.09468727057854495,
      "grad_norm": 1.9295979738235474,
      "learning_rate": 0.00019781395614347415,
      "loss": 1.778,
      "step": 2644
    },
    {
      "epoch": 0.09472308270811323,
      "grad_norm": 1.6828997135162354,
      "learning_rate": 0.0001978115434764527,
      "loss": 1.602,
      "step": 2645
    },
    {
      "epoch": 0.09475889483768152,
      "grad_norm": 1.602980375289917,
      "learning_rate": 0.00019780912949350307,
      "loss": 1.5946,
      "step": 2646
    },
    {
      "epoch": 0.09479470696724981,
      "grad_norm": 1.2852801084518433,
      "learning_rate": 0.00019780671419465776,
      "loss": 1.3936,
      "step": 2647
    },
    {
      "epoch": 0.0948305190968181,
      "grad_norm": 3.3350985050201416,
      "learning_rate": 0.00019780429757994928,
      "loss": 1.587,
      "step": 2648
    },
    {
      "epoch": 0.09486633122638638,
      "grad_norm": 1.130829930305481,
      "learning_rate": 0.00019780187964941011,
      "loss": 1.479,
      "step": 2649
    },
    {
      "epoch": 0.09490214335595466,
      "grad_norm": 1.76201593875885,
      "learning_rate": 0.00019779946040307284,
      "loss": 1.5925,
      "step": 2650
    },
    {
      "epoch": 0.09493795548552295,
      "grad_norm": 2.2327420711517334,
      "learning_rate": 0.00019779703984096998,
      "loss": 1.6929,
      "step": 2651
    },
    {
      "epoch": 0.09497376761509123,
      "grad_norm": 2.0052194595336914,
      "learning_rate": 0.00019779461796313408,
      "loss": 1.4586,
      "step": 2652
    },
    {
      "epoch": 0.09500957974465951,
      "grad_norm": 1.6383410692214966,
      "learning_rate": 0.00019779219476959777,
      "loss": 1.4045,
      "step": 2653
    },
    {
      "epoch": 0.0950453918742278,
      "grad_norm": 2.5757429599761963,
      "learning_rate": 0.0001977897702603936,
      "loss": 1.7264,
      "step": 2654
    },
    {
      "epoch": 0.09508120400379609,
      "grad_norm": 2.200880289077759,
      "learning_rate": 0.0001977873444355542,
      "loss": 1.7437,
      "step": 2655
    },
    {
      "epoch": 0.09511701613336437,
      "grad_norm": 2.070451498031616,
      "learning_rate": 0.00019778491729511224,
      "loss": 1.634,
      "step": 2656
    },
    {
      "epoch": 0.09515282826293266,
      "grad_norm": 1.4518322944641113,
      "learning_rate": 0.00019778248883910035,
      "loss": 1.7425,
      "step": 2657
    },
    {
      "epoch": 0.09518864039250094,
      "grad_norm": 1.1998945474624634,
      "learning_rate": 0.0001977800590675512,
      "loss": 1.4792,
      "step": 2658
    },
    {
      "epoch": 0.09522445252206922,
      "grad_norm": 1.4403979778289795,
      "learning_rate": 0.0001977776279804975,
      "loss": 1.5377,
      "step": 2659
    },
    {
      "epoch": 0.0952602646516375,
      "grad_norm": 1.753049373626709,
      "learning_rate": 0.0001977751955779719,
      "loss": 1.1242,
      "step": 2660
    },
    {
      "epoch": 0.09529607678120579,
      "grad_norm": 2.60457706451416,
      "learning_rate": 0.00019777276186000716,
      "loss": 1.3977,
      "step": 2661
    },
    {
      "epoch": 0.09533188891077408,
      "grad_norm": 2.1619999408721924,
      "learning_rate": 0.00019777032682663607,
      "loss": 1.8488,
      "step": 2662
    },
    {
      "epoch": 0.09536770104034237,
      "grad_norm": 1.300704002380371,
      "learning_rate": 0.00019776789047789133,
      "loss": 1.7853,
      "step": 2663
    },
    {
      "epoch": 0.09540351316991065,
      "grad_norm": 1.6537222862243652,
      "learning_rate": 0.0001977654528138057,
      "loss": 1.5626,
      "step": 2664
    },
    {
      "epoch": 0.09543932529947893,
      "grad_norm": 1.322972059249878,
      "learning_rate": 0.00019776301383441207,
      "loss": 1.5224,
      "step": 2665
    },
    {
      "epoch": 0.09547513742904722,
      "grad_norm": 1.761858582496643,
      "learning_rate": 0.00019776057353974315,
      "loss": 1.7065,
      "step": 2666
    },
    {
      "epoch": 0.0955109495586155,
      "grad_norm": 1.886605143547058,
      "learning_rate": 0.00019775813192983183,
      "loss": 1.2394,
      "step": 2667
    },
    {
      "epoch": 0.09554676168818378,
      "grad_norm": 2.51098370552063,
      "learning_rate": 0.00019775568900471096,
      "loss": 1.3918,
      "step": 2668
    },
    {
      "epoch": 0.09558257381775208,
      "grad_norm": 2.1192142963409424,
      "learning_rate": 0.00019775324476441336,
      "loss": 1.4485,
      "step": 2669
    },
    {
      "epoch": 0.09561838594732036,
      "grad_norm": 2.188304901123047,
      "learning_rate": 0.00019775079920897196,
      "loss": 1.7083,
      "step": 2670
    },
    {
      "epoch": 0.09565419807688864,
      "grad_norm": 1.3558300733566284,
      "learning_rate": 0.00019774835233841965,
      "loss": 1.7486,
      "step": 2671
    },
    {
      "epoch": 0.09569001020645693,
      "grad_norm": 1.7164361476898193,
      "learning_rate": 0.00019774590415278933,
      "loss": 1.4583,
      "step": 2672
    },
    {
      "epoch": 0.09572582233602521,
      "grad_norm": 1.419765830039978,
      "learning_rate": 0.00019774345465211398,
      "loss": 2.0221,
      "step": 2673
    },
    {
      "epoch": 0.09576163446559349,
      "grad_norm": 1.728421926498413,
      "learning_rate": 0.00019774100383642651,
      "loss": 1.3956,
      "step": 2674
    },
    {
      "epoch": 0.09579744659516178,
      "grad_norm": 2.3539481163024902,
      "learning_rate": 0.0001977385517057599,
      "loss": 1.6057,
      "step": 2675
    },
    {
      "epoch": 0.09583325872473007,
      "grad_norm": 2.140263319015503,
      "learning_rate": 0.00019773609826014718,
      "loss": 1.7721,
      "step": 2676
    },
    {
      "epoch": 0.09586907085429835,
      "grad_norm": 1.8717950582504272,
      "learning_rate": 0.0001977336434996213,
      "loss": 1.3382,
      "step": 2677
    },
    {
      "epoch": 0.09590488298386664,
      "grad_norm": 1.4856847524642944,
      "learning_rate": 0.00019773118742421532,
      "loss": 1.4095,
      "step": 2678
    },
    {
      "epoch": 0.09594069511343492,
      "grad_norm": 2.1832220554351807,
      "learning_rate": 0.00019772873003396228,
      "loss": 1.6058,
      "step": 2679
    },
    {
      "epoch": 0.0959765072430032,
      "grad_norm": 1.603605031967163,
      "learning_rate": 0.00019772627132889526,
      "loss": 1.5258,
      "step": 2680
    },
    {
      "epoch": 0.09601231937257149,
      "grad_norm": 1.6799068450927734,
      "learning_rate": 0.00019772381130904728,
      "loss": 1.6569,
      "step": 2681
    },
    {
      "epoch": 0.09604813150213977,
      "grad_norm": 1.3893667459487915,
      "learning_rate": 0.0001977213499744515,
      "loss": 1.6977,
      "step": 2682
    },
    {
      "epoch": 0.09608394363170807,
      "grad_norm": 2.7440836429595947,
      "learning_rate": 0.00019771888732514098,
      "loss": 1.495,
      "step": 2683
    },
    {
      "epoch": 0.09611975576127635,
      "grad_norm": 1.2910890579223633,
      "learning_rate": 0.00019771642336114892,
      "loss": 1.1927,
      "step": 2684
    },
    {
      "epoch": 0.09615556789084463,
      "grad_norm": 1.6925691366195679,
      "learning_rate": 0.0001977139580825084,
      "loss": 1.2957,
      "step": 2685
    },
    {
      "epoch": 0.09619138002041291,
      "grad_norm": 2.3331246376037598,
      "learning_rate": 0.0001977114914892526,
      "loss": 1.7952,
      "step": 2686
    },
    {
      "epoch": 0.0962271921499812,
      "grad_norm": 2.0909788608551025,
      "learning_rate": 0.00019770902358141478,
      "loss": 1.4007,
      "step": 2687
    },
    {
      "epoch": 0.09626300427954948,
      "grad_norm": 1.5967774391174316,
      "learning_rate": 0.00019770655435902805,
      "loss": 1.4887,
      "step": 2688
    },
    {
      "epoch": 0.09629881640911776,
      "grad_norm": 1.9621723890304565,
      "learning_rate": 0.00019770408382212564,
      "loss": 1.1734,
      "step": 2689
    },
    {
      "epoch": 0.09633462853868606,
      "grad_norm": 1.2921861410140991,
      "learning_rate": 0.00019770161197074084,
      "loss": 1.6445,
      "step": 2690
    },
    {
      "epoch": 0.09637044066825434,
      "grad_norm": 1.2130435705184937,
      "learning_rate": 0.00019769913880490688,
      "loss": 1.6201,
      "step": 2691
    },
    {
      "epoch": 0.09640625279782263,
      "grad_norm": 1.7521440982818604,
      "learning_rate": 0.000197696664324657,
      "loss": 1.6516,
      "step": 2692
    },
    {
      "epoch": 0.09644206492739091,
      "grad_norm": 1.5702006816864014,
      "learning_rate": 0.00019769418853002454,
      "loss": 1.2543,
      "step": 2693
    },
    {
      "epoch": 0.09647787705695919,
      "grad_norm": 1.7723594903945923,
      "learning_rate": 0.0001976917114210428,
      "loss": 1.5552,
      "step": 2694
    },
    {
      "epoch": 0.09651368918652747,
      "grad_norm": 1.4246082305908203,
      "learning_rate": 0.00019768923299774506,
      "loss": 1.6263,
      "step": 2695
    },
    {
      "epoch": 0.09654950131609576,
      "grad_norm": 1.6311163902282715,
      "learning_rate": 0.00019768675326016475,
      "loss": 1.5833,
      "step": 2696
    },
    {
      "epoch": 0.09658531344566405,
      "grad_norm": 1.1841009855270386,
      "learning_rate": 0.00019768427220833514,
      "loss": 1.3689,
      "step": 2697
    },
    {
      "epoch": 0.09662112557523234,
      "grad_norm": 1.5306531190872192,
      "learning_rate": 0.00019768178984228967,
      "loss": 1.8065,
      "step": 2698
    },
    {
      "epoch": 0.09665693770480062,
      "grad_norm": 1.5561325550079346,
      "learning_rate": 0.00019767930616206174,
      "loss": 1.6343,
      "step": 2699
    },
    {
      "epoch": 0.0966927498343689,
      "grad_norm": 2.01039457321167,
      "learning_rate": 0.00019767682116768472,
      "loss": 1.4323,
      "step": 2700
    },
    {
      "epoch": 0.09672856196393718,
      "grad_norm": 1.3528622388839722,
      "learning_rate": 0.00019767433485919206,
      "loss": 1.5957,
      "step": 2701
    },
    {
      "epoch": 0.09676437409350547,
      "grad_norm": 1.388312816619873,
      "learning_rate": 0.0001976718472366172,
      "loss": 1.5354,
      "step": 2702
    },
    {
      "epoch": 0.09680018622307375,
      "grad_norm": 1.7916830778121948,
      "learning_rate": 0.00019766935829999363,
      "loss": 1.7827,
      "step": 2703
    },
    {
      "epoch": 0.09683599835264203,
      "grad_norm": 2.035825729370117,
      "learning_rate": 0.00019766686804935488,
      "loss": 1.52,
      "step": 2704
    },
    {
      "epoch": 0.09687181048221033,
      "grad_norm": 1.4044291973114014,
      "learning_rate": 0.00019766437648473435,
      "loss": 1.778,
      "step": 2705
    },
    {
      "epoch": 0.09690762261177861,
      "grad_norm": 1.3465924263000488,
      "learning_rate": 0.00019766188360616563,
      "loss": 1.6834,
      "step": 2706
    },
    {
      "epoch": 0.0969434347413469,
      "grad_norm": 1.3629565238952637,
      "learning_rate": 0.00019765938941368222,
      "loss": 1.7601,
      "step": 2707
    },
    {
      "epoch": 0.09697924687091518,
      "grad_norm": 1.55349600315094,
      "learning_rate": 0.00019765689390731773,
      "loss": 1.8671,
      "step": 2708
    },
    {
      "epoch": 0.09701505900048346,
      "grad_norm": 1.5286920070648193,
      "learning_rate": 0.0001976543970871057,
      "loss": 1.6034,
      "step": 2709
    },
    {
      "epoch": 0.09705087113005174,
      "grad_norm": 1.5274664163589478,
      "learning_rate": 0.0001976518989530797,
      "loss": 1.7297,
      "step": 2710
    },
    {
      "epoch": 0.09708668325962003,
      "grad_norm": 1.5408055782318115,
      "learning_rate": 0.00019764939950527336,
      "loss": 1.5561,
      "step": 2711
    },
    {
      "epoch": 0.09712249538918832,
      "grad_norm": 1.8871009349822998,
      "learning_rate": 0.0001976468987437203,
      "loss": 1.5792,
      "step": 2712
    },
    {
      "epoch": 0.0971583075187566,
      "grad_norm": 1.9408537149429321,
      "learning_rate": 0.0001976443966684542,
      "loss": 1.6264,
      "step": 2713
    },
    {
      "epoch": 0.09719411964832489,
      "grad_norm": 1.510097861289978,
      "learning_rate": 0.00019764189327950869,
      "loss": 1.4198,
      "step": 2714
    },
    {
      "epoch": 0.09722993177789317,
      "grad_norm": 2.1380035877227783,
      "learning_rate": 0.00019763938857691744,
      "loss": 1.4642,
      "step": 2715
    },
    {
      "epoch": 0.09726574390746145,
      "grad_norm": 1.4172416925430298,
      "learning_rate": 0.00019763688256071418,
      "loss": 1.9416,
      "step": 2716
    },
    {
      "epoch": 0.09730155603702974,
      "grad_norm": 1.5462101697921753,
      "learning_rate": 0.0001976343752309326,
      "loss": 1.573,
      "step": 2717
    },
    {
      "epoch": 0.09733736816659802,
      "grad_norm": 1.9838913679122925,
      "learning_rate": 0.00019763186658760645,
      "loss": 1.7277,
      "step": 2718
    },
    {
      "epoch": 0.09737318029616632,
      "grad_norm": 1.450179100036621,
      "learning_rate": 0.00019762935663076946,
      "loss": 1.7552,
      "step": 2719
    },
    {
      "epoch": 0.0974089924257346,
      "grad_norm": 1.3784503936767578,
      "learning_rate": 0.00019762684536045542,
      "loss": 1.443,
      "step": 2720
    },
    {
      "epoch": 0.09744480455530288,
      "grad_norm": 2.3167805671691895,
      "learning_rate": 0.00019762433277669807,
      "loss": 1.8773,
      "step": 2721
    },
    {
      "epoch": 0.09748061668487117,
      "grad_norm": 2.927114486694336,
      "learning_rate": 0.00019762181887953128,
      "loss": 1.7365,
      "step": 2722
    },
    {
      "epoch": 0.09751642881443945,
      "grad_norm": 2.3608787059783936,
      "learning_rate": 0.00019761930366898883,
      "loss": 1.6947,
      "step": 2723
    },
    {
      "epoch": 0.09755224094400773,
      "grad_norm": 1.6629176139831543,
      "learning_rate": 0.0001976167871451046,
      "loss": 1.6643,
      "step": 2724
    },
    {
      "epoch": 0.09758805307357601,
      "grad_norm": 1.806070327758789,
      "learning_rate": 0.00019761426930791238,
      "loss": 2.0347,
      "step": 2725
    },
    {
      "epoch": 0.09762386520314431,
      "grad_norm": 2.381452798843384,
      "learning_rate": 0.00019761175015744605,
      "loss": 1.5381,
      "step": 2726
    },
    {
      "epoch": 0.0976596773327126,
      "grad_norm": 1.5285437107086182,
      "learning_rate": 0.0001976092296937396,
      "loss": 1.9806,
      "step": 2727
    },
    {
      "epoch": 0.09769548946228088,
      "grad_norm": 1.324546456336975,
      "learning_rate": 0.00019760670791682685,
      "loss": 1.7321,
      "step": 2728
    },
    {
      "epoch": 0.09773130159184916,
      "grad_norm": 1.9074602127075195,
      "learning_rate": 0.00019760418482674173,
      "loss": 1.6216,
      "step": 2729
    },
    {
      "epoch": 0.09776711372141744,
      "grad_norm": 1.6975016593933105,
      "learning_rate": 0.0001976016604235182,
      "loss": 1.7029,
      "step": 2730
    },
    {
      "epoch": 0.09780292585098573,
      "grad_norm": 1.9731216430664062,
      "learning_rate": 0.00019759913470719024,
      "loss": 1.5059,
      "step": 2731
    },
    {
      "epoch": 0.09783873798055401,
      "grad_norm": 1.32267165184021,
      "learning_rate": 0.00019759660767779184,
      "loss": 1.4878,
      "step": 2732
    },
    {
      "epoch": 0.0978745501101223,
      "grad_norm": 1.2084392309188843,
      "learning_rate": 0.00019759407933535693,
      "loss": 1.6624,
      "step": 2733
    },
    {
      "epoch": 0.09791036223969059,
      "grad_norm": 2.081899404525757,
      "learning_rate": 0.0001975915496799196,
      "loss": 1.7721,
      "step": 2734
    },
    {
      "epoch": 0.09794617436925887,
      "grad_norm": 2.2579267024993896,
      "learning_rate": 0.00019758901871151383,
      "loss": 1.3182,
      "step": 2735
    },
    {
      "epoch": 0.09798198649882715,
      "grad_norm": 1.7588104009628296,
      "learning_rate": 0.00019758648643017373,
      "loss": 1.2959,
      "step": 2736
    },
    {
      "epoch": 0.09801779862839544,
      "grad_norm": 1.9816704988479614,
      "learning_rate": 0.0001975839528359333,
      "loss": 1.4635,
      "step": 2737
    },
    {
      "epoch": 0.09805361075796372,
      "grad_norm": 1.833520770072937,
      "learning_rate": 0.00019758141792882667,
      "loss": 1.4191,
      "step": 2738
    },
    {
      "epoch": 0.098089422887532,
      "grad_norm": 1.7605247497558594,
      "learning_rate": 0.00019757888170888793,
      "loss": 1.3866,
      "step": 2739
    },
    {
      "epoch": 0.0981252350171003,
      "grad_norm": 2.2199249267578125,
      "learning_rate": 0.0001975763441761512,
      "loss": 1.7087,
      "step": 2740
    },
    {
      "epoch": 0.09816104714666858,
      "grad_norm": 1.999975562095642,
      "learning_rate": 0.00019757380533065065,
      "loss": 1.8079,
      "step": 2741
    },
    {
      "epoch": 0.09819685927623686,
      "grad_norm": 1.5539270639419556,
      "learning_rate": 0.00019757126517242038,
      "loss": 1.7708,
      "step": 2742
    },
    {
      "epoch": 0.09823267140580515,
      "grad_norm": 1.6064941883087158,
      "learning_rate": 0.0001975687237014946,
      "loss": 1.7572,
      "step": 2743
    },
    {
      "epoch": 0.09826848353537343,
      "grad_norm": 1.6434097290039062,
      "learning_rate": 0.0001975661809179075,
      "loss": 1.6864,
      "step": 2744
    },
    {
      "epoch": 0.09830429566494171,
      "grad_norm": 1.3108112812042236,
      "learning_rate": 0.0001975636368216933,
      "loss": 1.4519,
      "step": 2745
    },
    {
      "epoch": 0.09834010779451,
      "grad_norm": 1.4596123695373535,
      "learning_rate": 0.0001975610914128862,
      "loss": 1.8373,
      "step": 2746
    },
    {
      "epoch": 0.09837591992407829,
      "grad_norm": 1.9536592960357666,
      "learning_rate": 0.00019755854469152045,
      "loss": 1.2695,
      "step": 2747
    },
    {
      "epoch": 0.09841173205364658,
      "grad_norm": 1.4774552583694458,
      "learning_rate": 0.00019755599665763037,
      "loss": 1.4301,
      "step": 2748
    },
    {
      "epoch": 0.09844754418321486,
      "grad_norm": 1.8306077718734741,
      "learning_rate": 0.00019755344731125013,
      "loss": 1.7258,
      "step": 2749
    },
    {
      "epoch": 0.09848335631278314,
      "grad_norm": 1.3312674760818481,
      "learning_rate": 0.00019755089665241413,
      "loss": 1.6966,
      "step": 2750
    },
    {
      "epoch": 0.09851916844235142,
      "grad_norm": 1.9812297821044922,
      "learning_rate": 0.00019754834468115664,
      "loss": 1.7699,
      "step": 2751
    },
    {
      "epoch": 0.0985549805719197,
      "grad_norm": 1.5077332258224487,
      "learning_rate": 0.00019754579139751198,
      "loss": 1.82,
      "step": 2752
    },
    {
      "epoch": 0.09859079270148799,
      "grad_norm": 1.7618402242660522,
      "learning_rate": 0.00019754323680151457,
      "loss": 1.4261,
      "step": 2753
    },
    {
      "epoch": 0.09862660483105627,
      "grad_norm": 2.45039439201355,
      "learning_rate": 0.00019754068089319869,
      "loss": 1.6046,
      "step": 2754
    },
    {
      "epoch": 0.09866241696062457,
      "grad_norm": 1.6822230815887451,
      "learning_rate": 0.00019753812367259878,
      "loss": 1.4612,
      "step": 2755
    },
    {
      "epoch": 0.09869822909019285,
      "grad_norm": 1.5892387628555298,
      "learning_rate": 0.00019753556513974922,
      "loss": 1.5376,
      "step": 2756
    },
    {
      "epoch": 0.09873404121976113,
      "grad_norm": 1.8028465509414673,
      "learning_rate": 0.00019753300529468446,
      "loss": 1.6662,
      "step": 2757
    },
    {
      "epoch": 0.09876985334932942,
      "grad_norm": 1.4199618101119995,
      "learning_rate": 0.00019753044413743892,
      "loss": 1.7844,
      "step": 2758
    },
    {
      "epoch": 0.0988056654788977,
      "grad_norm": 2.244751453399658,
      "learning_rate": 0.00019752788166804702,
      "loss": 1.7328,
      "step": 2759
    },
    {
      "epoch": 0.09884147760846598,
      "grad_norm": 1.673805594444275,
      "learning_rate": 0.0001975253178865433,
      "loss": 1.6421,
      "step": 2760
    },
    {
      "epoch": 0.09887728973803427,
      "grad_norm": 1.4417093992233276,
      "learning_rate": 0.00019752275279296227,
      "loss": 1.6409,
      "step": 2761
    },
    {
      "epoch": 0.09891310186760256,
      "grad_norm": 1.2416033744812012,
      "learning_rate": 0.00019752018638733836,
      "loss": 1.673,
      "step": 2762
    },
    {
      "epoch": 0.09894891399717085,
      "grad_norm": 1.481547236442566,
      "learning_rate": 0.00019751761866970612,
      "loss": 1.7885,
      "step": 2763
    },
    {
      "epoch": 0.09898472612673913,
      "grad_norm": 1.45943284034729,
      "learning_rate": 0.00019751504964010016,
      "loss": 1.9124,
      "step": 2764
    },
    {
      "epoch": 0.09902053825630741,
      "grad_norm": 1.5683441162109375,
      "learning_rate": 0.00019751247929855495,
      "loss": 1.3896,
      "step": 2765
    },
    {
      "epoch": 0.0990563503858757,
      "grad_norm": 1.5302742719650269,
      "learning_rate": 0.0001975099076451051,
      "loss": 1.5202,
      "step": 2766
    },
    {
      "epoch": 0.09909216251544398,
      "grad_norm": 2.1754040718078613,
      "learning_rate": 0.00019750733467978525,
      "loss": 1.6422,
      "step": 2767
    },
    {
      "epoch": 0.09912797464501226,
      "grad_norm": 2.2145087718963623,
      "learning_rate": 0.00019750476040262998,
      "loss": 2.0339,
      "step": 2768
    },
    {
      "epoch": 0.09916378677458056,
      "grad_norm": 1.463472843170166,
      "learning_rate": 0.00019750218481367392,
      "loss": 1.6032,
      "step": 2769
    },
    {
      "epoch": 0.09919959890414884,
      "grad_norm": 1.4679081439971924,
      "learning_rate": 0.00019749960791295174,
      "loss": 1.702,
      "step": 2770
    },
    {
      "epoch": 0.09923541103371712,
      "grad_norm": 1.6745630502700806,
      "learning_rate": 0.0001974970297004981,
      "loss": 1.872,
      "step": 2771
    },
    {
      "epoch": 0.0992712231632854,
      "grad_norm": 1.3749425411224365,
      "learning_rate": 0.0001974944501763477,
      "loss": 1.5595,
      "step": 2772
    },
    {
      "epoch": 0.09930703529285369,
      "grad_norm": 1.8216643333435059,
      "learning_rate": 0.0001974918693405352,
      "loss": 1.5808,
      "step": 2773
    },
    {
      "epoch": 0.09934284742242197,
      "grad_norm": 1.7504353523254395,
      "learning_rate": 0.0001974892871930954,
      "loss": 1.6087,
      "step": 2774
    },
    {
      "epoch": 0.09937865955199025,
      "grad_norm": 1.5744881629943848,
      "learning_rate": 0.00019748670373406294,
      "loss": 1.7731,
      "step": 2775
    },
    {
      "epoch": 0.09941447168155855,
      "grad_norm": 2.4953949451446533,
      "learning_rate": 0.00019748411896347267,
      "loss": 1.6093,
      "step": 2776
    },
    {
      "epoch": 0.09945028381112683,
      "grad_norm": 1.4541890621185303,
      "learning_rate": 0.00019748153288135932,
      "loss": 1.7734,
      "step": 2777
    },
    {
      "epoch": 0.09948609594069512,
      "grad_norm": 1.250166654586792,
      "learning_rate": 0.0001974789454877577,
      "loss": 1.3861,
      "step": 2778
    },
    {
      "epoch": 0.0995219080702634,
      "grad_norm": 1.8420655727386475,
      "learning_rate": 0.0001974763567827026,
      "loss": 1.5565,
      "step": 2779
    },
    {
      "epoch": 0.09955772019983168,
      "grad_norm": 1.5533993244171143,
      "learning_rate": 0.00019747376676622878,
      "loss": 1.5603,
      "step": 2780
    },
    {
      "epoch": 0.09959353232939996,
      "grad_norm": 1.4248117208480835,
      "learning_rate": 0.00019747117543837125,
      "loss": 1.8146,
      "step": 2781
    },
    {
      "epoch": 0.09962934445896825,
      "grad_norm": 1.520105242729187,
      "learning_rate": 0.00019746858279916476,
      "loss": 1.3316,
      "step": 2782
    },
    {
      "epoch": 0.09966515658853654,
      "grad_norm": 1.8025250434875488,
      "learning_rate": 0.0001974659888486442,
      "loss": 1.903,
      "step": 2783
    },
    {
      "epoch": 0.09970096871810483,
      "grad_norm": 1.3632969856262207,
      "learning_rate": 0.0001974633935868445,
      "loss": 1.41,
      "step": 2784
    },
    {
      "epoch": 0.09973678084767311,
      "grad_norm": 1.2988522052764893,
      "learning_rate": 0.00019746079701380055,
      "loss": 1.6359,
      "step": 2785
    },
    {
      "epoch": 0.09977259297724139,
      "grad_norm": 2.9737942218780518,
      "learning_rate": 0.00019745819912954732,
      "loss": 1.5823,
      "step": 2786
    },
    {
      "epoch": 0.09980840510680968,
      "grad_norm": 1.4505422115325928,
      "learning_rate": 0.00019745559993411966,
      "loss": 1.5382,
      "step": 2787
    },
    {
      "epoch": 0.09984421723637796,
      "grad_norm": 1.6142643690109253,
      "learning_rate": 0.00019745299942755266,
      "loss": 1.6859,
      "step": 2788
    },
    {
      "epoch": 0.09988002936594624,
      "grad_norm": 1.6316615343093872,
      "learning_rate": 0.00019745039760988127,
      "loss": 1.6503,
      "step": 2789
    },
    {
      "epoch": 0.09991584149551454,
      "grad_norm": 2.2728500366210938,
      "learning_rate": 0.00019744779448114047,
      "loss": 1.2704,
      "step": 2790
    },
    {
      "epoch": 0.09995165362508282,
      "grad_norm": 1.512316346168518,
      "learning_rate": 0.00019744519004136527,
      "loss": 1.269,
      "step": 2791
    },
    {
      "epoch": 0.0999874657546511,
      "grad_norm": 1.5691752433776855,
      "learning_rate": 0.00019744258429059075,
      "loss": 1.6717,
      "step": 2792
    },
    {
      "epoch": 0.10002327788421939,
      "grad_norm": 1.5791586637496948,
      "learning_rate": 0.00019743997722885198,
      "loss": 1.6228,
      "step": 2793
    },
    {
      "epoch": 0.10005909001378767,
      "grad_norm": 1.5813857316970825,
      "learning_rate": 0.00019743736885618395,
      "loss": 1.3792,
      "step": 2794
    },
    {
      "epoch": 0.10009490214335595,
      "grad_norm": 2.116391897201538,
      "learning_rate": 0.00019743475917262187,
      "loss": 1.5369,
      "step": 2795
    },
    {
      "epoch": 0.10013071427292423,
      "grad_norm": 1.7743563652038574,
      "learning_rate": 0.00019743214817820074,
      "loss": 1.677,
      "step": 2796
    },
    {
      "epoch": 0.10016652640249253,
      "grad_norm": 1.637967586517334,
      "learning_rate": 0.00019742953587295573,
      "loss": 1.6838,
      "step": 2797
    },
    {
      "epoch": 0.10020233853206081,
      "grad_norm": 1.7225396633148193,
      "learning_rate": 0.000197426922256922,
      "loss": 1.7073,
      "step": 2798
    },
    {
      "epoch": 0.1002381506616291,
      "grad_norm": 1.4051920175552368,
      "learning_rate": 0.00019742430733013473,
      "loss": 1.3682,
      "step": 2799
    },
    {
      "epoch": 0.10027396279119738,
      "grad_norm": 1.5960766077041626,
      "learning_rate": 0.00019742169109262904,
      "loss": 1.8122,
      "step": 2800
    },
    {
      "epoch": 0.10030977492076566,
      "grad_norm": 1.5492359399795532,
      "learning_rate": 0.00019741907354444018,
      "loss": 1.656,
      "step": 2801
    },
    {
      "epoch": 0.10034558705033395,
      "grad_norm": 1.6034247875213623,
      "learning_rate": 0.00019741645468560336,
      "loss": 1.8924,
      "step": 2802
    },
    {
      "epoch": 0.10038139917990223,
      "grad_norm": 2.0048415660858154,
      "learning_rate": 0.00019741383451615376,
      "loss": 1.6452,
      "step": 2803
    },
    {
      "epoch": 0.10041721130947051,
      "grad_norm": 1.9224190711975098,
      "learning_rate": 0.0001974112130361267,
      "loss": 1.747,
      "step": 2804
    },
    {
      "epoch": 0.10045302343903881,
      "grad_norm": 1.7946642637252808,
      "learning_rate": 0.0001974085902455574,
      "loss": 1.6644,
      "step": 2805
    },
    {
      "epoch": 0.10048883556860709,
      "grad_norm": 1.3062515258789062,
      "learning_rate": 0.0001974059661444812,
      "loss": 1.6385,
      "step": 2806
    },
    {
      "epoch": 0.10052464769817537,
      "grad_norm": 1.9280710220336914,
      "learning_rate": 0.00019740334073293334,
      "loss": 1.7102,
      "step": 2807
    },
    {
      "epoch": 0.10056045982774366,
      "grad_norm": 1.6444439888000488,
      "learning_rate": 0.0001974007140109492,
      "loss": 1.645,
      "step": 2808
    },
    {
      "epoch": 0.10059627195731194,
      "grad_norm": 2.063418388366699,
      "learning_rate": 0.00019739808597856405,
      "loss": 1.9611,
      "step": 2809
    },
    {
      "epoch": 0.10063208408688022,
      "grad_norm": 1.3895442485809326,
      "learning_rate": 0.0001973954566358133,
      "loss": 1.5975,
      "step": 2810
    },
    {
      "epoch": 0.1006678962164485,
      "grad_norm": 1.1396634578704834,
      "learning_rate": 0.0001973928259827323,
      "loss": 1.5006,
      "step": 2811
    },
    {
      "epoch": 0.1007037083460168,
      "grad_norm": 2.067750930786133,
      "learning_rate": 0.0001973901940193565,
      "loss": 1.4993,
      "step": 2812
    },
    {
      "epoch": 0.10073952047558508,
      "grad_norm": 2.5891499519348145,
      "learning_rate": 0.00019738756074572127,
      "loss": 1.7023,
      "step": 2813
    },
    {
      "epoch": 0.10077533260515337,
      "grad_norm": 1.5110852718353271,
      "learning_rate": 0.00019738492616186198,
      "loss": 1.7311,
      "step": 2814
    },
    {
      "epoch": 0.10081114473472165,
      "grad_norm": 1.5566610097885132,
      "learning_rate": 0.00019738229026781414,
      "loss": 1.4559,
      "step": 2815
    },
    {
      "epoch": 0.10084695686428993,
      "grad_norm": 1.310876488685608,
      "learning_rate": 0.00019737965306361322,
      "loss": 1.5475,
      "step": 2816
    },
    {
      "epoch": 0.10088276899385822,
      "grad_norm": 1.5883020162582397,
      "learning_rate": 0.00019737701454929468,
      "loss": 1.6171,
      "step": 2817
    },
    {
      "epoch": 0.1009185811234265,
      "grad_norm": 1.3175413608551025,
      "learning_rate": 0.000197374374724894,
      "loss": 1.6509,
      "step": 2818
    },
    {
      "epoch": 0.1009543932529948,
      "grad_norm": 1.287886619567871,
      "learning_rate": 0.0001973717335904467,
      "loss": 1.5315,
      "step": 2819
    },
    {
      "epoch": 0.10099020538256308,
      "grad_norm": 2.7122435569763184,
      "learning_rate": 0.00019736909114598833,
      "loss": 1.8696,
      "step": 2820
    },
    {
      "epoch": 0.10102601751213136,
      "grad_norm": 1.4318008422851562,
      "learning_rate": 0.00019736644739155445,
      "loss": 1.4318,
      "step": 2821
    },
    {
      "epoch": 0.10106182964169964,
      "grad_norm": 1.3406039476394653,
      "learning_rate": 0.00019736380232718062,
      "loss": 1.6633,
      "step": 2822
    },
    {
      "epoch": 0.10109764177126793,
      "grad_norm": 1.6335771083831787,
      "learning_rate": 0.00019736115595290238,
      "loss": 1.7371,
      "step": 2823
    },
    {
      "epoch": 0.10113345390083621,
      "grad_norm": 1.8981465101242065,
      "learning_rate": 0.00019735850826875542,
      "loss": 1.596,
      "step": 2824
    },
    {
      "epoch": 0.10116926603040449,
      "grad_norm": 1.0624903440475464,
      "learning_rate": 0.0001973558592747753,
      "loss": 1.4223,
      "step": 2825
    },
    {
      "epoch": 0.10120507815997279,
      "grad_norm": 1.5947264432907104,
      "learning_rate": 0.00019735320897099764,
      "loss": 1.9196,
      "step": 2826
    },
    {
      "epoch": 0.10124089028954107,
      "grad_norm": 1.1546331644058228,
      "learning_rate": 0.00019735055735745817,
      "loss": 1.6058,
      "step": 2827
    },
    {
      "epoch": 0.10127670241910935,
      "grad_norm": 2.101027011871338,
      "learning_rate": 0.0001973479044341925,
      "loss": 1.6675,
      "step": 2828
    },
    {
      "epoch": 0.10131251454867764,
      "grad_norm": 2.380446672439575,
      "learning_rate": 0.00019734525020123639,
      "loss": 1.8698,
      "step": 2829
    },
    {
      "epoch": 0.10134832667824592,
      "grad_norm": 2.061709403991699,
      "learning_rate": 0.00019734259465862546,
      "loss": 1.6063,
      "step": 2830
    },
    {
      "epoch": 0.1013841388078142,
      "grad_norm": 1.974861741065979,
      "learning_rate": 0.0001973399378063955,
      "loss": 1.3884,
      "step": 2831
    },
    {
      "epoch": 0.10141995093738249,
      "grad_norm": 1.3313992023468018,
      "learning_rate": 0.00019733727964458221,
      "loss": 1.4632,
      "step": 2832
    },
    {
      "epoch": 0.10145576306695078,
      "grad_norm": 1.6623539924621582,
      "learning_rate": 0.00019733462017322142,
      "loss": 1.8361,
      "step": 2833
    },
    {
      "epoch": 0.10149157519651907,
      "grad_norm": 1.7105774879455566,
      "learning_rate": 0.00019733195939234882,
      "loss": 1.7204,
      "step": 2834
    },
    {
      "epoch": 0.10152738732608735,
      "grad_norm": 1.2900364398956299,
      "learning_rate": 0.00019732929730200031,
      "loss": 1.6921,
      "step": 2835
    },
    {
      "epoch": 0.10156319945565563,
      "grad_norm": 1.5834048986434937,
      "learning_rate": 0.00019732663390221162,
      "loss": 1.876,
      "step": 2836
    },
    {
      "epoch": 0.10159901158522391,
      "grad_norm": 1.4001473188400269,
      "learning_rate": 0.0001973239691930186,
      "loss": 1.6452,
      "step": 2837
    },
    {
      "epoch": 0.1016348237147922,
      "grad_norm": 2.6770882606506348,
      "learning_rate": 0.00019732130317445714,
      "loss": 1.6828,
      "step": 2838
    },
    {
      "epoch": 0.10167063584436048,
      "grad_norm": 1.2641379833221436,
      "learning_rate": 0.00019731863584656308,
      "loss": 1.5711,
      "step": 2839
    },
    {
      "epoch": 0.10170644797392878,
      "grad_norm": 1.3252092599868774,
      "learning_rate": 0.0001973159672093723,
      "loss": 1.8273,
      "step": 2840
    },
    {
      "epoch": 0.10174226010349706,
      "grad_norm": 1.4117766618728638,
      "learning_rate": 0.00019731329726292073,
      "loss": 1.4361,
      "step": 2841
    },
    {
      "epoch": 0.10177807223306534,
      "grad_norm": 1.7400825023651123,
      "learning_rate": 0.00019731062600724424,
      "loss": 1.5425,
      "step": 2842
    },
    {
      "epoch": 0.10181388436263362,
      "grad_norm": 1.2664161920547485,
      "learning_rate": 0.0001973079534423788,
      "loss": 1.7458,
      "step": 2843
    },
    {
      "epoch": 0.10184969649220191,
      "grad_norm": 1.4911015033721924,
      "learning_rate": 0.00019730527956836035,
      "loss": 1.6534,
      "step": 2844
    },
    {
      "epoch": 0.10188550862177019,
      "grad_norm": 1.494632601737976,
      "learning_rate": 0.00019730260438522492,
      "loss": 1.4687,
      "step": 2845
    },
    {
      "epoch": 0.10192132075133847,
      "grad_norm": 2.710315227508545,
      "learning_rate": 0.00019729992789300845,
      "loss": 1.6687,
      "step": 2846
    },
    {
      "epoch": 0.10195713288090676,
      "grad_norm": 1.292092204093933,
      "learning_rate": 0.00019729725009174693,
      "loss": 1.6175,
      "step": 2847
    },
    {
      "epoch": 0.10199294501047505,
      "grad_norm": 1.6304455995559692,
      "learning_rate": 0.00019729457098147647,
      "loss": 1.4029,
      "step": 2848
    },
    {
      "epoch": 0.10202875714004334,
      "grad_norm": 1.7625586986541748,
      "learning_rate": 0.000197291890562233,
      "loss": 1.9333,
      "step": 2849
    },
    {
      "epoch": 0.10206456926961162,
      "grad_norm": 2.0807228088378906,
      "learning_rate": 0.00019728920883405263,
      "loss": 1.5159,
      "step": 2850
    },
    {
      "epoch": 0.1021003813991799,
      "grad_norm": 1.6510517597198486,
      "learning_rate": 0.00019728652579697152,
      "loss": 1.5753,
      "step": 2851
    },
    {
      "epoch": 0.10213619352874818,
      "grad_norm": 1.9343862533569336,
      "learning_rate": 0.00019728384145102564,
      "loss": 1.615,
      "step": 2852
    },
    {
      "epoch": 0.10217200565831647,
      "grad_norm": 1.403560996055603,
      "learning_rate": 0.00019728115579625117,
      "loss": 1.4167,
      "step": 2853
    },
    {
      "epoch": 0.10220781778788475,
      "grad_norm": 1.5644052028656006,
      "learning_rate": 0.00019727846883268427,
      "loss": 1.7456,
      "step": 2854
    },
    {
      "epoch": 0.10224362991745305,
      "grad_norm": 1.6829103231430054,
      "learning_rate": 0.00019727578056036101,
      "loss": 1.7578,
      "step": 2855
    },
    {
      "epoch": 0.10227944204702133,
      "grad_norm": 1.861466646194458,
      "learning_rate": 0.0001972730909793176,
      "loss": 1.5786,
      "step": 2856
    },
    {
      "epoch": 0.10231525417658961,
      "grad_norm": 1.6972744464874268,
      "learning_rate": 0.00019727040008959024,
      "loss": 1.6002,
      "step": 2857
    },
    {
      "epoch": 0.1023510663061579,
      "grad_norm": 1.5097100734710693,
      "learning_rate": 0.00019726770789121512,
      "loss": 1.5889,
      "step": 2858
    },
    {
      "epoch": 0.10238687843572618,
      "grad_norm": 1.354004979133606,
      "learning_rate": 0.00019726501438422842,
      "loss": 1.6782,
      "step": 2859
    },
    {
      "epoch": 0.10242269056529446,
      "grad_norm": 1.700211763381958,
      "learning_rate": 0.00019726231956866645,
      "loss": 1.3758,
      "step": 2860
    },
    {
      "epoch": 0.10245850269486274,
      "grad_norm": 2.7783384323120117,
      "learning_rate": 0.00019725962344456543,
      "loss": 1.5725,
      "step": 2861
    },
    {
      "epoch": 0.10249431482443104,
      "grad_norm": 1.8978086709976196,
      "learning_rate": 0.00019725692601196162,
      "loss": 1.6209,
      "step": 2862
    },
    {
      "epoch": 0.10253012695399932,
      "grad_norm": 2.073197841644287,
      "learning_rate": 0.00019725422727089132,
      "loss": 1.3348,
      "step": 2863
    },
    {
      "epoch": 0.1025659390835676,
      "grad_norm": 1.9262858629226685,
      "learning_rate": 0.00019725152722139085,
      "loss": 1.8382,
      "step": 2864
    },
    {
      "epoch": 0.10260175121313589,
      "grad_norm": 1.9772917032241821,
      "learning_rate": 0.00019724882586349653,
      "loss": 1.7399,
      "step": 2865
    },
    {
      "epoch": 0.10263756334270417,
      "grad_norm": 1.9056910276412964,
      "learning_rate": 0.00019724612319724468,
      "loss": 1.5189,
      "step": 2866
    },
    {
      "epoch": 0.10267337547227245,
      "grad_norm": 1.3025951385498047,
      "learning_rate": 0.0001972434192226717,
      "loss": 1.6968,
      "step": 2867
    },
    {
      "epoch": 0.10270918760184074,
      "grad_norm": 1.772749900817871,
      "learning_rate": 0.00019724071393981393,
      "loss": 1.3956,
      "step": 2868
    },
    {
      "epoch": 0.10274499973140903,
      "grad_norm": 2.921922445297241,
      "learning_rate": 0.00019723800734870782,
      "loss": 1.7918,
      "step": 2869
    },
    {
      "epoch": 0.10278081186097732,
      "grad_norm": 1.5922387838363647,
      "learning_rate": 0.00019723529944938974,
      "loss": 1.8264,
      "step": 2870
    },
    {
      "epoch": 0.1028166239905456,
      "grad_norm": 1.159803867340088,
      "learning_rate": 0.00019723259024189612,
      "loss": 1.4858,
      "step": 2871
    },
    {
      "epoch": 0.10285243612011388,
      "grad_norm": 1.4431113004684448,
      "learning_rate": 0.0001972298797262634,
      "loss": 1.6408,
      "step": 2872
    },
    {
      "epoch": 0.10288824824968217,
      "grad_norm": 1.711513876914978,
      "learning_rate": 0.0001972271679025281,
      "loss": 1.6958,
      "step": 2873
    },
    {
      "epoch": 0.10292406037925045,
      "grad_norm": 1.448328971862793,
      "learning_rate": 0.00019722445477072666,
      "loss": 1.4791,
      "step": 2874
    },
    {
      "epoch": 0.10295987250881873,
      "grad_norm": 1.213527798652649,
      "learning_rate": 0.00019722174033089557,
      "loss": 1.5364,
      "step": 2875
    },
    {
      "epoch": 0.10299568463838703,
      "grad_norm": 1.4906288385391235,
      "learning_rate": 0.0001972190245830714,
      "loss": 1.4892,
      "step": 2876
    },
    {
      "epoch": 0.10303149676795531,
      "grad_norm": 1.639464020729065,
      "learning_rate": 0.00019721630752729064,
      "loss": 1.7839,
      "step": 2877
    },
    {
      "epoch": 0.1030673088975236,
      "grad_norm": 2.0337164402008057,
      "learning_rate": 0.00019721358916358986,
      "loss": 1.5573,
      "step": 2878
    },
    {
      "epoch": 0.10310312102709188,
      "grad_norm": 1.2845540046691895,
      "learning_rate": 0.00019721086949200565,
      "loss": 1.6272,
      "step": 2879
    },
    {
      "epoch": 0.10313893315666016,
      "grad_norm": 1.317577838897705,
      "learning_rate": 0.00019720814851257457,
      "loss": 1.5472,
      "step": 2880
    },
    {
      "epoch": 0.10317474528622844,
      "grad_norm": 1.3626635074615479,
      "learning_rate": 0.00019720542622533323,
      "loss": 1.6389,
      "step": 2881
    },
    {
      "epoch": 0.10321055741579672,
      "grad_norm": 1.4564090967178345,
      "learning_rate": 0.0001972027026303183,
      "loss": 1.5697,
      "step": 2882
    },
    {
      "epoch": 0.10324636954536502,
      "grad_norm": 1.459882140159607,
      "learning_rate": 0.00019719997772756637,
      "loss": 1.4884,
      "step": 2883
    },
    {
      "epoch": 0.1032821816749333,
      "grad_norm": 1.6414357423782349,
      "learning_rate": 0.00019719725151711413,
      "loss": 2.0682,
      "step": 2884
    },
    {
      "epoch": 0.10331799380450159,
      "grad_norm": 1.225942850112915,
      "learning_rate": 0.00019719452399899823,
      "loss": 1.3942,
      "step": 2885
    },
    {
      "epoch": 0.10335380593406987,
      "grad_norm": 1.67031991481781,
      "learning_rate": 0.00019719179517325538,
      "loss": 1.4331,
      "step": 2886
    },
    {
      "epoch": 0.10338961806363815,
      "grad_norm": 2.3105432987213135,
      "learning_rate": 0.00019718906503992233,
      "loss": 1.7211,
      "step": 2887
    },
    {
      "epoch": 0.10342543019320644,
      "grad_norm": 1.6651228666305542,
      "learning_rate": 0.00019718633359903573,
      "loss": 1.3568,
      "step": 2888
    },
    {
      "epoch": 0.10346124232277472,
      "grad_norm": 2.5842862129211426,
      "learning_rate": 0.00019718360085063238,
      "loss": 1.6142,
      "step": 2889
    },
    {
      "epoch": 0.10349705445234302,
      "grad_norm": 1.3132528066635132,
      "learning_rate": 0.00019718086679474905,
      "loss": 1.5248,
      "step": 2890
    },
    {
      "epoch": 0.1035328665819113,
      "grad_norm": 1.5135085582733154,
      "learning_rate": 0.0001971781314314225,
      "loss": 1.4578,
      "step": 2891
    },
    {
      "epoch": 0.10356867871147958,
      "grad_norm": 1.8713313341140747,
      "learning_rate": 0.00019717539476068959,
      "loss": 1.5403,
      "step": 2892
    },
    {
      "epoch": 0.10360449084104786,
      "grad_norm": 1.3445090055465698,
      "learning_rate": 0.00019717265678258702,
      "loss": 1.6932,
      "step": 2893
    },
    {
      "epoch": 0.10364030297061615,
      "grad_norm": 2.2110817432403564,
      "learning_rate": 0.00019716991749715174,
      "loss": 1.6663,
      "step": 2894
    },
    {
      "epoch": 0.10367611510018443,
      "grad_norm": 1.4373912811279297,
      "learning_rate": 0.00019716717690442055,
      "loss": 1.6872,
      "step": 2895
    },
    {
      "epoch": 0.10371192722975271,
      "grad_norm": 1.3239036798477173,
      "learning_rate": 0.00019716443500443034,
      "loss": 1.4225,
      "step": 2896
    },
    {
      "epoch": 0.103747739359321,
      "grad_norm": 1.6182363033294678,
      "learning_rate": 0.00019716169179721799,
      "loss": 1.7373,
      "step": 2897
    },
    {
      "epoch": 0.10378355148888929,
      "grad_norm": 1.3850167989730835,
      "learning_rate": 0.00019715894728282037,
      "loss": 1.5345,
      "step": 2898
    },
    {
      "epoch": 0.10381936361845757,
      "grad_norm": 1.4673134088516235,
      "learning_rate": 0.00019715620146127448,
      "loss": 1.5932,
      "step": 2899
    },
    {
      "epoch": 0.10385517574802586,
      "grad_norm": 1.7944717407226562,
      "learning_rate": 0.0001971534543326172,
      "loss": 1.6443,
      "step": 2900
    },
    {
      "epoch": 0.10389098787759414,
      "grad_norm": 1.4910426139831543,
      "learning_rate": 0.0001971507058968855,
      "loss": 1.7324,
      "step": 2901
    },
    {
      "epoch": 0.10392680000716242,
      "grad_norm": 2.181478261947632,
      "learning_rate": 0.00019714795615411644,
      "loss": 1.6161,
      "step": 2902
    },
    {
      "epoch": 0.1039626121367307,
      "grad_norm": 1.2145978212356567,
      "learning_rate": 0.00019714520510434686,
      "loss": 1.5633,
      "step": 2903
    },
    {
      "epoch": 0.10399842426629899,
      "grad_norm": 1.4202697277069092,
      "learning_rate": 0.0001971424527476139,
      "loss": 1.3827,
      "step": 2904
    },
    {
      "epoch": 0.10403423639586729,
      "grad_norm": 2.2743844985961914,
      "learning_rate": 0.0001971396990839545,
      "loss": 1.6353,
      "step": 2905
    },
    {
      "epoch": 0.10407004852543557,
      "grad_norm": 1.5119394063949585,
      "learning_rate": 0.0001971369441134058,
      "loss": 1.4954,
      "step": 2906
    },
    {
      "epoch": 0.10410586065500385,
      "grad_norm": 2.039295196533203,
      "learning_rate": 0.00019713418783600477,
      "loss": 1.4147,
      "step": 2907
    },
    {
      "epoch": 0.10414167278457213,
      "grad_norm": 1.9588508605957031,
      "learning_rate": 0.00019713143025178856,
      "loss": 1.7196,
      "step": 2908
    },
    {
      "epoch": 0.10417748491414042,
      "grad_norm": 1.776458978652954,
      "learning_rate": 0.00019712867136079427,
      "loss": 1.5977,
      "step": 2909
    },
    {
      "epoch": 0.1042132970437087,
      "grad_norm": 1.6060348749160767,
      "learning_rate": 0.00019712591116305896,
      "loss": 1.8558,
      "step": 2910
    },
    {
      "epoch": 0.10424910917327698,
      "grad_norm": 1.828564167022705,
      "learning_rate": 0.0001971231496586198,
      "loss": 1.6377,
      "step": 2911
    },
    {
      "epoch": 0.10428492130284528,
      "grad_norm": 2.2467479705810547,
      "learning_rate": 0.00019712038684751394,
      "loss": 1.6065,
      "step": 2912
    },
    {
      "epoch": 0.10432073343241356,
      "grad_norm": 1.4483753442764282,
      "learning_rate": 0.0001971176227297786,
      "loss": 1.4403,
      "step": 2913
    },
    {
      "epoch": 0.10435654556198184,
      "grad_norm": 1.480074405670166,
      "learning_rate": 0.00019711485730545086,
      "loss": 1.4702,
      "step": 2914
    },
    {
      "epoch": 0.10439235769155013,
      "grad_norm": 2.7046217918395996,
      "learning_rate": 0.000197112090574568,
      "loss": 1.5385,
      "step": 2915
    },
    {
      "epoch": 0.10442816982111841,
      "grad_norm": 1.4356821775436401,
      "learning_rate": 0.00019710932253716722,
      "loss": 1.2486,
      "step": 2916
    },
    {
      "epoch": 0.1044639819506867,
      "grad_norm": 1.5573153495788574,
      "learning_rate": 0.00019710655319328578,
      "loss": 1.8457,
      "step": 2917
    },
    {
      "epoch": 0.10449979408025498,
      "grad_norm": 1.470349669456482,
      "learning_rate": 0.00019710378254296092,
      "loss": 2.0201,
      "step": 2918
    },
    {
      "epoch": 0.10453560620982327,
      "grad_norm": 1.6238548755645752,
      "learning_rate": 0.00019710101058622993,
      "loss": 1.279,
      "step": 2919
    },
    {
      "epoch": 0.10457141833939156,
      "grad_norm": 1.5568904876708984,
      "learning_rate": 0.00019709823732313008,
      "loss": 1.6239,
      "step": 2920
    },
    {
      "epoch": 0.10460723046895984,
      "grad_norm": 2.827681064605713,
      "learning_rate": 0.0001970954627536987,
      "loss": 1.8256,
      "step": 2921
    },
    {
      "epoch": 0.10464304259852812,
      "grad_norm": 1.9860001802444458,
      "learning_rate": 0.00019709268687797312,
      "loss": 1.6688,
      "step": 2922
    },
    {
      "epoch": 0.1046788547280964,
      "grad_norm": 1.3693649768829346,
      "learning_rate": 0.00019708990969599067,
      "loss": 1.4567,
      "step": 2923
    },
    {
      "epoch": 0.10471466685766469,
      "grad_norm": 1.5036529302597046,
      "learning_rate": 0.00019708713120778873,
      "loss": 1.7974,
      "step": 2924
    },
    {
      "epoch": 0.10475047898723297,
      "grad_norm": 1.446396827697754,
      "learning_rate": 0.00019708435141340465,
      "loss": 1.5247,
      "step": 2925
    },
    {
      "epoch": 0.10478629111680127,
      "grad_norm": 1.1922216415405273,
      "learning_rate": 0.00019708157031287588,
      "loss": 1.8338,
      "step": 2926
    },
    {
      "epoch": 0.10482210324636955,
      "grad_norm": 1.8486875295639038,
      "learning_rate": 0.0001970787879062398,
      "loss": 1.359,
      "step": 2927
    },
    {
      "epoch": 0.10485791537593783,
      "grad_norm": 1.4665194749832153,
      "learning_rate": 0.00019707600419353383,
      "loss": 1.6474,
      "step": 2928
    },
    {
      "epoch": 0.10489372750550612,
      "grad_norm": 1.7647480964660645,
      "learning_rate": 0.00019707321917479547,
      "loss": 1.6358,
      "step": 2929
    },
    {
      "epoch": 0.1049295396350744,
      "grad_norm": 1.694216012954712,
      "learning_rate": 0.00019707043285006214,
      "loss": 1.618,
      "step": 2930
    },
    {
      "epoch": 0.10496535176464268,
      "grad_norm": 1.8111251592636108,
      "learning_rate": 0.00019706764521937138,
      "loss": 1.6625,
      "step": 2931
    },
    {
      "epoch": 0.10500116389421096,
      "grad_norm": 1.5972100496292114,
      "learning_rate": 0.00019706485628276062,
      "loss": 1.5874,
      "step": 2932
    },
    {
      "epoch": 0.10503697602377926,
      "grad_norm": 1.9588450193405151,
      "learning_rate": 0.00019706206604026746,
      "loss": 1.5872,
      "step": 2933
    },
    {
      "epoch": 0.10507278815334754,
      "grad_norm": 1.1568310260772705,
      "learning_rate": 0.00019705927449192937,
      "loss": 1.4907,
      "step": 2934
    },
    {
      "epoch": 0.10510860028291583,
      "grad_norm": 1.5908012390136719,
      "learning_rate": 0.00019705648163778397,
      "loss": 1.5814,
      "step": 2935
    },
    {
      "epoch": 0.10514441241248411,
      "grad_norm": 1.7401422262191772,
      "learning_rate": 0.00019705368747786878,
      "loss": 1.9683,
      "step": 2936
    },
    {
      "epoch": 0.10518022454205239,
      "grad_norm": 1.7295998334884644,
      "learning_rate": 0.00019705089201222143,
      "loss": 1.3394,
      "step": 2937
    },
    {
      "epoch": 0.10521603667162067,
      "grad_norm": 1.5581072568893433,
      "learning_rate": 0.00019704809524087952,
      "loss": 1.7824,
      "step": 2938
    },
    {
      "epoch": 0.10525184880118896,
      "grad_norm": 1.2611238956451416,
      "learning_rate": 0.00019704529716388068,
      "loss": 1.5758,
      "step": 2939
    },
    {
      "epoch": 0.10528766093075725,
      "grad_norm": 1.469044804573059,
      "learning_rate": 0.00019704249778126253,
      "loss": 1.5392,
      "step": 2940
    },
    {
      "epoch": 0.10532347306032554,
      "grad_norm": 1.7068703174591064,
      "learning_rate": 0.00019703969709306273,
      "loss": 1.6083,
      "step": 2941
    },
    {
      "epoch": 0.10535928518989382,
      "grad_norm": 1.3602216243743896,
      "learning_rate": 0.000197036895099319,
      "loss": 1.9103,
      "step": 2942
    },
    {
      "epoch": 0.1053950973194621,
      "grad_norm": 1.29122793674469,
      "learning_rate": 0.000197034091800069,
      "loss": 1.9111,
      "step": 2943
    },
    {
      "epoch": 0.10543090944903039,
      "grad_norm": 1.4069241285324097,
      "learning_rate": 0.00019703128719535047,
      "loss": 1.6298,
      "step": 2944
    },
    {
      "epoch": 0.10546672157859867,
      "grad_norm": 1.6360414028167725,
      "learning_rate": 0.00019702848128520112,
      "loss": 1.5577,
      "step": 2945
    },
    {
      "epoch": 0.10550253370816695,
      "grad_norm": 2.210772752761841,
      "learning_rate": 0.00019702567406965874,
      "loss": 1.8814,
      "step": 2946
    },
    {
      "epoch": 0.10553834583773523,
      "grad_norm": 1.9793405532836914,
      "learning_rate": 0.00019702286554876107,
      "loss": 1.8386,
      "step": 2947
    },
    {
      "epoch": 0.10557415796730353,
      "grad_norm": 1.657800316810608,
      "learning_rate": 0.00019702005572254586,
      "loss": 1.603,
      "step": 2948
    },
    {
      "epoch": 0.10560997009687181,
      "grad_norm": 2.0030272006988525,
      "learning_rate": 0.00019701724459105096,
      "loss": 1.615,
      "step": 2949
    },
    {
      "epoch": 0.1056457822264401,
      "grad_norm": 1.4756288528442383,
      "learning_rate": 0.0001970144321543142,
      "loss": 1.713,
      "step": 2950
    },
    {
      "epoch": 0.10568159435600838,
      "grad_norm": 1.9561413526535034,
      "learning_rate": 0.00019701161841237337,
      "loss": 1.5548,
      "step": 2951
    },
    {
      "epoch": 0.10571740648557666,
      "grad_norm": 2.031585216522217,
      "learning_rate": 0.00019700880336526635,
      "loss": 1.9009,
      "step": 2952
    },
    {
      "epoch": 0.10575321861514494,
      "grad_norm": 1.3856836557388306,
      "learning_rate": 0.000197005987013031,
      "loss": 1.9266,
      "step": 2953
    },
    {
      "epoch": 0.10578903074471323,
      "grad_norm": 1.4809298515319824,
      "learning_rate": 0.00019700316935570525,
      "loss": 1.7208,
      "step": 2954
    },
    {
      "epoch": 0.10582484287428152,
      "grad_norm": 1.3451876640319824,
      "learning_rate": 0.00019700035039332697,
      "loss": 1.6572,
      "step": 2955
    },
    {
      "epoch": 0.10586065500384981,
      "grad_norm": 1.6890588998794556,
      "learning_rate": 0.00019699753012593412,
      "loss": 1.6136,
      "step": 2956
    },
    {
      "epoch": 0.10589646713341809,
      "grad_norm": 1.5177700519561768,
      "learning_rate": 0.0001969947085535646,
      "loss": 1.2336,
      "step": 2957
    },
    {
      "epoch": 0.10593227926298637,
      "grad_norm": 2.2147626876831055,
      "learning_rate": 0.00019699188567625639,
      "loss": 1.7097,
      "step": 2958
    },
    {
      "epoch": 0.10596809139255466,
      "grad_norm": 1.2537357807159424,
      "learning_rate": 0.00019698906149404746,
      "loss": 1.6603,
      "step": 2959
    },
    {
      "epoch": 0.10600390352212294,
      "grad_norm": 1.5088328123092651,
      "learning_rate": 0.00019698623600697583,
      "loss": 1.6384,
      "step": 2960
    },
    {
      "epoch": 0.10603971565169122,
      "grad_norm": 1.7185791730880737,
      "learning_rate": 0.0001969834092150795,
      "loss": 1.9186,
      "step": 2961
    },
    {
      "epoch": 0.10607552778125952,
      "grad_norm": 1.3850542306900024,
      "learning_rate": 0.0001969805811183965,
      "loss": 1.7726,
      "step": 2962
    },
    {
      "epoch": 0.1061113399108278,
      "grad_norm": 1.59743332862854,
      "learning_rate": 0.00019697775171696486,
      "loss": 1.6747,
      "step": 2963
    },
    {
      "epoch": 0.10614715204039608,
      "grad_norm": 1.8517948389053345,
      "learning_rate": 0.00019697492101082266,
      "loss": 1.6965,
      "step": 2964
    },
    {
      "epoch": 0.10618296416996437,
      "grad_norm": 1.5334497690200806,
      "learning_rate": 0.000196972089000008,
      "loss": 1.6025,
      "step": 2965
    },
    {
      "epoch": 0.10621877629953265,
      "grad_norm": 1.2686636447906494,
      "learning_rate": 0.00019696925568455894,
      "loss": 1.5366,
      "step": 2966
    },
    {
      "epoch": 0.10625458842910093,
      "grad_norm": 1.518660306930542,
      "learning_rate": 0.00019696642106451368,
      "loss": 1.8495,
      "step": 2967
    },
    {
      "epoch": 0.10629040055866922,
      "grad_norm": 1.365371584892273,
      "learning_rate": 0.00019696358513991027,
      "loss": 1.8094,
      "step": 2968
    },
    {
      "epoch": 0.10632621268823751,
      "grad_norm": 1.5992499589920044,
      "learning_rate": 0.0001969607479107869,
      "loss": 1.6623,
      "step": 2969
    },
    {
      "epoch": 0.1063620248178058,
      "grad_norm": 1.7085697650909424,
      "learning_rate": 0.00019695790937718176,
      "loss": 1.8765,
      "step": 2970
    },
    {
      "epoch": 0.10639783694737408,
      "grad_norm": 2.420846700668335,
      "learning_rate": 0.00019695506953913298,
      "loss": 2.0682,
      "step": 2971
    },
    {
      "epoch": 0.10643364907694236,
      "grad_norm": 2.160022735595703,
      "learning_rate": 0.0001969522283966788,
      "loss": 1.9651,
      "step": 2972
    },
    {
      "epoch": 0.10646946120651064,
      "grad_norm": 1.9768903255462646,
      "learning_rate": 0.00019694938594985747,
      "loss": 1.5112,
      "step": 2973
    },
    {
      "epoch": 0.10650527333607893,
      "grad_norm": 1.8344073295593262,
      "learning_rate": 0.00019694654219870722,
      "loss": 1.7646,
      "step": 2974
    },
    {
      "epoch": 0.10654108546564721,
      "grad_norm": 1.4202852249145508,
      "learning_rate": 0.00019694369714326625,
      "loss": 1.6141,
      "step": 2975
    },
    {
      "epoch": 0.1065768975952155,
      "grad_norm": 1.8785535097122192,
      "learning_rate": 0.00019694085078357293,
      "loss": 1.7087,
      "step": 2976
    },
    {
      "epoch": 0.10661270972478379,
      "grad_norm": 1.394516944885254,
      "learning_rate": 0.00019693800311966549,
      "loss": 1.5792,
      "step": 2977
    },
    {
      "epoch": 0.10664852185435207,
      "grad_norm": 1.6537799835205078,
      "learning_rate": 0.00019693515415158223,
      "loss": 1.6773,
      "step": 2978
    },
    {
      "epoch": 0.10668433398392035,
      "grad_norm": 1.922308325767517,
      "learning_rate": 0.00019693230387936154,
      "loss": 1.7615,
      "step": 2979
    },
    {
      "epoch": 0.10672014611348864,
      "grad_norm": 1.586806058883667,
      "learning_rate": 0.00019692945230304174,
      "loss": 1.7742,
      "step": 2980
    },
    {
      "epoch": 0.10675595824305692,
      "grad_norm": 2.630951166152954,
      "learning_rate": 0.00019692659942266118,
      "loss": 2.197,
      "step": 2981
    },
    {
      "epoch": 0.1067917703726252,
      "grad_norm": 1.6085656881332397,
      "learning_rate": 0.00019692374523825823,
      "loss": 1.4823,
      "step": 2982
    },
    {
      "epoch": 0.1068275825021935,
      "grad_norm": 1.7773700952529907,
      "learning_rate": 0.00019692088974987133,
      "loss": 1.8165,
      "step": 2983
    },
    {
      "epoch": 0.10686339463176178,
      "grad_norm": 1.6625611782073975,
      "learning_rate": 0.0001969180329575389,
      "loss": 1.6183,
      "step": 2984
    },
    {
      "epoch": 0.10689920676133007,
      "grad_norm": 3.302849292755127,
      "learning_rate": 0.0001969151748612993,
      "loss": 1.8682,
      "step": 2985
    },
    {
      "epoch": 0.10693501889089835,
      "grad_norm": 1.2780555486679077,
      "learning_rate": 0.00019691231546119107,
      "loss": 1.3038,
      "step": 2986
    },
    {
      "epoch": 0.10697083102046663,
      "grad_norm": 1.9470473527908325,
      "learning_rate": 0.00019690945475725266,
      "loss": 1.4459,
      "step": 2987
    },
    {
      "epoch": 0.10700664315003491,
      "grad_norm": 1.5665614604949951,
      "learning_rate": 0.0001969065927495225,
      "loss": 1.6486,
      "step": 2988
    },
    {
      "epoch": 0.1070424552796032,
      "grad_norm": 1.4052174091339111,
      "learning_rate": 0.00019690372943803914,
      "loss": 1.7583,
      "step": 2989
    },
    {
      "epoch": 0.1070782674091715,
      "grad_norm": 2.11560320854187,
      "learning_rate": 0.00019690086482284112,
      "loss": 1.4927,
      "step": 2990
    },
    {
      "epoch": 0.10711407953873978,
      "grad_norm": 1.2599753141403198,
      "learning_rate": 0.00019689799890396694,
      "loss": 1.5102,
      "step": 2991
    },
    {
      "epoch": 0.10714989166830806,
      "grad_norm": 2.6709375381469727,
      "learning_rate": 0.0001968951316814552,
      "loss": 1.7188,
      "step": 2992
    },
    {
      "epoch": 0.10718570379787634,
      "grad_norm": 2.060213565826416,
      "learning_rate": 0.0001968922631553444,
      "loss": 1.5915,
      "step": 2993
    },
    {
      "epoch": 0.10722151592744462,
      "grad_norm": 1.579801082611084,
      "learning_rate": 0.00019688939332567325,
      "loss": 1.5262,
      "step": 2994
    },
    {
      "epoch": 0.10725732805701291,
      "grad_norm": 1.5131771564483643,
      "learning_rate": 0.00019688652219248021,
      "loss": 1.2072,
      "step": 2995
    },
    {
      "epoch": 0.10729314018658119,
      "grad_norm": 1.1854443550109863,
      "learning_rate": 0.00019688364975580406,
      "loss": 1.556,
      "step": 2996
    },
    {
      "epoch": 0.10732895231614947,
      "grad_norm": 1.2820968627929688,
      "learning_rate": 0.00019688077601568332,
      "loss": 1.4393,
      "step": 2997
    },
    {
      "epoch": 0.10736476444571777,
      "grad_norm": 1.5455936193466187,
      "learning_rate": 0.00019687790097215675,
      "loss": 1.6942,
      "step": 2998
    },
    {
      "epoch": 0.10740057657528605,
      "grad_norm": 1.7165623903274536,
      "learning_rate": 0.00019687502462526296,
      "loss": 1.8651,
      "step": 2999
    },
    {
      "epoch": 0.10743638870485434,
      "grad_norm": 1.641993522644043,
      "learning_rate": 0.00019687214697504068,
      "loss": 1.7097,
      "step": 3000
    },
    {
      "epoch": 0.10747220083442262,
      "grad_norm": 1.5092154741287231,
      "learning_rate": 0.00019686926802152862,
      "loss": 1.6165,
      "step": 3001
    },
    {
      "epoch": 0.1075080129639909,
      "grad_norm": 1.4830560684204102,
      "learning_rate": 0.0001968663877647655,
      "loss": 1.6553,
      "step": 3002
    },
    {
      "epoch": 0.10754382509355918,
      "grad_norm": 1.6284724473953247,
      "learning_rate": 0.0001968635062047901,
      "loss": 1.7557,
      "step": 3003
    },
    {
      "epoch": 0.10757963722312747,
      "grad_norm": 1.5482624769210815,
      "learning_rate": 0.00019686062334164114,
      "loss": 1.531,
      "step": 3004
    },
    {
      "epoch": 0.10761544935269576,
      "grad_norm": 1.963331937789917,
      "learning_rate": 0.00019685773917535747,
      "loss": 1.5902,
      "step": 3005
    },
    {
      "epoch": 0.10765126148226405,
      "grad_norm": 1.459810495376587,
      "learning_rate": 0.00019685485370597781,
      "loss": 1.6869,
      "step": 3006
    },
    {
      "epoch": 0.10768707361183233,
      "grad_norm": 1.9397635459899902,
      "learning_rate": 0.00019685196693354108,
      "loss": 1.9816,
      "step": 3007
    },
    {
      "epoch": 0.10772288574140061,
      "grad_norm": 1.316998839378357,
      "learning_rate": 0.00019684907885808602,
      "loss": 1.4956,
      "step": 3008
    },
    {
      "epoch": 0.1077586978709689,
      "grad_norm": 1.8278248310089111,
      "learning_rate": 0.00019684618947965157,
      "loss": 1.7439,
      "step": 3009
    },
    {
      "epoch": 0.10779451000053718,
      "grad_norm": 1.5549050569534302,
      "learning_rate": 0.00019684329879827655,
      "loss": 1.7759,
      "step": 3010
    },
    {
      "epoch": 0.10783032213010546,
      "grad_norm": 1.860183596611023,
      "learning_rate": 0.00019684040681399988,
      "loss": 1.7717,
      "step": 3011
    },
    {
      "epoch": 0.10786613425967376,
      "grad_norm": 1.4960808753967285,
      "learning_rate": 0.0001968375135268604,
      "loss": 1.8338,
      "step": 3012
    },
    {
      "epoch": 0.10790194638924204,
      "grad_norm": 2.194607734680176,
      "learning_rate": 0.00019683461893689713,
      "loss": 1.3071,
      "step": 3013
    },
    {
      "epoch": 0.10793775851881032,
      "grad_norm": 1.5040713548660278,
      "learning_rate": 0.00019683172304414895,
      "loss": 1.9321,
      "step": 3014
    },
    {
      "epoch": 0.1079735706483786,
      "grad_norm": 1.7146315574645996,
      "learning_rate": 0.00019682882584865486,
      "loss": 1.6084,
      "step": 3015
    },
    {
      "epoch": 0.10800938277794689,
      "grad_norm": 1.6277421712875366,
      "learning_rate": 0.0001968259273504538,
      "loss": 1.296,
      "step": 3016
    },
    {
      "epoch": 0.10804519490751517,
      "grad_norm": 3.6533689498901367,
      "learning_rate": 0.0001968230275495848,
      "loss": 1.5586,
      "step": 3017
    },
    {
      "epoch": 0.10808100703708345,
      "grad_norm": 1.6158928871154785,
      "learning_rate": 0.00019682012644608684,
      "loss": 1.9503,
      "step": 3018
    },
    {
      "epoch": 0.10811681916665175,
      "grad_norm": 1.6434288024902344,
      "learning_rate": 0.000196817224039999,
      "loss": 1.5552,
      "step": 3019
    },
    {
      "epoch": 0.10815263129622003,
      "grad_norm": 1.395654320716858,
      "learning_rate": 0.00019681432033136025,
      "loss": 1.6248,
      "step": 3020
    },
    {
      "epoch": 0.10818844342578832,
      "grad_norm": 1.5744826793670654,
      "learning_rate": 0.00019681141532020973,
      "loss": 1.4838,
      "step": 3021
    },
    {
      "epoch": 0.1082242555553566,
      "grad_norm": 1.7468857765197754,
      "learning_rate": 0.00019680850900658648,
      "loss": 1.8954,
      "step": 3022
    },
    {
      "epoch": 0.10826006768492488,
      "grad_norm": 2.2361481189727783,
      "learning_rate": 0.00019680560139052962,
      "loss": 1.5789,
      "step": 3023
    },
    {
      "epoch": 0.10829587981449317,
      "grad_norm": 1.6525112390518188,
      "learning_rate": 0.00019680269247207826,
      "loss": 1.705,
      "step": 3024
    },
    {
      "epoch": 0.10833169194406145,
      "grad_norm": 1.5771716833114624,
      "learning_rate": 0.00019679978225127154,
      "loss": 1.5043,
      "step": 3025
    },
    {
      "epoch": 0.10836750407362974,
      "grad_norm": 1.917914867401123,
      "learning_rate": 0.00019679687072814863,
      "loss": 1.8099,
      "step": 3026
    },
    {
      "epoch": 0.10840331620319803,
      "grad_norm": 1.7994537353515625,
      "learning_rate": 0.00019679395790274867,
      "loss": 1.8708,
      "step": 3027
    },
    {
      "epoch": 0.10843912833276631,
      "grad_norm": 1.5328046083450317,
      "learning_rate": 0.00019679104377511085,
      "loss": 1.6289,
      "step": 3028
    },
    {
      "epoch": 0.1084749404623346,
      "grad_norm": 1.5930429697036743,
      "learning_rate": 0.0001967881283452744,
      "loss": 1.6851,
      "step": 3029
    },
    {
      "epoch": 0.10851075259190288,
      "grad_norm": 1.4349143505096436,
      "learning_rate": 0.00019678521161327854,
      "loss": 1.7717,
      "step": 3030
    },
    {
      "epoch": 0.10854656472147116,
      "grad_norm": 1.4315009117126465,
      "learning_rate": 0.0001967822935791625,
      "loss": 1.8928,
      "step": 3031
    },
    {
      "epoch": 0.10858237685103944,
      "grad_norm": 1.4336453676223755,
      "learning_rate": 0.0001967793742429655,
      "loss": 1.9097,
      "step": 3032
    },
    {
      "epoch": 0.10861818898060774,
      "grad_norm": 1.6382025480270386,
      "learning_rate": 0.00019677645360472693,
      "loss": 1.6168,
      "step": 3033
    },
    {
      "epoch": 0.10865400111017602,
      "grad_norm": 2.577434778213501,
      "learning_rate": 0.00019677353166448595,
      "loss": 1.7761,
      "step": 3034
    },
    {
      "epoch": 0.1086898132397443,
      "grad_norm": 1.9029229879379272,
      "learning_rate": 0.00019677060842228193,
      "loss": 1.594,
      "step": 3035
    },
    {
      "epoch": 0.10872562536931259,
      "grad_norm": 1.9777616262435913,
      "learning_rate": 0.00019676768387815423,
      "loss": 1.8283,
      "step": 3036
    },
    {
      "epoch": 0.10876143749888087,
      "grad_norm": 1.5141868591308594,
      "learning_rate": 0.00019676475803214217,
      "loss": 1.3587,
      "step": 3037
    },
    {
      "epoch": 0.10879724962844915,
      "grad_norm": 1.609864354133606,
      "learning_rate": 0.0001967618308842851,
      "loss": 1.7622,
      "step": 3038
    },
    {
      "epoch": 0.10883306175801744,
      "grad_norm": 2.1204848289489746,
      "learning_rate": 0.00019675890243462237,
      "loss": 1.6448,
      "step": 3039
    },
    {
      "epoch": 0.10886887388758573,
      "grad_norm": 1.144871473312378,
      "learning_rate": 0.00019675597268319344,
      "loss": 1.6545,
      "step": 3040
    },
    {
      "epoch": 0.10890468601715401,
      "grad_norm": 1.7205898761749268,
      "learning_rate": 0.00019675304163003772,
      "loss": 1.7451,
      "step": 3041
    },
    {
      "epoch": 0.1089404981467223,
      "grad_norm": 1.9791665077209473,
      "learning_rate": 0.00019675010927519462,
      "loss": 1.871,
      "step": 3042
    },
    {
      "epoch": 0.10897631027629058,
      "grad_norm": 1.382914423942566,
      "learning_rate": 0.0001967471756187036,
      "loss": 1.6498,
      "step": 3043
    },
    {
      "epoch": 0.10901212240585886,
      "grad_norm": 2.2286922931671143,
      "learning_rate": 0.0001967442406606041,
      "loss": 1.5953,
      "step": 3044
    },
    {
      "epoch": 0.10904793453542715,
      "grad_norm": 1.5717313289642334,
      "learning_rate": 0.00019674130440093567,
      "loss": 1.7754,
      "step": 3045
    },
    {
      "epoch": 0.10908374666499543,
      "grad_norm": 1.2834662199020386,
      "learning_rate": 0.00019673836683973777,
      "loss": 1.546,
      "step": 3046
    },
    {
      "epoch": 0.10911955879456371,
      "grad_norm": 2.3408381938934326,
      "learning_rate": 0.00019673542797704992,
      "loss": 1.6439,
      "step": 3047
    },
    {
      "epoch": 0.10915537092413201,
      "grad_norm": 2.09751558303833,
      "learning_rate": 0.00019673248781291167,
      "loss": 1.7954,
      "step": 3048
    },
    {
      "epoch": 0.10919118305370029,
      "grad_norm": 1.5235075950622559,
      "learning_rate": 0.00019672954634736257,
      "loss": 2.0121,
      "step": 3049
    },
    {
      "epoch": 0.10922699518326857,
      "grad_norm": 2.3699159622192383,
      "learning_rate": 0.00019672660358044218,
      "loss": 1.5584,
      "step": 3050
    },
    {
      "epoch": 0.10926280731283686,
      "grad_norm": 1.5464084148406982,
      "learning_rate": 0.00019672365951219013,
      "loss": 1.2083,
      "step": 3051
    },
    {
      "epoch": 0.10929861944240514,
      "grad_norm": 1.4740662574768066,
      "learning_rate": 0.00019672071414264598,
      "loss": 1.6632,
      "step": 3052
    },
    {
      "epoch": 0.10933443157197342,
      "grad_norm": 1.5872660875320435,
      "learning_rate": 0.0001967177674718494,
      "loss": 1.7066,
      "step": 3053
    },
    {
      "epoch": 0.1093702437015417,
      "grad_norm": 1.4718883037567139,
      "learning_rate": 0.00019671481949984002,
      "loss": 1.7492,
      "step": 3054
    },
    {
      "epoch": 0.10940605583111,
      "grad_norm": 1.7044392824172974,
      "learning_rate": 0.0001967118702266575,
      "loss": 1.3809,
      "step": 3055
    },
    {
      "epoch": 0.10944186796067829,
      "grad_norm": 1.6310226917266846,
      "learning_rate": 0.0001967089196523415,
      "loss": 1.8413,
      "step": 3056
    },
    {
      "epoch": 0.10947768009024657,
      "grad_norm": 1.6687250137329102,
      "learning_rate": 0.00019670596777693176,
      "loss": 1.4604,
      "step": 3057
    },
    {
      "epoch": 0.10951349221981485,
      "grad_norm": 1.9419200420379639,
      "learning_rate": 0.00019670301460046795,
      "loss": 1.7134,
      "step": 3058
    },
    {
      "epoch": 0.10954930434938313,
      "grad_norm": 1.624732255935669,
      "learning_rate": 0.0001967000601229898,
      "loss": 1.6658,
      "step": 3059
    },
    {
      "epoch": 0.10958511647895142,
      "grad_norm": 2.9077885150909424,
      "learning_rate": 0.00019669710434453707,
      "loss": 1.8957,
      "step": 3060
    },
    {
      "epoch": 0.1096209286085197,
      "grad_norm": 1.5916715860366821,
      "learning_rate": 0.00019669414726514956,
      "loss": 1.8633,
      "step": 3061
    },
    {
      "epoch": 0.109656740738088,
      "grad_norm": 1.584800124168396,
      "learning_rate": 0.000196691188884867,
      "loss": 1.3014,
      "step": 3062
    },
    {
      "epoch": 0.10969255286765628,
      "grad_norm": 1.7780625820159912,
      "learning_rate": 0.00019668822920372922,
      "loss": 1.7299,
      "step": 3063
    },
    {
      "epoch": 0.10972836499722456,
      "grad_norm": 1.7964451313018799,
      "learning_rate": 0.00019668526822177605,
      "loss": 1.5913,
      "step": 3064
    },
    {
      "epoch": 0.10976417712679284,
      "grad_norm": 1.9102052450180054,
      "learning_rate": 0.00019668230593904734,
      "loss": 1.3066,
      "step": 3065
    },
    {
      "epoch": 0.10979998925636113,
      "grad_norm": 1.4517691135406494,
      "learning_rate": 0.00019667934235558285,
      "loss": 1.4668,
      "step": 3066
    },
    {
      "epoch": 0.10983580138592941,
      "grad_norm": 1.3721327781677246,
      "learning_rate": 0.00019667637747142257,
      "loss": 1.8355,
      "step": 3067
    },
    {
      "epoch": 0.10987161351549769,
      "grad_norm": 2.1370456218719482,
      "learning_rate": 0.0001966734112866063,
      "loss": 1.386,
      "step": 3068
    },
    {
      "epoch": 0.10990742564506599,
      "grad_norm": 1.892778992652893,
      "learning_rate": 0.00019667044380117398,
      "loss": 1.4856,
      "step": 3069
    },
    {
      "epoch": 0.10994323777463427,
      "grad_norm": 1.4679338932037354,
      "learning_rate": 0.00019666747501516553,
      "loss": 1.7188,
      "step": 3070
    },
    {
      "epoch": 0.10997904990420256,
      "grad_norm": 1.373197317123413,
      "learning_rate": 0.00019666450492862093,
      "loss": 1.3631,
      "step": 3071
    },
    {
      "epoch": 0.11001486203377084,
      "grad_norm": 1.1402403116226196,
      "learning_rate": 0.0001966615335415801,
      "loss": 1.5227,
      "step": 3072
    },
    {
      "epoch": 0.11005067416333912,
      "grad_norm": 1.6488516330718994,
      "learning_rate": 0.000196658560854083,
      "loss": 1.7497,
      "step": 3073
    },
    {
      "epoch": 0.1100864862929074,
      "grad_norm": 1.3037346601486206,
      "learning_rate": 0.00019665558686616965,
      "loss": 1.8412,
      "step": 3074
    },
    {
      "epoch": 0.11012229842247569,
      "grad_norm": 1.519590139389038,
      "learning_rate": 0.00019665261157788004,
      "loss": 1.7108,
      "step": 3075
    },
    {
      "epoch": 0.11015811055204398,
      "grad_norm": 1.3518164157867432,
      "learning_rate": 0.00019664963498925423,
      "loss": 1.6536,
      "step": 3076
    },
    {
      "epoch": 0.11019392268161227,
      "grad_norm": 3.0712289810180664,
      "learning_rate": 0.00019664665710033226,
      "loss": 1.6875,
      "step": 3077
    },
    {
      "epoch": 0.11022973481118055,
      "grad_norm": 1.832874059677124,
      "learning_rate": 0.0001966436779111542,
      "loss": 1.8305,
      "step": 3078
    },
    {
      "epoch": 0.11026554694074883,
      "grad_norm": 1.384520411491394,
      "learning_rate": 0.00019664069742176006,
      "loss": 1.6199,
      "step": 3079
    },
    {
      "epoch": 0.11030135907031711,
      "grad_norm": 1.634929895401001,
      "learning_rate": 0.00019663771563219006,
      "loss": 1.6368,
      "step": 3080
    },
    {
      "epoch": 0.1103371711998854,
      "grad_norm": 1.5089815855026245,
      "learning_rate": 0.00019663473254248417,
      "loss": 1.8269,
      "step": 3081
    },
    {
      "epoch": 0.11037298332945368,
      "grad_norm": 1.8302271366119385,
      "learning_rate": 0.00019663174815268266,
      "loss": 1.7455,
      "step": 3082
    },
    {
      "epoch": 0.11040879545902198,
      "grad_norm": 1.9949239492416382,
      "learning_rate": 0.0001966287624628256,
      "loss": 1.8868,
      "step": 3083
    },
    {
      "epoch": 0.11044460758859026,
      "grad_norm": 1.749061107635498,
      "learning_rate": 0.0001966257754729532,
      "loss": 1.9831,
      "step": 3084
    },
    {
      "epoch": 0.11048041971815854,
      "grad_norm": 1.5313340425491333,
      "learning_rate": 0.00019662278718310562,
      "loss": 1.6813,
      "step": 3085
    },
    {
      "epoch": 0.11051623184772683,
      "grad_norm": 2.523949146270752,
      "learning_rate": 0.0001966197975933231,
      "loss": 1.4592,
      "step": 3086
    },
    {
      "epoch": 0.11055204397729511,
      "grad_norm": 1.696785807609558,
      "learning_rate": 0.0001966168067036458,
      "loss": 1.4651,
      "step": 3087
    },
    {
      "epoch": 0.11058785610686339,
      "grad_norm": 2.151517391204834,
      "learning_rate": 0.000196613814514114,
      "loss": 1.7816,
      "step": 3088
    },
    {
      "epoch": 0.11062366823643167,
      "grad_norm": 1.1579859256744385,
      "learning_rate": 0.00019661082102476795,
      "loss": 1.4892,
      "step": 3089
    },
    {
      "epoch": 0.11065948036599997,
      "grad_norm": 1.6856324672698975,
      "learning_rate": 0.00019660782623564792,
      "loss": 1.6944,
      "step": 3090
    },
    {
      "epoch": 0.11069529249556825,
      "grad_norm": 1.9886382818222046,
      "learning_rate": 0.0001966048301467942,
      "loss": 1.5386,
      "step": 3091
    },
    {
      "epoch": 0.11073110462513654,
      "grad_norm": 1.2798269987106323,
      "learning_rate": 0.0001966018327582471,
      "loss": 1.5971,
      "step": 3092
    },
    {
      "epoch": 0.11076691675470482,
      "grad_norm": 1.2525361776351929,
      "learning_rate": 0.00019659883407004697,
      "loss": 1.6204,
      "step": 3093
    },
    {
      "epoch": 0.1108027288842731,
      "grad_norm": 1.73500657081604,
      "learning_rate": 0.00019659583408223412,
      "loss": 1.7557,
      "step": 3094
    },
    {
      "epoch": 0.11083854101384139,
      "grad_norm": 1.7497897148132324,
      "learning_rate": 0.00019659283279484891,
      "loss": 1.7348,
      "step": 3095
    },
    {
      "epoch": 0.11087435314340967,
      "grad_norm": 1.3274827003479004,
      "learning_rate": 0.00019658983020793175,
      "loss": 1.4419,
      "step": 3096
    },
    {
      "epoch": 0.11091016527297795,
      "grad_norm": 2.2989590167999268,
      "learning_rate": 0.000196586826321523,
      "loss": 1.9745,
      "step": 3097
    },
    {
      "epoch": 0.11094597740254625,
      "grad_norm": 2.2532341480255127,
      "learning_rate": 0.0001965838211356631,
      "loss": 1.8956,
      "step": 3098
    },
    {
      "epoch": 0.11098178953211453,
      "grad_norm": 2.0518624782562256,
      "learning_rate": 0.00019658081465039246,
      "loss": 1.681,
      "step": 3099
    },
    {
      "epoch": 0.11101760166168281,
      "grad_norm": 1.6704341173171997,
      "learning_rate": 0.00019657780686575157,
      "loss": 1.7184,
      "step": 3100
    },
    {
      "epoch": 0.1110534137912511,
      "grad_norm": 1.0749001502990723,
      "learning_rate": 0.00019657479778178083,
      "loss": 1.6137,
      "step": 3101
    },
    {
      "epoch": 0.11108922592081938,
      "grad_norm": 1.7141863107681274,
      "learning_rate": 0.00019657178739852075,
      "loss": 1.484,
      "step": 3102
    },
    {
      "epoch": 0.11112503805038766,
      "grad_norm": 1.4108766317367554,
      "learning_rate": 0.00019656877571601187,
      "loss": 1.8166,
      "step": 3103
    },
    {
      "epoch": 0.11116085017995594,
      "grad_norm": 1.4703015089035034,
      "learning_rate": 0.00019656576273429467,
      "loss": 1.796,
      "step": 3104
    },
    {
      "epoch": 0.11119666230952424,
      "grad_norm": 1.2321605682373047,
      "learning_rate": 0.0001965627484534097,
      "loss": 1.6408,
      "step": 3105
    },
    {
      "epoch": 0.11123247443909252,
      "grad_norm": 1.6735345125198364,
      "learning_rate": 0.0001965597328733975,
      "loss": 1.6995,
      "step": 3106
    },
    {
      "epoch": 0.1112682865686608,
      "grad_norm": 1.1838451623916626,
      "learning_rate": 0.00019655671599429865,
      "loss": 1.663,
      "step": 3107
    },
    {
      "epoch": 0.11130409869822909,
      "grad_norm": 1.8111246824264526,
      "learning_rate": 0.0001965536978161537,
      "loss": 1.7964,
      "step": 3108
    },
    {
      "epoch": 0.11133991082779737,
      "grad_norm": 2.363241195678711,
      "learning_rate": 0.00019655067833900333,
      "loss": 1.5979,
      "step": 3109
    },
    {
      "epoch": 0.11137572295736566,
      "grad_norm": 1.3892968893051147,
      "learning_rate": 0.00019654765756288813,
      "loss": 1.9266,
      "step": 3110
    },
    {
      "epoch": 0.11141153508693394,
      "grad_norm": 2.5642900466918945,
      "learning_rate": 0.00019654463548784873,
      "loss": 1.6784,
      "step": 3111
    },
    {
      "epoch": 0.11144734721650223,
      "grad_norm": 1.9628474712371826,
      "learning_rate": 0.00019654161211392576,
      "loss": 1.5582,
      "step": 3112
    },
    {
      "epoch": 0.11148315934607052,
      "grad_norm": 1.6404321193695068,
      "learning_rate": 0.00019653858744115996,
      "loss": 1.4472,
      "step": 3113
    },
    {
      "epoch": 0.1115189714756388,
      "grad_norm": 1.7177790403366089,
      "learning_rate": 0.00019653556146959197,
      "loss": 1.4041,
      "step": 3114
    },
    {
      "epoch": 0.11155478360520708,
      "grad_norm": 1.340905785560608,
      "learning_rate": 0.00019653253419926254,
      "loss": 1.7599,
      "step": 3115
    },
    {
      "epoch": 0.11159059573477537,
      "grad_norm": 1.997868537902832,
      "learning_rate": 0.00019652950563021237,
      "loss": 1.6725,
      "step": 3116
    },
    {
      "epoch": 0.11162640786434365,
      "grad_norm": 1.5993092060089111,
      "learning_rate": 0.00019652647576248223,
      "loss": 1.6371,
      "step": 3117
    },
    {
      "epoch": 0.11166221999391193,
      "grad_norm": 1.9976245164871216,
      "learning_rate": 0.00019652344459611287,
      "loss": 1.8941,
      "step": 3118
    },
    {
      "epoch": 0.11169803212348023,
      "grad_norm": 1.9197741746902466,
      "learning_rate": 0.00019652041213114504,
      "loss": 1.9326,
      "step": 3119
    },
    {
      "epoch": 0.11173384425304851,
      "grad_norm": 1.5483609437942505,
      "learning_rate": 0.0001965173783676196,
      "loss": 1.3853,
      "step": 3120
    },
    {
      "epoch": 0.1117696563826168,
      "grad_norm": 1.5547926425933838,
      "learning_rate": 0.0001965143433055773,
      "loss": 1.769,
      "step": 3121
    },
    {
      "epoch": 0.11180546851218508,
      "grad_norm": 1.8922780752182007,
      "learning_rate": 0.00019651130694505904,
      "loss": 1.7482,
      "step": 3122
    },
    {
      "epoch": 0.11184128064175336,
      "grad_norm": 1.4488351345062256,
      "learning_rate": 0.00019650826928610564,
      "loss": 1.8088,
      "step": 3123
    },
    {
      "epoch": 0.11187709277132164,
      "grad_norm": 1.4341516494750977,
      "learning_rate": 0.00019650523032875791,
      "loss": 1.361,
      "step": 3124
    },
    {
      "epoch": 0.11191290490088993,
      "grad_norm": 1.3100446462631226,
      "learning_rate": 0.00019650219007305686,
      "loss": 1.53,
      "step": 3125
    },
    {
      "epoch": 0.11194871703045822,
      "grad_norm": 1.9011170864105225,
      "learning_rate": 0.00019649914851904327,
      "loss": 1.5894,
      "step": 3126
    },
    {
      "epoch": 0.1119845291600265,
      "grad_norm": 1.6128778457641602,
      "learning_rate": 0.0001964961056667581,
      "loss": 1.7066,
      "step": 3127
    },
    {
      "epoch": 0.11202034128959479,
      "grad_norm": 1.8269716501235962,
      "learning_rate": 0.00019649306151624235,
      "loss": 1.4852,
      "step": 3128
    },
    {
      "epoch": 0.11205615341916307,
      "grad_norm": 1.937235713005066,
      "learning_rate": 0.0001964900160675369,
      "loss": 1.5758,
      "step": 3129
    },
    {
      "epoch": 0.11209196554873135,
      "grad_norm": 1.8874437808990479,
      "learning_rate": 0.00019648696932068272,
      "loss": 1.98,
      "step": 3130
    },
    {
      "epoch": 0.11212777767829964,
      "grad_norm": 1.6971936225891113,
      "learning_rate": 0.0001964839212757209,
      "loss": 1.5301,
      "step": 3131
    },
    {
      "epoch": 0.11216358980786792,
      "grad_norm": 1.5790627002716064,
      "learning_rate": 0.00019648087193269232,
      "loss": 1.6124,
      "step": 3132
    },
    {
      "epoch": 0.11219940193743622,
      "grad_norm": 1.3589112758636475,
      "learning_rate": 0.00019647782129163805,
      "loss": 1.6747,
      "step": 3133
    },
    {
      "epoch": 0.1122352140670045,
      "grad_norm": 1.376923680305481,
      "learning_rate": 0.00019647476935259916,
      "loss": 1.5895,
      "step": 3134
    },
    {
      "epoch": 0.11227102619657278,
      "grad_norm": 2.1595699787139893,
      "learning_rate": 0.0001964717161156167,
      "loss": 1.5737,
      "step": 3135
    },
    {
      "epoch": 0.11230683832614106,
      "grad_norm": 1.3219338655471802,
      "learning_rate": 0.00019646866158073173,
      "loss": 1.5445,
      "step": 3136
    },
    {
      "epoch": 0.11234265045570935,
      "grad_norm": 1.44645357131958,
      "learning_rate": 0.00019646560574798535,
      "loss": 1.2135,
      "step": 3137
    },
    {
      "epoch": 0.11237846258527763,
      "grad_norm": 1.1569671630859375,
      "learning_rate": 0.0001964625486174187,
      "loss": 1.3581,
      "step": 3138
    },
    {
      "epoch": 0.11241427471484591,
      "grad_norm": 1.5105891227722168,
      "learning_rate": 0.00019645949018907283,
      "loss": 1.5849,
      "step": 3139
    },
    {
      "epoch": 0.11245008684441421,
      "grad_norm": 1.767600655555725,
      "learning_rate": 0.000196456430462989,
      "loss": 1.3581,
      "step": 3140
    },
    {
      "epoch": 0.11248589897398249,
      "grad_norm": 1.3685157299041748,
      "learning_rate": 0.00019645336943920828,
      "loss": 1.7175,
      "step": 3141
    },
    {
      "epoch": 0.11252171110355078,
      "grad_norm": 1.4004133939743042,
      "learning_rate": 0.00019645030711777192,
      "loss": 1.2914,
      "step": 3142
    },
    {
      "epoch": 0.11255752323311906,
      "grad_norm": 1.5391340255737305,
      "learning_rate": 0.0001964472434987211,
      "loss": 1.7839,
      "step": 3143
    },
    {
      "epoch": 0.11259333536268734,
      "grad_norm": 1.2137620449066162,
      "learning_rate": 0.00019644417858209702,
      "loss": 1.5005,
      "step": 3144
    },
    {
      "epoch": 0.11262914749225562,
      "grad_norm": 1.874804139137268,
      "learning_rate": 0.00019644111236794088,
      "loss": 1.8163,
      "step": 3145
    },
    {
      "epoch": 0.1126649596218239,
      "grad_norm": 1.817853569984436,
      "learning_rate": 0.000196438044856294,
      "loss": 1.6768,
      "step": 3146
    },
    {
      "epoch": 0.11270077175139219,
      "grad_norm": 1.3849821090698242,
      "learning_rate": 0.0001964349760471976,
      "loss": 1.5519,
      "step": 3147
    },
    {
      "epoch": 0.11273658388096049,
      "grad_norm": 1.526769995689392,
      "learning_rate": 0.00019643190594069302,
      "loss": 1.939,
      "step": 3148
    },
    {
      "epoch": 0.11277239601052877,
      "grad_norm": 2.315960645675659,
      "learning_rate": 0.00019642883453682152,
      "loss": 1.8201,
      "step": 3149
    },
    {
      "epoch": 0.11280820814009705,
      "grad_norm": 1.7398210763931274,
      "learning_rate": 0.00019642576183562444,
      "loss": 1.4776,
      "step": 3150
    },
    {
      "epoch": 0.11284402026966533,
      "grad_norm": 1.768480658531189,
      "learning_rate": 0.00019642268783714312,
      "loss": 1.9995,
      "step": 3151
    },
    {
      "epoch": 0.11287983239923362,
      "grad_norm": 1.9272271394729614,
      "learning_rate": 0.0001964196125414189,
      "loss": 1.6427,
      "step": 3152
    },
    {
      "epoch": 0.1129156445288019,
      "grad_norm": 1.451150894165039,
      "learning_rate": 0.0001964165359484932,
      "loss": 1.6079,
      "step": 3153
    },
    {
      "epoch": 0.11295145665837018,
      "grad_norm": 1.640161395072937,
      "learning_rate": 0.00019641345805840733,
      "loss": 1.5618,
      "step": 3154
    },
    {
      "epoch": 0.11298726878793848,
      "grad_norm": 1.5714844465255737,
      "learning_rate": 0.00019641037887120277,
      "loss": 1.7151,
      "step": 3155
    },
    {
      "epoch": 0.11302308091750676,
      "grad_norm": 1.5246632099151611,
      "learning_rate": 0.00019640729838692092,
      "loss": 1.425,
      "step": 3156
    },
    {
      "epoch": 0.11305889304707505,
      "grad_norm": 1.6431224346160889,
      "learning_rate": 0.00019640421660560323,
      "loss": 1.6421,
      "step": 3157
    },
    {
      "epoch": 0.11309470517664333,
      "grad_norm": 1.2764418125152588,
      "learning_rate": 0.00019640113352729116,
      "loss": 1.5922,
      "step": 3158
    },
    {
      "epoch": 0.11313051730621161,
      "grad_norm": 1.4131271839141846,
      "learning_rate": 0.00019639804915202617,
      "loss": 1.5558,
      "step": 3159
    },
    {
      "epoch": 0.1131663294357799,
      "grad_norm": 1.229662537574768,
      "learning_rate": 0.0001963949634798498,
      "loss": 1.5588,
      "step": 3160
    },
    {
      "epoch": 0.11320214156534818,
      "grad_norm": 1.4876227378845215,
      "learning_rate": 0.0001963918765108035,
      "loss": 1.4488,
      "step": 3161
    },
    {
      "epoch": 0.11323795369491647,
      "grad_norm": 1.2596564292907715,
      "learning_rate": 0.00019638878824492886,
      "loss": 1.4179,
      "step": 3162
    },
    {
      "epoch": 0.11327376582448476,
      "grad_norm": 1.4421182870864868,
      "learning_rate": 0.0001963856986822674,
      "loss": 1.3701,
      "step": 3163
    },
    {
      "epoch": 0.11330957795405304,
      "grad_norm": 1.4923369884490967,
      "learning_rate": 0.00019638260782286072,
      "loss": 1.3928,
      "step": 3164
    },
    {
      "epoch": 0.11334539008362132,
      "grad_norm": 1.701507806777954,
      "learning_rate": 0.00019637951566675035,
      "loss": 1.8446,
      "step": 3165
    },
    {
      "epoch": 0.1133812022131896,
      "grad_norm": 1.7031058073043823,
      "learning_rate": 0.00019637642221397792,
      "loss": 1.6509,
      "step": 3166
    },
    {
      "epoch": 0.11341701434275789,
      "grad_norm": 1.7792160511016846,
      "learning_rate": 0.00019637332746458506,
      "loss": 1.7611,
      "step": 3167
    },
    {
      "epoch": 0.11345282647232617,
      "grad_norm": 2.105922222137451,
      "learning_rate": 0.00019637023141861338,
      "loss": 1.716,
      "step": 3168
    },
    {
      "epoch": 0.11348863860189447,
      "grad_norm": 2.09403920173645,
      "learning_rate": 0.00019636713407610455,
      "loss": 1.5287,
      "step": 3169
    },
    {
      "epoch": 0.11352445073146275,
      "grad_norm": 1.582002878189087,
      "learning_rate": 0.0001963640354371002,
      "loss": 2.0743,
      "step": 3170
    },
    {
      "epoch": 0.11356026286103103,
      "grad_norm": 1.5988606214523315,
      "learning_rate": 0.00019636093550164208,
      "loss": 1.6238,
      "step": 3171
    },
    {
      "epoch": 0.11359607499059932,
      "grad_norm": 1.5815061330795288,
      "learning_rate": 0.00019635783426977187,
      "loss": 2.0104,
      "step": 3172
    },
    {
      "epoch": 0.1136318871201676,
      "grad_norm": 1.422127366065979,
      "learning_rate": 0.00019635473174153128,
      "loss": 1.7212,
      "step": 3173
    },
    {
      "epoch": 0.11366769924973588,
      "grad_norm": 1.4015790224075317,
      "learning_rate": 0.00019635162791696212,
      "loss": 1.672,
      "step": 3174
    },
    {
      "epoch": 0.11370351137930416,
      "grad_norm": 1.4458675384521484,
      "learning_rate": 0.00019634852279610602,
      "loss": 1.6764,
      "step": 3175
    },
    {
      "epoch": 0.11373932350887246,
      "grad_norm": 2.8560314178466797,
      "learning_rate": 0.00019634541637900487,
      "loss": 1.6277,
      "step": 3176
    },
    {
      "epoch": 0.11377513563844074,
      "grad_norm": 1.758179783821106,
      "learning_rate": 0.0001963423086657004,
      "loss": 1.4349,
      "step": 3177
    },
    {
      "epoch": 0.11381094776800903,
      "grad_norm": 1.4987274408340454,
      "learning_rate": 0.00019633919965623444,
      "loss": 1.7514,
      "step": 3178
    },
    {
      "epoch": 0.11384675989757731,
      "grad_norm": 1.8860074281692505,
      "learning_rate": 0.0001963360893506488,
      "loss": 1.4095,
      "step": 3179
    },
    {
      "epoch": 0.11388257202714559,
      "grad_norm": 1.5295324325561523,
      "learning_rate": 0.0001963329777489854,
      "loss": 1.479,
      "step": 3180
    },
    {
      "epoch": 0.11391838415671388,
      "grad_norm": 1.5854120254516602,
      "learning_rate": 0.00019632986485128602,
      "loss": 1.6353,
      "step": 3181
    },
    {
      "epoch": 0.11395419628628216,
      "grad_norm": 1.5801342725753784,
      "learning_rate": 0.00019632675065759254,
      "loss": 1.4634,
      "step": 3182
    },
    {
      "epoch": 0.11399000841585046,
      "grad_norm": 1.8172426223754883,
      "learning_rate": 0.0001963236351679469,
      "loss": 1.7716,
      "step": 3183
    },
    {
      "epoch": 0.11402582054541874,
      "grad_norm": 1.4854059219360352,
      "learning_rate": 0.00019632051838239099,
      "loss": 1.6442,
      "step": 3184
    },
    {
      "epoch": 0.11406163267498702,
      "grad_norm": 1.682544231414795,
      "learning_rate": 0.00019631740030096677,
      "loss": 1.9143,
      "step": 3185
    },
    {
      "epoch": 0.1140974448045553,
      "grad_norm": 1.281859040260315,
      "learning_rate": 0.00019631428092371612,
      "loss": 1.5915,
      "step": 3186
    },
    {
      "epoch": 0.11413325693412359,
      "grad_norm": 1.3597787618637085,
      "learning_rate": 0.00019631116025068112,
      "loss": 1.6747,
      "step": 3187
    },
    {
      "epoch": 0.11416906906369187,
      "grad_norm": 1.622419834136963,
      "learning_rate": 0.00019630803828190368,
      "loss": 1.848,
      "step": 3188
    },
    {
      "epoch": 0.11420488119326015,
      "grad_norm": 1.744403600692749,
      "learning_rate": 0.00019630491501742577,
      "loss": 1.5405,
      "step": 3189
    },
    {
      "epoch": 0.11424069332282845,
      "grad_norm": 1.3016765117645264,
      "learning_rate": 0.00019630179045728946,
      "loss": 1.5965,
      "step": 3190
    },
    {
      "epoch": 0.11427650545239673,
      "grad_norm": 1.3310681581497192,
      "learning_rate": 0.00019629866460153683,
      "loss": 1.5548,
      "step": 3191
    },
    {
      "epoch": 0.11431231758196501,
      "grad_norm": 2.2537686824798584,
      "learning_rate": 0.00019629553745020983,
      "loss": 1.2023,
      "step": 3192
    },
    {
      "epoch": 0.1143481297115333,
      "grad_norm": 1.3856066465377808,
      "learning_rate": 0.00019629240900335062,
      "loss": 1.5476,
      "step": 3193
    },
    {
      "epoch": 0.11438394184110158,
      "grad_norm": 1.8118864297866821,
      "learning_rate": 0.00019628927926100125,
      "loss": 1.4654,
      "step": 3194
    },
    {
      "epoch": 0.11441975397066986,
      "grad_norm": 1.674796223640442,
      "learning_rate": 0.0001962861482232038,
      "loss": 1.549,
      "step": 3195
    },
    {
      "epoch": 0.11445556610023815,
      "grad_norm": 1.317910075187683,
      "learning_rate": 0.00019628301589000047,
      "loss": 1.3573,
      "step": 3196
    },
    {
      "epoch": 0.11449137822980643,
      "grad_norm": 2.1432268619537354,
      "learning_rate": 0.00019627988226143334,
      "loss": 1.477,
      "step": 3197
    },
    {
      "epoch": 0.11452719035937473,
      "grad_norm": 1.5118340253829956,
      "learning_rate": 0.00019627674733754458,
      "loss": 1.5706,
      "step": 3198
    },
    {
      "epoch": 0.11456300248894301,
      "grad_norm": 1.699642539024353,
      "learning_rate": 0.00019627361111837637,
      "loss": 1.4643,
      "step": 3199
    },
    {
      "epoch": 0.11459881461851129,
      "grad_norm": 1.292831301689148,
      "learning_rate": 0.00019627047360397092,
      "loss": 1.8203,
      "step": 3200
    },
    {
      "epoch": 0.11463462674807957,
      "grad_norm": 1.6774110794067383,
      "learning_rate": 0.00019626733479437042,
      "loss": 1.9212,
      "step": 3201
    },
    {
      "epoch": 0.11467043887764786,
      "grad_norm": 1.5225844383239746,
      "learning_rate": 0.0001962641946896171,
      "loss": 1.6182,
      "step": 3202
    },
    {
      "epoch": 0.11470625100721614,
      "grad_norm": 2.3992981910705566,
      "learning_rate": 0.0001962610532897532,
      "loss": 1.9434,
      "step": 3203
    },
    {
      "epoch": 0.11474206313678442,
      "grad_norm": 1.9705089330673218,
      "learning_rate": 0.00019625791059482106,
      "loss": 1.4532,
      "step": 3204
    },
    {
      "epoch": 0.11477787526635272,
      "grad_norm": 2.258025884628296,
      "learning_rate": 0.00019625476660486285,
      "loss": 1.3525,
      "step": 3205
    },
    {
      "epoch": 0.114813687395921,
      "grad_norm": 1.7987005710601807,
      "learning_rate": 0.0001962516213199209,
      "loss": 1.7462,
      "step": 3206
    },
    {
      "epoch": 0.11484949952548928,
      "grad_norm": 1.720633864402771,
      "learning_rate": 0.00019624847474003756,
      "loss": 1.7263,
      "step": 3207
    },
    {
      "epoch": 0.11488531165505757,
      "grad_norm": 1.8208423852920532,
      "learning_rate": 0.00019624532686525513,
      "loss": 1.5331,
      "step": 3208
    },
    {
      "epoch": 0.11492112378462585,
      "grad_norm": 1.2464954853057861,
      "learning_rate": 0.000196242177695616,
      "loss": 1.4239,
      "step": 3209
    },
    {
      "epoch": 0.11495693591419413,
      "grad_norm": 1.676194429397583,
      "learning_rate": 0.0001962390272311625,
      "loss": 1.8397,
      "step": 3210
    },
    {
      "epoch": 0.11499274804376242,
      "grad_norm": 1.4178789854049683,
      "learning_rate": 0.00019623587547193703,
      "loss": 1.525,
      "step": 3211
    },
    {
      "epoch": 0.11502856017333071,
      "grad_norm": 1.856972098350525,
      "learning_rate": 0.00019623272241798198,
      "loss": 1.9044,
      "step": 3212
    },
    {
      "epoch": 0.115064372302899,
      "grad_norm": 2.108940362930298,
      "learning_rate": 0.0001962295680693398,
      "loss": 1.7134,
      "step": 3213
    },
    {
      "epoch": 0.11510018443246728,
      "grad_norm": 1.6334648132324219,
      "learning_rate": 0.0001962264124260529,
      "loss": 1.5537,
      "step": 3214
    },
    {
      "epoch": 0.11513599656203556,
      "grad_norm": 1.7307848930358887,
      "learning_rate": 0.00019622325548816373,
      "loss": 1.7397,
      "step": 3215
    },
    {
      "epoch": 0.11517180869160384,
      "grad_norm": 1.7003225088119507,
      "learning_rate": 0.0001962200972557148,
      "loss": 1.7081,
      "step": 3216
    },
    {
      "epoch": 0.11520762082117213,
      "grad_norm": 1.5408225059509277,
      "learning_rate": 0.00019621693772874855,
      "loss": 1.3395,
      "step": 3217
    },
    {
      "epoch": 0.11524343295074041,
      "grad_norm": 1.5445457696914673,
      "learning_rate": 0.00019621377690730754,
      "loss": 1.5551,
      "step": 3218
    },
    {
      "epoch": 0.1152792450803087,
      "grad_norm": 2.64323353767395,
      "learning_rate": 0.00019621061479143425,
      "loss": 1.662,
      "step": 3219
    },
    {
      "epoch": 0.11531505720987699,
      "grad_norm": 1.4125784635543823,
      "learning_rate": 0.00019620745138117124,
      "loss": 1.498,
      "step": 3220
    },
    {
      "epoch": 0.11535086933944527,
      "grad_norm": 1.3482048511505127,
      "learning_rate": 0.00019620428667656108,
      "loss": 1.6449,
      "step": 3221
    },
    {
      "epoch": 0.11538668146901356,
      "grad_norm": 1.5159589052200317,
      "learning_rate": 0.00019620112067764636,
      "loss": 1.7357,
      "step": 3222
    },
    {
      "epoch": 0.11542249359858184,
      "grad_norm": 1.3427600860595703,
      "learning_rate": 0.0001961979533844696,
      "loss": 1.4208,
      "step": 3223
    },
    {
      "epoch": 0.11545830572815012,
      "grad_norm": 2.6894454956054688,
      "learning_rate": 0.0001961947847970735,
      "loss": 1.7024,
      "step": 3224
    },
    {
      "epoch": 0.1154941178577184,
      "grad_norm": 2.293968677520752,
      "learning_rate": 0.00019619161491550065,
      "loss": 1.2496,
      "step": 3225
    },
    {
      "epoch": 0.1155299299872867,
      "grad_norm": 2.2852323055267334,
      "learning_rate": 0.00019618844373979372,
      "loss": 1.6852,
      "step": 3226
    },
    {
      "epoch": 0.11556574211685498,
      "grad_norm": 1.5715569257736206,
      "learning_rate": 0.0001961852712699953,
      "loss": 1.4904,
      "step": 3227
    },
    {
      "epoch": 0.11560155424642327,
      "grad_norm": 1.5239616632461548,
      "learning_rate": 0.00019618209750614813,
      "loss": 1.4977,
      "step": 3228
    },
    {
      "epoch": 0.11563736637599155,
      "grad_norm": 1.4947649240493774,
      "learning_rate": 0.00019617892244829495,
      "loss": 1.5866,
      "step": 3229
    },
    {
      "epoch": 0.11567317850555983,
      "grad_norm": 1.623744010925293,
      "learning_rate": 0.0001961757460964784,
      "loss": 1.5114,
      "step": 3230
    },
    {
      "epoch": 0.11570899063512811,
      "grad_norm": 1.7694364786148071,
      "learning_rate": 0.00019617256845074125,
      "loss": 1.6273,
      "step": 3231
    },
    {
      "epoch": 0.1157448027646964,
      "grad_norm": 1.5195114612579346,
      "learning_rate": 0.00019616938951112623,
      "loss": 1.3845,
      "step": 3232
    },
    {
      "epoch": 0.1157806148942647,
      "grad_norm": 1.6811881065368652,
      "learning_rate": 0.00019616620927767614,
      "loss": 1.3784,
      "step": 3233
    },
    {
      "epoch": 0.11581642702383298,
      "grad_norm": 1.8384026288986206,
      "learning_rate": 0.00019616302775043377,
      "loss": 1.6615,
      "step": 3234
    },
    {
      "epoch": 0.11585223915340126,
      "grad_norm": 1.5395166873931885,
      "learning_rate": 0.00019615984492944187,
      "loss": 1.7524,
      "step": 3235
    },
    {
      "epoch": 0.11588805128296954,
      "grad_norm": 1.3467326164245605,
      "learning_rate": 0.00019615666081474332,
      "loss": 1.6619,
      "step": 3236
    },
    {
      "epoch": 0.11592386341253783,
      "grad_norm": 1.4491621255874634,
      "learning_rate": 0.00019615347540638092,
      "loss": 1.5844,
      "step": 3237
    },
    {
      "epoch": 0.11595967554210611,
      "grad_norm": 2.1446890830993652,
      "learning_rate": 0.00019615028870439752,
      "loss": 1.6285,
      "step": 3238
    },
    {
      "epoch": 0.11599548767167439,
      "grad_norm": 2.09091854095459,
      "learning_rate": 0.00019614710070883602,
      "loss": 1.5445,
      "step": 3239
    },
    {
      "epoch": 0.11603129980124269,
      "grad_norm": 1.3230056762695312,
      "learning_rate": 0.00019614391141973934,
      "loss": 1.5391,
      "step": 3240
    },
    {
      "epoch": 0.11606711193081097,
      "grad_norm": 1.5421323776245117,
      "learning_rate": 0.00019614072083715028,
      "loss": 1.4999,
      "step": 3241
    },
    {
      "epoch": 0.11610292406037925,
      "grad_norm": 1.7281304597854614,
      "learning_rate": 0.00019613752896111187,
      "loss": 1.5916,
      "step": 3242
    },
    {
      "epoch": 0.11613873618994754,
      "grad_norm": 2.072479724884033,
      "learning_rate": 0.00019613433579166706,
      "loss": 1.922,
      "step": 3243
    },
    {
      "epoch": 0.11617454831951582,
      "grad_norm": 1.6558066606521606,
      "learning_rate": 0.0001961311413288587,
      "loss": 1.6862,
      "step": 3244
    },
    {
      "epoch": 0.1162103604490841,
      "grad_norm": 2.008986711502075,
      "learning_rate": 0.00019612794557272983,
      "loss": 1.5542,
      "step": 3245
    },
    {
      "epoch": 0.11624617257865238,
      "grad_norm": 1.22840416431427,
      "learning_rate": 0.00019612474852332348,
      "loss": 1.4519,
      "step": 3246
    },
    {
      "epoch": 0.11628198470822067,
      "grad_norm": 1.6340930461883545,
      "learning_rate": 0.00019612155018068264,
      "loss": 1.5469,
      "step": 3247
    },
    {
      "epoch": 0.11631779683778896,
      "grad_norm": 2.318631887435913,
      "learning_rate": 0.00019611835054485032,
      "loss": 1.7494,
      "step": 3248
    },
    {
      "epoch": 0.11635360896735725,
      "grad_norm": 1.6667400598526,
      "learning_rate": 0.00019611514961586957,
      "loss": 1.3893,
      "step": 3249
    },
    {
      "epoch": 0.11638942109692553,
      "grad_norm": 1.4194461107254028,
      "learning_rate": 0.00019611194739378344,
      "loss": 1.8322,
      "step": 3250
    },
    {
      "epoch": 0.11642523322649381,
      "grad_norm": 1.7950221300125122,
      "learning_rate": 0.00019610874387863508,
      "loss": 1.4541,
      "step": 3251
    },
    {
      "epoch": 0.1164610453560621,
      "grad_norm": 1.682098627090454,
      "learning_rate": 0.00019610553907046748,
      "loss": 1.6563,
      "step": 3252
    },
    {
      "epoch": 0.11649685748563038,
      "grad_norm": 2.29400372505188,
      "learning_rate": 0.0001961023329693239,
      "loss": 1.7833,
      "step": 3253
    },
    {
      "epoch": 0.11653266961519866,
      "grad_norm": 1.656288981437683,
      "learning_rate": 0.00019609912557524734,
      "loss": 1.5191,
      "step": 3254
    },
    {
      "epoch": 0.11656848174476696,
      "grad_norm": 1.3235529661178589,
      "learning_rate": 0.000196095916888281,
      "loss": 1.539,
      "step": 3255
    },
    {
      "epoch": 0.11660429387433524,
      "grad_norm": 1.794012188911438,
      "learning_rate": 0.00019609270690846807,
      "loss": 1.7339,
      "step": 3256
    },
    {
      "epoch": 0.11664010600390352,
      "grad_norm": 1.4943112134933472,
      "learning_rate": 0.00019608949563585174,
      "loss": 1.5602,
      "step": 3257
    },
    {
      "epoch": 0.1166759181334718,
      "grad_norm": 2.628278970718384,
      "learning_rate": 0.00019608628307047517,
      "loss": 1.7145,
      "step": 3258
    },
    {
      "epoch": 0.11671173026304009,
      "grad_norm": 1.359542965888977,
      "learning_rate": 0.0001960830692123816,
      "loss": 1.3189,
      "step": 3259
    },
    {
      "epoch": 0.11674754239260837,
      "grad_norm": 1.5178779363632202,
      "learning_rate": 0.00019607985406161425,
      "loss": 1.527,
      "step": 3260
    },
    {
      "epoch": 0.11678335452217666,
      "grad_norm": 1.6285661458969116,
      "learning_rate": 0.00019607663761821644,
      "loss": 1.8477,
      "step": 3261
    },
    {
      "epoch": 0.11681916665174495,
      "grad_norm": 1.4759529829025269,
      "learning_rate": 0.0001960734198822314,
      "loss": 1.6696,
      "step": 3262
    },
    {
      "epoch": 0.11685497878131323,
      "grad_norm": 1.8590339422225952,
      "learning_rate": 0.0001960702008537024,
      "loss": 1.5988,
      "step": 3263
    },
    {
      "epoch": 0.11689079091088152,
      "grad_norm": 2.2385072708129883,
      "learning_rate": 0.00019606698053267277,
      "loss": 1.7862,
      "step": 3264
    },
    {
      "epoch": 0.1169266030404498,
      "grad_norm": 1.7401012182235718,
      "learning_rate": 0.00019606375891918583,
      "loss": 1.8007,
      "step": 3265
    },
    {
      "epoch": 0.11696241517001808,
      "grad_norm": 1.3024067878723145,
      "learning_rate": 0.00019606053601328496,
      "loss": 1.5707,
      "step": 3266
    },
    {
      "epoch": 0.11699822729958637,
      "grad_norm": 2.206632375717163,
      "learning_rate": 0.00019605731181501342,
      "loss": 1.4116,
      "step": 3267
    },
    {
      "epoch": 0.11703403942915465,
      "grad_norm": 1.3507251739501953,
      "learning_rate": 0.00019605408632441474,
      "loss": 1.6662,
      "step": 3268
    },
    {
      "epoch": 0.11706985155872295,
      "grad_norm": 1.4132224321365356,
      "learning_rate": 0.00019605085954153218,
      "loss": 1.7236,
      "step": 3269
    },
    {
      "epoch": 0.11710566368829123,
      "grad_norm": 1.848435878753662,
      "learning_rate": 0.00019604763146640922,
      "loss": 1.774,
      "step": 3270
    },
    {
      "epoch": 0.11714147581785951,
      "grad_norm": 1.356055736541748,
      "learning_rate": 0.00019604440209908925,
      "loss": 1.9218,
      "step": 3271
    },
    {
      "epoch": 0.1171772879474278,
      "grad_norm": 1.9709981679916382,
      "learning_rate": 0.00019604117143961575,
      "loss": 1.5058,
      "step": 3272
    },
    {
      "epoch": 0.11721310007699608,
      "grad_norm": 1.8755549192428589,
      "learning_rate": 0.00019603793948803216,
      "loss": 1.4719,
      "step": 3273
    },
    {
      "epoch": 0.11724891220656436,
      "grad_norm": 2.5782339572906494,
      "learning_rate": 0.000196034706244382,
      "loss": 1.5972,
      "step": 3274
    },
    {
      "epoch": 0.11728472433613264,
      "grad_norm": 1.4497991800308228,
      "learning_rate": 0.0001960314717087087,
      "loss": 1.4209,
      "step": 3275
    },
    {
      "epoch": 0.11732053646570094,
      "grad_norm": 1.4412992000579834,
      "learning_rate": 0.00019602823588105585,
      "loss": 1.517,
      "step": 3276
    },
    {
      "epoch": 0.11735634859526922,
      "grad_norm": 1.4440698623657227,
      "learning_rate": 0.000196024998761467,
      "loss": 1.5662,
      "step": 3277
    },
    {
      "epoch": 0.1173921607248375,
      "grad_norm": 3.968775987625122,
      "learning_rate": 0.00019602176034998556,
      "loss": 2.1371,
      "step": 3278
    },
    {
      "epoch": 0.11742797285440579,
      "grad_norm": 1.4765177965164185,
      "learning_rate": 0.00019601852064665524,
      "loss": 1.6333,
      "step": 3279
    },
    {
      "epoch": 0.11746378498397407,
      "grad_norm": 1.5928294658660889,
      "learning_rate": 0.0001960152796515196,
      "loss": 1.6321,
      "step": 3280
    },
    {
      "epoch": 0.11749959711354235,
      "grad_norm": 1.4952478408813477,
      "learning_rate": 0.00019601203736462219,
      "loss": 1.7204,
      "step": 3281
    },
    {
      "epoch": 0.11753540924311064,
      "grad_norm": 2.540161609649658,
      "learning_rate": 0.00019600879378600666,
      "loss": 1.6557,
      "step": 3282
    },
    {
      "epoch": 0.11757122137267893,
      "grad_norm": 1.3990188837051392,
      "learning_rate": 0.0001960055489157167,
      "loss": 1.6378,
      "step": 3283
    },
    {
      "epoch": 0.11760703350224722,
      "grad_norm": 1.4970078468322754,
      "learning_rate": 0.00019600230275379588,
      "loss": 1.7807,
      "step": 3284
    },
    {
      "epoch": 0.1176428456318155,
      "grad_norm": 1.9799847602844238,
      "learning_rate": 0.0001959990553002879,
      "loss": 1.6429,
      "step": 3285
    },
    {
      "epoch": 0.11767865776138378,
      "grad_norm": 1.5458717346191406,
      "learning_rate": 0.0001959958065552365,
      "loss": 1.8378,
      "step": 3286
    },
    {
      "epoch": 0.11771446989095206,
      "grad_norm": 1.1473616361618042,
      "learning_rate": 0.0001959925565186853,
      "loss": 1.4595,
      "step": 3287
    },
    {
      "epoch": 0.11775028202052035,
      "grad_norm": 1.4953563213348389,
      "learning_rate": 0.00019598930519067813,
      "loss": 1.6126,
      "step": 3288
    },
    {
      "epoch": 0.11778609415008863,
      "grad_norm": 1.4678362607955933,
      "learning_rate": 0.00019598605257125864,
      "loss": 1.492,
      "step": 3289
    },
    {
      "epoch": 0.11782190627965693,
      "grad_norm": 2.026249408721924,
      "learning_rate": 0.0001959827986604706,
      "loss": 1.6692,
      "step": 3290
    },
    {
      "epoch": 0.11785771840922521,
      "grad_norm": 1.462710976600647,
      "learning_rate": 0.00019597954345835787,
      "loss": 1.4171,
      "step": 3291
    },
    {
      "epoch": 0.11789353053879349,
      "grad_norm": 1.6118189096450806,
      "learning_rate": 0.00019597628696496418,
      "loss": 1.7512,
      "step": 3292
    },
    {
      "epoch": 0.11792934266836178,
      "grad_norm": 1.621272087097168,
      "learning_rate": 0.0001959730291803333,
      "loss": 1.9068,
      "step": 3293
    },
    {
      "epoch": 0.11796515479793006,
      "grad_norm": 1.9732425212860107,
      "learning_rate": 0.00019596977010450915,
      "loss": 1.4781,
      "step": 3294
    },
    {
      "epoch": 0.11800096692749834,
      "grad_norm": 1.2705248594284058,
      "learning_rate": 0.00019596650973753555,
      "loss": 1.9466,
      "step": 3295
    },
    {
      "epoch": 0.11803677905706662,
      "grad_norm": 2.1659154891967773,
      "learning_rate": 0.00019596324807945632,
      "loss": 1.526,
      "step": 3296
    },
    {
      "epoch": 0.1180725911866349,
      "grad_norm": 1.6250512599945068,
      "learning_rate": 0.00019595998513031537,
      "loss": 1.6662,
      "step": 3297
    },
    {
      "epoch": 0.1181084033162032,
      "grad_norm": 3.29927659034729,
      "learning_rate": 0.00019595672089015663,
      "loss": 1.7103,
      "step": 3298
    },
    {
      "epoch": 0.11814421544577149,
      "grad_norm": 1.6981123685836792,
      "learning_rate": 0.00019595345535902394,
      "loss": 1.2357,
      "step": 3299
    },
    {
      "epoch": 0.11818002757533977,
      "grad_norm": 1.8738186359405518,
      "learning_rate": 0.0001959501885369613,
      "loss": 1.6669,
      "step": 3300
    },
    {
      "epoch": 0.11821583970490805,
      "grad_norm": 2.400287628173828,
      "learning_rate": 0.00019594692042401263,
      "loss": 1.6469,
      "step": 3301
    },
    {
      "epoch": 0.11825165183447633,
      "grad_norm": 2.324725389480591,
      "learning_rate": 0.00019594365102022193,
      "loss": 1.8658,
      "step": 3302
    },
    {
      "epoch": 0.11828746396404462,
      "grad_norm": 1.470152735710144,
      "learning_rate": 0.00019594038032563315,
      "loss": 1.7331,
      "step": 3303
    },
    {
      "epoch": 0.1183232760936129,
      "grad_norm": 1.8536076545715332,
      "learning_rate": 0.0001959371083402903,
      "loss": 1.6881,
      "step": 3304
    },
    {
      "epoch": 0.1183590882231812,
      "grad_norm": 1.580949306488037,
      "learning_rate": 0.00019593383506423743,
      "loss": 1.5505,
      "step": 3305
    },
    {
      "epoch": 0.11839490035274948,
      "grad_norm": 1.4899119138717651,
      "learning_rate": 0.00019593056049751852,
      "loss": 1.5014,
      "step": 3306
    },
    {
      "epoch": 0.11843071248231776,
      "grad_norm": 1.3627156019210815,
      "learning_rate": 0.0001959272846401777,
      "loss": 1.502,
      "step": 3307
    },
    {
      "epoch": 0.11846652461188605,
      "grad_norm": 2.0167927742004395,
      "learning_rate": 0.000195924007492259,
      "loss": 1.6057,
      "step": 3308
    },
    {
      "epoch": 0.11850233674145433,
      "grad_norm": 2.2905960083007812,
      "learning_rate": 0.00019592072905380648,
      "loss": 1.6687,
      "step": 3309
    },
    {
      "epoch": 0.11853814887102261,
      "grad_norm": 2.684022903442383,
      "learning_rate": 0.00019591744932486428,
      "loss": 1.6046,
      "step": 3310
    },
    {
      "epoch": 0.1185739610005909,
      "grad_norm": 1.62175452709198,
      "learning_rate": 0.00019591416830547657,
      "loss": 1.5254,
      "step": 3311
    },
    {
      "epoch": 0.11860977313015919,
      "grad_norm": 1.9423574209213257,
      "learning_rate": 0.0001959108859956874,
      "loss": 1.6332,
      "step": 3312
    },
    {
      "epoch": 0.11864558525972747,
      "grad_norm": 2.0534324645996094,
      "learning_rate": 0.00019590760239554097,
      "loss": 1.6084,
      "step": 3313
    },
    {
      "epoch": 0.11868139738929576,
      "grad_norm": 1.9046193361282349,
      "learning_rate": 0.00019590431750508153,
      "loss": 1.585,
      "step": 3314
    },
    {
      "epoch": 0.11871720951886404,
      "grad_norm": 1.8338521718978882,
      "learning_rate": 0.00019590103132435314,
      "loss": 1.8236,
      "step": 3315
    },
    {
      "epoch": 0.11875302164843232,
      "grad_norm": 1.7030360698699951,
      "learning_rate": 0.00019589774385340007,
      "loss": 1.716,
      "step": 3316
    },
    {
      "epoch": 0.1187888337780006,
      "grad_norm": 2.5421273708343506,
      "learning_rate": 0.0001958944550922666,
      "loss": 1.6587,
      "step": 3317
    },
    {
      "epoch": 0.11882464590756889,
      "grad_norm": 1.5858945846557617,
      "learning_rate": 0.0001958911650409969,
      "loss": 1.4128,
      "step": 3318
    },
    {
      "epoch": 0.11886045803713718,
      "grad_norm": 1.6112788915634155,
      "learning_rate": 0.0001958878736996353,
      "loss": 1.5822,
      "step": 3319
    },
    {
      "epoch": 0.11889627016670547,
      "grad_norm": 1.3577380180358887,
      "learning_rate": 0.00019588458106822602,
      "loss": 1.5731,
      "step": 3320
    },
    {
      "epoch": 0.11893208229627375,
      "grad_norm": 1.960248351097107,
      "learning_rate": 0.00019588128714681337,
      "loss": 1.6318,
      "step": 3321
    },
    {
      "epoch": 0.11896789442584203,
      "grad_norm": 1.8858251571655273,
      "learning_rate": 0.0001958779919354417,
      "loss": 1.8675,
      "step": 3322
    },
    {
      "epoch": 0.11900370655541032,
      "grad_norm": 1.6242177486419678,
      "learning_rate": 0.00019587469543415532,
      "loss": 1.5496,
      "step": 3323
    },
    {
      "epoch": 0.1190395186849786,
      "grad_norm": 1.738443374633789,
      "learning_rate": 0.00019587139764299857,
      "loss": 1.614,
      "step": 3324
    },
    {
      "epoch": 0.11907533081454688,
      "grad_norm": 1.9253852367401123,
      "learning_rate": 0.00019586809856201586,
      "loss": 1.8159,
      "step": 3325
    },
    {
      "epoch": 0.11911114294411518,
      "grad_norm": 1.6207756996154785,
      "learning_rate": 0.00019586479819125153,
      "loss": 1.3749,
      "step": 3326
    },
    {
      "epoch": 0.11914695507368346,
      "grad_norm": 1.6630465984344482,
      "learning_rate": 0.00019586149653074997,
      "loss": 1.3199,
      "step": 3327
    },
    {
      "epoch": 0.11918276720325174,
      "grad_norm": 1.2815748453140259,
      "learning_rate": 0.00019585819358055567,
      "loss": 1.6019,
      "step": 3328
    },
    {
      "epoch": 0.11921857933282003,
      "grad_norm": 2.2043771743774414,
      "learning_rate": 0.00019585488934071302,
      "loss": 1.7924,
      "step": 3329
    },
    {
      "epoch": 0.11925439146238831,
      "grad_norm": 1.9640440940856934,
      "learning_rate": 0.00019585158381126645,
      "loss": 1.4117,
      "step": 3330
    },
    {
      "epoch": 0.11929020359195659,
      "grad_norm": 1.911688208580017,
      "learning_rate": 0.00019584827699226044,
      "loss": 1.6101,
      "step": 3331
    },
    {
      "epoch": 0.11932601572152488,
      "grad_norm": 1.7753043174743652,
      "learning_rate": 0.00019584496888373955,
      "loss": 1.353,
      "step": 3332
    },
    {
      "epoch": 0.11936182785109317,
      "grad_norm": 1.6217472553253174,
      "learning_rate": 0.00019584165948574822,
      "loss": 1.6926,
      "step": 3333
    },
    {
      "epoch": 0.11939763998066145,
      "grad_norm": 1.9915986061096191,
      "learning_rate": 0.00019583834879833097,
      "loss": 1.3721,
      "step": 3334
    },
    {
      "epoch": 0.11943345211022974,
      "grad_norm": 2.1937718391418457,
      "learning_rate": 0.0001958350368215324,
      "loss": 1.526,
      "step": 3335
    },
    {
      "epoch": 0.11946926423979802,
      "grad_norm": 1.4065150022506714,
      "learning_rate": 0.00019583172355539698,
      "loss": 1.5645,
      "step": 3336
    },
    {
      "epoch": 0.1195050763693663,
      "grad_norm": 1.9046217203140259,
      "learning_rate": 0.00019582840899996936,
      "loss": 1.4022,
      "step": 3337
    },
    {
      "epoch": 0.11954088849893459,
      "grad_norm": 1.5891504287719727,
      "learning_rate": 0.00019582509315529408,
      "loss": 1.7865,
      "step": 3338
    },
    {
      "epoch": 0.11957670062850287,
      "grad_norm": 2.062664270401001,
      "learning_rate": 0.0001958217760214158,
      "loss": 1.8436,
      "step": 3339
    },
    {
      "epoch": 0.11961251275807117,
      "grad_norm": 1.4193004369735718,
      "learning_rate": 0.00019581845759837914,
      "loss": 1.5553,
      "step": 3340
    },
    {
      "epoch": 0.11964832488763945,
      "grad_norm": 1.8979054689407349,
      "learning_rate": 0.0001958151378862287,
      "loss": 1.4616,
      "step": 3341
    },
    {
      "epoch": 0.11968413701720773,
      "grad_norm": 1.922569990158081,
      "learning_rate": 0.00019581181688500918,
      "loss": 1.7021,
      "step": 3342
    },
    {
      "epoch": 0.11971994914677601,
      "grad_norm": 1.8692396879196167,
      "learning_rate": 0.00019580849459476527,
      "loss": 1.5885,
      "step": 3343
    },
    {
      "epoch": 0.1197557612763443,
      "grad_norm": 1.5603022575378418,
      "learning_rate": 0.00019580517101554164,
      "loss": 1.4791,
      "step": 3344
    },
    {
      "epoch": 0.11979157340591258,
      "grad_norm": 1.3422173261642456,
      "learning_rate": 0.00019580184614738299,
      "loss": 1.626,
      "step": 3345
    },
    {
      "epoch": 0.11982738553548086,
      "grad_norm": 1.8114674091339111,
      "learning_rate": 0.0001957985199903341,
      "loss": 1.5095,
      "step": 3346
    },
    {
      "epoch": 0.11986319766504915,
      "grad_norm": 1.594317078590393,
      "learning_rate": 0.00019579519254443967,
      "loss": 1.7669,
      "step": 3347
    },
    {
      "epoch": 0.11989900979461744,
      "grad_norm": 1.1270697116851807,
      "learning_rate": 0.00019579186380974455,
      "loss": 1.3657,
      "step": 3348
    },
    {
      "epoch": 0.11993482192418572,
      "grad_norm": 1.7906159162521362,
      "learning_rate": 0.0001957885337862934,
      "loss": 1.7797,
      "step": 3349
    },
    {
      "epoch": 0.11997063405375401,
      "grad_norm": 1.6569995880126953,
      "learning_rate": 0.00019578520247413113,
      "loss": 1.7222,
      "step": 3350
    },
    {
      "epoch": 0.12000644618332229,
      "grad_norm": 1.458554744720459,
      "learning_rate": 0.0001957818698733025,
      "loss": 1.7073,
      "step": 3351
    },
    {
      "epoch": 0.12004225831289057,
      "grad_norm": 1.9747871160507202,
      "learning_rate": 0.00019577853598385235,
      "loss": 1.495,
      "step": 3352
    },
    {
      "epoch": 0.12007807044245886,
      "grad_norm": 1.7646852731704712,
      "learning_rate": 0.00019577520080582556,
      "loss": 1.6767,
      "step": 3353
    },
    {
      "epoch": 0.12011388257202714,
      "grad_norm": 1.4243603944778442,
      "learning_rate": 0.00019577186433926698,
      "loss": 1.8968,
      "step": 3354
    },
    {
      "epoch": 0.12014969470159544,
      "grad_norm": 1.5458492040634155,
      "learning_rate": 0.00019576852658422146,
      "loss": 1.7073,
      "step": 3355
    },
    {
      "epoch": 0.12018550683116372,
      "grad_norm": 2.496417760848999,
      "learning_rate": 0.000195765187540734,
      "loss": 2.0073,
      "step": 3356
    },
    {
      "epoch": 0.120221318960732,
      "grad_norm": 2.2858903408050537,
      "learning_rate": 0.00019576184720884946,
      "loss": 1.893,
      "step": 3357
    },
    {
      "epoch": 0.12025713109030028,
      "grad_norm": 1.6809884309768677,
      "learning_rate": 0.00019575850558861278,
      "loss": 1.597,
      "step": 3358
    },
    {
      "epoch": 0.12029294321986857,
      "grad_norm": 2.288938283920288,
      "learning_rate": 0.00019575516268006892,
      "loss": 1.594,
      "step": 3359
    },
    {
      "epoch": 0.12032875534943685,
      "grad_norm": 1.5342596769332886,
      "learning_rate": 0.00019575181848326289,
      "loss": 1.4432,
      "step": 3360
    },
    {
      "epoch": 0.12036456747900513,
      "grad_norm": 1.6115968227386475,
      "learning_rate": 0.00019574847299823965,
      "loss": 1.6213,
      "step": 3361
    },
    {
      "epoch": 0.12040037960857343,
      "grad_norm": 1.2919033765792847,
      "learning_rate": 0.00019574512622504416,
      "loss": 1.6027,
      "step": 3362
    },
    {
      "epoch": 0.12043619173814171,
      "grad_norm": 1.2876076698303223,
      "learning_rate": 0.00019574177816372154,
      "loss": 1.541,
      "step": 3363
    },
    {
      "epoch": 0.12047200386771,
      "grad_norm": 1.2301689386367798,
      "learning_rate": 0.0001957384288143168,
      "loss": 1.733,
      "step": 3364
    },
    {
      "epoch": 0.12050781599727828,
      "grad_norm": 2.5293233394622803,
      "learning_rate": 0.000195735078176875,
      "loss": 1.7144,
      "step": 3365
    },
    {
      "epoch": 0.12054362812684656,
      "grad_norm": 2.2179653644561768,
      "learning_rate": 0.0001957317262514412,
      "loss": 1.673,
      "step": 3366
    },
    {
      "epoch": 0.12057944025641484,
      "grad_norm": 2.00532603263855,
      "learning_rate": 0.00019572837303806048,
      "loss": 1.7745,
      "step": 3367
    },
    {
      "epoch": 0.12061525238598313,
      "grad_norm": 1.687103271484375,
      "learning_rate": 0.00019572501853677802,
      "loss": 1.4363,
      "step": 3368
    },
    {
      "epoch": 0.12065106451555142,
      "grad_norm": 1.7787339687347412,
      "learning_rate": 0.0001957216627476389,
      "loss": 1.5124,
      "step": 3369
    },
    {
      "epoch": 0.1206868766451197,
      "grad_norm": 2.2750205993652344,
      "learning_rate": 0.0001957183056706883,
      "loss": 1.6114,
      "step": 3370
    },
    {
      "epoch": 0.12072268877468799,
      "grad_norm": 2.279995918273926,
      "learning_rate": 0.0001957149473059713,
      "loss": 1.5944,
      "step": 3371
    },
    {
      "epoch": 0.12075850090425627,
      "grad_norm": 1.3898695707321167,
      "learning_rate": 0.0001957115876535332,
      "loss": 1.7184,
      "step": 3372
    },
    {
      "epoch": 0.12079431303382455,
      "grad_norm": 1.9735623598098755,
      "learning_rate": 0.00019570822671341915,
      "loss": 1.6018,
      "step": 3373
    },
    {
      "epoch": 0.12083012516339284,
      "grad_norm": 1.697853684425354,
      "learning_rate": 0.00019570486448567437,
      "loss": 1.6601,
      "step": 3374
    },
    {
      "epoch": 0.12086593729296112,
      "grad_norm": 1.8338154554367065,
      "learning_rate": 0.00019570150097034404,
      "loss": 1.7257,
      "step": 3375
    },
    {
      "epoch": 0.12090174942252942,
      "grad_norm": 1.4777474403381348,
      "learning_rate": 0.0001956981361674735,
      "loss": 1.7592,
      "step": 3376
    },
    {
      "epoch": 0.1209375615520977,
      "grad_norm": 1.6526165008544922,
      "learning_rate": 0.00019569477007710798,
      "loss": 1.714,
      "step": 3377
    },
    {
      "epoch": 0.12097337368166598,
      "grad_norm": 1.5610233545303345,
      "learning_rate": 0.00019569140269929276,
      "loss": 1.7781,
      "step": 3378
    },
    {
      "epoch": 0.12100918581123427,
      "grad_norm": 1.9586926698684692,
      "learning_rate": 0.00019568803403407315,
      "loss": 1.7737,
      "step": 3379
    },
    {
      "epoch": 0.12104499794080255,
      "grad_norm": 1.7385483980178833,
      "learning_rate": 0.00019568466408149447,
      "loss": 1.3575,
      "step": 3380
    },
    {
      "epoch": 0.12108081007037083,
      "grad_norm": 1.839732050895691,
      "learning_rate": 0.00019568129284160203,
      "loss": 1.5505,
      "step": 3381
    },
    {
      "epoch": 0.12111662219993911,
      "grad_norm": 1.285846471786499,
      "learning_rate": 0.00019567792031444125,
      "loss": 1.4664,
      "step": 3382
    },
    {
      "epoch": 0.12115243432950741,
      "grad_norm": 1.6386572122573853,
      "learning_rate": 0.00019567454650005749,
      "loss": 1.6598,
      "step": 3383
    },
    {
      "epoch": 0.1211882464590757,
      "grad_norm": 1.7839397192001343,
      "learning_rate": 0.00019567117139849605,
      "loss": 1.7173,
      "step": 3384
    },
    {
      "epoch": 0.12122405858864398,
      "grad_norm": 1.632829189300537,
      "learning_rate": 0.00019566779500980247,
      "loss": 1.3093,
      "step": 3385
    },
    {
      "epoch": 0.12125987071821226,
      "grad_norm": 1.8660407066345215,
      "learning_rate": 0.00019566441733402207,
      "loss": 1.2346,
      "step": 3386
    },
    {
      "epoch": 0.12129568284778054,
      "grad_norm": 1.885238766670227,
      "learning_rate": 0.00019566103837120036,
      "loss": 1.3926,
      "step": 3387
    },
    {
      "epoch": 0.12133149497734882,
      "grad_norm": 1.5017540454864502,
      "learning_rate": 0.00019565765812138274,
      "loss": 1.4036,
      "step": 3388
    },
    {
      "epoch": 0.12136730710691711,
      "grad_norm": 1.5726633071899414,
      "learning_rate": 0.00019565427658461474,
      "loss": 1.624,
      "step": 3389
    },
    {
      "epoch": 0.1214031192364854,
      "grad_norm": 1.7805520296096802,
      "learning_rate": 0.00019565089376094184,
      "loss": 1.7118,
      "step": 3390
    },
    {
      "epoch": 0.12143893136605369,
      "grad_norm": 1.9089933633804321,
      "learning_rate": 0.0001956475096504095,
      "loss": 1.7371,
      "step": 3391
    },
    {
      "epoch": 0.12147474349562197,
      "grad_norm": 1.7601195573806763,
      "learning_rate": 0.00019564412425306338,
      "loss": 1.9506,
      "step": 3392
    },
    {
      "epoch": 0.12151055562519025,
      "grad_norm": 1.944966197013855,
      "learning_rate": 0.00019564073756894889,
      "loss": 1.8287,
      "step": 3393
    },
    {
      "epoch": 0.12154636775475854,
      "grad_norm": 2.336825370788574,
      "learning_rate": 0.00019563734959811163,
      "loss": 1.5147,
      "step": 3394
    },
    {
      "epoch": 0.12158217988432682,
      "grad_norm": 2.415855646133423,
      "learning_rate": 0.00019563396034059724,
      "loss": 1.4703,
      "step": 3395
    },
    {
      "epoch": 0.1216179920138951,
      "grad_norm": 1.4387824535369873,
      "learning_rate": 0.00019563056979645123,
      "loss": 1.4512,
      "step": 3396
    },
    {
      "epoch": 0.12165380414346338,
      "grad_norm": 2.0192601680755615,
      "learning_rate": 0.00019562717796571929,
      "loss": 1.536,
      "step": 3397
    },
    {
      "epoch": 0.12168961627303168,
      "grad_norm": 1.5521689653396606,
      "learning_rate": 0.00019562378484844697,
      "loss": 1.6861,
      "step": 3398
    },
    {
      "epoch": 0.12172542840259996,
      "grad_norm": 1.3953551054000854,
      "learning_rate": 0.00019562039044468,
      "loss": 1.713,
      "step": 3399
    },
    {
      "epoch": 0.12176124053216825,
      "grad_norm": 2.014117956161499,
      "learning_rate": 0.00019561699475446401,
      "loss": 1.4568,
      "step": 3400
    },
    {
      "epoch": 0.12179705266173653,
      "grad_norm": 1.4220621585845947,
      "learning_rate": 0.00019561359777784472,
      "loss": 1.4072,
      "step": 3401
    },
    {
      "epoch": 0.12183286479130481,
      "grad_norm": 1.6513803005218506,
      "learning_rate": 0.0001956101995148678,
      "loss": 1.6897,
      "step": 3402
    },
    {
      "epoch": 0.1218686769208731,
      "grad_norm": 1.61029052734375,
      "learning_rate": 0.00019560679996557894,
      "loss": 1.6601,
      "step": 3403
    },
    {
      "epoch": 0.12190448905044138,
      "grad_norm": 1.2982370853424072,
      "learning_rate": 0.00019560339913002396,
      "loss": 1.6648,
      "step": 3404
    },
    {
      "epoch": 0.12194030118000967,
      "grad_norm": 1.9422756433486938,
      "learning_rate": 0.00019559999700824852,
      "loss": 1.8368,
      "step": 3405
    },
    {
      "epoch": 0.12197611330957796,
      "grad_norm": 1.4068212509155273,
      "learning_rate": 0.00019559659360029845,
      "loss": 1.681,
      "step": 3406
    },
    {
      "epoch": 0.12201192543914624,
      "grad_norm": 2.2662994861602783,
      "learning_rate": 0.0001955931889062195,
      "loss": 1.7048,
      "step": 3407
    },
    {
      "epoch": 0.12204773756871452,
      "grad_norm": 1.8388558626174927,
      "learning_rate": 0.00019558978292605754,
      "loss": 1.4593,
      "step": 3408
    },
    {
      "epoch": 0.1220835496982828,
      "grad_norm": 2.5157718658447266,
      "learning_rate": 0.00019558637565985834,
      "loss": 1.5091,
      "step": 3409
    },
    {
      "epoch": 0.12211936182785109,
      "grad_norm": 1.798851728439331,
      "learning_rate": 0.00019558296710766774,
      "loss": 1.5104,
      "step": 3410
    },
    {
      "epoch": 0.12215517395741937,
      "grad_norm": 1.3769530057907104,
      "learning_rate": 0.00019557955726953163,
      "loss": 1.516,
      "step": 3411
    },
    {
      "epoch": 0.12219098608698767,
      "grad_norm": 1.429823398590088,
      "learning_rate": 0.00019557614614549586,
      "loss": 1.6896,
      "step": 3412
    },
    {
      "epoch": 0.12222679821655595,
      "grad_norm": 2.1437742710113525,
      "learning_rate": 0.00019557273373560632,
      "loss": 1.8087,
      "step": 3413
    },
    {
      "epoch": 0.12226261034612423,
      "grad_norm": 1.2321090698242188,
      "learning_rate": 0.00019556932003990892,
      "loss": 1.331,
      "step": 3414
    },
    {
      "epoch": 0.12229842247569252,
      "grad_norm": 1.5394012928009033,
      "learning_rate": 0.0001955659050584496,
      "loss": 1.6945,
      "step": 3415
    },
    {
      "epoch": 0.1223342346052608,
      "grad_norm": 1.67526376247406,
      "learning_rate": 0.0001955624887912743,
      "loss": 1.3599,
      "step": 3416
    },
    {
      "epoch": 0.12237004673482908,
      "grad_norm": 1.7456480264663696,
      "learning_rate": 0.00019555907123842902,
      "loss": 1.6028,
      "step": 3417
    },
    {
      "epoch": 0.12240585886439737,
      "grad_norm": 1.8807971477508545,
      "learning_rate": 0.00019555565239995966,
      "loss": 1.4512,
      "step": 3418
    },
    {
      "epoch": 0.12244167099396566,
      "grad_norm": 1.4449418783187866,
      "learning_rate": 0.00019555223227591225,
      "loss": 1.6143,
      "step": 3419
    },
    {
      "epoch": 0.12247748312353395,
      "grad_norm": 1.2141433954238892,
      "learning_rate": 0.0001955488108663328,
      "loss": 1.6618,
      "step": 3420
    },
    {
      "epoch": 0.12251329525310223,
      "grad_norm": 1.769530177116394,
      "learning_rate": 0.00019554538817126739,
      "loss": 1.5924,
      "step": 3421
    },
    {
      "epoch": 0.12254910738267051,
      "grad_norm": 1.607627272605896,
      "learning_rate": 0.000195541964190762,
      "loss": 1.64,
      "step": 3422
    },
    {
      "epoch": 0.1225849195122388,
      "grad_norm": 2.0021603107452393,
      "learning_rate": 0.00019553853892486273,
      "loss": 1.5966,
      "step": 3423
    },
    {
      "epoch": 0.12262073164180708,
      "grad_norm": 1.4065313339233398,
      "learning_rate": 0.00019553511237361564,
      "loss": 1.7312,
      "step": 3424
    },
    {
      "epoch": 0.12265654377137536,
      "grad_norm": 2.360572338104248,
      "learning_rate": 0.00019553168453706685,
      "loss": 1.6099,
      "step": 3425
    },
    {
      "epoch": 0.12269235590094366,
      "grad_norm": 1.807173728942871,
      "learning_rate": 0.00019552825541526247,
      "loss": 1.4694,
      "step": 3426
    },
    {
      "epoch": 0.12272816803051194,
      "grad_norm": 1.3030319213867188,
      "learning_rate": 0.00019552482500824865,
      "loss": 1.8069,
      "step": 3427
    },
    {
      "epoch": 0.12276398016008022,
      "grad_norm": 1.7905423641204834,
      "learning_rate": 0.0001955213933160715,
      "loss": 1.5229,
      "step": 3428
    },
    {
      "epoch": 0.1227997922896485,
      "grad_norm": 2.1524462699890137,
      "learning_rate": 0.00019551796033877726,
      "loss": 1.3083,
      "step": 3429
    },
    {
      "epoch": 0.12283560441921679,
      "grad_norm": 1.6406817436218262,
      "learning_rate": 0.00019551452607641205,
      "loss": 1.8781,
      "step": 3430
    },
    {
      "epoch": 0.12287141654878507,
      "grad_norm": 1.5875840187072754,
      "learning_rate": 0.0001955110905290221,
      "loss": 1.9099,
      "step": 3431
    },
    {
      "epoch": 0.12290722867835335,
      "grad_norm": 1.7724831104278564,
      "learning_rate": 0.00019550765369665362,
      "loss": 1.4552,
      "step": 3432
    },
    {
      "epoch": 0.12294304080792165,
      "grad_norm": 1.4344149827957153,
      "learning_rate": 0.00019550421557935286,
      "loss": 1.7283,
      "step": 3433
    },
    {
      "epoch": 0.12297885293748993,
      "grad_norm": 1.5454742908477783,
      "learning_rate": 0.00019550077617716606,
      "loss": 1.7259,
      "step": 3434
    },
    {
      "epoch": 0.12301466506705822,
      "grad_norm": 1.3281104564666748,
      "learning_rate": 0.00019549733549013954,
      "loss": 1.674,
      "step": 3435
    },
    {
      "epoch": 0.1230504771966265,
      "grad_norm": 1.5898027420043945,
      "learning_rate": 0.0001954938935183195,
      "loss": 1.7844,
      "step": 3436
    },
    {
      "epoch": 0.12308628932619478,
      "grad_norm": 1.7751864194869995,
      "learning_rate": 0.00019549045026175232,
      "loss": 1.6153,
      "step": 3437
    },
    {
      "epoch": 0.12312210145576306,
      "grad_norm": 1.5239864587783813,
      "learning_rate": 0.00019548700572048433,
      "loss": 1.5856,
      "step": 3438
    },
    {
      "epoch": 0.12315791358533135,
      "grad_norm": 1.585067629814148,
      "learning_rate": 0.00019548355989456182,
      "loss": 1.5034,
      "step": 3439
    },
    {
      "epoch": 0.12319372571489964,
      "grad_norm": 1.449587345123291,
      "learning_rate": 0.0001954801127840312,
      "loss": 1.4557,
      "step": 3440
    },
    {
      "epoch": 0.12322953784446793,
      "grad_norm": 1.8552396297454834,
      "learning_rate": 0.00019547666438893879,
      "loss": 1.4238,
      "step": 3441
    },
    {
      "epoch": 0.12326534997403621,
      "grad_norm": 1.5274230241775513,
      "learning_rate": 0.00019547321470933103,
      "loss": 1.4542,
      "step": 3442
    },
    {
      "epoch": 0.12330116210360449,
      "grad_norm": 1.3685420751571655,
      "learning_rate": 0.00019546976374525433,
      "loss": 1.5404,
      "step": 3443
    },
    {
      "epoch": 0.12333697423317277,
      "grad_norm": 1.6194121837615967,
      "learning_rate": 0.0001954663114967551,
      "loss": 1.2244,
      "step": 3444
    },
    {
      "epoch": 0.12337278636274106,
      "grad_norm": 1.4230560064315796,
      "learning_rate": 0.0001954628579638798,
      "loss": 1.0127,
      "step": 3445
    },
    {
      "epoch": 0.12340859849230934,
      "grad_norm": 1.3849716186523438,
      "learning_rate": 0.0001954594031466749,
      "loss": 1.425,
      "step": 3446
    },
    {
      "epoch": 0.12344441062187762,
      "grad_norm": 1.335742473602295,
      "learning_rate": 0.00019545594704518682,
      "loss": 1.7514,
      "step": 3447
    },
    {
      "epoch": 0.12348022275144592,
      "grad_norm": 1.256226897239685,
      "learning_rate": 0.00019545248965946216,
      "loss": 1.6755,
      "step": 3448
    },
    {
      "epoch": 0.1235160348810142,
      "grad_norm": 1.8748904466629028,
      "learning_rate": 0.00019544903098954732,
      "loss": 1.516,
      "step": 3449
    },
    {
      "epoch": 0.12355184701058249,
      "grad_norm": 1.9898053407669067,
      "learning_rate": 0.0001954455710354889,
      "loss": 1.5757,
      "step": 3450
    },
    {
      "epoch": 0.12358765914015077,
      "grad_norm": 2.141721725463867,
      "learning_rate": 0.00019544210979733343,
      "loss": 1.9781,
      "step": 3451
    },
    {
      "epoch": 0.12362347126971905,
      "grad_norm": 1.461661696434021,
      "learning_rate": 0.0001954386472751275,
      "loss": 1.8818,
      "step": 3452
    },
    {
      "epoch": 0.12365928339928733,
      "grad_norm": 2.044703245162964,
      "learning_rate": 0.0001954351834689177,
      "loss": 1.8225,
      "step": 3453
    },
    {
      "epoch": 0.12369509552885562,
      "grad_norm": 2.4182326793670654,
      "learning_rate": 0.0001954317183787506,
      "loss": 1.6916,
      "step": 3454
    },
    {
      "epoch": 0.12373090765842391,
      "grad_norm": 1.5588754415512085,
      "learning_rate": 0.00019542825200467279,
      "loss": 1.7322,
      "step": 3455
    },
    {
      "epoch": 0.1237667197879922,
      "grad_norm": 1.1768525838851929,
      "learning_rate": 0.00019542478434673096,
      "loss": 1.5594,
      "step": 3456
    },
    {
      "epoch": 0.12380253191756048,
      "grad_norm": 1.7234669923782349,
      "learning_rate": 0.00019542131540497174,
      "loss": 1.7051,
      "step": 3457
    },
    {
      "epoch": 0.12383834404712876,
      "grad_norm": 1.9066272974014282,
      "learning_rate": 0.00019541784517944182,
      "loss": 1.7679,
      "step": 3458
    },
    {
      "epoch": 0.12387415617669705,
      "grad_norm": 1.688675045967102,
      "learning_rate": 0.0001954143736701879,
      "loss": 1.2632,
      "step": 3459
    },
    {
      "epoch": 0.12390996830626533,
      "grad_norm": 1.6447263956069946,
      "learning_rate": 0.0001954109008772566,
      "loss": 1.7467,
      "step": 3460
    },
    {
      "epoch": 0.12394578043583361,
      "grad_norm": 1.7118195295333862,
      "learning_rate": 0.00019540742680069473,
      "loss": 1.6544,
      "step": 3461
    },
    {
      "epoch": 0.12398159256540191,
      "grad_norm": 2.0889880657196045,
      "learning_rate": 0.000195403951440549,
      "loss": 1.8585,
      "step": 3462
    },
    {
      "epoch": 0.12401740469497019,
      "grad_norm": 1.5095477104187012,
      "learning_rate": 0.00019540047479686616,
      "loss": 1.5888,
      "step": 3463
    },
    {
      "epoch": 0.12405321682453847,
      "grad_norm": 1.3710155487060547,
      "learning_rate": 0.00019539699686969302,
      "loss": 1.4073,
      "step": 3464
    },
    {
      "epoch": 0.12408902895410676,
      "grad_norm": 1.261867642402649,
      "learning_rate": 0.0001953935176590763,
      "loss": 1.7322,
      "step": 3465
    },
    {
      "epoch": 0.12412484108367504,
      "grad_norm": 1.395007848739624,
      "learning_rate": 0.00019539003716506287,
      "loss": 1.5827,
      "step": 3466
    },
    {
      "epoch": 0.12416065321324332,
      "grad_norm": 1.8088184595108032,
      "learning_rate": 0.0001953865553876995,
      "loss": 2.0731,
      "step": 3467
    },
    {
      "epoch": 0.1241964653428116,
      "grad_norm": 2.2658514976501465,
      "learning_rate": 0.00019538307232703313,
      "loss": 1.6118,
      "step": 3468
    },
    {
      "epoch": 0.1242322774723799,
      "grad_norm": 1.3344290256500244,
      "learning_rate": 0.0001953795879831105,
      "loss": 1.6023,
      "step": 3469
    },
    {
      "epoch": 0.12426808960194818,
      "grad_norm": 1.413332462310791,
      "learning_rate": 0.00019537610235597857,
      "loss": 1.5516,
      "step": 3470
    },
    {
      "epoch": 0.12430390173151647,
      "grad_norm": 1.3064593076705933,
      "learning_rate": 0.00019537261544568421,
      "loss": 1.3519,
      "step": 3471
    },
    {
      "epoch": 0.12433971386108475,
      "grad_norm": 4.2692060470581055,
      "learning_rate": 0.00019536912725227432,
      "loss": 1.7083,
      "step": 3472
    },
    {
      "epoch": 0.12437552599065303,
      "grad_norm": 1.505118489265442,
      "learning_rate": 0.00019536563777579585,
      "loss": 1.597,
      "step": 3473
    },
    {
      "epoch": 0.12441133812022132,
      "grad_norm": 1.7473787069320679,
      "learning_rate": 0.0001953621470162957,
      "loss": 1.7033,
      "step": 3474
    },
    {
      "epoch": 0.1244471502497896,
      "grad_norm": 1.8221546411514282,
      "learning_rate": 0.00019535865497382094,
      "loss": 1.4684,
      "step": 3475
    },
    {
      "epoch": 0.1244829623793579,
      "grad_norm": 1.8060365915298462,
      "learning_rate": 0.00019535516164841842,
      "loss": 1.6477,
      "step": 3476
    },
    {
      "epoch": 0.12451877450892618,
      "grad_norm": 1.790130853652954,
      "learning_rate": 0.00019535166704013522,
      "loss": 1.5853,
      "step": 3477
    },
    {
      "epoch": 0.12455458663849446,
      "grad_norm": 2.9032599925994873,
      "learning_rate": 0.00019534817114901833,
      "loss": 1.7533,
      "step": 3478
    },
    {
      "epoch": 0.12459039876806274,
      "grad_norm": 1.4391133785247803,
      "learning_rate": 0.0001953446739751148,
      "loss": 1.6817,
      "step": 3479
    },
    {
      "epoch": 0.12462621089763103,
      "grad_norm": 1.8639107942581177,
      "learning_rate": 0.00019534117551847166,
      "loss": 1.6716,
      "step": 3480
    },
    {
      "epoch": 0.12466202302719931,
      "grad_norm": 1.7020066976547241,
      "learning_rate": 0.000195337675779136,
      "loss": 1.7672,
      "step": 3481
    },
    {
      "epoch": 0.12469783515676759,
      "grad_norm": 2.505441188812256,
      "learning_rate": 0.00019533417475715487,
      "loss": 1.6561,
      "step": 3482
    },
    {
      "epoch": 0.12473364728633589,
      "grad_norm": 1.5666439533233643,
      "learning_rate": 0.0001953306724525754,
      "loss": 1.6065,
      "step": 3483
    },
    {
      "epoch": 0.12476945941590417,
      "grad_norm": 1.5428463220596313,
      "learning_rate": 0.00019532716886544468,
      "loss": 1.4111,
      "step": 3484
    },
    {
      "epoch": 0.12480527154547245,
      "grad_norm": 1.5650322437286377,
      "learning_rate": 0.0001953236639958099,
      "loss": 1.5319,
      "step": 3485
    },
    {
      "epoch": 0.12484108367504074,
      "grad_norm": 1.1498960256576538,
      "learning_rate": 0.00019532015784371818,
      "loss": 1.8039,
      "step": 3486
    },
    {
      "epoch": 0.12487689580460902,
      "grad_norm": 1.6411759853363037,
      "learning_rate": 0.00019531665040921668,
      "loss": 1.812,
      "step": 3487
    },
    {
      "epoch": 0.1249127079341773,
      "grad_norm": 1.3807140588760376,
      "learning_rate": 0.00019531314169235259,
      "loss": 1.1548,
      "step": 3488
    },
    {
      "epoch": 0.12494852006374559,
      "grad_norm": 1.5563311576843262,
      "learning_rate": 0.00019530963169317312,
      "loss": 1.2558,
      "step": 3489
    },
    {
      "epoch": 0.12498433219331388,
      "grad_norm": 1.7699103355407715,
      "learning_rate": 0.0001953061204117255,
      "loss": 1.6063,
      "step": 3490
    },
    {
      "epoch": 0.12502014432288217,
      "grad_norm": 1.5972926616668701,
      "learning_rate": 0.00019530260784805697,
      "loss": 1.3708,
      "step": 3491
    },
    {
      "epoch": 0.12505595645245043,
      "grad_norm": 1.3855398893356323,
      "learning_rate": 0.00019529909400221475,
      "loss": 1.3582,
      "step": 3492
    },
    {
      "epoch": 0.12509176858201873,
      "grad_norm": 1.3573073148727417,
      "learning_rate": 0.00019529557887424618,
      "loss": 1.8298,
      "step": 3493
    },
    {
      "epoch": 0.12512758071158703,
      "grad_norm": 1.5848575830459595,
      "learning_rate": 0.00019529206246419854,
      "loss": 1.5766,
      "step": 3494
    },
    {
      "epoch": 0.1251633928411553,
      "grad_norm": 1.3935266733169556,
      "learning_rate": 0.00019528854477211908,
      "loss": 1.5095,
      "step": 3495
    },
    {
      "epoch": 0.1251992049707236,
      "grad_norm": 1.3671562671661377,
      "learning_rate": 0.0001952850257980552,
      "loss": 1.5969,
      "step": 3496
    },
    {
      "epoch": 0.12523501710029186,
      "grad_norm": 1.2846484184265137,
      "learning_rate": 0.00019528150554205419,
      "loss": 1.4534,
      "step": 3497
    },
    {
      "epoch": 0.12527082922986016,
      "grad_norm": 1.6687588691711426,
      "learning_rate": 0.00019527798400416338,
      "loss": 1.5611,
      "step": 3498
    },
    {
      "epoch": 0.12530664135942843,
      "grad_norm": 1.379241704940796,
      "learning_rate": 0.00019527446118443025,
      "loss": 1.4491,
      "step": 3499
    },
    {
      "epoch": 0.12534245348899672,
      "grad_norm": 1.4438045024871826,
      "learning_rate": 0.00019527093708290215,
      "loss": 1.6858,
      "step": 3500
    },
    {
      "epoch": 0.12537826561856502,
      "grad_norm": 1.3740583658218384,
      "learning_rate": 0.00019526741169962643,
      "loss": 1.8549,
      "step": 3501
    },
    {
      "epoch": 0.1254140777481333,
      "grad_norm": 2.178952932357788,
      "learning_rate": 0.00019526388503465062,
      "loss": 1.8759,
      "step": 3502
    },
    {
      "epoch": 0.1254498898777016,
      "grad_norm": 1.6737861633300781,
      "learning_rate": 0.00019526035708802207,
      "loss": 1.5266,
      "step": 3503
    },
    {
      "epoch": 0.12548570200726986,
      "grad_norm": 1.7721725702285767,
      "learning_rate": 0.00019525682785978833,
      "loss": 1.7037,
      "step": 3504
    },
    {
      "epoch": 0.12552151413683815,
      "grad_norm": 1.7052780389785767,
      "learning_rate": 0.00019525329734999683,
      "loss": 1.7986,
      "step": 3505
    },
    {
      "epoch": 0.12555732626640642,
      "grad_norm": 1.8097865581512451,
      "learning_rate": 0.0001952497655586951,
      "loss": 1.397,
      "step": 3506
    },
    {
      "epoch": 0.12559313839597472,
      "grad_norm": 1.6215720176696777,
      "learning_rate": 0.00019524623248593062,
      "loss": 1.4687,
      "step": 3507
    },
    {
      "epoch": 0.12562895052554302,
      "grad_norm": 1.4546492099761963,
      "learning_rate": 0.00019524269813175096,
      "loss": 1.3458,
      "step": 3508
    },
    {
      "epoch": 0.12566476265511128,
      "grad_norm": 2.1986570358276367,
      "learning_rate": 0.00019523916249620363,
      "loss": 1.5225,
      "step": 3509
    },
    {
      "epoch": 0.12570057478467958,
      "grad_norm": 3.118424892425537,
      "learning_rate": 0.0001952356255793362,
      "loss": 1.679,
      "step": 3510
    },
    {
      "epoch": 0.12573638691424785,
      "grad_norm": 1.657961368560791,
      "learning_rate": 0.00019523208738119632,
      "loss": 1.482,
      "step": 3511
    },
    {
      "epoch": 0.12577219904381615,
      "grad_norm": 1.8398610353469849,
      "learning_rate": 0.00019522854790183152,
      "loss": 1.649,
      "step": 3512
    },
    {
      "epoch": 0.12580801117338442,
      "grad_norm": 1.3558249473571777,
      "learning_rate": 0.00019522500714128942,
      "loss": 1.5596,
      "step": 3513
    },
    {
      "epoch": 0.1258438233029527,
      "grad_norm": 1.444551706314087,
      "learning_rate": 0.0001952214650996177,
      "loss": 1.3201,
      "step": 3514
    },
    {
      "epoch": 0.125879635432521,
      "grad_norm": 1.4728481769561768,
      "learning_rate": 0.000195217921776864,
      "loss": 1.6087,
      "step": 3515
    },
    {
      "epoch": 0.12591544756208928,
      "grad_norm": 1.3532110452651978,
      "learning_rate": 0.000195214377173076,
      "loss": 1.6132,
      "step": 3516
    },
    {
      "epoch": 0.12595125969165757,
      "grad_norm": 1.7189764976501465,
      "learning_rate": 0.00019521083128830137,
      "loss": 1.508,
      "step": 3517
    },
    {
      "epoch": 0.12598707182122584,
      "grad_norm": 1.3803995847702026,
      "learning_rate": 0.0001952072841225878,
      "loss": 1.7025,
      "step": 3518
    },
    {
      "epoch": 0.12602288395079414,
      "grad_norm": 1.9265297651290894,
      "learning_rate": 0.00019520373567598304,
      "loss": 1.4115,
      "step": 3519
    },
    {
      "epoch": 0.1260586960803624,
      "grad_norm": 2.2055609226226807,
      "learning_rate": 0.0001952001859485348,
      "loss": 1.8266,
      "step": 3520
    },
    {
      "epoch": 0.1260945082099307,
      "grad_norm": 1.7643824815750122,
      "learning_rate": 0.0001951966349402909,
      "loss": 1.5893,
      "step": 3521
    },
    {
      "epoch": 0.12613032033949897,
      "grad_norm": 2.026763677597046,
      "learning_rate": 0.00019519308265129903,
      "loss": 1.4778,
      "step": 3522
    },
    {
      "epoch": 0.12616613246906727,
      "grad_norm": 1.664231538772583,
      "learning_rate": 0.00019518952908160705,
      "loss": 1.6736,
      "step": 3523
    },
    {
      "epoch": 0.12620194459863557,
      "grad_norm": 1.765260934829712,
      "learning_rate": 0.00019518597423126273,
      "loss": 1.665,
      "step": 3524
    },
    {
      "epoch": 0.12623775672820384,
      "grad_norm": 2.3637137413024902,
      "learning_rate": 0.0001951824181003139,
      "loss": 1.3418,
      "step": 3525
    },
    {
      "epoch": 0.12627356885777213,
      "grad_norm": 1.4499986171722412,
      "learning_rate": 0.00019517886068880843,
      "loss": 1.6264,
      "step": 3526
    },
    {
      "epoch": 0.1263093809873404,
      "grad_norm": 1.7675611972808838,
      "learning_rate": 0.00019517530199679415,
      "loss": 1.6114,
      "step": 3527
    },
    {
      "epoch": 0.1263451931169087,
      "grad_norm": 1.6797280311584473,
      "learning_rate": 0.00019517174202431895,
      "loss": 1.4893,
      "step": 3528
    },
    {
      "epoch": 0.12638100524647697,
      "grad_norm": 1.2751450538635254,
      "learning_rate": 0.00019516818077143071,
      "loss": 1.6812,
      "step": 3529
    },
    {
      "epoch": 0.12641681737604527,
      "grad_norm": 2.196251630783081,
      "learning_rate": 0.00019516461823817737,
      "loss": 1.3997,
      "step": 3530
    },
    {
      "epoch": 0.12645262950561356,
      "grad_norm": 2.467259645462036,
      "learning_rate": 0.00019516105442460684,
      "loss": 1.4916,
      "step": 3531
    },
    {
      "epoch": 0.12648844163518183,
      "grad_norm": 1.8031963109970093,
      "learning_rate": 0.0001951574893307671,
      "loss": 1.8051,
      "step": 3532
    },
    {
      "epoch": 0.12652425376475013,
      "grad_norm": 1.3685870170593262,
      "learning_rate": 0.00019515392295670604,
      "loss": 1.4656,
      "step": 3533
    },
    {
      "epoch": 0.1265600658943184,
      "grad_norm": 1.3247365951538086,
      "learning_rate": 0.00019515035530247172,
      "loss": 1.5153,
      "step": 3534
    },
    {
      "epoch": 0.1265958780238867,
      "grad_norm": 1.341870665550232,
      "learning_rate": 0.0001951467863681121,
      "loss": 1.5531,
      "step": 3535
    },
    {
      "epoch": 0.12663169015345496,
      "grad_norm": 1.5447837114334106,
      "learning_rate": 0.00019514321615367517,
      "loss": 1.8168,
      "step": 3536
    },
    {
      "epoch": 0.12666750228302326,
      "grad_norm": 1.3823332786560059,
      "learning_rate": 0.000195139644659209,
      "loss": 1.6153,
      "step": 3537
    },
    {
      "epoch": 0.12670331441259156,
      "grad_norm": 1.7624273300170898,
      "learning_rate": 0.00019513607188476168,
      "loss": 1.5328,
      "step": 3538
    },
    {
      "epoch": 0.12673912654215982,
      "grad_norm": 1.4588311910629272,
      "learning_rate": 0.00019513249783038118,
      "loss": 1.7005,
      "step": 3539
    },
    {
      "epoch": 0.12677493867172812,
      "grad_norm": 2.036802291870117,
      "learning_rate": 0.00019512892249611566,
      "loss": 1.8805,
      "step": 3540
    },
    {
      "epoch": 0.1268107508012964,
      "grad_norm": 2.2579665184020996,
      "learning_rate": 0.00019512534588201318,
      "loss": 1.3699,
      "step": 3541
    },
    {
      "epoch": 0.1268465629308647,
      "grad_norm": 2.078040599822998,
      "learning_rate": 0.00019512176798812189,
      "loss": 1.6924,
      "step": 3542
    },
    {
      "epoch": 0.12688237506043296,
      "grad_norm": 1.5221152305603027,
      "learning_rate": 0.0001951181888144899,
      "loss": 1.8049,
      "step": 3543
    },
    {
      "epoch": 0.12691818719000125,
      "grad_norm": 1.3810714483261108,
      "learning_rate": 0.00019511460836116537,
      "loss": 1.6305,
      "step": 3544
    },
    {
      "epoch": 0.12695399931956955,
      "grad_norm": 1.5885570049285889,
      "learning_rate": 0.00019511102662819648,
      "loss": 1.6947,
      "step": 3545
    },
    {
      "epoch": 0.12698981144913782,
      "grad_norm": 2.6954009532928467,
      "learning_rate": 0.0001951074436156314,
      "loss": 1.5527,
      "step": 3546
    },
    {
      "epoch": 0.12702562357870611,
      "grad_norm": 2.2410197257995605,
      "learning_rate": 0.00019510385932351837,
      "loss": 1.5252,
      "step": 3547
    },
    {
      "epoch": 0.12706143570827438,
      "grad_norm": 1.5985091924667358,
      "learning_rate": 0.00019510027375190556,
      "loss": 1.6583,
      "step": 3548
    },
    {
      "epoch": 0.12709724783784268,
      "grad_norm": 1.7822718620300293,
      "learning_rate": 0.00019509668690084126,
      "loss": 1.5017,
      "step": 3549
    },
    {
      "epoch": 0.12713305996741095,
      "grad_norm": 1.6479328870773315,
      "learning_rate": 0.00019509309877037369,
      "loss": 1.4035,
      "step": 3550
    },
    {
      "epoch": 0.12716887209697925,
      "grad_norm": 2.219636917114258,
      "learning_rate": 0.00019508950936055115,
      "loss": 1.8554,
      "step": 3551
    },
    {
      "epoch": 0.12720468422654754,
      "grad_norm": 2.312058925628662,
      "learning_rate": 0.0001950859186714219,
      "loss": 1.6133,
      "step": 3552
    },
    {
      "epoch": 0.1272404963561158,
      "grad_norm": 1.3517247438430786,
      "learning_rate": 0.00019508232670303427,
      "loss": 1.5924,
      "step": 3553
    },
    {
      "epoch": 0.1272763084856841,
      "grad_norm": 1.835817813873291,
      "learning_rate": 0.00019507873345543658,
      "loss": 1.4991,
      "step": 3554
    },
    {
      "epoch": 0.12731212061525238,
      "grad_norm": 1.174762487411499,
      "learning_rate": 0.00019507513892867717,
      "loss": 1.5491,
      "step": 3555
    },
    {
      "epoch": 0.12734793274482067,
      "grad_norm": 1.6903138160705566,
      "learning_rate": 0.0001950715431228044,
      "loss": 1.526,
      "step": 3556
    },
    {
      "epoch": 0.12738374487438894,
      "grad_norm": 2.1673574447631836,
      "learning_rate": 0.0001950679460378667,
      "loss": 1.7603,
      "step": 3557
    },
    {
      "epoch": 0.12741955700395724,
      "grad_norm": 1.705032229423523,
      "learning_rate": 0.00019506434767391237,
      "loss": 1.7594,
      "step": 3558
    },
    {
      "epoch": 0.12745536913352554,
      "grad_norm": 1.64605712890625,
      "learning_rate": 0.00019506074803098987,
      "loss": 1.9528,
      "step": 3559
    },
    {
      "epoch": 0.1274911812630938,
      "grad_norm": 1.3241467475891113,
      "learning_rate": 0.00019505714710914764,
      "loss": 1.5991,
      "step": 3560
    },
    {
      "epoch": 0.1275269933926621,
      "grad_norm": 2.728872299194336,
      "learning_rate": 0.0001950535449084341,
      "loss": 1.6143,
      "step": 3561
    },
    {
      "epoch": 0.12756280552223037,
      "grad_norm": 1.7435359954833984,
      "learning_rate": 0.0001950499414288977,
      "loss": 1.4116,
      "step": 3562
    },
    {
      "epoch": 0.12759861765179867,
      "grad_norm": 1.7278475761413574,
      "learning_rate": 0.000195046336670587,
      "loss": 1.6202,
      "step": 3563
    },
    {
      "epoch": 0.12763442978136694,
      "grad_norm": 1.4856594800949097,
      "learning_rate": 0.0001950427306335504,
      "loss": 1.85,
      "step": 3564
    },
    {
      "epoch": 0.12767024191093523,
      "grad_norm": 1.2689180374145508,
      "learning_rate": 0.00019503912331783648,
      "loss": 1.5936,
      "step": 3565
    },
    {
      "epoch": 0.12770605404050353,
      "grad_norm": 2.304764747619629,
      "learning_rate": 0.00019503551472349373,
      "loss": 1.5894,
      "step": 3566
    },
    {
      "epoch": 0.1277418661700718,
      "grad_norm": 1.8257946968078613,
      "learning_rate": 0.0001950319048505707,
      "loss": 1.3694,
      "step": 3567
    },
    {
      "epoch": 0.1277776782996401,
      "grad_norm": 1.283836841583252,
      "learning_rate": 0.000195028293699116,
      "loss": 1.537,
      "step": 3568
    },
    {
      "epoch": 0.12781349042920837,
      "grad_norm": 2.9690396785736084,
      "learning_rate": 0.0001950246812691782,
      "loss": 1.5115,
      "step": 3569
    },
    {
      "epoch": 0.12784930255877666,
      "grad_norm": 2.2096030712127686,
      "learning_rate": 0.00019502106756080583,
      "loss": 1.8623,
      "step": 3570
    },
    {
      "epoch": 0.12788511468834493,
      "grad_norm": 1.6043503284454346,
      "learning_rate": 0.00019501745257404762,
      "loss": 1.5075,
      "step": 3571
    },
    {
      "epoch": 0.12792092681791323,
      "grad_norm": 1.8550293445587158,
      "learning_rate": 0.00019501383630895211,
      "loss": 1.6319,
      "step": 3572
    },
    {
      "epoch": 0.12795673894748152,
      "grad_norm": 1.1330084800720215,
      "learning_rate": 0.00019501021876556802,
      "loss": 1.7045,
      "step": 3573
    },
    {
      "epoch": 0.1279925510770498,
      "grad_norm": 3.4932503700256348,
      "learning_rate": 0.00019500659994394398,
      "loss": 1.5967,
      "step": 3574
    },
    {
      "epoch": 0.1280283632066181,
      "grad_norm": 2.301506280899048,
      "learning_rate": 0.0001950029798441287,
      "loss": 1.7702,
      "step": 3575
    },
    {
      "epoch": 0.12806417533618636,
      "grad_norm": 1.4132013320922852,
      "learning_rate": 0.00019499935846617084,
      "loss": 1.8051,
      "step": 3576
    },
    {
      "epoch": 0.12809998746575466,
      "grad_norm": 1.813435435295105,
      "learning_rate": 0.0001949957358101192,
      "loss": 1.5849,
      "step": 3577
    },
    {
      "epoch": 0.12813579959532292,
      "grad_norm": 1.2465287446975708,
      "learning_rate": 0.00019499211187602242,
      "loss": 1.5895,
      "step": 3578
    },
    {
      "epoch": 0.12817161172489122,
      "grad_norm": 1.4146263599395752,
      "learning_rate": 0.0001949884866639293,
      "loss": 1.9067,
      "step": 3579
    },
    {
      "epoch": 0.12820742385445952,
      "grad_norm": 1.7771110534667969,
      "learning_rate": 0.00019498486017388865,
      "loss": 1.59,
      "step": 3580
    },
    {
      "epoch": 0.1282432359840278,
      "grad_norm": 2.2220492362976074,
      "learning_rate": 0.00019498123240594924,
      "loss": 1.4147,
      "step": 3581
    },
    {
      "epoch": 0.12827904811359608,
      "grad_norm": 2.194570779800415,
      "learning_rate": 0.00019497760336015984,
      "loss": 1.7702,
      "step": 3582
    },
    {
      "epoch": 0.12831486024316435,
      "grad_norm": 1.2068736553192139,
      "learning_rate": 0.0001949739730365693,
      "loss": 1.6503,
      "step": 3583
    },
    {
      "epoch": 0.12835067237273265,
      "grad_norm": 1.3843286037445068,
      "learning_rate": 0.0001949703414352265,
      "loss": 1.6336,
      "step": 3584
    },
    {
      "epoch": 0.12838648450230092,
      "grad_norm": 1.7587592601776123,
      "learning_rate": 0.0001949667085561802,
      "loss": 1.6254,
      "step": 3585
    },
    {
      "epoch": 0.12842229663186921,
      "grad_norm": 2.3941118717193604,
      "learning_rate": 0.00019496307439947937,
      "loss": 1.7615,
      "step": 3586
    },
    {
      "epoch": 0.1284581087614375,
      "grad_norm": 1.4519942998886108,
      "learning_rate": 0.00019495943896517286,
      "loss": 1.3183,
      "step": 3587
    },
    {
      "epoch": 0.12849392089100578,
      "grad_norm": 1.6823445558547974,
      "learning_rate": 0.0001949558022533096,
      "loss": 1.4198,
      "step": 3588
    },
    {
      "epoch": 0.12852973302057408,
      "grad_norm": 1.5577532052993774,
      "learning_rate": 0.00019495216426393847,
      "loss": 1.6609,
      "step": 3589
    },
    {
      "epoch": 0.12856554515014235,
      "grad_norm": 1.4570772647857666,
      "learning_rate": 0.0001949485249971085,
      "loss": 1.5659,
      "step": 3590
    },
    {
      "epoch": 0.12860135727971064,
      "grad_norm": 1.4191895723342896,
      "learning_rate": 0.00019494488445286856,
      "loss": 1.4708,
      "step": 3591
    },
    {
      "epoch": 0.1286371694092789,
      "grad_norm": 1.8212336301803589,
      "learning_rate": 0.00019494124263126766,
      "loss": 1.5248,
      "step": 3592
    },
    {
      "epoch": 0.1286729815388472,
      "grad_norm": 1.7171404361724854,
      "learning_rate": 0.00019493759953235484,
      "loss": 1.7069,
      "step": 3593
    },
    {
      "epoch": 0.1287087936684155,
      "grad_norm": 1.546820878982544,
      "learning_rate": 0.00019493395515617908,
      "loss": 1.3911,
      "step": 3594
    },
    {
      "epoch": 0.12874460579798377,
      "grad_norm": 1.5426114797592163,
      "learning_rate": 0.00019493030950278937,
      "loss": 1.7267,
      "step": 3595
    },
    {
      "epoch": 0.12878041792755207,
      "grad_norm": 1.4243674278259277,
      "learning_rate": 0.00019492666257223484,
      "loss": 1.8226,
      "step": 3596
    },
    {
      "epoch": 0.12881623005712034,
      "grad_norm": 1.364989161491394,
      "learning_rate": 0.00019492301436456447,
      "loss": 1.6399,
      "step": 3597
    },
    {
      "epoch": 0.12885204218668864,
      "grad_norm": 2.2802083492279053,
      "learning_rate": 0.00019491936487982744,
      "loss": 1.6192,
      "step": 3598
    },
    {
      "epoch": 0.1288878543162569,
      "grad_norm": 1.3957548141479492,
      "learning_rate": 0.00019491571411807274,
      "loss": 1.6369,
      "step": 3599
    },
    {
      "epoch": 0.1289236664458252,
      "grad_norm": 1.914682149887085,
      "learning_rate": 0.00019491206207934955,
      "loss": 1.7215,
      "step": 3600
    },
    {
      "epoch": 0.1289594785753935,
      "grad_norm": 2.625183582305908,
      "learning_rate": 0.00019490840876370703,
      "loss": 1.367,
      "step": 3601
    },
    {
      "epoch": 0.12899529070496177,
      "grad_norm": 1.3641928434371948,
      "learning_rate": 0.00019490475417119425,
      "loss": 1.7807,
      "step": 3602
    },
    {
      "epoch": 0.12903110283453006,
      "grad_norm": 1.6439179182052612,
      "learning_rate": 0.00019490109830186042,
      "loss": 1.4354,
      "step": 3603
    },
    {
      "epoch": 0.12906691496409833,
      "grad_norm": 1.4001127481460571,
      "learning_rate": 0.00019489744115575475,
      "loss": 1.5427,
      "step": 3604
    },
    {
      "epoch": 0.12910272709366663,
      "grad_norm": 1.2241640090942383,
      "learning_rate": 0.00019489378273292643,
      "loss": 1.4639,
      "step": 3605
    },
    {
      "epoch": 0.1291385392232349,
      "grad_norm": 1.4079821109771729,
      "learning_rate": 0.00019489012303342462,
      "loss": 1.6541,
      "step": 3606
    },
    {
      "epoch": 0.1291743513528032,
      "grad_norm": 2.0953893661499023,
      "learning_rate": 0.00019488646205729864,
      "loss": 1.5438,
      "step": 3607
    },
    {
      "epoch": 0.1292101634823715,
      "grad_norm": 2.2923357486724854,
      "learning_rate": 0.00019488279980459772,
      "loss": 1.6912,
      "step": 3608
    },
    {
      "epoch": 0.12924597561193976,
      "grad_norm": 1.4191970825195312,
      "learning_rate": 0.00019487913627537108,
      "loss": 1.8267,
      "step": 3609
    },
    {
      "epoch": 0.12928178774150806,
      "grad_norm": 2.6938745975494385,
      "learning_rate": 0.00019487547146966808,
      "loss": 1.7512,
      "step": 3610
    },
    {
      "epoch": 0.12931759987107633,
      "grad_norm": 2.0451958179473877,
      "learning_rate": 0.00019487180538753796,
      "loss": 1.7794,
      "step": 3611
    },
    {
      "epoch": 0.12935341200064462,
      "grad_norm": 1.861045241355896,
      "learning_rate": 0.0001948681380290301,
      "loss": 1.8672,
      "step": 3612
    },
    {
      "epoch": 0.1293892241302129,
      "grad_norm": 1.9906060695648193,
      "learning_rate": 0.0001948644693941938,
      "loss": 1.4666,
      "step": 3613
    },
    {
      "epoch": 0.1294250362597812,
      "grad_norm": 2.1408982276916504,
      "learning_rate": 0.00019486079948307844,
      "loss": 1.5275,
      "step": 3614
    },
    {
      "epoch": 0.12946084838934946,
      "grad_norm": 1.689831018447876,
      "learning_rate": 0.00019485712829573338,
      "loss": 1.6761,
      "step": 3615
    },
    {
      "epoch": 0.12949666051891776,
      "grad_norm": 1.6339595317840576,
      "learning_rate": 0.000194853455832208,
      "loss": 1.477,
      "step": 3616
    },
    {
      "epoch": 0.12953247264848605,
      "grad_norm": 1.2168866395950317,
      "learning_rate": 0.00019484978209255175,
      "loss": 1.1758,
      "step": 3617
    },
    {
      "epoch": 0.12956828477805432,
      "grad_norm": 3.002838134765625,
      "learning_rate": 0.00019484610707681403,
      "loss": 1.5886,
      "step": 3618
    },
    {
      "epoch": 0.12960409690762262,
      "grad_norm": 1.7156847715377808,
      "learning_rate": 0.00019484243078504428,
      "loss": 1.5088,
      "step": 3619
    },
    {
      "epoch": 0.1296399090371909,
      "grad_norm": 1.3578027486801147,
      "learning_rate": 0.00019483875321729194,
      "loss": 1.5774,
      "step": 3620
    },
    {
      "epoch": 0.12967572116675918,
      "grad_norm": 1.3243157863616943,
      "learning_rate": 0.00019483507437360653,
      "loss": 1.7064,
      "step": 3621
    },
    {
      "epoch": 0.12971153329632745,
      "grad_norm": 2.185899257659912,
      "learning_rate": 0.0001948313942540375,
      "loss": 1.7627,
      "step": 3622
    },
    {
      "epoch": 0.12974734542589575,
      "grad_norm": 1.9846500158309937,
      "learning_rate": 0.00019482771285863438,
      "loss": 1.5378,
      "step": 3623
    },
    {
      "epoch": 0.12978315755546405,
      "grad_norm": 2.1353330612182617,
      "learning_rate": 0.00019482403018744674,
      "loss": 1.6318,
      "step": 3624
    },
    {
      "epoch": 0.12981896968503231,
      "grad_norm": 1.6027969121932983,
      "learning_rate": 0.00019482034624052408,
      "loss": 1.935,
      "step": 3625
    },
    {
      "epoch": 0.1298547818146006,
      "grad_norm": 1.7356218099594116,
      "learning_rate": 0.00019481666101791594,
      "loss": 1.593,
      "step": 3626
    },
    {
      "epoch": 0.12989059394416888,
      "grad_norm": 1.551705002784729,
      "learning_rate": 0.00019481297451967195,
      "loss": 1.7691,
      "step": 3627
    },
    {
      "epoch": 0.12992640607373718,
      "grad_norm": 2.0163631439208984,
      "learning_rate": 0.0001948092867458417,
      "loss": 1.4032,
      "step": 3628
    },
    {
      "epoch": 0.12996221820330545,
      "grad_norm": 1.6839312314987183,
      "learning_rate": 0.00019480559769647477,
      "loss": 1.8452,
      "step": 3629
    },
    {
      "epoch": 0.12999803033287374,
      "grad_norm": 1.5859227180480957,
      "learning_rate": 0.00019480190737162083,
      "loss": 1.5909,
      "step": 3630
    },
    {
      "epoch": 0.13003384246244204,
      "grad_norm": 1.7990963459014893,
      "learning_rate": 0.0001947982157713295,
      "loss": 1.5817,
      "step": 3631
    },
    {
      "epoch": 0.1300696545920103,
      "grad_norm": 1.4903086423873901,
      "learning_rate": 0.00019479452289565048,
      "loss": 1.8922,
      "step": 3632
    },
    {
      "epoch": 0.1301054667215786,
      "grad_norm": 1.210903525352478,
      "learning_rate": 0.00019479082874463338,
      "loss": 1.5365,
      "step": 3633
    },
    {
      "epoch": 0.13014127885114687,
      "grad_norm": 1.3731626272201538,
      "learning_rate": 0.000194787133318328,
      "loss": 1.7997,
      "step": 3634
    },
    {
      "epoch": 0.13017709098071517,
      "grad_norm": 1.7975584268569946,
      "learning_rate": 0.000194783436616784,
      "loss": 1.4367,
      "step": 3635
    },
    {
      "epoch": 0.13021290311028344,
      "grad_norm": 1.2568280696868896,
      "learning_rate": 0.00019477973864005113,
      "loss": 1.8788,
      "step": 3636
    },
    {
      "epoch": 0.13024871523985174,
      "grad_norm": 1.7492560148239136,
      "learning_rate": 0.0001947760393881791,
      "loss": 1.3877,
      "step": 3637
    },
    {
      "epoch": 0.13028452736942003,
      "grad_norm": 2.4165797233581543,
      "learning_rate": 0.00019477233886121772,
      "loss": 1.7056,
      "step": 3638
    },
    {
      "epoch": 0.1303203394989883,
      "grad_norm": 1.5210332870483398,
      "learning_rate": 0.00019476863705921677,
      "loss": 1.8208,
      "step": 3639
    },
    {
      "epoch": 0.1303561516285566,
      "grad_norm": 1.419649600982666,
      "learning_rate": 0.00019476493398222608,
      "loss": 1.4835,
      "step": 3640
    },
    {
      "epoch": 0.13039196375812487,
      "grad_norm": 2.1755502223968506,
      "learning_rate": 0.0001947612296302954,
      "loss": 1.441,
      "step": 3641
    },
    {
      "epoch": 0.13042777588769316,
      "grad_norm": 1.332960844039917,
      "learning_rate": 0.00019475752400347464,
      "loss": 1.6522,
      "step": 3642
    },
    {
      "epoch": 0.13046358801726143,
      "grad_norm": 1.6385389566421509,
      "learning_rate": 0.00019475381710181363,
      "loss": 1.3294,
      "step": 3643
    },
    {
      "epoch": 0.13049940014682973,
      "grad_norm": 2.0935511589050293,
      "learning_rate": 0.0001947501089253622,
      "loss": 1.9683,
      "step": 3644
    },
    {
      "epoch": 0.13053521227639803,
      "grad_norm": 1.5718079805374146,
      "learning_rate": 0.00019474639947417028,
      "loss": 1.4642,
      "step": 3645
    },
    {
      "epoch": 0.1305710244059663,
      "grad_norm": 1.8546572923660278,
      "learning_rate": 0.0001947426887482878,
      "loss": 1.6069,
      "step": 3646
    },
    {
      "epoch": 0.1306068365355346,
      "grad_norm": 1.3184876441955566,
      "learning_rate": 0.0001947389767477646,
      "loss": 1.666,
      "step": 3647
    },
    {
      "epoch": 0.13064264866510286,
      "grad_norm": 1.41603422164917,
      "learning_rate": 0.00019473526347265073,
      "loss": 1.588,
      "step": 3648
    },
    {
      "epoch": 0.13067846079467116,
      "grad_norm": 1.5340843200683594,
      "learning_rate": 0.00019473154892299608,
      "loss": 1.848,
      "step": 3649
    },
    {
      "epoch": 0.13071427292423943,
      "grad_norm": 2.2189242839813232,
      "learning_rate": 0.00019472783309885057,
      "loss": 1.9326,
      "step": 3650
    },
    {
      "epoch": 0.13075008505380772,
      "grad_norm": 2.0368833541870117,
      "learning_rate": 0.0001947241160002643,
      "loss": 1.7715,
      "step": 3651
    },
    {
      "epoch": 0.13078589718337602,
      "grad_norm": 1.550632357597351,
      "learning_rate": 0.00019472039762728728,
      "loss": 1.5666,
      "step": 3652
    },
    {
      "epoch": 0.1308217093129443,
      "grad_norm": 1.8009685277938843,
      "learning_rate": 0.00019471667797996944,
      "loss": 1.5932,
      "step": 3653
    },
    {
      "epoch": 0.1308575214425126,
      "grad_norm": 1.6331686973571777,
      "learning_rate": 0.00019471295705836088,
      "loss": 1.4965,
      "step": 3654
    },
    {
      "epoch": 0.13089333357208086,
      "grad_norm": 1.6497012376785278,
      "learning_rate": 0.00019470923486251165,
      "loss": 1.638,
      "step": 3655
    },
    {
      "epoch": 0.13092914570164915,
      "grad_norm": 1.526160717010498,
      "learning_rate": 0.00019470551139247184,
      "loss": 1.7185,
      "step": 3656
    },
    {
      "epoch": 0.13096495783121742,
      "grad_norm": 1.3786871433258057,
      "learning_rate": 0.00019470178664829154,
      "loss": 1.6264,
      "step": 3657
    },
    {
      "epoch": 0.13100076996078572,
      "grad_norm": 1.2619601488113403,
      "learning_rate": 0.00019469806063002082,
      "loss": 1.5839,
      "step": 3658
    },
    {
      "epoch": 0.13103658209035401,
      "grad_norm": 2.778254270553589,
      "learning_rate": 0.0001946943333377099,
      "loss": 1.5246,
      "step": 3659
    },
    {
      "epoch": 0.13107239421992228,
      "grad_norm": 1.5448185205459595,
      "learning_rate": 0.00019469060477140886,
      "loss": 1.4715,
      "step": 3660
    },
    {
      "epoch": 0.13110820634949058,
      "grad_norm": 1.5126322507858276,
      "learning_rate": 0.00019468687493116784,
      "loss": 1.6236,
      "step": 3661
    },
    {
      "epoch": 0.13114401847905885,
      "grad_norm": 2.1006526947021484,
      "learning_rate": 0.00019468314381703708,
      "loss": 1.8905,
      "step": 3662
    },
    {
      "epoch": 0.13117983060862715,
      "grad_norm": 1.9070483446121216,
      "learning_rate": 0.00019467941142906674,
      "loss": 1.6353,
      "step": 3663
    },
    {
      "epoch": 0.13121564273819541,
      "grad_norm": 1.40544855594635,
      "learning_rate": 0.00019467567776730707,
      "loss": 1.5716,
      "step": 3664
    },
    {
      "epoch": 0.1312514548677637,
      "grad_norm": 1.3862591981887817,
      "learning_rate": 0.00019467194283180828,
      "loss": 1.5309,
      "step": 3665
    },
    {
      "epoch": 0.131287266997332,
      "grad_norm": 1.2076510190963745,
      "learning_rate": 0.0001946682066226206,
      "loss": 1.585,
      "step": 3666
    },
    {
      "epoch": 0.13132307912690028,
      "grad_norm": 2.244685411453247,
      "learning_rate": 0.0001946644691397943,
      "loss": 1.6969,
      "step": 3667
    },
    {
      "epoch": 0.13135889125646857,
      "grad_norm": 1.1528452634811401,
      "learning_rate": 0.00019466073038337968,
      "loss": 1.2363,
      "step": 3668
    },
    {
      "epoch": 0.13139470338603684,
      "grad_norm": 1.7110793590545654,
      "learning_rate": 0.00019465699035342706,
      "loss": 1.6413,
      "step": 3669
    },
    {
      "epoch": 0.13143051551560514,
      "grad_norm": 1.4075368642807007,
      "learning_rate": 0.00019465324904998672,
      "loss": 1.6554,
      "step": 3670
    },
    {
      "epoch": 0.1314663276451734,
      "grad_norm": 1.4957016706466675,
      "learning_rate": 0.000194649506473109,
      "loss": 1.6191,
      "step": 3671
    },
    {
      "epoch": 0.1315021397747417,
      "grad_norm": 1.4293609857559204,
      "learning_rate": 0.00019464576262284426,
      "loss": 1.3673,
      "step": 3672
    },
    {
      "epoch": 0.13153795190431,
      "grad_norm": 1.4260005950927734,
      "learning_rate": 0.00019464201749924288,
      "loss": 1.7648,
      "step": 3673
    },
    {
      "epoch": 0.13157376403387827,
      "grad_norm": 1.628811240196228,
      "learning_rate": 0.00019463827110235523,
      "loss": 1.5441,
      "step": 3674
    },
    {
      "epoch": 0.13160957616344657,
      "grad_norm": 2.7041733264923096,
      "learning_rate": 0.00019463452343223173,
      "loss": 1.7595,
      "step": 3675
    },
    {
      "epoch": 0.13164538829301484,
      "grad_norm": 4.314607620239258,
      "learning_rate": 0.00019463077448892278,
      "loss": 1.4025,
      "step": 3676
    },
    {
      "epoch": 0.13168120042258313,
      "grad_norm": 1.7181106805801392,
      "learning_rate": 0.0001946270242724788,
      "loss": 1.495,
      "step": 3677
    },
    {
      "epoch": 0.1317170125521514,
      "grad_norm": 1.686544418334961,
      "learning_rate": 0.0001946232727829503,
      "loss": 1.727,
      "step": 3678
    },
    {
      "epoch": 0.1317528246817197,
      "grad_norm": 1.6624183654785156,
      "learning_rate": 0.00019461952002038771,
      "loss": 1.7125,
      "step": 3679
    },
    {
      "epoch": 0.131788636811288,
      "grad_norm": 2.172788381576538,
      "learning_rate": 0.0001946157659848415,
      "loss": 2.2068,
      "step": 3680
    },
    {
      "epoch": 0.13182444894085626,
      "grad_norm": 1.7004787921905518,
      "learning_rate": 0.00019461201067636226,
      "loss": 1.8583,
      "step": 3681
    },
    {
      "epoch": 0.13186026107042456,
      "grad_norm": 1.7210794687271118,
      "learning_rate": 0.00019460825409500042,
      "loss": 1.7554,
      "step": 3682
    },
    {
      "epoch": 0.13189607319999283,
      "grad_norm": 1.6098159551620483,
      "learning_rate": 0.00019460449624080655,
      "loss": 1.6517,
      "step": 3683
    },
    {
      "epoch": 0.13193188532956113,
      "grad_norm": 2.078428030014038,
      "learning_rate": 0.00019460073711383125,
      "loss": 1.4448,
      "step": 3684
    },
    {
      "epoch": 0.1319676974591294,
      "grad_norm": 1.5644986629486084,
      "learning_rate": 0.00019459697671412503,
      "loss": 1.7364,
      "step": 3685
    },
    {
      "epoch": 0.1320035095886977,
      "grad_norm": 1.9476972818374634,
      "learning_rate": 0.0001945932150417385,
      "loss": 1.4418,
      "step": 3686
    },
    {
      "epoch": 0.132039321718266,
      "grad_norm": 2.0070180892944336,
      "learning_rate": 0.0001945894520967223,
      "loss": 1.8668,
      "step": 3687
    },
    {
      "epoch": 0.13207513384783426,
      "grad_norm": 2.056800127029419,
      "learning_rate": 0.00019458568787912703,
      "loss": 1.6064,
      "step": 3688
    },
    {
      "epoch": 0.13211094597740256,
      "grad_norm": 1.6706658601760864,
      "learning_rate": 0.00019458192238900335,
      "loss": 1.5147,
      "step": 3689
    },
    {
      "epoch": 0.13214675810697082,
      "grad_norm": 1.309191346168518,
      "learning_rate": 0.00019457815562640187,
      "loss": 1.1176,
      "step": 3690
    },
    {
      "epoch": 0.13218257023653912,
      "grad_norm": 1.6902896165847778,
      "learning_rate": 0.00019457438759137334,
      "loss": 1.4857,
      "step": 3691
    },
    {
      "epoch": 0.1322183823661074,
      "grad_norm": 1.9160804748535156,
      "learning_rate": 0.00019457061828396838,
      "loss": 1.4037,
      "step": 3692
    },
    {
      "epoch": 0.1322541944956757,
      "grad_norm": 1.564273476600647,
      "learning_rate": 0.00019456684770423777,
      "loss": 1.6198,
      "step": 3693
    },
    {
      "epoch": 0.13229000662524398,
      "grad_norm": 1.7293407917022705,
      "learning_rate": 0.00019456307585223218,
      "loss": 1.5622,
      "step": 3694
    },
    {
      "epoch": 0.13232581875481225,
      "grad_norm": 1.3648868799209595,
      "learning_rate": 0.00019455930272800243,
      "loss": 1.507,
      "step": 3695
    },
    {
      "epoch": 0.13236163088438055,
      "grad_norm": 1.312423825263977,
      "learning_rate": 0.00019455552833159918,
      "loss": 1.7455,
      "step": 3696
    },
    {
      "epoch": 0.13239744301394882,
      "grad_norm": 1.8402488231658936,
      "learning_rate": 0.00019455175266307328,
      "loss": 1.6363,
      "step": 3697
    },
    {
      "epoch": 0.13243325514351711,
      "grad_norm": 1.179580807685852,
      "learning_rate": 0.00019454797572247552,
      "loss": 1.6652,
      "step": 3698
    },
    {
      "epoch": 0.13246906727308538,
      "grad_norm": 1.4366496801376343,
      "learning_rate": 0.0001945441975098567,
      "loss": 1.7087,
      "step": 3699
    },
    {
      "epoch": 0.13250487940265368,
      "grad_norm": 1.816963791847229,
      "learning_rate": 0.00019454041802526766,
      "loss": 1.7501,
      "step": 3700
    },
    {
      "epoch": 0.13254069153222198,
      "grad_norm": 1.6543554067611694,
      "learning_rate": 0.00019453663726875923,
      "loss": 1.6763,
      "step": 3701
    },
    {
      "epoch": 0.13257650366179025,
      "grad_norm": 1.448135256767273,
      "learning_rate": 0.0001945328552403823,
      "loss": 1.6681,
      "step": 3702
    },
    {
      "epoch": 0.13261231579135854,
      "grad_norm": 1.8160815238952637,
      "learning_rate": 0.00019452907194018776,
      "loss": 1.5419,
      "step": 3703
    },
    {
      "epoch": 0.1326481279209268,
      "grad_norm": 1.41280198097229,
      "learning_rate": 0.00019452528736822646,
      "loss": 1.7881,
      "step": 3704
    },
    {
      "epoch": 0.1326839400504951,
      "grad_norm": 1.6581374406814575,
      "learning_rate": 0.00019452150152454936,
      "loss": 2.032,
      "step": 3705
    },
    {
      "epoch": 0.13271975218006338,
      "grad_norm": 1.9039281606674194,
      "learning_rate": 0.0001945177144092074,
      "loss": 1.3711,
      "step": 3706
    },
    {
      "epoch": 0.13275556430963167,
      "grad_norm": 2.1422500610351562,
      "learning_rate": 0.0001945139260222515,
      "loss": 1.5075,
      "step": 3707
    },
    {
      "epoch": 0.13279137643919997,
      "grad_norm": 2.3555521965026855,
      "learning_rate": 0.00019451013636373262,
      "loss": 1.6693,
      "step": 3708
    },
    {
      "epoch": 0.13282718856876824,
      "grad_norm": 1.4542937278747559,
      "learning_rate": 0.00019450634543370177,
      "loss": 1.6695,
      "step": 3709
    },
    {
      "epoch": 0.13286300069833654,
      "grad_norm": 1.687001347541809,
      "learning_rate": 0.00019450255323220995,
      "loss": 1.5687,
      "step": 3710
    },
    {
      "epoch": 0.1328988128279048,
      "grad_norm": 1.497179388999939,
      "learning_rate": 0.00019449875975930818,
      "loss": 2.1073,
      "step": 3711
    },
    {
      "epoch": 0.1329346249574731,
      "grad_norm": 2.0575437545776367,
      "learning_rate": 0.00019449496501504747,
      "loss": 1.7413,
      "step": 3712
    },
    {
      "epoch": 0.13297043708704137,
      "grad_norm": 1.6596516370773315,
      "learning_rate": 0.0001944911689994789,
      "loss": 1.4723,
      "step": 3713
    },
    {
      "epoch": 0.13300624921660967,
      "grad_norm": 1.3024641275405884,
      "learning_rate": 0.0001944873717126536,
      "loss": 1.4606,
      "step": 3714
    },
    {
      "epoch": 0.13304206134617794,
      "grad_norm": 1.9574958086013794,
      "learning_rate": 0.00019448357315462255,
      "loss": 1.7971,
      "step": 3715
    },
    {
      "epoch": 0.13307787347574623,
      "grad_norm": 1.5044560432434082,
      "learning_rate": 0.00019447977332543687,
      "loss": 1.4529,
      "step": 3716
    },
    {
      "epoch": 0.13311368560531453,
      "grad_norm": 2.2105259895324707,
      "learning_rate": 0.00019447597222514772,
      "loss": 1.9443,
      "step": 3717
    },
    {
      "epoch": 0.1331494977348828,
      "grad_norm": 2.5844597816467285,
      "learning_rate": 0.00019447216985380626,
      "loss": 1.3201,
      "step": 3718
    },
    {
      "epoch": 0.1331853098644511,
      "grad_norm": 1.6001721620559692,
      "learning_rate": 0.0001944683662114636,
      "loss": 1.6278,
      "step": 3719
    },
    {
      "epoch": 0.13322112199401936,
      "grad_norm": 1.7248762845993042,
      "learning_rate": 0.00019446456129817093,
      "loss": 1.6102,
      "step": 3720
    },
    {
      "epoch": 0.13325693412358766,
      "grad_norm": 1.187751054763794,
      "learning_rate": 0.00019446075511397943,
      "loss": 1.4295,
      "step": 3721
    },
    {
      "epoch": 0.13329274625315593,
      "grad_norm": 2.0701000690460205,
      "learning_rate": 0.0001944569476589403,
      "loss": 1.7675,
      "step": 3722
    },
    {
      "epoch": 0.13332855838272423,
      "grad_norm": 1.876980185508728,
      "learning_rate": 0.00019445313893310482,
      "loss": 1.7526,
      "step": 3723
    },
    {
      "epoch": 0.13336437051229252,
      "grad_norm": 2.1986751556396484,
      "learning_rate": 0.00019444932893652417,
      "loss": 1.3216,
      "step": 3724
    },
    {
      "epoch": 0.1334001826418608,
      "grad_norm": 2.8376963138580322,
      "learning_rate": 0.00019444551766924963,
      "loss": 1.9192,
      "step": 3725
    },
    {
      "epoch": 0.1334359947714291,
      "grad_norm": 1.2646297216415405,
      "learning_rate": 0.00019444170513133248,
      "loss": 1.6767,
      "step": 3726
    },
    {
      "epoch": 0.13347180690099736,
      "grad_norm": 1.789505124092102,
      "learning_rate": 0.00019443789132282403,
      "loss": 1.7196,
      "step": 3727
    },
    {
      "epoch": 0.13350761903056566,
      "grad_norm": 1.2923089265823364,
      "learning_rate": 0.0001944340762437755,
      "loss": 1.2456,
      "step": 3728
    },
    {
      "epoch": 0.13354343116013392,
      "grad_norm": 2.154646635055542,
      "learning_rate": 0.00019443025989423834,
      "loss": 1.5132,
      "step": 3729
    },
    {
      "epoch": 0.13357924328970222,
      "grad_norm": 1.5187957286834717,
      "learning_rate": 0.00019442644227426383,
      "loss": 1.5191,
      "step": 3730
    },
    {
      "epoch": 0.13361505541927052,
      "grad_norm": 1.3650364875793457,
      "learning_rate": 0.00019442262338390337,
      "loss": 1.3481,
      "step": 3731
    },
    {
      "epoch": 0.1336508675488388,
      "grad_norm": 2.002079963684082,
      "learning_rate": 0.00019441880322320824,
      "loss": 1.5562,
      "step": 3732
    },
    {
      "epoch": 0.13368667967840708,
      "grad_norm": 1.4634652137756348,
      "learning_rate": 0.00019441498179222997,
      "loss": 1.8203,
      "step": 3733
    },
    {
      "epoch": 0.13372249180797535,
      "grad_norm": 2.2729015350341797,
      "learning_rate": 0.00019441115909101986,
      "loss": 1.4976,
      "step": 3734
    },
    {
      "epoch": 0.13375830393754365,
      "grad_norm": 1.93559992313385,
      "learning_rate": 0.0001944073351196294,
      "loss": 1.4768,
      "step": 3735
    },
    {
      "epoch": 0.13379411606711192,
      "grad_norm": 1.4857501983642578,
      "learning_rate": 0.00019440350987811003,
      "loss": 1.5598,
      "step": 3736
    },
    {
      "epoch": 0.13382992819668021,
      "grad_norm": 1.494903564453125,
      "learning_rate": 0.0001943996833665132,
      "loss": 1.5704,
      "step": 3737
    },
    {
      "epoch": 0.1338657403262485,
      "grad_norm": 1.9642722606658936,
      "learning_rate": 0.0001943958555848904,
      "loss": 1.855,
      "step": 3738
    },
    {
      "epoch": 0.13390155245581678,
      "grad_norm": 2.271066904067993,
      "learning_rate": 0.00019439202653329313,
      "loss": 1.9326,
      "step": 3739
    },
    {
      "epoch": 0.13393736458538508,
      "grad_norm": 1.980333924293518,
      "learning_rate": 0.00019438819621177289,
      "loss": 1.9477,
      "step": 3740
    },
    {
      "epoch": 0.13397317671495335,
      "grad_norm": 1.602170467376709,
      "learning_rate": 0.00019438436462038125,
      "loss": 1.434,
      "step": 3741
    },
    {
      "epoch": 0.13400898884452164,
      "grad_norm": 2.0307059288024902,
      "learning_rate": 0.00019438053175916968,
      "loss": 1.5559,
      "step": 3742
    },
    {
      "epoch": 0.1340448009740899,
      "grad_norm": 2.126970052719116,
      "learning_rate": 0.00019437669762818985,
      "loss": 1.8426,
      "step": 3743
    },
    {
      "epoch": 0.1340806131036582,
      "grad_norm": 2.3032383918762207,
      "learning_rate": 0.00019437286222749326,
      "loss": 1.7689,
      "step": 3744
    },
    {
      "epoch": 0.1341164252332265,
      "grad_norm": 1.8578029870986938,
      "learning_rate": 0.00019436902555713153,
      "loss": 1.8085,
      "step": 3745
    },
    {
      "epoch": 0.13415223736279477,
      "grad_norm": 1.2118780612945557,
      "learning_rate": 0.00019436518761715632,
      "loss": 1.6056,
      "step": 3746
    },
    {
      "epoch": 0.13418804949236307,
      "grad_norm": 2.220696210861206,
      "learning_rate": 0.0001943613484076192,
      "loss": 1.4916,
      "step": 3747
    },
    {
      "epoch": 0.13422386162193134,
      "grad_norm": 1.5976279973983765,
      "learning_rate": 0.0001943575079285719,
      "loss": 1.6949,
      "step": 3748
    },
    {
      "epoch": 0.13425967375149964,
      "grad_norm": 1.7551590204238892,
      "learning_rate": 0.000194353666180066,
      "loss": 1.5253,
      "step": 3749
    },
    {
      "epoch": 0.1342954858810679,
      "grad_norm": 1.204624891281128,
      "learning_rate": 0.00019434982316215326,
      "loss": 1.5708,
      "step": 3750
    },
    {
      "epoch": 0.1343312980106362,
      "grad_norm": 1.5681825876235962,
      "learning_rate": 0.00019434597887488532,
      "loss": 1.4955,
      "step": 3751
    },
    {
      "epoch": 0.1343671101402045,
      "grad_norm": 1.9825056791305542,
      "learning_rate": 0.00019434213331831398,
      "loss": 1.8734,
      "step": 3752
    },
    {
      "epoch": 0.13440292226977277,
      "grad_norm": 1.5142807960510254,
      "learning_rate": 0.00019433828649249087,
      "loss": 1.4563,
      "step": 3753
    },
    {
      "epoch": 0.13443873439934106,
      "grad_norm": 1.6803392171859741,
      "learning_rate": 0.00019433443839746785,
      "loss": 1.6086,
      "step": 3754
    },
    {
      "epoch": 0.13447454652890933,
      "grad_norm": 1.265977144241333,
      "learning_rate": 0.00019433058903329663,
      "loss": 1.7027,
      "step": 3755
    },
    {
      "epoch": 0.13451035865847763,
      "grad_norm": 2.002089500427246,
      "learning_rate": 0.00019432673840002898,
      "loss": 1.8738,
      "step": 3756
    },
    {
      "epoch": 0.1345461707880459,
      "grad_norm": 1.433565378189087,
      "learning_rate": 0.00019432288649771676,
      "loss": 1.3665,
      "step": 3757
    },
    {
      "epoch": 0.1345819829176142,
      "grad_norm": 1.1154769659042358,
      "learning_rate": 0.0001943190333264118,
      "loss": 1.6499,
      "step": 3758
    },
    {
      "epoch": 0.1346177950471825,
      "grad_norm": 1.8043376207351685,
      "learning_rate": 0.0001943151788861659,
      "loss": 1.6887,
      "step": 3759
    },
    {
      "epoch": 0.13465360717675076,
      "grad_norm": 1.4192312955856323,
      "learning_rate": 0.0001943113231770309,
      "loss": 1.3811,
      "step": 3760
    },
    {
      "epoch": 0.13468941930631906,
      "grad_norm": 1.6293889284133911,
      "learning_rate": 0.0001943074661990587,
      "loss": 1.9191,
      "step": 3761
    },
    {
      "epoch": 0.13472523143588733,
      "grad_norm": 1.9068212509155273,
      "learning_rate": 0.0001943036079523012,
      "loss": 1.6092,
      "step": 3762
    },
    {
      "epoch": 0.13476104356545562,
      "grad_norm": 1.3996855020523071,
      "learning_rate": 0.00019429974843681032,
      "loss": 1.4419,
      "step": 3763
    },
    {
      "epoch": 0.1347968556950239,
      "grad_norm": 1.6263465881347656,
      "learning_rate": 0.0001942958876526379,
      "loss": 1.7413,
      "step": 3764
    },
    {
      "epoch": 0.1348326678245922,
      "grad_norm": 2.075838327407837,
      "learning_rate": 0.000194292025599836,
      "loss": 1.5274,
      "step": 3765
    },
    {
      "epoch": 0.1348684799541605,
      "grad_norm": 2.207247734069824,
      "learning_rate": 0.00019428816227845652,
      "loss": 1.9061,
      "step": 3766
    },
    {
      "epoch": 0.13490429208372876,
      "grad_norm": 1.719639778137207,
      "learning_rate": 0.0001942842976885514,
      "loss": 1.6978,
      "step": 3767
    },
    {
      "epoch": 0.13494010421329705,
      "grad_norm": 1.641737461090088,
      "learning_rate": 0.00019428043183017274,
      "loss": 1.807,
      "step": 3768
    },
    {
      "epoch": 0.13497591634286532,
      "grad_norm": 4.124390125274658,
      "learning_rate": 0.00019427656470337242,
      "loss": 1.825,
      "step": 3769
    },
    {
      "epoch": 0.13501172847243362,
      "grad_norm": 1.5481641292572021,
      "learning_rate": 0.00019427269630820258,
      "loss": 1.5993,
      "step": 3770
    },
    {
      "epoch": 0.1350475406020019,
      "grad_norm": 1.4922468662261963,
      "learning_rate": 0.00019426882664471515,
      "loss": 1.5663,
      "step": 3771
    },
    {
      "epoch": 0.13508335273157018,
      "grad_norm": 1.7647703886032104,
      "learning_rate": 0.00019426495571296234,
      "loss": 1.6529,
      "step": 3772
    },
    {
      "epoch": 0.13511916486113848,
      "grad_norm": 1.8872132301330566,
      "learning_rate": 0.00019426108351299607,
      "loss": 1.8358,
      "step": 3773
    },
    {
      "epoch": 0.13515497699070675,
      "grad_norm": 1.7756673097610474,
      "learning_rate": 0.00019425721004486852,
      "loss": 1.8599,
      "step": 3774
    },
    {
      "epoch": 0.13519078912027505,
      "grad_norm": 1.4053908586502075,
      "learning_rate": 0.00019425333530863182,
      "loss": 1.4622,
      "step": 3775
    },
    {
      "epoch": 0.13522660124984331,
      "grad_norm": 1.7477492094039917,
      "learning_rate": 0.00019424945930433807,
      "loss": 1.8449,
      "step": 3776
    },
    {
      "epoch": 0.1352624133794116,
      "grad_norm": 1.654231309890747,
      "learning_rate": 0.0001942455820320394,
      "loss": 1.4055,
      "step": 3777
    },
    {
      "epoch": 0.13529822550897988,
      "grad_norm": 1.1583056449890137,
      "learning_rate": 0.00019424170349178802,
      "loss": 1.6276,
      "step": 3778
    },
    {
      "epoch": 0.13533403763854818,
      "grad_norm": 1.3777235746383667,
      "learning_rate": 0.00019423782368363604,
      "loss": 1.471,
      "step": 3779
    },
    {
      "epoch": 0.13536984976811647,
      "grad_norm": 1.4694525003433228,
      "learning_rate": 0.00019423394260763573,
      "loss": 1.4806,
      "step": 3780
    },
    {
      "epoch": 0.13540566189768474,
      "grad_norm": 1.5091458559036255,
      "learning_rate": 0.00019423006026383926,
      "loss": 1.7003,
      "step": 3781
    },
    {
      "epoch": 0.13544147402725304,
      "grad_norm": 1.9628468751907349,
      "learning_rate": 0.0001942261766522989,
      "loss": 1.8922,
      "step": 3782
    },
    {
      "epoch": 0.1354772861568213,
      "grad_norm": 1.4558885097503662,
      "learning_rate": 0.00019422229177306686,
      "loss": 1.4158,
      "step": 3783
    },
    {
      "epoch": 0.1355130982863896,
      "grad_norm": 2.17854380607605,
      "learning_rate": 0.0001942184056261954,
      "loss": 1.5416,
      "step": 3784
    },
    {
      "epoch": 0.13554891041595787,
      "grad_norm": 1.5657025575637817,
      "learning_rate": 0.00019421451821173685,
      "loss": 1.768,
      "step": 3785
    },
    {
      "epoch": 0.13558472254552617,
      "grad_norm": 1.424683928489685,
      "learning_rate": 0.0001942106295297435,
      "loss": 1.6767,
      "step": 3786
    },
    {
      "epoch": 0.13562053467509447,
      "grad_norm": 1.3735336065292358,
      "learning_rate": 0.00019420673958026762,
      "loss": 1.6295,
      "step": 3787
    },
    {
      "epoch": 0.13565634680466274,
      "grad_norm": 1.9221537113189697,
      "learning_rate": 0.0001942028483633616,
      "loss": 1.9549,
      "step": 3788
    },
    {
      "epoch": 0.13569215893423103,
      "grad_norm": 1.564569354057312,
      "learning_rate": 0.00019419895587907777,
      "loss": 1.6037,
      "step": 3789
    },
    {
      "epoch": 0.1357279710637993,
      "grad_norm": 1.569985032081604,
      "learning_rate": 0.0001941950621274685,
      "loss": 1.8659,
      "step": 3790
    },
    {
      "epoch": 0.1357637831933676,
      "grad_norm": 1.9685579538345337,
      "learning_rate": 0.00019419116710858614,
      "loss": 1.7621,
      "step": 3791
    },
    {
      "epoch": 0.13579959532293587,
      "grad_norm": 1.6017086505889893,
      "learning_rate": 0.00019418727082248316,
      "loss": 1.4238,
      "step": 3792
    },
    {
      "epoch": 0.13583540745250416,
      "grad_norm": 1.52402925491333,
      "learning_rate": 0.00019418337326921193,
      "loss": 1.4991,
      "step": 3793
    },
    {
      "epoch": 0.13587121958207246,
      "grad_norm": 1.6972289085388184,
      "learning_rate": 0.0001941794744488249,
      "loss": 1.7073,
      "step": 3794
    },
    {
      "epoch": 0.13590703171164073,
      "grad_norm": 1.2555439472198486,
      "learning_rate": 0.0001941755743613745,
      "loss": 1.6406,
      "step": 3795
    },
    {
      "epoch": 0.13594284384120903,
      "grad_norm": 2.1183104515075684,
      "learning_rate": 0.00019417167300691328,
      "loss": 1.6005,
      "step": 3796
    },
    {
      "epoch": 0.1359786559707773,
      "grad_norm": 1.3762356042861938,
      "learning_rate": 0.00019416777038549362,
      "loss": 1.3172,
      "step": 3797
    },
    {
      "epoch": 0.1360144681003456,
      "grad_norm": 1.4184659719467163,
      "learning_rate": 0.00019416386649716812,
      "loss": 1.5459,
      "step": 3798
    },
    {
      "epoch": 0.13605028022991386,
      "grad_norm": 1.5762983560562134,
      "learning_rate": 0.0001941599613419892,
      "loss": 1.3168,
      "step": 3799
    },
    {
      "epoch": 0.13608609235948216,
      "grad_norm": 2.1073853969573975,
      "learning_rate": 0.00019415605492000953,
      "loss": 1.9473,
      "step": 3800
    },
    {
      "epoch": 0.13612190448905045,
      "grad_norm": 1.605509638786316,
      "learning_rate": 0.00019415214723128154,
      "loss": 1.5105,
      "step": 3801
    },
    {
      "epoch": 0.13615771661861872,
      "grad_norm": 2.1277925968170166,
      "learning_rate": 0.0001941482382758579,
      "loss": 1.621,
      "step": 3802
    },
    {
      "epoch": 0.13619352874818702,
      "grad_norm": 1.3043378591537476,
      "learning_rate": 0.00019414432805379113,
      "loss": 1.389,
      "step": 3803
    },
    {
      "epoch": 0.1362293408777553,
      "grad_norm": 2.173734664916992,
      "learning_rate": 0.00019414041656513385,
      "loss": 1.4075,
      "step": 3804
    },
    {
      "epoch": 0.1362651530073236,
      "grad_norm": 1.3967981338500977,
      "learning_rate": 0.0001941365038099387,
      "loss": 1.708,
      "step": 3805
    },
    {
      "epoch": 0.13630096513689186,
      "grad_norm": 1.6393210887908936,
      "learning_rate": 0.00019413258978825834,
      "loss": 1.6744,
      "step": 3806
    },
    {
      "epoch": 0.13633677726646015,
      "grad_norm": 1.854275107383728,
      "learning_rate": 0.0001941286745001454,
      "loss": 1.5661,
      "step": 3807
    },
    {
      "epoch": 0.13637258939602845,
      "grad_norm": 1.3104915618896484,
      "learning_rate": 0.00019412475794565256,
      "loss": 1.3521,
      "step": 3808
    },
    {
      "epoch": 0.13640840152559672,
      "grad_norm": 1.714754343032837,
      "learning_rate": 0.00019412084012483249,
      "loss": 1.662,
      "step": 3809
    },
    {
      "epoch": 0.13644421365516501,
      "grad_norm": 2.037449359893799,
      "learning_rate": 0.00019411692103773795,
      "loss": 1.4034,
      "step": 3810
    },
    {
      "epoch": 0.13648002578473328,
      "grad_norm": 1.4741345643997192,
      "learning_rate": 0.00019411300068442167,
      "loss": 1.7963,
      "step": 3811
    },
    {
      "epoch": 0.13651583791430158,
      "grad_norm": 1.921120524406433,
      "learning_rate": 0.0001941090790649363,
      "loss": 1.6161,
      "step": 3812
    },
    {
      "epoch": 0.13655165004386985,
      "grad_norm": 1.4448705911636353,
      "learning_rate": 0.00019410515617933468,
      "loss": 1.2904,
      "step": 3813
    },
    {
      "epoch": 0.13658746217343815,
      "grad_norm": 1.8026163578033447,
      "learning_rate": 0.0001941012320276696,
      "loss": 1.9121,
      "step": 3814
    },
    {
      "epoch": 0.13662327430300641,
      "grad_norm": 1.457109808921814,
      "learning_rate": 0.0001940973066099938,
      "loss": 1.4424,
      "step": 3815
    },
    {
      "epoch": 0.1366590864325747,
      "grad_norm": 2.4692842960357666,
      "learning_rate": 0.00019409337992636015,
      "loss": 1.5736,
      "step": 3816
    },
    {
      "epoch": 0.136694898562143,
      "grad_norm": 2.4115750789642334,
      "learning_rate": 0.0001940894519768214,
      "loss": 1.4381,
      "step": 3817
    },
    {
      "epoch": 0.13673071069171128,
      "grad_norm": 1.905058741569519,
      "learning_rate": 0.00019408552276143045,
      "loss": 1.8125,
      "step": 3818
    },
    {
      "epoch": 0.13676652282127957,
      "grad_norm": 1.6240928173065186,
      "learning_rate": 0.00019408159228024018,
      "loss": 1.5705,
      "step": 3819
    },
    {
      "epoch": 0.13680233495084784,
      "grad_norm": 1.6123714447021484,
      "learning_rate": 0.00019407766053330342,
      "loss": 1.4053,
      "step": 3820
    },
    {
      "epoch": 0.13683814708041614,
      "grad_norm": 1.5208178758621216,
      "learning_rate": 0.00019407372752067308,
      "loss": 1.3531,
      "step": 3821
    },
    {
      "epoch": 0.1368739592099844,
      "grad_norm": 1.6093798875808716,
      "learning_rate": 0.0001940697932424021,
      "loss": 1.5774,
      "step": 3822
    },
    {
      "epoch": 0.1369097713395527,
      "grad_norm": 1.2067844867706299,
      "learning_rate": 0.0001940658576985434,
      "loss": 1.4985,
      "step": 3823
    },
    {
      "epoch": 0.136945583469121,
      "grad_norm": 1.312362551689148,
      "learning_rate": 0.0001940619208891499,
      "loss": 1.6762,
      "step": 3824
    },
    {
      "epoch": 0.13698139559868927,
      "grad_norm": 1.438751459121704,
      "learning_rate": 0.0001940579828142746,
      "loss": 1.6947,
      "step": 3825
    },
    {
      "epoch": 0.13701720772825757,
      "grad_norm": 1.2554978132247925,
      "learning_rate": 0.00019405404347397047,
      "loss": 1.5329,
      "step": 3826
    },
    {
      "epoch": 0.13705301985782584,
      "grad_norm": 1.9145128726959229,
      "learning_rate": 0.0001940501028682905,
      "loss": 1.7587,
      "step": 3827
    },
    {
      "epoch": 0.13708883198739413,
      "grad_norm": 1.6156679391860962,
      "learning_rate": 0.00019404616099728773,
      "loss": 1.5496,
      "step": 3828
    },
    {
      "epoch": 0.1371246441169624,
      "grad_norm": 1.4059901237487793,
      "learning_rate": 0.00019404221786101513,
      "loss": 1.4649,
      "step": 3829
    },
    {
      "epoch": 0.1371604562465307,
      "grad_norm": 1.4880584478378296,
      "learning_rate": 0.0001940382734595258,
      "loss": 1.5342,
      "step": 3830
    },
    {
      "epoch": 0.137196268376099,
      "grad_norm": 2.3095836639404297,
      "learning_rate": 0.00019403432779287286,
      "loss": 1.7837,
      "step": 3831
    },
    {
      "epoch": 0.13723208050566726,
      "grad_norm": 1.6353516578674316,
      "learning_rate": 0.00019403038086110926,
      "loss": 1.7732,
      "step": 3832
    },
    {
      "epoch": 0.13726789263523556,
      "grad_norm": 1.5463842153549194,
      "learning_rate": 0.00019402643266428822,
      "loss": 1.7355,
      "step": 3833
    },
    {
      "epoch": 0.13730370476480383,
      "grad_norm": 2.119309186935425,
      "learning_rate": 0.00019402248320246282,
      "loss": 2.0842,
      "step": 3834
    },
    {
      "epoch": 0.13733951689437213,
      "grad_norm": 1.909306287765503,
      "learning_rate": 0.00019401853247568614,
      "loss": 1.5683,
      "step": 3835
    },
    {
      "epoch": 0.1373753290239404,
      "grad_norm": 1.564666509628296,
      "learning_rate": 0.00019401458048401145,
      "loss": 1.3599,
      "step": 3836
    },
    {
      "epoch": 0.1374111411535087,
      "grad_norm": 1.7415882349014282,
      "learning_rate": 0.0001940106272274918,
      "loss": 1.4302,
      "step": 3837
    },
    {
      "epoch": 0.137446953283077,
      "grad_norm": 1.8725720643997192,
      "learning_rate": 0.00019400667270618046,
      "loss": 1.3951,
      "step": 3838
    },
    {
      "epoch": 0.13748276541264526,
      "grad_norm": 1.7075797319412231,
      "learning_rate": 0.00019400271692013058,
      "loss": 1.9641,
      "step": 3839
    },
    {
      "epoch": 0.13751857754221355,
      "grad_norm": 2.7534332275390625,
      "learning_rate": 0.0001939987598693954,
      "loss": 1.6384,
      "step": 3840
    },
    {
      "epoch": 0.13755438967178182,
      "grad_norm": 2.2370193004608154,
      "learning_rate": 0.00019399480155402813,
      "loss": 1.57,
      "step": 3841
    },
    {
      "epoch": 0.13759020180135012,
      "grad_norm": 1.1581882238388062,
      "learning_rate": 0.0001939908419740821,
      "loss": 1.664,
      "step": 3842
    },
    {
      "epoch": 0.1376260139309184,
      "grad_norm": 1.3554637432098389,
      "learning_rate": 0.0001939868811296105,
      "loss": 1.2336,
      "step": 3843
    },
    {
      "epoch": 0.1376618260604867,
      "grad_norm": 1.4079585075378418,
      "learning_rate": 0.00019398291902066666,
      "loss": 1.5603,
      "step": 3844
    },
    {
      "epoch": 0.13769763819005498,
      "grad_norm": 1.8186653852462769,
      "learning_rate": 0.00019397895564730386,
      "loss": 1.3144,
      "step": 3845
    },
    {
      "epoch": 0.13773345031962325,
      "grad_norm": 2.0624194145202637,
      "learning_rate": 0.00019397499100957542,
      "loss": 1.974,
      "step": 3846
    },
    {
      "epoch": 0.13776926244919155,
      "grad_norm": 2.5842959880828857,
      "learning_rate": 0.00019397102510753473,
      "loss": 1.5397,
      "step": 3847
    },
    {
      "epoch": 0.13780507457875982,
      "grad_norm": 1.7426347732543945,
      "learning_rate": 0.0001939670579412351,
      "loss": 1.8237,
      "step": 3848
    },
    {
      "epoch": 0.13784088670832811,
      "grad_norm": 1.6106834411621094,
      "learning_rate": 0.00019396308951072992,
      "loss": 1.6794,
      "step": 3849
    },
    {
      "epoch": 0.13787669883789638,
      "grad_norm": 2.3994948863983154,
      "learning_rate": 0.00019395911981607254,
      "loss": 1.4646,
      "step": 3850
    },
    {
      "epoch": 0.13791251096746468,
      "grad_norm": 1.4603257179260254,
      "learning_rate": 0.00019395514885731644,
      "loss": 1.5009,
      "step": 3851
    },
    {
      "epoch": 0.13794832309703298,
      "grad_norm": 2.004852056503296,
      "learning_rate": 0.000193951176634515,
      "loss": 1.6899,
      "step": 3852
    },
    {
      "epoch": 0.13798413522660125,
      "grad_norm": 1.9660876989364624,
      "learning_rate": 0.00019394720314772166,
      "loss": 1.7038,
      "step": 3853
    },
    {
      "epoch": 0.13801994735616954,
      "grad_norm": 1.419790506362915,
      "learning_rate": 0.00019394322839698988,
      "loss": 1.8069,
      "step": 3854
    },
    {
      "epoch": 0.1380557594857378,
      "grad_norm": 1.6597111225128174,
      "learning_rate": 0.00019393925238237313,
      "loss": 1.8452,
      "step": 3855
    },
    {
      "epoch": 0.1380915716153061,
      "grad_norm": 1.4487488269805908,
      "learning_rate": 0.00019393527510392494,
      "loss": 1.7923,
      "step": 3856
    },
    {
      "epoch": 0.13812738374487438,
      "grad_norm": 1.6511144638061523,
      "learning_rate": 0.0001939312965616988,
      "loss": 1.6654,
      "step": 3857
    },
    {
      "epoch": 0.13816319587444267,
      "grad_norm": 1.8880691528320312,
      "learning_rate": 0.0001939273167557482,
      "loss": 1.666,
      "step": 3858
    },
    {
      "epoch": 0.13819900800401097,
      "grad_norm": 1.333387017250061,
      "learning_rate": 0.00019392333568612672,
      "loss": 1.6742,
      "step": 3859
    },
    {
      "epoch": 0.13823482013357924,
      "grad_norm": 1.4086413383483887,
      "learning_rate": 0.00019391935335288788,
      "loss": 1.5357,
      "step": 3860
    },
    {
      "epoch": 0.13827063226314754,
      "grad_norm": 2.5079047679901123,
      "learning_rate": 0.00019391536975608533,
      "loss": 1.6305,
      "step": 3861
    },
    {
      "epoch": 0.1383064443927158,
      "grad_norm": 1.7378469705581665,
      "learning_rate": 0.0001939113848957726,
      "loss": 1.2147,
      "step": 3862
    },
    {
      "epoch": 0.1383422565222841,
      "grad_norm": 2.0225167274475098,
      "learning_rate": 0.00019390739877200335,
      "loss": 1.5807,
      "step": 3863
    },
    {
      "epoch": 0.13837806865185237,
      "grad_norm": 1.6210397481918335,
      "learning_rate": 0.00019390341138483117,
      "loss": 1.6289,
      "step": 3864
    },
    {
      "epoch": 0.13841388078142067,
      "grad_norm": 1.4240684509277344,
      "learning_rate": 0.0001938994227343097,
      "loss": 1.7307,
      "step": 3865
    },
    {
      "epoch": 0.13844969291098896,
      "grad_norm": 1.4018386602401733,
      "learning_rate": 0.00019389543282049263,
      "loss": 1.4844,
      "step": 3866
    },
    {
      "epoch": 0.13848550504055723,
      "grad_norm": 1.4546011686325073,
      "learning_rate": 0.0001938914416434336,
      "loss": 1.6839,
      "step": 3867
    },
    {
      "epoch": 0.13852131717012553,
      "grad_norm": 1.5144541263580322,
      "learning_rate": 0.00019388744920318638,
      "loss": 1.6554,
      "step": 3868
    },
    {
      "epoch": 0.1385571292996938,
      "grad_norm": 1.9731286764144897,
      "learning_rate": 0.00019388345549980462,
      "loss": 1.8439,
      "step": 3869
    },
    {
      "epoch": 0.1385929414292621,
      "grad_norm": 1.4650535583496094,
      "learning_rate": 0.00019387946053334206,
      "loss": 1.836,
      "step": 3870
    },
    {
      "epoch": 0.13862875355883036,
      "grad_norm": 2.14497447013855,
      "learning_rate": 0.00019387546430385246,
      "loss": 1.5419,
      "step": 3871
    },
    {
      "epoch": 0.13866456568839866,
      "grad_norm": 1.929789423942566,
      "learning_rate": 0.00019387146681138957,
      "loss": 1.6334,
      "step": 3872
    },
    {
      "epoch": 0.13870037781796696,
      "grad_norm": 1.5620695352554321,
      "learning_rate": 0.00019386746805600717,
      "loss": 1.5802,
      "step": 3873
    },
    {
      "epoch": 0.13873618994753523,
      "grad_norm": 1.579016089439392,
      "learning_rate": 0.00019386346803775909,
      "loss": 1.5735,
      "step": 3874
    },
    {
      "epoch": 0.13877200207710352,
      "grad_norm": 2.0494682788848877,
      "learning_rate": 0.00019385946675669913,
      "loss": 1.6366,
      "step": 3875
    },
    {
      "epoch": 0.1388078142066718,
      "grad_norm": 1.5577737092971802,
      "learning_rate": 0.0001938554642128811,
      "loss": 1.5314,
      "step": 3876
    },
    {
      "epoch": 0.1388436263362401,
      "grad_norm": 1.546035647392273,
      "learning_rate": 0.00019385146040635886,
      "loss": 1.6867,
      "step": 3877
    },
    {
      "epoch": 0.13887943846580836,
      "grad_norm": 1.4088118076324463,
      "learning_rate": 0.00019384745533718628,
      "loss": 1.2657,
      "step": 3878
    },
    {
      "epoch": 0.13891525059537665,
      "grad_norm": 2.4216670989990234,
      "learning_rate": 0.00019384344900541723,
      "loss": 1.448,
      "step": 3879
    },
    {
      "epoch": 0.13895106272494495,
      "grad_norm": 2.032590627670288,
      "learning_rate": 0.00019383944141110565,
      "loss": 1.607,
      "step": 3880
    },
    {
      "epoch": 0.13898687485451322,
      "grad_norm": 1.4707528352737427,
      "learning_rate": 0.00019383543255430542,
      "loss": 1.3947,
      "step": 3881
    },
    {
      "epoch": 0.13902268698408152,
      "grad_norm": 1.8237826824188232,
      "learning_rate": 0.00019383142243507048,
      "loss": 2.0583,
      "step": 3882
    },
    {
      "epoch": 0.13905849911364979,
      "grad_norm": 1.9654194116592407,
      "learning_rate": 0.00019382741105345482,
      "loss": 1.6511,
      "step": 3883
    },
    {
      "epoch": 0.13909431124321808,
      "grad_norm": 1.2782480716705322,
      "learning_rate": 0.0001938233984095123,
      "loss": 1.8372,
      "step": 3884
    },
    {
      "epoch": 0.13913012337278635,
      "grad_norm": 1.2892425060272217,
      "learning_rate": 0.00019381938450329704,
      "loss": 1.8873,
      "step": 3885
    },
    {
      "epoch": 0.13916593550235465,
      "grad_norm": 1.9381794929504395,
      "learning_rate": 0.00019381536933486295,
      "loss": 1.8379,
      "step": 3886
    },
    {
      "epoch": 0.13920174763192295,
      "grad_norm": 1.7396278381347656,
      "learning_rate": 0.0001938113529042641,
      "loss": 1.7356,
      "step": 3887
    },
    {
      "epoch": 0.13923755976149121,
      "grad_norm": 1.13014817237854,
      "learning_rate": 0.0001938073352115545,
      "loss": 1.4796,
      "step": 3888
    },
    {
      "epoch": 0.1392733718910595,
      "grad_norm": 1.9644767045974731,
      "learning_rate": 0.00019380331625678821,
      "loss": 1.7187,
      "step": 3889
    },
    {
      "epoch": 0.13930918402062778,
      "grad_norm": 1.3784605264663696,
      "learning_rate": 0.00019379929604001927,
      "loss": 1.5731,
      "step": 3890
    },
    {
      "epoch": 0.13934499615019608,
      "grad_norm": 2.058908700942993,
      "learning_rate": 0.00019379527456130183,
      "loss": 1.7203,
      "step": 3891
    },
    {
      "epoch": 0.13938080827976435,
      "grad_norm": 1.9625152349472046,
      "learning_rate": 0.00019379125182068994,
      "loss": 1.4671,
      "step": 3892
    },
    {
      "epoch": 0.13941662040933264,
      "grad_norm": 2.0643458366394043,
      "learning_rate": 0.00019378722781823772,
      "loss": 1.6485,
      "step": 3893
    },
    {
      "epoch": 0.13945243253890094,
      "grad_norm": 1.778398871421814,
      "learning_rate": 0.00019378320255399934,
      "loss": 1.6492,
      "step": 3894
    },
    {
      "epoch": 0.1394882446684692,
      "grad_norm": 1.6145321130752563,
      "learning_rate": 0.00019377917602802897,
      "loss": 1.6214,
      "step": 3895
    },
    {
      "epoch": 0.1395240567980375,
      "grad_norm": 2.243457794189453,
      "learning_rate": 0.00019377514824038073,
      "loss": 1.9855,
      "step": 3896
    },
    {
      "epoch": 0.13955986892760577,
      "grad_norm": 2.51615047454834,
      "learning_rate": 0.00019377111919110883,
      "loss": 2.0496,
      "step": 3897
    },
    {
      "epoch": 0.13959568105717407,
      "grad_norm": 1.4384043216705322,
      "learning_rate": 0.00019376708888026747,
      "loss": 1.7474,
      "step": 3898
    },
    {
      "epoch": 0.13963149318674234,
      "grad_norm": 2.0922720432281494,
      "learning_rate": 0.0001937630573079109,
      "loss": 1.5602,
      "step": 3899
    },
    {
      "epoch": 0.13966730531631064,
      "grad_norm": 1.4650601148605347,
      "learning_rate": 0.0001937590244740933,
      "loss": 1.3567,
      "step": 3900
    },
    {
      "epoch": 0.13970311744587893,
      "grad_norm": 1.8897500038146973,
      "learning_rate": 0.000193754990378869,
      "loss": 1.6301,
      "step": 3901
    },
    {
      "epoch": 0.1397389295754472,
      "grad_norm": 1.7540127038955688,
      "learning_rate": 0.00019375095502229223,
      "loss": 1.4029,
      "step": 3902
    },
    {
      "epoch": 0.1397747417050155,
      "grad_norm": 1.8361109495162964,
      "learning_rate": 0.0001937469184044173,
      "loss": 1.6593,
      "step": 3903
    },
    {
      "epoch": 0.13981055383458377,
      "grad_norm": 1.9301716089248657,
      "learning_rate": 0.0001937428805252985,
      "loss": 1.3712,
      "step": 3904
    },
    {
      "epoch": 0.13984636596415206,
      "grad_norm": 1.518929123878479,
      "learning_rate": 0.00019373884138499018,
      "loss": 1.5331,
      "step": 3905
    },
    {
      "epoch": 0.13988217809372033,
      "grad_norm": 1.6392289400100708,
      "learning_rate": 0.00019373480098354665,
      "loss": 1.5763,
      "step": 3906
    },
    {
      "epoch": 0.13991799022328863,
      "grad_norm": 1.7571324110031128,
      "learning_rate": 0.00019373075932102227,
      "loss": 1.6839,
      "step": 3907
    },
    {
      "epoch": 0.13995380235285693,
      "grad_norm": 1.5957682132720947,
      "learning_rate": 0.00019372671639747145,
      "loss": 1.7607,
      "step": 3908
    },
    {
      "epoch": 0.1399896144824252,
      "grad_norm": 1.305310845375061,
      "learning_rate": 0.00019372267221294854,
      "loss": 1.679,
      "step": 3909
    },
    {
      "epoch": 0.1400254266119935,
      "grad_norm": 1.584241271018982,
      "learning_rate": 0.00019371862676750796,
      "loss": 1.6078,
      "step": 3910
    },
    {
      "epoch": 0.14006123874156176,
      "grad_norm": 1.5107513666152954,
      "learning_rate": 0.00019371458006120417,
      "loss": 1.7136,
      "step": 3911
    },
    {
      "epoch": 0.14009705087113006,
      "grad_norm": 1.7936160564422607,
      "learning_rate": 0.00019371053209409157,
      "loss": 1.5745,
      "step": 3912
    },
    {
      "epoch": 0.14013286300069833,
      "grad_norm": 2.0149476528167725,
      "learning_rate": 0.00019370648286622466,
      "loss": 1.3959,
      "step": 3913
    },
    {
      "epoch": 0.14016867513026662,
      "grad_norm": 1.791279673576355,
      "learning_rate": 0.00019370243237765787,
      "loss": 1.2805,
      "step": 3914
    },
    {
      "epoch": 0.1402044872598349,
      "grad_norm": 1.2142484188079834,
      "learning_rate": 0.00019369838062844577,
      "loss": 1.6181,
      "step": 3915
    },
    {
      "epoch": 0.1402402993894032,
      "grad_norm": 1.787173867225647,
      "learning_rate": 0.00019369432761864278,
      "loss": 1.5593,
      "step": 3916
    },
    {
      "epoch": 0.14027611151897149,
      "grad_norm": 1.2792410850524902,
      "learning_rate": 0.00019369027334830346,
      "loss": 1.5291,
      "step": 3917
    },
    {
      "epoch": 0.14031192364853975,
      "grad_norm": 1.5265432596206665,
      "learning_rate": 0.00019368621781748238,
      "loss": 1.7252,
      "step": 3918
    },
    {
      "epoch": 0.14034773577810805,
      "grad_norm": 1.5280210971832275,
      "learning_rate": 0.0001936821610262341,
      "loss": 1.6944,
      "step": 3919
    },
    {
      "epoch": 0.14038354790767632,
      "grad_norm": 1.6217960119247437,
      "learning_rate": 0.00019367810297461313,
      "loss": 1.4182,
      "step": 3920
    },
    {
      "epoch": 0.14041936003724462,
      "grad_norm": 1.484212875366211,
      "learning_rate": 0.00019367404366267416,
      "loss": 1.4251,
      "step": 3921
    },
    {
      "epoch": 0.14045517216681289,
      "grad_norm": 1.711181402206421,
      "learning_rate": 0.0001936699830904718,
      "loss": 1.7121,
      "step": 3922
    },
    {
      "epoch": 0.14049098429638118,
      "grad_norm": 1.52628755569458,
      "learning_rate": 0.00019366592125806057,
      "loss": 1.5609,
      "step": 3923
    },
    {
      "epoch": 0.14052679642594948,
      "grad_norm": 1.480896234512329,
      "learning_rate": 0.00019366185816549524,
      "loss": 1.4094,
      "step": 3924
    },
    {
      "epoch": 0.14056260855551775,
      "grad_norm": 1.4317753314971924,
      "learning_rate": 0.0001936577938128304,
      "loss": 1.6869,
      "step": 3925
    },
    {
      "epoch": 0.14059842068508605,
      "grad_norm": 2.088813543319702,
      "learning_rate": 0.00019365372820012077,
      "loss": 1.4735,
      "step": 3926
    },
    {
      "epoch": 0.14063423281465431,
      "grad_norm": 1.6882331371307373,
      "learning_rate": 0.00019364966132742102,
      "loss": 1.7869,
      "step": 3927
    },
    {
      "epoch": 0.1406700449442226,
      "grad_norm": 2.3925135135650635,
      "learning_rate": 0.00019364559319478585,
      "loss": 1.5204,
      "step": 3928
    },
    {
      "epoch": 0.14070585707379088,
      "grad_norm": 1.3298742771148682,
      "learning_rate": 0.00019364152380227007,
      "loss": 1.3791,
      "step": 3929
    },
    {
      "epoch": 0.14074166920335918,
      "grad_norm": 2.0857841968536377,
      "learning_rate": 0.00019363745314992836,
      "loss": 1.6491,
      "step": 3930
    },
    {
      "epoch": 0.14077748133292747,
      "grad_norm": 1.499003291130066,
      "learning_rate": 0.00019363338123781548,
      "loss": 1.5969,
      "step": 3931
    },
    {
      "epoch": 0.14081329346249574,
      "grad_norm": 1.8543964624404907,
      "learning_rate": 0.00019362930806598625,
      "loss": 1.5286,
      "step": 3932
    },
    {
      "epoch": 0.14084910559206404,
      "grad_norm": 1.2838963270187378,
      "learning_rate": 0.00019362523363449546,
      "loss": 1.3795,
      "step": 3933
    },
    {
      "epoch": 0.1408849177216323,
      "grad_norm": 1.6175882816314697,
      "learning_rate": 0.0001936211579433979,
      "loss": 1.5956,
      "step": 3934
    },
    {
      "epoch": 0.1409207298512006,
      "grad_norm": 1.3847663402557373,
      "learning_rate": 0.00019361708099274844,
      "loss": 1.7353,
      "step": 3935
    },
    {
      "epoch": 0.14095654198076887,
      "grad_norm": 1.8384443521499634,
      "learning_rate": 0.00019361300278260193,
      "loss": 2.0406,
      "step": 3936
    },
    {
      "epoch": 0.14099235411033717,
      "grad_norm": 1.939363718032837,
      "learning_rate": 0.00019360892331301316,
      "loss": 1.7603,
      "step": 3937
    },
    {
      "epoch": 0.14102816623990547,
      "grad_norm": 1.2415847778320312,
      "learning_rate": 0.00019360484258403713,
      "loss": 1.7003,
      "step": 3938
    },
    {
      "epoch": 0.14106397836947374,
      "grad_norm": 3.8942644596099854,
      "learning_rate": 0.00019360076059572867,
      "loss": 1.3907,
      "step": 3939
    },
    {
      "epoch": 0.14109979049904203,
      "grad_norm": 1.311605453491211,
      "learning_rate": 0.0001935966773481427,
      "loss": 1.7552,
      "step": 3940
    },
    {
      "epoch": 0.1411356026286103,
      "grad_norm": 1.316116452217102,
      "learning_rate": 0.00019359259284133418,
      "loss": 1.4707,
      "step": 3941
    },
    {
      "epoch": 0.1411714147581786,
      "grad_norm": 1.855420470237732,
      "learning_rate": 0.00019358850707535804,
      "loss": 1.445,
      "step": 3942
    },
    {
      "epoch": 0.14120722688774687,
      "grad_norm": 1.1825437545776367,
      "learning_rate": 0.00019358442005026926,
      "loss": 1.3708,
      "step": 3943
    },
    {
      "epoch": 0.14124303901731516,
      "grad_norm": 1.6518439054489136,
      "learning_rate": 0.0001935803317661228,
      "loss": 1.6204,
      "step": 3944
    },
    {
      "epoch": 0.14127885114688346,
      "grad_norm": 1.6733884811401367,
      "learning_rate": 0.0001935762422229737,
      "loss": 1.7793,
      "step": 3945
    },
    {
      "epoch": 0.14131466327645173,
      "grad_norm": 1.8255501985549927,
      "learning_rate": 0.00019357215142087699,
      "loss": 1.3537,
      "step": 3946
    },
    {
      "epoch": 0.14135047540602003,
      "grad_norm": 1.6905624866485596,
      "learning_rate": 0.0001935680593598877,
      "loss": 1.4878,
      "step": 3947
    },
    {
      "epoch": 0.1413862875355883,
      "grad_norm": 1.3700306415557861,
      "learning_rate": 0.00019356396604006083,
      "loss": 1.3288,
      "step": 3948
    },
    {
      "epoch": 0.1414220996651566,
      "grad_norm": 1.2963467836380005,
      "learning_rate": 0.00019355987146145147,
      "loss": 1.5802,
      "step": 3949
    },
    {
      "epoch": 0.14145791179472486,
      "grad_norm": 1.2930432558059692,
      "learning_rate": 0.00019355577562411473,
      "loss": 1.592,
      "step": 3950
    },
    {
      "epoch": 0.14149372392429316,
      "grad_norm": 1.4524013996124268,
      "learning_rate": 0.00019355167852810575,
      "loss": 1.6285,
      "step": 3951
    },
    {
      "epoch": 0.14152953605386145,
      "grad_norm": 1.399037480354309,
      "learning_rate": 0.00019354758017347957,
      "loss": 1.5855,
      "step": 3952
    },
    {
      "epoch": 0.14156534818342972,
      "grad_norm": 1.4224966764450073,
      "learning_rate": 0.00019354348056029136,
      "loss": 1.9141,
      "step": 3953
    },
    {
      "epoch": 0.14160116031299802,
      "grad_norm": 1.995407223701477,
      "learning_rate": 0.0001935393796885963,
      "loss": 1.6105,
      "step": 3954
    },
    {
      "epoch": 0.1416369724425663,
      "grad_norm": 1.7793134450912476,
      "learning_rate": 0.00019353527755844953,
      "loss": 1.6095,
      "step": 3955
    },
    {
      "epoch": 0.14167278457213459,
      "grad_norm": 1.9654982089996338,
      "learning_rate": 0.00019353117416990627,
      "loss": 1.6372,
      "step": 3956
    },
    {
      "epoch": 0.14170859670170285,
      "grad_norm": 1.6730225086212158,
      "learning_rate": 0.0001935270695230217,
      "loss": 1.3259,
      "step": 3957
    },
    {
      "epoch": 0.14174440883127115,
      "grad_norm": 1.6628282070159912,
      "learning_rate": 0.00019352296361785105,
      "loss": 1.6113,
      "step": 3958
    },
    {
      "epoch": 0.14178022096083945,
      "grad_norm": 1.5034743547439575,
      "learning_rate": 0.00019351885645444957,
      "loss": 1.5158,
      "step": 3959
    },
    {
      "epoch": 0.14181603309040772,
      "grad_norm": 2.691425323486328,
      "learning_rate": 0.0001935147480328725,
      "loss": 1.6924,
      "step": 3960
    },
    {
      "epoch": 0.141851845219976,
      "grad_norm": 1.750555157661438,
      "learning_rate": 0.0001935106383531751,
      "loss": 1.6288,
      "step": 3961
    },
    {
      "epoch": 0.14188765734954428,
      "grad_norm": 2.0469725131988525,
      "learning_rate": 0.00019350652741541272,
      "loss": 1.7488,
      "step": 3962
    },
    {
      "epoch": 0.14192346947911258,
      "grad_norm": 1.605979323387146,
      "learning_rate": 0.00019350241521964062,
      "loss": 1.6378,
      "step": 3963
    },
    {
      "epoch": 0.14195928160868085,
      "grad_norm": 1.7067230939865112,
      "learning_rate": 0.00019349830176591408,
      "loss": 1.778,
      "step": 3964
    },
    {
      "epoch": 0.14199509373824915,
      "grad_norm": 2.101348400115967,
      "learning_rate": 0.00019349418705428854,
      "loss": 1.4173,
      "step": 3965
    },
    {
      "epoch": 0.14203090586781744,
      "grad_norm": 1.571287989616394,
      "learning_rate": 0.0001934900710848193,
      "loss": 1.6081,
      "step": 3966
    },
    {
      "epoch": 0.1420667179973857,
      "grad_norm": 1.2210115194320679,
      "learning_rate": 0.00019348595385756178,
      "loss": 1.1689,
      "step": 3967
    },
    {
      "epoch": 0.142102530126954,
      "grad_norm": 1.7547118663787842,
      "learning_rate": 0.00019348183537257131,
      "loss": 1.652,
      "step": 3968
    },
    {
      "epoch": 0.14213834225652228,
      "grad_norm": 2.7394344806671143,
      "learning_rate": 0.00019347771562990332,
      "loss": 1.8632,
      "step": 3969
    },
    {
      "epoch": 0.14217415438609057,
      "grad_norm": 1.5795948505401611,
      "learning_rate": 0.00019347359462961326,
      "loss": 1.5562,
      "step": 3970
    },
    {
      "epoch": 0.14220996651565884,
      "grad_norm": 1.3725916147232056,
      "learning_rate": 0.00019346947237175655,
      "loss": 1.4961,
      "step": 3971
    },
    {
      "epoch": 0.14224577864522714,
      "grad_norm": 1.5435025691986084,
      "learning_rate": 0.00019346534885638866,
      "loss": 1.877,
      "step": 3972
    },
    {
      "epoch": 0.14228159077479544,
      "grad_norm": 1.6002931594848633,
      "learning_rate": 0.00019346122408356507,
      "loss": 1.6769,
      "step": 3973
    },
    {
      "epoch": 0.1423174029043637,
      "grad_norm": 1.736573576927185,
      "learning_rate": 0.00019345709805334123,
      "loss": 1.5778,
      "step": 3974
    },
    {
      "epoch": 0.142353215033932,
      "grad_norm": 1.5255703926086426,
      "learning_rate": 0.00019345297076577272,
      "loss": 1.8338,
      "step": 3975
    },
    {
      "epoch": 0.14238902716350027,
      "grad_norm": 1.9835999011993408,
      "learning_rate": 0.00019344884222091503,
      "loss": 1.6607,
      "step": 3976
    },
    {
      "epoch": 0.14242483929306857,
      "grad_norm": 1.5903600454330444,
      "learning_rate": 0.00019344471241882372,
      "loss": 1.5658,
      "step": 3977
    },
    {
      "epoch": 0.14246065142263684,
      "grad_norm": 1.982418179512024,
      "learning_rate": 0.0001934405813595543,
      "loss": 1.7213,
      "step": 3978
    },
    {
      "epoch": 0.14249646355220513,
      "grad_norm": 1.5569443702697754,
      "learning_rate": 0.00019343644904316242,
      "loss": 1.6512,
      "step": 3979
    },
    {
      "epoch": 0.14253227568177343,
      "grad_norm": 1.252939224243164,
      "learning_rate": 0.0001934323154697036,
      "loss": 1.307,
      "step": 3980
    },
    {
      "epoch": 0.1425680878113417,
      "grad_norm": 1.7318528890609741,
      "learning_rate": 0.00019342818063923357,
      "loss": 1.6283,
      "step": 3981
    },
    {
      "epoch": 0.14260389994091,
      "grad_norm": 1.3830797672271729,
      "learning_rate": 0.00019342404455180784,
      "loss": 1.5416,
      "step": 3982
    },
    {
      "epoch": 0.14263971207047826,
      "grad_norm": 1.8082152605056763,
      "learning_rate": 0.00019341990720748208,
      "loss": 1.4456,
      "step": 3983
    },
    {
      "epoch": 0.14267552420004656,
      "grad_norm": 1.605903148651123,
      "learning_rate": 0.000193415768606312,
      "loss": 1.9392,
      "step": 3984
    },
    {
      "epoch": 0.14271133632961483,
      "grad_norm": 1.7564297914505005,
      "learning_rate": 0.00019341162874835326,
      "loss": 1.9917,
      "step": 3985
    },
    {
      "epoch": 0.14274714845918313,
      "grad_norm": 1.354068398475647,
      "learning_rate": 0.00019340748763366152,
      "loss": 1.6126,
      "step": 3986
    },
    {
      "epoch": 0.14278296058875142,
      "grad_norm": 1.6402426958084106,
      "learning_rate": 0.00019340334526229253,
      "loss": 1.7782,
      "step": 3987
    },
    {
      "epoch": 0.1428187727183197,
      "grad_norm": 1.6833535432815552,
      "learning_rate": 0.00019339920163430202,
      "loss": 1.6701,
      "step": 3988
    },
    {
      "epoch": 0.142854584847888,
      "grad_norm": 2.3586952686309814,
      "learning_rate": 0.0001933950567497457,
      "loss": 1.6699,
      "step": 3989
    },
    {
      "epoch": 0.14289039697745626,
      "grad_norm": 2.0771920680999756,
      "learning_rate": 0.0001933909106086794,
      "loss": 1.6299,
      "step": 3990
    },
    {
      "epoch": 0.14292620910702455,
      "grad_norm": 1.3921229839324951,
      "learning_rate": 0.00019338676321115883,
      "loss": 1.4079,
      "step": 3991
    },
    {
      "epoch": 0.14296202123659282,
      "grad_norm": 1.5394184589385986,
      "learning_rate": 0.00019338261455723984,
      "loss": 1.3872,
      "step": 3992
    },
    {
      "epoch": 0.14299783336616112,
      "grad_norm": 1.6472933292388916,
      "learning_rate": 0.00019337846464697825,
      "loss": 1.3228,
      "step": 3993
    },
    {
      "epoch": 0.14303364549572942,
      "grad_norm": 1.6850037574768066,
      "learning_rate": 0.00019337431348042983,
      "loss": 1.5189,
      "step": 3994
    },
    {
      "epoch": 0.14306945762529769,
      "grad_norm": 3.3073441982269287,
      "learning_rate": 0.00019337016105765048,
      "loss": 1.5217,
      "step": 3995
    },
    {
      "epoch": 0.14310526975486598,
      "grad_norm": 1.2370121479034424,
      "learning_rate": 0.00019336600737869603,
      "loss": 1.89,
      "step": 3996
    },
    {
      "epoch": 0.14314108188443425,
      "grad_norm": 1.7099319696426392,
      "learning_rate": 0.00019336185244362244,
      "loss": 1.5129,
      "step": 3997
    },
    {
      "epoch": 0.14317689401400255,
      "grad_norm": 1.7119805812835693,
      "learning_rate": 0.0001933576962524855,
      "loss": 1.3498,
      "step": 3998
    },
    {
      "epoch": 0.14321270614357082,
      "grad_norm": 1.3768866062164307,
      "learning_rate": 0.0001933535388053412,
      "loss": 1.4905,
      "step": 3999
    },
    {
      "epoch": 0.1432485182731391,
      "grad_norm": 2.029284715652466,
      "learning_rate": 0.00019334938010224546,
      "loss": 1.8039,
      "step": 4000
    },
    {
      "epoch": 0.1432843304027074,
      "grad_norm": 1.4423243999481201,
      "learning_rate": 0.0001933452201432542,
      "loss": 1.7993,
      "step": 4001
    },
    {
      "epoch": 0.14332014253227568,
      "grad_norm": 1.3633579015731812,
      "learning_rate": 0.00019334105892842342,
      "loss": 1.5531,
      "step": 4002
    },
    {
      "epoch": 0.14335595466184398,
      "grad_norm": 1.5188398361206055,
      "learning_rate": 0.00019333689645780912,
      "loss": 1.7493,
      "step": 4003
    },
    {
      "epoch": 0.14339176679141225,
      "grad_norm": 1.5770578384399414,
      "learning_rate": 0.00019333273273146721,
      "loss": 1.5965,
      "step": 4004
    },
    {
      "epoch": 0.14342757892098054,
      "grad_norm": 1.7399095296859741,
      "learning_rate": 0.00019332856774945383,
      "loss": 1.6626,
      "step": 4005
    },
    {
      "epoch": 0.1434633910505488,
      "grad_norm": 1.923545241355896,
      "learning_rate": 0.00019332440151182493,
      "loss": 1.245,
      "step": 4006
    },
    {
      "epoch": 0.1434992031801171,
      "grad_norm": 1.3939712047576904,
      "learning_rate": 0.00019332023401863658,
      "loss": 1.3326,
      "step": 4007
    },
    {
      "epoch": 0.1435350153096854,
      "grad_norm": 1.7758983373641968,
      "learning_rate": 0.00019331606526994488,
      "loss": 1.603,
      "step": 4008
    },
    {
      "epoch": 0.14357082743925367,
      "grad_norm": 1.7037540674209595,
      "learning_rate": 0.0001933118952658059,
      "loss": 1.5085,
      "step": 4009
    },
    {
      "epoch": 0.14360663956882197,
      "grad_norm": 1.764907956123352,
      "learning_rate": 0.00019330772400627573,
      "loss": 1.1214,
      "step": 4010
    },
    {
      "epoch": 0.14364245169839024,
      "grad_norm": 1.9461796283721924,
      "learning_rate": 0.00019330355149141046,
      "loss": 1.5362,
      "step": 4011
    },
    {
      "epoch": 0.14367826382795854,
      "grad_norm": 2.126535415649414,
      "learning_rate": 0.00019329937772126626,
      "loss": 1.9574,
      "step": 4012
    },
    {
      "epoch": 0.1437140759575268,
      "grad_norm": 1.2146317958831787,
      "learning_rate": 0.0001932952026958993,
      "loss": 1.3589,
      "step": 4013
    },
    {
      "epoch": 0.1437498880870951,
      "grad_norm": 1.6272441148757935,
      "learning_rate": 0.00019329102641536575,
      "loss": 1.4191,
      "step": 4014
    },
    {
      "epoch": 0.14378570021666337,
      "grad_norm": 1.698939561843872,
      "learning_rate": 0.00019328684887972173,
      "loss": 1.6507,
      "step": 4015
    },
    {
      "epoch": 0.14382151234623167,
      "grad_norm": 2.0913240909576416,
      "learning_rate": 0.00019328267008902352,
      "loss": 1.5051,
      "step": 4016
    },
    {
      "epoch": 0.14385732447579996,
      "grad_norm": 1.6525379419326782,
      "learning_rate": 0.00019327849004332728,
      "loss": 1.7334,
      "step": 4017
    },
    {
      "epoch": 0.14389313660536823,
      "grad_norm": 1.996968388557434,
      "learning_rate": 0.0001932743087426893,
      "loss": 1.6867,
      "step": 4018
    },
    {
      "epoch": 0.14392894873493653,
      "grad_norm": 1.8276373147964478,
      "learning_rate": 0.00019327012618716583,
      "loss": 1.6823,
      "step": 4019
    },
    {
      "epoch": 0.1439647608645048,
      "grad_norm": 1.7384955883026123,
      "learning_rate": 0.00019326594237681311,
      "loss": 1.9046,
      "step": 4020
    },
    {
      "epoch": 0.1440005729940731,
      "grad_norm": 1.7132830619812012,
      "learning_rate": 0.00019326175731168742,
      "loss": 1.8533,
      "step": 4021
    },
    {
      "epoch": 0.14403638512364136,
      "grad_norm": 1.467646837234497,
      "learning_rate": 0.00019325757099184507,
      "loss": 1.2429,
      "step": 4022
    },
    {
      "epoch": 0.14407219725320966,
      "grad_norm": 1.7912276983261108,
      "learning_rate": 0.00019325338341734245,
      "loss": 1.685,
      "step": 4023
    },
    {
      "epoch": 0.14410800938277796,
      "grad_norm": 1.2599592208862305,
      "learning_rate": 0.00019324919458823582,
      "loss": 1.5783,
      "step": 4024
    },
    {
      "epoch": 0.14414382151234623,
      "grad_norm": 2.230529308319092,
      "learning_rate": 0.00019324500450458153,
      "loss": 1.7649,
      "step": 4025
    },
    {
      "epoch": 0.14417963364191452,
      "grad_norm": 1.7477943897247314,
      "learning_rate": 0.000193240813166436,
      "loss": 1.5066,
      "step": 4026
    },
    {
      "epoch": 0.1442154457714828,
      "grad_norm": 2.1602365970611572,
      "learning_rate": 0.0001932366205738556,
      "loss": 1.3825,
      "step": 4027
    },
    {
      "epoch": 0.1442512579010511,
      "grad_norm": 1.7186968326568604,
      "learning_rate": 0.00019323242672689676,
      "loss": 1.6016,
      "step": 4028
    },
    {
      "epoch": 0.14428707003061936,
      "grad_norm": 1.2913117408752441,
      "learning_rate": 0.00019322823162561586,
      "loss": 1.6547,
      "step": 4029
    },
    {
      "epoch": 0.14432288216018765,
      "grad_norm": 2.078582525253296,
      "learning_rate": 0.00019322403527006937,
      "loss": 1.5363,
      "step": 4030
    },
    {
      "epoch": 0.14435869428975595,
      "grad_norm": 1.4359171390533447,
      "learning_rate": 0.00019321983766031373,
      "loss": 1.8056,
      "step": 4031
    },
    {
      "epoch": 0.14439450641932422,
      "grad_norm": 1.497952938079834,
      "learning_rate": 0.00019321563879640542,
      "loss": 1.421,
      "step": 4032
    },
    {
      "epoch": 0.14443031854889252,
      "grad_norm": 1.6174252033233643,
      "learning_rate": 0.00019321143867840091,
      "loss": 1.5336,
      "step": 4033
    },
    {
      "epoch": 0.14446613067846079,
      "grad_norm": 1.4377790689468384,
      "learning_rate": 0.00019320723730635676,
      "loss": 1.8566,
      "step": 4034
    },
    {
      "epoch": 0.14450194280802908,
      "grad_norm": 2.628340482711792,
      "learning_rate": 0.00019320303468032944,
      "loss": 1.6271,
      "step": 4035
    },
    {
      "epoch": 0.14453775493759735,
      "grad_norm": 1.5452464818954468,
      "learning_rate": 0.00019319883080037552,
      "loss": 1.7898,
      "step": 4036
    },
    {
      "epoch": 0.14457356706716565,
      "grad_norm": 1.258744478225708,
      "learning_rate": 0.00019319462566655155,
      "loss": 1.727,
      "step": 4037
    },
    {
      "epoch": 0.14460937919673394,
      "grad_norm": 1.481390357017517,
      "learning_rate": 0.0001931904192789141,
      "loss": 1.6695,
      "step": 4038
    },
    {
      "epoch": 0.1446451913263022,
      "grad_norm": 1.921749234199524,
      "learning_rate": 0.00019318621163751974,
      "loss": 1.4772,
      "step": 4039
    },
    {
      "epoch": 0.1446810034558705,
      "grad_norm": 1.3307794332504272,
      "learning_rate": 0.00019318200274242515,
      "loss": 1.7675,
      "step": 4040
    },
    {
      "epoch": 0.14471681558543878,
      "grad_norm": 1.9500494003295898,
      "learning_rate": 0.0001931777925936869,
      "loss": 1.2512,
      "step": 4041
    },
    {
      "epoch": 0.14475262771500708,
      "grad_norm": 1.4262527227401733,
      "learning_rate": 0.00019317358119136163,
      "loss": 1.5668,
      "step": 4042
    },
    {
      "epoch": 0.14478843984457535,
      "grad_norm": 2.077671766281128,
      "learning_rate": 0.000193169368535506,
      "loss": 1.6646,
      "step": 4043
    },
    {
      "epoch": 0.14482425197414364,
      "grad_norm": 2.521920680999756,
      "learning_rate": 0.00019316515462617672,
      "loss": 1.9527,
      "step": 4044
    },
    {
      "epoch": 0.14486006410371194,
      "grad_norm": 2.3683269023895264,
      "learning_rate": 0.00019316093946343044,
      "loss": 1.3807,
      "step": 4045
    },
    {
      "epoch": 0.1448958762332802,
      "grad_norm": 2.656454086303711,
      "learning_rate": 0.00019315672304732388,
      "loss": 1.38,
      "step": 4046
    },
    {
      "epoch": 0.1449316883628485,
      "grad_norm": 2.149718999862671,
      "learning_rate": 0.0001931525053779138,
      "loss": 1.4746,
      "step": 4047
    },
    {
      "epoch": 0.14496750049241677,
      "grad_norm": 1.8854492902755737,
      "learning_rate": 0.00019314828645525692,
      "loss": 1.5098,
      "step": 4048
    },
    {
      "epoch": 0.14500331262198507,
      "grad_norm": 1.668682336807251,
      "learning_rate": 0.00019314406627940996,
      "loss": 1.3973,
      "step": 4049
    },
    {
      "epoch": 0.14503912475155334,
      "grad_norm": 2.1312713623046875,
      "learning_rate": 0.00019313984485042976,
      "loss": 1.6263,
      "step": 4050
    },
    {
      "epoch": 0.14507493688112164,
      "grad_norm": 1.5961402654647827,
      "learning_rate": 0.0001931356221683731,
      "loss": 1.6585,
      "step": 4051
    },
    {
      "epoch": 0.14511074901068993,
      "grad_norm": 1.9257919788360596,
      "learning_rate": 0.00019313139823329677,
      "loss": 1.5084,
      "step": 4052
    },
    {
      "epoch": 0.1451465611402582,
      "grad_norm": 1.2645249366760254,
      "learning_rate": 0.00019312717304525762,
      "loss": 1.4746,
      "step": 4053
    },
    {
      "epoch": 0.1451823732698265,
      "grad_norm": 2.637812614440918,
      "learning_rate": 0.00019312294660431246,
      "loss": 2.0501,
      "step": 4054
    },
    {
      "epoch": 0.14521818539939477,
      "grad_norm": 1.7504860162734985,
      "learning_rate": 0.00019311871891051818,
      "loss": 1.5642,
      "step": 4055
    },
    {
      "epoch": 0.14525399752896306,
      "grad_norm": 1.7490657567977905,
      "learning_rate": 0.00019311448996393163,
      "loss": 1.4698,
      "step": 4056
    },
    {
      "epoch": 0.14528980965853133,
      "grad_norm": 1.9256107807159424,
      "learning_rate": 0.00019311025976460978,
      "loss": 1.5543,
      "step": 4057
    },
    {
      "epoch": 0.14532562178809963,
      "grad_norm": 1.8809314966201782,
      "learning_rate": 0.00019310602831260944,
      "loss": 1.6738,
      "step": 4058
    },
    {
      "epoch": 0.14536143391766793,
      "grad_norm": 1.5549511909484863,
      "learning_rate": 0.0001931017956079876,
      "loss": 1.8064,
      "step": 4059
    },
    {
      "epoch": 0.1453972460472362,
      "grad_norm": 1.6281778812408447,
      "learning_rate": 0.0001930975616508012,
      "loss": 1.6228,
      "step": 4060
    },
    {
      "epoch": 0.1454330581768045,
      "grad_norm": 1.7993215322494507,
      "learning_rate": 0.00019309332644110722,
      "loss": 1.9817,
      "step": 4061
    },
    {
      "epoch": 0.14546887030637276,
      "grad_norm": 1.8530840873718262,
      "learning_rate": 0.0001930890899789626,
      "loss": 1.3957,
      "step": 4062
    },
    {
      "epoch": 0.14550468243594106,
      "grad_norm": 1.63814115524292,
      "learning_rate": 0.0001930848522644243,
      "loss": 1.6432,
      "step": 4063
    },
    {
      "epoch": 0.14554049456550933,
      "grad_norm": 1.4699954986572266,
      "learning_rate": 0.00019308061329754942,
      "loss": 1.6721,
      "step": 4064
    },
    {
      "epoch": 0.14557630669507762,
      "grad_norm": 1.9068433046340942,
      "learning_rate": 0.00019307637307839498,
      "loss": 1.4541,
      "step": 4065
    },
    {
      "epoch": 0.14561211882464592,
      "grad_norm": 2.003532886505127,
      "learning_rate": 0.00019307213160701798,
      "loss": 1.5527,
      "step": 4066
    },
    {
      "epoch": 0.1456479309542142,
      "grad_norm": 1.656280755996704,
      "learning_rate": 0.0001930678888834755,
      "loss": 1.4751,
      "step": 4067
    },
    {
      "epoch": 0.14568374308378249,
      "grad_norm": 1.5825142860412598,
      "learning_rate": 0.00019306364490782462,
      "loss": 1.749,
      "step": 4068
    },
    {
      "epoch": 0.14571955521335075,
      "grad_norm": 1.796554446220398,
      "learning_rate": 0.00019305939968012245,
      "loss": 1.6165,
      "step": 4069
    },
    {
      "epoch": 0.14575536734291905,
      "grad_norm": 1.4555829763412476,
      "learning_rate": 0.00019305515320042611,
      "loss": 1.5499,
      "step": 4070
    },
    {
      "epoch": 0.14579117947248732,
      "grad_norm": 1.407891869544983,
      "learning_rate": 0.00019305090546879267,
      "loss": 1.4587,
      "step": 4071
    },
    {
      "epoch": 0.14582699160205562,
      "grad_norm": 1.5201910734176636,
      "learning_rate": 0.00019304665648527935,
      "loss": 1.4193,
      "step": 4072
    },
    {
      "epoch": 0.1458628037316239,
      "grad_norm": 1.4494929313659668,
      "learning_rate": 0.00019304240624994328,
      "loss": 1.7287,
      "step": 4073
    },
    {
      "epoch": 0.14589861586119218,
      "grad_norm": 2.1447548866271973,
      "learning_rate": 0.00019303815476284168,
      "loss": 1.4123,
      "step": 4074
    },
    {
      "epoch": 0.14593442799076048,
      "grad_norm": 2.7469265460968018,
      "learning_rate": 0.0001930339020240317,
      "loss": 1.7491,
      "step": 4075
    },
    {
      "epoch": 0.14597024012032875,
      "grad_norm": 1.5956698656082153,
      "learning_rate": 0.00019302964803357057,
      "loss": 1.7761,
      "step": 4076
    },
    {
      "epoch": 0.14600605224989704,
      "grad_norm": 1.6789740324020386,
      "learning_rate": 0.00019302539279151553,
      "loss": 1.571,
      "step": 4077
    },
    {
      "epoch": 0.1460418643794653,
      "grad_norm": 1.620974063873291,
      "learning_rate": 0.00019302113629792383,
      "loss": 1.1998,
      "step": 4078
    },
    {
      "epoch": 0.1460776765090336,
      "grad_norm": 1.7867196798324585,
      "learning_rate": 0.0001930168785528527,
      "loss": 1.6286,
      "step": 4079
    },
    {
      "epoch": 0.1461134886386019,
      "grad_norm": 1.3706187009811401,
      "learning_rate": 0.00019301261955635948,
      "loss": 1.7489,
      "step": 4080
    },
    {
      "epoch": 0.14614930076817018,
      "grad_norm": 1.409914493560791,
      "learning_rate": 0.00019300835930850143,
      "loss": 1.466,
      "step": 4081
    },
    {
      "epoch": 0.14618511289773847,
      "grad_norm": 1.6765789985656738,
      "learning_rate": 0.0001930040978093359,
      "loss": 1.4924,
      "step": 4082
    },
    {
      "epoch": 0.14622092502730674,
      "grad_norm": 1.7411587238311768,
      "learning_rate": 0.00019299983505892016,
      "loss": 1.467,
      "step": 4083
    },
    {
      "epoch": 0.14625673715687504,
      "grad_norm": 2.216538906097412,
      "learning_rate": 0.00019299557105731166,
      "loss": 1.4325,
      "step": 4084
    },
    {
      "epoch": 0.1462925492864433,
      "grad_norm": 1.6220505237579346,
      "learning_rate": 0.00019299130580456765,
      "loss": 1.5235,
      "step": 4085
    },
    {
      "epoch": 0.1463283614160116,
      "grad_norm": 1.5487260818481445,
      "learning_rate": 0.0001929870393007456,
      "loss": 1.7881,
      "step": 4086
    },
    {
      "epoch": 0.1463641735455799,
      "grad_norm": 1.7539433240890503,
      "learning_rate": 0.00019298277154590284,
      "loss": 1.3377,
      "step": 4087
    },
    {
      "epoch": 0.14639998567514817,
      "grad_norm": 1.9030964374542236,
      "learning_rate": 0.0001929785025400969,
      "loss": 1.5991,
      "step": 4088
    },
    {
      "epoch": 0.14643579780471647,
      "grad_norm": 1.8642268180847168,
      "learning_rate": 0.0001929742322833851,
      "loss": 1.5565,
      "step": 4089
    },
    {
      "epoch": 0.14647160993428474,
      "grad_norm": 2.047663450241089,
      "learning_rate": 0.00019296996077582492,
      "loss": 1.4049,
      "step": 4090
    },
    {
      "epoch": 0.14650742206385303,
      "grad_norm": 1.8870108127593994,
      "learning_rate": 0.00019296568801747385,
      "loss": 1.6984,
      "step": 4091
    },
    {
      "epoch": 0.1465432341934213,
      "grad_norm": 1.8283346891403198,
      "learning_rate": 0.00019296141400838938,
      "loss": 1.5549,
      "step": 4092
    },
    {
      "epoch": 0.1465790463229896,
      "grad_norm": 1.5765265226364136,
      "learning_rate": 0.00019295713874862896,
      "loss": 1.3471,
      "step": 4093
    },
    {
      "epoch": 0.1466148584525579,
      "grad_norm": 1.2447913885116577,
      "learning_rate": 0.0001929528622382502,
      "loss": 1.5457,
      "step": 4094
    },
    {
      "epoch": 0.14665067058212616,
      "grad_norm": 1.6574128866195679,
      "learning_rate": 0.00019294858447731054,
      "loss": 1.4687,
      "step": 4095
    },
    {
      "epoch": 0.14668648271169446,
      "grad_norm": 2.061941385269165,
      "learning_rate": 0.0001929443054658676,
      "loss": 1.8521,
      "step": 4096
    },
    {
      "epoch": 0.14672229484126273,
      "grad_norm": 1.5628901720046997,
      "learning_rate": 0.00019294002520397888,
      "loss": 1.6355,
      "step": 4097
    },
    {
      "epoch": 0.14675810697083103,
      "grad_norm": 1.9049891233444214,
      "learning_rate": 0.000192935743691702,
      "loss": 1.5275,
      "step": 4098
    },
    {
      "epoch": 0.1467939191003993,
      "grad_norm": 1.6401816606521606,
      "learning_rate": 0.00019293146092909462,
      "loss": 1.5944,
      "step": 4099
    },
    {
      "epoch": 0.1468297312299676,
      "grad_norm": 1.786863923072815,
      "learning_rate": 0.00019292717691621428,
      "loss": 1.579,
      "step": 4100
    },
    {
      "epoch": 0.1468655433595359,
      "grad_norm": 1.6657721996307373,
      "learning_rate": 0.00019292289165311863,
      "loss": 1.6496,
      "step": 4101
    },
    {
      "epoch": 0.14690135548910416,
      "grad_norm": 1.5139931440353394,
      "learning_rate": 0.00019291860513986534,
      "loss": 1.8056,
      "step": 4102
    },
    {
      "epoch": 0.14693716761867245,
      "grad_norm": 2.9963998794555664,
      "learning_rate": 0.0001929143173765121,
      "loss": 1.6924,
      "step": 4103
    },
    {
      "epoch": 0.14697297974824072,
      "grad_norm": 1.4028496742248535,
      "learning_rate": 0.00019291002836311654,
      "loss": 1.7265,
      "step": 4104
    },
    {
      "epoch": 0.14700879187780902,
      "grad_norm": 1.964406132698059,
      "learning_rate": 0.0001929057380997364,
      "loss": 1.3505,
      "step": 4105
    },
    {
      "epoch": 0.1470446040073773,
      "grad_norm": 1.6631462574005127,
      "learning_rate": 0.0001929014465864294,
      "loss": 1.5789,
      "step": 4106
    },
    {
      "epoch": 0.14708041613694559,
      "grad_norm": 1.476067066192627,
      "learning_rate": 0.00019289715382325327,
      "loss": 1.4211,
      "step": 4107
    },
    {
      "epoch": 0.14711622826651388,
      "grad_norm": 2.226459264755249,
      "learning_rate": 0.00019289285981026577,
      "loss": 1.8789,
      "step": 4108
    },
    {
      "epoch": 0.14715204039608215,
      "grad_norm": 1.4748985767364502,
      "learning_rate": 0.00019288856454752464,
      "loss": 1.4772,
      "step": 4109
    },
    {
      "epoch": 0.14718785252565045,
      "grad_norm": 1.3329429626464844,
      "learning_rate": 0.0001928842680350877,
      "loss": 1.4615,
      "step": 4110
    },
    {
      "epoch": 0.14722366465521872,
      "grad_norm": 1.498295783996582,
      "learning_rate": 0.00019287997027301275,
      "loss": 1.5852,
      "step": 4111
    },
    {
      "epoch": 0.147259476784787,
      "grad_norm": 1.641626238822937,
      "learning_rate": 0.00019287567126135763,
      "loss": 1.3915,
      "step": 4112
    },
    {
      "epoch": 0.14729528891435528,
      "grad_norm": 1.6395761966705322,
      "learning_rate": 0.00019287137100018013,
      "loss": 2.055,
      "step": 4113
    },
    {
      "epoch": 0.14733110104392358,
      "grad_norm": 1.7023566961288452,
      "learning_rate": 0.00019286706948953812,
      "loss": 1.8299,
      "step": 4114
    },
    {
      "epoch": 0.14736691317349185,
      "grad_norm": 1.3379522562026978,
      "learning_rate": 0.00019286276672948952,
      "loss": 1.6071,
      "step": 4115
    },
    {
      "epoch": 0.14740272530306014,
      "grad_norm": 1.9350535869598389,
      "learning_rate": 0.00019285846272009213,
      "loss": 1.64,
      "step": 4116
    },
    {
      "epoch": 0.14743853743262844,
      "grad_norm": 1.9311000108718872,
      "learning_rate": 0.00019285415746140392,
      "loss": 1.7087,
      "step": 4117
    },
    {
      "epoch": 0.1474743495621967,
      "grad_norm": 1.5109360218048096,
      "learning_rate": 0.0001928498509534828,
      "loss": 1.5686,
      "step": 4118
    },
    {
      "epoch": 0.147510161691765,
      "grad_norm": 1.3077818155288696,
      "learning_rate": 0.0001928455431963867,
      "loss": 1.488,
      "step": 4119
    },
    {
      "epoch": 0.14754597382133328,
      "grad_norm": 1.6297913789749146,
      "learning_rate": 0.00019284123419017357,
      "loss": 1.8312,
      "step": 4120
    },
    {
      "epoch": 0.14758178595090157,
      "grad_norm": 1.9442793130874634,
      "learning_rate": 0.0001928369239349014,
      "loss": 1.6814,
      "step": 4121
    },
    {
      "epoch": 0.14761759808046984,
      "grad_norm": 1.7997331619262695,
      "learning_rate": 0.00019283261243062817,
      "loss": 1.6751,
      "step": 4122
    },
    {
      "epoch": 0.14765341021003814,
      "grad_norm": 1.4392348527908325,
      "learning_rate": 0.0001928282996774119,
      "loss": 1.5335,
      "step": 4123
    },
    {
      "epoch": 0.14768922233960644,
      "grad_norm": 1.7808204889297485,
      "learning_rate": 0.00019282398567531058,
      "loss": 1.5752,
      "step": 4124
    },
    {
      "epoch": 0.1477250344691747,
      "grad_norm": 1.8829363584518433,
      "learning_rate": 0.00019281967042438227,
      "loss": 1.5472,
      "step": 4125
    },
    {
      "epoch": 0.147760846598743,
      "grad_norm": 1.8297572135925293,
      "learning_rate": 0.000192815353924685,
      "loss": 1.6218,
      "step": 4126
    },
    {
      "epoch": 0.14779665872831127,
      "grad_norm": 1.4455066919326782,
      "learning_rate": 0.0001928110361762769,
      "loss": 1.5734,
      "step": 4127
    },
    {
      "epoch": 0.14783247085787957,
      "grad_norm": 1.5757455825805664,
      "learning_rate": 0.000192806717179216,
      "loss": 1.561,
      "step": 4128
    },
    {
      "epoch": 0.14786828298744784,
      "grad_norm": 1.614942193031311,
      "learning_rate": 0.00019280239693356048,
      "loss": 1.5828,
      "step": 4129
    },
    {
      "epoch": 0.14790409511701613,
      "grad_norm": 1.358445405960083,
      "learning_rate": 0.0001927980754393684,
      "loss": 1.7682,
      "step": 4130
    },
    {
      "epoch": 0.14793990724658443,
      "grad_norm": 2.172544240951538,
      "learning_rate": 0.00019279375269669785,
      "loss": 1.5469,
      "step": 4131
    },
    {
      "epoch": 0.1479757193761527,
      "grad_norm": 1.9252666234970093,
      "learning_rate": 0.00019278942870560713,
      "loss": 1.8746,
      "step": 4132
    },
    {
      "epoch": 0.148011531505721,
      "grad_norm": 1.5914863348007202,
      "learning_rate": 0.0001927851034661543,
      "loss": 1.3982,
      "step": 4133
    },
    {
      "epoch": 0.14804734363528926,
      "grad_norm": 1.750622272491455,
      "learning_rate": 0.0001927807769783976,
      "loss": 1.4232,
      "step": 4134
    },
    {
      "epoch": 0.14808315576485756,
      "grad_norm": 2.3878767490386963,
      "learning_rate": 0.0001927764492423952,
      "loss": 1.6743,
      "step": 4135
    },
    {
      "epoch": 0.14811896789442583,
      "grad_norm": 1.5392639636993408,
      "learning_rate": 0.0001927721202582054,
      "loss": 1.5216,
      "step": 4136
    },
    {
      "epoch": 0.14815478002399413,
      "grad_norm": 1.3905284404754639,
      "learning_rate": 0.00019276779002588634,
      "loss": 1.6186,
      "step": 4137
    },
    {
      "epoch": 0.14819059215356242,
      "grad_norm": 1.1709767580032349,
      "learning_rate": 0.00019276345854549634,
      "loss": 1.6602,
      "step": 4138
    },
    {
      "epoch": 0.1482264042831307,
      "grad_norm": 1.6870055198669434,
      "learning_rate": 0.00019275912581709367,
      "loss": 1.8051,
      "step": 4139
    },
    {
      "epoch": 0.148262216412699,
      "grad_norm": 1.4576029777526855,
      "learning_rate": 0.0001927547918407366,
      "loss": 1.6819,
      "step": 4140
    },
    {
      "epoch": 0.14829802854226726,
      "grad_norm": 1.3755351305007935,
      "learning_rate": 0.00019275045661648344,
      "loss": 1.6577,
      "step": 4141
    },
    {
      "epoch": 0.14833384067183555,
      "grad_norm": 1.6038774251937866,
      "learning_rate": 0.00019274612014439258,
      "loss": 1.6699,
      "step": 4142
    },
    {
      "epoch": 0.14836965280140382,
      "grad_norm": 1.760777473449707,
      "learning_rate": 0.00019274178242452224,
      "loss": 1.6629,
      "step": 4143
    },
    {
      "epoch": 0.14840546493097212,
      "grad_norm": 2.301041841506958,
      "learning_rate": 0.0001927374434569309,
      "loss": 1.7042,
      "step": 4144
    },
    {
      "epoch": 0.14844127706054042,
      "grad_norm": 1.588143229484558,
      "learning_rate": 0.00019273310324167687,
      "loss": 1.8193,
      "step": 4145
    },
    {
      "epoch": 0.14847708919010869,
      "grad_norm": 1.4113539457321167,
      "learning_rate": 0.00019272876177881852,
      "loss": 1.3863,
      "step": 4146
    },
    {
      "epoch": 0.14851290131967698,
      "grad_norm": 2.251722812652588,
      "learning_rate": 0.00019272441906841432,
      "loss": 1.5958,
      "step": 4147
    },
    {
      "epoch": 0.14854871344924525,
      "grad_norm": 1.7772836685180664,
      "learning_rate": 0.00019272007511052266,
      "loss": 1.7736,
      "step": 4148
    },
    {
      "epoch": 0.14858452557881355,
      "grad_norm": 1.554674744606018,
      "learning_rate": 0.000192715729905202,
      "loss": 1.6975,
      "step": 4149
    },
    {
      "epoch": 0.14862033770838182,
      "grad_norm": 1.403333306312561,
      "learning_rate": 0.00019271138345251077,
      "loss": 1.7412,
      "step": 4150
    },
    {
      "epoch": 0.1486561498379501,
      "grad_norm": 1.7260535955429077,
      "learning_rate": 0.00019270703575250748,
      "loss": 1.5755,
      "step": 4151
    },
    {
      "epoch": 0.1486919619675184,
      "grad_norm": 2.66255784034729,
      "learning_rate": 0.0001927026868052506,
      "loss": 1.34,
      "step": 4152
    },
    {
      "epoch": 0.14872777409708668,
      "grad_norm": 1.3532320261001587,
      "learning_rate": 0.00019269833661079866,
      "loss": 1.6399,
      "step": 4153
    },
    {
      "epoch": 0.14876358622665498,
      "grad_norm": 3.2480742931365967,
      "learning_rate": 0.00019269398516921015,
      "loss": 2.3012,
      "step": 4154
    },
    {
      "epoch": 0.14879939835622324,
      "grad_norm": 1.9341964721679688,
      "learning_rate": 0.00019268963248054367,
      "loss": 1.818,
      "step": 4155
    },
    {
      "epoch": 0.14883521048579154,
      "grad_norm": 1.7220872640609741,
      "learning_rate": 0.00019268527854485773,
      "loss": 1.5048,
      "step": 4156
    },
    {
      "epoch": 0.1488710226153598,
      "grad_norm": 1.5572283267974854,
      "learning_rate": 0.0001926809233622109,
      "loss": 1.9191,
      "step": 4157
    },
    {
      "epoch": 0.1489068347449281,
      "grad_norm": 2.1267168521881104,
      "learning_rate": 0.0001926765669326618,
      "loss": 1.3807,
      "step": 4158
    },
    {
      "epoch": 0.1489426468744964,
      "grad_norm": 2.1267030239105225,
      "learning_rate": 0.00019267220925626907,
      "loss": 1.7096,
      "step": 4159
    },
    {
      "epoch": 0.14897845900406467,
      "grad_norm": 1.8242478370666504,
      "learning_rate": 0.00019266785033309128,
      "loss": 1.2927,
      "step": 4160
    },
    {
      "epoch": 0.14901427113363297,
      "grad_norm": 1.9160438776016235,
      "learning_rate": 0.0001926634901631871,
      "loss": 1.6361,
      "step": 4161
    },
    {
      "epoch": 0.14905008326320124,
      "grad_norm": 2.731289863586426,
      "learning_rate": 0.00019265912874661515,
      "loss": 1.3703,
      "step": 4162
    },
    {
      "epoch": 0.14908589539276954,
      "grad_norm": 1.2577362060546875,
      "learning_rate": 0.0001926547660834342,
      "loss": 1.7034,
      "step": 4163
    },
    {
      "epoch": 0.1491217075223378,
      "grad_norm": 2.197948932647705,
      "learning_rate": 0.00019265040217370286,
      "loss": 1.4072,
      "step": 4164
    },
    {
      "epoch": 0.1491575196519061,
      "grad_norm": 1.9935760498046875,
      "learning_rate": 0.0001926460370174799,
      "loss": 1.3064,
      "step": 4165
    },
    {
      "epoch": 0.1491933317814744,
      "grad_norm": 1.7972018718719482,
      "learning_rate": 0.00019264167061482397,
      "loss": 1.4971,
      "step": 4166
    },
    {
      "epoch": 0.14922914391104267,
      "grad_norm": 1.4733848571777344,
      "learning_rate": 0.0001926373029657939,
      "loss": 1.6895,
      "step": 4167
    },
    {
      "epoch": 0.14926495604061096,
      "grad_norm": 1.9356968402862549,
      "learning_rate": 0.00019263293407044838,
      "loss": 1.792,
      "step": 4168
    },
    {
      "epoch": 0.14930076817017923,
      "grad_norm": 1.5070140361785889,
      "learning_rate": 0.00019262856392884625,
      "loss": 1.6962,
      "step": 4169
    },
    {
      "epoch": 0.14933658029974753,
      "grad_norm": 1.9698539972305298,
      "learning_rate": 0.00019262419254104628,
      "loss": 1.7214,
      "step": 4170
    },
    {
      "epoch": 0.1493723924293158,
      "grad_norm": 1.7777533531188965,
      "learning_rate": 0.00019261981990710723,
      "loss": 1.6434,
      "step": 4171
    },
    {
      "epoch": 0.1494082045588841,
      "grad_norm": 1.478132963180542,
      "learning_rate": 0.000192615446027088,
      "loss": 1.493,
      "step": 4172
    },
    {
      "epoch": 0.1494440166884524,
      "grad_norm": 1.4548566341400146,
      "learning_rate": 0.00019261107090104743,
      "loss": 1.5991,
      "step": 4173
    },
    {
      "epoch": 0.14947982881802066,
      "grad_norm": 1.5640240907669067,
      "learning_rate": 0.00019260669452904433,
      "loss": 1.3423,
      "step": 4174
    },
    {
      "epoch": 0.14951564094758896,
      "grad_norm": 1.3845349550247192,
      "learning_rate": 0.00019260231691113763,
      "loss": 1.8214,
      "step": 4175
    },
    {
      "epoch": 0.14955145307715723,
      "grad_norm": 3.009308099746704,
      "learning_rate": 0.00019259793804738619,
      "loss": 1.6082,
      "step": 4176
    },
    {
      "epoch": 0.14958726520672552,
      "grad_norm": 1.3763492107391357,
      "learning_rate": 0.00019259355793784893,
      "loss": 1.6317,
      "step": 4177
    },
    {
      "epoch": 0.1496230773362938,
      "grad_norm": 1.6250135898590088,
      "learning_rate": 0.00019258917658258483,
      "loss": 1.7019,
      "step": 4178
    },
    {
      "epoch": 0.1496588894658621,
      "grad_norm": 1.358526349067688,
      "learning_rate": 0.00019258479398165273,
      "loss": 1.6496,
      "step": 4179
    },
    {
      "epoch": 0.14969470159543039,
      "grad_norm": 2.3600802421569824,
      "learning_rate": 0.00019258041013511167,
      "loss": 1.5639,
      "step": 4180
    },
    {
      "epoch": 0.14973051372499865,
      "grad_norm": 2.442429542541504,
      "learning_rate": 0.00019257602504302063,
      "loss": 1.8827,
      "step": 4181
    },
    {
      "epoch": 0.14976632585456695,
      "grad_norm": 2.7102713584899902,
      "learning_rate": 0.0001925716387054386,
      "loss": 1.8546,
      "step": 4182
    },
    {
      "epoch": 0.14980213798413522,
      "grad_norm": 2.178633689880371,
      "learning_rate": 0.00019256725112242455,
      "loss": 2.0248,
      "step": 4183
    },
    {
      "epoch": 0.14983795011370352,
      "grad_norm": 1.9340856075286865,
      "learning_rate": 0.00019256286229403754,
      "loss": 1.5944,
      "step": 4184
    },
    {
      "epoch": 0.14987376224327179,
      "grad_norm": 1.5424754619598389,
      "learning_rate": 0.00019255847222033663,
      "loss": 1.711,
      "step": 4185
    },
    {
      "epoch": 0.14990957437284008,
      "grad_norm": 2.0131659507751465,
      "learning_rate": 0.00019255408090138086,
      "loss": 1.7849,
      "step": 4186
    },
    {
      "epoch": 0.14994538650240838,
      "grad_norm": 1.3488818407058716,
      "learning_rate": 0.00019254968833722934,
      "loss": 1.567,
      "step": 4187
    },
    {
      "epoch": 0.14998119863197665,
      "grad_norm": 1.4595369100570679,
      "learning_rate": 0.0001925452945279411,
      "loss": 1.5943,
      "step": 4188
    },
    {
      "epoch": 0.15001701076154494,
      "grad_norm": 2.130014419555664,
      "learning_rate": 0.00019254089947357534,
      "loss": 1.6016,
      "step": 4189
    },
    {
      "epoch": 0.1500528228911132,
      "grad_norm": 1.4605454206466675,
      "learning_rate": 0.00019253650317419113,
      "loss": 1.6543,
      "step": 4190
    },
    {
      "epoch": 0.1500886350206815,
      "grad_norm": 1.6974399089813232,
      "learning_rate": 0.0001925321056298476,
      "loss": 1.473,
      "step": 4191
    },
    {
      "epoch": 0.15012444715024978,
      "grad_norm": 1.9349365234375,
      "learning_rate": 0.000192527706840604,
      "loss": 1.4904,
      "step": 4192
    },
    {
      "epoch": 0.15016025927981808,
      "grad_norm": 1.4719817638397217,
      "learning_rate": 0.00019252330680651945,
      "loss": 1.6002,
      "step": 4193
    },
    {
      "epoch": 0.15019607140938637,
      "grad_norm": 1.5579456090927124,
      "learning_rate": 0.0001925189055276531,
      "loss": 1.2423,
      "step": 4194
    },
    {
      "epoch": 0.15023188353895464,
      "grad_norm": 1.1475658416748047,
      "learning_rate": 0.00019251450300406426,
      "loss": 1.4654,
      "step": 4195
    },
    {
      "epoch": 0.15026769566852294,
      "grad_norm": 2.370964527130127,
      "learning_rate": 0.00019251009923581213,
      "loss": 1.782,
      "step": 4196
    },
    {
      "epoch": 0.1503035077980912,
      "grad_norm": 1.2127045392990112,
      "learning_rate": 0.0001925056942229559,
      "loss": 1.5602,
      "step": 4197
    },
    {
      "epoch": 0.1503393199276595,
      "grad_norm": 1.3405200242996216,
      "learning_rate": 0.00019250128796555492,
      "loss": 1.7569,
      "step": 4198
    },
    {
      "epoch": 0.15037513205722777,
      "grad_norm": 1.5121461153030396,
      "learning_rate": 0.0001924968804636684,
      "loss": 1.5301,
      "step": 4199
    },
    {
      "epoch": 0.15041094418679607,
      "grad_norm": 1.8312376737594604,
      "learning_rate": 0.0001924924717173557,
      "loss": 1.5413,
      "step": 4200
    },
    {
      "epoch": 0.15044675631636437,
      "grad_norm": 1.5825523138046265,
      "learning_rate": 0.00019248806172667606,
      "loss": 1.7082,
      "step": 4201
    },
    {
      "epoch": 0.15048256844593264,
      "grad_norm": 1.664043664932251,
      "learning_rate": 0.00019248365049168888,
      "loss": 1.6665,
      "step": 4202
    },
    {
      "epoch": 0.15051838057550093,
      "grad_norm": 1.4566935300827026,
      "learning_rate": 0.00019247923801245345,
      "loss": 1.4875,
      "step": 4203
    },
    {
      "epoch": 0.1505541927050692,
      "grad_norm": 1.3645408153533936,
      "learning_rate": 0.0001924748242890292,
      "loss": 1.578,
      "step": 4204
    },
    {
      "epoch": 0.1505900048346375,
      "grad_norm": 1.360119104385376,
      "learning_rate": 0.00019247040932147546,
      "loss": 1.6422,
      "step": 4205
    },
    {
      "epoch": 0.15062581696420577,
      "grad_norm": 1.9559032917022705,
      "learning_rate": 0.00019246599310985163,
      "loss": 1.64,
      "step": 4206
    },
    {
      "epoch": 0.15066162909377406,
      "grad_norm": 1.3689755201339722,
      "learning_rate": 0.0001924615756542171,
      "loss": 1.5453,
      "step": 4207
    },
    {
      "epoch": 0.15069744122334236,
      "grad_norm": 2.615208864212036,
      "learning_rate": 0.0001924571569546314,
      "loss": 1.7152,
      "step": 4208
    },
    {
      "epoch": 0.15073325335291063,
      "grad_norm": 1.3856520652770996,
      "learning_rate": 0.00019245273701115387,
      "loss": 1.5176,
      "step": 4209
    },
    {
      "epoch": 0.15076906548247893,
      "grad_norm": 1.9545027017593384,
      "learning_rate": 0.00019244831582384406,
      "loss": 1.38,
      "step": 4210
    },
    {
      "epoch": 0.1508048776120472,
      "grad_norm": 1.9589897394180298,
      "learning_rate": 0.0001924438933927614,
      "loss": 1.5865,
      "step": 4211
    },
    {
      "epoch": 0.1508406897416155,
      "grad_norm": 1.9281854629516602,
      "learning_rate": 0.00019243946971796535,
      "loss": 1.2587,
      "step": 4212
    },
    {
      "epoch": 0.15087650187118376,
      "grad_norm": 2.179154396057129,
      "learning_rate": 0.00019243504479951552,
      "loss": 1.6455,
      "step": 4213
    },
    {
      "epoch": 0.15091231400075206,
      "grad_norm": 1.6948529481887817,
      "learning_rate": 0.00019243061863747138,
      "loss": 1.5106,
      "step": 4214
    },
    {
      "epoch": 0.15094812613032033,
      "grad_norm": 1.5750123262405396,
      "learning_rate": 0.0001924261912318925,
      "loss": 1.7466,
      "step": 4215
    },
    {
      "epoch": 0.15098393825988862,
      "grad_norm": 1.5252922773361206,
      "learning_rate": 0.00019242176258283845,
      "loss": 1.4174,
      "step": 4216
    },
    {
      "epoch": 0.15101975038945692,
      "grad_norm": 1.4911702871322632,
      "learning_rate": 0.00019241733269036878,
      "loss": 1.792,
      "step": 4217
    },
    {
      "epoch": 0.1510555625190252,
      "grad_norm": 2.6139893531799316,
      "learning_rate": 0.0001924129015545431,
      "loss": 1.496,
      "step": 4218
    },
    {
      "epoch": 0.15109137464859348,
      "grad_norm": 1.4543615579605103,
      "learning_rate": 0.00019240846917542107,
      "loss": 1.3647,
      "step": 4219
    },
    {
      "epoch": 0.15112718677816175,
      "grad_norm": 2.1806602478027344,
      "learning_rate": 0.00019240403555306225,
      "loss": 1.8613,
      "step": 4220
    },
    {
      "epoch": 0.15116299890773005,
      "grad_norm": 1.576529622077942,
      "learning_rate": 0.00019239960068752633,
      "loss": 1.5466,
      "step": 4221
    },
    {
      "epoch": 0.15119881103729832,
      "grad_norm": 1.780757188796997,
      "learning_rate": 0.00019239516457887298,
      "loss": 1.732,
      "step": 4222
    },
    {
      "epoch": 0.15123462316686662,
      "grad_norm": 1.6536751985549927,
      "learning_rate": 0.00019239072722716186,
      "loss": 1.5458,
      "step": 4223
    },
    {
      "epoch": 0.1512704352964349,
      "grad_norm": 3.4212424755096436,
      "learning_rate": 0.0001923862886324527,
      "loss": 1.6249,
      "step": 4224
    },
    {
      "epoch": 0.15130624742600318,
      "grad_norm": 2.4516830444335938,
      "learning_rate": 0.00019238184879480518,
      "loss": 1.722,
      "step": 4225
    },
    {
      "epoch": 0.15134205955557148,
      "grad_norm": 1.8271950483322144,
      "learning_rate": 0.00019237740771427906,
      "loss": 1.3643,
      "step": 4226
    },
    {
      "epoch": 0.15137787168513975,
      "grad_norm": 2.7619059085845947,
      "learning_rate": 0.00019237296539093408,
      "loss": 1.786,
      "step": 4227
    },
    {
      "epoch": 0.15141368381470804,
      "grad_norm": 2.0190656185150146,
      "learning_rate": 0.00019236852182482998,
      "loss": 1.1294,
      "step": 4228
    },
    {
      "epoch": 0.1514494959442763,
      "grad_norm": 2.5435988903045654,
      "learning_rate": 0.0001923640770160266,
      "loss": 1.965,
      "step": 4229
    },
    {
      "epoch": 0.1514853080738446,
      "grad_norm": 1.3937095403671265,
      "learning_rate": 0.00019235963096458366,
      "loss": 1.7719,
      "step": 4230
    },
    {
      "epoch": 0.1515211202034129,
      "grad_norm": 1.7603644132614136,
      "learning_rate": 0.00019235518367056106,
      "loss": 1.9055,
      "step": 4231
    },
    {
      "epoch": 0.15155693233298118,
      "grad_norm": 1.535352349281311,
      "learning_rate": 0.0001923507351340186,
      "loss": 1.7588,
      "step": 4232
    },
    {
      "epoch": 0.15159274446254947,
      "grad_norm": 1.4259074926376343,
      "learning_rate": 0.00019234628535501607,
      "loss": 1.5549,
      "step": 4233
    },
    {
      "epoch": 0.15162855659211774,
      "grad_norm": 1.2318248748779297,
      "learning_rate": 0.00019234183433361344,
      "loss": 1.6689,
      "step": 4234
    },
    {
      "epoch": 0.15166436872168604,
      "grad_norm": 1.5642166137695312,
      "learning_rate": 0.0001923373820698705,
      "loss": 1.5143,
      "step": 4235
    },
    {
      "epoch": 0.1517001808512543,
      "grad_norm": 2.0307435989379883,
      "learning_rate": 0.00019233292856384723,
      "loss": 1.6097,
      "step": 4236
    },
    {
      "epoch": 0.1517359929808226,
      "grad_norm": 1.8250125646591187,
      "learning_rate": 0.00019232847381560347,
      "loss": 1.7707,
      "step": 4237
    },
    {
      "epoch": 0.1517718051103909,
      "grad_norm": 1.4493552446365356,
      "learning_rate": 0.00019232401782519923,
      "loss": 1.683,
      "step": 4238
    },
    {
      "epoch": 0.15180761723995917,
      "grad_norm": 1.6072957515716553,
      "learning_rate": 0.0001923195605926944,
      "loss": 1.5616,
      "step": 4239
    },
    {
      "epoch": 0.15184342936952747,
      "grad_norm": 2.072479724884033,
      "learning_rate": 0.00019231510211814896,
      "loss": 1.5442,
      "step": 4240
    },
    {
      "epoch": 0.15187924149909574,
      "grad_norm": 2.20919132232666,
      "learning_rate": 0.0001923106424016229,
      "loss": 1.8976,
      "step": 4241
    },
    {
      "epoch": 0.15191505362866403,
      "grad_norm": 1.0948854684829712,
      "learning_rate": 0.00019230618144317624,
      "loss": 1.4725,
      "step": 4242
    },
    {
      "epoch": 0.1519508657582323,
      "grad_norm": 1.5968300104141235,
      "learning_rate": 0.00019230171924286896,
      "loss": 1.4642,
      "step": 4243
    },
    {
      "epoch": 0.1519866778878006,
      "grad_norm": 1.6487561464309692,
      "learning_rate": 0.0001922972558007611,
      "loss": 1.7243,
      "step": 4244
    },
    {
      "epoch": 0.1520224900173689,
      "grad_norm": 1.4044201374053955,
      "learning_rate": 0.00019229279111691272,
      "loss": 1.6046,
      "step": 4245
    },
    {
      "epoch": 0.15205830214693716,
      "grad_norm": 1.3864117860794067,
      "learning_rate": 0.0001922883251913839,
      "loss": 1.6748,
      "step": 4246
    },
    {
      "epoch": 0.15209411427650546,
      "grad_norm": 2.1303157806396484,
      "learning_rate": 0.00019228385802423469,
      "loss": 1.7468,
      "step": 4247
    },
    {
      "epoch": 0.15212992640607373,
      "grad_norm": 2.085751533508301,
      "learning_rate": 0.0001922793896155252,
      "loss": 1.9624,
      "step": 4248
    },
    {
      "epoch": 0.15216573853564203,
      "grad_norm": 1.5157827138900757,
      "learning_rate": 0.00019227491996531558,
      "loss": 1.6942,
      "step": 4249
    },
    {
      "epoch": 0.1522015506652103,
      "grad_norm": 2.0757040977478027,
      "learning_rate": 0.00019227044907366595,
      "loss": 1.4183,
      "step": 4250
    },
    {
      "epoch": 0.1522373627947786,
      "grad_norm": 1.8636008501052856,
      "learning_rate": 0.00019226597694063638,
      "loss": 1.5945,
      "step": 4251
    },
    {
      "epoch": 0.1522731749243469,
      "grad_norm": 1.7461687326431274,
      "learning_rate": 0.0001922615035662872,
      "loss": 1.5256,
      "step": 4252
    },
    {
      "epoch": 0.15230898705391516,
      "grad_norm": 1.3671756982803345,
      "learning_rate": 0.00019225702895067843,
      "loss": 1.4619,
      "step": 4253
    },
    {
      "epoch": 0.15234479918348345,
      "grad_norm": 1.5393552780151367,
      "learning_rate": 0.00019225255309387036,
      "loss": 1.728,
      "step": 4254
    },
    {
      "epoch": 0.15238061131305172,
      "grad_norm": 2.7406222820281982,
      "learning_rate": 0.00019224807599592318,
      "loss": 1.7323,
      "step": 4255
    },
    {
      "epoch": 0.15241642344262002,
      "grad_norm": 1.4855775833129883,
      "learning_rate": 0.00019224359765689713,
      "loss": 1.7204,
      "step": 4256
    },
    {
      "epoch": 0.1524522355721883,
      "grad_norm": 1.3277029991149902,
      "learning_rate": 0.00019223911807685244,
      "loss": 1.5232,
      "step": 4257
    },
    {
      "epoch": 0.15248804770175658,
      "grad_norm": 1.8301455974578857,
      "learning_rate": 0.00019223463725584944,
      "loss": 1.5253,
      "step": 4258
    },
    {
      "epoch": 0.15252385983132488,
      "grad_norm": 1.7147241830825806,
      "learning_rate": 0.00019223015519394834,
      "loss": 1.6035,
      "step": 4259
    },
    {
      "epoch": 0.15255967196089315,
      "grad_norm": 2.0909276008605957,
      "learning_rate": 0.00019222567189120947,
      "loss": 1.3304,
      "step": 4260
    },
    {
      "epoch": 0.15259548409046145,
      "grad_norm": 2.0987610816955566,
      "learning_rate": 0.00019222118734769317,
      "loss": 1.3313,
      "step": 4261
    },
    {
      "epoch": 0.15263129622002972,
      "grad_norm": 2.262599229812622,
      "learning_rate": 0.00019221670156345971,
      "loss": 1.5674,
      "step": 4262
    },
    {
      "epoch": 0.152667108349598,
      "grad_norm": 1.5876708030700684,
      "learning_rate": 0.00019221221453856954,
      "loss": 1.7479,
      "step": 4263
    },
    {
      "epoch": 0.15270292047916628,
      "grad_norm": 1.6196600198745728,
      "learning_rate": 0.00019220772627308292,
      "loss": 1.5288,
      "step": 4264
    },
    {
      "epoch": 0.15273873260873458,
      "grad_norm": 1.6314212083816528,
      "learning_rate": 0.00019220323676706028,
      "loss": 1.6841,
      "step": 4265
    },
    {
      "epoch": 0.15277454473830288,
      "grad_norm": 2.1747360229492188,
      "learning_rate": 0.00019219874602056204,
      "loss": 1.8169,
      "step": 4266
    },
    {
      "epoch": 0.15281035686787114,
      "grad_norm": 1.5729329586029053,
      "learning_rate": 0.0001921942540336486,
      "loss": 1.5527,
      "step": 4267
    },
    {
      "epoch": 0.15284616899743944,
      "grad_norm": 2.512627124786377,
      "learning_rate": 0.00019218976080638043,
      "loss": 2.074,
      "step": 4268
    },
    {
      "epoch": 0.1528819811270077,
      "grad_norm": 1.6818349361419678,
      "learning_rate": 0.0001921852663388179,
      "loss": 1.5731,
      "step": 4269
    },
    {
      "epoch": 0.152917793256576,
      "grad_norm": 1.7092454433441162,
      "learning_rate": 0.0001921807706310215,
      "loss": 1.5929,
      "step": 4270
    },
    {
      "epoch": 0.15295360538614428,
      "grad_norm": 1.1693865060806274,
      "learning_rate": 0.00019217627368305176,
      "loss": 1.7299,
      "step": 4271
    },
    {
      "epoch": 0.15298941751571257,
      "grad_norm": 1.8270220756530762,
      "learning_rate": 0.0001921717754949692,
      "loss": 1.6689,
      "step": 4272
    },
    {
      "epoch": 0.15302522964528087,
      "grad_norm": 1.5601493120193481,
      "learning_rate": 0.00019216727606683425,
      "loss": 1.5714,
      "step": 4273
    },
    {
      "epoch": 0.15306104177484914,
      "grad_norm": 1.6041169166564941,
      "learning_rate": 0.00019216277539870752,
      "loss": 1.5596,
      "step": 4274
    },
    {
      "epoch": 0.15309685390441743,
      "grad_norm": 1.5292716026306152,
      "learning_rate": 0.00019215827349064948,
      "loss": 1.6235,
      "step": 4275
    },
    {
      "epoch": 0.1531326660339857,
      "grad_norm": 2.1486856937408447,
      "learning_rate": 0.00019215377034272074,
      "loss": 1.7512,
      "step": 4276
    },
    {
      "epoch": 0.153168478163554,
      "grad_norm": 1.281849980354309,
      "learning_rate": 0.00019214926595498196,
      "loss": 1.5935,
      "step": 4277
    },
    {
      "epoch": 0.15320429029312227,
      "grad_norm": 2.1205649375915527,
      "learning_rate": 0.0001921447603274936,
      "loss": 1.6512,
      "step": 4278
    },
    {
      "epoch": 0.15324010242269057,
      "grad_norm": 1.811474323272705,
      "learning_rate": 0.0001921402534603164,
      "loss": 1.7293,
      "step": 4279
    },
    {
      "epoch": 0.15327591455225886,
      "grad_norm": 1.8452061414718628,
      "learning_rate": 0.00019213574535351092,
      "loss": 1.6552,
      "step": 4280
    },
    {
      "epoch": 0.15331172668182713,
      "grad_norm": 2.1379551887512207,
      "learning_rate": 0.00019213123600713783,
      "loss": 1.6141,
      "step": 4281
    },
    {
      "epoch": 0.15334753881139543,
      "grad_norm": 1.5196058750152588,
      "learning_rate": 0.0001921267254212578,
      "loss": 1.5217,
      "step": 4282
    },
    {
      "epoch": 0.1533833509409637,
      "grad_norm": 1.5301889181137085,
      "learning_rate": 0.00019212221359593152,
      "loss": 1.2763,
      "step": 4283
    },
    {
      "epoch": 0.153419163070532,
      "grad_norm": 1.7867982387542725,
      "learning_rate": 0.00019211770053121968,
      "loss": 1.7346,
      "step": 4284
    },
    {
      "epoch": 0.15345497520010026,
      "grad_norm": 2.244570016860962,
      "learning_rate": 0.000192113186227183,
      "loss": 1.5217,
      "step": 4285
    },
    {
      "epoch": 0.15349078732966856,
      "grad_norm": 2.4065821170806885,
      "learning_rate": 0.0001921086706838822,
      "loss": 1.4745,
      "step": 4286
    },
    {
      "epoch": 0.15352659945923686,
      "grad_norm": 1.6467057466506958,
      "learning_rate": 0.0001921041539013781,
      "loss": 1.5964,
      "step": 4287
    },
    {
      "epoch": 0.15356241158880513,
      "grad_norm": 1.5443044900894165,
      "learning_rate": 0.00019209963587973138,
      "loss": 1.8898,
      "step": 4288
    },
    {
      "epoch": 0.15359822371837342,
      "grad_norm": 2.1365907192230225,
      "learning_rate": 0.00019209511661900285,
      "loss": 1.4256,
      "step": 4289
    },
    {
      "epoch": 0.1536340358479417,
      "grad_norm": 1.4355337619781494,
      "learning_rate": 0.00019209059611925336,
      "loss": 1.4913,
      "step": 4290
    },
    {
      "epoch": 0.15366984797751,
      "grad_norm": 1.4051859378814697,
      "learning_rate": 0.00019208607438054364,
      "loss": 1.547,
      "step": 4291
    },
    {
      "epoch": 0.15370566010707826,
      "grad_norm": 1.3148490190505981,
      "learning_rate": 0.0001920815514029346,
      "loss": 1.7636,
      "step": 4292
    },
    {
      "epoch": 0.15374147223664655,
      "grad_norm": 1.518978476524353,
      "learning_rate": 0.00019207702718648705,
      "loss": 1.4276,
      "step": 4293
    },
    {
      "epoch": 0.15377728436621485,
      "grad_norm": 1.5426995754241943,
      "learning_rate": 0.00019207250173126187,
      "loss": 1.5931,
      "step": 4294
    },
    {
      "epoch": 0.15381309649578312,
      "grad_norm": 1.3305436372756958,
      "learning_rate": 0.00019206797503731996,
      "loss": 1.5414,
      "step": 4295
    },
    {
      "epoch": 0.15384890862535142,
      "grad_norm": 1.3565562963485718,
      "learning_rate": 0.0001920634471047222,
      "loss": 1.4051,
      "step": 4296
    },
    {
      "epoch": 0.15388472075491968,
      "grad_norm": 1.4529234170913696,
      "learning_rate": 0.0001920589179335295,
      "loss": 1.7174,
      "step": 4297
    },
    {
      "epoch": 0.15392053288448798,
      "grad_norm": 1.6896700859069824,
      "learning_rate": 0.00019205438752380283,
      "loss": 1.5858,
      "step": 4298
    },
    {
      "epoch": 0.15395634501405625,
      "grad_norm": 2.139944314956665,
      "learning_rate": 0.00019204985587560307,
      "loss": 1.871,
      "step": 4299
    },
    {
      "epoch": 0.15399215714362455,
      "grad_norm": 2.2128217220306396,
      "learning_rate": 0.00019204532298899127,
      "loss": 1.7976,
      "step": 4300
    },
    {
      "epoch": 0.15402796927319284,
      "grad_norm": 1.833006501197815,
      "learning_rate": 0.0001920407888640284,
      "loss": 1.8781,
      "step": 4301
    },
    {
      "epoch": 0.1540637814027611,
      "grad_norm": 1.8609815835952759,
      "learning_rate": 0.00019203625350077541,
      "loss": 1.89,
      "step": 4302
    },
    {
      "epoch": 0.1540995935323294,
      "grad_norm": 1.195251703262329,
      "learning_rate": 0.00019203171689929333,
      "loss": 1.6183,
      "step": 4303
    },
    {
      "epoch": 0.15413540566189768,
      "grad_norm": 1.212415099143982,
      "learning_rate": 0.00019202717905964325,
      "loss": 1.4988,
      "step": 4304
    },
    {
      "epoch": 0.15417121779146598,
      "grad_norm": 1.2291085720062256,
      "learning_rate": 0.00019202263998188617,
      "loss": 1.6022,
      "step": 4305
    },
    {
      "epoch": 0.15420702992103424,
      "grad_norm": 1.3216090202331543,
      "learning_rate": 0.00019201809966608316,
      "loss": 1.5615,
      "step": 4306
    },
    {
      "epoch": 0.15424284205060254,
      "grad_norm": 2.6345887184143066,
      "learning_rate": 0.0001920135581122953,
      "loss": 1.4008,
      "step": 4307
    },
    {
      "epoch": 0.1542786541801708,
      "grad_norm": 2.237443208694458,
      "learning_rate": 0.00019200901532058376,
      "loss": 1.5771,
      "step": 4308
    },
    {
      "epoch": 0.1543144663097391,
      "grad_norm": 1.5229663848876953,
      "learning_rate": 0.00019200447129100954,
      "loss": 1.4401,
      "step": 4309
    },
    {
      "epoch": 0.1543502784393074,
      "grad_norm": 1.4248111248016357,
      "learning_rate": 0.00019199992602363385,
      "loss": 1.6815,
      "step": 4310
    },
    {
      "epoch": 0.15438609056887567,
      "grad_norm": 1.6188117265701294,
      "learning_rate": 0.00019199537951851788,
      "loss": 1.7073,
      "step": 4311
    },
    {
      "epoch": 0.15442190269844397,
      "grad_norm": 1.5151900053024292,
      "learning_rate": 0.0001919908317757227,
      "loss": 1.5666,
      "step": 4312
    },
    {
      "epoch": 0.15445771482801224,
      "grad_norm": 2.0137276649475098,
      "learning_rate": 0.00019198628279530952,
      "loss": 1.9482,
      "step": 4313
    },
    {
      "epoch": 0.15449352695758053,
      "grad_norm": 2.2357749938964844,
      "learning_rate": 0.00019198173257733961,
      "loss": 1.5095,
      "step": 4314
    },
    {
      "epoch": 0.1545293390871488,
      "grad_norm": 1.722032070159912,
      "learning_rate": 0.00019197718112187409,
      "loss": 1.2756,
      "step": 4315
    },
    {
      "epoch": 0.1545651512167171,
      "grad_norm": 1.9286119937896729,
      "learning_rate": 0.00019197262842897425,
      "loss": 1.2332,
      "step": 4316
    },
    {
      "epoch": 0.1546009633462854,
      "grad_norm": 1.6464585065841675,
      "learning_rate": 0.00019196807449870133,
      "loss": 1.768,
      "step": 4317
    },
    {
      "epoch": 0.15463677547585367,
      "grad_norm": 1.3432539701461792,
      "learning_rate": 0.00019196351933111662,
      "loss": 1.3892,
      "step": 4318
    },
    {
      "epoch": 0.15467258760542196,
      "grad_norm": 1.8034002780914307,
      "learning_rate": 0.00019195896292628138,
      "loss": 1.5942,
      "step": 4319
    },
    {
      "epoch": 0.15470839973499023,
      "grad_norm": 2.2822370529174805,
      "learning_rate": 0.00019195440528425688,
      "loss": 1.6211,
      "step": 4320
    },
    {
      "epoch": 0.15474421186455853,
      "grad_norm": 1.6064921617507935,
      "learning_rate": 0.00019194984640510447,
      "loss": 1.4712,
      "step": 4321
    },
    {
      "epoch": 0.1547800239941268,
      "grad_norm": 1.5366289615631104,
      "learning_rate": 0.00019194528628888554,
      "loss": 1.847,
      "step": 4322
    },
    {
      "epoch": 0.1548158361236951,
      "grad_norm": 1.3502074480056763,
      "learning_rate": 0.00019194072493566134,
      "loss": 1.9242,
      "step": 4323
    },
    {
      "epoch": 0.1548516482532634,
      "grad_norm": 1.394463300704956,
      "learning_rate": 0.00019193616234549328,
      "loss": 1.4996,
      "step": 4324
    },
    {
      "epoch": 0.15488746038283166,
      "grad_norm": 1.9019322395324707,
      "learning_rate": 0.00019193159851844276,
      "loss": 1.5957,
      "step": 4325
    },
    {
      "epoch": 0.15492327251239996,
      "grad_norm": 1.9424875974655151,
      "learning_rate": 0.00019192703345457114,
      "loss": 1.6297,
      "step": 4326
    },
    {
      "epoch": 0.15495908464196823,
      "grad_norm": 1.5669649839401245,
      "learning_rate": 0.00019192246715393988,
      "loss": 1.7531,
      "step": 4327
    },
    {
      "epoch": 0.15499489677153652,
      "grad_norm": 1.4756217002868652,
      "learning_rate": 0.0001919178996166104,
      "loss": 1.7796,
      "step": 4328
    },
    {
      "epoch": 0.1550307089011048,
      "grad_norm": 1.9263927936553955,
      "learning_rate": 0.00019191333084264412,
      "loss": 1.907,
      "step": 4329
    },
    {
      "epoch": 0.1550665210306731,
      "grad_norm": 1.990947961807251,
      "learning_rate": 0.00019190876083210258,
      "loss": 1.7693,
      "step": 4330
    },
    {
      "epoch": 0.15510233316024138,
      "grad_norm": 1.4774898290634155,
      "learning_rate": 0.00019190418958504716,
      "loss": 1.6509,
      "step": 4331
    },
    {
      "epoch": 0.15513814528980965,
      "grad_norm": 1.6892197132110596,
      "learning_rate": 0.00019189961710153948,
      "loss": 1.7677,
      "step": 4332
    },
    {
      "epoch": 0.15517395741937795,
      "grad_norm": 1.8601287603378296,
      "learning_rate": 0.00019189504338164095,
      "loss": 1.5163,
      "step": 4333
    },
    {
      "epoch": 0.15520976954894622,
      "grad_norm": 2.1573426723480225,
      "learning_rate": 0.00019189046842541316,
      "loss": 1.9389,
      "step": 4334
    },
    {
      "epoch": 0.15524558167851452,
      "grad_norm": 2.2533397674560547,
      "learning_rate": 0.00019188589223291763,
      "loss": 1.6179,
      "step": 4335
    },
    {
      "epoch": 0.15528139380808278,
      "grad_norm": 1.462187647819519,
      "learning_rate": 0.00019188131480421595,
      "loss": 1.5003,
      "step": 4336
    },
    {
      "epoch": 0.15531720593765108,
      "grad_norm": 1.378979206085205,
      "learning_rate": 0.0001918767361393697,
      "loss": 1.674,
      "step": 4337
    },
    {
      "epoch": 0.15535301806721938,
      "grad_norm": 1.7389543056488037,
      "learning_rate": 0.00019187215623844053,
      "loss": 1.5242,
      "step": 4338
    },
    {
      "epoch": 0.15538883019678765,
      "grad_norm": 1.7959777116775513,
      "learning_rate": 0.00019186757510148995,
      "loss": 1.5021,
      "step": 4339
    },
    {
      "epoch": 0.15542464232635594,
      "grad_norm": 1.8340651988983154,
      "learning_rate": 0.00019186299272857965,
      "loss": 1.6096,
      "step": 4340
    },
    {
      "epoch": 0.1554604544559242,
      "grad_norm": 1.2857201099395752,
      "learning_rate": 0.0001918584091197713,
      "loss": 1.5341,
      "step": 4341
    },
    {
      "epoch": 0.1554962665854925,
      "grad_norm": 1.4234918355941772,
      "learning_rate": 0.00019185382427512653,
      "loss": 1.3857,
      "step": 4342
    },
    {
      "epoch": 0.15553207871506078,
      "grad_norm": 1.4376306533813477,
      "learning_rate": 0.00019184923819470703,
      "loss": 1.683,
      "step": 4343
    },
    {
      "epoch": 0.15556789084462908,
      "grad_norm": 1.7839604616165161,
      "learning_rate": 0.0001918446508785745,
      "loss": 1.4875,
      "step": 4344
    },
    {
      "epoch": 0.15560370297419737,
      "grad_norm": 2.093585968017578,
      "learning_rate": 0.00019184006232679068,
      "loss": 1.708,
      "step": 4345
    },
    {
      "epoch": 0.15563951510376564,
      "grad_norm": 1.6490429639816284,
      "learning_rate": 0.00019183547253941733,
      "loss": 1.4749,
      "step": 4346
    },
    {
      "epoch": 0.15567532723333394,
      "grad_norm": 1.4258878231048584,
      "learning_rate": 0.0001918308815165161,
      "loss": 1.5839,
      "step": 4347
    },
    {
      "epoch": 0.1557111393629022,
      "grad_norm": 1.3602529764175415,
      "learning_rate": 0.0001918262892581488,
      "loss": 1.4586,
      "step": 4348
    },
    {
      "epoch": 0.1557469514924705,
      "grad_norm": 2.043414354324341,
      "learning_rate": 0.00019182169576437724,
      "loss": 1.6374,
      "step": 4349
    },
    {
      "epoch": 0.15578276362203877,
      "grad_norm": 1.3405836820602417,
      "learning_rate": 0.00019181710103526321,
      "loss": 1.7475,
      "step": 4350
    },
    {
      "epoch": 0.15581857575160707,
      "grad_norm": 1.6657061576843262,
      "learning_rate": 0.00019181250507086854,
      "loss": 1.4629,
      "step": 4351
    },
    {
      "epoch": 0.15585438788117537,
      "grad_norm": 1.7391798496246338,
      "learning_rate": 0.00019180790787125504,
      "loss": 1.6842,
      "step": 4352
    },
    {
      "epoch": 0.15589020001074363,
      "grad_norm": 1.8803669214248657,
      "learning_rate": 0.00019180330943648454,
      "loss": 1.7868,
      "step": 4353
    },
    {
      "epoch": 0.15592601214031193,
      "grad_norm": 1.6519395112991333,
      "learning_rate": 0.00019179870976661895,
      "loss": 1.5024,
      "step": 4354
    },
    {
      "epoch": 0.1559618242698802,
      "grad_norm": 1.9108158349990845,
      "learning_rate": 0.0001917941088617201,
      "loss": 1.5709,
      "step": 4355
    },
    {
      "epoch": 0.1559976363994485,
      "grad_norm": 1.362729787826538,
      "learning_rate": 0.00019178950672184996,
      "loss": 1.6681,
      "step": 4356
    },
    {
      "epoch": 0.15603344852901677,
      "grad_norm": 1.1817349195480347,
      "learning_rate": 0.0001917849033470704,
      "loss": 1.6116,
      "step": 4357
    },
    {
      "epoch": 0.15606926065858506,
      "grad_norm": 1.8846251964569092,
      "learning_rate": 0.00019178029873744335,
      "loss": 1.6507,
      "step": 4358
    },
    {
      "epoch": 0.15610507278815336,
      "grad_norm": 1.3459134101867676,
      "learning_rate": 0.00019177569289303078,
      "loss": 1.5352,
      "step": 4359
    },
    {
      "epoch": 0.15614088491772163,
      "grad_norm": 1.8612664937973022,
      "learning_rate": 0.00019177108581389462,
      "loss": 1.7641,
      "step": 4360
    },
    {
      "epoch": 0.15617669704728993,
      "grad_norm": 2.042421817779541,
      "learning_rate": 0.0001917664775000969,
      "loss": 1.5105,
      "step": 4361
    },
    {
      "epoch": 0.1562125091768582,
      "grad_norm": 1.3293653726577759,
      "learning_rate": 0.00019176186795169956,
      "loss": 1.5479,
      "step": 4362
    },
    {
      "epoch": 0.1562483213064265,
      "grad_norm": 1.579801321029663,
      "learning_rate": 0.0001917572571687647,
      "loss": 1.5348,
      "step": 4363
    },
    {
      "epoch": 0.15628413343599476,
      "grad_norm": 1.945090889930725,
      "learning_rate": 0.00019175264515135427,
      "loss": 1.7108,
      "step": 4364
    },
    {
      "epoch": 0.15631994556556306,
      "grad_norm": 1.8579822778701782,
      "learning_rate": 0.00019174803189953035,
      "loss": 2.0244,
      "step": 4365
    },
    {
      "epoch": 0.15635575769513135,
      "grad_norm": 1.6764092445373535,
      "learning_rate": 0.00019174341741335504,
      "loss": 1.4705,
      "step": 4366
    },
    {
      "epoch": 0.15639156982469962,
      "grad_norm": 1.4812837839126587,
      "learning_rate": 0.00019173880169289035,
      "loss": 1.6117,
      "step": 4367
    },
    {
      "epoch": 0.15642738195426792,
      "grad_norm": 1.7119643688201904,
      "learning_rate": 0.00019173418473819844,
      "loss": 1.6282,
      "step": 4368
    },
    {
      "epoch": 0.1564631940838362,
      "grad_norm": 2.8325753211975098,
      "learning_rate": 0.0001917295665493414,
      "loss": 1.4901,
      "step": 4369
    },
    {
      "epoch": 0.15649900621340448,
      "grad_norm": 1.717469334602356,
      "learning_rate": 0.00019172494712638136,
      "loss": 1.7564,
      "step": 4370
    },
    {
      "epoch": 0.15653481834297275,
      "grad_norm": 3.0501840114593506,
      "learning_rate": 0.0001917203264693805,
      "loss": 1.5696,
      "step": 4371
    },
    {
      "epoch": 0.15657063047254105,
      "grad_norm": 2.0719053745269775,
      "learning_rate": 0.0001917157045784009,
      "loss": 1.5447,
      "step": 4372
    },
    {
      "epoch": 0.15660644260210935,
      "grad_norm": 2.070021390914917,
      "learning_rate": 0.00019171108145350484,
      "loss": 1.8622,
      "step": 4373
    },
    {
      "epoch": 0.15664225473167762,
      "grad_norm": 1.451671838760376,
      "learning_rate": 0.00019170645709475447,
      "loss": 1.3056,
      "step": 4374
    },
    {
      "epoch": 0.1566780668612459,
      "grad_norm": 1.7899963855743408,
      "learning_rate": 0.00019170183150221201,
      "loss": 1.3817,
      "step": 4375
    },
    {
      "epoch": 0.15671387899081418,
      "grad_norm": 2.0643718242645264,
      "learning_rate": 0.00019169720467593972,
      "loss": 1.751,
      "step": 4376
    },
    {
      "epoch": 0.15674969112038248,
      "grad_norm": 1.62465238571167,
      "learning_rate": 0.0001916925766159998,
      "loss": 1.5946,
      "step": 4377
    },
    {
      "epoch": 0.15678550324995075,
      "grad_norm": 1.3463857173919678,
      "learning_rate": 0.0001916879473224545,
      "loss": 1.3212,
      "step": 4378
    },
    {
      "epoch": 0.15682131537951904,
      "grad_norm": 1.371046543121338,
      "learning_rate": 0.00019168331679536623,
      "loss": 1.8971,
      "step": 4379
    },
    {
      "epoch": 0.15685712750908734,
      "grad_norm": 1.3901337385177612,
      "learning_rate": 0.00019167868503479712,
      "loss": 1.6782,
      "step": 4380
    },
    {
      "epoch": 0.1568929396386556,
      "grad_norm": 1.584809422492981,
      "learning_rate": 0.00019167405204080956,
      "loss": 1.5364,
      "step": 4381
    },
    {
      "epoch": 0.1569287517682239,
      "grad_norm": 1.3930258750915527,
      "learning_rate": 0.00019166941781346592,
      "loss": 1.2805,
      "step": 4382
    },
    {
      "epoch": 0.15696456389779218,
      "grad_norm": 2.2190914154052734,
      "learning_rate": 0.0001916647823528285,
      "loss": 1.8246,
      "step": 4383
    },
    {
      "epoch": 0.15700037602736047,
      "grad_norm": 2.376316547393799,
      "learning_rate": 0.00019166014565895966,
      "loss": 1.8629,
      "step": 4384
    },
    {
      "epoch": 0.15703618815692874,
      "grad_norm": 1.8719273805618286,
      "learning_rate": 0.0001916555077319218,
      "loss": 1.642,
      "step": 4385
    },
    {
      "epoch": 0.15707200028649704,
      "grad_norm": 1.8520594835281372,
      "learning_rate": 0.0001916508685717773,
      "loss": 1.6127,
      "step": 4386
    },
    {
      "epoch": 0.15710781241606533,
      "grad_norm": 1.3201779127120972,
      "learning_rate": 0.0001916462281785886,
      "loss": 1.5689,
      "step": 4387
    },
    {
      "epoch": 0.1571436245456336,
      "grad_norm": 2.1866772174835205,
      "learning_rate": 0.0001916415865524181,
      "loss": 1.6067,
      "step": 4388
    },
    {
      "epoch": 0.1571794366752019,
      "grad_norm": 1.5497702360153198,
      "learning_rate": 0.00019163694369332825,
      "loss": 1.3363,
      "step": 4389
    },
    {
      "epoch": 0.15721524880477017,
      "grad_norm": 1.541739583015442,
      "learning_rate": 0.00019163229960138156,
      "loss": 1.3411,
      "step": 4390
    },
    {
      "epoch": 0.15725106093433847,
      "grad_norm": 1.7324460744857788,
      "learning_rate": 0.00019162765427664045,
      "loss": 1.6654,
      "step": 4391
    },
    {
      "epoch": 0.15728687306390673,
      "grad_norm": 1.6018710136413574,
      "learning_rate": 0.00019162300771916746,
      "loss": 1.6184,
      "step": 4392
    },
    {
      "epoch": 0.15732268519347503,
      "grad_norm": 1.3000634908676147,
      "learning_rate": 0.00019161835992902507,
      "loss": 1.5956,
      "step": 4393
    },
    {
      "epoch": 0.15735849732304333,
      "grad_norm": 1.8652485609054565,
      "learning_rate": 0.00019161371090627583,
      "loss": 1.5652,
      "step": 4394
    },
    {
      "epoch": 0.1573943094526116,
      "grad_norm": 1.4492467641830444,
      "learning_rate": 0.00019160906065098228,
      "loss": 1.4706,
      "step": 4395
    },
    {
      "epoch": 0.1574301215821799,
      "grad_norm": 1.5934580564498901,
      "learning_rate": 0.00019160440916320698,
      "loss": 1.6347,
      "step": 4396
    },
    {
      "epoch": 0.15746593371174816,
      "grad_norm": 1.5383509397506714,
      "learning_rate": 0.00019159975644301256,
      "loss": 1.7649,
      "step": 4397
    },
    {
      "epoch": 0.15750174584131646,
      "grad_norm": 1.6854779720306396,
      "learning_rate": 0.00019159510249046154,
      "loss": 1.6996,
      "step": 4398
    },
    {
      "epoch": 0.15753755797088473,
      "grad_norm": 1.5617847442626953,
      "learning_rate": 0.00019159044730561656,
      "loss": 1.7046,
      "step": 4399
    },
    {
      "epoch": 0.15757337010045303,
      "grad_norm": 1.5053625106811523,
      "learning_rate": 0.00019158579088854026,
      "loss": 1.6323,
      "step": 4400
    },
    {
      "epoch": 0.15760918223002132,
      "grad_norm": 1.4751994609832764,
      "learning_rate": 0.0001915811332392953,
      "loss": 1.6127,
      "step": 4401
    },
    {
      "epoch": 0.1576449943595896,
      "grad_norm": 1.9280856847763062,
      "learning_rate": 0.00019157647435794428,
      "loss": 1.2965,
      "step": 4402
    },
    {
      "epoch": 0.1576808064891579,
      "grad_norm": 2.083721160888672,
      "learning_rate": 0.00019157181424454996,
      "loss": 1.899,
      "step": 4403
    },
    {
      "epoch": 0.15771661861872616,
      "grad_norm": 1.2885043621063232,
      "learning_rate": 0.00019156715289917497,
      "loss": 1.5413,
      "step": 4404
    },
    {
      "epoch": 0.15775243074829445,
      "grad_norm": 1.6024374961853027,
      "learning_rate": 0.0001915624903218821,
      "loss": 1.4269,
      "step": 4405
    },
    {
      "epoch": 0.15778824287786272,
      "grad_norm": 2.1277174949645996,
      "learning_rate": 0.00019155782651273398,
      "loss": 1.9484,
      "step": 4406
    },
    {
      "epoch": 0.15782405500743102,
      "grad_norm": 1.5377131700515747,
      "learning_rate": 0.00019155316147179342,
      "loss": 1.5133,
      "step": 4407
    },
    {
      "epoch": 0.1578598671369993,
      "grad_norm": 1.6273584365844727,
      "learning_rate": 0.00019154849519912318,
      "loss": 1.7354,
      "step": 4408
    },
    {
      "epoch": 0.15789567926656758,
      "grad_norm": 2.1824471950531006,
      "learning_rate": 0.00019154382769478602,
      "loss": 1.6386,
      "step": 4409
    },
    {
      "epoch": 0.15793149139613588,
      "grad_norm": 1.4899530410766602,
      "learning_rate": 0.00019153915895884474,
      "loss": 1.5148,
      "step": 4410
    },
    {
      "epoch": 0.15796730352570415,
      "grad_norm": 1.584786057472229,
      "learning_rate": 0.00019153448899136212,
      "loss": 1.3899,
      "step": 4411
    },
    {
      "epoch": 0.15800311565527245,
      "grad_norm": 1.3985209465026855,
      "learning_rate": 0.00019152981779240106,
      "loss": 1.7417,
      "step": 4412
    },
    {
      "epoch": 0.15803892778484072,
      "grad_norm": 1.8065505027770996,
      "learning_rate": 0.00019152514536202437,
      "loss": 1.1059,
      "step": 4413
    },
    {
      "epoch": 0.158074739914409,
      "grad_norm": 1.7250769138336182,
      "learning_rate": 0.0001915204717002949,
      "loss": 1.819,
      "step": 4414
    },
    {
      "epoch": 0.15811055204397728,
      "grad_norm": 1.400039553642273,
      "learning_rate": 0.00019151579680727553,
      "loss": 1.586,
      "step": 4415
    },
    {
      "epoch": 0.15814636417354558,
      "grad_norm": 2.593348264694214,
      "learning_rate": 0.00019151112068302917,
      "loss": 1.9007,
      "step": 4416
    },
    {
      "epoch": 0.15818217630311388,
      "grad_norm": 1.6109260320663452,
      "learning_rate": 0.0001915064433276187,
      "loss": 1.4546,
      "step": 4417
    },
    {
      "epoch": 0.15821798843268214,
      "grad_norm": 1.6492670774459839,
      "learning_rate": 0.0001915017647411071,
      "loss": 1.5743,
      "step": 4418
    },
    {
      "epoch": 0.15825380056225044,
      "grad_norm": 2.3043372631073,
      "learning_rate": 0.00019149708492355728,
      "loss": 2.1891,
      "step": 4419
    },
    {
      "epoch": 0.1582896126918187,
      "grad_norm": 1.7343984842300415,
      "learning_rate": 0.0001914924038750322,
      "loss": 1.817,
      "step": 4420
    },
    {
      "epoch": 0.158325424821387,
      "grad_norm": 1.7777115106582642,
      "learning_rate": 0.00019148772159559486,
      "loss": 1.4785,
      "step": 4421
    },
    {
      "epoch": 0.15836123695095528,
      "grad_norm": 1.8788021802902222,
      "learning_rate": 0.00019148303808530818,
      "loss": 1.4365,
      "step": 4422
    },
    {
      "epoch": 0.15839704908052357,
      "grad_norm": 1.6759637594223022,
      "learning_rate": 0.00019147835334423527,
      "loss": 1.5425,
      "step": 4423
    },
    {
      "epoch": 0.15843286121009187,
      "grad_norm": 1.5373001098632812,
      "learning_rate": 0.0001914736673724391,
      "loss": 1.4686,
      "step": 4424
    },
    {
      "epoch": 0.15846867333966014,
      "grad_norm": 1.998205542564392,
      "learning_rate": 0.00019146898016998273,
      "loss": 1.7219,
      "step": 4425
    },
    {
      "epoch": 0.15850448546922843,
      "grad_norm": 1.5967192649841309,
      "learning_rate": 0.00019146429173692923,
      "loss": 1.4809,
      "step": 4426
    },
    {
      "epoch": 0.1585402975987967,
      "grad_norm": 1.7499881982803345,
      "learning_rate": 0.00019145960207334165,
      "loss": 1.4854,
      "step": 4427
    },
    {
      "epoch": 0.158576109728365,
      "grad_norm": 1.3718316555023193,
      "learning_rate": 0.00019145491117928312,
      "loss": 1.7586,
      "step": 4428
    },
    {
      "epoch": 0.15861192185793327,
      "grad_norm": 1.6473380327224731,
      "learning_rate": 0.00019145021905481673,
      "loss": 1.8481,
      "step": 4429
    },
    {
      "epoch": 0.15864773398750157,
      "grad_norm": 1.416988730430603,
      "learning_rate": 0.00019144552570000558,
      "loss": 1.5625,
      "step": 4430
    },
    {
      "epoch": 0.15868354611706986,
      "grad_norm": 1.5032716989517212,
      "learning_rate": 0.00019144083111491284,
      "loss": 1.511,
      "step": 4431
    },
    {
      "epoch": 0.15871935824663813,
      "grad_norm": 1.6411312818527222,
      "learning_rate": 0.0001914361352996017,
      "loss": 1.3405,
      "step": 4432
    },
    {
      "epoch": 0.15875517037620643,
      "grad_norm": 1.9214622974395752,
      "learning_rate": 0.00019143143825413526,
      "loss": 1.6389,
      "step": 4433
    },
    {
      "epoch": 0.1587909825057747,
      "grad_norm": 1.5103355646133423,
      "learning_rate": 0.00019142673997857678,
      "loss": 1.4921,
      "step": 4434
    },
    {
      "epoch": 0.158826794635343,
      "grad_norm": 1.3821022510528564,
      "learning_rate": 0.00019142204047298945,
      "loss": 1.4057,
      "step": 4435
    },
    {
      "epoch": 0.15886260676491126,
      "grad_norm": 2.0215821266174316,
      "learning_rate": 0.00019141733973743644,
      "loss": 1.7836,
      "step": 4436
    },
    {
      "epoch": 0.15889841889447956,
      "grad_norm": 1.9990155696868896,
      "learning_rate": 0.0001914126377719811,
      "loss": 1.508,
      "step": 4437
    },
    {
      "epoch": 0.15893423102404786,
      "grad_norm": 1.4154623746871948,
      "learning_rate": 0.00019140793457668665,
      "loss": 1.4155,
      "step": 4438
    },
    {
      "epoch": 0.15897004315361613,
      "grad_norm": 1.843180775642395,
      "learning_rate": 0.0001914032301516163,
      "loss": 1.6476,
      "step": 4439
    },
    {
      "epoch": 0.15900585528318442,
      "grad_norm": 1.4247946739196777,
      "learning_rate": 0.0001913985244968334,
      "loss": 1.4814,
      "step": 4440
    },
    {
      "epoch": 0.1590416674127527,
      "grad_norm": 1.6840612888336182,
      "learning_rate": 0.00019139381761240127,
      "loss": 1.5899,
      "step": 4441
    },
    {
      "epoch": 0.159077479542321,
      "grad_norm": 1.6597590446472168,
      "learning_rate": 0.00019138910949838321,
      "loss": 1.4572,
      "step": 4442
    },
    {
      "epoch": 0.15911329167188926,
      "grad_norm": 1.8821818828582764,
      "learning_rate": 0.0001913844001548425,
      "loss": 1.6773,
      "step": 4443
    },
    {
      "epoch": 0.15914910380145755,
      "grad_norm": 1.239434838294983,
      "learning_rate": 0.00019137968958184265,
      "loss": 1.5669,
      "step": 4444
    },
    {
      "epoch": 0.15918491593102585,
      "grad_norm": 1.371330738067627,
      "learning_rate": 0.00019137497777944691,
      "loss": 1.2934,
      "step": 4445
    },
    {
      "epoch": 0.15922072806059412,
      "grad_norm": 2.062408447265625,
      "learning_rate": 0.00019137026474771874,
      "loss": 1.3111,
      "step": 4446
    },
    {
      "epoch": 0.15925654019016242,
      "grad_norm": 2.3759512901306152,
      "learning_rate": 0.00019136555048672145,
      "loss": 1.4451,
      "step": 4447
    },
    {
      "epoch": 0.15929235231973068,
      "grad_norm": 2.019629716873169,
      "learning_rate": 0.0001913608349965186,
      "loss": 1.6197,
      "step": 4448
    },
    {
      "epoch": 0.15932816444929898,
      "grad_norm": 1.272171139717102,
      "learning_rate": 0.0001913561182771735,
      "loss": 1.6714,
      "step": 4449
    },
    {
      "epoch": 0.15936397657886725,
      "grad_norm": 1.7505909204483032,
      "learning_rate": 0.00019135140032874973,
      "loss": 1.5886,
      "step": 4450
    },
    {
      "epoch": 0.15939978870843555,
      "grad_norm": 1.6725510358810425,
      "learning_rate": 0.00019134668115131068,
      "loss": 1.6745,
      "step": 4451
    },
    {
      "epoch": 0.15943560083800384,
      "grad_norm": 1.7542790174484253,
      "learning_rate": 0.00019134196074491988,
      "loss": 1.584,
      "step": 4452
    },
    {
      "epoch": 0.1594714129675721,
      "grad_norm": 1.595931053161621,
      "learning_rate": 0.00019133723910964078,
      "loss": 1.5579,
      "step": 4453
    },
    {
      "epoch": 0.1595072250971404,
      "grad_norm": 2.000361204147339,
      "learning_rate": 0.00019133251624553696,
      "loss": 1.5491,
      "step": 4454
    },
    {
      "epoch": 0.15954303722670868,
      "grad_norm": 1.6233510971069336,
      "learning_rate": 0.00019132779215267197,
      "loss": 1.7638,
      "step": 4455
    },
    {
      "epoch": 0.15957884935627697,
      "grad_norm": 1.5333667993545532,
      "learning_rate": 0.00019132306683110933,
      "loss": 1.4675,
      "step": 4456
    },
    {
      "epoch": 0.15961466148584524,
      "grad_norm": 1.6372523307800293,
      "learning_rate": 0.0001913183402809126,
      "loss": 1.8686,
      "step": 4457
    },
    {
      "epoch": 0.15965047361541354,
      "grad_norm": 1.6973719596862793,
      "learning_rate": 0.00019131361250214541,
      "loss": 1.4094,
      "step": 4458
    },
    {
      "epoch": 0.15968628574498184,
      "grad_norm": 1.5486109256744385,
      "learning_rate": 0.00019130888349487134,
      "loss": 1.2735,
      "step": 4459
    },
    {
      "epoch": 0.1597220978745501,
      "grad_norm": 1.7678258419036865,
      "learning_rate": 0.00019130415325915406,
      "loss": 1.454,
      "step": 4460
    },
    {
      "epoch": 0.1597579100041184,
      "grad_norm": 1.7122492790222168,
      "learning_rate": 0.00019129942179505713,
      "loss": 1.3684,
      "step": 4461
    },
    {
      "epoch": 0.15979372213368667,
      "grad_norm": 1.4079463481903076,
      "learning_rate": 0.00019129468910264428,
      "loss": 1.6309,
      "step": 4462
    },
    {
      "epoch": 0.15982953426325497,
      "grad_norm": 2.084505319595337,
      "learning_rate": 0.00019128995518197912,
      "loss": 1.6246,
      "step": 4463
    },
    {
      "epoch": 0.15986534639282324,
      "grad_norm": 1.141055941581726,
      "learning_rate": 0.00019128522003312537,
      "loss": 1.2627,
      "step": 4464
    },
    {
      "epoch": 0.15990115852239153,
      "grad_norm": 1.7960704565048218,
      "learning_rate": 0.00019128048365614676,
      "loss": 1.5314,
      "step": 4465
    },
    {
      "epoch": 0.15993697065195983,
      "grad_norm": 1.8227843046188354,
      "learning_rate": 0.00019127574605110693,
      "loss": 1.8467,
      "step": 4466
    },
    {
      "epoch": 0.1599727827815281,
      "grad_norm": 1.508379340171814,
      "learning_rate": 0.00019127100721806975,
      "loss": 1.6528,
      "step": 4467
    },
    {
      "epoch": 0.1600085949110964,
      "grad_norm": 1.5757975578308105,
      "learning_rate": 0.00019126626715709885,
      "loss": 1.25,
      "step": 4468
    },
    {
      "epoch": 0.16004440704066467,
      "grad_norm": 1.5202853679656982,
      "learning_rate": 0.00019126152586825806,
      "loss": 1.4711,
      "step": 4469
    },
    {
      "epoch": 0.16008021917023296,
      "grad_norm": 1.781907558441162,
      "learning_rate": 0.00019125678335161117,
      "loss": 1.4699,
      "step": 4470
    },
    {
      "epoch": 0.16011603129980123,
      "grad_norm": 1.2357120513916016,
      "learning_rate": 0.00019125203960722198,
      "loss": 1.5946,
      "step": 4471
    },
    {
      "epoch": 0.16015184342936953,
      "grad_norm": 1.6788946390151978,
      "learning_rate": 0.00019124729463515427,
      "loss": 1.719,
      "step": 4472
    },
    {
      "epoch": 0.16018765555893782,
      "grad_norm": 1.5136687755584717,
      "learning_rate": 0.00019124254843547195,
      "loss": 1.5618,
      "step": 4473
    },
    {
      "epoch": 0.1602234676885061,
      "grad_norm": 1.810028314590454,
      "learning_rate": 0.0001912378010082388,
      "loss": 1.5755,
      "step": 4474
    },
    {
      "epoch": 0.1602592798180744,
      "grad_norm": 4.77049446105957,
      "learning_rate": 0.00019123305235351873,
      "loss": 1.4884,
      "step": 4475
    },
    {
      "epoch": 0.16029509194764266,
      "grad_norm": 1.4742318391799927,
      "learning_rate": 0.00019122830247137563,
      "loss": 1.3881,
      "step": 4476
    },
    {
      "epoch": 0.16033090407721096,
      "grad_norm": 2.340515375137329,
      "learning_rate": 0.00019122355136187342,
      "loss": 1.8475,
      "step": 4477
    },
    {
      "epoch": 0.16036671620677923,
      "grad_norm": 1.6513471603393555,
      "learning_rate": 0.00019121879902507595,
      "loss": 1.6164,
      "step": 4478
    },
    {
      "epoch": 0.16040252833634752,
      "grad_norm": 1.3290749788284302,
      "learning_rate": 0.00019121404546104724,
      "loss": 1.6316,
      "step": 4479
    },
    {
      "epoch": 0.16043834046591582,
      "grad_norm": 1.52192223072052,
      "learning_rate": 0.00019120929066985122,
      "loss": 1.2523,
      "step": 4480
    },
    {
      "epoch": 0.1604741525954841,
      "grad_norm": 2.044299602508545,
      "learning_rate": 0.0001912045346515518,
      "loss": 1.4638,
      "step": 4481
    },
    {
      "epoch": 0.16050996472505238,
      "grad_norm": 1.4951385259628296,
      "learning_rate": 0.00019119977740621305,
      "loss": 1.476,
      "step": 4482
    },
    {
      "epoch": 0.16054577685462065,
      "grad_norm": 1.5891962051391602,
      "learning_rate": 0.0001911950189338989,
      "loss": 1.1198,
      "step": 4483
    },
    {
      "epoch": 0.16058158898418895,
      "grad_norm": 1.8488585948944092,
      "learning_rate": 0.00019119025923467343,
      "loss": 1.5125,
      "step": 4484
    },
    {
      "epoch": 0.16061740111375722,
      "grad_norm": 2.9444382190704346,
      "learning_rate": 0.00019118549830860065,
      "loss": 1.4347,
      "step": 4485
    },
    {
      "epoch": 0.16065321324332552,
      "grad_norm": 2.6969316005706787,
      "learning_rate": 0.0001911807361557446,
      "loss": 1.4744,
      "step": 4486
    },
    {
      "epoch": 0.1606890253728938,
      "grad_norm": 1.69938063621521,
      "learning_rate": 0.00019117597277616932,
      "loss": 1.5696,
      "step": 4487
    },
    {
      "epoch": 0.16072483750246208,
      "grad_norm": 1.9011529684066772,
      "learning_rate": 0.00019117120816993899,
      "loss": 1.6207,
      "step": 4488
    },
    {
      "epoch": 0.16076064963203038,
      "grad_norm": 1.8051317930221558,
      "learning_rate": 0.00019116644233711764,
      "loss": 1.5994,
      "step": 4489
    },
    {
      "epoch": 0.16079646176159865,
      "grad_norm": 1.7188615798950195,
      "learning_rate": 0.0001911616752777694,
      "loss": 1.7194,
      "step": 4490
    },
    {
      "epoch": 0.16083227389116694,
      "grad_norm": 1.4117614030838013,
      "learning_rate": 0.0001911569069919584,
      "loss": 1.666,
      "step": 4491
    },
    {
      "epoch": 0.1608680860207352,
      "grad_norm": 1.8532236814498901,
      "learning_rate": 0.00019115213747974882,
      "loss": 1.2589,
      "step": 4492
    },
    {
      "epoch": 0.1609038981503035,
      "grad_norm": 1.4849985837936401,
      "learning_rate": 0.0001911473667412048,
      "loss": 1.5491,
      "step": 4493
    },
    {
      "epoch": 0.1609397102798718,
      "grad_norm": 1.5304160118103027,
      "learning_rate": 0.00019114259477639057,
      "loss": 1.656,
      "step": 4494
    },
    {
      "epoch": 0.16097552240944007,
      "grad_norm": 1.8665733337402344,
      "learning_rate": 0.00019113782158537024,
      "loss": 1.447,
      "step": 4495
    },
    {
      "epoch": 0.16101133453900837,
      "grad_norm": 1.4369174242019653,
      "learning_rate": 0.0001911330471682081,
      "loss": 1.3837,
      "step": 4496
    },
    {
      "epoch": 0.16104714666857664,
      "grad_norm": 1.2710061073303223,
      "learning_rate": 0.00019112827152496835,
      "loss": 1.7857,
      "step": 4497
    },
    {
      "epoch": 0.16108295879814494,
      "grad_norm": 1.0543251037597656,
      "learning_rate": 0.00019112349465571525,
      "loss": 1.6002,
      "step": 4498
    },
    {
      "epoch": 0.1611187709277132,
      "grad_norm": 1.3556406497955322,
      "learning_rate": 0.0001911187165605131,
      "loss": 1.5894,
      "step": 4499
    },
    {
      "epoch": 0.1611545830572815,
      "grad_norm": 1.9065412282943726,
      "learning_rate": 0.00019111393723942615,
      "loss": 1.5288,
      "step": 4500
    },
    {
      "epoch": 0.1611903951868498,
      "grad_norm": 1.592268705368042,
      "learning_rate": 0.00019110915669251868,
      "loss": 1.5041,
      "step": 4501
    },
    {
      "epoch": 0.16122620731641807,
      "grad_norm": 1.5915298461914062,
      "learning_rate": 0.00019110437491985505,
      "loss": 1.9453,
      "step": 4502
    },
    {
      "epoch": 0.16126201944598637,
      "grad_norm": 2.055844306945801,
      "learning_rate": 0.00019109959192149955,
      "loss": 1.405,
      "step": 4503
    },
    {
      "epoch": 0.16129783157555463,
      "grad_norm": 1.386273980140686,
      "learning_rate": 0.0001910948076975166,
      "loss": 1.3482,
      "step": 4504
    },
    {
      "epoch": 0.16133364370512293,
      "grad_norm": 1.3630104064941406,
      "learning_rate": 0.00019109002224797046,
      "loss": 1.8159,
      "step": 4505
    },
    {
      "epoch": 0.1613694558346912,
      "grad_norm": 1.701704978942871,
      "learning_rate": 0.00019108523557292558,
      "loss": 1.5629,
      "step": 4506
    },
    {
      "epoch": 0.1614052679642595,
      "grad_norm": 1.483626127243042,
      "learning_rate": 0.00019108044767244636,
      "loss": 1.6765,
      "step": 4507
    },
    {
      "epoch": 0.16144108009382777,
      "grad_norm": 1.3464930057525635,
      "learning_rate": 0.0001910756585465972,
      "loss": 1.7481,
      "step": 4508
    },
    {
      "epoch": 0.16147689222339606,
      "grad_norm": 1.5823675394058228,
      "learning_rate": 0.0001910708681954425,
      "loss": 1.5115,
      "step": 4509
    },
    {
      "epoch": 0.16151270435296436,
      "grad_norm": 1.8616117238998413,
      "learning_rate": 0.00019106607661904682,
      "loss": 1.6684,
      "step": 4510
    },
    {
      "epoch": 0.16154851648253263,
      "grad_norm": 1.4790433645248413,
      "learning_rate": 0.00019106128381747448,
      "loss": 1.528,
      "step": 4511
    },
    {
      "epoch": 0.16158432861210092,
      "grad_norm": 1.3721879720687866,
      "learning_rate": 0.00019105648979079006,
      "loss": 1.6033,
      "step": 4512
    },
    {
      "epoch": 0.1616201407416692,
      "grad_norm": 1.1864506006240845,
      "learning_rate": 0.000191051694539058,
      "loss": 1.5174,
      "step": 4513
    },
    {
      "epoch": 0.1616559528712375,
      "grad_norm": 1.2634984254837036,
      "learning_rate": 0.0001910468980623428,
      "loss": 1.5987,
      "step": 4514
    },
    {
      "epoch": 0.16169176500080576,
      "grad_norm": 2.0381622314453125,
      "learning_rate": 0.0001910421003607091,
      "loss": 1.6299,
      "step": 4515
    },
    {
      "epoch": 0.16172757713037406,
      "grad_norm": 1.4539151191711426,
      "learning_rate": 0.00019103730143422135,
      "loss": 1.8375,
      "step": 4516
    },
    {
      "epoch": 0.16176338925994235,
      "grad_norm": 1.7480626106262207,
      "learning_rate": 0.00019103250128294413,
      "loss": 1.608,
      "step": 4517
    },
    {
      "epoch": 0.16179920138951062,
      "grad_norm": 1.822675347328186,
      "learning_rate": 0.00019102769990694208,
      "loss": 1.5056,
      "step": 4518
    },
    {
      "epoch": 0.16183501351907892,
      "grad_norm": 1.5207996368408203,
      "learning_rate": 0.00019102289730627968,
      "loss": 1.5263,
      "step": 4519
    },
    {
      "epoch": 0.1618708256486472,
      "grad_norm": 2.3780176639556885,
      "learning_rate": 0.0001910180934810216,
      "loss": 1.581,
      "step": 4520
    },
    {
      "epoch": 0.16190663777821548,
      "grad_norm": 2.004570484161377,
      "learning_rate": 0.0001910132884312325,
      "loss": 1.853,
      "step": 4521
    },
    {
      "epoch": 0.16194244990778375,
      "grad_norm": 1.402849793434143,
      "learning_rate": 0.00019100848215697705,
      "loss": 1.5757,
      "step": 4522
    },
    {
      "epoch": 0.16197826203735205,
      "grad_norm": 1.324273943901062,
      "learning_rate": 0.00019100367465831983,
      "loss": 1.5214,
      "step": 4523
    },
    {
      "epoch": 0.16201407416692035,
      "grad_norm": 1.5259591341018677,
      "learning_rate": 0.00019099886593532554,
      "loss": 1.4345,
      "step": 4524
    },
    {
      "epoch": 0.16204988629648862,
      "grad_norm": 1.5737764835357666,
      "learning_rate": 0.00019099405598805888,
      "loss": 1.5467,
      "step": 4525
    },
    {
      "epoch": 0.1620856984260569,
      "grad_norm": 1.5883179903030396,
      "learning_rate": 0.0001909892448165846,
      "loss": 1.553,
      "step": 4526
    },
    {
      "epoch": 0.16212151055562518,
      "grad_norm": 1.3239035606384277,
      "learning_rate": 0.0001909844324209674,
      "loss": 1.5387,
      "step": 4527
    },
    {
      "epoch": 0.16215732268519348,
      "grad_norm": 1.957148790359497,
      "learning_rate": 0.00019097961880127203,
      "loss": 1.5983,
      "step": 4528
    },
    {
      "epoch": 0.16219313481476175,
      "grad_norm": 1.2890499830245972,
      "learning_rate": 0.0001909748039575632,
      "loss": 1.6041,
      "step": 4529
    },
    {
      "epoch": 0.16222894694433004,
      "grad_norm": 1.505007266998291,
      "learning_rate": 0.00019096998788990574,
      "loss": 1.3721,
      "step": 4530
    },
    {
      "epoch": 0.16226475907389834,
      "grad_norm": 1.8835853338241577,
      "learning_rate": 0.00019096517059836448,
      "loss": 1.8609,
      "step": 4531
    },
    {
      "epoch": 0.1623005712034666,
      "grad_norm": 1.509743571281433,
      "learning_rate": 0.00019096035208300416,
      "loss": 1.5606,
      "step": 4532
    },
    {
      "epoch": 0.1623363833330349,
      "grad_norm": 1.2899866104125977,
      "learning_rate": 0.00019095553234388962,
      "loss": 1.7222,
      "step": 4533
    },
    {
      "epoch": 0.16237219546260317,
      "grad_norm": 1.679097294807434,
      "learning_rate": 0.00019095071138108575,
      "loss": 1.5859,
      "step": 4534
    },
    {
      "epoch": 0.16240800759217147,
      "grad_norm": 1.3049614429473877,
      "learning_rate": 0.00019094588919465734,
      "loss": 1.4242,
      "step": 4535
    },
    {
      "epoch": 0.16244381972173974,
      "grad_norm": 2.140238046646118,
      "learning_rate": 0.0001909410657846693,
      "loss": 1.7717,
      "step": 4536
    },
    {
      "epoch": 0.16247963185130804,
      "grad_norm": 1.8646196126937866,
      "learning_rate": 0.00019093624115118656,
      "loss": 1.5395,
      "step": 4537
    },
    {
      "epoch": 0.16251544398087633,
      "grad_norm": 1.333450198173523,
      "learning_rate": 0.00019093141529427396,
      "loss": 1.5125,
      "step": 4538
    },
    {
      "epoch": 0.1625512561104446,
      "grad_norm": 1.132122278213501,
      "learning_rate": 0.00019092658821399648,
      "loss": 1.4367,
      "step": 4539
    },
    {
      "epoch": 0.1625870682400129,
      "grad_norm": 1.6168757677078247,
      "learning_rate": 0.00019092175991041905,
      "loss": 1.6936,
      "step": 4540
    },
    {
      "epoch": 0.16262288036958117,
      "grad_norm": 1.564820647239685,
      "learning_rate": 0.0001909169303836066,
      "loss": 1.645,
      "step": 4541
    },
    {
      "epoch": 0.16265869249914947,
      "grad_norm": 1.6114917993545532,
      "learning_rate": 0.00019091209963362416,
      "loss": 1.4573,
      "step": 4542
    },
    {
      "epoch": 0.16269450462871773,
      "grad_norm": 2.2553696632385254,
      "learning_rate": 0.00019090726766053667,
      "loss": 1.7323,
      "step": 4543
    },
    {
      "epoch": 0.16273031675828603,
      "grad_norm": 1.2172681093215942,
      "learning_rate": 0.00019090243446440915,
      "loss": 1.5023,
      "step": 4544
    },
    {
      "epoch": 0.16276612888785433,
      "grad_norm": 1.509480357170105,
      "learning_rate": 0.0001908976000453066,
      "loss": 1.6955,
      "step": 4545
    },
    {
      "epoch": 0.1628019410174226,
      "grad_norm": 1.7382971048355103,
      "learning_rate": 0.00019089276440329415,
      "loss": 1.4946,
      "step": 4546
    },
    {
      "epoch": 0.1628377531469909,
      "grad_norm": 2.00982928276062,
      "learning_rate": 0.00019088792753843675,
      "loss": 1.7347,
      "step": 4547
    },
    {
      "epoch": 0.16287356527655916,
      "grad_norm": 1.5432096719741821,
      "learning_rate": 0.00019088308945079956,
      "loss": 1.8206,
      "step": 4548
    },
    {
      "epoch": 0.16290937740612746,
      "grad_norm": 1.4766311645507812,
      "learning_rate": 0.00019087825014044762,
      "loss": 1.6214,
      "step": 4549
    },
    {
      "epoch": 0.16294518953569573,
      "grad_norm": 2.6827495098114014,
      "learning_rate": 0.00019087340960744604,
      "loss": 1.6215,
      "step": 4550
    },
    {
      "epoch": 0.16298100166526402,
      "grad_norm": 1.2720814943313599,
      "learning_rate": 0.00019086856785185992,
      "loss": 1.5734,
      "step": 4551
    },
    {
      "epoch": 0.16301681379483232,
      "grad_norm": 1.7849035263061523,
      "learning_rate": 0.0001908637248737545,
      "loss": 1.5433,
      "step": 4552
    },
    {
      "epoch": 0.1630526259244006,
      "grad_norm": 1.7423144578933716,
      "learning_rate": 0.00019085888067319485,
      "loss": 1.4252,
      "step": 4553
    },
    {
      "epoch": 0.1630884380539689,
      "grad_norm": 2.395569324493408,
      "learning_rate": 0.00019085403525024612,
      "loss": 1.5359,
      "step": 4554
    },
    {
      "epoch": 0.16312425018353716,
      "grad_norm": 1.3138221502304077,
      "learning_rate": 0.00019084918860497356,
      "loss": 1.4022,
      "step": 4555
    },
    {
      "epoch": 0.16316006231310545,
      "grad_norm": 1.5145282745361328,
      "learning_rate": 0.00019084434073744238,
      "loss": 1.108,
      "step": 4556
    },
    {
      "epoch": 0.16319587444267372,
      "grad_norm": 1.376220703125,
      "learning_rate": 0.00019083949164771773,
      "loss": 1.4966,
      "step": 4557
    },
    {
      "epoch": 0.16323168657224202,
      "grad_norm": 1.3827080726623535,
      "learning_rate": 0.00019083464133586492,
      "loss": 1.6534,
      "step": 4558
    },
    {
      "epoch": 0.16326749870181032,
      "grad_norm": 1.524684190750122,
      "learning_rate": 0.00019082978980194918,
      "loss": 1.7182,
      "step": 4559
    },
    {
      "epoch": 0.16330331083137858,
      "grad_norm": 1.8181837797164917,
      "learning_rate": 0.00019082493704603576,
      "loss": 1.631,
      "step": 4560
    },
    {
      "epoch": 0.16333912296094688,
      "grad_norm": 1.2133936882019043,
      "learning_rate": 0.00019082008306819001,
      "loss": 1.7072,
      "step": 4561
    },
    {
      "epoch": 0.16337493509051515,
      "grad_norm": 1.3571362495422363,
      "learning_rate": 0.00019081522786847717,
      "loss": 1.6741,
      "step": 4562
    },
    {
      "epoch": 0.16341074722008345,
      "grad_norm": 1.4400066137313843,
      "learning_rate": 0.0001908103714469626,
      "loss": 1.394,
      "step": 4563
    },
    {
      "epoch": 0.16344655934965172,
      "grad_norm": 2.2084951400756836,
      "learning_rate": 0.00019080551380371157,
      "loss": 1.7241,
      "step": 4564
    },
    {
      "epoch": 0.16348237147922,
      "grad_norm": 1.2983593940734863,
      "learning_rate": 0.0001908006549387895,
      "loss": 1.4786,
      "step": 4565
    },
    {
      "epoch": 0.1635181836087883,
      "grad_norm": 1.8585529327392578,
      "learning_rate": 0.00019079579485226176,
      "loss": 1.5445,
      "step": 4566
    },
    {
      "epoch": 0.16355399573835658,
      "grad_norm": 1.7613205909729004,
      "learning_rate": 0.0001907909335441937,
      "loss": 1.54,
      "step": 4567
    },
    {
      "epoch": 0.16358980786792487,
      "grad_norm": 1.85093355178833,
      "learning_rate": 0.00019078607101465078,
      "loss": 1.5251,
      "step": 4568
    },
    {
      "epoch": 0.16362561999749314,
      "grad_norm": 2.1224894523620605,
      "learning_rate": 0.00019078120726369834,
      "loss": 1.6325,
      "step": 4569
    },
    {
      "epoch": 0.16366143212706144,
      "grad_norm": 1.862607717514038,
      "learning_rate": 0.00019077634229140188,
      "loss": 1.4159,
      "step": 4570
    },
    {
      "epoch": 0.1636972442566297,
      "grad_norm": 1.5398863554000854,
      "learning_rate": 0.0001907714760978268,
      "loss": 1.6412,
      "step": 4571
    },
    {
      "epoch": 0.163733056386198,
      "grad_norm": 1.6998611688613892,
      "learning_rate": 0.0001907666086830386,
      "loss": 1.6833,
      "step": 4572
    },
    {
      "epoch": 0.1637688685157663,
      "grad_norm": 1.262229561805725,
      "learning_rate": 0.0001907617400471028,
      "loss": 1.3874,
      "step": 4573
    },
    {
      "epoch": 0.16380468064533457,
      "grad_norm": 1.8390412330627441,
      "learning_rate": 0.00019075687019008483,
      "loss": 1.7563,
      "step": 4574
    },
    {
      "epoch": 0.16384049277490287,
      "grad_norm": 1.7868350744247437,
      "learning_rate": 0.00019075199911205024,
      "loss": 1.5629,
      "step": 4575
    },
    {
      "epoch": 0.16387630490447114,
      "grad_norm": 1.949973464012146,
      "learning_rate": 0.00019074712681306456,
      "loss": 1.9253,
      "step": 4576
    },
    {
      "epoch": 0.16391211703403943,
      "grad_norm": 1.4536129236221313,
      "learning_rate": 0.00019074225329319337,
      "loss": 1.5995,
      "step": 4577
    },
    {
      "epoch": 0.1639479291636077,
      "grad_norm": 1.5427803993225098,
      "learning_rate": 0.00019073737855250218,
      "loss": 1.5241,
      "step": 4578
    },
    {
      "epoch": 0.163983741293176,
      "grad_norm": 1.3218435049057007,
      "learning_rate": 0.00019073250259105663,
      "loss": 1.6316,
      "step": 4579
    },
    {
      "epoch": 0.1640195534227443,
      "grad_norm": 1.3886061906814575,
      "learning_rate": 0.00019072762540892226,
      "loss": 1.5922,
      "step": 4580
    },
    {
      "epoch": 0.16405536555231257,
      "grad_norm": 1.7809525728225708,
      "learning_rate": 0.00019072274700616474,
      "loss": 1.951,
      "step": 4581
    },
    {
      "epoch": 0.16409117768188086,
      "grad_norm": 1.7092533111572266,
      "learning_rate": 0.00019071786738284968,
      "loss": 1.4504,
      "step": 4582
    },
    {
      "epoch": 0.16412698981144913,
      "grad_norm": 1.6752238273620605,
      "learning_rate": 0.00019071298653904276,
      "loss": 1.6263,
      "step": 4583
    },
    {
      "epoch": 0.16416280194101743,
      "grad_norm": 1.514328122138977,
      "learning_rate": 0.00019070810447480957,
      "loss": 1.6375,
      "step": 4584
    },
    {
      "epoch": 0.1641986140705857,
      "grad_norm": 1.2303587198257446,
      "learning_rate": 0.00019070322119021588,
      "loss": 1.4602,
      "step": 4585
    },
    {
      "epoch": 0.164234426200154,
      "grad_norm": 2.415630578994751,
      "learning_rate": 0.00019069833668532732,
      "loss": 1.6044,
      "step": 4586
    },
    {
      "epoch": 0.1642702383297223,
      "grad_norm": 1.291359543800354,
      "learning_rate": 0.00019069345096020966,
      "loss": 1.5712,
      "step": 4587
    },
    {
      "epoch": 0.16430605045929056,
      "grad_norm": 1.2763800621032715,
      "learning_rate": 0.00019068856401492857,
      "loss": 1.6667,
      "step": 4588
    },
    {
      "epoch": 0.16434186258885886,
      "grad_norm": 1.4778591394424438,
      "learning_rate": 0.00019068367584954986,
      "loss": 1.3605,
      "step": 4589
    },
    {
      "epoch": 0.16437767471842712,
      "grad_norm": 1.095234990119934,
      "learning_rate": 0.00019067878646413923,
      "loss": 1.5766,
      "step": 4590
    },
    {
      "epoch": 0.16441348684799542,
      "grad_norm": 1.2776097059249878,
      "learning_rate": 0.0001906738958587625,
      "loss": 1.5516,
      "step": 4591
    },
    {
      "epoch": 0.1644492989775637,
      "grad_norm": 1.524300456047058,
      "learning_rate": 0.00019066900403348551,
      "loss": 1.4556,
      "step": 4592
    },
    {
      "epoch": 0.164485111107132,
      "grad_norm": 1.618364691734314,
      "learning_rate": 0.000190664110988374,
      "loss": 1.6515,
      "step": 4593
    },
    {
      "epoch": 0.16452092323670028,
      "grad_norm": 2.83412766456604,
      "learning_rate": 0.00019065921672349384,
      "loss": 1.6118,
      "step": 4594
    },
    {
      "epoch": 0.16455673536626855,
      "grad_norm": 1.6449576616287231,
      "learning_rate": 0.00019065432123891083,
      "loss": 1.5997,
      "step": 4595
    },
    {
      "epoch": 0.16459254749583685,
      "grad_norm": 1.6248342990875244,
      "learning_rate": 0.00019064942453469086,
      "loss": 1.6638,
      "step": 4596
    },
    {
      "epoch": 0.16462835962540512,
      "grad_norm": 2.3031158447265625,
      "learning_rate": 0.0001906445266108998,
      "loss": 1.7916,
      "step": 4597
    },
    {
      "epoch": 0.16466417175497342,
      "grad_norm": 1.7779685258865356,
      "learning_rate": 0.0001906396274676036,
      "loss": 1.4638,
      "step": 4598
    },
    {
      "epoch": 0.16469998388454168,
      "grad_norm": 1.4672049283981323,
      "learning_rate": 0.00019063472710486814,
      "loss": 1.6929,
      "step": 4599
    },
    {
      "epoch": 0.16473579601410998,
      "grad_norm": 1.7978802919387817,
      "learning_rate": 0.0001906298255227593,
      "loss": 1.6518,
      "step": 4600
    },
    {
      "epoch": 0.16477160814367828,
      "grad_norm": 1.8508810997009277,
      "learning_rate": 0.00019062492272134307,
      "loss": 1.4408,
      "step": 4601
    },
    {
      "epoch": 0.16480742027324655,
      "grad_norm": 1.8320178985595703,
      "learning_rate": 0.0001906200187006854,
      "loss": 1.5152,
      "step": 4602
    },
    {
      "epoch": 0.16484323240281484,
      "grad_norm": 1.634345531463623,
      "learning_rate": 0.0001906151134608523,
      "loss": 1.6666,
      "step": 4603
    },
    {
      "epoch": 0.1648790445323831,
      "grad_norm": 1.9694124460220337,
      "learning_rate": 0.0001906102070019097,
      "loss": 1.6432,
      "step": 4604
    },
    {
      "epoch": 0.1649148566619514,
      "grad_norm": 2.6619157791137695,
      "learning_rate": 0.00019060529932392366,
      "loss": 1.4241,
      "step": 4605
    },
    {
      "epoch": 0.16495066879151968,
      "grad_norm": 2.111093282699585,
      "learning_rate": 0.00019060039042696016,
      "loss": 1.5643,
      "step": 4606
    },
    {
      "epoch": 0.16498648092108797,
      "grad_norm": 1.5327450037002563,
      "learning_rate": 0.00019059548031108528,
      "loss": 1.507,
      "step": 4607
    },
    {
      "epoch": 0.16502229305065624,
      "grad_norm": 2.1236813068389893,
      "learning_rate": 0.0001905905689763651,
      "loss": 1.3831,
      "step": 4608
    },
    {
      "epoch": 0.16505810518022454,
      "grad_norm": 1.4303845167160034,
      "learning_rate": 0.00019058565642286567,
      "loss": 1.1834,
      "step": 4609
    },
    {
      "epoch": 0.16509391730979284,
      "grad_norm": 1.145059585571289,
      "learning_rate": 0.00019058074265065303,
      "loss": 1.4526,
      "step": 4610
    },
    {
      "epoch": 0.1651297294393611,
      "grad_norm": 2.205949068069458,
      "learning_rate": 0.00019057582765979341,
      "loss": 1.4746,
      "step": 4611
    },
    {
      "epoch": 0.1651655415689294,
      "grad_norm": 1.787413477897644,
      "learning_rate": 0.00019057091145035281,
      "loss": 1.5251,
      "step": 4612
    },
    {
      "epoch": 0.16520135369849767,
      "grad_norm": 1.5059257745742798,
      "learning_rate": 0.00019056599402239742,
      "loss": 1.5406,
      "step": 4613
    },
    {
      "epoch": 0.16523716582806597,
      "grad_norm": 1.2862091064453125,
      "learning_rate": 0.00019056107537599343,
      "loss": 1.5867,
      "step": 4614
    },
    {
      "epoch": 0.16527297795763424,
      "grad_norm": 1.640185832977295,
      "learning_rate": 0.000190556155511207,
      "loss": 1.9078,
      "step": 4615
    },
    {
      "epoch": 0.16530879008720253,
      "grad_norm": 2.0387260913848877,
      "learning_rate": 0.00019055123442810427,
      "loss": 1.6649,
      "step": 4616
    },
    {
      "epoch": 0.16534460221677083,
      "grad_norm": 1.985213041305542,
      "learning_rate": 0.00019054631212675152,
      "loss": 1.2348,
      "step": 4617
    },
    {
      "epoch": 0.1653804143463391,
      "grad_norm": 1.5583789348602295,
      "learning_rate": 0.00019054138860721492,
      "loss": 1.7628,
      "step": 4618
    },
    {
      "epoch": 0.1654162264759074,
      "grad_norm": 2.322619915008545,
      "learning_rate": 0.00019053646386956073,
      "loss": 1.385,
      "step": 4619
    },
    {
      "epoch": 0.16545203860547567,
      "grad_norm": 1.4217337369918823,
      "learning_rate": 0.00019053153791385516,
      "loss": 1.5673,
      "step": 4620
    },
    {
      "epoch": 0.16548785073504396,
      "grad_norm": 2.825695276260376,
      "learning_rate": 0.00019052661074016458,
      "loss": 1.5057,
      "step": 4621
    },
    {
      "epoch": 0.16552366286461223,
      "grad_norm": 1.6821850538253784,
      "learning_rate": 0.0001905216823485552,
      "loss": 1.4592,
      "step": 4622
    },
    {
      "epoch": 0.16555947499418053,
      "grad_norm": 1.3392614126205444,
      "learning_rate": 0.00019051675273909336,
      "loss": 1.7243,
      "step": 4623
    },
    {
      "epoch": 0.16559528712374882,
      "grad_norm": 1.9273262023925781,
      "learning_rate": 0.00019051182191184537,
      "loss": 1.3487,
      "step": 4624
    },
    {
      "epoch": 0.1656310992533171,
      "grad_norm": 1.8148261308670044,
      "learning_rate": 0.00019050688986687754,
      "loss": 1.5975,
      "step": 4625
    },
    {
      "epoch": 0.1656669113828854,
      "grad_norm": 1.3820042610168457,
      "learning_rate": 0.00019050195660425627,
      "loss": 1.465,
      "step": 4626
    },
    {
      "epoch": 0.16570272351245366,
      "grad_norm": 1.627036690711975,
      "learning_rate": 0.00019049702212404793,
      "loss": 1.3036,
      "step": 4627
    },
    {
      "epoch": 0.16573853564202196,
      "grad_norm": 1.241610050201416,
      "learning_rate": 0.00019049208642631885,
      "loss": 1.6127,
      "step": 4628
    },
    {
      "epoch": 0.16577434777159022,
      "grad_norm": 1.7306376695632935,
      "learning_rate": 0.00019048714951113552,
      "loss": 1.775,
      "step": 4629
    },
    {
      "epoch": 0.16581015990115852,
      "grad_norm": 1.5658414363861084,
      "learning_rate": 0.00019048221137856427,
      "loss": 1.8022,
      "step": 4630
    },
    {
      "epoch": 0.16584597203072682,
      "grad_norm": 1.4824405908584595,
      "learning_rate": 0.0001904772720286716,
      "loss": 1.5788,
      "step": 4631
    },
    {
      "epoch": 0.1658817841602951,
      "grad_norm": 1.6523913145065308,
      "learning_rate": 0.00019047233146152393,
      "loss": 1.4036,
      "step": 4632
    },
    {
      "epoch": 0.16591759628986338,
      "grad_norm": 1.4244657754898071,
      "learning_rate": 0.00019046738967718778,
      "loss": 1.6885,
      "step": 4633
    },
    {
      "epoch": 0.16595340841943165,
      "grad_norm": 1.679327368736267,
      "learning_rate": 0.00019046244667572957,
      "loss": 1.6083,
      "step": 4634
    },
    {
      "epoch": 0.16598922054899995,
      "grad_norm": 1.6196266412734985,
      "learning_rate": 0.00019045750245721583,
      "loss": 1.3341,
      "step": 4635
    },
    {
      "epoch": 0.16602503267856822,
      "grad_norm": 2.6375415325164795,
      "learning_rate": 0.00019045255702171307,
      "loss": 1.7689,
      "step": 4636
    },
    {
      "epoch": 0.16606084480813652,
      "grad_norm": 1.9008333683013916,
      "learning_rate": 0.00019044761036928783,
      "loss": 1.6576,
      "step": 4637
    },
    {
      "epoch": 0.1660966569377048,
      "grad_norm": 2.3040616512298584,
      "learning_rate": 0.00019044266250000668,
      "loss": 1.7239,
      "step": 4638
    },
    {
      "epoch": 0.16613246906727308,
      "grad_norm": 2.4334566593170166,
      "learning_rate": 0.00019043771341393614,
      "loss": 1.7195,
      "step": 4639
    },
    {
      "epoch": 0.16616828119684138,
      "grad_norm": 1.575951337814331,
      "learning_rate": 0.00019043276311114283,
      "loss": 1.6378,
      "step": 4640
    },
    {
      "epoch": 0.16620409332640965,
      "grad_norm": 2.337109327316284,
      "learning_rate": 0.00019042781159169336,
      "loss": 1.6753,
      "step": 4641
    },
    {
      "epoch": 0.16623990545597794,
      "grad_norm": 1.561690092086792,
      "learning_rate": 0.00019042285885565433,
      "loss": 1.6058,
      "step": 4642
    },
    {
      "epoch": 0.1662757175855462,
      "grad_norm": 1.7889922857284546,
      "learning_rate": 0.00019041790490309233,
      "loss": 1.1429,
      "step": 4643
    },
    {
      "epoch": 0.1663115297151145,
      "grad_norm": 1.3240993022918701,
      "learning_rate": 0.00019041294973407412,
      "loss": 1.641,
      "step": 4644
    },
    {
      "epoch": 0.1663473418446828,
      "grad_norm": 2.112696409225464,
      "learning_rate": 0.00019040799334866626,
      "loss": 1.8524,
      "step": 4645
    },
    {
      "epoch": 0.16638315397425107,
      "grad_norm": 1.8621598482131958,
      "learning_rate": 0.00019040303574693545,
      "loss": 1.724,
      "step": 4646
    },
    {
      "epoch": 0.16641896610381937,
      "grad_norm": 1.5603737831115723,
      "learning_rate": 0.00019039807692894842,
      "loss": 1.4181,
      "step": 4647
    },
    {
      "epoch": 0.16645477823338764,
      "grad_norm": 1.5125117301940918,
      "learning_rate": 0.00019039311689477185,
      "loss": 1.628,
      "step": 4648
    },
    {
      "epoch": 0.16649059036295594,
      "grad_norm": 1.4379773139953613,
      "learning_rate": 0.0001903881556444725,
      "loss": 1.5146,
      "step": 4649
    },
    {
      "epoch": 0.1665264024925242,
      "grad_norm": 2.8193790912628174,
      "learning_rate": 0.00019038319317811714,
      "loss": 1.4629,
      "step": 4650
    },
    {
      "epoch": 0.1665622146220925,
      "grad_norm": 1.4412678480148315,
      "learning_rate": 0.00019037822949577248,
      "loss": 1.2369,
      "step": 4651
    },
    {
      "epoch": 0.1665980267516608,
      "grad_norm": 1.3727669715881348,
      "learning_rate": 0.00019037326459750534,
      "loss": 1.6593,
      "step": 4652
    },
    {
      "epoch": 0.16663383888122907,
      "grad_norm": 1.2576712369918823,
      "learning_rate": 0.00019036829848338246,
      "loss": 1.4383,
      "step": 4653
    },
    {
      "epoch": 0.16666965101079736,
      "grad_norm": 1.55268132686615,
      "learning_rate": 0.00019036333115347073,
      "loss": 1.6322,
      "step": 4654
    },
    {
      "epoch": 0.16670546314036563,
      "grad_norm": 1.6275209188461304,
      "learning_rate": 0.00019035836260783691,
      "loss": 1.4125,
      "step": 4655
    },
    {
      "epoch": 0.16674127526993393,
      "grad_norm": 1.3718266487121582,
      "learning_rate": 0.00019035339284654787,
      "loss": 1.3941,
      "step": 4656
    },
    {
      "epoch": 0.1667770873995022,
      "grad_norm": 1.859714388847351,
      "learning_rate": 0.0001903484218696705,
      "loss": 1.5406,
      "step": 4657
    },
    {
      "epoch": 0.1668128995290705,
      "grad_norm": 1.5116417407989502,
      "learning_rate": 0.00019034344967727165,
      "loss": 1.6361,
      "step": 4658
    },
    {
      "epoch": 0.1668487116586388,
      "grad_norm": 1.9010447263717651,
      "learning_rate": 0.0001903384762694182,
      "loss": 1.7493,
      "step": 4659
    },
    {
      "epoch": 0.16688452378820706,
      "grad_norm": 1.5057936906814575,
      "learning_rate": 0.0001903335016461771,
      "loss": 1.3977,
      "step": 4660
    },
    {
      "epoch": 0.16692033591777536,
      "grad_norm": 1.8938969373703003,
      "learning_rate": 0.00019032852580761527,
      "loss": 1.6976,
      "step": 4661
    },
    {
      "epoch": 0.16695614804734363,
      "grad_norm": 1.1892553567886353,
      "learning_rate": 0.00019032354875379962,
      "loss": 1.2591,
      "step": 4662
    },
    {
      "epoch": 0.16699196017691192,
      "grad_norm": 1.9757999181747437,
      "learning_rate": 0.00019031857048479713,
      "loss": 1.305,
      "step": 4663
    },
    {
      "epoch": 0.1670277723064802,
      "grad_norm": 1.8454691171646118,
      "learning_rate": 0.00019031359100067478,
      "loss": 1.7795,
      "step": 4664
    },
    {
      "epoch": 0.1670635844360485,
      "grad_norm": 2.3742315769195557,
      "learning_rate": 0.00019030861030149956,
      "loss": 1.4219,
      "step": 4665
    },
    {
      "epoch": 0.1670993965656168,
      "grad_norm": 1.6165446043014526,
      "learning_rate": 0.00019030362838733846,
      "loss": 1.3658,
      "step": 4666
    },
    {
      "epoch": 0.16713520869518506,
      "grad_norm": 1.2845289707183838,
      "learning_rate": 0.00019029864525825857,
      "loss": 1.5783,
      "step": 4667
    },
    {
      "epoch": 0.16717102082475335,
      "grad_norm": 1.3073277473449707,
      "learning_rate": 0.00019029366091432684,
      "loss": 1.6852,
      "step": 4668
    },
    {
      "epoch": 0.16720683295432162,
      "grad_norm": 1.8444633483886719,
      "learning_rate": 0.0001902886753556104,
      "loss": 1.3777,
      "step": 4669
    },
    {
      "epoch": 0.16724264508388992,
      "grad_norm": 1.9535983800888062,
      "learning_rate": 0.0001902836885821763,
      "loss": 1.5784,
      "step": 4670
    },
    {
      "epoch": 0.1672784572134582,
      "grad_norm": 1.5261894464492798,
      "learning_rate": 0.00019027870059409158,
      "loss": 1.3801,
      "step": 4671
    },
    {
      "epoch": 0.16731426934302648,
      "grad_norm": 1.40703284740448,
      "learning_rate": 0.00019027371139142342,
      "loss": 1.4457,
      "step": 4672
    },
    {
      "epoch": 0.16735008147259478,
      "grad_norm": 1.388258695602417,
      "learning_rate": 0.00019026872097423894,
      "loss": 1.3877,
      "step": 4673
    },
    {
      "epoch": 0.16738589360216305,
      "grad_norm": 1.8786894083023071,
      "learning_rate": 0.00019026372934260525,
      "loss": 1.318,
      "step": 4674
    },
    {
      "epoch": 0.16742170573173135,
      "grad_norm": 1.9680266380310059,
      "learning_rate": 0.0001902587364965895,
      "loss": 1.36,
      "step": 4675
    },
    {
      "epoch": 0.16745751786129962,
      "grad_norm": 1.4114983081817627,
      "learning_rate": 0.00019025374243625888,
      "loss": 1.7629,
      "step": 4676
    },
    {
      "epoch": 0.1674933299908679,
      "grad_norm": 1.4164817333221436,
      "learning_rate": 0.0001902487471616806,
      "loss": 1.935,
      "step": 4677
    },
    {
      "epoch": 0.16752914212043618,
      "grad_norm": 1.973514437675476,
      "learning_rate": 0.00019024375067292181,
      "loss": 1.3236,
      "step": 4678
    },
    {
      "epoch": 0.16756495425000448,
      "grad_norm": 1.940759301185608,
      "learning_rate": 0.00019023875297004977,
      "loss": 1.7045,
      "step": 4679
    },
    {
      "epoch": 0.16760076637957277,
      "grad_norm": 1.5399664640426636,
      "learning_rate": 0.0001902337540531317,
      "loss": 1.8139,
      "step": 4680
    },
    {
      "epoch": 0.16763657850914104,
      "grad_norm": 1.957043170928955,
      "learning_rate": 0.00019022875392223486,
      "loss": 1.6672,
      "step": 4681
    },
    {
      "epoch": 0.16767239063870934,
      "grad_norm": 1.677278757095337,
      "learning_rate": 0.00019022375257742656,
      "loss": 1.8589,
      "step": 4682
    },
    {
      "epoch": 0.1677082027682776,
      "grad_norm": 1.3992509841918945,
      "learning_rate": 0.000190218750018774,
      "loss": 1.3775,
      "step": 4683
    },
    {
      "epoch": 0.1677440148978459,
      "grad_norm": 3.1266071796417236,
      "learning_rate": 0.00019021374624634456,
      "loss": 1.5442,
      "step": 4684
    },
    {
      "epoch": 0.16777982702741417,
      "grad_norm": 1.9289811849594116,
      "learning_rate": 0.0001902087412602055,
      "loss": 1.4118,
      "step": 4685
    },
    {
      "epoch": 0.16781563915698247,
      "grad_norm": 1.3202590942382812,
      "learning_rate": 0.00019020373506042424,
      "loss": 1.3978,
      "step": 4686
    },
    {
      "epoch": 0.16785145128655077,
      "grad_norm": 1.4680845737457275,
      "learning_rate": 0.00019019872764706804,
      "loss": 1.8426,
      "step": 4687
    },
    {
      "epoch": 0.16788726341611904,
      "grad_norm": 1.5509357452392578,
      "learning_rate": 0.00019019371902020434,
      "loss": 1.6945,
      "step": 4688
    },
    {
      "epoch": 0.16792307554568733,
      "grad_norm": 2.0995917320251465,
      "learning_rate": 0.00019018870917990045,
      "loss": 1.9977,
      "step": 4689
    },
    {
      "epoch": 0.1679588876752556,
      "grad_norm": 1.4658147096633911,
      "learning_rate": 0.00019018369812622384,
      "loss": 1.7161,
      "step": 4690
    },
    {
      "epoch": 0.1679946998048239,
      "grad_norm": 1.6705026626586914,
      "learning_rate": 0.0001901786858592419,
      "loss": 1.464,
      "step": 4691
    },
    {
      "epoch": 0.16803051193439217,
      "grad_norm": 2.0256388187408447,
      "learning_rate": 0.00019017367237902206,
      "loss": 1.2567,
      "step": 4692
    },
    {
      "epoch": 0.16806632406396046,
      "grad_norm": 1.6162030696868896,
      "learning_rate": 0.00019016865768563176,
      "loss": 1.7855,
      "step": 4693
    },
    {
      "epoch": 0.16810213619352876,
      "grad_norm": 1.8909845352172852,
      "learning_rate": 0.0001901636417791385,
      "loss": 1.7189,
      "step": 4694
    },
    {
      "epoch": 0.16813794832309703,
      "grad_norm": 1.6075421571731567,
      "learning_rate": 0.00019015862465960974,
      "loss": 1.7273,
      "step": 4695
    },
    {
      "epoch": 0.16817376045266533,
      "grad_norm": 1.5975663661956787,
      "learning_rate": 0.00019015360632711298,
      "loss": 1.8074,
      "step": 4696
    },
    {
      "epoch": 0.1682095725822336,
      "grad_norm": 1.947001576423645,
      "learning_rate": 0.00019014858678171573,
      "loss": 1.5853,
      "step": 4697
    },
    {
      "epoch": 0.1682453847118019,
      "grad_norm": 2.0930190086364746,
      "learning_rate": 0.0001901435660234855,
      "loss": 1.6109,
      "step": 4698
    },
    {
      "epoch": 0.16828119684137016,
      "grad_norm": 1.5381509065628052,
      "learning_rate": 0.00019013854405248992,
      "loss": 1.6881,
      "step": 4699
    },
    {
      "epoch": 0.16831700897093846,
      "grad_norm": 1.4447983503341675,
      "learning_rate": 0.0001901335208687965,
      "loss": 1.5999,
      "step": 4700
    },
    {
      "epoch": 0.16835282110050676,
      "grad_norm": 1.7236464023590088,
      "learning_rate": 0.00019012849647247277,
      "loss": 1.3609,
      "step": 4701
    },
    {
      "epoch": 0.16838863323007502,
      "grad_norm": 1.4596633911132812,
      "learning_rate": 0.0001901234708635864,
      "loss": 1.5683,
      "step": 4702
    },
    {
      "epoch": 0.16842444535964332,
      "grad_norm": 1.497183084487915,
      "learning_rate": 0.00019011844404220497,
      "loss": 1.502,
      "step": 4703
    },
    {
      "epoch": 0.1684602574892116,
      "grad_norm": 1.4851388931274414,
      "learning_rate": 0.00019011341600839616,
      "loss": 1.4912,
      "step": 4704
    },
    {
      "epoch": 0.1684960696187799,
      "grad_norm": 1.5417276620864868,
      "learning_rate": 0.00019010838676222755,
      "loss": 1.7476,
      "step": 4705
    },
    {
      "epoch": 0.16853188174834816,
      "grad_norm": 2.275900363922119,
      "learning_rate": 0.00019010335630376682,
      "loss": 1.2967,
      "step": 4706
    },
    {
      "epoch": 0.16856769387791645,
      "grad_norm": 1.881659984588623,
      "learning_rate": 0.00019009832463308168,
      "loss": 1.9507,
      "step": 4707
    },
    {
      "epoch": 0.16860350600748472,
      "grad_norm": 2.0647614002227783,
      "learning_rate": 0.00019009329175023978,
      "loss": 1.6887,
      "step": 4708
    },
    {
      "epoch": 0.16863931813705302,
      "grad_norm": 1.464247226715088,
      "learning_rate": 0.00019008825765530886,
      "loss": 1.6486,
      "step": 4709
    },
    {
      "epoch": 0.16867513026662131,
      "grad_norm": 1.4748529195785522,
      "learning_rate": 0.00019008322234835662,
      "loss": 1.5222,
      "step": 4710
    },
    {
      "epoch": 0.16871094239618958,
      "grad_norm": 1.4888544082641602,
      "learning_rate": 0.00019007818582945086,
      "loss": 1.3543,
      "step": 4711
    },
    {
      "epoch": 0.16874675452575788,
      "grad_norm": 1.683009386062622,
      "learning_rate": 0.00019007314809865928,
      "loss": 1.7678,
      "step": 4712
    },
    {
      "epoch": 0.16878256665532615,
      "grad_norm": 2.0621697902679443,
      "learning_rate": 0.00019006810915604967,
      "loss": 1.7165,
      "step": 4713
    },
    {
      "epoch": 0.16881837878489445,
      "grad_norm": 2.0385336875915527,
      "learning_rate": 0.00019006306900168983,
      "loss": 1.6773,
      "step": 4714
    },
    {
      "epoch": 0.16885419091446272,
      "grad_norm": 1.4429692029953003,
      "learning_rate": 0.00019005802763564757,
      "loss": 1.3534,
      "step": 4715
    },
    {
      "epoch": 0.168890003044031,
      "grad_norm": 1.8206114768981934,
      "learning_rate": 0.00019005298505799073,
      "loss": 1.6491,
      "step": 4716
    },
    {
      "epoch": 0.1689258151735993,
      "grad_norm": 1.4741144180297852,
      "learning_rate": 0.00019004794126878713,
      "loss": 1.8381,
      "step": 4717
    },
    {
      "epoch": 0.16896162730316758,
      "grad_norm": 1.921205759048462,
      "learning_rate": 0.00019004289626810462,
      "loss": 1.6051,
      "step": 4718
    },
    {
      "epoch": 0.16899743943273587,
      "grad_norm": 1.435024619102478,
      "learning_rate": 0.00019003785005601112,
      "loss": 1.5402,
      "step": 4719
    },
    {
      "epoch": 0.16903325156230414,
      "grad_norm": 1.4565465450286865,
      "learning_rate": 0.00019003280263257447,
      "loss": 1.6903,
      "step": 4720
    },
    {
      "epoch": 0.16906906369187244,
      "grad_norm": 1.364829182624817,
      "learning_rate": 0.0001900277539978626,
      "loss": 1.5461,
      "step": 4721
    },
    {
      "epoch": 0.1691048758214407,
      "grad_norm": 1.5410417318344116,
      "learning_rate": 0.0001900227041519434,
      "loss": 1.4193,
      "step": 4722
    },
    {
      "epoch": 0.169140687951009,
      "grad_norm": 1.8465443849563599,
      "learning_rate": 0.00019001765309488487,
      "loss": 1.8538,
      "step": 4723
    },
    {
      "epoch": 0.1691765000805773,
      "grad_norm": 1.8663067817687988,
      "learning_rate": 0.00019001260082675492,
      "loss": 1.8133,
      "step": 4724
    },
    {
      "epoch": 0.16921231221014557,
      "grad_norm": 1.7289997339248657,
      "learning_rate": 0.00019000754734762153,
      "loss": 1.6781,
      "step": 4725
    },
    {
      "epoch": 0.16924812433971387,
      "grad_norm": 1.981075644493103,
      "learning_rate": 0.0001900024926575527,
      "loss": 1.5013,
      "step": 4726
    },
    {
      "epoch": 0.16928393646928214,
      "grad_norm": 1.7916004657745361,
      "learning_rate": 0.0001899974367566164,
      "loss": 1.5879,
      "step": 4727
    },
    {
      "epoch": 0.16931974859885043,
      "grad_norm": 2.0685083866119385,
      "learning_rate": 0.00018999237964488074,
      "loss": 1.428,
      "step": 4728
    },
    {
      "epoch": 0.1693555607284187,
      "grad_norm": 1.6859298944473267,
      "learning_rate": 0.0001899873213224136,
      "loss": 1.5873,
      "step": 4729
    },
    {
      "epoch": 0.169391372857987,
      "grad_norm": 1.8945560455322266,
      "learning_rate": 0.0001899822617892832,
      "loss": 1.8325,
      "step": 4730
    },
    {
      "epoch": 0.1694271849875553,
      "grad_norm": 1.4337220191955566,
      "learning_rate": 0.0001899772010455575,
      "loss": 1.5093,
      "step": 4731
    },
    {
      "epoch": 0.16946299711712356,
      "grad_norm": 1.2983472347259521,
      "learning_rate": 0.00018997213909130464,
      "loss": 1.2281,
      "step": 4732
    },
    {
      "epoch": 0.16949880924669186,
      "grad_norm": 2.4921884536743164,
      "learning_rate": 0.0001899670759265927,
      "loss": 1.8038,
      "step": 4733
    },
    {
      "epoch": 0.16953462137626013,
      "grad_norm": 2.076310634613037,
      "learning_rate": 0.00018996201155148983,
      "loss": 1.4219,
      "step": 4734
    },
    {
      "epoch": 0.16957043350582843,
      "grad_norm": 1.6718064546585083,
      "learning_rate": 0.0001899569459660641,
      "loss": 1.8481,
      "step": 4735
    },
    {
      "epoch": 0.1696062456353967,
      "grad_norm": 1.3863492012023926,
      "learning_rate": 0.0001899518791703837,
      "loss": 1.4694,
      "step": 4736
    },
    {
      "epoch": 0.169642057764965,
      "grad_norm": 1.388734221458435,
      "learning_rate": 0.0001899468111645168,
      "loss": 1.6998,
      "step": 4737
    },
    {
      "epoch": 0.1696778698945333,
      "grad_norm": 2.2914044857025146,
      "learning_rate": 0.00018994174194853161,
      "loss": 1.5079,
      "step": 4738
    },
    {
      "epoch": 0.16971368202410156,
      "grad_norm": 1.6915277242660522,
      "learning_rate": 0.00018993667152249625,
      "loss": 1.5748,
      "step": 4739
    },
    {
      "epoch": 0.16974949415366986,
      "grad_norm": 1.5572466850280762,
      "learning_rate": 0.00018993159988647901,
      "loss": 1.8074,
      "step": 4740
    },
    {
      "epoch": 0.16978530628323812,
      "grad_norm": 1.6799663305282593,
      "learning_rate": 0.0001899265270405481,
      "loss": 1.6571,
      "step": 4741
    },
    {
      "epoch": 0.16982111841280642,
      "grad_norm": 1.6444222927093506,
      "learning_rate": 0.00018992145298477175,
      "loss": 1.4765,
      "step": 4742
    },
    {
      "epoch": 0.1698569305423747,
      "grad_norm": 2.0587503910064697,
      "learning_rate": 0.00018991637771921825,
      "loss": 2.0098,
      "step": 4743
    },
    {
      "epoch": 0.169892742671943,
      "grad_norm": 1.695487141609192,
      "learning_rate": 0.00018991130124395585,
      "loss": 1.8735,
      "step": 4744
    },
    {
      "epoch": 0.16992855480151128,
      "grad_norm": 1.8474738597869873,
      "learning_rate": 0.0001899062235590529,
      "loss": 1.5526,
      "step": 4745
    },
    {
      "epoch": 0.16996436693107955,
      "grad_norm": 1.5542140007019043,
      "learning_rate": 0.00018990114466457768,
      "loss": 1.7612,
      "step": 4746
    },
    {
      "epoch": 0.17000017906064785,
      "grad_norm": 1.6513017416000366,
      "learning_rate": 0.0001898960645605985,
      "loss": 1.5931,
      "step": 4747
    },
    {
      "epoch": 0.17003599119021612,
      "grad_norm": 2.3618006706237793,
      "learning_rate": 0.00018989098324718375,
      "loss": 1.5834,
      "step": 4748
    },
    {
      "epoch": 0.17007180331978441,
      "grad_norm": 1.3786261081695557,
      "learning_rate": 0.00018988590072440176,
      "loss": 1.6837,
      "step": 4749
    },
    {
      "epoch": 0.17010761544935268,
      "grad_norm": 1.8375169038772583,
      "learning_rate": 0.00018988081699232095,
      "loss": 1.726,
      "step": 4750
    },
    {
      "epoch": 0.17014342757892098,
      "grad_norm": 1.6182897090911865,
      "learning_rate": 0.00018987573205100965,
      "loss": 1.6424,
      "step": 4751
    },
    {
      "epoch": 0.17017923970848928,
      "grad_norm": 1.8573373556137085,
      "learning_rate": 0.00018987064590053634,
      "loss": 2.111,
      "step": 4752
    },
    {
      "epoch": 0.17021505183805755,
      "grad_norm": 1.8004847764968872,
      "learning_rate": 0.0001898655585409694,
      "loss": 1.5171,
      "step": 4753
    },
    {
      "epoch": 0.17025086396762584,
      "grad_norm": 1.9292305707931519,
      "learning_rate": 0.00018986046997237726,
      "loss": 1.4811,
      "step": 4754
    },
    {
      "epoch": 0.1702866760971941,
      "grad_norm": 1.3681687116622925,
      "learning_rate": 0.00018985538019482842,
      "loss": 1.6534,
      "step": 4755
    },
    {
      "epoch": 0.1703224882267624,
      "grad_norm": 2.1036245822906494,
      "learning_rate": 0.00018985028920839137,
      "loss": 1.6907,
      "step": 4756
    },
    {
      "epoch": 0.17035830035633068,
      "grad_norm": 1.4106781482696533,
      "learning_rate": 0.00018984519701313455,
      "loss": 1.5812,
      "step": 4757
    },
    {
      "epoch": 0.17039411248589897,
      "grad_norm": 1.4277902841567993,
      "learning_rate": 0.0001898401036091265,
      "loss": 1.5905,
      "step": 4758
    },
    {
      "epoch": 0.17042992461546727,
      "grad_norm": 1.5859612226486206,
      "learning_rate": 0.00018983500899643577,
      "loss": 1.9137,
      "step": 4759
    },
    {
      "epoch": 0.17046573674503554,
      "grad_norm": 1.3639755249023438,
      "learning_rate": 0.00018982991317513084,
      "loss": 1.4867,
      "step": 4760
    },
    {
      "epoch": 0.17050154887460384,
      "grad_norm": 2.493811845779419,
      "learning_rate": 0.0001898248161452803,
      "loss": 1.4949,
      "step": 4761
    },
    {
      "epoch": 0.1705373610041721,
      "grad_norm": 1.4881577491760254,
      "learning_rate": 0.00018981971790695275,
      "loss": 1.5378,
      "step": 4762
    },
    {
      "epoch": 0.1705731731337404,
      "grad_norm": 1.687936544418335,
      "learning_rate": 0.0001898146184602167,
      "loss": 1.6787,
      "step": 4763
    },
    {
      "epoch": 0.17060898526330867,
      "grad_norm": 1.3773003816604614,
      "learning_rate": 0.0001898095178051409,
      "loss": 1.4484,
      "step": 4764
    },
    {
      "epoch": 0.17064479739287697,
      "grad_norm": 1.2580550909042358,
      "learning_rate": 0.0001898044159417938,
      "loss": 1.5878,
      "step": 4765
    },
    {
      "epoch": 0.17068060952244526,
      "grad_norm": 1.4113050699234009,
      "learning_rate": 0.00018979931287024416,
      "loss": 1.2658,
      "step": 4766
    },
    {
      "epoch": 0.17071642165201353,
      "grad_norm": 2.062539577484131,
      "learning_rate": 0.00018979420859056062,
      "loss": 1.7766,
      "step": 4767
    },
    {
      "epoch": 0.17075223378158183,
      "grad_norm": 2.4932713508605957,
      "learning_rate": 0.0001897891031028118,
      "loss": 1.9452,
      "step": 4768
    },
    {
      "epoch": 0.1707880459111501,
      "grad_norm": 1.9760795831680298,
      "learning_rate": 0.0001897839964070664,
      "loss": 1.4249,
      "step": 4769
    },
    {
      "epoch": 0.1708238580407184,
      "grad_norm": 2.0726683139801025,
      "learning_rate": 0.00018977888850339319,
      "loss": 1.7297,
      "step": 4770
    },
    {
      "epoch": 0.17085967017028666,
      "grad_norm": 1.313335657119751,
      "learning_rate": 0.0001897737793918608,
      "loss": 1.5473,
      "step": 4771
    },
    {
      "epoch": 0.17089548229985496,
      "grad_norm": 1.480515956878662,
      "learning_rate": 0.00018976866907253803,
      "loss": 1.2051,
      "step": 4772
    },
    {
      "epoch": 0.17093129442942326,
      "grad_norm": 1.6562939882278442,
      "learning_rate": 0.0001897635575454936,
      "loss": 1.6465,
      "step": 4773
    },
    {
      "epoch": 0.17096710655899153,
      "grad_norm": 1.5738717317581177,
      "learning_rate": 0.0001897584448107963,
      "loss": 1.7648,
      "step": 4774
    },
    {
      "epoch": 0.17100291868855982,
      "grad_norm": 1.9761050939559937,
      "learning_rate": 0.0001897533308685149,
      "loss": 1.6079,
      "step": 4775
    },
    {
      "epoch": 0.1710387308181281,
      "grad_norm": 1.6524931192398071,
      "learning_rate": 0.00018974821571871822,
      "loss": 1.5807,
      "step": 4776
    },
    {
      "epoch": 0.1710745429476964,
      "grad_norm": 1.2474535703659058,
      "learning_rate": 0.00018974309936147502,
      "loss": 1.3729,
      "step": 4777
    },
    {
      "epoch": 0.17111035507726466,
      "grad_norm": 1.4474204778671265,
      "learning_rate": 0.00018973798179685423,
      "loss": 1.4076,
      "step": 4778
    },
    {
      "epoch": 0.17114616720683296,
      "grad_norm": 1.3599224090576172,
      "learning_rate": 0.0001897328630249246,
      "loss": 1.5103,
      "step": 4779
    },
    {
      "epoch": 0.17118197933640125,
      "grad_norm": 1.535048484802246,
      "learning_rate": 0.0001897277430457551,
      "loss": 1.3841,
      "step": 4780
    },
    {
      "epoch": 0.17121779146596952,
      "grad_norm": 1.823050856590271,
      "learning_rate": 0.00018972262185941452,
      "loss": 1.7171,
      "step": 4781
    },
    {
      "epoch": 0.17125360359553782,
      "grad_norm": 2.32497239112854,
      "learning_rate": 0.00018971749946597178,
      "loss": 1.4991,
      "step": 4782
    },
    {
      "epoch": 0.1712894157251061,
      "grad_norm": 1.427659511566162,
      "learning_rate": 0.00018971237586549587,
      "loss": 1.5099,
      "step": 4783
    },
    {
      "epoch": 0.17132522785467438,
      "grad_norm": 1.4457180500030518,
      "learning_rate": 0.00018970725105805562,
      "loss": 1.5058,
      "step": 4784
    },
    {
      "epoch": 0.17136103998424265,
      "grad_norm": 1.7552998065948486,
      "learning_rate": 0.00018970212504372004,
      "loss": 1.5277,
      "step": 4785
    },
    {
      "epoch": 0.17139685211381095,
      "grad_norm": 1.6383708715438843,
      "learning_rate": 0.00018969699782255808,
      "loss": 1.8856,
      "step": 4786
    },
    {
      "epoch": 0.17143266424337925,
      "grad_norm": 1.7544535398483276,
      "learning_rate": 0.0001896918693946387,
      "loss": 1.4428,
      "step": 4787
    },
    {
      "epoch": 0.17146847637294751,
      "grad_norm": 1.2920863628387451,
      "learning_rate": 0.0001896867397600309,
      "loss": 1.7034,
      "step": 4788
    },
    {
      "epoch": 0.1715042885025158,
      "grad_norm": 2.2302498817443848,
      "learning_rate": 0.00018968160891880373,
      "loss": 1.9788,
      "step": 4789
    },
    {
      "epoch": 0.17154010063208408,
      "grad_norm": 1.1154465675354004,
      "learning_rate": 0.00018967647687102618,
      "loss": 1.2938,
      "step": 4790
    },
    {
      "epoch": 0.17157591276165238,
      "grad_norm": 2.1247408390045166,
      "learning_rate": 0.00018967134361676732,
      "loss": 1.5288,
      "step": 4791
    },
    {
      "epoch": 0.17161172489122065,
      "grad_norm": 1.328395128250122,
      "learning_rate": 0.00018966620915609618,
      "loss": 1.5552,
      "step": 4792
    },
    {
      "epoch": 0.17164753702078894,
      "grad_norm": 2.552074432373047,
      "learning_rate": 0.00018966107348908188,
      "loss": 1.4236,
      "step": 4793
    },
    {
      "epoch": 0.17168334915035724,
      "grad_norm": 1.6310802698135376,
      "learning_rate": 0.00018965593661579347,
      "loss": 1.6218,
      "step": 4794
    },
    {
      "epoch": 0.1717191612799255,
      "grad_norm": 1.940028429031372,
      "learning_rate": 0.00018965079853630007,
      "loss": 1.7298,
      "step": 4795
    },
    {
      "epoch": 0.1717549734094938,
      "grad_norm": 1.7597527503967285,
      "learning_rate": 0.00018964565925067085,
      "loss": 1.7404,
      "step": 4796
    },
    {
      "epoch": 0.17179078553906207,
      "grad_norm": 1.7288353443145752,
      "learning_rate": 0.0001896405187589749,
      "loss": 1.4283,
      "step": 4797
    },
    {
      "epoch": 0.17182659766863037,
      "grad_norm": 1.8926104307174683,
      "learning_rate": 0.0001896353770612814,
      "loss": 1.7908,
      "step": 4798
    },
    {
      "epoch": 0.17186240979819864,
      "grad_norm": 2.0064892768859863,
      "learning_rate": 0.00018963023415765956,
      "loss": 1.5197,
      "step": 4799
    },
    {
      "epoch": 0.17189822192776694,
      "grad_norm": 1.9775582551956177,
      "learning_rate": 0.00018962509004817846,
      "loss": 1.7128,
      "step": 4800
    },
    {
      "epoch": 0.17193403405733523,
      "grad_norm": 1.5334006547927856,
      "learning_rate": 0.00018961994473290744,
      "loss": 1.5295,
      "step": 4801
    },
    {
      "epoch": 0.1719698461869035,
      "grad_norm": 1.7124851942062378,
      "learning_rate": 0.00018961479821191562,
      "loss": 1.5951,
      "step": 4802
    },
    {
      "epoch": 0.1720056583164718,
      "grad_norm": 1.847869634628296,
      "learning_rate": 0.00018960965048527232,
      "loss": 1.6884,
      "step": 4803
    },
    {
      "epoch": 0.17204147044604007,
      "grad_norm": 1.9723440408706665,
      "learning_rate": 0.00018960450155304677,
      "loss": 1.4846,
      "step": 4804
    },
    {
      "epoch": 0.17207728257560836,
      "grad_norm": 1.7727763652801514,
      "learning_rate": 0.00018959935141530821,
      "loss": 1.5497,
      "step": 4805
    },
    {
      "epoch": 0.17211309470517663,
      "grad_norm": 1.8374532461166382,
      "learning_rate": 0.00018959420007212593,
      "loss": 1.2231,
      "step": 4806
    },
    {
      "epoch": 0.17214890683474493,
      "grad_norm": 1.577351450920105,
      "learning_rate": 0.0001895890475235693,
      "loss": 1.5661,
      "step": 4807
    },
    {
      "epoch": 0.1721847189643132,
      "grad_norm": 1.4347150325775146,
      "learning_rate": 0.00018958389376970758,
      "loss": 1.644,
      "step": 4808
    },
    {
      "epoch": 0.1722205310938815,
      "grad_norm": 1.2892824411392212,
      "learning_rate": 0.00018957873881061014,
      "loss": 1.407,
      "step": 4809
    },
    {
      "epoch": 0.1722563432234498,
      "grad_norm": 1.7509137392044067,
      "learning_rate": 0.00018957358264634627,
      "loss": 1.8629,
      "step": 4810
    },
    {
      "epoch": 0.17229215535301806,
      "grad_norm": 1.6735416650772095,
      "learning_rate": 0.0001895684252769854,
      "loss": 1.3715,
      "step": 4811
    },
    {
      "epoch": 0.17232796748258636,
      "grad_norm": 1.627367615699768,
      "learning_rate": 0.00018956326670259695,
      "loss": 1.3836,
      "step": 4812
    },
    {
      "epoch": 0.17236377961215463,
      "grad_norm": 1.8383601903915405,
      "learning_rate": 0.00018955810692325025,
      "loss": 1.6612,
      "step": 4813
    },
    {
      "epoch": 0.17239959174172292,
      "grad_norm": 1.8072162866592407,
      "learning_rate": 0.00018955294593901476,
      "loss": 1.7922,
      "step": 4814
    },
    {
      "epoch": 0.1724354038712912,
      "grad_norm": 1.667475938796997,
      "learning_rate": 0.00018954778374995988,
      "loss": 1.4375,
      "step": 4815
    },
    {
      "epoch": 0.1724712160008595,
      "grad_norm": 1.460464358329773,
      "learning_rate": 0.00018954262035615505,
      "loss": 1.8729,
      "step": 4816
    },
    {
      "epoch": 0.1725070281304278,
      "grad_norm": 1.6498515605926514,
      "learning_rate": 0.0001895374557576698,
      "loss": 1.7033,
      "step": 4817
    },
    {
      "epoch": 0.17254284025999606,
      "grad_norm": 1.6040598154067993,
      "learning_rate": 0.00018953228995457355,
      "loss": 1.5721,
      "step": 4818
    },
    {
      "epoch": 0.17257865238956435,
      "grad_norm": 1.5885403156280518,
      "learning_rate": 0.00018952712294693585,
      "loss": 1.6447,
      "step": 4819
    },
    {
      "epoch": 0.17261446451913262,
      "grad_norm": 1.9566494226455688,
      "learning_rate": 0.0001895219547348262,
      "loss": 1.316,
      "step": 4820
    },
    {
      "epoch": 0.17265027664870092,
      "grad_norm": 2.1236560344696045,
      "learning_rate": 0.0001895167853183141,
      "loss": 1.8277,
      "step": 4821
    },
    {
      "epoch": 0.1726860887782692,
      "grad_norm": 2.2870054244995117,
      "learning_rate": 0.00018951161469746915,
      "loss": 1.4881,
      "step": 4822
    },
    {
      "epoch": 0.17272190090783748,
      "grad_norm": 1.455335259437561,
      "learning_rate": 0.00018950644287236084,
      "loss": 1.8649,
      "step": 4823
    },
    {
      "epoch": 0.17275771303740578,
      "grad_norm": 1.4912785291671753,
      "learning_rate": 0.00018950126984305885,
      "loss": 1.4861,
      "step": 4824
    },
    {
      "epoch": 0.17279352516697405,
      "grad_norm": 1.774718165397644,
      "learning_rate": 0.0001894960956096327,
      "loss": 1.4081,
      "step": 4825
    },
    {
      "epoch": 0.17282933729654235,
      "grad_norm": 1.8963184356689453,
      "learning_rate": 0.000189490920172152,
      "loss": 1.7869,
      "step": 4826
    },
    {
      "epoch": 0.17286514942611061,
      "grad_norm": 1.4983773231506348,
      "learning_rate": 0.00018948574353068643,
      "loss": 1.7589,
      "step": 4827
    },
    {
      "epoch": 0.1729009615556789,
      "grad_norm": 1.7496103048324585,
      "learning_rate": 0.0001894805656853056,
      "loss": 1.6636,
      "step": 4828
    },
    {
      "epoch": 0.17293677368524718,
      "grad_norm": 1.442915678024292,
      "learning_rate": 0.00018947538663607918,
      "loss": 1.2341,
      "step": 4829
    },
    {
      "epoch": 0.17297258581481548,
      "grad_norm": 2.0862679481506348,
      "learning_rate": 0.00018947020638307687,
      "loss": 1.6988,
      "step": 4830
    },
    {
      "epoch": 0.17300839794438377,
      "grad_norm": 1.8809422254562378,
      "learning_rate": 0.0001894650249263683,
      "loss": 1.7472,
      "step": 4831
    },
    {
      "epoch": 0.17304421007395204,
      "grad_norm": 1.4286421537399292,
      "learning_rate": 0.00018945984226602326,
      "loss": 1.7746,
      "step": 4832
    },
    {
      "epoch": 0.17308002220352034,
      "grad_norm": 1.5385067462921143,
      "learning_rate": 0.00018945465840211143,
      "loss": 1.7762,
      "step": 4833
    },
    {
      "epoch": 0.1731158343330886,
      "grad_norm": 1.8565199375152588,
      "learning_rate": 0.00018944947333470252,
      "loss": 1.6201,
      "step": 4834
    },
    {
      "epoch": 0.1731516464626569,
      "grad_norm": 1.7545292377471924,
      "learning_rate": 0.0001894442870638664,
      "loss": 1.468,
      "step": 4835
    },
    {
      "epoch": 0.17318745859222517,
      "grad_norm": 2.246614456176758,
      "learning_rate": 0.00018943909958967273,
      "loss": 1.7679,
      "step": 4836
    },
    {
      "epoch": 0.17322327072179347,
      "grad_norm": 1.7241929769515991,
      "learning_rate": 0.00018943391091219133,
      "loss": 1.6106,
      "step": 4837
    },
    {
      "epoch": 0.17325908285136177,
      "grad_norm": 1.5218449831008911,
      "learning_rate": 0.00018942872103149206,
      "loss": 1.7442,
      "step": 4838
    },
    {
      "epoch": 0.17329489498093004,
      "grad_norm": 1.9982221126556396,
      "learning_rate": 0.00018942352994764464,
      "loss": 1.6432,
      "step": 4839
    },
    {
      "epoch": 0.17333070711049833,
      "grad_norm": 2.3945224285125732,
      "learning_rate": 0.00018941833766071903,
      "loss": 1.7619,
      "step": 4840
    },
    {
      "epoch": 0.1733665192400666,
      "grad_norm": 1.7721819877624512,
      "learning_rate": 0.00018941314417078502,
      "loss": 1.5272,
      "step": 4841
    },
    {
      "epoch": 0.1734023313696349,
      "grad_norm": 1.349071979522705,
      "learning_rate": 0.00018940794947791247,
      "loss": 1.3999,
      "step": 4842
    },
    {
      "epoch": 0.17343814349920317,
      "grad_norm": 1.555430293083191,
      "learning_rate": 0.0001894027535821713,
      "loss": 1.5955,
      "step": 4843
    },
    {
      "epoch": 0.17347395562877146,
      "grad_norm": 1.8791706562042236,
      "learning_rate": 0.0001893975564836314,
      "loss": 1.5432,
      "step": 4844
    },
    {
      "epoch": 0.17350976775833976,
      "grad_norm": 2.6219892501831055,
      "learning_rate": 0.00018939235818236268,
      "loss": 2.0688,
      "step": 4845
    },
    {
      "epoch": 0.17354557988790803,
      "grad_norm": 2.0889711380004883,
      "learning_rate": 0.00018938715867843512,
      "loss": 1.4781,
      "step": 4846
    },
    {
      "epoch": 0.17358139201747633,
      "grad_norm": 1.7666796445846558,
      "learning_rate": 0.0001893819579719186,
      "loss": 1.4148,
      "step": 4847
    },
    {
      "epoch": 0.1736172041470446,
      "grad_norm": 2.062335968017578,
      "learning_rate": 0.00018937675606288317,
      "loss": 1.4671,
      "step": 4848
    },
    {
      "epoch": 0.1736530162766129,
      "grad_norm": 1.5806844234466553,
      "learning_rate": 0.00018937155295139878,
      "loss": 1.7065,
      "step": 4849
    },
    {
      "epoch": 0.17368882840618116,
      "grad_norm": 1.7525285482406616,
      "learning_rate": 0.00018936634863753537,
      "loss": 1.5066,
      "step": 4850
    },
    {
      "epoch": 0.17372464053574946,
      "grad_norm": 1.3670426607131958,
      "learning_rate": 0.00018936114312136307,
      "loss": 1.5515,
      "step": 4851
    },
    {
      "epoch": 0.17376045266531776,
      "grad_norm": 1.4752345085144043,
      "learning_rate": 0.0001893559364029518,
      "loss": 1.4657,
      "step": 4852
    },
    {
      "epoch": 0.17379626479488602,
      "grad_norm": 1.8495032787322998,
      "learning_rate": 0.00018935072848237172,
      "loss": 1.6514,
      "step": 4853
    },
    {
      "epoch": 0.17383207692445432,
      "grad_norm": 2.1843106746673584,
      "learning_rate": 0.00018934551935969284,
      "loss": 1.4489,
      "step": 4854
    },
    {
      "epoch": 0.1738678890540226,
      "grad_norm": 1.5473827123641968,
      "learning_rate": 0.00018934030903498518,
      "loss": 1.5436,
      "step": 4855
    },
    {
      "epoch": 0.1739037011835909,
      "grad_norm": 1.4357026815414429,
      "learning_rate": 0.00018933509750831897,
      "loss": 1.6933,
      "step": 4856
    },
    {
      "epoch": 0.17393951331315916,
      "grad_norm": 1.5008283853530884,
      "learning_rate": 0.00018932988477976423,
      "loss": 1.6496,
      "step": 4857
    },
    {
      "epoch": 0.17397532544272745,
      "grad_norm": 1.748734474182129,
      "learning_rate": 0.0001893246708493911,
      "loss": 1.6295,
      "step": 4858
    },
    {
      "epoch": 0.17401113757229575,
      "grad_norm": 2.9435081481933594,
      "learning_rate": 0.0001893194557172698,
      "loss": 1.3209,
      "step": 4859
    },
    {
      "epoch": 0.17404694970186402,
      "grad_norm": 1.4894545078277588,
      "learning_rate": 0.0001893142393834704,
      "loss": 1.7977,
      "step": 4860
    },
    {
      "epoch": 0.17408276183143231,
      "grad_norm": 2.0931038856506348,
      "learning_rate": 0.00018930902184806313,
      "loss": 1.9863,
      "step": 4861
    },
    {
      "epoch": 0.17411857396100058,
      "grad_norm": 1.7065221071243286,
      "learning_rate": 0.00018930380311111815,
      "loss": 1.4548,
      "step": 4862
    },
    {
      "epoch": 0.17415438609056888,
      "grad_norm": 1.5173895359039307,
      "learning_rate": 0.0001892985831727057,
      "loss": 1.607,
      "step": 4863
    },
    {
      "epoch": 0.17419019822013715,
      "grad_norm": 1.4496746063232422,
      "learning_rate": 0.000189293362032896,
      "loss": 1.4503,
      "step": 4864
    },
    {
      "epoch": 0.17422601034970545,
      "grad_norm": 1.7697582244873047,
      "learning_rate": 0.00018928813969175932,
      "loss": 1.6072,
      "step": 4865
    },
    {
      "epoch": 0.17426182247927374,
      "grad_norm": 1.6433504819869995,
      "learning_rate": 0.00018928291614936585,
      "loss": 1.7286,
      "step": 4866
    },
    {
      "epoch": 0.174297634608842,
      "grad_norm": 1.8499150276184082,
      "learning_rate": 0.00018927769140578593,
      "loss": 1.5328,
      "step": 4867
    },
    {
      "epoch": 0.1743334467384103,
      "grad_norm": 2.2025258541107178,
      "learning_rate": 0.00018927246546108985,
      "loss": 1.755,
      "step": 4868
    },
    {
      "epoch": 0.17436925886797858,
      "grad_norm": 2.001037120819092,
      "learning_rate": 0.00018926723831534789,
      "loss": 1.4364,
      "step": 4869
    },
    {
      "epoch": 0.17440507099754687,
      "grad_norm": 1.4569237232208252,
      "learning_rate": 0.00018926200996863038,
      "loss": 1.8061,
      "step": 4870
    },
    {
      "epoch": 0.17444088312711514,
      "grad_norm": 1.2852599620819092,
      "learning_rate": 0.00018925678042100766,
      "loss": 1.2263,
      "step": 4871
    },
    {
      "epoch": 0.17447669525668344,
      "grad_norm": 1.772127628326416,
      "learning_rate": 0.0001892515496725501,
      "loss": 1.4574,
      "step": 4872
    },
    {
      "epoch": 0.17451250738625174,
      "grad_norm": 1.3789273500442505,
      "learning_rate": 0.00018924631772332807,
      "loss": 1.4344,
      "step": 4873
    },
    {
      "epoch": 0.17454831951582,
      "grad_norm": 1.595746397972107,
      "learning_rate": 0.00018924108457341195,
      "loss": 1.9308,
      "step": 4874
    },
    {
      "epoch": 0.1745841316453883,
      "grad_norm": 1.9921436309814453,
      "learning_rate": 0.00018923585022287214,
      "loss": 1.5553,
      "step": 4875
    },
    {
      "epoch": 0.17461994377495657,
      "grad_norm": 1.2532308101654053,
      "learning_rate": 0.00018923061467177908,
      "loss": 1.5337,
      "step": 4876
    },
    {
      "epoch": 0.17465575590452487,
      "grad_norm": 1.189430832862854,
      "learning_rate": 0.0001892253779202032,
      "loss": 1.4088,
      "step": 4877
    },
    {
      "epoch": 0.17469156803409314,
      "grad_norm": 2.0476043224334717,
      "learning_rate": 0.00018922013996821492,
      "loss": 1.2659,
      "step": 4878
    },
    {
      "epoch": 0.17472738016366143,
      "grad_norm": 2.3595757484436035,
      "learning_rate": 0.0001892149008158848,
      "loss": 1.4595,
      "step": 4879
    },
    {
      "epoch": 0.17476319229322973,
      "grad_norm": 1.562600016593933,
      "learning_rate": 0.00018920966046328324,
      "loss": 1.5422,
      "step": 4880
    },
    {
      "epoch": 0.174799004422798,
      "grad_norm": 1.5140035152435303,
      "learning_rate": 0.00018920441891048077,
      "loss": 1.628,
      "step": 4881
    },
    {
      "epoch": 0.1748348165523663,
      "grad_norm": 1.220018744468689,
      "learning_rate": 0.00018919917615754792,
      "loss": 1.4796,
      "step": 4882
    },
    {
      "epoch": 0.17487062868193456,
      "grad_norm": 1.8878225088119507,
      "learning_rate": 0.00018919393220455518,
      "loss": 1.2749,
      "step": 4883
    },
    {
      "epoch": 0.17490644081150286,
      "grad_norm": 1.5586538314819336,
      "learning_rate": 0.00018918868705157318,
      "loss": 1.4179,
      "step": 4884
    },
    {
      "epoch": 0.17494225294107113,
      "grad_norm": 1.4380314350128174,
      "learning_rate": 0.00018918344069867243,
      "loss": 1.5866,
      "step": 4885
    },
    {
      "epoch": 0.17497806507063943,
      "grad_norm": 3.5423998832702637,
      "learning_rate": 0.00018917819314592351,
      "loss": 1.9557,
      "step": 4886
    },
    {
      "epoch": 0.17501387720020772,
      "grad_norm": 1.9124844074249268,
      "learning_rate": 0.00018917294439339705,
      "loss": 1.517,
      "step": 4887
    },
    {
      "epoch": 0.175049689329776,
      "grad_norm": 1.3516485691070557,
      "learning_rate": 0.0001891676944411636,
      "loss": 1.5067,
      "step": 4888
    },
    {
      "epoch": 0.1750855014593443,
      "grad_norm": 1.858694076538086,
      "learning_rate": 0.00018916244328929388,
      "loss": 1.4675,
      "step": 4889
    },
    {
      "epoch": 0.17512131358891256,
      "grad_norm": 1.8786976337432861,
      "learning_rate": 0.00018915719093785848,
      "loss": 1.5355,
      "step": 4890
    },
    {
      "epoch": 0.17515712571848085,
      "grad_norm": 1.3482768535614014,
      "learning_rate": 0.00018915193738692812,
      "loss": 1.6203,
      "step": 4891
    },
    {
      "epoch": 0.17519293784804912,
      "grad_norm": 1.3710525035858154,
      "learning_rate": 0.00018914668263657342,
      "loss": 1.3993,
      "step": 4892
    },
    {
      "epoch": 0.17522874997761742,
      "grad_norm": 1.5821046829223633,
      "learning_rate": 0.00018914142668686505,
      "loss": 1.5148,
      "step": 4893
    },
    {
      "epoch": 0.17526456210718572,
      "grad_norm": 2.2228548526763916,
      "learning_rate": 0.0001891361695378738,
      "loss": 1.9352,
      "step": 4894
    },
    {
      "epoch": 0.175300374236754,
      "grad_norm": 3.116131544113159,
      "learning_rate": 0.00018913091118967037,
      "loss": 1.8747,
      "step": 4895
    },
    {
      "epoch": 0.17533618636632228,
      "grad_norm": 1.6172984838485718,
      "learning_rate": 0.00018912565164232552,
      "loss": 1.3994,
      "step": 4896
    },
    {
      "epoch": 0.17537199849589055,
      "grad_norm": 1.4550762176513672,
      "learning_rate": 0.00018912039089590997,
      "loss": 1.4414,
      "step": 4897
    },
    {
      "epoch": 0.17540781062545885,
      "grad_norm": 2.1344704627990723,
      "learning_rate": 0.00018911512895049452,
      "loss": 1.6229,
      "step": 4898
    },
    {
      "epoch": 0.17544362275502712,
      "grad_norm": 1.5842924118041992,
      "learning_rate": 0.00018910986580614997,
      "loss": 1.325,
      "step": 4899
    },
    {
      "epoch": 0.17547943488459541,
      "grad_norm": 1.6697801351547241,
      "learning_rate": 0.00018910460146294707,
      "loss": 1.5925,
      "step": 4900
    },
    {
      "epoch": 0.1755152470141637,
      "grad_norm": 1.509250521659851,
      "learning_rate": 0.00018909933592095674,
      "loss": 1.4572,
      "step": 4901
    },
    {
      "epoch": 0.17555105914373198,
      "grad_norm": 1.783841609954834,
      "learning_rate": 0.0001890940691802498,
      "loss": 1.6234,
      "step": 4902
    },
    {
      "epoch": 0.17558687127330028,
      "grad_norm": 1.7891788482666016,
      "learning_rate": 0.00018908880124089702,
      "loss": 1.458,
      "step": 4903
    },
    {
      "epoch": 0.17562268340286855,
      "grad_norm": 2.8010456562042236,
      "learning_rate": 0.0001890835321029694,
      "loss": 1.5074,
      "step": 4904
    },
    {
      "epoch": 0.17565849553243684,
      "grad_norm": 1.094394564628601,
      "learning_rate": 0.00018907826176653772,
      "loss": 1.3326,
      "step": 4905
    },
    {
      "epoch": 0.1756943076620051,
      "grad_norm": 1.3641574382781982,
      "learning_rate": 0.00018907299023167293,
      "loss": 1.7497,
      "step": 4906
    },
    {
      "epoch": 0.1757301197915734,
      "grad_norm": 1.4977530241012573,
      "learning_rate": 0.00018906771749844595,
      "loss": 1.5956,
      "step": 4907
    },
    {
      "epoch": 0.17576593192114168,
      "grad_norm": 1.44773268699646,
      "learning_rate": 0.00018906244356692775,
      "loss": 1.7386,
      "step": 4908
    },
    {
      "epoch": 0.17580174405070997,
      "grad_norm": 1.46619713306427,
      "learning_rate": 0.00018905716843718926,
      "loss": 1.6829,
      "step": 4909
    },
    {
      "epoch": 0.17583755618027827,
      "grad_norm": 4.56627893447876,
      "learning_rate": 0.00018905189210930142,
      "loss": 1.5221,
      "step": 4910
    },
    {
      "epoch": 0.17587336830984654,
      "grad_norm": 1.9148731231689453,
      "learning_rate": 0.00018904661458333524,
      "loss": 1.5686,
      "step": 4911
    },
    {
      "epoch": 0.17590918043941484,
      "grad_norm": 1.6884065866470337,
      "learning_rate": 0.00018904133585936173,
      "loss": 1.531,
      "step": 4912
    },
    {
      "epoch": 0.1759449925689831,
      "grad_norm": 1.323333740234375,
      "learning_rate": 0.0001890360559374519,
      "loss": 1.7735,
      "step": 4913
    },
    {
      "epoch": 0.1759808046985514,
      "grad_norm": 2.1087372303009033,
      "learning_rate": 0.00018903077481767676,
      "loss": 1.305,
      "step": 4914
    },
    {
      "epoch": 0.17601661682811967,
      "grad_norm": 2.0046706199645996,
      "learning_rate": 0.00018902549250010743,
      "loss": 1.5707,
      "step": 4915
    },
    {
      "epoch": 0.17605242895768797,
      "grad_norm": 2.4085426330566406,
      "learning_rate": 0.0001890202089848149,
      "loss": 1.6196,
      "step": 4916
    },
    {
      "epoch": 0.17608824108725626,
      "grad_norm": 1.8258897066116333,
      "learning_rate": 0.00018901492427187032,
      "loss": 1.5813,
      "step": 4917
    },
    {
      "epoch": 0.17612405321682453,
      "grad_norm": 1.4927031993865967,
      "learning_rate": 0.0001890096383613447,
      "loss": 1.6213,
      "step": 4918
    },
    {
      "epoch": 0.17615986534639283,
      "grad_norm": 1.136582612991333,
      "learning_rate": 0.00018900435125330923,
      "loss": 1.4248,
      "step": 4919
    },
    {
      "epoch": 0.1761956774759611,
      "grad_norm": 1.3010669946670532,
      "learning_rate": 0.00018899906294783504,
      "loss": 1.6543,
      "step": 4920
    },
    {
      "epoch": 0.1762314896055294,
      "grad_norm": 1.9157588481903076,
      "learning_rate": 0.00018899377344499328,
      "loss": 1.5228,
      "step": 4921
    },
    {
      "epoch": 0.17626730173509766,
      "grad_norm": 2.504683494567871,
      "learning_rate": 0.00018898848274485506,
      "loss": 2.0403,
      "step": 4922
    },
    {
      "epoch": 0.17630311386466596,
      "grad_norm": 1.409203052520752,
      "learning_rate": 0.00018898319084749158,
      "loss": 1.2702,
      "step": 4923
    },
    {
      "epoch": 0.17633892599423426,
      "grad_norm": 2.0909388065338135,
      "learning_rate": 0.00018897789775297404,
      "loss": 1.8806,
      "step": 4924
    },
    {
      "epoch": 0.17637473812380253,
      "grad_norm": 1.3817533254623413,
      "learning_rate": 0.0001889726034613737,
      "loss": 1.5544,
      "step": 4925
    },
    {
      "epoch": 0.17641055025337082,
      "grad_norm": 1.9548993110656738,
      "learning_rate": 0.00018896730797276175,
      "loss": 1.6944,
      "step": 4926
    },
    {
      "epoch": 0.1764463623829391,
      "grad_norm": 1.3529974222183228,
      "learning_rate": 0.00018896201128720938,
      "loss": 1.471,
      "step": 4927
    },
    {
      "epoch": 0.1764821745125074,
      "grad_norm": 1.5662931203842163,
      "learning_rate": 0.00018895671340478796,
      "loss": 1.5855,
      "step": 4928
    },
    {
      "epoch": 0.17651798664207566,
      "grad_norm": 1.9040552377700806,
      "learning_rate": 0.00018895141432556867,
      "loss": 1.4631,
      "step": 4929
    },
    {
      "epoch": 0.17655379877164395,
      "grad_norm": 2.3355555534362793,
      "learning_rate": 0.00018894611404962283,
      "loss": 1.6409,
      "step": 4930
    },
    {
      "epoch": 0.17658961090121225,
      "grad_norm": 1.3861925601959229,
      "learning_rate": 0.0001889408125770218,
      "loss": 1.5785,
      "step": 4931
    },
    {
      "epoch": 0.17662542303078052,
      "grad_norm": 1.3609862327575684,
      "learning_rate": 0.00018893550990783684,
      "loss": 1.4329,
      "step": 4932
    },
    {
      "epoch": 0.17666123516034882,
      "grad_norm": 1.739033818244934,
      "learning_rate": 0.00018893020604213932,
      "loss": 1.6007,
      "step": 4933
    },
    {
      "epoch": 0.1766970472899171,
      "grad_norm": 2.2016334533691406,
      "learning_rate": 0.00018892490098000055,
      "loss": 1.4416,
      "step": 4934
    },
    {
      "epoch": 0.17673285941948538,
      "grad_norm": 1.7724460363388062,
      "learning_rate": 0.00018891959472149198,
      "loss": 1.4571,
      "step": 4935
    },
    {
      "epoch": 0.17676867154905365,
      "grad_norm": 2.670675754547119,
      "learning_rate": 0.00018891428726668495,
      "loss": 1.5139,
      "step": 4936
    },
    {
      "epoch": 0.17680448367862195,
      "grad_norm": 1.693932294845581,
      "learning_rate": 0.00018890897861565086,
      "loss": 2.0678,
      "step": 4937
    },
    {
      "epoch": 0.17684029580819025,
      "grad_norm": 1.6531952619552612,
      "learning_rate": 0.00018890366876846119,
      "loss": 1.7776,
      "step": 4938
    },
    {
      "epoch": 0.17687610793775851,
      "grad_norm": 2.104870080947876,
      "learning_rate": 0.00018889835772518731,
      "loss": 1.5346,
      "step": 4939
    },
    {
      "epoch": 0.1769119200673268,
      "grad_norm": 1.7648040056228638,
      "learning_rate": 0.00018889304548590067,
      "loss": 1.8308,
      "step": 4940
    },
    {
      "epoch": 0.17694773219689508,
      "grad_norm": 1.3178796768188477,
      "learning_rate": 0.00018888773205067282,
      "loss": 1.7215,
      "step": 4941
    },
    {
      "epoch": 0.17698354432646338,
      "grad_norm": 1.2029863595962524,
      "learning_rate": 0.00018888241741957514,
      "loss": 1.5563,
      "step": 4942
    },
    {
      "epoch": 0.17701935645603165,
      "grad_norm": 2.1883883476257324,
      "learning_rate": 0.00018887710159267923,
      "loss": 1.7515,
      "step": 4943
    },
    {
      "epoch": 0.17705516858559994,
      "grad_norm": 1.2338509559631348,
      "learning_rate": 0.00018887178457005653,
      "loss": 1.4304,
      "step": 4944
    },
    {
      "epoch": 0.17709098071516824,
      "grad_norm": 1.7541881799697876,
      "learning_rate": 0.00018886646635177864,
      "loss": 1.7429,
      "step": 4945
    },
    {
      "epoch": 0.1771267928447365,
      "grad_norm": 1.4852969646453857,
      "learning_rate": 0.00018886114693791704,
      "loss": 1.4427,
      "step": 4946
    },
    {
      "epoch": 0.1771626049743048,
      "grad_norm": 1.5804656744003296,
      "learning_rate": 0.00018885582632854333,
      "loss": 1.431,
      "step": 4947
    },
    {
      "epoch": 0.17719841710387307,
      "grad_norm": 1.3191192150115967,
      "learning_rate": 0.00018885050452372912,
      "loss": 1.5655,
      "step": 4948
    },
    {
      "epoch": 0.17723422923344137,
      "grad_norm": 2.596315383911133,
      "learning_rate": 0.00018884518152354596,
      "loss": 1.496,
      "step": 4949
    },
    {
      "epoch": 0.17727004136300964,
      "grad_norm": 1.8313298225402832,
      "learning_rate": 0.00018883985732806547,
      "loss": 1.5148,
      "step": 4950
    },
    {
      "epoch": 0.17730585349257794,
      "grad_norm": 1.5022848844528198,
      "learning_rate": 0.00018883453193735932,
      "loss": 1.4678,
      "step": 4951
    },
    {
      "epoch": 0.17734166562214623,
      "grad_norm": 1.8129926919937134,
      "learning_rate": 0.00018882920535149913,
      "loss": 1.5432,
      "step": 4952
    },
    {
      "epoch": 0.1773774777517145,
      "grad_norm": 1.5845561027526855,
      "learning_rate": 0.00018882387757055655,
      "loss": 1.9524,
      "step": 4953
    },
    {
      "epoch": 0.1774132898812828,
      "grad_norm": 1.0808457136154175,
      "learning_rate": 0.00018881854859460328,
      "loss": 1.643,
      "step": 4954
    },
    {
      "epoch": 0.17744910201085107,
      "grad_norm": 1.7996083498001099,
      "learning_rate": 0.00018881321842371103,
      "loss": 1.6271,
      "step": 4955
    },
    {
      "epoch": 0.17748491414041936,
      "grad_norm": 1.7075645923614502,
      "learning_rate": 0.00018880788705795144,
      "loss": 1.8461,
      "step": 4956
    },
    {
      "epoch": 0.17752072626998763,
      "grad_norm": 1.3339205980300903,
      "learning_rate": 0.00018880255449739634,
      "loss": 1.5211,
      "step": 4957
    },
    {
      "epoch": 0.17755653839955593,
      "grad_norm": 1.3496900796890259,
      "learning_rate": 0.00018879722074211736,
      "loss": 1.5706,
      "step": 4958
    },
    {
      "epoch": 0.17759235052912423,
      "grad_norm": 1.2548480033874512,
      "learning_rate": 0.00018879188579218635,
      "loss": 1.5067,
      "step": 4959
    },
    {
      "epoch": 0.1776281626586925,
      "grad_norm": 1.7974036931991577,
      "learning_rate": 0.000188786549647675,
      "loss": 1.38,
      "step": 4960
    },
    {
      "epoch": 0.1776639747882608,
      "grad_norm": 1.8285739421844482,
      "learning_rate": 0.0001887812123086552,
      "loss": 1.8508,
      "step": 4961
    },
    {
      "epoch": 0.17769978691782906,
      "grad_norm": 1.8094350099563599,
      "learning_rate": 0.0001887758737751987,
      "loss": 1.4213,
      "step": 4962
    },
    {
      "epoch": 0.17773559904739736,
      "grad_norm": 1.3000043630599976,
      "learning_rate": 0.00018877053404737734,
      "loss": 1.6764,
      "step": 4963
    },
    {
      "epoch": 0.17777141117696563,
      "grad_norm": 1.428836703300476,
      "learning_rate": 0.00018876519312526293,
      "loss": 1.6104,
      "step": 4964
    },
    {
      "epoch": 0.17780722330653392,
      "grad_norm": 1.7616479396820068,
      "learning_rate": 0.00018875985100892738,
      "loss": 1.2714,
      "step": 4965
    },
    {
      "epoch": 0.17784303543610222,
      "grad_norm": 1.2152507305145264,
      "learning_rate": 0.0001887545076984425,
      "loss": 1.6792,
      "step": 4966
    },
    {
      "epoch": 0.1778788475656705,
      "grad_norm": 1.5839030742645264,
      "learning_rate": 0.0001887491631938802,
      "loss": 1.2338,
      "step": 4967
    },
    {
      "epoch": 0.17791465969523879,
      "grad_norm": 3.108227014541626,
      "learning_rate": 0.0001887438174953124,
      "loss": 1.7161,
      "step": 4968
    },
    {
      "epoch": 0.17795047182480705,
      "grad_norm": 1.628882884979248,
      "learning_rate": 0.000188738470602811,
      "loss": 1.9082,
      "step": 4969
    },
    {
      "epoch": 0.17798628395437535,
      "grad_norm": 1.6627739667892456,
      "learning_rate": 0.00018873312251644793,
      "loss": 1.5598,
      "step": 4970
    },
    {
      "epoch": 0.17802209608394362,
      "grad_norm": 1.4799766540527344,
      "learning_rate": 0.00018872777323629514,
      "loss": 1.3478,
      "step": 4971
    },
    {
      "epoch": 0.17805790821351192,
      "grad_norm": 2.2663679122924805,
      "learning_rate": 0.00018872242276242465,
      "loss": 1.8249,
      "step": 4972
    },
    {
      "epoch": 0.17809372034308021,
      "grad_norm": 1.7324576377868652,
      "learning_rate": 0.0001887170710949084,
      "loss": 1.9787,
      "step": 4973
    },
    {
      "epoch": 0.17812953247264848,
      "grad_norm": 1.4873781204223633,
      "learning_rate": 0.00018871171823381836,
      "loss": 1.9714,
      "step": 4974
    },
    {
      "epoch": 0.17816534460221678,
      "grad_norm": 1.8465592861175537,
      "learning_rate": 0.00018870636417922662,
      "loss": 1.4609,
      "step": 4975
    },
    {
      "epoch": 0.17820115673178505,
      "grad_norm": 2.0763323307037354,
      "learning_rate": 0.00018870100893120516,
      "loss": 1.8321,
      "step": 4976
    },
    {
      "epoch": 0.17823696886135335,
      "grad_norm": 1.5704597234725952,
      "learning_rate": 0.00018869565248982607,
      "loss": 1.4162,
      "step": 4977
    },
    {
      "epoch": 0.17827278099092161,
      "grad_norm": 1.4963994026184082,
      "learning_rate": 0.00018869029485516135,
      "loss": 1.2262,
      "step": 4978
    },
    {
      "epoch": 0.1783085931204899,
      "grad_norm": 1.768388271331787,
      "learning_rate": 0.0001886849360272831,
      "loss": 1.5942,
      "step": 4979
    },
    {
      "epoch": 0.1783444052500582,
      "grad_norm": 1.202764868736267,
      "learning_rate": 0.00018867957600626344,
      "loss": 1.7769,
      "step": 4980
    },
    {
      "epoch": 0.17838021737962648,
      "grad_norm": 1.4093730449676514,
      "learning_rate": 0.0001886742147921745,
      "loss": 1.6073,
      "step": 4981
    },
    {
      "epoch": 0.17841602950919477,
      "grad_norm": 1.5123848915100098,
      "learning_rate": 0.0001886688523850884,
      "loss": 1.5704,
      "step": 4982
    },
    {
      "epoch": 0.17845184163876304,
      "grad_norm": 1.5131558179855347,
      "learning_rate": 0.0001886634887850772,
      "loss": 1.7871,
      "step": 4983
    },
    {
      "epoch": 0.17848765376833134,
      "grad_norm": 1.548387050628662,
      "learning_rate": 0.00018865812399221317,
      "loss": 1.6158,
      "step": 4984
    },
    {
      "epoch": 0.1785234658978996,
      "grad_norm": 1.310096263885498,
      "learning_rate": 0.00018865275800656844,
      "loss": 1.6574,
      "step": 4985
    },
    {
      "epoch": 0.1785592780274679,
      "grad_norm": 1.429352045059204,
      "learning_rate": 0.00018864739082821518,
      "loss": 1.6101,
      "step": 4986
    },
    {
      "epoch": 0.1785950901570362,
      "grad_norm": 1.6432214975357056,
      "learning_rate": 0.00018864202245722566,
      "loss": 1.7259,
      "step": 4987
    },
    {
      "epoch": 0.17863090228660447,
      "grad_norm": 1.8350661993026733,
      "learning_rate": 0.00018863665289367204,
      "loss": 1.6486,
      "step": 4988
    },
    {
      "epoch": 0.17866671441617277,
      "grad_norm": 1.4883739948272705,
      "learning_rate": 0.0001886312821376266,
      "loss": 1.5008,
      "step": 4989
    },
    {
      "epoch": 0.17870252654574104,
      "grad_norm": 1.758785605430603,
      "learning_rate": 0.00018862591018916155,
      "loss": 1.5733,
      "step": 4990
    },
    {
      "epoch": 0.17873833867530933,
      "grad_norm": 1.595152497291565,
      "learning_rate": 0.00018862053704834925,
      "loss": 1.4687,
      "step": 4991
    },
    {
      "epoch": 0.1787741508048776,
      "grad_norm": 2.3240230083465576,
      "learning_rate": 0.00018861516271526191,
      "loss": 1.4684,
      "step": 4992
    },
    {
      "epoch": 0.1788099629344459,
      "grad_norm": 1.5655384063720703,
      "learning_rate": 0.00018860978718997185,
      "loss": 1.8508,
      "step": 4993
    },
    {
      "epoch": 0.1788457750640142,
      "grad_norm": 1.9290118217468262,
      "learning_rate": 0.00018860441047255144,
      "loss": 1.6378,
      "step": 4994
    },
    {
      "epoch": 0.17888158719358246,
      "grad_norm": 1.8785892724990845,
      "learning_rate": 0.00018859903256307297,
      "loss": 1.8559,
      "step": 4995
    },
    {
      "epoch": 0.17891739932315076,
      "grad_norm": 1.6702314615249634,
      "learning_rate": 0.00018859365346160877,
      "loss": 1.6963,
      "step": 4996
    },
    {
      "epoch": 0.17895321145271903,
      "grad_norm": 2.097177743911743,
      "learning_rate": 0.00018858827316823126,
      "loss": 1.4722,
      "step": 4997
    },
    {
      "epoch": 0.17898902358228733,
      "grad_norm": 1.577393889427185,
      "learning_rate": 0.0001885828916830128,
      "loss": 1.5131,
      "step": 4998
    },
    {
      "epoch": 0.1790248357118556,
      "grad_norm": 2.048555374145508,
      "learning_rate": 0.00018857750900602583,
      "loss": 1.8445,
      "step": 4999
    },
    {
      "epoch": 0.1790606478414239,
      "grad_norm": 1.83433997631073,
      "learning_rate": 0.00018857212513734268,
      "loss": 1.348,
      "step": 5000
    },
    {
      "epoch": 0.17909645997099216,
      "grad_norm": 1.6511319875717163,
      "learning_rate": 0.00018856674007703585,
      "loss": 1.7947,
      "step": 5001
    },
    {
      "epoch": 0.17913227210056046,
      "grad_norm": 1.9509414434432983,
      "learning_rate": 0.0001885613538251778,
      "loss": 1.6304,
      "step": 5002
    },
    {
      "epoch": 0.17916808423012875,
      "grad_norm": 1.4391506910324097,
      "learning_rate": 0.00018855596638184095,
      "loss": 1.5117,
      "step": 5003
    },
    {
      "epoch": 0.17920389635969702,
      "grad_norm": 1.9456406831741333,
      "learning_rate": 0.0001885505777470978,
      "loss": 1.6562,
      "step": 5004
    },
    {
      "epoch": 0.17923970848926532,
      "grad_norm": 1.2658982276916504,
      "learning_rate": 0.00018854518792102084,
      "loss": 1.4669,
      "step": 5005
    },
    {
      "epoch": 0.1792755206188336,
      "grad_norm": 1.4607932567596436,
      "learning_rate": 0.00018853979690368263,
      "loss": 1.8091,
      "step": 5006
    },
    {
      "epoch": 0.17931133274840189,
      "grad_norm": 1.4325487613677979,
      "learning_rate": 0.0001885344046951556,
      "loss": 1.6542,
      "step": 5007
    },
    {
      "epoch": 0.17934714487797015,
      "grad_norm": 1.477703332901001,
      "learning_rate": 0.00018852901129551243,
      "loss": 1.5072,
      "step": 5008
    },
    {
      "epoch": 0.17938295700753845,
      "grad_norm": 2.172150135040283,
      "learning_rate": 0.00018852361670482556,
      "loss": 1.7455,
      "step": 5009
    },
    {
      "epoch": 0.17941876913710675,
      "grad_norm": 1.7485425472259521,
      "learning_rate": 0.00018851822092316763,
      "loss": 1.8077,
      "step": 5010
    },
    {
      "epoch": 0.17945458126667502,
      "grad_norm": 1.443163514137268,
      "learning_rate": 0.00018851282395061122,
      "loss": 1.3942,
      "step": 5011
    },
    {
      "epoch": 0.17949039339624331,
      "grad_norm": 1.8629781007766724,
      "learning_rate": 0.00018850742578722894,
      "loss": 1.6342,
      "step": 5012
    },
    {
      "epoch": 0.17952620552581158,
      "grad_norm": 2.062121629714966,
      "learning_rate": 0.00018850202643309338,
      "loss": 1.6435,
      "step": 5013
    },
    {
      "epoch": 0.17956201765537988,
      "grad_norm": 1.654205083847046,
      "learning_rate": 0.00018849662588827723,
      "loss": 1.4998,
      "step": 5014
    },
    {
      "epoch": 0.17959782978494815,
      "grad_norm": 2.017118453979492,
      "learning_rate": 0.00018849122415285315,
      "loss": 1.6594,
      "step": 5015
    },
    {
      "epoch": 0.17963364191451645,
      "grad_norm": 1.514277696609497,
      "learning_rate": 0.00018848582122689376,
      "loss": 1.5694,
      "step": 5016
    },
    {
      "epoch": 0.17966945404408474,
      "grad_norm": 1.299932599067688,
      "learning_rate": 0.0001884804171104718,
      "loss": 1.6352,
      "step": 5017
    },
    {
      "epoch": 0.179705266173653,
      "grad_norm": 2.4143078327178955,
      "learning_rate": 0.00018847501180365995,
      "loss": 1.9637,
      "step": 5018
    },
    {
      "epoch": 0.1797410783032213,
      "grad_norm": 1.4530065059661865,
      "learning_rate": 0.000188469605306531,
      "loss": 1.6145,
      "step": 5019
    },
    {
      "epoch": 0.17977689043278958,
      "grad_norm": 1.9624786376953125,
      "learning_rate": 0.00018846419761915753,
      "loss": 1.3612,
      "step": 5020
    },
    {
      "epoch": 0.17981270256235787,
      "grad_norm": 1.5147650241851807,
      "learning_rate": 0.00018845878874161249,
      "loss": 1.127,
      "step": 5021
    },
    {
      "epoch": 0.17984851469192614,
      "grad_norm": 1.562705636024475,
      "learning_rate": 0.00018845337867396848,
      "loss": 1.6414,
      "step": 5022
    },
    {
      "epoch": 0.17988432682149444,
      "grad_norm": 1.8817349672317505,
      "learning_rate": 0.0001884479674162984,
      "loss": 1.8362,
      "step": 5023
    },
    {
      "epoch": 0.17992013895106274,
      "grad_norm": 1.2586948871612549,
      "learning_rate": 0.00018844255496867497,
      "loss": 1.4049,
      "step": 5024
    },
    {
      "epoch": 0.179955951080631,
      "grad_norm": 1.6511366367340088,
      "learning_rate": 0.00018843714133117106,
      "loss": 1.6536,
      "step": 5025
    },
    {
      "epoch": 0.1799917632101993,
      "grad_norm": 1.656224012374878,
      "learning_rate": 0.0001884317265038595,
      "loss": 1.6308,
      "step": 5026
    },
    {
      "epoch": 0.18002757533976757,
      "grad_norm": 2.134478807449341,
      "learning_rate": 0.00018842631048681313,
      "loss": 1.3212,
      "step": 5027
    },
    {
      "epoch": 0.18006338746933587,
      "grad_norm": 1.6021968126296997,
      "learning_rate": 0.00018842089328010482,
      "loss": 1.7667,
      "step": 5028
    },
    {
      "epoch": 0.18009919959890414,
      "grad_norm": 2.7165796756744385,
      "learning_rate": 0.00018841547488380745,
      "loss": 1.7146,
      "step": 5029
    },
    {
      "epoch": 0.18013501172847243,
      "grad_norm": 1.9333155155181885,
      "learning_rate": 0.00018841005529799388,
      "loss": 1.6024,
      "step": 5030
    },
    {
      "epoch": 0.18017082385804073,
      "grad_norm": 1.365800380706787,
      "learning_rate": 0.00018840463452273707,
      "loss": 1.4516,
      "step": 5031
    },
    {
      "epoch": 0.180206635987609,
      "grad_norm": 1.4623908996582031,
      "learning_rate": 0.00018839921255810996,
      "loss": 1.6984,
      "step": 5032
    },
    {
      "epoch": 0.1802424481171773,
      "grad_norm": 1.6998884677886963,
      "learning_rate": 0.00018839378940418544,
      "loss": 1.4623,
      "step": 5033
    },
    {
      "epoch": 0.18027826024674556,
      "grad_norm": 2.1139144897460938,
      "learning_rate": 0.00018838836506103652,
      "loss": 1.5678,
      "step": 5034
    },
    {
      "epoch": 0.18031407237631386,
      "grad_norm": 2.4886868000030518,
      "learning_rate": 0.00018838293952873616,
      "loss": 1.8086,
      "step": 5035
    },
    {
      "epoch": 0.18034988450588213,
      "grad_norm": 1.4761202335357666,
      "learning_rate": 0.0001883775128073573,
      "loss": 1.8566,
      "step": 5036
    },
    {
      "epoch": 0.18038569663545043,
      "grad_norm": 1.5243207216262817,
      "learning_rate": 0.0001883720848969731,
      "loss": 1.5345,
      "step": 5037
    },
    {
      "epoch": 0.18042150876501872,
      "grad_norm": 2.0490150451660156,
      "learning_rate": 0.00018836665579765642,
      "loss": 1.3139,
      "step": 5038
    },
    {
      "epoch": 0.180457320894587,
      "grad_norm": 1.9801737070083618,
      "learning_rate": 0.0001883612255094804,
      "loss": 1.6783,
      "step": 5039
    },
    {
      "epoch": 0.1804931330241553,
      "grad_norm": 1.588937520980835,
      "learning_rate": 0.00018835579403251806,
      "loss": 1.4642,
      "step": 5040
    },
    {
      "epoch": 0.18052894515372356,
      "grad_norm": 1.7436013221740723,
      "learning_rate": 0.00018835036136684248,
      "loss": 1.7724,
      "step": 5041
    },
    {
      "epoch": 0.18056475728329185,
      "grad_norm": 1.4341626167297363,
      "learning_rate": 0.00018834492751252678,
      "loss": 1.5823,
      "step": 5042
    },
    {
      "epoch": 0.18060056941286012,
      "grad_norm": 1.5773844718933105,
      "learning_rate": 0.000188339492469644,
      "loss": 1.5664,
      "step": 5043
    },
    {
      "epoch": 0.18063638154242842,
      "grad_norm": 2.491431474685669,
      "learning_rate": 0.0001883340562382673,
      "loss": 2.0062,
      "step": 5044
    },
    {
      "epoch": 0.18067219367199672,
      "grad_norm": 1.8883206844329834,
      "learning_rate": 0.00018832861881846983,
      "loss": 1.6656,
      "step": 5045
    },
    {
      "epoch": 0.18070800580156499,
      "grad_norm": 1.7385650873184204,
      "learning_rate": 0.00018832318021032472,
      "loss": 1.637,
      "step": 5046
    },
    {
      "epoch": 0.18074381793113328,
      "grad_norm": 1.5121614933013916,
      "learning_rate": 0.00018831774041390517,
      "loss": 1.1038,
      "step": 5047
    },
    {
      "epoch": 0.18077963006070155,
      "grad_norm": 1.1999789476394653,
      "learning_rate": 0.00018831229942928434,
      "loss": 1.5409,
      "step": 5048
    },
    {
      "epoch": 0.18081544219026985,
      "grad_norm": 1.249335527420044,
      "learning_rate": 0.0001883068572565354,
      "loss": 1.3306,
      "step": 5049
    },
    {
      "epoch": 0.18085125431983812,
      "grad_norm": 1.4928827285766602,
      "learning_rate": 0.00018830141389573166,
      "loss": 1.699,
      "step": 5050
    },
    {
      "epoch": 0.18088706644940641,
      "grad_norm": 1.2425192594528198,
      "learning_rate": 0.00018829596934694624,
      "loss": 1.6774,
      "step": 5051
    },
    {
      "epoch": 0.1809228785789747,
      "grad_norm": 1.6975491046905518,
      "learning_rate": 0.0001882905236102525,
      "loss": 1.6176,
      "step": 5052
    },
    {
      "epoch": 0.18095869070854298,
      "grad_norm": 1.3166990280151367,
      "learning_rate": 0.0001882850766857236,
      "loss": 1.3818,
      "step": 5053
    },
    {
      "epoch": 0.18099450283811128,
      "grad_norm": 1.856614351272583,
      "learning_rate": 0.0001882796285734329,
      "loss": 1.5886,
      "step": 5054
    },
    {
      "epoch": 0.18103031496767955,
      "grad_norm": 1.364363670349121,
      "learning_rate": 0.0001882741792734537,
      "loss": 1.4846,
      "step": 5055
    },
    {
      "epoch": 0.18106612709724784,
      "grad_norm": 1.5802984237670898,
      "learning_rate": 0.00018826872878585925,
      "loss": 1.5028,
      "step": 5056
    },
    {
      "epoch": 0.1811019392268161,
      "grad_norm": 1.4129685163497925,
      "learning_rate": 0.0001882632771107229,
      "loss": 1.6633,
      "step": 5057
    },
    {
      "epoch": 0.1811377513563844,
      "grad_norm": 1.0658329725265503,
      "learning_rate": 0.00018825782424811802,
      "loss": 1.4006,
      "step": 5058
    },
    {
      "epoch": 0.1811735634859527,
      "grad_norm": 1.5389435291290283,
      "learning_rate": 0.00018825237019811796,
      "loss": 1.7775,
      "step": 5059
    },
    {
      "epoch": 0.18120937561552097,
      "grad_norm": 1.2066320180892944,
      "learning_rate": 0.0001882469149607961,
      "loss": 1.5404,
      "step": 5060
    },
    {
      "epoch": 0.18124518774508927,
      "grad_norm": 1.2888737916946411,
      "learning_rate": 0.00018824145853622582,
      "loss": 1.6816,
      "step": 5061
    },
    {
      "epoch": 0.18128099987465754,
      "grad_norm": 1.9192713499069214,
      "learning_rate": 0.00018823600092448054,
      "loss": 1.6496,
      "step": 5062
    },
    {
      "epoch": 0.18131681200422584,
      "grad_norm": 1.499048113822937,
      "learning_rate": 0.0001882305421256337,
      "loss": 1.4216,
      "step": 5063
    },
    {
      "epoch": 0.1813526241337941,
      "grad_norm": 1.7019413709640503,
      "learning_rate": 0.0001882250821397587,
      "loss": 1.5245,
      "step": 5064
    },
    {
      "epoch": 0.1813884362633624,
      "grad_norm": 1.9641376733779907,
      "learning_rate": 0.00018821962096692905,
      "loss": 1.622,
      "step": 5065
    },
    {
      "epoch": 0.1814242483929307,
      "grad_norm": 2.3913800716400146,
      "learning_rate": 0.00018821415860721818,
      "loss": 1.2029,
      "step": 5066
    },
    {
      "epoch": 0.18146006052249897,
      "grad_norm": 1.3694190979003906,
      "learning_rate": 0.0001882086950606996,
      "loss": 1.6648,
      "step": 5067
    },
    {
      "epoch": 0.18149587265206726,
      "grad_norm": 1.5379868745803833,
      "learning_rate": 0.0001882032303274468,
      "loss": 1.4357,
      "step": 5068
    },
    {
      "epoch": 0.18153168478163553,
      "grad_norm": 1.7823456525802612,
      "learning_rate": 0.0001881977644075333,
      "loss": 1.4252,
      "step": 5069
    },
    {
      "epoch": 0.18156749691120383,
      "grad_norm": 1.8627572059631348,
      "learning_rate": 0.00018819229730103267,
      "loss": 1.6875,
      "step": 5070
    },
    {
      "epoch": 0.1816033090407721,
      "grad_norm": 1.8082793951034546,
      "learning_rate": 0.00018818682900801842,
      "loss": 1.451,
      "step": 5071
    },
    {
      "epoch": 0.1816391211703404,
      "grad_norm": 1.3875577449798584,
      "learning_rate": 0.00018818135952856414,
      "loss": 1.6158,
      "step": 5072
    },
    {
      "epoch": 0.1816749332999087,
      "grad_norm": 1.2505030632019043,
      "learning_rate": 0.00018817588886274345,
      "loss": 1.7009,
      "step": 5073
    },
    {
      "epoch": 0.18171074542947696,
      "grad_norm": 1.660299301147461,
      "learning_rate": 0.00018817041701062987,
      "loss": 1.666,
      "step": 5074
    },
    {
      "epoch": 0.18174655755904526,
      "grad_norm": 1.854568600654602,
      "learning_rate": 0.00018816494397229708,
      "loss": 1.6123,
      "step": 5075
    },
    {
      "epoch": 0.18178236968861353,
      "grad_norm": 2.6197986602783203,
      "learning_rate": 0.0001881594697478187,
      "loss": 1.5637,
      "step": 5076
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 2.150740623474121,
      "learning_rate": 0.00018815399433726835,
      "loss": 1.6449,
      "step": 5077
    },
    {
      "epoch": 0.1818539939477501,
      "grad_norm": 3.9189467430114746,
      "learning_rate": 0.00018814851774071972,
      "loss": 1.6728,
      "step": 5078
    },
    {
      "epoch": 0.1818898060773184,
      "grad_norm": 1.6135289669036865,
      "learning_rate": 0.00018814303995824653,
      "loss": 1.6037,
      "step": 5079
    },
    {
      "epoch": 0.18192561820688669,
      "grad_norm": 1.3914783000946045,
      "learning_rate": 0.00018813756098992238,
      "loss": 1.5109,
      "step": 5080
    },
    {
      "epoch": 0.18196143033645495,
      "grad_norm": 1.7650524377822876,
      "learning_rate": 0.00018813208083582106,
      "loss": 1.44,
      "step": 5081
    },
    {
      "epoch": 0.18199724246602325,
      "grad_norm": 1.4060581922531128,
      "learning_rate": 0.00018812659949601627,
      "loss": 1.5721,
      "step": 5082
    },
    {
      "epoch": 0.18203305459559152,
      "grad_norm": 1.284258246421814,
      "learning_rate": 0.00018812111697058174,
      "loss": 1.5441,
      "step": 5083
    },
    {
      "epoch": 0.18206886672515982,
      "grad_norm": 2.5756237506866455,
      "learning_rate": 0.0001881156332595913,
      "loss": 1.463,
      "step": 5084
    },
    {
      "epoch": 0.18210467885472809,
      "grad_norm": 1.4949012994766235,
      "learning_rate": 0.00018811014836311865,
      "loss": 1.4694,
      "step": 5085
    },
    {
      "epoch": 0.18214049098429638,
      "grad_norm": 1.3897700309753418,
      "learning_rate": 0.00018810466228123758,
      "loss": 1.5629,
      "step": 5086
    },
    {
      "epoch": 0.18217630311386468,
      "grad_norm": 1.605708360671997,
      "learning_rate": 0.00018809917501402197,
      "loss": 1.4193,
      "step": 5087
    },
    {
      "epoch": 0.18221211524343295,
      "grad_norm": 1.3950620889663696,
      "learning_rate": 0.00018809368656154556,
      "loss": 1.4448,
      "step": 5088
    },
    {
      "epoch": 0.18224792737300125,
      "grad_norm": 1.950050950050354,
      "learning_rate": 0.00018808819692388225,
      "loss": 1.4997,
      "step": 5089
    },
    {
      "epoch": 0.18228373950256951,
      "grad_norm": 1.9414753913879395,
      "learning_rate": 0.00018808270610110584,
      "loss": 1.3708,
      "step": 5090
    },
    {
      "epoch": 0.1823195516321378,
      "grad_norm": 1.3951416015625,
      "learning_rate": 0.0001880772140932903,
      "loss": 1.6992,
      "step": 5091
    },
    {
      "epoch": 0.18235536376170608,
      "grad_norm": 1.6158993244171143,
      "learning_rate": 0.00018807172090050945,
      "loss": 1.6278,
      "step": 5092
    },
    {
      "epoch": 0.18239117589127438,
      "grad_norm": 1.6947451829910278,
      "learning_rate": 0.00018806622652283713,
      "loss": 1.5909,
      "step": 5093
    },
    {
      "epoch": 0.18242698802084267,
      "grad_norm": 1.4441710710525513,
      "learning_rate": 0.0001880607309603474,
      "loss": 1.4248,
      "step": 5094
    },
    {
      "epoch": 0.18246280015041094,
      "grad_norm": 1.387609839439392,
      "learning_rate": 0.00018805523421311411,
      "loss": 1.6221,
      "step": 5095
    },
    {
      "epoch": 0.18249861227997924,
      "grad_norm": 1.3998216390609741,
      "learning_rate": 0.0001880497362812112,
      "loss": 1.7985,
      "step": 5096
    },
    {
      "epoch": 0.1825344244095475,
      "grad_norm": 2.511861562728882,
      "learning_rate": 0.00018804423716471268,
      "loss": 1.4662,
      "step": 5097
    },
    {
      "epoch": 0.1825702365391158,
      "grad_norm": 1.66793954372406,
      "learning_rate": 0.00018803873686369253,
      "loss": 1.3022,
      "step": 5098
    },
    {
      "epoch": 0.18260604866868407,
      "grad_norm": 1.4244165420532227,
      "learning_rate": 0.00018803323537822472,
      "loss": 1.5109,
      "step": 5099
    },
    {
      "epoch": 0.18264186079825237,
      "grad_norm": 1.2689194679260254,
      "learning_rate": 0.00018802773270838329,
      "loss": 1.4388,
      "step": 5100
    },
    {
      "epoch": 0.18267767292782064,
      "grad_norm": 1.5355168581008911,
      "learning_rate": 0.0001880222288542423,
      "loss": 1.3934,
      "step": 5101
    },
    {
      "epoch": 0.18271348505738894,
      "grad_norm": 1.695492148399353,
      "learning_rate": 0.0001880167238158757,
      "loss": 1.6925,
      "step": 5102
    },
    {
      "epoch": 0.18274929718695723,
      "grad_norm": 1.5693861246109009,
      "learning_rate": 0.00018801121759335764,
      "loss": 1.6833,
      "step": 5103
    },
    {
      "epoch": 0.1827851093165255,
      "grad_norm": 1.766347050666809,
      "learning_rate": 0.0001880057101867622,
      "loss": 1.4459,
      "step": 5104
    },
    {
      "epoch": 0.1828209214460938,
      "grad_norm": 1.3743687868118286,
      "learning_rate": 0.00018800020159616342,
      "loss": 1.7515,
      "step": 5105
    },
    {
      "epoch": 0.18285673357566207,
      "grad_norm": 1.5413109064102173,
      "learning_rate": 0.00018799469182163544,
      "loss": 1.5712,
      "step": 5106
    },
    {
      "epoch": 0.18289254570523036,
      "grad_norm": 2.0827620029449463,
      "learning_rate": 0.00018798918086325236,
      "loss": 1.8366,
      "step": 5107
    },
    {
      "epoch": 0.18292835783479863,
      "grad_norm": 1.954444408416748,
      "learning_rate": 0.0001879836687210884,
      "loss": 1.8004,
      "step": 5108
    },
    {
      "epoch": 0.18296416996436693,
      "grad_norm": 1.6314371824264526,
      "learning_rate": 0.00018797815539521763,
      "loss": 1.6223,
      "step": 5109
    },
    {
      "epoch": 0.18299998209393523,
      "grad_norm": 1.643937349319458,
      "learning_rate": 0.00018797264088571427,
      "loss": 1.9055,
      "step": 5110
    },
    {
      "epoch": 0.1830357942235035,
      "grad_norm": 2.121596574783325,
      "learning_rate": 0.0001879671251926525,
      "loss": 1.7183,
      "step": 5111
    },
    {
      "epoch": 0.1830716063530718,
      "grad_norm": 1.6858540773391724,
      "learning_rate": 0.00018796160831610655,
      "loss": 1.7874,
      "step": 5112
    },
    {
      "epoch": 0.18310741848264006,
      "grad_norm": 1.651357889175415,
      "learning_rate": 0.00018795609025615062,
      "loss": 1.6096,
      "step": 5113
    },
    {
      "epoch": 0.18314323061220836,
      "grad_norm": 1.636696457862854,
      "learning_rate": 0.00018795057101285895,
      "loss": 1.7267,
      "step": 5114
    },
    {
      "epoch": 0.18317904274177663,
      "grad_norm": 2.019561529159546,
      "learning_rate": 0.0001879450505863058,
      "loss": 1.5251,
      "step": 5115
    },
    {
      "epoch": 0.18321485487134492,
      "grad_norm": 1.834036946296692,
      "learning_rate": 0.00018793952897656544,
      "loss": 1.4337,
      "step": 5116
    },
    {
      "epoch": 0.18325066700091322,
      "grad_norm": 1.3989968299865723,
      "learning_rate": 0.00018793400618371213,
      "loss": 1.2123,
      "step": 5117
    },
    {
      "epoch": 0.1832864791304815,
      "grad_norm": 1.2780959606170654,
      "learning_rate": 0.0001879284822078202,
      "loss": 1.4671,
      "step": 5118
    },
    {
      "epoch": 0.18332229126004979,
      "grad_norm": 2.15604567527771,
      "learning_rate": 0.00018792295704896396,
      "loss": 1.5538,
      "step": 5119
    },
    {
      "epoch": 0.18335810338961805,
      "grad_norm": 1.4170936346054077,
      "learning_rate": 0.00018791743070721776,
      "loss": 1.9131,
      "step": 5120
    },
    {
      "epoch": 0.18339391551918635,
      "grad_norm": 1.7687115669250488,
      "learning_rate": 0.0001879119031826559,
      "loss": 1.6419,
      "step": 5121
    },
    {
      "epoch": 0.18342972764875462,
      "grad_norm": 1.4007076025009155,
      "learning_rate": 0.00018790637447535283,
      "loss": 1.701,
      "step": 5122
    },
    {
      "epoch": 0.18346553977832292,
      "grad_norm": 1.8551342487335205,
      "learning_rate": 0.00018790084458538285,
      "loss": 1.7679,
      "step": 5123
    },
    {
      "epoch": 0.1835013519078912,
      "grad_norm": 1.933009386062622,
      "learning_rate": 0.0001878953135128204,
      "loss": 1.6205,
      "step": 5124
    },
    {
      "epoch": 0.18353716403745948,
      "grad_norm": 1.9234230518341064,
      "learning_rate": 0.00018788978125773987,
      "loss": 1.4022,
      "step": 5125
    },
    {
      "epoch": 0.18357297616702778,
      "grad_norm": 1.5829157829284668,
      "learning_rate": 0.0001878842478202157,
      "loss": 1.4791,
      "step": 5126
    },
    {
      "epoch": 0.18360878829659605,
      "grad_norm": 1.4500535726547241,
      "learning_rate": 0.00018787871320032236,
      "loss": 1.6684,
      "step": 5127
    },
    {
      "epoch": 0.18364460042616434,
      "grad_norm": 2.3648834228515625,
      "learning_rate": 0.0001878731773981343,
      "loss": 1.9178,
      "step": 5128
    },
    {
      "epoch": 0.18368041255573261,
      "grad_norm": 2.026014566421509,
      "learning_rate": 0.00018786764041372594,
      "loss": 1.7238,
      "step": 5129
    },
    {
      "epoch": 0.1837162246853009,
      "grad_norm": 2.3626532554626465,
      "learning_rate": 0.00018786210224717184,
      "loss": 1.6493,
      "step": 5130
    },
    {
      "epoch": 0.1837520368148692,
      "grad_norm": 1.9708654880523682,
      "learning_rate": 0.0001878565628985465,
      "loss": 1.7803,
      "step": 5131
    },
    {
      "epoch": 0.18378784894443748,
      "grad_norm": 2.1790413856506348,
      "learning_rate": 0.00018785102236792444,
      "loss": 1.5954,
      "step": 5132
    },
    {
      "epoch": 0.18382366107400577,
      "grad_norm": 1.556118369102478,
      "learning_rate": 0.00018784548065538018,
      "loss": 1.7467,
      "step": 5133
    },
    {
      "epoch": 0.18385947320357404,
      "grad_norm": 1.646291971206665,
      "learning_rate": 0.0001878399377609883,
      "loss": 1.3901,
      "step": 5134
    },
    {
      "epoch": 0.18389528533314234,
      "grad_norm": 1.8888112306594849,
      "learning_rate": 0.00018783439368482335,
      "loss": 1.8621,
      "step": 5135
    },
    {
      "epoch": 0.1839310974627106,
      "grad_norm": 1.5839077234268188,
      "learning_rate": 0.00018782884842695992,
      "loss": 1.3945,
      "step": 5136
    },
    {
      "epoch": 0.1839669095922789,
      "grad_norm": 1.7424887418746948,
      "learning_rate": 0.00018782330198747265,
      "loss": 1.5983,
      "step": 5137
    },
    {
      "epoch": 0.1840027217218472,
      "grad_norm": 1.7385514974594116,
      "learning_rate": 0.0001878177543664361,
      "loss": 1.8294,
      "step": 5138
    },
    {
      "epoch": 0.18403853385141547,
      "grad_norm": 1.573832392692566,
      "learning_rate": 0.00018781220556392497,
      "loss": 1.6636,
      "step": 5139
    },
    {
      "epoch": 0.18407434598098377,
      "grad_norm": 2.315880537033081,
      "learning_rate": 0.00018780665558001388,
      "loss": 1.6846,
      "step": 5140
    },
    {
      "epoch": 0.18411015811055204,
      "grad_norm": 1.4497591257095337,
      "learning_rate": 0.00018780110441477752,
      "loss": 1.4998,
      "step": 5141
    },
    {
      "epoch": 0.18414597024012033,
      "grad_norm": 1.6672792434692383,
      "learning_rate": 0.00018779555206829054,
      "loss": 1.6283,
      "step": 5142
    },
    {
      "epoch": 0.1841817823696886,
      "grad_norm": 1.4344167709350586,
      "learning_rate": 0.00018778999854062765,
      "loss": 1.3823,
      "step": 5143
    },
    {
      "epoch": 0.1842175944992569,
      "grad_norm": 1.8563214540481567,
      "learning_rate": 0.00018778444383186357,
      "loss": 1.5636,
      "step": 5144
    },
    {
      "epoch": 0.1842534066288252,
      "grad_norm": 1.5528720617294312,
      "learning_rate": 0.00018777888794207302,
      "loss": 1.5238,
      "step": 5145
    },
    {
      "epoch": 0.18428921875839346,
      "grad_norm": 1.825433373451233,
      "learning_rate": 0.0001877733308713308,
      "loss": 1.3471,
      "step": 5146
    },
    {
      "epoch": 0.18432503088796176,
      "grad_norm": 1.9533745050430298,
      "learning_rate": 0.00018776777261971162,
      "loss": 1.5054,
      "step": 5147
    },
    {
      "epoch": 0.18436084301753003,
      "grad_norm": 2.1287336349487305,
      "learning_rate": 0.00018776221318729026,
      "loss": 1.8556,
      "step": 5148
    },
    {
      "epoch": 0.18439665514709833,
      "grad_norm": 3.0732877254486084,
      "learning_rate": 0.00018775665257414153,
      "loss": 1.6433,
      "step": 5149
    },
    {
      "epoch": 0.1844324672766666,
      "grad_norm": 1.8303265571594238,
      "learning_rate": 0.00018775109078034022,
      "loss": 1.4745,
      "step": 5150
    },
    {
      "epoch": 0.1844682794062349,
      "grad_norm": 1.7558493614196777,
      "learning_rate": 0.00018774552780596117,
      "loss": 1.6524,
      "step": 5151
    },
    {
      "epoch": 0.1845040915358032,
      "grad_norm": 1.354691505432129,
      "learning_rate": 0.00018773996365107926,
      "loss": 1.0427,
      "step": 5152
    },
    {
      "epoch": 0.18453990366537146,
      "grad_norm": 1.8739055395126343,
      "learning_rate": 0.00018773439831576929,
      "loss": 1.2968,
      "step": 5153
    },
    {
      "epoch": 0.18457571579493975,
      "grad_norm": 1.5816646814346313,
      "learning_rate": 0.00018772883180010616,
      "loss": 1.5516,
      "step": 5154
    },
    {
      "epoch": 0.18461152792450802,
      "grad_norm": 1.3845489025115967,
      "learning_rate": 0.0001877232641041648,
      "loss": 1.3547,
      "step": 5155
    },
    {
      "epoch": 0.18464734005407632,
      "grad_norm": 1.0942845344543457,
      "learning_rate": 0.00018771769522802004,
      "loss": 1.6058,
      "step": 5156
    },
    {
      "epoch": 0.1846831521836446,
      "grad_norm": 2.4216606616973877,
      "learning_rate": 0.00018771212517174686,
      "loss": 1.8014,
      "step": 5157
    },
    {
      "epoch": 0.18471896431321289,
      "grad_norm": 2.278639554977417,
      "learning_rate": 0.00018770655393542012,
      "loss": 1.574,
      "step": 5158
    },
    {
      "epoch": 0.18475477644278118,
      "grad_norm": 1.3788586854934692,
      "learning_rate": 0.0001877009815191149,
      "loss": 1.4745,
      "step": 5159
    },
    {
      "epoch": 0.18479058857234945,
      "grad_norm": 1.4675894975662231,
      "learning_rate": 0.00018769540792290608,
      "loss": 1.5777,
      "step": 5160
    },
    {
      "epoch": 0.18482640070191775,
      "grad_norm": 1.862516164779663,
      "learning_rate": 0.00018768983314686866,
      "loss": 1.5242,
      "step": 5161
    },
    {
      "epoch": 0.18486221283148602,
      "grad_norm": 1.9851064682006836,
      "learning_rate": 0.00018768425719107765,
      "loss": 1.5826,
      "step": 5162
    },
    {
      "epoch": 0.1848980249610543,
      "grad_norm": 1.6969146728515625,
      "learning_rate": 0.00018767868005560806,
      "loss": 1.6246,
      "step": 5163
    },
    {
      "epoch": 0.18493383709062258,
      "grad_norm": 1.3387763500213623,
      "learning_rate": 0.0001876731017405349,
      "loss": 1.4774,
      "step": 5164
    },
    {
      "epoch": 0.18496964922019088,
      "grad_norm": 1.9716835021972656,
      "learning_rate": 0.0001876675222459333,
      "loss": 1.3885,
      "step": 5165
    },
    {
      "epoch": 0.18500546134975918,
      "grad_norm": 3.473733901977539,
      "learning_rate": 0.0001876619415718782,
      "loss": 1.4541,
      "step": 5166
    },
    {
      "epoch": 0.18504127347932744,
      "grad_norm": 1.877964973449707,
      "learning_rate": 0.00018765635971844483,
      "loss": 1.563,
      "step": 5167
    },
    {
      "epoch": 0.18507708560889574,
      "grad_norm": 2.445122718811035,
      "learning_rate": 0.00018765077668570816,
      "loss": 1.7484,
      "step": 5168
    },
    {
      "epoch": 0.185112897738464,
      "grad_norm": 1.877567172050476,
      "learning_rate": 0.00018764519247374336,
      "loss": 1.5732,
      "step": 5169
    },
    {
      "epoch": 0.1851487098680323,
      "grad_norm": 2.211268663406372,
      "learning_rate": 0.00018763960708262557,
      "loss": 1.684,
      "step": 5170
    },
    {
      "epoch": 0.18518452199760058,
      "grad_norm": 2.7970688343048096,
      "learning_rate": 0.0001876340205124299,
      "loss": 1.4222,
      "step": 5171
    },
    {
      "epoch": 0.18522033412716887,
      "grad_norm": 1.5880224704742432,
      "learning_rate": 0.00018762843276323151,
      "loss": 1.824,
      "step": 5172
    },
    {
      "epoch": 0.18525614625673717,
      "grad_norm": 1.6042425632476807,
      "learning_rate": 0.0001876228438351056,
      "loss": 1.6025,
      "step": 5173
    },
    {
      "epoch": 0.18529195838630544,
      "grad_norm": 1.544500470161438,
      "learning_rate": 0.00018761725372812735,
      "loss": 1.7088,
      "step": 5174
    },
    {
      "epoch": 0.18532777051587374,
      "grad_norm": 2.324397325515747,
      "learning_rate": 0.00018761166244237197,
      "loss": 1.5568,
      "step": 5175
    },
    {
      "epoch": 0.185363582645442,
      "grad_norm": 3.774132013320923,
      "learning_rate": 0.00018760606997791468,
      "loss": 1.2232,
      "step": 5176
    },
    {
      "epoch": 0.1853993947750103,
      "grad_norm": 1.6885324716567993,
      "learning_rate": 0.00018760047633483074,
      "loss": 1.7845,
      "step": 5177
    },
    {
      "epoch": 0.18543520690457857,
      "grad_norm": 1.3358042240142822,
      "learning_rate": 0.00018759488151319539,
      "loss": 1.4645,
      "step": 5178
    },
    {
      "epoch": 0.18547101903414687,
      "grad_norm": 1.6287620067596436,
      "learning_rate": 0.00018758928551308385,
      "loss": 1.5811,
      "step": 5179
    },
    {
      "epoch": 0.18550683116371516,
      "grad_norm": 1.919089674949646,
      "learning_rate": 0.0001875836883345715,
      "loss": 1.4477,
      "step": 5180
    },
    {
      "epoch": 0.18554264329328343,
      "grad_norm": 1.681052803993225,
      "learning_rate": 0.00018757808997773358,
      "loss": 1.3891,
      "step": 5181
    },
    {
      "epoch": 0.18557845542285173,
      "grad_norm": 1.7119311094284058,
      "learning_rate": 0.00018757249044264542,
      "loss": 1.5623,
      "step": 5182
    },
    {
      "epoch": 0.18561426755242,
      "grad_norm": 1.74424147605896,
      "learning_rate": 0.00018756688972938239,
      "loss": 1.5426,
      "step": 5183
    },
    {
      "epoch": 0.1856500796819883,
      "grad_norm": 1.6085398197174072,
      "learning_rate": 0.0001875612878380198,
      "loss": 1.9637,
      "step": 5184
    },
    {
      "epoch": 0.18568589181155656,
      "grad_norm": 1.232580542564392,
      "learning_rate": 0.00018755568476863302,
      "loss": 1.7271,
      "step": 5185
    },
    {
      "epoch": 0.18572170394112486,
      "grad_norm": 1.6930204629898071,
      "learning_rate": 0.00018755008052129743,
      "loss": 1.7984,
      "step": 5186
    },
    {
      "epoch": 0.18575751607069316,
      "grad_norm": 2.1576590538024902,
      "learning_rate": 0.00018754447509608847,
      "loss": 1.378,
      "step": 5187
    },
    {
      "epoch": 0.18579332820026143,
      "grad_norm": 1.799008846282959,
      "learning_rate": 0.0001875388684930815,
      "loss": 1.2764,
      "step": 5188
    },
    {
      "epoch": 0.18582914032982972,
      "grad_norm": 1.5881048440933228,
      "learning_rate": 0.00018753326071235197,
      "loss": 1.6597,
      "step": 5189
    },
    {
      "epoch": 0.185864952459398,
      "grad_norm": 2.208606719970703,
      "learning_rate": 0.00018752765175397533,
      "loss": 1.6626,
      "step": 5190
    },
    {
      "epoch": 0.1859007645889663,
      "grad_norm": 1.4758398532867432,
      "learning_rate": 0.00018752204161802706,
      "loss": 1.6252,
      "step": 5191
    },
    {
      "epoch": 0.18593657671853456,
      "grad_norm": 1.7125110626220703,
      "learning_rate": 0.00018751643030458256,
      "loss": 1.5789,
      "step": 5192
    },
    {
      "epoch": 0.18597238884810285,
      "grad_norm": 2.279865264892578,
      "learning_rate": 0.00018751081781371743,
      "loss": 1.5543,
      "step": 5193
    },
    {
      "epoch": 0.18600820097767115,
      "grad_norm": 1.559960126876831,
      "learning_rate": 0.00018750520414550711,
      "loss": 1.4844,
      "step": 5194
    },
    {
      "epoch": 0.18604401310723942,
      "grad_norm": 1.2859904766082764,
      "learning_rate": 0.00018749958930002717,
      "loss": 1.5597,
      "step": 5195
    },
    {
      "epoch": 0.18607982523680772,
      "grad_norm": 2.042282819747925,
      "learning_rate": 0.00018749397327735308,
      "loss": 1.2362,
      "step": 5196
    },
    {
      "epoch": 0.18611563736637599,
      "grad_norm": 2.044483184814453,
      "learning_rate": 0.00018748835607756045,
      "loss": 1.631,
      "step": 5197
    },
    {
      "epoch": 0.18615144949594428,
      "grad_norm": 1.2718511819839478,
      "learning_rate": 0.00018748273770072485,
      "loss": 1.5658,
      "step": 5198
    },
    {
      "epoch": 0.18618726162551255,
      "grad_norm": 1.6728475093841553,
      "learning_rate": 0.00018747711814692185,
      "loss": 1.897,
      "step": 5199
    },
    {
      "epoch": 0.18622307375508085,
      "grad_norm": 1.7691491842269897,
      "learning_rate": 0.00018747149741622706,
      "loss": 1.6203,
      "step": 5200
    },
    {
      "epoch": 0.18625888588464912,
      "grad_norm": 1.5065457820892334,
      "learning_rate": 0.0001874658755087161,
      "loss": 1.805,
      "step": 5201
    },
    {
      "epoch": 0.1862946980142174,
      "grad_norm": 1.8144469261169434,
      "learning_rate": 0.00018746025242446463,
      "loss": 1.5208,
      "step": 5202
    },
    {
      "epoch": 0.1863305101437857,
      "grad_norm": 1.7388842105865479,
      "learning_rate": 0.00018745462816354826,
      "loss": 1.4172,
      "step": 5203
    },
    {
      "epoch": 0.18636632227335398,
      "grad_norm": 1.3475241661071777,
      "learning_rate": 0.0001874490027260427,
      "loss": 1.6845,
      "step": 5204
    },
    {
      "epoch": 0.18640213440292228,
      "grad_norm": 2.016415596008301,
      "learning_rate": 0.0001874433761120236,
      "loss": 1.4557,
      "step": 5205
    },
    {
      "epoch": 0.18643794653249054,
      "grad_norm": 1.5078966617584229,
      "learning_rate": 0.00018743774832156667,
      "loss": 1.5403,
      "step": 5206
    },
    {
      "epoch": 0.18647375866205884,
      "grad_norm": 1.3218117952346802,
      "learning_rate": 0.0001874321193547476,
      "loss": 1.5534,
      "step": 5207
    },
    {
      "epoch": 0.1865095707916271,
      "grad_norm": 1.9156447649002075,
      "learning_rate": 0.00018742648921164215,
      "loss": 1.4724,
      "step": 5208
    },
    {
      "epoch": 0.1865453829211954,
      "grad_norm": 2.4503142833709717,
      "learning_rate": 0.00018742085789232607,
      "loss": 1.4975,
      "step": 5209
    },
    {
      "epoch": 0.1865811950507637,
      "grad_norm": 1.653955101966858,
      "learning_rate": 0.0001874152253968751,
      "loss": 1.5087,
      "step": 5210
    },
    {
      "epoch": 0.18661700718033197,
      "grad_norm": 1.6351701021194458,
      "learning_rate": 0.00018740959172536506,
      "loss": 1.471,
      "step": 5211
    },
    {
      "epoch": 0.18665281930990027,
      "grad_norm": 1.942446231842041,
      "learning_rate": 0.0001874039568778717,
      "loss": 1.5784,
      "step": 5212
    },
    {
      "epoch": 0.18668863143946854,
      "grad_norm": 1.5409560203552246,
      "learning_rate": 0.0001873983208544708,
      "loss": 1.7789,
      "step": 5213
    },
    {
      "epoch": 0.18672444356903684,
      "grad_norm": 1.5599496364593506,
      "learning_rate": 0.00018739268365523828,
      "loss": 1.3787,
      "step": 5214
    },
    {
      "epoch": 0.1867602556986051,
      "grad_norm": 2.1475934982299805,
      "learning_rate": 0.00018738704528024994,
      "loss": 1.8485,
      "step": 5215
    },
    {
      "epoch": 0.1867960678281734,
      "grad_norm": 1.8166731595993042,
      "learning_rate": 0.00018738140572958155,
      "loss": 1.5655,
      "step": 5216
    },
    {
      "epoch": 0.1868318799577417,
      "grad_norm": 2.244706630706787,
      "learning_rate": 0.00018737576500330914,
      "loss": 1.4576,
      "step": 5217
    },
    {
      "epoch": 0.18686769208730997,
      "grad_norm": 1.2443066835403442,
      "learning_rate": 0.00018737012310150847,
      "loss": 1.3797,
      "step": 5218
    },
    {
      "epoch": 0.18690350421687826,
      "grad_norm": 1.4576160907745361,
      "learning_rate": 0.00018736448002425554,
      "loss": 1.4719,
      "step": 5219
    },
    {
      "epoch": 0.18693931634644653,
      "grad_norm": 1.2665461301803589,
      "learning_rate": 0.00018735883577162619,
      "loss": 1.5914,
      "step": 5220
    },
    {
      "epoch": 0.18697512847601483,
      "grad_norm": 1.9680484533309937,
      "learning_rate": 0.0001873531903436964,
      "loss": 1.7045,
      "step": 5221
    },
    {
      "epoch": 0.1870109406055831,
      "grad_norm": 1.387949824333191,
      "learning_rate": 0.00018734754374054207,
      "loss": 1.4011,
      "step": 5222
    },
    {
      "epoch": 0.1870467527351514,
      "grad_norm": 1.9003175497055054,
      "learning_rate": 0.0001873418959622393,
      "loss": 1.443,
      "step": 5223
    },
    {
      "epoch": 0.1870825648647197,
      "grad_norm": 1.8332993984222412,
      "learning_rate": 0.0001873362470088639,
      "loss": 1.7366,
      "step": 5224
    },
    {
      "epoch": 0.18711837699428796,
      "grad_norm": 1.9775099754333496,
      "learning_rate": 0.00018733059688049198,
      "loss": 1.574,
      "step": 5225
    },
    {
      "epoch": 0.18715418912385626,
      "grad_norm": 2.4329521656036377,
      "learning_rate": 0.00018732494557719952,
      "loss": 1.8489,
      "step": 5226
    },
    {
      "epoch": 0.18719000125342453,
      "grad_norm": 1.4437566995620728,
      "learning_rate": 0.00018731929309906254,
      "loss": 1.4442,
      "step": 5227
    },
    {
      "epoch": 0.18722581338299282,
      "grad_norm": 1.4024875164031982,
      "learning_rate": 0.00018731363944615717,
      "loss": 1.2043,
      "step": 5228
    },
    {
      "epoch": 0.1872616255125611,
      "grad_norm": 1.692596197128296,
      "learning_rate": 0.00018730798461855938,
      "loss": 1.4935,
      "step": 5229
    },
    {
      "epoch": 0.1872974376421294,
      "grad_norm": 1.579940676689148,
      "learning_rate": 0.00018730232861634524,
      "loss": 1.5019,
      "step": 5230
    },
    {
      "epoch": 0.18733324977169769,
      "grad_norm": 1.8518500328063965,
      "learning_rate": 0.0001872966714395909,
      "loss": 1.5695,
      "step": 5231
    },
    {
      "epoch": 0.18736906190126595,
      "grad_norm": 1.7087852954864502,
      "learning_rate": 0.00018729101308837245,
      "loss": 1.844,
      "step": 5232
    },
    {
      "epoch": 0.18740487403083425,
      "grad_norm": 1.8843986988067627,
      "learning_rate": 0.000187285353562766,
      "loss": 1.4705,
      "step": 5233
    },
    {
      "epoch": 0.18744068616040252,
      "grad_norm": 1.5346555709838867,
      "learning_rate": 0.00018727969286284776,
      "loss": 1.4533,
      "step": 5234
    },
    {
      "epoch": 0.18747649828997082,
      "grad_norm": 1.526907205581665,
      "learning_rate": 0.0001872740309886938,
      "loss": 1.6521,
      "step": 5235
    },
    {
      "epoch": 0.18751231041953909,
      "grad_norm": 2.257167100906372,
      "learning_rate": 0.00018726836794038035,
      "loss": 1.4737,
      "step": 5236
    },
    {
      "epoch": 0.18754812254910738,
      "grad_norm": 2.471599817276001,
      "learning_rate": 0.00018726270371798357,
      "loss": 1.5256,
      "step": 5237
    },
    {
      "epoch": 0.18758393467867568,
      "grad_norm": 1.6534587144851685,
      "learning_rate": 0.00018725703832157966,
      "loss": 1.7539,
      "step": 5238
    },
    {
      "epoch": 0.18761974680824395,
      "grad_norm": 1.419296383857727,
      "learning_rate": 0.00018725137175124482,
      "loss": 1.5579,
      "step": 5239
    },
    {
      "epoch": 0.18765555893781224,
      "grad_norm": 1.8545119762420654,
      "learning_rate": 0.0001872457040070554,
      "loss": 1.6909,
      "step": 5240
    },
    {
      "epoch": 0.1876913710673805,
      "grad_norm": 1.6722489595413208,
      "learning_rate": 0.0001872400350890875,
      "loss": 1.4801,
      "step": 5241
    },
    {
      "epoch": 0.1877271831969488,
      "grad_norm": 1.4976879358291626,
      "learning_rate": 0.00018723436499741748,
      "loss": 1.3575,
      "step": 5242
    },
    {
      "epoch": 0.18776299532651708,
      "grad_norm": 1.9052339792251587,
      "learning_rate": 0.0001872286937321216,
      "loss": 1.2981,
      "step": 5243
    },
    {
      "epoch": 0.18779880745608538,
      "grad_norm": 1.9531632661819458,
      "learning_rate": 0.00018722302129327618,
      "loss": 1.4758,
      "step": 5244
    },
    {
      "epoch": 0.18783461958565367,
      "grad_norm": 1.5285097360610962,
      "learning_rate": 0.0001872173476809575,
      "loss": 1.39,
      "step": 5245
    },
    {
      "epoch": 0.18787043171522194,
      "grad_norm": 2.232374906539917,
      "learning_rate": 0.00018721167289524195,
      "loss": 1.7108,
      "step": 5246
    },
    {
      "epoch": 0.18790624384479024,
      "grad_norm": 1.175122618675232,
      "learning_rate": 0.0001872059969362058,
      "loss": 1.578,
      "step": 5247
    },
    {
      "epoch": 0.1879420559743585,
      "grad_norm": 1.2910648584365845,
      "learning_rate": 0.00018720031980392544,
      "loss": 1.4514,
      "step": 5248
    },
    {
      "epoch": 0.1879778681039268,
      "grad_norm": 2.2122230529785156,
      "learning_rate": 0.0001871946414984773,
      "loss": 1.5838,
      "step": 5249
    },
    {
      "epoch": 0.18801368023349507,
      "grad_norm": 1.6212471723556519,
      "learning_rate": 0.00018718896201993767,
      "loss": 1.103,
      "step": 5250
    },
    {
      "epoch": 0.18804949236306337,
      "grad_norm": 1.6619919538497925,
      "learning_rate": 0.00018718328136838305,
      "loss": 1.4881,
      "step": 5251
    },
    {
      "epoch": 0.18808530449263167,
      "grad_norm": 1.6589463949203491,
      "learning_rate": 0.00018717759954388986,
      "loss": 1.4259,
      "step": 5252
    },
    {
      "epoch": 0.18812111662219994,
      "grad_norm": 1.471097707748413,
      "learning_rate": 0.00018717191654653452,
      "loss": 1.6129,
      "step": 5253
    },
    {
      "epoch": 0.18815692875176823,
      "grad_norm": 1.1274768114089966,
      "learning_rate": 0.00018716623237639347,
      "loss": 1.4241,
      "step": 5254
    },
    {
      "epoch": 0.1881927408813365,
      "grad_norm": 1.8204070329666138,
      "learning_rate": 0.00018716054703354318,
      "loss": 1.6479,
      "step": 5255
    },
    {
      "epoch": 0.1882285530109048,
      "grad_norm": 2.424353837966919,
      "learning_rate": 0.0001871548605180602,
      "loss": 1.5044,
      "step": 5256
    },
    {
      "epoch": 0.18826436514047307,
      "grad_norm": 1.7272015810012817,
      "learning_rate": 0.00018714917283002094,
      "loss": 1.9286,
      "step": 5257
    },
    {
      "epoch": 0.18830017727004136,
      "grad_norm": 1.4467058181762695,
      "learning_rate": 0.000187143483969502,
      "loss": 1.7475,
      "step": 5258
    },
    {
      "epoch": 0.18833598939960966,
      "grad_norm": 1.6046894788742065,
      "learning_rate": 0.00018713779393657993,
      "loss": 1.5029,
      "step": 5259
    },
    {
      "epoch": 0.18837180152917793,
      "grad_norm": 1.3539988994598389,
      "learning_rate": 0.00018713210273133118,
      "loss": 1.5756,
      "step": 5260
    },
    {
      "epoch": 0.18840761365874623,
      "grad_norm": 1.6449440717697144,
      "learning_rate": 0.00018712641035383243,
      "loss": 1.4489,
      "step": 5261
    },
    {
      "epoch": 0.1884434257883145,
      "grad_norm": 1.5281622409820557,
      "learning_rate": 0.00018712071680416017,
      "loss": 1.6258,
      "step": 5262
    },
    {
      "epoch": 0.1884792379178828,
      "grad_norm": 1.7914692163467407,
      "learning_rate": 0.00018711502208239108,
      "loss": 1.2722,
      "step": 5263
    },
    {
      "epoch": 0.18851505004745106,
      "grad_norm": 1.9788846969604492,
      "learning_rate": 0.0001871093261886017,
      "loss": 1.3999,
      "step": 5264
    },
    {
      "epoch": 0.18855086217701936,
      "grad_norm": 1.66983962059021,
      "learning_rate": 0.00018710362912286872,
      "loss": 1.7335,
      "step": 5265
    },
    {
      "epoch": 0.18858667430658765,
      "grad_norm": 2.1798553466796875,
      "learning_rate": 0.00018709793088526877,
      "loss": 1.5971,
      "step": 5266
    },
    {
      "epoch": 0.18862248643615592,
      "grad_norm": 2.0857560634613037,
      "learning_rate": 0.0001870922314758785,
      "loss": 1.4487,
      "step": 5267
    },
    {
      "epoch": 0.18865829856572422,
      "grad_norm": 1.8404710292816162,
      "learning_rate": 0.0001870865308947746,
      "loss": 1.5343,
      "step": 5268
    },
    {
      "epoch": 0.1886941106952925,
      "grad_norm": 1.9581575393676758,
      "learning_rate": 0.00018708082914203376,
      "loss": 1.4325,
      "step": 5269
    },
    {
      "epoch": 0.18872992282486079,
      "grad_norm": 1.5912389755249023,
      "learning_rate": 0.0001870751262177327,
      "loss": 1.503,
      "step": 5270
    },
    {
      "epoch": 0.18876573495442905,
      "grad_norm": 2.4967665672302246,
      "learning_rate": 0.00018706942212194812,
      "loss": 1.4733,
      "step": 5271
    },
    {
      "epoch": 0.18880154708399735,
      "grad_norm": 1.365278959274292,
      "learning_rate": 0.0001870637168547568,
      "loss": 1.6569,
      "step": 5272
    },
    {
      "epoch": 0.18883735921356565,
      "grad_norm": 1.912571668624878,
      "learning_rate": 0.00018705801041623546,
      "loss": 1.7692,
      "step": 5273
    },
    {
      "epoch": 0.18887317134313392,
      "grad_norm": 2.0421035289764404,
      "learning_rate": 0.0001870523028064609,
      "loss": 1.4651,
      "step": 5274
    },
    {
      "epoch": 0.1889089834727022,
      "grad_norm": 1.6724461317062378,
      "learning_rate": 0.00018704659402550986,
      "loss": 1.6736,
      "step": 5275
    },
    {
      "epoch": 0.18894479560227048,
      "grad_norm": 2.037821054458618,
      "learning_rate": 0.0001870408840734592,
      "loss": 1.492,
      "step": 5276
    },
    {
      "epoch": 0.18898060773183878,
      "grad_norm": 1.94366455078125,
      "learning_rate": 0.00018703517295038573,
      "loss": 1.3209,
      "step": 5277
    },
    {
      "epoch": 0.18901641986140705,
      "grad_norm": 1.6621423959732056,
      "learning_rate": 0.00018702946065636623,
      "loss": 1.4488,
      "step": 5278
    },
    {
      "epoch": 0.18905223199097534,
      "grad_norm": 2.5038325786590576,
      "learning_rate": 0.00018702374719147766,
      "loss": 1.2706,
      "step": 5279
    },
    {
      "epoch": 0.18908804412054364,
      "grad_norm": 1.4643855094909668,
      "learning_rate": 0.00018701803255579677,
      "loss": 1.7987,
      "step": 5280
    },
    {
      "epoch": 0.1891238562501119,
      "grad_norm": 1.5782333612442017,
      "learning_rate": 0.00018701231674940054,
      "loss": 1.7855,
      "step": 5281
    },
    {
      "epoch": 0.1891596683796802,
      "grad_norm": 1.8929588794708252,
      "learning_rate": 0.0001870065997723658,
      "loss": 1.6753,
      "step": 5282
    },
    {
      "epoch": 0.18919548050924848,
      "grad_norm": 1.534611463546753,
      "learning_rate": 0.00018700088162476952,
      "loss": 1.439,
      "step": 5283
    },
    {
      "epoch": 0.18923129263881677,
      "grad_norm": 1.16856849193573,
      "learning_rate": 0.00018699516230668856,
      "loss": 1.6246,
      "step": 5284
    },
    {
      "epoch": 0.18926710476838504,
      "grad_norm": 1.5402215719223022,
      "learning_rate": 0.00018698944181819993,
      "loss": 1.3181,
      "step": 5285
    },
    {
      "epoch": 0.18930291689795334,
      "grad_norm": 1.398967981338501,
      "learning_rate": 0.00018698372015938058,
      "loss": 1.9008,
      "step": 5286
    },
    {
      "epoch": 0.18933872902752164,
      "grad_norm": 1.6031793355941772,
      "learning_rate": 0.00018697799733030746,
      "loss": 1.5932,
      "step": 5287
    },
    {
      "epoch": 0.1893745411570899,
      "grad_norm": 1.5183573961257935,
      "learning_rate": 0.00018697227333105756,
      "loss": 1.6072,
      "step": 5288
    },
    {
      "epoch": 0.1894103532866582,
      "grad_norm": 2.281196117401123,
      "learning_rate": 0.00018696654816170795,
      "loss": 1.6894,
      "step": 5289
    },
    {
      "epoch": 0.18944616541622647,
      "grad_norm": 1.6974093914031982,
      "learning_rate": 0.0001869608218223356,
      "loss": 1.2243,
      "step": 5290
    },
    {
      "epoch": 0.18948197754579477,
      "grad_norm": 1.6468058824539185,
      "learning_rate": 0.0001869550943130175,
      "loss": 1.2508,
      "step": 5291
    },
    {
      "epoch": 0.18951778967536304,
      "grad_norm": 1.3340821266174316,
      "learning_rate": 0.00018694936563383086,
      "loss": 1.3316,
      "step": 5292
    },
    {
      "epoch": 0.18955360180493133,
      "grad_norm": 1.515095591545105,
      "learning_rate": 0.00018694363578485262,
      "loss": 1.6026,
      "step": 5293
    },
    {
      "epoch": 0.18958941393449963,
      "grad_norm": 1.9505254030227661,
      "learning_rate": 0.00018693790476615992,
      "loss": 1.5093,
      "step": 5294
    },
    {
      "epoch": 0.1896252260640679,
      "grad_norm": 1.6551986932754517,
      "learning_rate": 0.00018693217257782985,
      "loss": 1.2679,
      "step": 5295
    },
    {
      "epoch": 0.1896610381936362,
      "grad_norm": 1.6577612161636353,
      "learning_rate": 0.00018692643921993952,
      "loss": 1.583,
      "step": 5296
    },
    {
      "epoch": 0.18969685032320446,
      "grad_norm": 1.423334002494812,
      "learning_rate": 0.0001869207046925661,
      "loss": 1.508,
      "step": 5297
    },
    {
      "epoch": 0.18973266245277276,
      "grad_norm": 1.5805872678756714,
      "learning_rate": 0.0001869149689957867,
      "loss": 1.7373,
      "step": 5298
    },
    {
      "epoch": 0.18976847458234103,
      "grad_norm": 1.982264518737793,
      "learning_rate": 0.0001869092321296785,
      "loss": 1.4873,
      "step": 5299
    },
    {
      "epoch": 0.18980428671190933,
      "grad_norm": 1.3803657293319702,
      "learning_rate": 0.00018690349409431872,
      "loss": 1.347,
      "step": 5300
    },
    {
      "epoch": 0.1898400988414776,
      "grad_norm": 1.725968360900879,
      "learning_rate": 0.00018689775488978452,
      "loss": 1.6085,
      "step": 5301
    },
    {
      "epoch": 0.1898759109710459,
      "grad_norm": 2.5847837924957275,
      "learning_rate": 0.0001868920145161531,
      "loss": 1.5907,
      "step": 5302
    },
    {
      "epoch": 0.1899117231006142,
      "grad_norm": 1.841369390487671,
      "learning_rate": 0.0001868862729735017,
      "loss": 1.8148,
      "step": 5303
    },
    {
      "epoch": 0.18994753523018246,
      "grad_norm": 2.197054147720337,
      "learning_rate": 0.00018688053026190757,
      "loss": 1.2587,
      "step": 5304
    },
    {
      "epoch": 0.18998334735975075,
      "grad_norm": 1.7132951021194458,
      "learning_rate": 0.000186874786381448,
      "loss": 1.5339,
      "step": 5305
    },
    {
      "epoch": 0.19001915948931902,
      "grad_norm": 1.6471575498580933,
      "learning_rate": 0.0001868690413322002,
      "loss": 1.5682,
      "step": 5306
    },
    {
      "epoch": 0.19005497161888732,
      "grad_norm": 1.9123592376708984,
      "learning_rate": 0.00018686329511424153,
      "loss": 1.7367,
      "step": 5307
    },
    {
      "epoch": 0.1900907837484556,
      "grad_norm": 2.1285805702209473,
      "learning_rate": 0.00018685754772764928,
      "loss": 1.7673,
      "step": 5308
    },
    {
      "epoch": 0.19012659587802389,
      "grad_norm": 1.352774739265442,
      "learning_rate": 0.00018685179917250072,
      "loss": 1.6098,
      "step": 5309
    },
    {
      "epoch": 0.19016240800759218,
      "grad_norm": 1.3797439336776733,
      "learning_rate": 0.0001868460494488733,
      "loss": 1.5094,
      "step": 5310
    },
    {
      "epoch": 0.19019822013716045,
      "grad_norm": 2.124342918395996,
      "learning_rate": 0.00018684029855684425,
      "loss": 1.6705,
      "step": 5311
    },
    {
      "epoch": 0.19023403226672875,
      "grad_norm": 1.2310181856155396,
      "learning_rate": 0.00018683454649649103,
      "loss": 1.4928,
      "step": 5312
    },
    {
      "epoch": 0.19026984439629702,
      "grad_norm": 1.4215664863586426,
      "learning_rate": 0.00018682879326789098,
      "loss": 1.6286,
      "step": 5313
    },
    {
      "epoch": 0.1903056565258653,
      "grad_norm": 1.8718503713607788,
      "learning_rate": 0.00018682303887112154,
      "loss": 1.4652,
      "step": 5314
    },
    {
      "epoch": 0.19034146865543358,
      "grad_norm": 2.483689308166504,
      "learning_rate": 0.00018681728330626008,
      "loss": 1.8905,
      "step": 5315
    },
    {
      "epoch": 0.19037728078500188,
      "grad_norm": 1.4977774620056152,
      "learning_rate": 0.00018681152657338404,
      "loss": 1.8194,
      "step": 5316
    },
    {
      "epoch": 0.19041309291457018,
      "grad_norm": 1.7027922868728638,
      "learning_rate": 0.00018680576867257095,
      "loss": 1.6571,
      "step": 5317
    },
    {
      "epoch": 0.19044890504413844,
      "grad_norm": 1.6885054111480713,
      "learning_rate": 0.00018680000960389818,
      "loss": 1.4111,
      "step": 5318
    },
    {
      "epoch": 0.19048471717370674,
      "grad_norm": 1.3242998123168945,
      "learning_rate": 0.00018679424936744323,
      "loss": 1.3061,
      "step": 5319
    },
    {
      "epoch": 0.190520529303275,
      "grad_norm": 2.3705389499664307,
      "learning_rate": 0.00018678848796328362,
      "loss": 1.2376,
      "step": 5320
    },
    {
      "epoch": 0.1905563414328433,
      "grad_norm": 1.4017654657363892,
      "learning_rate": 0.00018678272539149687,
      "loss": 1.2629,
      "step": 5321
    },
    {
      "epoch": 0.19059215356241158,
      "grad_norm": 1.6854712963104248,
      "learning_rate": 0.00018677696165216048,
      "loss": 1.6078,
      "step": 5322
    },
    {
      "epoch": 0.19062796569197987,
      "grad_norm": 1.8608250617980957,
      "learning_rate": 0.000186771196745352,
      "loss": 1.7155,
      "step": 5323
    },
    {
      "epoch": 0.19066377782154817,
      "grad_norm": 1.982079029083252,
      "learning_rate": 0.000186765430671149,
      "loss": 1.529,
      "step": 5324
    },
    {
      "epoch": 0.19069958995111644,
      "grad_norm": 1.904109239578247,
      "learning_rate": 0.00018675966342962904,
      "loss": 1.5579,
      "step": 5325
    },
    {
      "epoch": 0.19073540208068474,
      "grad_norm": 1.7409393787384033,
      "learning_rate": 0.00018675389502086976,
      "loss": 1.6718,
      "step": 5326
    },
    {
      "epoch": 0.190771214210253,
      "grad_norm": 1.7952065467834473,
      "learning_rate": 0.00018674812544494865,
      "loss": 1.6698,
      "step": 5327
    },
    {
      "epoch": 0.1908070263398213,
      "grad_norm": 2.428290367126465,
      "learning_rate": 0.00018674235470194348,
      "loss": 1.7883,
      "step": 5328
    },
    {
      "epoch": 0.19084283846938957,
      "grad_norm": 1.6777397394180298,
      "learning_rate": 0.0001867365827919318,
      "loss": 1.6194,
      "step": 5329
    },
    {
      "epoch": 0.19087865059895787,
      "grad_norm": 2.2093400955200195,
      "learning_rate": 0.00018673080971499126,
      "loss": 1.4838,
      "step": 5330
    },
    {
      "epoch": 0.19091446272852616,
      "grad_norm": 1.6835341453552246,
      "learning_rate": 0.00018672503547119957,
      "loss": 1.4022,
      "step": 5331
    },
    {
      "epoch": 0.19095027485809443,
      "grad_norm": 1.4746954441070557,
      "learning_rate": 0.00018671926006063442,
      "loss": 1.2283,
      "step": 5332
    },
    {
      "epoch": 0.19098608698766273,
      "grad_norm": 2.170846939086914,
      "learning_rate": 0.00018671348348337343,
      "loss": 1.8196,
      "step": 5333
    },
    {
      "epoch": 0.191021899117231,
      "grad_norm": 1.571633219718933,
      "learning_rate": 0.00018670770573949442,
      "loss": 1.5886,
      "step": 5334
    },
    {
      "epoch": 0.1910577112467993,
      "grad_norm": 1.9823776483535767,
      "learning_rate": 0.00018670192682907505,
      "loss": 1.6207,
      "step": 5335
    },
    {
      "epoch": 0.19109352337636756,
      "grad_norm": 1.2739793062210083,
      "learning_rate": 0.00018669614675219308,
      "loss": 1.5255,
      "step": 5336
    },
    {
      "epoch": 0.19112933550593586,
      "grad_norm": 1.8321658372879028,
      "learning_rate": 0.0001866903655089263,
      "loss": 1.7024,
      "step": 5337
    },
    {
      "epoch": 0.19116514763550416,
      "grad_norm": 1.5952870845794678,
      "learning_rate": 0.00018668458309935247,
      "loss": 1.5503,
      "step": 5338
    },
    {
      "epoch": 0.19120095976507243,
      "grad_norm": 1.4116911888122559,
      "learning_rate": 0.0001866787995235494,
      "loss": 1.8354,
      "step": 5339
    },
    {
      "epoch": 0.19123677189464072,
      "grad_norm": 1.6893408298492432,
      "learning_rate": 0.00018667301478159489,
      "loss": 1.7062,
      "step": 5340
    },
    {
      "epoch": 0.191272584024209,
      "grad_norm": 1.6460639238357544,
      "learning_rate": 0.00018666722887356673,
      "loss": 1.7274,
      "step": 5341
    },
    {
      "epoch": 0.1913083961537773,
      "grad_norm": 1.753871202468872,
      "learning_rate": 0.00018666144179954283,
      "loss": 2.059,
      "step": 5342
    },
    {
      "epoch": 0.19134420828334556,
      "grad_norm": 1.788332223892212,
      "learning_rate": 0.00018665565355960103,
      "loss": 1.2346,
      "step": 5343
    },
    {
      "epoch": 0.19138002041291385,
      "grad_norm": 1.3304243087768555,
      "learning_rate": 0.0001866498641538192,
      "loss": 1.7645,
      "step": 5344
    },
    {
      "epoch": 0.19141583254248215,
      "grad_norm": 1.1889979839324951,
      "learning_rate": 0.00018664407358227517,
      "loss": 1.5125,
      "step": 5345
    },
    {
      "epoch": 0.19145164467205042,
      "grad_norm": 2.2318155765533447,
      "learning_rate": 0.0001866382818450469,
      "loss": 1.4744,
      "step": 5346
    },
    {
      "epoch": 0.19148745680161872,
      "grad_norm": 1.7189171314239502,
      "learning_rate": 0.00018663248894221232,
      "loss": 1.466,
      "step": 5347
    },
    {
      "epoch": 0.19152326893118699,
      "grad_norm": 1.3835846185684204,
      "learning_rate": 0.00018662669487384936,
      "loss": 1.6789,
      "step": 5348
    },
    {
      "epoch": 0.19155908106075528,
      "grad_norm": 2.947704315185547,
      "learning_rate": 0.00018662089964003594,
      "loss": 1.5365,
      "step": 5349
    },
    {
      "epoch": 0.19159489319032355,
      "grad_norm": 1.4203107357025146,
      "learning_rate": 0.00018661510324085003,
      "loss": 1.6378,
      "step": 5350
    },
    {
      "epoch": 0.19163070531989185,
      "grad_norm": 1.2650010585784912,
      "learning_rate": 0.00018660930567636968,
      "loss": 1.4558,
      "step": 5351
    },
    {
      "epoch": 0.19166651744946014,
      "grad_norm": 1.485657811164856,
      "learning_rate": 0.00018660350694667282,
      "loss": 1.5388,
      "step": 5352
    },
    {
      "epoch": 0.1917023295790284,
      "grad_norm": 1.648183822631836,
      "learning_rate": 0.00018659770705183748,
      "loss": 1.3599,
      "step": 5353
    },
    {
      "epoch": 0.1917381417085967,
      "grad_norm": 1.3641210794448853,
      "learning_rate": 0.00018659190599194168,
      "loss": 1.336,
      "step": 5354
    },
    {
      "epoch": 0.19177395383816498,
      "grad_norm": 1.6528867483139038,
      "learning_rate": 0.0001865861037670635,
      "loss": 1.5748,
      "step": 5355
    },
    {
      "epoch": 0.19180976596773328,
      "grad_norm": 1.54402756690979,
      "learning_rate": 0.00018658030037728098,
      "loss": 1.495,
      "step": 5356
    },
    {
      "epoch": 0.19184557809730154,
      "grad_norm": 1.9251580238342285,
      "learning_rate": 0.00018657449582267218,
      "loss": 1.5426,
      "step": 5357
    },
    {
      "epoch": 0.19188139022686984,
      "grad_norm": 2.060964345932007,
      "learning_rate": 0.00018656869010331523,
      "loss": 1.7168,
      "step": 5358
    },
    {
      "epoch": 0.19191720235643814,
      "grad_norm": 1.657238483428955,
      "learning_rate": 0.00018656288321928824,
      "loss": 1.2452,
      "step": 5359
    },
    {
      "epoch": 0.1919530144860064,
      "grad_norm": 1.5768673419952393,
      "learning_rate": 0.0001865570751706693,
      "loss": 1.5858,
      "step": 5360
    },
    {
      "epoch": 0.1919888266155747,
      "grad_norm": 1.583263635635376,
      "learning_rate": 0.00018655126595753654,
      "loss": 1.2867,
      "step": 5361
    },
    {
      "epoch": 0.19202463874514297,
      "grad_norm": 1.5505911111831665,
      "learning_rate": 0.00018654545557996816,
      "loss": 1.5733,
      "step": 5362
    },
    {
      "epoch": 0.19206045087471127,
      "grad_norm": 1.305277705192566,
      "learning_rate": 0.0001865396440380423,
      "loss": 1.6863,
      "step": 5363
    },
    {
      "epoch": 0.19209626300427954,
      "grad_norm": 2.091987133026123,
      "learning_rate": 0.00018653383133183718,
      "loss": 1.9752,
      "step": 5364
    },
    {
      "epoch": 0.19213207513384783,
      "grad_norm": 1.7626006603240967,
      "learning_rate": 0.00018652801746143097,
      "loss": 1.6712,
      "step": 5365
    },
    {
      "epoch": 0.19216788726341613,
      "grad_norm": 1.4710909128189087,
      "learning_rate": 0.00018652220242690187,
      "loss": 1.7792,
      "step": 5366
    },
    {
      "epoch": 0.1922036993929844,
      "grad_norm": 1.8461323976516724,
      "learning_rate": 0.00018651638622832817,
      "loss": 1.3979,
      "step": 5367
    },
    {
      "epoch": 0.1922395115225527,
      "grad_norm": 1.7423018217086792,
      "learning_rate": 0.00018651056886578808,
      "loss": 1.5811,
      "step": 5368
    },
    {
      "epoch": 0.19227532365212097,
      "grad_norm": 2.2182414531707764,
      "learning_rate": 0.00018650475033935992,
      "loss": 1.6277,
      "step": 5369
    },
    {
      "epoch": 0.19231113578168926,
      "grad_norm": 1.7264537811279297,
      "learning_rate": 0.00018649893064912187,
      "loss": 1.6134,
      "step": 5370
    },
    {
      "epoch": 0.19234694791125753,
      "grad_norm": 2.0035877227783203,
      "learning_rate": 0.00018649310979515228,
      "loss": 1.6803,
      "step": 5371
    },
    {
      "epoch": 0.19238276004082583,
      "grad_norm": 1.4782098531723022,
      "learning_rate": 0.0001864872877775295,
      "loss": 1.7665,
      "step": 5372
    },
    {
      "epoch": 0.19241857217039413,
      "grad_norm": 1.79098641872406,
      "learning_rate": 0.00018648146459633182,
      "loss": 1.3728,
      "step": 5373
    },
    {
      "epoch": 0.1924543842999624,
      "grad_norm": 1.7012203931808472,
      "learning_rate": 0.00018647564025163756,
      "loss": 1.6528,
      "step": 5374
    },
    {
      "epoch": 0.1924901964295307,
      "grad_norm": 1.1748199462890625,
      "learning_rate": 0.00018646981474352515,
      "loss": 1.5076,
      "step": 5375
    },
    {
      "epoch": 0.19252600855909896,
      "grad_norm": 1.2763925790786743,
      "learning_rate": 0.0001864639880720729,
      "loss": 1.4572,
      "step": 5376
    },
    {
      "epoch": 0.19256182068866726,
      "grad_norm": 1.897787094116211,
      "learning_rate": 0.0001864581602373592,
      "loss": 1.6753,
      "step": 5377
    },
    {
      "epoch": 0.19259763281823553,
      "grad_norm": 1.900539755821228,
      "learning_rate": 0.00018645233123946252,
      "loss": 1.3106,
      "step": 5378
    },
    {
      "epoch": 0.19263344494780382,
      "grad_norm": 1.413638949394226,
      "learning_rate": 0.0001864465010784612,
      "loss": 1.4018,
      "step": 5379
    },
    {
      "epoch": 0.19266925707737212,
      "grad_norm": 1.6315624713897705,
      "learning_rate": 0.00018644066975443373,
      "loss": 1.6498,
      "step": 5380
    },
    {
      "epoch": 0.1927050692069404,
      "grad_norm": 1.668734073638916,
      "learning_rate": 0.00018643483726745857,
      "loss": 1.2368,
      "step": 5381
    },
    {
      "epoch": 0.19274088133650868,
      "grad_norm": 2.507664203643799,
      "learning_rate": 0.00018642900361761413,
      "loss": 1.5805,
      "step": 5382
    },
    {
      "epoch": 0.19277669346607695,
      "grad_norm": 2.0644752979278564,
      "learning_rate": 0.00018642316880497893,
      "loss": 1.6608,
      "step": 5383
    },
    {
      "epoch": 0.19281250559564525,
      "grad_norm": 2.1488327980041504,
      "learning_rate": 0.00018641733282963153,
      "loss": 1.9823,
      "step": 5384
    },
    {
      "epoch": 0.19284831772521352,
      "grad_norm": 1.6034685373306274,
      "learning_rate": 0.00018641149569165034,
      "loss": 1.5834,
      "step": 5385
    },
    {
      "epoch": 0.19288412985478182,
      "grad_norm": 1.9179775714874268,
      "learning_rate": 0.00018640565739111393,
      "loss": 1.534,
      "step": 5386
    },
    {
      "epoch": 0.1929199419843501,
      "grad_norm": 1.6921823024749756,
      "learning_rate": 0.0001863998179281009,
      "loss": 1.5869,
      "step": 5387
    },
    {
      "epoch": 0.19295575411391838,
      "grad_norm": 1.5110647678375244,
      "learning_rate": 0.0001863939773026897,
      "loss": 1.4647,
      "step": 5388
    },
    {
      "epoch": 0.19299156624348668,
      "grad_norm": 1.7196460962295532,
      "learning_rate": 0.00018638813551495901,
      "loss": 1.7801,
      "step": 5389
    },
    {
      "epoch": 0.19302737837305495,
      "grad_norm": 1.4753451347351074,
      "learning_rate": 0.0001863822925649874,
      "loss": 1.8287,
      "step": 5390
    },
    {
      "epoch": 0.19306319050262324,
      "grad_norm": 1.5066410303115845,
      "learning_rate": 0.00018637644845285344,
      "loss": 1.4221,
      "step": 5391
    },
    {
      "epoch": 0.1930990026321915,
      "grad_norm": 1.68746817111969,
      "learning_rate": 0.00018637060317863583,
      "loss": 1.9171,
      "step": 5392
    },
    {
      "epoch": 0.1931348147617598,
      "grad_norm": 1.6595805883407593,
      "learning_rate": 0.00018636475674241315,
      "loss": 1.6928,
      "step": 5393
    },
    {
      "epoch": 0.1931706268913281,
      "grad_norm": 1.706078290939331,
      "learning_rate": 0.00018635890914426404,
      "loss": 1.618,
      "step": 5394
    },
    {
      "epoch": 0.19320643902089638,
      "grad_norm": 1.5801364183425903,
      "learning_rate": 0.00018635306038426724,
      "loss": 1.5143,
      "step": 5395
    },
    {
      "epoch": 0.19324225115046467,
      "grad_norm": 2.3066155910491943,
      "learning_rate": 0.00018634721046250139,
      "loss": 1.4796,
      "step": 5396
    },
    {
      "epoch": 0.19327806328003294,
      "grad_norm": 1.6973354816436768,
      "learning_rate": 0.00018634135937904518,
      "loss": 1.7408,
      "step": 5397
    },
    {
      "epoch": 0.19331387540960124,
      "grad_norm": 1.4387764930725098,
      "learning_rate": 0.00018633550713397737,
      "loss": 1.4763,
      "step": 5398
    },
    {
      "epoch": 0.1933496875391695,
      "grad_norm": 1.51828134059906,
      "learning_rate": 0.00018632965372737665,
      "loss": 1.71,
      "step": 5399
    },
    {
      "epoch": 0.1933854996687378,
      "grad_norm": 2.4643683433532715,
      "learning_rate": 0.00018632379915932185,
      "loss": 1.6933,
      "step": 5400
    },
    {
      "epoch": 0.19342131179830607,
      "grad_norm": 1.6648520231246948,
      "learning_rate": 0.00018631794342989163,
      "loss": 1.2457,
      "step": 5401
    },
    {
      "epoch": 0.19345712392787437,
      "grad_norm": 1.4097614288330078,
      "learning_rate": 0.00018631208653916486,
      "loss": 1.503,
      "step": 5402
    },
    {
      "epoch": 0.19349293605744267,
      "grad_norm": 1.4005558490753174,
      "learning_rate": 0.0001863062284872203,
      "loss": 1.2093,
      "step": 5403
    },
    {
      "epoch": 0.19352874818701093,
      "grad_norm": 1.770641803741455,
      "learning_rate": 0.00018630036927413672,
      "loss": 1.409,
      "step": 5404
    },
    {
      "epoch": 0.19356456031657923,
      "grad_norm": 1.9547629356384277,
      "learning_rate": 0.00018629450889999302,
      "loss": 1.5299,
      "step": 5405
    },
    {
      "epoch": 0.1936003724461475,
      "grad_norm": 1.4061657190322876,
      "learning_rate": 0.000186288647364868,
      "loss": 1.4272,
      "step": 5406
    },
    {
      "epoch": 0.1936361845757158,
      "grad_norm": 2.2210681438446045,
      "learning_rate": 0.00018628278466884055,
      "loss": 1.4814,
      "step": 5407
    },
    {
      "epoch": 0.19367199670528407,
      "grad_norm": 1.4909852743148804,
      "learning_rate": 0.00018627692081198954,
      "loss": 1.4706,
      "step": 5408
    },
    {
      "epoch": 0.19370780883485236,
      "grad_norm": 1.9401264190673828,
      "learning_rate": 0.00018627105579439382,
      "loss": 1.2785,
      "step": 5409
    },
    {
      "epoch": 0.19374362096442066,
      "grad_norm": 1.8306653499603271,
      "learning_rate": 0.00018626518961613236,
      "loss": 1.3549,
      "step": 5410
    },
    {
      "epoch": 0.19377943309398893,
      "grad_norm": 1.7504512071609497,
      "learning_rate": 0.00018625932227728402,
      "loss": 1.6292,
      "step": 5411
    },
    {
      "epoch": 0.19381524522355723,
      "grad_norm": 1.2952179908752441,
      "learning_rate": 0.00018625345377792777,
      "loss": 1.5527,
      "step": 5412
    },
    {
      "epoch": 0.1938510573531255,
      "grad_norm": 1.9367178678512573,
      "learning_rate": 0.0001862475841181426,
      "loss": 1.611,
      "step": 5413
    },
    {
      "epoch": 0.1938868694826938,
      "grad_norm": 2.24019455909729,
      "learning_rate": 0.00018624171329800738,
      "loss": 1.7476,
      "step": 5414
    },
    {
      "epoch": 0.19392268161226206,
      "grad_norm": 1.6576284170150757,
      "learning_rate": 0.00018623584131760118,
      "loss": 1.7381,
      "step": 5415
    },
    {
      "epoch": 0.19395849374183036,
      "grad_norm": 1.9396398067474365,
      "learning_rate": 0.00018622996817700295,
      "loss": 1.5697,
      "step": 5416
    },
    {
      "epoch": 0.19399430587139865,
      "grad_norm": 1.9461170434951782,
      "learning_rate": 0.00018622409387629175,
      "loss": 1.4934,
      "step": 5417
    },
    {
      "epoch": 0.19403011800096692,
      "grad_norm": 1.8357774019241333,
      "learning_rate": 0.0001862182184155466,
      "loss": 1.7385,
      "step": 5418
    },
    {
      "epoch": 0.19406593013053522,
      "grad_norm": 1.4890779256820679,
      "learning_rate": 0.00018621234179484647,
      "loss": 1.5282,
      "step": 5419
    },
    {
      "epoch": 0.1941017422601035,
      "grad_norm": 2.412010431289673,
      "learning_rate": 0.00018620646401427054,
      "loss": 1.7779,
      "step": 5420
    },
    {
      "epoch": 0.19413755438967178,
      "grad_norm": 2.704232692718506,
      "learning_rate": 0.00018620058507389783,
      "loss": 1.5965,
      "step": 5421
    },
    {
      "epoch": 0.19417336651924005,
      "grad_norm": 1.7113075256347656,
      "learning_rate": 0.00018619470497380745,
      "loss": 1.7815,
      "step": 5422
    },
    {
      "epoch": 0.19420917864880835,
      "grad_norm": 1.8636949062347412,
      "learning_rate": 0.00018618882371407847,
      "loss": 1.6106,
      "step": 5423
    },
    {
      "epoch": 0.19424499077837665,
      "grad_norm": 1.3164920806884766,
      "learning_rate": 0.00018618294129479007,
      "loss": 1.3721,
      "step": 5424
    },
    {
      "epoch": 0.19428080290794492,
      "grad_norm": 1.5221353769302368,
      "learning_rate": 0.00018617705771602132,
      "loss": 1.4453,
      "step": 5425
    },
    {
      "epoch": 0.1943166150375132,
      "grad_norm": 2.4712367057800293,
      "learning_rate": 0.00018617117297785145,
      "loss": 1.9291,
      "step": 5426
    },
    {
      "epoch": 0.19435242716708148,
      "grad_norm": 1.8102853298187256,
      "learning_rate": 0.00018616528708035958,
      "loss": 1.6105,
      "step": 5427
    },
    {
      "epoch": 0.19438823929664978,
      "grad_norm": 1.8387972116470337,
      "learning_rate": 0.00018615940002362496,
      "loss": 1.4088,
      "step": 5428
    },
    {
      "epoch": 0.19442405142621805,
      "grad_norm": 1.8663088083267212,
      "learning_rate": 0.0001861535118077267,
      "loss": 1.5058,
      "step": 5429
    },
    {
      "epoch": 0.19445986355578634,
      "grad_norm": 2.4539215564727783,
      "learning_rate": 0.0001861476224327441,
      "loss": 1.8223,
      "step": 5430
    },
    {
      "epoch": 0.19449567568535464,
      "grad_norm": 1.7828904390335083,
      "learning_rate": 0.00018614173189875636,
      "loss": 1.4433,
      "step": 5431
    },
    {
      "epoch": 0.1945314878149229,
      "grad_norm": 2.0361180305480957,
      "learning_rate": 0.0001861358402058427,
      "loss": 1.5065,
      "step": 5432
    },
    {
      "epoch": 0.1945672999444912,
      "grad_norm": 1.5484020709991455,
      "learning_rate": 0.00018612994735408246,
      "loss": 1.7755,
      "step": 5433
    },
    {
      "epoch": 0.19460311207405948,
      "grad_norm": 1.351025938987732,
      "learning_rate": 0.00018612405334355488,
      "loss": 1.5675,
      "step": 5434
    },
    {
      "epoch": 0.19463892420362777,
      "grad_norm": 1.856544852256775,
      "learning_rate": 0.00018611815817433925,
      "loss": 1.8301,
      "step": 5435
    },
    {
      "epoch": 0.19467473633319604,
      "grad_norm": 1.3813837766647339,
      "learning_rate": 0.00018611226184651484,
      "loss": 1.7018,
      "step": 5436
    },
    {
      "epoch": 0.19471054846276434,
      "grad_norm": 1.4931066036224365,
      "learning_rate": 0.00018610636436016106,
      "loss": 1.5255,
      "step": 5437
    },
    {
      "epoch": 0.19474636059233263,
      "grad_norm": 1.5089569091796875,
      "learning_rate": 0.00018610046571535723,
      "loss": 1.4895,
      "step": 5438
    },
    {
      "epoch": 0.1947821727219009,
      "grad_norm": 1.7368457317352295,
      "learning_rate": 0.00018609456591218266,
      "loss": 1.1445,
      "step": 5439
    },
    {
      "epoch": 0.1948179848514692,
      "grad_norm": 1.6913045644760132,
      "learning_rate": 0.0001860886649507168,
      "loss": 1.5326,
      "step": 5440
    },
    {
      "epoch": 0.19485379698103747,
      "grad_norm": 1.6324273347854614,
      "learning_rate": 0.00018608276283103896,
      "loss": 1.7677,
      "step": 5441
    },
    {
      "epoch": 0.19488960911060577,
      "grad_norm": 1.6233586072921753,
      "learning_rate": 0.0001860768595532286,
      "loss": 1.1984,
      "step": 5442
    },
    {
      "epoch": 0.19492542124017403,
      "grad_norm": 1.4947038888931274,
      "learning_rate": 0.00018607095511736515,
      "loss": 1.7798,
      "step": 5443
    },
    {
      "epoch": 0.19496123336974233,
      "grad_norm": 1.6145468950271606,
      "learning_rate": 0.00018606504952352798,
      "loss": 1.4282,
      "step": 5444
    },
    {
      "epoch": 0.19499704549931063,
      "grad_norm": 2.4696900844573975,
      "learning_rate": 0.00018605914277179664,
      "loss": 1.8904,
      "step": 5445
    },
    {
      "epoch": 0.1950328576288789,
      "grad_norm": 1.766887903213501,
      "learning_rate": 0.00018605323486225049,
      "loss": 1.6336,
      "step": 5446
    },
    {
      "epoch": 0.1950686697584472,
      "grad_norm": 1.9337714910507202,
      "learning_rate": 0.00018604732579496908,
      "loss": 1.5009,
      "step": 5447
    },
    {
      "epoch": 0.19510448188801546,
      "grad_norm": 1.4915785789489746,
      "learning_rate": 0.0001860414155700319,
      "loss": 1.552,
      "step": 5448
    },
    {
      "epoch": 0.19514029401758376,
      "grad_norm": 1.450589656829834,
      "learning_rate": 0.00018603550418751845,
      "loss": 1.4491,
      "step": 5449
    },
    {
      "epoch": 0.19517610614715203,
      "grad_norm": 2.733760118484497,
      "learning_rate": 0.0001860295916475083,
      "loss": 1.539,
      "step": 5450
    },
    {
      "epoch": 0.19521191827672033,
      "grad_norm": 1.459010362625122,
      "learning_rate": 0.00018602367795008093,
      "loss": 1.7806,
      "step": 5451
    },
    {
      "epoch": 0.19524773040628862,
      "grad_norm": 1.4586896896362305,
      "learning_rate": 0.00018601776309531593,
      "loss": 1.4384,
      "step": 5452
    },
    {
      "epoch": 0.1952835425358569,
      "grad_norm": 1.6691211462020874,
      "learning_rate": 0.00018601184708329292,
      "loss": 1.5412,
      "step": 5453
    },
    {
      "epoch": 0.1953193546654252,
      "grad_norm": 2.4468443393707275,
      "learning_rate": 0.00018600592991409141,
      "loss": 1.5801,
      "step": 5454
    },
    {
      "epoch": 0.19535516679499346,
      "grad_norm": 2.1524693965911865,
      "learning_rate": 0.00018600001158779108,
      "loss": 1.6428,
      "step": 5455
    },
    {
      "epoch": 0.19539097892456175,
      "grad_norm": 1.6459256410598755,
      "learning_rate": 0.00018599409210447152,
      "loss": 1.6684,
      "step": 5456
    },
    {
      "epoch": 0.19542679105413002,
      "grad_norm": 1.3564072847366333,
      "learning_rate": 0.0001859881714642124,
      "loss": 1.6664,
      "step": 5457
    },
    {
      "epoch": 0.19546260318369832,
      "grad_norm": 1.9026246070861816,
      "learning_rate": 0.00018598224966709332,
      "loss": 1.6176,
      "step": 5458
    },
    {
      "epoch": 0.19549841531326662,
      "grad_norm": 1.4654345512390137,
      "learning_rate": 0.00018597632671319398,
      "loss": 1.2719,
      "step": 5459
    },
    {
      "epoch": 0.19553422744283488,
      "grad_norm": 1.4894424676895142,
      "learning_rate": 0.0001859704026025941,
      "loss": 1.519,
      "step": 5460
    },
    {
      "epoch": 0.19557003957240318,
      "grad_norm": 1.355757713317871,
      "learning_rate": 0.0001859644773353733,
      "loss": 1.4525,
      "step": 5461
    },
    {
      "epoch": 0.19560585170197145,
      "grad_norm": 1.8707811832427979,
      "learning_rate": 0.00018595855091161137,
      "loss": 1.4504,
      "step": 5462
    },
    {
      "epoch": 0.19564166383153975,
      "grad_norm": 1.3696109056472778,
      "learning_rate": 0.00018595262333138802,
      "loss": 1.7364,
      "step": 5463
    },
    {
      "epoch": 0.19567747596110802,
      "grad_norm": 1.756722092628479,
      "learning_rate": 0.000185946694594783,
      "loss": 1.9198,
      "step": 5464
    },
    {
      "epoch": 0.1957132880906763,
      "grad_norm": 1.6089026927947998,
      "learning_rate": 0.000185940764701876,
      "loss": 1.4237,
      "step": 5465
    },
    {
      "epoch": 0.1957491002202446,
      "grad_norm": 2.058664560317993,
      "learning_rate": 0.00018593483365274694,
      "loss": 1.4842,
      "step": 5466
    },
    {
      "epoch": 0.19578491234981288,
      "grad_norm": 1.3057408332824707,
      "learning_rate": 0.00018592890144747553,
      "loss": 1.4789,
      "step": 5467
    },
    {
      "epoch": 0.19582072447938118,
      "grad_norm": 2.4167938232421875,
      "learning_rate": 0.00018592296808614156,
      "loss": 1.9637,
      "step": 5468
    },
    {
      "epoch": 0.19585653660894944,
      "grad_norm": 1.7090920209884644,
      "learning_rate": 0.0001859170335688249,
      "loss": 1.3604,
      "step": 5469
    },
    {
      "epoch": 0.19589234873851774,
      "grad_norm": 1.6871206760406494,
      "learning_rate": 0.0001859110978956054,
      "loss": 1.3423,
      "step": 5470
    },
    {
      "epoch": 0.195928160868086,
      "grad_norm": 1.3249101638793945,
      "learning_rate": 0.00018590516106656288,
      "loss": 1.6105,
      "step": 5471
    },
    {
      "epoch": 0.1959639729976543,
      "grad_norm": 2.58074688911438,
      "learning_rate": 0.00018589922308177723,
      "loss": 1.5327,
      "step": 5472
    },
    {
      "epoch": 0.1959997851272226,
      "grad_norm": 1.6249611377716064,
      "learning_rate": 0.0001858932839413283,
      "loss": 1.4159,
      "step": 5473
    },
    {
      "epoch": 0.19603559725679087,
      "grad_norm": 1.391710877418518,
      "learning_rate": 0.0001858873436452961,
      "loss": 1.2536,
      "step": 5474
    },
    {
      "epoch": 0.19607140938635917,
      "grad_norm": 1.7863149642944336,
      "learning_rate": 0.0001858814021937604,
      "loss": 1.2555,
      "step": 5475
    },
    {
      "epoch": 0.19610722151592744,
      "grad_norm": 2.6330697536468506,
      "learning_rate": 0.0001858754595868013,
      "loss": 1.3862,
      "step": 5476
    },
    {
      "epoch": 0.19614303364549573,
      "grad_norm": 1.5848230123519897,
      "learning_rate": 0.0001858695158244986,
      "loss": 1.36,
      "step": 5477
    },
    {
      "epoch": 0.196178845775064,
      "grad_norm": 2.69571852684021,
      "learning_rate": 0.00018586357090693233,
      "loss": 2.119,
      "step": 5478
    },
    {
      "epoch": 0.1962146579046323,
      "grad_norm": 1.6064741611480713,
      "learning_rate": 0.0001858576248341825,
      "loss": 1.4919,
      "step": 5479
    },
    {
      "epoch": 0.1962504700342006,
      "grad_norm": 3.198639154434204,
      "learning_rate": 0.00018585167760632905,
      "loss": 1.6805,
      "step": 5480
    },
    {
      "epoch": 0.19628628216376887,
      "grad_norm": 1.5669358968734741,
      "learning_rate": 0.00018584572922345202,
      "loss": 1.42,
      "step": 5481
    },
    {
      "epoch": 0.19632209429333716,
      "grad_norm": 1.6380369663238525,
      "learning_rate": 0.00018583977968563144,
      "loss": 1.4726,
      "step": 5482
    },
    {
      "epoch": 0.19635790642290543,
      "grad_norm": 1.8254414796829224,
      "learning_rate": 0.00018583382899294736,
      "loss": 1.4393,
      "step": 5483
    },
    {
      "epoch": 0.19639371855247373,
      "grad_norm": 1.700395941734314,
      "learning_rate": 0.00018582787714547982,
      "loss": 1.3454,
      "step": 5484
    },
    {
      "epoch": 0.196429530682042,
      "grad_norm": 1.4598270654678345,
      "learning_rate": 0.0001858219241433089,
      "loss": 1.7212,
      "step": 5485
    },
    {
      "epoch": 0.1964653428116103,
      "grad_norm": 1.7634484767913818,
      "learning_rate": 0.0001858159699865147,
      "loss": 1.47,
      "step": 5486
    },
    {
      "epoch": 0.1965011549411786,
      "grad_norm": 1.5036978721618652,
      "learning_rate": 0.00018581001467517734,
      "loss": 1.8402,
      "step": 5487
    },
    {
      "epoch": 0.19653696707074686,
      "grad_norm": 2.055551767349243,
      "learning_rate": 0.00018580405820937688,
      "loss": 2.0814,
      "step": 5488
    },
    {
      "epoch": 0.19657277920031516,
      "grad_norm": 1.3088536262512207,
      "learning_rate": 0.0001857981005891935,
      "loss": 1.5177,
      "step": 5489
    },
    {
      "epoch": 0.19660859132988343,
      "grad_norm": 1.9697926044464111,
      "learning_rate": 0.00018579214181470736,
      "loss": 1.3903,
      "step": 5490
    },
    {
      "epoch": 0.19664440345945172,
      "grad_norm": 1.503617286682129,
      "learning_rate": 0.00018578618188599863,
      "loss": 1.5119,
      "step": 5491
    },
    {
      "epoch": 0.19668021558902,
      "grad_norm": 2.178367853164673,
      "learning_rate": 0.00018578022080314747,
      "loss": 1.3386,
      "step": 5492
    },
    {
      "epoch": 0.1967160277185883,
      "grad_norm": 1.7492157220840454,
      "learning_rate": 0.00018577425856623408,
      "loss": 1.5961,
      "step": 5493
    },
    {
      "epoch": 0.19675183984815658,
      "grad_norm": 1.7134050130844116,
      "learning_rate": 0.00018576829517533868,
      "loss": 1.3217,
      "step": 5494
    },
    {
      "epoch": 0.19678765197772485,
      "grad_norm": 1.4412091970443726,
      "learning_rate": 0.00018576233063054151,
      "loss": 1.5938,
      "step": 5495
    },
    {
      "epoch": 0.19682346410729315,
      "grad_norm": 1.636104702949524,
      "learning_rate": 0.00018575636493192282,
      "loss": 1.8974,
      "step": 5496
    },
    {
      "epoch": 0.19685927623686142,
      "grad_norm": 1.2458416223526,
      "learning_rate": 0.00018575039807956282,
      "loss": 1.6478,
      "step": 5497
    },
    {
      "epoch": 0.19689508836642972,
      "grad_norm": 2.1648898124694824,
      "learning_rate": 0.00018574443007354186,
      "loss": 1.4979,
      "step": 5498
    },
    {
      "epoch": 0.19693090049599798,
      "grad_norm": 1.4280565977096558,
      "learning_rate": 0.00018573846091394017,
      "loss": 1.4956,
      "step": 5499
    },
    {
      "epoch": 0.19696671262556628,
      "grad_norm": 1.6333988904953003,
      "learning_rate": 0.00018573249060083812,
      "loss": 1.5715,
      "step": 5500
    },
    {
      "epoch": 0.19700252475513455,
      "grad_norm": 1.6430987119674683,
      "learning_rate": 0.00018572651913431596,
      "loss": 1.699,
      "step": 5501
    },
    {
      "epoch": 0.19703833688470285,
      "grad_norm": 1.433403730392456,
      "learning_rate": 0.00018572054651445408,
      "loss": 1.5459,
      "step": 5502
    },
    {
      "epoch": 0.19707414901427114,
      "grad_norm": 2.47914981842041,
      "learning_rate": 0.00018571457274133279,
      "loss": 1.7359,
      "step": 5503
    },
    {
      "epoch": 0.1971099611438394,
      "grad_norm": 1.688639760017395,
      "learning_rate": 0.0001857085978150325,
      "loss": 1.7368,
      "step": 5504
    },
    {
      "epoch": 0.1971457732734077,
      "grad_norm": 1.4147371053695679,
      "learning_rate": 0.0001857026217356336,
      "loss": 1.4653,
      "step": 5505
    },
    {
      "epoch": 0.19718158540297598,
      "grad_norm": 1.7496812343597412,
      "learning_rate": 0.00018569664450321645,
      "loss": 1.5381,
      "step": 5506
    },
    {
      "epoch": 0.19721739753254428,
      "grad_norm": 1.178428292274475,
      "learning_rate": 0.00018569066611786152,
      "loss": 1.559,
      "step": 5507
    },
    {
      "epoch": 0.19725320966211254,
      "grad_norm": 1.813152551651001,
      "learning_rate": 0.00018568468657964918,
      "loss": 1.4128,
      "step": 5508
    },
    {
      "epoch": 0.19728902179168084,
      "grad_norm": 1.3909872770309448,
      "learning_rate": 0.00018567870588865994,
      "loss": 1.6463,
      "step": 5509
    },
    {
      "epoch": 0.19732483392124914,
      "grad_norm": 1.5748597383499146,
      "learning_rate": 0.0001856727240449742,
      "loss": 1.7144,
      "step": 5510
    },
    {
      "epoch": 0.1973606460508174,
      "grad_norm": 1.5605316162109375,
      "learning_rate": 0.0001856667410486725,
      "loss": 1.8678,
      "step": 5511
    },
    {
      "epoch": 0.1973964581803857,
      "grad_norm": 1.2146022319793701,
      "learning_rate": 0.00018566075689983527,
      "loss": 1.7029,
      "step": 5512
    },
    {
      "epoch": 0.19743227030995397,
      "grad_norm": 1.2470415830612183,
      "learning_rate": 0.00018565477159854306,
      "loss": 1.7266,
      "step": 5513
    },
    {
      "epoch": 0.19746808243952227,
      "grad_norm": 1.8456315994262695,
      "learning_rate": 0.00018564878514487637,
      "loss": 1.4474,
      "step": 5514
    },
    {
      "epoch": 0.19750389456909054,
      "grad_norm": 1.6164729595184326,
      "learning_rate": 0.0001856427975389158,
      "loss": 1.8401,
      "step": 5515
    },
    {
      "epoch": 0.19753970669865883,
      "grad_norm": 2.2162938117980957,
      "learning_rate": 0.00018563680878074182,
      "loss": 1.4896,
      "step": 5516
    },
    {
      "epoch": 0.19757551882822713,
      "grad_norm": 1.71847403049469,
      "learning_rate": 0.00018563081887043505,
      "loss": 1.5033,
      "step": 5517
    },
    {
      "epoch": 0.1976113309577954,
      "grad_norm": 1.6429033279418945,
      "learning_rate": 0.00018562482780807606,
      "loss": 1.4215,
      "step": 5518
    },
    {
      "epoch": 0.1976471430873637,
      "grad_norm": 1.78325355052948,
      "learning_rate": 0.00018561883559374548,
      "loss": 1.4578,
      "step": 5519
    },
    {
      "epoch": 0.19768295521693197,
      "grad_norm": 2.1721127033233643,
      "learning_rate": 0.0001856128422275239,
      "loss": 1.6051,
      "step": 5520
    },
    {
      "epoch": 0.19771876734650026,
      "grad_norm": 1.7439571619033813,
      "learning_rate": 0.00018560684770949198,
      "loss": 1.6344,
      "step": 5521
    },
    {
      "epoch": 0.19775457947606853,
      "grad_norm": 2.096254348754883,
      "learning_rate": 0.0001856008520397303,
      "loss": 1.6342,
      "step": 5522
    },
    {
      "epoch": 0.19779039160563683,
      "grad_norm": 2.046941041946411,
      "learning_rate": 0.00018559485521831958,
      "loss": 1.6274,
      "step": 5523
    },
    {
      "epoch": 0.19782620373520513,
      "grad_norm": 1.634363055229187,
      "learning_rate": 0.00018558885724534054,
      "loss": 1.4636,
      "step": 5524
    },
    {
      "epoch": 0.1978620158647734,
      "grad_norm": 1.593102216720581,
      "learning_rate": 0.00018558285812087378,
      "loss": 1.6564,
      "step": 5525
    },
    {
      "epoch": 0.1978978279943417,
      "grad_norm": 1.5818394422531128,
      "learning_rate": 0.0001855768578450001,
      "loss": 1.6222,
      "step": 5526
    },
    {
      "epoch": 0.19793364012390996,
      "grad_norm": 1.5421195030212402,
      "learning_rate": 0.00018557085641780018,
      "loss": 1.6816,
      "step": 5527
    },
    {
      "epoch": 0.19796945225347826,
      "grad_norm": 1.7133044004440308,
      "learning_rate": 0.0001855648538393547,
      "loss": 1.7907,
      "step": 5528
    },
    {
      "epoch": 0.19800526438304653,
      "grad_norm": 3.0813517570495605,
      "learning_rate": 0.00018555885010974454,
      "loss": 1.4835,
      "step": 5529
    },
    {
      "epoch": 0.19804107651261482,
      "grad_norm": 2.0917038917541504,
      "learning_rate": 0.00018555284522905042,
      "loss": 1.6455,
      "step": 5530
    },
    {
      "epoch": 0.19807688864218312,
      "grad_norm": 1.5909149646759033,
      "learning_rate": 0.00018554683919735313,
      "loss": 1.6325,
      "step": 5531
    },
    {
      "epoch": 0.1981127007717514,
      "grad_norm": 1.382629156112671,
      "learning_rate": 0.0001855408320147334,
      "loss": 1.5228,
      "step": 5532
    },
    {
      "epoch": 0.19814851290131968,
      "grad_norm": 1.8413569927215576,
      "learning_rate": 0.00018553482368127217,
      "loss": 1.6239,
      "step": 5533
    },
    {
      "epoch": 0.19818432503088795,
      "grad_norm": 1.6792410612106323,
      "learning_rate": 0.0001855288141970502,
      "loss": 1.5107,
      "step": 5534
    },
    {
      "epoch": 0.19822013716045625,
      "grad_norm": 1.9996005296707153,
      "learning_rate": 0.00018552280356214838,
      "loss": 1.5626,
      "step": 5535
    },
    {
      "epoch": 0.19825594929002452,
      "grad_norm": 1.782104253768921,
      "learning_rate": 0.00018551679177664755,
      "loss": 1.4933,
      "step": 5536
    },
    {
      "epoch": 0.19829176141959282,
      "grad_norm": 1.5002424716949463,
      "learning_rate": 0.0001855107788406286,
      "loss": 1.6503,
      "step": 5537
    },
    {
      "epoch": 0.1983275735491611,
      "grad_norm": 1.3976023197174072,
      "learning_rate": 0.0001855047647541724,
      "loss": 1.5177,
      "step": 5538
    },
    {
      "epoch": 0.19836338567872938,
      "grad_norm": 1.560597538948059,
      "learning_rate": 0.00018549874951735988,
      "loss": 1.7587,
      "step": 5539
    },
    {
      "epoch": 0.19839919780829768,
      "grad_norm": 1.5787992477416992,
      "learning_rate": 0.00018549273313027198,
      "loss": 1.5259,
      "step": 5540
    },
    {
      "epoch": 0.19843500993786595,
      "grad_norm": 1.5140466690063477,
      "learning_rate": 0.00018548671559298963,
      "loss": 1.6429,
      "step": 5541
    },
    {
      "epoch": 0.19847082206743424,
      "grad_norm": 1.7304792404174805,
      "learning_rate": 0.00018548069690559383,
      "loss": 1.6498,
      "step": 5542
    },
    {
      "epoch": 0.1985066341970025,
      "grad_norm": 2.0761163234710693,
      "learning_rate": 0.00018547467706816546,
      "loss": 1.4024,
      "step": 5543
    },
    {
      "epoch": 0.1985424463265708,
      "grad_norm": 1.838121771812439,
      "learning_rate": 0.00018546865608078559,
      "loss": 1.7005,
      "step": 5544
    },
    {
      "epoch": 0.1985782584561391,
      "grad_norm": 1.4359781742095947,
      "learning_rate": 0.0001854626339435352,
      "loss": 1.7055,
      "step": 5545
    },
    {
      "epoch": 0.19861407058570738,
      "grad_norm": 1.474090576171875,
      "learning_rate": 0.0001854566106564953,
      "loss": 1.4552,
      "step": 5546
    },
    {
      "epoch": 0.19864988271527567,
      "grad_norm": 1.9340765476226807,
      "learning_rate": 0.00018545058621974693,
      "loss": 1.5708,
      "step": 5547
    },
    {
      "epoch": 0.19868569484484394,
      "grad_norm": 1.570233941078186,
      "learning_rate": 0.00018544456063337116,
      "loss": 1.4817,
      "step": 5548
    },
    {
      "epoch": 0.19872150697441224,
      "grad_norm": 1.671566128730774,
      "learning_rate": 0.00018543853389744905,
      "loss": 1.593,
      "step": 5549
    },
    {
      "epoch": 0.1987573191039805,
      "grad_norm": 1.457958459854126,
      "learning_rate": 0.00018543250601206165,
      "loss": 1.7928,
      "step": 5550
    },
    {
      "epoch": 0.1987931312335488,
      "grad_norm": 1.4195009469985962,
      "learning_rate": 0.00018542647697729009,
      "loss": 1.4611,
      "step": 5551
    },
    {
      "epoch": 0.1988289433631171,
      "grad_norm": 1.9061682224273682,
      "learning_rate": 0.00018542044679321549,
      "loss": 1.582,
      "step": 5552
    },
    {
      "epoch": 0.19886475549268537,
      "grad_norm": 1.399892807006836,
      "learning_rate": 0.00018541441545991892,
      "loss": 1.5558,
      "step": 5553
    },
    {
      "epoch": 0.19890056762225367,
      "grad_norm": 1.7510415315628052,
      "learning_rate": 0.00018540838297748162,
      "loss": 1.3359,
      "step": 5554
    },
    {
      "epoch": 0.19893637975182193,
      "grad_norm": 1.2389678955078125,
      "learning_rate": 0.0001854023493459847,
      "loss": 1.3077,
      "step": 5555
    },
    {
      "epoch": 0.19897219188139023,
      "grad_norm": 1.1171600818634033,
      "learning_rate": 0.00018539631456550927,
      "loss": 1.1792,
      "step": 5556
    },
    {
      "epoch": 0.1990080040109585,
      "grad_norm": 1.5652157068252563,
      "learning_rate": 0.00018539027863613664,
      "loss": 1.5377,
      "step": 5557
    },
    {
      "epoch": 0.1990438161405268,
      "grad_norm": 1.341874361038208,
      "learning_rate": 0.0001853842415579479,
      "loss": 1.6419,
      "step": 5558
    },
    {
      "epoch": 0.1990796282700951,
      "grad_norm": 2.2000255584716797,
      "learning_rate": 0.0001853782033310244,
      "loss": 1.5106,
      "step": 5559
    },
    {
      "epoch": 0.19911544039966336,
      "grad_norm": 1.4743818044662476,
      "learning_rate": 0.00018537216395544723,
      "loss": 1.5688,
      "step": 5560
    },
    {
      "epoch": 0.19915125252923166,
      "grad_norm": 2.2175040245056152,
      "learning_rate": 0.00018536612343129778,
      "loss": 1.505,
      "step": 5561
    },
    {
      "epoch": 0.19918706465879993,
      "grad_norm": 1.6958733797073364,
      "learning_rate": 0.0001853600817586572,
      "loss": 1.762,
      "step": 5562
    },
    {
      "epoch": 0.19922287678836822,
      "grad_norm": 1.9428437948226929,
      "learning_rate": 0.00018535403893760684,
      "loss": 1.6962,
      "step": 5563
    },
    {
      "epoch": 0.1992586889179365,
      "grad_norm": 1.8468763828277588,
      "learning_rate": 0.00018534799496822802,
      "loss": 1.464,
      "step": 5564
    },
    {
      "epoch": 0.1992945010475048,
      "grad_norm": 1.587731122970581,
      "learning_rate": 0.00018534194985060198,
      "loss": 1.6521,
      "step": 5565
    },
    {
      "epoch": 0.1993303131770731,
      "grad_norm": 1.6171441078186035,
      "learning_rate": 0.0001853359035848101,
      "loss": 1.6767,
      "step": 5566
    },
    {
      "epoch": 0.19936612530664136,
      "grad_norm": 2.446106195449829,
      "learning_rate": 0.0001853298561709337,
      "loss": 1.5723,
      "step": 5567
    },
    {
      "epoch": 0.19940193743620965,
      "grad_norm": 1.455934762954712,
      "learning_rate": 0.0001853238076090542,
      "loss": 1.6693,
      "step": 5568
    },
    {
      "epoch": 0.19943774956577792,
      "grad_norm": 1.3956012725830078,
      "learning_rate": 0.00018531775789925288,
      "loss": 1.5087,
      "step": 5569
    },
    {
      "epoch": 0.19947356169534622,
      "grad_norm": 1.2480905055999756,
      "learning_rate": 0.00018531170704161117,
      "loss": 1.4799,
      "step": 5570
    },
    {
      "epoch": 0.1995093738249145,
      "grad_norm": 2.1647627353668213,
      "learning_rate": 0.00018530565503621052,
      "loss": 1.5694,
      "step": 5571
    },
    {
      "epoch": 0.19954518595448278,
      "grad_norm": 2.104330062866211,
      "learning_rate": 0.00018529960188313233,
      "loss": 1.3568,
      "step": 5572
    },
    {
      "epoch": 0.19958099808405108,
      "grad_norm": 1.5048644542694092,
      "learning_rate": 0.000185293547582458,
      "loss": 1.4253,
      "step": 5573
    },
    {
      "epoch": 0.19961681021361935,
      "grad_norm": 1.904709815979004,
      "learning_rate": 0.000185287492134269,
      "loss": 1.7133,
      "step": 5574
    },
    {
      "epoch": 0.19965262234318765,
      "grad_norm": 2.1871891021728516,
      "learning_rate": 0.00018528143553864682,
      "loss": 1.4204,
      "step": 5575
    },
    {
      "epoch": 0.19968843447275592,
      "grad_norm": 1.9296875,
      "learning_rate": 0.00018527537779567294,
      "loss": 1.8226,
      "step": 5576
    },
    {
      "epoch": 0.1997242466023242,
      "grad_norm": 1.6656574010849,
      "learning_rate": 0.00018526931890542882,
      "loss": 1.4225,
      "step": 5577
    },
    {
      "epoch": 0.19976005873189248,
      "grad_norm": 1.660765290260315,
      "learning_rate": 0.00018526325886799601,
      "loss": 1.3672,
      "step": 5578
    },
    {
      "epoch": 0.19979587086146078,
      "grad_norm": 1.7996066808700562,
      "learning_rate": 0.00018525719768345606,
      "loss": 1.7538,
      "step": 5579
    },
    {
      "epoch": 0.19983168299102907,
      "grad_norm": 2.2600836753845215,
      "learning_rate": 0.00018525113535189047,
      "loss": 2.0307,
      "step": 5580
    },
    {
      "epoch": 0.19986749512059734,
      "grad_norm": 1.627731204032898,
      "learning_rate": 0.00018524507187338082,
      "loss": 1.7304,
      "step": 5581
    },
    {
      "epoch": 0.19990330725016564,
      "grad_norm": 1.4446642398834229,
      "learning_rate": 0.00018523900724800872,
      "loss": 1.4868,
      "step": 5582
    },
    {
      "epoch": 0.1999391193797339,
      "grad_norm": 1.7544971704483032,
      "learning_rate": 0.00018523294147585568,
      "loss": 1.373,
      "step": 5583
    },
    {
      "epoch": 0.1999749315093022,
      "grad_norm": 1.6773810386657715,
      "learning_rate": 0.00018522687455700337,
      "loss": 1.5266,
      "step": 5584
    },
    {
      "epoch": 0.20001074363887048,
      "grad_norm": 1.7085083723068237,
      "learning_rate": 0.0001852208064915334,
      "loss": 1.6517,
      "step": 5585
    },
    {
      "epoch": 0.20004655576843877,
      "grad_norm": 1.6179802417755127,
      "learning_rate": 0.00018521473727952742,
      "loss": 1.577,
      "step": 5586
    },
    {
      "epoch": 0.20008236789800707,
      "grad_norm": 1.6990835666656494,
      "learning_rate": 0.00018520866692106703,
      "loss": 1.7635,
      "step": 5587
    },
    {
      "epoch": 0.20011818002757534,
      "grad_norm": 1.5353055000305176,
      "learning_rate": 0.00018520259541623398,
      "loss": 1.6554,
      "step": 5588
    },
    {
      "epoch": 0.20015399215714363,
      "grad_norm": 1.858586072921753,
      "learning_rate": 0.0001851965227651099,
      "loss": 1.5469,
      "step": 5589
    },
    {
      "epoch": 0.2001898042867119,
      "grad_norm": 1.5021321773529053,
      "learning_rate": 0.00018519044896777648,
      "loss": 1.8475,
      "step": 5590
    },
    {
      "epoch": 0.2002256164162802,
      "grad_norm": 1.355085015296936,
      "learning_rate": 0.0001851843740243155,
      "loss": 1.6449,
      "step": 5591
    },
    {
      "epoch": 0.20026142854584847,
      "grad_norm": 1.4706250429153442,
      "learning_rate": 0.00018517829793480861,
      "loss": 1.6929,
      "step": 5592
    },
    {
      "epoch": 0.20029724067541677,
      "grad_norm": 2.08127498626709,
      "learning_rate": 0.0001851722206993376,
      "loss": 1.5465,
      "step": 5593
    },
    {
      "epoch": 0.20033305280498506,
      "grad_norm": 2.2466256618499756,
      "learning_rate": 0.00018516614231798423,
      "loss": 1.8191,
      "step": 5594
    },
    {
      "epoch": 0.20036886493455333,
      "grad_norm": 1.2826673984527588,
      "learning_rate": 0.00018516006279083026,
      "loss": 1.7799,
      "step": 5595
    },
    {
      "epoch": 0.20040467706412163,
      "grad_norm": 1.394559621810913,
      "learning_rate": 0.0001851539821179575,
      "loss": 1.4522,
      "step": 5596
    },
    {
      "epoch": 0.2004404891936899,
      "grad_norm": 1.4629602432250977,
      "learning_rate": 0.00018514790029944777,
      "loss": 1.7201,
      "step": 5597
    },
    {
      "epoch": 0.2004763013232582,
      "grad_norm": 1.7223575115203857,
      "learning_rate": 0.00018514181733538285,
      "loss": 1.4502,
      "step": 5598
    },
    {
      "epoch": 0.20051211345282646,
      "grad_norm": 1.4611260890960693,
      "learning_rate": 0.00018513573322584463,
      "loss": 1.6936,
      "step": 5599
    },
    {
      "epoch": 0.20054792558239476,
      "grad_norm": 1.4663982391357422,
      "learning_rate": 0.0001851296479709149,
      "loss": 1.7965,
      "step": 5600
    },
    {
      "epoch": 0.20058373771196303,
      "grad_norm": 1.651157259941101,
      "learning_rate": 0.00018512356157067558,
      "loss": 1.5611,
      "step": 5601
    },
    {
      "epoch": 0.20061954984153132,
      "grad_norm": 1.538913369178772,
      "learning_rate": 0.00018511747402520857,
      "loss": 1.1651,
      "step": 5602
    },
    {
      "epoch": 0.20065536197109962,
      "grad_norm": 1.3776566982269287,
      "learning_rate": 0.0001851113853345957,
      "loss": 1.676,
      "step": 5603
    },
    {
      "epoch": 0.2006911741006679,
      "grad_norm": 1.6608750820159912,
      "learning_rate": 0.00018510529549891895,
      "loss": 1.6722,
      "step": 5604
    },
    {
      "epoch": 0.2007269862302362,
      "grad_norm": 2.144692897796631,
      "learning_rate": 0.00018509920451826022,
      "loss": 1.2254,
      "step": 5605
    },
    {
      "epoch": 0.20076279835980446,
      "grad_norm": 1.9433279037475586,
      "learning_rate": 0.00018509311239270145,
      "loss": 1.4043,
      "step": 5606
    },
    {
      "epoch": 0.20079861048937275,
      "grad_norm": 1.8131729364395142,
      "learning_rate": 0.00018508701912232464,
      "loss": 1.4766,
      "step": 5607
    },
    {
      "epoch": 0.20083442261894102,
      "grad_norm": 1.691941261291504,
      "learning_rate": 0.00018508092470721175,
      "loss": 1.3085,
      "step": 5608
    },
    {
      "epoch": 0.20087023474850932,
      "grad_norm": 1.7415412664413452,
      "learning_rate": 0.0001850748291474447,
      "loss": 1.6581,
      "step": 5609
    },
    {
      "epoch": 0.20090604687807762,
      "grad_norm": 1.5415544509887695,
      "learning_rate": 0.00018506873244310563,
      "loss": 1.3757,
      "step": 5610
    },
    {
      "epoch": 0.20094185900764588,
      "grad_norm": 1.8270162343978882,
      "learning_rate": 0.00018506263459427648,
      "loss": 1.4954,
      "step": 5611
    },
    {
      "epoch": 0.20097767113721418,
      "grad_norm": 1.324442982673645,
      "learning_rate": 0.00018505653560103928,
      "loss": 1.5829,
      "step": 5612
    },
    {
      "epoch": 0.20101348326678245,
      "grad_norm": 1.567414402961731,
      "learning_rate": 0.00018505043546347612,
      "loss": 1.3768,
      "step": 5613
    },
    {
      "epoch": 0.20104929539635075,
      "grad_norm": 1.504832148551941,
      "learning_rate": 0.00018504433418166908,
      "loss": 1.4962,
      "step": 5614
    },
    {
      "epoch": 0.20108510752591902,
      "grad_norm": 1.508852481842041,
      "learning_rate": 0.00018503823175570021,
      "loss": 1.6095,
      "step": 5615
    },
    {
      "epoch": 0.2011209196554873,
      "grad_norm": 1.5696232318878174,
      "learning_rate": 0.00018503212818565161,
      "loss": 1.3963,
      "step": 5616
    },
    {
      "epoch": 0.2011567317850556,
      "grad_norm": 1.629343867301941,
      "learning_rate": 0.00018502602347160544,
      "loss": 1.4996,
      "step": 5617
    },
    {
      "epoch": 0.20119254391462388,
      "grad_norm": 1.793253779411316,
      "learning_rate": 0.00018501991761364376,
      "loss": 1.4764,
      "step": 5618
    },
    {
      "epoch": 0.20122835604419217,
      "grad_norm": 1.5109243392944336,
      "learning_rate": 0.00018501381061184876,
      "loss": 1.71,
      "step": 5619
    },
    {
      "epoch": 0.20126416817376044,
      "grad_norm": 1.6845890283584595,
      "learning_rate": 0.0001850077024663026,
      "loss": 1.411,
      "step": 5620
    },
    {
      "epoch": 0.20129998030332874,
      "grad_norm": 1.3381202220916748,
      "learning_rate": 0.00018500159317708749,
      "loss": 1.9312,
      "step": 5621
    },
    {
      "epoch": 0.201335792432897,
      "grad_norm": 1.788730502128601,
      "learning_rate": 0.00018499548274428557,
      "loss": 1.4369,
      "step": 5622
    },
    {
      "epoch": 0.2013716045624653,
      "grad_norm": 1.5795271396636963,
      "learning_rate": 0.00018498937116797904,
      "loss": 1.4818,
      "step": 5623
    },
    {
      "epoch": 0.2014074166920336,
      "grad_norm": 1.4016799926757812,
      "learning_rate": 0.0001849832584482502,
      "loss": 1.6474,
      "step": 5624
    },
    {
      "epoch": 0.20144322882160187,
      "grad_norm": 1.7416629791259766,
      "learning_rate": 0.00018497714458518122,
      "loss": 1.6052,
      "step": 5625
    },
    {
      "epoch": 0.20147904095117017,
      "grad_norm": 1.3918477296829224,
      "learning_rate": 0.00018497102957885434,
      "loss": 1.349,
      "step": 5626
    },
    {
      "epoch": 0.20151485308073844,
      "grad_norm": 1.4549763202667236,
      "learning_rate": 0.0001849649134293519,
      "loss": 1.5576,
      "step": 5627
    },
    {
      "epoch": 0.20155066521030673,
      "grad_norm": 1.955651044845581,
      "learning_rate": 0.00018495879613675612,
      "loss": 1.4563,
      "step": 5628
    },
    {
      "epoch": 0.201586477339875,
      "grad_norm": 1.9450856447219849,
      "learning_rate": 0.00018495267770114935,
      "loss": 1.3856,
      "step": 5629
    },
    {
      "epoch": 0.2016222894694433,
      "grad_norm": 1.2987256050109863,
      "learning_rate": 0.00018494655812261387,
      "loss": 1.5538,
      "step": 5630
    },
    {
      "epoch": 0.2016581015990116,
      "grad_norm": 1.7535088062286377,
      "learning_rate": 0.00018494043740123202,
      "loss": 1.7828,
      "step": 5631
    },
    {
      "epoch": 0.20169391372857987,
      "grad_norm": 1.5175572633743286,
      "learning_rate": 0.00018493431553708614,
      "loss": 1.6587,
      "step": 5632
    },
    {
      "epoch": 0.20172972585814816,
      "grad_norm": 1.82554292678833,
      "learning_rate": 0.0001849281925302586,
      "loss": 1.6141,
      "step": 5633
    },
    {
      "epoch": 0.20176553798771643,
      "grad_norm": 1.5592626333236694,
      "learning_rate": 0.0001849220683808318,
      "loss": 1.5767,
      "step": 5634
    },
    {
      "epoch": 0.20180135011728473,
      "grad_norm": 1.3777804374694824,
      "learning_rate": 0.00018491594308888814,
      "loss": 1.4966,
      "step": 5635
    },
    {
      "epoch": 0.201837162246853,
      "grad_norm": 1.6428275108337402,
      "learning_rate": 0.00018490981665450994,
      "loss": 1.6563,
      "step": 5636
    },
    {
      "epoch": 0.2018729743764213,
      "grad_norm": 1.416022777557373,
      "learning_rate": 0.00018490368907777974,
      "loss": 1.8694,
      "step": 5637
    },
    {
      "epoch": 0.2019087865059896,
      "grad_norm": 1.6186103820800781,
      "learning_rate": 0.0001848975603587799,
      "loss": 1.5395,
      "step": 5638
    },
    {
      "epoch": 0.20194459863555786,
      "grad_norm": 1.5777395963668823,
      "learning_rate": 0.00018489143049759286,
      "loss": 1.6282,
      "step": 5639
    },
    {
      "epoch": 0.20198041076512616,
      "grad_norm": 1.8486963510513306,
      "learning_rate": 0.00018488529949430116,
      "loss": 1.5511,
      "step": 5640
    },
    {
      "epoch": 0.20201622289469442,
      "grad_norm": 2.059964418411255,
      "learning_rate": 0.00018487916734898722,
      "loss": 1.5742,
      "step": 5641
    },
    {
      "epoch": 0.20205203502426272,
      "grad_norm": 2.6553430557250977,
      "learning_rate": 0.0001848730340617336,
      "loss": 1.5722,
      "step": 5642
    },
    {
      "epoch": 0.202087847153831,
      "grad_norm": 1.6141916513442993,
      "learning_rate": 0.00018486689963262277,
      "loss": 1.6907,
      "step": 5643
    },
    {
      "epoch": 0.2021236592833993,
      "grad_norm": 1.7546532154083252,
      "learning_rate": 0.00018486076406173726,
      "loss": 1.4523,
      "step": 5644
    },
    {
      "epoch": 0.20215947141296758,
      "grad_norm": 1.4749776124954224,
      "learning_rate": 0.00018485462734915966,
      "loss": 1.5986,
      "step": 5645
    },
    {
      "epoch": 0.20219528354253585,
      "grad_norm": 1.5348141193389893,
      "learning_rate": 0.0001848484894949725,
      "loss": 1.7698,
      "step": 5646
    },
    {
      "epoch": 0.20223109567210415,
      "grad_norm": 1.4728552103042603,
      "learning_rate": 0.00018484235049925836,
      "loss": 1.6859,
      "step": 5647
    },
    {
      "epoch": 0.20226690780167242,
      "grad_norm": 2.210272789001465,
      "learning_rate": 0.00018483621036209983,
      "loss": 1.907,
      "step": 5648
    },
    {
      "epoch": 0.20230271993124072,
      "grad_norm": 1.673674464225769,
      "learning_rate": 0.0001848300690835795,
      "loss": 1.665,
      "step": 5649
    },
    {
      "epoch": 0.20233853206080898,
      "grad_norm": 1.225216269493103,
      "learning_rate": 0.00018482392666378003,
      "loss": 1.5294,
      "step": 5650
    },
    {
      "epoch": 0.20237434419037728,
      "grad_norm": 1.159347414970398,
      "learning_rate": 0.00018481778310278405,
      "loss": 1.4917,
      "step": 5651
    },
    {
      "epoch": 0.20241015631994558,
      "grad_norm": 1.6550319194793701,
      "learning_rate": 0.0001848116384006742,
      "loss": 1.6163,
      "step": 5652
    },
    {
      "epoch": 0.20244596844951385,
      "grad_norm": 1.5937591791152954,
      "learning_rate": 0.00018480549255753313,
      "loss": 1.4051,
      "step": 5653
    },
    {
      "epoch": 0.20248178057908214,
      "grad_norm": 1.5911157131195068,
      "learning_rate": 0.0001847993455734436,
      "loss": 1.6417,
      "step": 5654
    },
    {
      "epoch": 0.2025175927086504,
      "grad_norm": 1.6294200420379639,
      "learning_rate": 0.00018479319744848821,
      "loss": 1.7678,
      "step": 5655
    },
    {
      "epoch": 0.2025534048382187,
      "grad_norm": 2.7911853790283203,
      "learning_rate": 0.00018478704818274976,
      "loss": 1.693,
      "step": 5656
    },
    {
      "epoch": 0.20258921696778698,
      "grad_norm": 1.8016701936721802,
      "learning_rate": 0.00018478089777631092,
      "loss": 1.608,
      "step": 5657
    },
    {
      "epoch": 0.20262502909735527,
      "grad_norm": 1.3736586570739746,
      "learning_rate": 0.00018477474622925449,
      "loss": 1.5401,
      "step": 5658
    },
    {
      "epoch": 0.20266084122692357,
      "grad_norm": 1.949476718902588,
      "learning_rate": 0.00018476859354166317,
      "loss": 1.7005,
      "step": 5659
    },
    {
      "epoch": 0.20269665335649184,
      "grad_norm": 1.9096870422363281,
      "learning_rate": 0.0001847624397136198,
      "loss": 1.6647,
      "step": 5660
    },
    {
      "epoch": 0.20273246548606014,
      "grad_norm": 1.571459412574768,
      "learning_rate": 0.0001847562847452071,
      "loss": 1.6983,
      "step": 5661
    },
    {
      "epoch": 0.2027682776156284,
      "grad_norm": 1.6884174346923828,
      "learning_rate": 0.0001847501286365079,
      "loss": 1.4849,
      "step": 5662
    },
    {
      "epoch": 0.2028040897451967,
      "grad_norm": 1.2368390560150146,
      "learning_rate": 0.00018474397138760508,
      "loss": 1.5239,
      "step": 5663
    },
    {
      "epoch": 0.20283990187476497,
      "grad_norm": 2.1719744205474854,
      "learning_rate": 0.00018473781299858146,
      "loss": 1.7039,
      "step": 5664
    },
    {
      "epoch": 0.20287571400433327,
      "grad_norm": 1.8494501113891602,
      "learning_rate": 0.0001847316534695198,
      "loss": 1.8157,
      "step": 5665
    },
    {
      "epoch": 0.20291152613390157,
      "grad_norm": 1.4276621341705322,
      "learning_rate": 0.0001847254928005031,
      "loss": 1.8173,
      "step": 5666
    },
    {
      "epoch": 0.20294733826346983,
      "grad_norm": 1.520430088043213,
      "learning_rate": 0.00018471933099161415,
      "loss": 1.6106,
      "step": 5667
    },
    {
      "epoch": 0.20298315039303813,
      "grad_norm": 1.8959459066390991,
      "learning_rate": 0.00018471316804293594,
      "loss": 1.7846,
      "step": 5668
    },
    {
      "epoch": 0.2030189625226064,
      "grad_norm": 1.4433817863464355,
      "learning_rate": 0.00018470700395455125,
      "loss": 1.9018,
      "step": 5669
    },
    {
      "epoch": 0.2030547746521747,
      "grad_norm": 1.5954314470291138,
      "learning_rate": 0.00018470083872654312,
      "loss": 1.2725,
      "step": 5670
    },
    {
      "epoch": 0.20309058678174297,
      "grad_norm": 1.5294626951217651,
      "learning_rate": 0.00018469467235899444,
      "loss": 1.5162,
      "step": 5671
    },
    {
      "epoch": 0.20312639891131126,
      "grad_norm": 1.519560694694519,
      "learning_rate": 0.00018468850485198822,
      "loss": 1.5687,
      "step": 5672
    },
    {
      "epoch": 0.20316221104087956,
      "grad_norm": 2.257767915725708,
      "learning_rate": 0.00018468233620560739,
      "loss": 1.3524,
      "step": 5673
    },
    {
      "epoch": 0.20319802317044783,
      "grad_norm": 1.5276037454605103,
      "learning_rate": 0.00018467616641993498,
      "loss": 1.6337,
      "step": 5674
    },
    {
      "epoch": 0.20323383530001612,
      "grad_norm": 1.8103950023651123,
      "learning_rate": 0.00018466999549505392,
      "loss": 1.4483,
      "step": 5675
    },
    {
      "epoch": 0.2032696474295844,
      "grad_norm": 1.5021846294403076,
      "learning_rate": 0.00018466382343104734,
      "loss": 1.6137,
      "step": 5676
    },
    {
      "epoch": 0.2033054595591527,
      "grad_norm": 1.6001121997833252,
      "learning_rate": 0.00018465765022799823,
      "loss": 1.3985,
      "step": 5677
    },
    {
      "epoch": 0.20334127168872096,
      "grad_norm": 1.44736647605896,
      "learning_rate": 0.00018465147588598958,
      "loss": 1.2853,
      "step": 5678
    },
    {
      "epoch": 0.20337708381828926,
      "grad_norm": 1.746927261352539,
      "learning_rate": 0.00018464530040510456,
      "loss": 1.6917,
      "step": 5679
    },
    {
      "epoch": 0.20341289594785755,
      "grad_norm": 1.9499015808105469,
      "learning_rate": 0.0001846391237854262,
      "loss": 1.7338,
      "step": 5680
    },
    {
      "epoch": 0.20344870807742582,
      "grad_norm": 1.527923583984375,
      "learning_rate": 0.0001846329460270376,
      "loss": 1.5749,
      "step": 5681
    },
    {
      "epoch": 0.20348452020699412,
      "grad_norm": 1.8569077253341675,
      "learning_rate": 0.0001846267671300219,
      "loss": 1.7941,
      "step": 5682
    },
    {
      "epoch": 0.2035203323365624,
      "grad_norm": 1.7201156616210938,
      "learning_rate": 0.00018462058709446216,
      "loss": 1.7085,
      "step": 5683
    },
    {
      "epoch": 0.20355614446613068,
      "grad_norm": 3.1380867958068848,
      "learning_rate": 0.00018461440592044165,
      "loss": 1.887,
      "step": 5684
    },
    {
      "epoch": 0.20359195659569895,
      "grad_norm": 1.3634907007217407,
      "learning_rate": 0.00018460822360804338,
      "loss": 1.6883,
      "step": 5685
    },
    {
      "epoch": 0.20362776872526725,
      "grad_norm": 1.4703376293182373,
      "learning_rate": 0.00018460204015735064,
      "loss": 1.5107,
      "step": 5686
    },
    {
      "epoch": 0.20366358085483555,
      "grad_norm": 1.7667032480239868,
      "learning_rate": 0.00018459585556844656,
      "loss": 1.829,
      "step": 5687
    },
    {
      "epoch": 0.20369939298440382,
      "grad_norm": 2.221970796585083,
      "learning_rate": 0.00018458966984141438,
      "loss": 1.3804,
      "step": 5688
    },
    {
      "epoch": 0.2037352051139721,
      "grad_norm": 2.146343231201172,
      "learning_rate": 0.00018458348297633727,
      "loss": 1.6312,
      "step": 5689
    },
    {
      "epoch": 0.20377101724354038,
      "grad_norm": 1.7487691640853882,
      "learning_rate": 0.00018457729497329853,
      "loss": 1.38,
      "step": 5690
    },
    {
      "epoch": 0.20380682937310868,
      "grad_norm": 1.7649887800216675,
      "learning_rate": 0.0001845711058323814,
      "loss": 1.6627,
      "step": 5691
    },
    {
      "epoch": 0.20384264150267695,
      "grad_norm": 1.3033711910247803,
      "learning_rate": 0.0001845649155536691,
      "loss": 1.6029,
      "step": 5692
    },
    {
      "epoch": 0.20387845363224524,
      "grad_norm": 1.5455960035324097,
      "learning_rate": 0.00018455872413724496,
      "loss": 1.573,
      "step": 5693
    },
    {
      "epoch": 0.2039142657618135,
      "grad_norm": 1.7799946069717407,
      "learning_rate": 0.00018455253158319225,
      "loss": 1.4815,
      "step": 5694
    },
    {
      "epoch": 0.2039500778913818,
      "grad_norm": 1.9104725122451782,
      "learning_rate": 0.00018454633789159427,
      "loss": 1.684,
      "step": 5695
    },
    {
      "epoch": 0.2039858900209501,
      "grad_norm": 1.4825611114501953,
      "learning_rate": 0.0001845401430625344,
      "loss": 1.5976,
      "step": 5696
    },
    {
      "epoch": 0.20402170215051837,
      "grad_norm": 1.7644739151000977,
      "learning_rate": 0.00018453394709609598,
      "loss": 2.0426,
      "step": 5697
    },
    {
      "epoch": 0.20405751428008667,
      "grad_norm": 1.5432175397872925,
      "learning_rate": 0.0001845277499923623,
      "loss": 1.8283,
      "step": 5698
    },
    {
      "epoch": 0.20409332640965494,
      "grad_norm": 2.520534038543701,
      "learning_rate": 0.0001845215517514168,
      "loss": 1.7116,
      "step": 5699
    },
    {
      "epoch": 0.20412913853922324,
      "grad_norm": 1.333713173866272,
      "learning_rate": 0.0001845153523733428,
      "loss": 1.5942,
      "step": 5700
    },
    {
      "epoch": 0.2041649506687915,
      "grad_norm": 1.654269814491272,
      "learning_rate": 0.00018450915185822382,
      "loss": 1.8312,
      "step": 5701
    },
    {
      "epoch": 0.2042007627983598,
      "grad_norm": 1.8361196517944336,
      "learning_rate": 0.00018450295020614317,
      "loss": 1.592,
      "step": 5702
    },
    {
      "epoch": 0.2042365749279281,
      "grad_norm": 1.52461576461792,
      "learning_rate": 0.00018449674741718433,
      "loss": 1.8999,
      "step": 5703
    },
    {
      "epoch": 0.20427238705749637,
      "grad_norm": 1.7082089185714722,
      "learning_rate": 0.00018449054349143072,
      "loss": 1.4507,
      "step": 5704
    },
    {
      "epoch": 0.20430819918706467,
      "grad_norm": 1.8044872283935547,
      "learning_rate": 0.0001844843384289659,
      "loss": 1.9495,
      "step": 5705
    },
    {
      "epoch": 0.20434401131663293,
      "grad_norm": 2.0654454231262207,
      "learning_rate": 0.00018447813222987323,
      "loss": 2.0662,
      "step": 5706
    },
    {
      "epoch": 0.20437982344620123,
      "grad_norm": 2.090245008468628,
      "learning_rate": 0.00018447192489423625,
      "loss": 1.7459,
      "step": 5707
    },
    {
      "epoch": 0.2044156355757695,
      "grad_norm": 1.994872808456421,
      "learning_rate": 0.00018446571642213852,
      "loss": 2.0094,
      "step": 5708
    },
    {
      "epoch": 0.2044514477053378,
      "grad_norm": 1.3152283430099487,
      "learning_rate": 0.0001844595068136635,
      "loss": 1.6219,
      "step": 5709
    },
    {
      "epoch": 0.2044872598349061,
      "grad_norm": 1.7344331741333008,
      "learning_rate": 0.0001844532960688948,
      "loss": 1.5429,
      "step": 5710
    },
    {
      "epoch": 0.20452307196447436,
      "grad_norm": 1.7377667427062988,
      "learning_rate": 0.0001844470841879159,
      "loss": 1.6329,
      "step": 5711
    },
    {
      "epoch": 0.20455888409404266,
      "grad_norm": 1.2515347003936768,
      "learning_rate": 0.00018444087117081042,
      "loss": 1.5964,
      "step": 5712
    },
    {
      "epoch": 0.20459469622361093,
      "grad_norm": 1.538106918334961,
      "learning_rate": 0.00018443465701766196,
      "loss": 1.323,
      "step": 5713
    },
    {
      "epoch": 0.20463050835317922,
      "grad_norm": 1.9756258726119995,
      "learning_rate": 0.0001844284417285541,
      "loss": 1.729,
      "step": 5714
    },
    {
      "epoch": 0.2046663204827475,
      "grad_norm": 1.4631593227386475,
      "learning_rate": 0.00018442222530357043,
      "loss": 1.7244,
      "step": 5715
    },
    {
      "epoch": 0.2047021326123158,
      "grad_norm": 1.497786045074463,
      "learning_rate": 0.00018441600774279465,
      "loss": 1.6359,
      "step": 5716
    },
    {
      "epoch": 0.2047379447418841,
      "grad_norm": 1.5431510210037231,
      "learning_rate": 0.00018440978904631032,
      "loss": 1.3245,
      "step": 5717
    },
    {
      "epoch": 0.20477375687145236,
      "grad_norm": 1.3295104503631592,
      "learning_rate": 0.00018440356921420122,
      "loss": 1.5629,
      "step": 5718
    },
    {
      "epoch": 0.20480956900102065,
      "grad_norm": 1.2782552242279053,
      "learning_rate": 0.00018439734824655092,
      "loss": 1.3756,
      "step": 5719
    },
    {
      "epoch": 0.20484538113058892,
      "grad_norm": 1.6534874439239502,
      "learning_rate": 0.00018439112614344322,
      "loss": 1.3338,
      "step": 5720
    },
    {
      "epoch": 0.20488119326015722,
      "grad_norm": 1.5594265460968018,
      "learning_rate": 0.0001843849029049617,
      "loss": 1.5184,
      "step": 5721
    },
    {
      "epoch": 0.2049170053897255,
      "grad_norm": 1.6822896003723145,
      "learning_rate": 0.00018437867853119023,
      "loss": 1.4627,
      "step": 5722
    },
    {
      "epoch": 0.20495281751929378,
      "grad_norm": 2.03576397895813,
      "learning_rate": 0.00018437245302221244,
      "loss": 1.646,
      "step": 5723
    },
    {
      "epoch": 0.20498862964886208,
      "grad_norm": 1.5040608644485474,
      "learning_rate": 0.00018436622637811215,
      "loss": 1.596,
      "step": 5724
    },
    {
      "epoch": 0.20502444177843035,
      "grad_norm": 1.8175115585327148,
      "learning_rate": 0.0001843599985989731,
      "loss": 1.4633,
      "step": 5725
    },
    {
      "epoch": 0.20506025390799865,
      "grad_norm": 2.4232726097106934,
      "learning_rate": 0.0001843537696848791,
      "loss": 1.5774,
      "step": 5726
    },
    {
      "epoch": 0.20509606603756692,
      "grad_norm": 1.5448403358459473,
      "learning_rate": 0.0001843475396359139,
      "loss": 1.234,
      "step": 5727
    },
    {
      "epoch": 0.2051318781671352,
      "grad_norm": 1.332551121711731,
      "learning_rate": 0.00018434130845216138,
      "loss": 1.4809,
      "step": 5728
    },
    {
      "epoch": 0.20516769029670348,
      "grad_norm": 1.351884365081787,
      "learning_rate": 0.00018433507613370534,
      "loss": 1.569,
      "step": 5729
    },
    {
      "epoch": 0.20520350242627178,
      "grad_norm": 1.5869837999343872,
      "learning_rate": 0.00018432884268062964,
      "loss": 1.7137,
      "step": 5730
    },
    {
      "epoch": 0.20523931455584007,
      "grad_norm": 1.690000295639038,
      "learning_rate": 0.00018432260809301816,
      "loss": 1.5449,
      "step": 5731
    },
    {
      "epoch": 0.20527512668540834,
      "grad_norm": 1.3282488584518433,
      "learning_rate": 0.00018431637237095472,
      "loss": 1.3576,
      "step": 5732
    },
    {
      "epoch": 0.20531093881497664,
      "grad_norm": 2.1671175956726074,
      "learning_rate": 0.00018431013551452327,
      "loss": 1.5015,
      "step": 5733
    },
    {
      "epoch": 0.2053467509445449,
      "grad_norm": 1.4850959777832031,
      "learning_rate": 0.0001843038975238077,
      "loss": 1.2529,
      "step": 5734
    },
    {
      "epoch": 0.2053825630741132,
      "grad_norm": 1.4230232238769531,
      "learning_rate": 0.00018429765839889193,
      "loss": 1.3751,
      "step": 5735
    },
    {
      "epoch": 0.20541837520368147,
      "grad_norm": 1.605311632156372,
      "learning_rate": 0.0001842914181398599,
      "loss": 1.6034,
      "step": 5736
    },
    {
      "epoch": 0.20545418733324977,
      "grad_norm": 1.7204928398132324,
      "learning_rate": 0.00018428517674679557,
      "loss": 1.3672,
      "step": 5737
    },
    {
      "epoch": 0.20548999946281807,
      "grad_norm": 1.6784799098968506,
      "learning_rate": 0.0001842789342197829,
      "loss": 1.5203,
      "step": 5738
    },
    {
      "epoch": 0.20552581159238634,
      "grad_norm": 1.939692497253418,
      "learning_rate": 0.00018427269055890588,
      "loss": 1.9134,
      "step": 5739
    },
    {
      "epoch": 0.20556162372195463,
      "grad_norm": 1.4052371978759766,
      "learning_rate": 0.00018426644576424855,
      "loss": 1.5995,
      "step": 5740
    },
    {
      "epoch": 0.2055974358515229,
      "grad_norm": 1.1106784343719482,
      "learning_rate": 0.00018426019983589482,
      "loss": 1.5371,
      "step": 5741
    },
    {
      "epoch": 0.2056332479810912,
      "grad_norm": 1.7211228609085083,
      "learning_rate": 0.00018425395277392882,
      "loss": 1.6038,
      "step": 5742
    },
    {
      "epoch": 0.20566906011065947,
      "grad_norm": 1.6395955085754395,
      "learning_rate": 0.0001842477045784346,
      "loss": 1.4582,
      "step": 5743
    },
    {
      "epoch": 0.20570487224022777,
      "grad_norm": 1.9705967903137207,
      "learning_rate": 0.00018424145524949614,
      "loss": 1.7143,
      "step": 5744
    },
    {
      "epoch": 0.20574068436979606,
      "grad_norm": 3.6009767055511475,
      "learning_rate": 0.00018423520478719758,
      "loss": 1.5917,
      "step": 5745
    },
    {
      "epoch": 0.20577649649936433,
      "grad_norm": 1.7975378036499023,
      "learning_rate": 0.00018422895319162298,
      "loss": 1.4714,
      "step": 5746
    },
    {
      "epoch": 0.20581230862893263,
      "grad_norm": 2.167494535446167,
      "learning_rate": 0.0001842227004628565,
      "loss": 1.4169,
      "step": 5747
    },
    {
      "epoch": 0.2058481207585009,
      "grad_norm": 2.0247933864593506,
      "learning_rate": 0.00018421644660098217,
      "loss": 1.6403,
      "step": 5748
    },
    {
      "epoch": 0.2058839328880692,
      "grad_norm": 1.7096855640411377,
      "learning_rate": 0.00018421019160608424,
      "loss": 1.5459,
      "step": 5749
    },
    {
      "epoch": 0.20591974501763746,
      "grad_norm": 2.594543933868408,
      "learning_rate": 0.00018420393547824676,
      "loss": 1.4367,
      "step": 5750
    },
    {
      "epoch": 0.20595555714720576,
      "grad_norm": 2.801156520843506,
      "learning_rate": 0.000184197678217554,
      "loss": 1.5627,
      "step": 5751
    },
    {
      "epoch": 0.20599136927677406,
      "grad_norm": 1.9084726572036743,
      "learning_rate": 0.00018419141982409001,
      "loss": 1.3815,
      "step": 5752
    },
    {
      "epoch": 0.20602718140634232,
      "grad_norm": 1.531064748764038,
      "learning_rate": 0.00018418516029793916,
      "loss": 1.639,
      "step": 5753
    },
    {
      "epoch": 0.20606299353591062,
      "grad_norm": 1.4335132837295532,
      "learning_rate": 0.00018417889963918548,
      "loss": 1.849,
      "step": 5754
    },
    {
      "epoch": 0.2060988056654789,
      "grad_norm": 1.271621584892273,
      "learning_rate": 0.00018417263784791335,
      "loss": 1.7477,
      "step": 5755
    },
    {
      "epoch": 0.2061346177950472,
      "grad_norm": 1.3262522220611572,
      "learning_rate": 0.0001841663749242069,
      "loss": 1.7865,
      "step": 5756
    },
    {
      "epoch": 0.20617042992461546,
      "grad_norm": 1.6832629442214966,
      "learning_rate": 0.0001841601108681505,
      "loss": 1.3944,
      "step": 5757
    },
    {
      "epoch": 0.20620624205418375,
      "grad_norm": 1.295717477798462,
      "learning_rate": 0.00018415384567982833,
      "loss": 1.6328,
      "step": 5758
    },
    {
      "epoch": 0.20624205418375205,
      "grad_norm": 1.605613350868225,
      "learning_rate": 0.0001841475793593247,
      "loss": 1.4366,
      "step": 5759
    },
    {
      "epoch": 0.20627786631332032,
      "grad_norm": 1.8435920476913452,
      "learning_rate": 0.00018414131190672394,
      "loss": 1.5914,
      "step": 5760
    },
    {
      "epoch": 0.20631367844288862,
      "grad_norm": 1.62568199634552,
      "learning_rate": 0.00018413504332211037,
      "loss": 1.3896,
      "step": 5761
    },
    {
      "epoch": 0.20634949057245688,
      "grad_norm": 1.6363476514816284,
      "learning_rate": 0.00018412877360556834,
      "loss": 1.624,
      "step": 5762
    },
    {
      "epoch": 0.20638530270202518,
      "grad_norm": 1.4820305109024048,
      "learning_rate": 0.00018412250275718218,
      "loss": 1.5039,
      "step": 5763
    },
    {
      "epoch": 0.20642111483159345,
      "grad_norm": 1.8757277727127075,
      "learning_rate": 0.00018411623077703624,
      "loss": 1.7367,
      "step": 5764
    },
    {
      "epoch": 0.20645692696116175,
      "grad_norm": 2.476335287094116,
      "learning_rate": 0.0001841099576652149,
      "loss": 1.8845,
      "step": 5765
    },
    {
      "epoch": 0.20649273909073004,
      "grad_norm": 1.4801583290100098,
      "learning_rate": 0.00018410368342180263,
      "loss": 1.4245,
      "step": 5766
    },
    {
      "epoch": 0.2065285512202983,
      "grad_norm": 1.6733896732330322,
      "learning_rate": 0.00018409740804688373,
      "loss": 1.3733,
      "step": 5767
    },
    {
      "epoch": 0.2065643633498666,
      "grad_norm": 1.4926596879959106,
      "learning_rate": 0.0001840911315405427,
      "loss": 1.4559,
      "step": 5768
    },
    {
      "epoch": 0.20660017547943488,
      "grad_norm": 1.447648048400879,
      "learning_rate": 0.00018408485390286397,
      "loss": 1.5615,
      "step": 5769
    },
    {
      "epoch": 0.20663598760900317,
      "grad_norm": 1.9112826585769653,
      "learning_rate": 0.00018407857513393197,
      "loss": 1.665,
      "step": 5770
    },
    {
      "epoch": 0.20667179973857144,
      "grad_norm": 2.436891794204712,
      "learning_rate": 0.00018407229523383122,
      "loss": 1.681,
      "step": 5771
    },
    {
      "epoch": 0.20670761186813974,
      "grad_norm": 1.7354581356048584,
      "learning_rate": 0.00018406601420264618,
      "loss": 1.4212,
      "step": 5772
    },
    {
      "epoch": 0.20674342399770804,
      "grad_norm": 1.4902445077896118,
      "learning_rate": 0.00018405973204046135,
      "loss": 1.6882,
      "step": 5773
    },
    {
      "epoch": 0.2067792361272763,
      "grad_norm": 1.775509238243103,
      "learning_rate": 0.00018405344874736126,
      "loss": 1.3206,
      "step": 5774
    },
    {
      "epoch": 0.2068150482568446,
      "grad_norm": 1.1771329641342163,
      "learning_rate": 0.00018404716432343044,
      "loss": 1.4992,
      "step": 5775
    },
    {
      "epoch": 0.20685086038641287,
      "grad_norm": 1.741278052330017,
      "learning_rate": 0.0001840408787687534,
      "loss": 1.2352,
      "step": 5776
    },
    {
      "epoch": 0.20688667251598117,
      "grad_norm": 2.0680439472198486,
      "learning_rate": 0.0001840345920834148,
      "loss": 1.5474,
      "step": 5777
    },
    {
      "epoch": 0.20692248464554944,
      "grad_norm": 1.8485260009765625,
      "learning_rate": 0.00018402830426749914,
      "loss": 1.3399,
      "step": 5778
    },
    {
      "epoch": 0.20695829677511773,
      "grad_norm": 1.583206057548523,
      "learning_rate": 0.00018402201532109102,
      "loss": 1.6553,
      "step": 5779
    },
    {
      "epoch": 0.20699410890468603,
      "grad_norm": 1.9147899150848389,
      "learning_rate": 0.00018401572524427505,
      "loss": 1.9019,
      "step": 5780
    },
    {
      "epoch": 0.2070299210342543,
      "grad_norm": 1.735785961151123,
      "learning_rate": 0.0001840094340371359,
      "loss": 1.778,
      "step": 5781
    },
    {
      "epoch": 0.2070657331638226,
      "grad_norm": 1.9432623386383057,
      "learning_rate": 0.00018400314169975818,
      "loss": 1.5151,
      "step": 5782
    },
    {
      "epoch": 0.20710154529339087,
      "grad_norm": 1.6666193008422852,
      "learning_rate": 0.00018399684823222653,
      "loss": 1.2603,
      "step": 5783
    },
    {
      "epoch": 0.20713735742295916,
      "grad_norm": 1.642317771911621,
      "learning_rate": 0.00018399055363462562,
      "loss": 1.4213,
      "step": 5784
    },
    {
      "epoch": 0.20717316955252743,
      "grad_norm": 2.0764431953430176,
      "learning_rate": 0.0001839842579070402,
      "loss": 1.278,
      "step": 5785
    },
    {
      "epoch": 0.20720898168209573,
      "grad_norm": 1.3511927127838135,
      "learning_rate": 0.0001839779610495549,
      "loss": 1.4598,
      "step": 5786
    },
    {
      "epoch": 0.20724479381166402,
      "grad_norm": 1.897929072380066,
      "learning_rate": 0.00018397166306225444,
      "loss": 1.6758,
      "step": 5787
    },
    {
      "epoch": 0.2072806059412323,
      "grad_norm": 1.5352063179016113,
      "learning_rate": 0.00018396536394522359,
      "loss": 1.655,
      "step": 5788
    },
    {
      "epoch": 0.2073164180708006,
      "grad_norm": 1.7674046754837036,
      "learning_rate": 0.00018395906369854704,
      "loss": 1.4136,
      "step": 5789
    },
    {
      "epoch": 0.20735223020036886,
      "grad_norm": 1.2308050394058228,
      "learning_rate": 0.00018395276232230964,
      "loss": 1.4874,
      "step": 5790
    },
    {
      "epoch": 0.20738804232993716,
      "grad_norm": 1.8490521907806396,
      "learning_rate": 0.00018394645981659608,
      "loss": 1.4895,
      "step": 5791
    },
    {
      "epoch": 0.20742385445950542,
      "grad_norm": 1.8193918466567993,
      "learning_rate": 0.00018394015618149122,
      "loss": 1.3781,
      "step": 5792
    },
    {
      "epoch": 0.20745966658907372,
      "grad_norm": 1.4143835306167603,
      "learning_rate": 0.00018393385141707977,
      "loss": 1.5742,
      "step": 5793
    },
    {
      "epoch": 0.207495478718642,
      "grad_norm": 1.3196909427642822,
      "learning_rate": 0.00018392754552344666,
      "loss": 1.3337,
      "step": 5794
    },
    {
      "epoch": 0.2075312908482103,
      "grad_norm": 1.3538028001785278,
      "learning_rate": 0.00018392123850067668,
      "loss": 1.5709,
      "step": 5795
    },
    {
      "epoch": 0.20756710297777858,
      "grad_norm": 1.751012921333313,
      "learning_rate": 0.00018391493034885468,
      "loss": 1.4149,
      "step": 5796
    },
    {
      "epoch": 0.20760291510734685,
      "grad_norm": 1.4756646156311035,
      "learning_rate": 0.00018390862106806554,
      "loss": 1.5291,
      "step": 5797
    },
    {
      "epoch": 0.20763872723691515,
      "grad_norm": 1.8159650564193726,
      "learning_rate": 0.00018390231065839414,
      "loss": 1.5351,
      "step": 5798
    },
    {
      "epoch": 0.20767453936648342,
      "grad_norm": 1.8627691268920898,
      "learning_rate": 0.00018389599911992538,
      "loss": 1.4553,
      "step": 5799
    },
    {
      "epoch": 0.20771035149605171,
      "grad_norm": 1.5494425296783447,
      "learning_rate": 0.00018388968645274416,
      "loss": 1.4181,
      "step": 5800
    },
    {
      "epoch": 0.20774616362561998,
      "grad_norm": 1.2609368562698364,
      "learning_rate": 0.00018388337265693542,
      "loss": 1.474,
      "step": 5801
    },
    {
      "epoch": 0.20778197575518828,
      "grad_norm": 1.9432854652404785,
      "learning_rate": 0.0001838770577325841,
      "loss": 1.5919,
      "step": 5802
    },
    {
      "epoch": 0.20781778788475658,
      "grad_norm": 1.5743569135665894,
      "learning_rate": 0.00018387074167977517,
      "loss": 1.4351,
      "step": 5803
    },
    {
      "epoch": 0.20785360001432485,
      "grad_norm": 1.721057653427124,
      "learning_rate": 0.00018386442449859358,
      "loss": 1.538,
      "step": 5804
    },
    {
      "epoch": 0.20788941214389314,
      "grad_norm": 2.5399715900421143,
      "learning_rate": 0.00018385810618912435,
      "loss": 1.5953,
      "step": 5805
    },
    {
      "epoch": 0.2079252242734614,
      "grad_norm": 1.6732019186019897,
      "learning_rate": 0.00018385178675145246,
      "loss": 1.371,
      "step": 5806
    },
    {
      "epoch": 0.2079610364030297,
      "grad_norm": 1.5420591831207275,
      "learning_rate": 0.00018384546618566296,
      "loss": 1.3819,
      "step": 5807
    },
    {
      "epoch": 0.20799684853259798,
      "grad_norm": 2.665620803833008,
      "learning_rate": 0.00018383914449184084,
      "loss": 1.7292,
      "step": 5808
    },
    {
      "epoch": 0.20803266066216627,
      "grad_norm": 1.8281484842300415,
      "learning_rate": 0.0001838328216700712,
      "loss": 1.3194,
      "step": 5809
    },
    {
      "epoch": 0.20806847279173457,
      "grad_norm": 3.399212598800659,
      "learning_rate": 0.00018382649772043908,
      "loss": 1.9008,
      "step": 5810
    },
    {
      "epoch": 0.20810428492130284,
      "grad_norm": 1.844014286994934,
      "learning_rate": 0.00018382017264302955,
      "loss": 1.7792,
      "step": 5811
    },
    {
      "epoch": 0.20814009705087114,
      "grad_norm": 1.7047966718673706,
      "learning_rate": 0.0001838138464379277,
      "loss": 1.3427,
      "step": 5812
    },
    {
      "epoch": 0.2081759091804394,
      "grad_norm": 1.635097861289978,
      "learning_rate": 0.0001838075191052187,
      "loss": 1.8786,
      "step": 5813
    },
    {
      "epoch": 0.2082117213100077,
      "grad_norm": 1.8636630773544312,
      "learning_rate": 0.0001838011906449876,
      "loss": 1.6942,
      "step": 5814
    },
    {
      "epoch": 0.20824753343957597,
      "grad_norm": 1.397093653678894,
      "learning_rate": 0.0001837948610573196,
      "loss": 1.4005,
      "step": 5815
    },
    {
      "epoch": 0.20828334556914427,
      "grad_norm": 1.8946696519851685,
      "learning_rate": 0.0001837885303422998,
      "loss": 1.748,
      "step": 5816
    },
    {
      "epoch": 0.20831915769871256,
      "grad_norm": 1.4679687023162842,
      "learning_rate": 0.00018378219850001345,
      "loss": 1.4857,
      "step": 5817
    },
    {
      "epoch": 0.20835496982828083,
      "grad_norm": 1.5090714693069458,
      "learning_rate": 0.00018377586553054565,
      "loss": 1.7174,
      "step": 5818
    },
    {
      "epoch": 0.20839078195784913,
      "grad_norm": 1.7879695892333984,
      "learning_rate": 0.00018376953143398167,
      "loss": 1.6207,
      "step": 5819
    },
    {
      "epoch": 0.2084265940874174,
      "grad_norm": 2.3632588386535645,
      "learning_rate": 0.00018376319621040668,
      "loss": 1.4637,
      "step": 5820
    },
    {
      "epoch": 0.2084624062169857,
      "grad_norm": 2.151449680328369,
      "learning_rate": 0.00018375685985990594,
      "loss": 2.0277,
      "step": 5821
    },
    {
      "epoch": 0.20849821834655397,
      "grad_norm": 1.7777212858200073,
      "learning_rate": 0.00018375052238256466,
      "loss": 1.7095,
      "step": 5822
    },
    {
      "epoch": 0.20853403047612226,
      "grad_norm": 2.0421993732452393,
      "learning_rate": 0.00018374418377846817,
      "loss": 1.8641,
      "step": 5823
    },
    {
      "epoch": 0.20856984260569056,
      "grad_norm": 1.6348986625671387,
      "learning_rate": 0.0001837378440477017,
      "loss": 1.6932,
      "step": 5824
    },
    {
      "epoch": 0.20860565473525883,
      "grad_norm": 1.481870174407959,
      "learning_rate": 0.00018373150319035055,
      "loss": 1.6623,
      "step": 5825
    },
    {
      "epoch": 0.20864146686482712,
      "grad_norm": 1.5054895877838135,
      "learning_rate": 0.00018372516120650003,
      "loss": 1.579,
      "step": 5826
    },
    {
      "epoch": 0.2086772789943954,
      "grad_norm": 1.6486055850982666,
      "learning_rate": 0.00018371881809623545,
      "loss": 1.4551,
      "step": 5827
    },
    {
      "epoch": 0.2087130911239637,
      "grad_norm": 1.4928501844406128,
      "learning_rate": 0.0001837124738596422,
      "loss": 1.8089,
      "step": 5828
    },
    {
      "epoch": 0.20874890325353196,
      "grad_norm": 1.265228509902954,
      "learning_rate": 0.00018370612849680557,
      "loss": 1.6432,
      "step": 5829
    },
    {
      "epoch": 0.20878471538310026,
      "grad_norm": 1.9759631156921387,
      "learning_rate": 0.00018369978200781094,
      "loss": 1.6254,
      "step": 5830
    },
    {
      "epoch": 0.20882052751266855,
      "grad_norm": 1.7568730115890503,
      "learning_rate": 0.00018369343439274372,
      "loss": 1.8442,
      "step": 5831
    },
    {
      "epoch": 0.20885633964223682,
      "grad_norm": 1.8063440322875977,
      "learning_rate": 0.0001836870856516893,
      "loss": 1.4209,
      "step": 5832
    },
    {
      "epoch": 0.20889215177180512,
      "grad_norm": 1.3789457082748413,
      "learning_rate": 0.0001836807357847331,
      "loss": 1.4017,
      "step": 5833
    },
    {
      "epoch": 0.2089279639013734,
      "grad_norm": 1.9040539264678955,
      "learning_rate": 0.00018367438479196055,
      "loss": 1.7246,
      "step": 5834
    },
    {
      "epoch": 0.20896377603094168,
      "grad_norm": 2.3700289726257324,
      "learning_rate": 0.00018366803267345704,
      "loss": 1.5969,
      "step": 5835
    },
    {
      "epoch": 0.20899958816050995,
      "grad_norm": 1.2375210523605347,
      "learning_rate": 0.0001836616794293081,
      "loss": 1.5293,
      "step": 5836
    },
    {
      "epoch": 0.20903540029007825,
      "grad_norm": 1.7296556234359741,
      "learning_rate": 0.00018365532505959918,
      "loss": 1.5033,
      "step": 5837
    },
    {
      "epoch": 0.20907121241964655,
      "grad_norm": 1.7073261737823486,
      "learning_rate": 0.00018364896956441577,
      "loss": 1.6766,
      "step": 5838
    },
    {
      "epoch": 0.20910702454921481,
      "grad_norm": 1.5448671579360962,
      "learning_rate": 0.00018364261294384336,
      "loss": 1.5826,
      "step": 5839
    },
    {
      "epoch": 0.2091428366787831,
      "grad_norm": 1.3705118894577026,
      "learning_rate": 0.0001836362551979675,
      "loss": 1.5987,
      "step": 5840
    },
    {
      "epoch": 0.20917864880835138,
      "grad_norm": 1.785382866859436,
      "learning_rate": 0.00018362989632687374,
      "loss": 1.7167,
      "step": 5841
    },
    {
      "epoch": 0.20921446093791968,
      "grad_norm": 1.6626938581466675,
      "learning_rate": 0.00018362353633064754,
      "loss": 1.5908,
      "step": 5842
    },
    {
      "epoch": 0.20925027306748795,
      "grad_norm": 1.831031084060669,
      "learning_rate": 0.00018361717520937458,
      "loss": 1.8772,
      "step": 5843
    },
    {
      "epoch": 0.20928608519705624,
      "grad_norm": 1.589511752128601,
      "learning_rate": 0.00018361081296314037,
      "loss": 1.815,
      "step": 5844
    },
    {
      "epoch": 0.20932189732662454,
      "grad_norm": 1.4295203685760498,
      "learning_rate": 0.0001836044495920305,
      "loss": 1.4799,
      "step": 5845
    },
    {
      "epoch": 0.2093577094561928,
      "grad_norm": 1.112631916999817,
      "learning_rate": 0.00018359808509613062,
      "loss": 1.3803,
      "step": 5846
    },
    {
      "epoch": 0.2093935215857611,
      "grad_norm": 1.5145599842071533,
      "learning_rate": 0.00018359171947552631,
      "loss": 1.3917,
      "step": 5847
    },
    {
      "epoch": 0.20942933371532937,
      "grad_norm": 2.381051778793335,
      "learning_rate": 0.00018358535273030327,
      "loss": 1.5099,
      "step": 5848
    },
    {
      "epoch": 0.20946514584489767,
      "grad_norm": 1.7032874822616577,
      "learning_rate": 0.0001835789848605471,
      "loss": 1.3714,
      "step": 5849
    },
    {
      "epoch": 0.20950095797446594,
      "grad_norm": 1.7090039253234863,
      "learning_rate": 0.00018357261586634353,
      "loss": 1.5838,
      "step": 5850
    },
    {
      "epoch": 0.20953677010403424,
      "grad_norm": 1.4482029676437378,
      "learning_rate": 0.00018356624574777822,
      "loss": 1.5396,
      "step": 5851
    },
    {
      "epoch": 0.20957258223360253,
      "grad_norm": 1.815578579902649,
      "learning_rate": 0.0001835598745049368,
      "loss": 1.5823,
      "step": 5852
    },
    {
      "epoch": 0.2096083943631708,
      "grad_norm": 1.842309594154358,
      "learning_rate": 0.00018355350213790513,
      "loss": 1.5635,
      "step": 5853
    },
    {
      "epoch": 0.2096442064927391,
      "grad_norm": 1.4999361038208008,
      "learning_rate": 0.00018354712864676885,
      "loss": 1.5074,
      "step": 5854
    },
    {
      "epoch": 0.20968001862230737,
      "grad_norm": 1.7008081674575806,
      "learning_rate": 0.00018354075403161367,
      "loss": 1.369,
      "step": 5855
    },
    {
      "epoch": 0.20971583075187566,
      "grad_norm": 2.609595537185669,
      "learning_rate": 0.00018353437829252543,
      "loss": 1.3887,
      "step": 5856
    },
    {
      "epoch": 0.20975164288144393,
      "grad_norm": 1.640960454940796,
      "learning_rate": 0.00018352800142958992,
      "loss": 1.4278,
      "step": 5857
    },
    {
      "epoch": 0.20978745501101223,
      "grad_norm": 1.3580068349838257,
      "learning_rate": 0.00018352162344289284,
      "loss": 1.4415,
      "step": 5858
    },
    {
      "epoch": 0.20982326714058053,
      "grad_norm": 1.7644946575164795,
      "learning_rate": 0.0001835152443325201,
      "loss": 1.592,
      "step": 5859
    },
    {
      "epoch": 0.2098590792701488,
      "grad_norm": 1.3494611978530884,
      "learning_rate": 0.00018350886409855744,
      "loss": 1.6329,
      "step": 5860
    },
    {
      "epoch": 0.2098948913997171,
      "grad_norm": 1.8251352310180664,
      "learning_rate": 0.00018350248274109077,
      "loss": 1.8776,
      "step": 5861
    },
    {
      "epoch": 0.20993070352928536,
      "grad_norm": 1.4748811721801758,
      "learning_rate": 0.00018349610026020585,
      "loss": 1.4748,
      "step": 5862
    },
    {
      "epoch": 0.20996651565885366,
      "grad_norm": 1.604453206062317,
      "learning_rate": 0.00018348971665598865,
      "loss": 1.2366,
      "step": 5863
    },
    {
      "epoch": 0.21000232778842193,
      "grad_norm": 1.736939787864685,
      "learning_rate": 0.000183483331928525,
      "loss": 1.5625,
      "step": 5864
    },
    {
      "epoch": 0.21003813991799022,
      "grad_norm": 1.6238269805908203,
      "learning_rate": 0.00018347694607790077,
      "loss": 1.5814,
      "step": 5865
    },
    {
      "epoch": 0.21007395204755852,
      "grad_norm": 1.4518622159957886,
      "learning_rate": 0.00018347055910420193,
      "loss": 1.6499,
      "step": 5866
    },
    {
      "epoch": 0.2101097641771268,
      "grad_norm": 2.65487003326416,
      "learning_rate": 0.0001834641710075144,
      "loss": 1.417,
      "step": 5867
    },
    {
      "epoch": 0.2101455763066951,
      "grad_norm": 2.1541175842285156,
      "learning_rate": 0.0001834577817879241,
      "loss": 1.1873,
      "step": 5868
    },
    {
      "epoch": 0.21018138843626336,
      "grad_norm": 1.5349715948104858,
      "learning_rate": 0.000183451391445517,
      "loss": 1.6339,
      "step": 5869
    },
    {
      "epoch": 0.21021720056583165,
      "grad_norm": 1.4636268615722656,
      "learning_rate": 0.00018344499998037907,
      "loss": 1.4769,
      "step": 5870
    },
    {
      "epoch": 0.21025301269539992,
      "grad_norm": 1.8863812685012817,
      "learning_rate": 0.0001834386073925963,
      "loss": 1.6514,
      "step": 5871
    },
    {
      "epoch": 0.21028882482496822,
      "grad_norm": 1.8188995122909546,
      "learning_rate": 0.0001834322136822547,
      "loss": 1.2138,
      "step": 5872
    },
    {
      "epoch": 0.21032463695453651,
      "grad_norm": 1.6070584058761597,
      "learning_rate": 0.00018342581884944027,
      "loss": 1.5714,
      "step": 5873
    },
    {
      "epoch": 0.21036044908410478,
      "grad_norm": 1.322187900543213,
      "learning_rate": 0.0001834194228942391,
      "loss": 1.4738,
      "step": 5874
    },
    {
      "epoch": 0.21039626121367308,
      "grad_norm": 1.4101799726486206,
      "learning_rate": 0.00018341302581673715,
      "loss": 1.256,
      "step": 5875
    },
    {
      "epoch": 0.21043207334324135,
      "grad_norm": 2.0768353939056396,
      "learning_rate": 0.00018340662761702055,
      "loss": 1.7106,
      "step": 5876
    },
    {
      "epoch": 0.21046788547280965,
      "grad_norm": 1.6466964483261108,
      "learning_rate": 0.00018340022829517537,
      "loss": 1.7091,
      "step": 5877
    },
    {
      "epoch": 0.21050369760237791,
      "grad_norm": 2.0485565662384033,
      "learning_rate": 0.00018339382785128767,
      "loss": 1.7706,
      "step": 5878
    },
    {
      "epoch": 0.2105395097319462,
      "grad_norm": 1.6183522939682007,
      "learning_rate": 0.00018338742628544363,
      "loss": 1.6943,
      "step": 5879
    },
    {
      "epoch": 0.2105753218615145,
      "grad_norm": 1.507336974143982,
      "learning_rate": 0.0001833810235977293,
      "loss": 1.6523,
      "step": 5880
    },
    {
      "epoch": 0.21061113399108278,
      "grad_norm": 1.335508942604065,
      "learning_rate": 0.00018337461978823084,
      "loss": 1.5629,
      "step": 5881
    },
    {
      "epoch": 0.21064694612065107,
      "grad_norm": 1.6823132038116455,
      "learning_rate": 0.00018336821485703445,
      "loss": 1.3857,
      "step": 5882
    },
    {
      "epoch": 0.21068275825021934,
      "grad_norm": 1.4094301462173462,
      "learning_rate": 0.00018336180880422625,
      "loss": 1.7403,
      "step": 5883
    },
    {
      "epoch": 0.21071857037978764,
      "grad_norm": 2.0397469997406006,
      "learning_rate": 0.00018335540162989244,
      "loss": 1.7507,
      "step": 5884
    },
    {
      "epoch": 0.2107543825093559,
      "grad_norm": 1.7900669574737549,
      "learning_rate": 0.00018334899333411926,
      "loss": 1.6038,
      "step": 5885
    },
    {
      "epoch": 0.2107901946389242,
      "grad_norm": 1.5094026327133179,
      "learning_rate": 0.00018334258391699285,
      "loss": 1.65,
      "step": 5886
    },
    {
      "epoch": 0.2108260067684925,
      "grad_norm": 1.6741819381713867,
      "learning_rate": 0.00018333617337859946,
      "loss": 1.6157,
      "step": 5887
    },
    {
      "epoch": 0.21086181889806077,
      "grad_norm": 1.4904005527496338,
      "learning_rate": 0.00018332976171902537,
      "loss": 1.6095,
      "step": 5888
    },
    {
      "epoch": 0.21089763102762907,
      "grad_norm": 1.66598379611969,
      "learning_rate": 0.00018332334893835683,
      "loss": 1.3666,
      "step": 5889
    },
    {
      "epoch": 0.21093344315719734,
      "grad_norm": 1.658950686454773,
      "learning_rate": 0.00018331693503668013,
      "loss": 1.6743,
      "step": 5890
    },
    {
      "epoch": 0.21096925528676563,
      "grad_norm": 1.507813811302185,
      "learning_rate": 0.00018331052001408152,
      "loss": 1.3427,
      "step": 5891
    },
    {
      "epoch": 0.2110050674163339,
      "grad_norm": 1.7073756456375122,
      "learning_rate": 0.0001833041038706473,
      "loss": 1.5811,
      "step": 5892
    },
    {
      "epoch": 0.2110408795459022,
      "grad_norm": 1.5660537481307983,
      "learning_rate": 0.00018329768660646384,
      "loss": 1.5551,
      "step": 5893
    },
    {
      "epoch": 0.21107669167547047,
      "grad_norm": 1.3659303188323975,
      "learning_rate": 0.00018329126822161747,
      "loss": 1.469,
      "step": 5894
    },
    {
      "epoch": 0.21111250380503876,
      "grad_norm": 1.2605630159378052,
      "learning_rate": 0.0001832848487161945,
      "loss": 1.5376,
      "step": 5895
    },
    {
      "epoch": 0.21114831593460706,
      "grad_norm": 1.5032516717910767,
      "learning_rate": 0.00018327842809028134,
      "loss": 1.4326,
      "step": 5896
    },
    {
      "epoch": 0.21118412806417533,
      "grad_norm": 1.8132591247558594,
      "learning_rate": 0.00018327200634396434,
      "loss": 1.614,
      "step": 5897
    },
    {
      "epoch": 0.21121994019374363,
      "grad_norm": 1.4367175102233887,
      "learning_rate": 0.0001832655834773299,
      "loss": 1.4983,
      "step": 5898
    },
    {
      "epoch": 0.2112557523233119,
      "grad_norm": 1.4380050897598267,
      "learning_rate": 0.00018325915949046444,
      "loss": 1.3839,
      "step": 5899
    },
    {
      "epoch": 0.2112915644528802,
      "grad_norm": 1.1733310222625732,
      "learning_rate": 0.00018325273438345437,
      "loss": 1.582,
      "step": 5900
    },
    {
      "epoch": 0.21132737658244846,
      "grad_norm": 1.7513288259506226,
      "learning_rate": 0.0001832463081563862,
      "loss": 1.5806,
      "step": 5901
    },
    {
      "epoch": 0.21136318871201676,
      "grad_norm": 1.4104002714157104,
      "learning_rate": 0.00018323988080934628,
      "loss": 1.6676,
      "step": 5902
    },
    {
      "epoch": 0.21139900084158506,
      "grad_norm": 1.9451051950454712,
      "learning_rate": 0.00018323345234242118,
      "loss": 1.5106,
      "step": 5903
    },
    {
      "epoch": 0.21143481297115332,
      "grad_norm": 1.9381672143936157,
      "learning_rate": 0.0001832270227556973,
      "loss": 1.0622,
      "step": 5904
    },
    {
      "epoch": 0.21147062510072162,
      "grad_norm": 1.0977705717086792,
      "learning_rate": 0.0001832205920492612,
      "loss": 1.4634,
      "step": 5905
    },
    {
      "epoch": 0.2115064372302899,
      "grad_norm": 2.0630342960357666,
      "learning_rate": 0.0001832141602231994,
      "loss": 1.5579,
      "step": 5906
    },
    {
      "epoch": 0.2115422493598582,
      "grad_norm": 1.915184497833252,
      "learning_rate": 0.0001832077272775984,
      "loss": 1.8376,
      "step": 5907
    },
    {
      "epoch": 0.21157806148942646,
      "grad_norm": 2.1756672859191895,
      "learning_rate": 0.0001832012932125448,
      "loss": 1.6073,
      "step": 5908
    },
    {
      "epoch": 0.21161387361899475,
      "grad_norm": 1.440247654914856,
      "learning_rate": 0.00018319485802812503,
      "loss": 1.5361,
      "step": 5909
    },
    {
      "epoch": 0.21164968574856305,
      "grad_norm": 1.6484793424606323,
      "learning_rate": 0.00018318842172442582,
      "loss": 1.7188,
      "step": 5910
    },
    {
      "epoch": 0.21168549787813132,
      "grad_norm": 2.225233793258667,
      "learning_rate": 0.0001831819843015337,
      "loss": 1.5521,
      "step": 5911
    },
    {
      "epoch": 0.21172131000769961,
      "grad_norm": 1.6229103803634644,
      "learning_rate": 0.00018317554575953527,
      "loss": 1.4822,
      "step": 5912
    },
    {
      "epoch": 0.21175712213726788,
      "grad_norm": 2.2891218662261963,
      "learning_rate": 0.00018316910609851713,
      "loss": 1.3342,
      "step": 5913
    },
    {
      "epoch": 0.21179293426683618,
      "grad_norm": 1.2021150588989258,
      "learning_rate": 0.00018316266531856598,
      "loss": 1.6548,
      "step": 5914
    },
    {
      "epoch": 0.21182874639640445,
      "grad_norm": 1.3572343587875366,
      "learning_rate": 0.00018315622341976844,
      "loss": 1.6756,
      "step": 5915
    },
    {
      "epoch": 0.21186455852597275,
      "grad_norm": 1.7767938375473022,
      "learning_rate": 0.0001831497804022112,
      "loss": 1.398,
      "step": 5916
    },
    {
      "epoch": 0.21190037065554104,
      "grad_norm": 2.0373289585113525,
      "learning_rate": 0.00018314333626598089,
      "loss": 1.4118,
      "step": 5917
    },
    {
      "epoch": 0.2119361827851093,
      "grad_norm": 1.7908626794815063,
      "learning_rate": 0.0001831368910111642,
      "loss": 1.4941,
      "step": 5918
    },
    {
      "epoch": 0.2119719949146776,
      "grad_norm": 1.8968188762664795,
      "learning_rate": 0.00018313044463784793,
      "loss": 1.4038,
      "step": 5919
    },
    {
      "epoch": 0.21200780704424588,
      "grad_norm": 1.641762137413025,
      "learning_rate": 0.00018312399714611876,
      "loss": 1.649,
      "step": 5920
    },
    {
      "epoch": 0.21204361917381417,
      "grad_norm": 1.3661926984786987,
      "learning_rate": 0.00018311754853606344,
      "loss": 1.7253,
      "step": 5921
    },
    {
      "epoch": 0.21207943130338244,
      "grad_norm": 1.5163073539733887,
      "learning_rate": 0.00018311109880776868,
      "loss": 1.6616,
      "step": 5922
    },
    {
      "epoch": 0.21211524343295074,
      "grad_norm": 1.510380506515503,
      "learning_rate": 0.00018310464796132133,
      "loss": 1.5898,
      "step": 5923
    },
    {
      "epoch": 0.21215105556251904,
      "grad_norm": 1.931257963180542,
      "learning_rate": 0.0001830981959968081,
      "loss": 1.5899,
      "step": 5924
    },
    {
      "epoch": 0.2121868676920873,
      "grad_norm": 1.626753330230713,
      "learning_rate": 0.00018309174291431587,
      "loss": 1.5853,
      "step": 5925
    },
    {
      "epoch": 0.2122226798216556,
      "grad_norm": 1.351036548614502,
      "learning_rate": 0.00018308528871393138,
      "loss": 1.3402,
      "step": 5926
    },
    {
      "epoch": 0.21225849195122387,
      "grad_norm": 1.6709541082382202,
      "learning_rate": 0.00018307883339574153,
      "loss": 1.544,
      "step": 5927
    },
    {
      "epoch": 0.21229430408079217,
      "grad_norm": 3.2044754028320312,
      "learning_rate": 0.00018307237695983314,
      "loss": 1.5532,
      "step": 5928
    },
    {
      "epoch": 0.21233011621036044,
      "grad_norm": 1.6904844045639038,
      "learning_rate": 0.00018306591940629307,
      "loss": 1.4901,
      "step": 5929
    },
    {
      "epoch": 0.21236592833992873,
      "grad_norm": 1.53884756565094,
      "learning_rate": 0.00018305946073520822,
      "loss": 1.5167,
      "step": 5930
    },
    {
      "epoch": 0.21240174046949703,
      "grad_norm": 1.9038699865341187,
      "learning_rate": 0.00018305300094666543,
      "loss": 1.5806,
      "step": 5931
    },
    {
      "epoch": 0.2124375525990653,
      "grad_norm": 1.951897382736206,
      "learning_rate": 0.00018304654004075167,
      "loss": 1.7285,
      "step": 5932
    },
    {
      "epoch": 0.2124733647286336,
      "grad_norm": 1.620422124862671,
      "learning_rate": 0.0001830400780175538,
      "loss": 1.2822,
      "step": 5933
    },
    {
      "epoch": 0.21250917685820186,
      "grad_norm": 1.8646143674850464,
      "learning_rate": 0.00018303361487715883,
      "loss": 1.5526,
      "step": 5934
    },
    {
      "epoch": 0.21254498898777016,
      "grad_norm": 1.271231770515442,
      "learning_rate": 0.00018302715061965365,
      "loss": 1.9238,
      "step": 5935
    },
    {
      "epoch": 0.21258080111733843,
      "grad_norm": 1.5239760875701904,
      "learning_rate": 0.00018302068524512528,
      "loss": 1.327,
      "step": 5936
    },
    {
      "epoch": 0.21261661324690673,
      "grad_norm": 1.642659068107605,
      "learning_rate": 0.00018301421875366067,
      "loss": 1.6107,
      "step": 5937
    },
    {
      "epoch": 0.21265242537647502,
      "grad_norm": 1.5333079099655151,
      "learning_rate": 0.00018300775114534683,
      "loss": 1.6014,
      "step": 5938
    },
    {
      "epoch": 0.2126882375060433,
      "grad_norm": 1.676763892173767,
      "learning_rate": 0.00018300128242027078,
      "loss": 1.3634,
      "step": 5939
    },
    {
      "epoch": 0.2127240496356116,
      "grad_norm": 1.5230528116226196,
      "learning_rate": 0.00018299481257851952,
      "loss": 1.8291,
      "step": 5940
    },
    {
      "epoch": 0.21275986176517986,
      "grad_norm": 1.4611763954162598,
      "learning_rate": 0.00018298834162018012,
      "loss": 1.6594,
      "step": 5941
    },
    {
      "epoch": 0.21279567389474816,
      "grad_norm": 1.8587379455566406,
      "learning_rate": 0.00018298186954533962,
      "loss": 1.4631,
      "step": 5942
    },
    {
      "epoch": 0.21283148602431642,
      "grad_norm": 2.7046003341674805,
      "learning_rate": 0.00018297539635408512,
      "loss": 1.5129,
      "step": 5943
    },
    {
      "epoch": 0.21286729815388472,
      "grad_norm": 2.1789839267730713,
      "learning_rate": 0.00018296892204650367,
      "loss": 1.6845,
      "step": 5944
    },
    {
      "epoch": 0.21290311028345302,
      "grad_norm": 1.4094992876052856,
      "learning_rate": 0.00018296244662268241,
      "loss": 1.7135,
      "step": 5945
    },
    {
      "epoch": 0.2129389224130213,
      "grad_norm": 1.2647565603256226,
      "learning_rate": 0.00018295597008270847,
      "loss": 1.6716,
      "step": 5946
    },
    {
      "epoch": 0.21297473454258958,
      "grad_norm": 1.6723589897155762,
      "learning_rate": 0.00018294949242666895,
      "loss": 1.7836,
      "step": 5947
    },
    {
      "epoch": 0.21301054667215785,
      "grad_norm": 1.859749674797058,
      "learning_rate": 0.00018294301365465095,
      "loss": 1.6115,
      "step": 5948
    },
    {
      "epoch": 0.21304635880172615,
      "grad_norm": 1.2979680299758911,
      "learning_rate": 0.00018293653376674177,
      "loss": 1.3994,
      "step": 5949
    },
    {
      "epoch": 0.21308217093129442,
      "grad_norm": 1.4515433311462402,
      "learning_rate": 0.00018293005276302844,
      "loss": 1.3845,
      "step": 5950
    },
    {
      "epoch": 0.21311798306086271,
      "grad_norm": 1.7454895973205566,
      "learning_rate": 0.00018292357064359828,
      "loss": 1.7939,
      "step": 5951
    },
    {
      "epoch": 0.213153795190431,
      "grad_norm": 1.893728494644165,
      "learning_rate": 0.0001829170874085384,
      "loss": 1.5585,
      "step": 5952
    },
    {
      "epoch": 0.21318960731999928,
      "grad_norm": 1.5000876188278198,
      "learning_rate": 0.00018291060305793608,
      "loss": 1.524,
      "step": 5953
    },
    {
      "epoch": 0.21322541944956758,
      "grad_norm": 1.9762799739837646,
      "learning_rate": 0.00018290411759187855,
      "loss": 1.6497,
      "step": 5954
    },
    {
      "epoch": 0.21326123157913585,
      "grad_norm": 2.6302709579467773,
      "learning_rate": 0.00018289763101045302,
      "loss": 1.6087,
      "step": 5955
    },
    {
      "epoch": 0.21329704370870414,
      "grad_norm": 2.3435487747192383,
      "learning_rate": 0.00018289114331374685,
      "loss": 1.4392,
      "step": 5956
    },
    {
      "epoch": 0.2133328558382724,
      "grad_norm": 1.2770248651504517,
      "learning_rate": 0.00018288465450184722,
      "loss": 1.2857,
      "step": 5957
    },
    {
      "epoch": 0.2133686679678407,
      "grad_norm": 1.952136754989624,
      "learning_rate": 0.0001828781645748415,
      "loss": 1.5773,
      "step": 5958
    },
    {
      "epoch": 0.213404480097409,
      "grad_norm": 1.2606817483901978,
      "learning_rate": 0.00018287167353281698,
      "loss": 1.3636,
      "step": 5959
    },
    {
      "epoch": 0.21344029222697727,
      "grad_norm": 1.316763162612915,
      "learning_rate": 0.000182865181375861,
      "loss": 1.2342,
      "step": 5960
    },
    {
      "epoch": 0.21347610435654557,
      "grad_norm": 1.371982216835022,
      "learning_rate": 0.0001828586881040609,
      "loss": 1.6322,
      "step": 5961
    },
    {
      "epoch": 0.21351191648611384,
      "grad_norm": 1.4569306373596191,
      "learning_rate": 0.00018285219371750398,
      "loss": 1.389,
      "step": 5962
    },
    {
      "epoch": 0.21354772861568214,
      "grad_norm": 1.64267897605896,
      "learning_rate": 0.0001828456982162777,
      "loss": 1.734,
      "step": 5963
    },
    {
      "epoch": 0.2135835407452504,
      "grad_norm": 2.057509660720825,
      "learning_rate": 0.0001828392016004694,
      "loss": 1.8128,
      "step": 5964
    },
    {
      "epoch": 0.2136193528748187,
      "grad_norm": 1.5392036437988281,
      "learning_rate": 0.00018283270387016654,
      "loss": 1.5686,
      "step": 5965
    },
    {
      "epoch": 0.213655165004387,
      "grad_norm": 2.590017557144165,
      "learning_rate": 0.00018282620502545647,
      "loss": 1.8044,
      "step": 5966
    },
    {
      "epoch": 0.21369097713395527,
      "grad_norm": 1.446028709411621,
      "learning_rate": 0.00018281970506642663,
      "loss": 1.3209,
      "step": 5967
    },
    {
      "epoch": 0.21372678926352356,
      "grad_norm": 2.1150851249694824,
      "learning_rate": 0.0001828132039931645,
      "loss": 1.6727,
      "step": 5968
    },
    {
      "epoch": 0.21376260139309183,
      "grad_norm": 1.4321050643920898,
      "learning_rate": 0.00018280670180575754,
      "loss": 1.4389,
      "step": 5969
    },
    {
      "epoch": 0.21379841352266013,
      "grad_norm": 1.3660449981689453,
      "learning_rate": 0.00018280019850429321,
      "loss": 1.4298,
      "step": 5970
    },
    {
      "epoch": 0.2138342256522284,
      "grad_norm": 1.6467535495758057,
      "learning_rate": 0.000182793694088859,
      "loss": 1.4494,
      "step": 5971
    },
    {
      "epoch": 0.2138700377817967,
      "grad_norm": 2.1352715492248535,
      "learning_rate": 0.00018278718855954247,
      "loss": 1.7657,
      "step": 5972
    },
    {
      "epoch": 0.213905849911365,
      "grad_norm": 2.0386393070220947,
      "learning_rate": 0.00018278068191643107,
      "loss": 1.709,
      "step": 5973
    },
    {
      "epoch": 0.21394166204093326,
      "grad_norm": 1.3228604793548584,
      "learning_rate": 0.0001827741741596124,
      "loss": 1.4663,
      "step": 5974
    },
    {
      "epoch": 0.21397747417050156,
      "grad_norm": 2.2697250843048096,
      "learning_rate": 0.00018276766528917398,
      "loss": 1.8581,
      "step": 5975
    },
    {
      "epoch": 0.21401328630006983,
      "grad_norm": 1.8088141679763794,
      "learning_rate": 0.00018276115530520336,
      "loss": 1.7364,
      "step": 5976
    },
    {
      "epoch": 0.21404909842963812,
      "grad_norm": 1.4815630912780762,
      "learning_rate": 0.0001827546442077882,
      "loss": 1.3892,
      "step": 5977
    },
    {
      "epoch": 0.2140849105592064,
      "grad_norm": 1.7200106382369995,
      "learning_rate": 0.000182748131997016,
      "loss": 1.4714,
      "step": 5978
    },
    {
      "epoch": 0.2141207226887747,
      "grad_norm": 1.388688087463379,
      "learning_rate": 0.00018274161867297447,
      "loss": 1.7146,
      "step": 5979
    },
    {
      "epoch": 0.214156534818343,
      "grad_norm": 1.3790088891983032,
      "learning_rate": 0.00018273510423575117,
      "loss": 1.5785,
      "step": 5980
    },
    {
      "epoch": 0.21419234694791126,
      "grad_norm": 1.8050322532653809,
      "learning_rate": 0.00018272858868543374,
      "loss": 1.3374,
      "step": 5981
    },
    {
      "epoch": 0.21422815907747955,
      "grad_norm": 1.851999282836914,
      "learning_rate": 0.00018272207202210986,
      "loss": 1.4933,
      "step": 5982
    },
    {
      "epoch": 0.21426397120704782,
      "grad_norm": 1.7498927116394043,
      "learning_rate": 0.00018271555424586723,
      "loss": 1.8221,
      "step": 5983
    },
    {
      "epoch": 0.21429978333661612,
      "grad_norm": 1.9699684381484985,
      "learning_rate": 0.0001827090353567935,
      "loss": 1.563,
      "step": 5984
    },
    {
      "epoch": 0.2143355954661844,
      "grad_norm": 1.985451102256775,
      "learning_rate": 0.0001827025153549764,
      "loss": 1.8659,
      "step": 5985
    },
    {
      "epoch": 0.21437140759575268,
      "grad_norm": 1.738781213760376,
      "learning_rate": 0.00018269599424050362,
      "loss": 1.582,
      "step": 5986
    },
    {
      "epoch": 0.21440721972532098,
      "grad_norm": 1.600334644317627,
      "learning_rate": 0.00018268947201346291,
      "loss": 1.6137,
      "step": 5987
    },
    {
      "epoch": 0.21444303185488925,
      "grad_norm": 1.34315824508667,
      "learning_rate": 0.00018268294867394204,
      "loss": 1.4509,
      "step": 5988
    },
    {
      "epoch": 0.21447884398445755,
      "grad_norm": 2.184556007385254,
      "learning_rate": 0.00018267642422202873,
      "loss": 1.7843,
      "step": 5989
    },
    {
      "epoch": 0.21451465611402581,
      "grad_norm": 1.9627654552459717,
      "learning_rate": 0.00018266989865781076,
      "loss": 1.7417,
      "step": 5990
    },
    {
      "epoch": 0.2145504682435941,
      "grad_norm": 1.7299060821533203,
      "learning_rate": 0.00018266337198137594,
      "loss": 1.652,
      "step": 5991
    },
    {
      "epoch": 0.21458628037316238,
      "grad_norm": 1.5557527542114258,
      "learning_rate": 0.00018265684419281213,
      "loss": 1.6432,
      "step": 5992
    },
    {
      "epoch": 0.21462209250273068,
      "grad_norm": 1.7059578895568848,
      "learning_rate": 0.00018265031529220705,
      "loss": 1.456,
      "step": 5993
    },
    {
      "epoch": 0.21465790463229895,
      "grad_norm": 1.6850507259368896,
      "learning_rate": 0.0001826437852796486,
      "loss": 1.5577,
      "step": 5994
    },
    {
      "epoch": 0.21469371676186724,
      "grad_norm": 2.0986411571502686,
      "learning_rate": 0.00018263725415522462,
      "loss": 1.5896,
      "step": 5995
    },
    {
      "epoch": 0.21472952889143554,
      "grad_norm": 2.174347400665283,
      "learning_rate": 0.000182630721919023,
      "loss": 1.6033,
      "step": 5996
    },
    {
      "epoch": 0.2147653410210038,
      "grad_norm": 1.433610439300537,
      "learning_rate": 0.00018262418857113157,
      "loss": 1.6622,
      "step": 5997
    },
    {
      "epoch": 0.2148011531505721,
      "grad_norm": 1.3677752017974854,
      "learning_rate": 0.00018261765411163827,
      "loss": 1.6868,
      "step": 5998
    },
    {
      "epoch": 0.21483696528014037,
      "grad_norm": 1.4604482650756836,
      "learning_rate": 0.000182611118540631,
      "loss": 1.3887,
      "step": 5999
    },
    {
      "epoch": 0.21487277740970867,
      "grad_norm": 1.462753415107727,
      "learning_rate": 0.00018260458185819772,
      "loss": 1.5631,
      "step": 6000
    },
    {
      "epoch": 0.21490858953927694,
      "grad_norm": 1.279866099357605,
      "learning_rate": 0.00018259804406442633,
      "loss": 1.3168,
      "step": 6001
    },
    {
      "epoch": 0.21494440166884524,
      "grad_norm": 2.1085100173950195,
      "learning_rate": 0.0001825915051594048,
      "loss": 1.8449,
      "step": 6002
    },
    {
      "epoch": 0.21498021379841353,
      "grad_norm": 1.6899045705795288,
      "learning_rate": 0.0001825849651432211,
      "loss": 1.4615,
      "step": 6003
    },
    {
      "epoch": 0.2150160259279818,
      "grad_norm": 1.6363755464553833,
      "learning_rate": 0.0001825784240159632,
      "loss": 1.4947,
      "step": 6004
    },
    {
      "epoch": 0.2150518380575501,
      "grad_norm": 1.7822414636611938,
      "learning_rate": 0.00018257188177771914,
      "loss": 1.579,
      "step": 6005
    },
    {
      "epoch": 0.21508765018711837,
      "grad_norm": 2.806736469268799,
      "learning_rate": 0.00018256533842857695,
      "loss": 1.4047,
      "step": 6006
    },
    {
      "epoch": 0.21512346231668666,
      "grad_norm": 1.5889925956726074,
      "learning_rate": 0.0001825587939686246,
      "loss": 1.563,
      "step": 6007
    },
    {
      "epoch": 0.21515927444625493,
      "grad_norm": 1.2758655548095703,
      "learning_rate": 0.00018255224839795018,
      "loss": 1.6386,
      "step": 6008
    },
    {
      "epoch": 0.21519508657582323,
      "grad_norm": 1.890386939048767,
      "learning_rate": 0.00018254570171664174,
      "loss": 1.4298,
      "step": 6009
    },
    {
      "epoch": 0.21523089870539153,
      "grad_norm": 1.240126609802246,
      "learning_rate": 0.00018253915392478737,
      "loss": 1.4986,
      "step": 6010
    },
    {
      "epoch": 0.2152667108349598,
      "grad_norm": 2.025054454803467,
      "learning_rate": 0.00018253260502247513,
      "loss": 1.7933,
      "step": 6011
    },
    {
      "epoch": 0.2153025229645281,
      "grad_norm": 1.7431395053863525,
      "learning_rate": 0.00018252605500979316,
      "loss": 1.8384,
      "step": 6012
    },
    {
      "epoch": 0.21533833509409636,
      "grad_norm": 1.3581688404083252,
      "learning_rate": 0.00018251950388682958,
      "loss": 1.4347,
      "step": 6013
    },
    {
      "epoch": 0.21537414722366466,
      "grad_norm": 1.4468944072723389,
      "learning_rate": 0.0001825129516536725,
      "loss": 1.5021,
      "step": 6014
    },
    {
      "epoch": 0.21540995935323293,
      "grad_norm": 2.2720768451690674,
      "learning_rate": 0.0001825063983104101,
      "loss": 1.7324,
      "step": 6015
    },
    {
      "epoch": 0.21544577148280122,
      "grad_norm": 2.472119092941284,
      "learning_rate": 0.00018249984385713055,
      "loss": 1.905,
      "step": 6016
    },
    {
      "epoch": 0.21548158361236952,
      "grad_norm": 1.8467071056365967,
      "learning_rate": 0.000182493288293922,
      "loss": 1.4867,
      "step": 6017
    },
    {
      "epoch": 0.2155173957419378,
      "grad_norm": 1.5712488889694214,
      "learning_rate": 0.00018248673162087268,
      "loss": 1.5598,
      "step": 6018
    },
    {
      "epoch": 0.2155532078715061,
      "grad_norm": 1.773179054260254,
      "learning_rate": 0.00018248017383807076,
      "loss": 1.539,
      "step": 6019
    },
    {
      "epoch": 0.21558902000107436,
      "grad_norm": 1.6343525648117065,
      "learning_rate": 0.0001824736149456045,
      "loss": 1.6218,
      "step": 6020
    },
    {
      "epoch": 0.21562483213064265,
      "grad_norm": 1.5364235639572144,
      "learning_rate": 0.00018246705494356214,
      "loss": 1.3762,
      "step": 6021
    },
    {
      "epoch": 0.21566064426021092,
      "grad_norm": 1.940355896949768,
      "learning_rate": 0.00018246049383203192,
      "loss": 1.4093,
      "step": 6022
    },
    {
      "epoch": 0.21569645638977922,
      "grad_norm": 1.5513060092926025,
      "learning_rate": 0.00018245393161110215,
      "loss": 1.0682,
      "step": 6023
    },
    {
      "epoch": 0.21573226851934751,
      "grad_norm": 1.6082549095153809,
      "learning_rate": 0.00018244736828086107,
      "loss": 1.2319,
      "step": 6024
    },
    {
      "epoch": 0.21576808064891578,
      "grad_norm": 1.8053920269012451,
      "learning_rate": 0.00018244080384139698,
      "loss": 1.7213,
      "step": 6025
    },
    {
      "epoch": 0.21580389277848408,
      "grad_norm": 1.4318716526031494,
      "learning_rate": 0.00018243423829279824,
      "loss": 1.5302,
      "step": 6026
    },
    {
      "epoch": 0.21583970490805235,
      "grad_norm": 1.3476557731628418,
      "learning_rate": 0.00018242767163515318,
      "loss": 1.3816,
      "step": 6027
    },
    {
      "epoch": 0.21587551703762065,
      "grad_norm": 2.339296817779541,
      "learning_rate": 0.00018242110386855007,
      "loss": 1.2601,
      "step": 6028
    },
    {
      "epoch": 0.21591132916718891,
      "grad_norm": 1.4866342544555664,
      "learning_rate": 0.00018241453499307734,
      "loss": 1.4542,
      "step": 6029
    },
    {
      "epoch": 0.2159471412967572,
      "grad_norm": 2.2102510929107666,
      "learning_rate": 0.00018240796500882338,
      "loss": 1.8524,
      "step": 6030
    },
    {
      "epoch": 0.2159829534263255,
      "grad_norm": 1.8730422258377075,
      "learning_rate": 0.0001824013939158765,
      "loss": 1.7469,
      "step": 6031
    },
    {
      "epoch": 0.21601876555589378,
      "grad_norm": 1.2295218706130981,
      "learning_rate": 0.0001823948217143252,
      "loss": 1.5667,
      "step": 6032
    },
    {
      "epoch": 0.21605457768546207,
      "grad_norm": 1.4555342197418213,
      "learning_rate": 0.00018238824840425785,
      "loss": 1.3013,
      "step": 6033
    },
    {
      "epoch": 0.21609038981503034,
      "grad_norm": 2.010502576828003,
      "learning_rate": 0.00018238167398576286,
      "loss": 1.5193,
      "step": 6034
    },
    {
      "epoch": 0.21612620194459864,
      "grad_norm": 2.1535098552703857,
      "learning_rate": 0.00018237509845892873,
      "loss": 1.6717,
      "step": 6035
    },
    {
      "epoch": 0.2161620140741669,
      "grad_norm": 1.6601369380950928,
      "learning_rate": 0.00018236852182384393,
      "loss": 1.5295,
      "step": 6036
    },
    {
      "epoch": 0.2161978262037352,
      "grad_norm": 1.641852617263794,
      "learning_rate": 0.00018236194408059685,
      "loss": 1.1745,
      "step": 6037
    },
    {
      "epoch": 0.2162336383333035,
      "grad_norm": 1.5132910013198853,
      "learning_rate": 0.00018235536522927611,
      "loss": 1.6211,
      "step": 6038
    },
    {
      "epoch": 0.21626945046287177,
      "grad_norm": 1.4342702627182007,
      "learning_rate": 0.00018234878526997015,
      "loss": 1.371,
      "step": 6039
    },
    {
      "epoch": 0.21630526259244007,
      "grad_norm": 1.345091462135315,
      "learning_rate": 0.0001823422042027675,
      "loss": 1.5545,
      "step": 6040
    },
    {
      "epoch": 0.21634107472200834,
      "grad_norm": 2.147703170776367,
      "learning_rate": 0.0001823356220277567,
      "loss": 1.5573,
      "step": 6041
    },
    {
      "epoch": 0.21637688685157663,
      "grad_norm": 1.943217396736145,
      "learning_rate": 0.00018232903874502632,
      "loss": 1.3352,
      "step": 6042
    },
    {
      "epoch": 0.2164126989811449,
      "grad_norm": 1.6899324655532837,
      "learning_rate": 0.00018232245435466493,
      "loss": 1.4203,
      "step": 6043
    },
    {
      "epoch": 0.2164485111107132,
      "grad_norm": 1.9825857877731323,
      "learning_rate": 0.0001823158688567611,
      "loss": 1.591,
      "step": 6044
    },
    {
      "epoch": 0.2164843232402815,
      "grad_norm": 1.5396236181259155,
      "learning_rate": 0.00018230928225140342,
      "loss": 1.8448,
      "step": 6045
    },
    {
      "epoch": 0.21652013536984976,
      "grad_norm": 1.9984638690948486,
      "learning_rate": 0.00018230269453868052,
      "loss": 1.417,
      "step": 6046
    },
    {
      "epoch": 0.21655594749941806,
      "grad_norm": 1.8226370811462402,
      "learning_rate": 0.00018229610571868102,
      "loss": 2.0727,
      "step": 6047
    },
    {
      "epoch": 0.21659175962898633,
      "grad_norm": 1.4168521165847778,
      "learning_rate": 0.0001822895157914936,
      "loss": 1.7747,
      "step": 6048
    },
    {
      "epoch": 0.21662757175855463,
      "grad_norm": 1.5113242864608765,
      "learning_rate": 0.00018228292475720687,
      "loss": 1.7142,
      "step": 6049
    },
    {
      "epoch": 0.2166633838881229,
      "grad_norm": 1.5562756061553955,
      "learning_rate": 0.00018227633261590955,
      "loss": 1.6209,
      "step": 6050
    },
    {
      "epoch": 0.2166991960176912,
      "grad_norm": 1.5761768817901611,
      "learning_rate": 0.00018226973936769027,
      "loss": 1.6547,
      "step": 6051
    },
    {
      "epoch": 0.2167350081472595,
      "grad_norm": 1.78731369972229,
      "learning_rate": 0.0001822631450126378,
      "loss": 1.3805,
      "step": 6052
    },
    {
      "epoch": 0.21677082027682776,
      "grad_norm": 1.4102073907852173,
      "learning_rate": 0.00018225654955084079,
      "loss": 1.562,
      "step": 6053
    },
    {
      "epoch": 0.21680663240639605,
      "grad_norm": 1.5794233083724976,
      "learning_rate": 0.00018224995298238804,
      "loss": 1.6456,
      "step": 6054
    },
    {
      "epoch": 0.21684244453596432,
      "grad_norm": 1.8132075071334839,
      "learning_rate": 0.00018224335530736825,
      "loss": 1.9961,
      "step": 6055
    },
    {
      "epoch": 0.21687825666553262,
      "grad_norm": 1.6012133359909058,
      "learning_rate": 0.0001822367565258702,
      "loss": 1.7669,
      "step": 6056
    },
    {
      "epoch": 0.2169140687951009,
      "grad_norm": 1.3363044261932373,
      "learning_rate": 0.0001822301566379827,
      "loss": 1.6712,
      "step": 6057
    },
    {
      "epoch": 0.2169498809246692,
      "grad_norm": 1.529463291168213,
      "learning_rate": 0.00018222355564379448,
      "loss": 1.3612,
      "step": 6058
    },
    {
      "epoch": 0.21698569305423748,
      "grad_norm": 1.4082953929901123,
      "learning_rate": 0.00018221695354339435,
      "loss": 1.4312,
      "step": 6059
    },
    {
      "epoch": 0.21702150518380575,
      "grad_norm": 1.7950927019119263,
      "learning_rate": 0.00018221035033687123,
      "loss": 1.5232,
      "step": 6060
    },
    {
      "epoch": 0.21705731731337405,
      "grad_norm": 1.8718847036361694,
      "learning_rate": 0.00018220374602431386,
      "loss": 1.7111,
      "step": 6061
    },
    {
      "epoch": 0.21709312944294232,
      "grad_norm": 1.4784345626831055,
      "learning_rate": 0.0001821971406058111,
      "loss": 1.6418,
      "step": 6062
    },
    {
      "epoch": 0.21712894157251061,
      "grad_norm": 1.6133242845535278,
      "learning_rate": 0.00018219053408145185,
      "loss": 1.3773,
      "step": 6063
    },
    {
      "epoch": 0.21716475370207888,
      "grad_norm": 1.6696078777313232,
      "learning_rate": 0.000182183926451325,
      "loss": 1.6862,
      "step": 6064
    },
    {
      "epoch": 0.21720056583164718,
      "grad_norm": 1.7715810537338257,
      "learning_rate": 0.00018217731771551942,
      "loss": 1.734,
      "step": 6065
    },
    {
      "epoch": 0.21723637796121548,
      "grad_norm": 1.6767003536224365,
      "learning_rate": 0.00018217070787412404,
      "loss": 1.4907,
      "step": 6066
    },
    {
      "epoch": 0.21727219009078375,
      "grad_norm": 1.7085485458374023,
      "learning_rate": 0.00018216409692722779,
      "loss": 1.8883,
      "step": 6067
    },
    {
      "epoch": 0.21730800222035204,
      "grad_norm": 1.3934434652328491,
      "learning_rate": 0.00018215748487491958,
      "loss": 1.3403,
      "step": 6068
    },
    {
      "epoch": 0.2173438143499203,
      "grad_norm": 2.090029001235962,
      "learning_rate": 0.00018215087171728837,
      "loss": 1.7042,
      "step": 6069
    },
    {
      "epoch": 0.2173796264794886,
      "grad_norm": 2.287304639816284,
      "learning_rate": 0.00018214425745442317,
      "loss": 1.7766,
      "step": 6070
    },
    {
      "epoch": 0.21741543860905688,
      "grad_norm": 1.7092816829681396,
      "learning_rate": 0.00018213764208641292,
      "loss": 1.4644,
      "step": 6071
    },
    {
      "epoch": 0.21745125073862517,
      "grad_norm": 1.7535308599472046,
      "learning_rate": 0.00018213102561334668,
      "loss": 1.39,
      "step": 6072
    },
    {
      "epoch": 0.21748706286819347,
      "grad_norm": 2.193551778793335,
      "learning_rate": 0.00018212440803531342,
      "loss": 1.5143,
      "step": 6073
    },
    {
      "epoch": 0.21752287499776174,
      "grad_norm": 1.7536066770553589,
      "learning_rate": 0.00018211778935240219,
      "loss": 1.617,
      "step": 6074
    },
    {
      "epoch": 0.21755868712733004,
      "grad_norm": 1.5110563039779663,
      "learning_rate": 0.00018211116956470203,
      "loss": 1.6147,
      "step": 6075
    },
    {
      "epoch": 0.2175944992568983,
      "grad_norm": 1.4865481853485107,
      "learning_rate": 0.00018210454867230195,
      "loss": 1.6124,
      "step": 6076
    },
    {
      "epoch": 0.2176303113864666,
      "grad_norm": 1.9881222248077393,
      "learning_rate": 0.00018209792667529112,
      "loss": 1.5487,
      "step": 6077
    },
    {
      "epoch": 0.21766612351603487,
      "grad_norm": 1.733006477355957,
      "learning_rate": 0.00018209130357375858,
      "loss": 1.7815,
      "step": 6078
    },
    {
      "epoch": 0.21770193564560317,
      "grad_norm": 1.2904987335205078,
      "learning_rate": 0.00018208467936779347,
      "loss": 1.8821,
      "step": 6079
    },
    {
      "epoch": 0.21773774777517146,
      "grad_norm": 2.602548837661743,
      "learning_rate": 0.00018207805405748482,
      "loss": 1.4122,
      "step": 6080
    },
    {
      "epoch": 0.21777355990473973,
      "grad_norm": 1.7389355897903442,
      "learning_rate": 0.00018207142764292187,
      "loss": 1.4373,
      "step": 6081
    },
    {
      "epoch": 0.21780937203430803,
      "grad_norm": 1.316788911819458,
      "learning_rate": 0.00018206480012419372,
      "loss": 1.6428,
      "step": 6082
    },
    {
      "epoch": 0.2178451841638763,
      "grad_norm": 1.8363261222839355,
      "learning_rate": 0.0001820581715013895,
      "loss": 1.4394,
      "step": 6083
    },
    {
      "epoch": 0.2178809962934446,
      "grad_norm": 1.57510507106781,
      "learning_rate": 0.0001820515417745985,
      "loss": 1.4141,
      "step": 6084
    },
    {
      "epoch": 0.21791680842301286,
      "grad_norm": 1.3749953508377075,
      "learning_rate": 0.0001820449109439098,
      "loss": 1.5511,
      "step": 6085
    },
    {
      "epoch": 0.21795262055258116,
      "grad_norm": 1.5707284212112427,
      "learning_rate": 0.00018203827900941264,
      "loss": 1.7035,
      "step": 6086
    },
    {
      "epoch": 0.21798843268214946,
      "grad_norm": 1.7157187461853027,
      "learning_rate": 0.0001820316459711963,
      "loss": 1.599,
      "step": 6087
    },
    {
      "epoch": 0.21802424481171773,
      "grad_norm": 1.5144253969192505,
      "learning_rate": 0.0001820250118293499,
      "loss": 1.9309,
      "step": 6088
    },
    {
      "epoch": 0.21806005694128602,
      "grad_norm": 1.6175367832183838,
      "learning_rate": 0.00018201837658396287,
      "loss": 1.6032,
      "step": 6089
    },
    {
      "epoch": 0.2180958690708543,
      "grad_norm": 1.6035107374191284,
      "learning_rate": 0.00018201174023512433,
      "loss": 1.4334,
      "step": 6090
    },
    {
      "epoch": 0.2181316812004226,
      "grad_norm": 1.8502047061920166,
      "learning_rate": 0.0001820051027829236,
      "loss": 1.354,
      "step": 6091
    },
    {
      "epoch": 0.21816749332999086,
      "grad_norm": 1.380337119102478,
      "learning_rate": 0.00018199846422745002,
      "loss": 1.6663,
      "step": 6092
    },
    {
      "epoch": 0.21820330545955915,
      "grad_norm": 1.2796369791030884,
      "learning_rate": 0.00018199182456879286,
      "loss": 1.5156,
      "step": 6093
    },
    {
      "epoch": 0.21823911758912742,
      "grad_norm": 1.8583347797393799,
      "learning_rate": 0.00018198518380704143,
      "loss": 1.896,
      "step": 6094
    },
    {
      "epoch": 0.21827492971869572,
      "grad_norm": 1.7820069789886475,
      "learning_rate": 0.00018197854194228517,
      "loss": 1.8372,
      "step": 6095
    },
    {
      "epoch": 0.21831074184826402,
      "grad_norm": 2.359118938446045,
      "learning_rate": 0.00018197189897461332,
      "loss": 1.5367,
      "step": 6096
    },
    {
      "epoch": 0.2183465539778323,
      "grad_norm": 1.677788496017456,
      "learning_rate": 0.00018196525490411534,
      "loss": 1.5914,
      "step": 6097
    },
    {
      "epoch": 0.21838236610740058,
      "grad_norm": 1.7194525003433228,
      "learning_rate": 0.00018195860973088058,
      "loss": 1.6376,
      "step": 6098
    },
    {
      "epoch": 0.21841817823696885,
      "grad_norm": 1.8285008668899536,
      "learning_rate": 0.00018195196345499842,
      "loss": 1.8412,
      "step": 6099
    },
    {
      "epoch": 0.21845399036653715,
      "grad_norm": 2.399778366088867,
      "learning_rate": 0.00018194531607655833,
      "loss": 1.3773,
      "step": 6100
    },
    {
      "epoch": 0.21848980249610542,
      "grad_norm": 1.259798288345337,
      "learning_rate": 0.0001819386675956497,
      "loss": 1.3212,
      "step": 6101
    },
    {
      "epoch": 0.21852561462567371,
      "grad_norm": 1.758084774017334,
      "learning_rate": 0.000181932018012362,
      "loss": 1.6877,
      "step": 6102
    },
    {
      "epoch": 0.218561426755242,
      "grad_norm": 1.4051774740219116,
      "learning_rate": 0.00018192536732678468,
      "loss": 1.4873,
      "step": 6103
    },
    {
      "epoch": 0.21859723888481028,
      "grad_norm": 1.8246309757232666,
      "learning_rate": 0.00018191871553900718,
      "loss": 1.7631,
      "step": 6104
    },
    {
      "epoch": 0.21863305101437858,
      "grad_norm": 1.6398242712020874,
      "learning_rate": 0.00018191206264911908,
      "loss": 1.3941,
      "step": 6105
    },
    {
      "epoch": 0.21866886314394685,
      "grad_norm": 2.2544310092926025,
      "learning_rate": 0.0001819054086572098,
      "loss": 1.7397,
      "step": 6106
    },
    {
      "epoch": 0.21870467527351514,
      "grad_norm": 1.5466108322143555,
      "learning_rate": 0.00018189875356336893,
      "loss": 1.5326,
      "step": 6107
    },
    {
      "epoch": 0.2187404874030834,
      "grad_norm": 1.4698082208633423,
      "learning_rate": 0.00018189209736768595,
      "loss": 1.4511,
      "step": 6108
    },
    {
      "epoch": 0.2187762995326517,
      "grad_norm": 1.9529378414154053,
      "learning_rate": 0.00018188544007025043,
      "loss": 1.8635,
      "step": 6109
    },
    {
      "epoch": 0.21881211166222,
      "grad_norm": 1.4893909692764282,
      "learning_rate": 0.00018187878167115197,
      "loss": 1.6932,
      "step": 6110
    },
    {
      "epoch": 0.21884792379178827,
      "grad_norm": 1.3069435358047485,
      "learning_rate": 0.00018187212217048008,
      "loss": 1.5887,
      "step": 6111
    },
    {
      "epoch": 0.21888373592135657,
      "grad_norm": 1.8581095933914185,
      "learning_rate": 0.00018186546156832444,
      "loss": 1.7178,
      "step": 6112
    },
    {
      "epoch": 0.21891954805092484,
      "grad_norm": 1.3659560680389404,
      "learning_rate": 0.00018185879986477456,
      "loss": 1.2872,
      "step": 6113
    },
    {
      "epoch": 0.21895536018049314,
      "grad_norm": 1.5570614337921143,
      "learning_rate": 0.00018185213705992014,
      "loss": 1.5665,
      "step": 6114
    },
    {
      "epoch": 0.2189911723100614,
      "grad_norm": 1.6068415641784668,
      "learning_rate": 0.00018184547315385082,
      "loss": 1.869,
      "step": 6115
    },
    {
      "epoch": 0.2190269844396297,
      "grad_norm": 1.2946780920028687,
      "learning_rate": 0.0001818388081466562,
      "loss": 1.4014,
      "step": 6116
    },
    {
      "epoch": 0.219062796569198,
      "grad_norm": 1.327081322669983,
      "learning_rate": 0.00018183214203842601,
      "loss": 1.8673,
      "step": 6117
    },
    {
      "epoch": 0.21909860869876627,
      "grad_norm": 1.738240122795105,
      "learning_rate": 0.00018182547482924988,
      "loss": 1.406,
      "step": 6118
    },
    {
      "epoch": 0.21913442082833456,
      "grad_norm": 2.0041754245758057,
      "learning_rate": 0.00018181880651921755,
      "loss": 1.6267,
      "step": 6119
    },
    {
      "epoch": 0.21917023295790283,
      "grad_norm": 1.6936300992965698,
      "learning_rate": 0.0001818121371084187,
      "loss": 1.741,
      "step": 6120
    },
    {
      "epoch": 0.21920604508747113,
      "grad_norm": 1.2888492345809937,
      "learning_rate": 0.00018180546659694307,
      "loss": 1.5933,
      "step": 6121
    },
    {
      "epoch": 0.2192418572170394,
      "grad_norm": 1.3462530374526978,
      "learning_rate": 0.0001817987949848804,
      "loss": 1.6054,
      "step": 6122
    },
    {
      "epoch": 0.2192776693466077,
      "grad_norm": 2.644855260848999,
      "learning_rate": 0.0001817921222723205,
      "loss": 1.7208,
      "step": 6123
    },
    {
      "epoch": 0.219313481476176,
      "grad_norm": 1.3023197650909424,
      "learning_rate": 0.00018178544845935308,
      "loss": 1.6019,
      "step": 6124
    },
    {
      "epoch": 0.21934929360574426,
      "grad_norm": 1.9801610708236694,
      "learning_rate": 0.00018177877354606797,
      "loss": 1.4413,
      "step": 6125
    },
    {
      "epoch": 0.21938510573531256,
      "grad_norm": 1.6524903774261475,
      "learning_rate": 0.00018177209753255492,
      "loss": 1.4086,
      "step": 6126
    },
    {
      "epoch": 0.21942091786488083,
      "grad_norm": 1.5688244104385376,
      "learning_rate": 0.00018176542041890376,
      "loss": 1.5965,
      "step": 6127
    },
    {
      "epoch": 0.21945672999444912,
      "grad_norm": 1.5412373542785645,
      "learning_rate": 0.00018175874220520438,
      "loss": 1.4583,
      "step": 6128
    },
    {
      "epoch": 0.2194925421240174,
      "grad_norm": 1.9981822967529297,
      "learning_rate": 0.00018175206289154655,
      "loss": 1.5695,
      "step": 6129
    },
    {
      "epoch": 0.2195283542535857,
      "grad_norm": 1.5729684829711914,
      "learning_rate": 0.00018174538247802015,
      "loss": 1.1199,
      "step": 6130
    },
    {
      "epoch": 0.21956416638315399,
      "grad_norm": 1.5781065225601196,
      "learning_rate": 0.00018173870096471512,
      "loss": 1.7226,
      "step": 6131
    },
    {
      "epoch": 0.21959997851272225,
      "grad_norm": 1.6586942672729492,
      "learning_rate": 0.00018173201835172128,
      "loss": 1.8269,
      "step": 6132
    },
    {
      "epoch": 0.21963579064229055,
      "grad_norm": 1.3786426782608032,
      "learning_rate": 0.00018172533463912857,
      "loss": 1.4942,
      "step": 6133
    },
    {
      "epoch": 0.21967160277185882,
      "grad_norm": 1.72449791431427,
      "learning_rate": 0.00018171864982702692,
      "loss": 1.2589,
      "step": 6134
    },
    {
      "epoch": 0.21970741490142712,
      "grad_norm": 1.6442193984985352,
      "learning_rate": 0.0001817119639155062,
      "loss": 1.3379,
      "step": 6135
    },
    {
      "epoch": 0.21974322703099539,
      "grad_norm": 1.8619656562805176,
      "learning_rate": 0.00018170527690465643,
      "loss": 1.5929,
      "step": 6136
    },
    {
      "epoch": 0.21977903916056368,
      "grad_norm": 1.7643412351608276,
      "learning_rate": 0.00018169858879456757,
      "loss": 1.5387,
      "step": 6137
    },
    {
      "epoch": 0.21981485129013198,
      "grad_norm": 1.723496437072754,
      "learning_rate": 0.00018169189958532953,
      "loss": 1.6536,
      "step": 6138
    },
    {
      "epoch": 0.21985066341970025,
      "grad_norm": 2.1918203830718994,
      "learning_rate": 0.0001816852092770324,
      "loss": 1.6178,
      "step": 6139
    },
    {
      "epoch": 0.21988647554926855,
      "grad_norm": 1.7967193126678467,
      "learning_rate": 0.00018167851786976612,
      "loss": 1.4998,
      "step": 6140
    },
    {
      "epoch": 0.21992228767883681,
      "grad_norm": 1.3858860731124878,
      "learning_rate": 0.00018167182536362074,
      "loss": 1.4789,
      "step": 6141
    },
    {
      "epoch": 0.2199580998084051,
      "grad_norm": 1.7353756427764893,
      "learning_rate": 0.00018166513175868633,
      "loss": 1.4175,
      "step": 6142
    },
    {
      "epoch": 0.21999391193797338,
      "grad_norm": 1.6364984512329102,
      "learning_rate": 0.0001816584370550529,
      "loss": 1.2205,
      "step": 6143
    },
    {
      "epoch": 0.22002972406754168,
      "grad_norm": 2.147890567779541,
      "learning_rate": 0.00018165174125281053,
      "loss": 1.4915,
      "step": 6144
    },
    {
      "epoch": 0.22006553619710997,
      "grad_norm": 1.781660556793213,
      "learning_rate": 0.0001816450443520493,
      "loss": 1.4868,
      "step": 6145
    },
    {
      "epoch": 0.22010134832667824,
      "grad_norm": 1.9566551446914673,
      "learning_rate": 0.00018163834635285931,
      "loss": 1.2187,
      "step": 6146
    },
    {
      "epoch": 0.22013716045624654,
      "grad_norm": 1.7707067728042603,
      "learning_rate": 0.00018163164725533068,
      "loss": 1.7138,
      "step": 6147
    },
    {
      "epoch": 0.2201729725858148,
      "grad_norm": 3.547220468521118,
      "learning_rate": 0.0001816249470595535,
      "loss": 1.4808,
      "step": 6148
    },
    {
      "epoch": 0.2202087847153831,
      "grad_norm": 2.155532121658325,
      "learning_rate": 0.000181618245765618,
      "loss": 1.5946,
      "step": 6149
    },
    {
      "epoch": 0.22024459684495137,
      "grad_norm": 2.2041211128234863,
      "learning_rate": 0.00018161154337361426,
      "loss": 1.5524,
      "step": 6150
    },
    {
      "epoch": 0.22028040897451967,
      "grad_norm": 1.5323847532272339,
      "learning_rate": 0.00018160483988363248,
      "loss": 1.6958,
      "step": 6151
    },
    {
      "epoch": 0.22031622110408797,
      "grad_norm": 1.5857058763504028,
      "learning_rate": 0.00018159813529576284,
      "loss": 1.9383,
      "step": 6152
    },
    {
      "epoch": 0.22035203323365624,
      "grad_norm": 1.6870907545089722,
      "learning_rate": 0.0001815914296100955,
      "loss": 1.7842,
      "step": 6153
    },
    {
      "epoch": 0.22038784536322453,
      "grad_norm": 1.9299302101135254,
      "learning_rate": 0.00018158472282672078,
      "loss": 1.8147,
      "step": 6154
    },
    {
      "epoch": 0.2204236574927928,
      "grad_norm": 1.6161481142044067,
      "learning_rate": 0.00018157801494572885,
      "loss": 1.6953,
      "step": 6155
    },
    {
      "epoch": 0.2204594696223611,
      "grad_norm": 1.6391772031784058,
      "learning_rate": 0.00018157130596720996,
      "loss": 1.6505,
      "step": 6156
    },
    {
      "epoch": 0.22049528175192937,
      "grad_norm": 1.602982759475708,
      "learning_rate": 0.0001815645958912543,
      "loss": 1.3066,
      "step": 6157
    },
    {
      "epoch": 0.22053109388149766,
      "grad_norm": 2.5656521320343018,
      "learning_rate": 0.0001815578847179523,
      "loss": 1.6071,
      "step": 6158
    },
    {
      "epoch": 0.22056690601106596,
      "grad_norm": 1.99832284450531,
      "learning_rate": 0.0001815511724473941,
      "loss": 1.7087,
      "step": 6159
    },
    {
      "epoch": 0.22060271814063423,
      "grad_norm": 2.7880725860595703,
      "learning_rate": 0.0001815444590796701,
      "loss": 1.4331,
      "step": 6160
    },
    {
      "epoch": 0.22063853027020253,
      "grad_norm": 2.317629337310791,
      "learning_rate": 0.00018153774461487058,
      "loss": 1.5764,
      "step": 6161
    },
    {
      "epoch": 0.2206743423997708,
      "grad_norm": 1.6050294637680054,
      "learning_rate": 0.00018153102905308589,
      "loss": 1.8059,
      "step": 6162
    },
    {
      "epoch": 0.2207101545293391,
      "grad_norm": 1.231059193611145,
      "learning_rate": 0.00018152431239440637,
      "loss": 1.6119,
      "step": 6163
    },
    {
      "epoch": 0.22074596665890736,
      "grad_norm": 1.5513490438461304,
      "learning_rate": 0.00018151759463892235,
      "loss": 1.6632,
      "step": 6164
    },
    {
      "epoch": 0.22078177878847566,
      "grad_norm": 1.990567684173584,
      "learning_rate": 0.00018151087578672427,
      "loss": 1.4703,
      "step": 6165
    },
    {
      "epoch": 0.22081759091804395,
      "grad_norm": 1.8540167808532715,
      "learning_rate": 0.00018150415583790253,
      "loss": 1.3144,
      "step": 6166
    },
    {
      "epoch": 0.22085340304761222,
      "grad_norm": 1.6363404989242554,
      "learning_rate": 0.00018149743479254745,
      "loss": 1.7437,
      "step": 6167
    },
    {
      "epoch": 0.22088921517718052,
      "grad_norm": 1.5769498348236084,
      "learning_rate": 0.00018149071265074955,
      "loss": 1.6028,
      "step": 6168
    },
    {
      "epoch": 0.2209250273067488,
      "grad_norm": 1.8275872468948364,
      "learning_rate": 0.0001814839894125992,
      "loss": 1.1498,
      "step": 6169
    },
    {
      "epoch": 0.22096083943631709,
      "grad_norm": 1.8236889839172363,
      "learning_rate": 0.0001814772650781869,
      "loss": 1.728,
      "step": 6170
    },
    {
      "epoch": 0.22099665156588535,
      "grad_norm": 2.3265621662139893,
      "learning_rate": 0.0001814705396476031,
      "loss": 1.3853,
      "step": 6171
    },
    {
      "epoch": 0.22103246369545365,
      "grad_norm": 1.8353328704833984,
      "learning_rate": 0.00018146381312093826,
      "loss": 1.526,
      "step": 6172
    },
    {
      "epoch": 0.22106827582502195,
      "grad_norm": 1.6437424421310425,
      "learning_rate": 0.00018145708549828287,
      "loss": 1.8824,
      "step": 6173
    },
    {
      "epoch": 0.22110408795459022,
      "grad_norm": 1.395546317100525,
      "learning_rate": 0.00018145035677972753,
      "loss": 1.7179,
      "step": 6174
    },
    {
      "epoch": 0.2211399000841585,
      "grad_norm": 1.4064130783081055,
      "learning_rate": 0.00018144362696536267,
      "loss": 1.5391,
      "step": 6175
    },
    {
      "epoch": 0.22117571221372678,
      "grad_norm": 2.1587369441986084,
      "learning_rate": 0.00018143689605527885,
      "loss": 1.1614,
      "step": 6176
    },
    {
      "epoch": 0.22121152434329508,
      "grad_norm": 1.8785380125045776,
      "learning_rate": 0.00018143016404956669,
      "loss": 1.1919,
      "step": 6177
    },
    {
      "epoch": 0.22124733647286335,
      "grad_norm": 1.5250529050827026,
      "learning_rate": 0.00018142343094831667,
      "loss": 1.7925,
      "step": 6178
    },
    {
      "epoch": 0.22128314860243165,
      "grad_norm": 1.775303840637207,
      "learning_rate": 0.0001814166967516194,
      "loss": 1.5447,
      "step": 6179
    },
    {
      "epoch": 0.22131896073199994,
      "grad_norm": 1.2775332927703857,
      "learning_rate": 0.00018140996145956552,
      "loss": 1.1843,
      "step": 6180
    },
    {
      "epoch": 0.2213547728615682,
      "grad_norm": 1.5202330350875854,
      "learning_rate": 0.00018140322507224563,
      "loss": 1.4731,
      "step": 6181
    },
    {
      "epoch": 0.2213905849911365,
      "grad_norm": 1.5258673429489136,
      "learning_rate": 0.00018139648758975032,
      "loss": 1.2282,
      "step": 6182
    },
    {
      "epoch": 0.22142639712070478,
      "grad_norm": 2.3224363327026367,
      "learning_rate": 0.00018138974901217027,
      "loss": 1.9237,
      "step": 6183
    },
    {
      "epoch": 0.22146220925027307,
      "grad_norm": 1.9350823163986206,
      "learning_rate": 0.00018138300933959615,
      "loss": 1.67,
      "step": 6184
    },
    {
      "epoch": 0.22149802137984134,
      "grad_norm": 1.8837072849273682,
      "learning_rate": 0.0001813762685721186,
      "loss": 1.4745,
      "step": 6185
    },
    {
      "epoch": 0.22153383350940964,
      "grad_norm": 1.769679307937622,
      "learning_rate": 0.00018136952670982833,
      "loss": 1.3859,
      "step": 6186
    },
    {
      "epoch": 0.22156964563897794,
      "grad_norm": 1.521981120109558,
      "learning_rate": 0.00018136278375281605,
      "loss": 1.5587,
      "step": 6187
    },
    {
      "epoch": 0.2216054577685462,
      "grad_norm": 1.510655403137207,
      "learning_rate": 0.00018135603970117242,
      "loss": 1.5694,
      "step": 6188
    },
    {
      "epoch": 0.2216412698981145,
      "grad_norm": 1.176619529724121,
      "learning_rate": 0.00018134929455498828,
      "loss": 1.48,
      "step": 6189
    },
    {
      "epoch": 0.22167708202768277,
      "grad_norm": 2.1085398197174072,
      "learning_rate": 0.0001813425483143543,
      "loss": 1.8069,
      "step": 6190
    },
    {
      "epoch": 0.22171289415725107,
      "grad_norm": 1.5834347009658813,
      "learning_rate": 0.00018133580097936123,
      "loss": 1.5686,
      "step": 6191
    },
    {
      "epoch": 0.22174870628681934,
      "grad_norm": 2.538158893585205,
      "learning_rate": 0.00018132905255009986,
      "loss": 1.6837,
      "step": 6192
    },
    {
      "epoch": 0.22178451841638763,
      "grad_norm": 1.5322290658950806,
      "learning_rate": 0.00018132230302666104,
      "loss": 1.6239,
      "step": 6193
    },
    {
      "epoch": 0.2218203305459559,
      "grad_norm": 2.453597068786621,
      "learning_rate": 0.0001813155524091355,
      "loss": 1.3409,
      "step": 6194
    },
    {
      "epoch": 0.2218561426755242,
      "grad_norm": 1.7916405200958252,
      "learning_rate": 0.00018130880069761412,
      "loss": 1.8466,
      "step": 6195
    },
    {
      "epoch": 0.2218919548050925,
      "grad_norm": 1.4588032960891724,
      "learning_rate": 0.00018130204789218769,
      "loss": 1.7948,
      "step": 6196
    },
    {
      "epoch": 0.22192776693466076,
      "grad_norm": 1.7840462923049927,
      "learning_rate": 0.00018129529399294706,
      "loss": 1.9397,
      "step": 6197
    },
    {
      "epoch": 0.22196357906422906,
      "grad_norm": 1.4719196557998657,
      "learning_rate": 0.00018128853899998312,
      "loss": 1.3483,
      "step": 6198
    },
    {
      "epoch": 0.22199939119379733,
      "grad_norm": 1.5863127708435059,
      "learning_rate": 0.00018128178291338678,
      "loss": 1.7581,
      "step": 6199
    },
    {
      "epoch": 0.22203520332336563,
      "grad_norm": 1.4045363664627075,
      "learning_rate": 0.00018127502573324887,
      "loss": 1.4722,
      "step": 6200
    },
    {
      "epoch": 0.2220710154529339,
      "grad_norm": 1.8057374954223633,
      "learning_rate": 0.00018126826745966032,
      "loss": 1.7084,
      "step": 6201
    },
    {
      "epoch": 0.2221068275825022,
      "grad_norm": 1.5089365243911743,
      "learning_rate": 0.00018126150809271208,
      "loss": 1.678,
      "step": 6202
    },
    {
      "epoch": 0.2221426397120705,
      "grad_norm": 1.7062714099884033,
      "learning_rate": 0.00018125474763249505,
      "loss": 1.7745,
      "step": 6203
    },
    {
      "epoch": 0.22217845184163876,
      "grad_norm": 2.236283540725708,
      "learning_rate": 0.00018124798607910018,
      "loss": 1.6298,
      "step": 6204
    },
    {
      "epoch": 0.22221426397120705,
      "grad_norm": 1.514384150505066,
      "learning_rate": 0.0001812412234326185,
      "loss": 1.5133,
      "step": 6205
    },
    {
      "epoch": 0.22225007610077532,
      "grad_norm": 1.520835041999817,
      "learning_rate": 0.00018123445969314095,
      "loss": 1.8003,
      "step": 6206
    },
    {
      "epoch": 0.22228588823034362,
      "grad_norm": 2.0588576793670654,
      "learning_rate": 0.00018122769486075854,
      "loss": 1.6132,
      "step": 6207
    },
    {
      "epoch": 0.2223217003599119,
      "grad_norm": 1.1865562200546265,
      "learning_rate": 0.00018122092893556224,
      "loss": 1.5413,
      "step": 6208
    },
    {
      "epoch": 0.22235751248948019,
      "grad_norm": 1.4218183755874634,
      "learning_rate": 0.0001812141619176431,
      "loss": 1.5644,
      "step": 6209
    },
    {
      "epoch": 0.22239332461904848,
      "grad_norm": 1.5924785137176514,
      "learning_rate": 0.00018120739380709218,
      "loss": 1.6539,
      "step": 6210
    },
    {
      "epoch": 0.22242913674861675,
      "grad_norm": 1.443805456161499,
      "learning_rate": 0.00018120062460400056,
      "loss": 1.3789,
      "step": 6211
    },
    {
      "epoch": 0.22246494887818505,
      "grad_norm": 1.639870285987854,
      "learning_rate": 0.00018119385430845925,
      "loss": 1.5284,
      "step": 6212
    },
    {
      "epoch": 0.22250076100775332,
      "grad_norm": 1.666635274887085,
      "learning_rate": 0.00018118708292055936,
      "loss": 1.8486,
      "step": 6213
    },
    {
      "epoch": 0.2225365731373216,
      "grad_norm": 2.6075167655944824,
      "learning_rate": 0.00018118031044039198,
      "loss": 1.778,
      "step": 6214
    },
    {
      "epoch": 0.22257238526688988,
      "grad_norm": 1.558547019958496,
      "learning_rate": 0.00018117353686804825,
      "loss": 1.4225,
      "step": 6215
    },
    {
      "epoch": 0.22260819739645818,
      "grad_norm": 1.4020191431045532,
      "learning_rate": 0.00018116676220361933,
      "loss": 1.5796,
      "step": 6216
    },
    {
      "epoch": 0.22264400952602648,
      "grad_norm": 2.120662212371826,
      "learning_rate": 0.00018115998644719627,
      "loss": 1.9943,
      "step": 6217
    },
    {
      "epoch": 0.22267982165559475,
      "grad_norm": 1.3515700101852417,
      "learning_rate": 0.0001811532095988703,
      "loss": 1.3811,
      "step": 6218
    },
    {
      "epoch": 0.22271563378516304,
      "grad_norm": 1.45658278465271,
      "learning_rate": 0.00018114643165873258,
      "loss": 1.5035,
      "step": 6219
    },
    {
      "epoch": 0.2227514459147313,
      "grad_norm": 1.6097304821014404,
      "learning_rate": 0.00018113965262687426,
      "loss": 1.524,
      "step": 6220
    },
    {
      "epoch": 0.2227872580442996,
      "grad_norm": 1.5943652391433716,
      "learning_rate": 0.00018113287250338662,
      "loss": 1.3009,
      "step": 6221
    },
    {
      "epoch": 0.22282307017386788,
      "grad_norm": 1.5684243440628052,
      "learning_rate": 0.0001811260912883608,
      "loss": 1.4074,
      "step": 6222
    },
    {
      "epoch": 0.22285888230343617,
      "grad_norm": 1.7942955493927002,
      "learning_rate": 0.0001811193089818881,
      "loss": 1.5083,
      "step": 6223
    },
    {
      "epoch": 0.22289469443300447,
      "grad_norm": 2.2910614013671875,
      "learning_rate": 0.0001811125255840597,
      "loss": 1.6178,
      "step": 6224
    },
    {
      "epoch": 0.22293050656257274,
      "grad_norm": 1.332371473312378,
      "learning_rate": 0.00018110574109496692,
      "loss": 1.4492,
      "step": 6225
    },
    {
      "epoch": 0.22296631869214104,
      "grad_norm": 1.8471901416778564,
      "learning_rate": 0.000181098955514701,
      "loss": 1.9102,
      "step": 6226
    },
    {
      "epoch": 0.2230021308217093,
      "grad_norm": 1.6000697612762451,
      "learning_rate": 0.00018109216884335325,
      "loss": 1.6102,
      "step": 6227
    },
    {
      "epoch": 0.2230379429512776,
      "grad_norm": 2.0699243545532227,
      "learning_rate": 0.00018108538108101496,
      "loss": 1.8961,
      "step": 6228
    },
    {
      "epoch": 0.22307375508084587,
      "grad_norm": 1.5482474565505981,
      "learning_rate": 0.00018107859222777747,
      "loss": 1.6179,
      "step": 6229
    },
    {
      "epoch": 0.22310956721041417,
      "grad_norm": 1.4630217552185059,
      "learning_rate": 0.0001810718022837321,
      "loss": 1.3147,
      "step": 6230
    },
    {
      "epoch": 0.22314537933998246,
      "grad_norm": 1.7212188243865967,
      "learning_rate": 0.00018106501124897024,
      "loss": 1.6115,
      "step": 6231
    },
    {
      "epoch": 0.22318119146955073,
      "grad_norm": 1.2010308504104614,
      "learning_rate": 0.00018105821912358318,
      "loss": 1.5784,
      "step": 6232
    },
    {
      "epoch": 0.22321700359911903,
      "grad_norm": 2.4285857677459717,
      "learning_rate": 0.00018105142590766235,
      "loss": 1.3719,
      "step": 6233
    },
    {
      "epoch": 0.2232528157286873,
      "grad_norm": 1.9801949262619019,
      "learning_rate": 0.00018104463160129912,
      "loss": 1.5422,
      "step": 6234
    },
    {
      "epoch": 0.2232886278582556,
      "grad_norm": 1.3408149480819702,
      "learning_rate": 0.00018103783620458495,
      "loss": 1.3496,
      "step": 6235
    },
    {
      "epoch": 0.22332443998782386,
      "grad_norm": 1.6955066919326782,
      "learning_rate": 0.0001810310397176112,
      "loss": 1.6563,
      "step": 6236
    },
    {
      "epoch": 0.22336025211739216,
      "grad_norm": 1.4549084901809692,
      "learning_rate": 0.0001810242421404693,
      "loss": 1.4026,
      "step": 6237
    },
    {
      "epoch": 0.22339606424696046,
      "grad_norm": 1.6285228729248047,
      "learning_rate": 0.00018101744347325078,
      "loss": 1.511,
      "step": 6238
    },
    {
      "epoch": 0.22343187637652873,
      "grad_norm": 1.6430824995040894,
      "learning_rate": 0.00018101064371604705,
      "loss": 1.3705,
      "step": 6239
    },
    {
      "epoch": 0.22346768850609702,
      "grad_norm": 2.0069923400878906,
      "learning_rate": 0.0001810038428689496,
      "loss": 1.7298,
      "step": 6240
    },
    {
      "epoch": 0.2235035006356653,
      "grad_norm": 1.5702707767486572,
      "learning_rate": 0.00018099704093204997,
      "loss": 1.7249,
      "step": 6241
    },
    {
      "epoch": 0.2235393127652336,
      "grad_norm": 1.7741975784301758,
      "learning_rate": 0.00018099023790543956,
      "loss": 1.6276,
      "step": 6242
    },
    {
      "epoch": 0.22357512489480186,
      "grad_norm": 2.8160290718078613,
      "learning_rate": 0.00018098343378921002,
      "loss": 1.3841,
      "step": 6243
    },
    {
      "epoch": 0.22361093702437015,
      "grad_norm": 1.9434565305709839,
      "learning_rate": 0.00018097662858345282,
      "loss": 1.2333,
      "step": 6244
    },
    {
      "epoch": 0.22364674915393845,
      "grad_norm": 1.4331356287002563,
      "learning_rate": 0.00018096982228825957,
      "loss": 1.6064,
      "step": 6245
    },
    {
      "epoch": 0.22368256128350672,
      "grad_norm": 1.8122775554656982,
      "learning_rate": 0.00018096301490372175,
      "loss": 1.5847,
      "step": 6246
    },
    {
      "epoch": 0.22371837341307502,
      "grad_norm": 1.7992037534713745,
      "learning_rate": 0.00018095620642993106,
      "loss": 1.5136,
      "step": 6247
    },
    {
      "epoch": 0.22375418554264329,
      "grad_norm": 2.3840503692626953,
      "learning_rate": 0.000180949396866979,
      "loss": 1.4274,
      "step": 6248
    },
    {
      "epoch": 0.22378999767221158,
      "grad_norm": 1.6450737714767456,
      "learning_rate": 0.00018094258621495724,
      "loss": 1.5595,
      "step": 6249
    },
    {
      "epoch": 0.22382580980177985,
      "grad_norm": 1.2827435731887817,
      "learning_rate": 0.00018093577447395737,
      "loss": 1.6185,
      "step": 6250
    },
    {
      "epoch": 0.22386162193134815,
      "grad_norm": 2.183157444000244,
      "learning_rate": 0.00018092896164407108,
      "loss": 1.3872,
      "step": 6251
    },
    {
      "epoch": 0.22389743406091644,
      "grad_norm": 2.581634521484375,
      "learning_rate": 0.00018092214772538994,
      "loss": 1.6593,
      "step": 6252
    },
    {
      "epoch": 0.2239332461904847,
      "grad_norm": 1.961072325706482,
      "learning_rate": 0.00018091533271800576,
      "loss": 1.5356,
      "step": 6253
    },
    {
      "epoch": 0.223969058320053,
      "grad_norm": 2.0827033519744873,
      "learning_rate": 0.00018090851662201011,
      "loss": 1.5634,
      "step": 6254
    },
    {
      "epoch": 0.22400487044962128,
      "grad_norm": 1.3981750011444092,
      "learning_rate": 0.00018090169943749476,
      "loss": 1.6067,
      "step": 6255
    },
    {
      "epoch": 0.22404068257918958,
      "grad_norm": 1.7544121742248535,
      "learning_rate": 0.00018089488116455137,
      "loss": 1.7618,
      "step": 6256
    },
    {
      "epoch": 0.22407649470875785,
      "grad_norm": 1.768568754196167,
      "learning_rate": 0.00018088806180327174,
      "loss": 1.7251,
      "step": 6257
    },
    {
      "epoch": 0.22411230683832614,
      "grad_norm": 1.7023133039474487,
      "learning_rate": 0.00018088124135374754,
      "loss": 1.3179,
      "step": 6258
    },
    {
      "epoch": 0.22414811896789444,
      "grad_norm": 1.566927433013916,
      "learning_rate": 0.00018087441981607056,
      "loss": 1.3815,
      "step": 6259
    },
    {
      "epoch": 0.2241839310974627,
      "grad_norm": 1.7557883262634277,
      "learning_rate": 0.00018086759719033261,
      "loss": 1.6312,
      "step": 6260
    },
    {
      "epoch": 0.224219743227031,
      "grad_norm": 1.8031694889068604,
      "learning_rate": 0.00018086077347662544,
      "loss": 1.3539,
      "step": 6261
    },
    {
      "epoch": 0.22425555535659927,
      "grad_norm": 1.501671552658081,
      "learning_rate": 0.00018085394867504087,
      "loss": 1.7447,
      "step": 6262
    },
    {
      "epoch": 0.22429136748616757,
      "grad_norm": 1.342667579650879,
      "learning_rate": 0.00018084712278567072,
      "loss": 1.5844,
      "step": 6263
    },
    {
      "epoch": 0.22432717961573584,
      "grad_norm": 2.0589170455932617,
      "learning_rate": 0.00018084029580860679,
      "loss": 2.0065,
      "step": 6264
    },
    {
      "epoch": 0.22436299174530414,
      "grad_norm": 1.7275969982147217,
      "learning_rate": 0.000180833467743941,
      "loss": 1.4656,
      "step": 6265
    },
    {
      "epoch": 0.22439880387487243,
      "grad_norm": 1.4489154815673828,
      "learning_rate": 0.00018082663859176514,
      "loss": 1.4036,
      "step": 6266
    },
    {
      "epoch": 0.2244346160044407,
      "grad_norm": 1.4381591081619263,
      "learning_rate": 0.00018081980835217115,
      "loss": 1.5086,
      "step": 6267
    },
    {
      "epoch": 0.224470428134009,
      "grad_norm": 1.5848140716552734,
      "learning_rate": 0.00018081297702525083,
      "loss": 1.88,
      "step": 6268
    },
    {
      "epoch": 0.22450624026357727,
      "grad_norm": 1.8268117904663086,
      "learning_rate": 0.0001808061446110962,
      "loss": 1.713,
      "step": 6269
    },
    {
      "epoch": 0.22454205239314556,
      "grad_norm": 2.428276538848877,
      "learning_rate": 0.0001807993111097991,
      "loss": 1.7823,
      "step": 6270
    },
    {
      "epoch": 0.22457786452271383,
      "grad_norm": 1.898605465888977,
      "learning_rate": 0.0001807924765214515,
      "loss": 1.4447,
      "step": 6271
    },
    {
      "epoch": 0.22461367665228213,
      "grad_norm": 1.346147894859314,
      "learning_rate": 0.00018078564084614534,
      "loss": 1.4593,
      "step": 6272
    },
    {
      "epoch": 0.22464948878185043,
      "grad_norm": 1.8311480283737183,
      "learning_rate": 0.0001807788040839726,
      "loss": 1.6737,
      "step": 6273
    },
    {
      "epoch": 0.2246853009114187,
      "grad_norm": 1.3992347717285156,
      "learning_rate": 0.0001807719662350252,
      "loss": 1.5412,
      "step": 6274
    },
    {
      "epoch": 0.224721113040987,
      "grad_norm": 1.5278624296188354,
      "learning_rate": 0.00018076512729939522,
      "loss": 1.6334,
      "step": 6275
    },
    {
      "epoch": 0.22475692517055526,
      "grad_norm": 1.8901277780532837,
      "learning_rate": 0.00018075828727717464,
      "loss": 1.8401,
      "step": 6276
    },
    {
      "epoch": 0.22479273730012356,
      "grad_norm": 2.278604745864868,
      "learning_rate": 0.00018075144616845544,
      "loss": 1.7158,
      "step": 6277
    },
    {
      "epoch": 0.22482854942969183,
      "grad_norm": 1.5036617517471313,
      "learning_rate": 0.00018074460397332973,
      "loss": 1.7371,
      "step": 6278
    },
    {
      "epoch": 0.22486436155926012,
      "grad_norm": 1.6316254138946533,
      "learning_rate": 0.00018073776069188954,
      "loss": 1.6364,
      "step": 6279
    },
    {
      "epoch": 0.22490017368882842,
      "grad_norm": 1.824449062347412,
      "learning_rate": 0.0001807309163242269,
      "loss": 1.2859,
      "step": 6280
    },
    {
      "epoch": 0.2249359858183967,
      "grad_norm": 1.523816466331482,
      "learning_rate": 0.0001807240708704339,
      "loss": 1.6968,
      "step": 6281
    },
    {
      "epoch": 0.22497179794796499,
      "grad_norm": 1.8160182237625122,
      "learning_rate": 0.0001807172243306027,
      "loss": 1.5923,
      "step": 6282
    },
    {
      "epoch": 0.22500761007753325,
      "grad_norm": 1.6295522451400757,
      "learning_rate": 0.00018071037670482532,
      "loss": 1.5968,
      "step": 6283
    },
    {
      "epoch": 0.22504342220710155,
      "grad_norm": 1.463212251663208,
      "learning_rate": 0.00018070352799319395,
      "loss": 1.5347,
      "step": 6284
    },
    {
      "epoch": 0.22507923433666982,
      "grad_norm": 2.812549114227295,
      "learning_rate": 0.0001806966781958007,
      "loss": 1.654,
      "step": 6285
    },
    {
      "epoch": 0.22511504646623812,
      "grad_norm": 2.011024236679077,
      "learning_rate": 0.00018068982731273773,
      "loss": 1.9352,
      "step": 6286
    },
    {
      "epoch": 0.2251508585958064,
      "grad_norm": 1.4859150648117065,
      "learning_rate": 0.00018068297534409725,
      "loss": 1.6877,
      "step": 6287
    },
    {
      "epoch": 0.22518667072537468,
      "grad_norm": 1.5548865795135498,
      "learning_rate": 0.00018067612228997137,
      "loss": 1.53,
      "step": 6288
    },
    {
      "epoch": 0.22522248285494298,
      "grad_norm": 1.7031159400939941,
      "learning_rate": 0.00018066926815045236,
      "loss": 1.8067,
      "step": 6289
    },
    {
      "epoch": 0.22525829498451125,
      "grad_norm": 1.9277985095977783,
      "learning_rate": 0.00018066241292563238,
      "loss": 1.3285,
      "step": 6290
    },
    {
      "epoch": 0.22529410711407954,
      "grad_norm": 1.4769515991210938,
      "learning_rate": 0.00018065555661560368,
      "loss": 1.6458,
      "step": 6291
    },
    {
      "epoch": 0.2253299192436478,
      "grad_norm": 2.014338970184326,
      "learning_rate": 0.00018064869922045852,
      "loss": 1.4688,
      "step": 6292
    },
    {
      "epoch": 0.2253657313732161,
      "grad_norm": 2.711336851119995,
      "learning_rate": 0.00018064184074028915,
      "loss": 1.5906,
      "step": 6293
    },
    {
      "epoch": 0.22540154350278438,
      "grad_norm": 1.8387864828109741,
      "learning_rate": 0.0001806349811751878,
      "loss": 1.6775,
      "step": 6294
    },
    {
      "epoch": 0.22543735563235268,
      "grad_norm": 1.484028935432434,
      "learning_rate": 0.00018062812052524683,
      "loss": 1.4705,
      "step": 6295
    },
    {
      "epoch": 0.22547316776192097,
      "grad_norm": 1.9155514240264893,
      "learning_rate": 0.00018062125879055846,
      "loss": 1.6196,
      "step": 6296
    },
    {
      "epoch": 0.22550897989148924,
      "grad_norm": 2.07671856880188,
      "learning_rate": 0.00018061439597121508,
      "loss": 1.6635,
      "step": 6297
    },
    {
      "epoch": 0.22554479202105754,
      "grad_norm": 1.2981393337249756,
      "learning_rate": 0.000180607532067309,
      "loss": 1.3212,
      "step": 6298
    },
    {
      "epoch": 0.2255806041506258,
      "grad_norm": 1.5452812910079956,
      "learning_rate": 0.0001806006670789325,
      "loss": 1.7199,
      "step": 6299
    },
    {
      "epoch": 0.2256164162801941,
      "grad_norm": 2.042917251586914,
      "learning_rate": 0.00018059380100617802,
      "loss": 1.561,
      "step": 6300
    },
    {
      "epoch": 0.22565222840976237,
      "grad_norm": 2.002063035964966,
      "learning_rate": 0.0001805869338491379,
      "loss": 1.4581,
      "step": 6301
    },
    {
      "epoch": 0.22568804053933067,
      "grad_norm": 1.6189521551132202,
      "learning_rate": 0.00018058006560790453,
      "loss": 1.2429,
      "step": 6302
    },
    {
      "epoch": 0.22572385266889897,
      "grad_norm": 1.7002862691879272,
      "learning_rate": 0.00018057319628257034,
      "loss": 1.6999,
      "step": 6303
    },
    {
      "epoch": 0.22575966479846724,
      "grad_norm": 1.913028597831726,
      "learning_rate": 0.0001805663258732277,
      "loss": 1.9182,
      "step": 6304
    },
    {
      "epoch": 0.22579547692803553,
      "grad_norm": 1.3990799188613892,
      "learning_rate": 0.0001805594543799691,
      "loss": 1.6439,
      "step": 6305
    },
    {
      "epoch": 0.2258312890576038,
      "grad_norm": 1.9262644052505493,
      "learning_rate": 0.00018055258180288696,
      "loss": 1.6578,
      "step": 6306
    },
    {
      "epoch": 0.2258671011871721,
      "grad_norm": 1.1598010063171387,
      "learning_rate": 0.0001805457081420737,
      "loss": 1.645,
      "step": 6307
    },
    {
      "epoch": 0.22590291331674037,
      "grad_norm": 2.582381248474121,
      "learning_rate": 0.00018053883339762183,
      "loss": 1.9006,
      "step": 6308
    },
    {
      "epoch": 0.22593872544630866,
      "grad_norm": 1.5848309993743896,
      "learning_rate": 0.00018053195756962388,
      "loss": 1.4253,
      "step": 6309
    },
    {
      "epoch": 0.22597453757587696,
      "grad_norm": 1.6140457391738892,
      "learning_rate": 0.0001805250806581723,
      "loss": 1.3751,
      "step": 6310
    },
    {
      "epoch": 0.22601034970544523,
      "grad_norm": 1.7949855327606201,
      "learning_rate": 0.00018051820266335963,
      "loss": 1.4448,
      "step": 6311
    },
    {
      "epoch": 0.22604616183501353,
      "grad_norm": 1.4919768571853638,
      "learning_rate": 0.0001805113235852784,
      "loss": 1.4016,
      "step": 6312
    },
    {
      "epoch": 0.2260819739645818,
      "grad_norm": 1.2824842929840088,
      "learning_rate": 0.00018050444342402114,
      "loss": 1.3802,
      "step": 6313
    },
    {
      "epoch": 0.2261177860941501,
      "grad_norm": 1.514653205871582,
      "learning_rate": 0.0001804975621796805,
      "loss": 1.4011,
      "step": 6314
    },
    {
      "epoch": 0.22615359822371836,
      "grad_norm": 1.4949651956558228,
      "learning_rate": 0.00018049067985234895,
      "loss": 1.3139,
      "step": 6315
    },
    {
      "epoch": 0.22618941035328666,
      "grad_norm": 1.9180841445922852,
      "learning_rate": 0.00018048379644211915,
      "loss": 1.4127,
      "step": 6316
    },
    {
      "epoch": 0.22622522248285495,
      "grad_norm": 1.3751518726348877,
      "learning_rate": 0.00018047691194908368,
      "loss": 1.4633,
      "step": 6317
    },
    {
      "epoch": 0.22626103461242322,
      "grad_norm": 2.1771159172058105,
      "learning_rate": 0.00018047002637333517,
      "loss": 1.6793,
      "step": 6318
    },
    {
      "epoch": 0.22629684674199152,
      "grad_norm": 1.4497225284576416,
      "learning_rate": 0.00018046313971496622,
      "loss": 1.4479,
      "step": 6319
    },
    {
      "epoch": 0.2263326588715598,
      "grad_norm": 1.9885940551757812,
      "learning_rate": 0.00018045625197406957,
      "loss": 1.4871,
      "step": 6320
    },
    {
      "epoch": 0.22636847100112809,
      "grad_norm": 1.246687412261963,
      "learning_rate": 0.00018044936315073779,
      "loss": 1.6281,
      "step": 6321
    },
    {
      "epoch": 0.22640428313069635,
      "grad_norm": 3.3700385093688965,
      "learning_rate": 0.0001804424732450636,
      "loss": 1.9536,
      "step": 6322
    },
    {
      "epoch": 0.22644009526026465,
      "grad_norm": 1.9437425136566162,
      "learning_rate": 0.0001804355822571397,
      "loss": 1.7087,
      "step": 6323
    },
    {
      "epoch": 0.22647590738983295,
      "grad_norm": 1.4002232551574707,
      "learning_rate": 0.00018042869018705882,
      "loss": 1.37,
      "step": 6324
    },
    {
      "epoch": 0.22651171951940122,
      "grad_norm": 1.4852545261383057,
      "learning_rate": 0.00018042179703491365,
      "loss": 1.4754,
      "step": 6325
    },
    {
      "epoch": 0.2265475316489695,
      "grad_norm": 1.689438819885254,
      "learning_rate": 0.00018041490280079693,
      "loss": 1.6379,
      "step": 6326
    },
    {
      "epoch": 0.22658334377853778,
      "grad_norm": 1.4079216718673706,
      "learning_rate": 0.00018040800748480142,
      "loss": 1.4282,
      "step": 6327
    },
    {
      "epoch": 0.22661915590810608,
      "grad_norm": 1.5791819095611572,
      "learning_rate": 0.00018040111108701988,
      "loss": 1.7066,
      "step": 6328
    },
    {
      "epoch": 0.22665496803767435,
      "grad_norm": 1.8294531106948853,
      "learning_rate": 0.00018039421360754513,
      "loss": 1.7546,
      "step": 6329
    },
    {
      "epoch": 0.22669078016724264,
      "grad_norm": 1.7738605737686157,
      "learning_rate": 0.0001803873150464699,
      "loss": 1.3616,
      "step": 6330
    },
    {
      "epoch": 0.22672659229681094,
      "grad_norm": 1.9782477617263794,
      "learning_rate": 0.00018038041540388705,
      "loss": 1.3159,
      "step": 6331
    },
    {
      "epoch": 0.2267624044263792,
      "grad_norm": 1.4635730981826782,
      "learning_rate": 0.00018037351467988942,
      "loss": 1.4511,
      "step": 6332
    },
    {
      "epoch": 0.2267982165559475,
      "grad_norm": 1.2722338438034058,
      "learning_rate": 0.0001803666128745698,
      "loss": 1.2824,
      "step": 6333
    },
    {
      "epoch": 0.22683402868551578,
      "grad_norm": 1.7284460067749023,
      "learning_rate": 0.00018035970998802106,
      "loss": 1.5667,
      "step": 6334
    },
    {
      "epoch": 0.22686984081508407,
      "grad_norm": 2.6332521438598633,
      "learning_rate": 0.0001803528060203361,
      "loss": 1.4429,
      "step": 6335
    },
    {
      "epoch": 0.22690565294465234,
      "grad_norm": 1.378933310508728,
      "learning_rate": 0.00018034590097160778,
      "loss": 1.5293,
      "step": 6336
    },
    {
      "epoch": 0.22694146507422064,
      "grad_norm": 1.7138888835906982,
      "learning_rate": 0.000180338994841929,
      "loss": 1.4708,
      "step": 6337
    },
    {
      "epoch": 0.22697727720378894,
      "grad_norm": 1.4144269227981567,
      "learning_rate": 0.00018033208763139266,
      "loss": 1.578,
      "step": 6338
    },
    {
      "epoch": 0.2270130893333572,
      "grad_norm": 1.4831655025482178,
      "learning_rate": 0.0001803251793400917,
      "loss": 1.6398,
      "step": 6339
    },
    {
      "epoch": 0.2270489014629255,
      "grad_norm": 1.3129905462265015,
      "learning_rate": 0.0001803182699681191,
      "loss": 1.4506,
      "step": 6340
    },
    {
      "epoch": 0.22708471359249377,
      "grad_norm": 1.9662981033325195,
      "learning_rate": 0.00018031135951556774,
      "loss": 1.3558,
      "step": 6341
    },
    {
      "epoch": 0.22712052572206207,
      "grad_norm": 2.3339595794677734,
      "learning_rate": 0.00018030444798253066,
      "loss": 1.4351,
      "step": 6342
    },
    {
      "epoch": 0.22715633785163034,
      "grad_norm": 1.4761534929275513,
      "learning_rate": 0.0001802975353691008,
      "loss": 1.6159,
      "step": 6343
    },
    {
      "epoch": 0.22719214998119863,
      "grad_norm": 1.3196064233779907,
      "learning_rate": 0.00018029062167537117,
      "loss": 1.6138,
      "step": 6344
    },
    {
      "epoch": 0.22722796211076693,
      "grad_norm": 2.0419766902923584,
      "learning_rate": 0.00018028370690143482,
      "loss": 1.8339,
      "step": 6345
    },
    {
      "epoch": 0.2272637742403352,
      "grad_norm": 1.5514767169952393,
      "learning_rate": 0.00018027679104738473,
      "loss": 1.3189,
      "step": 6346
    },
    {
      "epoch": 0.2272995863699035,
      "grad_norm": 2.7501816749572754,
      "learning_rate": 0.00018026987411331398,
      "loss": 1.8378,
      "step": 6347
    },
    {
      "epoch": 0.22733539849947176,
      "grad_norm": 1.7505801916122437,
      "learning_rate": 0.0001802629560993156,
      "loss": 1.1969,
      "step": 6348
    },
    {
      "epoch": 0.22737121062904006,
      "grad_norm": 1.4281238317489624,
      "learning_rate": 0.0001802560370054827,
      "loss": 1.78,
      "step": 6349
    },
    {
      "epoch": 0.22740702275860833,
      "grad_norm": 1.6368708610534668,
      "learning_rate": 0.00018024911683190833,
      "loss": 1.3031,
      "step": 6350
    },
    {
      "epoch": 0.22744283488817663,
      "grad_norm": 1.8474522829055786,
      "learning_rate": 0.0001802421955786856,
      "loss": 1.8179,
      "step": 6351
    },
    {
      "epoch": 0.22747864701774492,
      "grad_norm": 1.4447168111801147,
      "learning_rate": 0.00018023527324590764,
      "loss": 1.4314,
      "step": 6352
    },
    {
      "epoch": 0.2275144591473132,
      "grad_norm": 1.9680529832839966,
      "learning_rate": 0.0001802283498336676,
      "loss": 1.5552,
      "step": 6353
    },
    {
      "epoch": 0.2275502712768815,
      "grad_norm": 1.547568917274475,
      "learning_rate": 0.00018022142534205858,
      "loss": 1.5545,
      "step": 6354
    },
    {
      "epoch": 0.22758608340644976,
      "grad_norm": 1.4191595315933228,
      "learning_rate": 0.00018021449977117374,
      "loss": 1.359,
      "step": 6355
    },
    {
      "epoch": 0.22762189553601805,
      "grad_norm": 1.7385724782943726,
      "learning_rate": 0.00018020757312110628,
      "loss": 1.6338,
      "step": 6356
    },
    {
      "epoch": 0.22765770766558632,
      "grad_norm": 1.5614619255065918,
      "learning_rate": 0.0001802006453919494,
      "loss": 1.7259,
      "step": 6357
    },
    {
      "epoch": 0.22769351979515462,
      "grad_norm": 1.4145011901855469,
      "learning_rate": 0.0001801937165837963,
      "loss": 1.7155,
      "step": 6358
    },
    {
      "epoch": 0.22772933192472292,
      "grad_norm": 1.572713017463684,
      "learning_rate": 0.00018018678669674015,
      "loss": 1.7646,
      "step": 6359
    },
    {
      "epoch": 0.22776514405429119,
      "grad_norm": 1.4809653759002686,
      "learning_rate": 0.00018017985573087425,
      "loss": 1.4841,
      "step": 6360
    },
    {
      "epoch": 0.22780095618385948,
      "grad_norm": 2.632725715637207,
      "learning_rate": 0.0001801729236862918,
      "loss": 1.9187,
      "step": 6361
    },
    {
      "epoch": 0.22783676831342775,
      "grad_norm": 1.5056543350219727,
      "learning_rate": 0.0001801659905630861,
      "loss": 1.6402,
      "step": 6362
    },
    {
      "epoch": 0.22787258044299605,
      "grad_norm": 1.3668396472930908,
      "learning_rate": 0.00018015905636135037,
      "loss": 1.4198,
      "step": 6363
    },
    {
      "epoch": 0.22790839257256432,
      "grad_norm": 1.650078296661377,
      "learning_rate": 0.00018015212108117793,
      "loss": 1.8745,
      "step": 6364
    },
    {
      "epoch": 0.2279442047021326,
      "grad_norm": 2.297715902328491,
      "learning_rate": 0.0001801451847226621,
      "loss": 1.8726,
      "step": 6365
    },
    {
      "epoch": 0.2279800168317009,
      "grad_norm": 1.448528528213501,
      "learning_rate": 0.0001801382472858962,
      "loss": 1.5484,
      "step": 6366
    },
    {
      "epoch": 0.22801582896126918,
      "grad_norm": 1.730246901512146,
      "learning_rate": 0.00018013130877097357,
      "loss": 1.4851,
      "step": 6367
    },
    {
      "epoch": 0.22805164109083748,
      "grad_norm": 1.6730163097381592,
      "learning_rate": 0.0001801243691779875,
      "loss": 1.522,
      "step": 6368
    },
    {
      "epoch": 0.22808745322040574,
      "grad_norm": 1.6910569667816162,
      "learning_rate": 0.00018011742850703146,
      "loss": 1.4104,
      "step": 6369
    },
    {
      "epoch": 0.22812326534997404,
      "grad_norm": 1.7938308715820312,
      "learning_rate": 0.00018011048675819872,
      "loss": 1.4716,
      "step": 6370
    },
    {
      "epoch": 0.2281590774795423,
      "grad_norm": 1.5565799474716187,
      "learning_rate": 0.0001801035439315827,
      "loss": 1.5583,
      "step": 6371
    },
    {
      "epoch": 0.2281948896091106,
      "grad_norm": 1.8678364753723145,
      "learning_rate": 0.00018009660002727684,
      "loss": 1.5625,
      "step": 6372
    },
    {
      "epoch": 0.2282307017386789,
      "grad_norm": 1.3503905534744263,
      "learning_rate": 0.00018008965504537455,
      "loss": 1.5808,
      "step": 6373
    },
    {
      "epoch": 0.22826651386824717,
      "grad_norm": 1.5878171920776367,
      "learning_rate": 0.00018008270898596927,
      "loss": 1.7811,
      "step": 6374
    },
    {
      "epoch": 0.22830232599781547,
      "grad_norm": 1.4853515625,
      "learning_rate": 0.00018007576184915443,
      "loss": 1.5756,
      "step": 6375
    },
    {
      "epoch": 0.22833813812738374,
      "grad_norm": 1.5101004838943481,
      "learning_rate": 0.00018006881363502348,
      "loss": 1.3801,
      "step": 6376
    },
    {
      "epoch": 0.22837395025695204,
      "grad_norm": 1.6938331127166748,
      "learning_rate": 0.00018006186434366996,
      "loss": 1.7725,
      "step": 6377
    },
    {
      "epoch": 0.2284097623865203,
      "grad_norm": 2.058910846710205,
      "learning_rate": 0.0001800549139751873,
      "loss": 1.5681,
      "step": 6378
    },
    {
      "epoch": 0.2284455745160886,
      "grad_norm": 1.5402759313583374,
      "learning_rate": 0.00018004796252966908,
      "loss": 1.4617,
      "step": 6379
    },
    {
      "epoch": 0.2284813866456569,
      "grad_norm": 1.5306357145309448,
      "learning_rate": 0.00018004101000720872,
      "loss": 1.7913,
      "step": 6380
    },
    {
      "epoch": 0.22851719877522517,
      "grad_norm": 1.2929790019989014,
      "learning_rate": 0.00018003405640789987,
      "loss": 1.553,
      "step": 6381
    },
    {
      "epoch": 0.22855301090479346,
      "grad_norm": 2.393298387527466,
      "learning_rate": 0.00018002710173183596,
      "loss": 1.6883,
      "step": 6382
    },
    {
      "epoch": 0.22858882303436173,
      "grad_norm": 2.0982320308685303,
      "learning_rate": 0.00018002014597911066,
      "loss": 1.6216,
      "step": 6383
    },
    {
      "epoch": 0.22862463516393003,
      "grad_norm": 1.3045873641967773,
      "learning_rate": 0.00018001318914981753,
      "loss": 1.4899,
      "step": 6384
    },
    {
      "epoch": 0.2286604472934983,
      "grad_norm": 2.3285608291625977,
      "learning_rate": 0.00018000623124405014,
      "loss": 1.4443,
      "step": 6385
    },
    {
      "epoch": 0.2286962594230666,
      "grad_norm": 1.8748106956481934,
      "learning_rate": 0.0001799992722619021,
      "loss": 1.2426,
      "step": 6386
    },
    {
      "epoch": 0.22873207155263486,
      "grad_norm": 1.3914124965667725,
      "learning_rate": 0.00017999231220346707,
      "loss": 1.5566,
      "step": 6387
    },
    {
      "epoch": 0.22876788368220316,
      "grad_norm": 1.7344447374343872,
      "learning_rate": 0.00017998535106883862,
      "loss": 1.052,
      "step": 6388
    },
    {
      "epoch": 0.22880369581177146,
      "grad_norm": 1.5599722862243652,
      "learning_rate": 0.00017997838885811047,
      "loss": 1.6529,
      "step": 6389
    },
    {
      "epoch": 0.22883950794133973,
      "grad_norm": 2.0522711277008057,
      "learning_rate": 0.00017997142557137625,
      "loss": 1.4925,
      "step": 6390
    },
    {
      "epoch": 0.22887532007090802,
      "grad_norm": 1.712795376777649,
      "learning_rate": 0.00017996446120872967,
      "loss": 1.7085,
      "step": 6391
    },
    {
      "epoch": 0.2289111322004763,
      "grad_norm": 1.9475706815719604,
      "learning_rate": 0.00017995749577026443,
      "loss": 1.6277,
      "step": 6392
    },
    {
      "epoch": 0.2289469443300446,
      "grad_norm": 1.562954068183899,
      "learning_rate": 0.0001799505292560742,
      "loss": 1.8237,
      "step": 6393
    },
    {
      "epoch": 0.22898275645961286,
      "grad_norm": 1.4970970153808594,
      "learning_rate": 0.00017994356166625271,
      "loss": 1.6119,
      "step": 6394
    },
    {
      "epoch": 0.22901856858918115,
      "grad_norm": 1.8240835666656494,
      "learning_rate": 0.0001799365930008937,
      "loss": 1.7235,
      "step": 6395
    },
    {
      "epoch": 0.22905438071874945,
      "grad_norm": 1.410115361213684,
      "learning_rate": 0.000179929623260091,
      "loss": 1.429,
      "step": 6396
    },
    {
      "epoch": 0.22909019284831772,
      "grad_norm": 1.732579231262207,
      "learning_rate": 0.0001799226524439383,
      "loss": 1.6388,
      "step": 6397
    },
    {
      "epoch": 0.22912600497788602,
      "grad_norm": 2.0173988342285156,
      "learning_rate": 0.0001799156805525294,
      "loss": 1.6149,
      "step": 6398
    },
    {
      "epoch": 0.22916181710745429,
      "grad_norm": 1.4980366230010986,
      "learning_rate": 0.00017990870758595811,
      "loss": 1.5596,
      "step": 6399
    },
    {
      "epoch": 0.22919762923702258,
      "grad_norm": 2.1325318813323975,
      "learning_rate": 0.0001799017335443182,
      "loss": 1.5557,
      "step": 6400
    },
    {
      "epoch": 0.22923344136659085,
      "grad_norm": 1.7884522676467896,
      "learning_rate": 0.00017989475842770358,
      "loss": 1.6517,
      "step": 6401
    },
    {
      "epoch": 0.22926925349615915,
      "grad_norm": 2.133835554122925,
      "learning_rate": 0.00017988778223620799,
      "loss": 1.7446,
      "step": 6402
    },
    {
      "epoch": 0.22930506562572744,
      "grad_norm": 1.8718260526657104,
      "learning_rate": 0.0001798808049699254,
      "loss": 1.7916,
      "step": 6403
    },
    {
      "epoch": 0.2293408777552957,
      "grad_norm": 1.6703238487243652,
      "learning_rate": 0.00017987382662894955,
      "loss": 1.5125,
      "step": 6404
    },
    {
      "epoch": 0.229376689884864,
      "grad_norm": 1.7600868940353394,
      "learning_rate": 0.00017986684721337442,
      "loss": 1.5224,
      "step": 6405
    },
    {
      "epoch": 0.22941250201443228,
      "grad_norm": 2.184349298477173,
      "learning_rate": 0.00017985986672329392,
      "loss": 1.5551,
      "step": 6406
    },
    {
      "epoch": 0.22944831414400058,
      "grad_norm": 1.4052538871765137,
      "learning_rate": 0.0001798528851588019,
      "loss": 1.7645,
      "step": 6407
    },
    {
      "epoch": 0.22948412627356884,
      "grad_norm": 1.7552518844604492,
      "learning_rate": 0.0001798459025199923,
      "loss": 1.5787,
      "step": 6408
    },
    {
      "epoch": 0.22951993840313714,
      "grad_norm": 1.7903473377227783,
      "learning_rate": 0.00017983891880695907,
      "loss": 1.6544,
      "step": 6409
    },
    {
      "epoch": 0.22955575053270544,
      "grad_norm": 1.63605797290802,
      "learning_rate": 0.00017983193401979616,
      "loss": 1.7336,
      "step": 6410
    },
    {
      "epoch": 0.2295915626622737,
      "grad_norm": 1.7833870649337769,
      "learning_rate": 0.0001798249481585976,
      "loss": 1.425,
      "step": 6411
    },
    {
      "epoch": 0.229627374791842,
      "grad_norm": 1.5450332164764404,
      "learning_rate": 0.0001798179612234573,
      "loss": 1.3434,
      "step": 6412
    },
    {
      "epoch": 0.22966318692141027,
      "grad_norm": 1.3541749715805054,
      "learning_rate": 0.0001798109732144693,
      "loss": 1.5277,
      "step": 6413
    },
    {
      "epoch": 0.22969899905097857,
      "grad_norm": 2.7870755195617676,
      "learning_rate": 0.0001798039841317276,
      "loss": 1.5438,
      "step": 6414
    },
    {
      "epoch": 0.22973481118054684,
      "grad_norm": 1.9439632892608643,
      "learning_rate": 0.00017979699397532625,
      "loss": 1.6845,
      "step": 6415
    },
    {
      "epoch": 0.22977062331011514,
      "grad_norm": 1.472992181777954,
      "learning_rate": 0.00017979000274535926,
      "loss": 1.59,
      "step": 6416
    },
    {
      "epoch": 0.22980643543968343,
      "grad_norm": 1.9579699039459229,
      "learning_rate": 0.0001797830104419207,
      "loss": 1.3821,
      "step": 6417
    },
    {
      "epoch": 0.2298422475692517,
      "grad_norm": 1.2323920726776123,
      "learning_rate": 0.00017977601706510465,
      "loss": 1.5868,
      "step": 6418
    },
    {
      "epoch": 0.22987805969882,
      "grad_norm": 1.9308135509490967,
      "learning_rate": 0.0001797690226150052,
      "loss": 1.5732,
      "step": 6419
    },
    {
      "epoch": 0.22991387182838827,
      "grad_norm": 1.283857822418213,
      "learning_rate": 0.00017976202709171643,
      "loss": 1.3582,
      "step": 6420
    },
    {
      "epoch": 0.22994968395795656,
      "grad_norm": 1.7926145792007446,
      "learning_rate": 0.0001797550304953325,
      "loss": 1.542,
      "step": 6421
    },
    {
      "epoch": 0.22998549608752483,
      "grad_norm": 2.011465072631836,
      "learning_rate": 0.00017974803282594747,
      "loss": 1.3821,
      "step": 6422
    },
    {
      "epoch": 0.23002130821709313,
      "grad_norm": 1.5812910795211792,
      "learning_rate": 0.00017974103408365557,
      "loss": 1.5659,
      "step": 6423
    },
    {
      "epoch": 0.23005712034666143,
      "grad_norm": 1.8776155710220337,
      "learning_rate": 0.00017973403426855088,
      "loss": 1.7368,
      "step": 6424
    },
    {
      "epoch": 0.2300929324762297,
      "grad_norm": 1.4800854921340942,
      "learning_rate": 0.00017972703338072762,
      "loss": 1.6797,
      "step": 6425
    },
    {
      "epoch": 0.230128744605798,
      "grad_norm": 1.9689767360687256,
      "learning_rate": 0.00017972003142027997,
      "loss": 2.1101,
      "step": 6426
    },
    {
      "epoch": 0.23016455673536626,
      "grad_norm": 1.2748472690582275,
      "learning_rate": 0.00017971302838730213,
      "loss": 1.5971,
      "step": 6427
    },
    {
      "epoch": 0.23020036886493456,
      "grad_norm": 1.3918734788894653,
      "learning_rate": 0.00017970602428188834,
      "loss": 1.4222,
      "step": 6428
    },
    {
      "epoch": 0.23023618099450283,
      "grad_norm": 1.911766529083252,
      "learning_rate": 0.00017969901910413276,
      "loss": 1.3167,
      "step": 6429
    },
    {
      "epoch": 0.23027199312407112,
      "grad_norm": 1.62160325050354,
      "learning_rate": 0.00017969201285412972,
      "loss": 1.6437,
      "step": 6430
    },
    {
      "epoch": 0.23030780525363942,
      "grad_norm": 1.9006664752960205,
      "learning_rate": 0.0001796850055319734,
      "loss": 1.3403,
      "step": 6431
    },
    {
      "epoch": 0.2303436173832077,
      "grad_norm": 3.353529691696167,
      "learning_rate": 0.00017967799713775815,
      "loss": 1.7944,
      "step": 6432
    },
    {
      "epoch": 0.23037942951277599,
      "grad_norm": 2.1918132305145264,
      "learning_rate": 0.00017967098767157822,
      "loss": 1.7478,
      "step": 6433
    },
    {
      "epoch": 0.23041524164234425,
      "grad_norm": 1.9099441766738892,
      "learning_rate": 0.00017966397713352792,
      "loss": 1.2382,
      "step": 6434
    },
    {
      "epoch": 0.23045105377191255,
      "grad_norm": 1.6920967102050781,
      "learning_rate": 0.00017965696552370156,
      "loss": 1.4024,
      "step": 6435
    },
    {
      "epoch": 0.23048686590148082,
      "grad_norm": 1.8513035774230957,
      "learning_rate": 0.00017964995284219348,
      "loss": 1.4771,
      "step": 6436
    },
    {
      "epoch": 0.23052267803104912,
      "grad_norm": 2.202528476715088,
      "learning_rate": 0.00017964293908909803,
      "loss": 1.9447,
      "step": 6437
    },
    {
      "epoch": 0.2305584901606174,
      "grad_norm": 1.4865727424621582,
      "learning_rate": 0.00017963592426450956,
      "loss": 1.4472,
      "step": 6438
    },
    {
      "epoch": 0.23059430229018568,
      "grad_norm": 1.3838335275650024,
      "learning_rate": 0.0001796289083685225,
      "loss": 1.7319,
      "step": 6439
    },
    {
      "epoch": 0.23063011441975398,
      "grad_norm": 1.7137506008148193,
      "learning_rate": 0.0001796218914012311,
      "loss": 1.6542,
      "step": 6440
    },
    {
      "epoch": 0.23066592654932225,
      "grad_norm": 1.3480849266052246,
      "learning_rate": 0.0001796148733627299,
      "loss": 1.6549,
      "step": 6441
    },
    {
      "epoch": 0.23070173867889054,
      "grad_norm": 1.6950236558914185,
      "learning_rate": 0.00017960785425311332,
      "loss": 1.503,
      "step": 6442
    },
    {
      "epoch": 0.2307375508084588,
      "grad_norm": 2.1359996795654297,
      "learning_rate": 0.0001796008340724757,
      "loss": 1.5654,
      "step": 6443
    },
    {
      "epoch": 0.2307733629380271,
      "grad_norm": 1.839069128036499,
      "learning_rate": 0.00017959381282091152,
      "loss": 1.5582,
      "step": 6444
    },
    {
      "epoch": 0.2308091750675954,
      "grad_norm": 2.33095383644104,
      "learning_rate": 0.0001795867904985153,
      "loss": 1.7936,
      "step": 6445
    },
    {
      "epoch": 0.23084498719716368,
      "grad_norm": 1.4011579751968384,
      "learning_rate": 0.00017957976710538144,
      "loss": 1.7156,
      "step": 6446
    },
    {
      "epoch": 0.23088079932673197,
      "grad_norm": 1.6660329103469849,
      "learning_rate": 0.00017957274264160448,
      "loss": 1.8214,
      "step": 6447
    },
    {
      "epoch": 0.23091661145630024,
      "grad_norm": 1.4621220827102661,
      "learning_rate": 0.0001795657171072789,
      "loss": 1.6597,
      "step": 6448
    },
    {
      "epoch": 0.23095242358586854,
      "grad_norm": 1.8586870431900024,
      "learning_rate": 0.00017955869050249925,
      "loss": 2.0127,
      "step": 6449
    },
    {
      "epoch": 0.2309882357154368,
      "grad_norm": 1.8170366287231445,
      "learning_rate": 0.00017955166282736002,
      "loss": 1.6028,
      "step": 6450
    },
    {
      "epoch": 0.2310240478450051,
      "grad_norm": 1.4119073152542114,
      "learning_rate": 0.0001795446340819558,
      "loss": 1.1546,
      "step": 6451
    },
    {
      "epoch": 0.2310598599745734,
      "grad_norm": 1.4222018718719482,
      "learning_rate": 0.0001795376042663811,
      "loss": 1.3849,
      "step": 6452
    },
    {
      "epoch": 0.23109567210414167,
      "grad_norm": 1.6756181716918945,
      "learning_rate": 0.00017953057338073055,
      "loss": 1.6725,
      "step": 6453
    },
    {
      "epoch": 0.23113148423370997,
      "grad_norm": 1.8242745399475098,
      "learning_rate": 0.00017952354142509872,
      "loss": 1.5498,
      "step": 6454
    },
    {
      "epoch": 0.23116729636327824,
      "grad_norm": 1.50627601146698,
      "learning_rate": 0.00017951650839958023,
      "loss": 1.4107,
      "step": 6455
    },
    {
      "epoch": 0.23120310849284653,
      "grad_norm": 1.5063188076019287,
      "learning_rate": 0.0001795094743042697,
      "loss": 1.5205,
      "step": 6456
    },
    {
      "epoch": 0.2312389206224148,
      "grad_norm": 1.6867979764938354,
      "learning_rate": 0.00017950243913926171,
      "loss": 1.4739,
      "step": 6457
    },
    {
      "epoch": 0.2312747327519831,
      "grad_norm": 2.8445029258728027,
      "learning_rate": 0.000179495402904651,
      "loss": 1.8396,
      "step": 6458
    },
    {
      "epoch": 0.2313105448815514,
      "grad_norm": 1.4332222938537598,
      "learning_rate": 0.00017948836560053216,
      "loss": 1.3423,
      "step": 6459
    },
    {
      "epoch": 0.23134635701111966,
      "grad_norm": 1.8352477550506592,
      "learning_rate": 0.00017948132722699992,
      "loss": 1.7802,
      "step": 6460
    },
    {
      "epoch": 0.23138216914068796,
      "grad_norm": 1.4083185195922852,
      "learning_rate": 0.0001794742877841489,
      "loss": 1.6408,
      "step": 6461
    },
    {
      "epoch": 0.23141798127025623,
      "grad_norm": 1.7609177827835083,
      "learning_rate": 0.00017946724727207388,
      "loss": 1.3707,
      "step": 6462
    },
    {
      "epoch": 0.23145379339982453,
      "grad_norm": 2.6420741081237793,
      "learning_rate": 0.00017946020569086955,
      "loss": 1.2318,
      "step": 6463
    },
    {
      "epoch": 0.2314896055293928,
      "grad_norm": 1.4578214883804321,
      "learning_rate": 0.00017945316304063066,
      "loss": 1.3948,
      "step": 6464
    },
    {
      "epoch": 0.2315254176589611,
      "grad_norm": 1.539219617843628,
      "learning_rate": 0.00017944611932145193,
      "loss": 1.5818,
      "step": 6465
    },
    {
      "epoch": 0.2315612297885294,
      "grad_norm": 2.507035493850708,
      "learning_rate": 0.0001794390745334281,
      "loss": 1.1414,
      "step": 6466
    },
    {
      "epoch": 0.23159704191809766,
      "grad_norm": 2.5263805389404297,
      "learning_rate": 0.00017943202867665408,
      "loss": 1.7751,
      "step": 6467
    },
    {
      "epoch": 0.23163285404766595,
      "grad_norm": 1.9388482570648193,
      "learning_rate": 0.00017942498175122453,
      "loss": 1.9844,
      "step": 6468
    },
    {
      "epoch": 0.23166866617723422,
      "grad_norm": 1.8652827739715576,
      "learning_rate": 0.0001794179337572343,
      "loss": 1.4795,
      "step": 6469
    },
    {
      "epoch": 0.23170447830680252,
      "grad_norm": 1.9985748529434204,
      "learning_rate": 0.0001794108846947782,
      "loss": 1.7811,
      "step": 6470
    },
    {
      "epoch": 0.2317402904363708,
      "grad_norm": 2.3072433471679688,
      "learning_rate": 0.00017940383456395109,
      "loss": 1.4727,
      "step": 6471
    },
    {
      "epoch": 0.23177610256593908,
      "grad_norm": 1.2530908584594727,
      "learning_rate": 0.00017939678336484783,
      "loss": 1.459,
      "step": 6472
    },
    {
      "epoch": 0.23181191469550738,
      "grad_norm": 1.7884855270385742,
      "learning_rate": 0.00017938973109756323,
      "loss": 1.7551,
      "step": 6473
    },
    {
      "epoch": 0.23184772682507565,
      "grad_norm": 2.037320613861084,
      "learning_rate": 0.00017938267776219225,
      "loss": 1.684,
      "step": 6474
    },
    {
      "epoch": 0.23188353895464395,
      "grad_norm": 1.3687164783477783,
      "learning_rate": 0.00017937562335882968,
      "loss": 1.7375,
      "step": 6475
    },
    {
      "epoch": 0.23191935108421222,
      "grad_norm": 1.592024564743042,
      "learning_rate": 0.00017936856788757055,
      "loss": 1.5635,
      "step": 6476
    },
    {
      "epoch": 0.2319551632137805,
      "grad_norm": 1.433759331703186,
      "learning_rate": 0.00017936151134850966,
      "loss": 1.6694,
      "step": 6477
    },
    {
      "epoch": 0.23199097534334878,
      "grad_norm": 2.6087303161621094,
      "learning_rate": 0.00017935445374174202,
      "loss": 1.5391,
      "step": 6478
    },
    {
      "epoch": 0.23202678747291708,
      "grad_norm": 1.7297744750976562,
      "learning_rate": 0.0001793473950673626,
      "loss": 1.6245,
      "step": 6479
    },
    {
      "epoch": 0.23206259960248538,
      "grad_norm": 1.6509273052215576,
      "learning_rate": 0.00017934033532546632,
      "loss": 1.6673,
      "step": 6480
    },
    {
      "epoch": 0.23209841173205364,
      "grad_norm": 1.5647467374801636,
      "learning_rate": 0.00017933327451614812,
      "loss": 1.6559,
      "step": 6481
    },
    {
      "epoch": 0.23213422386162194,
      "grad_norm": 1.4652289152145386,
      "learning_rate": 0.0001793262126395031,
      "loss": 1.5293,
      "step": 6482
    },
    {
      "epoch": 0.2321700359911902,
      "grad_norm": 1.4504691362380981,
      "learning_rate": 0.00017931914969562617,
      "loss": 1.5716,
      "step": 6483
    },
    {
      "epoch": 0.2322058481207585,
      "grad_norm": 1.9844943284988403,
      "learning_rate": 0.0001793120856846124,
      "loss": 1.6779,
      "step": 6484
    },
    {
      "epoch": 0.23224166025032678,
      "grad_norm": 1.2851874828338623,
      "learning_rate": 0.00017930502060655682,
      "loss": 1.4328,
      "step": 6485
    },
    {
      "epoch": 0.23227747237989507,
      "grad_norm": 1.9430853128433228,
      "learning_rate": 0.0001792979544615545,
      "loss": 1.7059,
      "step": 6486
    },
    {
      "epoch": 0.23231328450946334,
      "grad_norm": 1.595819115638733,
      "learning_rate": 0.00017929088724970052,
      "loss": 1.567,
      "step": 6487
    },
    {
      "epoch": 0.23234909663903164,
      "grad_norm": 2.0435125827789307,
      "learning_rate": 0.0001792838189710899,
      "loss": 1.8018,
      "step": 6488
    },
    {
      "epoch": 0.23238490876859993,
      "grad_norm": 1.225368618965149,
      "learning_rate": 0.00017927674962581774,
      "loss": 1.3603,
      "step": 6489
    },
    {
      "epoch": 0.2324207208981682,
      "grad_norm": 1.9997037649154663,
      "learning_rate": 0.0001792696792139792,
      "loss": 1.4549,
      "step": 6490
    },
    {
      "epoch": 0.2324565330277365,
      "grad_norm": 1.4556210041046143,
      "learning_rate": 0.0001792626077356694,
      "loss": 1.5237,
      "step": 6491
    },
    {
      "epoch": 0.23249234515730477,
      "grad_norm": 1.36371648311615,
      "learning_rate": 0.0001792555351909834,
      "loss": 1.7727,
      "step": 6492
    },
    {
      "epoch": 0.23252815728687307,
      "grad_norm": 1.272359848022461,
      "learning_rate": 0.0001792484615800164,
      "loss": 1.4476,
      "step": 6493
    },
    {
      "epoch": 0.23256396941644134,
      "grad_norm": 1.8688451051712036,
      "learning_rate": 0.00017924138690286366,
      "loss": 1.6197,
      "step": 6494
    },
    {
      "epoch": 0.23259978154600963,
      "grad_norm": 1.4307365417480469,
      "learning_rate": 0.0001792343111596202,
      "loss": 1.5767,
      "step": 6495
    },
    {
      "epoch": 0.23263559367557793,
      "grad_norm": 1.3361865282058716,
      "learning_rate": 0.00017922723435038131,
      "loss": 1.2674,
      "step": 6496
    },
    {
      "epoch": 0.2326714058051462,
      "grad_norm": 1.858230710029602,
      "learning_rate": 0.00017922015647524217,
      "loss": 1.6375,
      "step": 6497
    },
    {
      "epoch": 0.2327072179347145,
      "grad_norm": 1.5478800535202026,
      "learning_rate": 0.00017921307753429803,
      "loss": 1.2073,
      "step": 6498
    },
    {
      "epoch": 0.23274303006428276,
      "grad_norm": 1.2986811399459839,
      "learning_rate": 0.00017920599752764408,
      "loss": 1.2458,
      "step": 6499
    },
    {
      "epoch": 0.23277884219385106,
      "grad_norm": 1.7942970991134644,
      "learning_rate": 0.00017919891645537563,
      "loss": 1.5128,
      "step": 6500
    },
    {
      "epoch": 0.23281465432341933,
      "grad_norm": 2.231881618499756,
      "learning_rate": 0.0001791918343175879,
      "loss": 1.7035,
      "step": 6501
    },
    {
      "epoch": 0.23285046645298763,
      "grad_norm": 1.8615409135818481,
      "learning_rate": 0.0001791847511143762,
      "loss": 1.269,
      "step": 6502
    },
    {
      "epoch": 0.23288627858255592,
      "grad_norm": 1.8193210363388062,
      "learning_rate": 0.0001791776668458358,
      "loss": 1.7327,
      "step": 6503
    },
    {
      "epoch": 0.2329220907121242,
      "grad_norm": 1.4053817987442017,
      "learning_rate": 0.00017917058151206204,
      "loss": 1.6057,
      "step": 6504
    },
    {
      "epoch": 0.2329579028416925,
      "grad_norm": 1.530077576637268,
      "learning_rate": 0.00017916349511315022,
      "loss": 1.6364,
      "step": 6505
    },
    {
      "epoch": 0.23299371497126076,
      "grad_norm": 1.6833380460739136,
      "learning_rate": 0.0001791564076491957,
      "loss": 1.5841,
      "step": 6506
    },
    {
      "epoch": 0.23302952710082905,
      "grad_norm": 1.6770578622817993,
      "learning_rate": 0.0001791493191202938,
      "loss": 1.8875,
      "step": 6507
    },
    {
      "epoch": 0.23306533923039732,
      "grad_norm": 2.6495304107666016,
      "learning_rate": 0.00017914222952653992,
      "loss": 1.6587,
      "step": 6508
    },
    {
      "epoch": 0.23310115135996562,
      "grad_norm": 1.6221790313720703,
      "learning_rate": 0.00017913513886802943,
      "loss": 1.5729,
      "step": 6509
    },
    {
      "epoch": 0.23313696348953392,
      "grad_norm": 1.6502245664596558,
      "learning_rate": 0.0001791280471448577,
      "loss": 1.6398,
      "step": 6510
    },
    {
      "epoch": 0.23317277561910218,
      "grad_norm": 1.5571110248565674,
      "learning_rate": 0.00017912095435712017,
      "loss": 1.515,
      "step": 6511
    },
    {
      "epoch": 0.23320858774867048,
      "grad_norm": 1.4226124286651611,
      "learning_rate": 0.0001791138605049123,
      "loss": 1.7522,
      "step": 6512
    },
    {
      "epoch": 0.23324439987823875,
      "grad_norm": 1.4621069431304932,
      "learning_rate": 0.00017910676558832944,
      "loss": 1.6877,
      "step": 6513
    },
    {
      "epoch": 0.23328021200780705,
      "grad_norm": 2.131011724472046,
      "learning_rate": 0.0001790996696074671,
      "loss": 1.6325,
      "step": 6514
    },
    {
      "epoch": 0.23331602413737532,
      "grad_norm": 1.4561113119125366,
      "learning_rate": 0.00017909257256242076,
      "loss": 1.3617,
      "step": 6515
    },
    {
      "epoch": 0.2333518362669436,
      "grad_norm": 1.3390988111495972,
      "learning_rate": 0.00017908547445328585,
      "loss": 1.6789,
      "step": 6516
    },
    {
      "epoch": 0.2333876483965119,
      "grad_norm": 1.5033286809921265,
      "learning_rate": 0.0001790783752801579,
      "loss": 1.547,
      "step": 6517
    },
    {
      "epoch": 0.23342346052608018,
      "grad_norm": 1.6072916984558105,
      "learning_rate": 0.00017907127504313241,
      "loss": 1.3535,
      "step": 6518
    },
    {
      "epoch": 0.23345927265564848,
      "grad_norm": 1.4692038297653198,
      "learning_rate": 0.00017906417374230493,
      "loss": 1.5247,
      "step": 6519
    },
    {
      "epoch": 0.23349508478521674,
      "grad_norm": 1.9015370607376099,
      "learning_rate": 0.00017905707137777098,
      "loss": 1.353,
      "step": 6520
    },
    {
      "epoch": 0.23353089691478504,
      "grad_norm": 1.8074142932891846,
      "learning_rate": 0.00017904996794962608,
      "loss": 1.4386,
      "step": 6521
    },
    {
      "epoch": 0.2335667090443533,
      "grad_norm": 1.8867095708847046,
      "learning_rate": 0.00017904286345796582,
      "loss": 1.5851,
      "step": 6522
    },
    {
      "epoch": 0.2336025211739216,
      "grad_norm": 1.6819953918457031,
      "learning_rate": 0.00017903575790288585,
      "loss": 1.5244,
      "step": 6523
    },
    {
      "epoch": 0.2336383333034899,
      "grad_norm": 2.2761800289154053,
      "learning_rate": 0.00017902865128448166,
      "loss": 1.8374,
      "step": 6524
    },
    {
      "epoch": 0.23367414543305817,
      "grad_norm": 1.5938763618469238,
      "learning_rate": 0.00017902154360284893,
      "loss": 1.3266,
      "step": 6525
    },
    {
      "epoch": 0.23370995756262647,
      "grad_norm": 1.7620958089828491,
      "learning_rate": 0.00017901443485808324,
      "loss": 1.6188,
      "step": 6526
    },
    {
      "epoch": 0.23374576969219474,
      "grad_norm": 1.395570158958435,
      "learning_rate": 0.00017900732505028025,
      "loss": 1.5055,
      "step": 6527
    },
    {
      "epoch": 0.23378158182176303,
      "grad_norm": 1.467537522315979,
      "learning_rate": 0.00017900021417953564,
      "loss": 1.5082,
      "step": 6528
    },
    {
      "epoch": 0.2338173939513313,
      "grad_norm": 1.5979235172271729,
      "learning_rate": 0.000178993102245945,
      "loss": 1.6297,
      "step": 6529
    },
    {
      "epoch": 0.2338532060808996,
      "grad_norm": 1.9237210750579834,
      "learning_rate": 0.0001789859892496041,
      "loss": 1.4545,
      "step": 6530
    },
    {
      "epoch": 0.2338890182104679,
      "grad_norm": 1.3320746421813965,
      "learning_rate": 0.00017897887519060862,
      "loss": 1.3044,
      "step": 6531
    },
    {
      "epoch": 0.23392483034003617,
      "grad_norm": 1.386905312538147,
      "learning_rate": 0.0001789717600690542,
      "loss": 1.6688,
      "step": 6532
    },
    {
      "epoch": 0.23396064246960446,
      "grad_norm": 1.8010025024414062,
      "learning_rate": 0.00017896464388503664,
      "loss": 1.2744,
      "step": 6533
    },
    {
      "epoch": 0.23399645459917273,
      "grad_norm": 1.6420273780822754,
      "learning_rate": 0.00017895752663865167,
      "loss": 1.6643,
      "step": 6534
    },
    {
      "epoch": 0.23403226672874103,
      "grad_norm": 1.6220277547836304,
      "learning_rate": 0.00017895040832999502,
      "loss": 1.6328,
      "step": 6535
    },
    {
      "epoch": 0.2340680788583093,
      "grad_norm": 1.4307856559753418,
      "learning_rate": 0.00017894328895916244,
      "loss": 1.6593,
      "step": 6536
    },
    {
      "epoch": 0.2341038909878776,
      "grad_norm": 1.9479830265045166,
      "learning_rate": 0.00017893616852624974,
      "loss": 1.4733,
      "step": 6537
    },
    {
      "epoch": 0.2341397031174459,
      "grad_norm": 1.6316691637039185,
      "learning_rate": 0.00017892904703135272,
      "loss": 1.6507,
      "step": 6538
    },
    {
      "epoch": 0.23417551524701416,
      "grad_norm": 1.9461170434951782,
      "learning_rate": 0.0001789219244745672,
      "loss": 0.9729,
      "step": 6539
    },
    {
      "epoch": 0.23421132737658246,
      "grad_norm": 1.3413337469100952,
      "learning_rate": 0.00017891480085598896,
      "loss": 1.6239,
      "step": 6540
    },
    {
      "epoch": 0.23424713950615073,
      "grad_norm": 1.4433578252792358,
      "learning_rate": 0.00017890767617571388,
      "loss": 1.4859,
      "step": 6541
    },
    {
      "epoch": 0.23428295163571902,
      "grad_norm": 1.8441306352615356,
      "learning_rate": 0.00017890055043383782,
      "loss": 1.5165,
      "step": 6542
    },
    {
      "epoch": 0.2343187637652873,
      "grad_norm": 1.71013343334198,
      "learning_rate": 0.0001788934236304566,
      "loss": 1.5745,
      "step": 6543
    },
    {
      "epoch": 0.2343545758948556,
      "grad_norm": 2.440901517868042,
      "learning_rate": 0.00017888629576566614,
      "loss": 2.0483,
      "step": 6544
    },
    {
      "epoch": 0.23439038802442388,
      "grad_norm": 1.9219796657562256,
      "learning_rate": 0.00017887916683956233,
      "loss": 1.4539,
      "step": 6545
    },
    {
      "epoch": 0.23442620015399215,
      "grad_norm": 1.5786051750183105,
      "learning_rate": 0.0001788720368522411,
      "loss": 1.3918,
      "step": 6546
    },
    {
      "epoch": 0.23446201228356045,
      "grad_norm": 1.9648666381835938,
      "learning_rate": 0.0001788649058037983,
      "loss": 1.5756,
      "step": 6547
    },
    {
      "epoch": 0.23449782441312872,
      "grad_norm": 1.5821706056594849,
      "learning_rate": 0.00017885777369432994,
      "loss": 1.402,
      "step": 6548
    },
    {
      "epoch": 0.23453363654269702,
      "grad_norm": 1.3339437246322632,
      "learning_rate": 0.000178850640523932,
      "loss": 1.5828,
      "step": 6549
    },
    {
      "epoch": 0.23456944867226528,
      "grad_norm": 1.375614047050476,
      "learning_rate": 0.00017884350629270035,
      "loss": 1.6135,
      "step": 6550
    },
    {
      "epoch": 0.23460526080183358,
      "grad_norm": 1.6413074731826782,
      "learning_rate": 0.00017883637100073104,
      "loss": 1.1794,
      "step": 6551
    },
    {
      "epoch": 0.23464107293140188,
      "grad_norm": 1.5169514417648315,
      "learning_rate": 0.00017882923464812006,
      "loss": 1.1396,
      "step": 6552
    },
    {
      "epoch": 0.23467688506097015,
      "grad_norm": 1.6196491718292236,
      "learning_rate": 0.00017882209723496338,
      "loss": 1.7209,
      "step": 6553
    },
    {
      "epoch": 0.23471269719053844,
      "grad_norm": 1.7131088972091675,
      "learning_rate": 0.00017881495876135708,
      "loss": 1.6069,
      "step": 6554
    },
    {
      "epoch": 0.2347485093201067,
      "grad_norm": 1.6686679124832153,
      "learning_rate": 0.00017880781922739717,
      "loss": 1.9669,
      "step": 6555
    },
    {
      "epoch": 0.234784321449675,
      "grad_norm": 1.7967392206192017,
      "learning_rate": 0.0001788006786331797,
      "loss": 1.5043,
      "step": 6556
    },
    {
      "epoch": 0.23482013357924328,
      "grad_norm": 1.6955111026763916,
      "learning_rate": 0.00017879353697880073,
      "loss": 1.6294,
      "step": 6557
    },
    {
      "epoch": 0.23485594570881158,
      "grad_norm": 1.5779500007629395,
      "learning_rate": 0.00017878639426435638,
      "loss": 1.3588,
      "step": 6558
    },
    {
      "epoch": 0.23489175783837987,
      "grad_norm": 1.3038164377212524,
      "learning_rate": 0.00017877925048994273,
      "loss": 1.5094,
      "step": 6559
    },
    {
      "epoch": 0.23492756996794814,
      "grad_norm": 1.5466959476470947,
      "learning_rate": 0.00017877210565565586,
      "loss": 1.7958,
      "step": 6560
    },
    {
      "epoch": 0.23496338209751644,
      "grad_norm": 1.521091103553772,
      "learning_rate": 0.0001787649597615919,
      "loss": 1.6854,
      "step": 6561
    },
    {
      "epoch": 0.2349991942270847,
      "grad_norm": 1.5606482028961182,
      "learning_rate": 0.00017875781280784705,
      "loss": 1.4973,
      "step": 6562
    },
    {
      "epoch": 0.235035006356653,
      "grad_norm": 2.0361833572387695,
      "learning_rate": 0.0001787506647945174,
      "loss": 1.3534,
      "step": 6563
    },
    {
      "epoch": 0.23507081848622127,
      "grad_norm": 2.3265650272369385,
      "learning_rate": 0.00017874351572169913,
      "loss": 1.3959,
      "step": 6564
    },
    {
      "epoch": 0.23510663061578957,
      "grad_norm": 1.6580675840377808,
      "learning_rate": 0.00017873636558948846,
      "loss": 1.4783,
      "step": 6565
    },
    {
      "epoch": 0.23514244274535787,
      "grad_norm": 1.493943691253662,
      "learning_rate": 0.00017872921439798152,
      "loss": 1.6773,
      "step": 6566
    },
    {
      "epoch": 0.23517825487492613,
      "grad_norm": 1.4382308721542358,
      "learning_rate": 0.00017872206214727455,
      "loss": 1.4732,
      "step": 6567
    },
    {
      "epoch": 0.23521406700449443,
      "grad_norm": 1.3880226612091064,
      "learning_rate": 0.0001787149088374638,
      "loss": 1.462,
      "step": 6568
    },
    {
      "epoch": 0.2352498791340627,
      "grad_norm": 1.3058699369430542,
      "learning_rate": 0.00017870775446864547,
      "loss": 1.5238,
      "step": 6569
    },
    {
      "epoch": 0.235285691263631,
      "grad_norm": 2.446697235107422,
      "learning_rate": 0.00017870059904091584,
      "loss": 1.4423,
      "step": 6570
    },
    {
      "epoch": 0.23532150339319927,
      "grad_norm": 1.8130626678466797,
      "learning_rate": 0.00017869344255437117,
      "loss": 1.5468,
      "step": 6571
    },
    {
      "epoch": 0.23535731552276756,
      "grad_norm": 1.4712576866149902,
      "learning_rate": 0.00017868628500910773,
      "loss": 1.6648,
      "step": 6572
    },
    {
      "epoch": 0.23539312765233586,
      "grad_norm": 1.2141464948654175,
      "learning_rate": 0.00017867912640522182,
      "loss": 1.7291,
      "step": 6573
    },
    {
      "epoch": 0.23542893978190413,
      "grad_norm": 1.6197760105133057,
      "learning_rate": 0.00017867196674280976,
      "loss": 1.9246,
      "step": 6574
    },
    {
      "epoch": 0.23546475191147243,
      "grad_norm": 1.6172927618026733,
      "learning_rate": 0.00017866480602196787,
      "loss": 1.3133,
      "step": 6575
    },
    {
      "epoch": 0.2355005640410407,
      "grad_norm": 1.7807902097702026,
      "learning_rate": 0.00017865764424279248,
      "loss": 1.4431,
      "step": 6576
    },
    {
      "epoch": 0.235536376170609,
      "grad_norm": 1.82480788230896,
      "learning_rate": 0.00017865048140537995,
      "loss": 1.7821,
      "step": 6577
    },
    {
      "epoch": 0.23557218830017726,
      "grad_norm": 1.434335708618164,
      "learning_rate": 0.00017864331750982665,
      "loss": 1.4632,
      "step": 6578
    },
    {
      "epoch": 0.23560800042974556,
      "grad_norm": 1.5445455312728882,
      "learning_rate": 0.0001786361525562289,
      "loss": 1.634,
      "step": 6579
    },
    {
      "epoch": 0.23564381255931385,
      "grad_norm": 1.4568474292755127,
      "learning_rate": 0.0001786289865446832,
      "loss": 1.5565,
      "step": 6580
    },
    {
      "epoch": 0.23567962468888212,
      "grad_norm": 1.866904377937317,
      "learning_rate": 0.00017862181947528592,
      "loss": 1.5269,
      "step": 6581
    },
    {
      "epoch": 0.23571543681845042,
      "grad_norm": 2.703174352645874,
      "learning_rate": 0.00017861465134813348,
      "loss": 1.5168,
      "step": 6582
    },
    {
      "epoch": 0.2357512489480187,
      "grad_norm": 1.7558021545410156,
      "learning_rate": 0.00017860748216332227,
      "loss": 1.4795,
      "step": 6583
    },
    {
      "epoch": 0.23578706107758698,
      "grad_norm": 1.4444247484207153,
      "learning_rate": 0.00017860031192094882,
      "loss": 1.4356,
      "step": 6584
    },
    {
      "epoch": 0.23582287320715525,
      "grad_norm": 1.3739668130874634,
      "learning_rate": 0.00017859314062110954,
      "loss": 1.4568,
      "step": 6585
    },
    {
      "epoch": 0.23585868533672355,
      "grad_norm": 1.9809393882751465,
      "learning_rate": 0.00017858596826390093,
      "loss": 1.6483,
      "step": 6586
    },
    {
      "epoch": 0.23589449746629182,
      "grad_norm": 1.643008828163147,
      "learning_rate": 0.0001785787948494195,
      "loss": 1.7832,
      "step": 6587
    },
    {
      "epoch": 0.23593030959586012,
      "grad_norm": 1.5805420875549316,
      "learning_rate": 0.00017857162037776173,
      "loss": 1.5009,
      "step": 6588
    },
    {
      "epoch": 0.2359661217254284,
      "grad_norm": 1.6239664554595947,
      "learning_rate": 0.0001785644448490242,
      "loss": 1.5902,
      "step": 6589
    },
    {
      "epoch": 0.23600193385499668,
      "grad_norm": 1.4822232723236084,
      "learning_rate": 0.00017855726826330334,
      "loss": 1.5981,
      "step": 6590
    },
    {
      "epoch": 0.23603774598456498,
      "grad_norm": 1.701465129852295,
      "learning_rate": 0.00017855009062069582,
      "loss": 1.4967,
      "step": 6591
    },
    {
      "epoch": 0.23607355811413325,
      "grad_norm": 2.413892984390259,
      "learning_rate": 0.00017854291192129812,
      "loss": 1.6526,
      "step": 6592
    },
    {
      "epoch": 0.23610937024370154,
      "grad_norm": 1.274664282798767,
      "learning_rate": 0.00017853573216520684,
      "loss": 1.6259,
      "step": 6593
    },
    {
      "epoch": 0.2361451823732698,
      "grad_norm": 1.5732338428497314,
      "learning_rate": 0.00017852855135251864,
      "loss": 1.2351,
      "step": 6594
    },
    {
      "epoch": 0.2361809945028381,
      "grad_norm": 1.7816098928451538,
      "learning_rate": 0.00017852136948333006,
      "loss": 1.6006,
      "step": 6595
    },
    {
      "epoch": 0.2362168066324064,
      "grad_norm": 1.5788737535476685,
      "learning_rate": 0.00017851418655773772,
      "loss": 1.5484,
      "step": 6596
    },
    {
      "epoch": 0.23625261876197468,
      "grad_norm": 1.6438406705856323,
      "learning_rate": 0.00017850700257583828,
      "loss": 1.704,
      "step": 6597
    },
    {
      "epoch": 0.23628843089154297,
      "grad_norm": 1.7204110622406006,
      "learning_rate": 0.00017849981753772836,
      "loss": 1.6303,
      "step": 6598
    },
    {
      "epoch": 0.23632424302111124,
      "grad_norm": 1.7380869388580322,
      "learning_rate": 0.0001784926314435047,
      "loss": 1.7262,
      "step": 6599
    },
    {
      "epoch": 0.23636005515067954,
      "grad_norm": 2.4021716117858887,
      "learning_rate": 0.00017848544429326392,
      "loss": 1.4997,
      "step": 6600
    },
    {
      "epoch": 0.2363958672802478,
      "grad_norm": 2.6693549156188965,
      "learning_rate": 0.00017847825608710273,
      "loss": 1.4439,
      "step": 6601
    },
    {
      "epoch": 0.2364316794098161,
      "grad_norm": 1.5793838500976562,
      "learning_rate": 0.00017847106682511782,
      "loss": 1.519,
      "step": 6602
    },
    {
      "epoch": 0.2364674915393844,
      "grad_norm": 2.942883253097534,
      "learning_rate": 0.00017846387650740592,
      "loss": 1.8072,
      "step": 6603
    },
    {
      "epoch": 0.23650330366895267,
      "grad_norm": 1.5325664281845093,
      "learning_rate": 0.00017845668513406378,
      "loss": 1.3556,
      "step": 6604
    },
    {
      "epoch": 0.23653911579852097,
      "grad_norm": 1.4211392402648926,
      "learning_rate": 0.00017844949270518816,
      "loss": 1.2079,
      "step": 6605
    },
    {
      "epoch": 0.23657492792808923,
      "grad_norm": 1.4556447267532349,
      "learning_rate": 0.00017844229922087582,
      "loss": 1.477,
      "step": 6606
    },
    {
      "epoch": 0.23661074005765753,
      "grad_norm": 2.2137489318847656,
      "learning_rate": 0.00017843510468122347,
      "loss": 1.6519,
      "step": 6607
    },
    {
      "epoch": 0.2366465521872258,
      "grad_norm": 1.6451637744903564,
      "learning_rate": 0.00017842790908632802,
      "loss": 1.4015,
      "step": 6608
    },
    {
      "epoch": 0.2366823643167941,
      "grad_norm": 1.9870620965957642,
      "learning_rate": 0.00017842071243628617,
      "loss": 1.3847,
      "step": 6609
    },
    {
      "epoch": 0.2367181764463624,
      "grad_norm": 1.9691691398620605,
      "learning_rate": 0.0001784135147311948,
      "loss": 1.5867,
      "step": 6610
    },
    {
      "epoch": 0.23675398857593066,
      "grad_norm": 1.8930854797363281,
      "learning_rate": 0.00017840631597115076,
      "loss": 1.7622,
      "step": 6611
    },
    {
      "epoch": 0.23678980070549896,
      "grad_norm": 1.7381625175476074,
      "learning_rate": 0.00017839911615625086,
      "loss": 1.5428,
      "step": 6612
    },
    {
      "epoch": 0.23682561283506723,
      "grad_norm": 1.6592036485671997,
      "learning_rate": 0.00017839191528659198,
      "loss": 1.6095,
      "step": 6613
    },
    {
      "epoch": 0.23686142496463553,
      "grad_norm": 1.5722460746765137,
      "learning_rate": 0.000178384713362271,
      "loss": 1.529,
      "step": 6614
    },
    {
      "epoch": 0.2368972370942038,
      "grad_norm": 1.8130706548690796,
      "learning_rate": 0.00017837751038338482,
      "loss": 1.619,
      "step": 6615
    },
    {
      "epoch": 0.2369330492237721,
      "grad_norm": 1.874606728553772,
      "learning_rate": 0.00017837030635003032,
      "loss": 1.675,
      "step": 6616
    },
    {
      "epoch": 0.2369688613533404,
      "grad_norm": 1.4484257698059082,
      "learning_rate": 0.00017836310126230444,
      "loss": 1.4822,
      "step": 6617
    },
    {
      "epoch": 0.23700467348290866,
      "grad_norm": 1.4785351753234863,
      "learning_rate": 0.00017835589512030413,
      "loss": 1.6452,
      "step": 6618
    },
    {
      "epoch": 0.23704048561247695,
      "grad_norm": 1.5347731113433838,
      "learning_rate": 0.00017834868792412632,
      "loss": 1.4137,
      "step": 6619
    },
    {
      "epoch": 0.23707629774204522,
      "grad_norm": 1.8363127708435059,
      "learning_rate": 0.00017834147967386797,
      "loss": 1.3594,
      "step": 6620
    },
    {
      "epoch": 0.23711210987161352,
      "grad_norm": 2.004275321960449,
      "learning_rate": 0.00017833427036962604,
      "loss": 1.6062,
      "step": 6621
    },
    {
      "epoch": 0.2371479220011818,
      "grad_norm": 1.5115212202072144,
      "learning_rate": 0.0001783270600114976,
      "loss": 1.4327,
      "step": 6622
    },
    {
      "epoch": 0.23718373413075008,
      "grad_norm": 1.6097291707992554,
      "learning_rate": 0.00017831984859957955,
      "loss": 1.4477,
      "step": 6623
    },
    {
      "epoch": 0.23721954626031838,
      "grad_norm": 1.896243929862976,
      "learning_rate": 0.00017831263613396898,
      "loss": 1.7308,
      "step": 6624
    },
    {
      "epoch": 0.23725535838988665,
      "grad_norm": 2.4240522384643555,
      "learning_rate": 0.0001783054226147629,
      "loss": 1.4856,
      "step": 6625
    },
    {
      "epoch": 0.23729117051945495,
      "grad_norm": 1.8697086572647095,
      "learning_rate": 0.0001782982080420584,
      "loss": 1.3909,
      "step": 6626
    },
    {
      "epoch": 0.23732698264902322,
      "grad_norm": 1.623146891593933,
      "learning_rate": 0.00017829099241595245,
      "loss": 1.6915,
      "step": 6627
    },
    {
      "epoch": 0.2373627947785915,
      "grad_norm": 1.2264447212219238,
      "learning_rate": 0.00017828377573654225,
      "loss": 1.7342,
      "step": 6628
    },
    {
      "epoch": 0.23739860690815978,
      "grad_norm": 1.5856890678405762,
      "learning_rate": 0.00017827655800392478,
      "loss": 1.7977,
      "step": 6629
    },
    {
      "epoch": 0.23743441903772808,
      "grad_norm": 1.787320852279663,
      "learning_rate": 0.00017826933921819723,
      "loss": 1.4515,
      "step": 6630
    },
    {
      "epoch": 0.23747023116729638,
      "grad_norm": 1.6283372640609741,
      "learning_rate": 0.00017826211937945665,
      "loss": 1.9511,
      "step": 6631
    },
    {
      "epoch": 0.23750604329686464,
      "grad_norm": 1.7102324962615967,
      "learning_rate": 0.00017825489848780022,
      "loss": 1.6524,
      "step": 6632
    },
    {
      "epoch": 0.23754185542643294,
      "grad_norm": 2.1773719787597656,
      "learning_rate": 0.00017824767654332505,
      "loss": 1.7596,
      "step": 6633
    },
    {
      "epoch": 0.2375776675560012,
      "grad_norm": 1.3473145961761475,
      "learning_rate": 0.00017824045354612836,
      "loss": 1.2803,
      "step": 6634
    },
    {
      "epoch": 0.2376134796855695,
      "grad_norm": 1.546225666999817,
      "learning_rate": 0.00017823322949630727,
      "loss": 1.5292,
      "step": 6635
    },
    {
      "epoch": 0.23764929181513778,
      "grad_norm": 1.7746154069900513,
      "learning_rate": 0.000178226004393959,
      "loss": 1.4978,
      "step": 6636
    },
    {
      "epoch": 0.23768510394470607,
      "grad_norm": 1.3727898597717285,
      "learning_rate": 0.0001782187782391807,
      "loss": 1.3965,
      "step": 6637
    },
    {
      "epoch": 0.23772091607427437,
      "grad_norm": 2.398136615753174,
      "learning_rate": 0.0001782115510320697,
      "loss": 1.827,
      "step": 6638
    },
    {
      "epoch": 0.23775672820384264,
      "grad_norm": 1.6898165941238403,
      "learning_rate": 0.00017820432277272313,
      "loss": 1.5812,
      "step": 6639
    },
    {
      "epoch": 0.23779254033341093,
      "grad_norm": 1.2213449478149414,
      "learning_rate": 0.00017819709346123826,
      "loss": 1.6483,
      "step": 6640
    },
    {
      "epoch": 0.2378283524629792,
      "grad_norm": 1.784114956855774,
      "learning_rate": 0.0001781898630977124,
      "loss": 1.5349,
      "step": 6641
    },
    {
      "epoch": 0.2378641645925475,
      "grad_norm": 1.8772135972976685,
      "learning_rate": 0.00017818263168224276,
      "loss": 1.5394,
      "step": 6642
    },
    {
      "epoch": 0.23789997672211577,
      "grad_norm": 2.150813341140747,
      "learning_rate": 0.0001781753992149267,
      "loss": 1.554,
      "step": 6643
    },
    {
      "epoch": 0.23793578885168407,
      "grad_norm": 1.4352753162384033,
      "learning_rate": 0.00017816816569586144,
      "loss": 1.7107,
      "step": 6644
    },
    {
      "epoch": 0.23797160098125236,
      "grad_norm": 1.6923801898956299,
      "learning_rate": 0.00017816093112514437,
      "loss": 1.5776,
      "step": 6645
    },
    {
      "epoch": 0.23800741311082063,
      "grad_norm": 1.5510084629058838,
      "learning_rate": 0.00017815369550287278,
      "loss": 1.5203,
      "step": 6646
    },
    {
      "epoch": 0.23804322524038893,
      "grad_norm": 1.1947929859161377,
      "learning_rate": 0.00017814645882914402,
      "loss": 1.6705,
      "step": 6647
    },
    {
      "epoch": 0.2380790373699572,
      "grad_norm": 1.9524680376052856,
      "learning_rate": 0.00017813922110405548,
      "loss": 1.7717,
      "step": 6648
    },
    {
      "epoch": 0.2381148494995255,
      "grad_norm": 1.318559169769287,
      "learning_rate": 0.00017813198232770447,
      "loss": 1.445,
      "step": 6649
    },
    {
      "epoch": 0.23815066162909376,
      "grad_norm": 1.6523274183273315,
      "learning_rate": 0.00017812474250018844,
      "loss": 1.4655,
      "step": 6650
    },
    {
      "epoch": 0.23818647375866206,
      "grad_norm": 1.3412474393844604,
      "learning_rate": 0.00017811750162160478,
      "loss": 1.4733,
      "step": 6651
    },
    {
      "epoch": 0.23822228588823036,
      "grad_norm": 1.291676640510559,
      "learning_rate": 0.00017811025969205092,
      "loss": 1.3368,
      "step": 6652
    },
    {
      "epoch": 0.23825809801779863,
      "grad_norm": 1.5550686120986938,
      "learning_rate": 0.00017810301671162426,
      "loss": 1.7387,
      "step": 6653
    },
    {
      "epoch": 0.23829391014736692,
      "grad_norm": 1.3425484895706177,
      "learning_rate": 0.00017809577268042224,
      "loss": 1.5973,
      "step": 6654
    },
    {
      "epoch": 0.2383297222769352,
      "grad_norm": 1.645659327507019,
      "learning_rate": 0.00017808852759854235,
      "loss": 1.4145,
      "step": 6655
    },
    {
      "epoch": 0.2383655344065035,
      "grad_norm": 1.3277592658996582,
      "learning_rate": 0.00017808128146608204,
      "loss": 1.4578,
      "step": 6656
    },
    {
      "epoch": 0.23840134653607176,
      "grad_norm": 1.8686926364898682,
      "learning_rate": 0.0001780740342831388,
      "loss": 1.5942,
      "step": 6657
    },
    {
      "epoch": 0.23843715866564005,
      "grad_norm": 1.322493553161621,
      "learning_rate": 0.00017806678604981012,
      "loss": 1.2065,
      "step": 6658
    },
    {
      "epoch": 0.23847297079520835,
      "grad_norm": 1.6285942792892456,
      "learning_rate": 0.00017805953676619356,
      "loss": 1.6118,
      "step": 6659
    },
    {
      "epoch": 0.23850878292477662,
      "grad_norm": 1.799084186553955,
      "learning_rate": 0.00017805228643238662,
      "loss": 1.5807,
      "step": 6660
    },
    {
      "epoch": 0.23854459505434492,
      "grad_norm": 2.529820680618286,
      "learning_rate": 0.00017804503504848684,
      "loss": 1.4212,
      "step": 6661
    },
    {
      "epoch": 0.23858040718391318,
      "grad_norm": 2.7669854164123535,
      "learning_rate": 0.00017803778261459181,
      "loss": 1.5821,
      "step": 6662
    },
    {
      "epoch": 0.23861621931348148,
      "grad_norm": 1.924968957901001,
      "learning_rate": 0.00017803052913079905,
      "loss": 1.3714,
      "step": 6663
    },
    {
      "epoch": 0.23865203144304975,
      "grad_norm": 1.2607649564743042,
      "learning_rate": 0.0001780232745972062,
      "loss": 1.3695,
      "step": 6664
    },
    {
      "epoch": 0.23868784357261805,
      "grad_norm": 1.7694482803344727,
      "learning_rate": 0.00017801601901391078,
      "loss": 1.433,
      "step": 6665
    },
    {
      "epoch": 0.23872365570218634,
      "grad_norm": 1.6368229389190674,
      "learning_rate": 0.0001780087623810105,
      "loss": 1.8103,
      "step": 6666
    },
    {
      "epoch": 0.2387594678317546,
      "grad_norm": 1.6224684715270996,
      "learning_rate": 0.00017800150469860293,
      "loss": 1.7134,
      "step": 6667
    },
    {
      "epoch": 0.2387952799613229,
      "grad_norm": 1.4222160577774048,
      "learning_rate": 0.00017799424596678573,
      "loss": 1.5797,
      "step": 6668
    },
    {
      "epoch": 0.23883109209089118,
      "grad_norm": 1.7973840236663818,
      "learning_rate": 0.00017798698618565653,
      "loss": 1.552,
      "step": 6669
    },
    {
      "epoch": 0.23886690422045948,
      "grad_norm": 1.7325069904327393,
      "learning_rate": 0.0001779797253553131,
      "loss": 1.9692,
      "step": 6670
    },
    {
      "epoch": 0.23890271635002774,
      "grad_norm": 1.5968601703643799,
      "learning_rate": 0.000177972463475853,
      "loss": 1.68,
      "step": 6671
    },
    {
      "epoch": 0.23893852847959604,
      "grad_norm": 1.8423383235931396,
      "learning_rate": 0.000177965200547374,
      "loss": 1.8313,
      "step": 6672
    },
    {
      "epoch": 0.23897434060916434,
      "grad_norm": 2.4150888919830322,
      "learning_rate": 0.00017795793656997377,
      "loss": 1.3579,
      "step": 6673
    },
    {
      "epoch": 0.2390101527387326,
      "grad_norm": 1.754241943359375,
      "learning_rate": 0.00017795067154375007,
      "loss": 1.4763,
      "step": 6674
    },
    {
      "epoch": 0.2390459648683009,
      "grad_norm": 1.7057687044143677,
      "learning_rate": 0.00017794340546880064,
      "loss": 1.5102,
      "step": 6675
    },
    {
      "epoch": 0.23908177699786917,
      "grad_norm": 1.922208309173584,
      "learning_rate": 0.00017793613834522326,
      "loss": 1.4778,
      "step": 6676
    },
    {
      "epoch": 0.23911758912743747,
      "grad_norm": 1.225993037223816,
      "learning_rate": 0.0001779288701731156,
      "loss": 1.5574,
      "step": 6677
    },
    {
      "epoch": 0.23915340125700574,
      "grad_norm": 2.4143288135528564,
      "learning_rate": 0.00017792160095257556,
      "loss": 1.6166,
      "step": 6678
    },
    {
      "epoch": 0.23918921338657403,
      "grad_norm": 1.4441640377044678,
      "learning_rate": 0.00017791433068370087,
      "loss": 1.5373,
      "step": 6679
    },
    {
      "epoch": 0.23922502551614233,
      "grad_norm": 1.7802202701568604,
      "learning_rate": 0.00017790705936658938,
      "loss": 1.1986,
      "step": 6680
    },
    {
      "epoch": 0.2392608376457106,
      "grad_norm": 2.0310826301574707,
      "learning_rate": 0.00017789978700133888,
      "loss": 1.8544,
      "step": 6681
    },
    {
      "epoch": 0.2392966497752789,
      "grad_norm": 1.6153504848480225,
      "learning_rate": 0.00017789251358804725,
      "loss": 1.5214,
      "step": 6682
    },
    {
      "epoch": 0.23933246190484717,
      "grad_norm": 1.5683387517929077,
      "learning_rate": 0.00017788523912681231,
      "loss": 1.5966,
      "step": 6683
    },
    {
      "epoch": 0.23936827403441546,
      "grad_norm": 2.497631072998047,
      "learning_rate": 0.00017787796361773197,
      "loss": 1.4201,
      "step": 6684
    },
    {
      "epoch": 0.23940408616398373,
      "grad_norm": 1.5542875528335571,
      "learning_rate": 0.00017787068706090405,
      "loss": 1.6915,
      "step": 6685
    },
    {
      "epoch": 0.23943989829355203,
      "grad_norm": 1.6513056755065918,
      "learning_rate": 0.0001778634094564265,
      "loss": 1.4809,
      "step": 6686
    },
    {
      "epoch": 0.2394757104231203,
      "grad_norm": 2.319539785385132,
      "learning_rate": 0.0001778561308043972,
      "loss": 1.4589,
      "step": 6687
    },
    {
      "epoch": 0.2395115225526886,
      "grad_norm": 1.8904774188995361,
      "learning_rate": 0.00017784885110491412,
      "loss": 1.6562,
      "step": 6688
    },
    {
      "epoch": 0.2395473346822569,
      "grad_norm": 1.582964539527893,
      "learning_rate": 0.00017784157035807515,
      "loss": 1.6561,
      "step": 6689
    },
    {
      "epoch": 0.23958314681182516,
      "grad_norm": 1.625274896621704,
      "learning_rate": 0.00017783428856397825,
      "loss": 1.4374,
      "step": 6690
    },
    {
      "epoch": 0.23961895894139346,
      "grad_norm": 1.5694506168365479,
      "learning_rate": 0.00017782700572272137,
      "loss": 1.6288,
      "step": 6691
    },
    {
      "epoch": 0.23965477107096173,
      "grad_norm": 2.2098121643066406,
      "learning_rate": 0.00017781972183440254,
      "loss": 1.6151,
      "step": 6692
    },
    {
      "epoch": 0.23969058320053002,
      "grad_norm": 2.150939464569092,
      "learning_rate": 0.00017781243689911973,
      "loss": 1.6039,
      "step": 6693
    },
    {
      "epoch": 0.2397263953300983,
      "grad_norm": 1.4807977676391602,
      "learning_rate": 0.00017780515091697096,
      "loss": 1.7675,
      "step": 6694
    },
    {
      "epoch": 0.2397622074596666,
      "grad_norm": 1.5989789962768555,
      "learning_rate": 0.00017779786388805424,
      "loss": 1.3923,
      "step": 6695
    },
    {
      "epoch": 0.23979801958923488,
      "grad_norm": 1.3950860500335693,
      "learning_rate": 0.00017779057581246763,
      "loss": 1.4481,
      "step": 6696
    },
    {
      "epoch": 0.23983383171880315,
      "grad_norm": 1.7480555772781372,
      "learning_rate": 0.00017778328669030918,
      "loss": 2.0431,
      "step": 6697
    },
    {
      "epoch": 0.23986964384837145,
      "grad_norm": 1.7169839143753052,
      "learning_rate": 0.0001777759965216769,
      "loss": 1.5416,
      "step": 6698
    },
    {
      "epoch": 0.23990545597793972,
      "grad_norm": 1.6872042417526245,
      "learning_rate": 0.0001777687053066689,
      "loss": 1.726,
      "step": 6699
    },
    {
      "epoch": 0.23994126810750802,
      "grad_norm": 1.3428165912628174,
      "learning_rate": 0.00017776141304538332,
      "loss": 1.4027,
      "step": 6700
    },
    {
      "epoch": 0.23997708023707628,
      "grad_norm": 1.719918131828308,
      "learning_rate": 0.00017775411973791822,
      "loss": 1.6605,
      "step": 6701
    },
    {
      "epoch": 0.24001289236664458,
      "grad_norm": 1.745969533920288,
      "learning_rate": 0.00017774682538437175,
      "loss": 1.4542,
      "step": 6702
    },
    {
      "epoch": 0.24004870449621288,
      "grad_norm": 1.530691146850586,
      "learning_rate": 0.00017773952998484204,
      "loss": 1.5051,
      "step": 6703
    },
    {
      "epoch": 0.24008451662578115,
      "grad_norm": 1.7143809795379639,
      "learning_rate": 0.0001777322335394272,
      "loss": 1.7822,
      "step": 6704
    },
    {
      "epoch": 0.24012032875534944,
      "grad_norm": 2.3987367153167725,
      "learning_rate": 0.00017772493604822543,
      "loss": 1.5643,
      "step": 6705
    },
    {
      "epoch": 0.2401561408849177,
      "grad_norm": 1.9206353425979614,
      "learning_rate": 0.00017771763751133488,
      "loss": 1.6011,
      "step": 6706
    },
    {
      "epoch": 0.240191953014486,
      "grad_norm": 1.3068255186080933,
      "learning_rate": 0.0001777103379288538,
      "loss": 1.3736,
      "step": 6707
    },
    {
      "epoch": 0.24022776514405428,
      "grad_norm": 1.405949592590332,
      "learning_rate": 0.00017770303730088035,
      "loss": 1.5704,
      "step": 6708
    },
    {
      "epoch": 0.24026357727362257,
      "grad_norm": 2.038421392440796,
      "learning_rate": 0.00017769573562751275,
      "loss": 1.5038,
      "step": 6709
    },
    {
      "epoch": 0.24029938940319087,
      "grad_norm": 1.7716072797775269,
      "learning_rate": 0.0001776884329088493,
      "loss": 1.7277,
      "step": 6710
    },
    {
      "epoch": 0.24033520153275914,
      "grad_norm": 1.2767550945281982,
      "learning_rate": 0.00017768112914498817,
      "loss": 1.2572,
      "step": 6711
    },
    {
      "epoch": 0.24037101366232744,
      "grad_norm": 2.466627836227417,
      "learning_rate": 0.00017767382433602762,
      "loss": 1.4945,
      "step": 6712
    },
    {
      "epoch": 0.2404068257918957,
      "grad_norm": 1.978790521621704,
      "learning_rate": 0.00017766651848206597,
      "loss": 1.8079,
      "step": 6713
    },
    {
      "epoch": 0.240442637921464,
      "grad_norm": 1.4621291160583496,
      "learning_rate": 0.00017765921158320152,
      "loss": 1.6702,
      "step": 6714
    },
    {
      "epoch": 0.24047845005103227,
      "grad_norm": 1.7895176410675049,
      "learning_rate": 0.00017765190363953253,
      "loss": 1.2385,
      "step": 6715
    },
    {
      "epoch": 0.24051426218060057,
      "grad_norm": 1.5651522874832153,
      "learning_rate": 0.00017764459465115736,
      "loss": 1.6956,
      "step": 6716
    },
    {
      "epoch": 0.24055007431016887,
      "grad_norm": 1.6989222764968872,
      "learning_rate": 0.0001776372846181743,
      "loss": 1.5337,
      "step": 6717
    },
    {
      "epoch": 0.24058588643973713,
      "grad_norm": 2.1151821613311768,
      "learning_rate": 0.00017762997354068172,
      "loss": 1.8988,
      "step": 6718
    },
    {
      "epoch": 0.24062169856930543,
      "grad_norm": 2.0804238319396973,
      "learning_rate": 0.00017762266141877796,
      "loss": 1.4169,
      "step": 6719
    },
    {
      "epoch": 0.2406575106988737,
      "grad_norm": 1.9376277923583984,
      "learning_rate": 0.00017761534825256144,
      "loss": 1.5997,
      "step": 6720
    },
    {
      "epoch": 0.240693322828442,
      "grad_norm": 1.5434688329696655,
      "learning_rate": 0.00017760803404213052,
      "loss": 1.0933,
      "step": 6721
    },
    {
      "epoch": 0.24072913495801027,
      "grad_norm": 2.0929014682769775,
      "learning_rate": 0.00017760071878758363,
      "loss": 1.4231,
      "step": 6722
    },
    {
      "epoch": 0.24076494708757856,
      "grad_norm": 2.1523287296295166,
      "learning_rate": 0.00017759340248901917,
      "loss": 1.7028,
      "step": 6723
    },
    {
      "epoch": 0.24080075921714686,
      "grad_norm": 1.568102240562439,
      "learning_rate": 0.00017758608514653555,
      "loss": 1.4582,
      "step": 6724
    },
    {
      "epoch": 0.24083657134671513,
      "grad_norm": 1.5287367105484009,
      "learning_rate": 0.00017757876676023125,
      "loss": 1.503,
      "step": 6725
    },
    {
      "epoch": 0.24087238347628342,
      "grad_norm": 1.5015430450439453,
      "learning_rate": 0.0001775714473302047,
      "loss": 1.4269,
      "step": 6726
    },
    {
      "epoch": 0.2409081956058517,
      "grad_norm": 1.2125335931777954,
      "learning_rate": 0.0001775641268565544,
      "loss": 1.5207,
      "step": 6727
    },
    {
      "epoch": 0.24094400773542,
      "grad_norm": 2.623624086380005,
      "learning_rate": 0.0001775568053393788,
      "loss": 1.7906,
      "step": 6728
    },
    {
      "epoch": 0.24097981986498826,
      "grad_norm": 1.626744270324707,
      "learning_rate": 0.00017754948277877642,
      "loss": 1.6469,
      "step": 6729
    },
    {
      "epoch": 0.24101563199455656,
      "grad_norm": 1.9941731691360474,
      "learning_rate": 0.0001775421591748458,
      "loss": 1.5711,
      "step": 6730
    },
    {
      "epoch": 0.24105144412412485,
      "grad_norm": 1.4977132081985474,
      "learning_rate": 0.00017753483452768545,
      "loss": 1.5056,
      "step": 6731
    },
    {
      "epoch": 0.24108725625369312,
      "grad_norm": 1.9533582925796509,
      "learning_rate": 0.0001775275088373939,
      "loss": 1.5436,
      "step": 6732
    },
    {
      "epoch": 0.24112306838326142,
      "grad_norm": 2.0875680446624756,
      "learning_rate": 0.00017752018210406972,
      "loss": 1.8286,
      "step": 6733
    },
    {
      "epoch": 0.2411588805128297,
      "grad_norm": 1.3487203121185303,
      "learning_rate": 0.00017751285432781152,
      "loss": 1.483,
      "step": 6734
    },
    {
      "epoch": 0.24119469264239798,
      "grad_norm": 1.8331043720245361,
      "learning_rate": 0.00017750552550871782,
      "loss": 1.6495,
      "step": 6735
    },
    {
      "epoch": 0.24123050477196625,
      "grad_norm": 1.6515965461730957,
      "learning_rate": 0.00017749819564688725,
      "loss": 1.3278,
      "step": 6736
    },
    {
      "epoch": 0.24126631690153455,
      "grad_norm": 1.197119116783142,
      "learning_rate": 0.00017749086474241844,
      "loss": 1.5107,
      "step": 6737
    },
    {
      "epoch": 0.24130212903110285,
      "grad_norm": 2.1350796222686768,
      "learning_rate": 0.00017748353279540999,
      "loss": 1.705,
      "step": 6738
    },
    {
      "epoch": 0.24133794116067112,
      "grad_norm": 1.517443060874939,
      "learning_rate": 0.00017747619980596055,
      "loss": 1.6448,
      "step": 6739
    },
    {
      "epoch": 0.2413737532902394,
      "grad_norm": 1.9738852977752686,
      "learning_rate": 0.00017746886577416876,
      "loss": 1.4471,
      "step": 6740
    },
    {
      "epoch": 0.24140956541980768,
      "grad_norm": 1.6398792266845703,
      "learning_rate": 0.00017746153070013335,
      "loss": 1.7618,
      "step": 6741
    },
    {
      "epoch": 0.24144537754937598,
      "grad_norm": 1.5546917915344238,
      "learning_rate": 0.00017745419458395294,
      "loss": 1.7391,
      "step": 6742
    },
    {
      "epoch": 0.24148118967894425,
      "grad_norm": 1.7060317993164062,
      "learning_rate": 0.00017744685742572625,
      "loss": 1.9142,
      "step": 6743
    },
    {
      "epoch": 0.24151700180851254,
      "grad_norm": 1.6105141639709473,
      "learning_rate": 0.000177439519225552,
      "loss": 1.5765,
      "step": 6744
    },
    {
      "epoch": 0.24155281393808084,
      "grad_norm": 1.5093984603881836,
      "learning_rate": 0.0001774321799835289,
      "loss": 1.6196,
      "step": 6745
    },
    {
      "epoch": 0.2415886260676491,
      "grad_norm": 1.6497421264648438,
      "learning_rate": 0.00017742483969975572,
      "loss": 1.575,
      "step": 6746
    },
    {
      "epoch": 0.2416244381972174,
      "grad_norm": 1.5175936222076416,
      "learning_rate": 0.00017741749837433117,
      "loss": 1.6146,
      "step": 6747
    },
    {
      "epoch": 0.24166025032678567,
      "grad_norm": 1.6494495868682861,
      "learning_rate": 0.00017741015600735403,
      "loss": 1.4422,
      "step": 6748
    },
    {
      "epoch": 0.24169606245635397,
      "grad_norm": 1.5587164163589478,
      "learning_rate": 0.0001774028125989231,
      "loss": 1.7001,
      "step": 6749
    },
    {
      "epoch": 0.24173187458592224,
      "grad_norm": 2.176522970199585,
      "learning_rate": 0.00017739546814913722,
      "loss": 1.6531,
      "step": 6750
    },
    {
      "epoch": 0.24176768671549054,
      "grad_norm": 1.236741065979004,
      "learning_rate": 0.00017738812265809508,
      "loss": 1.5294,
      "step": 6751
    },
    {
      "epoch": 0.24180349884505883,
      "grad_norm": 1.8967229127883911,
      "learning_rate": 0.0001773807761258956,
      "loss": 1.4595,
      "step": 6752
    },
    {
      "epoch": 0.2418393109746271,
      "grad_norm": 2.09911847114563,
      "learning_rate": 0.0001773734285526376,
      "loss": 1.4782,
      "step": 6753
    },
    {
      "epoch": 0.2418751231041954,
      "grad_norm": 1.3407038450241089,
      "learning_rate": 0.0001773660799384199,
      "loss": 1.4959,
      "step": 6754
    },
    {
      "epoch": 0.24191093523376367,
      "grad_norm": 1.8099561929702759,
      "learning_rate": 0.0001773587302833414,
      "loss": 1.2301,
      "step": 6755
    },
    {
      "epoch": 0.24194674736333197,
      "grad_norm": 1.691125750541687,
      "learning_rate": 0.000177351379587501,
      "loss": 1.3265,
      "step": 6756
    },
    {
      "epoch": 0.24198255949290023,
      "grad_norm": 1.1911656856536865,
      "learning_rate": 0.0001773440278509975,
      "loss": 1.7656,
      "step": 6757
    },
    {
      "epoch": 0.24201837162246853,
      "grad_norm": 1.8270931243896484,
      "learning_rate": 0.00017733667507392991,
      "loss": 1.5717,
      "step": 6758
    },
    {
      "epoch": 0.24205418375203683,
      "grad_norm": 2.1644065380096436,
      "learning_rate": 0.00017732932125639713,
      "loss": 1.6089,
      "step": 6759
    },
    {
      "epoch": 0.2420899958816051,
      "grad_norm": 1.8169310092926025,
      "learning_rate": 0.00017732196639849804,
      "loss": 1.6283,
      "step": 6760
    },
    {
      "epoch": 0.2421258080111734,
      "grad_norm": 1.8802064657211304,
      "learning_rate": 0.0001773146105003317,
      "loss": 1.5327,
      "step": 6761
    },
    {
      "epoch": 0.24216162014074166,
      "grad_norm": 1.2530100345611572,
      "learning_rate": 0.00017730725356199692,
      "loss": 1.553,
      "step": 6762
    },
    {
      "epoch": 0.24219743227030996,
      "grad_norm": 1.9466575384140015,
      "learning_rate": 0.0001772998955835928,
      "loss": 1.7341,
      "step": 6763
    },
    {
      "epoch": 0.24223324439987823,
      "grad_norm": 1.4589053392410278,
      "learning_rate": 0.00017729253656521832,
      "loss": 1.4459,
      "step": 6764
    },
    {
      "epoch": 0.24226905652944652,
      "grad_norm": 1.6372430324554443,
      "learning_rate": 0.00017728517650697243,
      "loss": 1.6402,
      "step": 6765
    },
    {
      "epoch": 0.24230486865901482,
      "grad_norm": 1.8554078340530396,
      "learning_rate": 0.0001772778154089542,
      "loss": 1.511,
      "step": 6766
    },
    {
      "epoch": 0.2423406807885831,
      "grad_norm": 1.3996329307556152,
      "learning_rate": 0.0001772704532712626,
      "loss": 1.6167,
      "step": 6767
    },
    {
      "epoch": 0.2423764929181514,
      "grad_norm": 1.9406613111495972,
      "learning_rate": 0.00017726309009399676,
      "loss": 1.6832,
      "step": 6768
    },
    {
      "epoch": 0.24241230504771966,
      "grad_norm": 2.206716537475586,
      "learning_rate": 0.0001772557258772557,
      "loss": 1.6634,
      "step": 6769
    },
    {
      "epoch": 0.24244811717728795,
      "grad_norm": 1.4321078062057495,
      "learning_rate": 0.0001772483606211385,
      "loss": 1.4673,
      "step": 6770
    },
    {
      "epoch": 0.24248392930685622,
      "grad_norm": 1.2396024465560913,
      "learning_rate": 0.00017724099432574425,
      "loss": 1.5565,
      "step": 6771
    },
    {
      "epoch": 0.24251974143642452,
      "grad_norm": 1.692460060119629,
      "learning_rate": 0.00017723362699117206,
      "loss": 1.8142,
      "step": 6772
    },
    {
      "epoch": 0.24255555356599282,
      "grad_norm": 1.9822334051132202,
      "learning_rate": 0.00017722625861752103,
      "loss": 1.6546,
      "step": 6773
    },
    {
      "epoch": 0.24259136569556108,
      "grad_norm": 1.494781255722046,
      "learning_rate": 0.0001772188892048903,
      "loss": 1.4441,
      "step": 6774
    },
    {
      "epoch": 0.24262717782512938,
      "grad_norm": 2.4597997665405273,
      "learning_rate": 0.00017721151875337907,
      "loss": 2.0706,
      "step": 6775
    },
    {
      "epoch": 0.24266298995469765,
      "grad_norm": 1.3796907663345337,
      "learning_rate": 0.00017720414726308642,
      "loss": 1.5801,
      "step": 6776
    },
    {
      "epoch": 0.24269880208426595,
      "grad_norm": 1.4934757947921753,
      "learning_rate": 0.00017719677473411154,
      "loss": 1.3307,
      "step": 6777
    },
    {
      "epoch": 0.24273461421383422,
      "grad_norm": 1.4481245279312134,
      "learning_rate": 0.00017718940116655363,
      "loss": 1.7379,
      "step": 6778
    },
    {
      "epoch": 0.2427704263434025,
      "grad_norm": 1.3896671533584595,
      "learning_rate": 0.00017718202656051194,
      "loss": 1.7022,
      "step": 6779
    },
    {
      "epoch": 0.2428062384729708,
      "grad_norm": 1.709444284439087,
      "learning_rate": 0.0001771746509160856,
      "loss": 1.4188,
      "step": 6780
    },
    {
      "epoch": 0.24284205060253908,
      "grad_norm": 1.5271943807601929,
      "learning_rate": 0.00017716727423337388,
      "loss": 1.4017,
      "step": 6781
    },
    {
      "epoch": 0.24287786273210737,
      "grad_norm": 1.7868486642837524,
      "learning_rate": 0.00017715989651247602,
      "loss": 1.6336,
      "step": 6782
    },
    {
      "epoch": 0.24291367486167564,
      "grad_norm": 2.2249338626861572,
      "learning_rate": 0.0001771525177534913,
      "loss": 1.4794,
      "step": 6783
    },
    {
      "epoch": 0.24294948699124394,
      "grad_norm": 1.2865608930587769,
      "learning_rate": 0.00017714513795651898,
      "loss": 1.4339,
      "step": 6784
    },
    {
      "epoch": 0.2429852991208122,
      "grad_norm": 1.402057409286499,
      "learning_rate": 0.00017713775712165832,
      "loss": 1.705,
      "step": 6785
    },
    {
      "epoch": 0.2430211112503805,
      "grad_norm": 1.5345312356948853,
      "learning_rate": 0.00017713037524900863,
      "loss": 1.6735,
      "step": 6786
    },
    {
      "epoch": 0.24305692337994877,
      "grad_norm": 1.4884132146835327,
      "learning_rate": 0.00017712299233866923,
      "loss": 1.5185,
      "step": 6787
    },
    {
      "epoch": 0.24309273550951707,
      "grad_norm": 2.092890977859497,
      "learning_rate": 0.0001771156083907395,
      "loss": 1.8618,
      "step": 6788
    },
    {
      "epoch": 0.24312854763908537,
      "grad_norm": 1.263377070426941,
      "learning_rate": 0.0001771082234053187,
      "loss": 1.5741,
      "step": 6789
    },
    {
      "epoch": 0.24316435976865364,
      "grad_norm": 1.8555021286010742,
      "learning_rate": 0.0001771008373825062,
      "loss": 1.4518,
      "step": 6790
    },
    {
      "epoch": 0.24320017189822193,
      "grad_norm": 1.67750084400177,
      "learning_rate": 0.0001770934503224014,
      "loss": 1.3955,
      "step": 6791
    },
    {
      "epoch": 0.2432359840277902,
      "grad_norm": 1.6171958446502686,
      "learning_rate": 0.00017708606222510367,
      "loss": 1.3687,
      "step": 6792
    },
    {
      "epoch": 0.2432717961573585,
      "grad_norm": 2.1286425590515137,
      "learning_rate": 0.0001770786730907124,
      "loss": 1.5347,
      "step": 6793
    },
    {
      "epoch": 0.24330760828692677,
      "grad_norm": 1.4012260437011719,
      "learning_rate": 0.00017707128291932702,
      "loss": 1.4511,
      "step": 6794
    },
    {
      "epoch": 0.24334342041649507,
      "grad_norm": 2.556178569793701,
      "learning_rate": 0.00017706389171104694,
      "loss": 1.8041,
      "step": 6795
    },
    {
      "epoch": 0.24337923254606336,
      "grad_norm": 2.018625259399414,
      "learning_rate": 0.00017705649946597157,
      "loss": 1.6411,
      "step": 6796
    },
    {
      "epoch": 0.24341504467563163,
      "grad_norm": 1.3456376791000366,
      "learning_rate": 0.00017704910618420044,
      "loss": 1.537,
      "step": 6797
    },
    {
      "epoch": 0.24345085680519993,
      "grad_norm": 1.720267653465271,
      "learning_rate": 0.00017704171186583295,
      "loss": 1.6337,
      "step": 6798
    },
    {
      "epoch": 0.2434866689347682,
      "grad_norm": 1.793156623840332,
      "learning_rate": 0.00017703431651096862,
      "loss": 1.9494,
      "step": 6799
    },
    {
      "epoch": 0.2435224810643365,
      "grad_norm": 2.458571195602417,
      "learning_rate": 0.00017702692011970693,
      "loss": 1.347,
      "step": 6800
    },
    {
      "epoch": 0.24355829319390476,
      "grad_norm": 2.3884029388427734,
      "learning_rate": 0.00017701952269214737,
      "loss": 1.5462,
      "step": 6801
    },
    {
      "epoch": 0.24359410532347306,
      "grad_norm": 2.0176198482513428,
      "learning_rate": 0.00017701212422838948,
      "loss": 1.5606,
      "step": 6802
    },
    {
      "epoch": 0.24362991745304136,
      "grad_norm": 2.216850519180298,
      "learning_rate": 0.00017700472472853283,
      "loss": 1.7962,
      "step": 6803
    },
    {
      "epoch": 0.24366572958260962,
      "grad_norm": 1.417450189590454,
      "learning_rate": 0.00017699732419267688,
      "loss": 1.4538,
      "step": 6804
    },
    {
      "epoch": 0.24370154171217792,
      "grad_norm": 2.4908945560455322,
      "learning_rate": 0.0001769899226209213,
      "loss": 1.3105,
      "step": 6805
    },
    {
      "epoch": 0.2437373538417462,
      "grad_norm": 1.8872681856155396,
      "learning_rate": 0.0001769825200133656,
      "loss": 1.6064,
      "step": 6806
    },
    {
      "epoch": 0.2437731659713145,
      "grad_norm": 1.579154133796692,
      "learning_rate": 0.00017697511637010938,
      "loss": 1.2872,
      "step": 6807
    },
    {
      "epoch": 0.24380897810088276,
      "grad_norm": 2.0289571285247803,
      "learning_rate": 0.0001769677116912523,
      "loss": 1.4756,
      "step": 6808
    },
    {
      "epoch": 0.24384479023045105,
      "grad_norm": 1.60190749168396,
      "learning_rate": 0.00017696030597689393,
      "loss": 1.501,
      "step": 6809
    },
    {
      "epoch": 0.24388060236001935,
      "grad_norm": 1.8626149892807007,
      "learning_rate": 0.00017695289922713389,
      "loss": 1.4009,
      "step": 6810
    },
    {
      "epoch": 0.24391641448958762,
      "grad_norm": 1.8006678819656372,
      "learning_rate": 0.00017694549144207185,
      "loss": 1.6877,
      "step": 6811
    },
    {
      "epoch": 0.24395222661915592,
      "grad_norm": 1.565643310546875,
      "learning_rate": 0.0001769380826218075,
      "loss": 1.6743,
      "step": 6812
    },
    {
      "epoch": 0.24398803874872418,
      "grad_norm": 1.7469196319580078,
      "learning_rate": 0.00017693067276644049,
      "loss": 1.6445,
      "step": 6813
    },
    {
      "epoch": 0.24402385087829248,
      "grad_norm": 1.8169236183166504,
      "learning_rate": 0.00017692326187607052,
      "loss": 1.4561,
      "step": 6814
    },
    {
      "epoch": 0.24405966300786075,
      "grad_norm": 1.2375288009643555,
      "learning_rate": 0.00017691584995079725,
      "loss": 1.2383,
      "step": 6815
    },
    {
      "epoch": 0.24409547513742905,
      "grad_norm": 1.8225083351135254,
      "learning_rate": 0.00017690843699072045,
      "loss": 1.6827,
      "step": 6816
    },
    {
      "epoch": 0.24413128726699734,
      "grad_norm": 1.7936789989471436,
      "learning_rate": 0.00017690102299593985,
      "loss": 1.7272,
      "step": 6817
    },
    {
      "epoch": 0.2441670993965656,
      "grad_norm": 1.4096508026123047,
      "learning_rate": 0.00017689360796655515,
      "loss": 1.5091,
      "step": 6818
    },
    {
      "epoch": 0.2442029115261339,
      "grad_norm": 1.3659213781356812,
      "learning_rate": 0.00017688619190266616,
      "loss": 1.6082,
      "step": 6819
    },
    {
      "epoch": 0.24423872365570218,
      "grad_norm": 1.6657557487487793,
      "learning_rate": 0.00017687877480437262,
      "loss": 1.5673,
      "step": 6820
    },
    {
      "epoch": 0.24427453578527047,
      "grad_norm": 1.459875226020813,
      "learning_rate": 0.00017687135667177436,
      "loss": 1.6767,
      "step": 6821
    },
    {
      "epoch": 0.24431034791483874,
      "grad_norm": 1.7852752208709717,
      "learning_rate": 0.00017686393750497112,
      "loss": 1.5721,
      "step": 6822
    },
    {
      "epoch": 0.24434616004440704,
      "grad_norm": 1.7470086812973022,
      "learning_rate": 0.0001768565173040628,
      "loss": 1.3168,
      "step": 6823
    },
    {
      "epoch": 0.24438197217397534,
      "grad_norm": 2.435948371887207,
      "learning_rate": 0.0001768490960691491,
      "loss": 1.7345,
      "step": 6824
    },
    {
      "epoch": 0.2444177843035436,
      "grad_norm": 1.1866915225982666,
      "learning_rate": 0.00017684167380033002,
      "loss": 1.6174,
      "step": 6825
    },
    {
      "epoch": 0.2444535964331119,
      "grad_norm": 1.7603780031204224,
      "learning_rate": 0.00017683425049770527,
      "loss": 1.6618,
      "step": 6826
    },
    {
      "epoch": 0.24448940856268017,
      "grad_norm": 1.6207976341247559,
      "learning_rate": 0.00017682682616137484,
      "loss": 1.5455,
      "step": 6827
    },
    {
      "epoch": 0.24452522069224847,
      "grad_norm": 1.5510258674621582,
      "learning_rate": 0.00017681940079143855,
      "loss": 1.5884,
      "step": 6828
    },
    {
      "epoch": 0.24456103282181674,
      "grad_norm": 1.862894892692566,
      "learning_rate": 0.0001768119743879963,
      "loss": 1.4781,
      "step": 6829
    },
    {
      "epoch": 0.24459684495138503,
      "grad_norm": 1.6464170217514038,
      "learning_rate": 0.00017680454695114802,
      "loss": 1.7318,
      "step": 6830
    },
    {
      "epoch": 0.24463265708095333,
      "grad_norm": 1.6220334768295288,
      "learning_rate": 0.00017679711848099362,
      "loss": 1.4225,
      "step": 6831
    },
    {
      "epoch": 0.2446684692105216,
      "grad_norm": 1.9520400762557983,
      "learning_rate": 0.0001767896889776331,
      "loss": 1.8062,
      "step": 6832
    },
    {
      "epoch": 0.2447042813400899,
      "grad_norm": 1.7717475891113281,
      "learning_rate": 0.00017678225844116628,
      "loss": 1.5246,
      "step": 6833
    },
    {
      "epoch": 0.24474009346965817,
      "grad_norm": 1.6127135753631592,
      "learning_rate": 0.00017677482687169328,
      "loss": 1.6002,
      "step": 6834
    },
    {
      "epoch": 0.24477590559922646,
      "grad_norm": 1.3745635747909546,
      "learning_rate": 0.000176767394269314,
      "loss": 1.6012,
      "step": 6835
    },
    {
      "epoch": 0.24481171772879473,
      "grad_norm": 1.3935896158218384,
      "learning_rate": 0.00017675996063412844,
      "loss": 1.5386,
      "step": 6836
    },
    {
      "epoch": 0.24484752985836303,
      "grad_norm": 1.7197144031524658,
      "learning_rate": 0.00017675252596623665,
      "loss": 1.7675,
      "step": 6837
    },
    {
      "epoch": 0.24488334198793132,
      "grad_norm": 1.8122268915176392,
      "learning_rate": 0.00017674509026573864,
      "loss": 1.7868,
      "step": 6838
    },
    {
      "epoch": 0.2449191541174996,
      "grad_norm": 1.5085411071777344,
      "learning_rate": 0.00017673765353273438,
      "loss": 1.6055,
      "step": 6839
    },
    {
      "epoch": 0.2449549662470679,
      "grad_norm": 2.0807392597198486,
      "learning_rate": 0.00017673021576732404,
      "loss": 1.5603,
      "step": 6840
    },
    {
      "epoch": 0.24499077837663616,
      "grad_norm": 2.034391164779663,
      "learning_rate": 0.00017672277696960756,
      "loss": 1.2525,
      "step": 6841
    },
    {
      "epoch": 0.24502659050620446,
      "grad_norm": 1.4948173761367798,
      "learning_rate": 0.0001767153371396851,
      "loss": 1.6105,
      "step": 6842
    },
    {
      "epoch": 0.24506240263577272,
      "grad_norm": 1.3254578113555908,
      "learning_rate": 0.00017670789627765676,
      "loss": 1.5827,
      "step": 6843
    },
    {
      "epoch": 0.24509821476534102,
      "grad_norm": 1.611053228378296,
      "learning_rate": 0.0001767004543836226,
      "loss": 1.7118,
      "step": 6844
    },
    {
      "epoch": 0.24513402689490932,
      "grad_norm": 1.7157574892044067,
      "learning_rate": 0.00017669301145768277,
      "loss": 1.6883,
      "step": 6845
    },
    {
      "epoch": 0.2451698390244776,
      "grad_norm": 1.8786518573760986,
      "learning_rate": 0.0001766855674999374,
      "loss": 1.5183,
      "step": 6846
    },
    {
      "epoch": 0.24520565115404588,
      "grad_norm": 1.7475110292434692,
      "learning_rate": 0.00017667812251048664,
      "loss": 1.5028,
      "step": 6847
    },
    {
      "epoch": 0.24524146328361415,
      "grad_norm": 1.989743709564209,
      "learning_rate": 0.00017667067648943064,
      "loss": 1.8715,
      "step": 6848
    },
    {
      "epoch": 0.24527727541318245,
      "grad_norm": 1.4464224576950073,
      "learning_rate": 0.00017666322943686957,
      "loss": 1.577,
      "step": 6849
    },
    {
      "epoch": 0.24531308754275072,
      "grad_norm": 2.7265331745147705,
      "learning_rate": 0.00017665578135290364,
      "loss": 1.4392,
      "step": 6850
    },
    {
      "epoch": 0.24534889967231902,
      "grad_norm": 2.068549156188965,
      "learning_rate": 0.00017664833223763306,
      "loss": 1.5901,
      "step": 6851
    },
    {
      "epoch": 0.2453847118018873,
      "grad_norm": 1.8748308420181274,
      "learning_rate": 0.00017664088209115805,
      "loss": 1.581,
      "step": 6852
    },
    {
      "epoch": 0.24542052393145558,
      "grad_norm": 1.8242149353027344,
      "learning_rate": 0.00017663343091357881,
      "loss": 1.8448,
      "step": 6853
    },
    {
      "epoch": 0.24545633606102388,
      "grad_norm": 1.4797003269195557,
      "learning_rate": 0.00017662597870499562,
      "loss": 1.4438,
      "step": 6854
    },
    {
      "epoch": 0.24549214819059215,
      "grad_norm": 1.5570054054260254,
      "learning_rate": 0.00017661852546550875,
      "loss": 1.6526,
      "step": 6855
    },
    {
      "epoch": 0.24552796032016044,
      "grad_norm": 1.3332992792129517,
      "learning_rate": 0.00017661107119521842,
      "loss": 1.5923,
      "step": 6856
    },
    {
      "epoch": 0.2455637724497287,
      "grad_norm": 2.3711464405059814,
      "learning_rate": 0.00017660361589422497,
      "loss": 1.4343,
      "step": 6857
    },
    {
      "epoch": 0.245599584579297,
      "grad_norm": 1.5084023475646973,
      "learning_rate": 0.00017659615956262865,
      "loss": 1.7877,
      "step": 6858
    },
    {
      "epoch": 0.2456353967088653,
      "grad_norm": 1.477497935295105,
      "learning_rate": 0.00017658870220052983,
      "loss": 1.5231,
      "step": 6859
    },
    {
      "epoch": 0.24567120883843357,
      "grad_norm": 1.6098153591156006,
      "learning_rate": 0.00017658124380802882,
      "loss": 1.4208,
      "step": 6860
    },
    {
      "epoch": 0.24570702096800187,
      "grad_norm": 1.4886384010314941,
      "learning_rate": 0.00017657378438522593,
      "loss": 1.6053,
      "step": 6861
    },
    {
      "epoch": 0.24574283309757014,
      "grad_norm": 1.6936547756195068,
      "learning_rate": 0.00017656632393222156,
      "loss": 1.426,
      "step": 6862
    },
    {
      "epoch": 0.24577864522713844,
      "grad_norm": 1.6345033645629883,
      "learning_rate": 0.00017655886244911603,
      "loss": 1.5935,
      "step": 6863
    },
    {
      "epoch": 0.2458144573567067,
      "grad_norm": 1.7725507020950317,
      "learning_rate": 0.00017655139993600982,
      "loss": 1.3166,
      "step": 6864
    },
    {
      "epoch": 0.245850269486275,
      "grad_norm": 1.4159860610961914,
      "learning_rate": 0.0001765439363930032,
      "loss": 1.5238,
      "step": 6865
    },
    {
      "epoch": 0.2458860816158433,
      "grad_norm": 1.7116299867630005,
      "learning_rate": 0.00017653647182019671,
      "loss": 1.7032,
      "step": 6866
    },
    {
      "epoch": 0.24592189374541157,
      "grad_norm": 2.9299378395080566,
      "learning_rate": 0.0001765290062176907,
      "loss": 1.5626,
      "step": 6867
    },
    {
      "epoch": 0.24595770587497987,
      "grad_norm": 1.746596336364746,
      "learning_rate": 0.00017652153958558562,
      "loss": 1.3949,
      "step": 6868
    },
    {
      "epoch": 0.24599351800454813,
      "grad_norm": 1.607316017150879,
      "learning_rate": 0.00017651407192398195,
      "loss": 1.4019,
      "step": 6869
    },
    {
      "epoch": 0.24602933013411643,
      "grad_norm": 1.8668955564498901,
      "learning_rate": 0.0001765066032329801,
      "loss": 1.5584,
      "step": 6870
    },
    {
      "epoch": 0.2460651422636847,
      "grad_norm": 1.5562617778778076,
      "learning_rate": 0.0001764991335126806,
      "loss": 1.4447,
      "step": 6871
    },
    {
      "epoch": 0.246100954393253,
      "grad_norm": 1.6031650304794312,
      "learning_rate": 0.000176491662763184,
      "loss": 1.603,
      "step": 6872
    },
    {
      "epoch": 0.2461367665228213,
      "grad_norm": 1.7676963806152344,
      "learning_rate": 0.0001764841909845907,
      "loss": 1.6027,
      "step": 6873
    },
    {
      "epoch": 0.24617257865238956,
      "grad_norm": 1.4618767499923706,
      "learning_rate": 0.00017647671817700122,
      "loss": 1.149,
      "step": 6874
    },
    {
      "epoch": 0.24620839078195786,
      "grad_norm": 1.703264832496643,
      "learning_rate": 0.00017646924434051617,
      "loss": 1.9018,
      "step": 6875
    },
    {
      "epoch": 0.24624420291152613,
      "grad_norm": 1.5716698169708252,
      "learning_rate": 0.0001764617694752361,
      "loss": 1.3323,
      "step": 6876
    },
    {
      "epoch": 0.24628001504109442,
      "grad_norm": 1.4548003673553467,
      "learning_rate": 0.00017645429358126156,
      "loss": 1.6873,
      "step": 6877
    },
    {
      "epoch": 0.2463158271706627,
      "grad_norm": 1.5308620929718018,
      "learning_rate": 0.0001764468166586931,
      "loss": 1.5299,
      "step": 6878
    },
    {
      "epoch": 0.246351639300231,
      "grad_norm": 1.557874083518982,
      "learning_rate": 0.00017643933870763133,
      "loss": 1.4238,
      "step": 6879
    },
    {
      "epoch": 0.2463874514297993,
      "grad_norm": 2.1845991611480713,
      "learning_rate": 0.00017643185972817684,
      "loss": 1.5245,
      "step": 6880
    },
    {
      "epoch": 0.24642326355936756,
      "grad_norm": 1.7284934520721436,
      "learning_rate": 0.0001764243797204303,
      "loss": 1.3871,
      "step": 6881
    },
    {
      "epoch": 0.24645907568893585,
      "grad_norm": 1.6655116081237793,
      "learning_rate": 0.0001764168986844923,
      "loss": 1.6806,
      "step": 6882
    },
    {
      "epoch": 0.24649488781850412,
      "grad_norm": 2.0219454765319824,
      "learning_rate": 0.00017640941662046345,
      "loss": 1.6322,
      "step": 6883
    },
    {
      "epoch": 0.24653069994807242,
      "grad_norm": 1.672935962677002,
      "learning_rate": 0.00017640193352844454,
      "loss": 1.532,
      "step": 6884
    },
    {
      "epoch": 0.2465665120776407,
      "grad_norm": 1.3515428304672241,
      "learning_rate": 0.00017639444940853612,
      "loss": 1.466,
      "step": 6885
    },
    {
      "epoch": 0.24660232420720898,
      "grad_norm": 1.778982162475586,
      "learning_rate": 0.00017638696426083893,
      "loss": 1.67,
      "step": 6886
    },
    {
      "epoch": 0.24663813633677725,
      "grad_norm": 1.437366008758545,
      "learning_rate": 0.00017637947808545369,
      "loss": 1.282,
      "step": 6887
    },
    {
      "epoch": 0.24667394846634555,
      "grad_norm": 1.3975059986114502,
      "learning_rate": 0.00017637199088248106,
      "loss": 1.7724,
      "step": 6888
    },
    {
      "epoch": 0.24670976059591385,
      "grad_norm": 1.6595691442489624,
      "learning_rate": 0.00017636450265202185,
      "loss": 1.5012,
      "step": 6889
    },
    {
      "epoch": 0.24674557272548212,
      "grad_norm": 1.3656624555587769,
      "learning_rate": 0.00017635701339417672,
      "loss": 1.3337,
      "step": 6890
    },
    {
      "epoch": 0.2467813848550504,
      "grad_norm": 1.6880717277526855,
      "learning_rate": 0.0001763495231090465,
      "loss": 1.8932,
      "step": 6891
    },
    {
      "epoch": 0.24681719698461868,
      "grad_norm": 1.7251242399215698,
      "learning_rate": 0.0001763420317967319,
      "loss": 1.3787,
      "step": 6892
    },
    {
      "epoch": 0.24685300911418698,
      "grad_norm": 1.4598206281661987,
      "learning_rate": 0.00017633453945733373,
      "loss": 1.3369,
      "step": 6893
    },
    {
      "epoch": 0.24688882124375525,
      "grad_norm": 2.0298573970794678,
      "learning_rate": 0.00017632704609095283,
      "loss": 1.7565,
      "step": 6894
    },
    {
      "epoch": 0.24692463337332354,
      "grad_norm": 1.364871859550476,
      "learning_rate": 0.00017631955169768998,
      "loss": 1.6694,
      "step": 6895
    },
    {
      "epoch": 0.24696044550289184,
      "grad_norm": 1.5108788013458252,
      "learning_rate": 0.00017631205627764598,
      "loss": 1.6086,
      "step": 6896
    },
    {
      "epoch": 0.2469962576324601,
      "grad_norm": 1.313675045967102,
      "learning_rate": 0.0001763045598309217,
      "loss": 1.7611,
      "step": 6897
    },
    {
      "epoch": 0.2470320697620284,
      "grad_norm": 1.818166732788086,
      "learning_rate": 0.00017629706235761802,
      "loss": 1.6188,
      "step": 6898
    },
    {
      "epoch": 0.24706788189159667,
      "grad_norm": 1.8856993913650513,
      "learning_rate": 0.00017628956385783577,
      "loss": 1.7872,
      "step": 6899
    },
    {
      "epoch": 0.24710369402116497,
      "grad_norm": 1.811854362487793,
      "learning_rate": 0.00017628206433167583,
      "loss": 1.6524,
      "step": 6900
    },
    {
      "epoch": 0.24713950615073324,
      "grad_norm": 1.3475021123886108,
      "learning_rate": 0.00017627456377923911,
      "loss": 1.4307,
      "step": 6901
    },
    {
      "epoch": 0.24717531828030154,
      "grad_norm": 1.7281248569488525,
      "learning_rate": 0.00017626706220062654,
      "loss": 1.6302,
      "step": 6902
    },
    {
      "epoch": 0.24721113040986983,
      "grad_norm": 2.6777141094207764,
      "learning_rate": 0.00017625955959593904,
      "loss": 1.6385,
      "step": 6903
    },
    {
      "epoch": 0.2472469425394381,
      "grad_norm": 1.415355920791626,
      "learning_rate": 0.0001762520559652775,
      "loss": 1.5923,
      "step": 6904
    },
    {
      "epoch": 0.2472827546690064,
      "grad_norm": 1.761030912399292,
      "learning_rate": 0.00017624455130874292,
      "loss": 1.6887,
      "step": 6905
    },
    {
      "epoch": 0.24731856679857467,
      "grad_norm": 1.5250886678695679,
      "learning_rate": 0.00017623704562643624,
      "loss": 1.7323,
      "step": 6906
    },
    {
      "epoch": 0.24735437892814297,
      "grad_norm": 1.85284423828125,
      "learning_rate": 0.00017622953891845847,
      "loss": 1.4674,
      "step": 6907
    },
    {
      "epoch": 0.24739019105771123,
      "grad_norm": 2.0614492893218994,
      "learning_rate": 0.00017622203118491055,
      "loss": 1.7599,
      "step": 6908
    },
    {
      "epoch": 0.24742600318727953,
      "grad_norm": 2.1460821628570557,
      "learning_rate": 0.00017621452242589354,
      "loss": 1.437,
      "step": 6909
    },
    {
      "epoch": 0.24746181531684783,
      "grad_norm": 1.4736385345458984,
      "learning_rate": 0.00017620701264150845,
      "loss": 1.7396,
      "step": 6910
    },
    {
      "epoch": 0.2474976274464161,
      "grad_norm": 1.7290641069412231,
      "learning_rate": 0.0001761995018318563,
      "loss": 1.508,
      "step": 6911
    },
    {
      "epoch": 0.2475334395759844,
      "grad_norm": 1.502830982208252,
      "learning_rate": 0.00017619198999703812,
      "loss": 1.455,
      "step": 6912
    },
    {
      "epoch": 0.24756925170555266,
      "grad_norm": 1.5334092378616333,
      "learning_rate": 0.00017618447713715503,
      "loss": 1.6101,
      "step": 6913
    },
    {
      "epoch": 0.24760506383512096,
      "grad_norm": 1.4206371307373047,
      "learning_rate": 0.00017617696325230805,
      "loss": 1.6806,
      "step": 6914
    },
    {
      "epoch": 0.24764087596468923,
      "grad_norm": 1.6280468702316284,
      "learning_rate": 0.0001761694483425983,
      "loss": 1.4521,
      "step": 6915
    },
    {
      "epoch": 0.24767668809425752,
      "grad_norm": 1.273987054824829,
      "learning_rate": 0.00017616193240812687,
      "loss": 1.5572,
      "step": 6916
    },
    {
      "epoch": 0.24771250022382582,
      "grad_norm": 1.521125078201294,
      "learning_rate": 0.00017615441544899488,
      "loss": 1.6782,
      "step": 6917
    },
    {
      "epoch": 0.2477483123533941,
      "grad_norm": 1.4935173988342285,
      "learning_rate": 0.00017614689746530345,
      "loss": 1.7162,
      "step": 6918
    },
    {
      "epoch": 0.2477841244829624,
      "grad_norm": 1.9287409782409668,
      "learning_rate": 0.00017613937845715376,
      "loss": 1.5272,
      "step": 6919
    },
    {
      "epoch": 0.24781993661253066,
      "grad_norm": 1.6758594512939453,
      "learning_rate": 0.00017613185842464693,
      "loss": 1.7733,
      "step": 6920
    },
    {
      "epoch": 0.24785574874209895,
      "grad_norm": 2.1583969593048096,
      "learning_rate": 0.00017612433736788417,
      "loss": 1.6371,
      "step": 6921
    },
    {
      "epoch": 0.24789156087166722,
      "grad_norm": 1.667779803276062,
      "learning_rate": 0.0001761168152869666,
      "loss": 1.5351,
      "step": 6922
    },
    {
      "epoch": 0.24792737300123552,
      "grad_norm": 1.3906711339950562,
      "learning_rate": 0.00017610929218199553,
      "loss": 1.6684,
      "step": 6923
    },
    {
      "epoch": 0.24796318513080381,
      "grad_norm": 1.3722862005233765,
      "learning_rate": 0.00017610176805307206,
      "loss": 1.411,
      "step": 6924
    },
    {
      "epoch": 0.24799899726037208,
      "grad_norm": 1.9369726181030273,
      "learning_rate": 0.00017609424290029746,
      "loss": 1.5838,
      "step": 6925
    },
    {
      "epoch": 0.24803480938994038,
      "grad_norm": 2.327025890350342,
      "learning_rate": 0.000176086716723773,
      "loss": 1.6616,
      "step": 6926
    },
    {
      "epoch": 0.24807062151950865,
      "grad_norm": 1.8831285238265991,
      "learning_rate": 0.0001760791895235999,
      "loss": 1.2508,
      "step": 6927
    },
    {
      "epoch": 0.24810643364907695,
      "grad_norm": 1.6104234457015991,
      "learning_rate": 0.00017607166129987944,
      "loss": 1.5941,
      "step": 6928
    },
    {
      "epoch": 0.24814224577864522,
      "grad_norm": 1.5727404356002808,
      "learning_rate": 0.0001760641320527129,
      "loss": 1.4856,
      "step": 6929
    },
    {
      "epoch": 0.2481780579082135,
      "grad_norm": 2.3073596954345703,
      "learning_rate": 0.00017605660178220158,
      "loss": 1.8495,
      "step": 6930
    },
    {
      "epoch": 0.2482138700377818,
      "grad_norm": 1.5391161441802979,
      "learning_rate": 0.0001760490704884468,
      "loss": 1.7637,
      "step": 6931
    },
    {
      "epoch": 0.24824968216735008,
      "grad_norm": 2.0064144134521484,
      "learning_rate": 0.00017604153817154985,
      "loss": 1.8597,
      "step": 6932
    },
    {
      "epoch": 0.24828549429691837,
      "grad_norm": 1.7063194513320923,
      "learning_rate": 0.00017603400483161212,
      "loss": 1.2984,
      "step": 6933
    },
    {
      "epoch": 0.24832130642648664,
      "grad_norm": 1.3737378120422363,
      "learning_rate": 0.0001760264704687349,
      "loss": 1.5487,
      "step": 6934
    },
    {
      "epoch": 0.24835711855605494,
      "grad_norm": 1.4802964925765991,
      "learning_rate": 0.00017601893508301962,
      "loss": 1.2851,
      "step": 6935
    },
    {
      "epoch": 0.2483929306856232,
      "grad_norm": 1.6147825717926025,
      "learning_rate": 0.0001760113986745676,
      "loss": 1.7134,
      "step": 6936
    },
    {
      "epoch": 0.2484287428151915,
      "grad_norm": 2.331864833831787,
      "learning_rate": 0.00017600386124348028,
      "loss": 1.3899,
      "step": 6937
    },
    {
      "epoch": 0.2484645549447598,
      "grad_norm": 1.2941458225250244,
      "learning_rate": 0.00017599632278985904,
      "loss": 1.5272,
      "step": 6938
    },
    {
      "epoch": 0.24850036707432807,
      "grad_norm": 1.85355544090271,
      "learning_rate": 0.00017598878331380528,
      "loss": 1.6488,
      "step": 6939
    },
    {
      "epoch": 0.24853617920389637,
      "grad_norm": 1.6847429275512695,
      "learning_rate": 0.0001759812428154205,
      "loss": 1.4094,
      "step": 6940
    },
    {
      "epoch": 0.24857199133346464,
      "grad_norm": 2.6703569889068604,
      "learning_rate": 0.00017597370129480606,
      "loss": 1.2444,
      "step": 6941
    },
    {
      "epoch": 0.24860780346303293,
      "grad_norm": 1.4878178834915161,
      "learning_rate": 0.00017596615875206347,
      "loss": 1.6358,
      "step": 6942
    },
    {
      "epoch": 0.2486436155926012,
      "grad_norm": 1.5205473899841309,
      "learning_rate": 0.00017595861518729424,
      "loss": 1.5765,
      "step": 6943
    },
    {
      "epoch": 0.2486794277221695,
      "grad_norm": 1.6328760385513306,
      "learning_rate": 0.00017595107060059984,
      "loss": 1.5,
      "step": 6944
    },
    {
      "epoch": 0.2487152398517378,
      "grad_norm": 1.4762791395187378,
      "learning_rate": 0.0001759435249920817,
      "loss": 1.9516,
      "step": 6945
    },
    {
      "epoch": 0.24875105198130606,
      "grad_norm": 1.660315752029419,
      "learning_rate": 0.0001759359783618414,
      "loss": 1.6335,
      "step": 6946
    },
    {
      "epoch": 0.24878686411087436,
      "grad_norm": 1.6185013055801392,
      "learning_rate": 0.00017592843070998049,
      "loss": 1.7045,
      "step": 6947
    },
    {
      "epoch": 0.24882267624044263,
      "grad_norm": 1.4820398092269897,
      "learning_rate": 0.00017592088203660045,
      "loss": 1.4543,
      "step": 6948
    },
    {
      "epoch": 0.24885848837001093,
      "grad_norm": 1.3168247938156128,
      "learning_rate": 0.00017591333234180293,
      "loss": 1.5607,
      "step": 6949
    },
    {
      "epoch": 0.2488943004995792,
      "grad_norm": 1.3358012437820435,
      "learning_rate": 0.0001759057816256894,
      "loss": 1.452,
      "step": 6950
    },
    {
      "epoch": 0.2489301126291475,
      "grad_norm": 2.118440866470337,
      "learning_rate": 0.00017589822988836148,
      "loss": 1.5474,
      "step": 6951
    },
    {
      "epoch": 0.2489659247587158,
      "grad_norm": 2.4116692543029785,
      "learning_rate": 0.00017589067712992082,
      "loss": 1.3814,
      "step": 6952
    },
    {
      "epoch": 0.24900173688828406,
      "grad_norm": 1.8422210216522217,
      "learning_rate": 0.00017588312335046897,
      "loss": 1.6097,
      "step": 6953
    },
    {
      "epoch": 0.24903754901785236,
      "grad_norm": 1.4137614965438843,
      "learning_rate": 0.00017587556855010755,
      "loss": 1.7229,
      "step": 6954
    },
    {
      "epoch": 0.24907336114742062,
      "grad_norm": 1.4702421426773071,
      "learning_rate": 0.00017586801272893827,
      "loss": 1.7011,
      "step": 6955
    },
    {
      "epoch": 0.24910917327698892,
      "grad_norm": 1.7273403406143188,
      "learning_rate": 0.00017586045588706273,
      "loss": 1.2071,
      "step": 6956
    },
    {
      "epoch": 0.2491449854065572,
      "grad_norm": 1.629349946975708,
      "learning_rate": 0.0001758528980245826,
      "loss": 1.6223,
      "step": 6957
    },
    {
      "epoch": 0.2491807975361255,
      "grad_norm": 1.4688459634780884,
      "learning_rate": 0.00017584533914159956,
      "loss": 1.2623,
      "step": 6958
    },
    {
      "epoch": 0.24921660966569378,
      "grad_norm": 1.8785773515701294,
      "learning_rate": 0.00017583777923821533,
      "loss": 1.5056,
      "step": 6959
    },
    {
      "epoch": 0.24925242179526205,
      "grad_norm": 1.7873069047927856,
      "learning_rate": 0.0001758302183145316,
      "loss": 1.5735,
      "step": 6960
    },
    {
      "epoch": 0.24928823392483035,
      "grad_norm": 1.8960673809051514,
      "learning_rate": 0.00017582265637065012,
      "loss": 1.7953,
      "step": 6961
    },
    {
      "epoch": 0.24932404605439862,
      "grad_norm": 1.3856395483016968,
      "learning_rate": 0.00017581509340667257,
      "loss": 1.5991,
      "step": 6962
    },
    {
      "epoch": 0.24935985818396691,
      "grad_norm": 1.5916454792022705,
      "learning_rate": 0.00017580752942270077,
      "loss": 1.7537,
      "step": 6963
    },
    {
      "epoch": 0.24939567031353518,
      "grad_norm": 1.445228099822998,
      "learning_rate": 0.0001757999644188364,
      "loss": 1.6487,
      "step": 6964
    },
    {
      "epoch": 0.24943148244310348,
      "grad_norm": 1.5743353366851807,
      "learning_rate": 0.0001757923983951813,
      "loss": 1.5899,
      "step": 6965
    },
    {
      "epoch": 0.24946729457267178,
      "grad_norm": 1.4328770637512207,
      "learning_rate": 0.00017578483135183726,
      "loss": 1.6986,
      "step": 6966
    },
    {
      "epoch": 0.24950310670224005,
      "grad_norm": 1.5256279706954956,
      "learning_rate": 0.00017577726328890604,
      "loss": 1.3874,
      "step": 6967
    },
    {
      "epoch": 0.24953891883180834,
      "grad_norm": 1.6930445432662964,
      "learning_rate": 0.0001757696942064895,
      "loss": 1.5721,
      "step": 6968
    },
    {
      "epoch": 0.2495747309613766,
      "grad_norm": 1.8914119005203247,
      "learning_rate": 0.00017576212410468949,
      "loss": 1.3895,
      "step": 6969
    },
    {
      "epoch": 0.2496105430909449,
      "grad_norm": 1.2605080604553223,
      "learning_rate": 0.00017575455298360782,
      "loss": 1.562,
      "step": 6970
    },
    {
      "epoch": 0.24964635522051318,
      "grad_norm": 1.4235647916793823,
      "learning_rate": 0.00017574698084334633,
      "loss": 1.6511,
      "step": 6971
    },
    {
      "epoch": 0.24968216735008147,
      "grad_norm": 1.5089948177337646,
      "learning_rate": 0.00017573940768400692,
      "loss": 1.2899,
      "step": 6972
    },
    {
      "epoch": 0.24971797947964977,
      "grad_norm": 1.1215814352035522,
      "learning_rate": 0.00017573183350569148,
      "loss": 1.2685,
      "step": 6973
    },
    {
      "epoch": 0.24975379160921804,
      "grad_norm": 1.6614004373550415,
      "learning_rate": 0.00017572425830850193,
      "loss": 1.4867,
      "step": 6974
    },
    {
      "epoch": 0.24978960373878634,
      "grad_norm": 1.3425636291503906,
      "learning_rate": 0.00017571668209254013,
      "loss": 1.6014,
      "step": 6975
    },
    {
      "epoch": 0.2498254158683546,
      "grad_norm": 1.719012975692749,
      "learning_rate": 0.00017570910485790805,
      "loss": 1.3348,
      "step": 6976
    },
    {
      "epoch": 0.2498612279979229,
      "grad_norm": 1.7479561567306519,
      "learning_rate": 0.00017570152660470765,
      "loss": 1.7141,
      "step": 6977
    },
    {
      "epoch": 0.24989704012749117,
      "grad_norm": 1.8728541135787964,
      "learning_rate": 0.00017569394733304083,
      "loss": 1.7683,
      "step": 6978
    },
    {
      "epoch": 0.24993285225705947,
      "grad_norm": 1.8570350408554077,
      "learning_rate": 0.00017568636704300958,
      "loss": 1.7784,
      "step": 6979
    },
    {
      "epoch": 0.24996866438662776,
      "grad_norm": 1.8610103130340576,
      "learning_rate": 0.0001756787857347159,
      "loss": 1.416,
      "step": 6980
    },
    {
      "epoch": 0.25000447651619606,
      "grad_norm": 1.5104469060897827,
      "learning_rate": 0.00017567120340826177,
      "loss": 1.6459,
      "step": 6981
    },
    {
      "epoch": 0.25004028864576433,
      "grad_norm": 1.4502947330474854,
      "learning_rate": 0.0001756636200637492,
      "loss": 1.3919,
      "step": 6982
    },
    {
      "epoch": 0.2500761007753326,
      "grad_norm": 1.8482967615127563,
      "learning_rate": 0.00017565603570128023,
      "loss": 1.5629,
      "step": 6983
    },
    {
      "epoch": 0.25011191290490087,
      "grad_norm": 1.9667346477508545,
      "learning_rate": 0.0001756484503209569,
      "loss": 1.6214,
      "step": 6984
    },
    {
      "epoch": 0.2501477250344692,
      "grad_norm": 1.8211997747421265,
      "learning_rate": 0.00017564086392288125,
      "loss": 1.8636,
      "step": 6985
    },
    {
      "epoch": 0.25018353716403746,
      "grad_norm": 1.3446124792099,
      "learning_rate": 0.00017563327650715535,
      "loss": 1.8394,
      "step": 6986
    },
    {
      "epoch": 0.25021934929360573,
      "grad_norm": 1.7953028678894043,
      "learning_rate": 0.00017562568807388126,
      "loss": 1.6372,
      "step": 6987
    },
    {
      "epoch": 0.25025516142317406,
      "grad_norm": 1.6460068225860596,
      "learning_rate": 0.0001756180986231611,
      "loss": 1.5185,
      "step": 6988
    },
    {
      "epoch": 0.2502909735527423,
      "grad_norm": 1.467795491218567,
      "learning_rate": 0.00017561050815509695,
      "loss": 1.5313,
      "step": 6989
    },
    {
      "epoch": 0.2503267856823106,
      "grad_norm": 1.5542267560958862,
      "learning_rate": 0.00017560291666979095,
      "loss": 1.1941,
      "step": 6990
    },
    {
      "epoch": 0.25036259781187886,
      "grad_norm": 1.5457571744918823,
      "learning_rate": 0.00017559532416734524,
      "loss": 1.7399,
      "step": 6991
    },
    {
      "epoch": 0.2503984099414472,
      "grad_norm": 1.6578049659729004,
      "learning_rate": 0.00017558773064786193,
      "loss": 1.4826,
      "step": 6992
    },
    {
      "epoch": 0.25043422207101546,
      "grad_norm": 2.136821746826172,
      "learning_rate": 0.0001755801361114432,
      "loss": 1.8302,
      "step": 6993
    },
    {
      "epoch": 0.2504700342005837,
      "grad_norm": 1.3996917009353638,
      "learning_rate": 0.00017557254055819126,
      "loss": 1.5399,
      "step": 6994
    },
    {
      "epoch": 0.25050584633015205,
      "grad_norm": 1.3739955425262451,
      "learning_rate": 0.00017556494398820823,
      "loss": 1.3365,
      "step": 6995
    },
    {
      "epoch": 0.2505416584597203,
      "grad_norm": 1.558883786201477,
      "learning_rate": 0.0001755573464015964,
      "loss": 1.6076,
      "step": 6996
    },
    {
      "epoch": 0.2505774705892886,
      "grad_norm": 1.6406570672988892,
      "learning_rate": 0.00017554974779845792,
      "loss": 1.8316,
      "step": 6997
    },
    {
      "epoch": 0.25061328271885686,
      "grad_norm": 1.7725262641906738,
      "learning_rate": 0.000175542148178895,
      "loss": 1.7184,
      "step": 6998
    },
    {
      "epoch": 0.2506490948484252,
      "grad_norm": 1.7781455516815186,
      "learning_rate": 0.00017553454754300996,
      "loss": 1.2366,
      "step": 6999
    },
    {
      "epoch": 0.25068490697799345,
      "grad_norm": 1.753017783164978,
      "learning_rate": 0.000175526945890905,
      "loss": 1.6132,
      "step": 7000
    },
    {
      "epoch": 0.2507207191075617,
      "grad_norm": 2.3922290802001953,
      "learning_rate": 0.0001755193432226824,
      "loss": 1.5879,
      "step": 7001
    },
    {
      "epoch": 0.25075653123713004,
      "grad_norm": 1.8872179985046387,
      "learning_rate": 0.00017551173953844445,
      "loss": 1.4223,
      "step": 7002
    },
    {
      "epoch": 0.2507923433666983,
      "grad_norm": 1.5126007795333862,
      "learning_rate": 0.00017550413483829344,
      "loss": 1.7051,
      "step": 7003
    },
    {
      "epoch": 0.2508281554962666,
      "grad_norm": 1.5426768064498901,
      "learning_rate": 0.0001754965291223317,
      "loss": 1.3291,
      "step": 7004
    },
    {
      "epoch": 0.25086396762583485,
      "grad_norm": 1.5832163095474243,
      "learning_rate": 0.00017548892239066156,
      "loss": 1.26,
      "step": 7005
    },
    {
      "epoch": 0.2508997797554032,
      "grad_norm": 1.4844779968261719,
      "learning_rate": 0.00017548131464338533,
      "loss": 1.697,
      "step": 7006
    },
    {
      "epoch": 0.25093559188497144,
      "grad_norm": 1.5662215948104858,
      "learning_rate": 0.00017547370588060537,
      "loss": 1.3258,
      "step": 7007
    },
    {
      "epoch": 0.2509714040145397,
      "grad_norm": 1.3159148693084717,
      "learning_rate": 0.00017546609610242405,
      "loss": 1.589,
      "step": 7008
    },
    {
      "epoch": 0.25100721614410804,
      "grad_norm": 1.476625919342041,
      "learning_rate": 0.00017545848530894377,
      "loss": 1.715,
      "step": 7009
    },
    {
      "epoch": 0.2510430282736763,
      "grad_norm": 1.64826238155365,
      "learning_rate": 0.0001754508735002669,
      "loss": 1.8173,
      "step": 7010
    },
    {
      "epoch": 0.2510788404032446,
      "grad_norm": 1.6059377193450928,
      "learning_rate": 0.00017544326067649583,
      "loss": 1.6255,
      "step": 7011
    },
    {
      "epoch": 0.25111465253281284,
      "grad_norm": 1.6135188341140747,
      "learning_rate": 0.00017543564683773302,
      "loss": 1.6396,
      "step": 7012
    },
    {
      "epoch": 0.25115046466238117,
      "grad_norm": 1.242232322692871,
      "learning_rate": 0.00017542803198408087,
      "loss": 1.6733,
      "step": 7013
    },
    {
      "epoch": 0.25118627679194944,
      "grad_norm": 2.01194167137146,
      "learning_rate": 0.00017542041611564186,
      "loss": 1.4317,
      "step": 7014
    },
    {
      "epoch": 0.2512220889215177,
      "grad_norm": 1.904948353767395,
      "learning_rate": 0.00017541279923251844,
      "loss": 1.6848,
      "step": 7015
    },
    {
      "epoch": 0.25125790105108603,
      "grad_norm": 1.638850212097168,
      "learning_rate": 0.00017540518133481308,
      "loss": 1.3428,
      "step": 7016
    },
    {
      "epoch": 0.2512937131806543,
      "grad_norm": 2.054460287094116,
      "learning_rate": 0.00017539756242262826,
      "loss": 1.5478,
      "step": 7017
    },
    {
      "epoch": 0.25132952531022257,
      "grad_norm": 1.5042341947555542,
      "learning_rate": 0.0001753899424960665,
      "loss": 1.6176,
      "step": 7018
    },
    {
      "epoch": 0.25136533743979084,
      "grad_norm": 1.8969051837921143,
      "learning_rate": 0.0001753823215552303,
      "loss": 1.3305,
      "step": 7019
    },
    {
      "epoch": 0.25140114956935916,
      "grad_norm": 1.6631965637207031,
      "learning_rate": 0.00017537469960022221,
      "loss": 1.7607,
      "step": 7020
    },
    {
      "epoch": 0.25143696169892743,
      "grad_norm": 1.4155299663543701,
      "learning_rate": 0.00017536707663114477,
      "loss": 1.4719,
      "step": 7021
    },
    {
      "epoch": 0.2514727738284957,
      "grad_norm": 1.9903786182403564,
      "learning_rate": 0.00017535945264810052,
      "loss": 1.5096,
      "step": 7022
    },
    {
      "epoch": 0.251508585958064,
      "grad_norm": 1.4660794734954834,
      "learning_rate": 0.00017535182765119204,
      "loss": 1.3804,
      "step": 7023
    },
    {
      "epoch": 0.2515443980876323,
      "grad_norm": 1.6268247365951538,
      "learning_rate": 0.00017534420164052193,
      "loss": 1.4592,
      "step": 7024
    },
    {
      "epoch": 0.25158021021720056,
      "grad_norm": 1.503402590751648,
      "learning_rate": 0.00017533657461619274,
      "loss": 1.4945,
      "step": 7025
    },
    {
      "epoch": 0.25161602234676883,
      "grad_norm": 1.4199435710906982,
      "learning_rate": 0.00017532894657830715,
      "loss": 1.7763,
      "step": 7026
    },
    {
      "epoch": 0.25165183447633716,
      "grad_norm": 1.6909997463226318,
      "learning_rate": 0.00017532131752696776,
      "loss": 1.546,
      "step": 7027
    },
    {
      "epoch": 0.2516876466059054,
      "grad_norm": 1.7875890731811523,
      "learning_rate": 0.00017531368746227718,
      "loss": 1.501,
      "step": 7028
    },
    {
      "epoch": 0.2517234587354737,
      "grad_norm": 2.4746861457824707,
      "learning_rate": 0.00017530605638433805,
      "loss": 1.576,
      "step": 7029
    },
    {
      "epoch": 0.251759270865042,
      "grad_norm": 1.6079216003417969,
      "learning_rate": 0.00017529842429325312,
      "loss": 1.7127,
      "step": 7030
    },
    {
      "epoch": 0.2517950829946103,
      "grad_norm": 1.7867952585220337,
      "learning_rate": 0.00017529079118912502,
      "loss": 1.1966,
      "step": 7031
    },
    {
      "epoch": 0.25183089512417856,
      "grad_norm": 1.4089163541793823,
      "learning_rate": 0.00017528315707205643,
      "loss": 1.7839,
      "step": 7032
    },
    {
      "epoch": 0.2518667072537468,
      "grad_norm": 1.644866704940796,
      "learning_rate": 0.00017527552194215005,
      "loss": 1.4532,
      "step": 7033
    },
    {
      "epoch": 0.25190251938331515,
      "grad_norm": 1.627656102180481,
      "learning_rate": 0.00017526788579950864,
      "loss": 1.7095,
      "step": 7034
    },
    {
      "epoch": 0.2519383315128834,
      "grad_norm": 1.9081600904464722,
      "learning_rate": 0.0001752602486442349,
      "loss": 1.6353,
      "step": 7035
    },
    {
      "epoch": 0.2519741436424517,
      "grad_norm": 1.579540729522705,
      "learning_rate": 0.0001752526104764316,
      "loss": 1.4417,
      "step": 7036
    },
    {
      "epoch": 0.25200995577201996,
      "grad_norm": 1.7195303440093994,
      "learning_rate": 0.0001752449712962015,
      "loss": 1.3465,
      "step": 7037
    },
    {
      "epoch": 0.2520457679015883,
      "grad_norm": 1.6513937711715698,
      "learning_rate": 0.00017523733110364736,
      "loss": 1.3911,
      "step": 7038
    },
    {
      "epoch": 0.25208158003115655,
      "grad_norm": 1.918748140335083,
      "learning_rate": 0.000175229689898872,
      "loss": 1.728,
      "step": 7039
    },
    {
      "epoch": 0.2521173921607248,
      "grad_norm": 1.678743600845337,
      "learning_rate": 0.00017522204768197818,
      "loss": 1.3992,
      "step": 7040
    },
    {
      "epoch": 0.25215320429029314,
      "grad_norm": 1.795161247253418,
      "learning_rate": 0.00017521440445306875,
      "loss": 1.6584,
      "step": 7041
    },
    {
      "epoch": 0.2521890164198614,
      "grad_norm": 1.783471941947937,
      "learning_rate": 0.00017520676021224652,
      "loss": 1.5853,
      "step": 7042
    },
    {
      "epoch": 0.2522248285494297,
      "grad_norm": 1.6953623294830322,
      "learning_rate": 0.00017519911495961435,
      "loss": 1.6603,
      "step": 7043
    },
    {
      "epoch": 0.25226064067899795,
      "grad_norm": 1.5812227725982666,
      "learning_rate": 0.0001751914686952751,
      "loss": 1.6276,
      "step": 7044
    },
    {
      "epoch": 0.2522964528085663,
      "grad_norm": 1.6534466743469238,
      "learning_rate": 0.0001751838214193316,
      "loss": 1.6614,
      "step": 7045
    },
    {
      "epoch": 0.25233226493813454,
      "grad_norm": 1.963584303855896,
      "learning_rate": 0.0001751761731318868,
      "loss": 1.3825,
      "step": 7046
    },
    {
      "epoch": 0.2523680770677028,
      "grad_norm": 1.7353453636169434,
      "learning_rate": 0.00017516852383304353,
      "loss": 1.3718,
      "step": 7047
    },
    {
      "epoch": 0.25240388919727114,
      "grad_norm": 1.7376201152801514,
      "learning_rate": 0.00017516087352290472,
      "loss": 1.6218,
      "step": 7048
    },
    {
      "epoch": 0.2524397013268394,
      "grad_norm": 1.3723000288009644,
      "learning_rate": 0.00017515322220157333,
      "loss": 1.234,
      "step": 7049
    },
    {
      "epoch": 0.2524755134564077,
      "grad_norm": 1.7133162021636963,
      "learning_rate": 0.0001751455698691523,
      "loss": 1.1578,
      "step": 7050
    },
    {
      "epoch": 0.25251132558597594,
      "grad_norm": 1.6886383295059204,
      "learning_rate": 0.00017513791652574453,
      "loss": 1.6405,
      "step": 7051
    },
    {
      "epoch": 0.25254713771554427,
      "grad_norm": 1.5092798471450806,
      "learning_rate": 0.00017513026217145302,
      "loss": 1.7148,
      "step": 7052
    },
    {
      "epoch": 0.25258294984511254,
      "grad_norm": 1.4195395708084106,
      "learning_rate": 0.00017512260680638072,
      "loss": 1.4879,
      "step": 7053
    },
    {
      "epoch": 0.2526187619746808,
      "grad_norm": 1.9386016130447388,
      "learning_rate": 0.00017511495043063066,
      "loss": 1.7046,
      "step": 7054
    },
    {
      "epoch": 0.25265457410424913,
      "grad_norm": 1.3876835107803345,
      "learning_rate": 0.00017510729304430584,
      "loss": 1.5976,
      "step": 7055
    },
    {
      "epoch": 0.2526903862338174,
      "grad_norm": 1.3243073225021362,
      "learning_rate": 0.00017509963464750928,
      "loss": 1.4122,
      "step": 7056
    },
    {
      "epoch": 0.25272619836338567,
      "grad_norm": 2.481052875518799,
      "learning_rate": 0.000175091975240344,
      "loss": 1.6062,
      "step": 7057
    },
    {
      "epoch": 0.25276201049295394,
      "grad_norm": 1.4827232360839844,
      "learning_rate": 0.00017508431482291304,
      "loss": 1.3834,
      "step": 7058
    },
    {
      "epoch": 0.25279782262252226,
      "grad_norm": 1.5816888809204102,
      "learning_rate": 0.0001750766533953195,
      "loss": 1.8041,
      "step": 7059
    },
    {
      "epoch": 0.25283363475209053,
      "grad_norm": 1.6099169254302979,
      "learning_rate": 0.00017506899095766641,
      "loss": 1.5866,
      "step": 7060
    },
    {
      "epoch": 0.2528694468816588,
      "grad_norm": 1.552196979522705,
      "learning_rate": 0.0001750613275100569,
      "loss": 1.3113,
      "step": 7061
    },
    {
      "epoch": 0.2529052590112271,
      "grad_norm": 1.621914029121399,
      "learning_rate": 0.00017505366305259402,
      "loss": 1.4599,
      "step": 7062
    },
    {
      "epoch": 0.2529410711407954,
      "grad_norm": 1.5911298990249634,
      "learning_rate": 0.00017504599758538095,
      "loss": 1.3794,
      "step": 7063
    },
    {
      "epoch": 0.25297688327036366,
      "grad_norm": 1.4210929870605469,
      "learning_rate": 0.00017503833110852078,
      "loss": 1.3875,
      "step": 7064
    },
    {
      "epoch": 0.25301269539993193,
      "grad_norm": 2.398237705230713,
      "learning_rate": 0.00017503066362211663,
      "loss": 1.6252,
      "step": 7065
    },
    {
      "epoch": 0.25304850752950026,
      "grad_norm": 1.9180196523666382,
      "learning_rate": 0.00017502299512627172,
      "loss": 1.6482,
      "step": 7066
    },
    {
      "epoch": 0.2530843196590685,
      "grad_norm": 1.9644347429275513,
      "learning_rate": 0.00017501532562108916,
      "loss": 1.3261,
      "step": 7067
    },
    {
      "epoch": 0.2531201317886368,
      "grad_norm": 1.9443340301513672,
      "learning_rate": 0.00017500765510667217,
      "loss": 1.5519,
      "step": 7068
    },
    {
      "epoch": 0.2531559439182051,
      "grad_norm": 2.6250624656677246,
      "learning_rate": 0.0001749999835831239,
      "loss": 1.7498,
      "step": 7069
    },
    {
      "epoch": 0.2531917560477734,
      "grad_norm": 1.7860009670257568,
      "learning_rate": 0.00017499231105054763,
      "loss": 1.4288,
      "step": 7070
    },
    {
      "epoch": 0.25322756817734166,
      "grad_norm": 1.8363738059997559,
      "learning_rate": 0.00017498463750904652,
      "loss": 1.1333,
      "step": 7071
    },
    {
      "epoch": 0.2532633803069099,
      "grad_norm": 1.4646892547607422,
      "learning_rate": 0.00017497696295872385,
      "loss": 1.7276,
      "step": 7072
    },
    {
      "epoch": 0.25329919243647825,
      "grad_norm": 1.5846657752990723,
      "learning_rate": 0.00017496928739968288,
      "loss": 1.693,
      "step": 7073
    },
    {
      "epoch": 0.2533350045660465,
      "grad_norm": 1.4653985500335693,
      "learning_rate": 0.0001749616108320268,
      "loss": 1.4689,
      "step": 7074
    },
    {
      "epoch": 0.2533708166956148,
      "grad_norm": 1.4848623275756836,
      "learning_rate": 0.000174953933255859,
      "loss": 1.4579,
      "step": 7075
    },
    {
      "epoch": 0.2534066288251831,
      "grad_norm": 1.226401448249817,
      "learning_rate": 0.0001749462546712827,
      "loss": 1.2865,
      "step": 7076
    },
    {
      "epoch": 0.2534424409547514,
      "grad_norm": 1.9527426958084106,
      "learning_rate": 0.00017493857507840116,
      "loss": 1.3185,
      "step": 7077
    },
    {
      "epoch": 0.25347825308431965,
      "grad_norm": 2.5343284606933594,
      "learning_rate": 0.0001749308944773178,
      "loss": 1.7735,
      "step": 7078
    },
    {
      "epoch": 0.2535140652138879,
      "grad_norm": 1.6038436889648438,
      "learning_rate": 0.0001749232128681359,
      "loss": 1.5201,
      "step": 7079
    },
    {
      "epoch": 0.25354987734345624,
      "grad_norm": 1.7659517526626587,
      "learning_rate": 0.00017491553025095882,
      "loss": 1.8174,
      "step": 7080
    },
    {
      "epoch": 0.2535856894730245,
      "grad_norm": 1.3074805736541748,
      "learning_rate": 0.00017490784662588992,
      "loss": 1.4389,
      "step": 7081
    },
    {
      "epoch": 0.2536215016025928,
      "grad_norm": 1.4077818393707275,
      "learning_rate": 0.00017490016199303256,
      "loss": 1.4818,
      "step": 7082
    },
    {
      "epoch": 0.2536573137321611,
      "grad_norm": 1.844086766242981,
      "learning_rate": 0.00017489247635249012,
      "loss": 1.7697,
      "step": 7083
    },
    {
      "epoch": 0.2536931258617294,
      "grad_norm": 1.78142249584198,
      "learning_rate": 0.00017488478970436604,
      "loss": 1.5957,
      "step": 7084
    },
    {
      "epoch": 0.25372893799129764,
      "grad_norm": 1.8254570960998535,
      "learning_rate": 0.0001748771020487637,
      "loss": 1.4769,
      "step": 7085
    },
    {
      "epoch": 0.2537647501208659,
      "grad_norm": 1.7204885482788086,
      "learning_rate": 0.00017486941338578653,
      "loss": 1.4355,
      "step": 7086
    },
    {
      "epoch": 0.25380056225043424,
      "grad_norm": 1.4895552396774292,
      "learning_rate": 0.000174861723715538,
      "loss": 1.407,
      "step": 7087
    },
    {
      "epoch": 0.2538363743800025,
      "grad_norm": 2.2478761672973633,
      "learning_rate": 0.0001748540330381215,
      "loss": 1.7522,
      "step": 7088
    },
    {
      "epoch": 0.2538721865095708,
      "grad_norm": 1.7592345476150513,
      "learning_rate": 0.00017484634135364057,
      "loss": 1.3917,
      "step": 7089
    },
    {
      "epoch": 0.2539079986391391,
      "grad_norm": 1.845581293106079,
      "learning_rate": 0.00017483864866219868,
      "loss": 1.4701,
      "step": 7090
    },
    {
      "epoch": 0.25394381076870737,
      "grad_norm": 1.259600043296814,
      "learning_rate": 0.00017483095496389928,
      "loss": 1.203,
      "step": 7091
    },
    {
      "epoch": 0.25397962289827564,
      "grad_norm": 1.8078480958938599,
      "learning_rate": 0.0001748232602588459,
      "loss": 1.5376,
      "step": 7092
    },
    {
      "epoch": 0.2540154350278439,
      "grad_norm": 1.4246379137039185,
      "learning_rate": 0.0001748155645471421,
      "loss": 1.7493,
      "step": 7093
    },
    {
      "epoch": 0.25405124715741223,
      "grad_norm": 1.7548315525054932,
      "learning_rate": 0.00017480786782889137,
      "loss": 1.3788,
      "step": 7094
    },
    {
      "epoch": 0.2540870592869805,
      "grad_norm": 1.760630488395691,
      "learning_rate": 0.00017480017010419724,
      "loss": 1.6053,
      "step": 7095
    },
    {
      "epoch": 0.25412287141654877,
      "grad_norm": 1.9188206195831299,
      "learning_rate": 0.00017479247137316335,
      "loss": 1.5271,
      "step": 7096
    },
    {
      "epoch": 0.2541586835461171,
      "grad_norm": 1.4333956241607666,
      "learning_rate": 0.0001747847716358932,
      "loss": 1.3216,
      "step": 7097
    },
    {
      "epoch": 0.25419449567568536,
      "grad_norm": 1.5908639430999756,
      "learning_rate": 0.00017477707089249043,
      "loss": 1.3639,
      "step": 7098
    },
    {
      "epoch": 0.25423030780525363,
      "grad_norm": 1.311259388923645,
      "learning_rate": 0.00017476936914305862,
      "loss": 1.6077,
      "step": 7099
    },
    {
      "epoch": 0.2542661199348219,
      "grad_norm": 1.6787432432174683,
      "learning_rate": 0.00017476166638770142,
      "loss": 1.6371,
      "step": 7100
    },
    {
      "epoch": 0.2543019320643902,
      "grad_norm": 1.8875246047973633,
      "learning_rate": 0.0001747539626265224,
      "loss": 1.5801,
      "step": 7101
    },
    {
      "epoch": 0.2543377441939585,
      "grad_norm": 2.0725271701812744,
      "learning_rate": 0.00017474625785962524,
      "loss": 1.6117,
      "step": 7102
    },
    {
      "epoch": 0.25437355632352676,
      "grad_norm": 2.2691195011138916,
      "learning_rate": 0.00017473855208711362,
      "loss": 1.6867,
      "step": 7103
    },
    {
      "epoch": 0.2544093684530951,
      "grad_norm": 1.4735385179519653,
      "learning_rate": 0.00017473084530909117,
      "loss": 1.5192,
      "step": 7104
    },
    {
      "epoch": 0.25444518058266336,
      "grad_norm": 1.6299970149993896,
      "learning_rate": 0.0001747231375256616,
      "loss": 1.74,
      "step": 7105
    },
    {
      "epoch": 0.2544809927122316,
      "grad_norm": 1.8156439065933228,
      "learning_rate": 0.00017471542873692862,
      "loss": 1.6018,
      "step": 7106
    },
    {
      "epoch": 0.2545168048417999,
      "grad_norm": 1.7628570795059204,
      "learning_rate": 0.0001747077189429959,
      "loss": 1.3708,
      "step": 7107
    },
    {
      "epoch": 0.2545526169713682,
      "grad_norm": 1.5150426626205444,
      "learning_rate": 0.00017470000814396718,
      "loss": 1.7994,
      "step": 7108
    },
    {
      "epoch": 0.2545884291009365,
      "grad_norm": 2.202345371246338,
      "learning_rate": 0.0001746922963399462,
      "loss": 1.3141,
      "step": 7109
    },
    {
      "epoch": 0.25462424123050476,
      "grad_norm": 1.6175556182861328,
      "learning_rate": 0.00017468458353103676,
      "loss": 1.4706,
      "step": 7110
    },
    {
      "epoch": 0.2546600533600731,
      "grad_norm": 1.6333051919937134,
      "learning_rate": 0.00017467686971734257,
      "loss": 1.621,
      "step": 7111
    },
    {
      "epoch": 0.25469586548964135,
      "grad_norm": 1.4622938632965088,
      "learning_rate": 0.0001746691548989674,
      "loss": 1.1874,
      "step": 7112
    },
    {
      "epoch": 0.2547316776192096,
      "grad_norm": 1.556118369102478,
      "learning_rate": 0.00017466143907601508,
      "loss": 1.5941,
      "step": 7113
    },
    {
      "epoch": 0.2547674897487779,
      "grad_norm": 1.436787724494934,
      "learning_rate": 0.00017465372224858937,
      "loss": 1.5353,
      "step": 7114
    },
    {
      "epoch": 0.2548033018783462,
      "grad_norm": 1.2040411233901978,
      "learning_rate": 0.00017464600441679417,
      "loss": 1.5374,
      "step": 7115
    },
    {
      "epoch": 0.2548391140079145,
      "grad_norm": 1.6581120491027832,
      "learning_rate": 0.0001746382855807333,
      "loss": 1.5132,
      "step": 7116
    },
    {
      "epoch": 0.25487492613748275,
      "grad_norm": 1.7110474109649658,
      "learning_rate": 0.0001746305657405105,
      "loss": 1.5188,
      "step": 7117
    },
    {
      "epoch": 0.2549107382670511,
      "grad_norm": 1.6720296144485474,
      "learning_rate": 0.00017462284489622973,
      "loss": 1.4921,
      "step": 7118
    },
    {
      "epoch": 0.25494655039661934,
      "grad_norm": 1.5299689769744873,
      "learning_rate": 0.00017461512304799484,
      "loss": 1.4806,
      "step": 7119
    },
    {
      "epoch": 0.2549823625261876,
      "grad_norm": 1.5790822505950928,
      "learning_rate": 0.0001746074001959097,
      "loss": 1.2113,
      "step": 7120
    },
    {
      "epoch": 0.2550181746557559,
      "grad_norm": 2.029690980911255,
      "learning_rate": 0.00017459967634007826,
      "loss": 1.6939,
      "step": 7121
    },
    {
      "epoch": 0.2550539867853242,
      "grad_norm": 1.6915394067764282,
      "learning_rate": 0.00017459195148060438,
      "loss": 1.2217,
      "step": 7122
    },
    {
      "epoch": 0.2550897989148925,
      "grad_norm": 2.6420986652374268,
      "learning_rate": 0.00017458422561759203,
      "loss": 2.0125,
      "step": 7123
    },
    {
      "epoch": 0.25512561104446074,
      "grad_norm": 1.7383835315704346,
      "learning_rate": 0.0001745764987511451,
      "loss": 1.2076,
      "step": 7124
    },
    {
      "epoch": 0.25516142317402907,
      "grad_norm": 1.5973914861679077,
      "learning_rate": 0.0001745687708813676,
      "loss": 1.5747,
      "step": 7125
    },
    {
      "epoch": 0.25519723530359734,
      "grad_norm": 1.8514518737792969,
      "learning_rate": 0.00017456104200836347,
      "loss": 1.3596,
      "step": 7126
    },
    {
      "epoch": 0.2552330474331656,
      "grad_norm": 1.7249946594238281,
      "learning_rate": 0.00017455331213223668,
      "loss": 1.7315,
      "step": 7127
    },
    {
      "epoch": 0.2552688595627339,
      "grad_norm": 1.592330813407898,
      "learning_rate": 0.00017454558125309125,
      "loss": 1.4371,
      "step": 7128
    },
    {
      "epoch": 0.2553046716923022,
      "grad_norm": 4.213657379150391,
      "learning_rate": 0.00017453784937103122,
      "loss": 1.464,
      "step": 7129
    },
    {
      "epoch": 0.25534048382187047,
      "grad_norm": 1.4696651697158813,
      "learning_rate": 0.00017453011648616053,
      "loss": 1.4912,
      "step": 7130
    },
    {
      "epoch": 0.25537629595143874,
      "grad_norm": 1.5843778848648071,
      "learning_rate": 0.00017452238259858327,
      "loss": 1.6334,
      "step": 7131
    },
    {
      "epoch": 0.25541210808100706,
      "grad_norm": 1.2764482498168945,
      "learning_rate": 0.00017451464770840348,
      "loss": 1.5292,
      "step": 7132
    },
    {
      "epoch": 0.25544792021057533,
      "grad_norm": 1.7480367422103882,
      "learning_rate": 0.00017450691181572522,
      "loss": 1.387,
      "step": 7133
    },
    {
      "epoch": 0.2554837323401436,
      "grad_norm": 2.2001123428344727,
      "learning_rate": 0.00017449917492065256,
      "loss": 1.4539,
      "step": 7134
    },
    {
      "epoch": 0.25551954446971187,
      "grad_norm": 1.604820728302002,
      "learning_rate": 0.0001744914370232896,
      "loss": 1.5517,
      "step": 7135
    },
    {
      "epoch": 0.2555553565992802,
      "grad_norm": 1.7131909132003784,
      "learning_rate": 0.00017448369812374045,
      "loss": 1.4457,
      "step": 7136
    },
    {
      "epoch": 0.25559116872884846,
      "grad_norm": 1.6699351072311401,
      "learning_rate": 0.00017447595822210924,
      "loss": 1.2495,
      "step": 7137
    },
    {
      "epoch": 0.25562698085841673,
      "grad_norm": 1.6325026750564575,
      "learning_rate": 0.00017446821731850008,
      "loss": 1.4587,
      "step": 7138
    },
    {
      "epoch": 0.25566279298798505,
      "grad_norm": 1.5442523956298828,
      "learning_rate": 0.00017446047541301707,
      "loss": 1.4748,
      "step": 7139
    },
    {
      "epoch": 0.2556986051175533,
      "grad_norm": 2.9096853733062744,
      "learning_rate": 0.00017445273250576442,
      "loss": 1.6572,
      "step": 7140
    },
    {
      "epoch": 0.2557344172471216,
      "grad_norm": 1.796773910522461,
      "learning_rate": 0.0001744449885968463,
      "loss": 1.4552,
      "step": 7141
    },
    {
      "epoch": 0.25577022937668986,
      "grad_norm": 2.095973253250122,
      "learning_rate": 0.00017443724368636693,
      "loss": 1.4722,
      "step": 7142
    },
    {
      "epoch": 0.2558060415062582,
      "grad_norm": 1.7989848852157593,
      "learning_rate": 0.00017442949777443038,
      "loss": 1.7267,
      "step": 7143
    },
    {
      "epoch": 0.25584185363582646,
      "grad_norm": 1.8527088165283203,
      "learning_rate": 0.000174421750861141,
      "loss": 1.5778,
      "step": 7144
    },
    {
      "epoch": 0.2558776657653947,
      "grad_norm": 1.6256322860717773,
      "learning_rate": 0.00017441400294660294,
      "loss": 1.6681,
      "step": 7145
    },
    {
      "epoch": 0.25591347789496305,
      "grad_norm": 1.3841743469238281,
      "learning_rate": 0.00017440625403092045,
      "loss": 1.5331,
      "step": 7146
    },
    {
      "epoch": 0.2559492900245313,
      "grad_norm": 1.9727689027786255,
      "learning_rate": 0.00017439850411419782,
      "loss": 1.5752,
      "step": 7147
    },
    {
      "epoch": 0.2559851021540996,
      "grad_norm": 1.6270530223846436,
      "learning_rate": 0.00017439075319653928,
      "loss": 1.7994,
      "step": 7148
    },
    {
      "epoch": 0.25602091428366786,
      "grad_norm": 1.7818320989608765,
      "learning_rate": 0.0001743830012780491,
      "loss": 1.6047,
      "step": 7149
    },
    {
      "epoch": 0.2560567264132362,
      "grad_norm": 1.760941505432129,
      "learning_rate": 0.00017437524835883157,
      "loss": 1.6809,
      "step": 7150
    },
    {
      "epoch": 0.25609253854280445,
      "grad_norm": 1.2127779722213745,
      "learning_rate": 0.00017436749443899103,
      "loss": 1.6395,
      "step": 7151
    },
    {
      "epoch": 0.2561283506723727,
      "grad_norm": 2.064774751663208,
      "learning_rate": 0.00017435973951863179,
      "loss": 1.5165,
      "step": 7152
    },
    {
      "epoch": 0.25616416280194104,
      "grad_norm": 2.9734225273132324,
      "learning_rate": 0.0001743519835978581,
      "loss": 1.3607,
      "step": 7153
    },
    {
      "epoch": 0.2561999749315093,
      "grad_norm": 1.5963677167892456,
      "learning_rate": 0.00017434422667677446,
      "loss": 1.6676,
      "step": 7154
    },
    {
      "epoch": 0.2562357870610776,
      "grad_norm": 1.5437045097351074,
      "learning_rate": 0.00017433646875548512,
      "loss": 1.583,
      "step": 7155
    },
    {
      "epoch": 0.25627159919064585,
      "grad_norm": 1.6270062923431396,
      "learning_rate": 0.0001743287098340945,
      "loss": 1.4786,
      "step": 7156
    },
    {
      "epoch": 0.2563074113202142,
      "grad_norm": 1.7923583984375,
      "learning_rate": 0.00017432094991270692,
      "loss": 1.6218,
      "step": 7157
    },
    {
      "epoch": 0.25634322344978244,
      "grad_norm": 2.1703848838806152,
      "learning_rate": 0.00017431318899142686,
      "loss": 2.0572,
      "step": 7158
    },
    {
      "epoch": 0.2563790355793507,
      "grad_norm": 1.547512173652649,
      "learning_rate": 0.0001743054270703587,
      "loss": 1.4179,
      "step": 7159
    },
    {
      "epoch": 0.25641484770891904,
      "grad_norm": 1.7714463472366333,
      "learning_rate": 0.00017429766414960685,
      "loss": 1.5927,
      "step": 7160
    },
    {
      "epoch": 0.2564506598384873,
      "grad_norm": 2.190034866333008,
      "learning_rate": 0.0001742899002292758,
      "loss": 1.8858,
      "step": 7161
    },
    {
      "epoch": 0.2564864719680556,
      "grad_norm": 1.3032276630401611,
      "learning_rate": 0.00017428213530946995,
      "loss": 1.6218,
      "step": 7162
    },
    {
      "epoch": 0.25652228409762384,
      "grad_norm": 1.72031569480896,
      "learning_rate": 0.00017427436939029378,
      "loss": 1.6231,
      "step": 7163
    },
    {
      "epoch": 0.25655809622719217,
      "grad_norm": 1.9104535579681396,
      "learning_rate": 0.00017426660247185177,
      "loss": 1.4184,
      "step": 7164
    },
    {
      "epoch": 0.25659390835676044,
      "grad_norm": 1.665347933769226,
      "learning_rate": 0.0001742588345542484,
      "loss": 1.9099,
      "step": 7165
    },
    {
      "epoch": 0.2566297204863287,
      "grad_norm": 1.2468254566192627,
      "learning_rate": 0.00017425106563758824,
      "loss": 1.458,
      "step": 7166
    },
    {
      "epoch": 0.25666553261589703,
      "grad_norm": 2.2506954669952393,
      "learning_rate": 0.00017424329572197578,
      "loss": 1.6085,
      "step": 7167
    },
    {
      "epoch": 0.2567013447454653,
      "grad_norm": 1.5501458644866943,
      "learning_rate": 0.0001742355248075155,
      "loss": 1.4801,
      "step": 7168
    },
    {
      "epoch": 0.25673715687503357,
      "grad_norm": 1.3393751382827759,
      "learning_rate": 0.00017422775289431202,
      "loss": 1.4539,
      "step": 7169
    },
    {
      "epoch": 0.25677296900460184,
      "grad_norm": 1.88332200050354,
      "learning_rate": 0.00017421997998246985,
      "loss": 1.6451,
      "step": 7170
    },
    {
      "epoch": 0.25680878113417016,
      "grad_norm": 2.4323949813842773,
      "learning_rate": 0.0001742122060720936,
      "loss": 1.7289,
      "step": 7171
    },
    {
      "epoch": 0.25684459326373843,
      "grad_norm": 2.0443127155303955,
      "learning_rate": 0.00017420443116328784,
      "loss": 1.3284,
      "step": 7172
    },
    {
      "epoch": 0.2568804053933067,
      "grad_norm": 1.3344660997390747,
      "learning_rate": 0.0001741966552561572,
      "loss": 1.3553,
      "step": 7173
    },
    {
      "epoch": 0.256916217522875,
      "grad_norm": 1.9592489004135132,
      "learning_rate": 0.00017418887835080624,
      "loss": 1.3423,
      "step": 7174
    },
    {
      "epoch": 0.2569520296524433,
      "grad_norm": 1.8439921140670776,
      "learning_rate": 0.0001741811004473396,
      "loss": 1.4173,
      "step": 7175
    },
    {
      "epoch": 0.25698784178201156,
      "grad_norm": 2.6069836616516113,
      "learning_rate": 0.000174173321545862,
      "loss": 1.42,
      "step": 7176
    },
    {
      "epoch": 0.25702365391157983,
      "grad_norm": 1.5937824249267578,
      "learning_rate": 0.000174165541646478,
      "loss": 1.5705,
      "step": 7177
    },
    {
      "epoch": 0.25705946604114815,
      "grad_norm": 1.4017707109451294,
      "learning_rate": 0.0001741577607492923,
      "loss": 1.4923,
      "step": 7178
    },
    {
      "epoch": 0.2570952781707164,
      "grad_norm": 1.6453132629394531,
      "learning_rate": 0.00017414997885440957,
      "loss": 1.6204,
      "step": 7179
    },
    {
      "epoch": 0.2571310903002847,
      "grad_norm": 1.494167685508728,
      "learning_rate": 0.00017414219596193455,
      "loss": 1.2734,
      "step": 7180
    },
    {
      "epoch": 0.257166902429853,
      "grad_norm": 1.562565565109253,
      "learning_rate": 0.0001741344120719719,
      "loss": 1.327,
      "step": 7181
    },
    {
      "epoch": 0.2572027145594213,
      "grad_norm": 1.6081397533416748,
      "learning_rate": 0.00017412662718462637,
      "loss": 1.546,
      "step": 7182
    },
    {
      "epoch": 0.25723852668898955,
      "grad_norm": 1.5741753578186035,
      "learning_rate": 0.00017411884130000271,
      "loss": 1.6006,
      "step": 7183
    },
    {
      "epoch": 0.2572743388185578,
      "grad_norm": 2.1187093257904053,
      "learning_rate": 0.00017411105441820563,
      "loss": 1.5594,
      "step": 7184
    },
    {
      "epoch": 0.25731015094812615,
      "grad_norm": 1.4211379289627075,
      "learning_rate": 0.0001741032665393399,
      "loss": 1.6873,
      "step": 7185
    },
    {
      "epoch": 0.2573459630776944,
      "grad_norm": 1.706182837486267,
      "learning_rate": 0.00017409547766351034,
      "loss": 1.3723,
      "step": 7186
    },
    {
      "epoch": 0.2573817752072627,
      "grad_norm": 1.8475958108901978,
      "learning_rate": 0.00017408768779082165,
      "loss": 1.4974,
      "step": 7187
    },
    {
      "epoch": 0.257417587336831,
      "grad_norm": 1.5614365339279175,
      "learning_rate": 0.00017407989692137872,
      "loss": 1.5942,
      "step": 7188
    },
    {
      "epoch": 0.2574533994663993,
      "grad_norm": 2.806450843811035,
      "learning_rate": 0.0001740721050552863,
      "loss": 1.6668,
      "step": 7189
    },
    {
      "epoch": 0.25748921159596755,
      "grad_norm": 2.1845903396606445,
      "learning_rate": 0.0001740643121926493,
      "loss": 1.4023,
      "step": 7190
    },
    {
      "epoch": 0.2575250237255358,
      "grad_norm": 1.9280097484588623,
      "learning_rate": 0.0001740565183335725,
      "loss": 1.4899,
      "step": 7191
    },
    {
      "epoch": 0.25756083585510414,
      "grad_norm": 1.400450348854065,
      "learning_rate": 0.00017404872347816076,
      "loss": 1.2537,
      "step": 7192
    },
    {
      "epoch": 0.2575966479846724,
      "grad_norm": 2.4562292098999023,
      "learning_rate": 0.00017404092762651898,
      "loss": 1.501,
      "step": 7193
    },
    {
      "epoch": 0.2576324601142407,
      "grad_norm": 2.2671611309051514,
      "learning_rate": 0.000174033130778752,
      "loss": 1.4855,
      "step": 7194
    },
    {
      "epoch": 0.257668272243809,
      "grad_norm": 1.6012718677520752,
      "learning_rate": 0.00017402533293496477,
      "loss": 1.4119,
      "step": 7195
    },
    {
      "epoch": 0.2577040843733773,
      "grad_norm": 1.4546482563018799,
      "learning_rate": 0.00017401753409526216,
      "loss": 1.6494,
      "step": 7196
    },
    {
      "epoch": 0.25773989650294554,
      "grad_norm": 2.1474809646606445,
      "learning_rate": 0.0001740097342597491,
      "loss": 1.6661,
      "step": 7197
    },
    {
      "epoch": 0.2577757086325138,
      "grad_norm": 1.6500005722045898,
      "learning_rate": 0.0001740019334285305,
      "loss": 1.4682,
      "step": 7198
    },
    {
      "epoch": 0.25781152076208214,
      "grad_norm": 1.741763710975647,
      "learning_rate": 0.0001739941316017114,
      "loss": 1.4744,
      "step": 7199
    },
    {
      "epoch": 0.2578473328916504,
      "grad_norm": 2.0389153957366943,
      "learning_rate": 0.00017398632877939666,
      "loss": 1.584,
      "step": 7200
    },
    {
      "epoch": 0.2578831450212187,
      "grad_norm": 1.478593111038208,
      "learning_rate": 0.00017397852496169134,
      "loss": 1.4449,
      "step": 7201
    },
    {
      "epoch": 0.257918957150787,
      "grad_norm": 2.4067537784576416,
      "learning_rate": 0.00017397072014870037,
      "loss": 1.3542,
      "step": 7202
    },
    {
      "epoch": 0.25795476928035527,
      "grad_norm": 2.244966506958008,
      "learning_rate": 0.00017396291434052877,
      "loss": 1.4444,
      "step": 7203
    },
    {
      "epoch": 0.25799058140992354,
      "grad_norm": 1.5638600587844849,
      "learning_rate": 0.00017395510753728157,
      "loss": 1.441,
      "step": 7204
    },
    {
      "epoch": 0.2580263935394918,
      "grad_norm": 1.8143665790557861,
      "learning_rate": 0.0001739472997390638,
      "loss": 1.7743,
      "step": 7205
    },
    {
      "epoch": 0.25806220566906013,
      "grad_norm": 1.868913173675537,
      "learning_rate": 0.00017393949094598047,
      "loss": 1.3007,
      "step": 7206
    },
    {
      "epoch": 0.2580980177986284,
      "grad_norm": 1.5457756519317627,
      "learning_rate": 0.00017393168115813673,
      "loss": 1.6073,
      "step": 7207
    },
    {
      "epoch": 0.25813382992819667,
      "grad_norm": 1.3162868022918701,
      "learning_rate": 0.0001739238703756375,
      "loss": 1.5155,
      "step": 7208
    },
    {
      "epoch": 0.258169642057765,
      "grad_norm": 1.955180287361145,
      "learning_rate": 0.00017391605859858798,
      "loss": 1.5373,
      "step": 7209
    },
    {
      "epoch": 0.25820545418733326,
      "grad_norm": 1.7129416465759277,
      "learning_rate": 0.00017390824582709326,
      "loss": 1.8984,
      "step": 7210
    },
    {
      "epoch": 0.25824126631690153,
      "grad_norm": 1.3623757362365723,
      "learning_rate": 0.0001739004320612584,
      "loss": 1.5941,
      "step": 7211
    },
    {
      "epoch": 0.2582770784464698,
      "grad_norm": 1.5919731855392456,
      "learning_rate": 0.00017389261730118858,
      "loss": 1.6802,
      "step": 7212
    },
    {
      "epoch": 0.2583128905760381,
      "grad_norm": 1.5978248119354248,
      "learning_rate": 0.0001738848015469889,
      "loss": 1.4566,
      "step": 7213
    },
    {
      "epoch": 0.2583487027056064,
      "grad_norm": 2.2260446548461914,
      "learning_rate": 0.0001738769847987645,
      "loss": 1.2414,
      "step": 7214
    },
    {
      "epoch": 0.25838451483517466,
      "grad_norm": 1.7300844192504883,
      "learning_rate": 0.0001738691670566206,
      "loss": 1.3088,
      "step": 7215
    },
    {
      "epoch": 0.258420326964743,
      "grad_norm": 1.403152346611023,
      "learning_rate": 0.0001738613483206623,
      "loss": 1.607,
      "step": 7216
    },
    {
      "epoch": 0.25845613909431125,
      "grad_norm": 1.4186724424362183,
      "learning_rate": 0.00017385352859099483,
      "loss": 1.6773,
      "step": 7217
    },
    {
      "epoch": 0.2584919512238795,
      "grad_norm": 1.7511612176895142,
      "learning_rate": 0.00017384570786772345,
      "loss": 1.8717,
      "step": 7218
    },
    {
      "epoch": 0.2585277633534478,
      "grad_norm": 2.500760793685913,
      "learning_rate": 0.00017383788615095327,
      "loss": 1.6079,
      "step": 7219
    },
    {
      "epoch": 0.2585635754830161,
      "grad_norm": 1.5331605672836304,
      "learning_rate": 0.0001738300634407896,
      "loss": 1.4821,
      "step": 7220
    },
    {
      "epoch": 0.2585993876125844,
      "grad_norm": 1.4042764902114868,
      "learning_rate": 0.00017382223973733767,
      "loss": 1.5336,
      "step": 7221
    },
    {
      "epoch": 0.25863519974215265,
      "grad_norm": 1.2958499193191528,
      "learning_rate": 0.0001738144150407027,
      "loss": 1.5881,
      "step": 7222
    },
    {
      "epoch": 0.258671011871721,
      "grad_norm": 1.417944073677063,
      "learning_rate": 0.00017380658935099,
      "loss": 1.6748,
      "step": 7223
    },
    {
      "epoch": 0.25870682400128925,
      "grad_norm": 1.7564271688461304,
      "learning_rate": 0.00017379876266830486,
      "loss": 1.5369,
      "step": 7224
    },
    {
      "epoch": 0.2587426361308575,
      "grad_norm": 1.4452077150344849,
      "learning_rate": 0.00017379093499275258,
      "loss": 1.6696,
      "step": 7225
    },
    {
      "epoch": 0.2587784482604258,
      "grad_norm": 1.0980790853500366,
      "learning_rate": 0.00017378310632443843,
      "loss": 1.2389,
      "step": 7226
    },
    {
      "epoch": 0.2588142603899941,
      "grad_norm": 1.6883370876312256,
      "learning_rate": 0.00017377527666346772,
      "loss": 1.6746,
      "step": 7227
    },
    {
      "epoch": 0.2588500725195624,
      "grad_norm": 1.3964970111846924,
      "learning_rate": 0.00017376744600994587,
      "loss": 1.6275,
      "step": 7228
    },
    {
      "epoch": 0.25888588464913065,
      "grad_norm": 2.0311527252197266,
      "learning_rate": 0.00017375961436397818,
      "loss": 1.3165,
      "step": 7229
    },
    {
      "epoch": 0.2589216967786989,
      "grad_norm": 1.7014087438583374,
      "learning_rate": 0.00017375178172567002,
      "loss": 1.5707,
      "step": 7230
    },
    {
      "epoch": 0.25895750890826724,
      "grad_norm": 1.977447748184204,
      "learning_rate": 0.00017374394809512676,
      "loss": 1.5756,
      "step": 7231
    },
    {
      "epoch": 0.2589933210378355,
      "grad_norm": 1.3852441310882568,
      "learning_rate": 0.0001737361134724538,
      "loss": 1.4648,
      "step": 7232
    },
    {
      "epoch": 0.2590291331674038,
      "grad_norm": 1.8260564804077148,
      "learning_rate": 0.00017372827785775655,
      "loss": 1.8661,
      "step": 7233
    },
    {
      "epoch": 0.2590649452969721,
      "grad_norm": 1.7901363372802734,
      "learning_rate": 0.00017372044125114045,
      "loss": 1.3408,
      "step": 7234
    },
    {
      "epoch": 0.2591007574265404,
      "grad_norm": 2.0273873805999756,
      "learning_rate": 0.0001737126036527109,
      "loss": 1.5693,
      "step": 7235
    },
    {
      "epoch": 0.25913656955610864,
      "grad_norm": 1.6030428409576416,
      "learning_rate": 0.00017370476506257333,
      "loss": 1.444,
      "step": 7236
    },
    {
      "epoch": 0.2591723816856769,
      "grad_norm": 1.4109402894973755,
      "learning_rate": 0.0001736969254808332,
      "loss": 1.3457,
      "step": 7237
    },
    {
      "epoch": 0.25920819381524524,
      "grad_norm": 2.0388617515563965,
      "learning_rate": 0.00017368908490759605,
      "loss": 1.7262,
      "step": 7238
    },
    {
      "epoch": 0.2592440059448135,
      "grad_norm": 2.0132298469543457,
      "learning_rate": 0.00017368124334296727,
      "loss": 1.3662,
      "step": 7239
    },
    {
      "epoch": 0.2592798180743818,
      "grad_norm": 1.9762897491455078,
      "learning_rate": 0.00017367340078705242,
      "loss": 1.5578,
      "step": 7240
    },
    {
      "epoch": 0.2593156302039501,
      "grad_norm": 2.113196611404419,
      "learning_rate": 0.000173665557239957,
      "loss": 1.7907,
      "step": 7241
    },
    {
      "epoch": 0.25935144233351837,
      "grad_norm": 2.036367416381836,
      "learning_rate": 0.00017365771270178652,
      "loss": 1.6257,
      "step": 7242
    },
    {
      "epoch": 0.25938725446308664,
      "grad_norm": 1.8975567817687988,
      "learning_rate": 0.00017364986717264652,
      "loss": 1.7479,
      "step": 7243
    },
    {
      "epoch": 0.2594230665926549,
      "grad_norm": 1.5035462379455566,
      "learning_rate": 0.00017364202065264258,
      "loss": 1.7363,
      "step": 7244
    },
    {
      "epoch": 0.25945887872222323,
      "grad_norm": 2.175387382507324,
      "learning_rate": 0.00017363417314188024,
      "loss": 1.8906,
      "step": 7245
    },
    {
      "epoch": 0.2594946908517915,
      "grad_norm": 1.3993414640426636,
      "learning_rate": 0.00017362632464046506,
      "loss": 1.5836,
      "step": 7246
    },
    {
      "epoch": 0.25953050298135977,
      "grad_norm": 1.9026535749435425,
      "learning_rate": 0.00017361847514850266,
      "loss": 1.5866,
      "step": 7247
    },
    {
      "epoch": 0.2595663151109281,
      "grad_norm": 2.123619318008423,
      "learning_rate": 0.00017361062466609867,
      "loss": 1.1663,
      "step": 7248
    },
    {
      "epoch": 0.25960212724049636,
      "grad_norm": 1.8966612815856934,
      "learning_rate": 0.00017360277319335865,
      "loss": 1.5483,
      "step": 7249
    },
    {
      "epoch": 0.25963793937006463,
      "grad_norm": 1.919511079788208,
      "learning_rate": 0.00017359492073038826,
      "loss": 1.7607,
      "step": 7250
    },
    {
      "epoch": 0.2596737514996329,
      "grad_norm": 1.6821178197860718,
      "learning_rate": 0.00017358706727729311,
      "loss": 1.3796,
      "step": 7251
    },
    {
      "epoch": 0.2597095636292012,
      "grad_norm": 1.4946900606155396,
      "learning_rate": 0.00017357921283417892,
      "loss": 1.7299,
      "step": 7252
    },
    {
      "epoch": 0.2597453757587695,
      "grad_norm": 1.5954385995864868,
      "learning_rate": 0.00017357135740115137,
      "loss": 1.4342,
      "step": 7253
    },
    {
      "epoch": 0.25978118788833776,
      "grad_norm": 1.8522142171859741,
      "learning_rate": 0.00017356350097831605,
      "loss": 1.4027,
      "step": 7254
    },
    {
      "epoch": 0.2598170000179061,
      "grad_norm": 1.701069712638855,
      "learning_rate": 0.00017355564356577873,
      "loss": 1.6356,
      "step": 7255
    },
    {
      "epoch": 0.25985281214747435,
      "grad_norm": 2.1512773036956787,
      "learning_rate": 0.00017354778516364512,
      "loss": 1.4857,
      "step": 7256
    },
    {
      "epoch": 0.2598886242770426,
      "grad_norm": 2.0113370418548584,
      "learning_rate": 0.00017353992577202093,
      "loss": 1.7957,
      "step": 7257
    },
    {
      "epoch": 0.2599244364066109,
      "grad_norm": 1.7597367763519287,
      "learning_rate": 0.00017353206539101186,
      "loss": 1.4799,
      "step": 7258
    },
    {
      "epoch": 0.2599602485361792,
      "grad_norm": 1.4501522779464722,
      "learning_rate": 0.00017352420402072375,
      "loss": 1.4439,
      "step": 7259
    },
    {
      "epoch": 0.2599960606657475,
      "grad_norm": 1.3950390815734863,
      "learning_rate": 0.00017351634166126227,
      "loss": 1.6883,
      "step": 7260
    },
    {
      "epoch": 0.26003187279531575,
      "grad_norm": 1.6042085886001587,
      "learning_rate": 0.00017350847831273329,
      "loss": 1.7091,
      "step": 7261
    },
    {
      "epoch": 0.2600676849248841,
      "grad_norm": 1.632921814918518,
      "learning_rate": 0.00017350061397524252,
      "loss": 1.541,
      "step": 7262
    },
    {
      "epoch": 0.26010349705445235,
      "grad_norm": 1.503475546836853,
      "learning_rate": 0.0001734927486488958,
      "loss": 1.5553,
      "step": 7263
    },
    {
      "epoch": 0.2601393091840206,
      "grad_norm": 1.7809571027755737,
      "learning_rate": 0.00017348488233379897,
      "loss": 1.8013,
      "step": 7264
    },
    {
      "epoch": 0.2601751213135889,
      "grad_norm": 1.4528809785842896,
      "learning_rate": 0.0001734770150300578,
      "loss": 1.5241,
      "step": 7265
    },
    {
      "epoch": 0.2602109334431572,
      "grad_norm": 1.5344305038452148,
      "learning_rate": 0.00017346914673777822,
      "loss": 1.7202,
      "step": 7266
    },
    {
      "epoch": 0.2602467455727255,
      "grad_norm": 1.6437627077102661,
      "learning_rate": 0.000173461277457066,
      "loss": 1.4926,
      "step": 7267
    },
    {
      "epoch": 0.26028255770229375,
      "grad_norm": 1.5676155090332031,
      "learning_rate": 0.00017345340718802704,
      "loss": 1.6236,
      "step": 7268
    },
    {
      "epoch": 0.2603183698318621,
      "grad_norm": 2.03763747215271,
      "learning_rate": 0.00017344553593076726,
      "loss": 1.3897,
      "step": 7269
    },
    {
      "epoch": 0.26035418196143034,
      "grad_norm": 2.440800666809082,
      "learning_rate": 0.00017343766368539253,
      "loss": 1.7987,
      "step": 7270
    },
    {
      "epoch": 0.2603899940909986,
      "grad_norm": 1.6920621395111084,
      "learning_rate": 0.00017342979045200876,
      "loss": 1.8625,
      "step": 7271
    },
    {
      "epoch": 0.2604258062205669,
      "grad_norm": 1.7777220010757446,
      "learning_rate": 0.00017342191623072187,
      "loss": 1.375,
      "step": 7272
    },
    {
      "epoch": 0.2604616183501352,
      "grad_norm": 1.4054484367370605,
      "learning_rate": 0.00017341404102163782,
      "loss": 1.64,
      "step": 7273
    },
    {
      "epoch": 0.2604974304797035,
      "grad_norm": 2.3189139366149902,
      "learning_rate": 0.00017340616482486253,
      "loss": 1.6583,
      "step": 7274
    },
    {
      "epoch": 0.26053324260927174,
      "grad_norm": 1.4085335731506348,
      "learning_rate": 0.00017339828764050198,
      "loss": 1.3233,
      "step": 7275
    },
    {
      "epoch": 0.26056905473884007,
      "grad_norm": 1.6071890592575073,
      "learning_rate": 0.00017339040946866217,
      "loss": 1.6617,
      "step": 7276
    },
    {
      "epoch": 0.26060486686840834,
      "grad_norm": 1.9245442152023315,
      "learning_rate": 0.00017338253030944905,
      "loss": 1.7055,
      "step": 7277
    },
    {
      "epoch": 0.2606406789979766,
      "grad_norm": 2.0860061645507812,
      "learning_rate": 0.00017337465016296864,
      "loss": 1.5268,
      "step": 7278
    },
    {
      "epoch": 0.2606764911275449,
      "grad_norm": 1.7437604665756226,
      "learning_rate": 0.00017336676902932695,
      "loss": 1.4931,
      "step": 7279
    },
    {
      "epoch": 0.2607123032571132,
      "grad_norm": 2.2297158241271973,
      "learning_rate": 0.00017335888690863,
      "loss": 1.4163,
      "step": 7280
    },
    {
      "epoch": 0.26074811538668147,
      "grad_norm": 1.6739795207977295,
      "learning_rate": 0.00017335100380098392,
      "loss": 1.6585,
      "step": 7281
    },
    {
      "epoch": 0.26078392751624974,
      "grad_norm": 1.6758095026016235,
      "learning_rate": 0.00017334311970649465,
      "loss": 1.5066,
      "step": 7282
    },
    {
      "epoch": 0.26081973964581806,
      "grad_norm": 1.510539174079895,
      "learning_rate": 0.00017333523462526832,
      "loss": 1.5937,
      "step": 7283
    },
    {
      "epoch": 0.26085555177538633,
      "grad_norm": 2.0457797050476074,
      "learning_rate": 0.000173327348557411,
      "loss": 1.6134,
      "step": 7284
    },
    {
      "epoch": 0.2608913639049546,
      "grad_norm": 1.2520917654037476,
      "learning_rate": 0.00017331946150302878,
      "loss": 1.4669,
      "step": 7285
    },
    {
      "epoch": 0.26092717603452287,
      "grad_norm": 1.4760613441467285,
      "learning_rate": 0.00017331157346222779,
      "loss": 1.3615,
      "step": 7286
    },
    {
      "epoch": 0.2609629881640912,
      "grad_norm": 1.5453470945358276,
      "learning_rate": 0.00017330368443511417,
      "loss": 1.6023,
      "step": 7287
    },
    {
      "epoch": 0.26099880029365946,
      "grad_norm": 1.4087209701538086,
      "learning_rate": 0.00017329579442179401,
      "loss": 1.7286,
      "step": 7288
    },
    {
      "epoch": 0.26103461242322773,
      "grad_norm": 1.5778745412826538,
      "learning_rate": 0.00017328790342237347,
      "loss": 1.4194,
      "step": 7289
    },
    {
      "epoch": 0.26107042455279605,
      "grad_norm": 1.4336007833480835,
      "learning_rate": 0.00017328001143695874,
      "loss": 1.2318,
      "step": 7290
    },
    {
      "epoch": 0.2611062366823643,
      "grad_norm": 1.543596625328064,
      "learning_rate": 0.00017327211846565596,
      "loss": 1.5086,
      "step": 7291
    },
    {
      "epoch": 0.2611420488119326,
      "grad_norm": 1.4977967739105225,
      "learning_rate": 0.0001732642245085714,
      "loss": 1.5653,
      "step": 7292
    },
    {
      "epoch": 0.26117786094150086,
      "grad_norm": 1.346598744392395,
      "learning_rate": 0.00017325632956581113,
      "loss": 1.5548,
      "step": 7293
    },
    {
      "epoch": 0.2612136730710692,
      "grad_norm": 2.323775053024292,
      "learning_rate": 0.00017324843363748148,
      "loss": 1.521,
      "step": 7294
    },
    {
      "epoch": 0.26124948520063745,
      "grad_norm": 1.5547338724136353,
      "learning_rate": 0.00017324053672368862,
      "loss": 1.5629,
      "step": 7295
    },
    {
      "epoch": 0.2612852973302057,
      "grad_norm": 1.9380877017974854,
      "learning_rate": 0.0001732326388245388,
      "loss": 1.5277,
      "step": 7296
    },
    {
      "epoch": 0.26132110945977405,
      "grad_norm": 2.077476739883423,
      "learning_rate": 0.00017322473994013833,
      "loss": 1.4113,
      "step": 7297
    },
    {
      "epoch": 0.2613569215893423,
      "grad_norm": 1.7269058227539062,
      "learning_rate": 0.00017321684007059343,
      "loss": 1.4078,
      "step": 7298
    },
    {
      "epoch": 0.2613927337189106,
      "grad_norm": 1.5610759258270264,
      "learning_rate": 0.00017320893921601036,
      "loss": 1.348,
      "step": 7299
    },
    {
      "epoch": 0.26142854584847885,
      "grad_norm": 1.4764721393585205,
      "learning_rate": 0.00017320103737649548,
      "loss": 1.3946,
      "step": 7300
    },
    {
      "epoch": 0.2614643579780472,
      "grad_norm": 1.443997859954834,
      "learning_rate": 0.00017319313455215504,
      "loss": 1.4114,
      "step": 7301
    },
    {
      "epoch": 0.26150017010761545,
      "grad_norm": 1.4743808507919312,
      "learning_rate": 0.00017318523074309538,
      "loss": 1.7602,
      "step": 7302
    },
    {
      "epoch": 0.2615359822371837,
      "grad_norm": 1.5797048807144165,
      "learning_rate": 0.00017317732594942286,
      "loss": 1.5341,
      "step": 7303
    },
    {
      "epoch": 0.26157179436675204,
      "grad_norm": 2.011399745941162,
      "learning_rate": 0.0001731694201712438,
      "loss": 1.5281,
      "step": 7304
    },
    {
      "epoch": 0.2616076064963203,
      "grad_norm": 3.038586378097534,
      "learning_rate": 0.0001731615134086646,
      "loss": 1.4805,
      "step": 7305
    },
    {
      "epoch": 0.2616434186258886,
      "grad_norm": 1.6645811796188354,
      "learning_rate": 0.00017315360566179158,
      "loss": 1.4753,
      "step": 7306
    },
    {
      "epoch": 0.26167923075545685,
      "grad_norm": 1.668992519378662,
      "learning_rate": 0.00017314569693073115,
      "loss": 1.545,
      "step": 7307
    },
    {
      "epoch": 0.2617150428850252,
      "grad_norm": 1.6997525691986084,
      "learning_rate": 0.00017313778721558975,
      "loss": 1.5749,
      "step": 7308
    },
    {
      "epoch": 0.26175085501459344,
      "grad_norm": 1.5419024229049683,
      "learning_rate": 0.00017312987651647374,
      "loss": 1.457,
      "step": 7309
    },
    {
      "epoch": 0.2617866671441617,
      "grad_norm": 1.712589144706726,
      "learning_rate": 0.0001731219648334896,
      "loss": 2.0299,
      "step": 7310
    },
    {
      "epoch": 0.26182247927373004,
      "grad_norm": 1.5588375329971313,
      "learning_rate": 0.00017311405216674373,
      "loss": 1.5899,
      "step": 7311
    },
    {
      "epoch": 0.2618582914032983,
      "grad_norm": 1.906366229057312,
      "learning_rate": 0.00017310613851634257,
      "loss": 1.698,
      "step": 7312
    },
    {
      "epoch": 0.2618941035328666,
      "grad_norm": 1.9092109203338623,
      "learning_rate": 0.00017309822388239266,
      "loss": 1.6362,
      "step": 7313
    },
    {
      "epoch": 0.26192991566243484,
      "grad_norm": 1.797682285308838,
      "learning_rate": 0.0001730903082650004,
      "loss": 1.3929,
      "step": 7314
    },
    {
      "epoch": 0.26196572779200317,
      "grad_norm": 2.1721010208129883,
      "learning_rate": 0.00017308239166427232,
      "loss": 1.9409,
      "step": 7315
    },
    {
      "epoch": 0.26200153992157144,
      "grad_norm": 2.2614290714263916,
      "learning_rate": 0.00017307447408031497,
      "loss": 1.5107,
      "step": 7316
    },
    {
      "epoch": 0.2620373520511397,
      "grad_norm": 1.8231735229492188,
      "learning_rate": 0.0001730665555132348,
      "loss": 1.4254,
      "step": 7317
    },
    {
      "epoch": 0.26207316418070803,
      "grad_norm": 1.7644098997116089,
      "learning_rate": 0.00017305863596313837,
      "loss": 1.4696,
      "step": 7318
    },
    {
      "epoch": 0.2621089763102763,
      "grad_norm": 1.7434821128845215,
      "learning_rate": 0.00017305071543013227,
      "loss": 1.522,
      "step": 7319
    },
    {
      "epoch": 0.26214478843984457,
      "grad_norm": 1.7708624601364136,
      "learning_rate": 0.000173042793914323,
      "loss": 1.4897,
      "step": 7320
    },
    {
      "epoch": 0.26218060056941284,
      "grad_norm": 1.6023200750350952,
      "learning_rate": 0.00017303487141581716,
      "loss": 1.58,
      "step": 7321
    },
    {
      "epoch": 0.26221641269898116,
      "grad_norm": 2.101994752883911,
      "learning_rate": 0.0001730269479347213,
      "loss": 1.7245,
      "step": 7322
    },
    {
      "epoch": 0.26225222482854943,
      "grad_norm": 1.9376888275146484,
      "learning_rate": 0.00017301902347114208,
      "loss": 1.5611,
      "step": 7323
    },
    {
      "epoch": 0.2622880369581177,
      "grad_norm": 2.2252988815307617,
      "learning_rate": 0.0001730110980251861,
      "loss": 1.7032,
      "step": 7324
    },
    {
      "epoch": 0.262323849087686,
      "grad_norm": 1.6127136945724487,
      "learning_rate": 0.00017300317159695995,
      "loss": 1.4226,
      "step": 7325
    },
    {
      "epoch": 0.2623596612172543,
      "grad_norm": 2.258042812347412,
      "learning_rate": 0.0001729952441865703,
      "loss": 1.6989,
      "step": 7326
    },
    {
      "epoch": 0.26239547334682256,
      "grad_norm": 1.4718269109725952,
      "learning_rate": 0.0001729873157941238,
      "loss": 1.5979,
      "step": 7327
    },
    {
      "epoch": 0.26243128547639083,
      "grad_norm": 1.5999681949615479,
      "learning_rate": 0.00017297938641972716,
      "loss": 1.4861,
      "step": 7328
    },
    {
      "epoch": 0.26246709760595915,
      "grad_norm": 1.6622743606567383,
      "learning_rate": 0.00017297145606348695,
      "loss": 1.4849,
      "step": 7329
    },
    {
      "epoch": 0.2625029097355274,
      "grad_norm": 2.1193180084228516,
      "learning_rate": 0.00017296352472550994,
      "loss": 1.8168,
      "step": 7330
    },
    {
      "epoch": 0.2625387218650957,
      "grad_norm": 1.6978873014450073,
      "learning_rate": 0.00017295559240590282,
      "loss": 1.9455,
      "step": 7331
    },
    {
      "epoch": 0.262574533994664,
      "grad_norm": 1.5294241905212402,
      "learning_rate": 0.00017294765910477234,
      "loss": 1.5882,
      "step": 7332
    },
    {
      "epoch": 0.2626103461242323,
      "grad_norm": 1.2453467845916748,
      "learning_rate": 0.00017293972482222515,
      "loss": 1.4419,
      "step": 7333
    },
    {
      "epoch": 0.26264615825380055,
      "grad_norm": 2.2070870399475098,
      "learning_rate": 0.00017293178955836807,
      "loss": 1.7747,
      "step": 7334
    },
    {
      "epoch": 0.2626819703833688,
      "grad_norm": 1.5115678310394287,
      "learning_rate": 0.00017292385331330786,
      "loss": 1.6529,
      "step": 7335
    },
    {
      "epoch": 0.26271778251293715,
      "grad_norm": 1.4143116474151611,
      "learning_rate": 0.00017291591608715123,
      "loss": 1.5566,
      "step": 7336
    },
    {
      "epoch": 0.2627535946425054,
      "grad_norm": 1.4451357126235962,
      "learning_rate": 0.00017290797788000503,
      "loss": 1.6816,
      "step": 7337
    },
    {
      "epoch": 0.2627894067720737,
      "grad_norm": 1.6054461002349854,
      "learning_rate": 0.00017290003869197603,
      "loss": 1.7646,
      "step": 7338
    },
    {
      "epoch": 0.262825218901642,
      "grad_norm": 1.4832358360290527,
      "learning_rate": 0.00017289209852317102,
      "loss": 1.2788,
      "step": 7339
    },
    {
      "epoch": 0.2628610310312103,
      "grad_norm": 1.7397518157958984,
      "learning_rate": 0.00017288415737369689,
      "loss": 1.5923,
      "step": 7340
    },
    {
      "epoch": 0.26289684316077855,
      "grad_norm": 2.040109157562256,
      "learning_rate": 0.0001728762152436604,
      "loss": 1.5015,
      "step": 7341
    },
    {
      "epoch": 0.2629326552903468,
      "grad_norm": 1.6289827823638916,
      "learning_rate": 0.00017286827213316844,
      "loss": 1.283,
      "step": 7342
    },
    {
      "epoch": 0.26296846741991514,
      "grad_norm": 1.4795184135437012,
      "learning_rate": 0.0001728603280423279,
      "loss": 1.6067,
      "step": 7343
    },
    {
      "epoch": 0.2630042795494834,
      "grad_norm": 1.4530529975891113,
      "learning_rate": 0.00017285238297124562,
      "loss": 1.3256,
      "step": 7344
    },
    {
      "epoch": 0.2630400916790517,
      "grad_norm": 2.1068873405456543,
      "learning_rate": 0.00017284443692002846,
      "loss": 1.8416,
      "step": 7345
    },
    {
      "epoch": 0.26307590380862,
      "grad_norm": 1.2771168947219849,
      "learning_rate": 0.00017283648988878343,
      "loss": 1.5879,
      "step": 7346
    },
    {
      "epoch": 0.2631117159381883,
      "grad_norm": 1.4236646890640259,
      "learning_rate": 0.00017282854187761735,
      "loss": 1.8003,
      "step": 7347
    },
    {
      "epoch": 0.26314752806775654,
      "grad_norm": 1.7822551727294922,
      "learning_rate": 0.00017282059288663715,
      "loss": 1.5744,
      "step": 7348
    },
    {
      "epoch": 0.2631833401973248,
      "grad_norm": 1.6378618478775024,
      "learning_rate": 0.00017281264291594983,
      "loss": 1.69,
      "step": 7349
    },
    {
      "epoch": 0.26321915232689314,
      "grad_norm": 1.6540489196777344,
      "learning_rate": 0.00017280469196566235,
      "loss": 1.3977,
      "step": 7350
    },
    {
      "epoch": 0.2632549644564614,
      "grad_norm": 1.8168818950653076,
      "learning_rate": 0.0001727967400358816,
      "loss": 1.5793,
      "step": 7351
    },
    {
      "epoch": 0.2632907765860297,
      "grad_norm": 2.310772657394409,
      "learning_rate": 0.00017278878712671464,
      "loss": 1.8026,
      "step": 7352
    },
    {
      "epoch": 0.263326588715598,
      "grad_norm": 1.4462745189666748,
      "learning_rate": 0.00017278083323826846,
      "loss": 1.6332,
      "step": 7353
    },
    {
      "epoch": 0.26336240084516627,
      "grad_norm": 1.3352108001708984,
      "learning_rate": 0.00017277287837065002,
      "loss": 1.2135,
      "step": 7354
    },
    {
      "epoch": 0.26339821297473454,
      "grad_norm": 2.1556811332702637,
      "learning_rate": 0.0001727649225239664,
      "loss": 1.4066,
      "step": 7355
    },
    {
      "epoch": 0.2634340251043028,
      "grad_norm": 1.7638002634048462,
      "learning_rate": 0.00017275696569832457,
      "loss": 1.623,
      "step": 7356
    },
    {
      "epoch": 0.26346983723387113,
      "grad_norm": 1.8131543397903442,
      "learning_rate": 0.00017274900789383165,
      "loss": 1.4694,
      "step": 7357
    },
    {
      "epoch": 0.2635056493634394,
      "grad_norm": 1.7827945947647095,
      "learning_rate": 0.0001727410491105946,
      "loss": 1.3291,
      "step": 7358
    },
    {
      "epoch": 0.26354146149300767,
      "grad_norm": 2.088897943496704,
      "learning_rate": 0.00017273308934872064,
      "loss": 1.4516,
      "step": 7359
    },
    {
      "epoch": 0.263577273622576,
      "grad_norm": 1.7818443775177002,
      "learning_rate": 0.00017272512860831674,
      "loss": 1.8525,
      "step": 7360
    },
    {
      "epoch": 0.26361308575214426,
      "grad_norm": 1.5051300525665283,
      "learning_rate": 0.00017271716688949007,
      "loss": 1.5285,
      "step": 7361
    },
    {
      "epoch": 0.26364889788171253,
      "grad_norm": 2.0163896083831787,
      "learning_rate": 0.0001727092041923477,
      "loss": 1.5589,
      "step": 7362
    },
    {
      "epoch": 0.2636847100112808,
      "grad_norm": 1.6455193758010864,
      "learning_rate": 0.00017270124051699682,
      "loss": 1.673,
      "step": 7363
    },
    {
      "epoch": 0.2637205221408491,
      "grad_norm": 1.561238408088684,
      "learning_rate": 0.00017269327586354446,
      "loss": 1.6594,
      "step": 7364
    },
    {
      "epoch": 0.2637563342704174,
      "grad_norm": 1.765101671218872,
      "learning_rate": 0.00017268531023209788,
      "loss": 1.5668,
      "step": 7365
    },
    {
      "epoch": 0.26379214639998566,
      "grad_norm": 2.3557276725769043,
      "learning_rate": 0.0001726773436227642,
      "loss": 1.5615,
      "step": 7366
    },
    {
      "epoch": 0.263827958529554,
      "grad_norm": 1.725612759590149,
      "learning_rate": 0.0001726693760356506,
      "loss": 1.5138,
      "step": 7367
    },
    {
      "epoch": 0.26386377065912225,
      "grad_norm": 1.6361420154571533,
      "learning_rate": 0.0001726614074708643,
      "loss": 1.3698,
      "step": 7368
    },
    {
      "epoch": 0.2638995827886905,
      "grad_norm": 1.775638461112976,
      "learning_rate": 0.00017265343792851248,
      "loss": 1.3577,
      "step": 7369
    },
    {
      "epoch": 0.2639353949182588,
      "grad_norm": 1.387298345565796,
      "learning_rate": 0.00017264546740870234,
      "loss": 1.5771,
      "step": 7370
    },
    {
      "epoch": 0.2639712070478271,
      "grad_norm": 1.341925859451294,
      "learning_rate": 0.0001726374959115412,
      "loss": 1.3746,
      "step": 7371
    },
    {
      "epoch": 0.2640070191773954,
      "grad_norm": 1.5934443473815918,
      "learning_rate": 0.0001726295234371362,
      "loss": 1.5295,
      "step": 7372
    },
    {
      "epoch": 0.26404283130696365,
      "grad_norm": 1.5405175685882568,
      "learning_rate": 0.00017262154998559466,
      "loss": 1.5476,
      "step": 7373
    },
    {
      "epoch": 0.264078643436532,
      "grad_norm": 1.6285407543182373,
      "learning_rate": 0.00017261357555702387,
      "loss": 1.565,
      "step": 7374
    },
    {
      "epoch": 0.26411445556610025,
      "grad_norm": 2.251279592514038,
      "learning_rate": 0.00017260560015153106,
      "loss": 1.7117,
      "step": 7375
    },
    {
      "epoch": 0.2641502676956685,
      "grad_norm": 2.265822410583496,
      "learning_rate": 0.00017259762376922356,
      "loss": 2.0113,
      "step": 7376
    },
    {
      "epoch": 0.2641860798252368,
      "grad_norm": 2.2183547019958496,
      "learning_rate": 0.00017258964641020868,
      "loss": 1.5966,
      "step": 7377
    },
    {
      "epoch": 0.2642218919548051,
      "grad_norm": 2.0988476276397705,
      "learning_rate": 0.0001725816680745937,
      "loss": 1.3639,
      "step": 7378
    },
    {
      "epoch": 0.2642577040843734,
      "grad_norm": 2.5295467376708984,
      "learning_rate": 0.00017257368876248604,
      "loss": 1.4807,
      "step": 7379
    },
    {
      "epoch": 0.26429351621394165,
      "grad_norm": 1.4437577724456787,
      "learning_rate": 0.000172565708473993,
      "loss": 1.4974,
      "step": 7380
    },
    {
      "epoch": 0.26432932834351,
      "grad_norm": 1.2854939699172974,
      "learning_rate": 0.00017255772720922195,
      "loss": 1.6244,
      "step": 7381
    },
    {
      "epoch": 0.26436514047307824,
      "grad_norm": 1.3018090724945068,
      "learning_rate": 0.0001725497449682803,
      "loss": 1.3882,
      "step": 7382
    },
    {
      "epoch": 0.2644009526026465,
      "grad_norm": 1.5124176740646362,
      "learning_rate": 0.00017254176175127538,
      "loss": 1.4241,
      "step": 7383
    },
    {
      "epoch": 0.2644367647322148,
      "grad_norm": 2.2638661861419678,
      "learning_rate": 0.00017253377755831466,
      "loss": 1.7341,
      "step": 7384
    },
    {
      "epoch": 0.2644725768617831,
      "grad_norm": 1.42527174949646,
      "learning_rate": 0.00017252579238950552,
      "loss": 1.5593,
      "step": 7385
    },
    {
      "epoch": 0.2645083889913514,
      "grad_norm": 1.981040596961975,
      "learning_rate": 0.00017251780624495536,
      "loss": 1.482,
      "step": 7386
    },
    {
      "epoch": 0.26454420112091964,
      "grad_norm": 2.1076390743255615,
      "learning_rate": 0.0001725098191247717,
      "loss": 1.5527,
      "step": 7387
    },
    {
      "epoch": 0.26458001325048797,
      "grad_norm": 1.287087321281433,
      "learning_rate": 0.00017250183102906195,
      "loss": 1.7802,
      "step": 7388
    },
    {
      "epoch": 0.26461582538005624,
      "grad_norm": 1.3365637063980103,
      "learning_rate": 0.00017249384195793357,
      "loss": 1.4736,
      "step": 7389
    },
    {
      "epoch": 0.2646516375096245,
      "grad_norm": 1.6480307579040527,
      "learning_rate": 0.000172485851911494,
      "loss": 1.8273,
      "step": 7390
    },
    {
      "epoch": 0.2646874496391928,
      "grad_norm": 2.1490418910980225,
      "learning_rate": 0.00017247786088985087,
      "loss": 1.751,
      "step": 7391
    },
    {
      "epoch": 0.2647232617687611,
      "grad_norm": 1.840654730796814,
      "learning_rate": 0.0001724698688931116,
      "loss": 1.772,
      "step": 7392
    },
    {
      "epoch": 0.26475907389832937,
      "grad_norm": 3.2278695106506348,
      "learning_rate": 0.0001724618759213837,
      "loss": 1.7355,
      "step": 7393
    },
    {
      "epoch": 0.26479488602789764,
      "grad_norm": 2.007066249847412,
      "learning_rate": 0.00017245388197477477,
      "loss": 1.4662,
      "step": 7394
    },
    {
      "epoch": 0.26483069815746596,
      "grad_norm": 1.8525476455688477,
      "learning_rate": 0.0001724458870533923,
      "loss": 1.6683,
      "step": 7395
    },
    {
      "epoch": 0.26486651028703423,
      "grad_norm": 1.865721583366394,
      "learning_rate": 0.00017243789115734383,
      "loss": 1.7475,
      "step": 7396
    },
    {
      "epoch": 0.2649023224166025,
      "grad_norm": 1.5090208053588867,
      "learning_rate": 0.00017242989428673701,
      "loss": 1.5835,
      "step": 7397
    },
    {
      "epoch": 0.26493813454617077,
      "grad_norm": 2.2835018634796143,
      "learning_rate": 0.0001724218964416794,
      "loss": 1.4297,
      "step": 7398
    },
    {
      "epoch": 0.2649739466757391,
      "grad_norm": 1.7916940450668335,
      "learning_rate": 0.00017241389762227857,
      "loss": 1.666,
      "step": 7399
    },
    {
      "epoch": 0.26500975880530736,
      "grad_norm": 1.830520510673523,
      "learning_rate": 0.00017240589782864215,
      "loss": 1.5415,
      "step": 7400
    },
    {
      "epoch": 0.26504557093487563,
      "grad_norm": 1.5442086458206177,
      "learning_rate": 0.00017239789706087778,
      "loss": 1.6078,
      "step": 7401
    },
    {
      "epoch": 0.26508138306444395,
      "grad_norm": 1.3873393535614014,
      "learning_rate": 0.0001723898953190931,
      "loss": 1.6903,
      "step": 7402
    },
    {
      "epoch": 0.2651171951940122,
      "grad_norm": 1.6361714601516724,
      "learning_rate": 0.00017238189260339573,
      "loss": 1.6274,
      "step": 7403
    },
    {
      "epoch": 0.2651530073235805,
      "grad_norm": 1.1806063652038574,
      "learning_rate": 0.00017237388891389336,
      "loss": 1.5139,
      "step": 7404
    },
    {
      "epoch": 0.26518881945314876,
      "grad_norm": 1.8152517080307007,
      "learning_rate": 0.0001723658842506937,
      "loss": 1.3687,
      "step": 7405
    },
    {
      "epoch": 0.2652246315827171,
      "grad_norm": 1.5166256427764893,
      "learning_rate": 0.0001723578786139044,
      "loss": 1.3072,
      "step": 7406
    },
    {
      "epoch": 0.26526044371228535,
      "grad_norm": 1.715145468711853,
      "learning_rate": 0.00017234987200363317,
      "loss": 1.4742,
      "step": 7407
    },
    {
      "epoch": 0.2652962558418536,
      "grad_norm": 2.2993531227111816,
      "learning_rate": 0.00017234186441998777,
      "loss": 1.5068,
      "step": 7408
    },
    {
      "epoch": 0.26533206797142195,
      "grad_norm": 1.8281476497650146,
      "learning_rate": 0.00017233385586307588,
      "loss": 1.6021,
      "step": 7409
    },
    {
      "epoch": 0.2653678801009902,
      "grad_norm": 1.5883283615112305,
      "learning_rate": 0.00017232584633300522,
      "loss": 1.6,
      "step": 7410
    },
    {
      "epoch": 0.2654036922305585,
      "grad_norm": 1.610416293144226,
      "learning_rate": 0.00017231783582988367,
      "loss": 1.6896,
      "step": 7411
    },
    {
      "epoch": 0.26543950436012675,
      "grad_norm": 1.5972161293029785,
      "learning_rate": 0.00017230982435381887,
      "loss": 1.7697,
      "step": 7412
    },
    {
      "epoch": 0.2654753164896951,
      "grad_norm": 1.538904070854187,
      "learning_rate": 0.00017230181190491862,
      "loss": 1.5154,
      "step": 7413
    },
    {
      "epoch": 0.26551112861926335,
      "grad_norm": 1.459111213684082,
      "learning_rate": 0.0001722937984832908,
      "loss": 1.6878,
      "step": 7414
    },
    {
      "epoch": 0.2655469407488316,
      "grad_norm": 1.890040636062622,
      "learning_rate": 0.0001722857840890432,
      "loss": 1.3518,
      "step": 7415
    },
    {
      "epoch": 0.26558275287839994,
      "grad_norm": 1.6970199346542358,
      "learning_rate": 0.00017227776872228359,
      "loss": 1.5614,
      "step": 7416
    },
    {
      "epoch": 0.2656185650079682,
      "grad_norm": 1.4791874885559082,
      "learning_rate": 0.00017226975238311982,
      "loss": 1.4532,
      "step": 7417
    },
    {
      "epoch": 0.2656543771375365,
      "grad_norm": 1.8702563047409058,
      "learning_rate": 0.00017226173507165976,
      "loss": 1.7689,
      "step": 7418
    },
    {
      "epoch": 0.26569018926710475,
      "grad_norm": 1.5942208766937256,
      "learning_rate": 0.0001722537167880113,
      "loss": 1.7556,
      "step": 7419
    },
    {
      "epoch": 0.2657260013966731,
      "grad_norm": 1.6000324487686157,
      "learning_rate": 0.00017224569753228225,
      "loss": 1.5437,
      "step": 7420
    },
    {
      "epoch": 0.26576181352624134,
      "grad_norm": 2.1063296794891357,
      "learning_rate": 0.00017223767730458053,
      "loss": 1.7174,
      "step": 7421
    },
    {
      "epoch": 0.2657976256558096,
      "grad_norm": 1.3740330934524536,
      "learning_rate": 0.00017222965610501405,
      "loss": 1.4734,
      "step": 7422
    },
    {
      "epoch": 0.26583343778537794,
      "grad_norm": 1.5399854183197021,
      "learning_rate": 0.00017222163393369071,
      "loss": 1.5247,
      "step": 7423
    },
    {
      "epoch": 0.2658692499149462,
      "grad_norm": 1.5186257362365723,
      "learning_rate": 0.00017221361079071846,
      "loss": 1.901,
      "step": 7424
    },
    {
      "epoch": 0.2659050620445145,
      "grad_norm": 1.7701181173324585,
      "learning_rate": 0.00017220558667620518,
      "loss": 1.4393,
      "step": 7425
    },
    {
      "epoch": 0.26594087417408274,
      "grad_norm": 1.7692437171936035,
      "learning_rate": 0.0001721975615902589,
      "loss": 1.5675,
      "step": 7426
    },
    {
      "epoch": 0.26597668630365107,
      "grad_norm": 2.3157219886779785,
      "learning_rate": 0.00017218953553298759,
      "loss": 1.5851,
      "step": 7427
    },
    {
      "epoch": 0.26601249843321934,
      "grad_norm": 1.5588358640670776,
      "learning_rate": 0.00017218150850449915,
      "loss": 1.2724,
      "step": 7428
    },
    {
      "epoch": 0.2660483105627876,
      "grad_norm": 1.2565510272979736,
      "learning_rate": 0.00017217348050490162,
      "loss": 1.6188,
      "step": 7429
    },
    {
      "epoch": 0.2660841226923559,
      "grad_norm": 1.3198089599609375,
      "learning_rate": 0.00017216545153430303,
      "loss": 1.4822,
      "step": 7430
    },
    {
      "epoch": 0.2661199348219242,
      "grad_norm": 1.2001327276229858,
      "learning_rate": 0.00017215742159281137,
      "loss": 1.4967,
      "step": 7431
    },
    {
      "epoch": 0.26615574695149247,
      "grad_norm": 1.8182188272476196,
      "learning_rate": 0.00017214939068053468,
      "loss": 1.6206,
      "step": 7432
    },
    {
      "epoch": 0.26619155908106074,
      "grad_norm": 1.916847586631775,
      "learning_rate": 0.000172141358797581,
      "loss": 1.6809,
      "step": 7433
    },
    {
      "epoch": 0.26622737121062906,
      "grad_norm": 1.1985223293304443,
      "learning_rate": 0.0001721333259440584,
      "loss": 1.4388,
      "step": 7434
    },
    {
      "epoch": 0.26626318334019733,
      "grad_norm": 1.6115180253982544,
      "learning_rate": 0.00017212529212007492,
      "loss": 1.6658,
      "step": 7435
    },
    {
      "epoch": 0.2662989954697656,
      "grad_norm": 1.7065064907073975,
      "learning_rate": 0.0001721172573257387,
      "loss": 1.5012,
      "step": 7436
    },
    {
      "epoch": 0.26633480759933387,
      "grad_norm": 1.26376211643219,
      "learning_rate": 0.0001721092215611578,
      "loss": 1.0781,
      "step": 7437
    },
    {
      "epoch": 0.2663706197289022,
      "grad_norm": 1.3036880493164062,
      "learning_rate": 0.00017210118482644036,
      "loss": 1.5433,
      "step": 7438
    },
    {
      "epoch": 0.26640643185847046,
      "grad_norm": 1.7557674646377563,
      "learning_rate": 0.00017209314712169445,
      "loss": 1.449,
      "step": 7439
    },
    {
      "epoch": 0.26644224398803873,
      "grad_norm": 1.509422779083252,
      "learning_rate": 0.00017208510844702823,
      "loss": 1.6469,
      "step": 7440
    },
    {
      "epoch": 0.26647805611760705,
      "grad_norm": 1.7672717571258545,
      "learning_rate": 0.00017207706880254987,
      "loss": 1.6361,
      "step": 7441
    },
    {
      "epoch": 0.2665138682471753,
      "grad_norm": 1.363512396812439,
      "learning_rate": 0.00017206902818836756,
      "loss": 1.2467,
      "step": 7442
    },
    {
      "epoch": 0.2665496803767436,
      "grad_norm": 2.0854268074035645,
      "learning_rate": 0.00017206098660458937,
      "loss": 1.596,
      "step": 7443
    },
    {
      "epoch": 0.26658549250631186,
      "grad_norm": 1.7014752626419067,
      "learning_rate": 0.00017205294405132362,
      "loss": 1.4051,
      "step": 7444
    },
    {
      "epoch": 0.2666213046358802,
      "grad_norm": 1.4514315128326416,
      "learning_rate": 0.00017204490052867842,
      "loss": 1.3514,
      "step": 7445
    },
    {
      "epoch": 0.26665711676544845,
      "grad_norm": 1.4428623914718628,
      "learning_rate": 0.00017203685603676202,
      "loss": 1.6256,
      "step": 7446
    },
    {
      "epoch": 0.2666929288950167,
      "grad_norm": 2.695467233657837,
      "learning_rate": 0.0001720288105756826,
      "loss": 1.8272,
      "step": 7447
    },
    {
      "epoch": 0.26672874102458505,
      "grad_norm": 1.776180386543274,
      "learning_rate": 0.0001720207641455485,
      "loss": 1.6105,
      "step": 7448
    },
    {
      "epoch": 0.2667645531541533,
      "grad_norm": 1.5230605602264404,
      "learning_rate": 0.0001720127167464679,
      "loss": 1.6002,
      "step": 7449
    },
    {
      "epoch": 0.2668003652837216,
      "grad_norm": 1.4507273435592651,
      "learning_rate": 0.00017200466837854908,
      "loss": 1.6159,
      "step": 7450
    },
    {
      "epoch": 0.26683617741328985,
      "grad_norm": 2.1067185401916504,
      "learning_rate": 0.00017199661904190037,
      "loss": 1.3856,
      "step": 7451
    },
    {
      "epoch": 0.2668719895428582,
      "grad_norm": 1.6292303800582886,
      "learning_rate": 0.00017198856873662996,
      "loss": 1.5911,
      "step": 7452
    },
    {
      "epoch": 0.26690780167242645,
      "grad_norm": 1.3270217180252075,
      "learning_rate": 0.00017198051746284624,
      "loss": 1.4468,
      "step": 7453
    },
    {
      "epoch": 0.2669436138019947,
      "grad_norm": 1.3355484008789062,
      "learning_rate": 0.00017197246522065752,
      "loss": 1.4923,
      "step": 7454
    },
    {
      "epoch": 0.26697942593156304,
      "grad_norm": 1.5357935428619385,
      "learning_rate": 0.00017196441201017208,
      "loss": 1.2817,
      "step": 7455
    },
    {
      "epoch": 0.2670152380611313,
      "grad_norm": 1.796885371208191,
      "learning_rate": 0.00017195635783149834,
      "loss": 1.7462,
      "step": 7456
    },
    {
      "epoch": 0.2670510501906996,
      "grad_norm": 2.4231655597686768,
      "learning_rate": 0.0001719483026847446,
      "loss": 1.9039,
      "step": 7457
    },
    {
      "epoch": 0.26708686232026785,
      "grad_norm": 1.710451602935791,
      "learning_rate": 0.00017194024657001927,
      "loss": 1.5666,
      "step": 7458
    },
    {
      "epoch": 0.2671226744498362,
      "grad_norm": 1.5234856605529785,
      "learning_rate": 0.0001719321894874307,
      "loss": 1.6446,
      "step": 7459
    },
    {
      "epoch": 0.26715848657940444,
      "grad_norm": 1.9876848459243774,
      "learning_rate": 0.00017192413143708735,
      "loss": 1.6329,
      "step": 7460
    },
    {
      "epoch": 0.2671942987089727,
      "grad_norm": 1.7578610181808472,
      "learning_rate": 0.00017191607241909753,
      "loss": 1.5056,
      "step": 7461
    },
    {
      "epoch": 0.26723011083854104,
      "grad_norm": 1.4033610820770264,
      "learning_rate": 0.00017190801243356977,
      "loss": 1.4246,
      "step": 7462
    },
    {
      "epoch": 0.2672659229681093,
      "grad_norm": 1.780724048614502,
      "learning_rate": 0.0001718999514806124,
      "loss": 1.3211,
      "step": 7463
    },
    {
      "epoch": 0.2673017350976776,
      "grad_norm": 2.1150660514831543,
      "learning_rate": 0.000171891889560334,
      "loss": 1.5886,
      "step": 7464
    },
    {
      "epoch": 0.26733754722724584,
      "grad_norm": 1.8331183195114136,
      "learning_rate": 0.0001718838266728429,
      "loss": 1.4792,
      "step": 7465
    },
    {
      "epoch": 0.26737335935681417,
      "grad_norm": 1.8556559085845947,
      "learning_rate": 0.00017187576281824766,
      "loss": 1.5132,
      "step": 7466
    },
    {
      "epoch": 0.26740917148638244,
      "grad_norm": 1.7563835382461548,
      "learning_rate": 0.00017186769799665673,
      "loss": 1.5768,
      "step": 7467
    },
    {
      "epoch": 0.2674449836159507,
      "grad_norm": 2.372431516647339,
      "learning_rate": 0.00017185963220817864,
      "loss": 1.5395,
      "step": 7468
    },
    {
      "epoch": 0.26748079574551903,
      "grad_norm": 1.086047649383545,
      "learning_rate": 0.0001718515654529219,
      "loss": 1.3656,
      "step": 7469
    },
    {
      "epoch": 0.2675166078750873,
      "grad_norm": 1.8703947067260742,
      "learning_rate": 0.000171843497730995,
      "loss": 1.6605,
      "step": 7470
    },
    {
      "epoch": 0.26755242000465557,
      "grad_norm": 1.627095103263855,
      "learning_rate": 0.00017183542904250656,
      "loss": 1.6894,
      "step": 7471
    },
    {
      "epoch": 0.26758823213422384,
      "grad_norm": 1.8694849014282227,
      "learning_rate": 0.00017182735938756506,
      "loss": 1.4633,
      "step": 7472
    },
    {
      "epoch": 0.26762404426379216,
      "grad_norm": 1.7789463996887207,
      "learning_rate": 0.00017181928876627907,
      "loss": 1.5519,
      "step": 7473
    },
    {
      "epoch": 0.26765985639336043,
      "grad_norm": 2.8911938667297363,
      "learning_rate": 0.0001718112171787572,
      "loss": 1.5961,
      "step": 7474
    },
    {
      "epoch": 0.2676956685229287,
      "grad_norm": 1.5189895629882812,
      "learning_rate": 0.000171803144625108,
      "loss": 1.401,
      "step": 7475
    },
    {
      "epoch": 0.267731480652497,
      "grad_norm": 1.9463037252426147,
      "learning_rate": 0.00017179507110544014,
      "loss": 1.8663,
      "step": 7476
    },
    {
      "epoch": 0.2677672927820653,
      "grad_norm": 1.4575629234313965,
      "learning_rate": 0.0001717869966198622,
      "loss": 1.5934,
      "step": 7477
    },
    {
      "epoch": 0.26780310491163356,
      "grad_norm": 1.911528468132019,
      "learning_rate": 0.00017177892116848284,
      "loss": 1.572,
      "step": 7478
    },
    {
      "epoch": 0.26783891704120183,
      "grad_norm": 1.543802261352539,
      "learning_rate": 0.00017177084475141069,
      "loss": 1.7237,
      "step": 7479
    },
    {
      "epoch": 0.26787472917077015,
      "grad_norm": 1.4773614406585693,
      "learning_rate": 0.0001717627673687544,
      "loss": 1.6445,
      "step": 7480
    },
    {
      "epoch": 0.2679105413003384,
      "grad_norm": 1.8544567823410034,
      "learning_rate": 0.0001717546890206226,
      "loss": 1.2569,
      "step": 7481
    },
    {
      "epoch": 0.2679463534299067,
      "grad_norm": 1.9463386535644531,
      "learning_rate": 0.00017174660970712403,
      "loss": 1.5379,
      "step": 7482
    },
    {
      "epoch": 0.267982165559475,
      "grad_norm": 2.0309228897094727,
      "learning_rate": 0.00017173852942836739,
      "loss": 1.6161,
      "step": 7483
    },
    {
      "epoch": 0.2680179776890433,
      "grad_norm": 1.7174644470214844,
      "learning_rate": 0.00017173044818446137,
      "loss": 1.6121,
      "step": 7484
    },
    {
      "epoch": 0.26805378981861155,
      "grad_norm": 1.8857377767562866,
      "learning_rate": 0.00017172236597551467,
      "loss": 1.6238,
      "step": 7485
    },
    {
      "epoch": 0.2680896019481798,
      "grad_norm": 1.3329565525054932,
      "learning_rate": 0.0001717142828016361,
      "loss": 1.7221,
      "step": 7486
    },
    {
      "epoch": 0.26812541407774815,
      "grad_norm": 1.682708978652954,
      "learning_rate": 0.00017170619866293434,
      "loss": 1.1758,
      "step": 7487
    },
    {
      "epoch": 0.2681612262073164,
      "grad_norm": 1.6127010583877563,
      "learning_rate": 0.00017169811355951815,
      "loss": 1.5131,
      "step": 7488
    },
    {
      "epoch": 0.2681970383368847,
      "grad_norm": 1.7014356851577759,
      "learning_rate": 0.0001716900274914963,
      "loss": 1.3939,
      "step": 7489
    },
    {
      "epoch": 0.268232850466453,
      "grad_norm": 1.85292387008667,
      "learning_rate": 0.00017168194045897767,
      "loss": 1.434,
      "step": 7490
    },
    {
      "epoch": 0.2682686625960213,
      "grad_norm": 1.6793440580368042,
      "learning_rate": 0.000171673852462071,
      "loss": 1.3301,
      "step": 7491
    },
    {
      "epoch": 0.26830447472558955,
      "grad_norm": 1.561973214149475,
      "learning_rate": 0.00017166576350088506,
      "loss": 1.2769,
      "step": 7492
    },
    {
      "epoch": 0.2683402868551578,
      "grad_norm": 1.6289607286453247,
      "learning_rate": 0.0001716576735755287,
      "loss": 1.4536,
      "step": 7493
    },
    {
      "epoch": 0.26837609898472614,
      "grad_norm": 1.7525798082351685,
      "learning_rate": 0.00017164958268611077,
      "loss": 1.3704,
      "step": 7494
    },
    {
      "epoch": 0.2684119111142944,
      "grad_norm": 1.8854219913482666,
      "learning_rate": 0.00017164149083274017,
      "loss": 1.6748,
      "step": 7495
    },
    {
      "epoch": 0.2684477232438627,
      "grad_norm": 1.3732788562774658,
      "learning_rate": 0.0001716333980155257,
      "loss": 1.5534,
      "step": 7496
    },
    {
      "epoch": 0.268483535373431,
      "grad_norm": 1.6843669414520264,
      "learning_rate": 0.00017162530423457626,
      "loss": 1.7485,
      "step": 7497
    },
    {
      "epoch": 0.2685193475029993,
      "grad_norm": 1.746842861175537,
      "learning_rate": 0.00017161720949000075,
      "loss": 1.7009,
      "step": 7498
    },
    {
      "epoch": 0.26855515963256754,
      "grad_norm": 2.6704607009887695,
      "learning_rate": 0.00017160911378190808,
      "loss": 1.4903,
      "step": 7499
    },
    {
      "epoch": 0.2685909717621358,
      "grad_norm": 1.9854313135147095,
      "learning_rate": 0.00017160101711040713,
      "loss": 1.7229,
      "step": 7500
    },
    {
      "epoch": 0.26862678389170414,
      "grad_norm": 2.014387845993042,
      "learning_rate": 0.00017159291947560682,
      "loss": 1.5854,
      "step": 7501
    },
    {
      "epoch": 0.2686625960212724,
      "grad_norm": 1.4004807472229004,
      "learning_rate": 0.00017158482087761617,
      "loss": 1.5235,
      "step": 7502
    },
    {
      "epoch": 0.2686984081508407,
      "grad_norm": 1.545235276222229,
      "learning_rate": 0.0001715767213165441,
      "loss": 1.6951,
      "step": 7503
    },
    {
      "epoch": 0.268734220280409,
      "grad_norm": 1.4803507328033447,
      "learning_rate": 0.00017156862079249953,
      "loss": 1.7183,
      "step": 7504
    },
    {
      "epoch": 0.26877003240997727,
      "grad_norm": 1.6288284063339233,
      "learning_rate": 0.00017156051930559155,
      "loss": 1.6609,
      "step": 7505
    },
    {
      "epoch": 0.26880584453954554,
      "grad_norm": 1.784579873085022,
      "learning_rate": 0.00017155241685592903,
      "loss": 1.6495,
      "step": 7506
    },
    {
      "epoch": 0.2688416566691138,
      "grad_norm": 1.2053595781326294,
      "learning_rate": 0.00017154431344362106,
      "loss": 1.4678,
      "step": 7507
    },
    {
      "epoch": 0.26887746879868213,
      "grad_norm": 1.9320080280303955,
      "learning_rate": 0.00017153620906877666,
      "loss": 1.6605,
      "step": 7508
    },
    {
      "epoch": 0.2689132809282504,
      "grad_norm": 1.4452803134918213,
      "learning_rate": 0.00017152810373150478,
      "loss": 1.6794,
      "step": 7509
    },
    {
      "epoch": 0.26894909305781867,
      "grad_norm": 1.9542020559310913,
      "learning_rate": 0.00017151999743191456,
      "loss": 1.4616,
      "step": 7510
    },
    {
      "epoch": 0.268984905187387,
      "grad_norm": 2.002528667449951,
      "learning_rate": 0.00017151189017011503,
      "loss": 1.7598,
      "step": 7511
    },
    {
      "epoch": 0.26902071731695526,
      "grad_norm": 1.4809521436691284,
      "learning_rate": 0.00017150378194621529,
      "loss": 1.5741,
      "step": 7512
    },
    {
      "epoch": 0.26905652944652353,
      "grad_norm": 2.5902462005615234,
      "learning_rate": 0.0001714956727603244,
      "loss": 1.5187,
      "step": 7513
    },
    {
      "epoch": 0.2690923415760918,
      "grad_norm": 1.7048165798187256,
      "learning_rate": 0.0001714875626125514,
      "loss": 1.5129,
      "step": 7514
    },
    {
      "epoch": 0.2691281537056601,
      "grad_norm": 1.714754343032837,
      "learning_rate": 0.0001714794515030055,
      "loss": 1.9192,
      "step": 7515
    },
    {
      "epoch": 0.2691639658352284,
      "grad_norm": 1.5746515989303589,
      "learning_rate": 0.00017147133943179577,
      "loss": 1.5068,
      "step": 7516
    },
    {
      "epoch": 0.26919977796479666,
      "grad_norm": 1.5344665050506592,
      "learning_rate": 0.00017146322639903137,
      "loss": 1.6429,
      "step": 7517
    },
    {
      "epoch": 0.269235590094365,
      "grad_norm": 1.722382664680481,
      "learning_rate": 0.00017145511240482142,
      "loss": 1.3101,
      "step": 7518
    },
    {
      "epoch": 0.26927140222393325,
      "grad_norm": 1.7240898609161377,
      "learning_rate": 0.00017144699744927507,
      "loss": 1.657,
      "step": 7519
    },
    {
      "epoch": 0.2693072143535015,
      "grad_norm": 2.000774383544922,
      "learning_rate": 0.0001714388815325016,
      "loss": 1.4911,
      "step": 7520
    },
    {
      "epoch": 0.2693430264830698,
      "grad_norm": 2.6266844272613525,
      "learning_rate": 0.0001714307646546101,
      "loss": 1.7797,
      "step": 7521
    },
    {
      "epoch": 0.2693788386126381,
      "grad_norm": 1.6065651178359985,
      "learning_rate": 0.00017142264681570978,
      "loss": 1.6816,
      "step": 7522
    },
    {
      "epoch": 0.2694146507422064,
      "grad_norm": 1.504185438156128,
      "learning_rate": 0.00017141452801590988,
      "loss": 1.2781,
      "step": 7523
    },
    {
      "epoch": 0.26945046287177465,
      "grad_norm": 1.5775504112243652,
      "learning_rate": 0.00017140640825531967,
      "loss": 1.6476,
      "step": 7524
    },
    {
      "epoch": 0.269486275001343,
      "grad_norm": 1.482826828956604,
      "learning_rate": 0.0001713982875340483,
      "loss": 1.7394,
      "step": 7525
    },
    {
      "epoch": 0.26952208713091125,
      "grad_norm": 2.138429641723633,
      "learning_rate": 0.00017139016585220512,
      "loss": 1.6288,
      "step": 7526
    },
    {
      "epoch": 0.2695578992604795,
      "grad_norm": 1.5291169881820679,
      "learning_rate": 0.0001713820432098993,
      "loss": 1.3784,
      "step": 7527
    },
    {
      "epoch": 0.2695937113900478,
      "grad_norm": 1.7900766134262085,
      "learning_rate": 0.00017137391960724013,
      "loss": 1.4586,
      "step": 7528
    },
    {
      "epoch": 0.2696295235196161,
      "grad_norm": 1.9882597923278809,
      "learning_rate": 0.000171365795044337,
      "loss": 1.5581,
      "step": 7529
    },
    {
      "epoch": 0.2696653356491844,
      "grad_norm": 1.6502845287322998,
      "learning_rate": 0.00017135766952129913,
      "loss": 1.4082,
      "step": 7530
    },
    {
      "epoch": 0.26970114777875265,
      "grad_norm": 1.6091289520263672,
      "learning_rate": 0.00017134954303823588,
      "loss": 1.5386,
      "step": 7531
    },
    {
      "epoch": 0.269736959908321,
      "grad_norm": 1.4166127443313599,
      "learning_rate": 0.00017134141559525654,
      "loss": 1.4206,
      "step": 7532
    },
    {
      "epoch": 0.26977277203788924,
      "grad_norm": 1.5068105459213257,
      "learning_rate": 0.00017133328719247048,
      "loss": 1.3298,
      "step": 7533
    },
    {
      "epoch": 0.2698085841674575,
      "grad_norm": 1.52097749710083,
      "learning_rate": 0.00017132515782998704,
      "loss": 1.3535,
      "step": 7534
    },
    {
      "epoch": 0.2698443962970258,
      "grad_norm": 1.7753351926803589,
      "learning_rate": 0.00017131702750791564,
      "loss": 1.2874,
      "step": 7535
    },
    {
      "epoch": 0.2698802084265941,
      "grad_norm": 1.4764755964279175,
      "learning_rate": 0.0001713088962263656,
      "loss": 1.6138,
      "step": 7536
    },
    {
      "epoch": 0.2699160205561624,
      "grad_norm": 1.4922411441802979,
      "learning_rate": 0.00017130076398544635,
      "loss": 1.7891,
      "step": 7537
    },
    {
      "epoch": 0.26995183268573064,
      "grad_norm": 1.7627573013305664,
      "learning_rate": 0.0001712926307852673,
      "loss": 1.4984,
      "step": 7538
    },
    {
      "epoch": 0.26998764481529897,
      "grad_norm": 1.7921463251113892,
      "learning_rate": 0.00017128449662593786,
      "loss": 1.5461,
      "step": 7539
    },
    {
      "epoch": 0.27002345694486724,
      "grad_norm": 1.4542369842529297,
      "learning_rate": 0.00017127636150756747,
      "loss": 1.7536,
      "step": 7540
    },
    {
      "epoch": 0.2700592690744355,
      "grad_norm": 1.5525248050689697,
      "learning_rate": 0.00017126822543026555,
      "loss": 1.6489,
      "step": 7541
    },
    {
      "epoch": 0.2700950812040038,
      "grad_norm": 2.08571720123291,
      "learning_rate": 0.00017126008839414163,
      "loss": 1.8435,
      "step": 7542
    },
    {
      "epoch": 0.2701308933335721,
      "grad_norm": 1.3870586156845093,
      "learning_rate": 0.00017125195039930508,
      "loss": 1.3148,
      "step": 7543
    },
    {
      "epoch": 0.27016670546314037,
      "grad_norm": 3.187460422515869,
      "learning_rate": 0.0001712438114458655,
      "loss": 1.6014,
      "step": 7544
    },
    {
      "epoch": 0.27020251759270864,
      "grad_norm": 1.4459199905395508,
      "learning_rate": 0.00017123567153393233,
      "loss": 1.4049,
      "step": 7545
    },
    {
      "epoch": 0.27023832972227696,
      "grad_norm": 2.3334615230560303,
      "learning_rate": 0.00017122753066361508,
      "loss": 1.542,
      "step": 7546
    },
    {
      "epoch": 0.27027414185184523,
      "grad_norm": 1.4789403676986694,
      "learning_rate": 0.00017121938883502328,
      "loss": 1.7571,
      "step": 7547
    },
    {
      "epoch": 0.2703099539814135,
      "grad_norm": 1.5586892366409302,
      "learning_rate": 0.00017121124604826645,
      "loss": 1.6486,
      "step": 7548
    },
    {
      "epoch": 0.27034576611098177,
      "grad_norm": 1.6130740642547607,
      "learning_rate": 0.00017120310230345418,
      "loss": 1.3414,
      "step": 7549
    },
    {
      "epoch": 0.2703815782405501,
      "grad_norm": 1.3961517810821533,
      "learning_rate": 0.000171194957600696,
      "loss": 1.4985,
      "step": 7550
    },
    {
      "epoch": 0.27041739037011836,
      "grad_norm": 1.2499228715896606,
      "learning_rate": 0.00017118681194010153,
      "loss": 1.2782,
      "step": 7551
    },
    {
      "epoch": 0.27045320249968663,
      "grad_norm": 2.04305362701416,
      "learning_rate": 0.0001711786653217803,
      "loss": 1.409,
      "step": 7552
    },
    {
      "epoch": 0.27048901462925495,
      "grad_norm": 1.6010410785675049,
      "learning_rate": 0.00017117051774584194,
      "loss": 1.324,
      "step": 7553
    },
    {
      "epoch": 0.2705248267588232,
      "grad_norm": 1.6431407928466797,
      "learning_rate": 0.00017116236921239607,
      "loss": 1.4098,
      "step": 7554
    },
    {
      "epoch": 0.2705606388883915,
      "grad_norm": 1.8824855089187622,
      "learning_rate": 0.00017115421972155234,
      "loss": 1.498,
      "step": 7555
    },
    {
      "epoch": 0.27059645101795976,
      "grad_norm": 1.4907480478286743,
      "learning_rate": 0.00017114606927342036,
      "loss": 1.4708,
      "step": 7556
    },
    {
      "epoch": 0.2706322631475281,
      "grad_norm": 1.5933283567428589,
      "learning_rate": 0.0001711379178681098,
      "loss": 1.3108,
      "step": 7557
    },
    {
      "epoch": 0.27066807527709635,
      "grad_norm": 1.7678366899490356,
      "learning_rate": 0.00017112976550573026,
      "loss": 1.6437,
      "step": 7558
    },
    {
      "epoch": 0.2707038874066646,
      "grad_norm": 1.8513715267181396,
      "learning_rate": 0.00017112161218639152,
      "loss": 1.6796,
      "step": 7559
    },
    {
      "epoch": 0.27073969953623295,
      "grad_norm": 1.9389971494674683,
      "learning_rate": 0.00017111345791020324,
      "loss": 1.5358,
      "step": 7560
    },
    {
      "epoch": 0.2707755116658012,
      "grad_norm": 1.4210546016693115,
      "learning_rate": 0.0001711053026772751,
      "loss": 1.546,
      "step": 7561
    },
    {
      "epoch": 0.2708113237953695,
      "grad_norm": 2.9331257343292236,
      "learning_rate": 0.00017109714648771683,
      "loss": 1.4633,
      "step": 7562
    },
    {
      "epoch": 0.27084713592493775,
      "grad_norm": 2.1464321613311768,
      "learning_rate": 0.00017108898934163814,
      "loss": 1.9307,
      "step": 7563
    },
    {
      "epoch": 0.2708829480545061,
      "grad_norm": 1.5199549198150635,
      "learning_rate": 0.0001710808312391488,
      "loss": 1.1444,
      "step": 7564
    },
    {
      "epoch": 0.27091876018407435,
      "grad_norm": 1.5869702100753784,
      "learning_rate": 0.0001710726721803586,
      "loss": 1.3033,
      "step": 7565
    },
    {
      "epoch": 0.2709545723136426,
      "grad_norm": 1.7587640285491943,
      "learning_rate": 0.00017106451216537723,
      "loss": 1.5064,
      "step": 7566
    },
    {
      "epoch": 0.27099038444321094,
      "grad_norm": 1.4721653461456299,
      "learning_rate": 0.00017105635119431457,
      "loss": 1.4631,
      "step": 7567
    },
    {
      "epoch": 0.2710261965727792,
      "grad_norm": 1.7774546146392822,
      "learning_rate": 0.0001710481892672803,
      "loss": 1.5663,
      "step": 7568
    },
    {
      "epoch": 0.2710620087023475,
      "grad_norm": 1.6262949705123901,
      "learning_rate": 0.00017104002638438433,
      "loss": 1.7109,
      "step": 7569
    },
    {
      "epoch": 0.27109782083191575,
      "grad_norm": 1.9079837799072266,
      "learning_rate": 0.00017103186254573642,
      "loss": 1.5319,
      "step": 7570
    },
    {
      "epoch": 0.2711336329614841,
      "grad_norm": 2.4740493297576904,
      "learning_rate": 0.00017102369775144643,
      "loss": 1.4753,
      "step": 7571
    },
    {
      "epoch": 0.27116944509105234,
      "grad_norm": 1.6783044338226318,
      "learning_rate": 0.0001710155320016242,
      "loss": 1.6086,
      "step": 7572
    },
    {
      "epoch": 0.2712052572206206,
      "grad_norm": 1.438635230064392,
      "learning_rate": 0.00017100736529637958,
      "loss": 1.4598,
      "step": 7573
    },
    {
      "epoch": 0.27124106935018893,
      "grad_norm": 1.3612160682678223,
      "learning_rate": 0.0001709991976358225,
      "loss": 1.766,
      "step": 7574
    },
    {
      "epoch": 0.2712768814797572,
      "grad_norm": 2.465081214904785,
      "learning_rate": 0.00017099102902006275,
      "loss": 1.5949,
      "step": 7575
    },
    {
      "epoch": 0.2713126936093255,
      "grad_norm": 2.009199380874634,
      "learning_rate": 0.00017098285944921028,
      "loss": 1.9,
      "step": 7576
    },
    {
      "epoch": 0.27134850573889374,
      "grad_norm": 1.7520637512207031,
      "learning_rate": 0.00017097468892337503,
      "loss": 1.6389,
      "step": 7577
    },
    {
      "epoch": 0.27138431786846207,
      "grad_norm": 1.6367089748382568,
      "learning_rate": 0.00017096651744266686,
      "loss": 1.7878,
      "step": 7578
    },
    {
      "epoch": 0.27142012999803034,
      "grad_norm": 1.7914072275161743,
      "learning_rate": 0.00017095834500719574,
      "loss": 1.5384,
      "step": 7579
    },
    {
      "epoch": 0.2714559421275986,
      "grad_norm": 1.9473427534103394,
      "learning_rate": 0.00017095017161707164,
      "loss": 1.5749,
      "step": 7580
    },
    {
      "epoch": 0.27149175425716693,
      "grad_norm": 1.9787449836730957,
      "learning_rate": 0.00017094199727240447,
      "loss": 1.6192,
      "step": 7581
    },
    {
      "epoch": 0.2715275663867352,
      "grad_norm": 1.2734183073043823,
      "learning_rate": 0.00017093382197330427,
      "loss": 1.3988,
      "step": 7582
    },
    {
      "epoch": 0.27156337851630347,
      "grad_norm": 1.445749044418335,
      "learning_rate": 0.00017092564571988096,
      "loss": 1.4681,
      "step": 7583
    },
    {
      "epoch": 0.27159919064587174,
      "grad_norm": 1.8988313674926758,
      "learning_rate": 0.0001709174685122446,
      "loss": 1.3218,
      "step": 7584
    },
    {
      "epoch": 0.27163500277544006,
      "grad_norm": 1.7073367834091187,
      "learning_rate": 0.00017090929035050513,
      "loss": 1.6704,
      "step": 7585
    },
    {
      "epoch": 0.27167081490500833,
      "grad_norm": 1.4656105041503906,
      "learning_rate": 0.00017090111123477266,
      "loss": 1.4475,
      "step": 7586
    },
    {
      "epoch": 0.2717066270345766,
      "grad_norm": 2.0253167152404785,
      "learning_rate": 0.0001708929311651572,
      "loss": 1.7153,
      "step": 7587
    },
    {
      "epoch": 0.2717424391641449,
      "grad_norm": 1.6637212038040161,
      "learning_rate": 0.0001708847501417688,
      "loss": 1.4888,
      "step": 7588
    },
    {
      "epoch": 0.2717782512937132,
      "grad_norm": 1.6039817333221436,
      "learning_rate": 0.00017087656816471754,
      "loss": 1.2485,
      "step": 7589
    },
    {
      "epoch": 0.27181406342328146,
      "grad_norm": 1.811334490776062,
      "learning_rate": 0.00017086838523411343,
      "loss": 1.7239,
      "step": 7590
    },
    {
      "epoch": 0.27184987555284973,
      "grad_norm": 1.2404311895370483,
      "learning_rate": 0.00017086020135006664,
      "loss": 1.2436,
      "step": 7591
    },
    {
      "epoch": 0.27188568768241805,
      "grad_norm": 1.3640433549880981,
      "learning_rate": 0.00017085201651268722,
      "loss": 1.4845,
      "step": 7592
    },
    {
      "epoch": 0.2719214998119863,
      "grad_norm": 2.0693013668060303,
      "learning_rate": 0.00017084383072208534,
      "loss": 1.7518,
      "step": 7593
    },
    {
      "epoch": 0.2719573119415546,
      "grad_norm": 1.524749994277954,
      "learning_rate": 0.00017083564397837108,
      "loss": 1.6015,
      "step": 7594
    },
    {
      "epoch": 0.2719931240711229,
      "grad_norm": 1.4143753051757812,
      "learning_rate": 0.00017082745628165463,
      "loss": 1.6716,
      "step": 7595
    },
    {
      "epoch": 0.2720289362006912,
      "grad_norm": 1.6675052642822266,
      "learning_rate": 0.0001708192676320461,
      "loss": 1.3527,
      "step": 7596
    },
    {
      "epoch": 0.27206474833025945,
      "grad_norm": 1.9079591035842896,
      "learning_rate": 0.00017081107802965564,
      "loss": 1.601,
      "step": 7597
    },
    {
      "epoch": 0.2721005604598277,
      "grad_norm": 1.47043776512146,
      "learning_rate": 0.0001708028874745935,
      "loss": 1.6689,
      "step": 7598
    },
    {
      "epoch": 0.27213637258939605,
      "grad_norm": 1.9490792751312256,
      "learning_rate": 0.0001707946959669698,
      "loss": 1.3982,
      "step": 7599
    },
    {
      "epoch": 0.2721721847189643,
      "grad_norm": 1.7489452362060547,
      "learning_rate": 0.00017078650350689482,
      "loss": 1.4841,
      "step": 7600
    },
    {
      "epoch": 0.2722079968485326,
      "grad_norm": 1.3652783632278442,
      "learning_rate": 0.00017077831009447878,
      "loss": 1.4928,
      "step": 7601
    },
    {
      "epoch": 0.2722438089781009,
      "grad_norm": 1.3833941221237183,
      "learning_rate": 0.00017077011572983183,
      "loss": 1.2949,
      "step": 7602
    },
    {
      "epoch": 0.2722796211076692,
      "grad_norm": 1.9464653730392456,
      "learning_rate": 0.00017076192041306425,
      "loss": 1.6098,
      "step": 7603
    },
    {
      "epoch": 0.27231543323723745,
      "grad_norm": 1.9903013706207275,
      "learning_rate": 0.00017075372414428633,
      "loss": 1.9585,
      "step": 7604
    },
    {
      "epoch": 0.2723512453668057,
      "grad_norm": 1.821960210800171,
      "learning_rate": 0.00017074552692360832,
      "loss": 1.3332,
      "step": 7605
    },
    {
      "epoch": 0.27238705749637404,
      "grad_norm": 2.120266914367676,
      "learning_rate": 0.00017073732875114045,
      "loss": 1.465,
      "step": 7606
    },
    {
      "epoch": 0.2724228696259423,
      "grad_norm": 1.6445937156677246,
      "learning_rate": 0.0001707291296269931,
      "loss": 1.5826,
      "step": 7607
    },
    {
      "epoch": 0.2724586817555106,
      "grad_norm": 1.6000337600708008,
      "learning_rate": 0.00017072092955127657,
      "loss": 1.4611,
      "step": 7608
    },
    {
      "epoch": 0.2724944938850789,
      "grad_norm": 1.9415369033813477,
      "learning_rate": 0.00017071272852410113,
      "loss": 1.5284,
      "step": 7609
    },
    {
      "epoch": 0.2725303060146472,
      "grad_norm": 1.6123709678649902,
      "learning_rate": 0.00017070452654557717,
      "loss": 1.5851,
      "step": 7610
    },
    {
      "epoch": 0.27256611814421544,
      "grad_norm": 1.5228267908096313,
      "learning_rate": 0.00017069632361581497,
      "loss": 1.8117,
      "step": 7611
    },
    {
      "epoch": 0.2726019302737837,
      "grad_norm": 1.7121150493621826,
      "learning_rate": 0.00017068811973492497,
      "loss": 1.6039,
      "step": 7612
    },
    {
      "epoch": 0.27263774240335203,
      "grad_norm": 1.8781545162200928,
      "learning_rate": 0.00017067991490301744,
      "loss": 1.5765,
      "step": 7613
    },
    {
      "epoch": 0.2726735545329203,
      "grad_norm": 2.0894546508789062,
      "learning_rate": 0.00017067170912020286,
      "loss": 1.5713,
      "step": 7614
    },
    {
      "epoch": 0.2727093666624886,
      "grad_norm": 1.5253492593765259,
      "learning_rate": 0.0001706635023865916,
      "loss": 1.4626,
      "step": 7615
    },
    {
      "epoch": 0.2727451787920569,
      "grad_norm": 1.6403868198394775,
      "learning_rate": 0.00017065529470229403,
      "loss": 1.5156,
      "step": 7616
    },
    {
      "epoch": 0.27278099092162517,
      "grad_norm": 1.4445627927780151,
      "learning_rate": 0.00017064708606742067,
      "loss": 1.7814,
      "step": 7617
    },
    {
      "epoch": 0.27281680305119343,
      "grad_norm": 1.6161423921585083,
      "learning_rate": 0.00017063887648208185,
      "loss": 1.8426,
      "step": 7618
    },
    {
      "epoch": 0.2728526151807617,
      "grad_norm": 2.159152030944824,
      "learning_rate": 0.00017063066594638805,
      "loss": 1.6711,
      "step": 7619
    },
    {
      "epoch": 0.27288842731033003,
      "grad_norm": 1.585121989250183,
      "learning_rate": 0.0001706224544604498,
      "loss": 1.6355,
      "step": 7620
    },
    {
      "epoch": 0.2729242394398983,
      "grad_norm": 2.4911491870880127,
      "learning_rate": 0.00017061424202437748,
      "loss": 1.8799,
      "step": 7621
    },
    {
      "epoch": 0.27296005156946657,
      "grad_norm": 1.3590527772903442,
      "learning_rate": 0.00017060602863828165,
      "loss": 1.3842,
      "step": 7622
    },
    {
      "epoch": 0.2729958636990349,
      "grad_norm": 1.6807591915130615,
      "learning_rate": 0.00017059781430227275,
      "loss": 1.8145,
      "step": 7623
    },
    {
      "epoch": 0.27303167582860316,
      "grad_norm": 1.9049506187438965,
      "learning_rate": 0.00017058959901646134,
      "loss": 1.5613,
      "step": 7624
    },
    {
      "epoch": 0.27306748795817143,
      "grad_norm": 1.9668833017349243,
      "learning_rate": 0.00017058138278095792,
      "loss": 1.3905,
      "step": 7625
    },
    {
      "epoch": 0.2731033000877397,
      "grad_norm": 1.8975080251693726,
      "learning_rate": 0.00017057316559587307,
      "loss": 1.6166,
      "step": 7626
    },
    {
      "epoch": 0.273139112217308,
      "grad_norm": 1.7440699338912964,
      "learning_rate": 0.00017056494746131725,
      "loss": 1.5235,
      "step": 7627
    },
    {
      "epoch": 0.2731749243468763,
      "grad_norm": 1.5745868682861328,
      "learning_rate": 0.00017055672837740113,
      "loss": 1.4461,
      "step": 7628
    },
    {
      "epoch": 0.27321073647644456,
      "grad_norm": 1.4910067319869995,
      "learning_rate": 0.00017054850834423522,
      "loss": 1.6589,
      "step": 7629
    },
    {
      "epoch": 0.27324654860601283,
      "grad_norm": 1.9283604621887207,
      "learning_rate": 0.00017054028736193013,
      "loss": 1.3877,
      "step": 7630
    },
    {
      "epoch": 0.27328236073558115,
      "grad_norm": 1.5712474584579468,
      "learning_rate": 0.00017053206543059647,
      "loss": 1.6484,
      "step": 7631
    },
    {
      "epoch": 0.2733181728651494,
      "grad_norm": 1.4814081192016602,
      "learning_rate": 0.00017052384255034485,
      "loss": 1.58,
      "step": 7632
    },
    {
      "epoch": 0.2733539849947177,
      "grad_norm": 2.332150936126709,
      "learning_rate": 0.00017051561872128592,
      "loss": 1.4022,
      "step": 7633
    },
    {
      "epoch": 0.273389797124286,
      "grad_norm": 1.6437220573425293,
      "learning_rate": 0.00017050739394353028,
      "loss": 1.4863,
      "step": 7634
    },
    {
      "epoch": 0.2734256092538543,
      "grad_norm": 2.036505937576294,
      "learning_rate": 0.00017049916821718861,
      "loss": 1.5958,
      "step": 7635
    },
    {
      "epoch": 0.27346142138342255,
      "grad_norm": 1.2595261335372925,
      "learning_rate": 0.00017049094154237155,
      "loss": 1.4669,
      "step": 7636
    },
    {
      "epoch": 0.2734972335129908,
      "grad_norm": 1.5755056142807007,
      "learning_rate": 0.0001704827139191898,
      "loss": 1.6032,
      "step": 7637
    },
    {
      "epoch": 0.27353304564255915,
      "grad_norm": 1.5014159679412842,
      "learning_rate": 0.00017047448534775406,
      "loss": 1.6173,
      "step": 7638
    },
    {
      "epoch": 0.2735688577721274,
      "grad_norm": 1.7377209663391113,
      "learning_rate": 0.00017046625582817503,
      "loss": 1.4034,
      "step": 7639
    },
    {
      "epoch": 0.2736046699016957,
      "grad_norm": 1.9546681642532349,
      "learning_rate": 0.00017045802536056344,
      "loss": 1.7961,
      "step": 7640
    },
    {
      "epoch": 0.273640482031264,
      "grad_norm": 1.5216879844665527,
      "learning_rate": 0.00017044979394502995,
      "loss": 1.6273,
      "step": 7641
    },
    {
      "epoch": 0.2736762941608323,
      "grad_norm": 2.2344608306884766,
      "learning_rate": 0.0001704415615816854,
      "loss": 1.5878,
      "step": 7642
    },
    {
      "epoch": 0.27371210629040055,
      "grad_norm": 1.5303434133529663,
      "learning_rate": 0.0001704333282706405,
      "loss": 1.5562,
      "step": 7643
    },
    {
      "epoch": 0.2737479184199688,
      "grad_norm": 2.065340280532837,
      "learning_rate": 0.00017042509401200598,
      "loss": 1.6453,
      "step": 7644
    },
    {
      "epoch": 0.27378373054953714,
      "grad_norm": 1.7603098154067993,
      "learning_rate": 0.00017041685880589272,
      "loss": 1.7976,
      "step": 7645
    },
    {
      "epoch": 0.2738195426791054,
      "grad_norm": 1.29735267162323,
      "learning_rate": 0.0001704086226524114,
      "loss": 1.4381,
      "step": 7646
    },
    {
      "epoch": 0.2738553548086737,
      "grad_norm": 1.7243160009384155,
      "learning_rate": 0.0001704003855516729,
      "loss": 1.6664,
      "step": 7647
    },
    {
      "epoch": 0.273891166938242,
      "grad_norm": 2.5377590656280518,
      "learning_rate": 0.00017039214750378805,
      "loss": 1.7078,
      "step": 7648
    },
    {
      "epoch": 0.2739269790678103,
      "grad_norm": 2.229074478149414,
      "learning_rate": 0.00017038390850886766,
      "loss": 1.4732,
      "step": 7649
    },
    {
      "epoch": 0.27396279119737854,
      "grad_norm": 1.3973140716552734,
      "learning_rate": 0.00017037566856702255,
      "loss": 1.3579,
      "step": 7650
    },
    {
      "epoch": 0.2739986033269468,
      "grad_norm": 1.4385688304901123,
      "learning_rate": 0.00017036742767836355,
      "loss": 1.6784,
      "step": 7651
    },
    {
      "epoch": 0.27403441545651513,
      "grad_norm": 1.4046844244003296,
      "learning_rate": 0.00017035918584300163,
      "loss": 1.4343,
      "step": 7652
    },
    {
      "epoch": 0.2740702275860834,
      "grad_norm": 1.4777336120605469,
      "learning_rate": 0.00017035094306104762,
      "loss": 1.752,
      "step": 7653
    },
    {
      "epoch": 0.2741060397156517,
      "grad_norm": 1.6677403450012207,
      "learning_rate": 0.0001703426993326124,
      "loss": 1.5784,
      "step": 7654
    },
    {
      "epoch": 0.27414185184522,
      "grad_norm": 1.8168667554855347,
      "learning_rate": 0.0001703344546578069,
      "loss": 1.7299,
      "step": 7655
    },
    {
      "epoch": 0.27417766397478827,
      "grad_norm": 1.6721216440200806,
      "learning_rate": 0.00017032620903674207,
      "loss": 1.5574,
      "step": 7656
    },
    {
      "epoch": 0.27421347610435653,
      "grad_norm": 1.7680671215057373,
      "learning_rate": 0.0001703179624695288,
      "loss": 1.3656,
      "step": 7657
    },
    {
      "epoch": 0.2742492882339248,
      "grad_norm": 2.2017982006073,
      "learning_rate": 0.00017030971495627802,
      "loss": 1.5316,
      "step": 7658
    },
    {
      "epoch": 0.27428510036349313,
      "grad_norm": 1.4911608695983887,
      "learning_rate": 0.00017030146649710072,
      "loss": 1.6055,
      "step": 7659
    },
    {
      "epoch": 0.2743209124930614,
      "grad_norm": 1.9482650756835938,
      "learning_rate": 0.00017029321709210787,
      "loss": 1.7288,
      "step": 7660
    },
    {
      "epoch": 0.27435672462262967,
      "grad_norm": 1.816979169845581,
      "learning_rate": 0.00017028496674141051,
      "loss": 1.1743,
      "step": 7661
    },
    {
      "epoch": 0.274392536752198,
      "grad_norm": 1.9508891105651855,
      "learning_rate": 0.0001702767154451195,
      "loss": 1.5504,
      "step": 7662
    },
    {
      "epoch": 0.27442834888176626,
      "grad_norm": 1.4764585494995117,
      "learning_rate": 0.000170268463203346,
      "loss": 1.5337,
      "step": 7663
    },
    {
      "epoch": 0.27446416101133453,
      "grad_norm": 1.4491671323776245,
      "learning_rate": 0.00017026021001620095,
      "loss": 1.6598,
      "step": 7664
    },
    {
      "epoch": 0.2744999731409028,
      "grad_norm": 1.9225811958312988,
      "learning_rate": 0.00017025195588379538,
      "loss": 1.4929,
      "step": 7665
    },
    {
      "epoch": 0.2745357852704711,
      "grad_norm": 1.355978012084961,
      "learning_rate": 0.0001702437008062404,
      "loss": 1.1573,
      "step": 7666
    },
    {
      "epoch": 0.2745715974000394,
      "grad_norm": 1.796846866607666,
      "learning_rate": 0.00017023544478364698,
      "loss": 1.3795,
      "step": 7667
    },
    {
      "epoch": 0.27460740952960766,
      "grad_norm": 1.581048846244812,
      "learning_rate": 0.0001702271878161263,
      "loss": 1.365,
      "step": 7668
    },
    {
      "epoch": 0.274643221659176,
      "grad_norm": 1.55429208278656,
      "learning_rate": 0.0001702189299037894,
      "loss": 1.5851,
      "step": 7669
    },
    {
      "epoch": 0.27467903378874425,
      "grad_norm": 1.660404920578003,
      "learning_rate": 0.00017021067104674734,
      "loss": 2.0028,
      "step": 7670
    },
    {
      "epoch": 0.2747148459183125,
      "grad_norm": 2.531282663345337,
      "learning_rate": 0.00017020241124511128,
      "loss": 1.6508,
      "step": 7671
    },
    {
      "epoch": 0.2747506580478808,
      "grad_norm": 1.8213257789611816,
      "learning_rate": 0.0001701941504989923,
      "loss": 1.4164,
      "step": 7672
    },
    {
      "epoch": 0.2747864701774491,
      "grad_norm": 1.5226930379867554,
      "learning_rate": 0.00017018588880850162,
      "loss": 1.7048,
      "step": 7673
    },
    {
      "epoch": 0.2748222823070174,
      "grad_norm": 1.236707329750061,
      "learning_rate": 0.0001701776261737503,
      "loss": 1.4652,
      "step": 7674
    },
    {
      "epoch": 0.27485809443658565,
      "grad_norm": 1.5473086833953857,
      "learning_rate": 0.00017016936259484953,
      "loss": 1.5635,
      "step": 7675
    },
    {
      "epoch": 0.274893906566154,
      "grad_norm": 1.5289727449417114,
      "learning_rate": 0.00017016109807191056,
      "loss": 1.5221,
      "step": 7676
    },
    {
      "epoch": 0.27492971869572225,
      "grad_norm": 1.2557204961776733,
      "learning_rate": 0.00017015283260504447,
      "loss": 1.5362,
      "step": 7677
    },
    {
      "epoch": 0.2749655308252905,
      "grad_norm": 1.6917392015457153,
      "learning_rate": 0.00017014456619436253,
      "loss": 1.5005,
      "step": 7678
    },
    {
      "epoch": 0.2750013429548588,
      "grad_norm": 1.553214192390442,
      "learning_rate": 0.00017013629883997594,
      "loss": 1.6225,
      "step": 7679
    },
    {
      "epoch": 0.2750371550844271,
      "grad_norm": 1.3000906705856323,
      "learning_rate": 0.00017012803054199587,
      "loss": 1.6618,
      "step": 7680
    },
    {
      "epoch": 0.2750729672139954,
      "grad_norm": 1.7456116676330566,
      "learning_rate": 0.00017011976130053367,
      "loss": 1.4712,
      "step": 7681
    },
    {
      "epoch": 0.27510877934356365,
      "grad_norm": 1.459320068359375,
      "learning_rate": 0.00017011149111570051,
      "loss": 1.6217,
      "step": 7682
    },
    {
      "epoch": 0.27514459147313197,
      "grad_norm": 2.354222297668457,
      "learning_rate": 0.00017010321998760762,
      "loss": 1.6253,
      "step": 7683
    },
    {
      "epoch": 0.27518040360270024,
      "grad_norm": 1.62624192237854,
      "learning_rate": 0.0001700949479163664,
      "loss": 1.4179,
      "step": 7684
    },
    {
      "epoch": 0.2752162157322685,
      "grad_norm": 1.8651334047317505,
      "learning_rate": 0.00017008667490208803,
      "loss": 1.4862,
      "step": 7685
    },
    {
      "epoch": 0.2752520278618368,
      "grad_norm": 1.6018571853637695,
      "learning_rate": 0.00017007840094488387,
      "loss": 1.7953,
      "step": 7686
    },
    {
      "epoch": 0.2752878399914051,
      "grad_norm": 1.353165626525879,
      "learning_rate": 0.00017007012604486525,
      "loss": 1.6099,
      "step": 7687
    },
    {
      "epoch": 0.2753236521209734,
      "grad_norm": 1.3602633476257324,
      "learning_rate": 0.0001700618502021434,
      "loss": 1.5703,
      "step": 7688
    },
    {
      "epoch": 0.27535946425054164,
      "grad_norm": 1.2508971691131592,
      "learning_rate": 0.00017005357341682979,
      "loss": 1.6003,
      "step": 7689
    },
    {
      "epoch": 0.27539527638010997,
      "grad_norm": 1.8892033100128174,
      "learning_rate": 0.0001700452956890357,
      "loss": 1.7102,
      "step": 7690
    },
    {
      "epoch": 0.27543108850967823,
      "grad_norm": 1.4698898792266846,
      "learning_rate": 0.0001700370170188725,
      "loss": 1.6364,
      "step": 7691
    },
    {
      "epoch": 0.2754669006392465,
      "grad_norm": 1.6167817115783691,
      "learning_rate": 0.00017002873740645157,
      "loss": 1.7467,
      "step": 7692
    },
    {
      "epoch": 0.2755027127688148,
      "grad_norm": 1.442500352859497,
      "learning_rate": 0.00017002045685188431,
      "loss": 1.4752,
      "step": 7693
    },
    {
      "epoch": 0.2755385248983831,
      "grad_norm": 1.7867764234542847,
      "learning_rate": 0.00017001217535528215,
      "loss": 1.5237,
      "step": 7694
    },
    {
      "epoch": 0.27557433702795137,
      "grad_norm": 1.4577064514160156,
      "learning_rate": 0.00017000389291675644,
      "loss": 1.683,
      "step": 7695
    },
    {
      "epoch": 0.27561014915751963,
      "grad_norm": 1.7021827697753906,
      "learning_rate": 0.00016999560953641867,
      "loss": 1.7309,
      "step": 7696
    },
    {
      "epoch": 0.27564596128708796,
      "grad_norm": 1.3312740325927734,
      "learning_rate": 0.00016998732521438024,
      "loss": 1.2283,
      "step": 7697
    },
    {
      "epoch": 0.27568177341665623,
      "grad_norm": 1.972090482711792,
      "learning_rate": 0.00016997903995075265,
      "loss": 1.3293,
      "step": 7698
    },
    {
      "epoch": 0.2757175855462245,
      "grad_norm": 1.9187233448028564,
      "learning_rate": 0.00016997075374564733,
      "loss": 1.3078,
      "step": 7699
    },
    {
      "epoch": 0.27575339767579277,
      "grad_norm": 1.3226217031478882,
      "learning_rate": 0.00016996246659917578,
      "loss": 1.3863,
      "step": 7700
    },
    {
      "epoch": 0.2757892098053611,
      "grad_norm": 2.0401716232299805,
      "learning_rate": 0.0001699541785114495,
      "loss": 1.4268,
      "step": 7701
    },
    {
      "epoch": 0.27582502193492936,
      "grad_norm": 1.7860567569732666,
      "learning_rate": 0.00016994588948257997,
      "loss": 1.358,
      "step": 7702
    },
    {
      "epoch": 0.27586083406449763,
      "grad_norm": 1.6787052154541016,
      "learning_rate": 0.0001699375995126787,
      "loss": 1.6829,
      "step": 7703
    },
    {
      "epoch": 0.27589664619406595,
      "grad_norm": 1.7264974117279053,
      "learning_rate": 0.00016992930860185726,
      "loss": 1.5136,
      "step": 7704
    },
    {
      "epoch": 0.2759324583236342,
      "grad_norm": 1.6791878938674927,
      "learning_rate": 0.0001699210167502272,
      "loss": 1.3216,
      "step": 7705
    },
    {
      "epoch": 0.2759682704532025,
      "grad_norm": 1.2904554605484009,
      "learning_rate": 0.00016991272395790007,
      "loss": 1.3571,
      "step": 7706
    },
    {
      "epoch": 0.27600408258277076,
      "grad_norm": 2.196139335632324,
      "learning_rate": 0.00016990443022498735,
      "loss": 1.5075,
      "step": 7707
    },
    {
      "epoch": 0.2760398947123391,
      "grad_norm": 1.6233375072479248,
      "learning_rate": 0.0001698961355516007,
      "loss": 1.5892,
      "step": 7708
    },
    {
      "epoch": 0.27607570684190735,
      "grad_norm": 1.7111704349517822,
      "learning_rate": 0.00016988783993785177,
      "loss": 1.6294,
      "step": 7709
    },
    {
      "epoch": 0.2761115189714756,
      "grad_norm": 1.633927345275879,
      "learning_rate": 0.00016987954338385202,
      "loss": 1.3068,
      "step": 7710
    },
    {
      "epoch": 0.27614733110104395,
      "grad_norm": 1.6509286165237427,
      "learning_rate": 0.0001698712458897132,
      "loss": 1.5484,
      "step": 7711
    },
    {
      "epoch": 0.2761831432306122,
      "grad_norm": 1.8952949047088623,
      "learning_rate": 0.0001698629474555469,
      "loss": 1.4961,
      "step": 7712
    },
    {
      "epoch": 0.2762189553601805,
      "grad_norm": 2.397242546081543,
      "learning_rate": 0.00016985464808146473,
      "loss": 1.8338,
      "step": 7713
    },
    {
      "epoch": 0.27625476748974875,
      "grad_norm": 1.9433248043060303,
      "learning_rate": 0.0001698463477675784,
      "loss": 1.5395,
      "step": 7714
    },
    {
      "epoch": 0.2762905796193171,
      "grad_norm": 2.032348394393921,
      "learning_rate": 0.00016983804651399956,
      "loss": 1.5128,
      "step": 7715
    },
    {
      "epoch": 0.27632639174888535,
      "grad_norm": 2.3550405502319336,
      "learning_rate": 0.00016982974432083986,
      "loss": 1.4864,
      "step": 7716
    },
    {
      "epoch": 0.2763622038784536,
      "grad_norm": 1.410760521888733,
      "learning_rate": 0.00016982144118821103,
      "loss": 1.6236,
      "step": 7717
    },
    {
      "epoch": 0.27639801600802194,
      "grad_norm": 1.9511107206344604,
      "learning_rate": 0.0001698131371162248,
      "loss": 1.7513,
      "step": 7718
    },
    {
      "epoch": 0.2764338281375902,
      "grad_norm": 1.6737018823623657,
      "learning_rate": 0.00016980483210499286,
      "loss": 1.5436,
      "step": 7719
    },
    {
      "epoch": 0.2764696402671585,
      "grad_norm": 1.807410478591919,
      "learning_rate": 0.00016979652615462692,
      "loss": 1.773,
      "step": 7720
    },
    {
      "epoch": 0.27650545239672675,
      "grad_norm": 1.6671122312545776,
      "learning_rate": 0.00016978821926523873,
      "loss": 1.5451,
      "step": 7721
    },
    {
      "epoch": 0.27654126452629507,
      "grad_norm": 1.9303603172302246,
      "learning_rate": 0.00016977991143694014,
      "loss": 1.6225,
      "step": 7722
    },
    {
      "epoch": 0.27657707665586334,
      "grad_norm": 1.4524554014205933,
      "learning_rate": 0.00016977160266984283,
      "loss": 1.4761,
      "step": 7723
    },
    {
      "epoch": 0.2766128887854316,
      "grad_norm": 1.716299295425415,
      "learning_rate": 0.00016976329296405855,
      "loss": 1.3507,
      "step": 7724
    },
    {
      "epoch": 0.27664870091499993,
      "grad_norm": 1.3396046161651611,
      "learning_rate": 0.0001697549823196992,
      "loss": 1.7153,
      "step": 7725
    },
    {
      "epoch": 0.2766845130445682,
      "grad_norm": 1.991355299949646,
      "learning_rate": 0.00016974667073687655,
      "loss": 1.648,
      "step": 7726
    },
    {
      "epoch": 0.27672032517413647,
      "grad_norm": 1.623319149017334,
      "learning_rate": 0.00016973835821570236,
      "loss": 1.5506,
      "step": 7727
    },
    {
      "epoch": 0.27675613730370474,
      "grad_norm": 1.4190425872802734,
      "learning_rate": 0.00016973004475628856,
      "loss": 1.4248,
      "step": 7728
    },
    {
      "epoch": 0.27679194943327307,
      "grad_norm": 1.6432379484176636,
      "learning_rate": 0.00016972173035874693,
      "loss": 1.7124,
      "step": 7729
    },
    {
      "epoch": 0.27682776156284133,
      "grad_norm": 1.4056947231292725,
      "learning_rate": 0.00016971341502318936,
      "loss": 1.4815,
      "step": 7730
    },
    {
      "epoch": 0.2768635736924096,
      "grad_norm": 1.9628760814666748,
      "learning_rate": 0.00016970509874972774,
      "loss": 1.7848,
      "step": 7731
    },
    {
      "epoch": 0.27689938582197793,
      "grad_norm": 1.8493858575820923,
      "learning_rate": 0.0001696967815384739,
      "loss": 1.6528,
      "step": 7732
    },
    {
      "epoch": 0.2769351979515462,
      "grad_norm": 1.7261875867843628,
      "learning_rate": 0.0001696884633895398,
      "loss": 1.5981,
      "step": 7733
    },
    {
      "epoch": 0.27697101008111447,
      "grad_norm": 1.6887271404266357,
      "learning_rate": 0.00016968014430303728,
      "loss": 1.6599,
      "step": 7734
    },
    {
      "epoch": 0.27700682221068273,
      "grad_norm": 1.777239203453064,
      "learning_rate": 0.0001696718242790783,
      "loss": 1.4756,
      "step": 7735
    },
    {
      "epoch": 0.27704263434025106,
      "grad_norm": 1.546141266822815,
      "learning_rate": 0.0001696635033177748,
      "loss": 1.551,
      "step": 7736
    },
    {
      "epoch": 0.27707844646981933,
      "grad_norm": 1.5476120710372925,
      "learning_rate": 0.00016965518141923874,
      "loss": 1.6505,
      "step": 7737
    },
    {
      "epoch": 0.2771142585993876,
      "grad_norm": 1.809834361076355,
      "learning_rate": 0.00016964685858358202,
      "loss": 1.8532,
      "step": 7738
    },
    {
      "epoch": 0.2771500707289559,
      "grad_norm": 2.0858800411224365,
      "learning_rate": 0.0001696385348109167,
      "loss": 1.7213,
      "step": 7739
    },
    {
      "epoch": 0.2771858828585242,
      "grad_norm": 1.580114722251892,
      "learning_rate": 0.0001696302101013547,
      "loss": 1.6748,
      "step": 7740
    },
    {
      "epoch": 0.27722169498809246,
      "grad_norm": 1.333909273147583,
      "learning_rate": 0.00016962188445500807,
      "loss": 1.632,
      "step": 7741
    },
    {
      "epoch": 0.27725750711766073,
      "grad_norm": 1.7078574895858765,
      "learning_rate": 0.00016961355787198875,
      "loss": 1.7498,
      "step": 7742
    },
    {
      "epoch": 0.27729331924722905,
      "grad_norm": 1.3296380043029785,
      "learning_rate": 0.00016960523035240883,
      "loss": 1.5679,
      "step": 7743
    },
    {
      "epoch": 0.2773291313767973,
      "grad_norm": 1.2145541906356812,
      "learning_rate": 0.0001695969018963803,
      "loss": 1.5629,
      "step": 7744
    },
    {
      "epoch": 0.2773649435063656,
      "grad_norm": 1.5315901041030884,
      "learning_rate": 0.00016958857250401525,
      "loss": 1.7075,
      "step": 7745
    },
    {
      "epoch": 0.2774007556359339,
      "grad_norm": 1.5283602476119995,
      "learning_rate": 0.0001695802421754257,
      "loss": 1.7226,
      "step": 7746
    },
    {
      "epoch": 0.2774365677655022,
      "grad_norm": 1.7161786556243896,
      "learning_rate": 0.00016957191091072376,
      "loss": 1.4375,
      "step": 7747
    },
    {
      "epoch": 0.27747237989507045,
      "grad_norm": 1.6843448877334595,
      "learning_rate": 0.0001695635787100215,
      "loss": 1.3808,
      "step": 7748
    },
    {
      "epoch": 0.2775081920246387,
      "grad_norm": 1.336690068244934,
      "learning_rate": 0.000169555245573431,
      "loss": 1.3474,
      "step": 7749
    },
    {
      "epoch": 0.27754400415420705,
      "grad_norm": 1.7416538000106812,
      "learning_rate": 0.0001695469115010644,
      "loss": 1.2874,
      "step": 7750
    },
    {
      "epoch": 0.2775798162837753,
      "grad_norm": 1.2422138452529907,
      "learning_rate": 0.00016953857649303381,
      "loss": 1.6665,
      "step": 7751
    },
    {
      "epoch": 0.2776156284133436,
      "grad_norm": 1.4898710250854492,
      "learning_rate": 0.00016953024054945138,
      "loss": 1.3389,
      "step": 7752
    },
    {
      "epoch": 0.2776514405429119,
      "grad_norm": 2.7302865982055664,
      "learning_rate": 0.00016952190367042926,
      "loss": 1.5453,
      "step": 7753
    },
    {
      "epoch": 0.2776872526724802,
      "grad_norm": 3.756218671798706,
      "learning_rate": 0.0001695135658560796,
      "loss": 2.0241,
      "step": 7754
    },
    {
      "epoch": 0.27772306480204845,
      "grad_norm": 1.5968868732452393,
      "learning_rate": 0.00016950522710651455,
      "loss": 1.335,
      "step": 7755
    },
    {
      "epoch": 0.2777588769316167,
      "grad_norm": 1.5735868215560913,
      "learning_rate": 0.00016949688742184637,
      "loss": 1.5448,
      "step": 7756
    },
    {
      "epoch": 0.27779468906118504,
      "grad_norm": 1.4668728113174438,
      "learning_rate": 0.0001694885468021872,
      "loss": 1.5775,
      "step": 7757
    },
    {
      "epoch": 0.2778305011907533,
      "grad_norm": 1.89346182346344,
      "learning_rate": 0.00016948020524764924,
      "loss": 1.3931,
      "step": 7758
    },
    {
      "epoch": 0.2778663133203216,
      "grad_norm": 1.6620376110076904,
      "learning_rate": 0.00016947186275834475,
      "loss": 1.6433,
      "step": 7759
    },
    {
      "epoch": 0.2779021254498899,
      "grad_norm": 1.8120241165161133,
      "learning_rate": 0.00016946351933438595,
      "loss": 1.418,
      "step": 7760
    },
    {
      "epoch": 0.27793793757945817,
      "grad_norm": 1.5897480249404907,
      "learning_rate": 0.00016945517497588512,
      "loss": 1.5138,
      "step": 7761
    },
    {
      "epoch": 0.27797374970902644,
      "grad_norm": 2.4386966228485107,
      "learning_rate": 0.00016944682968295452,
      "loss": 1.5283,
      "step": 7762
    },
    {
      "epoch": 0.2780095618385947,
      "grad_norm": 1.8611398935317993,
      "learning_rate": 0.00016943848345570638,
      "loss": 1.6865,
      "step": 7763
    },
    {
      "epoch": 0.27804537396816303,
      "grad_norm": 1.441057562828064,
      "learning_rate": 0.00016943013629425302,
      "loss": 1.453,
      "step": 7764
    },
    {
      "epoch": 0.2780811860977313,
      "grad_norm": 1.4636635780334473,
      "learning_rate": 0.00016942178819870672,
      "loss": 1.2875,
      "step": 7765
    },
    {
      "epoch": 0.27811699822729957,
      "grad_norm": 1.454779028892517,
      "learning_rate": 0.00016941343916917982,
      "loss": 1.6621,
      "step": 7766
    },
    {
      "epoch": 0.2781528103568679,
      "grad_norm": 1.4269800186157227,
      "learning_rate": 0.00016940508920578463,
      "loss": 1.7387,
      "step": 7767
    },
    {
      "epoch": 0.27818862248643617,
      "grad_norm": 2.06813645362854,
      "learning_rate": 0.00016939673830863348,
      "loss": 1.4947,
      "step": 7768
    },
    {
      "epoch": 0.27822443461600443,
      "grad_norm": 1.9147825241088867,
      "learning_rate": 0.00016938838647783877,
      "loss": 1.5197,
      "step": 7769
    },
    {
      "epoch": 0.2782602467455727,
      "grad_norm": 2.2611420154571533,
      "learning_rate": 0.00016938003371351278,
      "loss": 1.7869,
      "step": 7770
    },
    {
      "epoch": 0.27829605887514103,
      "grad_norm": 2.2714648246765137,
      "learning_rate": 0.00016937168001576795,
      "loss": 1.4491,
      "step": 7771
    },
    {
      "epoch": 0.2783318710047093,
      "grad_norm": 1.5648062229156494,
      "learning_rate": 0.00016936332538471666,
      "loss": 1.3759,
      "step": 7772
    },
    {
      "epoch": 0.27836768313427757,
      "grad_norm": 1.8871150016784668,
      "learning_rate": 0.00016935496982047128,
      "loss": 1.9112,
      "step": 7773
    },
    {
      "epoch": 0.2784034952638459,
      "grad_norm": 1.4186395406723022,
      "learning_rate": 0.00016934661332314424,
      "loss": 1.7234,
      "step": 7774
    },
    {
      "epoch": 0.27843930739341416,
      "grad_norm": 1.551112174987793,
      "learning_rate": 0.000169338255892848,
      "loss": 1.5181,
      "step": 7775
    },
    {
      "epoch": 0.27847511952298243,
      "grad_norm": 1.92075514793396,
      "learning_rate": 0.00016932989752969495,
      "loss": 1.5922,
      "step": 7776
    },
    {
      "epoch": 0.2785109316525507,
      "grad_norm": 1.918872356414795,
      "learning_rate": 0.00016932153823379754,
      "loss": 1.6461,
      "step": 7777
    },
    {
      "epoch": 0.278546743782119,
      "grad_norm": 1.9076799154281616,
      "learning_rate": 0.00016931317800526828,
      "loss": 1.9765,
      "step": 7778
    },
    {
      "epoch": 0.2785825559116873,
      "grad_norm": 1.9921611547470093,
      "learning_rate": 0.0001693048168442196,
      "loss": 1.6124,
      "step": 7779
    },
    {
      "epoch": 0.27861836804125556,
      "grad_norm": 1.2875360250473022,
      "learning_rate": 0.000169296454750764,
      "loss": 1.4644,
      "step": 7780
    },
    {
      "epoch": 0.2786541801708239,
      "grad_norm": 1.5666717290878296,
      "learning_rate": 0.00016928809172501397,
      "loss": 1.8456,
      "step": 7781
    },
    {
      "epoch": 0.27868999230039215,
      "grad_norm": 2.1061506271362305,
      "learning_rate": 0.00016927972776708208,
      "loss": 1.3571,
      "step": 7782
    },
    {
      "epoch": 0.2787258044299604,
      "grad_norm": 1.7013448476791382,
      "learning_rate": 0.0001692713628770808,
      "loss": 1.1378,
      "step": 7783
    },
    {
      "epoch": 0.2787616165595287,
      "grad_norm": 2.2516982555389404,
      "learning_rate": 0.00016926299705512273,
      "loss": 1.2538,
      "step": 7784
    },
    {
      "epoch": 0.278797428689097,
      "grad_norm": 1.5621269941329956,
      "learning_rate": 0.0001692546303013203,
      "loss": 1.8246,
      "step": 7785
    },
    {
      "epoch": 0.2788332408186653,
      "grad_norm": 2.3607735633850098,
      "learning_rate": 0.0001692462626157862,
      "loss": 1.4497,
      "step": 7786
    },
    {
      "epoch": 0.27886905294823355,
      "grad_norm": 2.1050877571105957,
      "learning_rate": 0.00016923789399863294,
      "loss": 1.4979,
      "step": 7787
    },
    {
      "epoch": 0.2789048650778019,
      "grad_norm": 1.711255431175232,
      "learning_rate": 0.00016922952444997313,
      "loss": 1.5194,
      "step": 7788
    },
    {
      "epoch": 0.27894067720737015,
      "grad_norm": 1.6491427421569824,
      "learning_rate": 0.00016922115396991939,
      "loss": 1.4156,
      "step": 7789
    },
    {
      "epoch": 0.2789764893369384,
      "grad_norm": 1.5196592807769775,
      "learning_rate": 0.00016921278255858425,
      "loss": 1.4432,
      "step": 7790
    },
    {
      "epoch": 0.2790123014665067,
      "grad_norm": 1.4955500364303589,
      "learning_rate": 0.00016920441021608048,
      "loss": 1.806,
      "step": 7791
    },
    {
      "epoch": 0.279048113596075,
      "grad_norm": 1.6842684745788574,
      "learning_rate": 0.0001691960369425206,
      "loss": 1.628,
      "step": 7792
    },
    {
      "epoch": 0.2790839257256433,
      "grad_norm": 1.859779953956604,
      "learning_rate": 0.0001691876627380173,
      "loss": 1.4083,
      "step": 7793
    },
    {
      "epoch": 0.27911973785521155,
      "grad_norm": 1.8693050146102905,
      "learning_rate": 0.00016917928760268325,
      "loss": 1.6231,
      "step": 7794
    },
    {
      "epoch": 0.27915554998477987,
      "grad_norm": 1.3143736124038696,
      "learning_rate": 0.0001691709115366311,
      "loss": 1.5244,
      "step": 7795
    },
    {
      "epoch": 0.27919136211434814,
      "grad_norm": 1.5413780212402344,
      "learning_rate": 0.00016916253453997358,
      "loss": 1.6463,
      "step": 7796
    },
    {
      "epoch": 0.2792271742439164,
      "grad_norm": 1.8122531175613403,
      "learning_rate": 0.00016915415661282335,
      "loss": 1.5876,
      "step": 7797
    },
    {
      "epoch": 0.2792629863734847,
      "grad_norm": 2.2389962673187256,
      "learning_rate": 0.00016914577775529316,
      "loss": 1.3309,
      "step": 7798
    },
    {
      "epoch": 0.279298798503053,
      "grad_norm": 1.547379732131958,
      "learning_rate": 0.0001691373979674957,
      "loss": 1.9148,
      "step": 7799
    },
    {
      "epoch": 0.27933461063262127,
      "grad_norm": 1.9751660823822021,
      "learning_rate": 0.00016912901724954377,
      "loss": 1.7473,
      "step": 7800
    },
    {
      "epoch": 0.27937042276218954,
      "grad_norm": 1.5748672485351562,
      "learning_rate": 0.00016912063560155005,
      "loss": 1.5901,
      "step": 7801
    },
    {
      "epoch": 0.27940623489175787,
      "grad_norm": 2.0997776985168457,
      "learning_rate": 0.00016911225302362738,
      "loss": 1.5381,
      "step": 7802
    },
    {
      "epoch": 0.27944204702132613,
      "grad_norm": 2.0922765731811523,
      "learning_rate": 0.00016910386951588845,
      "loss": 1.7502,
      "step": 7803
    },
    {
      "epoch": 0.2794778591508944,
      "grad_norm": 1.5022773742675781,
      "learning_rate": 0.0001690954850784461,
      "loss": 1.9076,
      "step": 7804
    },
    {
      "epoch": 0.27951367128046267,
      "grad_norm": 1.3663434982299805,
      "learning_rate": 0.00016908709971141312,
      "loss": 1.5557,
      "step": 7805
    },
    {
      "epoch": 0.279549483410031,
      "grad_norm": 2.1710376739501953,
      "learning_rate": 0.00016907871341490235,
      "loss": 1.6564,
      "step": 7806
    },
    {
      "epoch": 0.27958529553959927,
      "grad_norm": 2.2670748233795166,
      "learning_rate": 0.00016907032618902661,
      "loss": 1.3921,
      "step": 7807
    },
    {
      "epoch": 0.27962110766916753,
      "grad_norm": 1.7037529945373535,
      "learning_rate": 0.00016906193803389868,
      "loss": 1.6198,
      "step": 7808
    },
    {
      "epoch": 0.27965691979873586,
      "grad_norm": 2.3128836154937744,
      "learning_rate": 0.00016905354894963147,
      "loss": 1.8036,
      "step": 7809
    },
    {
      "epoch": 0.27969273192830413,
      "grad_norm": 1.735877275466919,
      "learning_rate": 0.00016904515893633785,
      "loss": 1.5839,
      "step": 7810
    },
    {
      "epoch": 0.2797285440578724,
      "grad_norm": 1.5727331638336182,
      "learning_rate": 0.0001690367679941307,
      "loss": 1.4397,
      "step": 7811
    },
    {
      "epoch": 0.27976435618744067,
      "grad_norm": 1.5470871925354004,
      "learning_rate": 0.00016902837612312285,
      "loss": 1.8064,
      "step": 7812
    },
    {
      "epoch": 0.279800168317009,
      "grad_norm": 1.5642430782318115,
      "learning_rate": 0.00016901998332342726,
      "loss": 1.6186,
      "step": 7813
    },
    {
      "epoch": 0.27983598044657726,
      "grad_norm": 1.3714781999588013,
      "learning_rate": 0.00016901158959515682,
      "loss": 1.5868,
      "step": 7814
    },
    {
      "epoch": 0.27987179257614553,
      "grad_norm": 1.6559131145477295,
      "learning_rate": 0.00016900319493842446,
      "loss": 1.3897,
      "step": 7815
    },
    {
      "epoch": 0.27990760470571385,
      "grad_norm": 2.2328150272369385,
      "learning_rate": 0.00016899479935334307,
      "loss": 2.1561,
      "step": 7816
    },
    {
      "epoch": 0.2799434168352821,
      "grad_norm": 1.605811357498169,
      "learning_rate": 0.0001689864028400257,
      "loss": 1.535,
      "step": 7817
    },
    {
      "epoch": 0.2799792289648504,
      "grad_norm": 1.7465509176254272,
      "learning_rate": 0.00016897800539858527,
      "loss": 1.5555,
      "step": 7818
    },
    {
      "epoch": 0.28001504109441866,
      "grad_norm": 1.5020278692245483,
      "learning_rate": 0.00016896960702913476,
      "loss": 1.6375,
      "step": 7819
    },
    {
      "epoch": 0.280050853223987,
      "grad_norm": 1.255519151687622,
      "learning_rate": 0.00016896120773178712,
      "loss": 1.7075,
      "step": 7820
    },
    {
      "epoch": 0.28008666535355525,
      "grad_norm": 1.8907737731933594,
      "learning_rate": 0.00016895280750665542,
      "loss": 1.7681,
      "step": 7821
    },
    {
      "epoch": 0.2801224774831235,
      "grad_norm": 1.6590030193328857,
      "learning_rate": 0.0001689444063538526,
      "loss": 1.5601,
      "step": 7822
    },
    {
      "epoch": 0.2801582896126918,
      "grad_norm": 2.1344289779663086,
      "learning_rate": 0.00016893600427349173,
      "loss": 1.3434,
      "step": 7823
    },
    {
      "epoch": 0.2801941017422601,
      "grad_norm": 1.558802843093872,
      "learning_rate": 0.00016892760126568584,
      "loss": 1.4994,
      "step": 7824
    },
    {
      "epoch": 0.2802299138718284,
      "grad_norm": 1.822415828704834,
      "learning_rate": 0.00016891919733054802,
      "loss": 1.5621,
      "step": 7825
    },
    {
      "epoch": 0.28026572600139665,
      "grad_norm": 1.2544628381729126,
      "learning_rate": 0.00016891079246819128,
      "loss": 1.4667,
      "step": 7826
    },
    {
      "epoch": 0.280301538130965,
      "grad_norm": 1.3893107175827026,
      "learning_rate": 0.0001689023866787287,
      "loss": 1.6372,
      "step": 7827
    },
    {
      "epoch": 0.28033735026053325,
      "grad_norm": 1.3773599863052368,
      "learning_rate": 0.00016889397996227342,
      "loss": 1.5978,
      "step": 7828
    },
    {
      "epoch": 0.2803731623901015,
      "grad_norm": 1.4866869449615479,
      "learning_rate": 0.00016888557231893846,
      "loss": 1.5474,
      "step": 7829
    },
    {
      "epoch": 0.2804089745196698,
      "grad_norm": 1.25416100025177,
      "learning_rate": 0.00016887716374883703,
      "loss": 1.5429,
      "step": 7830
    },
    {
      "epoch": 0.2804447866492381,
      "grad_norm": 1.3411985635757446,
      "learning_rate": 0.0001688687542520822,
      "loss": 1.601,
      "step": 7831
    },
    {
      "epoch": 0.2804805987788064,
      "grad_norm": 1.2994372844696045,
      "learning_rate": 0.0001688603438287871,
      "loss": 1.4822,
      "step": 7832
    },
    {
      "epoch": 0.28051641090837465,
      "grad_norm": 2.017587900161743,
      "learning_rate": 0.00016885193247906488,
      "loss": 1.3413,
      "step": 7833
    },
    {
      "epoch": 0.28055222303794297,
      "grad_norm": 1.8447659015655518,
      "learning_rate": 0.00016884352020302875,
      "loss": 1.7149,
      "step": 7834
    },
    {
      "epoch": 0.28058803516751124,
      "grad_norm": 1.4341516494750977,
      "learning_rate": 0.00016883510700079182,
      "loss": 1.5375,
      "step": 7835
    },
    {
      "epoch": 0.2806238472970795,
      "grad_norm": 1.4541009664535522,
      "learning_rate": 0.00016882669287246734,
      "loss": 1.5869,
      "step": 7836
    },
    {
      "epoch": 0.2806596594266478,
      "grad_norm": 1.8895456790924072,
      "learning_rate": 0.0001688182778181685,
      "loss": 1.647,
      "step": 7837
    },
    {
      "epoch": 0.2806954715562161,
      "grad_norm": 2.3410048484802246,
      "learning_rate": 0.0001688098618380085,
      "loss": 1.3165,
      "step": 7838
    },
    {
      "epoch": 0.28073128368578437,
      "grad_norm": 1.4924992322921753,
      "learning_rate": 0.00016880144493210052,
      "loss": 1.551,
      "step": 7839
    },
    {
      "epoch": 0.28076709581535264,
      "grad_norm": 1.6559271812438965,
      "learning_rate": 0.00016879302710055792,
      "loss": 1.5487,
      "step": 7840
    },
    {
      "epoch": 0.28080290794492097,
      "grad_norm": 1.4658722877502441,
      "learning_rate": 0.0001687846083434938,
      "loss": 1.7468,
      "step": 7841
    },
    {
      "epoch": 0.28083872007448923,
      "grad_norm": 1.6023316383361816,
      "learning_rate": 0.00016877618866102155,
      "loss": 1.6474,
      "step": 7842
    },
    {
      "epoch": 0.2808745322040575,
      "grad_norm": 2.7606184482574463,
      "learning_rate": 0.0001687677680532544,
      "loss": 1.3463,
      "step": 7843
    },
    {
      "epoch": 0.28091034433362577,
      "grad_norm": 2.0281307697296143,
      "learning_rate": 0.00016875934652030563,
      "loss": 1.5334,
      "step": 7844
    },
    {
      "epoch": 0.2809461564631941,
      "grad_norm": 1.7015283107757568,
      "learning_rate": 0.00016875092406228853,
      "loss": 1.4474,
      "step": 7845
    },
    {
      "epoch": 0.28098196859276237,
      "grad_norm": 2.321729898452759,
      "learning_rate": 0.00016874250067931644,
      "loss": 1.7248,
      "step": 7846
    },
    {
      "epoch": 0.28101778072233063,
      "grad_norm": 1.3899301290512085,
      "learning_rate": 0.00016873407637150268,
      "loss": 1.5622,
      "step": 7847
    },
    {
      "epoch": 0.28105359285189896,
      "grad_norm": 2.0998470783233643,
      "learning_rate": 0.00016872565113896056,
      "loss": 1.318,
      "step": 7848
    },
    {
      "epoch": 0.28108940498146723,
      "grad_norm": 1.7129113674163818,
      "learning_rate": 0.00016871722498180346,
      "loss": 1.6792,
      "step": 7849
    },
    {
      "epoch": 0.2811252171110355,
      "grad_norm": 1.6951273679733276,
      "learning_rate": 0.00016870879790014474,
      "loss": 1.506,
      "step": 7850
    },
    {
      "epoch": 0.28116102924060377,
      "grad_norm": 2.1859934329986572,
      "learning_rate": 0.00016870036989409778,
      "loss": 1.6351,
      "step": 7851
    },
    {
      "epoch": 0.2811968413701721,
      "grad_norm": 2.049006938934326,
      "learning_rate": 0.00016869194096377597,
      "loss": 1.5945,
      "step": 7852
    },
    {
      "epoch": 0.28123265349974036,
      "grad_norm": 1.6370213031768799,
      "learning_rate": 0.00016868351110929268,
      "loss": 1.6024,
      "step": 7853
    },
    {
      "epoch": 0.28126846562930863,
      "grad_norm": 1.693914532661438,
      "learning_rate": 0.00016867508033076135,
      "loss": 1.2703,
      "step": 7854
    },
    {
      "epoch": 0.28130427775887695,
      "grad_norm": 1.6464945077896118,
      "learning_rate": 0.00016866664862829543,
      "loss": 1.5552,
      "step": 7855
    },
    {
      "epoch": 0.2813400898884452,
      "grad_norm": 1.5109620094299316,
      "learning_rate": 0.00016865821600200827,
      "loss": 1.7248,
      "step": 7856
    },
    {
      "epoch": 0.2813759020180135,
      "grad_norm": 1.7957203388214111,
      "learning_rate": 0.0001686497824520134,
      "loss": 1.8003,
      "step": 7857
    },
    {
      "epoch": 0.28141171414758176,
      "grad_norm": 1.7840375900268555,
      "learning_rate": 0.00016864134797842426,
      "loss": 1.4401,
      "step": 7858
    },
    {
      "epoch": 0.2814475262771501,
      "grad_norm": 1.9683198928833008,
      "learning_rate": 0.00016863291258135434,
      "loss": 1.4671,
      "step": 7859
    },
    {
      "epoch": 0.28148333840671835,
      "grad_norm": 1.9993942975997925,
      "learning_rate": 0.00016862447626091707,
      "loss": 1.7088,
      "step": 7860
    },
    {
      "epoch": 0.2815191505362866,
      "grad_norm": 1.76691472530365,
      "learning_rate": 0.00016861603901722601,
      "loss": 1.3172,
      "step": 7861
    },
    {
      "epoch": 0.28155496266585495,
      "grad_norm": 1.4511359930038452,
      "learning_rate": 0.00016860760085039467,
      "loss": 1.5278,
      "step": 7862
    },
    {
      "epoch": 0.2815907747954232,
      "grad_norm": 1.7963536977767944,
      "learning_rate": 0.00016859916176053657,
      "loss": 1.6062,
      "step": 7863
    },
    {
      "epoch": 0.2816265869249915,
      "grad_norm": 1.900141954421997,
      "learning_rate": 0.00016859072174776522,
      "loss": 1.3176,
      "step": 7864
    },
    {
      "epoch": 0.28166239905455975,
      "grad_norm": 1.7314437627792358,
      "learning_rate": 0.00016858228081219416,
      "loss": 1.5749,
      "step": 7865
    },
    {
      "epoch": 0.2816982111841281,
      "grad_norm": 2.220306396484375,
      "learning_rate": 0.000168573838953937,
      "loss": 1.5796,
      "step": 7866
    },
    {
      "epoch": 0.28173402331369635,
      "grad_norm": 1.7966647148132324,
      "learning_rate": 0.00016856539617310728,
      "loss": 1.5018,
      "step": 7867
    },
    {
      "epoch": 0.2817698354432646,
      "grad_norm": 1.964325189590454,
      "learning_rate": 0.0001685569524698186,
      "loss": 1.7829,
      "step": 7868
    },
    {
      "epoch": 0.28180564757283294,
      "grad_norm": 1.6538739204406738,
      "learning_rate": 0.00016854850784418457,
      "loss": 1.8246,
      "step": 7869
    },
    {
      "epoch": 0.2818414597024012,
      "grad_norm": 1.633487343788147,
      "learning_rate": 0.00016854006229631877,
      "loss": 1.674,
      "step": 7870
    },
    {
      "epoch": 0.2818772718319695,
      "grad_norm": 2.0294923782348633,
      "learning_rate": 0.00016853161582633486,
      "loss": 1.8599,
      "step": 7871
    },
    {
      "epoch": 0.28191308396153775,
      "grad_norm": 1.7625004053115845,
      "learning_rate": 0.00016852316843434645,
      "loss": 1.2676,
      "step": 7872
    },
    {
      "epoch": 0.28194889609110607,
      "grad_norm": 1.4743856191635132,
      "learning_rate": 0.0001685147201204672,
      "loss": 1.8254,
      "step": 7873
    },
    {
      "epoch": 0.28198470822067434,
      "grad_norm": 1.7030346393585205,
      "learning_rate": 0.00016850627088481077,
      "loss": 1.4622,
      "step": 7874
    },
    {
      "epoch": 0.2820205203502426,
      "grad_norm": 1.8332545757293701,
      "learning_rate": 0.0001684978207274908,
      "loss": 1.6707,
      "step": 7875
    },
    {
      "epoch": 0.28205633247981093,
      "grad_norm": 1.432371973991394,
      "learning_rate": 0.00016848936964862106,
      "loss": 1.4167,
      "step": 7876
    },
    {
      "epoch": 0.2820921446093792,
      "grad_norm": 1.7265770435333252,
      "learning_rate": 0.00016848091764831518,
      "loss": 1.4469,
      "step": 7877
    },
    {
      "epoch": 0.28212795673894747,
      "grad_norm": 3.417351484298706,
      "learning_rate": 0.00016847246472668684,
      "loss": 1.5023,
      "step": 7878
    },
    {
      "epoch": 0.28216376886851574,
      "grad_norm": 1.2689472436904907,
      "learning_rate": 0.00016846401088384987,
      "loss": 1.5067,
      "step": 7879
    },
    {
      "epoch": 0.28219958099808407,
      "grad_norm": 2.011307954788208,
      "learning_rate": 0.0001684555561199179,
      "loss": 1.605,
      "step": 7880
    },
    {
      "epoch": 0.28223539312765233,
      "grad_norm": 2.8057658672332764,
      "learning_rate": 0.00016844710043500478,
      "loss": 1.6136,
      "step": 7881
    },
    {
      "epoch": 0.2822712052572206,
      "grad_norm": 1.857756495475769,
      "learning_rate": 0.00016843864382922418,
      "loss": 1.6513,
      "step": 7882
    },
    {
      "epoch": 0.2823070173867889,
      "grad_norm": 1.6902923583984375,
      "learning_rate": 0.0001684301863026899,
      "loss": 1.29,
      "step": 7883
    },
    {
      "epoch": 0.2823428295163572,
      "grad_norm": 2.0259835720062256,
      "learning_rate": 0.00016842172785551572,
      "loss": 1.7171,
      "step": 7884
    },
    {
      "epoch": 0.28237864164592547,
      "grad_norm": 1.9104828834533691,
      "learning_rate": 0.00016841326848781546,
      "loss": 1.6705,
      "step": 7885
    },
    {
      "epoch": 0.28241445377549373,
      "grad_norm": 2.1479547023773193,
      "learning_rate": 0.00016840480819970294,
      "loss": 1.5749,
      "step": 7886
    },
    {
      "epoch": 0.28245026590506206,
      "grad_norm": 1.6750024557113647,
      "learning_rate": 0.00016839634699129197,
      "loss": 1.4534,
      "step": 7887
    },
    {
      "epoch": 0.28248607803463033,
      "grad_norm": 1.583740472793579,
      "learning_rate": 0.00016838788486269634,
      "loss": 1.3206,
      "step": 7888
    },
    {
      "epoch": 0.2825218901641986,
      "grad_norm": 1.3861204385757446,
      "learning_rate": 0.00016837942181402993,
      "loss": 1.3911,
      "step": 7889
    },
    {
      "epoch": 0.2825577022937669,
      "grad_norm": 1.6452444791793823,
      "learning_rate": 0.00016837095784540663,
      "loss": 1.5608,
      "step": 7890
    },
    {
      "epoch": 0.2825935144233352,
      "grad_norm": 1.4533544778823853,
      "learning_rate": 0.0001683624929569403,
      "loss": 1.7149,
      "step": 7891
    },
    {
      "epoch": 0.28262932655290346,
      "grad_norm": 1.6614084243774414,
      "learning_rate": 0.0001683540271487448,
      "loss": 1.4509,
      "step": 7892
    },
    {
      "epoch": 0.28266513868247173,
      "grad_norm": 1.9571036100387573,
      "learning_rate": 0.000168345560420934,
      "loss": 1.6858,
      "step": 7893
    },
    {
      "epoch": 0.28270095081204005,
      "grad_norm": 1.823486328125,
      "learning_rate": 0.00016833709277362186,
      "loss": 1.788,
      "step": 7894
    },
    {
      "epoch": 0.2827367629416083,
      "grad_norm": 1.6518898010253906,
      "learning_rate": 0.0001683286242069223,
      "loss": 1.4248,
      "step": 7895
    },
    {
      "epoch": 0.2827725750711766,
      "grad_norm": 1.626325011253357,
      "learning_rate": 0.00016832015472094923,
      "loss": 1.203,
      "step": 7896
    },
    {
      "epoch": 0.2828083872007449,
      "grad_norm": 1.368032693862915,
      "learning_rate": 0.0001683116843158166,
      "loss": 1.3924,
      "step": 7897
    },
    {
      "epoch": 0.2828441993303132,
      "grad_norm": 1.738690972328186,
      "learning_rate": 0.00016830321299163837,
      "loss": 1.7386,
      "step": 7898
    },
    {
      "epoch": 0.28288001145988145,
      "grad_norm": 1.9578567743301392,
      "learning_rate": 0.0001682947407485285,
      "loss": 1.4225,
      "step": 7899
    },
    {
      "epoch": 0.2829158235894497,
      "grad_norm": 2.169372320175171,
      "learning_rate": 0.00016828626758660104,
      "loss": 1.7834,
      "step": 7900
    },
    {
      "epoch": 0.28295163571901805,
      "grad_norm": 2.070221185684204,
      "learning_rate": 0.00016827779350596988,
      "loss": 1.6165,
      "step": 7901
    },
    {
      "epoch": 0.2829874478485863,
      "grad_norm": 1.5898553133010864,
      "learning_rate": 0.00016826931850674913,
      "loss": 1.7263,
      "step": 7902
    },
    {
      "epoch": 0.2830232599781546,
      "grad_norm": 1.5495221614837646,
      "learning_rate": 0.0001682608425890527,
      "loss": 1.4515,
      "step": 7903
    },
    {
      "epoch": 0.2830590721077229,
      "grad_norm": 1.3812370300292969,
      "learning_rate": 0.00016825236575299473,
      "loss": 1.3405,
      "step": 7904
    },
    {
      "epoch": 0.2830948842372912,
      "grad_norm": 1.676615595817566,
      "learning_rate": 0.0001682438879986892,
      "loss": 1.32,
      "step": 7905
    },
    {
      "epoch": 0.28313069636685945,
      "grad_norm": 1.6965440511703491,
      "learning_rate": 0.0001682354093262502,
      "loss": 1.5504,
      "step": 7906
    },
    {
      "epoch": 0.2831665084964277,
      "grad_norm": 1.978925108909607,
      "learning_rate": 0.00016822692973579177,
      "loss": 1.4036,
      "step": 7907
    },
    {
      "epoch": 0.28320232062599604,
      "grad_norm": 2.1448557376861572,
      "learning_rate": 0.000168218449227428,
      "loss": 1.7598,
      "step": 7908
    },
    {
      "epoch": 0.2832381327555643,
      "grad_norm": 1.9719595909118652,
      "learning_rate": 0.00016820996780127302,
      "loss": 1.3712,
      "step": 7909
    },
    {
      "epoch": 0.2832739448851326,
      "grad_norm": 1.57223379611969,
      "learning_rate": 0.00016820148545744089,
      "loss": 1.5271,
      "step": 7910
    },
    {
      "epoch": 0.2833097570147009,
      "grad_norm": 1.6346989870071411,
      "learning_rate": 0.00016819300219604572,
      "loss": 1.6543,
      "step": 7911
    },
    {
      "epoch": 0.28334556914426917,
      "grad_norm": 1.9346641302108765,
      "learning_rate": 0.00016818451801720169,
      "loss": 1.643,
      "step": 7912
    },
    {
      "epoch": 0.28338138127383744,
      "grad_norm": 1.87030827999115,
      "learning_rate": 0.00016817603292102292,
      "loss": 1.7069,
      "step": 7913
    },
    {
      "epoch": 0.2834171934034057,
      "grad_norm": 1.740334391593933,
      "learning_rate": 0.00016816754690762356,
      "loss": 1.7116,
      "step": 7914
    },
    {
      "epoch": 0.28345300553297403,
      "grad_norm": 2.0844287872314453,
      "learning_rate": 0.0001681590599771178,
      "loss": 1.5254,
      "step": 7915
    },
    {
      "epoch": 0.2834888176625423,
      "grad_norm": 1.4731221199035645,
      "learning_rate": 0.00016815057212961985,
      "loss": 1.5977,
      "step": 7916
    },
    {
      "epoch": 0.28352462979211057,
      "grad_norm": 1.638742208480835,
      "learning_rate": 0.0001681420833652438,
      "loss": 1.5354,
      "step": 7917
    },
    {
      "epoch": 0.2835604419216789,
      "grad_norm": 1.5350745916366577,
      "learning_rate": 0.00016813359368410394,
      "loss": 1.6585,
      "step": 7918
    },
    {
      "epoch": 0.28359625405124717,
      "grad_norm": 1.2498204708099365,
      "learning_rate": 0.00016812510308631445,
      "loss": 1.312,
      "step": 7919
    },
    {
      "epoch": 0.28363206618081543,
      "grad_norm": 1.752829670906067,
      "learning_rate": 0.00016811661157198956,
      "loss": 1.3958,
      "step": 7920
    },
    {
      "epoch": 0.2836678783103837,
      "grad_norm": 1.6889894008636475,
      "learning_rate": 0.00016810811914124354,
      "loss": 1.5042,
      "step": 7921
    },
    {
      "epoch": 0.283703690439952,
      "grad_norm": 2.0191566944122314,
      "learning_rate": 0.00016809962579419064,
      "loss": 1.523,
      "step": 7922
    },
    {
      "epoch": 0.2837395025695203,
      "grad_norm": 2.0825445652008057,
      "learning_rate": 0.0001680911315309451,
      "loss": 1.5253,
      "step": 7923
    },
    {
      "epoch": 0.28377531469908857,
      "grad_norm": 1.8054052591323853,
      "learning_rate": 0.00016808263635162123,
      "loss": 1.7837,
      "step": 7924
    },
    {
      "epoch": 0.2838111268286569,
      "grad_norm": 1.8791615962982178,
      "learning_rate": 0.0001680741402563333,
      "loss": 1.2332,
      "step": 7925
    },
    {
      "epoch": 0.28384693895822516,
      "grad_norm": 1.7158640623092651,
      "learning_rate": 0.00016806564324519565,
      "loss": 1.5696,
      "step": 7926
    },
    {
      "epoch": 0.28388275108779343,
      "grad_norm": 1.4893388748168945,
      "learning_rate": 0.00016805714531832253,
      "loss": 1.5782,
      "step": 7927
    },
    {
      "epoch": 0.2839185632173617,
      "grad_norm": 2.08494234085083,
      "learning_rate": 0.00016804864647582832,
      "loss": 1.8085,
      "step": 7928
    },
    {
      "epoch": 0.28395437534693,
      "grad_norm": 1.590214729309082,
      "learning_rate": 0.00016804014671782736,
      "loss": 1.7791,
      "step": 7929
    },
    {
      "epoch": 0.2839901874764983,
      "grad_norm": 1.7456499338150024,
      "learning_rate": 0.00016803164604443395,
      "loss": 1.2424,
      "step": 7930
    },
    {
      "epoch": 0.28402599960606656,
      "grad_norm": 2.2353994846343994,
      "learning_rate": 0.00016802314445576254,
      "loss": 1.2298,
      "step": 7931
    },
    {
      "epoch": 0.2840618117356349,
      "grad_norm": 1.718406319618225,
      "learning_rate": 0.00016801464195192746,
      "loss": 1.6377,
      "step": 7932
    },
    {
      "epoch": 0.28409762386520315,
      "grad_norm": 1.8352092504501343,
      "learning_rate": 0.00016800613853304311,
      "loss": 1.3559,
      "step": 7933
    },
    {
      "epoch": 0.2841334359947714,
      "grad_norm": 1.6804618835449219,
      "learning_rate": 0.00016799763419922387,
      "loss": 1.8663,
      "step": 7934
    },
    {
      "epoch": 0.2841692481243397,
      "grad_norm": 1.5867995023727417,
      "learning_rate": 0.00016798912895058416,
      "loss": 1.4769,
      "step": 7935
    },
    {
      "epoch": 0.284205060253908,
      "grad_norm": 1.8360810279846191,
      "learning_rate": 0.00016798062278723845,
      "loss": 1.6387,
      "step": 7936
    },
    {
      "epoch": 0.2842408723834763,
      "grad_norm": 1.573351263999939,
      "learning_rate": 0.00016797211570930115,
      "loss": 1.6323,
      "step": 7937
    },
    {
      "epoch": 0.28427668451304455,
      "grad_norm": 1.9696694612503052,
      "learning_rate": 0.0001679636077168867,
      "loss": 1.8065,
      "step": 7938
    },
    {
      "epoch": 0.2843124966426129,
      "grad_norm": 1.245781421661377,
      "learning_rate": 0.00016795509881010955,
      "loss": 1.4907,
      "step": 7939
    },
    {
      "epoch": 0.28434830877218115,
      "grad_norm": 1.5631426572799683,
      "learning_rate": 0.00016794658898908424,
      "loss": 1.5878,
      "step": 7940
    },
    {
      "epoch": 0.2843841209017494,
      "grad_norm": 2.6311140060424805,
      "learning_rate": 0.00016793807825392517,
      "loss": 1.6975,
      "step": 7941
    },
    {
      "epoch": 0.2844199330313177,
      "grad_norm": 1.682794213294983,
      "learning_rate": 0.00016792956660474694,
      "loss": 1.7496,
      "step": 7942
    },
    {
      "epoch": 0.284455745160886,
      "grad_norm": 1.3867418766021729,
      "learning_rate": 0.00016792105404166404,
      "loss": 1.6555,
      "step": 7943
    },
    {
      "epoch": 0.2844915572904543,
      "grad_norm": 2.540458917617798,
      "learning_rate": 0.00016791254056479092,
      "loss": 1.2869,
      "step": 7944
    },
    {
      "epoch": 0.28452736942002255,
      "grad_norm": 1.879603624343872,
      "learning_rate": 0.00016790402617424216,
      "loss": 1.645,
      "step": 7945
    },
    {
      "epoch": 0.28456318154959087,
      "grad_norm": 1.6672354936599731,
      "learning_rate": 0.00016789551087013232,
      "loss": 1.6987,
      "step": 7946
    },
    {
      "epoch": 0.28459899367915914,
      "grad_norm": 1.5219227075576782,
      "learning_rate": 0.00016788699465257597,
      "loss": 1.6638,
      "step": 7947
    },
    {
      "epoch": 0.2846348058087274,
      "grad_norm": 1.9220548868179321,
      "learning_rate": 0.00016787847752168769,
      "loss": 1.3491,
      "step": 7948
    },
    {
      "epoch": 0.2846706179382957,
      "grad_norm": 1.5979007482528687,
      "learning_rate": 0.00016786995947758204,
      "loss": 1.5428,
      "step": 7949
    },
    {
      "epoch": 0.284706430067864,
      "grad_norm": 1.6197651624679565,
      "learning_rate": 0.00016786144052037365,
      "loss": 1.416,
      "step": 7950
    },
    {
      "epoch": 0.28474224219743227,
      "grad_norm": 1.597205400466919,
      "learning_rate": 0.00016785292065017707,
      "loss": 1.3962,
      "step": 7951
    },
    {
      "epoch": 0.28477805432700054,
      "grad_norm": 2.123845100402832,
      "learning_rate": 0.000167844399867107,
      "loss": 1.3666,
      "step": 7952
    },
    {
      "epoch": 0.28481386645656886,
      "grad_norm": 1.9097387790679932,
      "learning_rate": 0.00016783587817127804,
      "loss": 1.3416,
      "step": 7953
    },
    {
      "epoch": 0.28484967858613713,
      "grad_norm": 1.6967304944992065,
      "learning_rate": 0.00016782735556280484,
      "loss": 1.6285,
      "step": 7954
    },
    {
      "epoch": 0.2848854907157054,
      "grad_norm": 1.3434966802597046,
      "learning_rate": 0.00016781883204180207,
      "loss": 1.5585,
      "step": 7955
    },
    {
      "epoch": 0.28492130284527367,
      "grad_norm": 2.0665807723999023,
      "learning_rate": 0.00016781030760838436,
      "loss": 1.9691,
      "step": 7956
    },
    {
      "epoch": 0.284957114974842,
      "grad_norm": 1.8598235845565796,
      "learning_rate": 0.00016780178226266646,
      "loss": 1.6914,
      "step": 7957
    },
    {
      "epoch": 0.28499292710441027,
      "grad_norm": 1.6443570852279663,
      "learning_rate": 0.00016779325600476303,
      "loss": 1.7151,
      "step": 7958
    },
    {
      "epoch": 0.28502873923397853,
      "grad_norm": 2.839176893234253,
      "learning_rate": 0.00016778472883478878,
      "loss": 1.4785,
      "step": 7959
    },
    {
      "epoch": 0.28506455136354686,
      "grad_norm": 1.6859476566314697,
      "learning_rate": 0.00016777620075285847,
      "loss": 1.3766,
      "step": 7960
    },
    {
      "epoch": 0.2851003634931151,
      "grad_norm": 2.1881790161132812,
      "learning_rate": 0.00016776767175908676,
      "loss": 1.8795,
      "step": 7961
    },
    {
      "epoch": 0.2851361756226834,
      "grad_norm": 1.856256127357483,
      "learning_rate": 0.00016775914185358846,
      "loss": 1.7113,
      "step": 7962
    },
    {
      "epoch": 0.28517198775225167,
      "grad_norm": 2.153108596801758,
      "learning_rate": 0.00016775061103647834,
      "loss": 1.6182,
      "step": 7963
    },
    {
      "epoch": 0.28520779988182,
      "grad_norm": 1.3774833679199219,
      "learning_rate": 0.00016774207930787108,
      "loss": 1.5698,
      "step": 7964
    },
    {
      "epoch": 0.28524361201138826,
      "grad_norm": 2.905151844024658,
      "learning_rate": 0.00016773354666788155,
      "loss": 1.2458,
      "step": 7965
    },
    {
      "epoch": 0.28527942414095653,
      "grad_norm": 1.9551284313201904,
      "learning_rate": 0.00016772501311662454,
      "loss": 1.6533,
      "step": 7966
    },
    {
      "epoch": 0.28531523627052485,
      "grad_norm": 1.6111496686935425,
      "learning_rate": 0.00016771647865421483,
      "loss": 1.5987,
      "step": 7967
    },
    {
      "epoch": 0.2853510484000931,
      "grad_norm": 1.438314437866211,
      "learning_rate": 0.00016770794328076726,
      "loss": 1.626,
      "step": 7968
    },
    {
      "epoch": 0.2853868605296614,
      "grad_norm": 1.7360272407531738,
      "learning_rate": 0.00016769940699639662,
      "loss": 1.5362,
      "step": 7969
    },
    {
      "epoch": 0.28542267265922966,
      "grad_norm": 1.3850613832473755,
      "learning_rate": 0.0001676908698012178,
      "loss": 1.7958,
      "step": 7970
    },
    {
      "epoch": 0.285458484788798,
      "grad_norm": 1.258893370628357,
      "learning_rate": 0.0001676823316953456,
      "loss": 1.0779,
      "step": 7971
    },
    {
      "epoch": 0.28549429691836625,
      "grad_norm": 1.8675183057785034,
      "learning_rate": 0.00016767379267889498,
      "loss": 1.5385,
      "step": 7972
    },
    {
      "epoch": 0.2855301090479345,
      "grad_norm": 1.3767646551132202,
      "learning_rate": 0.00016766525275198078,
      "loss": 1.6422,
      "step": 7973
    },
    {
      "epoch": 0.28556592117750285,
      "grad_norm": 1.6324293613433838,
      "learning_rate": 0.00016765671191471785,
      "loss": 1.5077,
      "step": 7974
    },
    {
      "epoch": 0.2856017333070711,
      "grad_norm": 1.361929178237915,
      "learning_rate": 0.00016764817016722114,
      "loss": 1.6237,
      "step": 7975
    },
    {
      "epoch": 0.2856375454366394,
      "grad_norm": 1.4396480321884155,
      "learning_rate": 0.00016763962750960558,
      "loss": 1.6242,
      "step": 7976
    },
    {
      "epoch": 0.28567335756620765,
      "grad_norm": 1.8851977586746216,
      "learning_rate": 0.00016763108394198605,
      "loss": 1.7705,
      "step": 7977
    },
    {
      "epoch": 0.285709169695776,
      "grad_norm": 1.8253602981567383,
      "learning_rate": 0.00016762253946447757,
      "loss": 1.6693,
      "step": 7978
    },
    {
      "epoch": 0.28574498182534425,
      "grad_norm": 1.959054946899414,
      "learning_rate": 0.000167613994077195,
      "loss": 1.499,
      "step": 7979
    },
    {
      "epoch": 0.2857807939549125,
      "grad_norm": 1.7831599712371826,
      "learning_rate": 0.00016760544778025337,
      "loss": 1.8024,
      "step": 7980
    },
    {
      "epoch": 0.28581660608448084,
      "grad_norm": 2.5011425018310547,
      "learning_rate": 0.00016759690057376769,
      "loss": 1.7597,
      "step": 7981
    },
    {
      "epoch": 0.2858524182140491,
      "grad_norm": 1.647387146949768,
      "learning_rate": 0.00016758835245785284,
      "loss": 1.6501,
      "step": 7982
    },
    {
      "epoch": 0.2858882303436174,
      "grad_norm": 1.6591894626617432,
      "learning_rate": 0.00016757980343262393,
      "loss": 1.8044,
      "step": 7983
    },
    {
      "epoch": 0.28592404247318565,
      "grad_norm": 1.6702752113342285,
      "learning_rate": 0.00016757125349819592,
      "loss": 1.8583,
      "step": 7984
    },
    {
      "epoch": 0.28595985460275397,
      "grad_norm": 1.6072109937667847,
      "learning_rate": 0.00016756270265468385,
      "loss": 1.392,
      "step": 7985
    },
    {
      "epoch": 0.28599566673232224,
      "grad_norm": 1.9452677965164185,
      "learning_rate": 0.00016755415090220278,
      "loss": 1.5837,
      "step": 7986
    },
    {
      "epoch": 0.2860314788618905,
      "grad_norm": 1.7112654447555542,
      "learning_rate": 0.00016754559824086774,
      "loss": 1.3601,
      "step": 7987
    },
    {
      "epoch": 0.28606729099145883,
      "grad_norm": 1.639289140701294,
      "learning_rate": 0.00016753704467079383,
      "loss": 1.7134,
      "step": 7988
    },
    {
      "epoch": 0.2861031031210271,
      "grad_norm": 1.4900970458984375,
      "learning_rate": 0.00016752849019209607,
      "loss": 1.4638,
      "step": 7989
    },
    {
      "epoch": 0.28613891525059537,
      "grad_norm": 1.5364689826965332,
      "learning_rate": 0.00016751993480488956,
      "loss": 1.6713,
      "step": 7990
    },
    {
      "epoch": 0.28617472738016364,
      "grad_norm": 1.6079754829406738,
      "learning_rate": 0.0001675113785092895,
      "loss": 1.8136,
      "step": 7991
    },
    {
      "epoch": 0.28621053950973196,
      "grad_norm": 1.4302982091903687,
      "learning_rate": 0.00016750282130541084,
      "loss": 1.4623,
      "step": 7992
    },
    {
      "epoch": 0.28624635163930023,
      "grad_norm": 1.5606993436813354,
      "learning_rate": 0.00016749426319336884,
      "loss": 1.5872,
      "step": 7993
    },
    {
      "epoch": 0.2862821637688685,
      "grad_norm": 1.3386200666427612,
      "learning_rate": 0.00016748570417327857,
      "loss": 1.5573,
      "step": 7994
    },
    {
      "epoch": 0.2863179758984368,
      "grad_norm": 2.041780710220337,
      "learning_rate": 0.0001674771442452552,
      "loss": 1.6269,
      "step": 7995
    },
    {
      "epoch": 0.2863537880280051,
      "grad_norm": 1.6531713008880615,
      "learning_rate": 0.0001674685834094139,
      "loss": 1.5946,
      "step": 7996
    },
    {
      "epoch": 0.28638960015757337,
      "grad_norm": 1.9872679710388184,
      "learning_rate": 0.00016746002166586984,
      "loss": 1.5185,
      "step": 7997
    },
    {
      "epoch": 0.28642541228714163,
      "grad_norm": 1.2475894689559937,
      "learning_rate": 0.00016745145901473819,
      "loss": 1.3904,
      "step": 7998
    },
    {
      "epoch": 0.28646122441670996,
      "grad_norm": 1.5508625507354736,
      "learning_rate": 0.0001674428954561342,
      "loss": 1.4844,
      "step": 7999
    },
    {
      "epoch": 0.2864970365462782,
      "grad_norm": 2.021813154220581,
      "learning_rate": 0.000167434330990173,
      "loss": 1.6274,
      "step": 8000
    },
    {
      "epoch": 0.2865328486758465,
      "grad_norm": 1.5985814332962036,
      "learning_rate": 0.0001674257656169699,
      "loss": 1.837,
      "step": 8001
    },
    {
      "epoch": 0.2865686608054148,
      "grad_norm": 1.310536503791809,
      "learning_rate": 0.00016741719933664008,
      "loss": 1.5341,
      "step": 8002
    },
    {
      "epoch": 0.2866044729349831,
      "grad_norm": 2.1644248962402344,
      "learning_rate": 0.00016740863214929883,
      "loss": 1.4358,
      "step": 8003
    },
    {
      "epoch": 0.28664028506455136,
      "grad_norm": 1.359682559967041,
      "learning_rate": 0.00016740006405506133,
      "loss": 1.4231,
      "step": 8004
    },
    {
      "epoch": 0.28667609719411963,
      "grad_norm": 1.2971854209899902,
      "learning_rate": 0.00016739149505404298,
      "loss": 1.2328,
      "step": 8005
    },
    {
      "epoch": 0.28671190932368795,
      "grad_norm": 1.5941914319992065,
      "learning_rate": 0.00016738292514635893,
      "loss": 1.3761,
      "step": 8006
    },
    {
      "epoch": 0.2867477214532562,
      "grad_norm": 1.3879280090332031,
      "learning_rate": 0.0001673743543321246,
      "loss": 1.5472,
      "step": 8007
    },
    {
      "epoch": 0.2867835335828245,
      "grad_norm": 2.025636911392212,
      "learning_rate": 0.00016736578261145518,
      "loss": 1.7866,
      "step": 8008
    },
    {
      "epoch": 0.2868193457123928,
      "grad_norm": 1.9905762672424316,
      "learning_rate": 0.00016735720998446607,
      "loss": 1.451,
      "step": 8009
    },
    {
      "epoch": 0.2868551578419611,
      "grad_norm": 2.0516388416290283,
      "learning_rate": 0.0001673486364512726,
      "loss": 1.5917,
      "step": 8010
    },
    {
      "epoch": 0.28689096997152935,
      "grad_norm": 1.2939916849136353,
      "learning_rate": 0.00016734006201199006,
      "loss": 1.2998,
      "step": 8011
    },
    {
      "epoch": 0.2869267821010976,
      "grad_norm": 1.6238223314285278,
      "learning_rate": 0.00016733148666673388,
      "loss": 1.3898,
      "step": 8012
    },
    {
      "epoch": 0.28696259423066595,
      "grad_norm": 2.0216140747070312,
      "learning_rate": 0.0001673229104156194,
      "loss": 1.5983,
      "step": 8013
    },
    {
      "epoch": 0.2869984063602342,
      "grad_norm": 1.5885149240493774,
      "learning_rate": 0.000167314333258762,
      "loss": 1.5381,
      "step": 8014
    },
    {
      "epoch": 0.2870342184898025,
      "grad_norm": 2.034376382827759,
      "learning_rate": 0.00016730575519627707,
      "loss": 1.442,
      "step": 8015
    },
    {
      "epoch": 0.2870700306193708,
      "grad_norm": 1.7644431591033936,
      "learning_rate": 0.00016729717622828002,
      "loss": 1.4961,
      "step": 8016
    },
    {
      "epoch": 0.2871058427489391,
      "grad_norm": 1.5184577703475952,
      "learning_rate": 0.00016728859635488626,
      "loss": 1.536,
      "step": 8017
    },
    {
      "epoch": 0.28714165487850735,
      "grad_norm": 1.647737741470337,
      "learning_rate": 0.00016728001557621126,
      "loss": 1.5551,
      "step": 8018
    },
    {
      "epoch": 0.2871774670080756,
      "grad_norm": 1.5250535011291504,
      "learning_rate": 0.00016727143389237042,
      "loss": 1.4027,
      "step": 8019
    },
    {
      "epoch": 0.28721327913764394,
      "grad_norm": 1.6826478242874146,
      "learning_rate": 0.0001672628513034792,
      "loss": 1.6263,
      "step": 8020
    },
    {
      "epoch": 0.2872490912672122,
      "grad_norm": 1.4399031400680542,
      "learning_rate": 0.0001672542678096531,
      "loss": 1.6192,
      "step": 8021
    },
    {
      "epoch": 0.2872849033967805,
      "grad_norm": 1.9906154870986938,
      "learning_rate": 0.00016724568341100758,
      "loss": 1.4139,
      "step": 8022
    },
    {
      "epoch": 0.28732071552634875,
      "grad_norm": 1.7506561279296875,
      "learning_rate": 0.0001672370981076581,
      "loss": 1.4831,
      "step": 8023
    },
    {
      "epoch": 0.28735652765591707,
      "grad_norm": 1.703379511833191,
      "learning_rate": 0.00016722851189972024,
      "loss": 1.5373,
      "step": 8024
    },
    {
      "epoch": 0.28739233978548534,
      "grad_norm": 1.5690869092941284,
      "learning_rate": 0.00016721992478730942,
      "loss": 1.2874,
      "step": 8025
    },
    {
      "epoch": 0.2874281519150536,
      "grad_norm": 1.3549939393997192,
      "learning_rate": 0.00016721133677054123,
      "loss": 1.3596,
      "step": 8026
    },
    {
      "epoch": 0.28746396404462193,
      "grad_norm": 1.2308684587478638,
      "learning_rate": 0.00016720274784953122,
      "loss": 1.1399,
      "step": 8027
    },
    {
      "epoch": 0.2874997761741902,
      "grad_norm": 1.5450139045715332,
      "learning_rate": 0.00016719415802439493,
      "loss": 1.5061,
      "step": 8028
    },
    {
      "epoch": 0.28753558830375847,
      "grad_norm": 1.6292673349380493,
      "learning_rate": 0.0001671855672952479,
      "loss": 1.4167,
      "step": 8029
    },
    {
      "epoch": 0.28757140043332674,
      "grad_norm": 2.0085721015930176,
      "learning_rate": 0.00016717697566220573,
      "loss": 1.6847,
      "step": 8030
    },
    {
      "epoch": 0.28760721256289506,
      "grad_norm": 1.688120722770691,
      "learning_rate": 0.00016716838312538402,
      "loss": 1.5717,
      "step": 8031
    },
    {
      "epoch": 0.28764302469246333,
      "grad_norm": 1.3048220872879028,
      "learning_rate": 0.00016715978968489834,
      "loss": 1.5839,
      "step": 8032
    },
    {
      "epoch": 0.2876788368220316,
      "grad_norm": 1.4469718933105469,
      "learning_rate": 0.0001671511953408643,
      "loss": 1.5844,
      "step": 8033
    },
    {
      "epoch": 0.2877146489515999,
      "grad_norm": 1.7788381576538086,
      "learning_rate": 0.0001671426000933976,
      "loss": 1.5919,
      "step": 8034
    },
    {
      "epoch": 0.2877504610811682,
      "grad_norm": 1.2445861101150513,
      "learning_rate": 0.00016713400394261378,
      "loss": 1.5703,
      "step": 8035
    },
    {
      "epoch": 0.28778627321073647,
      "grad_norm": 1.5097942352294922,
      "learning_rate": 0.00016712540688862854,
      "loss": 1.7577,
      "step": 8036
    },
    {
      "epoch": 0.28782208534030473,
      "grad_norm": 1.4741946458816528,
      "learning_rate": 0.0001671168089315575,
      "loss": 1.7221,
      "step": 8037
    },
    {
      "epoch": 0.28785789746987306,
      "grad_norm": 2.001183271408081,
      "learning_rate": 0.00016710821007151646,
      "loss": 1.6299,
      "step": 8038
    },
    {
      "epoch": 0.2878937095994413,
      "grad_norm": 2.835008144378662,
      "learning_rate": 0.00016709961030862092,
      "loss": 1.3872,
      "step": 8039
    },
    {
      "epoch": 0.2879295217290096,
      "grad_norm": 2.880967140197754,
      "learning_rate": 0.00016709100964298673,
      "loss": 1.097,
      "step": 8040
    },
    {
      "epoch": 0.2879653338585779,
      "grad_norm": 1.6155633926391602,
      "learning_rate": 0.00016708240807472956,
      "loss": 1.6969,
      "step": 8041
    },
    {
      "epoch": 0.2880011459881462,
      "grad_norm": 1.329376459121704,
      "learning_rate": 0.00016707380560396508,
      "loss": 1.7448,
      "step": 8042
    },
    {
      "epoch": 0.28803695811771446,
      "grad_norm": 2.0664658546447754,
      "learning_rate": 0.0001670652022308091,
      "loss": 1.4754,
      "step": 8043
    },
    {
      "epoch": 0.28807277024728273,
      "grad_norm": 1.5695208311080933,
      "learning_rate": 0.0001670565979553773,
      "loss": 1.6802,
      "step": 8044
    },
    {
      "epoch": 0.28810858237685105,
      "grad_norm": 1.879258155822754,
      "learning_rate": 0.0001670479927777855,
      "loss": 1.3501,
      "step": 8045
    },
    {
      "epoch": 0.2881443945064193,
      "grad_norm": 2.0454368591308594,
      "learning_rate": 0.0001670393866981494,
      "loss": 1.6627,
      "step": 8046
    },
    {
      "epoch": 0.2881802066359876,
      "grad_norm": 1.3214797973632812,
      "learning_rate": 0.00016703077971658487,
      "loss": 1.484,
      "step": 8047
    },
    {
      "epoch": 0.2882160187655559,
      "grad_norm": 1.7491061687469482,
      "learning_rate": 0.00016702217183320762,
      "loss": 1.7092,
      "step": 8048
    },
    {
      "epoch": 0.2882518308951242,
      "grad_norm": 1.3021421432495117,
      "learning_rate": 0.00016701356304813357,
      "loss": 1.4552,
      "step": 8049
    },
    {
      "epoch": 0.28828764302469245,
      "grad_norm": 1.3362562656402588,
      "learning_rate": 0.00016700495336147841,
      "loss": 1.4673,
      "step": 8050
    },
    {
      "epoch": 0.2883234551542607,
      "grad_norm": 1.692622184753418,
      "learning_rate": 0.00016699634277335805,
      "loss": 1.4147,
      "step": 8051
    },
    {
      "epoch": 0.28835926728382905,
      "grad_norm": 2.1186437606811523,
      "learning_rate": 0.00016698773128388832,
      "loss": 1.7805,
      "step": 8052
    },
    {
      "epoch": 0.2883950794133973,
      "grad_norm": 1.411199688911438,
      "learning_rate": 0.00016697911889318508,
      "loss": 1.4397,
      "step": 8053
    },
    {
      "epoch": 0.2884308915429656,
      "grad_norm": 1.3372883796691895,
      "learning_rate": 0.00016697050560136417,
      "loss": 1.3642,
      "step": 8054
    },
    {
      "epoch": 0.2884667036725339,
      "grad_norm": 1.3349575996398926,
      "learning_rate": 0.0001669618914085415,
      "loss": 1.6066,
      "step": 8055
    },
    {
      "epoch": 0.2885025158021022,
      "grad_norm": 1.5699543952941895,
      "learning_rate": 0.00016695327631483298,
      "loss": 1.9668,
      "step": 8056
    },
    {
      "epoch": 0.28853832793167045,
      "grad_norm": 1.399040699005127,
      "learning_rate": 0.00016694466032035447,
      "loss": 1.5224,
      "step": 8057
    },
    {
      "epoch": 0.2885741400612387,
      "grad_norm": 1.3568496704101562,
      "learning_rate": 0.0001669360434252219,
      "loss": 1.5744,
      "step": 8058
    },
    {
      "epoch": 0.28860995219080704,
      "grad_norm": 3.9023430347442627,
      "learning_rate": 0.00016692742562955123,
      "loss": 1.4893,
      "step": 8059
    },
    {
      "epoch": 0.2886457643203753,
      "grad_norm": 1.27892005443573,
      "learning_rate": 0.00016691880693345837,
      "loss": 1.5439,
      "step": 8060
    },
    {
      "epoch": 0.2886815764499436,
      "grad_norm": 1.1989108324050903,
      "learning_rate": 0.00016691018733705926,
      "loss": 1.4919,
      "step": 8061
    },
    {
      "epoch": 0.2887173885795119,
      "grad_norm": 1.627945899963379,
      "learning_rate": 0.00016690156684046991,
      "loss": 1.5184,
      "step": 8062
    },
    {
      "epoch": 0.28875320070908017,
      "grad_norm": 1.6738954782485962,
      "learning_rate": 0.00016689294544380628,
      "loss": 1.6135,
      "step": 8063
    },
    {
      "epoch": 0.28878901283864844,
      "grad_norm": 1.9493415355682373,
      "learning_rate": 0.00016688432314718434,
      "loss": 1.5187,
      "step": 8064
    },
    {
      "epoch": 0.2888248249682167,
      "grad_norm": 1.8097580671310425,
      "learning_rate": 0.0001668756999507201,
      "loss": 1.7367,
      "step": 8065
    },
    {
      "epoch": 0.28886063709778503,
      "grad_norm": 2.6163599491119385,
      "learning_rate": 0.00016686707585452962,
      "loss": 1.8102,
      "step": 8066
    },
    {
      "epoch": 0.2888964492273533,
      "grad_norm": 1.294272780418396,
      "learning_rate": 0.00016685845085872883,
      "loss": 1.2674,
      "step": 8067
    },
    {
      "epoch": 0.28893226135692157,
      "grad_norm": 1.7067070007324219,
      "learning_rate": 0.00016684982496343386,
      "loss": 1.9238,
      "step": 8068
    },
    {
      "epoch": 0.2889680734864899,
      "grad_norm": 1.9256243705749512,
      "learning_rate": 0.0001668411981687607,
      "loss": 1.2723,
      "step": 8069
    },
    {
      "epoch": 0.28900388561605816,
      "grad_norm": 1.8836833238601685,
      "learning_rate": 0.00016683257047482548,
      "loss": 1.9504,
      "step": 8070
    },
    {
      "epoch": 0.28903969774562643,
      "grad_norm": 1.2621315717697144,
      "learning_rate": 0.0001668239418817442,
      "loss": 1.5191,
      "step": 8071
    },
    {
      "epoch": 0.2890755098751947,
      "grad_norm": 1.624903678894043,
      "learning_rate": 0.000166815312389633,
      "loss": 1.3441,
      "step": 8072
    },
    {
      "epoch": 0.289111322004763,
      "grad_norm": 1.787150502204895,
      "learning_rate": 0.00016680668199860793,
      "loss": 1.7283,
      "step": 8073
    },
    {
      "epoch": 0.2891471341343313,
      "grad_norm": 1.5423107147216797,
      "learning_rate": 0.00016679805070878514,
      "loss": 1.5735,
      "step": 8074
    },
    {
      "epoch": 0.28918294626389957,
      "grad_norm": 1.5319451093673706,
      "learning_rate": 0.00016678941852028075,
      "loss": 1.754,
      "step": 8075
    },
    {
      "epoch": 0.2892187583934679,
      "grad_norm": 1.78415846824646,
      "learning_rate": 0.0001667807854332109,
      "loss": 1.6037,
      "step": 8076
    },
    {
      "epoch": 0.28925457052303616,
      "grad_norm": 1.5288525819778442,
      "learning_rate": 0.0001667721514476917,
      "loss": 1.7794,
      "step": 8077
    },
    {
      "epoch": 0.2892903826526044,
      "grad_norm": 1.83223295211792,
      "learning_rate": 0.0001667635165638393,
      "loss": 1.4731,
      "step": 8078
    },
    {
      "epoch": 0.2893261947821727,
      "grad_norm": 1.4526853561401367,
      "learning_rate": 0.00016675488078176994,
      "loss": 1.65,
      "step": 8079
    },
    {
      "epoch": 0.289362006911741,
      "grad_norm": 1.5872565507888794,
      "learning_rate": 0.00016674624410159978,
      "loss": 1.5926,
      "step": 8080
    },
    {
      "epoch": 0.2893978190413093,
      "grad_norm": 1.7032769918441772,
      "learning_rate": 0.000166737606523445,
      "loss": 1.8127,
      "step": 8081
    },
    {
      "epoch": 0.28943363117087756,
      "grad_norm": 1.5613534450531006,
      "learning_rate": 0.00016672896804742178,
      "loss": 1.8724,
      "step": 8082
    },
    {
      "epoch": 0.2894694433004459,
      "grad_norm": 1.4352047443389893,
      "learning_rate": 0.00016672032867364638,
      "loss": 1.6871,
      "step": 8083
    },
    {
      "epoch": 0.28950525543001415,
      "grad_norm": 1.5885288715362549,
      "learning_rate": 0.00016671168840223503,
      "loss": 1.5456,
      "step": 8084
    },
    {
      "epoch": 0.2895410675595824,
      "grad_norm": 1.3646939992904663,
      "learning_rate": 0.00016670304723330397,
      "loss": 1.3528,
      "step": 8085
    },
    {
      "epoch": 0.2895768796891507,
      "grad_norm": 2.0885651111602783,
      "learning_rate": 0.00016669440516696945,
      "loss": 1.3586,
      "step": 8086
    },
    {
      "epoch": 0.289612691818719,
      "grad_norm": 1.4972714185714722,
      "learning_rate": 0.0001666857622033477,
      "loss": 1.6482,
      "step": 8087
    },
    {
      "epoch": 0.2896485039482873,
      "grad_norm": 1.8946317434310913,
      "learning_rate": 0.00016667711834255505,
      "loss": 1.672,
      "step": 8088
    },
    {
      "epoch": 0.28968431607785555,
      "grad_norm": 1.687410593032837,
      "learning_rate": 0.0001666684735847078,
      "loss": 1.1767,
      "step": 8089
    },
    {
      "epoch": 0.2897201282074239,
      "grad_norm": 1.3974437713623047,
      "learning_rate": 0.00016665982792992226,
      "loss": 1.5356,
      "step": 8090
    },
    {
      "epoch": 0.28975594033699215,
      "grad_norm": 1.5038459300994873,
      "learning_rate": 0.00016665118137831468,
      "loss": 1.2653,
      "step": 8091
    },
    {
      "epoch": 0.2897917524665604,
      "grad_norm": 1.456130862236023,
      "learning_rate": 0.00016664253393000144,
      "loss": 1.8139,
      "step": 8092
    },
    {
      "epoch": 0.2898275645961287,
      "grad_norm": 1.6186647415161133,
      "learning_rate": 0.00016663388558509887,
      "loss": 1.5897,
      "step": 8093
    },
    {
      "epoch": 0.289863376725697,
      "grad_norm": 1.3481247425079346,
      "learning_rate": 0.00016662523634372334,
      "loss": 1.4714,
      "step": 8094
    },
    {
      "epoch": 0.2898991888552653,
      "grad_norm": 1.415844440460205,
      "learning_rate": 0.00016661658620599113,
      "loss": 1.6026,
      "step": 8095
    },
    {
      "epoch": 0.28993500098483355,
      "grad_norm": 2.165412187576294,
      "learning_rate": 0.00016660793517201875,
      "loss": 1.6079,
      "step": 8096
    },
    {
      "epoch": 0.28997081311440187,
      "grad_norm": 1.5296587944030762,
      "learning_rate": 0.00016659928324192248,
      "loss": 1.7825,
      "step": 8097
    },
    {
      "epoch": 0.29000662524397014,
      "grad_norm": 1.6164709329605103,
      "learning_rate": 0.0001665906304158188,
      "loss": 1.5188,
      "step": 8098
    },
    {
      "epoch": 0.2900424373735384,
      "grad_norm": 1.5287069082260132,
      "learning_rate": 0.00016658197669382405,
      "loss": 1.579,
      "step": 8099
    },
    {
      "epoch": 0.2900782495031067,
      "grad_norm": 2.6942615509033203,
      "learning_rate": 0.0001665733220760547,
      "loss": 1.3303,
      "step": 8100
    },
    {
      "epoch": 0.290114061632675,
      "grad_norm": 2.589672327041626,
      "learning_rate": 0.00016656466656262718,
      "loss": 1.4543,
      "step": 8101
    },
    {
      "epoch": 0.29014987376224327,
      "grad_norm": 1.4460633993148804,
      "learning_rate": 0.00016655601015365794,
      "loss": 1.6278,
      "step": 8102
    },
    {
      "epoch": 0.29018568589181154,
      "grad_norm": 1.2138534784317017,
      "learning_rate": 0.00016654735284926341,
      "loss": 1.57,
      "step": 8103
    },
    {
      "epoch": 0.29022149802137986,
      "grad_norm": 1.3130338191986084,
      "learning_rate": 0.00016653869464956008,
      "loss": 1.5317,
      "step": 8104
    },
    {
      "epoch": 0.29025731015094813,
      "grad_norm": 2.141824245452881,
      "learning_rate": 0.00016653003555466448,
      "loss": 1.3303,
      "step": 8105
    },
    {
      "epoch": 0.2902931222805164,
      "grad_norm": 1.5391119718551636,
      "learning_rate": 0.00016652137556469305,
      "loss": 1.3035,
      "step": 8106
    },
    {
      "epoch": 0.29032893441008467,
      "grad_norm": 2.6114346981048584,
      "learning_rate": 0.00016651271467976232,
      "loss": 1.6472,
      "step": 8107
    },
    {
      "epoch": 0.290364746539653,
      "grad_norm": 2.207442283630371,
      "learning_rate": 0.0001665040528999888,
      "loss": 1.5477,
      "step": 8108
    },
    {
      "epoch": 0.29040055866922126,
      "grad_norm": 1.4280242919921875,
      "learning_rate": 0.00016649539022548903,
      "loss": 1.5803,
      "step": 8109
    },
    {
      "epoch": 0.29043637079878953,
      "grad_norm": 1.6368767023086548,
      "learning_rate": 0.00016648672665637958,
      "loss": 1.4957,
      "step": 8110
    },
    {
      "epoch": 0.29047218292835786,
      "grad_norm": 1.8318397998809814,
      "learning_rate": 0.00016647806219277698,
      "loss": 1.4815,
      "step": 8111
    },
    {
      "epoch": 0.2905079950579261,
      "grad_norm": 2.205343723297119,
      "learning_rate": 0.0001664693968347978,
      "loss": 1.8598,
      "step": 8112
    },
    {
      "epoch": 0.2905438071874944,
      "grad_norm": 2.0023486614227295,
      "learning_rate": 0.00016646073058255862,
      "loss": 1.3708,
      "step": 8113
    },
    {
      "epoch": 0.29057961931706267,
      "grad_norm": 1.8614946603775024,
      "learning_rate": 0.00016645206343617603,
      "loss": 1.7015,
      "step": 8114
    },
    {
      "epoch": 0.290615431446631,
      "grad_norm": 1.6110183000564575,
      "learning_rate": 0.00016644339539576664,
      "loss": 1.5513,
      "step": 8115
    },
    {
      "epoch": 0.29065124357619926,
      "grad_norm": 1.5081207752227783,
      "learning_rate": 0.0001664347264614471,
      "loss": 1.6817,
      "step": 8116
    },
    {
      "epoch": 0.2906870557057675,
      "grad_norm": 1.4741617441177368,
      "learning_rate": 0.000166426056633334,
      "loss": 1.7121,
      "step": 8117
    },
    {
      "epoch": 0.29072286783533585,
      "grad_norm": 1.6981256008148193,
      "learning_rate": 0.00016641738591154396,
      "loss": 1.626,
      "step": 8118
    },
    {
      "epoch": 0.2907586799649041,
      "grad_norm": 1.4795587062835693,
      "learning_rate": 0.00016640871429619372,
      "loss": 1.5249,
      "step": 8119
    },
    {
      "epoch": 0.2907944920944724,
      "grad_norm": 1.8192358016967773,
      "learning_rate": 0.00016640004178739985,
      "loss": 1.5308,
      "step": 8120
    },
    {
      "epoch": 0.29083030422404066,
      "grad_norm": 1.5716443061828613,
      "learning_rate": 0.0001663913683852791,
      "loss": 1.1341,
      "step": 8121
    },
    {
      "epoch": 0.290866116353609,
      "grad_norm": 1.5312237739562988,
      "learning_rate": 0.00016638269408994808,
      "loss": 1.5524,
      "step": 8122
    },
    {
      "epoch": 0.29090192848317725,
      "grad_norm": 1.443957805633545,
      "learning_rate": 0.00016637401890152358,
      "loss": 1.4105,
      "step": 8123
    },
    {
      "epoch": 0.2909377406127455,
      "grad_norm": 1.9536917209625244,
      "learning_rate": 0.00016636534282012225,
      "loss": 1.4532,
      "step": 8124
    },
    {
      "epoch": 0.29097355274231385,
      "grad_norm": 2.228641986846924,
      "learning_rate": 0.00016635666584586083,
      "loss": 1.5409,
      "step": 8125
    },
    {
      "epoch": 0.2910093648718821,
      "grad_norm": 2.049088716506958,
      "learning_rate": 0.00016634798797885607,
      "loss": 1.6185,
      "step": 8126
    },
    {
      "epoch": 0.2910451770014504,
      "grad_norm": 1.800985336303711,
      "learning_rate": 0.00016633930921922474,
      "loss": 1.6347,
      "step": 8127
    },
    {
      "epoch": 0.29108098913101865,
      "grad_norm": 1.634249210357666,
      "learning_rate": 0.00016633062956708354,
      "loss": 1.4489,
      "step": 8128
    },
    {
      "epoch": 0.291116801260587,
      "grad_norm": 1.9593605995178223,
      "learning_rate": 0.0001663219490225493,
      "loss": 1.4183,
      "step": 8129
    },
    {
      "epoch": 0.29115261339015525,
      "grad_norm": 1.6916444301605225,
      "learning_rate": 0.0001663132675857388,
      "loss": 1.8659,
      "step": 8130
    },
    {
      "epoch": 0.2911884255197235,
      "grad_norm": 1.6571805477142334,
      "learning_rate": 0.0001663045852567688,
      "loss": 1.648,
      "step": 8131
    },
    {
      "epoch": 0.29122423764929184,
      "grad_norm": 1.6843384504318237,
      "learning_rate": 0.00016629590203575613,
      "loss": 1.2479,
      "step": 8132
    },
    {
      "epoch": 0.2912600497788601,
      "grad_norm": 1.5704199075698853,
      "learning_rate": 0.0001662872179228176,
      "loss": 1.6044,
      "step": 8133
    },
    {
      "epoch": 0.2912958619084284,
      "grad_norm": 2.2352945804595947,
      "learning_rate": 0.0001662785329180701,
      "loss": 1.6788,
      "step": 8134
    },
    {
      "epoch": 0.29133167403799665,
      "grad_norm": 1.5743358135223389,
      "learning_rate": 0.0001662698470216304,
      "loss": 1.4598,
      "step": 8135
    },
    {
      "epoch": 0.29136748616756497,
      "grad_norm": 1.7981449365615845,
      "learning_rate": 0.0001662611602336154,
      "loss": 1.5804,
      "step": 8136
    },
    {
      "epoch": 0.29140329829713324,
      "grad_norm": 1.3479498624801636,
      "learning_rate": 0.00016625247255414198,
      "loss": 1.7092,
      "step": 8137
    },
    {
      "epoch": 0.2914391104267015,
      "grad_norm": 1.1672223806381226,
      "learning_rate": 0.000166243783983327,
      "loss": 1.3365,
      "step": 8138
    },
    {
      "epoch": 0.29147492255626983,
      "grad_norm": 1.7496581077575684,
      "learning_rate": 0.00016623509452128732,
      "loss": 1.5851,
      "step": 8139
    },
    {
      "epoch": 0.2915107346858381,
      "grad_norm": 1.4864665269851685,
      "learning_rate": 0.00016622640416813988,
      "loss": 1.586,
      "step": 8140
    },
    {
      "epoch": 0.29154654681540637,
      "grad_norm": 1.4444739818572998,
      "learning_rate": 0.00016621771292400162,
      "loss": 1.5328,
      "step": 8141
    },
    {
      "epoch": 0.29158235894497464,
      "grad_norm": 1.6213339567184448,
      "learning_rate": 0.00016620902078898943,
      "loss": 1.2968,
      "step": 8142
    },
    {
      "epoch": 0.29161817107454296,
      "grad_norm": 1.5538610219955444,
      "learning_rate": 0.0001662003277632203,
      "loss": 1.5661,
      "step": 8143
    },
    {
      "epoch": 0.29165398320411123,
      "grad_norm": 2.134049654006958,
      "learning_rate": 0.0001661916338468111,
      "loss": 1.399,
      "step": 8144
    },
    {
      "epoch": 0.2916897953336795,
      "grad_norm": 2.0003087520599365,
      "learning_rate": 0.00016618293903987888,
      "loss": 1.6424,
      "step": 8145
    },
    {
      "epoch": 0.2917256074632478,
      "grad_norm": 2.1988790035247803,
      "learning_rate": 0.00016617424334254061,
      "loss": 1.5365,
      "step": 8146
    },
    {
      "epoch": 0.2917614195928161,
      "grad_norm": 1.5732612609863281,
      "learning_rate": 0.00016616554675491325,
      "loss": 1.8022,
      "step": 8147
    },
    {
      "epoch": 0.29179723172238436,
      "grad_norm": 1.4355159997940063,
      "learning_rate": 0.00016615684927711376,
      "loss": 1.4396,
      "step": 8148
    },
    {
      "epoch": 0.29183304385195263,
      "grad_norm": 1.6861153841018677,
      "learning_rate": 0.00016614815090925923,
      "loss": 1.5059,
      "step": 8149
    },
    {
      "epoch": 0.29186885598152096,
      "grad_norm": 1.6933960914611816,
      "learning_rate": 0.00016613945165146668,
      "loss": 1.6366,
      "step": 8150
    },
    {
      "epoch": 0.2919046681110892,
      "grad_norm": 1.724313735961914,
      "learning_rate": 0.00016613075150385308,
      "loss": 1.3966,
      "step": 8151
    },
    {
      "epoch": 0.2919404802406575,
      "grad_norm": 1.6139729022979736,
      "learning_rate": 0.00016612205046653554,
      "loss": 1.5069,
      "step": 8152
    },
    {
      "epoch": 0.2919762923702258,
      "grad_norm": 1.470168113708496,
      "learning_rate": 0.00016611334853963106,
      "loss": 1.2152,
      "step": 8153
    },
    {
      "epoch": 0.2920121044997941,
      "grad_norm": 1.4632880687713623,
      "learning_rate": 0.0001661046457232568,
      "loss": 1.2306,
      "step": 8154
    },
    {
      "epoch": 0.29204791662936236,
      "grad_norm": 1.752076268196106,
      "learning_rate": 0.00016609594201752982,
      "loss": 1.6839,
      "step": 8155
    },
    {
      "epoch": 0.2920837287589306,
      "grad_norm": 1.5564721822738647,
      "learning_rate": 0.00016608723742256719,
      "loss": 1.5811,
      "step": 8156
    },
    {
      "epoch": 0.29211954088849895,
      "grad_norm": 1.4910109043121338,
      "learning_rate": 0.00016607853193848597,
      "loss": 1.4991,
      "step": 8157
    },
    {
      "epoch": 0.2921553530180672,
      "grad_norm": 1.9942272901535034,
      "learning_rate": 0.0001660698255654034,
      "loss": 1.6296,
      "step": 8158
    },
    {
      "epoch": 0.2921911651476355,
      "grad_norm": 2.043497323989868,
      "learning_rate": 0.0001660611183034365,
      "loss": 1.7203,
      "step": 8159
    },
    {
      "epoch": 0.2922269772772038,
      "grad_norm": 1.740036129951477,
      "learning_rate": 0.00016605241015270247,
      "loss": 1.5992,
      "step": 8160
    },
    {
      "epoch": 0.2922627894067721,
      "grad_norm": 1.7537078857421875,
      "learning_rate": 0.0001660437011133185,
      "loss": 1.7468,
      "step": 8161
    },
    {
      "epoch": 0.29229860153634035,
      "grad_norm": 1.6161073446273804,
      "learning_rate": 0.0001660349911854017,
      "loss": 1.6065,
      "step": 8162
    },
    {
      "epoch": 0.2923344136659086,
      "grad_norm": 2.7794902324676514,
      "learning_rate": 0.0001660262803690693,
      "loss": 1.356,
      "step": 8163
    },
    {
      "epoch": 0.29237022579547695,
      "grad_norm": 1.438673496246338,
      "learning_rate": 0.00016601756866443845,
      "loss": 1.7374,
      "step": 8164
    },
    {
      "epoch": 0.2924060379250452,
      "grad_norm": 1.6174455881118774,
      "learning_rate": 0.00016600885607162636,
      "loss": 1.2237,
      "step": 8165
    },
    {
      "epoch": 0.2924418500546135,
      "grad_norm": 1.4131717681884766,
      "learning_rate": 0.00016600014259075024,
      "loss": 1.6027,
      "step": 8166
    },
    {
      "epoch": 0.2924776621841818,
      "grad_norm": 2.0395007133483887,
      "learning_rate": 0.00016599142822192736,
      "loss": 1.0541,
      "step": 8167
    },
    {
      "epoch": 0.2925134743137501,
      "grad_norm": 1.4359726905822754,
      "learning_rate": 0.00016598271296527494,
      "loss": 1.7906,
      "step": 8168
    },
    {
      "epoch": 0.29254928644331835,
      "grad_norm": 1.8879810571670532,
      "learning_rate": 0.00016597399682091024,
      "loss": 1.2775,
      "step": 8169
    },
    {
      "epoch": 0.2925850985728866,
      "grad_norm": 1.5420466661453247,
      "learning_rate": 0.00016596527978895046,
      "loss": 1.3398,
      "step": 8170
    },
    {
      "epoch": 0.29262091070245494,
      "grad_norm": 1.2738362550735474,
      "learning_rate": 0.00016595656186951297,
      "loss": 1.7642,
      "step": 8171
    },
    {
      "epoch": 0.2926567228320232,
      "grad_norm": 1.8173683881759644,
      "learning_rate": 0.00016594784306271502,
      "loss": 1.6041,
      "step": 8172
    },
    {
      "epoch": 0.2926925349615915,
      "grad_norm": 1.720488429069519,
      "learning_rate": 0.00016593912336867393,
      "loss": 1.3482,
      "step": 8173
    },
    {
      "epoch": 0.2927283470911598,
      "grad_norm": 2.0132124423980713,
      "learning_rate": 0.00016593040278750694,
      "loss": 1.5073,
      "step": 8174
    },
    {
      "epoch": 0.29276415922072807,
      "grad_norm": 1.5772103071212769,
      "learning_rate": 0.00016592168131933144,
      "loss": 1.519,
      "step": 8175
    },
    {
      "epoch": 0.29279997135029634,
      "grad_norm": 2.622490406036377,
      "learning_rate": 0.00016591295896426476,
      "loss": 1.2288,
      "step": 8176
    },
    {
      "epoch": 0.2928357834798646,
      "grad_norm": 1.7737879753112793,
      "learning_rate": 0.00016590423572242422,
      "loss": 1.5786,
      "step": 8177
    },
    {
      "epoch": 0.29287159560943293,
      "grad_norm": 1.4796202182769775,
      "learning_rate": 0.0001658955115939272,
      "loss": 1.3456,
      "step": 8178
    },
    {
      "epoch": 0.2929074077390012,
      "grad_norm": 2.105363368988037,
      "learning_rate": 0.00016588678657889112,
      "loss": 1.6187,
      "step": 8179
    },
    {
      "epoch": 0.29294321986856947,
      "grad_norm": 1.5410113334655762,
      "learning_rate": 0.00016587806067743327,
      "loss": 2.0069,
      "step": 8180
    },
    {
      "epoch": 0.2929790319981378,
      "grad_norm": 1.7439554929733276,
      "learning_rate": 0.00016586933388967109,
      "loss": 1.3726,
      "step": 8181
    },
    {
      "epoch": 0.29301484412770606,
      "grad_norm": 2.0335540771484375,
      "learning_rate": 0.000165860606215722,
      "loss": 1.5939,
      "step": 8182
    },
    {
      "epoch": 0.29305065625727433,
      "grad_norm": 1.8200868368148804,
      "learning_rate": 0.0001658518776557034,
      "loss": 1.2275,
      "step": 8183
    },
    {
      "epoch": 0.2930864683868426,
      "grad_norm": 1.4193452596664429,
      "learning_rate": 0.00016584314820973273,
      "loss": 1.4636,
      "step": 8184
    },
    {
      "epoch": 0.2931222805164109,
      "grad_norm": 1.3906633853912354,
      "learning_rate": 0.00016583441787792745,
      "loss": 1.3655,
      "step": 8185
    },
    {
      "epoch": 0.2931580926459792,
      "grad_norm": 1.5661295652389526,
      "learning_rate": 0.00016582568666040497,
      "loss": 1.8314,
      "step": 8186
    },
    {
      "epoch": 0.29319390477554746,
      "grad_norm": 1.2787010669708252,
      "learning_rate": 0.0001658169545572828,
      "loss": 1.4975,
      "step": 8187
    },
    {
      "epoch": 0.2932297169051158,
      "grad_norm": 1.9857454299926758,
      "learning_rate": 0.0001658082215686784,
      "loss": 1.3856,
      "step": 8188
    },
    {
      "epoch": 0.29326552903468406,
      "grad_norm": 1.7901902198791504,
      "learning_rate": 0.00016579948769470927,
      "loss": 1.8455,
      "step": 8189
    },
    {
      "epoch": 0.2933013411642523,
      "grad_norm": 1.7982251644134521,
      "learning_rate": 0.00016579075293549292,
      "loss": 1.3959,
      "step": 8190
    },
    {
      "epoch": 0.2933371532938206,
      "grad_norm": 1.3840776681900024,
      "learning_rate": 0.00016578201729114682,
      "loss": 1.5659,
      "step": 8191
    },
    {
      "epoch": 0.2933729654233889,
      "grad_norm": 1.91280996799469,
      "learning_rate": 0.00016577328076178855,
      "loss": 1.604,
      "step": 8192
    },
    {
      "epoch": 0.2934087775529572,
      "grad_norm": 1.5600473880767822,
      "learning_rate": 0.0001657645433475356,
      "loss": 1.3662,
      "step": 8193
    },
    {
      "epoch": 0.29344458968252546,
      "grad_norm": 1.603729248046875,
      "learning_rate": 0.0001657558050485056,
      "loss": 1.3253,
      "step": 8194
    },
    {
      "epoch": 0.2934804018120938,
      "grad_norm": 1.883900761604309,
      "learning_rate": 0.00016574706586481607,
      "loss": 1.2011,
      "step": 8195
    },
    {
      "epoch": 0.29351621394166205,
      "grad_norm": 2.2101593017578125,
      "learning_rate": 0.0001657383257965845,
      "loss": 1.7439,
      "step": 8196
    },
    {
      "epoch": 0.2935520260712303,
      "grad_norm": 2.143193006515503,
      "learning_rate": 0.0001657295848439286,
      "loss": 1.2733,
      "step": 8197
    },
    {
      "epoch": 0.2935878382007986,
      "grad_norm": 1.8261759281158447,
      "learning_rate": 0.00016572084300696594,
      "loss": 1.6156,
      "step": 8198
    },
    {
      "epoch": 0.2936236503303669,
      "grad_norm": 2.3701932430267334,
      "learning_rate": 0.0001657121002858141,
      "loss": 1.8107,
      "step": 8199
    },
    {
      "epoch": 0.2936594624599352,
      "grad_norm": 1.553908109664917,
      "learning_rate": 0.0001657033566805907,
      "loss": 1.2718,
      "step": 8200
    },
    {
      "epoch": 0.29369527458950345,
      "grad_norm": 1.8061621189117432,
      "learning_rate": 0.00016569461219141337,
      "loss": 1.8474,
      "step": 8201
    },
    {
      "epoch": 0.2937310867190718,
      "grad_norm": 3.017632484436035,
      "learning_rate": 0.00016568586681839982,
      "loss": 1.5226,
      "step": 8202
    },
    {
      "epoch": 0.29376689884864005,
      "grad_norm": 1.9157863855361938,
      "learning_rate": 0.00016567712056166762,
      "loss": 1.4055,
      "step": 8203
    },
    {
      "epoch": 0.2938027109782083,
      "grad_norm": 1.4582178592681885,
      "learning_rate": 0.0001656683734213345,
      "loss": 1.6516,
      "step": 8204
    },
    {
      "epoch": 0.2938385231077766,
      "grad_norm": 1.641868233680725,
      "learning_rate": 0.00016565962539751808,
      "loss": 1.6904,
      "step": 8205
    },
    {
      "epoch": 0.2938743352373449,
      "grad_norm": 1.7872939109802246,
      "learning_rate": 0.00016565087649033614,
      "loss": 1.3973,
      "step": 8206
    },
    {
      "epoch": 0.2939101473669132,
      "grad_norm": 1.8729043006896973,
      "learning_rate": 0.00016564212669990634,
      "loss": 1.4836,
      "step": 8207
    },
    {
      "epoch": 0.29394595949648145,
      "grad_norm": 1.4845998287200928,
      "learning_rate": 0.00016563337602634642,
      "loss": 1.6219,
      "step": 8208
    },
    {
      "epoch": 0.29398177162604977,
      "grad_norm": 1.3829447031021118,
      "learning_rate": 0.00016562462446977403,
      "loss": 1.5805,
      "step": 8209
    },
    {
      "epoch": 0.29401758375561804,
      "grad_norm": 1.5662988424301147,
      "learning_rate": 0.000165615872030307,
      "loss": 1.4469,
      "step": 8210
    },
    {
      "epoch": 0.2940533958851863,
      "grad_norm": 1.8096336126327515,
      "learning_rate": 0.00016560711870806303,
      "loss": 1.6272,
      "step": 8211
    },
    {
      "epoch": 0.2940892080147546,
      "grad_norm": 1.8783751726150513,
      "learning_rate": 0.00016559836450315992,
      "loss": 1.5174,
      "step": 8212
    },
    {
      "epoch": 0.2941250201443229,
      "grad_norm": 1.8585585355758667,
      "learning_rate": 0.00016558960941571543,
      "loss": 1.482,
      "step": 8213
    },
    {
      "epoch": 0.29416083227389117,
      "grad_norm": 1.624119758605957,
      "learning_rate": 0.00016558085344584736,
      "loss": 1.2782,
      "step": 8214
    },
    {
      "epoch": 0.29419664440345944,
      "grad_norm": 1.2839974164962769,
      "learning_rate": 0.00016557209659367347,
      "loss": 1.5674,
      "step": 8215
    },
    {
      "epoch": 0.29423245653302776,
      "grad_norm": 2.2805612087249756,
      "learning_rate": 0.00016556333885931162,
      "loss": 1.6619,
      "step": 8216
    },
    {
      "epoch": 0.29426826866259603,
      "grad_norm": 1.5599788427352905,
      "learning_rate": 0.00016555458024287964,
      "loss": 1.7332,
      "step": 8217
    },
    {
      "epoch": 0.2943040807921643,
      "grad_norm": 2.071613311767578,
      "learning_rate": 0.0001655458207444953,
      "loss": 1.7957,
      "step": 8218
    },
    {
      "epoch": 0.29433989292173257,
      "grad_norm": 1.492445945739746,
      "learning_rate": 0.0001655370603642765,
      "loss": 1.5455,
      "step": 8219
    },
    {
      "epoch": 0.2943757050513009,
      "grad_norm": 1.4632467031478882,
      "learning_rate": 0.0001655282991023411,
      "loss": 1.5157,
      "step": 8220
    },
    {
      "epoch": 0.29441151718086916,
      "grad_norm": 2.3645126819610596,
      "learning_rate": 0.000165519536958807,
      "loss": 1.7148,
      "step": 8221
    },
    {
      "epoch": 0.29444732931043743,
      "grad_norm": 1.6779077053070068,
      "learning_rate": 0.000165510773933792,
      "loss": 1.3695,
      "step": 8222
    },
    {
      "epoch": 0.2944831414400057,
      "grad_norm": 1.7853635549545288,
      "learning_rate": 0.00016550201002741403,
      "loss": 1.5484,
      "step": 8223
    },
    {
      "epoch": 0.294518953569574,
      "grad_norm": 2.639172077178955,
      "learning_rate": 0.00016549324523979102,
      "loss": 1.8244,
      "step": 8224
    },
    {
      "epoch": 0.2945547656991423,
      "grad_norm": 1.921919345855713,
      "learning_rate": 0.0001654844795710409,
      "loss": 1.7747,
      "step": 8225
    },
    {
      "epoch": 0.29459057782871056,
      "grad_norm": 1.9929426908493042,
      "learning_rate": 0.00016547571302128153,
      "loss": 1.6522,
      "step": 8226
    },
    {
      "epoch": 0.2946263899582789,
      "grad_norm": 1.6712838411331177,
      "learning_rate": 0.00016546694559063093,
      "loss": 1.5273,
      "step": 8227
    },
    {
      "epoch": 0.29466220208784716,
      "grad_norm": 1.7256261110305786,
      "learning_rate": 0.000165458177279207,
      "loss": 1.7917,
      "step": 8228
    },
    {
      "epoch": 0.2946980142174154,
      "grad_norm": 1.6912678480148315,
      "learning_rate": 0.00016544940808712775,
      "loss": 1.5481,
      "step": 8229
    },
    {
      "epoch": 0.2947338263469837,
      "grad_norm": 1.7609024047851562,
      "learning_rate": 0.00016544063801451114,
      "loss": 1.7015,
      "step": 8230
    },
    {
      "epoch": 0.294769638476552,
      "grad_norm": 2.914182662963867,
      "learning_rate": 0.00016543186706147514,
      "loss": 1.8822,
      "step": 8231
    },
    {
      "epoch": 0.2948054506061203,
      "grad_norm": 1.4983017444610596,
      "learning_rate": 0.00016542309522813779,
      "loss": 1.5277,
      "step": 8232
    },
    {
      "epoch": 0.29484126273568856,
      "grad_norm": 1.6739991903305054,
      "learning_rate": 0.00016541432251461705,
      "loss": 1.5439,
      "step": 8233
    },
    {
      "epoch": 0.2948770748652569,
      "grad_norm": 1.3391531705856323,
      "learning_rate": 0.000165405548921031,
      "loss": 1.3658,
      "step": 8234
    },
    {
      "epoch": 0.29491288699482515,
      "grad_norm": 1.941439151763916,
      "learning_rate": 0.0001653967744474977,
      "loss": 1.2919,
      "step": 8235
    },
    {
      "epoch": 0.2949486991243934,
      "grad_norm": 1.8408081531524658,
      "learning_rate": 0.00016538799909413508,
      "loss": 1.4787,
      "step": 8236
    },
    {
      "epoch": 0.2949845112539617,
      "grad_norm": 1.8048700094223022,
      "learning_rate": 0.00016537922286106134,
      "loss": 1.5778,
      "step": 8237
    },
    {
      "epoch": 0.29502032338353,
      "grad_norm": 2.9586312770843506,
      "learning_rate": 0.00016537044574839444,
      "loss": 1.5999,
      "step": 8238
    },
    {
      "epoch": 0.2950561355130983,
      "grad_norm": 1.534470796585083,
      "learning_rate": 0.00016536166775625252,
      "loss": 1.7927,
      "step": 8239
    },
    {
      "epoch": 0.29509194764266655,
      "grad_norm": 1.389212965965271,
      "learning_rate": 0.0001653528888847537,
      "loss": 1.4612,
      "step": 8240
    },
    {
      "epoch": 0.2951277597722349,
      "grad_norm": 1.5178078413009644,
      "learning_rate": 0.00016534410913401603,
      "loss": 1.491,
      "step": 8241
    },
    {
      "epoch": 0.29516357190180315,
      "grad_norm": 2.1112802028656006,
      "learning_rate": 0.0001653353285041577,
      "loss": 1.4394,
      "step": 8242
    },
    {
      "epoch": 0.2951993840313714,
      "grad_norm": 1.57144296169281,
      "learning_rate": 0.00016532654699529678,
      "loss": 1.7182,
      "step": 8243
    },
    {
      "epoch": 0.2952351961609397,
      "grad_norm": 1.8326722383499146,
      "learning_rate": 0.00016531776460755143,
      "loss": 1.4076,
      "step": 8244
    },
    {
      "epoch": 0.295271008290508,
      "grad_norm": 1.9306200742721558,
      "learning_rate": 0.0001653089813410398,
      "loss": 1.6902,
      "step": 8245
    },
    {
      "epoch": 0.2953068204200763,
      "grad_norm": 1.3576823472976685,
      "learning_rate": 0.00016530019719588007,
      "loss": 1.4347,
      "step": 8246
    },
    {
      "epoch": 0.29534263254964455,
      "grad_norm": 1.4128303527832031,
      "learning_rate": 0.00016529141217219045,
      "loss": 1.6012,
      "step": 8247
    },
    {
      "epoch": 0.29537844467921287,
      "grad_norm": 1.5558239221572876,
      "learning_rate": 0.00016528262627008906,
      "loss": 1.4238,
      "step": 8248
    },
    {
      "epoch": 0.29541425680878114,
      "grad_norm": 1.8013745546340942,
      "learning_rate": 0.00016527383948969416,
      "loss": 1.7537,
      "step": 8249
    },
    {
      "epoch": 0.2954500689383494,
      "grad_norm": 1.5508323907852173,
      "learning_rate": 0.00016526505183112394,
      "loss": 1.6048,
      "step": 8250
    },
    {
      "epoch": 0.2954858810679177,
      "grad_norm": 1.7320663928985596,
      "learning_rate": 0.00016525626329449668,
      "loss": 1.7845,
      "step": 8251
    },
    {
      "epoch": 0.295521693197486,
      "grad_norm": 1.934444785118103,
      "learning_rate": 0.0001652474738799305,
      "loss": 1.6607,
      "step": 8252
    },
    {
      "epoch": 0.29555750532705427,
      "grad_norm": 1.4806501865386963,
      "learning_rate": 0.00016523868358754378,
      "loss": 1.5327,
      "step": 8253
    },
    {
      "epoch": 0.29559331745662254,
      "grad_norm": 2.4541170597076416,
      "learning_rate": 0.00016522989241745469,
      "loss": 1.4625,
      "step": 8254
    },
    {
      "epoch": 0.29562912958619086,
      "grad_norm": 1.7268229722976685,
      "learning_rate": 0.00016522110036978153,
      "loss": 1.3799,
      "step": 8255
    },
    {
      "epoch": 0.29566494171575913,
      "grad_norm": 2.0476462841033936,
      "learning_rate": 0.0001652123074446426,
      "loss": 1.7806,
      "step": 8256
    },
    {
      "epoch": 0.2957007538453274,
      "grad_norm": 1.8949542045593262,
      "learning_rate": 0.00016520351364215623,
      "loss": 1.8599,
      "step": 8257
    },
    {
      "epoch": 0.29573656597489567,
      "grad_norm": 1.6571156978607178,
      "learning_rate": 0.00016519471896244063,
      "loss": 1.5826,
      "step": 8258
    },
    {
      "epoch": 0.295772378104464,
      "grad_norm": 1.580122947692871,
      "learning_rate": 0.00016518592340561422,
      "loss": 1.45,
      "step": 8259
    },
    {
      "epoch": 0.29580819023403226,
      "grad_norm": 1.4408222436904907,
      "learning_rate": 0.0001651771269717953,
      "loss": 1.3734,
      "step": 8260
    },
    {
      "epoch": 0.29584400236360053,
      "grad_norm": 1.6053831577301025,
      "learning_rate": 0.0001651683296611022,
      "loss": 1.7505,
      "step": 8261
    },
    {
      "epoch": 0.29587981449316886,
      "grad_norm": 1.855286717414856,
      "learning_rate": 0.0001651595314736533,
      "loss": 1.4902,
      "step": 8262
    },
    {
      "epoch": 0.2959156266227371,
      "grad_norm": 2.0580155849456787,
      "learning_rate": 0.00016515073240956692,
      "loss": 1.6515,
      "step": 8263
    },
    {
      "epoch": 0.2959514387523054,
      "grad_norm": 1.4917335510253906,
      "learning_rate": 0.0001651419324689615,
      "loss": 1.5483,
      "step": 8264
    },
    {
      "epoch": 0.29598725088187366,
      "grad_norm": 1.6483070850372314,
      "learning_rate": 0.00016513313165195538,
      "loss": 1.4299,
      "step": 8265
    },
    {
      "epoch": 0.296023063011442,
      "grad_norm": 1.8228284120559692,
      "learning_rate": 0.00016512432995866702,
      "loss": 1.4021,
      "step": 8266
    },
    {
      "epoch": 0.29605887514101026,
      "grad_norm": 1.4714165925979614,
      "learning_rate": 0.00016511552738921479,
      "loss": 1.7201,
      "step": 8267
    },
    {
      "epoch": 0.2960946872705785,
      "grad_norm": 1.59091317653656,
      "learning_rate": 0.0001651067239437171,
      "loss": 1.686,
      "step": 8268
    },
    {
      "epoch": 0.29613049940014685,
      "grad_norm": 1.409410834312439,
      "learning_rate": 0.00016509791962229247,
      "loss": 1.5451,
      "step": 8269
    },
    {
      "epoch": 0.2961663115297151,
      "grad_norm": 1.9544563293457031,
      "learning_rate": 0.0001650891144250593,
      "loss": 1.6566,
      "step": 8270
    },
    {
      "epoch": 0.2962021236592834,
      "grad_norm": 1.4585261344909668,
      "learning_rate": 0.00016508030835213605,
      "loss": 1.4464,
      "step": 8271
    },
    {
      "epoch": 0.29623793578885166,
      "grad_norm": 1.7657428979873657,
      "learning_rate": 0.00016507150140364116,
      "loss": 1.4297,
      "step": 8272
    },
    {
      "epoch": 0.29627374791842,
      "grad_norm": 1.381548285484314,
      "learning_rate": 0.0001650626935796932,
      "loss": 1.5166,
      "step": 8273
    },
    {
      "epoch": 0.29630956004798825,
      "grad_norm": 1.7373437881469727,
      "learning_rate": 0.00016505388488041058,
      "loss": 1.5385,
      "step": 8274
    },
    {
      "epoch": 0.2963453721775565,
      "grad_norm": 1.6844048500061035,
      "learning_rate": 0.0001650450753059119,
      "loss": 1.4545,
      "step": 8275
    },
    {
      "epoch": 0.29638118430712485,
      "grad_norm": 1.7241437435150146,
      "learning_rate": 0.00016503626485631561,
      "loss": 1.5535,
      "step": 8276
    },
    {
      "epoch": 0.2964169964366931,
      "grad_norm": 1.5121371746063232,
      "learning_rate": 0.00016502745353174026,
      "loss": 1.7588,
      "step": 8277
    },
    {
      "epoch": 0.2964528085662614,
      "grad_norm": 2.0365071296691895,
      "learning_rate": 0.0001650186413323044,
      "loss": 1.7113,
      "step": 8278
    },
    {
      "epoch": 0.29648862069582965,
      "grad_norm": 1.6844797134399414,
      "learning_rate": 0.0001650098282581266,
      "loss": 1.3356,
      "step": 8279
    },
    {
      "epoch": 0.296524432825398,
      "grad_norm": 1.9010940790176392,
      "learning_rate": 0.00016500101430932541,
      "loss": 1.7213,
      "step": 8280
    },
    {
      "epoch": 0.29656024495496625,
      "grad_norm": 1.565537452697754,
      "learning_rate": 0.00016499219948601943,
      "loss": 1.6433,
      "step": 8281
    },
    {
      "epoch": 0.2965960570845345,
      "grad_norm": 1.7080549001693726,
      "learning_rate": 0.00016498338378832724,
      "loss": 1.5829,
      "step": 8282
    },
    {
      "epoch": 0.29663186921410284,
      "grad_norm": 1.454037070274353,
      "learning_rate": 0.00016497456721636743,
      "loss": 1.5696,
      "step": 8283
    },
    {
      "epoch": 0.2966676813436711,
      "grad_norm": 1.5883538722991943,
      "learning_rate": 0.00016496574977025862,
      "loss": 1.5431,
      "step": 8284
    },
    {
      "epoch": 0.2967034934732394,
      "grad_norm": 1.2771259546279907,
      "learning_rate": 0.00016495693145011947,
      "loss": 1.2164,
      "step": 8285
    },
    {
      "epoch": 0.29673930560280765,
      "grad_norm": 1.6104718446731567,
      "learning_rate": 0.00016494811225606858,
      "loss": 1.3776,
      "step": 8286
    },
    {
      "epoch": 0.29677511773237597,
      "grad_norm": 1.967247724533081,
      "learning_rate": 0.00016493929218822467,
      "loss": 1.2856,
      "step": 8287
    },
    {
      "epoch": 0.29681092986194424,
      "grad_norm": 1.7795873880386353,
      "learning_rate": 0.0001649304712467063,
      "loss": 1.4519,
      "step": 8288
    },
    {
      "epoch": 0.2968467419915125,
      "grad_norm": 2.0310049057006836,
      "learning_rate": 0.00016492164943163217,
      "loss": 1.5672,
      "step": 8289
    },
    {
      "epoch": 0.29688255412108083,
      "grad_norm": 1.3638725280761719,
      "learning_rate": 0.00016491282674312103,
      "loss": 1.5978,
      "step": 8290
    },
    {
      "epoch": 0.2969183662506491,
      "grad_norm": 1.5186011791229248,
      "learning_rate": 0.00016490400318129153,
      "loss": 1.3685,
      "step": 8291
    },
    {
      "epoch": 0.29695417838021737,
      "grad_norm": 1.4526851177215576,
      "learning_rate": 0.0001648951787462624,
      "loss": 1.5908,
      "step": 8292
    },
    {
      "epoch": 0.29698999050978564,
      "grad_norm": 2.9446048736572266,
      "learning_rate": 0.0001648863534381523,
      "loss": 1.8091,
      "step": 8293
    },
    {
      "epoch": 0.29702580263935396,
      "grad_norm": 1.383347511291504,
      "learning_rate": 0.00016487752725708005,
      "loss": 1.5765,
      "step": 8294
    },
    {
      "epoch": 0.29706161476892223,
      "grad_norm": 1.7374597787857056,
      "learning_rate": 0.00016486870020316437,
      "loss": 1.4451,
      "step": 8295
    },
    {
      "epoch": 0.2970974268984905,
      "grad_norm": 1.5283315181732178,
      "learning_rate": 0.000164859872276524,
      "loss": 1.5174,
      "step": 8296
    },
    {
      "epoch": 0.2971332390280588,
      "grad_norm": 1.4629818201065063,
      "learning_rate": 0.0001648510434772777,
      "loss": 1.5145,
      "step": 8297
    },
    {
      "epoch": 0.2971690511576271,
      "grad_norm": 1.7839181423187256,
      "learning_rate": 0.00016484221380554424,
      "loss": 1.4526,
      "step": 8298
    },
    {
      "epoch": 0.29720486328719536,
      "grad_norm": 1.8995558023452759,
      "learning_rate": 0.00016483338326144244,
      "loss": 1.5883,
      "step": 8299
    },
    {
      "epoch": 0.29724067541676363,
      "grad_norm": 1.3919588327407837,
      "learning_rate": 0.0001648245518450911,
      "loss": 1.2238,
      "step": 8300
    },
    {
      "epoch": 0.29727648754633196,
      "grad_norm": 1.8809860944747925,
      "learning_rate": 0.00016481571955660903,
      "loss": 1.6602,
      "step": 8301
    },
    {
      "epoch": 0.2973122996759002,
      "grad_norm": 1.5680149793624878,
      "learning_rate": 0.0001648068863961151,
      "loss": 1.4787,
      "step": 8302
    },
    {
      "epoch": 0.2973481118054685,
      "grad_norm": 1.533979892730713,
      "learning_rate": 0.00016479805236372806,
      "loss": 1.3545,
      "step": 8303
    },
    {
      "epoch": 0.2973839239350368,
      "grad_norm": 1.733001708984375,
      "learning_rate": 0.00016478921745956686,
      "loss": 1.7839,
      "step": 8304
    },
    {
      "epoch": 0.2974197360646051,
      "grad_norm": 2.160454750061035,
      "learning_rate": 0.00016478038168375028,
      "loss": 1.7761,
      "step": 8305
    },
    {
      "epoch": 0.29745554819417336,
      "grad_norm": 1.3565226793289185,
      "learning_rate": 0.00016477154503639723,
      "loss": 1.5211,
      "step": 8306
    },
    {
      "epoch": 0.2974913603237416,
      "grad_norm": 1.4650872945785522,
      "learning_rate": 0.00016476270751762656,
      "loss": 1.1181,
      "step": 8307
    },
    {
      "epoch": 0.29752717245330995,
      "grad_norm": 1.9036059379577637,
      "learning_rate": 0.00016475386912755724,
      "loss": 1.4345,
      "step": 8308
    },
    {
      "epoch": 0.2975629845828782,
      "grad_norm": 2.0247349739074707,
      "learning_rate": 0.0001647450298663081,
      "loss": 1.7541,
      "step": 8309
    },
    {
      "epoch": 0.2975987967124465,
      "grad_norm": 1.7385376691818237,
      "learning_rate": 0.00016473618973399811,
      "loss": 1.5686,
      "step": 8310
    },
    {
      "epoch": 0.2976346088420148,
      "grad_norm": 1.8126300573349,
      "learning_rate": 0.00016472734873074622,
      "loss": 1.8558,
      "step": 8311
    },
    {
      "epoch": 0.2976704209715831,
      "grad_norm": 1.5259616374969482,
      "learning_rate": 0.00016471850685667133,
      "loss": 1.4215,
      "step": 8312
    },
    {
      "epoch": 0.29770623310115135,
      "grad_norm": 1.6709272861480713,
      "learning_rate": 0.0001647096641118924,
      "loss": 1.5873,
      "step": 8313
    },
    {
      "epoch": 0.2977420452307196,
      "grad_norm": 1.7461010217666626,
      "learning_rate": 0.00016470082049652843,
      "loss": 1.579,
      "step": 8314
    },
    {
      "epoch": 0.29777785736028795,
      "grad_norm": 1.4879776239395142,
      "learning_rate": 0.00016469197601069838,
      "loss": 1.4497,
      "step": 8315
    },
    {
      "epoch": 0.2978136694898562,
      "grad_norm": 1.4053375720977783,
      "learning_rate": 0.00016468313065452121,
      "loss": 1.5219,
      "step": 8316
    },
    {
      "epoch": 0.2978494816194245,
      "grad_norm": 1.5887116193771362,
      "learning_rate": 0.00016467428442811595,
      "loss": 1.3375,
      "step": 8317
    },
    {
      "epoch": 0.2978852937489928,
      "grad_norm": 2.0717251300811768,
      "learning_rate": 0.00016466543733160163,
      "loss": 1.7047,
      "step": 8318
    },
    {
      "epoch": 0.2979211058785611,
      "grad_norm": 1.4569612741470337,
      "learning_rate": 0.00016465658936509726,
      "loss": 1.4269,
      "step": 8319
    },
    {
      "epoch": 0.29795691800812935,
      "grad_norm": 1.4222359657287598,
      "learning_rate": 0.0001646477405287219,
      "loss": 1.5088,
      "step": 8320
    },
    {
      "epoch": 0.2979927301376976,
      "grad_norm": 1.5125815868377686,
      "learning_rate": 0.00016463889082259456,
      "loss": 1.5915,
      "step": 8321
    },
    {
      "epoch": 0.29802854226726594,
      "grad_norm": 1.9704197645187378,
      "learning_rate": 0.00016463004024683432,
      "loss": 1.6187,
      "step": 8322
    },
    {
      "epoch": 0.2980643543968342,
      "grad_norm": 1.6360664367675781,
      "learning_rate": 0.0001646211888015603,
      "loss": 1.3247,
      "step": 8323
    },
    {
      "epoch": 0.2981001665264025,
      "grad_norm": 1.4352376461029053,
      "learning_rate": 0.0001646123364868915,
      "loss": 1.3175,
      "step": 8324
    },
    {
      "epoch": 0.2981359786559708,
      "grad_norm": 1.6311535835266113,
      "learning_rate": 0.00016460348330294704,
      "loss": 1.2099,
      "step": 8325
    },
    {
      "epoch": 0.29817179078553907,
      "grad_norm": 1.5380768775939941,
      "learning_rate": 0.00016459462924984605,
      "loss": 1.782,
      "step": 8326
    },
    {
      "epoch": 0.29820760291510734,
      "grad_norm": 1.4307458400726318,
      "learning_rate": 0.00016458577432770766,
      "loss": 1.5807,
      "step": 8327
    },
    {
      "epoch": 0.2982434150446756,
      "grad_norm": 1.925248384475708,
      "learning_rate": 0.000164576918536651,
      "loss": 1.4601,
      "step": 8328
    },
    {
      "epoch": 0.29827922717424393,
      "grad_norm": 1.3756663799285889,
      "learning_rate": 0.0001645680618767952,
      "loss": 1.487,
      "step": 8329
    },
    {
      "epoch": 0.2983150393038122,
      "grad_norm": 1.7213997840881348,
      "learning_rate": 0.00016455920434825936,
      "loss": 1.681,
      "step": 8330
    },
    {
      "epoch": 0.29835085143338047,
      "grad_norm": 1.3830705881118774,
      "learning_rate": 0.00016455034595116278,
      "loss": 1.3369,
      "step": 8331
    },
    {
      "epoch": 0.2983866635629488,
      "grad_norm": 1.7282135486602783,
      "learning_rate": 0.00016454148668562454,
      "loss": 1.1979,
      "step": 8332
    },
    {
      "epoch": 0.29842247569251706,
      "grad_norm": 2.0980710983276367,
      "learning_rate": 0.0001645326265517638,
      "loss": 1.4047,
      "step": 8333
    },
    {
      "epoch": 0.29845828782208533,
      "grad_norm": 1.5762666463851929,
      "learning_rate": 0.00016452376554969983,
      "loss": 1.7234,
      "step": 8334
    },
    {
      "epoch": 0.2984940999516536,
      "grad_norm": 1.630379319190979,
      "learning_rate": 0.00016451490367955183,
      "loss": 1.5907,
      "step": 8335
    },
    {
      "epoch": 0.2985299120812219,
      "grad_norm": 2.5405523777008057,
      "learning_rate": 0.00016450604094143904,
      "loss": 1.6051,
      "step": 8336
    },
    {
      "epoch": 0.2985657242107902,
      "grad_norm": 1.3858790397644043,
      "learning_rate": 0.00016449717733548066,
      "loss": 1.5432,
      "step": 8337
    },
    {
      "epoch": 0.29860153634035846,
      "grad_norm": 1.749637246131897,
      "learning_rate": 0.00016448831286179595,
      "loss": 1.6572,
      "step": 8338
    },
    {
      "epoch": 0.2986373484699268,
      "grad_norm": 1.4837186336517334,
      "learning_rate": 0.00016447944752050417,
      "loss": 1.552,
      "step": 8339
    },
    {
      "epoch": 0.29867316059949506,
      "grad_norm": 1.4648246765136719,
      "learning_rate": 0.00016447058131172462,
      "loss": 1.4813,
      "step": 8340
    },
    {
      "epoch": 0.2987089727290633,
      "grad_norm": 1.9088497161865234,
      "learning_rate": 0.00016446171423557652,
      "loss": 1.6785,
      "step": 8341
    },
    {
      "epoch": 0.2987447848586316,
      "grad_norm": 1.8436561822891235,
      "learning_rate": 0.00016445284629217923,
      "loss": 1.6028,
      "step": 8342
    },
    {
      "epoch": 0.2987805969881999,
      "grad_norm": 1.7736629247665405,
      "learning_rate": 0.00016444397748165205,
      "loss": 1.672,
      "step": 8343
    },
    {
      "epoch": 0.2988164091177682,
      "grad_norm": 1.4750014543533325,
      "learning_rate": 0.00016443510780411423,
      "loss": 1.5656,
      "step": 8344
    },
    {
      "epoch": 0.29885222124733646,
      "grad_norm": 2.1682565212249756,
      "learning_rate": 0.0001644262372596852,
      "loss": 1.507,
      "step": 8345
    },
    {
      "epoch": 0.2988880333769048,
      "grad_norm": 1.6554548740386963,
      "learning_rate": 0.00016441736584848422,
      "loss": 1.4692,
      "step": 8346
    },
    {
      "epoch": 0.29892384550647305,
      "grad_norm": 1.3574022054672241,
      "learning_rate": 0.0001644084935706307,
      "loss": 1.4869,
      "step": 8347
    },
    {
      "epoch": 0.2989596576360413,
      "grad_norm": 1.7374560832977295,
      "learning_rate": 0.00016439962042624396,
      "loss": 1.7313,
      "step": 8348
    },
    {
      "epoch": 0.2989954697656096,
      "grad_norm": 1.4507265090942383,
      "learning_rate": 0.0001643907464154434,
      "loss": 1.569,
      "step": 8349
    },
    {
      "epoch": 0.2990312818951779,
      "grad_norm": 1.6869750022888184,
      "learning_rate": 0.00016438187153834842,
      "loss": 1.5569,
      "step": 8350
    },
    {
      "epoch": 0.2990670940247462,
      "grad_norm": 1.4608795642852783,
      "learning_rate": 0.0001643729957950784,
      "loss": 1.5137,
      "step": 8351
    },
    {
      "epoch": 0.29910290615431445,
      "grad_norm": 1.5899919271469116,
      "learning_rate": 0.00016436411918575275,
      "loss": 1.4346,
      "step": 8352
    },
    {
      "epoch": 0.2991387182838828,
      "grad_norm": 1.7465403079986572,
      "learning_rate": 0.00016435524171049088,
      "loss": 1.8228,
      "step": 8353
    },
    {
      "epoch": 0.29917453041345105,
      "grad_norm": 1.4676436185836792,
      "learning_rate": 0.00016434636336941228,
      "loss": 1.9146,
      "step": 8354
    },
    {
      "epoch": 0.2992103425430193,
      "grad_norm": 1.4257675409317017,
      "learning_rate": 0.00016433748416263633,
      "loss": 1.7837,
      "step": 8355
    },
    {
      "epoch": 0.2992461546725876,
      "grad_norm": 2.4013359546661377,
      "learning_rate": 0.00016432860409028253,
      "loss": 1.7198,
      "step": 8356
    },
    {
      "epoch": 0.2992819668021559,
      "grad_norm": 1.3581948280334473,
      "learning_rate": 0.00016431972315247037,
      "loss": 1.3561,
      "step": 8357
    },
    {
      "epoch": 0.2993177789317242,
      "grad_norm": 1.4917012453079224,
      "learning_rate": 0.00016431084134931927,
      "loss": 1.789,
      "step": 8358
    },
    {
      "epoch": 0.29935359106129245,
      "grad_norm": 1.5227102041244507,
      "learning_rate": 0.00016430195868094875,
      "loss": 1.6315,
      "step": 8359
    },
    {
      "epoch": 0.29938940319086077,
      "grad_norm": 1.2591630220413208,
      "learning_rate": 0.00016429307514747834,
      "loss": 1.6123,
      "step": 8360
    },
    {
      "epoch": 0.29942521532042904,
      "grad_norm": 1.3538967370986938,
      "learning_rate": 0.00016428419074902752,
      "loss": 1.4282,
      "step": 8361
    },
    {
      "epoch": 0.2994610274499973,
      "grad_norm": 2.2829110622406006,
      "learning_rate": 0.00016427530548571585,
      "loss": 1.5408,
      "step": 8362
    },
    {
      "epoch": 0.2994968395795656,
      "grad_norm": 1.5202713012695312,
      "learning_rate": 0.00016426641935766284,
      "loss": 1.5582,
      "step": 8363
    },
    {
      "epoch": 0.2995326517091339,
      "grad_norm": 1.6621960401535034,
      "learning_rate": 0.00016425753236498807,
      "loss": 1.3942,
      "step": 8364
    },
    {
      "epoch": 0.29956846383870217,
      "grad_norm": 1.912682294845581,
      "learning_rate": 0.00016424864450781108,
      "loss": 1.6684,
      "step": 8365
    },
    {
      "epoch": 0.29960427596827044,
      "grad_norm": 1.3282166719436646,
      "learning_rate": 0.00016423975578625142,
      "loss": 1.5807,
      "step": 8366
    },
    {
      "epoch": 0.29964008809783876,
      "grad_norm": 1.7995824813842773,
      "learning_rate": 0.00016423086620042879,
      "loss": 1.4495,
      "step": 8367
    },
    {
      "epoch": 0.29967590022740703,
      "grad_norm": 1.258223056793213,
      "learning_rate": 0.00016422197575046265,
      "loss": 1.7473,
      "step": 8368
    },
    {
      "epoch": 0.2997117123569753,
      "grad_norm": 1.346404790878296,
      "learning_rate": 0.00016421308443647265,
      "loss": 1.4512,
      "step": 8369
    },
    {
      "epoch": 0.29974752448654357,
      "grad_norm": 1.8389484882354736,
      "learning_rate": 0.00016420419225857846,
      "loss": 1.5317,
      "step": 8370
    },
    {
      "epoch": 0.2997833366161119,
      "grad_norm": 1.1908183097839355,
      "learning_rate": 0.00016419529921689967,
      "loss": 1.2523,
      "step": 8371
    },
    {
      "epoch": 0.29981914874568016,
      "grad_norm": 1.4290704727172852,
      "learning_rate": 0.00016418640531155597,
      "loss": 1.552,
      "step": 8372
    },
    {
      "epoch": 0.29985496087524843,
      "grad_norm": 1.5900216102600098,
      "learning_rate": 0.00016417751054266692,
      "loss": 1.4178,
      "step": 8373
    },
    {
      "epoch": 0.29989077300481676,
      "grad_norm": 1.5539711713790894,
      "learning_rate": 0.00016416861491035228,
      "loss": 1.1549,
      "step": 8374
    },
    {
      "epoch": 0.299926585134385,
      "grad_norm": 1.5073728561401367,
      "learning_rate": 0.0001641597184147317,
      "loss": 1.4495,
      "step": 8375
    },
    {
      "epoch": 0.2999623972639533,
      "grad_norm": 1.420148491859436,
      "learning_rate": 0.0001641508210559249,
      "loss": 1.381,
      "step": 8376
    },
    {
      "epoch": 0.29999820939352156,
      "grad_norm": 1.7865599393844604,
      "learning_rate": 0.00016414192283405147,
      "loss": 1.3243,
      "step": 8377
    },
    {
      "epoch": 0.3000340215230899,
      "grad_norm": 2.0387165546417236,
      "learning_rate": 0.00016413302374923124,
      "loss": 1.6688,
      "step": 8378
    },
    {
      "epoch": 0.30006983365265816,
      "grad_norm": 1.510023832321167,
      "learning_rate": 0.00016412412380158392,
      "loss": 1.726,
      "step": 8379
    },
    {
      "epoch": 0.3001056457822264,
      "grad_norm": 1.5913726091384888,
      "learning_rate": 0.00016411522299122924,
      "loss": 1.7165,
      "step": 8380
    },
    {
      "epoch": 0.30014145791179475,
      "grad_norm": 1.7233575582504272,
      "learning_rate": 0.0001641063213182869,
      "loss": 1.5722,
      "step": 8381
    },
    {
      "epoch": 0.300177270041363,
      "grad_norm": 1.4803311824798584,
      "learning_rate": 0.00016409741878287671,
      "loss": 1.2891,
      "step": 8382
    },
    {
      "epoch": 0.3002130821709313,
      "grad_norm": 2.3403327465057373,
      "learning_rate": 0.00016408851538511846,
      "loss": 1.494,
      "step": 8383
    },
    {
      "epoch": 0.30024889430049956,
      "grad_norm": 1.5356950759887695,
      "learning_rate": 0.0001640796111251319,
      "loss": 1.2206,
      "step": 8384
    },
    {
      "epoch": 0.3002847064300679,
      "grad_norm": 1.619539737701416,
      "learning_rate": 0.0001640707060030368,
      "loss": 1.5673,
      "step": 8385
    },
    {
      "epoch": 0.30032051855963615,
      "grad_norm": 1.3473477363586426,
      "learning_rate": 0.00016406180001895298,
      "loss": 1.7308,
      "step": 8386
    },
    {
      "epoch": 0.3003563306892044,
      "grad_norm": 1.37346351146698,
      "learning_rate": 0.00016405289317300033,
      "loss": 1.307,
      "step": 8387
    },
    {
      "epoch": 0.30039214281877274,
      "grad_norm": 1.9907050132751465,
      "learning_rate": 0.00016404398546529859,
      "loss": 1.7465,
      "step": 8388
    },
    {
      "epoch": 0.300427954948341,
      "grad_norm": 1.3930773735046387,
      "learning_rate": 0.00016403507689596763,
      "loss": 1.4954,
      "step": 8389
    },
    {
      "epoch": 0.3004637670779093,
      "grad_norm": 2.656479835510254,
      "learning_rate": 0.0001640261674651273,
      "loss": 1.4997,
      "step": 8390
    },
    {
      "epoch": 0.30049957920747755,
      "grad_norm": 2.1430234909057617,
      "learning_rate": 0.0001640172571728975,
      "loss": 1.6448,
      "step": 8391
    },
    {
      "epoch": 0.3005353913370459,
      "grad_norm": 1.5094202756881714,
      "learning_rate": 0.0001640083460193981,
      "loss": 1.3046,
      "step": 8392
    },
    {
      "epoch": 0.30057120346661415,
      "grad_norm": 1.5384578704833984,
      "learning_rate": 0.00016399943400474895,
      "loss": 1.3601,
      "step": 8393
    },
    {
      "epoch": 0.3006070155961824,
      "grad_norm": 1.3882454633712769,
      "learning_rate": 0.00016399052112906994,
      "loss": 1.3568,
      "step": 8394
    },
    {
      "epoch": 0.30064282772575074,
      "grad_norm": 1.2879197597503662,
      "learning_rate": 0.00016398160739248104,
      "loss": 1.1968,
      "step": 8395
    },
    {
      "epoch": 0.300678639855319,
      "grad_norm": 1.569235920906067,
      "learning_rate": 0.00016397269279510215,
      "loss": 1.4976,
      "step": 8396
    },
    {
      "epoch": 0.3007144519848873,
      "grad_norm": 1.489635705947876,
      "learning_rate": 0.00016396377733705317,
      "loss": 1.4407,
      "step": 8397
    },
    {
      "epoch": 0.30075026411445555,
      "grad_norm": 2.0916008949279785,
      "learning_rate": 0.00016395486101845408,
      "loss": 1.3873,
      "step": 8398
    },
    {
      "epoch": 0.30078607624402387,
      "grad_norm": 1.9567631483078003,
      "learning_rate": 0.00016394594383942486,
      "loss": 1.2718,
      "step": 8399
    },
    {
      "epoch": 0.30082188837359214,
      "grad_norm": 2.1201071739196777,
      "learning_rate": 0.00016393702580008542,
      "loss": 1.3483,
      "step": 8400
    },
    {
      "epoch": 0.3008577005031604,
      "grad_norm": 1.7831958532333374,
      "learning_rate": 0.00016392810690055577,
      "loss": 1.4788,
      "step": 8401
    },
    {
      "epoch": 0.30089351263272873,
      "grad_norm": 1.9678820371627808,
      "learning_rate": 0.00016391918714095592,
      "loss": 1.6395,
      "step": 8402
    },
    {
      "epoch": 0.300929324762297,
      "grad_norm": 1.531544804573059,
      "learning_rate": 0.00016391026652140585,
      "loss": 1.4363,
      "step": 8403
    },
    {
      "epoch": 0.30096513689186527,
      "grad_norm": 2.5241010189056396,
      "learning_rate": 0.00016390134504202557,
      "loss": 1.3736,
      "step": 8404
    },
    {
      "epoch": 0.30100094902143354,
      "grad_norm": 2.2505462169647217,
      "learning_rate": 0.00016389242270293514,
      "loss": 1.1755,
      "step": 8405
    },
    {
      "epoch": 0.30103676115100186,
      "grad_norm": 1.7323700189590454,
      "learning_rate": 0.00016388349950425456,
      "loss": 1.5693,
      "step": 8406
    },
    {
      "epoch": 0.30107257328057013,
      "grad_norm": 1.761555552482605,
      "learning_rate": 0.0001638745754461039,
      "loss": 1.3442,
      "step": 8407
    },
    {
      "epoch": 0.3011083854101384,
      "grad_norm": 1.8615801334381104,
      "learning_rate": 0.00016386565052860323,
      "loss": 1.6013,
      "step": 8408
    },
    {
      "epoch": 0.3011441975397067,
      "grad_norm": 1.4850836992263794,
      "learning_rate": 0.00016385672475187262,
      "loss": 1.4907,
      "step": 8409
    },
    {
      "epoch": 0.301180009669275,
      "grad_norm": 1.6146981716156006,
      "learning_rate": 0.00016384779811603214,
      "loss": 1.5575,
      "step": 8410
    },
    {
      "epoch": 0.30121582179884326,
      "grad_norm": 1.6145588159561157,
      "learning_rate": 0.0001638388706212019,
      "loss": 1.4367,
      "step": 8411
    },
    {
      "epoch": 0.30125163392841153,
      "grad_norm": 2.3963756561279297,
      "learning_rate": 0.000163829942267502,
      "loss": 1.5233,
      "step": 8412
    },
    {
      "epoch": 0.30128744605797986,
      "grad_norm": 1.9880651235580444,
      "learning_rate": 0.00016382101305505254,
      "loss": 1.676,
      "step": 8413
    },
    {
      "epoch": 0.3013232581875481,
      "grad_norm": 1.5233274698257446,
      "learning_rate": 0.0001638120829839737,
      "loss": 1.5215,
      "step": 8414
    },
    {
      "epoch": 0.3013590703171164,
      "grad_norm": 1.7098640203475952,
      "learning_rate": 0.00016380315205438554,
      "loss": 1.3578,
      "step": 8415
    },
    {
      "epoch": 0.3013948824466847,
      "grad_norm": 2.10298490524292,
      "learning_rate": 0.00016379422026640831,
      "loss": 1.5301,
      "step": 8416
    },
    {
      "epoch": 0.301430694576253,
      "grad_norm": 1.466875433921814,
      "learning_rate": 0.00016378528762016218,
      "loss": 1.2364,
      "step": 8417
    },
    {
      "epoch": 0.30146650670582126,
      "grad_norm": 1.7537953853607178,
      "learning_rate": 0.00016377635411576723,
      "loss": 1.5604,
      "step": 8418
    },
    {
      "epoch": 0.3015023188353895,
      "grad_norm": 2.448864221572876,
      "learning_rate": 0.00016376741975334368,
      "loss": 1.6991,
      "step": 8419
    },
    {
      "epoch": 0.30153813096495785,
      "grad_norm": 2.6458969116210938,
      "learning_rate": 0.0001637584845330118,
      "loss": 1.9327,
      "step": 8420
    },
    {
      "epoch": 0.3015739430945261,
      "grad_norm": 1.5955549478530884,
      "learning_rate": 0.00016374954845489175,
      "loss": 1.4872,
      "step": 8421
    },
    {
      "epoch": 0.3016097552240944,
      "grad_norm": 1.0321791172027588,
      "learning_rate": 0.00016374061151910372,
      "loss": 1.2884,
      "step": 8422
    },
    {
      "epoch": 0.30164556735366266,
      "grad_norm": 1.4548122882843018,
      "learning_rate": 0.000163731673725768,
      "loss": 1.3905,
      "step": 8423
    },
    {
      "epoch": 0.301681379483231,
      "grad_norm": 1.4599908590316772,
      "learning_rate": 0.00016372273507500481,
      "loss": 1.637,
      "step": 8424
    },
    {
      "epoch": 0.30171719161279925,
      "grad_norm": 1.4581354856491089,
      "learning_rate": 0.00016371379556693442,
      "loss": 1.4863,
      "step": 8425
    },
    {
      "epoch": 0.3017530037423675,
      "grad_norm": 1.6916433572769165,
      "learning_rate": 0.0001637048552016771,
      "loss": 1.3605,
      "step": 8426
    },
    {
      "epoch": 0.30178881587193584,
      "grad_norm": 1.710843801498413,
      "learning_rate": 0.00016369591397935314,
      "loss": 1.6055,
      "step": 8427
    },
    {
      "epoch": 0.3018246280015041,
      "grad_norm": 1.4895509481430054,
      "learning_rate": 0.0001636869719000828,
      "loss": 1.4946,
      "step": 8428
    },
    {
      "epoch": 0.3018604401310724,
      "grad_norm": 1.3431249856948853,
      "learning_rate": 0.0001636780289639864,
      "loss": 1.5975,
      "step": 8429
    },
    {
      "epoch": 0.30189625226064065,
      "grad_norm": 1.9175156354904175,
      "learning_rate": 0.00016366908517118428,
      "loss": 1.2682,
      "step": 8430
    },
    {
      "epoch": 0.301932064390209,
      "grad_norm": 1.7841920852661133,
      "learning_rate": 0.00016366014052179674,
      "loss": 1.1883,
      "step": 8431
    },
    {
      "epoch": 0.30196787651977725,
      "grad_norm": 2.296452760696411,
      "learning_rate": 0.0001636511950159441,
      "loss": 1.4858,
      "step": 8432
    },
    {
      "epoch": 0.3020036886493455,
      "grad_norm": 1.7015984058380127,
      "learning_rate": 0.00016364224865374677,
      "loss": 1.6448,
      "step": 8433
    },
    {
      "epoch": 0.30203950077891384,
      "grad_norm": 1.5796960592269897,
      "learning_rate": 0.00016363330143532508,
      "loss": 1.5356,
      "step": 8434
    },
    {
      "epoch": 0.3020753129084821,
      "grad_norm": 1.562314510345459,
      "learning_rate": 0.00016362435336079938,
      "loss": 1.503,
      "step": 8435
    },
    {
      "epoch": 0.3021111250380504,
      "grad_norm": 1.5993971824645996,
      "learning_rate": 0.00016361540443029008,
      "loss": 1.7656,
      "step": 8436
    },
    {
      "epoch": 0.30214693716761865,
      "grad_norm": 2.1478962898254395,
      "learning_rate": 0.00016360645464391754,
      "loss": 1.4395,
      "step": 8437
    },
    {
      "epoch": 0.30218274929718697,
      "grad_norm": 2.219712257385254,
      "learning_rate": 0.00016359750400180226,
      "loss": 1.7623,
      "step": 8438
    },
    {
      "epoch": 0.30221856142675524,
      "grad_norm": 1.7242192029953003,
      "learning_rate": 0.00016358855250406455,
      "loss": 1.3607,
      "step": 8439
    },
    {
      "epoch": 0.3022543735563235,
      "grad_norm": 1.4670357704162598,
      "learning_rate": 0.0001635796001508249,
      "loss": 1.6872,
      "step": 8440
    },
    {
      "epoch": 0.30229018568589183,
      "grad_norm": 1.7151837348937988,
      "learning_rate": 0.00016357064694220375,
      "loss": 1.7587,
      "step": 8441
    },
    {
      "epoch": 0.3023259978154601,
      "grad_norm": 2.2550930976867676,
      "learning_rate": 0.00016356169287832156,
      "loss": 1.598,
      "step": 8442
    },
    {
      "epoch": 0.30236180994502837,
      "grad_norm": 1.5630356073379517,
      "learning_rate": 0.00016355273795929875,
      "loss": 1.6793,
      "step": 8443
    },
    {
      "epoch": 0.30239762207459664,
      "grad_norm": 2.2968051433563232,
      "learning_rate": 0.00016354378218525584,
      "loss": 1.9349,
      "step": 8444
    },
    {
      "epoch": 0.30243343420416496,
      "grad_norm": 1.6883846521377563,
      "learning_rate": 0.00016353482555631334,
      "loss": 1.5545,
      "step": 8445
    },
    {
      "epoch": 0.30246924633373323,
      "grad_norm": 1.5997358560562134,
      "learning_rate": 0.00016352586807259168,
      "loss": 1.9512,
      "step": 8446
    },
    {
      "epoch": 0.3025050584633015,
      "grad_norm": 1.1980454921722412,
      "learning_rate": 0.00016351690973421138,
      "loss": 1.5382,
      "step": 8447
    },
    {
      "epoch": 0.3025408705928698,
      "grad_norm": 1.3863976001739502,
      "learning_rate": 0.00016350795054129305,
      "loss": 1.4965,
      "step": 8448
    },
    {
      "epoch": 0.3025766827224381,
      "grad_norm": 1.4191679954528809,
      "learning_rate": 0.00016349899049395713,
      "loss": 1.6027,
      "step": 8449
    },
    {
      "epoch": 0.30261249485200636,
      "grad_norm": 1.6715854406356812,
      "learning_rate": 0.0001634900295923242,
      "loss": 1.7932,
      "step": 8450
    },
    {
      "epoch": 0.30264830698157463,
      "grad_norm": 1.6888922452926636,
      "learning_rate": 0.00016348106783651482,
      "loss": 1.107,
      "step": 8451
    },
    {
      "epoch": 0.30268411911114296,
      "grad_norm": 2.2722463607788086,
      "learning_rate": 0.00016347210522664956,
      "loss": 1.847,
      "step": 8452
    },
    {
      "epoch": 0.3027199312407112,
      "grad_norm": 1.5517665147781372,
      "learning_rate": 0.000163463141762849,
      "loss": 1.4461,
      "step": 8453
    },
    {
      "epoch": 0.3027557433702795,
      "grad_norm": 1.6976096630096436,
      "learning_rate": 0.00016345417744523374,
      "loss": 1.5298,
      "step": 8454
    },
    {
      "epoch": 0.3027915554998478,
      "grad_norm": 1.631230354309082,
      "learning_rate": 0.00016344521227392437,
      "loss": 1.4179,
      "step": 8455
    },
    {
      "epoch": 0.3028273676294161,
      "grad_norm": 1.2763921022415161,
      "learning_rate": 0.00016343624624904151,
      "loss": 1.7387,
      "step": 8456
    },
    {
      "epoch": 0.30286317975898436,
      "grad_norm": 2.225421667098999,
      "learning_rate": 0.00016342727937070577,
      "loss": 1.4409,
      "step": 8457
    },
    {
      "epoch": 0.3028989918885526,
      "grad_norm": 2.22019100189209,
      "learning_rate": 0.0001634183116390378,
      "loss": 1.9976,
      "step": 8458
    },
    {
      "epoch": 0.30293480401812095,
      "grad_norm": 1.8093554973602295,
      "learning_rate": 0.00016340934305415823,
      "loss": 1.7154,
      "step": 8459
    },
    {
      "epoch": 0.3029706161476892,
      "grad_norm": 1.4893295764923096,
      "learning_rate": 0.00016340037361618778,
      "loss": 1.3436,
      "step": 8460
    },
    {
      "epoch": 0.3030064282772575,
      "grad_norm": 1.99727201461792,
      "learning_rate": 0.00016339140332524707,
      "loss": 1.6495,
      "step": 8461
    },
    {
      "epoch": 0.3030422404068258,
      "grad_norm": 1.6516386270523071,
      "learning_rate": 0.0001633824321814568,
      "loss": 1.7346,
      "step": 8462
    },
    {
      "epoch": 0.3030780525363941,
      "grad_norm": 1.309701919555664,
      "learning_rate": 0.00016337346018493768,
      "loss": 1.4401,
      "step": 8463
    },
    {
      "epoch": 0.30311386466596235,
      "grad_norm": 1.5844017267227173,
      "learning_rate": 0.00016336448733581037,
      "loss": 1.7372,
      "step": 8464
    },
    {
      "epoch": 0.3031496767955306,
      "grad_norm": 1.5818967819213867,
      "learning_rate": 0.00016335551363419562,
      "loss": 1.5704,
      "step": 8465
    },
    {
      "epoch": 0.30318548892509894,
      "grad_norm": 1.4946575164794922,
      "learning_rate": 0.00016334653908021415,
      "loss": 1.6854,
      "step": 8466
    },
    {
      "epoch": 0.3032213010546672,
      "grad_norm": 2.2354283332824707,
      "learning_rate": 0.00016333756367398674,
      "loss": 1.9091,
      "step": 8467
    },
    {
      "epoch": 0.3032571131842355,
      "grad_norm": 1.3000431060791016,
      "learning_rate": 0.00016332858741563408,
      "loss": 1.4807,
      "step": 8468
    },
    {
      "epoch": 0.3032929253138038,
      "grad_norm": 1.4877946376800537,
      "learning_rate": 0.00016331961030527698,
      "loss": 1.3538,
      "step": 8469
    },
    {
      "epoch": 0.3033287374433721,
      "grad_norm": 1.6280548572540283,
      "learning_rate": 0.00016331063234303618,
      "loss": 1.6621,
      "step": 8470
    },
    {
      "epoch": 0.30336454957294035,
      "grad_norm": 1.3985967636108398,
      "learning_rate": 0.0001633016535290325,
      "loss": 1.2751,
      "step": 8471
    },
    {
      "epoch": 0.3034003617025086,
      "grad_norm": 2.0573692321777344,
      "learning_rate": 0.00016329267386338674,
      "loss": 1.4004,
      "step": 8472
    },
    {
      "epoch": 0.30343617383207694,
      "grad_norm": 1.4988259077072144,
      "learning_rate": 0.0001632836933462197,
      "loss": 1.2686,
      "step": 8473
    },
    {
      "epoch": 0.3034719859616452,
      "grad_norm": 1.387237548828125,
      "learning_rate": 0.00016327471197765216,
      "loss": 1.4457,
      "step": 8474
    },
    {
      "epoch": 0.3035077980912135,
      "grad_norm": 2.1457505226135254,
      "learning_rate": 0.000163265729757805,
      "loss": 1.6394,
      "step": 8475
    },
    {
      "epoch": 0.3035436102207818,
      "grad_norm": 1.7140662670135498,
      "learning_rate": 0.00016325674668679906,
      "loss": 1.5671,
      "step": 8476
    },
    {
      "epoch": 0.30357942235035007,
      "grad_norm": 1.7790052890777588,
      "learning_rate": 0.00016324776276475518,
      "loss": 1.6923,
      "step": 8477
    },
    {
      "epoch": 0.30361523447991834,
      "grad_norm": 1.8531421422958374,
      "learning_rate": 0.0001632387779917943,
      "loss": 1.6007,
      "step": 8478
    },
    {
      "epoch": 0.3036510466094866,
      "grad_norm": 1.4344226121902466,
      "learning_rate": 0.00016322979236803713,
      "loss": 1.6599,
      "step": 8479
    },
    {
      "epoch": 0.30368685873905493,
      "grad_norm": 1.5738317966461182,
      "learning_rate": 0.00016322080589360472,
      "loss": 1.7954,
      "step": 8480
    },
    {
      "epoch": 0.3037226708686232,
      "grad_norm": 1.9308953285217285,
      "learning_rate": 0.0001632118185686179,
      "loss": 1.4939,
      "step": 8481
    },
    {
      "epoch": 0.30375848299819147,
      "grad_norm": 1.450674057006836,
      "learning_rate": 0.0001632028303931976,
      "loss": 1.4645,
      "step": 8482
    },
    {
      "epoch": 0.3037942951277598,
      "grad_norm": 1.5372213125228882,
      "learning_rate": 0.00016319384136746477,
      "loss": 1.6913,
      "step": 8483
    },
    {
      "epoch": 0.30383010725732806,
      "grad_norm": 1.1684625148773193,
      "learning_rate": 0.0001631848514915403,
      "loss": 1.5842,
      "step": 8484
    },
    {
      "epoch": 0.30386591938689633,
      "grad_norm": 1.4962623119354248,
      "learning_rate": 0.00016317586076554515,
      "loss": 1.6532,
      "step": 8485
    },
    {
      "epoch": 0.3039017315164646,
      "grad_norm": 1.7394448518753052,
      "learning_rate": 0.0001631668691896003,
      "loss": 1.5523,
      "step": 8486
    },
    {
      "epoch": 0.3039375436460329,
      "grad_norm": 2.025491237640381,
      "learning_rate": 0.00016315787676382667,
      "loss": 1.6246,
      "step": 8487
    },
    {
      "epoch": 0.3039733557756012,
      "grad_norm": 3.2317962646484375,
      "learning_rate": 0.0001631488834883453,
      "loss": 1.5146,
      "step": 8488
    },
    {
      "epoch": 0.30400916790516946,
      "grad_norm": 2.3871772289276123,
      "learning_rate": 0.00016313988936327717,
      "loss": 1.7163,
      "step": 8489
    },
    {
      "epoch": 0.3040449800347378,
      "grad_norm": 1.6114022731781006,
      "learning_rate": 0.00016313089438874326,
      "loss": 1.8252,
      "step": 8490
    },
    {
      "epoch": 0.30408079216430606,
      "grad_norm": 1.964486002922058,
      "learning_rate": 0.00016312189856486462,
      "loss": 1.7783,
      "step": 8491
    },
    {
      "epoch": 0.3041166042938743,
      "grad_norm": 1.4887510538101196,
      "learning_rate": 0.00016311290189176223,
      "loss": 1.5578,
      "step": 8492
    },
    {
      "epoch": 0.3041524164234426,
      "grad_norm": 2.208726167678833,
      "learning_rate": 0.00016310390436955716,
      "loss": 1.51,
      "step": 8493
    },
    {
      "epoch": 0.3041882285530109,
      "grad_norm": 1.4921742677688599,
      "learning_rate": 0.00016309490599837045,
      "loss": 1.77,
      "step": 8494
    },
    {
      "epoch": 0.3042240406825792,
      "grad_norm": 1.6540600061416626,
      "learning_rate": 0.00016308590677832315,
      "loss": 1.4772,
      "step": 8495
    },
    {
      "epoch": 0.30425985281214746,
      "grad_norm": 1.4199429750442505,
      "learning_rate": 0.0001630769067095364,
      "loss": 1.4281,
      "step": 8496
    },
    {
      "epoch": 0.3042956649417158,
      "grad_norm": 2.591513156890869,
      "learning_rate": 0.0001630679057921312,
      "loss": 1.4973,
      "step": 8497
    },
    {
      "epoch": 0.30433147707128405,
      "grad_norm": 1.456068515777588,
      "learning_rate": 0.0001630589040262287,
      "loss": 1.4531,
      "step": 8498
    },
    {
      "epoch": 0.3043672892008523,
      "grad_norm": 2.0415728092193604,
      "learning_rate": 0.00016304990141194996,
      "loss": 1.4656,
      "step": 8499
    },
    {
      "epoch": 0.3044031013304206,
      "grad_norm": 1.6179795265197754,
      "learning_rate": 0.00016304089794941614,
      "loss": 1.7035,
      "step": 8500
    },
    {
      "epoch": 0.3044389134599889,
      "grad_norm": 1.568784475326538,
      "learning_rate": 0.00016303189363874835,
      "loss": 1.3761,
      "step": 8501
    },
    {
      "epoch": 0.3044747255895572,
      "grad_norm": 1.5610696077346802,
      "learning_rate": 0.00016302288848006776,
      "loss": 1.3034,
      "step": 8502
    },
    {
      "epoch": 0.30451053771912545,
      "grad_norm": 2.025500774383545,
      "learning_rate": 0.00016301388247349545,
      "loss": 1.4422,
      "step": 8503
    },
    {
      "epoch": 0.3045463498486938,
      "grad_norm": 1.6458475589752197,
      "learning_rate": 0.00016300487561915266,
      "loss": 1.5523,
      "step": 8504
    },
    {
      "epoch": 0.30458216197826204,
      "grad_norm": 1.8756228685379028,
      "learning_rate": 0.00016299586791716054,
      "loss": 1.642,
      "step": 8505
    },
    {
      "epoch": 0.3046179741078303,
      "grad_norm": 1.4094475507736206,
      "learning_rate": 0.00016298685936764026,
      "loss": 1.3843,
      "step": 8506
    },
    {
      "epoch": 0.3046537862373986,
      "grad_norm": 1.7004481554031372,
      "learning_rate": 0.00016297784997071308,
      "loss": 1.4843,
      "step": 8507
    },
    {
      "epoch": 0.3046895983669669,
      "grad_norm": 1.459661602973938,
      "learning_rate": 0.00016296883972650013,
      "loss": 1.4667,
      "step": 8508
    },
    {
      "epoch": 0.3047254104965352,
      "grad_norm": 1.3875668048858643,
      "learning_rate": 0.00016295982863512266,
      "loss": 1.4236,
      "step": 8509
    },
    {
      "epoch": 0.30476122262610345,
      "grad_norm": 2.15937876701355,
      "learning_rate": 0.00016295081669670191,
      "loss": 1.7024,
      "step": 8510
    },
    {
      "epoch": 0.30479703475567177,
      "grad_norm": 1.7457246780395508,
      "learning_rate": 0.00016294180391135914,
      "loss": 1.3893,
      "step": 8511
    },
    {
      "epoch": 0.30483284688524004,
      "grad_norm": 1.7229722738265991,
      "learning_rate": 0.00016293279027921557,
      "loss": 1.4539,
      "step": 8512
    },
    {
      "epoch": 0.3048686590148083,
      "grad_norm": 2.1370930671691895,
      "learning_rate": 0.0001629237758003925,
      "loss": 1.7631,
      "step": 8513
    },
    {
      "epoch": 0.3049044711443766,
      "grad_norm": 1.49868905544281,
      "learning_rate": 0.00016291476047501115,
      "loss": 1.6417,
      "step": 8514
    },
    {
      "epoch": 0.3049402832739449,
      "grad_norm": 1.5884264707565308,
      "learning_rate": 0.0001629057443031929,
      "loss": 1.4924,
      "step": 8515
    },
    {
      "epoch": 0.30497609540351317,
      "grad_norm": 1.7246391773223877,
      "learning_rate": 0.000162896727285059,
      "loss": 1.4056,
      "step": 8516
    },
    {
      "epoch": 0.30501190753308144,
      "grad_norm": 1.9741127490997314,
      "learning_rate": 0.00016288770942073075,
      "loss": 1.6657,
      "step": 8517
    },
    {
      "epoch": 0.30504771966264976,
      "grad_norm": 1.5608097314834595,
      "learning_rate": 0.00016287869071032952,
      "loss": 1.6203,
      "step": 8518
    },
    {
      "epoch": 0.30508353179221803,
      "grad_norm": 1.6867769956588745,
      "learning_rate": 0.00016286967115397655,
      "loss": 1.4592,
      "step": 8519
    },
    {
      "epoch": 0.3051193439217863,
      "grad_norm": 1.8151174783706665,
      "learning_rate": 0.00016286065075179332,
      "loss": 1.7879,
      "step": 8520
    },
    {
      "epoch": 0.30515515605135457,
      "grad_norm": 2.2028374671936035,
      "learning_rate": 0.00016285162950390104,
      "loss": 1.4099,
      "step": 8521
    },
    {
      "epoch": 0.3051909681809229,
      "grad_norm": 2.5838005542755127,
      "learning_rate": 0.00016284260741042123,
      "loss": 1.4674,
      "step": 8522
    },
    {
      "epoch": 0.30522678031049116,
      "grad_norm": 1.9016218185424805,
      "learning_rate": 0.00016283358447147516,
      "loss": 1.8284,
      "step": 8523
    },
    {
      "epoch": 0.30526259244005943,
      "grad_norm": 2.4954068660736084,
      "learning_rate": 0.0001628245606871843,
      "loss": 1.5852,
      "step": 8524
    },
    {
      "epoch": 0.30529840456962776,
      "grad_norm": 2.0137617588043213,
      "learning_rate": 0.00016281553605766998,
      "loss": 1.7631,
      "step": 8525
    },
    {
      "epoch": 0.305334216699196,
      "grad_norm": 2.0389068126678467,
      "learning_rate": 0.00016280651058305363,
      "loss": 1.661,
      "step": 8526
    },
    {
      "epoch": 0.3053700288287643,
      "grad_norm": 1.6179884672164917,
      "learning_rate": 0.00016279748426345673,
      "loss": 1.5391,
      "step": 8527
    },
    {
      "epoch": 0.30540584095833256,
      "grad_norm": 1.4547618627548218,
      "learning_rate": 0.0001627884570990007,
      "loss": 1.6447,
      "step": 8528
    },
    {
      "epoch": 0.3054416530879009,
      "grad_norm": 1.4951916933059692,
      "learning_rate": 0.0001627794290898069,
      "loss": 1.714,
      "step": 8529
    },
    {
      "epoch": 0.30547746521746916,
      "grad_norm": 1.8673044443130493,
      "learning_rate": 0.00016277040023599692,
      "loss": 1.6272,
      "step": 8530
    },
    {
      "epoch": 0.3055132773470374,
      "grad_norm": 1.5993306636810303,
      "learning_rate": 0.00016276137053769217,
      "loss": 1.3022,
      "step": 8531
    },
    {
      "epoch": 0.30554908947660575,
      "grad_norm": 1.4907748699188232,
      "learning_rate": 0.0001627523399950141,
      "loss": 1.606,
      "step": 8532
    },
    {
      "epoch": 0.305584901606174,
      "grad_norm": 1.347849726676941,
      "learning_rate": 0.00016274330860808426,
      "loss": 1.5253,
      "step": 8533
    },
    {
      "epoch": 0.3056207137357423,
      "grad_norm": 1.296706199645996,
      "learning_rate": 0.00016273427637702415,
      "loss": 1.1995,
      "step": 8534
    },
    {
      "epoch": 0.30565652586531056,
      "grad_norm": 1.5187026262283325,
      "learning_rate": 0.00016272524330195525,
      "loss": 1.1413,
      "step": 8535
    },
    {
      "epoch": 0.3056923379948789,
      "grad_norm": 2.58811616897583,
      "learning_rate": 0.00016271620938299912,
      "loss": 1.7734,
      "step": 8536
    },
    {
      "epoch": 0.30572815012444715,
      "grad_norm": 1.5693305730819702,
      "learning_rate": 0.0001627071746202773,
      "loss": 1.6448,
      "step": 8537
    },
    {
      "epoch": 0.3057639622540154,
      "grad_norm": 1.9932421445846558,
      "learning_rate": 0.00016269813901391132,
      "loss": 1.6241,
      "step": 8538
    },
    {
      "epoch": 0.30579977438358374,
      "grad_norm": 2.0719470977783203,
      "learning_rate": 0.00016268910256402277,
      "loss": 1.5669,
      "step": 8539
    },
    {
      "epoch": 0.305835586513152,
      "grad_norm": 1.9695795774459839,
      "learning_rate": 0.00016268006527073322,
      "loss": 1.5724,
      "step": 8540
    },
    {
      "epoch": 0.3058713986427203,
      "grad_norm": 1.7500296831130981,
      "learning_rate": 0.00016267102713416417,
      "loss": 1.3741,
      "step": 8541
    },
    {
      "epoch": 0.30590721077228855,
      "grad_norm": 1.8860368728637695,
      "learning_rate": 0.00016266198815443738,
      "loss": 1.4433,
      "step": 8542
    },
    {
      "epoch": 0.3059430229018569,
      "grad_norm": 1.7154771089553833,
      "learning_rate": 0.00016265294833167434,
      "loss": 1.5552,
      "step": 8543
    },
    {
      "epoch": 0.30597883503142514,
      "grad_norm": 1.621751070022583,
      "learning_rate": 0.0001626439076659967,
      "loss": 1.4277,
      "step": 8544
    },
    {
      "epoch": 0.3060146471609934,
      "grad_norm": 1.4160710573196411,
      "learning_rate": 0.00016263486615752606,
      "loss": 1.2813,
      "step": 8545
    },
    {
      "epoch": 0.30605045929056174,
      "grad_norm": 1.513929009437561,
      "learning_rate": 0.00016262582380638407,
      "loss": 1.4992,
      "step": 8546
    },
    {
      "epoch": 0.30608627142013,
      "grad_norm": 1.4435629844665527,
      "learning_rate": 0.00016261678061269244,
      "loss": 1.5759,
      "step": 8547
    },
    {
      "epoch": 0.3061220835496983,
      "grad_norm": 1.9000530242919922,
      "learning_rate": 0.0001626077365765728,
      "loss": 1.575,
      "step": 8548
    },
    {
      "epoch": 0.30615789567926655,
      "grad_norm": 1.5278146266937256,
      "learning_rate": 0.00016259869169814678,
      "loss": 1.8735,
      "step": 8549
    },
    {
      "epoch": 0.30619370780883487,
      "grad_norm": 1.8656506538391113,
      "learning_rate": 0.00016258964597753615,
      "loss": 1.4926,
      "step": 8550
    },
    {
      "epoch": 0.30622951993840314,
      "grad_norm": 2.041372060775757,
      "learning_rate": 0.00016258059941486259,
      "loss": 1.2913,
      "step": 8551
    },
    {
      "epoch": 0.3062653320679714,
      "grad_norm": 1.8052326440811157,
      "learning_rate": 0.00016257155201024776,
      "loss": 1.5017,
      "step": 8552
    },
    {
      "epoch": 0.30630114419753973,
      "grad_norm": 2.2449746131896973,
      "learning_rate": 0.0001625625037638134,
      "loss": 1.531,
      "step": 8553
    },
    {
      "epoch": 0.306336956327108,
      "grad_norm": 1.6481692790985107,
      "learning_rate": 0.00016255345467568126,
      "loss": 1.5007,
      "step": 8554
    },
    {
      "epoch": 0.30637276845667627,
      "grad_norm": 2.4112162590026855,
      "learning_rate": 0.00016254440474597307,
      "loss": 1.6164,
      "step": 8555
    },
    {
      "epoch": 0.30640858058624454,
      "grad_norm": 1.6740220785140991,
      "learning_rate": 0.0001625353539748106,
      "loss": 1.4353,
      "step": 8556
    },
    {
      "epoch": 0.30644439271581286,
      "grad_norm": 2.5266458988189697,
      "learning_rate": 0.00016252630236231557,
      "loss": 1.3675,
      "step": 8557
    },
    {
      "epoch": 0.30648020484538113,
      "grad_norm": 1.4388402700424194,
      "learning_rate": 0.00016251724990860983,
      "loss": 1.6966,
      "step": 8558
    },
    {
      "epoch": 0.3065160169749494,
      "grad_norm": 1.701951026916504,
      "learning_rate": 0.00016250819661381516,
      "loss": 1.5309,
      "step": 8559
    },
    {
      "epoch": 0.3065518291045177,
      "grad_norm": 1.3928033113479614,
      "learning_rate": 0.0001624991424780533,
      "loss": 1.1926,
      "step": 8560
    },
    {
      "epoch": 0.306587641234086,
      "grad_norm": 1.6623480319976807,
      "learning_rate": 0.0001624900875014461,
      "loss": 1.5179,
      "step": 8561
    },
    {
      "epoch": 0.30662345336365426,
      "grad_norm": 1.2392915487289429,
      "learning_rate": 0.0001624810316841154,
      "loss": 1.6586,
      "step": 8562
    },
    {
      "epoch": 0.30665926549322253,
      "grad_norm": 1.4448860883712769,
      "learning_rate": 0.000162471975026183,
      "loss": 1.3141,
      "step": 8563
    },
    {
      "epoch": 0.30669507762279086,
      "grad_norm": 1.6523834466934204,
      "learning_rate": 0.0001624629175277707,
      "loss": 1.4945,
      "step": 8564
    },
    {
      "epoch": 0.3067308897523591,
      "grad_norm": 1.7651172876358032,
      "learning_rate": 0.0001624538591890005,
      "loss": 1.4052,
      "step": 8565
    },
    {
      "epoch": 0.3067667018819274,
      "grad_norm": 1.7851136922836304,
      "learning_rate": 0.00016244480000999416,
      "loss": 1.2272,
      "step": 8566
    },
    {
      "epoch": 0.3068025140114957,
      "grad_norm": 1.844192385673523,
      "learning_rate": 0.0001624357399908736,
      "loss": 1.9858,
      "step": 8567
    },
    {
      "epoch": 0.306838326141064,
      "grad_norm": 1.3948429822921753,
      "learning_rate": 0.00016242667913176064,
      "loss": 1.6287,
      "step": 8568
    },
    {
      "epoch": 0.30687413827063226,
      "grad_norm": 2.166774034500122,
      "learning_rate": 0.00016241761743277726,
      "loss": 1.682,
      "step": 8569
    },
    {
      "epoch": 0.3069099504002005,
      "grad_norm": 2.1550753116607666,
      "learning_rate": 0.00016240855489404535,
      "loss": 1.3612,
      "step": 8570
    },
    {
      "epoch": 0.30694576252976885,
      "grad_norm": 1.3740218877792358,
      "learning_rate": 0.00016239949151568688,
      "loss": 1.5035,
      "step": 8571
    },
    {
      "epoch": 0.3069815746593371,
      "grad_norm": 1.7257821559906006,
      "learning_rate": 0.0001623904272978237,
      "loss": 1.5221,
      "step": 8572
    },
    {
      "epoch": 0.3070173867889054,
      "grad_norm": 2.2711541652679443,
      "learning_rate": 0.00016238136224057777,
      "loss": 1.6717,
      "step": 8573
    },
    {
      "epoch": 0.3070531989184737,
      "grad_norm": 1.7106958627700806,
      "learning_rate": 0.00016237229634407112,
      "loss": 1.8193,
      "step": 8574
    },
    {
      "epoch": 0.307089011048042,
      "grad_norm": 1.976996898651123,
      "learning_rate": 0.0001623632296084257,
      "loss": 1.6273,
      "step": 8575
    },
    {
      "epoch": 0.30712482317761025,
      "grad_norm": 2.0148656368255615,
      "learning_rate": 0.0001623541620337634,
      "loss": 1.7886,
      "step": 8576
    },
    {
      "epoch": 0.3071606353071785,
      "grad_norm": 1.4920234680175781,
      "learning_rate": 0.00016234509362020633,
      "loss": 1.6572,
      "step": 8577
    },
    {
      "epoch": 0.30719644743674684,
      "grad_norm": 2.609459638595581,
      "learning_rate": 0.00016233602436787644,
      "loss": 1.4282,
      "step": 8578
    },
    {
      "epoch": 0.3072322595663151,
      "grad_norm": 1.3255739212036133,
      "learning_rate": 0.00016232695427689575,
      "loss": 1.5997,
      "step": 8579
    },
    {
      "epoch": 0.3072680716958834,
      "grad_norm": 2.2817599773406982,
      "learning_rate": 0.00016231788334738627,
      "loss": 1.4272,
      "step": 8580
    },
    {
      "epoch": 0.3073038838254517,
      "grad_norm": 1.5442206859588623,
      "learning_rate": 0.00016230881157947006,
      "loss": 1.5387,
      "step": 8581
    },
    {
      "epoch": 0.30733969595502,
      "grad_norm": 1.3098913431167603,
      "learning_rate": 0.00016229973897326919,
      "loss": 1.5564,
      "step": 8582
    },
    {
      "epoch": 0.30737550808458824,
      "grad_norm": 1.755606770515442,
      "learning_rate": 0.00016229066552890563,
      "loss": 1.6957,
      "step": 8583
    },
    {
      "epoch": 0.3074113202141565,
      "grad_norm": 1.984049916267395,
      "learning_rate": 0.0001622815912465016,
      "loss": 1.314,
      "step": 8584
    },
    {
      "epoch": 0.30744713234372484,
      "grad_norm": 1.4831622838974,
      "learning_rate": 0.00016227251612617902,
      "loss": 1.3996,
      "step": 8585
    },
    {
      "epoch": 0.3074829444732931,
      "grad_norm": 2.159074306488037,
      "learning_rate": 0.0001622634401680601,
      "loss": 1.7296,
      "step": 8586
    },
    {
      "epoch": 0.3075187566028614,
      "grad_norm": 1.5581837892532349,
      "learning_rate": 0.0001622543633722669,
      "loss": 1.3766,
      "step": 8587
    },
    {
      "epoch": 0.3075545687324297,
      "grad_norm": 1.5768377780914307,
      "learning_rate": 0.00016224528573892153,
      "loss": 1.5968,
      "step": 8588
    },
    {
      "epoch": 0.30759038086199797,
      "grad_norm": 1.7396345138549805,
      "learning_rate": 0.00016223620726814615,
      "loss": 1.461,
      "step": 8589
    },
    {
      "epoch": 0.30762619299156624,
      "grad_norm": 2.5793988704681396,
      "learning_rate": 0.00016222712796006285,
      "loss": 1.4118,
      "step": 8590
    },
    {
      "epoch": 0.3076620051211345,
      "grad_norm": 1.5253145694732666,
      "learning_rate": 0.00016221804781479384,
      "loss": 1.4673,
      "step": 8591
    },
    {
      "epoch": 0.30769781725070283,
      "grad_norm": 1.6572800874710083,
      "learning_rate": 0.00016220896683246126,
      "loss": 1.187,
      "step": 8592
    },
    {
      "epoch": 0.3077336293802711,
      "grad_norm": 2.2144687175750732,
      "learning_rate": 0.00016219988501318727,
      "loss": 1.6095,
      "step": 8593
    },
    {
      "epoch": 0.30776944150983937,
      "grad_norm": 2.0595297813415527,
      "learning_rate": 0.00016219080235709403,
      "loss": 1.8326,
      "step": 8594
    },
    {
      "epoch": 0.3078052536394077,
      "grad_norm": 1.579040288925171,
      "learning_rate": 0.0001621817188643038,
      "loss": 1.5968,
      "step": 8595
    },
    {
      "epoch": 0.30784106576897596,
      "grad_norm": 1.8620355129241943,
      "learning_rate": 0.0001621726345349387,
      "loss": 1.5707,
      "step": 8596
    },
    {
      "epoch": 0.30787687789854423,
      "grad_norm": 1.648568868637085,
      "learning_rate": 0.00016216354936912105,
      "loss": 1.6395,
      "step": 8597
    },
    {
      "epoch": 0.3079126900281125,
      "grad_norm": 1.642183542251587,
      "learning_rate": 0.000162154463366973,
      "loss": 1.7484,
      "step": 8598
    },
    {
      "epoch": 0.3079485021576808,
      "grad_norm": 1.5687962770462036,
      "learning_rate": 0.00016214537652861687,
      "loss": 1.1456,
      "step": 8599
    },
    {
      "epoch": 0.3079843142872491,
      "grad_norm": 1.3825608491897583,
      "learning_rate": 0.00016213628885417483,
      "loss": 1.4397,
      "step": 8600
    },
    {
      "epoch": 0.30802012641681736,
      "grad_norm": 1.5594370365142822,
      "learning_rate": 0.00016212720034376914,
      "loss": 1.6672,
      "step": 8601
    },
    {
      "epoch": 0.3080559385463857,
      "grad_norm": 1.4455797672271729,
      "learning_rate": 0.00016211811099752215,
      "loss": 1.5133,
      "step": 8602
    },
    {
      "epoch": 0.30809175067595396,
      "grad_norm": 1.868511438369751,
      "learning_rate": 0.00016210902081555605,
      "loss": 1.5547,
      "step": 8603
    },
    {
      "epoch": 0.3081275628055222,
      "grad_norm": 1.375169277191162,
      "learning_rate": 0.00016209992979799326,
      "loss": 1.4245,
      "step": 8604
    },
    {
      "epoch": 0.3081633749350905,
      "grad_norm": 2.5195486545562744,
      "learning_rate": 0.00016209083794495598,
      "loss": 1.76,
      "step": 8605
    },
    {
      "epoch": 0.3081991870646588,
      "grad_norm": 1.5071332454681396,
      "learning_rate": 0.00016208174525656656,
      "loss": 1.2215,
      "step": 8606
    },
    {
      "epoch": 0.3082349991942271,
      "grad_norm": 1.7664871215820312,
      "learning_rate": 0.00016207265173294734,
      "loss": 1.6029,
      "step": 8607
    },
    {
      "epoch": 0.30827081132379536,
      "grad_norm": 2.763017416000366,
      "learning_rate": 0.00016206355737422067,
      "loss": 1.5892,
      "step": 8608
    },
    {
      "epoch": 0.3083066234533637,
      "grad_norm": 1.5861375331878662,
      "learning_rate": 0.00016205446218050892,
      "loss": 1.0838,
      "step": 8609
    },
    {
      "epoch": 0.30834243558293195,
      "grad_norm": 1.6708229780197144,
      "learning_rate": 0.00016204536615193439,
      "loss": 1.7046,
      "step": 8610
    },
    {
      "epoch": 0.3083782477125002,
      "grad_norm": 1.6125863790512085,
      "learning_rate": 0.00016203626928861948,
      "loss": 1.6065,
      "step": 8611
    },
    {
      "epoch": 0.3084140598420685,
      "grad_norm": 2.4662134647369385,
      "learning_rate": 0.00016202717159068662,
      "loss": 1.631,
      "step": 8612
    },
    {
      "epoch": 0.3084498719716368,
      "grad_norm": 1.8300012350082397,
      "learning_rate": 0.00016201807305825817,
      "loss": 1.49,
      "step": 8613
    },
    {
      "epoch": 0.3084856841012051,
      "grad_norm": 1.6964879035949707,
      "learning_rate": 0.0001620089736914565,
      "loss": 1.3984,
      "step": 8614
    },
    {
      "epoch": 0.30852149623077335,
      "grad_norm": 1.575770616531372,
      "learning_rate": 0.0001619998734904041,
      "loss": 1.6827,
      "step": 8615
    },
    {
      "epoch": 0.3085573083603416,
      "grad_norm": 1.308549165725708,
      "learning_rate": 0.00016199077245522341,
      "loss": 1.5704,
      "step": 8616
    },
    {
      "epoch": 0.30859312048990994,
      "grad_norm": 2.0342214107513428,
      "learning_rate": 0.00016198167058603682,
      "loss": 1.6039,
      "step": 8617
    },
    {
      "epoch": 0.3086289326194782,
      "grad_norm": 1.666513442993164,
      "learning_rate": 0.0001619725678829668,
      "loss": 1.5431,
      "step": 8618
    },
    {
      "epoch": 0.3086647447490465,
      "grad_norm": 1.700683832168579,
      "learning_rate": 0.0001619634643461358,
      "loss": 1.6518,
      "step": 8619
    },
    {
      "epoch": 0.3087005568786148,
      "grad_norm": 2.255946159362793,
      "learning_rate": 0.00016195435997566632,
      "loss": 1.7746,
      "step": 8620
    },
    {
      "epoch": 0.3087363690081831,
      "grad_norm": 1.617912769317627,
      "learning_rate": 0.00016194525477168087,
      "loss": 1.6655,
      "step": 8621
    },
    {
      "epoch": 0.30877218113775134,
      "grad_norm": 2.162280321121216,
      "learning_rate": 0.0001619361487343019,
      "loss": 1.5503,
      "step": 8622
    },
    {
      "epoch": 0.3088079932673196,
      "grad_norm": 1.8481221199035645,
      "learning_rate": 0.00016192704186365195,
      "loss": 1.5285,
      "step": 8623
    },
    {
      "epoch": 0.30884380539688794,
      "grad_norm": 1.9384018182754517,
      "learning_rate": 0.00016191793415985353,
      "loss": 1.9594,
      "step": 8624
    },
    {
      "epoch": 0.3088796175264562,
      "grad_norm": 1.9377477169036865,
      "learning_rate": 0.00016190882562302914,
      "loss": 1.5948,
      "step": 8625
    },
    {
      "epoch": 0.3089154296560245,
      "grad_norm": 1.4632514715194702,
      "learning_rate": 0.0001618997162533014,
      "loss": 1.6144,
      "step": 8626
    },
    {
      "epoch": 0.3089512417855928,
      "grad_norm": 1.6265308856964111,
      "learning_rate": 0.0001618906060507928,
      "loss": 1.6253,
      "step": 8627
    },
    {
      "epoch": 0.30898705391516107,
      "grad_norm": 1.8380078077316284,
      "learning_rate": 0.00016188149501562596,
      "loss": 1.287,
      "step": 8628
    },
    {
      "epoch": 0.30902286604472934,
      "grad_norm": 1.5310187339782715,
      "learning_rate": 0.00016187238314792338,
      "loss": 1.3016,
      "step": 8629
    },
    {
      "epoch": 0.3090586781742976,
      "grad_norm": 1.4253685474395752,
      "learning_rate": 0.00016186327044780772,
      "loss": 1.4775,
      "step": 8630
    },
    {
      "epoch": 0.30909449030386593,
      "grad_norm": 2.2036783695220947,
      "learning_rate": 0.00016185415691540156,
      "loss": 1.5592,
      "step": 8631
    },
    {
      "epoch": 0.3091303024334342,
      "grad_norm": 2.154386281967163,
      "learning_rate": 0.0001618450425508275,
      "loss": 1.4297,
      "step": 8632
    },
    {
      "epoch": 0.30916611456300247,
      "grad_norm": 1.7909826040267944,
      "learning_rate": 0.00016183592735420817,
      "loss": 1.3135,
      "step": 8633
    },
    {
      "epoch": 0.3092019266925708,
      "grad_norm": 1.5986603498458862,
      "learning_rate": 0.0001618268113256662,
      "loss": 1.6078,
      "step": 8634
    },
    {
      "epoch": 0.30923773882213906,
      "grad_norm": 1.9522628784179688,
      "learning_rate": 0.0001618176944653242,
      "loss": 1.6146,
      "step": 8635
    },
    {
      "epoch": 0.30927355095170733,
      "grad_norm": 1.3854929208755493,
      "learning_rate": 0.0001618085767733049,
      "loss": 1.551,
      "step": 8636
    },
    {
      "epoch": 0.3093093630812756,
      "grad_norm": 1.41969895362854,
      "learning_rate": 0.0001617994582497309,
      "loss": 1.3639,
      "step": 8637
    },
    {
      "epoch": 0.3093451752108439,
      "grad_norm": 2.2371199131011963,
      "learning_rate": 0.00016179033889472493,
      "loss": 1.5884,
      "step": 8638
    },
    {
      "epoch": 0.3093809873404122,
      "grad_norm": 1.718151330947876,
      "learning_rate": 0.00016178121870840965,
      "loss": 1.5665,
      "step": 8639
    },
    {
      "epoch": 0.30941679946998046,
      "grad_norm": 1.7268788814544678,
      "learning_rate": 0.00016177209769090774,
      "loss": 1.6803,
      "step": 8640
    },
    {
      "epoch": 0.3094526115995488,
      "grad_norm": 1.3030462265014648,
      "learning_rate": 0.00016176297584234196,
      "loss": 1.5642,
      "step": 8641
    },
    {
      "epoch": 0.30948842372911706,
      "grad_norm": 1.4846490621566772,
      "learning_rate": 0.00016175385316283502,
      "loss": 1.4474,
      "step": 8642
    },
    {
      "epoch": 0.3095242358586853,
      "grad_norm": 1.95109224319458,
      "learning_rate": 0.00016174472965250965,
      "loss": 1.6802,
      "step": 8643
    },
    {
      "epoch": 0.3095600479882536,
      "grad_norm": 2.3213067054748535,
      "learning_rate": 0.00016173560531148855,
      "loss": 1.6874,
      "step": 8644
    },
    {
      "epoch": 0.3095958601178219,
      "grad_norm": 1.6281256675720215,
      "learning_rate": 0.0001617264801398945,
      "loss": 1.6978,
      "step": 8645
    },
    {
      "epoch": 0.3096316722473902,
      "grad_norm": 1.7163335084915161,
      "learning_rate": 0.0001617173541378503,
      "loss": 1.507,
      "step": 8646
    },
    {
      "epoch": 0.30966748437695846,
      "grad_norm": 2.341641426086426,
      "learning_rate": 0.0001617082273054787,
      "loss": 1.6396,
      "step": 8647
    },
    {
      "epoch": 0.3097032965065268,
      "grad_norm": 2.265071392059326,
      "learning_rate": 0.00016169909964290256,
      "loss": 1.383,
      "step": 8648
    },
    {
      "epoch": 0.30973910863609505,
      "grad_norm": 2.135606288909912,
      "learning_rate": 0.00016168997115024458,
      "loss": 1.6133,
      "step": 8649
    },
    {
      "epoch": 0.3097749207656633,
      "grad_norm": 1.6686729192733765,
      "learning_rate": 0.0001616808418276276,
      "loss": 1.4704,
      "step": 8650
    },
    {
      "epoch": 0.3098107328952316,
      "grad_norm": 1.7093509435653687,
      "learning_rate": 0.00016167171167517447,
      "loss": 1.6578,
      "step": 8651
    },
    {
      "epoch": 0.3098465450247999,
      "grad_norm": 1.4961060285568237,
      "learning_rate": 0.00016166258069300803,
      "loss": 1.5668,
      "step": 8652
    },
    {
      "epoch": 0.3098823571543682,
      "grad_norm": 1.3876949548721313,
      "learning_rate": 0.00016165344888125106,
      "loss": 1.6351,
      "step": 8653
    },
    {
      "epoch": 0.30991816928393645,
      "grad_norm": 1.5020776987075806,
      "learning_rate": 0.00016164431624002647,
      "loss": 1.4318,
      "step": 8654
    },
    {
      "epoch": 0.3099539814135048,
      "grad_norm": 1.5066617727279663,
      "learning_rate": 0.00016163518276945715,
      "loss": 1.6516,
      "step": 8655
    },
    {
      "epoch": 0.30998979354307304,
      "grad_norm": 1.377149224281311,
      "learning_rate": 0.00016162604846966594,
      "loss": 1.5357,
      "step": 8656
    },
    {
      "epoch": 0.3100256056726413,
      "grad_norm": 1.648619294166565,
      "learning_rate": 0.0001616169133407757,
      "loss": 1.9226,
      "step": 8657
    },
    {
      "epoch": 0.3100614178022096,
      "grad_norm": 1.3753312826156616,
      "learning_rate": 0.00016160777738290945,
      "loss": 1.5631,
      "step": 8658
    },
    {
      "epoch": 0.3100972299317779,
      "grad_norm": 1.6483653783798218,
      "learning_rate": 0.00016159864059618997,
      "loss": 1.762,
      "step": 8659
    },
    {
      "epoch": 0.3101330420613462,
      "grad_norm": 1.7791167497634888,
      "learning_rate": 0.00016158950298074022,
      "loss": 1.575,
      "step": 8660
    },
    {
      "epoch": 0.31016885419091444,
      "grad_norm": 2.1656835079193115,
      "learning_rate": 0.00016158036453668318,
      "loss": 1.6359,
      "step": 8661
    },
    {
      "epoch": 0.31020466632048277,
      "grad_norm": 1.743486762046814,
      "learning_rate": 0.00016157122526414176,
      "loss": 1.8204,
      "step": 8662
    },
    {
      "epoch": 0.31024047845005104,
      "grad_norm": 1.796557903289795,
      "learning_rate": 0.00016156208516323895,
      "loss": 1.3592,
      "step": 8663
    },
    {
      "epoch": 0.3102762905796193,
      "grad_norm": 1.2790051698684692,
      "learning_rate": 0.00016155294423409768,
      "loss": 1.476,
      "step": 8664
    },
    {
      "epoch": 0.3103121027091876,
      "grad_norm": 1.7429996728897095,
      "learning_rate": 0.00016154380247684094,
      "loss": 1.5231,
      "step": 8665
    },
    {
      "epoch": 0.3103479148387559,
      "grad_norm": 2.2776238918304443,
      "learning_rate": 0.00016153465989159172,
      "loss": 1.5724,
      "step": 8666
    },
    {
      "epoch": 0.31038372696832417,
      "grad_norm": 1.5032392740249634,
      "learning_rate": 0.00016152551647847304,
      "loss": 1.7274,
      "step": 8667
    },
    {
      "epoch": 0.31041953909789244,
      "grad_norm": 1.6818513870239258,
      "learning_rate": 0.00016151637223760785,
      "loss": 1.4738,
      "step": 8668
    },
    {
      "epoch": 0.31045535122746076,
      "grad_norm": 1.457304835319519,
      "learning_rate": 0.0001615072271691193,
      "loss": 1.5367,
      "step": 8669
    },
    {
      "epoch": 0.31049116335702903,
      "grad_norm": 1.749411702156067,
      "learning_rate": 0.00016149808127313025,
      "loss": 1.6854,
      "step": 8670
    },
    {
      "epoch": 0.3105269754865973,
      "grad_norm": 1.698837399482727,
      "learning_rate": 0.00016148893454976393,
      "loss": 1.5631,
      "step": 8671
    },
    {
      "epoch": 0.31056278761616557,
      "grad_norm": 1.9084186553955078,
      "learning_rate": 0.00016147978699914325,
      "loss": 1.3116,
      "step": 8672
    },
    {
      "epoch": 0.3105985997457339,
      "grad_norm": 1.5788426399230957,
      "learning_rate": 0.00016147063862139138,
      "loss": 1.5894,
      "step": 8673
    },
    {
      "epoch": 0.31063441187530216,
      "grad_norm": 1.705234169960022,
      "learning_rate": 0.00016146148941663136,
      "loss": 1.4553,
      "step": 8674
    },
    {
      "epoch": 0.31067022400487043,
      "grad_norm": 1.3608901500701904,
      "learning_rate": 0.00016145233938498626,
      "loss": 1.5954,
      "step": 8675
    },
    {
      "epoch": 0.31070603613443876,
      "grad_norm": 1.920832872390747,
      "learning_rate": 0.00016144318852657921,
      "loss": 1.6358,
      "step": 8676
    },
    {
      "epoch": 0.310741848264007,
      "grad_norm": 1.759215235710144,
      "learning_rate": 0.00016143403684153328,
      "loss": 1.6726,
      "step": 8677
    },
    {
      "epoch": 0.3107776603935753,
      "grad_norm": 1.8007652759552002,
      "learning_rate": 0.00016142488432997168,
      "loss": 1.6163,
      "step": 8678
    },
    {
      "epoch": 0.31081347252314356,
      "grad_norm": 2.198650598526001,
      "learning_rate": 0.00016141573099201744,
      "loss": 1.4736,
      "step": 8679
    },
    {
      "epoch": 0.3108492846527119,
      "grad_norm": 1.4647966623306274,
      "learning_rate": 0.00016140657682779384,
      "loss": 1.2301,
      "step": 8680
    },
    {
      "epoch": 0.31088509678228016,
      "grad_norm": 1.752133846282959,
      "learning_rate": 0.0001613974218374239,
      "loss": 1.558,
      "step": 8681
    },
    {
      "epoch": 0.3109209089118484,
      "grad_norm": 1.8326823711395264,
      "learning_rate": 0.00016138826602103085,
      "loss": 1.8069,
      "step": 8682
    },
    {
      "epoch": 0.31095672104141675,
      "grad_norm": 1.3511853218078613,
      "learning_rate": 0.00016137910937873788,
      "loss": 1.5138,
      "step": 8683
    },
    {
      "epoch": 0.310992533170985,
      "grad_norm": 1.8145192861557007,
      "learning_rate": 0.00016136995191066818,
      "loss": 1.7591,
      "step": 8684
    },
    {
      "epoch": 0.3110283453005533,
      "grad_norm": 1.5319371223449707,
      "learning_rate": 0.0001613607936169449,
      "loss": 1.8161,
      "step": 8685
    },
    {
      "epoch": 0.31106415743012156,
      "grad_norm": 1.9978666305541992,
      "learning_rate": 0.00016135163449769132,
      "loss": 1.4615,
      "step": 8686
    },
    {
      "epoch": 0.3110999695596899,
      "grad_norm": 2.378784418106079,
      "learning_rate": 0.0001613424745530306,
      "loss": 1.5786,
      "step": 8687
    },
    {
      "epoch": 0.31113578168925815,
      "grad_norm": 2.055859327316284,
      "learning_rate": 0.00016133331378308604,
      "loss": 1.6159,
      "step": 8688
    },
    {
      "epoch": 0.3111715938188264,
      "grad_norm": 1.796878695487976,
      "learning_rate": 0.00016132415218798085,
      "loss": 1.6153,
      "step": 8689
    },
    {
      "epoch": 0.31120740594839474,
      "grad_norm": 1.5123406648635864,
      "learning_rate": 0.0001613149897678383,
      "loss": 1.556,
      "step": 8690
    },
    {
      "epoch": 0.311243218077963,
      "grad_norm": 2.314534902572632,
      "learning_rate": 0.00016130582652278163,
      "loss": 1.4182,
      "step": 8691
    },
    {
      "epoch": 0.3112790302075313,
      "grad_norm": 1.2278296947479248,
      "learning_rate": 0.0001612966624529342,
      "loss": 1.6972,
      "step": 8692
    },
    {
      "epoch": 0.31131484233709955,
      "grad_norm": 1.6352144479751587,
      "learning_rate": 0.0001612874975584192,
      "loss": 1.4284,
      "step": 8693
    },
    {
      "epoch": 0.3113506544666679,
      "grad_norm": 1.5539416074752808,
      "learning_rate": 0.00016127833183936,
      "loss": 1.524,
      "step": 8694
    },
    {
      "epoch": 0.31138646659623614,
      "grad_norm": 2.1352217197418213,
      "learning_rate": 0.00016126916529587987,
      "loss": 1.5975,
      "step": 8695
    },
    {
      "epoch": 0.3114222787258044,
      "grad_norm": 1.6609954833984375,
      "learning_rate": 0.00016125999792810213,
      "loss": 1.26,
      "step": 8696
    },
    {
      "epoch": 0.31145809085537274,
      "grad_norm": 1.37396240234375,
      "learning_rate": 0.00016125082973615017,
      "loss": 1.2051,
      "step": 8697
    },
    {
      "epoch": 0.311493902984941,
      "grad_norm": 2.057098865509033,
      "learning_rate": 0.00016124166072014728,
      "loss": 1.5141,
      "step": 8698
    },
    {
      "epoch": 0.3115297151145093,
      "grad_norm": 1.9792077541351318,
      "learning_rate": 0.00016123249088021688,
      "loss": 1.6889,
      "step": 8699
    },
    {
      "epoch": 0.31156552724407754,
      "grad_norm": 1.6447113752365112,
      "learning_rate": 0.00016122332021648226,
      "loss": 1.4876,
      "step": 8700
    },
    {
      "epoch": 0.31160133937364587,
      "grad_norm": 1.9308607578277588,
      "learning_rate": 0.00016121414872906687,
      "loss": 1.4819,
      "step": 8701
    },
    {
      "epoch": 0.31163715150321414,
      "grad_norm": 1.500744342803955,
      "learning_rate": 0.00016120497641809408,
      "loss": 1.5922,
      "step": 8702
    },
    {
      "epoch": 0.3116729636327824,
      "grad_norm": 2.0981481075286865,
      "learning_rate": 0.00016119580328368725,
      "loss": 1.7535,
      "step": 8703
    },
    {
      "epoch": 0.31170877576235073,
      "grad_norm": 1.7200511693954468,
      "learning_rate": 0.0001611866293259698,
      "loss": 1.3421,
      "step": 8704
    },
    {
      "epoch": 0.311744587891919,
      "grad_norm": 1.9420268535614014,
      "learning_rate": 0.00016117745454506522,
      "loss": 1.446,
      "step": 8705
    },
    {
      "epoch": 0.31178040002148727,
      "grad_norm": 2.217729091644287,
      "learning_rate": 0.00016116827894109686,
      "loss": 1.2806,
      "step": 8706
    },
    {
      "epoch": 0.31181621215105554,
      "grad_norm": 1.7282236814498901,
      "learning_rate": 0.00016115910251418827,
      "loss": 1.5973,
      "step": 8707
    },
    {
      "epoch": 0.31185202428062386,
      "grad_norm": 1.4423658847808838,
      "learning_rate": 0.0001611499252644628,
      "loss": 1.5761,
      "step": 8708
    },
    {
      "epoch": 0.31188783641019213,
      "grad_norm": 1.5264742374420166,
      "learning_rate": 0.00016114074719204396,
      "loss": 1.5954,
      "step": 8709
    },
    {
      "epoch": 0.3119236485397604,
      "grad_norm": 2.1491074562072754,
      "learning_rate": 0.00016113156829705526,
      "loss": 1.5177,
      "step": 8710
    },
    {
      "epoch": 0.3119594606693287,
      "grad_norm": 1.6613092422485352,
      "learning_rate": 0.00016112238857962017,
      "loss": 1.6098,
      "step": 8711
    },
    {
      "epoch": 0.311995272798897,
      "grad_norm": 1.7055860757827759,
      "learning_rate": 0.00016111320803986217,
      "loss": 1.7401,
      "step": 8712
    },
    {
      "epoch": 0.31203108492846526,
      "grad_norm": 1.538976788520813,
      "learning_rate": 0.00016110402667790475,
      "loss": 1.7231,
      "step": 8713
    },
    {
      "epoch": 0.31206689705803353,
      "grad_norm": 2.240173816680908,
      "learning_rate": 0.0001610948444938715,
      "loss": 1.8553,
      "step": 8714
    },
    {
      "epoch": 0.31210270918760186,
      "grad_norm": 1.49239182472229,
      "learning_rate": 0.00016108566148788594,
      "loss": 1.2735,
      "step": 8715
    },
    {
      "epoch": 0.3121385213171701,
      "grad_norm": 2.0355594158172607,
      "learning_rate": 0.00016107647766007159,
      "loss": 1.5694,
      "step": 8716
    },
    {
      "epoch": 0.3121743334467384,
      "grad_norm": 2.6552181243896484,
      "learning_rate": 0.000161067293010552,
      "loss": 1.4701,
      "step": 8717
    },
    {
      "epoch": 0.3122101455763067,
      "grad_norm": 1.752543568611145,
      "learning_rate": 0.00016105810753945076,
      "loss": 1.6328,
      "step": 8718
    },
    {
      "epoch": 0.312245957705875,
      "grad_norm": 1.854783296585083,
      "learning_rate": 0.00016104892124689147,
      "loss": 1.723,
      "step": 8719
    },
    {
      "epoch": 0.31228176983544326,
      "grad_norm": 1.347758412361145,
      "learning_rate": 0.00016103973413299767,
      "loss": 1.4929,
      "step": 8720
    },
    {
      "epoch": 0.3123175819650115,
      "grad_norm": 1.979087471961975,
      "learning_rate": 0.00016103054619789298,
      "loss": 1.639,
      "step": 8721
    },
    {
      "epoch": 0.31235339409457985,
      "grad_norm": 2.2528393268585205,
      "learning_rate": 0.00016102135744170098,
      "loss": 1.7851,
      "step": 8722
    },
    {
      "epoch": 0.3123892062241481,
      "grad_norm": 2.4195220470428467,
      "learning_rate": 0.00016101216786454538,
      "loss": 1.6844,
      "step": 8723
    },
    {
      "epoch": 0.3124250183537164,
      "grad_norm": 2.5333340167999268,
      "learning_rate": 0.00016100297746654975,
      "loss": 1.7123,
      "step": 8724
    },
    {
      "epoch": 0.3124608304832847,
      "grad_norm": 1.6024668216705322,
      "learning_rate": 0.00016099378624783773,
      "loss": 1.7999,
      "step": 8725
    },
    {
      "epoch": 0.312496642612853,
      "grad_norm": 1.7024332284927368,
      "learning_rate": 0.00016098459420853302,
      "loss": 1.881,
      "step": 8726
    },
    {
      "epoch": 0.31253245474242125,
      "grad_norm": 2.0940849781036377,
      "learning_rate": 0.00016097540134875924,
      "loss": 1.5664,
      "step": 8727
    },
    {
      "epoch": 0.3125682668719895,
      "grad_norm": 1.422271728515625,
      "learning_rate": 0.00016096620766864011,
      "loss": 1.5992,
      "step": 8728
    },
    {
      "epoch": 0.31260407900155784,
      "grad_norm": 2.1082029342651367,
      "learning_rate": 0.00016095701316829925,
      "loss": 1.2487,
      "step": 8729
    },
    {
      "epoch": 0.3126398911311261,
      "grad_norm": 1.6328003406524658,
      "learning_rate": 0.00016094781784786044,
      "loss": 1.5948,
      "step": 8730
    },
    {
      "epoch": 0.3126757032606944,
      "grad_norm": 1.901384949684143,
      "learning_rate": 0.00016093862170744733,
      "loss": 1.5415,
      "step": 8731
    },
    {
      "epoch": 0.3127115153902627,
      "grad_norm": 1.7360332012176514,
      "learning_rate": 0.00016092942474718372,
      "loss": 1.7264,
      "step": 8732
    },
    {
      "epoch": 0.312747327519831,
      "grad_norm": 1.5709059238433838,
      "learning_rate": 0.00016092022696719327,
      "loss": 1.798,
      "step": 8733
    },
    {
      "epoch": 0.31278313964939924,
      "grad_norm": 1.9449841976165771,
      "learning_rate": 0.00016091102836759974,
      "loss": 1.915,
      "step": 8734
    },
    {
      "epoch": 0.3128189517789675,
      "grad_norm": 1.2012020349502563,
      "learning_rate": 0.00016090182894852687,
      "loss": 1.1363,
      "step": 8735
    },
    {
      "epoch": 0.31285476390853584,
      "grad_norm": 2.4898271560668945,
      "learning_rate": 0.0001608926287100985,
      "loss": 1.7112,
      "step": 8736
    },
    {
      "epoch": 0.3128905760381041,
      "grad_norm": 2.0908682346343994,
      "learning_rate": 0.00016088342765243832,
      "loss": 1.8249,
      "step": 8737
    },
    {
      "epoch": 0.3129263881676724,
      "grad_norm": 2.0061733722686768,
      "learning_rate": 0.00016087422577567016,
      "loss": 1.538,
      "step": 8738
    },
    {
      "epoch": 0.3129622002972407,
      "grad_norm": 1.3970668315887451,
      "learning_rate": 0.00016086502307991783,
      "loss": 1.4419,
      "step": 8739
    },
    {
      "epoch": 0.31299801242680897,
      "grad_norm": 1.6271003484725952,
      "learning_rate": 0.0001608558195653051,
      "loss": 1.2525,
      "step": 8740
    },
    {
      "epoch": 0.31303382455637724,
      "grad_norm": 1.6746578216552734,
      "learning_rate": 0.00016084661523195585,
      "loss": 1.7385,
      "step": 8741
    },
    {
      "epoch": 0.3130696366859455,
      "grad_norm": 2.0202746391296387,
      "learning_rate": 0.00016083741007999388,
      "loss": 1.5669,
      "step": 8742
    },
    {
      "epoch": 0.31310544881551383,
      "grad_norm": 1.568692684173584,
      "learning_rate": 0.00016082820410954297,
      "loss": 1.6994,
      "step": 8743
    },
    {
      "epoch": 0.3131412609450821,
      "grad_norm": 1.844053030014038,
      "learning_rate": 0.0001608189973207271,
      "loss": 1.8135,
      "step": 8744
    },
    {
      "epoch": 0.31317707307465037,
      "grad_norm": 3.1667978763580322,
      "learning_rate": 0.00016080978971367004,
      "loss": 1.8685,
      "step": 8745
    },
    {
      "epoch": 0.3132128852042187,
      "grad_norm": 2.173363447189331,
      "learning_rate": 0.00016080058128849572,
      "loss": 1.4853,
      "step": 8746
    },
    {
      "epoch": 0.31324869733378696,
      "grad_norm": 1.3892388343811035,
      "learning_rate": 0.000160791372045328,
      "loss": 1.3848,
      "step": 8747
    },
    {
      "epoch": 0.31328450946335523,
      "grad_norm": 1.9529330730438232,
      "learning_rate": 0.00016078216198429077,
      "loss": 1.4344,
      "step": 8748
    },
    {
      "epoch": 0.3133203215929235,
      "grad_norm": 1.5327616930007935,
      "learning_rate": 0.00016077295110550796,
      "loss": 1.4037,
      "step": 8749
    },
    {
      "epoch": 0.3133561337224918,
      "grad_norm": 1.5518343448638916,
      "learning_rate": 0.0001607637394091035,
      "loss": 1.4539,
      "step": 8750
    },
    {
      "epoch": 0.3133919458520601,
      "grad_norm": 1.8877317905426025,
      "learning_rate": 0.00016075452689520128,
      "loss": 1.7284,
      "step": 8751
    },
    {
      "epoch": 0.31342775798162836,
      "grad_norm": 2.0070106983184814,
      "learning_rate": 0.0001607453135639253,
      "loss": 1.7028,
      "step": 8752
    },
    {
      "epoch": 0.3134635701111967,
      "grad_norm": 1.7277096509933472,
      "learning_rate": 0.00016073609941539944,
      "loss": 1.3793,
      "step": 8753
    },
    {
      "epoch": 0.31349938224076496,
      "grad_norm": 2.4877145290374756,
      "learning_rate": 0.00016072688444974774,
      "loss": 1.5569,
      "step": 8754
    },
    {
      "epoch": 0.3135351943703332,
      "grad_norm": 1.5935946702957153,
      "learning_rate": 0.00016071766866709413,
      "loss": 1.3458,
      "step": 8755
    },
    {
      "epoch": 0.3135710064999015,
      "grad_norm": 1.6100068092346191,
      "learning_rate": 0.00016070845206756263,
      "loss": 1.7064,
      "step": 8756
    },
    {
      "epoch": 0.3136068186294698,
      "grad_norm": 1.5823991298675537,
      "learning_rate": 0.00016069923465127718,
      "loss": 1.6565,
      "step": 8757
    },
    {
      "epoch": 0.3136426307590381,
      "grad_norm": 1.5896168947219849,
      "learning_rate": 0.00016069001641836182,
      "loss": 1.2576,
      "step": 8758
    },
    {
      "epoch": 0.31367844288860636,
      "grad_norm": 1.5499836206436157,
      "learning_rate": 0.00016068079736894058,
      "loss": 1.642,
      "step": 8759
    },
    {
      "epoch": 0.3137142550181747,
      "grad_norm": 1.5305835008621216,
      "learning_rate": 0.00016067157750313752,
      "loss": 1.3835,
      "step": 8760
    },
    {
      "epoch": 0.31375006714774295,
      "grad_norm": 1.4311829805374146,
      "learning_rate": 0.00016066235682107662,
      "loss": 1.6501,
      "step": 8761
    },
    {
      "epoch": 0.3137858792773112,
      "grad_norm": 1.4986733198165894,
      "learning_rate": 0.00016065313532288196,
      "loss": 1.44,
      "step": 8762
    },
    {
      "epoch": 0.3138216914068795,
      "grad_norm": 1.811113953590393,
      "learning_rate": 0.0001606439130086776,
      "loss": 1.341,
      "step": 8763
    },
    {
      "epoch": 0.3138575035364478,
      "grad_norm": 1.6800071001052856,
      "learning_rate": 0.00016063468987858763,
      "loss": 1.6177,
      "step": 8764
    },
    {
      "epoch": 0.3138933156660161,
      "grad_norm": 1.6395264863967896,
      "learning_rate": 0.00016062546593273612,
      "loss": 1.3695,
      "step": 8765
    },
    {
      "epoch": 0.31392912779558435,
      "grad_norm": 1.7987147569656372,
      "learning_rate": 0.00016061624117124715,
      "loss": 1.634,
      "step": 8766
    },
    {
      "epoch": 0.3139649399251527,
      "grad_norm": 1.6628080606460571,
      "learning_rate": 0.00016060701559424484,
      "loss": 1.509,
      "step": 8767
    },
    {
      "epoch": 0.31400075205472094,
      "grad_norm": 1.6090434789657593,
      "learning_rate": 0.00016059778920185332,
      "loss": 1.3413,
      "step": 8768
    },
    {
      "epoch": 0.3140365641842892,
      "grad_norm": 1.616045355796814,
      "learning_rate": 0.0001605885619941967,
      "loss": 1.6345,
      "step": 8769
    },
    {
      "epoch": 0.3140723763138575,
      "grad_norm": 2.1155946254730225,
      "learning_rate": 0.00016057933397139914,
      "loss": 1.4609,
      "step": 8770
    },
    {
      "epoch": 0.3141081884434258,
      "grad_norm": 1.705063819885254,
      "learning_rate": 0.0001605701051335848,
      "loss": 1.677,
      "step": 8771
    },
    {
      "epoch": 0.3141440005729941,
      "grad_norm": 1.82172691822052,
      "learning_rate": 0.00016056087548087784,
      "loss": 1.5523,
      "step": 8772
    },
    {
      "epoch": 0.31417981270256234,
      "grad_norm": 1.3792582750320435,
      "learning_rate": 0.00016055164501340235,
      "loss": 1.5636,
      "step": 8773
    },
    {
      "epoch": 0.31421562483213067,
      "grad_norm": 1.869604468345642,
      "learning_rate": 0.00016054241373128264,
      "loss": 1.7203,
      "step": 8774
    },
    {
      "epoch": 0.31425143696169894,
      "grad_norm": 1.405261754989624,
      "learning_rate": 0.00016053318163464285,
      "loss": 1.5212,
      "step": 8775
    },
    {
      "epoch": 0.3142872490912672,
      "grad_norm": 1.8906536102294922,
      "learning_rate": 0.00016052394872360717,
      "loss": 1.6859,
      "step": 8776
    },
    {
      "epoch": 0.3143230612208355,
      "grad_norm": 1.6787978410720825,
      "learning_rate": 0.00016051471499829983,
      "loss": 1.5595,
      "step": 8777
    },
    {
      "epoch": 0.3143588733504038,
      "grad_norm": 1.543586254119873,
      "learning_rate": 0.00016050548045884501,
      "loss": 1.5273,
      "step": 8778
    },
    {
      "epoch": 0.31439468547997207,
      "grad_norm": 1.6794708967208862,
      "learning_rate": 0.00016049624510536704,
      "loss": 1.7836,
      "step": 8779
    },
    {
      "epoch": 0.31443049760954034,
      "grad_norm": 1.616531252861023,
      "learning_rate": 0.00016048700893799014,
      "loss": 1.3527,
      "step": 8780
    },
    {
      "epoch": 0.31446630973910866,
      "grad_norm": 1.3275195360183716,
      "learning_rate": 0.00016047777195683858,
      "loss": 1.3845,
      "step": 8781
    },
    {
      "epoch": 0.31450212186867693,
      "grad_norm": 2.247191905975342,
      "learning_rate": 0.00016046853416203655,
      "loss": 1.394,
      "step": 8782
    },
    {
      "epoch": 0.3145379339982452,
      "grad_norm": 1.556274175643921,
      "learning_rate": 0.0001604592955537084,
      "loss": 1.7258,
      "step": 8783
    },
    {
      "epoch": 0.31457374612781347,
      "grad_norm": 1.2704604864120483,
      "learning_rate": 0.00016045005613197843,
      "loss": 1.2194,
      "step": 8784
    },
    {
      "epoch": 0.3146095582573818,
      "grad_norm": 2.084599494934082,
      "learning_rate": 0.00016044081589697092,
      "loss": 1.7761,
      "step": 8785
    },
    {
      "epoch": 0.31464537038695006,
      "grad_norm": 1.5749155282974243,
      "learning_rate": 0.0001604315748488102,
      "loss": 1.4076,
      "step": 8786
    },
    {
      "epoch": 0.31468118251651833,
      "grad_norm": 1.5411909818649292,
      "learning_rate": 0.00016042233298762062,
      "loss": 1.5305,
      "step": 8787
    },
    {
      "epoch": 0.31471699464608666,
      "grad_norm": 1.7500187158584595,
      "learning_rate": 0.00016041309031352644,
      "loss": 1.3224,
      "step": 8788
    },
    {
      "epoch": 0.3147528067756549,
      "grad_norm": 1.9515713453292847,
      "learning_rate": 0.00016040384682665214,
      "loss": 1.5637,
      "step": 8789
    },
    {
      "epoch": 0.3147886189052232,
      "grad_norm": 1.4233375787734985,
      "learning_rate": 0.0001603946025271219,
      "loss": 1.6988,
      "step": 8790
    },
    {
      "epoch": 0.31482443103479146,
      "grad_norm": 1.8382158279418945,
      "learning_rate": 0.00016038535741506025,
      "loss": 1.5435,
      "step": 8791
    },
    {
      "epoch": 0.3148602431643598,
      "grad_norm": 1.6217607259750366,
      "learning_rate": 0.00016037611149059147,
      "loss": 1.4189,
      "step": 8792
    },
    {
      "epoch": 0.31489605529392806,
      "grad_norm": 1.7804478406906128,
      "learning_rate": 0.00016036686475384002,
      "loss": 1.3143,
      "step": 8793
    },
    {
      "epoch": 0.3149318674234963,
      "grad_norm": 1.3662558794021606,
      "learning_rate": 0.00016035761720493023,
      "loss": 1.4369,
      "step": 8794
    },
    {
      "epoch": 0.31496767955306465,
      "grad_norm": 1.6197288036346436,
      "learning_rate": 0.0001603483688439866,
      "loss": 1.4308,
      "step": 8795
    },
    {
      "epoch": 0.3150034916826329,
      "grad_norm": 2.1956467628479004,
      "learning_rate": 0.00016033911967113347,
      "loss": 1.4458,
      "step": 8796
    },
    {
      "epoch": 0.3150393038122012,
      "grad_norm": 1.8377485275268555,
      "learning_rate": 0.00016032986968649536,
      "loss": 1.6501,
      "step": 8797
    },
    {
      "epoch": 0.31507511594176946,
      "grad_norm": 2.408724546432495,
      "learning_rate": 0.00016032061889019662,
      "loss": 1.2958,
      "step": 8798
    },
    {
      "epoch": 0.3151109280713378,
      "grad_norm": 1.6144731044769287,
      "learning_rate": 0.00016031136728236184,
      "loss": 1.5546,
      "step": 8799
    },
    {
      "epoch": 0.31514674020090605,
      "grad_norm": 1.4665641784667969,
      "learning_rate": 0.00016030211486311533,
      "loss": 1.4769,
      "step": 8800
    },
    {
      "epoch": 0.3151825523304743,
      "grad_norm": 1.562260389328003,
      "learning_rate": 0.0001602928616325817,
      "loss": 1.8916,
      "step": 8801
    },
    {
      "epoch": 0.31521836446004264,
      "grad_norm": 1.9326006174087524,
      "learning_rate": 0.00016028360759088534,
      "loss": 1.6079,
      "step": 8802
    },
    {
      "epoch": 0.3152541765896109,
      "grad_norm": 1.7705698013305664,
      "learning_rate": 0.00016027435273815085,
      "loss": 1.4926,
      "step": 8803
    },
    {
      "epoch": 0.3152899887191792,
      "grad_norm": 1.473935842514038,
      "learning_rate": 0.00016026509707450266,
      "loss": 1.379,
      "step": 8804
    },
    {
      "epoch": 0.31532580084874745,
      "grad_norm": 3.5197129249572754,
      "learning_rate": 0.00016025584060006532,
      "loss": 1.5021,
      "step": 8805
    },
    {
      "epoch": 0.3153616129783158,
      "grad_norm": 1.7519930601119995,
      "learning_rate": 0.0001602465833149634,
      "loss": 1.5798,
      "step": 8806
    },
    {
      "epoch": 0.31539742510788404,
      "grad_norm": 1.7574677467346191,
      "learning_rate": 0.0001602373252193214,
      "loss": 1.5769,
      "step": 8807
    },
    {
      "epoch": 0.3154332372374523,
      "grad_norm": 1.7921092510223389,
      "learning_rate": 0.0001602280663132639,
      "loss": 1.31,
      "step": 8808
    },
    {
      "epoch": 0.31546904936702064,
      "grad_norm": 2.8339967727661133,
      "learning_rate": 0.00016021880659691546,
      "loss": 1.3994,
      "step": 8809
    },
    {
      "epoch": 0.3155048614965889,
      "grad_norm": 1.3064004182815552,
      "learning_rate": 0.00016020954607040065,
      "loss": 1.5455,
      "step": 8810
    },
    {
      "epoch": 0.3155406736261572,
      "grad_norm": 1.6051712036132812,
      "learning_rate": 0.00016020028473384402,
      "loss": 1.7627,
      "step": 8811
    },
    {
      "epoch": 0.31557648575572544,
      "grad_norm": 2.130141019821167,
      "learning_rate": 0.00016019102258737027,
      "loss": 1.3654,
      "step": 8812
    },
    {
      "epoch": 0.31561229788529377,
      "grad_norm": 1.2903608083724976,
      "learning_rate": 0.00016018175963110389,
      "loss": 1.4281,
      "step": 8813
    },
    {
      "epoch": 0.31564811001486204,
      "grad_norm": 2.988527536392212,
      "learning_rate": 0.00016017249586516963,
      "loss": 1.5691,
      "step": 8814
    },
    {
      "epoch": 0.3156839221444303,
      "grad_norm": 1.7782341241836548,
      "learning_rate": 0.000160163231289692,
      "loss": 1.6,
      "step": 8815
    },
    {
      "epoch": 0.3157197342739986,
      "grad_norm": 2.2343456745147705,
      "learning_rate": 0.00016015396590479575,
      "loss": 1.5319,
      "step": 8816
    },
    {
      "epoch": 0.3157555464035669,
      "grad_norm": 1.5296396017074585,
      "learning_rate": 0.00016014469971060543,
      "loss": 1.5057,
      "step": 8817
    },
    {
      "epoch": 0.31579135853313517,
      "grad_norm": 1.5707215070724487,
      "learning_rate": 0.0001601354327072458,
      "loss": 1.4725,
      "step": 8818
    },
    {
      "epoch": 0.31582717066270344,
      "grad_norm": 1.7088760137557983,
      "learning_rate": 0.00016012616489484148,
      "loss": 1.4094,
      "step": 8819
    },
    {
      "epoch": 0.31586298279227176,
      "grad_norm": 1.6975724697113037,
      "learning_rate": 0.00016011689627351712,
      "loss": 1.4261,
      "step": 8820
    },
    {
      "epoch": 0.31589879492184003,
      "grad_norm": 2.539350986480713,
      "learning_rate": 0.00016010762684339752,
      "loss": 1.4377,
      "step": 8821
    },
    {
      "epoch": 0.3159346070514083,
      "grad_norm": 1.2720330953598022,
      "learning_rate": 0.00016009835660460732,
      "loss": 1.4547,
      "step": 8822
    },
    {
      "epoch": 0.31597041918097657,
      "grad_norm": 1.6770910024642944,
      "learning_rate": 0.00016008908555727123,
      "loss": 1.469,
      "step": 8823
    },
    {
      "epoch": 0.3160062313105449,
      "grad_norm": 2.0010297298431396,
      "learning_rate": 0.00016007981370151406,
      "loss": 1.5004,
      "step": 8824
    },
    {
      "epoch": 0.31604204344011316,
      "grad_norm": 1.965331792831421,
      "learning_rate": 0.00016007054103746047,
      "loss": 1.59,
      "step": 8825
    },
    {
      "epoch": 0.31607785556968143,
      "grad_norm": 1.8841441869735718,
      "learning_rate": 0.00016006126756523524,
      "loss": 1.6099,
      "step": 8826
    },
    {
      "epoch": 0.31611366769924976,
      "grad_norm": 1.748795986175537,
      "learning_rate": 0.0001600519932849631,
      "loss": 1.3553,
      "step": 8827
    },
    {
      "epoch": 0.316149479828818,
      "grad_norm": 1.5516470670700073,
      "learning_rate": 0.00016004271819676887,
      "loss": 1.6262,
      "step": 8828
    },
    {
      "epoch": 0.3161852919583863,
      "grad_norm": 1.339726209640503,
      "learning_rate": 0.0001600334423007773,
      "loss": 1.4469,
      "step": 8829
    },
    {
      "epoch": 0.31622110408795456,
      "grad_norm": 1.936703085899353,
      "learning_rate": 0.0001600241655971132,
      "loss": 1.5799,
      "step": 8830
    },
    {
      "epoch": 0.3162569162175229,
      "grad_norm": 2.148743152618408,
      "learning_rate": 0.0001600148880859014,
      "loss": 1.5546,
      "step": 8831
    },
    {
      "epoch": 0.31629272834709116,
      "grad_norm": 2.121833562850952,
      "learning_rate": 0.0001600056097672667,
      "loss": 1.7325,
      "step": 8832
    },
    {
      "epoch": 0.3163285404766594,
      "grad_norm": 1.56114661693573,
      "learning_rate": 0.00015999633064133392,
      "loss": 1.3962,
      "step": 8833
    },
    {
      "epoch": 0.31636435260622775,
      "grad_norm": 1.717720866203308,
      "learning_rate": 0.0001599870507082279,
      "loss": 1.6272,
      "step": 8834
    },
    {
      "epoch": 0.316400164735796,
      "grad_norm": 1.254668116569519,
      "learning_rate": 0.0001599777699680735,
      "loss": 1.547,
      "step": 8835
    },
    {
      "epoch": 0.3164359768653643,
      "grad_norm": 1.9581760168075562,
      "learning_rate": 0.0001599684884209955,
      "loss": 1.5793,
      "step": 8836
    },
    {
      "epoch": 0.31647178899493256,
      "grad_norm": 1.534761905670166,
      "learning_rate": 0.00015995920606711893,
      "loss": 1.2339,
      "step": 8837
    },
    {
      "epoch": 0.3165076011245009,
      "grad_norm": 1.87445867061615,
      "learning_rate": 0.00015994992290656855,
      "loss": 1.6946,
      "step": 8838
    },
    {
      "epoch": 0.31654341325406915,
      "grad_norm": 1.2772846221923828,
      "learning_rate": 0.00015994063893946928,
      "loss": 1.609,
      "step": 8839
    },
    {
      "epoch": 0.3165792253836374,
      "grad_norm": 1.4138472080230713,
      "learning_rate": 0.000159931354165946,
      "loss": 1.6902,
      "step": 8840
    },
    {
      "epoch": 0.31661503751320574,
      "grad_norm": 1.8769257068634033,
      "learning_rate": 0.0001599220685861237,
      "loss": 1.3907,
      "step": 8841
    },
    {
      "epoch": 0.316650849642774,
      "grad_norm": 1.4753496646881104,
      "learning_rate": 0.00015991278220012727,
      "loss": 1.6795,
      "step": 8842
    },
    {
      "epoch": 0.3166866617723423,
      "grad_norm": 2.22638201713562,
      "learning_rate": 0.00015990349500808162,
      "loss": 1.538,
      "step": 8843
    },
    {
      "epoch": 0.31672247390191055,
      "grad_norm": 1.6586682796478271,
      "learning_rate": 0.00015989420701011171,
      "loss": 1.4434,
      "step": 8844
    },
    {
      "epoch": 0.3167582860314789,
      "grad_norm": 1.5222989320755005,
      "learning_rate": 0.0001598849182063425,
      "loss": 1.4534,
      "step": 8845
    },
    {
      "epoch": 0.31679409816104714,
      "grad_norm": 1.7269152402877808,
      "learning_rate": 0.00015987562859689898,
      "loss": 1.5415,
      "step": 8846
    },
    {
      "epoch": 0.3168299102906154,
      "grad_norm": 1.076930284500122,
      "learning_rate": 0.0001598663381819061,
      "loss": 1.4976,
      "step": 8847
    },
    {
      "epoch": 0.31686572242018374,
      "grad_norm": 1.478829026222229,
      "learning_rate": 0.00015985704696148885,
      "loss": 1.8695,
      "step": 8848
    },
    {
      "epoch": 0.316901534549752,
      "grad_norm": 1.7606927156448364,
      "learning_rate": 0.00015984775493577225,
      "loss": 1.5552,
      "step": 8849
    },
    {
      "epoch": 0.3169373466793203,
      "grad_norm": 1.9603705406188965,
      "learning_rate": 0.0001598384621048813,
      "loss": 1.4099,
      "step": 8850
    },
    {
      "epoch": 0.31697315880888854,
      "grad_norm": 1.4589382410049438,
      "learning_rate": 0.00015982916846894106,
      "loss": 1.4863,
      "step": 8851
    },
    {
      "epoch": 0.31700897093845687,
      "grad_norm": 1.6379142999649048,
      "learning_rate": 0.0001598198740280765,
      "loss": 1.6302,
      "step": 8852
    },
    {
      "epoch": 0.31704478306802514,
      "grad_norm": 1.3787086009979248,
      "learning_rate": 0.00015981057878241273,
      "loss": 1.2704,
      "step": 8853
    },
    {
      "epoch": 0.3170805951975934,
      "grad_norm": 1.5794938802719116,
      "learning_rate": 0.00015980128273207473,
      "loss": 1.6077,
      "step": 8854
    },
    {
      "epoch": 0.31711640732716173,
      "grad_norm": 1.6516194343566895,
      "learning_rate": 0.00015979198587718764,
      "loss": 1.4949,
      "step": 8855
    },
    {
      "epoch": 0.31715221945673,
      "grad_norm": 1.768387794494629,
      "learning_rate": 0.00015978268821787648,
      "loss": 2.0093,
      "step": 8856
    },
    {
      "epoch": 0.31718803158629827,
      "grad_norm": 1.8336304426193237,
      "learning_rate": 0.0001597733897542664,
      "loss": 1.8041,
      "step": 8857
    },
    {
      "epoch": 0.31722384371586654,
      "grad_norm": 1.474213719367981,
      "learning_rate": 0.0001597640904864824,
      "loss": 1.4373,
      "step": 8858
    },
    {
      "epoch": 0.31725965584543486,
      "grad_norm": 2.2122840881347656,
      "learning_rate": 0.00015975479041464974,
      "loss": 1.4047,
      "step": 8859
    },
    {
      "epoch": 0.31729546797500313,
      "grad_norm": 1.903743028640747,
      "learning_rate": 0.0001597454895388934,
      "loss": 1.3211,
      "step": 8860
    },
    {
      "epoch": 0.3173312801045714,
      "grad_norm": 1.590577483177185,
      "learning_rate": 0.00015973618785933858,
      "loss": 1.6325,
      "step": 8861
    },
    {
      "epoch": 0.3173670922341397,
      "grad_norm": 1.5541341304779053,
      "learning_rate": 0.00015972688537611038,
      "loss": 1.5876,
      "step": 8862
    },
    {
      "epoch": 0.317402904363708,
      "grad_norm": 1.4802968502044678,
      "learning_rate": 0.000159717582089334,
      "loss": 1.5479,
      "step": 8863
    },
    {
      "epoch": 0.31743871649327626,
      "grad_norm": 1.7659330368041992,
      "learning_rate": 0.0001597082779991346,
      "loss": 1.6208,
      "step": 8864
    },
    {
      "epoch": 0.31747452862284453,
      "grad_norm": 1.7611559629440308,
      "learning_rate": 0.0001596989731056373,
      "loss": 1.5296,
      "step": 8865
    },
    {
      "epoch": 0.31751034075241286,
      "grad_norm": 1.8107753992080688,
      "learning_rate": 0.00015968966740896736,
      "loss": 1.704,
      "step": 8866
    },
    {
      "epoch": 0.3175461528819811,
      "grad_norm": 1.5399701595306396,
      "learning_rate": 0.0001596803609092499,
      "loss": 1.4575,
      "step": 8867
    },
    {
      "epoch": 0.3175819650115494,
      "grad_norm": 1.4436416625976562,
      "learning_rate": 0.0001596710536066102,
      "loss": 1.1832,
      "step": 8868
    },
    {
      "epoch": 0.3176177771411177,
      "grad_norm": 1.613898754119873,
      "learning_rate": 0.00015966174550117342,
      "loss": 1.6416,
      "step": 8869
    },
    {
      "epoch": 0.317653589270686,
      "grad_norm": 1.4900312423706055,
      "learning_rate": 0.00015965243659306482,
      "loss": 1.4502,
      "step": 8870
    },
    {
      "epoch": 0.31768940140025426,
      "grad_norm": 1.5705586671829224,
      "learning_rate": 0.00015964312688240967,
      "loss": 1.8653,
      "step": 8871
    },
    {
      "epoch": 0.3177252135298225,
      "grad_norm": 2.103754997253418,
      "learning_rate": 0.00015963381636933312,
      "loss": 1.2867,
      "step": 8872
    },
    {
      "epoch": 0.31776102565939085,
      "grad_norm": 1.2797647714614868,
      "learning_rate": 0.00015962450505396051,
      "loss": 1.2829,
      "step": 8873
    },
    {
      "epoch": 0.3177968377889591,
      "grad_norm": 1.5395480394363403,
      "learning_rate": 0.00015961519293641714,
      "loss": 1.4187,
      "step": 8874
    },
    {
      "epoch": 0.3178326499185274,
      "grad_norm": 1.9138820171356201,
      "learning_rate": 0.0001596058800168282,
      "loss": 1.371,
      "step": 8875
    },
    {
      "epoch": 0.3178684620480957,
      "grad_norm": 1.351996898651123,
      "learning_rate": 0.00015959656629531904,
      "loss": 1.3799,
      "step": 8876
    },
    {
      "epoch": 0.317904274177664,
      "grad_norm": 1.498100757598877,
      "learning_rate": 0.00015958725177201495,
      "loss": 1.3474,
      "step": 8877
    },
    {
      "epoch": 0.31794008630723225,
      "grad_norm": 1.4798411130905151,
      "learning_rate": 0.0001595779364470413,
      "loss": 1.548,
      "step": 8878
    },
    {
      "epoch": 0.3179758984368005,
      "grad_norm": 1.4499621391296387,
      "learning_rate": 0.0001595686203205233,
      "loss": 1.7545,
      "step": 8879
    },
    {
      "epoch": 0.31801171056636884,
      "grad_norm": 1.3401906490325928,
      "learning_rate": 0.00015955930339258634,
      "loss": 1.5813,
      "step": 8880
    },
    {
      "epoch": 0.3180475226959371,
      "grad_norm": 1.535853624343872,
      "learning_rate": 0.00015954998566335583,
      "loss": 1.8776,
      "step": 8881
    },
    {
      "epoch": 0.3180833348255054,
      "grad_norm": 1.4423049688339233,
      "learning_rate": 0.00015954066713295707,
      "loss": 1.468,
      "step": 8882
    },
    {
      "epoch": 0.3181191469550737,
      "grad_norm": 1.5601565837860107,
      "learning_rate": 0.00015953134780151543,
      "loss": 1.6817,
      "step": 8883
    },
    {
      "epoch": 0.318154959084642,
      "grad_norm": 1.770087480545044,
      "learning_rate": 0.00015952202766915627,
      "loss": 1.8615,
      "step": 8884
    },
    {
      "epoch": 0.31819077121421024,
      "grad_norm": 1.8110880851745605,
      "learning_rate": 0.00015951270673600503,
      "loss": 1.1239,
      "step": 8885
    },
    {
      "epoch": 0.3182265833437785,
      "grad_norm": 1.4439191818237305,
      "learning_rate": 0.0001595033850021871,
      "loss": 1.377,
      "step": 8886
    },
    {
      "epoch": 0.31826239547334684,
      "grad_norm": 2.1991729736328125,
      "learning_rate": 0.00015949406246782785,
      "loss": 1.3046,
      "step": 8887
    },
    {
      "epoch": 0.3182982076029151,
      "grad_norm": 1.6699777841567993,
      "learning_rate": 0.00015948473913305274,
      "loss": 1.4169,
      "step": 8888
    },
    {
      "epoch": 0.3183340197324834,
      "grad_norm": 1.6651912927627563,
      "learning_rate": 0.00015947541499798721,
      "loss": 1.6165,
      "step": 8889
    },
    {
      "epoch": 0.3183698318620517,
      "grad_norm": 1.996559500694275,
      "learning_rate": 0.00015946609006275666,
      "loss": 1.3417,
      "step": 8890
    },
    {
      "epoch": 0.31840564399161997,
      "grad_norm": 1.8512824773788452,
      "learning_rate": 0.0001594567643274866,
      "loss": 1.5618,
      "step": 8891
    },
    {
      "epoch": 0.31844145612118824,
      "grad_norm": 2.2073004245758057,
      "learning_rate": 0.00015944743779230244,
      "loss": 1.4179,
      "step": 8892
    },
    {
      "epoch": 0.3184772682507565,
      "grad_norm": 1.4508863687515259,
      "learning_rate": 0.00015943811045732973,
      "loss": 1.4505,
      "step": 8893
    },
    {
      "epoch": 0.31851308038032483,
      "grad_norm": 2.824732542037964,
      "learning_rate": 0.00015942878232269388,
      "loss": 1.6473,
      "step": 8894
    },
    {
      "epoch": 0.3185488925098931,
      "grad_norm": 1.6162723302841187,
      "learning_rate": 0.00015941945338852044,
      "loss": 1.1406,
      "step": 8895
    },
    {
      "epoch": 0.31858470463946137,
      "grad_norm": 1.8925987482070923,
      "learning_rate": 0.0001594101236549349,
      "loss": 1.3241,
      "step": 8896
    },
    {
      "epoch": 0.3186205167690297,
      "grad_norm": 1.586099624633789,
      "learning_rate": 0.00015940079312206276,
      "loss": 1.3499,
      "step": 8897
    },
    {
      "epoch": 0.31865632889859796,
      "grad_norm": 1.7400895357131958,
      "learning_rate": 0.00015939146179002957,
      "loss": 1.5616,
      "step": 8898
    },
    {
      "epoch": 0.31869214102816623,
      "grad_norm": 1.7760902643203735,
      "learning_rate": 0.00015938212965896088,
      "loss": 1.4622,
      "step": 8899
    },
    {
      "epoch": 0.3187279531577345,
      "grad_norm": 1.524623990058899,
      "learning_rate": 0.00015937279672898223,
      "loss": 1.5722,
      "step": 8900
    },
    {
      "epoch": 0.3187637652873028,
      "grad_norm": 1.5912421941757202,
      "learning_rate": 0.0001593634630002192,
      "loss": 1.4204,
      "step": 8901
    },
    {
      "epoch": 0.3187995774168711,
      "grad_norm": 2.2556264400482178,
      "learning_rate": 0.00015935412847279735,
      "loss": 1.7675,
      "step": 8902
    },
    {
      "epoch": 0.31883538954643936,
      "grad_norm": 1.4288015365600586,
      "learning_rate": 0.00015934479314684224,
      "loss": 1.4929,
      "step": 8903
    },
    {
      "epoch": 0.3188712016760077,
      "grad_norm": 1.7608261108398438,
      "learning_rate": 0.00015933545702247952,
      "loss": 1.6204,
      "step": 8904
    },
    {
      "epoch": 0.31890701380557596,
      "grad_norm": 1.5694899559020996,
      "learning_rate": 0.00015932612009983475,
      "loss": 1.5306,
      "step": 8905
    },
    {
      "epoch": 0.3189428259351442,
      "grad_norm": 1.9936593770980835,
      "learning_rate": 0.00015931678237903353,
      "loss": 1.7889,
      "step": 8906
    },
    {
      "epoch": 0.3189786380647125,
      "grad_norm": 1.8415395021438599,
      "learning_rate": 0.00015930744386020152,
      "loss": 1.3515,
      "step": 8907
    },
    {
      "epoch": 0.3190144501942808,
      "grad_norm": 2.319857597351074,
      "learning_rate": 0.0001592981045434644,
      "loss": 1.338,
      "step": 8908
    },
    {
      "epoch": 0.3190502623238491,
      "grad_norm": 1.5549952983856201,
      "learning_rate": 0.0001592887644289477,
      "loss": 1.5115,
      "step": 8909
    },
    {
      "epoch": 0.31908607445341736,
      "grad_norm": 2.5317156314849854,
      "learning_rate": 0.0001592794235167772,
      "loss": 1.2207,
      "step": 8910
    },
    {
      "epoch": 0.3191218865829857,
      "grad_norm": 1.3313252925872803,
      "learning_rate": 0.00015927008180707854,
      "loss": 1.5237,
      "step": 8911
    },
    {
      "epoch": 0.31915769871255395,
      "grad_norm": 2.1463143825531006,
      "learning_rate": 0.00015926073929997735,
      "loss": 1.4248,
      "step": 8912
    },
    {
      "epoch": 0.3191935108421222,
      "grad_norm": 1.8104480504989624,
      "learning_rate": 0.00015925139599559939,
      "loss": 1.5752,
      "step": 8913
    },
    {
      "epoch": 0.3192293229716905,
      "grad_norm": 2.5905628204345703,
      "learning_rate": 0.0001592420518940703,
      "loss": 1.5152,
      "step": 8914
    },
    {
      "epoch": 0.3192651351012588,
      "grad_norm": 2.154703140258789,
      "learning_rate": 0.0001592327069955158,
      "loss": 1.8066,
      "step": 8915
    },
    {
      "epoch": 0.3193009472308271,
      "grad_norm": 1.2889586687088013,
      "learning_rate": 0.00015922336130006162,
      "loss": 1.2934,
      "step": 8916
    },
    {
      "epoch": 0.31933675936039535,
      "grad_norm": 1.5765522718429565,
      "learning_rate": 0.00015921401480783356,
      "loss": 1.7927,
      "step": 8917
    },
    {
      "epoch": 0.3193725714899637,
      "grad_norm": 1.3909289836883545,
      "learning_rate": 0.0001592046675189573,
      "loss": 1.4859,
      "step": 8918
    },
    {
      "epoch": 0.31940838361953194,
      "grad_norm": 1.4396029710769653,
      "learning_rate": 0.00015919531943355857,
      "loss": 1.3662,
      "step": 8919
    },
    {
      "epoch": 0.3194441957491002,
      "grad_norm": 2.2816967964172363,
      "learning_rate": 0.0001591859705517632,
      "loss": 1.678,
      "step": 8920
    },
    {
      "epoch": 0.3194800078786685,
      "grad_norm": 2.0098812580108643,
      "learning_rate": 0.00015917662087369693,
      "loss": 1.8597,
      "step": 8921
    },
    {
      "epoch": 0.3195158200082368,
      "grad_norm": 1.5743976831436157,
      "learning_rate": 0.0001591672703994856,
      "loss": 1.3576,
      "step": 8922
    },
    {
      "epoch": 0.3195516321378051,
      "grad_norm": 1.5148391723632812,
      "learning_rate": 0.00015915791912925493,
      "loss": 1.4273,
      "step": 8923
    },
    {
      "epoch": 0.31958744426737334,
      "grad_norm": 1.6980741024017334,
      "learning_rate": 0.00015914856706313076,
      "loss": 1.4308,
      "step": 8924
    },
    {
      "epoch": 0.31962325639694167,
      "grad_norm": 1.7159067392349243,
      "learning_rate": 0.00015913921420123892,
      "loss": 1.843,
      "step": 8925
    },
    {
      "epoch": 0.31965906852650994,
      "grad_norm": 1.428824782371521,
      "learning_rate": 0.00015912986054370524,
      "loss": 1.4445,
      "step": 8926
    },
    {
      "epoch": 0.3196948806560782,
      "grad_norm": 1.6858423948287964,
      "learning_rate": 0.00015912050609065556,
      "loss": 1.1656,
      "step": 8927
    },
    {
      "epoch": 0.3197306927856465,
      "grad_norm": 1.671278476715088,
      "learning_rate": 0.00015911115084221575,
      "loss": 1.7137,
      "step": 8928
    },
    {
      "epoch": 0.3197665049152148,
      "grad_norm": 2.173511505126953,
      "learning_rate": 0.00015910179479851163,
      "loss": 1.6492,
      "step": 8929
    },
    {
      "epoch": 0.31980231704478307,
      "grad_norm": 1.9193826913833618,
      "learning_rate": 0.0001590924379596691,
      "loss": 1.5464,
      "step": 8930
    },
    {
      "epoch": 0.31983812917435134,
      "grad_norm": 1.7963279485702515,
      "learning_rate": 0.00015908308032581406,
      "loss": 1.7011,
      "step": 8931
    },
    {
      "epoch": 0.31987394130391966,
      "grad_norm": 1.4274190664291382,
      "learning_rate": 0.00015907372189707237,
      "loss": 1.4982,
      "step": 8932
    },
    {
      "epoch": 0.31990975343348793,
      "grad_norm": 1.926442265510559,
      "learning_rate": 0.00015906436267356993,
      "loss": 1.3027,
      "step": 8933
    },
    {
      "epoch": 0.3199455655630562,
      "grad_norm": 1.577248454093933,
      "learning_rate": 0.00015905500265543272,
      "loss": 1.4747,
      "step": 8934
    },
    {
      "epoch": 0.31998137769262447,
      "grad_norm": 2.0698063373565674,
      "learning_rate": 0.0001590456418427866,
      "loss": 1.6706,
      "step": 8935
    },
    {
      "epoch": 0.3200171898221928,
      "grad_norm": 1.6844439506530762,
      "learning_rate": 0.00015903628023575755,
      "loss": 1.2918,
      "step": 8936
    },
    {
      "epoch": 0.32005300195176106,
      "grad_norm": 1.304545283317566,
      "learning_rate": 0.00015902691783447142,
      "loss": 1.4205,
      "step": 8937
    },
    {
      "epoch": 0.32008881408132933,
      "grad_norm": 1.6640790700912476,
      "learning_rate": 0.00015901755463905434,
      "loss": 1.5702,
      "step": 8938
    },
    {
      "epoch": 0.32012462621089766,
      "grad_norm": 1.569692611694336,
      "learning_rate": 0.00015900819064963218,
      "loss": 1.5811,
      "step": 8939
    },
    {
      "epoch": 0.3201604383404659,
      "grad_norm": 1.8186932802200317,
      "learning_rate": 0.00015899882586633093,
      "loss": 1.2104,
      "step": 8940
    },
    {
      "epoch": 0.3201962504700342,
      "grad_norm": 1.6925221681594849,
      "learning_rate": 0.00015898946028927656,
      "loss": 1.5124,
      "step": 8941
    },
    {
      "epoch": 0.32023206259960246,
      "grad_norm": 1.6941040754318237,
      "learning_rate": 0.0001589800939185951,
      "loss": 1.5075,
      "step": 8942
    },
    {
      "epoch": 0.3202678747291708,
      "grad_norm": 1.9452513456344604,
      "learning_rate": 0.00015897072675441254,
      "loss": 1.5764,
      "step": 8943
    },
    {
      "epoch": 0.32030368685873906,
      "grad_norm": 2.2057533264160156,
      "learning_rate": 0.00015896135879685494,
      "loss": 1.7398,
      "step": 8944
    },
    {
      "epoch": 0.3203394989883073,
      "grad_norm": 1.8561280965805054,
      "learning_rate": 0.0001589519900460483,
      "loss": 1.5916,
      "step": 8945
    },
    {
      "epoch": 0.32037531111787565,
      "grad_norm": 1.8078523874282837,
      "learning_rate": 0.00015894262050211868,
      "loss": 1.4903,
      "step": 8946
    },
    {
      "epoch": 0.3204111232474439,
      "grad_norm": 1.5678856372833252,
      "learning_rate": 0.00015893325016519213,
      "loss": 1.5585,
      "step": 8947
    },
    {
      "epoch": 0.3204469353770122,
      "grad_norm": 2.330113172531128,
      "learning_rate": 0.0001589238790353947,
      "loss": 1.6519,
      "step": 8948
    },
    {
      "epoch": 0.32048274750658046,
      "grad_norm": 1.4870229959487915,
      "learning_rate": 0.00015891450711285254,
      "loss": 1.569,
      "step": 8949
    },
    {
      "epoch": 0.3205185596361488,
      "grad_norm": 2.701037883758545,
      "learning_rate": 0.00015890513439769164,
      "loss": 1.8232,
      "step": 8950
    },
    {
      "epoch": 0.32055437176571705,
      "grad_norm": 1.4118682146072388,
      "learning_rate": 0.00015889576089003814,
      "loss": 1.373,
      "step": 8951
    },
    {
      "epoch": 0.3205901838952853,
      "grad_norm": 1.6131876707077026,
      "learning_rate": 0.00015888638659001815,
      "loss": 1.503,
      "step": 8952
    },
    {
      "epoch": 0.32062599602485364,
      "grad_norm": 2.0327181816101074,
      "learning_rate": 0.0001588770114977578,
      "loss": 1.5179,
      "step": 8953
    },
    {
      "epoch": 0.3206618081544219,
      "grad_norm": 1.8969197273254395,
      "learning_rate": 0.00015886763561338317,
      "loss": 1.8301,
      "step": 8954
    },
    {
      "epoch": 0.3206976202839902,
      "grad_norm": 1.4035693407058716,
      "learning_rate": 0.00015885825893702048,
      "loss": 1.6315,
      "step": 8955
    },
    {
      "epoch": 0.32073343241355845,
      "grad_norm": 1.4035834074020386,
      "learning_rate": 0.0001588488814687958,
      "loss": 1.5281,
      "step": 8956
    },
    {
      "epoch": 0.3207692445431268,
      "grad_norm": 1.6486170291900635,
      "learning_rate": 0.00015883950320883536,
      "loss": 1.3552,
      "step": 8957
    },
    {
      "epoch": 0.32080505667269504,
      "grad_norm": 1.6082379817962646,
      "learning_rate": 0.0001588301241572653,
      "loss": 1.5837,
      "step": 8958
    },
    {
      "epoch": 0.3208408688022633,
      "grad_norm": 1.3907922506332397,
      "learning_rate": 0.0001588207443142118,
      "loss": 1.6525,
      "step": 8959
    },
    {
      "epoch": 0.32087668093183164,
      "grad_norm": 1.9259198904037476,
      "learning_rate": 0.00015881136367980103,
      "loss": 1.6522,
      "step": 8960
    },
    {
      "epoch": 0.3209124930613999,
      "grad_norm": 1.92613685131073,
      "learning_rate": 0.00015880198225415925,
      "loss": 1.2868,
      "step": 8961
    },
    {
      "epoch": 0.3209483051909682,
      "grad_norm": 1.6421856880187988,
      "learning_rate": 0.00015879260003741265,
      "loss": 1.7828,
      "step": 8962
    },
    {
      "epoch": 0.32098411732053644,
      "grad_norm": 1.4043571949005127,
      "learning_rate": 0.00015878321702968745,
      "loss": 1.6435,
      "step": 8963
    },
    {
      "epoch": 0.32101992945010477,
      "grad_norm": 2.38551926612854,
      "learning_rate": 0.0001587738332311099,
      "loss": 2.2681,
      "step": 8964
    },
    {
      "epoch": 0.32105574157967304,
      "grad_norm": 1.5245349407196045,
      "learning_rate": 0.0001587644486418062,
      "loss": 1.5347,
      "step": 8965
    },
    {
      "epoch": 0.3210915537092413,
      "grad_norm": 1.5355474948883057,
      "learning_rate": 0.00015875506326190267,
      "loss": 1.2685,
      "step": 8966
    },
    {
      "epoch": 0.32112736583880963,
      "grad_norm": 1.4486896991729736,
      "learning_rate": 0.00015874567709152557,
      "loss": 1.221,
      "step": 8967
    },
    {
      "epoch": 0.3211631779683779,
      "grad_norm": 1.4446889162063599,
      "learning_rate": 0.00015873629013080114,
      "loss": 1.5916,
      "step": 8968
    },
    {
      "epoch": 0.32119899009794617,
      "grad_norm": 2.2588870525360107,
      "learning_rate": 0.0001587269023798557,
      "loss": 1.5266,
      "step": 8969
    },
    {
      "epoch": 0.32123480222751444,
      "grad_norm": 1.638065218925476,
      "learning_rate": 0.0001587175138388155,
      "loss": 1.5114,
      "step": 8970
    },
    {
      "epoch": 0.32127061435708276,
      "grad_norm": 2.1063241958618164,
      "learning_rate": 0.00015870812450780695,
      "loss": 1.5352,
      "step": 8971
    },
    {
      "epoch": 0.32130642648665103,
      "grad_norm": 1.6843229532241821,
      "learning_rate": 0.00015869873438695628,
      "loss": 1.4685,
      "step": 8972
    },
    {
      "epoch": 0.3213422386162193,
      "grad_norm": 1.2848390340805054,
      "learning_rate": 0.00015868934347638985,
      "loss": 1.3653,
      "step": 8973
    },
    {
      "epoch": 0.3213780507457876,
      "grad_norm": 1.7618560791015625,
      "learning_rate": 0.00015867995177623403,
      "loss": 1.5555,
      "step": 8974
    },
    {
      "epoch": 0.3214138628753559,
      "grad_norm": 2.0183522701263428,
      "learning_rate": 0.00015867055928661517,
      "loss": 1.6017,
      "step": 8975
    },
    {
      "epoch": 0.32144967500492416,
      "grad_norm": 1.6978468894958496,
      "learning_rate": 0.00015866116600765957,
      "loss": 1.493,
      "step": 8976
    },
    {
      "epoch": 0.32148548713449243,
      "grad_norm": 1.5752780437469482,
      "learning_rate": 0.00015865177193949366,
      "loss": 1.5721,
      "step": 8977
    },
    {
      "epoch": 0.32152129926406076,
      "grad_norm": 1.5918536186218262,
      "learning_rate": 0.0001586423770822438,
      "loss": 1.4768,
      "step": 8978
    },
    {
      "epoch": 0.321557111393629,
      "grad_norm": 1.8312736749649048,
      "learning_rate": 0.0001586329814360364,
      "loss": 1.6069,
      "step": 8979
    },
    {
      "epoch": 0.3215929235231973,
      "grad_norm": 1.5193839073181152,
      "learning_rate": 0.0001586235850009979,
      "loss": 1.386,
      "step": 8980
    },
    {
      "epoch": 0.3216287356527656,
      "grad_norm": 2.7475154399871826,
      "learning_rate": 0.00015861418777725467,
      "loss": 1.451,
      "step": 8981
    },
    {
      "epoch": 0.3216645477823339,
      "grad_norm": 1.462113618850708,
      "learning_rate": 0.00015860478976493313,
      "loss": 1.4151,
      "step": 8982
    },
    {
      "epoch": 0.32170035991190216,
      "grad_norm": 1.9317468404769897,
      "learning_rate": 0.00015859539096415976,
      "loss": 1.6592,
      "step": 8983
    },
    {
      "epoch": 0.3217361720414704,
      "grad_norm": 1.8942067623138428,
      "learning_rate": 0.000158585991375061,
      "loss": 1.643,
      "step": 8984
    },
    {
      "epoch": 0.32177198417103875,
      "grad_norm": 1.6340413093566895,
      "learning_rate": 0.00015857659099776327,
      "loss": 1.5538,
      "step": 8985
    },
    {
      "epoch": 0.321807796300607,
      "grad_norm": 1.7912198305130005,
      "learning_rate": 0.0001585671898323931,
      "loss": 1.7632,
      "step": 8986
    },
    {
      "epoch": 0.3218436084301753,
      "grad_norm": 1.6519590616226196,
      "learning_rate": 0.0001585577878790769,
      "loss": 1.1778,
      "step": 8987
    },
    {
      "epoch": 0.3218794205597436,
      "grad_norm": 1.6828880310058594,
      "learning_rate": 0.00015854838513794118,
      "loss": 1.7943,
      "step": 8988
    },
    {
      "epoch": 0.3219152326893119,
      "grad_norm": 1.6437028646469116,
      "learning_rate": 0.00015853898160911252,
      "loss": 1.8893,
      "step": 8989
    },
    {
      "epoch": 0.32195104481888015,
      "grad_norm": 1.3070831298828125,
      "learning_rate": 0.00015852957729271735,
      "loss": 1.1287,
      "step": 8990
    },
    {
      "epoch": 0.3219868569484484,
      "grad_norm": 1.915002703666687,
      "learning_rate": 0.00015852017218888218,
      "loss": 1.3972,
      "step": 8991
    },
    {
      "epoch": 0.32202266907801674,
      "grad_norm": 1.2701020240783691,
      "learning_rate": 0.0001585107662977336,
      "loss": 1.5031,
      "step": 8992
    },
    {
      "epoch": 0.322058481207585,
      "grad_norm": 1.6998685598373413,
      "learning_rate": 0.00015850135961939814,
      "loss": 1.2343,
      "step": 8993
    },
    {
      "epoch": 0.3220942933371533,
      "grad_norm": 1.3520300388336182,
      "learning_rate": 0.00015849195215400234,
      "loss": 1.6848,
      "step": 8994
    },
    {
      "epoch": 0.3221301054667216,
      "grad_norm": 1.4436218738555908,
      "learning_rate": 0.0001584825439016728,
      "loss": 1.5324,
      "step": 8995
    },
    {
      "epoch": 0.3221659175962899,
      "grad_norm": 2.294487953186035,
      "learning_rate": 0.00015847313486253603,
      "loss": 1.8497,
      "step": 8996
    },
    {
      "epoch": 0.32220172972585814,
      "grad_norm": 1.5929428339004517,
      "learning_rate": 0.0001584637250367187,
      "loss": 1.4359,
      "step": 8997
    },
    {
      "epoch": 0.3222375418554264,
      "grad_norm": 2.296165943145752,
      "learning_rate": 0.00015845431442434733,
      "loss": 1.8707,
      "step": 8998
    },
    {
      "epoch": 0.32227335398499474,
      "grad_norm": 1.5806833505630493,
      "learning_rate": 0.00015844490302554856,
      "loss": 1.6144,
      "step": 8999
    },
    {
      "epoch": 0.322309166114563,
      "grad_norm": 1.7950143814086914,
      "learning_rate": 0.00015843549084044903,
      "loss": 1.7005,
      "step": 9000
    },
    {
      "epoch": 0.3223449782441313,
      "grad_norm": 2.0502243041992188,
      "learning_rate": 0.0001584260778691753,
      "loss": 1.5181,
      "step": 9001
    },
    {
      "epoch": 0.3223807903736996,
      "grad_norm": 1.9354183673858643,
      "learning_rate": 0.00015841666411185411,
      "loss": 1.3239,
      "step": 9002
    },
    {
      "epoch": 0.32241660250326787,
      "grad_norm": 1.4576351642608643,
      "learning_rate": 0.000158407249568612,
      "loss": 1.46,
      "step": 9003
    },
    {
      "epoch": 0.32245241463283614,
      "grad_norm": 1.6081417798995972,
      "learning_rate": 0.00015839783423957576,
      "loss": 1.4387,
      "step": 9004
    },
    {
      "epoch": 0.3224882267624044,
      "grad_norm": 2.0253992080688477,
      "learning_rate": 0.00015838841812487194,
      "loss": 1.7281,
      "step": 9005
    },
    {
      "epoch": 0.32252403889197273,
      "grad_norm": 1.742729902267456,
      "learning_rate": 0.00015837900122462725,
      "loss": 1.3755,
      "step": 9006
    },
    {
      "epoch": 0.322559851021541,
      "grad_norm": 2.4512531757354736,
      "learning_rate": 0.00015836958353896845,
      "loss": 1.3772,
      "step": 9007
    },
    {
      "epoch": 0.32259566315110927,
      "grad_norm": 1.7238280773162842,
      "learning_rate": 0.00015836016506802218,
      "loss": 1.3526,
      "step": 9008
    },
    {
      "epoch": 0.3226314752806776,
      "grad_norm": 2.1217103004455566,
      "learning_rate": 0.00015835074581191516,
      "loss": 1.7,
      "step": 9009
    },
    {
      "epoch": 0.32266728741024586,
      "grad_norm": 1.888587474822998,
      "learning_rate": 0.00015834132577077412,
      "loss": 1.5873,
      "step": 9010
    },
    {
      "epoch": 0.32270309953981413,
      "grad_norm": 1.8031010627746582,
      "learning_rate": 0.00015833190494472582,
      "loss": 1.8754,
      "step": 9011
    },
    {
      "epoch": 0.3227389116693824,
      "grad_norm": 1.394250750541687,
      "learning_rate": 0.00015832248333389693,
      "loss": 1.5981,
      "step": 9012
    },
    {
      "epoch": 0.3227747237989507,
      "grad_norm": 1.8603804111480713,
      "learning_rate": 0.00015831306093841432,
      "loss": 1.4249,
      "step": 9013
    },
    {
      "epoch": 0.322810535928519,
      "grad_norm": 2.721745014190674,
      "learning_rate": 0.00015830363775840467,
      "loss": 1.3469,
      "step": 9014
    },
    {
      "epoch": 0.32284634805808726,
      "grad_norm": 1.7731702327728271,
      "learning_rate": 0.00015829421379399475,
      "loss": 1.742,
      "step": 9015
    },
    {
      "epoch": 0.32288216018765553,
      "grad_norm": 1.3356654644012451,
      "learning_rate": 0.00015828478904531142,
      "loss": 1.4635,
      "step": 9016
    },
    {
      "epoch": 0.32291797231722386,
      "grad_norm": 1.3644803762435913,
      "learning_rate": 0.0001582753635124814,
      "loss": 1.3199,
      "step": 9017
    },
    {
      "epoch": 0.3229537844467921,
      "grad_norm": 1.669406771659851,
      "learning_rate": 0.00015826593719563156,
      "loss": 1.3352,
      "step": 9018
    },
    {
      "epoch": 0.3229895965763604,
      "grad_norm": 1.548660397529602,
      "learning_rate": 0.0001582565100948887,
      "loss": 1.5234,
      "step": 9019
    },
    {
      "epoch": 0.3230254087059287,
      "grad_norm": 1.9580634832382202,
      "learning_rate": 0.00015824708221037965,
      "loss": 1.5996,
      "step": 9020
    },
    {
      "epoch": 0.323061220835497,
      "grad_norm": 1.9380831718444824,
      "learning_rate": 0.0001582376535422312,
      "loss": 1.245,
      "step": 9021
    },
    {
      "epoch": 0.32309703296506526,
      "grad_norm": 1.6162350177764893,
      "learning_rate": 0.00015822822409057024,
      "loss": 1.4507,
      "step": 9022
    },
    {
      "epoch": 0.3231328450946335,
      "grad_norm": 1.2511042356491089,
      "learning_rate": 0.00015821879385552367,
      "loss": 1.3691,
      "step": 9023
    },
    {
      "epoch": 0.32316865722420185,
      "grad_norm": 1.4707595109939575,
      "learning_rate": 0.00015820936283721834,
      "loss": 1.6036,
      "step": 9024
    },
    {
      "epoch": 0.3232044693537701,
      "grad_norm": 2.1324875354766846,
      "learning_rate": 0.00015819993103578106,
      "loss": 1.4427,
      "step": 9025
    },
    {
      "epoch": 0.3232402814833384,
      "grad_norm": 1.6112456321716309,
      "learning_rate": 0.0001581904984513388,
      "loss": 1.3933,
      "step": 9026
    },
    {
      "epoch": 0.3232760936129067,
      "grad_norm": 2.6338889598846436,
      "learning_rate": 0.00015818106508401847,
      "loss": 1.508,
      "step": 9027
    },
    {
      "epoch": 0.323311905742475,
      "grad_norm": 2.0561411380767822,
      "learning_rate": 0.00015817163093394693,
      "loss": 1.3873,
      "step": 9028
    },
    {
      "epoch": 0.32334771787204325,
      "grad_norm": 1.4188092947006226,
      "learning_rate": 0.00015816219600125114,
      "loss": 1.3726,
      "step": 9029
    },
    {
      "epoch": 0.3233835300016115,
      "grad_norm": 2.1878104209899902,
      "learning_rate": 0.00015815276028605807,
      "loss": 1.3186,
      "step": 9030
    },
    {
      "epoch": 0.32341934213117984,
      "grad_norm": 1.4584554433822632,
      "learning_rate": 0.00015814332378849457,
      "loss": 1.5844,
      "step": 9031
    },
    {
      "epoch": 0.3234551542607481,
      "grad_norm": 2.6032674312591553,
      "learning_rate": 0.00015813388650868766,
      "loss": 1.4215,
      "step": 9032
    },
    {
      "epoch": 0.3234909663903164,
      "grad_norm": 1.6895302534103394,
      "learning_rate": 0.00015812444844676428,
      "loss": 1.5978,
      "step": 9033
    },
    {
      "epoch": 0.3235267785198847,
      "grad_norm": 1.8651604652404785,
      "learning_rate": 0.00015811500960285143,
      "loss": 1.3851,
      "step": 9034
    },
    {
      "epoch": 0.323562590649453,
      "grad_norm": 1.7477426528930664,
      "learning_rate": 0.00015810556997707608,
      "loss": 1.2799,
      "step": 9035
    },
    {
      "epoch": 0.32359840277902124,
      "grad_norm": 1.9270200729370117,
      "learning_rate": 0.00015809612956956527,
      "loss": 1.335,
      "step": 9036
    },
    {
      "epoch": 0.3236342149085895,
      "grad_norm": 1.5941219329833984,
      "learning_rate": 0.00015808668838044595,
      "loss": 1.5451,
      "step": 9037
    },
    {
      "epoch": 0.32367002703815784,
      "grad_norm": 2.0007755756378174,
      "learning_rate": 0.00015807724640984518,
      "loss": 1.4461,
      "step": 9038
    },
    {
      "epoch": 0.3237058391677261,
      "grad_norm": 2.1980857849121094,
      "learning_rate": 0.00015806780365788998,
      "loss": 1.8456,
      "step": 9039
    },
    {
      "epoch": 0.3237416512972944,
      "grad_norm": 4.429500579833984,
      "learning_rate": 0.00015805836012470733,
      "loss": 1.8815,
      "step": 9040
    },
    {
      "epoch": 0.3237774634268627,
      "grad_norm": 1.9881982803344727,
      "learning_rate": 0.0001580489158104244,
      "loss": 1.5833,
      "step": 9041
    },
    {
      "epoch": 0.32381327555643097,
      "grad_norm": 2.5081233978271484,
      "learning_rate": 0.00015803947071516813,
      "loss": 1.2176,
      "step": 9042
    },
    {
      "epoch": 0.32384908768599924,
      "grad_norm": 1.551257848739624,
      "learning_rate": 0.00015803002483906568,
      "loss": 1.6953,
      "step": 9043
    },
    {
      "epoch": 0.3238848998155675,
      "grad_norm": 1.6371716260910034,
      "learning_rate": 0.0001580205781822441,
      "loss": 1.4023,
      "step": 9044
    },
    {
      "epoch": 0.32392071194513583,
      "grad_norm": 1.3484255075454712,
      "learning_rate": 0.00015801113074483046,
      "loss": 1.6748,
      "step": 9045
    },
    {
      "epoch": 0.3239565240747041,
      "grad_norm": 1.9768213033676147,
      "learning_rate": 0.0001580016825269519,
      "loss": 1.6158,
      "step": 9046
    },
    {
      "epoch": 0.32399233620427237,
      "grad_norm": 1.9486857652664185,
      "learning_rate": 0.00015799223352873555,
      "loss": 1.4215,
      "step": 9047
    },
    {
      "epoch": 0.3240281483338407,
      "grad_norm": 1.4396650791168213,
      "learning_rate": 0.00015798278375030845,
      "loss": 1.1456,
      "step": 9048
    },
    {
      "epoch": 0.32406396046340896,
      "grad_norm": 1.3435920476913452,
      "learning_rate": 0.0001579733331917978,
      "loss": 1.4729,
      "step": 9049
    },
    {
      "epoch": 0.32409977259297723,
      "grad_norm": 1.549360990524292,
      "learning_rate": 0.00015796388185333076,
      "loss": 1.4954,
      "step": 9050
    },
    {
      "epoch": 0.3241355847225455,
      "grad_norm": 1.8293956518173218,
      "learning_rate": 0.00015795442973503442,
      "loss": 1.6314,
      "step": 9051
    },
    {
      "epoch": 0.3241713968521138,
      "grad_norm": 1.7965165376663208,
      "learning_rate": 0.00015794497683703601,
      "loss": 1.9173,
      "step": 9052
    },
    {
      "epoch": 0.3242072089816821,
      "grad_norm": 1.5780364274978638,
      "learning_rate": 0.00015793552315946266,
      "loss": 1.5728,
      "step": 9053
    },
    {
      "epoch": 0.32424302111125036,
      "grad_norm": 3.517538547515869,
      "learning_rate": 0.00015792606870244162,
      "loss": 1.7913,
      "step": 9054
    },
    {
      "epoch": 0.3242788332408187,
      "grad_norm": 1.9409570693969727,
      "learning_rate": 0.0001579166134661,
      "loss": 1.3348,
      "step": 9055
    },
    {
      "epoch": 0.32431464537038696,
      "grad_norm": 1.7771880626678467,
      "learning_rate": 0.00015790715745056506,
      "loss": 1.6459,
      "step": 9056
    },
    {
      "epoch": 0.3243504574999552,
      "grad_norm": 1.7616101503372192,
      "learning_rate": 0.00015789770065596404,
      "loss": 1.7816,
      "step": 9057
    },
    {
      "epoch": 0.3243862696295235,
      "grad_norm": 1.777740478515625,
      "learning_rate": 0.00015788824308242408,
      "loss": 1.8904,
      "step": 9058
    },
    {
      "epoch": 0.3244220817590918,
      "grad_norm": 2.1632614135742188,
      "learning_rate": 0.00015787878473007253,
      "loss": 1.6462,
      "step": 9059
    },
    {
      "epoch": 0.3244578938886601,
      "grad_norm": 2.3858914375305176,
      "learning_rate": 0.00015786932559903657,
      "loss": 1.6606,
      "step": 9060
    },
    {
      "epoch": 0.32449370601822836,
      "grad_norm": 1.9479713439941406,
      "learning_rate": 0.00015785986568944352,
      "loss": 1.7255,
      "step": 9061
    },
    {
      "epoch": 0.3245295181477967,
      "grad_norm": 1.4364668130874634,
      "learning_rate": 0.00015785040500142057,
      "loss": 1.5002,
      "step": 9062
    },
    {
      "epoch": 0.32456533027736495,
      "grad_norm": 2.068448066711426,
      "learning_rate": 0.00015784094353509507,
      "loss": 1.4521,
      "step": 9063
    },
    {
      "epoch": 0.3246011424069332,
      "grad_norm": 2.5338597297668457,
      "learning_rate": 0.00015783148129059425,
      "loss": 1.5654,
      "step": 9064
    },
    {
      "epoch": 0.3246369545365015,
      "grad_norm": 2.1325583457946777,
      "learning_rate": 0.00015782201826804548,
      "loss": 1.6559,
      "step": 9065
    },
    {
      "epoch": 0.3246727666660698,
      "grad_norm": 1.665851354598999,
      "learning_rate": 0.000157812554467576,
      "loss": 1.3783,
      "step": 9066
    },
    {
      "epoch": 0.3247085787956381,
      "grad_norm": 1.4407624006271362,
      "learning_rate": 0.0001578030898893132,
      "loss": 1.6831,
      "step": 9067
    },
    {
      "epoch": 0.32474439092520635,
      "grad_norm": 1.782218098640442,
      "learning_rate": 0.00015779362453338438,
      "loss": 1.7928,
      "step": 9068
    },
    {
      "epoch": 0.3247802030547747,
      "grad_norm": 1.5871248245239258,
      "learning_rate": 0.0001577841583999169,
      "loss": 1.4532,
      "step": 9069
    },
    {
      "epoch": 0.32481601518434294,
      "grad_norm": 1.4326859712600708,
      "learning_rate": 0.00015777469148903808,
      "loss": 1.5741,
      "step": 9070
    },
    {
      "epoch": 0.3248518273139112,
      "grad_norm": 1.2834270000457764,
      "learning_rate": 0.00015776522380087532,
      "loss": 1.5228,
      "step": 9071
    },
    {
      "epoch": 0.3248876394434795,
      "grad_norm": 1.9993133544921875,
      "learning_rate": 0.00015775575533555602,
      "loss": 1.6291,
      "step": 9072
    },
    {
      "epoch": 0.3249234515730478,
      "grad_norm": 1.301542043685913,
      "learning_rate": 0.0001577462860932075,
      "loss": 1.4781,
      "step": 9073
    },
    {
      "epoch": 0.3249592637026161,
      "grad_norm": 1.7204660177230835,
      "learning_rate": 0.00015773681607395717,
      "loss": 1.2986,
      "step": 9074
    },
    {
      "epoch": 0.32499507583218434,
      "grad_norm": 2.050814628601074,
      "learning_rate": 0.0001577273452779325,
      "loss": 1.8979,
      "step": 9075
    },
    {
      "epoch": 0.32503088796175267,
      "grad_norm": 1.3881765604019165,
      "learning_rate": 0.00015771787370526084,
      "loss": 1.5927,
      "step": 9076
    },
    {
      "epoch": 0.32506670009132094,
      "grad_norm": 1.4091527462005615,
      "learning_rate": 0.0001577084013560696,
      "loss": 1.3812,
      "step": 9077
    },
    {
      "epoch": 0.3251025122208892,
      "grad_norm": 2.129033088684082,
      "learning_rate": 0.0001576989282304863,
      "loss": 1.4002,
      "step": 9078
    },
    {
      "epoch": 0.3251383243504575,
      "grad_norm": 1.6239925622940063,
      "learning_rate": 0.00015768945432863835,
      "loss": 1.4022,
      "step": 9079
    },
    {
      "epoch": 0.3251741364800258,
      "grad_norm": 2.6688060760498047,
      "learning_rate": 0.00015767997965065322,
      "loss": 1.7299,
      "step": 9080
    },
    {
      "epoch": 0.32520994860959407,
      "grad_norm": 1.6006053686141968,
      "learning_rate": 0.00015767050419665836,
      "loss": 1.4682,
      "step": 9081
    },
    {
      "epoch": 0.32524576073916234,
      "grad_norm": 1.31721830368042,
      "learning_rate": 0.00015766102796678123,
      "loss": 1.5401,
      "step": 9082
    },
    {
      "epoch": 0.32528157286873066,
      "grad_norm": 2.0746214389801025,
      "learning_rate": 0.00015765155096114934,
      "loss": 1.6972,
      "step": 9083
    },
    {
      "epoch": 0.32531738499829893,
      "grad_norm": 1.4532837867736816,
      "learning_rate": 0.00015764207317989023,
      "loss": 1.1788,
      "step": 9084
    },
    {
      "epoch": 0.3253531971278672,
      "grad_norm": 1.7890650033950806,
      "learning_rate": 0.00015763259462313136,
      "loss": 1.5181,
      "step": 9085
    },
    {
      "epoch": 0.32538900925743547,
      "grad_norm": 1.5378397703170776,
      "learning_rate": 0.00015762311529100024,
      "loss": 1.7676,
      "step": 9086
    },
    {
      "epoch": 0.3254248213870038,
      "grad_norm": 1.4087746143341064,
      "learning_rate": 0.00015761363518362447,
      "loss": 1.5524,
      "step": 9087
    },
    {
      "epoch": 0.32546063351657206,
      "grad_norm": 1.3853182792663574,
      "learning_rate": 0.00015760415430113157,
      "loss": 1.6659,
      "step": 9088
    },
    {
      "epoch": 0.32549644564614033,
      "grad_norm": 1.8238393068313599,
      "learning_rate": 0.00015759467264364905,
      "loss": 1.7841,
      "step": 9089
    },
    {
      "epoch": 0.32553225777570866,
      "grad_norm": 1.984938383102417,
      "learning_rate": 0.00015758519021130451,
      "loss": 1.5044,
      "step": 9090
    },
    {
      "epoch": 0.3255680699052769,
      "grad_norm": 1.9241176843643188,
      "learning_rate": 0.0001575757070042255,
      "loss": 1.658,
      "step": 9091
    },
    {
      "epoch": 0.3256038820348452,
      "grad_norm": 1.9488638639450073,
      "learning_rate": 0.00015756622302253966,
      "loss": 1.5738,
      "step": 9092
    },
    {
      "epoch": 0.32563969416441346,
      "grad_norm": 1.1886811256408691,
      "learning_rate": 0.0001575567382663745,
      "loss": 1.5968,
      "step": 9093
    },
    {
      "epoch": 0.3256755062939818,
      "grad_norm": 1.2945717573165894,
      "learning_rate": 0.00015754725273585767,
      "loss": 1.3245,
      "step": 9094
    },
    {
      "epoch": 0.32571131842355006,
      "grad_norm": 1.674601674079895,
      "learning_rate": 0.0001575377664311168,
      "loss": 1.521,
      "step": 9095
    },
    {
      "epoch": 0.3257471305531183,
      "grad_norm": 1.8002949953079224,
      "learning_rate": 0.00015752827935227952,
      "loss": 1.4273,
      "step": 9096
    },
    {
      "epoch": 0.32578294268268665,
      "grad_norm": 2.0485355854034424,
      "learning_rate": 0.00015751879149947343,
      "loss": 1.6217,
      "step": 9097
    },
    {
      "epoch": 0.3258187548122549,
      "grad_norm": 1.7260215282440186,
      "learning_rate": 0.0001575093028728262,
      "loss": 1.1519,
      "step": 9098
    },
    {
      "epoch": 0.3258545669418232,
      "grad_norm": 2.4711947441101074,
      "learning_rate": 0.00015749981347246549,
      "loss": 1.5758,
      "step": 9099
    },
    {
      "epoch": 0.32589037907139146,
      "grad_norm": 1.7342562675476074,
      "learning_rate": 0.00015749032329851894,
      "loss": 1.3542,
      "step": 9100
    },
    {
      "epoch": 0.3259261912009598,
      "grad_norm": 1.5560182332992554,
      "learning_rate": 0.00015748083235111424,
      "loss": 1.4425,
      "step": 9101
    },
    {
      "epoch": 0.32596200333052805,
      "grad_norm": 1.55160653591156,
      "learning_rate": 0.00015747134063037908,
      "loss": 1.7229,
      "step": 9102
    },
    {
      "epoch": 0.3259978154600963,
      "grad_norm": 1.796183466911316,
      "learning_rate": 0.0001574618481364412,
      "loss": 1.5933,
      "step": 9103
    },
    {
      "epoch": 0.32603362758966464,
      "grad_norm": 1.8876303434371948,
      "learning_rate": 0.00015745235486942826,
      "loss": 1.6456,
      "step": 9104
    },
    {
      "epoch": 0.3260694397192329,
      "grad_norm": 1.37518310546875,
      "learning_rate": 0.00015744286082946797,
      "loss": 1.3399,
      "step": 9105
    },
    {
      "epoch": 0.3261052518488012,
      "grad_norm": 1.4342260360717773,
      "learning_rate": 0.0001574333660166881,
      "loss": 1.5002,
      "step": 9106
    },
    {
      "epoch": 0.32614106397836945,
      "grad_norm": 1.8933452367782593,
      "learning_rate": 0.0001574238704312164,
      "loss": 1.2322,
      "step": 9107
    },
    {
      "epoch": 0.3261768761079378,
      "grad_norm": 1.6854076385498047,
      "learning_rate": 0.00015741437407318056,
      "loss": 1.6315,
      "step": 9108
    },
    {
      "epoch": 0.32621268823750604,
      "grad_norm": 1.8995707035064697,
      "learning_rate": 0.00015740487694270838,
      "loss": 1.4575,
      "step": 9109
    },
    {
      "epoch": 0.3262485003670743,
      "grad_norm": 2.63657283782959,
      "learning_rate": 0.00015739537903992765,
      "loss": 1.3891,
      "step": 9110
    },
    {
      "epoch": 0.32628431249664264,
      "grad_norm": 1.7770910263061523,
      "learning_rate": 0.0001573858803649661,
      "loss": 1.4237,
      "step": 9111
    },
    {
      "epoch": 0.3263201246262109,
      "grad_norm": 2.1773881912231445,
      "learning_rate": 0.00015737638091795157,
      "loss": 1.6328,
      "step": 9112
    },
    {
      "epoch": 0.3263559367557792,
      "grad_norm": 1.9160898923873901,
      "learning_rate": 0.00015736688069901183,
      "loss": 1.5335,
      "step": 9113
    },
    {
      "epoch": 0.32639174888534744,
      "grad_norm": 1.9974308013916016,
      "learning_rate": 0.00015735737970827473,
      "loss": 1.1659,
      "step": 9114
    },
    {
      "epoch": 0.32642756101491577,
      "grad_norm": 1.807265281677246,
      "learning_rate": 0.00015734787794586806,
      "loss": 1.5504,
      "step": 9115
    },
    {
      "epoch": 0.32646337314448404,
      "grad_norm": 1.8353880643844604,
      "learning_rate": 0.00015733837541191968,
      "loss": 1.6066,
      "step": 9116
    },
    {
      "epoch": 0.3264991852740523,
      "grad_norm": 1.676720380783081,
      "learning_rate": 0.00015732887210655742,
      "loss": 1.4618,
      "step": 9117
    },
    {
      "epoch": 0.32653499740362063,
      "grad_norm": 1.6320055723190308,
      "learning_rate": 0.00015731936802990912,
      "loss": 1.7968,
      "step": 9118
    },
    {
      "epoch": 0.3265708095331889,
      "grad_norm": 2.2749667167663574,
      "learning_rate": 0.00015730986318210265,
      "loss": 1.4673,
      "step": 9119
    },
    {
      "epoch": 0.32660662166275717,
      "grad_norm": 1.3211688995361328,
      "learning_rate": 0.00015730035756326592,
      "loss": 1.5625,
      "step": 9120
    },
    {
      "epoch": 0.32664243379232544,
      "grad_norm": 1.8194984197616577,
      "learning_rate": 0.00015729085117352674,
      "loss": 1.4496,
      "step": 9121
    },
    {
      "epoch": 0.32667824592189376,
      "grad_norm": 2.0746090412139893,
      "learning_rate": 0.00015728134401301312,
      "loss": 1.2436,
      "step": 9122
    },
    {
      "epoch": 0.32671405805146203,
      "grad_norm": 1.5881251096725464,
      "learning_rate": 0.0001572718360818529,
      "loss": 1.3511,
      "step": 9123
    },
    {
      "epoch": 0.3267498701810303,
      "grad_norm": 1.9258278608322144,
      "learning_rate": 0.00015726232738017397,
      "loss": 1.5968,
      "step": 9124
    },
    {
      "epoch": 0.3267856823105986,
      "grad_norm": 1.449415922164917,
      "learning_rate": 0.00015725281790810431,
      "loss": 1.5124,
      "step": 9125
    },
    {
      "epoch": 0.3268214944401669,
      "grad_norm": 1.274713158607483,
      "learning_rate": 0.00015724330766577182,
      "loss": 1.6398,
      "step": 9126
    },
    {
      "epoch": 0.32685730656973516,
      "grad_norm": 1.757870078086853,
      "learning_rate": 0.0001572337966533045,
      "loss": 1.8657,
      "step": 9127
    },
    {
      "epoch": 0.32689311869930343,
      "grad_norm": 1.3837286233901978,
      "learning_rate": 0.0001572242848708302,
      "loss": 1.5027,
      "step": 9128
    },
    {
      "epoch": 0.32692893082887176,
      "grad_norm": 2.0841174125671387,
      "learning_rate": 0.00015721477231847702,
      "loss": 1.5041,
      "step": 9129
    },
    {
      "epoch": 0.32696474295844,
      "grad_norm": 3.516465663909912,
      "learning_rate": 0.00015720525899637285,
      "loss": 1.6544,
      "step": 9130
    },
    {
      "epoch": 0.3270005550880083,
      "grad_norm": 1.4936449527740479,
      "learning_rate": 0.00015719574490464573,
      "loss": 1.5635,
      "step": 9131
    },
    {
      "epoch": 0.3270363672175766,
      "grad_norm": 2.2506232261657715,
      "learning_rate": 0.00015718623004342362,
      "loss": 1.719,
      "step": 9132
    },
    {
      "epoch": 0.3270721793471449,
      "grad_norm": 1.8259614706039429,
      "learning_rate": 0.00015717671441283458,
      "loss": 1.7564,
      "step": 9133
    },
    {
      "epoch": 0.32710799147671316,
      "grad_norm": 1.6399543285369873,
      "learning_rate": 0.0001571671980130066,
      "loss": 1.8228,
      "step": 9134
    },
    {
      "epoch": 0.3271438036062814,
      "grad_norm": 1.5702282190322876,
      "learning_rate": 0.00015715768084406765,
      "loss": 1.5844,
      "step": 9135
    },
    {
      "epoch": 0.32717961573584975,
      "grad_norm": 1.4354286193847656,
      "learning_rate": 0.0001571481629061459,
      "loss": 1.5932,
      "step": 9136
    },
    {
      "epoch": 0.327215427865418,
      "grad_norm": 1.8592087030410767,
      "learning_rate": 0.0001571386441993693,
      "loss": 1.686,
      "step": 9137
    },
    {
      "epoch": 0.3272512399949863,
      "grad_norm": 1.372339129447937,
      "learning_rate": 0.00015712912472386597,
      "loss": 1.682,
      "step": 9138
    },
    {
      "epoch": 0.3272870521245546,
      "grad_norm": 1.4004724025726318,
      "learning_rate": 0.00015711960447976393,
      "loss": 1.834,
      "step": 9139
    },
    {
      "epoch": 0.3273228642541229,
      "grad_norm": 1.7258496284484863,
      "learning_rate": 0.0001571100834671913,
      "loss": 1.549,
      "step": 9140
    },
    {
      "epoch": 0.32735867638369115,
      "grad_norm": 1.3979535102844238,
      "learning_rate": 0.00015710056168627618,
      "loss": 1.689,
      "step": 9141
    },
    {
      "epoch": 0.3273944885132594,
      "grad_norm": 2.24281907081604,
      "learning_rate": 0.00015709103913714664,
      "loss": 1.5228,
      "step": 9142
    },
    {
      "epoch": 0.32743030064282774,
      "grad_norm": 1.3384792804718018,
      "learning_rate": 0.0001570815158199308,
      "loss": 1.2907,
      "step": 9143
    },
    {
      "epoch": 0.327466112772396,
      "grad_norm": 1.8024048805236816,
      "learning_rate": 0.00015707199173475682,
      "loss": 1.4062,
      "step": 9144
    },
    {
      "epoch": 0.3275019249019643,
      "grad_norm": 2.6472220420837402,
      "learning_rate": 0.00015706246688175282,
      "loss": 1.2139,
      "step": 9145
    },
    {
      "epoch": 0.3275377370315326,
      "grad_norm": 1.628482699394226,
      "learning_rate": 0.00015705294126104692,
      "loss": 1.4098,
      "step": 9146
    },
    {
      "epoch": 0.3275735491611009,
      "grad_norm": 1.4141241312026978,
      "learning_rate": 0.00015704341487276726,
      "loss": 1.2319,
      "step": 9147
    },
    {
      "epoch": 0.32760936129066914,
      "grad_norm": 1.8727307319641113,
      "learning_rate": 0.00015703388771704205,
      "loss": 1.6832,
      "step": 9148
    },
    {
      "epoch": 0.3276451734202374,
      "grad_norm": 1.9481236934661865,
      "learning_rate": 0.00015702435979399946,
      "loss": 1.4832,
      "step": 9149
    },
    {
      "epoch": 0.32768098554980574,
      "grad_norm": 1.4255954027175903,
      "learning_rate": 0.00015701483110376762,
      "loss": 1.3543,
      "step": 9150
    },
    {
      "epoch": 0.327716797679374,
      "grad_norm": 2.0080718994140625,
      "learning_rate": 0.00015700530164647485,
      "loss": 1.6513,
      "step": 9151
    },
    {
      "epoch": 0.3277526098089423,
      "grad_norm": 1.8739054203033447,
      "learning_rate": 0.00015699577142224924,
      "loss": 1.1768,
      "step": 9152
    },
    {
      "epoch": 0.3277884219385106,
      "grad_norm": 1.5352083444595337,
      "learning_rate": 0.000156986240431219,
      "loss": 1.4779,
      "step": 9153
    },
    {
      "epoch": 0.32782423406807887,
      "grad_norm": 1.6548919677734375,
      "learning_rate": 0.00015697670867351247,
      "loss": 1.5669,
      "step": 9154
    },
    {
      "epoch": 0.32786004619764714,
      "grad_norm": 1.8532062768936157,
      "learning_rate": 0.0001569671761492578,
      "loss": 1.713,
      "step": 9155
    },
    {
      "epoch": 0.3278958583272154,
      "grad_norm": 1.4983744621276855,
      "learning_rate": 0.00015695764285858323,
      "loss": 1.7685,
      "step": 9156
    },
    {
      "epoch": 0.32793167045678373,
      "grad_norm": 1.8667110204696655,
      "learning_rate": 0.00015694810880161706,
      "loss": 1.3894,
      "step": 9157
    },
    {
      "epoch": 0.327967482586352,
      "grad_norm": 1.7034077644348145,
      "learning_rate": 0.00015693857397848756,
      "loss": 1.6046,
      "step": 9158
    },
    {
      "epoch": 0.32800329471592027,
      "grad_norm": 2.344395399093628,
      "learning_rate": 0.00015692903838932299,
      "loss": 1.5522,
      "step": 9159
    },
    {
      "epoch": 0.3280391068454886,
      "grad_norm": 2.167020797729492,
      "learning_rate": 0.00015691950203425162,
      "loss": 1.3704,
      "step": 9160
    },
    {
      "epoch": 0.32807491897505686,
      "grad_norm": 1.6182739734649658,
      "learning_rate": 0.0001569099649134018,
      "loss": 1.3054,
      "step": 9161
    },
    {
      "epoch": 0.32811073110462513,
      "grad_norm": 1.4349606037139893,
      "learning_rate": 0.0001569004270269018,
      "loss": 1.4676,
      "step": 9162
    },
    {
      "epoch": 0.3281465432341934,
      "grad_norm": 1.9794306755065918,
      "learning_rate": 0.00015689088837487995,
      "loss": 1.8758,
      "step": 9163
    },
    {
      "epoch": 0.3281823553637617,
      "grad_norm": 1.2538909912109375,
      "learning_rate": 0.00015688134895746459,
      "loss": 1.5755,
      "step": 9164
    },
    {
      "epoch": 0.32821816749333,
      "grad_norm": 2.244776487350464,
      "learning_rate": 0.000156871808774784,
      "loss": 1.6322,
      "step": 9165
    },
    {
      "epoch": 0.32825397962289826,
      "grad_norm": 1.843839406967163,
      "learning_rate": 0.00015686226782696662,
      "loss": 1.6366,
      "step": 9166
    },
    {
      "epoch": 0.3282897917524666,
      "grad_norm": 2.0505659580230713,
      "learning_rate": 0.0001568527261141408,
      "loss": 1.5395,
      "step": 9167
    },
    {
      "epoch": 0.32832560388203486,
      "grad_norm": 1.6145918369293213,
      "learning_rate": 0.00015684318363643485,
      "loss": 1.4893,
      "step": 9168
    },
    {
      "epoch": 0.3283614160116031,
      "grad_norm": 1.7534884214401245,
      "learning_rate": 0.0001568336403939772,
      "loss": 1.5143,
      "step": 9169
    },
    {
      "epoch": 0.3283972281411714,
      "grad_norm": 1.496999740600586,
      "learning_rate": 0.00015682409638689623,
      "loss": 1.6016,
      "step": 9170
    },
    {
      "epoch": 0.3284330402707397,
      "grad_norm": 1.4966206550598145,
      "learning_rate": 0.00015681455161532034,
      "loss": 1.311,
      "step": 9171
    },
    {
      "epoch": 0.328468852400308,
      "grad_norm": 1.686967372894287,
      "learning_rate": 0.00015680500607937793,
      "loss": 1.2972,
      "step": 9172
    },
    {
      "epoch": 0.32850466452987626,
      "grad_norm": 1.885436773300171,
      "learning_rate": 0.00015679545977919745,
      "loss": 1.7629,
      "step": 9173
    },
    {
      "epoch": 0.3285404766594446,
      "grad_norm": 1.4676927328109741,
      "learning_rate": 0.0001567859127149073,
      "loss": 1.7276,
      "step": 9174
    },
    {
      "epoch": 0.32857628878901285,
      "grad_norm": 1.650899887084961,
      "learning_rate": 0.00015677636488663595,
      "loss": 1.5655,
      "step": 9175
    },
    {
      "epoch": 0.3286121009185811,
      "grad_norm": 1.4728350639343262,
      "learning_rate": 0.00015676681629451185,
      "loss": 1.682,
      "step": 9176
    },
    {
      "epoch": 0.3286479130481494,
      "grad_norm": 1.5229905843734741,
      "learning_rate": 0.0001567572669386635,
      "loss": 1.628,
      "step": 9177
    },
    {
      "epoch": 0.3286837251777177,
      "grad_norm": 1.990729808807373,
      "learning_rate": 0.0001567477168192193,
      "loss": 1.9226,
      "step": 9178
    },
    {
      "epoch": 0.328719537307286,
      "grad_norm": 1.5891399383544922,
      "learning_rate": 0.00015673816593630776,
      "loss": 1.3932,
      "step": 9179
    },
    {
      "epoch": 0.32875534943685425,
      "grad_norm": 1.5668306350708008,
      "learning_rate": 0.00015672861429005737,
      "loss": 1.3301,
      "step": 9180
    },
    {
      "epoch": 0.3287911615664226,
      "grad_norm": 1.5943611860275269,
      "learning_rate": 0.00015671906188059672,
      "loss": 1.5854,
      "step": 9181
    },
    {
      "epoch": 0.32882697369599084,
      "grad_norm": 1.9587252140045166,
      "learning_rate": 0.0001567095087080542,
      "loss": 1.8593,
      "step": 9182
    },
    {
      "epoch": 0.3288627858255591,
      "grad_norm": 1.4713166952133179,
      "learning_rate": 0.00015669995477255838,
      "loss": 1.6878,
      "step": 9183
    },
    {
      "epoch": 0.3288985979551274,
      "grad_norm": 2.106905937194824,
      "learning_rate": 0.00015669040007423784,
      "loss": 1.5709,
      "step": 9184
    },
    {
      "epoch": 0.3289344100846957,
      "grad_norm": 2.200146198272705,
      "learning_rate": 0.00015668084461322108,
      "loss": 1.6029,
      "step": 9185
    },
    {
      "epoch": 0.328970222214264,
      "grad_norm": 1.6008678674697876,
      "learning_rate": 0.00015667128838963668,
      "loss": 1.6772,
      "step": 9186
    },
    {
      "epoch": 0.32900603434383224,
      "grad_norm": 1.828938603401184,
      "learning_rate": 0.00015666173140361315,
      "loss": 1.7683,
      "step": 9187
    },
    {
      "epoch": 0.32904184647340057,
      "grad_norm": 2.305866241455078,
      "learning_rate": 0.00015665217365527917,
      "loss": 1.7502,
      "step": 9188
    },
    {
      "epoch": 0.32907765860296884,
      "grad_norm": 2.0801093578338623,
      "learning_rate": 0.00015664261514476322,
      "loss": 1.3371,
      "step": 9189
    },
    {
      "epoch": 0.3291134707325371,
      "grad_norm": 1.4453667402267456,
      "learning_rate": 0.00015663305587219396,
      "loss": 1.477,
      "step": 9190
    },
    {
      "epoch": 0.3291492828621054,
      "grad_norm": 1.6667442321777344,
      "learning_rate": 0.00015662349583770002,
      "loss": 1.8024,
      "step": 9191
    },
    {
      "epoch": 0.3291850949916737,
      "grad_norm": 1.4693574905395508,
      "learning_rate": 0.00015661393504140994,
      "loss": 1.3105,
      "step": 9192
    },
    {
      "epoch": 0.32922090712124197,
      "grad_norm": 1.4278935194015503,
      "learning_rate": 0.0001566043734834524,
      "loss": 1.8824,
      "step": 9193
    },
    {
      "epoch": 0.32925671925081024,
      "grad_norm": 2.0216543674468994,
      "learning_rate": 0.00015659481116395604,
      "loss": 1.6132,
      "step": 9194
    },
    {
      "epoch": 0.32929253138037856,
      "grad_norm": 1.8408523797988892,
      "learning_rate": 0.0001565852480830495,
      "loss": 1.191,
      "step": 9195
    },
    {
      "epoch": 0.32932834350994683,
      "grad_norm": 1.845346450805664,
      "learning_rate": 0.00015657568424086145,
      "loss": 1.8479,
      "step": 9196
    },
    {
      "epoch": 0.3293641556395151,
      "grad_norm": 1.3020843267440796,
      "learning_rate": 0.0001565661196375205,
      "loss": 1.3387,
      "step": 9197
    },
    {
      "epoch": 0.32939996776908337,
      "grad_norm": 1.6613608598709106,
      "learning_rate": 0.00015655655427315542,
      "loss": 1.4907,
      "step": 9198
    },
    {
      "epoch": 0.3294357798986517,
      "grad_norm": 1.7599493265151978,
      "learning_rate": 0.00015654698814789484,
      "loss": 1.5625,
      "step": 9199
    },
    {
      "epoch": 0.32947159202821996,
      "grad_norm": 2.014738082885742,
      "learning_rate": 0.00015653742126186745,
      "loss": 1.6346,
      "step": 9200
    },
    {
      "epoch": 0.32950740415778823,
      "grad_norm": 1.7053759098052979,
      "learning_rate": 0.00015652785361520204,
      "loss": 1.7228,
      "step": 9201
    },
    {
      "epoch": 0.32954321628735656,
      "grad_norm": 1.6620781421661377,
      "learning_rate": 0.00015651828520802722,
      "loss": 1.5416,
      "step": 9202
    },
    {
      "epoch": 0.3295790284169248,
      "grad_norm": 1.5674805641174316,
      "learning_rate": 0.00015650871604047182,
      "loss": 1.6512,
      "step": 9203
    },
    {
      "epoch": 0.3296148405464931,
      "grad_norm": 1.6600102186203003,
      "learning_rate": 0.0001564991461126645,
      "loss": 1.8433,
      "step": 9204
    },
    {
      "epoch": 0.32965065267606136,
      "grad_norm": 2.119927167892456,
      "learning_rate": 0.00015648957542473406,
      "loss": 1.6382,
      "step": 9205
    },
    {
      "epoch": 0.3296864648056297,
      "grad_norm": 2.353123426437378,
      "learning_rate": 0.00015648000397680924,
      "loss": 1.4772,
      "step": 9206
    },
    {
      "epoch": 0.32972227693519796,
      "grad_norm": 2.0134799480438232,
      "learning_rate": 0.00015647043176901886,
      "loss": 1.6496,
      "step": 9207
    },
    {
      "epoch": 0.3297580890647662,
      "grad_norm": 1.255784034729004,
      "learning_rate": 0.00015646085880149162,
      "loss": 1.6315,
      "step": 9208
    },
    {
      "epoch": 0.3297939011943345,
      "grad_norm": 1.3342036008834839,
      "learning_rate": 0.00015645128507435637,
      "loss": 1.3743,
      "step": 9209
    },
    {
      "epoch": 0.3298297133239028,
      "grad_norm": 1.8134342432022095,
      "learning_rate": 0.00015644171058774192,
      "loss": 1.6786,
      "step": 9210
    },
    {
      "epoch": 0.3298655254534711,
      "grad_norm": 1.3972450494766235,
      "learning_rate": 0.000156432135341777,
      "loss": 1.6555,
      "step": 9211
    },
    {
      "epoch": 0.32990133758303936,
      "grad_norm": 1.5140109062194824,
      "learning_rate": 0.00015642255933659053,
      "loss": 1.5552,
      "step": 9212
    },
    {
      "epoch": 0.3299371497126077,
      "grad_norm": 1.5048795938491821,
      "learning_rate": 0.0001564129825723113,
      "loss": 1.4055,
      "step": 9213
    },
    {
      "epoch": 0.32997296184217595,
      "grad_norm": 2.151780605316162,
      "learning_rate": 0.00015640340504906818,
      "loss": 1.5771,
      "step": 9214
    },
    {
      "epoch": 0.3300087739717442,
      "grad_norm": 1.7445200681686401,
      "learning_rate": 0.00015639382676698997,
      "loss": 1.8844,
      "step": 9215
    },
    {
      "epoch": 0.3300445861013125,
      "grad_norm": 1.9518778324127197,
      "learning_rate": 0.00015638424772620554,
      "loss": 2.0622,
      "step": 9216
    },
    {
      "epoch": 0.3300803982308808,
      "grad_norm": 2.5693469047546387,
      "learning_rate": 0.00015637466792684383,
      "loss": 1.3957,
      "step": 9217
    },
    {
      "epoch": 0.3301162103604491,
      "grad_norm": 1.5530214309692383,
      "learning_rate": 0.00015636508736903366,
      "loss": 1.3895,
      "step": 9218
    },
    {
      "epoch": 0.33015202249001735,
      "grad_norm": 1.2154394388198853,
      "learning_rate": 0.00015635550605290396,
      "loss": 1.561,
      "step": 9219
    },
    {
      "epoch": 0.3301878346195857,
      "grad_norm": 1.5539181232452393,
      "learning_rate": 0.00015634592397858362,
      "loss": 1.2993,
      "step": 9220
    },
    {
      "epoch": 0.33022364674915394,
      "grad_norm": 2.4144694805145264,
      "learning_rate": 0.00015633634114620154,
      "loss": 2.0973,
      "step": 9221
    },
    {
      "epoch": 0.3302594588787222,
      "grad_norm": 1.9893107414245605,
      "learning_rate": 0.00015632675755588668,
      "loss": 2.0019,
      "step": 9222
    },
    {
      "epoch": 0.3302952710082905,
      "grad_norm": 1.4870593547821045,
      "learning_rate": 0.00015631717320776795,
      "loss": 1.1392,
      "step": 9223
    },
    {
      "epoch": 0.3303310831378588,
      "grad_norm": 1.4275274276733398,
      "learning_rate": 0.00015630758810197427,
      "loss": 1.6437,
      "step": 9224
    },
    {
      "epoch": 0.3303668952674271,
      "grad_norm": 1.9945303201675415,
      "learning_rate": 0.00015629800223863465,
      "loss": 1.5218,
      "step": 9225
    },
    {
      "epoch": 0.33040270739699534,
      "grad_norm": 1.4058603048324585,
      "learning_rate": 0.000156288415617878,
      "loss": 1.4086,
      "step": 9226
    },
    {
      "epoch": 0.33043851952656367,
      "grad_norm": 1.57683265209198,
      "learning_rate": 0.00015627882823983336,
      "loss": 1.6094,
      "step": 9227
    },
    {
      "epoch": 0.33047433165613194,
      "grad_norm": 3.029585361480713,
      "learning_rate": 0.00015626924010462968,
      "loss": 1.7547,
      "step": 9228
    },
    {
      "epoch": 0.3305101437857002,
      "grad_norm": 1.5655510425567627,
      "learning_rate": 0.00015625965121239592,
      "loss": 1.8381,
      "step": 9229
    },
    {
      "epoch": 0.3305459559152685,
      "grad_norm": 2.5353496074676514,
      "learning_rate": 0.00015625006156326117,
      "loss": 1.8627,
      "step": 9230
    },
    {
      "epoch": 0.3305817680448368,
      "grad_norm": 1.9974932670593262,
      "learning_rate": 0.00015624047115735435,
      "loss": 1.605,
      "step": 9231
    },
    {
      "epoch": 0.33061758017440507,
      "grad_norm": 2.7682440280914307,
      "learning_rate": 0.00015623087999480458,
      "loss": 1.9736,
      "step": 9232
    },
    {
      "epoch": 0.33065339230397334,
      "grad_norm": 1.4476059675216675,
      "learning_rate": 0.00015622128807574081,
      "loss": 1.7133,
      "step": 9233
    },
    {
      "epoch": 0.33068920443354166,
      "grad_norm": 1.8865472078323364,
      "learning_rate": 0.00015621169540029216,
      "loss": 1.4769,
      "step": 9234
    },
    {
      "epoch": 0.33072501656310993,
      "grad_norm": 2.4696617126464844,
      "learning_rate": 0.00015620210196858763,
      "loss": 1.5348,
      "step": 9235
    },
    {
      "epoch": 0.3307608286926782,
      "grad_norm": 1.3538154363632202,
      "learning_rate": 0.00015619250778075634,
      "loss": 1.5639,
      "step": 9236
    },
    {
      "epoch": 0.33079664082224647,
      "grad_norm": 1.380228042602539,
      "learning_rate": 0.00015618291283692735,
      "loss": 1.4208,
      "step": 9237
    },
    {
      "epoch": 0.3308324529518148,
      "grad_norm": 1.5593613386154175,
      "learning_rate": 0.0001561733171372297,
      "loss": 1.4575,
      "step": 9238
    },
    {
      "epoch": 0.33086826508138306,
      "grad_norm": 1.0856677293777466,
      "learning_rate": 0.00015616372068179255,
      "loss": 1.287,
      "step": 9239
    },
    {
      "epoch": 0.33090407721095133,
      "grad_norm": 1.7727956771850586,
      "learning_rate": 0.00015615412347074498,
      "loss": 1.4814,
      "step": 9240
    },
    {
      "epoch": 0.33093988934051966,
      "grad_norm": 2.7936644554138184,
      "learning_rate": 0.0001561445255042161,
      "loss": 1.4023,
      "step": 9241
    },
    {
      "epoch": 0.3309757014700879,
      "grad_norm": 1.5259954929351807,
      "learning_rate": 0.00015613492678233509,
      "loss": 1.4711,
      "step": 9242
    },
    {
      "epoch": 0.3310115135996562,
      "grad_norm": 1.6369067430496216,
      "learning_rate": 0.000156125327305231,
      "loss": 1.6207,
      "step": 9243
    },
    {
      "epoch": 0.33104732572922446,
      "grad_norm": 1.5903338193893433,
      "learning_rate": 0.00015611572707303307,
      "loss": 1.5504,
      "step": 9244
    },
    {
      "epoch": 0.3310831378587928,
      "grad_norm": 1.5647993087768555,
      "learning_rate": 0.00015610612608587035,
      "loss": 1.722,
      "step": 9245
    },
    {
      "epoch": 0.33111894998836106,
      "grad_norm": 1.9212912321090698,
      "learning_rate": 0.00015609652434387216,
      "loss": 1.2589,
      "step": 9246
    },
    {
      "epoch": 0.3311547621179293,
      "grad_norm": 1.876175045967102,
      "learning_rate": 0.00015608692184716753,
      "loss": 1.5151,
      "step": 9247
    },
    {
      "epoch": 0.33119057424749765,
      "grad_norm": 2.0057926177978516,
      "learning_rate": 0.00015607731859588575,
      "loss": 1.4391,
      "step": 9248
    },
    {
      "epoch": 0.3312263863770659,
      "grad_norm": 1.4844675064086914,
      "learning_rate": 0.00015606771459015598,
      "loss": 1.8257,
      "step": 9249
    },
    {
      "epoch": 0.3312621985066342,
      "grad_norm": 1.5540400743484497,
      "learning_rate": 0.00015605810983010743,
      "loss": 1.2488,
      "step": 9250
    },
    {
      "epoch": 0.33129801063620246,
      "grad_norm": 2.0735278129577637,
      "learning_rate": 0.0001560485043158693,
      "loss": 1.4661,
      "step": 9251
    },
    {
      "epoch": 0.3313338227657708,
      "grad_norm": 1.5882608890533447,
      "learning_rate": 0.00015603889804757085,
      "loss": 1.4818,
      "step": 9252
    },
    {
      "epoch": 0.33136963489533905,
      "grad_norm": 1.3272738456726074,
      "learning_rate": 0.00015602929102534132,
      "loss": 1.6939,
      "step": 9253
    },
    {
      "epoch": 0.3314054470249073,
      "grad_norm": 1.2422311305999756,
      "learning_rate": 0.00015601968324930997,
      "loss": 1.3206,
      "step": 9254
    },
    {
      "epoch": 0.33144125915447564,
      "grad_norm": 1.8080556392669678,
      "learning_rate": 0.000156010074719606,
      "loss": 1.2622,
      "step": 9255
    },
    {
      "epoch": 0.3314770712840439,
      "grad_norm": 1.7431944608688354,
      "learning_rate": 0.00015600046543635875,
      "loss": 1.4804,
      "step": 9256
    },
    {
      "epoch": 0.3315128834136122,
      "grad_norm": 1.3665114641189575,
      "learning_rate": 0.0001559908553996975,
      "loss": 1.6605,
      "step": 9257
    },
    {
      "epoch": 0.33154869554318045,
      "grad_norm": 1.7952216863632202,
      "learning_rate": 0.00015598124460975148,
      "loss": 1.6897,
      "step": 9258
    },
    {
      "epoch": 0.3315845076727488,
      "grad_norm": 1.934351921081543,
      "learning_rate": 0.00015597163306665002,
      "loss": 1.2434,
      "step": 9259
    },
    {
      "epoch": 0.33162031980231704,
      "grad_norm": 1.282073974609375,
      "learning_rate": 0.00015596202077052245,
      "loss": 1.1964,
      "step": 9260
    },
    {
      "epoch": 0.3316561319318853,
      "grad_norm": 1.3141311407089233,
      "learning_rate": 0.00015595240772149803,
      "loss": 1.2027,
      "step": 9261
    },
    {
      "epoch": 0.33169194406145364,
      "grad_norm": 1.557449221611023,
      "learning_rate": 0.0001559427939197062,
      "loss": 1.7988,
      "step": 9262
    },
    {
      "epoch": 0.3317277561910219,
      "grad_norm": 1.4467054605484009,
      "learning_rate": 0.0001559331793652762,
      "loss": 1.2495,
      "step": 9263
    },
    {
      "epoch": 0.3317635683205902,
      "grad_norm": 1.3926703929901123,
      "learning_rate": 0.00015592356405833745,
      "loss": 1.5037,
      "step": 9264
    },
    {
      "epoch": 0.33179938045015844,
      "grad_norm": 2.078423261642456,
      "learning_rate": 0.00015591394799901927,
      "loss": 1.6851,
      "step": 9265
    },
    {
      "epoch": 0.33183519257972677,
      "grad_norm": 2.431959867477417,
      "learning_rate": 0.00015590433118745106,
      "loss": 1.8476,
      "step": 9266
    },
    {
      "epoch": 0.33187100470929504,
      "grad_norm": 1.8334934711456299,
      "learning_rate": 0.00015589471362376217,
      "loss": 1.8419,
      "step": 9267
    },
    {
      "epoch": 0.3319068168388633,
      "grad_norm": 1.5225297212600708,
      "learning_rate": 0.00015588509530808199,
      "loss": 1.9179,
      "step": 9268
    },
    {
      "epoch": 0.33194262896843163,
      "grad_norm": 1.3365960121154785,
      "learning_rate": 0.00015587547624053993,
      "loss": 1.1894,
      "step": 9269
    },
    {
      "epoch": 0.3319784410979999,
      "grad_norm": 2.1068549156188965,
      "learning_rate": 0.00015586585642126543,
      "loss": 1.6613,
      "step": 9270
    },
    {
      "epoch": 0.33201425322756817,
      "grad_norm": 1.5305715799331665,
      "learning_rate": 0.00015585623585038792,
      "loss": 1.4709,
      "step": 9271
    },
    {
      "epoch": 0.33205006535713644,
      "grad_norm": 1.8630884885787964,
      "learning_rate": 0.00015584661452803676,
      "loss": 1.6649,
      "step": 9272
    },
    {
      "epoch": 0.33208587748670476,
      "grad_norm": 1.756103754043579,
      "learning_rate": 0.00015583699245434146,
      "loss": 1.5741,
      "step": 9273
    },
    {
      "epoch": 0.33212168961627303,
      "grad_norm": 1.6057642698287964,
      "learning_rate": 0.00015582736962943148,
      "loss": 1.6483,
      "step": 9274
    },
    {
      "epoch": 0.3321575017458413,
      "grad_norm": 1.2686374187469482,
      "learning_rate": 0.00015581774605343622,
      "loss": 1.4546,
      "step": 9275
    },
    {
      "epoch": 0.3321933138754096,
      "grad_norm": 1.5505086183547974,
      "learning_rate": 0.0001558081217264852,
      "loss": 1.6977,
      "step": 9276
    },
    {
      "epoch": 0.3322291260049779,
      "grad_norm": 1.148769736289978,
      "learning_rate": 0.00015579849664870788,
      "loss": 1.4565,
      "step": 9277
    },
    {
      "epoch": 0.33226493813454616,
      "grad_norm": 1.3341801166534424,
      "learning_rate": 0.00015578887082023373,
      "loss": 1.4761,
      "step": 9278
    },
    {
      "epoch": 0.33230075026411443,
      "grad_norm": 1.950057029724121,
      "learning_rate": 0.00015577924424119233,
      "loss": 1.4574,
      "step": 9279
    },
    {
      "epoch": 0.33233656239368276,
      "grad_norm": 1.5431779623031616,
      "learning_rate": 0.00015576961691171314,
      "loss": 1.4851,
      "step": 9280
    },
    {
      "epoch": 0.332372374523251,
      "grad_norm": 1.7088158130645752,
      "learning_rate": 0.0001557599888319257,
      "loss": 1.7068,
      "step": 9281
    },
    {
      "epoch": 0.3324081866528193,
      "grad_norm": 2.0593230724334717,
      "learning_rate": 0.00015575036000195952,
      "loss": 1.4745,
      "step": 9282
    },
    {
      "epoch": 0.3324439987823876,
      "grad_norm": 1.5324299335479736,
      "learning_rate": 0.00015574073042194417,
      "loss": 1.8138,
      "step": 9283
    },
    {
      "epoch": 0.3324798109119559,
      "grad_norm": 1.420212984085083,
      "learning_rate": 0.0001557311000920092,
      "loss": 1.5369,
      "step": 9284
    },
    {
      "epoch": 0.33251562304152416,
      "grad_norm": 2.050830364227295,
      "learning_rate": 0.00015572146901228414,
      "loss": 1.6046,
      "step": 9285
    },
    {
      "epoch": 0.3325514351710924,
      "grad_norm": 1.4355131387710571,
      "learning_rate": 0.0001557118371828986,
      "loss": 1.4598,
      "step": 9286
    },
    {
      "epoch": 0.33258724730066075,
      "grad_norm": 2.0514612197875977,
      "learning_rate": 0.00015570220460398216,
      "loss": 1.3749,
      "step": 9287
    },
    {
      "epoch": 0.332623059430229,
      "grad_norm": 1.370781421661377,
      "learning_rate": 0.00015569257127566441,
      "loss": 1.6945,
      "step": 9288
    },
    {
      "epoch": 0.3326588715597973,
      "grad_norm": 1.860186219215393,
      "learning_rate": 0.00015568293719807493,
      "loss": 1.5742,
      "step": 9289
    },
    {
      "epoch": 0.3326946836893656,
      "grad_norm": 1.4769200086593628,
      "learning_rate": 0.00015567330237134338,
      "loss": 1.6708,
      "step": 9290
    },
    {
      "epoch": 0.3327304958189339,
      "grad_norm": 1.4108153581619263,
      "learning_rate": 0.00015566366679559937,
      "loss": 1.5285,
      "step": 9291
    },
    {
      "epoch": 0.33276630794850215,
      "grad_norm": 1.5427703857421875,
      "learning_rate": 0.0001556540304709725,
      "loss": 1.3293,
      "step": 9292
    },
    {
      "epoch": 0.3328021200780704,
      "grad_norm": 1.8373152017593384,
      "learning_rate": 0.00015564439339759245,
      "loss": 1.6855,
      "step": 9293
    },
    {
      "epoch": 0.33283793220763874,
      "grad_norm": 2.347149133682251,
      "learning_rate": 0.00015563475557558887,
      "loss": 1.4346,
      "step": 9294
    },
    {
      "epoch": 0.332873744337207,
      "grad_norm": 2.9432506561279297,
      "learning_rate": 0.00015562511700509138,
      "loss": 1.3162,
      "step": 9295
    },
    {
      "epoch": 0.3329095564667753,
      "grad_norm": 2.5354115962982178,
      "learning_rate": 0.00015561547768622974,
      "loss": 1.7069,
      "step": 9296
    },
    {
      "epoch": 0.3329453685963436,
      "grad_norm": 1.4742679595947266,
      "learning_rate": 0.00015560583761913357,
      "loss": 1.2812,
      "step": 9297
    },
    {
      "epoch": 0.3329811807259119,
      "grad_norm": 1.7055310010910034,
      "learning_rate": 0.00015559619680393256,
      "loss": 1.54,
      "step": 9298
    },
    {
      "epoch": 0.33301699285548014,
      "grad_norm": 1.5904908180236816,
      "learning_rate": 0.00015558655524075646,
      "loss": 1.4334,
      "step": 9299
    },
    {
      "epoch": 0.3330528049850484,
      "grad_norm": 1.9072233438491821,
      "learning_rate": 0.00015557691292973494,
      "loss": 1.4061,
      "step": 9300
    },
    {
      "epoch": 0.33308861711461674,
      "grad_norm": 1.6825381517410278,
      "learning_rate": 0.0001555672698709978,
      "loss": 1.3886,
      "step": 9301
    },
    {
      "epoch": 0.333124429244185,
      "grad_norm": 1.6894252300262451,
      "learning_rate": 0.00015555762606467465,
      "loss": 1.4306,
      "step": 9302
    },
    {
      "epoch": 0.3331602413737533,
      "grad_norm": 1.9015343189239502,
      "learning_rate": 0.00015554798151089534,
      "loss": 1.3311,
      "step": 9303
    },
    {
      "epoch": 0.3331960535033216,
      "grad_norm": 2.412245512008667,
      "learning_rate": 0.00015553833620978957,
      "loss": 1.5272,
      "step": 9304
    },
    {
      "epoch": 0.33323186563288987,
      "grad_norm": 2.0586354732513428,
      "learning_rate": 0.00015552869016148714,
      "loss": 1.8866,
      "step": 9305
    },
    {
      "epoch": 0.33326767776245814,
      "grad_norm": 1.8574638366699219,
      "learning_rate": 0.0001555190433661178,
      "loss": 1.4357,
      "step": 9306
    },
    {
      "epoch": 0.3333034898920264,
      "grad_norm": 1.5744131803512573,
      "learning_rate": 0.00015550939582381135,
      "loss": 1.3504,
      "step": 9307
    },
    {
      "epoch": 0.33333930202159473,
      "grad_norm": 1.7387135028839111,
      "learning_rate": 0.00015549974753469763,
      "loss": 1.5314,
      "step": 9308
    },
    {
      "epoch": 0.333375114151163,
      "grad_norm": 1.4193501472473145,
      "learning_rate": 0.00015549009849890634,
      "loss": 1.3785,
      "step": 9309
    },
    {
      "epoch": 0.33341092628073127,
      "grad_norm": 1.4320520162582397,
      "learning_rate": 0.0001554804487165674,
      "loss": 1.7415,
      "step": 9310
    },
    {
      "epoch": 0.3334467384102996,
      "grad_norm": 2.080427408218384,
      "learning_rate": 0.00015547079818781055,
      "loss": 1.4669,
      "step": 9311
    },
    {
      "epoch": 0.33348255053986786,
      "grad_norm": 2.3720574378967285,
      "learning_rate": 0.00015546114691276567,
      "loss": 1.5753,
      "step": 9312
    },
    {
      "epoch": 0.33351836266943613,
      "grad_norm": 1.8366647958755493,
      "learning_rate": 0.0001554514948915626,
      "loss": 1.5028,
      "step": 9313
    },
    {
      "epoch": 0.3335541747990044,
      "grad_norm": 1.6955592632293701,
      "learning_rate": 0.00015544184212433116,
      "loss": 1.5171,
      "step": 9314
    },
    {
      "epoch": 0.3335899869285727,
      "grad_norm": 2.041065216064453,
      "learning_rate": 0.00015543218861120125,
      "loss": 1.2655,
      "step": 9315
    },
    {
      "epoch": 0.333625799058141,
      "grad_norm": 1.9579260349273682,
      "learning_rate": 0.00015542253435230278,
      "loss": 1.4517,
      "step": 9316
    },
    {
      "epoch": 0.33366161118770926,
      "grad_norm": 1.472667932510376,
      "learning_rate": 0.0001554128793477656,
      "loss": 1.2546,
      "step": 9317
    },
    {
      "epoch": 0.3336974233172776,
      "grad_norm": 1.5596768856048584,
      "learning_rate": 0.0001554032235977196,
      "loss": 1.5515,
      "step": 9318
    },
    {
      "epoch": 0.33373323544684586,
      "grad_norm": 1.527860164642334,
      "learning_rate": 0.0001553935671022947,
      "loss": 1.3611,
      "step": 9319
    },
    {
      "epoch": 0.3337690475764141,
      "grad_norm": 1.6964343786239624,
      "learning_rate": 0.00015538390986162082,
      "loss": 1.5114,
      "step": 9320
    },
    {
      "epoch": 0.3338048597059824,
      "grad_norm": 1.6963163614273071,
      "learning_rate": 0.00015537425187582785,
      "loss": 1.7816,
      "step": 9321
    },
    {
      "epoch": 0.3338406718355507,
      "grad_norm": 1.6652700901031494,
      "learning_rate": 0.00015536459314504573,
      "loss": 1.8709,
      "step": 9322
    },
    {
      "epoch": 0.333876483965119,
      "grad_norm": 1.7200714349746704,
      "learning_rate": 0.00015535493366940442,
      "loss": 1.7631,
      "step": 9323
    },
    {
      "epoch": 0.33391229609468726,
      "grad_norm": 2.2434933185577393,
      "learning_rate": 0.0001553452734490339,
      "loss": 1.6587,
      "step": 9324
    },
    {
      "epoch": 0.3339481082242556,
      "grad_norm": 1.4792317152023315,
      "learning_rate": 0.00015533561248406413,
      "loss": 1.5377,
      "step": 9325
    },
    {
      "epoch": 0.33398392035382385,
      "grad_norm": 1.9175893068313599,
      "learning_rate": 0.00015532595077462507,
      "loss": 1.6146,
      "step": 9326
    },
    {
      "epoch": 0.3340197324833921,
      "grad_norm": 3.432957649230957,
      "learning_rate": 0.0001553162883208467,
      "loss": 1.5732,
      "step": 9327
    },
    {
      "epoch": 0.3340555446129604,
      "grad_norm": 1.9333349466323853,
      "learning_rate": 0.00015530662512285902,
      "loss": 1.211,
      "step": 9328
    },
    {
      "epoch": 0.3340913567425287,
      "grad_norm": 2.2683115005493164,
      "learning_rate": 0.00015529696118079205,
      "loss": 1.5071,
      "step": 9329
    },
    {
      "epoch": 0.334127168872097,
      "grad_norm": 1.9926609992980957,
      "learning_rate": 0.00015528729649477574,
      "loss": 1.571,
      "step": 9330
    },
    {
      "epoch": 0.33416298100166525,
      "grad_norm": 1.6616355180740356,
      "learning_rate": 0.00015527763106494024,
      "loss": 1.4256,
      "step": 9331
    },
    {
      "epoch": 0.3341987931312336,
      "grad_norm": 1.9569220542907715,
      "learning_rate": 0.0001552679648914155,
      "loss": 1.8484,
      "step": 9332
    },
    {
      "epoch": 0.33423460526080184,
      "grad_norm": 1.5473068952560425,
      "learning_rate": 0.00015525829797433157,
      "loss": 1.405,
      "step": 9333
    },
    {
      "epoch": 0.3342704173903701,
      "grad_norm": 1.8051518201828003,
      "learning_rate": 0.00015524863031381853,
      "loss": 1.2997,
      "step": 9334
    },
    {
      "epoch": 0.3343062295199384,
      "grad_norm": 1.7402265071868896,
      "learning_rate": 0.00015523896191000643,
      "loss": 1.7362,
      "step": 9335
    },
    {
      "epoch": 0.3343420416495067,
      "grad_norm": 1.6411949396133423,
      "learning_rate": 0.00015522929276302536,
      "loss": 1.3214,
      "step": 9336
    },
    {
      "epoch": 0.334377853779075,
      "grad_norm": 1.3993909358978271,
      "learning_rate": 0.0001552196228730054,
      "loss": 1.3546,
      "step": 9337
    },
    {
      "epoch": 0.33441366590864324,
      "grad_norm": 2.0956623554229736,
      "learning_rate": 0.00015520995224007662,
      "loss": 1.5105,
      "step": 9338
    },
    {
      "epoch": 0.33444947803821157,
      "grad_norm": 1.8981349468231201,
      "learning_rate": 0.00015520028086436915,
      "loss": 1.4384,
      "step": 9339
    },
    {
      "epoch": 0.33448529016777984,
      "grad_norm": 2.153280258178711,
      "learning_rate": 0.00015519060874601313,
      "loss": 1.3638,
      "step": 9340
    },
    {
      "epoch": 0.3345211022973481,
      "grad_norm": 1.4576733112335205,
      "learning_rate": 0.00015518093588513863,
      "loss": 1.5844,
      "step": 9341
    },
    {
      "epoch": 0.3345569144269164,
      "grad_norm": 1.9353009462356567,
      "learning_rate": 0.0001551712622818758,
      "loss": 1.6606,
      "step": 9342
    },
    {
      "epoch": 0.3345927265564847,
      "grad_norm": 1.6368129253387451,
      "learning_rate": 0.00015516158793635486,
      "loss": 1.2273,
      "step": 9343
    },
    {
      "epoch": 0.33462853868605297,
      "grad_norm": 1.8436633348464966,
      "learning_rate": 0.00015515191284870588,
      "loss": 1.771,
      "step": 9344
    },
    {
      "epoch": 0.33466435081562124,
      "grad_norm": 1.6525850296020508,
      "learning_rate": 0.00015514223701905904,
      "loss": 1.375,
      "step": 9345
    },
    {
      "epoch": 0.33470016294518956,
      "grad_norm": 1.5455073118209839,
      "learning_rate": 0.00015513256044754457,
      "loss": 1.6118,
      "step": 9346
    },
    {
      "epoch": 0.33473597507475783,
      "grad_norm": 1.7292460203170776,
      "learning_rate": 0.00015512288313429258,
      "loss": 1.6382,
      "step": 9347
    },
    {
      "epoch": 0.3347717872043261,
      "grad_norm": 2.00976824760437,
      "learning_rate": 0.0001551132050794333,
      "loss": 1.4821,
      "step": 9348
    },
    {
      "epoch": 0.33480759933389437,
      "grad_norm": 2.144354820251465,
      "learning_rate": 0.00015510352628309693,
      "loss": 1.8286,
      "step": 9349
    },
    {
      "epoch": 0.3348434114634627,
      "grad_norm": 1.8691902160644531,
      "learning_rate": 0.00015509384674541372,
      "loss": 1.5248,
      "step": 9350
    },
    {
      "epoch": 0.33487922359303096,
      "grad_norm": 1.6684008836746216,
      "learning_rate": 0.00015508416646651385,
      "loss": 1.5576,
      "step": 9351
    },
    {
      "epoch": 0.33491503572259923,
      "grad_norm": 1.8261315822601318,
      "learning_rate": 0.0001550744854465276,
      "loss": 1.704,
      "step": 9352
    },
    {
      "epoch": 0.33495084785216755,
      "grad_norm": 2.0527100563049316,
      "learning_rate": 0.00015506480368558516,
      "loss": 1.7708,
      "step": 9353
    },
    {
      "epoch": 0.3349866599817358,
      "grad_norm": 1.6841418743133545,
      "learning_rate": 0.00015505512118381683,
      "loss": 1.365,
      "step": 9354
    },
    {
      "epoch": 0.3350224721113041,
      "grad_norm": 1.531112790107727,
      "learning_rate": 0.00015504543794135284,
      "loss": 1.527,
      "step": 9355
    },
    {
      "epoch": 0.33505828424087236,
      "grad_norm": 1.4720302820205688,
      "learning_rate": 0.00015503575395832352,
      "loss": 1.4721,
      "step": 9356
    },
    {
      "epoch": 0.3350940963704407,
      "grad_norm": 2.4081616401672363,
      "learning_rate": 0.00015502606923485906,
      "loss": 1.4112,
      "step": 9357
    },
    {
      "epoch": 0.33512990850000896,
      "grad_norm": 1.4726104736328125,
      "learning_rate": 0.00015501638377108987,
      "loss": 1.4118,
      "step": 9358
    },
    {
      "epoch": 0.3351657206295772,
      "grad_norm": 1.6351170539855957,
      "learning_rate": 0.00015500669756714618,
      "loss": 1.341,
      "step": 9359
    },
    {
      "epoch": 0.33520153275914555,
      "grad_norm": 2.084728717803955,
      "learning_rate": 0.0001549970106231583,
      "loss": 1.457,
      "step": 9360
    },
    {
      "epoch": 0.3352373448887138,
      "grad_norm": 1.7264491319656372,
      "learning_rate": 0.00015498732293925667,
      "loss": 1.5515,
      "step": 9361
    },
    {
      "epoch": 0.3352731570182821,
      "grad_norm": 1.6600877046585083,
      "learning_rate": 0.00015497763451557148,
      "loss": 1.5989,
      "step": 9362
    },
    {
      "epoch": 0.33530896914785036,
      "grad_norm": 1.4888222217559814,
      "learning_rate": 0.00015496794535223315,
      "loss": 1.0686,
      "step": 9363
    },
    {
      "epoch": 0.3353447812774187,
      "grad_norm": 2.209500312805176,
      "learning_rate": 0.000154958255449372,
      "loss": 1.722,
      "step": 9364
    },
    {
      "epoch": 0.33538059340698695,
      "grad_norm": 1.8588663339614868,
      "learning_rate": 0.00015494856480711844,
      "loss": 1.3165,
      "step": 9365
    },
    {
      "epoch": 0.3354164055365552,
      "grad_norm": 2.067023754119873,
      "learning_rate": 0.0001549388734256028,
      "loss": 1.8567,
      "step": 9366
    },
    {
      "epoch": 0.33545221766612354,
      "grad_norm": 1.5753854513168335,
      "learning_rate": 0.00015492918130495547,
      "loss": 1.2895,
      "step": 9367
    },
    {
      "epoch": 0.3354880297956918,
      "grad_norm": 1.8132989406585693,
      "learning_rate": 0.0001549194884453069,
      "loss": 1.4248,
      "step": 9368
    },
    {
      "epoch": 0.3355238419252601,
      "grad_norm": 1.5888770818710327,
      "learning_rate": 0.00015490979484678743,
      "loss": 1.2666,
      "step": 9369
    },
    {
      "epoch": 0.33555965405482835,
      "grad_norm": 2.0815927982330322,
      "learning_rate": 0.0001549001005095275,
      "loss": 1.6768,
      "step": 9370
    },
    {
      "epoch": 0.3355954661843967,
      "grad_norm": 1.6123459339141846,
      "learning_rate": 0.00015489040543365754,
      "loss": 1.4473,
      "step": 9371
    },
    {
      "epoch": 0.33563127831396494,
      "grad_norm": 1.461475133895874,
      "learning_rate": 0.00015488070961930796,
      "loss": 1.4638,
      "step": 9372
    },
    {
      "epoch": 0.3356670904435332,
      "grad_norm": 1.3855684995651245,
      "learning_rate": 0.00015487101306660924,
      "loss": 1.3953,
      "step": 9373
    },
    {
      "epoch": 0.33570290257310154,
      "grad_norm": 2.608703851699829,
      "learning_rate": 0.00015486131577569182,
      "loss": 1.5456,
      "step": 9374
    },
    {
      "epoch": 0.3357387147026698,
      "grad_norm": 1.4365531206130981,
      "learning_rate": 0.00015485161774668615,
      "loss": 1.5599,
      "step": 9375
    },
    {
      "epoch": 0.3357745268322381,
      "grad_norm": 1.4558398723602295,
      "learning_rate": 0.00015484191897972274,
      "loss": 1.4977,
      "step": 9376
    },
    {
      "epoch": 0.33581033896180634,
      "grad_norm": 1.7082833051681519,
      "learning_rate": 0.000154832219474932,
      "loss": 1.3407,
      "step": 9377
    },
    {
      "epoch": 0.33584615109137467,
      "grad_norm": 1.5729615688323975,
      "learning_rate": 0.00015482251923244452,
      "loss": 1.4501,
      "step": 9378
    },
    {
      "epoch": 0.33588196322094294,
      "grad_norm": 1.2978506088256836,
      "learning_rate": 0.00015481281825239072,
      "loss": 1.5467,
      "step": 9379
    },
    {
      "epoch": 0.3359177753505112,
      "grad_norm": 1.3954700231552124,
      "learning_rate": 0.00015480311653490124,
      "loss": 1.4613,
      "step": 9380
    },
    {
      "epoch": 0.33595358748007953,
      "grad_norm": 1.4414048194885254,
      "learning_rate": 0.00015479341408010643,
      "loss": 1.4413,
      "step": 9381
    },
    {
      "epoch": 0.3359893996096478,
      "grad_norm": 1.9647263288497925,
      "learning_rate": 0.00015478371088813696,
      "loss": 1.8244,
      "step": 9382
    },
    {
      "epoch": 0.33602521173921607,
      "grad_norm": 1.619997262954712,
      "learning_rate": 0.0001547740069591233,
      "loss": 1.6385,
      "step": 9383
    },
    {
      "epoch": 0.33606102386878434,
      "grad_norm": 1.9023948907852173,
      "learning_rate": 0.00015476430229319603,
      "loss": 1.5266,
      "step": 9384
    },
    {
      "epoch": 0.33609683599835266,
      "grad_norm": 1.7760009765625,
      "learning_rate": 0.00015475459689048572,
      "loss": 1.5373,
      "step": 9385
    },
    {
      "epoch": 0.33613264812792093,
      "grad_norm": 1.5633459091186523,
      "learning_rate": 0.00015474489075112296,
      "loss": 1.4418,
      "step": 9386
    },
    {
      "epoch": 0.3361684602574892,
      "grad_norm": 1.4270317554473877,
      "learning_rate": 0.00015473518387523825,
      "loss": 1.5547,
      "step": 9387
    },
    {
      "epoch": 0.3362042723870575,
      "grad_norm": 2.286064386367798,
      "learning_rate": 0.0001547254762629623,
      "loss": 1.5236,
      "step": 9388
    },
    {
      "epoch": 0.3362400845166258,
      "grad_norm": 1.7473195791244507,
      "learning_rate": 0.00015471576791442564,
      "loss": 1.5732,
      "step": 9389
    },
    {
      "epoch": 0.33627589664619406,
      "grad_norm": 1.81753408908844,
      "learning_rate": 0.00015470605882975891,
      "loss": 1.4656,
      "step": 9390
    },
    {
      "epoch": 0.33631170877576233,
      "grad_norm": 1.6820625066757202,
      "learning_rate": 0.00015469634900909271,
      "loss": 1.4767,
      "step": 9391
    },
    {
      "epoch": 0.33634752090533065,
      "grad_norm": 1.315958023071289,
      "learning_rate": 0.00015468663845255768,
      "loss": 1.5724,
      "step": 9392
    },
    {
      "epoch": 0.3363833330348989,
      "grad_norm": 2.7729685306549072,
      "learning_rate": 0.0001546769271602845,
      "loss": 1.4956,
      "step": 9393
    },
    {
      "epoch": 0.3364191451644672,
      "grad_norm": 2.246842622756958,
      "learning_rate": 0.0001546672151324038,
      "loss": 1.6252,
      "step": 9394
    },
    {
      "epoch": 0.3364549572940355,
      "grad_norm": 1.5574842691421509,
      "learning_rate": 0.0001546575023690462,
      "loss": 1.6173,
      "step": 9395
    },
    {
      "epoch": 0.3364907694236038,
      "grad_norm": 1.6773743629455566,
      "learning_rate": 0.00015464778887034242,
      "loss": 1.4211,
      "step": 9396
    },
    {
      "epoch": 0.33652658155317206,
      "grad_norm": 1.4379688501358032,
      "learning_rate": 0.0001546380746364231,
      "loss": 1.5853,
      "step": 9397
    },
    {
      "epoch": 0.3365623936827403,
      "grad_norm": 1.9088571071624756,
      "learning_rate": 0.00015462835966741903,
      "loss": 0.9463,
      "step": 9398
    },
    {
      "epoch": 0.33659820581230865,
      "grad_norm": 1.3112964630126953,
      "learning_rate": 0.0001546186439634608,
      "loss": 1.6045,
      "step": 9399
    },
    {
      "epoch": 0.3366340179418769,
      "grad_norm": 1.9358813762664795,
      "learning_rate": 0.0001546089275246792,
      "loss": 1.6525,
      "step": 9400
    },
    {
      "epoch": 0.3366698300714452,
      "grad_norm": 1.9341341257095337,
      "learning_rate": 0.00015459921035120488,
      "loss": 1.5339,
      "step": 9401
    },
    {
      "epoch": 0.3367056422010135,
      "grad_norm": 1.2685867547988892,
      "learning_rate": 0.00015458949244316866,
      "loss": 1.5571,
      "step": 9402
    },
    {
      "epoch": 0.3367414543305818,
      "grad_norm": 1.3416786193847656,
      "learning_rate": 0.00015457977380070118,
      "loss": 1.4485,
      "step": 9403
    },
    {
      "epoch": 0.33677726646015005,
      "grad_norm": 2.085193157196045,
      "learning_rate": 0.00015457005442393327,
      "loss": 1.656,
      "step": 9404
    },
    {
      "epoch": 0.3368130785897183,
      "grad_norm": 1.7671416997909546,
      "learning_rate": 0.00015456033431299567,
      "loss": 1.3032,
      "step": 9405
    },
    {
      "epoch": 0.33684889071928664,
      "grad_norm": 1.6520533561706543,
      "learning_rate": 0.00015455061346801916,
      "loss": 1.3252,
      "step": 9406
    },
    {
      "epoch": 0.3368847028488549,
      "grad_norm": 2.0870721340179443,
      "learning_rate": 0.00015454089188913454,
      "loss": 1.3519,
      "step": 9407
    },
    {
      "epoch": 0.3369205149784232,
      "grad_norm": 2.2233269214630127,
      "learning_rate": 0.00015453116957647254,
      "loss": 1.4443,
      "step": 9408
    },
    {
      "epoch": 0.33695632710799145,
      "grad_norm": 1.454346776008606,
      "learning_rate": 0.00015452144653016397,
      "loss": 1.5947,
      "step": 9409
    },
    {
      "epoch": 0.3369921392375598,
      "grad_norm": 1.5266408920288086,
      "learning_rate": 0.0001545117227503397,
      "loss": 1.456,
      "step": 9410
    },
    {
      "epoch": 0.33702795136712804,
      "grad_norm": 1.5770138502120972,
      "learning_rate": 0.00015450199823713047,
      "loss": 1.4124,
      "step": 9411
    },
    {
      "epoch": 0.3370637634966963,
      "grad_norm": 1.7103294134140015,
      "learning_rate": 0.0001544922729906672,
      "loss": 1.4488,
      "step": 9412
    },
    {
      "epoch": 0.33709957562626464,
      "grad_norm": 1.5208357572555542,
      "learning_rate": 0.00015448254701108067,
      "loss": 1.437,
      "step": 9413
    },
    {
      "epoch": 0.3371353877558329,
      "grad_norm": 1.9375684261322021,
      "learning_rate": 0.00015447282029850174,
      "loss": 1.2997,
      "step": 9414
    },
    {
      "epoch": 0.3371711998854012,
      "grad_norm": 1.5922077894210815,
      "learning_rate": 0.00015446309285306131,
      "loss": 1.5067,
      "step": 9415
    },
    {
      "epoch": 0.33720701201496944,
      "grad_norm": 1.5397335290908813,
      "learning_rate": 0.0001544533646748902,
      "loss": 1.3838,
      "step": 9416
    },
    {
      "epoch": 0.33724282414453777,
      "grad_norm": 1.709815263748169,
      "learning_rate": 0.00015444363576411929,
      "loss": 1.3891,
      "step": 9417
    },
    {
      "epoch": 0.33727863627410604,
      "grad_norm": 1.6586835384368896,
      "learning_rate": 0.00015443390612087952,
      "loss": 1.3357,
      "step": 9418
    },
    {
      "epoch": 0.3373144484036743,
      "grad_norm": 1.817344069480896,
      "learning_rate": 0.00015442417574530173,
      "loss": 1.5462,
      "step": 9419
    },
    {
      "epoch": 0.33735026053324263,
      "grad_norm": 1.4759312868118286,
      "learning_rate": 0.00015441444463751687,
      "loss": 1.6142,
      "step": 9420
    },
    {
      "epoch": 0.3373860726628109,
      "grad_norm": 1.9167733192443848,
      "learning_rate": 0.00015440471279765583,
      "loss": 1.5566,
      "step": 9421
    },
    {
      "epoch": 0.33742188479237917,
      "grad_norm": 1.3484073877334595,
      "learning_rate": 0.00015439498022584957,
      "loss": 1.4768,
      "step": 9422
    },
    {
      "epoch": 0.33745769692194744,
      "grad_norm": 1.864970088005066,
      "learning_rate": 0.00015438524692222902,
      "loss": 1.5269,
      "step": 9423
    },
    {
      "epoch": 0.33749350905151576,
      "grad_norm": 1.4716098308563232,
      "learning_rate": 0.00015437551288692512,
      "loss": 1.452,
      "step": 9424
    },
    {
      "epoch": 0.33752932118108403,
      "grad_norm": 2.067917823791504,
      "learning_rate": 0.00015436577812006884,
      "loss": 1.4989,
      "step": 9425
    },
    {
      "epoch": 0.3375651333106523,
      "grad_norm": 2.316983938217163,
      "learning_rate": 0.00015435604262179116,
      "loss": 1.2659,
      "step": 9426
    },
    {
      "epoch": 0.3376009454402206,
      "grad_norm": 2.3151962757110596,
      "learning_rate": 0.000154346306392223,
      "loss": 1.5562,
      "step": 9427
    },
    {
      "epoch": 0.3376367575697889,
      "grad_norm": 1.9097533226013184,
      "learning_rate": 0.00015433656943149543,
      "loss": 1.7194,
      "step": 9428
    },
    {
      "epoch": 0.33767256969935716,
      "grad_norm": 1.6860368251800537,
      "learning_rate": 0.00015432683173973935,
      "loss": 1.7439,
      "step": 9429
    },
    {
      "epoch": 0.33770838182892543,
      "grad_norm": 1.6204484701156616,
      "learning_rate": 0.0001543170933170859,
      "loss": 1.4332,
      "step": 9430
    },
    {
      "epoch": 0.33774419395849375,
      "grad_norm": 1.807880163192749,
      "learning_rate": 0.00015430735416366596,
      "loss": 1.7937,
      "step": 9431
    },
    {
      "epoch": 0.337780006088062,
      "grad_norm": 1.4650651216506958,
      "learning_rate": 0.00015429761427961065,
      "loss": 1.456,
      "step": 9432
    },
    {
      "epoch": 0.3378158182176303,
      "grad_norm": 1.7693088054656982,
      "learning_rate": 0.00015428787366505094,
      "loss": 1.6138,
      "step": 9433
    },
    {
      "epoch": 0.3378516303471986,
      "grad_norm": 2.2674636840820312,
      "learning_rate": 0.00015427813232011799,
      "loss": 1.6677,
      "step": 9434
    },
    {
      "epoch": 0.3378874424767669,
      "grad_norm": 1.9544172286987305,
      "learning_rate": 0.00015426839024494272,
      "loss": 1.6052,
      "step": 9435
    },
    {
      "epoch": 0.33792325460633515,
      "grad_norm": 2.344970703125,
      "learning_rate": 0.0001542586474396563,
      "loss": 1.4323,
      "step": 9436
    },
    {
      "epoch": 0.3379590667359034,
      "grad_norm": 2.032620906829834,
      "learning_rate": 0.00015424890390438974,
      "loss": 1.5969,
      "step": 9437
    },
    {
      "epoch": 0.33799487886547175,
      "grad_norm": 1.735715627670288,
      "learning_rate": 0.00015423915963927418,
      "loss": 1.4165,
      "step": 9438
    },
    {
      "epoch": 0.33803069099504,
      "grad_norm": 1.2005674839019775,
      "learning_rate": 0.00015422941464444064,
      "loss": 1.569,
      "step": 9439
    },
    {
      "epoch": 0.3380665031246083,
      "grad_norm": 1.4805797338485718,
      "learning_rate": 0.00015421966892002032,
      "loss": 1.5567,
      "step": 9440
    },
    {
      "epoch": 0.3381023152541766,
      "grad_norm": 1.743964672088623,
      "learning_rate": 0.00015420992246614428,
      "loss": 1.5592,
      "step": 9441
    },
    {
      "epoch": 0.3381381273837449,
      "grad_norm": 1.772685170173645,
      "learning_rate": 0.00015420017528294368,
      "loss": 1.4956,
      "step": 9442
    },
    {
      "epoch": 0.33817393951331315,
      "grad_norm": 1.8625121116638184,
      "learning_rate": 0.00015419042737054963,
      "loss": 1.8012,
      "step": 9443
    },
    {
      "epoch": 0.3382097516428814,
      "grad_norm": 1.4339003562927246,
      "learning_rate": 0.00015418067872909326,
      "loss": 1.5661,
      "step": 9444
    },
    {
      "epoch": 0.33824556377244974,
      "grad_norm": 1.6561118364334106,
      "learning_rate": 0.00015417092935870574,
      "loss": 1.7139,
      "step": 9445
    },
    {
      "epoch": 0.338281375902018,
      "grad_norm": 1.8280638456344604,
      "learning_rate": 0.00015416117925951827,
      "loss": 1.143,
      "step": 9446
    },
    {
      "epoch": 0.3383171880315863,
      "grad_norm": 1.4793591499328613,
      "learning_rate": 0.000154151428431662,
      "loss": 1.422,
      "step": 9447
    },
    {
      "epoch": 0.3383530001611546,
      "grad_norm": 1.537424921989441,
      "learning_rate": 0.00015414167687526805,
      "loss": 1.3919,
      "step": 9448
    },
    {
      "epoch": 0.3383888122907229,
      "grad_norm": 2.0304996967315674,
      "learning_rate": 0.00015413192459046772,
      "loss": 1.3456,
      "step": 9449
    },
    {
      "epoch": 0.33842462442029114,
      "grad_norm": 1.8802131414413452,
      "learning_rate": 0.00015412217157739216,
      "loss": 1.2977,
      "step": 9450
    },
    {
      "epoch": 0.3384604365498594,
      "grad_norm": 1.9226833581924438,
      "learning_rate": 0.00015411241783617262,
      "loss": 1.4789,
      "step": 9451
    },
    {
      "epoch": 0.33849624867942774,
      "grad_norm": 1.7963732481002808,
      "learning_rate": 0.0001541026633669403,
      "loss": 1.3647,
      "step": 9452
    },
    {
      "epoch": 0.338532060808996,
      "grad_norm": 2.629523515701294,
      "learning_rate": 0.0001540929081698264,
      "loss": 1.3783,
      "step": 9453
    },
    {
      "epoch": 0.3385678729385643,
      "grad_norm": 1.9142543077468872,
      "learning_rate": 0.00015408315224496222,
      "loss": 1.5034,
      "step": 9454
    },
    {
      "epoch": 0.3386036850681326,
      "grad_norm": 1.6466937065124512,
      "learning_rate": 0.00015407339559247895,
      "loss": 1.5608,
      "step": 9455
    },
    {
      "epoch": 0.33863949719770087,
      "grad_norm": 1.5068143606185913,
      "learning_rate": 0.00015406363821250793,
      "loss": 1.5739,
      "step": 9456
    },
    {
      "epoch": 0.33867530932726914,
      "grad_norm": 1.6713124513626099,
      "learning_rate": 0.00015405388010518038,
      "loss": 1.9615,
      "step": 9457
    },
    {
      "epoch": 0.3387111214568374,
      "grad_norm": 2.409848928451538,
      "learning_rate": 0.00015404412127062762,
      "loss": 1.7276,
      "step": 9458
    },
    {
      "epoch": 0.33874693358640573,
      "grad_norm": 1.9540921449661255,
      "learning_rate": 0.00015403436170898088,
      "loss": 1.6689,
      "step": 9459
    },
    {
      "epoch": 0.338782745715974,
      "grad_norm": 1.4777239561080933,
      "learning_rate": 0.00015402460142037154,
      "loss": 1.4505,
      "step": 9460
    },
    {
      "epoch": 0.33881855784554227,
      "grad_norm": 2.009676694869995,
      "learning_rate": 0.00015401484040493085,
      "loss": 1.6446,
      "step": 9461
    },
    {
      "epoch": 0.3388543699751106,
      "grad_norm": 1.903566598892212,
      "learning_rate": 0.00015400507866279018,
      "loss": 1.5648,
      "step": 9462
    },
    {
      "epoch": 0.33889018210467886,
      "grad_norm": 1.4451779127120972,
      "learning_rate": 0.0001539953161940808,
      "loss": 1.6952,
      "step": 9463
    },
    {
      "epoch": 0.33892599423424713,
      "grad_norm": 2.141256332397461,
      "learning_rate": 0.00015398555299893412,
      "loss": 1.4004,
      "step": 9464
    },
    {
      "epoch": 0.3389618063638154,
      "grad_norm": 1.5447112321853638,
      "learning_rate": 0.00015397578907748146,
      "loss": 1.6946,
      "step": 9465
    },
    {
      "epoch": 0.3389976184933837,
      "grad_norm": 1.8307271003723145,
      "learning_rate": 0.00015396602442985417,
      "loss": 1.4,
      "step": 9466
    },
    {
      "epoch": 0.339033430622952,
      "grad_norm": 1.5202934741973877,
      "learning_rate": 0.00015395625905618364,
      "loss": 1.52,
      "step": 9467
    },
    {
      "epoch": 0.33906924275252026,
      "grad_norm": 1.4197298288345337,
      "learning_rate": 0.00015394649295660123,
      "loss": 1.6863,
      "step": 9468
    },
    {
      "epoch": 0.3391050548820886,
      "grad_norm": 1.4701685905456543,
      "learning_rate": 0.00015393672613123836,
      "loss": 1.3284,
      "step": 9469
    },
    {
      "epoch": 0.33914086701165685,
      "grad_norm": 1.5090365409851074,
      "learning_rate": 0.0001539269585802264,
      "loss": 1.5528,
      "step": 9470
    },
    {
      "epoch": 0.3391766791412251,
      "grad_norm": 1.5357013940811157,
      "learning_rate": 0.0001539171903036968,
      "loss": 1.5706,
      "step": 9471
    },
    {
      "epoch": 0.3392124912707934,
      "grad_norm": 1.3371944427490234,
      "learning_rate": 0.0001539074213017809,
      "loss": 1.2997,
      "step": 9472
    },
    {
      "epoch": 0.3392483034003617,
      "grad_norm": 1.5634480714797974,
      "learning_rate": 0.00015389765157461022,
      "loss": 1.2902,
      "step": 9473
    },
    {
      "epoch": 0.33928411552993,
      "grad_norm": 1.4515788555145264,
      "learning_rate": 0.00015388788112231615,
      "loss": 1.2752,
      "step": 9474
    },
    {
      "epoch": 0.33931992765949825,
      "grad_norm": 1.9081002473831177,
      "learning_rate": 0.00015387810994503016,
      "loss": 1.4576,
      "step": 9475
    },
    {
      "epoch": 0.3393557397890666,
      "grad_norm": 1.6875176429748535,
      "learning_rate": 0.0001538683380428837,
      "loss": 1.1983,
      "step": 9476
    },
    {
      "epoch": 0.33939155191863485,
      "grad_norm": 2.4767041206359863,
      "learning_rate": 0.00015385856541600825,
      "loss": 1.47,
      "step": 9477
    },
    {
      "epoch": 0.3394273640482031,
      "grad_norm": 1.4980638027191162,
      "learning_rate": 0.00015384879206453524,
      "loss": 1.4585,
      "step": 9478
    },
    {
      "epoch": 0.3394631761777714,
      "grad_norm": 2.4069361686706543,
      "learning_rate": 0.00015383901798859622,
      "loss": 1.485,
      "step": 9479
    },
    {
      "epoch": 0.3394989883073397,
      "grad_norm": 1.3716156482696533,
      "learning_rate": 0.00015382924318832264,
      "loss": 1.4037,
      "step": 9480
    },
    {
      "epoch": 0.339534800436908,
      "grad_norm": 1.2617043256759644,
      "learning_rate": 0.00015381946766384602,
      "loss": 1.4763,
      "step": 9481
    },
    {
      "epoch": 0.33957061256647625,
      "grad_norm": 1.9273022413253784,
      "learning_rate": 0.0001538096914152979,
      "loss": 1.676,
      "step": 9482
    },
    {
      "epoch": 0.3396064246960446,
      "grad_norm": 1.8018248081207275,
      "learning_rate": 0.00015379991444280979,
      "loss": 1.6164,
      "step": 9483
    },
    {
      "epoch": 0.33964223682561284,
      "grad_norm": 1.717280387878418,
      "learning_rate": 0.00015379013674651323,
      "loss": 1.6003,
      "step": 9484
    },
    {
      "epoch": 0.3396780489551811,
      "grad_norm": 1.4812376499176025,
      "learning_rate": 0.00015378035832653975,
      "loss": 1.7848,
      "step": 9485
    },
    {
      "epoch": 0.3397138610847494,
      "grad_norm": 1.6170680522918701,
      "learning_rate": 0.00015377057918302097,
      "loss": 1.567,
      "step": 9486
    },
    {
      "epoch": 0.3397496732143177,
      "grad_norm": 1.3598394393920898,
      "learning_rate": 0.00015376079931608838,
      "loss": 1.5678,
      "step": 9487
    },
    {
      "epoch": 0.339785485343886,
      "grad_norm": 1.3419015407562256,
      "learning_rate": 0.00015375101872587357,
      "loss": 1.4495,
      "step": 9488
    },
    {
      "epoch": 0.33982129747345424,
      "grad_norm": 1.6549489498138428,
      "learning_rate": 0.00015374123741250815,
      "loss": 1.4173,
      "step": 9489
    },
    {
      "epoch": 0.33985710960302257,
      "grad_norm": 1.4437153339385986,
      "learning_rate": 0.00015373145537612369,
      "loss": 1.5415,
      "step": 9490
    },
    {
      "epoch": 0.33989292173259084,
      "grad_norm": 3.1351869106292725,
      "learning_rate": 0.00015372167261685178,
      "loss": 1.3433,
      "step": 9491
    },
    {
      "epoch": 0.3399287338621591,
      "grad_norm": 2.001826763153076,
      "learning_rate": 0.0001537118891348241,
      "loss": 1.3978,
      "step": 9492
    },
    {
      "epoch": 0.3399645459917274,
      "grad_norm": 1.6999043226242065,
      "learning_rate": 0.00015370210493017222,
      "loss": 1.4984,
      "step": 9493
    },
    {
      "epoch": 0.3400003581212957,
      "grad_norm": 2.0367825031280518,
      "learning_rate": 0.00015369232000302777,
      "loss": 1.7663,
      "step": 9494
    },
    {
      "epoch": 0.34003617025086397,
      "grad_norm": 2.383127212524414,
      "learning_rate": 0.00015368253435352246,
      "loss": 1.5093,
      "step": 9495
    },
    {
      "epoch": 0.34007198238043224,
      "grad_norm": 1.8723777532577515,
      "learning_rate": 0.00015367274798178788,
      "loss": 1.5293,
      "step": 9496
    },
    {
      "epoch": 0.34010779451000056,
      "grad_norm": 1.9914052486419678,
      "learning_rate": 0.0001536629608879557,
      "loss": 1.4791,
      "step": 9497
    },
    {
      "epoch": 0.34014360663956883,
      "grad_norm": 1.547864317893982,
      "learning_rate": 0.00015365317307215759,
      "loss": 1.7745,
      "step": 9498
    },
    {
      "epoch": 0.3401794187691371,
      "grad_norm": 2.0833046436309814,
      "learning_rate": 0.00015364338453452528,
      "loss": 1.5529,
      "step": 9499
    },
    {
      "epoch": 0.34021523089870537,
      "grad_norm": 1.4355981349945068,
      "learning_rate": 0.00015363359527519036,
      "loss": 1.5735,
      "step": 9500
    },
    {
      "epoch": 0.3402510430282737,
      "grad_norm": 2.000267267227173,
      "learning_rate": 0.00015362380529428466,
      "loss": 1.4826,
      "step": 9501
    },
    {
      "epoch": 0.34028685515784196,
      "grad_norm": 2.343449831008911,
      "learning_rate": 0.0001536140145919398,
      "loss": 1.479,
      "step": 9502
    },
    {
      "epoch": 0.34032266728741023,
      "grad_norm": 1.5564789772033691,
      "learning_rate": 0.00015360422316828754,
      "loss": 1.6355,
      "step": 9503
    },
    {
      "epoch": 0.34035847941697855,
      "grad_norm": 1.6426461935043335,
      "learning_rate": 0.0001535944310234596,
      "loss": 1.7457,
      "step": 9504
    },
    {
      "epoch": 0.3403942915465468,
      "grad_norm": 1.9233627319335938,
      "learning_rate": 0.0001535846381575877,
      "loss": 1.7404,
      "step": 9505
    },
    {
      "epoch": 0.3404301036761151,
      "grad_norm": 1.6464073657989502,
      "learning_rate": 0.00015357484457080366,
      "loss": 1.1418,
      "step": 9506
    },
    {
      "epoch": 0.34046591580568336,
      "grad_norm": 1.4640589952468872,
      "learning_rate": 0.00015356505026323917,
      "loss": 1.5402,
      "step": 9507
    },
    {
      "epoch": 0.3405017279352517,
      "grad_norm": 1.8246670961380005,
      "learning_rate": 0.00015355525523502603,
      "loss": 1.6663,
      "step": 9508
    },
    {
      "epoch": 0.34053754006481995,
      "grad_norm": 1.9067139625549316,
      "learning_rate": 0.00015354545948629598,
      "loss": 1.6515,
      "step": 9509
    },
    {
      "epoch": 0.3405733521943882,
      "grad_norm": 1.3062249422073364,
      "learning_rate": 0.00015353566301718087,
      "loss": 1.3262,
      "step": 9510
    },
    {
      "epoch": 0.34060916432395655,
      "grad_norm": 1.529366374015808,
      "learning_rate": 0.00015352586582781247,
      "loss": 1.5815,
      "step": 9511
    },
    {
      "epoch": 0.3406449764535248,
      "grad_norm": 1.1446207761764526,
      "learning_rate": 0.0001535160679183226,
      "loss": 1.3855,
      "step": 9512
    },
    {
      "epoch": 0.3406807885830931,
      "grad_norm": 1.392532467842102,
      "learning_rate": 0.00015350626928884307,
      "loss": 1.6394,
      "step": 9513
    },
    {
      "epoch": 0.34071660071266135,
      "grad_norm": 1.3918993473052979,
      "learning_rate": 0.00015349646993950567,
      "loss": 1.3734,
      "step": 9514
    },
    {
      "epoch": 0.3407524128422297,
      "grad_norm": 1.6025148630142212,
      "learning_rate": 0.00015348666987044228,
      "loss": 1.603,
      "step": 9515
    },
    {
      "epoch": 0.34078822497179795,
      "grad_norm": 1.2502943277359009,
      "learning_rate": 0.00015347686908178475,
      "loss": 1.2778,
      "step": 9516
    },
    {
      "epoch": 0.3408240371013662,
      "grad_norm": 1.4781215190887451,
      "learning_rate": 0.0001534670675736649,
      "loss": 1.2568,
      "step": 9517
    },
    {
      "epoch": 0.34085984923093454,
      "grad_norm": 1.4211047887802124,
      "learning_rate": 0.00015345726534621466,
      "loss": 1.7071,
      "step": 9518
    },
    {
      "epoch": 0.3408956613605028,
      "grad_norm": 1.5720802545547485,
      "learning_rate": 0.00015344746239956587,
      "loss": 1.6633,
      "step": 9519
    },
    {
      "epoch": 0.3409314734900711,
      "grad_norm": 1.248879075050354,
      "learning_rate": 0.00015343765873385037,
      "loss": 1.731,
      "step": 9520
    },
    {
      "epoch": 0.34096728561963935,
      "grad_norm": 1.5327391624450684,
      "learning_rate": 0.00015342785434920017,
      "loss": 1.3957,
      "step": 9521
    },
    {
      "epoch": 0.3410030977492077,
      "grad_norm": 1.6686993837356567,
      "learning_rate": 0.00015341804924574707,
      "loss": 1.5403,
      "step": 9522
    },
    {
      "epoch": 0.34103890987877594,
      "grad_norm": 1.3809535503387451,
      "learning_rate": 0.00015340824342362303,
      "loss": 1.6858,
      "step": 9523
    },
    {
      "epoch": 0.3410747220083442,
      "grad_norm": 1.8143247365951538,
      "learning_rate": 0.00015339843688295997,
      "loss": 1.5465,
      "step": 9524
    },
    {
      "epoch": 0.34111053413791254,
      "grad_norm": 1.9842431545257568,
      "learning_rate": 0.00015338862962388977,
      "loss": 1.294,
      "step": 9525
    },
    {
      "epoch": 0.3411463462674808,
      "grad_norm": 1.9405596256256104,
      "learning_rate": 0.0001533788216465445,
      "loss": 1.6388,
      "step": 9526
    },
    {
      "epoch": 0.3411821583970491,
      "grad_norm": 1.7968621253967285,
      "learning_rate": 0.00015336901295105596,
      "loss": 1.8472,
      "step": 9527
    },
    {
      "epoch": 0.34121797052661734,
      "grad_norm": 2.323481559753418,
      "learning_rate": 0.00015335920353755627,
      "loss": 1.5243,
      "step": 9528
    },
    {
      "epoch": 0.34125378265618567,
      "grad_norm": 3.580449342727661,
      "learning_rate": 0.00015334939340617726,
      "loss": 1.6631,
      "step": 9529
    },
    {
      "epoch": 0.34128959478575394,
      "grad_norm": 1.3738518953323364,
      "learning_rate": 0.00015333958255705102,
      "loss": 1.6034,
      "step": 9530
    },
    {
      "epoch": 0.3413254069153222,
      "grad_norm": 1.4165027141571045,
      "learning_rate": 0.00015332977099030953,
      "loss": 1.8126,
      "step": 9531
    },
    {
      "epoch": 0.34136121904489053,
      "grad_norm": 1.6622012853622437,
      "learning_rate": 0.0001533199587060847,
      "loss": 1.3838,
      "step": 9532
    },
    {
      "epoch": 0.3413970311744588,
      "grad_norm": 1.4645581245422363,
      "learning_rate": 0.0001533101457045086,
      "loss": 1.648,
      "step": 9533
    },
    {
      "epoch": 0.34143284330402707,
      "grad_norm": 1.7480566501617432,
      "learning_rate": 0.0001533003319857133,
      "loss": 1.3988,
      "step": 9534
    },
    {
      "epoch": 0.34146865543359534,
      "grad_norm": 1.2829252481460571,
      "learning_rate": 0.00015329051754983076,
      "loss": 1.313,
      "step": 9535
    },
    {
      "epoch": 0.34150446756316366,
      "grad_norm": 1.990755558013916,
      "learning_rate": 0.00015328070239699305,
      "loss": 1.4727,
      "step": 9536
    },
    {
      "epoch": 0.34154027969273193,
      "grad_norm": 1.4537103176116943,
      "learning_rate": 0.0001532708865273322,
      "loss": 1.4554,
      "step": 9537
    },
    {
      "epoch": 0.3415760918223002,
      "grad_norm": 1.6201869249343872,
      "learning_rate": 0.0001532610699409803,
      "loss": 1.5224,
      "step": 9538
    },
    {
      "epoch": 0.3416119039518685,
      "grad_norm": 1.5027778148651123,
      "learning_rate": 0.00015325125263806943,
      "loss": 1.4322,
      "step": 9539
    },
    {
      "epoch": 0.3416477160814368,
      "grad_norm": 1.5254660844802856,
      "learning_rate": 0.0001532414346187316,
      "loss": 1.5606,
      "step": 9540
    },
    {
      "epoch": 0.34168352821100506,
      "grad_norm": 1.727812647819519,
      "learning_rate": 0.000153231615883099,
      "loss": 1.6136,
      "step": 9541
    },
    {
      "epoch": 0.34171934034057333,
      "grad_norm": 2.322026491165161,
      "learning_rate": 0.0001532217964313036,
      "loss": 1.5319,
      "step": 9542
    },
    {
      "epoch": 0.34175515247014165,
      "grad_norm": 2.0224554538726807,
      "learning_rate": 0.00015321197626347766,
      "loss": 1.6205,
      "step": 9543
    },
    {
      "epoch": 0.3417909645997099,
      "grad_norm": 1.3443764448165894,
      "learning_rate": 0.00015320215537975313,
      "loss": 1.6137,
      "step": 9544
    },
    {
      "epoch": 0.3418267767292782,
      "grad_norm": 1.777479648590088,
      "learning_rate": 0.0001531923337802623,
      "loss": 1.7252,
      "step": 9545
    },
    {
      "epoch": 0.3418625888588465,
      "grad_norm": 1.4354276657104492,
      "learning_rate": 0.0001531825114651372,
      "loss": 1.5497,
      "step": 9546
    },
    {
      "epoch": 0.3418984009884148,
      "grad_norm": 2.5619523525238037,
      "learning_rate": 0.00015317268843451003,
      "loss": 1.5586,
      "step": 9547
    },
    {
      "epoch": 0.34193421311798305,
      "grad_norm": 1.8442054986953735,
      "learning_rate": 0.0001531628646885129,
      "loss": 1.6815,
      "step": 9548
    },
    {
      "epoch": 0.3419700252475513,
      "grad_norm": 1.897230625152588,
      "learning_rate": 0.000153153040227278,
      "loss": 1.4317,
      "step": 9549
    },
    {
      "epoch": 0.34200583737711965,
      "grad_norm": 1.4326685667037964,
      "learning_rate": 0.00015314321505093751,
      "loss": 1.6475,
      "step": 9550
    },
    {
      "epoch": 0.3420416495066879,
      "grad_norm": 2.543628692626953,
      "learning_rate": 0.00015313338915962362,
      "loss": 1.6623,
      "step": 9551
    },
    {
      "epoch": 0.3420774616362562,
      "grad_norm": 1.4500302076339722,
      "learning_rate": 0.0001531235625534685,
      "loss": 1.3383,
      "step": 9552
    },
    {
      "epoch": 0.3421132737658245,
      "grad_norm": 1.4041082859039307,
      "learning_rate": 0.00015311373523260437,
      "loss": 1.3654,
      "step": 9553
    },
    {
      "epoch": 0.3421490858953928,
      "grad_norm": 1.5999187231063843,
      "learning_rate": 0.00015310390719716348,
      "loss": 1.5783,
      "step": 9554
    },
    {
      "epoch": 0.34218489802496105,
      "grad_norm": 1.8230500221252441,
      "learning_rate": 0.000153094078447278,
      "loss": 1.4435,
      "step": 9555
    },
    {
      "epoch": 0.3422207101545293,
      "grad_norm": 2.0518381595611572,
      "learning_rate": 0.00015308424898308017,
      "loss": 1.5841,
      "step": 9556
    },
    {
      "epoch": 0.34225652228409764,
      "grad_norm": 1.2910943031311035,
      "learning_rate": 0.00015307441880470227,
      "loss": 1.5275,
      "step": 9557
    },
    {
      "epoch": 0.3422923344136659,
      "grad_norm": 1.3538620471954346,
      "learning_rate": 0.00015306458791227646,
      "loss": 1.685,
      "step": 9558
    },
    {
      "epoch": 0.3423281465432342,
      "grad_norm": 2.853557586669922,
      "learning_rate": 0.00015305475630593516,
      "loss": 1.5067,
      "step": 9559
    },
    {
      "epoch": 0.3423639586728025,
      "grad_norm": 2.141958713531494,
      "learning_rate": 0.00015304492398581046,
      "loss": 1.4039,
      "step": 9560
    },
    {
      "epoch": 0.3423997708023708,
      "grad_norm": 2.1874539852142334,
      "learning_rate": 0.0001530350909520348,
      "loss": 1.6577,
      "step": 9561
    },
    {
      "epoch": 0.34243558293193904,
      "grad_norm": 1.843515396118164,
      "learning_rate": 0.00015302525720474038,
      "loss": 1.6085,
      "step": 9562
    },
    {
      "epoch": 0.3424713950615073,
      "grad_norm": 2.459810733795166,
      "learning_rate": 0.00015301542274405948,
      "loss": 1.5343,
      "step": 9563
    },
    {
      "epoch": 0.34250720719107564,
      "grad_norm": 2.1196999549865723,
      "learning_rate": 0.00015300558757012448,
      "loss": 1.2307,
      "step": 9564
    },
    {
      "epoch": 0.3425430193206439,
      "grad_norm": 5.689758777618408,
      "learning_rate": 0.00015299575168306774,
      "loss": 1.5805,
      "step": 9565
    },
    {
      "epoch": 0.3425788314502122,
      "grad_norm": 1.7519400119781494,
      "learning_rate": 0.00015298591508302142,
      "loss": 1.3325,
      "step": 9566
    },
    {
      "epoch": 0.3426146435797805,
      "grad_norm": 1.4828169345855713,
      "learning_rate": 0.000152976077770118,
      "loss": 1.5663,
      "step": 9567
    },
    {
      "epoch": 0.34265045570934877,
      "grad_norm": 1.916289210319519,
      "learning_rate": 0.00015296623974448982,
      "loss": 1.7129,
      "step": 9568
    },
    {
      "epoch": 0.34268626783891704,
      "grad_norm": 1.342599630355835,
      "learning_rate": 0.00015295640100626914,
      "loss": 1.3287,
      "step": 9569
    },
    {
      "epoch": 0.3427220799684853,
      "grad_norm": 1.6580992937088013,
      "learning_rate": 0.00015294656155558843,
      "loss": 1.2133,
      "step": 9570
    },
    {
      "epoch": 0.34275789209805363,
      "grad_norm": 1.2530838251113892,
      "learning_rate": 0.00015293672139258003,
      "loss": 1.3984,
      "step": 9571
    },
    {
      "epoch": 0.3427937042276219,
      "grad_norm": 1.6452621221542358,
      "learning_rate": 0.00015292688051737633,
      "loss": 1.5749,
      "step": 9572
    },
    {
      "epoch": 0.34282951635719017,
      "grad_norm": 1.9408454895019531,
      "learning_rate": 0.0001529170389301097,
      "loss": 1.5734,
      "step": 9573
    },
    {
      "epoch": 0.3428653284867585,
      "grad_norm": 1.6352133750915527,
      "learning_rate": 0.00015290719663091262,
      "loss": 1.5879,
      "step": 9574
    },
    {
      "epoch": 0.34290114061632676,
      "grad_norm": 2.4092254638671875,
      "learning_rate": 0.00015289735361991743,
      "loss": 1.4012,
      "step": 9575
    },
    {
      "epoch": 0.34293695274589503,
      "grad_norm": 1.5429397821426392,
      "learning_rate": 0.00015288750989725657,
      "loss": 1.2422,
      "step": 9576
    },
    {
      "epoch": 0.3429727648754633,
      "grad_norm": 1.6832060813903809,
      "learning_rate": 0.00015287766546306247,
      "loss": 1.6365,
      "step": 9577
    },
    {
      "epoch": 0.3430085770050316,
      "grad_norm": 1.3857985734939575,
      "learning_rate": 0.00015286782031746763,
      "loss": 1.4832,
      "step": 9578
    },
    {
      "epoch": 0.3430443891345999,
      "grad_norm": 2.488194227218628,
      "learning_rate": 0.00015285797446060442,
      "loss": 1.8965,
      "step": 9579
    },
    {
      "epoch": 0.34308020126416816,
      "grad_norm": 1.5207929611206055,
      "learning_rate": 0.00015284812789260536,
      "loss": 1.5894,
      "step": 9580
    },
    {
      "epoch": 0.3431160133937365,
      "grad_norm": 1.9598724842071533,
      "learning_rate": 0.00015283828061360291,
      "loss": 1.423,
      "step": 9581
    },
    {
      "epoch": 0.34315182552330475,
      "grad_norm": 2.2789857387542725,
      "learning_rate": 0.00015282843262372955,
      "loss": 1.7208,
      "step": 9582
    },
    {
      "epoch": 0.343187637652873,
      "grad_norm": 1.4618828296661377,
      "learning_rate": 0.0001528185839231178,
      "loss": 1.7469,
      "step": 9583
    },
    {
      "epoch": 0.3432234497824413,
      "grad_norm": 1.8113200664520264,
      "learning_rate": 0.00015280873451190008,
      "loss": 1.5084,
      "step": 9584
    },
    {
      "epoch": 0.3432592619120096,
      "grad_norm": 2.0830557346343994,
      "learning_rate": 0.000152798884390209,
      "loss": 1.5144,
      "step": 9585
    },
    {
      "epoch": 0.3432950740415779,
      "grad_norm": 2.5192768573760986,
      "learning_rate": 0.000152789033558177,
      "loss": 1.6695,
      "step": 9586
    },
    {
      "epoch": 0.34333088617114615,
      "grad_norm": 2.1534531116485596,
      "learning_rate": 0.0001527791820159367,
      "loss": 1.5468,
      "step": 9587
    },
    {
      "epoch": 0.3433666983007145,
      "grad_norm": 1.605986475944519,
      "learning_rate": 0.00015276932976362052,
      "loss": 1.2892,
      "step": 9588
    },
    {
      "epoch": 0.34340251043028275,
      "grad_norm": 1.4528756141662598,
      "learning_rate": 0.00015275947680136112,
      "loss": 1.5642,
      "step": 9589
    },
    {
      "epoch": 0.343438322559851,
      "grad_norm": 1.5881379842758179,
      "learning_rate": 0.000152749623129291,
      "loss": 1.5147,
      "step": 9590
    },
    {
      "epoch": 0.3434741346894193,
      "grad_norm": 1.604173183441162,
      "learning_rate": 0.00015273976874754274,
      "loss": 1.5219,
      "step": 9591
    },
    {
      "epoch": 0.3435099468189876,
      "grad_norm": 1.7604161500930786,
      "learning_rate": 0.00015272991365624896,
      "loss": 1.727,
      "step": 9592
    },
    {
      "epoch": 0.3435457589485559,
      "grad_norm": 1.5851354598999023,
      "learning_rate": 0.00015272005785554215,
      "loss": 1.7405,
      "step": 9593
    },
    {
      "epoch": 0.34358157107812415,
      "grad_norm": 1.7585153579711914,
      "learning_rate": 0.000152710201345555,
      "loss": 1.5413,
      "step": 9594
    },
    {
      "epoch": 0.3436173832076925,
      "grad_norm": 2.0630173683166504,
      "learning_rate": 0.00015270034412642007,
      "loss": 1.7264,
      "step": 9595
    },
    {
      "epoch": 0.34365319533726074,
      "grad_norm": 2.172152042388916,
      "learning_rate": 0.00015269048619827,
      "loss": 1.7205,
      "step": 9596
    },
    {
      "epoch": 0.343689007466829,
      "grad_norm": 2.5975513458251953,
      "learning_rate": 0.0001526806275612374,
      "loss": 1.9148,
      "step": 9597
    },
    {
      "epoch": 0.3437248195963973,
      "grad_norm": 1.559423804283142,
      "learning_rate": 0.00015267076821545489,
      "loss": 1.6451,
      "step": 9598
    },
    {
      "epoch": 0.3437606317259656,
      "grad_norm": 2.140960693359375,
      "learning_rate": 0.00015266090816105514,
      "loss": 1.9031,
      "step": 9599
    },
    {
      "epoch": 0.3437964438555339,
      "grad_norm": 1.62228262424469,
      "learning_rate": 0.00015265104739817082,
      "loss": 1.8635,
      "step": 9600
    },
    {
      "epoch": 0.34383225598510214,
      "grad_norm": 1.7803810834884644,
      "learning_rate": 0.00015264118592693457,
      "loss": 1.2803,
      "step": 9601
    },
    {
      "epoch": 0.34386806811467047,
      "grad_norm": 1.7218365669250488,
      "learning_rate": 0.00015263132374747907,
      "loss": 1.3518,
      "step": 9602
    },
    {
      "epoch": 0.34390388024423874,
      "grad_norm": 1.9356237649917603,
      "learning_rate": 0.00015262146085993697,
      "loss": 1.7451,
      "step": 9603
    },
    {
      "epoch": 0.343939692373807,
      "grad_norm": 1.4498659372329712,
      "learning_rate": 0.00015261159726444098,
      "loss": 1.7193,
      "step": 9604
    },
    {
      "epoch": 0.3439755045033753,
      "grad_norm": 1.850898027420044,
      "learning_rate": 0.00015260173296112385,
      "loss": 1.4767,
      "step": 9605
    },
    {
      "epoch": 0.3440113166329436,
      "grad_norm": 1.6529873609542847,
      "learning_rate": 0.00015259186795011823,
      "loss": 1.9236,
      "step": 9606
    },
    {
      "epoch": 0.34404712876251187,
      "grad_norm": 2.0878450870513916,
      "learning_rate": 0.0001525820022315569,
      "loss": 1.684,
      "step": 9607
    },
    {
      "epoch": 0.34408294089208014,
      "grad_norm": 1.974679708480835,
      "learning_rate": 0.0001525721358055725,
      "loss": 1.4546,
      "step": 9608
    },
    {
      "epoch": 0.3441187530216484,
      "grad_norm": 1.576131820678711,
      "learning_rate": 0.0001525622686722979,
      "loss": 1.5243,
      "step": 9609
    },
    {
      "epoch": 0.34415456515121673,
      "grad_norm": 2.246279716491699,
      "learning_rate": 0.00015255240083186572,
      "loss": 1.2719,
      "step": 9610
    },
    {
      "epoch": 0.344190377280785,
      "grad_norm": 1.4744218587875366,
      "learning_rate": 0.00015254253228440877,
      "loss": 1.569,
      "step": 9611
    },
    {
      "epoch": 0.34422618941035327,
      "grad_norm": 2.0905468463897705,
      "learning_rate": 0.00015253266303005987,
      "loss": 1.3189,
      "step": 9612
    },
    {
      "epoch": 0.3442620015399216,
      "grad_norm": 1.9109582901000977,
      "learning_rate": 0.00015252279306895172,
      "loss": 1.3807,
      "step": 9613
    },
    {
      "epoch": 0.34429781366948986,
      "grad_norm": 1.441930890083313,
      "learning_rate": 0.00015251292240121714,
      "loss": 1.3186,
      "step": 9614
    },
    {
      "epoch": 0.34433362579905813,
      "grad_norm": 1.6397720575332642,
      "learning_rate": 0.0001525030510269889,
      "loss": 1.477,
      "step": 9615
    },
    {
      "epoch": 0.3443694379286264,
      "grad_norm": 1.3502668142318726,
      "learning_rate": 0.00015249317894639987,
      "loss": 1.5195,
      "step": 9616
    },
    {
      "epoch": 0.3444052500581947,
      "grad_norm": 1.8765751123428345,
      "learning_rate": 0.00015248330615958282,
      "loss": 1.4637,
      "step": 9617
    },
    {
      "epoch": 0.344441062187763,
      "grad_norm": 1.7291653156280518,
      "learning_rate": 0.00015247343266667061,
      "loss": 1.5855,
      "step": 9618
    },
    {
      "epoch": 0.34447687431733126,
      "grad_norm": 1.695115566253662,
      "learning_rate": 0.00015246355846779602,
      "loss": 1.8087,
      "step": 9619
    },
    {
      "epoch": 0.3445126864468996,
      "grad_norm": 1.4280116558074951,
      "learning_rate": 0.00015245368356309194,
      "loss": 1.3901,
      "step": 9620
    },
    {
      "epoch": 0.34454849857646785,
      "grad_norm": 1.5115997791290283,
      "learning_rate": 0.00015244380795269118,
      "loss": 1.7692,
      "step": 9621
    },
    {
      "epoch": 0.3445843107060361,
      "grad_norm": 1.878031849861145,
      "learning_rate": 0.00015243393163672664,
      "loss": 1.469,
      "step": 9622
    },
    {
      "epoch": 0.3446201228356044,
      "grad_norm": 1.791983723640442,
      "learning_rate": 0.00015242405461533118,
      "loss": 1.6334,
      "step": 9623
    },
    {
      "epoch": 0.3446559349651727,
      "grad_norm": 1.6626172065734863,
      "learning_rate": 0.0001524141768886377,
      "loss": 1.527,
      "step": 9624
    },
    {
      "epoch": 0.344691747094741,
      "grad_norm": 2.501840829849243,
      "learning_rate": 0.0001524042984567791,
      "loss": 1.6678,
      "step": 9625
    },
    {
      "epoch": 0.34472755922430925,
      "grad_norm": 1.8956242799758911,
      "learning_rate": 0.0001523944193198882,
      "loss": 1.4583,
      "step": 9626
    },
    {
      "epoch": 0.3447633713538776,
      "grad_norm": 2.0565409660339355,
      "learning_rate": 0.00015238453947809805,
      "loss": 1.697,
      "step": 9627
    },
    {
      "epoch": 0.34479918348344585,
      "grad_norm": 2.0155301094055176,
      "learning_rate": 0.00015237465893154143,
      "loss": 1.71,
      "step": 9628
    },
    {
      "epoch": 0.3448349956130141,
      "grad_norm": 2.654571056365967,
      "learning_rate": 0.00015236477768035137,
      "loss": 1.5797,
      "step": 9629
    },
    {
      "epoch": 0.3448708077425824,
      "grad_norm": 1.6048696041107178,
      "learning_rate": 0.00015235489572466078,
      "loss": 1.7246,
      "step": 9630
    },
    {
      "epoch": 0.3449066198721507,
      "grad_norm": 2.183603525161743,
      "learning_rate": 0.00015234501306460256,
      "loss": 1.3855,
      "step": 9631
    },
    {
      "epoch": 0.344942432001719,
      "grad_norm": 1.7944676876068115,
      "learning_rate": 0.0001523351297003097,
      "loss": 1.585,
      "step": 9632
    },
    {
      "epoch": 0.34497824413128725,
      "grad_norm": 1.8598835468292236,
      "learning_rate": 0.00015232524563191523,
      "loss": 1.3623,
      "step": 9633
    },
    {
      "epoch": 0.3450140562608556,
      "grad_norm": 2.6025619506835938,
      "learning_rate": 0.00015231536085955205,
      "loss": 1.6624,
      "step": 9634
    },
    {
      "epoch": 0.34504986839042384,
      "grad_norm": 1.3773020505905151,
      "learning_rate": 0.00015230547538335317,
      "loss": 1.68,
      "step": 9635
    },
    {
      "epoch": 0.3450856805199921,
      "grad_norm": 2.4558515548706055,
      "learning_rate": 0.00015229558920345162,
      "loss": 1.5141,
      "step": 9636
    },
    {
      "epoch": 0.3451214926495604,
      "grad_norm": 1.6440409421920776,
      "learning_rate": 0.00015228570231998033,
      "loss": 1.5429,
      "step": 9637
    },
    {
      "epoch": 0.3451573047791287,
      "grad_norm": 1.8336268663406372,
      "learning_rate": 0.00015227581473307238,
      "loss": 1.559,
      "step": 9638
    },
    {
      "epoch": 0.345193116908697,
      "grad_norm": 1.4943885803222656,
      "learning_rate": 0.00015226592644286075,
      "loss": 1.3127,
      "step": 9639
    },
    {
      "epoch": 0.34522892903826524,
      "grad_norm": 1.8755306005477905,
      "learning_rate": 0.00015225603744947852,
      "loss": 1.5688,
      "step": 9640
    },
    {
      "epoch": 0.34526474116783357,
      "grad_norm": 1.2759945392608643,
      "learning_rate": 0.0001522461477530587,
      "loss": 1.5486,
      "step": 9641
    },
    {
      "epoch": 0.34530055329740184,
      "grad_norm": 1.2741237878799438,
      "learning_rate": 0.00015223625735373436,
      "loss": 1.5329,
      "step": 9642
    },
    {
      "epoch": 0.3453363654269701,
      "grad_norm": 1.827254056930542,
      "learning_rate": 0.00015222636625163854,
      "loss": 1.7189,
      "step": 9643
    },
    {
      "epoch": 0.3453721775565384,
      "grad_norm": 1.470656394958496,
      "learning_rate": 0.00015221647444690437,
      "loss": 1.3983,
      "step": 9644
    },
    {
      "epoch": 0.3454079896861067,
      "grad_norm": 1.6272372007369995,
      "learning_rate": 0.00015220658193966489,
      "loss": 1.4919,
      "step": 9645
    },
    {
      "epoch": 0.34544380181567497,
      "grad_norm": 1.6300824880599976,
      "learning_rate": 0.00015219668873005314,
      "loss": 1.567,
      "step": 9646
    },
    {
      "epoch": 0.34547961394524324,
      "grad_norm": 1.6557114124298096,
      "learning_rate": 0.0001521867948182023,
      "loss": 1.5153,
      "step": 9647
    },
    {
      "epoch": 0.34551542607481156,
      "grad_norm": 1.9078369140625,
      "learning_rate": 0.00015217690020424547,
      "loss": 1.8161,
      "step": 9648
    },
    {
      "epoch": 0.34555123820437983,
      "grad_norm": 1.5560898780822754,
      "learning_rate": 0.00015216700488831573,
      "loss": 1.6721,
      "step": 9649
    },
    {
      "epoch": 0.3455870503339481,
      "grad_norm": 1.4022656679153442,
      "learning_rate": 0.00015215710887054622,
      "loss": 1.5221,
      "step": 9650
    },
    {
      "epoch": 0.34562286246351637,
      "grad_norm": 1.6788058280944824,
      "learning_rate": 0.00015214721215107011,
      "loss": 1.459,
      "step": 9651
    },
    {
      "epoch": 0.3456586745930847,
      "grad_norm": 1.7111127376556396,
      "learning_rate": 0.0001521373147300205,
      "loss": 1.3335,
      "step": 9652
    },
    {
      "epoch": 0.34569448672265296,
      "grad_norm": 1.7129048109054565,
      "learning_rate": 0.0001521274166075306,
      "loss": 1.5987,
      "step": 9653
    },
    {
      "epoch": 0.34573029885222123,
      "grad_norm": 1.3389177322387695,
      "learning_rate": 0.00015211751778373357,
      "loss": 1.424,
      "step": 9654
    },
    {
      "epoch": 0.34576611098178955,
      "grad_norm": 1.848729133605957,
      "learning_rate": 0.0001521076182587625,
      "loss": 1.7741,
      "step": 9655
    },
    {
      "epoch": 0.3458019231113578,
      "grad_norm": 1.724698781967163,
      "learning_rate": 0.0001520977180327507,
      "loss": 1.4231,
      "step": 9656
    },
    {
      "epoch": 0.3458377352409261,
      "grad_norm": 2.336831569671631,
      "learning_rate": 0.00015208781710583126,
      "loss": 1.2322,
      "step": 9657
    },
    {
      "epoch": 0.34587354737049436,
      "grad_norm": 2.277905225753784,
      "learning_rate": 0.00015207791547813744,
      "loss": 1.6277,
      "step": 9658
    },
    {
      "epoch": 0.3459093595000627,
      "grad_norm": 1.8792333602905273,
      "learning_rate": 0.00015206801314980245,
      "loss": 1.7224,
      "step": 9659
    },
    {
      "epoch": 0.34594517162963095,
      "grad_norm": 1.5436029434204102,
      "learning_rate": 0.00015205811012095952,
      "loss": 1.7911,
      "step": 9660
    },
    {
      "epoch": 0.3459809837591992,
      "grad_norm": 1.6252689361572266,
      "learning_rate": 0.00015204820639174184,
      "loss": 1.5968,
      "step": 9661
    },
    {
      "epoch": 0.34601679588876755,
      "grad_norm": 1.3083243370056152,
      "learning_rate": 0.00015203830196228272,
      "loss": 1.6488,
      "step": 9662
    },
    {
      "epoch": 0.3460526080183358,
      "grad_norm": 2.9727373123168945,
      "learning_rate": 0.00015202839683271536,
      "loss": 1.2147,
      "step": 9663
    },
    {
      "epoch": 0.3460884201479041,
      "grad_norm": 1.409303069114685,
      "learning_rate": 0.000152018491003173,
      "loss": 1.3595,
      "step": 9664
    },
    {
      "epoch": 0.34612423227747235,
      "grad_norm": 1.4419803619384766,
      "learning_rate": 0.00015200858447378897,
      "loss": 1.5398,
      "step": 9665
    },
    {
      "epoch": 0.3461600444070407,
      "grad_norm": 1.60395348072052,
      "learning_rate": 0.0001519986772446965,
      "loss": 1.44,
      "step": 9666
    },
    {
      "epoch": 0.34619585653660895,
      "grad_norm": 1.5209178924560547,
      "learning_rate": 0.00015198876931602894,
      "loss": 1.6457,
      "step": 9667
    },
    {
      "epoch": 0.3462316686661772,
      "grad_norm": 1.4209908246994019,
      "learning_rate": 0.00015197886068791952,
      "loss": 1.5196,
      "step": 9668
    },
    {
      "epoch": 0.34626748079574554,
      "grad_norm": 1.624914288520813,
      "learning_rate": 0.00015196895136050157,
      "loss": 1.3967,
      "step": 9669
    },
    {
      "epoch": 0.3463032929253138,
      "grad_norm": 1.4199450016021729,
      "learning_rate": 0.00015195904133390842,
      "loss": 1.2958,
      "step": 9670
    },
    {
      "epoch": 0.3463391050548821,
      "grad_norm": 1.8208658695220947,
      "learning_rate": 0.00015194913060827343,
      "loss": 1.4508,
      "step": 9671
    },
    {
      "epoch": 0.34637491718445035,
      "grad_norm": 1.6885290145874023,
      "learning_rate": 0.00015193921918372984,
      "loss": 1.4547,
      "step": 9672
    },
    {
      "epoch": 0.3464107293140187,
      "grad_norm": 1.8708484172821045,
      "learning_rate": 0.00015192930706041112,
      "loss": 1.5467,
      "step": 9673
    },
    {
      "epoch": 0.34644654144358694,
      "grad_norm": 1.600899338722229,
      "learning_rate": 0.00015191939423845049,
      "loss": 1.4948,
      "step": 9674
    },
    {
      "epoch": 0.3464823535731552,
      "grad_norm": 1.680953025817871,
      "learning_rate": 0.0001519094807179814,
      "loss": 1.5515,
      "step": 9675
    },
    {
      "epoch": 0.34651816570272354,
      "grad_norm": 2.119535207748413,
      "learning_rate": 0.00015189956649913722,
      "loss": 1.5226,
      "step": 9676
    },
    {
      "epoch": 0.3465539778322918,
      "grad_norm": 1.7617014646530151,
      "learning_rate": 0.0001518896515820513,
      "loss": 1.8543,
      "step": 9677
    },
    {
      "epoch": 0.3465897899618601,
      "grad_norm": 1.7265313863754272,
      "learning_rate": 0.00015187973596685706,
      "loss": 1.5269,
      "step": 9678
    },
    {
      "epoch": 0.34662560209142834,
      "grad_norm": 1.4291623830795288,
      "learning_rate": 0.0001518698196536879,
      "loss": 1.4145,
      "step": 9679
    },
    {
      "epoch": 0.34666141422099667,
      "grad_norm": 1.4286699295043945,
      "learning_rate": 0.00015185990264267725,
      "loss": 1.4807,
      "step": 9680
    },
    {
      "epoch": 0.34669722635056494,
      "grad_norm": 1.8250465393066406,
      "learning_rate": 0.00015184998493395846,
      "loss": 1.841,
      "step": 9681
    },
    {
      "epoch": 0.3467330384801332,
      "grad_norm": 1.9351611137390137,
      "learning_rate": 0.00015184006652766503,
      "loss": 1.4977,
      "step": 9682
    },
    {
      "epoch": 0.34676885060970153,
      "grad_norm": 2.0955495834350586,
      "learning_rate": 0.00015183014742393036,
      "loss": 1.6411,
      "step": 9683
    },
    {
      "epoch": 0.3468046627392698,
      "grad_norm": 2.191845178604126,
      "learning_rate": 0.0001518202276228879,
      "loss": 1.3476,
      "step": 9684
    },
    {
      "epoch": 0.34684047486883807,
      "grad_norm": 1.5463204383850098,
      "learning_rate": 0.00015181030712467113,
      "loss": 1.4183,
      "step": 9685
    },
    {
      "epoch": 0.34687628699840634,
      "grad_norm": 1.43691086769104,
      "learning_rate": 0.0001518003859294135,
      "loss": 1.643,
      "step": 9686
    },
    {
      "epoch": 0.34691209912797466,
      "grad_norm": 1.4780082702636719,
      "learning_rate": 0.00015179046403724852,
      "loss": 1.5861,
      "step": 9687
    },
    {
      "epoch": 0.34694791125754293,
      "grad_norm": 1.961837887763977,
      "learning_rate": 0.00015178054144830965,
      "loss": 1.7576,
      "step": 9688
    },
    {
      "epoch": 0.3469837233871112,
      "grad_norm": 1.8840779066085815,
      "learning_rate": 0.0001517706181627304,
      "loss": 1.3534,
      "step": 9689
    },
    {
      "epoch": 0.3470195355166795,
      "grad_norm": 2.285611629486084,
      "learning_rate": 0.0001517606941806442,
      "loss": 1.8627,
      "step": 9690
    },
    {
      "epoch": 0.3470553476462478,
      "grad_norm": 1.699450135231018,
      "learning_rate": 0.00015175076950218468,
      "loss": 1.3957,
      "step": 9691
    },
    {
      "epoch": 0.34709115977581606,
      "grad_norm": 1.8687032461166382,
      "learning_rate": 0.00015174084412748529,
      "loss": 1.4801,
      "step": 9692
    },
    {
      "epoch": 0.34712697190538433,
      "grad_norm": 1.4657593965530396,
      "learning_rate": 0.00015173091805667957,
      "loss": 1.5096,
      "step": 9693
    },
    {
      "epoch": 0.34716278403495265,
      "grad_norm": 1.5671417713165283,
      "learning_rate": 0.0001517209912899011,
      "loss": 1.6317,
      "step": 9694
    },
    {
      "epoch": 0.3471985961645209,
      "grad_norm": 1.8080321550369263,
      "learning_rate": 0.00015171106382728342,
      "loss": 1.2073,
      "step": 9695
    },
    {
      "epoch": 0.3472344082940892,
      "grad_norm": 1.7904155254364014,
      "learning_rate": 0.00015170113566896005,
      "loss": 1.3794,
      "step": 9696
    },
    {
      "epoch": 0.3472702204236575,
      "grad_norm": 1.9424508810043335,
      "learning_rate": 0.0001516912068150646,
      "loss": 1.6862,
      "step": 9697
    },
    {
      "epoch": 0.3473060325532258,
      "grad_norm": 1.6134363412857056,
      "learning_rate": 0.00015168127726573064,
      "loss": 1.5705,
      "step": 9698
    },
    {
      "epoch": 0.34734184468279405,
      "grad_norm": 2.201932430267334,
      "learning_rate": 0.00015167134702109177,
      "loss": 1.5044,
      "step": 9699
    },
    {
      "epoch": 0.3473776568123623,
      "grad_norm": 1.6935111284255981,
      "learning_rate": 0.00015166141608128158,
      "loss": 1.7495,
      "step": 9700
    },
    {
      "epoch": 0.34741346894193065,
      "grad_norm": 1.9058157205581665,
      "learning_rate": 0.0001516514844464336,
      "loss": 1.7792,
      "step": 9701
    },
    {
      "epoch": 0.3474492810714989,
      "grad_norm": 1.739087462425232,
      "learning_rate": 0.00015164155211668163,
      "loss": 1.1891,
      "step": 9702
    },
    {
      "epoch": 0.3474850932010672,
      "grad_norm": 2.0421783924102783,
      "learning_rate": 0.00015163161909215913,
      "loss": 1.4558,
      "step": 9703
    },
    {
      "epoch": 0.3475209053306355,
      "grad_norm": 1.9003093242645264,
      "learning_rate": 0.00015162168537299979,
      "loss": 1.7258,
      "step": 9704
    },
    {
      "epoch": 0.3475567174602038,
      "grad_norm": 1.799619197845459,
      "learning_rate": 0.00015161175095933729,
      "loss": 1.4118,
      "step": 9705
    },
    {
      "epoch": 0.34759252958977205,
      "grad_norm": 2.3882851600646973,
      "learning_rate": 0.00015160181585130523,
      "loss": 1.4103,
      "step": 9706
    },
    {
      "epoch": 0.3476283417193403,
      "grad_norm": 1.882601022720337,
      "learning_rate": 0.00015159188004903733,
      "loss": 1.6516,
      "step": 9707
    },
    {
      "epoch": 0.34766415384890864,
      "grad_norm": 2.743563413619995,
      "learning_rate": 0.0001515819435526672,
      "loss": 1.6921,
      "step": 9708
    },
    {
      "epoch": 0.3476999659784769,
      "grad_norm": 1.8046648502349854,
      "learning_rate": 0.00015157200636232857,
      "loss": 1.6462,
      "step": 9709
    },
    {
      "epoch": 0.3477357781080452,
      "grad_norm": 1.5821819305419922,
      "learning_rate": 0.0001515620684781551,
      "loss": 1.6373,
      "step": 9710
    },
    {
      "epoch": 0.3477715902376135,
      "grad_norm": 1.2368242740631104,
      "learning_rate": 0.00015155212990028053,
      "loss": 1.4887,
      "step": 9711
    },
    {
      "epoch": 0.3478074023671818,
      "grad_norm": 1.3953620195388794,
      "learning_rate": 0.00015154219062883854,
      "loss": 1.348,
      "step": 9712
    },
    {
      "epoch": 0.34784321449675004,
      "grad_norm": 1.7120468616485596,
      "learning_rate": 0.00015153225066396288,
      "loss": 1.4454,
      "step": 9713
    },
    {
      "epoch": 0.3478790266263183,
      "grad_norm": 1.358496904373169,
      "learning_rate": 0.00015152231000578723,
      "loss": 1.3551,
      "step": 9714
    },
    {
      "epoch": 0.34791483875588664,
      "grad_norm": 1.2302814722061157,
      "learning_rate": 0.00015151236865444537,
      "loss": 1.5733,
      "step": 9715
    },
    {
      "epoch": 0.3479506508854549,
      "grad_norm": 2.139761447906494,
      "learning_rate": 0.00015150242661007103,
      "loss": 1.3676,
      "step": 9716
    },
    {
      "epoch": 0.3479864630150232,
      "grad_norm": 2.1644511222839355,
      "learning_rate": 0.000151492483872798,
      "loss": 1.493,
      "step": 9717
    },
    {
      "epoch": 0.3480222751445915,
      "grad_norm": 2.0593531131744385,
      "learning_rate": 0.00015148254044276,
      "loss": 1.788,
      "step": 9718
    },
    {
      "epoch": 0.34805808727415977,
      "grad_norm": 1.453326940536499,
      "learning_rate": 0.00015147259632009082,
      "loss": 1.2916,
      "step": 9719
    },
    {
      "epoch": 0.34809389940372804,
      "grad_norm": 1.3180242776870728,
      "learning_rate": 0.00015146265150492428,
      "loss": 1.4568,
      "step": 9720
    },
    {
      "epoch": 0.3481297115332963,
      "grad_norm": 1.6377161741256714,
      "learning_rate": 0.0001514527059973941,
      "loss": 1.5079,
      "step": 9721
    },
    {
      "epoch": 0.34816552366286463,
      "grad_norm": 1.53775155544281,
      "learning_rate": 0.00015144275979763416,
      "loss": 1.7657,
      "step": 9722
    },
    {
      "epoch": 0.3482013357924329,
      "grad_norm": 1.615124225616455,
      "learning_rate": 0.0001514328129057782,
      "loss": 1.5548,
      "step": 9723
    },
    {
      "epoch": 0.34823714792200117,
      "grad_norm": 2.9852917194366455,
      "learning_rate": 0.00015142286532196018,
      "loss": 1.5382,
      "step": 9724
    },
    {
      "epoch": 0.3482729600515695,
      "grad_norm": 1.7540369033813477,
      "learning_rate": 0.00015141291704631374,
      "loss": 1.539,
      "step": 9725
    },
    {
      "epoch": 0.34830877218113776,
      "grad_norm": 1.644862174987793,
      "learning_rate": 0.00015140296807897289,
      "loss": 1.495,
      "step": 9726
    },
    {
      "epoch": 0.34834458431070603,
      "grad_norm": 1.7135051488876343,
      "learning_rate": 0.00015139301842007137,
      "loss": 1.623,
      "step": 9727
    },
    {
      "epoch": 0.3483803964402743,
      "grad_norm": 1.8585213422775269,
      "learning_rate": 0.0001513830680697431,
      "loss": 1.4913,
      "step": 9728
    },
    {
      "epoch": 0.3484162085698426,
      "grad_norm": 1.7377535104751587,
      "learning_rate": 0.0001513731170281219,
      "loss": 1.3441,
      "step": 9729
    },
    {
      "epoch": 0.3484520206994109,
      "grad_norm": 1.4166041612625122,
      "learning_rate": 0.00015136316529534168,
      "loss": 1.6572,
      "step": 9730
    },
    {
      "epoch": 0.34848783282897916,
      "grad_norm": 2.0903475284576416,
      "learning_rate": 0.00015135321287153636,
      "loss": 1.2165,
      "step": 9731
    },
    {
      "epoch": 0.3485236449585475,
      "grad_norm": 1.809444785118103,
      "learning_rate": 0.00015134325975683975,
      "loss": 1.5006,
      "step": 9732
    },
    {
      "epoch": 0.34855945708811575,
      "grad_norm": 2.5409252643585205,
      "learning_rate": 0.00015133330595138586,
      "loss": 1.9359,
      "step": 9733
    },
    {
      "epoch": 0.348595269217684,
      "grad_norm": 1.4193686246871948,
      "learning_rate": 0.00015132335145530854,
      "loss": 1.4533,
      "step": 9734
    },
    {
      "epoch": 0.3486310813472523,
      "grad_norm": 1.9613677263259888,
      "learning_rate": 0.00015131339626874173,
      "loss": 1.3445,
      "step": 9735
    },
    {
      "epoch": 0.3486668934768206,
      "grad_norm": 1.5214570760726929,
      "learning_rate": 0.00015130344039181935,
      "loss": 1.2932,
      "step": 9736
    },
    {
      "epoch": 0.3487027056063889,
      "grad_norm": 1.8563958406448364,
      "learning_rate": 0.00015129348382467535,
      "loss": 1.876,
      "step": 9737
    },
    {
      "epoch": 0.34873851773595715,
      "grad_norm": 1.4242786169052124,
      "learning_rate": 0.00015128352656744372,
      "loss": 1.6204,
      "step": 9738
    },
    {
      "epoch": 0.3487743298655255,
      "grad_norm": 2.609499931335449,
      "learning_rate": 0.00015127356862025835,
      "loss": 1.422,
      "step": 9739
    },
    {
      "epoch": 0.34881014199509375,
      "grad_norm": 1.4955037832260132,
      "learning_rate": 0.00015126360998325326,
      "loss": 1.4913,
      "step": 9740
    },
    {
      "epoch": 0.348845954124662,
      "grad_norm": 1.7982819080352783,
      "learning_rate": 0.00015125365065656244,
      "loss": 1.2715,
      "step": 9741
    },
    {
      "epoch": 0.3488817662542303,
      "grad_norm": 2.949068307876587,
      "learning_rate": 0.00015124369064031988,
      "loss": 1.8439,
      "step": 9742
    },
    {
      "epoch": 0.3489175783837986,
      "grad_norm": 1.6375192403793335,
      "learning_rate": 0.00015123372993465953,
      "loss": 1.7551,
      "step": 9743
    },
    {
      "epoch": 0.3489533905133669,
      "grad_norm": 1.3728387355804443,
      "learning_rate": 0.00015122376853971545,
      "loss": 1.5968,
      "step": 9744
    },
    {
      "epoch": 0.34898920264293515,
      "grad_norm": 2.1111888885498047,
      "learning_rate": 0.00015121380645562163,
      "loss": 1.5258,
      "step": 9745
    },
    {
      "epoch": 0.3490250147725035,
      "grad_norm": 1.3476275205612183,
      "learning_rate": 0.0001512038436825121,
      "loss": 1.368,
      "step": 9746
    },
    {
      "epoch": 0.34906082690207174,
      "grad_norm": 1.5277924537658691,
      "learning_rate": 0.0001511938802205209,
      "loss": 1.5331,
      "step": 9747
    },
    {
      "epoch": 0.34909663903164,
      "grad_norm": 1.4983168840408325,
      "learning_rate": 0.0001511839160697821,
      "loss": 1.3779,
      "step": 9748
    },
    {
      "epoch": 0.3491324511612083,
      "grad_norm": 1.4858800172805786,
      "learning_rate": 0.00015117395123042968,
      "loss": 1.2519,
      "step": 9749
    },
    {
      "epoch": 0.3491682632907766,
      "grad_norm": 1.6948373317718506,
      "learning_rate": 0.0001511639857025978,
      "loss": 1.7947,
      "step": 9750
    },
    {
      "epoch": 0.3492040754203449,
      "grad_norm": 1.306716799736023,
      "learning_rate": 0.0001511540194864205,
      "loss": 1.4975,
      "step": 9751
    },
    {
      "epoch": 0.34923988754991314,
      "grad_norm": 1.2586356401443481,
      "learning_rate": 0.0001511440525820318,
      "loss": 1.5344,
      "step": 9752
    },
    {
      "epoch": 0.34927569967948147,
      "grad_norm": 1.4947229623794556,
      "learning_rate": 0.0001511340849895659,
      "loss": 1.4292,
      "step": 9753
    },
    {
      "epoch": 0.34931151180904974,
      "grad_norm": 1.699182391166687,
      "learning_rate": 0.0001511241167091568,
      "loss": 1.6858,
      "step": 9754
    },
    {
      "epoch": 0.349347323938618,
      "grad_norm": 1.7800447940826416,
      "learning_rate": 0.0001511141477409387,
      "loss": 1.441,
      "step": 9755
    },
    {
      "epoch": 0.3493831360681863,
      "grad_norm": 1.2749764919281006,
      "learning_rate": 0.00015110417808504562,
      "loss": 1.682,
      "step": 9756
    },
    {
      "epoch": 0.3494189481977546,
      "grad_norm": 1.828865885734558,
      "learning_rate": 0.00015109420774161178,
      "loss": 1.5552,
      "step": 9757
    },
    {
      "epoch": 0.34945476032732287,
      "grad_norm": 1.4188942909240723,
      "learning_rate": 0.00015108423671077124,
      "loss": 1.6542,
      "step": 9758
    },
    {
      "epoch": 0.34949057245689114,
      "grad_norm": 1.8428118228912354,
      "learning_rate": 0.00015107426499265825,
      "loss": 1.5784,
      "step": 9759
    },
    {
      "epoch": 0.34952638458645946,
      "grad_norm": 1.4908447265625,
      "learning_rate": 0.00015106429258740687,
      "loss": 1.6085,
      "step": 9760
    },
    {
      "epoch": 0.34956219671602773,
      "grad_norm": 1.5967750549316406,
      "learning_rate": 0.00015105431949515133,
      "loss": 1.1694,
      "step": 9761
    },
    {
      "epoch": 0.349598008845596,
      "grad_norm": 1.6794275045394897,
      "learning_rate": 0.00015104434571602577,
      "loss": 1.4624,
      "step": 9762
    },
    {
      "epoch": 0.34963382097516427,
      "grad_norm": 1.981397271156311,
      "learning_rate": 0.00015103437125016432,
      "loss": 1.3633,
      "step": 9763
    },
    {
      "epoch": 0.3496696331047326,
      "grad_norm": 2.4205143451690674,
      "learning_rate": 0.0001510243960977013,
      "loss": 1.7959,
      "step": 9764
    },
    {
      "epoch": 0.34970544523430086,
      "grad_norm": 2.2734334468841553,
      "learning_rate": 0.00015101442025877083,
      "loss": 1.3619,
      "step": 9765
    },
    {
      "epoch": 0.34974125736386913,
      "grad_norm": 1.8504884243011475,
      "learning_rate": 0.00015100444373350715,
      "loss": 1.6551,
      "step": 9766
    },
    {
      "epoch": 0.34977706949343745,
      "grad_norm": 1.5831323862075806,
      "learning_rate": 0.00015099446652204446,
      "loss": 1.5193,
      "step": 9767
    },
    {
      "epoch": 0.3498128816230057,
      "grad_norm": 1.8388887643814087,
      "learning_rate": 0.00015098448862451702,
      "loss": 1.4238,
      "step": 9768
    },
    {
      "epoch": 0.349848693752574,
      "grad_norm": 1.5945453643798828,
      "learning_rate": 0.00015097451004105904,
      "loss": 1.4274,
      "step": 9769
    },
    {
      "epoch": 0.34988450588214226,
      "grad_norm": 1.5054142475128174,
      "learning_rate": 0.00015096453077180475,
      "loss": 1.506,
      "step": 9770
    },
    {
      "epoch": 0.3499203180117106,
      "grad_norm": 1.8652490377426147,
      "learning_rate": 0.00015095455081688846,
      "loss": 1.8354,
      "step": 9771
    },
    {
      "epoch": 0.34995613014127885,
      "grad_norm": 1.8523324728012085,
      "learning_rate": 0.0001509445701764444,
      "loss": 1.6814,
      "step": 9772
    },
    {
      "epoch": 0.3499919422708471,
      "grad_norm": 1.3865050077438354,
      "learning_rate": 0.00015093458885060687,
      "loss": 1.2962,
      "step": 9773
    },
    {
      "epoch": 0.35002775440041545,
      "grad_norm": 1.6977640390396118,
      "learning_rate": 0.00015092460683951015,
      "loss": 1.4121,
      "step": 9774
    },
    {
      "epoch": 0.3500635665299837,
      "grad_norm": 2.1910016536712646,
      "learning_rate": 0.00015091462414328855,
      "loss": 1.9099,
      "step": 9775
    },
    {
      "epoch": 0.350099378659552,
      "grad_norm": 2.3362877368927,
      "learning_rate": 0.00015090464076207634,
      "loss": 1.7061,
      "step": 9776
    },
    {
      "epoch": 0.35013519078912025,
      "grad_norm": 1.5573481321334839,
      "learning_rate": 0.0001508946566960079,
      "loss": 1.6889,
      "step": 9777
    },
    {
      "epoch": 0.3501710029186886,
      "grad_norm": 1.7748955488204956,
      "learning_rate": 0.0001508846719452174,
      "loss": 1.3411,
      "step": 9778
    },
    {
      "epoch": 0.35020681504825685,
      "grad_norm": 1.7088675498962402,
      "learning_rate": 0.00015087468650983935,
      "loss": 1.5765,
      "step": 9779
    },
    {
      "epoch": 0.3502426271778251,
      "grad_norm": 2.062631368637085,
      "learning_rate": 0.00015086470039000802,
      "loss": 1.5179,
      "step": 9780
    },
    {
      "epoch": 0.35027843930739344,
      "grad_norm": 1.7780126333236694,
      "learning_rate": 0.00015085471358585774,
      "loss": 1.5016,
      "step": 9781
    },
    {
      "epoch": 0.3503142514369617,
      "grad_norm": 1.4579542875289917,
      "learning_rate": 0.00015084472609752284,
      "loss": 1.5187,
      "step": 9782
    },
    {
      "epoch": 0.35035006356653,
      "grad_norm": 1.4404877424240112,
      "learning_rate": 0.0001508347379251378,
      "loss": 1.5978,
      "step": 9783
    },
    {
      "epoch": 0.35038587569609825,
      "grad_norm": 1.950384259223938,
      "learning_rate": 0.0001508247490688369,
      "loss": 1.4719,
      "step": 9784
    },
    {
      "epoch": 0.3504216878256666,
      "grad_norm": 1.659462571144104,
      "learning_rate": 0.0001508147595287546,
      "loss": 1.6149,
      "step": 9785
    },
    {
      "epoch": 0.35045749995523484,
      "grad_norm": 1.4204747676849365,
      "learning_rate": 0.00015080476930502522,
      "loss": 1.6546,
      "step": 9786
    },
    {
      "epoch": 0.3504933120848031,
      "grad_norm": 1.5785166025161743,
      "learning_rate": 0.0001507947783977832,
      "loss": 1.5731,
      "step": 9787
    },
    {
      "epoch": 0.35052912421437143,
      "grad_norm": 1.9538383483886719,
      "learning_rate": 0.00015078478680716299,
      "loss": 1.473,
      "step": 9788
    },
    {
      "epoch": 0.3505649363439397,
      "grad_norm": 1.3947443962097168,
      "learning_rate": 0.00015077479453329894,
      "loss": 1.327,
      "step": 9789
    },
    {
      "epoch": 0.350600748473508,
      "grad_norm": 2.454119920730591,
      "learning_rate": 0.0001507648015763256,
      "loss": 1.497,
      "step": 9790
    },
    {
      "epoch": 0.35063656060307624,
      "grad_norm": 2.188707113265991,
      "learning_rate": 0.00015075480793637724,
      "loss": 1.8418,
      "step": 9791
    },
    {
      "epoch": 0.35067237273264457,
      "grad_norm": 1.5564440488815308,
      "learning_rate": 0.00015074481361358844,
      "loss": 1.6697,
      "step": 9792
    },
    {
      "epoch": 0.35070818486221284,
      "grad_norm": 2.128994941711426,
      "learning_rate": 0.00015073481860809363,
      "loss": 1.3797,
      "step": 9793
    },
    {
      "epoch": 0.3507439969917811,
      "grad_norm": 1.9400798082351685,
      "learning_rate": 0.00015072482292002732,
      "loss": 1.5212,
      "step": 9794
    },
    {
      "epoch": 0.35077980912134943,
      "grad_norm": 1.5561938285827637,
      "learning_rate": 0.0001507148265495239,
      "loss": 1.4001,
      "step": 9795
    },
    {
      "epoch": 0.3508156212509177,
      "grad_norm": 1.46834135055542,
      "learning_rate": 0.00015070482949671794,
      "loss": 1.3285,
      "step": 9796
    },
    {
      "epoch": 0.35085143338048597,
      "grad_norm": 1.8120313882827759,
      "learning_rate": 0.0001506948317617439,
      "loss": 1.809,
      "step": 9797
    },
    {
      "epoch": 0.35088724551005424,
      "grad_norm": 1.6378916501998901,
      "learning_rate": 0.00015068483334473623,
      "loss": 1.6557,
      "step": 9798
    },
    {
      "epoch": 0.35092305763962256,
      "grad_norm": 1.4764701128005981,
      "learning_rate": 0.00015067483424582956,
      "loss": 1.5015,
      "step": 9799
    },
    {
      "epoch": 0.35095886976919083,
      "grad_norm": 1.7734726667404175,
      "learning_rate": 0.00015066483446515836,
      "loss": 1.2686,
      "step": 9800
    },
    {
      "epoch": 0.3509946818987591,
      "grad_norm": 1.5742462873458862,
      "learning_rate": 0.00015065483400285716,
      "loss": 1.6867,
      "step": 9801
    },
    {
      "epoch": 0.3510304940283274,
      "grad_norm": 1.916930079460144,
      "learning_rate": 0.00015064483285906052,
      "loss": 1.4477,
      "step": 9802
    },
    {
      "epoch": 0.3510663061578957,
      "grad_norm": 1.2710362672805786,
      "learning_rate": 0.00015063483103390296,
      "loss": 1.4712,
      "step": 9803
    },
    {
      "epoch": 0.35110211828746396,
      "grad_norm": 1.4401026964187622,
      "learning_rate": 0.00015062482852751908,
      "loss": 1.4466,
      "step": 9804
    },
    {
      "epoch": 0.35113793041703223,
      "grad_norm": 1.357136845588684,
      "learning_rate": 0.0001506148253400434,
      "loss": 1.4442,
      "step": 9805
    },
    {
      "epoch": 0.35117374254660055,
      "grad_norm": 1.5309224128723145,
      "learning_rate": 0.00015060482147161058,
      "loss": 1.4533,
      "step": 9806
    },
    {
      "epoch": 0.3512095546761688,
      "grad_norm": 1.6512397527694702,
      "learning_rate": 0.00015059481692235514,
      "loss": 1.596,
      "step": 9807
    },
    {
      "epoch": 0.3512453668057371,
      "grad_norm": 1.5348283052444458,
      "learning_rate": 0.0001505848116924117,
      "loss": 1.4335,
      "step": 9808
    },
    {
      "epoch": 0.35128117893530536,
      "grad_norm": 1.6342748403549194,
      "learning_rate": 0.00015057480578191485,
      "loss": 1.1626,
      "step": 9809
    },
    {
      "epoch": 0.3513169910648737,
      "grad_norm": 1.6329182386398315,
      "learning_rate": 0.00015056479919099927,
      "loss": 1.5846,
      "step": 9810
    },
    {
      "epoch": 0.35135280319444195,
      "grad_norm": 1.609890341758728,
      "learning_rate": 0.0001505547919197995,
      "loss": 1.5383,
      "step": 9811
    },
    {
      "epoch": 0.3513886153240102,
      "grad_norm": 1.6261221170425415,
      "learning_rate": 0.00015054478396845026,
      "loss": 1.6088,
      "step": 9812
    },
    {
      "epoch": 0.35142442745357855,
      "grad_norm": 1.727931261062622,
      "learning_rate": 0.00015053477533708608,
      "loss": 1.6369,
      "step": 9813
    },
    {
      "epoch": 0.3514602395831468,
      "grad_norm": 1.760061264038086,
      "learning_rate": 0.00015052476602584177,
      "loss": 1.5403,
      "step": 9814
    },
    {
      "epoch": 0.3514960517127151,
      "grad_norm": 1.8632763624191284,
      "learning_rate": 0.00015051475603485183,
      "loss": 1.1816,
      "step": 9815
    },
    {
      "epoch": 0.35153186384228335,
      "grad_norm": 1.4889694452285767,
      "learning_rate": 0.00015050474536425101,
      "loss": 1.5174,
      "step": 9816
    },
    {
      "epoch": 0.3515676759718517,
      "grad_norm": 1.9194355010986328,
      "learning_rate": 0.00015049473401417403,
      "loss": 1.6779,
      "step": 9817
    },
    {
      "epoch": 0.35160348810141995,
      "grad_norm": 1.6100517511367798,
      "learning_rate": 0.00015048472198475553,
      "loss": 1.5647,
      "step": 9818
    },
    {
      "epoch": 0.3516393002309882,
      "grad_norm": 1.5931763648986816,
      "learning_rate": 0.00015047470927613018,
      "loss": 1.6965,
      "step": 9819
    },
    {
      "epoch": 0.35167511236055654,
      "grad_norm": 1.833192229270935,
      "learning_rate": 0.00015046469588843276,
      "loss": 1.5959,
      "step": 9820
    },
    {
      "epoch": 0.3517109244901248,
      "grad_norm": 2.009188175201416,
      "learning_rate": 0.00015045468182179795,
      "loss": 1.8723,
      "step": 9821
    },
    {
      "epoch": 0.3517467366196931,
      "grad_norm": 1.9709876775741577,
      "learning_rate": 0.0001504446670763605,
      "loss": 1.677,
      "step": 9822
    },
    {
      "epoch": 0.35178254874926135,
      "grad_norm": 1.6057246923446655,
      "learning_rate": 0.0001504346516522551,
      "loss": 1.7295,
      "step": 9823
    },
    {
      "epoch": 0.3518183608788297,
      "grad_norm": 1.5857231616973877,
      "learning_rate": 0.00015042463554961648,
      "loss": 1.4851,
      "step": 9824
    },
    {
      "epoch": 0.35185417300839794,
      "grad_norm": 2.4867377281188965,
      "learning_rate": 0.0001504146187685795,
      "loss": 1.5757,
      "step": 9825
    },
    {
      "epoch": 0.3518899851379662,
      "grad_norm": 1.5649453401565552,
      "learning_rate": 0.0001504046013092788,
      "loss": 1.2162,
      "step": 9826
    },
    {
      "epoch": 0.35192579726753453,
      "grad_norm": 1.9253320693969727,
      "learning_rate": 0.00015039458317184923,
      "loss": 1.6233,
      "step": 9827
    },
    {
      "epoch": 0.3519616093971028,
      "grad_norm": 1.2841295003890991,
      "learning_rate": 0.00015038456435642554,
      "loss": 1.5658,
      "step": 9828
    },
    {
      "epoch": 0.3519974215266711,
      "grad_norm": 1.5589649677276611,
      "learning_rate": 0.0001503745448631425,
      "loss": 1.6228,
      "step": 9829
    },
    {
      "epoch": 0.35203323365623934,
      "grad_norm": 2.096522808074951,
      "learning_rate": 0.00015036452469213504,
      "loss": 1.5682,
      "step": 9830
    },
    {
      "epoch": 0.35206904578580767,
      "grad_norm": 1.6603925228118896,
      "learning_rate": 0.00015035450384353775,
      "loss": 1.6372,
      "step": 9831
    },
    {
      "epoch": 0.35210485791537594,
      "grad_norm": 1.6871788501739502,
      "learning_rate": 0.0001503444823174856,
      "loss": 1.7357,
      "step": 9832
    },
    {
      "epoch": 0.3521406700449442,
      "grad_norm": 1.7978812456130981,
      "learning_rate": 0.0001503344601141134,
      "loss": 1.3659,
      "step": 9833
    },
    {
      "epoch": 0.35217648217451253,
      "grad_norm": 1.4551135301589966,
      "learning_rate": 0.00015032443723355597,
      "loss": 1.5676,
      "step": 9834
    },
    {
      "epoch": 0.3522122943040808,
      "grad_norm": 2.247584819793701,
      "learning_rate": 0.0001503144136759481,
      "loss": 1.3566,
      "step": 9835
    },
    {
      "epoch": 0.35224810643364907,
      "grad_norm": 1.6647343635559082,
      "learning_rate": 0.00015030438944142475,
      "loss": 1.5347,
      "step": 9836
    },
    {
      "epoch": 0.35228391856321734,
      "grad_norm": 1.3498740196228027,
      "learning_rate": 0.0001502943645301207,
      "loss": 1.241,
      "step": 9837
    },
    {
      "epoch": 0.35231973069278566,
      "grad_norm": 1.8701205253601074,
      "learning_rate": 0.00015028433894217087,
      "loss": 1.4011,
      "step": 9838
    },
    {
      "epoch": 0.35235554282235393,
      "grad_norm": 1.4306402206420898,
      "learning_rate": 0.0001502743126777101,
      "loss": 1.493,
      "step": 9839
    },
    {
      "epoch": 0.3523913549519222,
      "grad_norm": 1.9978787899017334,
      "learning_rate": 0.0001502642857368733,
      "loss": 1.591,
      "step": 9840
    },
    {
      "epoch": 0.3524271670814905,
      "grad_norm": 1.7107425928115845,
      "learning_rate": 0.00015025425811979542,
      "loss": 1.2661,
      "step": 9841
    },
    {
      "epoch": 0.3524629792110588,
      "grad_norm": 1.2457308769226074,
      "learning_rate": 0.00015024422982661125,
      "loss": 1.5245,
      "step": 9842
    },
    {
      "epoch": 0.35249879134062706,
      "grad_norm": 1.4251497983932495,
      "learning_rate": 0.00015023420085745584,
      "loss": 1.5731,
      "step": 9843
    },
    {
      "epoch": 0.35253460347019533,
      "grad_norm": 2.0676839351654053,
      "learning_rate": 0.00015022417121246398,
      "loss": 1.4724,
      "step": 9844
    },
    {
      "epoch": 0.35257041559976365,
      "grad_norm": 1.7826956510543823,
      "learning_rate": 0.00015021414089177077,
      "loss": 1.7058,
      "step": 9845
    },
    {
      "epoch": 0.3526062277293319,
      "grad_norm": 1.7209420204162598,
      "learning_rate": 0.000150204109895511,
      "loss": 1.5886,
      "step": 9846
    },
    {
      "epoch": 0.3526420398589002,
      "grad_norm": 1.6267695426940918,
      "learning_rate": 0.00015019407822381973,
      "loss": 1.4277,
      "step": 9847
    },
    {
      "epoch": 0.3526778519884685,
      "grad_norm": 1.7787023782730103,
      "learning_rate": 0.00015018404587683186,
      "loss": 1.7895,
      "step": 9848
    },
    {
      "epoch": 0.3527136641180368,
      "grad_norm": 1.4573239088058472,
      "learning_rate": 0.0001501740128546824,
      "loss": 1.278,
      "step": 9849
    },
    {
      "epoch": 0.35274947624760505,
      "grad_norm": 1.3814666271209717,
      "learning_rate": 0.00015016397915750633,
      "loss": 1.2868,
      "step": 9850
    },
    {
      "epoch": 0.3527852883771733,
      "grad_norm": 2.118358850479126,
      "learning_rate": 0.0001501539447854386,
      "loss": 1.3829,
      "step": 9851
    },
    {
      "epoch": 0.35282110050674165,
      "grad_norm": 1.9069745540618896,
      "learning_rate": 0.00015014390973861424,
      "loss": 1.7754,
      "step": 9852
    },
    {
      "epoch": 0.3528569126363099,
      "grad_norm": 2.0276107788085938,
      "learning_rate": 0.00015013387401716823,
      "loss": 1.4873,
      "step": 9853
    },
    {
      "epoch": 0.3528927247658782,
      "grad_norm": 2.06880784034729,
      "learning_rate": 0.00015012383762123566,
      "loss": 1.729,
      "step": 9854
    },
    {
      "epoch": 0.3529285368954465,
      "grad_norm": 2.1052966117858887,
      "learning_rate": 0.0001501138005509515,
      "loss": 1.46,
      "step": 9855
    },
    {
      "epoch": 0.3529643490250148,
      "grad_norm": 1.9557901620864868,
      "learning_rate": 0.00015010376280645077,
      "loss": 1.4189,
      "step": 9856
    },
    {
      "epoch": 0.35300016115458305,
      "grad_norm": 1.9544841051101685,
      "learning_rate": 0.00015009372438786858,
      "loss": 1.7716,
      "step": 9857
    },
    {
      "epoch": 0.3530359732841513,
      "grad_norm": 1.9688549041748047,
      "learning_rate": 0.00015008368529533992,
      "loss": 1.7624,
      "step": 9858
    },
    {
      "epoch": 0.35307178541371964,
      "grad_norm": 1.3620836734771729,
      "learning_rate": 0.00015007364552899988,
      "loss": 1.6062,
      "step": 9859
    },
    {
      "epoch": 0.3531075975432879,
      "grad_norm": 1.8967136144638062,
      "learning_rate": 0.0001500636050889835,
      "loss": 1.1698,
      "step": 9860
    },
    {
      "epoch": 0.3531434096728562,
      "grad_norm": 1.6941111087799072,
      "learning_rate": 0.00015005356397542597,
      "loss": 1.7253,
      "step": 9861
    },
    {
      "epoch": 0.3531792218024245,
      "grad_norm": 1.463535189628601,
      "learning_rate": 0.00015004352218846222,
      "loss": 1.5704,
      "step": 9862
    },
    {
      "epoch": 0.3532150339319928,
      "grad_norm": 2.1995041370391846,
      "learning_rate": 0.00015003347972822746,
      "loss": 1.7306,
      "step": 9863
    },
    {
      "epoch": 0.35325084606156104,
      "grad_norm": 1.662667155265808,
      "learning_rate": 0.00015002343659485678,
      "loss": 1.6146,
      "step": 9864
    },
    {
      "epoch": 0.3532866581911293,
      "grad_norm": 1.8545045852661133,
      "learning_rate": 0.0001500133927884853,
      "loss": 1.7385,
      "step": 9865
    },
    {
      "epoch": 0.35332247032069763,
      "grad_norm": 1.3238962888717651,
      "learning_rate": 0.0001500033483092481,
      "loss": 1.477,
      "step": 9866
    },
    {
      "epoch": 0.3533582824502659,
      "grad_norm": 2.3159918785095215,
      "learning_rate": 0.00014999330315728035,
      "loss": 1.4599,
      "step": 9867
    },
    {
      "epoch": 0.3533940945798342,
      "grad_norm": 1.7593953609466553,
      "learning_rate": 0.00014998325733271722,
      "loss": 1.3809,
      "step": 9868
    },
    {
      "epoch": 0.3534299067094025,
      "grad_norm": 1.9399266242980957,
      "learning_rate": 0.0001499732108356938,
      "loss": 1.5886,
      "step": 9869
    },
    {
      "epoch": 0.35346571883897077,
      "grad_norm": 1.7394417524337769,
      "learning_rate": 0.00014996316366634532,
      "loss": 1.4151,
      "step": 9870
    },
    {
      "epoch": 0.35350153096853904,
      "grad_norm": 2.505788564682007,
      "learning_rate": 0.0001499531158248069,
      "loss": 1.7199,
      "step": 9871
    },
    {
      "epoch": 0.3535373430981073,
      "grad_norm": 1.7400318384170532,
      "learning_rate": 0.00014994306731121374,
      "loss": 1.7993,
      "step": 9872
    },
    {
      "epoch": 0.35357315522767563,
      "grad_norm": 1.9016368389129639,
      "learning_rate": 0.00014993301812570104,
      "loss": 1.54,
      "step": 9873
    },
    {
      "epoch": 0.3536089673572439,
      "grad_norm": 1.5940747261047363,
      "learning_rate": 0.00014992296826840402,
      "loss": 1.4284,
      "step": 9874
    },
    {
      "epoch": 0.35364477948681217,
      "grad_norm": 2.1987946033477783,
      "learning_rate": 0.00014991291773945782,
      "loss": 1.4855,
      "step": 9875
    },
    {
      "epoch": 0.3536805916163805,
      "grad_norm": 1.4558799266815186,
      "learning_rate": 0.0001499028665389977,
      "loss": 1.8473,
      "step": 9876
    },
    {
      "epoch": 0.35371640374594876,
      "grad_norm": 1.5671563148498535,
      "learning_rate": 0.00014989281466715887,
      "loss": 1.5182,
      "step": 9877
    },
    {
      "epoch": 0.35375221587551703,
      "grad_norm": 1.395065188407898,
      "learning_rate": 0.0001498827621240766,
      "loss": 1.3509,
      "step": 9878
    },
    {
      "epoch": 0.3537880280050853,
      "grad_norm": 1.8056888580322266,
      "learning_rate": 0.0001498727089098861,
      "loss": 1.779,
      "step": 9879
    },
    {
      "epoch": 0.3538238401346536,
      "grad_norm": 1.9839811325073242,
      "learning_rate": 0.00014986265502472262,
      "loss": 1.813,
      "step": 9880
    },
    {
      "epoch": 0.3538596522642219,
      "grad_norm": 1.4838645458221436,
      "learning_rate": 0.00014985260046872145,
      "loss": 1.1724,
      "step": 9881
    },
    {
      "epoch": 0.35389546439379016,
      "grad_norm": 1.5873481035232544,
      "learning_rate": 0.00014984254524201784,
      "loss": 1.3684,
      "step": 9882
    },
    {
      "epoch": 0.3539312765233585,
      "grad_norm": 1.2479912042617798,
      "learning_rate": 0.0001498324893447471,
      "loss": 1.1336,
      "step": 9883
    },
    {
      "epoch": 0.35396708865292675,
      "grad_norm": 1.6061489582061768,
      "learning_rate": 0.00014982243277704446,
      "loss": 1.5485,
      "step": 9884
    },
    {
      "epoch": 0.354002900782495,
      "grad_norm": 2.5165042877197266,
      "learning_rate": 0.0001498123755390453,
      "loss": 1.4389,
      "step": 9885
    },
    {
      "epoch": 0.3540387129120633,
      "grad_norm": 2.0060083866119385,
      "learning_rate": 0.00014980231763088482,
      "loss": 1.5283,
      "step": 9886
    },
    {
      "epoch": 0.3540745250416316,
      "grad_norm": 1.2914835214614868,
      "learning_rate": 0.00014979225905269842,
      "loss": 1.4375,
      "step": 9887
    },
    {
      "epoch": 0.3541103371711999,
      "grad_norm": 1.3552098274230957,
      "learning_rate": 0.0001497821998046214,
      "loss": 1.4385,
      "step": 9888
    },
    {
      "epoch": 0.35414614930076815,
      "grad_norm": 1.729246735572815,
      "learning_rate": 0.0001497721398867891,
      "loss": 1.4477,
      "step": 9889
    },
    {
      "epoch": 0.3541819614303365,
      "grad_norm": 1.430737018585205,
      "learning_rate": 0.00014976207929933688,
      "loss": 1.6311,
      "step": 9890
    },
    {
      "epoch": 0.35421777355990475,
      "grad_norm": 1.8276435136795044,
      "learning_rate": 0.00014975201804240005,
      "loss": 1.6353,
      "step": 9891
    },
    {
      "epoch": 0.354253585689473,
      "grad_norm": 1.814315915107727,
      "learning_rate": 0.00014974195611611402,
      "loss": 1.9347,
      "step": 9892
    },
    {
      "epoch": 0.3542893978190413,
      "grad_norm": 1.468151330947876,
      "learning_rate": 0.00014973189352061409,
      "loss": 1.3659,
      "step": 9893
    },
    {
      "epoch": 0.3543252099486096,
      "grad_norm": 2.0722012519836426,
      "learning_rate": 0.0001497218302560357,
      "loss": 1.5477,
      "step": 9894
    },
    {
      "epoch": 0.3543610220781779,
      "grad_norm": 1.7138365507125854,
      "learning_rate": 0.0001497117663225142,
      "loss": 1.6164,
      "step": 9895
    },
    {
      "epoch": 0.35439683420774615,
      "grad_norm": 1.5745422840118408,
      "learning_rate": 0.00014970170172018505,
      "loss": 1.566,
      "step": 9896
    },
    {
      "epoch": 0.35443264633731447,
      "grad_norm": 2.2554714679718018,
      "learning_rate": 0.00014969163644918358,
      "loss": 1.4819,
      "step": 9897
    },
    {
      "epoch": 0.35446845846688274,
      "grad_norm": 1.376289963722229,
      "learning_rate": 0.00014968157050964526,
      "loss": 1.4859,
      "step": 9898
    },
    {
      "epoch": 0.354504270596451,
      "grad_norm": 1.6688337326049805,
      "learning_rate": 0.00014967150390170547,
      "loss": 1.5748,
      "step": 9899
    },
    {
      "epoch": 0.3545400827260193,
      "grad_norm": 2.0181753635406494,
      "learning_rate": 0.0001496614366254997,
      "loss": 1.37,
      "step": 9900
    },
    {
      "epoch": 0.3545758948555876,
      "grad_norm": 1.6945894956588745,
      "learning_rate": 0.00014965136868116334,
      "loss": 1.5793,
      "step": 9901
    },
    {
      "epoch": 0.3546117069851559,
      "grad_norm": 1.9368959665298462,
      "learning_rate": 0.00014964130006883187,
      "loss": 1.6875,
      "step": 9902
    },
    {
      "epoch": 0.35464751911472414,
      "grad_norm": 1.573068380355835,
      "learning_rate": 0.00014963123078864073,
      "loss": 1.5334,
      "step": 9903
    },
    {
      "epoch": 0.35468333124429247,
      "grad_norm": 1.3953185081481934,
      "learning_rate": 0.0001496211608407254,
      "loss": 1.6489,
      "step": 9904
    },
    {
      "epoch": 0.35471914337386073,
      "grad_norm": 2.508678674697876,
      "learning_rate": 0.00014961109022522135,
      "loss": 1.9514,
      "step": 9905
    },
    {
      "epoch": 0.354754955503429,
      "grad_norm": 2.420624256134033,
      "learning_rate": 0.0001496010189422641,
      "loss": 1.4283,
      "step": 9906
    },
    {
      "epoch": 0.3547907676329973,
      "grad_norm": 1.3854237794876099,
      "learning_rate": 0.00014959094699198907,
      "loss": 1.2818,
      "step": 9907
    },
    {
      "epoch": 0.3548265797625656,
      "grad_norm": 1.5231170654296875,
      "learning_rate": 0.00014958087437453186,
      "loss": 1.4811,
      "step": 9908
    },
    {
      "epoch": 0.35486239189213387,
      "grad_norm": 1.3888393640518188,
      "learning_rate": 0.00014957080109002794,
      "loss": 1.406,
      "step": 9909
    },
    {
      "epoch": 0.35489820402170213,
      "grad_norm": 2.292565107345581,
      "learning_rate": 0.00014956072713861286,
      "loss": 1.6349,
      "step": 9910
    },
    {
      "epoch": 0.35493401615127046,
      "grad_norm": 2.0095014572143555,
      "learning_rate": 0.00014955065252042206,
      "loss": 1.6491,
      "step": 9911
    },
    {
      "epoch": 0.35496982828083873,
      "grad_norm": 2.346771240234375,
      "learning_rate": 0.00014954057723559115,
      "loss": 1.8228,
      "step": 9912
    },
    {
      "epoch": 0.355005640410407,
      "grad_norm": 1.9204994440078735,
      "learning_rate": 0.0001495305012842557,
      "loss": 1.6642,
      "step": 9913
    },
    {
      "epoch": 0.35504145253997527,
      "grad_norm": 1.8857648372650146,
      "learning_rate": 0.00014952042466655126,
      "loss": 1.5507,
      "step": 9914
    },
    {
      "epoch": 0.3550772646695436,
      "grad_norm": 1.6578903198242188,
      "learning_rate": 0.00014951034738261337,
      "loss": 1.7453,
      "step": 9915
    },
    {
      "epoch": 0.35511307679911186,
      "grad_norm": 1.5523202419281006,
      "learning_rate": 0.00014950026943257762,
      "loss": 1.6968,
      "step": 9916
    },
    {
      "epoch": 0.35514888892868013,
      "grad_norm": 2.2549266815185547,
      "learning_rate": 0.00014949019081657959,
      "loss": 1.512,
      "step": 9917
    },
    {
      "epoch": 0.35518470105824845,
      "grad_norm": 2.5637290477752686,
      "learning_rate": 0.00014948011153475491,
      "loss": 1.7135,
      "step": 9918
    },
    {
      "epoch": 0.3552205131878167,
      "grad_norm": 1.7439630031585693,
      "learning_rate": 0.0001494700315872391,
      "loss": 1.5241,
      "step": 9919
    },
    {
      "epoch": 0.355256325317385,
      "grad_norm": 2.2923781871795654,
      "learning_rate": 0.00014945995097416788,
      "loss": 1.5614,
      "step": 9920
    },
    {
      "epoch": 0.35529213744695326,
      "grad_norm": 1.4263309240341187,
      "learning_rate": 0.0001494498696956768,
      "loss": 1.618,
      "step": 9921
    },
    {
      "epoch": 0.3553279495765216,
      "grad_norm": 1.650952935218811,
      "learning_rate": 0.0001494397877519015,
      "loss": 1.531,
      "step": 9922
    },
    {
      "epoch": 0.35536376170608985,
      "grad_norm": 1.950554609298706,
      "learning_rate": 0.00014942970514297761,
      "loss": 1.3641,
      "step": 9923
    },
    {
      "epoch": 0.3553995738356581,
      "grad_norm": 1.464562177658081,
      "learning_rate": 0.00014941962186904083,
      "loss": 1.508,
      "step": 9924
    },
    {
      "epoch": 0.35543538596522645,
      "grad_norm": 1.9498792886734009,
      "learning_rate": 0.00014940953793022676,
      "loss": 1.7068,
      "step": 9925
    },
    {
      "epoch": 0.3554711980947947,
      "grad_norm": 2.579939126968384,
      "learning_rate": 0.00014939945332667108,
      "loss": 1.5399,
      "step": 9926
    },
    {
      "epoch": 0.355507010224363,
      "grad_norm": 1.616997480392456,
      "learning_rate": 0.00014938936805850955,
      "loss": 1.2796,
      "step": 9927
    },
    {
      "epoch": 0.35554282235393125,
      "grad_norm": 1.5822079181671143,
      "learning_rate": 0.0001493792821258777,
      "loss": 1.5981,
      "step": 9928
    },
    {
      "epoch": 0.3555786344834996,
      "grad_norm": 1.6461950540542603,
      "learning_rate": 0.00014936919552891134,
      "loss": 1.2901,
      "step": 9929
    },
    {
      "epoch": 0.35561444661306785,
      "grad_norm": 1.615821123123169,
      "learning_rate": 0.00014935910826774612,
      "loss": 1.7623,
      "step": 9930
    },
    {
      "epoch": 0.3556502587426361,
      "grad_norm": 2.009953260421753,
      "learning_rate": 0.0001493490203425178,
      "loss": 1.4007,
      "step": 9931
    },
    {
      "epoch": 0.35568607087220444,
      "grad_norm": 1.6070547103881836,
      "learning_rate": 0.00014933893175336202,
      "loss": 1.6903,
      "step": 9932
    },
    {
      "epoch": 0.3557218830017727,
      "grad_norm": 1.6759109497070312,
      "learning_rate": 0.0001493288425004146,
      "loss": 1.5817,
      "step": 9933
    },
    {
      "epoch": 0.355757695131341,
      "grad_norm": 2.2906291484832764,
      "learning_rate": 0.00014931875258381117,
      "loss": 1.7121,
      "step": 9934
    },
    {
      "epoch": 0.35579350726090925,
      "grad_norm": 1.4188705682754517,
      "learning_rate": 0.00014930866200368761,
      "loss": 1.81,
      "step": 9935
    },
    {
      "epoch": 0.35582931939047757,
      "grad_norm": 2.0052592754364014,
      "learning_rate": 0.0001492985707601796,
      "loss": 1.5919,
      "step": 9936
    },
    {
      "epoch": 0.35586513152004584,
      "grad_norm": 1.7482327222824097,
      "learning_rate": 0.00014928847885342287,
      "loss": 1.4017,
      "step": 9937
    },
    {
      "epoch": 0.3559009436496141,
      "grad_norm": 1.6666542291641235,
      "learning_rate": 0.00014927838628355327,
      "loss": 1.3715,
      "step": 9938
    },
    {
      "epoch": 0.35593675577918243,
      "grad_norm": 1.207485556602478,
      "learning_rate": 0.0001492682930507065,
      "loss": 1.3941,
      "step": 9939
    },
    {
      "epoch": 0.3559725679087507,
      "grad_norm": 1.869559645652771,
      "learning_rate": 0.00014925819915501847,
      "loss": 1.7862,
      "step": 9940
    },
    {
      "epoch": 0.356008380038319,
      "grad_norm": 1.9369616508483887,
      "learning_rate": 0.00014924810459662484,
      "loss": 1.6843,
      "step": 9941
    },
    {
      "epoch": 0.35604419216788724,
      "grad_norm": 2.2861275672912598,
      "learning_rate": 0.0001492380093756615,
      "loss": 1.3355,
      "step": 9942
    },
    {
      "epoch": 0.35608000429745557,
      "grad_norm": 1.4812921285629272,
      "learning_rate": 0.0001492279134922643,
      "loss": 1.4724,
      "step": 9943
    },
    {
      "epoch": 0.35611581642702383,
      "grad_norm": 1.890170931816101,
      "learning_rate": 0.000149217816946569,
      "loss": 1.403,
      "step": 9944
    },
    {
      "epoch": 0.3561516285565921,
      "grad_norm": 1.7531453371047974,
      "learning_rate": 0.00014920771973871147,
      "loss": 1.8111,
      "step": 9945
    },
    {
      "epoch": 0.35618744068616043,
      "grad_norm": 2.1455671787261963,
      "learning_rate": 0.00014919762186882754,
      "loss": 1.6263,
      "step": 9946
    },
    {
      "epoch": 0.3562232528157287,
      "grad_norm": 1.9717743396759033,
      "learning_rate": 0.00014918752333705303,
      "loss": 1.1631,
      "step": 9947
    },
    {
      "epoch": 0.35625906494529697,
      "grad_norm": 1.4446394443511963,
      "learning_rate": 0.00014917742414352386,
      "loss": 1.4625,
      "step": 9948
    },
    {
      "epoch": 0.35629487707486523,
      "grad_norm": 1.9484100341796875,
      "learning_rate": 0.00014916732428837593,
      "loss": 1.8046,
      "step": 9949
    },
    {
      "epoch": 0.35633068920443356,
      "grad_norm": 1.383619785308838,
      "learning_rate": 0.00014915722377174503,
      "loss": 1.3816,
      "step": 9950
    },
    {
      "epoch": 0.35636650133400183,
      "grad_norm": 1.6241815090179443,
      "learning_rate": 0.0001491471225937671,
      "loss": 1.5867,
      "step": 9951
    },
    {
      "epoch": 0.3564023134635701,
      "grad_norm": 1.8264425992965698,
      "learning_rate": 0.000149137020754578,
      "loss": 1.6515,
      "step": 9952
    },
    {
      "epoch": 0.3564381255931384,
      "grad_norm": 1.544845700263977,
      "learning_rate": 0.0001491269182543137,
      "loss": 1.5767,
      "step": 9953
    },
    {
      "epoch": 0.3564739377227067,
      "grad_norm": 2.761215925216675,
      "learning_rate": 0.0001491168150931101,
      "loss": 1.3875,
      "step": 9954
    },
    {
      "epoch": 0.35650974985227496,
      "grad_norm": 2.150071859359741,
      "learning_rate": 0.00014910671127110308,
      "loss": 1.3972,
      "step": 9955
    },
    {
      "epoch": 0.35654556198184323,
      "grad_norm": 1.8057886362075806,
      "learning_rate": 0.00014909660678842862,
      "loss": 1.4971,
      "step": 9956
    },
    {
      "epoch": 0.35658137411141155,
      "grad_norm": 1.5841161012649536,
      "learning_rate": 0.0001490865016452226,
      "loss": 1.6912,
      "step": 9957
    },
    {
      "epoch": 0.3566171862409798,
      "grad_norm": 1.4754817485809326,
      "learning_rate": 0.00014907639584162109,
      "loss": 1.6008,
      "step": 9958
    },
    {
      "epoch": 0.3566529983705481,
      "grad_norm": 1.5032598972320557,
      "learning_rate": 0.00014906628937775995,
      "loss": 1.6716,
      "step": 9959
    },
    {
      "epoch": 0.3566888105001164,
      "grad_norm": 1.5778634548187256,
      "learning_rate": 0.00014905618225377517,
      "loss": 1.2682,
      "step": 9960
    },
    {
      "epoch": 0.3567246226296847,
      "grad_norm": 2.124629259109497,
      "learning_rate": 0.00014904607446980273,
      "loss": 1.4935,
      "step": 9961
    },
    {
      "epoch": 0.35676043475925295,
      "grad_norm": 1.7733240127563477,
      "learning_rate": 0.00014903596602597864,
      "loss": 1.8838,
      "step": 9962
    },
    {
      "epoch": 0.3567962468888212,
      "grad_norm": 2.439082622528076,
      "learning_rate": 0.00014902585692243885,
      "loss": 1.5614,
      "step": 9963
    },
    {
      "epoch": 0.35683205901838955,
      "grad_norm": 1.8383203744888306,
      "learning_rate": 0.00014901574715931942,
      "loss": 1.386,
      "step": 9964
    },
    {
      "epoch": 0.3568678711479578,
      "grad_norm": 1.448671579360962,
      "learning_rate": 0.00014900563673675633,
      "loss": 1.3934,
      "step": 9965
    },
    {
      "epoch": 0.3569036832775261,
      "grad_norm": 1.9962146282196045,
      "learning_rate": 0.00014899552565488563,
      "loss": 1.7006,
      "step": 9966
    },
    {
      "epoch": 0.3569394954070944,
      "grad_norm": 1.699221134185791,
      "learning_rate": 0.0001489854139138433,
      "loss": 1.5871,
      "step": 9967
    },
    {
      "epoch": 0.3569753075366627,
      "grad_norm": 1.5800422430038452,
      "learning_rate": 0.00014897530151376545,
      "loss": 1.6377,
      "step": 9968
    },
    {
      "epoch": 0.35701111966623095,
      "grad_norm": 1.6919838190078735,
      "learning_rate": 0.00014896518845478805,
      "loss": 1.424,
      "step": 9969
    },
    {
      "epoch": 0.3570469317957992,
      "grad_norm": 1.6668004989624023,
      "learning_rate": 0.00014895507473704718,
      "loss": 1.6673,
      "step": 9970
    },
    {
      "epoch": 0.35708274392536754,
      "grad_norm": 1.9433567523956299,
      "learning_rate": 0.00014894496036067903,
      "loss": 1.3958,
      "step": 9971
    },
    {
      "epoch": 0.3571185560549358,
      "grad_norm": 2.3744451999664307,
      "learning_rate": 0.00014893484532581947,
      "loss": 1.5432,
      "step": 9972
    },
    {
      "epoch": 0.3571543681845041,
      "grad_norm": 1.429279088973999,
      "learning_rate": 0.00014892472963260475,
      "loss": 1.9036,
      "step": 9973
    },
    {
      "epoch": 0.3571901803140724,
      "grad_norm": 1.7184338569641113,
      "learning_rate": 0.00014891461328117087,
      "loss": 1.3598,
      "step": 9974
    },
    {
      "epoch": 0.35722599244364067,
      "grad_norm": 1.5399808883666992,
      "learning_rate": 0.00014890449627165398,
      "loss": 1.5129,
      "step": 9975
    },
    {
      "epoch": 0.35726180457320894,
      "grad_norm": 1.5570214986801147,
      "learning_rate": 0.00014889437860419013,
      "loss": 1.6888,
      "step": 9976
    },
    {
      "epoch": 0.3572976167027772,
      "grad_norm": 1.7206803560256958,
      "learning_rate": 0.00014888426027891553,
      "loss": 1.6131,
      "step": 9977
    },
    {
      "epoch": 0.35733342883234553,
      "grad_norm": 1.5917768478393555,
      "learning_rate": 0.00014887414129596623,
      "loss": 1.6743,
      "step": 9978
    },
    {
      "epoch": 0.3573692409619138,
      "grad_norm": 1.8771898746490479,
      "learning_rate": 0.00014886402165547845,
      "loss": 1.4912,
      "step": 9979
    },
    {
      "epoch": 0.3574050530914821,
      "grad_norm": 1.189568281173706,
      "learning_rate": 0.00014885390135758826,
      "loss": 1.4007,
      "step": 9980
    },
    {
      "epoch": 0.3574408652210504,
      "grad_norm": 1.9327512979507446,
      "learning_rate": 0.00014884378040243184,
      "loss": 1.6163,
      "step": 9981
    },
    {
      "epoch": 0.35747667735061867,
      "grad_norm": 1.1493306159973145,
      "learning_rate": 0.0001488336587901454,
      "loss": 1.4209,
      "step": 9982
    },
    {
      "epoch": 0.35751248948018693,
      "grad_norm": 1.4550294876098633,
      "learning_rate": 0.000148823536520865,
      "loss": 1.3286,
      "step": 9983
    },
    {
      "epoch": 0.3575483016097552,
      "grad_norm": 1.721457600593567,
      "learning_rate": 0.00014881341359472696,
      "loss": 1.6156,
      "step": 9984
    },
    {
      "epoch": 0.35758411373932353,
      "grad_norm": 1.9905215501785278,
      "learning_rate": 0.00014880329001186736,
      "loss": 1.5031,
      "step": 9985
    },
    {
      "epoch": 0.3576199258688918,
      "grad_norm": 1.5918910503387451,
      "learning_rate": 0.00014879316577242246,
      "loss": 1.4491,
      "step": 9986
    },
    {
      "epoch": 0.35765573799846007,
      "grad_norm": 2.3297674655914307,
      "learning_rate": 0.00014878304087652847,
      "loss": 1.5093,
      "step": 9987
    },
    {
      "epoch": 0.3576915501280284,
      "grad_norm": 1.8075636625289917,
      "learning_rate": 0.00014877291532432158,
      "loss": 1.4857,
      "step": 9988
    },
    {
      "epoch": 0.35772736225759666,
      "grad_norm": 1.7248746156692505,
      "learning_rate": 0.00014876278911593802,
      "loss": 1.6853,
      "step": 9989
    },
    {
      "epoch": 0.35776317438716493,
      "grad_norm": 1.792223334312439,
      "learning_rate": 0.00014875266225151403,
      "loss": 1.2533,
      "step": 9990
    },
    {
      "epoch": 0.3577989865167332,
      "grad_norm": 1.5456364154815674,
      "learning_rate": 0.00014874253473118586,
      "loss": 1.4788,
      "step": 9991
    },
    {
      "epoch": 0.3578347986463015,
      "grad_norm": 1.5730777978897095,
      "learning_rate": 0.00014873240655508975,
      "loss": 1.4009,
      "step": 9992
    },
    {
      "epoch": 0.3578706107758698,
      "grad_norm": 1.5028215646743774,
      "learning_rate": 0.00014872227772336197,
      "loss": 1.5448,
      "step": 9993
    },
    {
      "epoch": 0.35790642290543806,
      "grad_norm": 1.584532618522644,
      "learning_rate": 0.0001487121482361388,
      "loss": 1.4787,
      "step": 9994
    },
    {
      "epoch": 0.3579422350350064,
      "grad_norm": 1.7555292844772339,
      "learning_rate": 0.00014870201809355653,
      "loss": 1.3804,
      "step": 9995
    },
    {
      "epoch": 0.35797804716457465,
      "grad_norm": 1.8509167432785034,
      "learning_rate": 0.0001486918872957514,
      "loss": 1.4473,
      "step": 9996
    },
    {
      "epoch": 0.3580138592941429,
      "grad_norm": 1.639758825302124,
      "learning_rate": 0.00014868175584285974,
      "loss": 1.4596,
      "step": 9997
    },
    {
      "epoch": 0.3580496714237112,
      "grad_norm": 1.8261754512786865,
      "learning_rate": 0.00014867162373501786,
      "loss": 1.6244,
      "step": 9998
    },
    {
      "epoch": 0.3580854835532795,
      "grad_norm": 1.5407792329788208,
      "learning_rate": 0.00014866149097236204,
      "loss": 1.4145,
      "step": 9999
    },
    {
      "epoch": 0.3581212956828478,
      "grad_norm": 1.653843879699707,
      "learning_rate": 0.00014865135755502866,
      "loss": 1.3846,
      "step": 10000
    },
    {
      "epoch": 0.35815710781241605,
      "grad_norm": 1.4592039585113525,
      "learning_rate": 0.000148641223483154,
      "loss": 1.3947,
      "step": 10001
    },
    {
      "epoch": 0.3581929199419843,
      "grad_norm": 1.936527132987976,
      "learning_rate": 0.00014863108875687444,
      "loss": 1.2548,
      "step": 10002
    },
    {
      "epoch": 0.35822873207155265,
      "grad_norm": 1.7076057195663452,
      "learning_rate": 0.00014862095337632626,
      "loss": 1.3316,
      "step": 10003
    },
    {
      "epoch": 0.3582645442011209,
      "grad_norm": 1.6434259414672852,
      "learning_rate": 0.00014861081734164592,
      "loss": 1.6868,
      "step": 10004
    },
    {
      "epoch": 0.3583003563306892,
      "grad_norm": 1.6703202724456787,
      "learning_rate": 0.0001486006806529697,
      "loss": 1.4944,
      "step": 10005
    },
    {
      "epoch": 0.3583361684602575,
      "grad_norm": 1.4541386365890503,
      "learning_rate": 0.00014859054331043406,
      "loss": 1.6077,
      "step": 10006
    },
    {
      "epoch": 0.3583719805898258,
      "grad_norm": 1.5497562885284424,
      "learning_rate": 0.0001485804053141753,
      "loss": 1.5403,
      "step": 10007
    },
    {
      "epoch": 0.35840779271939405,
      "grad_norm": 1.54259192943573,
      "learning_rate": 0.00014857026666432988,
      "loss": 1.5234,
      "step": 10008
    },
    {
      "epoch": 0.3584436048489623,
      "grad_norm": 1.5208549499511719,
      "learning_rate": 0.00014856012736103413,
      "loss": 1.6961,
      "step": 10009
    },
    {
      "epoch": 0.35847941697853064,
      "grad_norm": 1.6700527667999268,
      "learning_rate": 0.00014854998740442454,
      "loss": 1.5871,
      "step": 10010
    },
    {
      "epoch": 0.3585152291080989,
      "grad_norm": 1.868094563484192,
      "learning_rate": 0.00014853984679463747,
      "loss": 1.6691,
      "step": 10011
    },
    {
      "epoch": 0.3585510412376672,
      "grad_norm": 2.0113749504089355,
      "learning_rate": 0.00014852970553180938,
      "loss": 1.1609,
      "step": 10012
    },
    {
      "epoch": 0.3585868533672355,
      "grad_norm": 1.3860893249511719,
      "learning_rate": 0.0001485195636160767,
      "loss": 1.4034,
      "step": 10013
    },
    {
      "epoch": 0.35862266549680377,
      "grad_norm": 2.6419312953948975,
      "learning_rate": 0.00014850942104757588,
      "loss": 1.5761,
      "step": 10014
    },
    {
      "epoch": 0.35865847762637204,
      "grad_norm": 1.6753020286560059,
      "learning_rate": 0.0001484992778264434,
      "loss": 1.5387,
      "step": 10015
    },
    {
      "epoch": 0.3586942897559403,
      "grad_norm": 1.7237919569015503,
      "learning_rate": 0.00014848913395281568,
      "loss": 1.4762,
      "step": 10016
    },
    {
      "epoch": 0.35873010188550863,
      "grad_norm": 1.547621488571167,
      "learning_rate": 0.00014847898942682922,
      "loss": 1.5384,
      "step": 10017
    },
    {
      "epoch": 0.3587659140150769,
      "grad_norm": 1.5596733093261719,
      "learning_rate": 0.00014846884424862044,
      "loss": 1.4112,
      "step": 10018
    },
    {
      "epoch": 0.35880172614464517,
      "grad_norm": 1.4373083114624023,
      "learning_rate": 0.0001484586984183259,
      "loss": 1.3855,
      "step": 10019
    },
    {
      "epoch": 0.3588375382742135,
      "grad_norm": 1.8072415590286255,
      "learning_rate": 0.0001484485519360821,
      "loss": 1.5663,
      "step": 10020
    },
    {
      "epoch": 0.35887335040378177,
      "grad_norm": 1.838571310043335,
      "learning_rate": 0.00014843840480202554,
      "loss": 1.401,
      "step": 10021
    },
    {
      "epoch": 0.35890916253335003,
      "grad_norm": 1.517883539199829,
      "learning_rate": 0.00014842825701629267,
      "loss": 1.2617,
      "step": 10022
    },
    {
      "epoch": 0.3589449746629183,
      "grad_norm": 1.988364577293396,
      "learning_rate": 0.0001484181085790201,
      "loss": 1.478,
      "step": 10023
    },
    {
      "epoch": 0.35898078679248663,
      "grad_norm": 1.5956732034683228,
      "learning_rate": 0.00014840795949034439,
      "loss": 1.5329,
      "step": 10024
    },
    {
      "epoch": 0.3590165989220549,
      "grad_norm": 1.7661099433898926,
      "learning_rate": 0.00014839780975040194,
      "loss": 1.465,
      "step": 10025
    },
    {
      "epoch": 0.35905241105162317,
      "grad_norm": 1.3681966066360474,
      "learning_rate": 0.00014838765935932944,
      "loss": 1.5103,
      "step": 10026
    },
    {
      "epoch": 0.3590882231811915,
      "grad_norm": 2.267117977142334,
      "learning_rate": 0.00014837750831726338,
      "loss": 1.52,
      "step": 10027
    },
    {
      "epoch": 0.35912403531075976,
      "grad_norm": 1.8167510032653809,
      "learning_rate": 0.00014836735662434035,
      "loss": 1.6152,
      "step": 10028
    },
    {
      "epoch": 0.35915984744032803,
      "grad_norm": 1.5171904563903809,
      "learning_rate": 0.00014835720428069693,
      "loss": 1.0415,
      "step": 10029
    },
    {
      "epoch": 0.3591956595698963,
      "grad_norm": 1.476212739944458,
      "learning_rate": 0.0001483470512864697,
      "loss": 1.345,
      "step": 10030
    },
    {
      "epoch": 0.3592314716994646,
      "grad_norm": 1.4677760601043701,
      "learning_rate": 0.00014833689764179523,
      "loss": 1.6914,
      "step": 10031
    },
    {
      "epoch": 0.3592672838290329,
      "grad_norm": 1.542527198791504,
      "learning_rate": 0.00014832674334681022,
      "loss": 1.5375,
      "step": 10032
    },
    {
      "epoch": 0.35930309595860116,
      "grad_norm": 1.7230689525604248,
      "learning_rate": 0.00014831658840165117,
      "loss": 1.832,
      "step": 10033
    },
    {
      "epoch": 0.3593389080881695,
      "grad_norm": 2.0981647968292236,
      "learning_rate": 0.00014830643280645472,
      "loss": 1.4325,
      "step": 10034
    },
    {
      "epoch": 0.35937472021773775,
      "grad_norm": 1.4687614440917969,
      "learning_rate": 0.00014829627656135757,
      "loss": 1.5059,
      "step": 10035
    },
    {
      "epoch": 0.359410532347306,
      "grad_norm": 1.6538559198379517,
      "learning_rate": 0.0001482861196664963,
      "loss": 1.4741,
      "step": 10036
    },
    {
      "epoch": 0.3594463444768743,
      "grad_norm": 1.6689916849136353,
      "learning_rate": 0.00014827596212200762,
      "loss": 1.5735,
      "step": 10037
    },
    {
      "epoch": 0.3594821566064426,
      "grad_norm": 1.9871309995651245,
      "learning_rate": 0.00014826580392802806,
      "loss": 1.8311,
      "step": 10038
    },
    {
      "epoch": 0.3595179687360109,
      "grad_norm": 2.3612310886383057,
      "learning_rate": 0.00014825564508469443,
      "loss": 1.7536,
      "step": 10039
    },
    {
      "epoch": 0.35955378086557915,
      "grad_norm": 1.9663517475128174,
      "learning_rate": 0.00014824548559214332,
      "loss": 1.4124,
      "step": 10040
    },
    {
      "epoch": 0.3595895929951475,
      "grad_norm": 1.631361961364746,
      "learning_rate": 0.0001482353254505114,
      "loss": 1.6325,
      "step": 10041
    },
    {
      "epoch": 0.35962540512471575,
      "grad_norm": 1.6884336471557617,
      "learning_rate": 0.00014822516465993546,
      "loss": 1.4287,
      "step": 10042
    },
    {
      "epoch": 0.359661217254284,
      "grad_norm": 1.5852062702178955,
      "learning_rate": 0.0001482150032205521,
      "loss": 1.6681,
      "step": 10043
    },
    {
      "epoch": 0.3596970293838523,
      "grad_norm": 1.4541549682617188,
      "learning_rate": 0.00014820484113249805,
      "loss": 1.3463,
      "step": 10044
    },
    {
      "epoch": 0.3597328415134206,
      "grad_norm": 1.5388379096984863,
      "learning_rate": 0.00014819467839591007,
      "loss": 1.5042,
      "step": 10045
    },
    {
      "epoch": 0.3597686536429889,
      "grad_norm": 1.8870503902435303,
      "learning_rate": 0.00014818451501092485,
      "loss": 1.5489,
      "step": 10046
    },
    {
      "epoch": 0.35980446577255715,
      "grad_norm": 1.8153746128082275,
      "learning_rate": 0.00014817435097767912,
      "loss": 1.6726,
      "step": 10047
    },
    {
      "epoch": 0.35984027790212547,
      "grad_norm": 1.4867844581604004,
      "learning_rate": 0.00014816418629630968,
      "loss": 1.3771,
      "step": 10048
    },
    {
      "epoch": 0.35987609003169374,
      "grad_norm": 1.6260762214660645,
      "learning_rate": 0.0001481540209669532,
      "loss": 1.5194,
      "step": 10049
    },
    {
      "epoch": 0.359911902161262,
      "grad_norm": 1.7411034107208252,
      "learning_rate": 0.0001481438549897465,
      "loss": 1.3861,
      "step": 10050
    },
    {
      "epoch": 0.3599477142908303,
      "grad_norm": 1.4617067575454712,
      "learning_rate": 0.00014813368836482632,
      "loss": 1.1129,
      "step": 10051
    },
    {
      "epoch": 0.3599835264203986,
      "grad_norm": 2.216752767562866,
      "learning_rate": 0.00014812352109232947,
      "loss": 1.3592,
      "step": 10052
    },
    {
      "epoch": 0.36001933854996687,
      "grad_norm": 1.7082031965255737,
      "learning_rate": 0.0001481133531723927,
      "loss": 1.2952,
      "step": 10053
    },
    {
      "epoch": 0.36005515067953514,
      "grad_norm": 2.7109858989715576,
      "learning_rate": 0.00014810318460515282,
      "loss": 1.9209,
      "step": 10054
    },
    {
      "epoch": 0.36009096280910347,
      "grad_norm": 1.6055452823638916,
      "learning_rate": 0.00014809301539074667,
      "loss": 1.3236,
      "step": 10055
    },
    {
      "epoch": 0.36012677493867173,
      "grad_norm": 1.9085619449615479,
      "learning_rate": 0.000148082845529311,
      "loss": 1.6722,
      "step": 10056
    },
    {
      "epoch": 0.36016258706824,
      "grad_norm": 1.3563839197158813,
      "learning_rate": 0.00014807267502098267,
      "loss": 1.4078,
      "step": 10057
    },
    {
      "epoch": 0.36019839919780827,
      "grad_norm": 1.5918482542037964,
      "learning_rate": 0.00014806250386589851,
      "loss": 1.5898,
      "step": 10058
    },
    {
      "epoch": 0.3602342113273766,
      "grad_norm": 1.934302806854248,
      "learning_rate": 0.0001480523320641954,
      "loss": 1.288,
      "step": 10059
    },
    {
      "epoch": 0.36027002345694487,
      "grad_norm": 2.7092247009277344,
      "learning_rate": 0.00014804215961601008,
      "loss": 1.4568,
      "step": 10060
    },
    {
      "epoch": 0.36030583558651313,
      "grad_norm": 1.5964667797088623,
      "learning_rate": 0.00014803198652147952,
      "loss": 1.8112,
      "step": 10061
    },
    {
      "epoch": 0.36034164771608146,
      "grad_norm": 1.7393267154693604,
      "learning_rate": 0.00014802181278074052,
      "loss": 1.5946,
      "step": 10062
    },
    {
      "epoch": 0.36037745984564973,
      "grad_norm": 1.6845488548278809,
      "learning_rate": 0.00014801163839392998,
      "loss": 1.544,
      "step": 10063
    },
    {
      "epoch": 0.360413271975218,
      "grad_norm": 1.5113362073898315,
      "learning_rate": 0.00014800146336118474,
      "loss": 1.5542,
      "step": 10064
    },
    {
      "epoch": 0.36044908410478627,
      "grad_norm": 1.7169773578643799,
      "learning_rate": 0.0001479912876826418,
      "loss": 1.5439,
      "step": 10065
    },
    {
      "epoch": 0.3604848962343546,
      "grad_norm": 1.4633331298828125,
      "learning_rate": 0.0001479811113584379,
      "loss": 1.595,
      "step": 10066
    },
    {
      "epoch": 0.36052070836392286,
      "grad_norm": 1.579038381576538,
      "learning_rate": 0.00014797093438871008,
      "loss": 1.4934,
      "step": 10067
    },
    {
      "epoch": 0.36055652049349113,
      "grad_norm": 1.9892122745513916,
      "learning_rate": 0.00014796075677359525,
      "loss": 1.5002,
      "step": 10068
    },
    {
      "epoch": 0.36059233262305945,
      "grad_norm": 1.6243116855621338,
      "learning_rate": 0.00014795057851323023,
      "loss": 1.5465,
      "step": 10069
    },
    {
      "epoch": 0.3606281447526277,
      "grad_norm": 1.6825838088989258,
      "learning_rate": 0.0001479403996077521,
      "loss": 1.163,
      "step": 10070
    },
    {
      "epoch": 0.360663956882196,
      "grad_norm": 2.1626789569854736,
      "learning_rate": 0.0001479302200572977,
      "loss": 1.389,
      "step": 10071
    },
    {
      "epoch": 0.36069976901176426,
      "grad_norm": 1.4115447998046875,
      "learning_rate": 0.00014792003986200403,
      "loss": 1.4016,
      "step": 10072
    },
    {
      "epoch": 0.3607355811413326,
      "grad_norm": 2.0680456161499023,
      "learning_rate": 0.000147909859022008,
      "loss": 1.4026,
      "step": 10073
    },
    {
      "epoch": 0.36077139327090085,
      "grad_norm": 1.1022719144821167,
      "learning_rate": 0.00014789967753744664,
      "loss": 1.3813,
      "step": 10074
    },
    {
      "epoch": 0.3608072054004691,
      "grad_norm": 1.9899226427078247,
      "learning_rate": 0.00014788949540845689,
      "loss": 1.4066,
      "step": 10075
    },
    {
      "epoch": 0.36084301753003745,
      "grad_norm": 1.577979564666748,
      "learning_rate": 0.0001478793126351758,
      "loss": 1.5398,
      "step": 10076
    },
    {
      "epoch": 0.3608788296596057,
      "grad_norm": 1.9945513010025024,
      "learning_rate": 0.00014786912921774028,
      "loss": 1.5577,
      "step": 10077
    },
    {
      "epoch": 0.360914641789174,
      "grad_norm": 1.5704258680343628,
      "learning_rate": 0.00014785894515628736,
      "loss": 1.6951,
      "step": 10078
    },
    {
      "epoch": 0.36095045391874225,
      "grad_norm": 1.760332703590393,
      "learning_rate": 0.0001478487604509541,
      "loss": 1.4839,
      "step": 10079
    },
    {
      "epoch": 0.3609862660483106,
      "grad_norm": 1.6892656087875366,
      "learning_rate": 0.00014783857510187743,
      "loss": 1.7012,
      "step": 10080
    },
    {
      "epoch": 0.36102207817787885,
      "grad_norm": 1.9408334493637085,
      "learning_rate": 0.00014782838910919449,
      "loss": 1.3263,
      "step": 10081
    },
    {
      "epoch": 0.3610578903074471,
      "grad_norm": 1.6504887342453003,
      "learning_rate": 0.00014781820247304227,
      "loss": 1.5619,
      "step": 10082
    },
    {
      "epoch": 0.36109370243701544,
      "grad_norm": 1.608626127243042,
      "learning_rate": 0.00014780801519355782,
      "loss": 1.461,
      "step": 10083
    },
    {
      "epoch": 0.3611295145665837,
      "grad_norm": 2.068258285522461,
      "learning_rate": 0.00014779782727087815,
      "loss": 1.5741,
      "step": 10084
    },
    {
      "epoch": 0.361165326696152,
      "grad_norm": 1.8851020336151123,
      "learning_rate": 0.0001477876387051404,
      "loss": 1.66,
      "step": 10085
    },
    {
      "epoch": 0.36120113882572025,
      "grad_norm": 1.5188018083572388,
      "learning_rate": 0.00014777744949648163,
      "loss": 1.4233,
      "step": 10086
    },
    {
      "epoch": 0.36123695095528857,
      "grad_norm": 1.492539405822754,
      "learning_rate": 0.00014776725964503888,
      "loss": 1.4101,
      "step": 10087
    },
    {
      "epoch": 0.36127276308485684,
      "grad_norm": 2.47213077545166,
      "learning_rate": 0.00014775706915094928,
      "loss": 1.7695,
      "step": 10088
    },
    {
      "epoch": 0.3613085752144251,
      "grad_norm": 3.236239194869995,
      "learning_rate": 0.0001477468780143499,
      "loss": 1.3959,
      "step": 10089
    },
    {
      "epoch": 0.36134438734399343,
      "grad_norm": 1.5189967155456543,
      "learning_rate": 0.00014773668623537786,
      "loss": 1.3773,
      "step": 10090
    },
    {
      "epoch": 0.3613801994735617,
      "grad_norm": 1.4003039598464966,
      "learning_rate": 0.0001477264938141703,
      "loss": 1.6206,
      "step": 10091
    },
    {
      "epoch": 0.36141601160312997,
      "grad_norm": 1.4691357612609863,
      "learning_rate": 0.00014771630075086434,
      "loss": 1.5497,
      "step": 10092
    },
    {
      "epoch": 0.36145182373269824,
      "grad_norm": 1.6265677213668823,
      "learning_rate": 0.00014770610704559708,
      "loss": 1.7204,
      "step": 10093
    },
    {
      "epoch": 0.36148763586226657,
      "grad_norm": 1.6044228076934814,
      "learning_rate": 0.0001476959126985057,
      "loss": 1.464,
      "step": 10094
    },
    {
      "epoch": 0.36152344799183483,
      "grad_norm": 1.8812122344970703,
      "learning_rate": 0.00014768571770972734,
      "loss": 1.5653,
      "step": 10095
    },
    {
      "epoch": 0.3615592601214031,
      "grad_norm": 1.898262858390808,
      "learning_rate": 0.00014767552207939913,
      "loss": 1.2871,
      "step": 10096
    },
    {
      "epoch": 0.3615950722509714,
      "grad_norm": 1.926661729812622,
      "learning_rate": 0.0001476653258076583,
      "loss": 1.5776,
      "step": 10097
    },
    {
      "epoch": 0.3616308843805397,
      "grad_norm": 1.504490852355957,
      "learning_rate": 0.00014765512889464198,
      "loss": 1.4083,
      "step": 10098
    },
    {
      "epoch": 0.36166669651010797,
      "grad_norm": 1.455038070678711,
      "learning_rate": 0.00014764493134048737,
      "loss": 1.553,
      "step": 10099
    },
    {
      "epoch": 0.36170250863967623,
      "grad_norm": 1.8379685878753662,
      "learning_rate": 0.00014763473314533166,
      "loss": 1.2926,
      "step": 10100
    },
    {
      "epoch": 0.36173832076924456,
      "grad_norm": 1.7743902206420898,
      "learning_rate": 0.0001476245343093121,
      "loss": 1.4345,
      "step": 10101
    },
    {
      "epoch": 0.36177413289881283,
      "grad_norm": 1.722127914428711,
      "learning_rate": 0.00014761433483256582,
      "loss": 1.528,
      "step": 10102
    },
    {
      "epoch": 0.3618099450283811,
      "grad_norm": 1.4296984672546387,
      "learning_rate": 0.00014760413471523012,
      "loss": 1.3692,
      "step": 10103
    },
    {
      "epoch": 0.3618457571579494,
      "grad_norm": 1.6753002405166626,
      "learning_rate": 0.00014759393395744215,
      "loss": 1.6153,
      "step": 10104
    },
    {
      "epoch": 0.3618815692875177,
      "grad_norm": 3.1370999813079834,
      "learning_rate": 0.00014758373255933924,
      "loss": 1.8066,
      "step": 10105
    },
    {
      "epoch": 0.36191738141708596,
      "grad_norm": 1.653550624847412,
      "learning_rate": 0.00014757353052105853,
      "loss": 1.2039,
      "step": 10106
    },
    {
      "epoch": 0.36195319354665423,
      "grad_norm": 2.0767953395843506,
      "learning_rate": 0.00014756332784273738,
      "loss": 1.678,
      "step": 10107
    },
    {
      "epoch": 0.36198900567622255,
      "grad_norm": 1.60899019241333,
      "learning_rate": 0.00014755312452451296,
      "loss": 1.8436,
      "step": 10108
    },
    {
      "epoch": 0.3620248178057908,
      "grad_norm": 1.545812726020813,
      "learning_rate": 0.0001475429205665226,
      "loss": 1.3691,
      "step": 10109
    },
    {
      "epoch": 0.3620606299353591,
      "grad_norm": 1.5063380002975464,
      "learning_rate": 0.0001475327159689036,
      "loss": 1.7184,
      "step": 10110
    },
    {
      "epoch": 0.3620964420649274,
      "grad_norm": 2.4832561016082764,
      "learning_rate": 0.0001475225107317932,
      "loss": 1.3295,
      "step": 10111
    },
    {
      "epoch": 0.3621322541944957,
      "grad_norm": 3.1396214962005615,
      "learning_rate": 0.00014751230485532873,
      "loss": 1.6655,
      "step": 10112
    },
    {
      "epoch": 0.36216806632406395,
      "grad_norm": 1.7851954698562622,
      "learning_rate": 0.00014750209833964747,
      "loss": 1.5574,
      "step": 10113
    },
    {
      "epoch": 0.3622038784536322,
      "grad_norm": 1.6681740283966064,
      "learning_rate": 0.00014749189118488677,
      "loss": 1.4129,
      "step": 10114
    },
    {
      "epoch": 0.36223969058320055,
      "grad_norm": 1.871593952178955,
      "learning_rate": 0.0001474816833911839,
      "loss": 1.536,
      "step": 10115
    },
    {
      "epoch": 0.3622755027127688,
      "grad_norm": 2.309457540512085,
      "learning_rate": 0.00014747147495867627,
      "loss": 1.3783,
      "step": 10116
    },
    {
      "epoch": 0.3623113148423371,
      "grad_norm": 2.2859158515930176,
      "learning_rate": 0.00014746126588750116,
      "loss": 1.6957,
      "step": 10117
    },
    {
      "epoch": 0.3623471269719054,
      "grad_norm": 1.851605772972107,
      "learning_rate": 0.00014745105617779594,
      "loss": 1.4928,
      "step": 10118
    },
    {
      "epoch": 0.3623829391014737,
      "grad_norm": 1.9329193830490112,
      "learning_rate": 0.00014744084582969793,
      "loss": 1.7967,
      "step": 10119
    },
    {
      "epoch": 0.36241875123104195,
      "grad_norm": 1.6493408679962158,
      "learning_rate": 0.00014743063484334455,
      "loss": 1.6187,
      "step": 10120
    },
    {
      "epoch": 0.3624545633606102,
      "grad_norm": 1.6800282001495361,
      "learning_rate": 0.00014742042321887322,
      "loss": 1.6358,
      "step": 10121
    },
    {
      "epoch": 0.36249037549017854,
      "grad_norm": 1.4881089925765991,
      "learning_rate": 0.00014741021095642117,
      "loss": 1.6944,
      "step": 10122
    },
    {
      "epoch": 0.3625261876197468,
      "grad_norm": 1.2354776859283447,
      "learning_rate": 0.00014739999805612596,
      "loss": 1.151,
      "step": 10123
    },
    {
      "epoch": 0.3625619997493151,
      "grad_norm": 1.8695483207702637,
      "learning_rate": 0.00014738978451812488,
      "loss": 1.4648,
      "step": 10124
    },
    {
      "epoch": 0.3625978118788834,
      "grad_norm": 1.4767268896102905,
      "learning_rate": 0.00014737957034255538,
      "loss": 1.5988,
      "step": 10125
    },
    {
      "epoch": 0.36263362400845167,
      "grad_norm": 1.4204144477844238,
      "learning_rate": 0.00014736935552955488,
      "loss": 1.4762,
      "step": 10126
    },
    {
      "epoch": 0.36266943613801994,
      "grad_norm": 1.6141799688339233,
      "learning_rate": 0.00014735914007926084,
      "loss": 1.4833,
      "step": 10127
    },
    {
      "epoch": 0.3627052482675882,
      "grad_norm": 2.05619740486145,
      "learning_rate": 0.0001473489239918106,
      "loss": 1.391,
      "step": 10128
    },
    {
      "epoch": 0.36274106039715653,
      "grad_norm": 1.3890151977539062,
      "learning_rate": 0.0001473387072673417,
      "loss": 1.5043,
      "step": 10129
    },
    {
      "epoch": 0.3627768725267248,
      "grad_norm": 1.4893358945846558,
      "learning_rate": 0.00014732848990599154,
      "loss": 1.4655,
      "step": 10130
    },
    {
      "epoch": 0.36281268465629307,
      "grad_norm": 1.4449459314346313,
      "learning_rate": 0.0001473182719078976,
      "loss": 1.5565,
      "step": 10131
    },
    {
      "epoch": 0.3628484967858614,
      "grad_norm": 1.4098438024520874,
      "learning_rate": 0.00014730805327319737,
      "loss": 1.5439,
      "step": 10132
    },
    {
      "epoch": 0.36288430891542967,
      "grad_norm": 1.4516136646270752,
      "learning_rate": 0.00014729783400202828,
      "loss": 1.5875,
      "step": 10133
    },
    {
      "epoch": 0.36292012104499793,
      "grad_norm": 2.057598114013672,
      "learning_rate": 0.00014728761409452785,
      "loss": 1.5132,
      "step": 10134
    },
    {
      "epoch": 0.3629559331745662,
      "grad_norm": 2.4684760570526123,
      "learning_rate": 0.00014727739355083357,
      "loss": 1.6289,
      "step": 10135
    },
    {
      "epoch": 0.3629917453041345,
      "grad_norm": 2.0605359077453613,
      "learning_rate": 0.00014726717237108293,
      "loss": 1.6129,
      "step": 10136
    },
    {
      "epoch": 0.3630275574337028,
      "grad_norm": 1.469491720199585,
      "learning_rate": 0.00014725695055541348,
      "loss": 1.4592,
      "step": 10137
    },
    {
      "epoch": 0.36306336956327107,
      "grad_norm": 1.547979474067688,
      "learning_rate": 0.00014724672810396272,
      "loss": 1.441,
      "step": 10138
    },
    {
      "epoch": 0.3630991816928394,
      "grad_norm": 1.272243618965149,
      "learning_rate": 0.00014723650501686817,
      "loss": 1.4066,
      "step": 10139
    },
    {
      "epoch": 0.36313499382240766,
      "grad_norm": 1.7786619663238525,
      "learning_rate": 0.00014722628129426734,
      "loss": 1.4726,
      "step": 10140
    },
    {
      "epoch": 0.36317080595197593,
      "grad_norm": 1.6617714166641235,
      "learning_rate": 0.0001472160569362979,
      "loss": 1.6447,
      "step": 10141
    },
    {
      "epoch": 0.3632066180815442,
      "grad_norm": 2.5883841514587402,
      "learning_rate": 0.0001472058319430972,
      "loss": 1.4975,
      "step": 10142
    },
    {
      "epoch": 0.3632424302111125,
      "grad_norm": 1.781213402748108,
      "learning_rate": 0.000147195606314803,
      "loss": 1.5525,
      "step": 10143
    },
    {
      "epoch": 0.3632782423406808,
      "grad_norm": 4.267208099365234,
      "learning_rate": 0.0001471853800515528,
      "loss": 1.6374,
      "step": 10144
    },
    {
      "epoch": 0.36331405447024906,
      "grad_norm": 1.460854172706604,
      "learning_rate": 0.00014717515315348413,
      "loss": 1.7116,
      "step": 10145
    },
    {
      "epoch": 0.3633498665998174,
      "grad_norm": 1.7079209089279175,
      "learning_rate": 0.00014716492562073466,
      "loss": 1.4572,
      "step": 10146
    },
    {
      "epoch": 0.36338567872938565,
      "grad_norm": 1.5507252216339111,
      "learning_rate": 0.00014715469745344196,
      "loss": 1.3672,
      "step": 10147
    },
    {
      "epoch": 0.3634214908589539,
      "grad_norm": 2.1776158809661865,
      "learning_rate": 0.00014714446865174362,
      "loss": 1.7472,
      "step": 10148
    },
    {
      "epoch": 0.3634573029885222,
      "grad_norm": 2.235363483428955,
      "learning_rate": 0.00014713423921577725,
      "loss": 1.4751,
      "step": 10149
    },
    {
      "epoch": 0.3634931151180905,
      "grad_norm": 2.331434726715088,
      "learning_rate": 0.0001471240091456805,
      "loss": 1.5584,
      "step": 10150
    },
    {
      "epoch": 0.3635289272476588,
      "grad_norm": 1.8828892707824707,
      "learning_rate": 0.00014711377844159099,
      "loss": 1.4935,
      "step": 10151
    },
    {
      "epoch": 0.36356473937722705,
      "grad_norm": 1.5459474325180054,
      "learning_rate": 0.00014710354710364637,
      "loss": 1.1762,
      "step": 10152
    },
    {
      "epoch": 0.3636005515067954,
      "grad_norm": 1.9720122814178467,
      "learning_rate": 0.00014709331513198425,
      "loss": 1.751,
      "step": 10153
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 1.7527331113815308,
      "learning_rate": 0.00014708308252674236,
      "loss": 1.3377,
      "step": 10154
    },
    {
      "epoch": 0.3636721757659319,
      "grad_norm": 1.449865698814392,
      "learning_rate": 0.0001470728492880583,
      "loss": 1.5698,
      "step": 10155
    },
    {
      "epoch": 0.3637079878955002,
      "grad_norm": 1.602096438407898,
      "learning_rate": 0.00014706261541606983,
      "loss": 1.5431,
      "step": 10156
    },
    {
      "epoch": 0.3637438000250685,
      "grad_norm": 1.5750555992126465,
      "learning_rate": 0.00014705238091091455,
      "loss": 1.5822,
      "step": 10157
    },
    {
      "epoch": 0.3637796121546368,
      "grad_norm": 1.5211970806121826,
      "learning_rate": 0.00014704214577273016,
      "loss": 1.4798,
      "step": 10158
    },
    {
      "epoch": 0.36381542428420505,
      "grad_norm": 1.606300950050354,
      "learning_rate": 0.00014703191000165438,
      "loss": 1.6259,
      "step": 10159
    },
    {
      "epoch": 0.36385123641377337,
      "grad_norm": 2.0974745750427246,
      "learning_rate": 0.00014702167359782493,
      "loss": 1.5262,
      "step": 10160
    },
    {
      "epoch": 0.36388704854334164,
      "grad_norm": 1.8589930534362793,
      "learning_rate": 0.0001470114365613795,
      "loss": 1.2566,
      "step": 10161
    },
    {
      "epoch": 0.3639228606729099,
      "grad_norm": 1.8257662057876587,
      "learning_rate": 0.00014700119889245582,
      "loss": 1.5583,
      "step": 10162
    },
    {
      "epoch": 0.3639586728024782,
      "grad_norm": 1.329789638519287,
      "learning_rate": 0.00014699096059119166,
      "loss": 1.4588,
      "step": 10163
    },
    {
      "epoch": 0.3639944849320465,
      "grad_norm": 1.6192381381988525,
      "learning_rate": 0.0001469807216577247,
      "loss": 1.6746,
      "step": 10164
    },
    {
      "epoch": 0.36403029706161477,
      "grad_norm": 3.251436948776245,
      "learning_rate": 0.0001469704820921928,
      "loss": 1.4111,
      "step": 10165
    },
    {
      "epoch": 0.36406610919118304,
      "grad_norm": 1.9543156623840332,
      "learning_rate": 0.00014696024189473362,
      "loss": 1.4814,
      "step": 10166
    },
    {
      "epoch": 0.36410192132075136,
      "grad_norm": 1.507798671722412,
      "learning_rate": 0.00014695000106548496,
      "loss": 1.4728,
      "step": 10167
    },
    {
      "epoch": 0.36413773345031963,
      "grad_norm": 2.0722126960754395,
      "learning_rate": 0.0001469397596045846,
      "loss": 1.5825,
      "step": 10168
    },
    {
      "epoch": 0.3641735455798879,
      "grad_norm": 2.018157720565796,
      "learning_rate": 0.0001469295175121703,
      "loss": 1.6858,
      "step": 10169
    },
    {
      "epoch": 0.36420935770945617,
      "grad_norm": 1.5147225856781006,
      "learning_rate": 0.00014691927478837987,
      "loss": 1.5474,
      "step": 10170
    },
    {
      "epoch": 0.3642451698390245,
      "grad_norm": 1.813011646270752,
      "learning_rate": 0.00014690903143335117,
      "loss": 1.454,
      "step": 10171
    },
    {
      "epoch": 0.36428098196859277,
      "grad_norm": 1.3621699810028076,
      "learning_rate": 0.00014689878744722192,
      "loss": 1.4512,
      "step": 10172
    },
    {
      "epoch": 0.36431679409816103,
      "grad_norm": 1.4398069381713867,
      "learning_rate": 0.00014688854283013001,
      "loss": 1.4105,
      "step": 10173
    },
    {
      "epoch": 0.36435260622772936,
      "grad_norm": 1.4563603401184082,
      "learning_rate": 0.0001468782975822132,
      "loss": 1.3174,
      "step": 10174
    },
    {
      "epoch": 0.3643884183572976,
      "grad_norm": 1.7026748657226562,
      "learning_rate": 0.0001468680517036094,
      "loss": 1.642,
      "step": 10175
    },
    {
      "epoch": 0.3644242304868659,
      "grad_norm": 1.738409399986267,
      "learning_rate": 0.0001468578051944564,
      "loss": 1.4977,
      "step": 10176
    },
    {
      "epoch": 0.36446004261643417,
      "grad_norm": 2.2785589694976807,
      "learning_rate": 0.00014684755805489206,
      "loss": 1.5793,
      "step": 10177
    },
    {
      "epoch": 0.3644958547460025,
      "grad_norm": 1.4905781745910645,
      "learning_rate": 0.0001468373102850543,
      "loss": 1.4412,
      "step": 10178
    },
    {
      "epoch": 0.36453166687557076,
      "grad_norm": 1.4216437339782715,
      "learning_rate": 0.0001468270618850809,
      "loss": 1.7233,
      "step": 10179
    },
    {
      "epoch": 0.36456747900513903,
      "grad_norm": 2.1683859825134277,
      "learning_rate": 0.0001468168128551098,
      "loss": 1.7547,
      "step": 10180
    },
    {
      "epoch": 0.36460329113470735,
      "grad_norm": 2.0887510776519775,
      "learning_rate": 0.00014680656319527886,
      "loss": 1.317,
      "step": 10181
    },
    {
      "epoch": 0.3646391032642756,
      "grad_norm": 1.405361533164978,
      "learning_rate": 0.00014679631290572602,
      "loss": 1.423,
      "step": 10182
    },
    {
      "epoch": 0.3646749153938439,
      "grad_norm": 1.915013313293457,
      "learning_rate": 0.00014678606198658916,
      "loss": 1.3918,
      "step": 10183
    },
    {
      "epoch": 0.36471072752341216,
      "grad_norm": 1.6929726600646973,
      "learning_rate": 0.00014677581043800615,
      "loss": 1.5082,
      "step": 10184
    },
    {
      "epoch": 0.3647465396529805,
      "grad_norm": 1.6973090171813965,
      "learning_rate": 0.00014676555826011496,
      "loss": 1.2838,
      "step": 10185
    },
    {
      "epoch": 0.36478235178254875,
      "grad_norm": 1.729513168334961,
      "learning_rate": 0.0001467553054530535,
      "loss": 1.9042,
      "step": 10186
    },
    {
      "epoch": 0.364818163912117,
      "grad_norm": 1.4617276191711426,
      "learning_rate": 0.00014674505201695971,
      "loss": 1.4484,
      "step": 10187
    },
    {
      "epoch": 0.36485397604168535,
      "grad_norm": 1.501288890838623,
      "learning_rate": 0.00014673479795197154,
      "loss": 1.1133,
      "step": 10188
    },
    {
      "epoch": 0.3648897881712536,
      "grad_norm": 1.6365547180175781,
      "learning_rate": 0.00014672454325822696,
      "loss": 1.5686,
      "step": 10189
    },
    {
      "epoch": 0.3649256003008219,
      "grad_norm": 1.5337694883346558,
      "learning_rate": 0.00014671428793586392,
      "loss": 1.3405,
      "step": 10190
    },
    {
      "epoch": 0.36496141243039015,
      "grad_norm": 3.3464503288269043,
      "learning_rate": 0.0001467040319850204,
      "loss": 1.5726,
      "step": 10191
    },
    {
      "epoch": 0.3649972245599585,
      "grad_norm": 1.6443172693252563,
      "learning_rate": 0.0001466937754058344,
      "loss": 1.6067,
      "step": 10192
    },
    {
      "epoch": 0.36503303668952675,
      "grad_norm": 1.2222121953964233,
      "learning_rate": 0.00014668351819844384,
      "loss": 1.3071,
      "step": 10193
    },
    {
      "epoch": 0.365068848819095,
      "grad_norm": 1.9840507507324219,
      "learning_rate": 0.00014667326036298675,
      "loss": 1.7483,
      "step": 10194
    },
    {
      "epoch": 0.36510466094866334,
      "grad_norm": 1.4311078786849976,
      "learning_rate": 0.00014666300189960116,
      "loss": 1.1324,
      "step": 10195
    },
    {
      "epoch": 0.3651404730782316,
      "grad_norm": 1.5631734132766724,
      "learning_rate": 0.00014665274280842508,
      "loss": 1.4826,
      "step": 10196
    },
    {
      "epoch": 0.3651762852077999,
      "grad_norm": 1.5126618146896362,
      "learning_rate": 0.0001466424830895965,
      "loss": 1.4611,
      "step": 10197
    },
    {
      "epoch": 0.36521209733736815,
      "grad_norm": 1.7029465436935425,
      "learning_rate": 0.00014663222274325353,
      "loss": 1.2322,
      "step": 10198
    },
    {
      "epoch": 0.36524790946693647,
      "grad_norm": 1.7353816032409668,
      "learning_rate": 0.0001466219617695341,
      "loss": 1.2907,
      "step": 10199
    },
    {
      "epoch": 0.36528372159650474,
      "grad_norm": 1.881523847579956,
      "learning_rate": 0.00014661170016857633,
      "loss": 1.6683,
      "step": 10200
    },
    {
      "epoch": 0.365319533726073,
      "grad_norm": 1.5477781295776367,
      "learning_rate": 0.00014660143794051827,
      "loss": 1.7577,
      "step": 10201
    },
    {
      "epoch": 0.3653553458556413,
      "grad_norm": 2.418309211730957,
      "learning_rate": 0.00014659117508549797,
      "loss": 1.4513,
      "step": 10202
    },
    {
      "epoch": 0.3653911579852096,
      "grad_norm": 1.7008320093154907,
      "learning_rate": 0.0001465809116036535,
      "loss": 1.6526,
      "step": 10203
    },
    {
      "epoch": 0.36542697011477787,
      "grad_norm": 1.8753713369369507,
      "learning_rate": 0.00014657064749512295,
      "loss": 1.8226,
      "step": 10204
    },
    {
      "epoch": 0.36546278224434614,
      "grad_norm": 1.8317325115203857,
      "learning_rate": 0.0001465603827600444,
      "loss": 1.5395,
      "step": 10205
    },
    {
      "epoch": 0.36549859437391446,
      "grad_norm": 1.8578588962554932,
      "learning_rate": 0.00014655011739855595,
      "loss": 1.5256,
      "step": 10206
    },
    {
      "epoch": 0.36553440650348273,
      "grad_norm": 1.3821361064910889,
      "learning_rate": 0.00014653985141079576,
      "loss": 1.4077,
      "step": 10207
    },
    {
      "epoch": 0.365570218633051,
      "grad_norm": 1.7334470748901367,
      "learning_rate": 0.00014652958479690185,
      "loss": 1.2891,
      "step": 10208
    },
    {
      "epoch": 0.36560603076261927,
      "grad_norm": 1.7649697065353394,
      "learning_rate": 0.00014651931755701246,
      "loss": 1.4587,
      "step": 10209
    },
    {
      "epoch": 0.3656418428921876,
      "grad_norm": 1.560272455215454,
      "learning_rate": 0.0001465090496912656,
      "loss": 1.2729,
      "step": 10210
    },
    {
      "epoch": 0.36567765502175587,
      "grad_norm": 2.465453624725342,
      "learning_rate": 0.0001464987811997995,
      "loss": 1.7935,
      "step": 10211
    },
    {
      "epoch": 0.36571346715132413,
      "grad_norm": 1.8573144674301147,
      "learning_rate": 0.00014648851208275224,
      "loss": 1.7883,
      "step": 10212
    },
    {
      "epoch": 0.36574927928089246,
      "grad_norm": 1.308641791343689,
      "learning_rate": 0.00014647824234026205,
      "loss": 1.2575,
      "step": 10213
    },
    {
      "epoch": 0.3657850914104607,
      "grad_norm": 2.2015023231506348,
      "learning_rate": 0.00014646797197246706,
      "loss": 1.302,
      "step": 10214
    },
    {
      "epoch": 0.365820903540029,
      "grad_norm": 1.6897261142730713,
      "learning_rate": 0.00014645770097950544,
      "loss": 1.527,
      "step": 10215
    },
    {
      "epoch": 0.36585671566959727,
      "grad_norm": 1.403554916381836,
      "learning_rate": 0.00014644742936151538,
      "loss": 1.525,
      "step": 10216
    },
    {
      "epoch": 0.3658925277991656,
      "grad_norm": 1.6537847518920898,
      "learning_rate": 0.00014643715711863507,
      "loss": 1.566,
      "step": 10217
    },
    {
      "epoch": 0.36592833992873386,
      "grad_norm": 1.7466622591018677,
      "learning_rate": 0.00014642688425100273,
      "loss": 1.8627,
      "step": 10218
    },
    {
      "epoch": 0.36596415205830213,
      "grad_norm": 1.805519700050354,
      "learning_rate": 0.00014641661075875652,
      "loss": 1.373,
      "step": 10219
    },
    {
      "epoch": 0.36599996418787045,
      "grad_norm": 1.6405415534973145,
      "learning_rate": 0.0001464063366420347,
      "loss": 1.498,
      "step": 10220
    },
    {
      "epoch": 0.3660357763174387,
      "grad_norm": 1.6992943286895752,
      "learning_rate": 0.00014639606190097547,
      "loss": 1.4758,
      "step": 10221
    },
    {
      "epoch": 0.366071588447007,
      "grad_norm": 1.6012146472930908,
      "learning_rate": 0.00014638578653571708,
      "loss": 1.5874,
      "step": 10222
    },
    {
      "epoch": 0.36610740057657526,
      "grad_norm": 1.8396260738372803,
      "learning_rate": 0.00014637551054639774,
      "loss": 1.5488,
      "step": 10223
    },
    {
      "epoch": 0.3661432127061436,
      "grad_norm": 2.8932039737701416,
      "learning_rate": 0.00014636523393315578,
      "loss": 1.3023,
      "step": 10224
    },
    {
      "epoch": 0.36617902483571185,
      "grad_norm": 1.3132829666137695,
      "learning_rate": 0.00014635495669612934,
      "loss": 1.5124,
      "step": 10225
    },
    {
      "epoch": 0.3662148369652801,
      "grad_norm": 1.9502285718917847,
      "learning_rate": 0.0001463446788354568,
      "loss": 1.6832,
      "step": 10226
    },
    {
      "epoch": 0.36625064909484845,
      "grad_norm": 1.5486946105957031,
      "learning_rate": 0.00014633440035127638,
      "loss": 1.5737,
      "step": 10227
    },
    {
      "epoch": 0.3662864612244167,
      "grad_norm": 1.6194562911987305,
      "learning_rate": 0.00014632412124372635,
      "loss": 1.3977,
      "step": 10228
    },
    {
      "epoch": 0.366322273353985,
      "grad_norm": 1.5558431148529053,
      "learning_rate": 0.00014631384151294507,
      "loss": 1.8146,
      "step": 10229
    },
    {
      "epoch": 0.36635808548355325,
      "grad_norm": 1.5941044092178345,
      "learning_rate": 0.00014630356115907073,
      "loss": 1.6239,
      "step": 10230
    },
    {
      "epoch": 0.3663938976131216,
      "grad_norm": 1.5189129114151,
      "learning_rate": 0.00014629328018224175,
      "loss": 1.4244,
      "step": 10231
    },
    {
      "epoch": 0.36642970974268985,
      "grad_norm": 1.5772870779037476,
      "learning_rate": 0.00014628299858259638,
      "loss": 1.4397,
      "step": 10232
    },
    {
      "epoch": 0.3664655218722581,
      "grad_norm": 1.4384115934371948,
      "learning_rate": 0.00014627271636027297,
      "loss": 1.5807,
      "step": 10233
    },
    {
      "epoch": 0.36650133400182644,
      "grad_norm": 1.3554954528808594,
      "learning_rate": 0.00014626243351540983,
      "loss": 1.106,
      "step": 10234
    },
    {
      "epoch": 0.3665371461313947,
      "grad_norm": 1.87827730178833,
      "learning_rate": 0.00014625215004814533,
      "loss": 1.7206,
      "step": 10235
    },
    {
      "epoch": 0.366572958260963,
      "grad_norm": 2.0337166786193848,
      "learning_rate": 0.00014624186595861785,
      "loss": 1.4494,
      "step": 10236
    },
    {
      "epoch": 0.36660877039053125,
      "grad_norm": 1.458093523979187,
      "learning_rate": 0.00014623158124696565,
      "loss": 1.6958,
      "step": 10237
    },
    {
      "epoch": 0.36664458252009957,
      "grad_norm": 1.5604604482650757,
      "learning_rate": 0.00014622129591332722,
      "loss": 1.6612,
      "step": 10238
    },
    {
      "epoch": 0.36668039464966784,
      "grad_norm": 1.7306214570999146,
      "learning_rate": 0.0001462110099578408,
      "loss": 1.6922,
      "step": 10239
    },
    {
      "epoch": 0.3667162067792361,
      "grad_norm": 1.6424694061279297,
      "learning_rate": 0.0001462007233806449,
      "loss": 1.4857,
      "step": 10240
    },
    {
      "epoch": 0.36675201890880443,
      "grad_norm": 1.6787950992584229,
      "learning_rate": 0.00014619043618187784,
      "loss": 1.0656,
      "step": 10241
    },
    {
      "epoch": 0.3667878310383727,
      "grad_norm": 2.0720412731170654,
      "learning_rate": 0.00014618014836167807,
      "loss": 1.6523,
      "step": 10242
    },
    {
      "epoch": 0.36682364316794097,
      "grad_norm": 1.870436191558838,
      "learning_rate": 0.00014616985992018394,
      "loss": 1.7099,
      "step": 10243
    },
    {
      "epoch": 0.36685945529750924,
      "grad_norm": 1.4863344430923462,
      "learning_rate": 0.00014615957085753394,
      "loss": 1.5977,
      "step": 10244
    },
    {
      "epoch": 0.36689526742707756,
      "grad_norm": 1.7124794721603394,
      "learning_rate": 0.00014614928117386643,
      "loss": 1.3756,
      "step": 10245
    },
    {
      "epoch": 0.36693107955664583,
      "grad_norm": 4.074779987335205,
      "learning_rate": 0.0001461389908693199,
      "loss": 1.6485,
      "step": 10246
    },
    {
      "epoch": 0.3669668916862141,
      "grad_norm": 1.4112684726715088,
      "learning_rate": 0.0001461286999440327,
      "loss": 1.0859,
      "step": 10247
    },
    {
      "epoch": 0.3670027038157824,
      "grad_norm": 1.4489010572433472,
      "learning_rate": 0.00014611840839814336,
      "loss": 1.3586,
      "step": 10248
    },
    {
      "epoch": 0.3670385159453507,
      "grad_norm": 1.4673880338668823,
      "learning_rate": 0.00014610811623179038,
      "loss": 1.3926,
      "step": 10249
    },
    {
      "epoch": 0.36707432807491897,
      "grad_norm": 1.5906109809875488,
      "learning_rate": 0.00014609782344511213,
      "loss": 1.2522,
      "step": 10250
    },
    {
      "epoch": 0.36711014020448723,
      "grad_norm": 1.204659342765808,
      "learning_rate": 0.0001460875300382471,
      "loss": 1.2822,
      "step": 10251
    },
    {
      "epoch": 0.36714595233405556,
      "grad_norm": 1.5988489389419556,
      "learning_rate": 0.00014607723601133384,
      "loss": 1.4988,
      "step": 10252
    },
    {
      "epoch": 0.3671817644636238,
      "grad_norm": 1.5991809368133545,
      "learning_rate": 0.00014606694136451082,
      "loss": 1.7249,
      "step": 10253
    },
    {
      "epoch": 0.3672175765931921,
      "grad_norm": 1.5101706981658936,
      "learning_rate": 0.0001460566460979165,
      "loss": 1.5289,
      "step": 10254
    },
    {
      "epoch": 0.3672533887227604,
      "grad_norm": 1.951116681098938,
      "learning_rate": 0.00014604635021168942,
      "loss": 1.5754,
      "step": 10255
    },
    {
      "epoch": 0.3672892008523287,
      "grad_norm": 2.0610105991363525,
      "learning_rate": 0.00014603605370596808,
      "loss": 1.1101,
      "step": 10256
    },
    {
      "epoch": 0.36732501298189696,
      "grad_norm": 1.4582083225250244,
      "learning_rate": 0.00014602575658089105,
      "loss": 1.5282,
      "step": 10257
    },
    {
      "epoch": 0.36736082511146523,
      "grad_norm": 1.7049942016601562,
      "learning_rate": 0.0001460154588365968,
      "loss": 1.4164,
      "step": 10258
    },
    {
      "epoch": 0.36739663724103355,
      "grad_norm": 1.7061821222305298,
      "learning_rate": 0.00014600516047322392,
      "loss": 1.6465,
      "step": 10259
    },
    {
      "epoch": 0.3674324493706018,
      "grad_norm": 2.0501723289489746,
      "learning_rate": 0.00014599486149091096,
      "loss": 1.4027,
      "step": 10260
    },
    {
      "epoch": 0.3674682615001701,
      "grad_norm": 1.5676246881484985,
      "learning_rate": 0.00014598456188979643,
      "loss": 1.7521,
      "step": 10261
    },
    {
      "epoch": 0.3675040736297384,
      "grad_norm": 1.6521248817443848,
      "learning_rate": 0.000145974261670019,
      "loss": 1.4259,
      "step": 10262
    },
    {
      "epoch": 0.3675398857593067,
      "grad_norm": 1.571914792060852,
      "learning_rate": 0.00014596396083171715,
      "loss": 1.7523,
      "step": 10263
    },
    {
      "epoch": 0.36757569788887495,
      "grad_norm": 1.5955133438110352,
      "learning_rate": 0.0001459536593750295,
      "loss": 1.3922,
      "step": 10264
    },
    {
      "epoch": 0.3676115100184432,
      "grad_norm": 1.9849090576171875,
      "learning_rate": 0.00014594335730009462,
      "loss": 1.5897,
      "step": 10265
    },
    {
      "epoch": 0.36764732214801155,
      "grad_norm": 1.507638692855835,
      "learning_rate": 0.00014593305460705114,
      "loss": 1.5146,
      "step": 10266
    },
    {
      "epoch": 0.3676831342775798,
      "grad_norm": 1.4899927377700806,
      "learning_rate": 0.00014592275129603766,
      "loss": 1.5611,
      "step": 10267
    },
    {
      "epoch": 0.3677189464071481,
      "grad_norm": 1.6360082626342773,
      "learning_rate": 0.00014591244736719282,
      "loss": 1.5844,
      "step": 10268
    },
    {
      "epoch": 0.3677547585367164,
      "grad_norm": 1.3872517347335815,
      "learning_rate": 0.00014590214282065518,
      "loss": 1.5452,
      "step": 10269
    },
    {
      "epoch": 0.3677905706662847,
      "grad_norm": 1.434220790863037,
      "learning_rate": 0.00014589183765656343,
      "loss": 1.3571,
      "step": 10270
    },
    {
      "epoch": 0.36782638279585295,
      "grad_norm": 1.7095938920974731,
      "learning_rate": 0.00014588153187505625,
      "loss": 1.4374,
      "step": 10271
    },
    {
      "epoch": 0.3678621949254212,
      "grad_norm": 1.4860492944717407,
      "learning_rate": 0.00014587122547627217,
      "loss": 1.5456,
      "step": 10272
    },
    {
      "epoch": 0.36789800705498954,
      "grad_norm": 1.56809663772583,
      "learning_rate": 0.00014586091846034997,
      "loss": 1.343,
      "step": 10273
    },
    {
      "epoch": 0.3679338191845578,
      "grad_norm": 1.3190221786499023,
      "learning_rate": 0.00014585061082742824,
      "loss": 1.5654,
      "step": 10274
    },
    {
      "epoch": 0.3679696313141261,
      "grad_norm": 1.5554225444793701,
      "learning_rate": 0.0001458403025776457,
      "loss": 1.6568,
      "step": 10275
    },
    {
      "epoch": 0.3680054434436944,
      "grad_norm": 1.8499597311019897,
      "learning_rate": 0.000145829993711141,
      "loss": 1.606,
      "step": 10276
    },
    {
      "epoch": 0.36804125557326267,
      "grad_norm": 1.9961286783218384,
      "learning_rate": 0.00014581968422805287,
      "loss": 1.6618,
      "step": 10277
    },
    {
      "epoch": 0.36807706770283094,
      "grad_norm": 1.6830213069915771,
      "learning_rate": 0.00014580937412852,
      "loss": 1.4959,
      "step": 10278
    },
    {
      "epoch": 0.3681128798323992,
      "grad_norm": 1.537229299545288,
      "learning_rate": 0.0001457990634126811,
      "loss": 1.5156,
      "step": 10279
    },
    {
      "epoch": 0.36814869196196753,
      "grad_norm": 1.7691799402236938,
      "learning_rate": 0.00014578875208067483,
      "loss": 1.2807,
      "step": 10280
    },
    {
      "epoch": 0.3681845040915358,
      "grad_norm": 1.5051552057266235,
      "learning_rate": 0.00014577844013264,
      "loss": 1.5256,
      "step": 10281
    },
    {
      "epoch": 0.36822031622110407,
      "grad_norm": 1.39759361743927,
      "learning_rate": 0.0001457681275687153,
      "loss": 1.6506,
      "step": 10282
    },
    {
      "epoch": 0.3682561283506724,
      "grad_norm": 1.8484997749328613,
      "learning_rate": 0.00014575781438903946,
      "loss": 1.5992,
      "step": 10283
    },
    {
      "epoch": 0.36829194048024066,
      "grad_norm": 1.4711723327636719,
      "learning_rate": 0.0001457475005937513,
      "loss": 1.5188,
      "step": 10284
    },
    {
      "epoch": 0.36832775260980893,
      "grad_norm": 2.3037071228027344,
      "learning_rate": 0.0001457371861829895,
      "loss": 1.5384,
      "step": 10285
    },
    {
      "epoch": 0.3683635647393772,
      "grad_norm": 1.4197289943695068,
      "learning_rate": 0.00014572687115689282,
      "loss": 1.4293,
      "step": 10286
    },
    {
      "epoch": 0.3683993768689455,
      "grad_norm": 1.5581529140472412,
      "learning_rate": 0.0001457165555156001,
      "loss": 1.2188,
      "step": 10287
    },
    {
      "epoch": 0.3684351889985138,
      "grad_norm": 1.9231946468353271,
      "learning_rate": 0.00014570623925925014,
      "loss": 1.1728,
      "step": 10288
    },
    {
      "epoch": 0.36847100112808207,
      "grad_norm": 1.9287388324737549,
      "learning_rate": 0.00014569592238798163,
      "loss": 1.4546,
      "step": 10289
    },
    {
      "epoch": 0.3685068132576504,
      "grad_norm": 2.1329243183135986,
      "learning_rate": 0.00014568560490193345,
      "loss": 1.4835,
      "step": 10290
    },
    {
      "epoch": 0.36854262538721866,
      "grad_norm": 1.5576348304748535,
      "learning_rate": 0.0001456752868012444,
      "loss": 1.4383,
      "step": 10291
    },
    {
      "epoch": 0.3685784375167869,
      "grad_norm": 1.8817024230957031,
      "learning_rate": 0.00014566496808605326,
      "loss": 1.4858,
      "step": 10292
    },
    {
      "epoch": 0.3686142496463552,
      "grad_norm": 1.5834522247314453,
      "learning_rate": 0.00014565464875649888,
      "loss": 1.5599,
      "step": 10293
    },
    {
      "epoch": 0.3686500617759235,
      "grad_norm": 1.5673562288284302,
      "learning_rate": 0.0001456443288127201,
      "loss": 1.4004,
      "step": 10294
    },
    {
      "epoch": 0.3686858739054918,
      "grad_norm": 1.6744060516357422,
      "learning_rate": 0.00014563400825485576,
      "loss": 1.5878,
      "step": 10295
    },
    {
      "epoch": 0.36872168603506006,
      "grad_norm": 1.609439492225647,
      "learning_rate": 0.00014562368708304467,
      "loss": 1.4532,
      "step": 10296
    },
    {
      "epoch": 0.3687574981646284,
      "grad_norm": 2.1061041355133057,
      "learning_rate": 0.00014561336529742575,
      "loss": 1.5246,
      "step": 10297
    },
    {
      "epoch": 0.36879331029419665,
      "grad_norm": 1.7930060625076294,
      "learning_rate": 0.00014560304289813785,
      "loss": 1.3858,
      "step": 10298
    },
    {
      "epoch": 0.3688291224237649,
      "grad_norm": 1.5626553297042847,
      "learning_rate": 0.0001455927198853198,
      "loss": 1.5151,
      "step": 10299
    },
    {
      "epoch": 0.3688649345533332,
      "grad_norm": 2.6362035274505615,
      "learning_rate": 0.00014558239625911052,
      "loss": 1.9309,
      "step": 10300
    },
    {
      "epoch": 0.3689007466829015,
      "grad_norm": 1.53107488155365,
      "learning_rate": 0.00014557207201964893,
      "loss": 1.4419,
      "step": 10301
    },
    {
      "epoch": 0.3689365588124698,
      "grad_norm": 2.428128957748413,
      "learning_rate": 0.00014556174716707384,
      "loss": 1.6143,
      "step": 10302
    },
    {
      "epoch": 0.36897237094203805,
      "grad_norm": 1.9132128953933716,
      "learning_rate": 0.00014555142170152423,
      "loss": 1.6772,
      "step": 10303
    },
    {
      "epoch": 0.3690081830716064,
      "grad_norm": 1.4515597820281982,
      "learning_rate": 0.00014554109562313903,
      "loss": 1.3292,
      "step": 10304
    },
    {
      "epoch": 0.36904399520117465,
      "grad_norm": 2.2293195724487305,
      "learning_rate": 0.00014553076893205708,
      "loss": 1.4524,
      "step": 10305
    },
    {
      "epoch": 0.3690798073307429,
      "grad_norm": 1.7687851190567017,
      "learning_rate": 0.00014552044162841743,
      "loss": 1.2241,
      "step": 10306
    },
    {
      "epoch": 0.3691156194603112,
      "grad_norm": 1.509236454963684,
      "learning_rate": 0.0001455101137123589,
      "loss": 1.6139,
      "step": 10307
    },
    {
      "epoch": 0.3691514315898795,
      "grad_norm": 1.5203906297683716,
      "learning_rate": 0.00014549978518402053,
      "loss": 1.646,
      "step": 10308
    },
    {
      "epoch": 0.3691872437194478,
      "grad_norm": 2.03283429145813,
      "learning_rate": 0.0001454894560435412,
      "loss": 1.3515,
      "step": 10309
    },
    {
      "epoch": 0.36922305584901605,
      "grad_norm": 1.2739007472991943,
      "learning_rate": 0.00014547912629105995,
      "loss": 1.3494,
      "step": 10310
    },
    {
      "epoch": 0.36925886797858437,
      "grad_norm": 1.7174979448318481,
      "learning_rate": 0.00014546879592671573,
      "loss": 1.5571,
      "step": 10311
    },
    {
      "epoch": 0.36929468010815264,
      "grad_norm": 1.639888048171997,
      "learning_rate": 0.00014545846495064748,
      "loss": 1.6965,
      "step": 10312
    },
    {
      "epoch": 0.3693304922377209,
      "grad_norm": 1.811215877532959,
      "learning_rate": 0.0001454481333629942,
      "loss": 1.5453,
      "step": 10313
    },
    {
      "epoch": 0.3693663043672892,
      "grad_norm": 2.070788621902466,
      "learning_rate": 0.00014543780116389496,
      "loss": 1.5838,
      "step": 10314
    },
    {
      "epoch": 0.3694021164968575,
      "grad_norm": 1.934891700744629,
      "learning_rate": 0.0001454274683534887,
      "loss": 1.2206,
      "step": 10315
    },
    {
      "epoch": 0.36943792862642577,
      "grad_norm": 2.129443645477295,
      "learning_rate": 0.00014541713493191444,
      "loss": 1.5475,
      "step": 10316
    },
    {
      "epoch": 0.36947374075599404,
      "grad_norm": 1.6835366487503052,
      "learning_rate": 0.00014540680089931125,
      "loss": 1.4861,
      "step": 10317
    },
    {
      "epoch": 0.36950955288556236,
      "grad_norm": 1.4131048917770386,
      "learning_rate": 0.00014539646625581805,
      "loss": 1.1092,
      "step": 10318
    },
    {
      "epoch": 0.36954536501513063,
      "grad_norm": 2.283271551132202,
      "learning_rate": 0.00014538613100157404,
      "loss": 1.4867,
      "step": 10319
    },
    {
      "epoch": 0.3695811771446989,
      "grad_norm": 1.9396806955337524,
      "learning_rate": 0.0001453757951367181,
      "loss": 1.6176,
      "step": 10320
    },
    {
      "epoch": 0.36961698927426717,
      "grad_norm": 1.417442798614502,
      "learning_rate": 0.00014536545866138941,
      "loss": 1.3439,
      "step": 10321
    },
    {
      "epoch": 0.3696528014038355,
      "grad_norm": 1.7332024574279785,
      "learning_rate": 0.000145355121575727,
      "loss": 1.4305,
      "step": 10322
    },
    {
      "epoch": 0.36968861353340376,
      "grad_norm": 1.3164316415786743,
      "learning_rate": 0.00014534478387986992,
      "loss": 1.3904,
      "step": 10323
    },
    {
      "epoch": 0.36972442566297203,
      "grad_norm": 2.070188045501709,
      "learning_rate": 0.0001453344455739573,
      "loss": 1.4226,
      "step": 10324
    },
    {
      "epoch": 0.36976023779254036,
      "grad_norm": 1.8793950080871582,
      "learning_rate": 0.0001453241066581281,
      "loss": 1.3511,
      "step": 10325
    },
    {
      "epoch": 0.3697960499221086,
      "grad_norm": 1.7022724151611328,
      "learning_rate": 0.0001453137671325216,
      "loss": 1.4012,
      "step": 10326
    },
    {
      "epoch": 0.3698318620516769,
      "grad_norm": 1.8162293434143066,
      "learning_rate": 0.00014530342699727676,
      "loss": 1.2831,
      "step": 10327
    },
    {
      "epoch": 0.36986767418124517,
      "grad_norm": 1.9489846229553223,
      "learning_rate": 0.0001452930862525328,
      "loss": 1.5896,
      "step": 10328
    },
    {
      "epoch": 0.3699034863108135,
      "grad_norm": 1.9331752061843872,
      "learning_rate": 0.00014528274489842872,
      "loss": 1.3076,
      "step": 10329
    },
    {
      "epoch": 0.36993929844038176,
      "grad_norm": 1.6232823133468628,
      "learning_rate": 0.00014527240293510377,
      "loss": 1.3819,
      "step": 10330
    },
    {
      "epoch": 0.36997511056995,
      "grad_norm": 2.704481840133667,
      "learning_rate": 0.000145262060362697,
      "loss": 1.7733,
      "step": 10331
    },
    {
      "epoch": 0.37001092269951835,
      "grad_norm": 1.7846752405166626,
      "learning_rate": 0.00014525171718134762,
      "loss": 1.5951,
      "step": 10332
    },
    {
      "epoch": 0.3700467348290866,
      "grad_norm": 1.3831145763397217,
      "learning_rate": 0.00014524137339119478,
      "loss": 1.5853,
      "step": 10333
    },
    {
      "epoch": 0.3700825469586549,
      "grad_norm": 2.4267773628234863,
      "learning_rate": 0.00014523102899237754,
      "loss": 1.3878,
      "step": 10334
    },
    {
      "epoch": 0.37011835908822316,
      "grad_norm": 1.7978699207305908,
      "learning_rate": 0.00014522068398503522,
      "loss": 1.6433,
      "step": 10335
    },
    {
      "epoch": 0.3701541712177915,
      "grad_norm": 1.7147051095962524,
      "learning_rate": 0.00014521033836930689,
      "loss": 1.0625,
      "step": 10336
    },
    {
      "epoch": 0.37018998334735975,
      "grad_norm": 2.743452310562134,
      "learning_rate": 0.0001451999921453318,
      "loss": 1.5351,
      "step": 10337
    },
    {
      "epoch": 0.370225795476928,
      "grad_norm": 1.5962437391281128,
      "learning_rate": 0.00014518964531324907,
      "loss": 1.4559,
      "step": 10338
    },
    {
      "epoch": 0.37026160760649635,
      "grad_norm": 1.8143364191055298,
      "learning_rate": 0.000145179297873198,
      "loss": 1.9998,
      "step": 10339
    },
    {
      "epoch": 0.3702974197360646,
      "grad_norm": 1.644858956336975,
      "learning_rate": 0.00014516894982531775,
      "loss": 1.4414,
      "step": 10340
    },
    {
      "epoch": 0.3703332318656329,
      "grad_norm": 1.4432719945907593,
      "learning_rate": 0.00014515860116974752,
      "loss": 1.5982,
      "step": 10341
    },
    {
      "epoch": 0.37036904399520115,
      "grad_norm": 1.5924537181854248,
      "learning_rate": 0.0001451482519066266,
      "loss": 1.3127,
      "step": 10342
    },
    {
      "epoch": 0.3704048561247695,
      "grad_norm": 1.6284945011138916,
      "learning_rate": 0.00014513790203609416,
      "loss": 1.2715,
      "step": 10343
    },
    {
      "epoch": 0.37044066825433775,
      "grad_norm": 1.8258150815963745,
      "learning_rate": 0.0001451275515582895,
      "loss": 1.7172,
      "step": 10344
    },
    {
      "epoch": 0.370476480383906,
      "grad_norm": 1.3459279537200928,
      "learning_rate": 0.0001451172004733518,
      "loss": 1.6076,
      "step": 10345
    },
    {
      "epoch": 0.37051229251347434,
      "grad_norm": 1.891798496246338,
      "learning_rate": 0.00014510684878142038,
      "loss": 1.7049,
      "step": 10346
    },
    {
      "epoch": 0.3705481046430426,
      "grad_norm": 1.2938027381896973,
      "learning_rate": 0.00014509649648263449,
      "loss": 1.6065,
      "step": 10347
    },
    {
      "epoch": 0.3705839167726109,
      "grad_norm": 1.8871210813522339,
      "learning_rate": 0.00014508614357713342,
      "loss": 1.4965,
      "step": 10348
    },
    {
      "epoch": 0.37061972890217915,
      "grad_norm": 1.41502046585083,
      "learning_rate": 0.00014507579006505642,
      "loss": 1.4462,
      "step": 10349
    },
    {
      "epoch": 0.37065554103174747,
      "grad_norm": 1.7434496879577637,
      "learning_rate": 0.00014506543594654288,
      "loss": 1.2558,
      "step": 10350
    },
    {
      "epoch": 0.37069135316131574,
      "grad_norm": 1.9141662120819092,
      "learning_rate": 0.00014505508122173198,
      "loss": 1.607,
      "step": 10351
    },
    {
      "epoch": 0.370727165290884,
      "grad_norm": 1.4460514783859253,
      "learning_rate": 0.00014504472589076307,
      "loss": 1.8445,
      "step": 10352
    },
    {
      "epoch": 0.37076297742045233,
      "grad_norm": 1.3187059164047241,
      "learning_rate": 0.00014503436995377548,
      "loss": 1.6002,
      "step": 10353
    },
    {
      "epoch": 0.3707987895500206,
      "grad_norm": 1.4016839265823364,
      "learning_rate": 0.00014502401341090853,
      "loss": 1.3959,
      "step": 10354
    },
    {
      "epoch": 0.37083460167958887,
      "grad_norm": 1.599345088005066,
      "learning_rate": 0.00014501365626230157,
      "loss": 1.9364,
      "step": 10355
    },
    {
      "epoch": 0.37087041380915714,
      "grad_norm": 2.149724006652832,
      "learning_rate": 0.00014500329850809394,
      "loss": 1.5702,
      "step": 10356
    },
    {
      "epoch": 0.37090622593872546,
      "grad_norm": 1.2839821577072144,
      "learning_rate": 0.00014499294014842494,
      "loss": 1.1688,
      "step": 10357
    },
    {
      "epoch": 0.37094203806829373,
      "grad_norm": 1.5366618633270264,
      "learning_rate": 0.000144982581183434,
      "loss": 1.2508,
      "step": 10358
    },
    {
      "epoch": 0.370977850197862,
      "grad_norm": 1.925222635269165,
      "learning_rate": 0.00014497222161326045,
      "loss": 1.4266,
      "step": 10359
    },
    {
      "epoch": 0.3710136623274303,
      "grad_norm": 1.6245794296264648,
      "learning_rate": 0.00014496186143804366,
      "loss": 1.5413,
      "step": 10360
    },
    {
      "epoch": 0.3710494744569986,
      "grad_norm": 1.6511958837509155,
      "learning_rate": 0.000144951500657923,
      "loss": 1.4162,
      "step": 10361
    },
    {
      "epoch": 0.37108528658656686,
      "grad_norm": 1.8239529132843018,
      "learning_rate": 0.00014494113927303792,
      "loss": 1.4576,
      "step": 10362
    },
    {
      "epoch": 0.37112109871613513,
      "grad_norm": 2.1355695724487305,
      "learning_rate": 0.00014493077728352778,
      "loss": 1.6973,
      "step": 10363
    },
    {
      "epoch": 0.37115691084570346,
      "grad_norm": 1.5001815557479858,
      "learning_rate": 0.00014492041468953194,
      "loss": 1.207,
      "step": 10364
    },
    {
      "epoch": 0.3711927229752717,
      "grad_norm": 1.4016623497009277,
      "learning_rate": 0.0001449100514911899,
      "loss": 1.3825,
      "step": 10365
    },
    {
      "epoch": 0.37122853510484,
      "grad_norm": 2.0167899131774902,
      "learning_rate": 0.00014489968768864107,
      "loss": 1.6569,
      "step": 10366
    },
    {
      "epoch": 0.3712643472344083,
      "grad_norm": 1.745682716369629,
      "learning_rate": 0.00014488932328202484,
      "loss": 1.574,
      "step": 10367
    },
    {
      "epoch": 0.3713001593639766,
      "grad_norm": 1.8677984476089478,
      "learning_rate": 0.00014487895827148067,
      "loss": 1.8146,
      "step": 10368
    },
    {
      "epoch": 0.37133597149354486,
      "grad_norm": 2.221468687057495,
      "learning_rate": 0.00014486859265714798,
      "loss": 1.5426,
      "step": 10369
    },
    {
      "epoch": 0.3713717836231131,
      "grad_norm": 1.536044716835022,
      "learning_rate": 0.00014485822643916626,
      "loss": 1.2835,
      "step": 10370
    },
    {
      "epoch": 0.37140759575268145,
      "grad_norm": 1.6062583923339844,
      "learning_rate": 0.00014484785961767498,
      "loss": 1.5294,
      "step": 10371
    },
    {
      "epoch": 0.3714434078822497,
      "grad_norm": 1.858888030052185,
      "learning_rate": 0.0001448374921928136,
      "loss": 1.4026,
      "step": 10372
    },
    {
      "epoch": 0.371479220011818,
      "grad_norm": 1.7808476686477661,
      "learning_rate": 0.00014482712416472157,
      "loss": 1.829,
      "step": 10373
    },
    {
      "epoch": 0.3715150321413863,
      "grad_norm": 1.4914846420288086,
      "learning_rate": 0.00014481675553353843,
      "loss": 1.2286,
      "step": 10374
    },
    {
      "epoch": 0.3715508442709546,
      "grad_norm": 1.6184641122817993,
      "learning_rate": 0.00014480638629940366,
      "loss": 1.7472,
      "step": 10375
    },
    {
      "epoch": 0.37158665640052285,
      "grad_norm": 1.7213737964630127,
      "learning_rate": 0.00014479601646245676,
      "loss": 1.4159,
      "step": 10376
    },
    {
      "epoch": 0.3716224685300911,
      "grad_norm": 1.4510760307312012,
      "learning_rate": 0.00014478564602283725,
      "loss": 1.4266,
      "step": 10377
    },
    {
      "epoch": 0.37165828065965945,
      "grad_norm": 1.8478851318359375,
      "learning_rate": 0.0001447752749806846,
      "loss": 1.5385,
      "step": 10378
    },
    {
      "epoch": 0.3716940927892277,
      "grad_norm": 1.6792529821395874,
      "learning_rate": 0.00014476490333613842,
      "loss": 1.5736,
      "step": 10379
    },
    {
      "epoch": 0.371729904918796,
      "grad_norm": 1.8026201725006104,
      "learning_rate": 0.00014475453108933817,
      "loss": 1.7353,
      "step": 10380
    },
    {
      "epoch": 0.3717657170483643,
      "grad_norm": 1.645272135734558,
      "learning_rate": 0.00014474415824042346,
      "loss": 1.2897,
      "step": 10381
    },
    {
      "epoch": 0.3718015291779326,
      "grad_norm": 1.5104624032974243,
      "learning_rate": 0.0001447337847895338,
      "loss": 1.7482,
      "step": 10382
    },
    {
      "epoch": 0.37183734130750085,
      "grad_norm": 2.0950675010681152,
      "learning_rate": 0.00014472341073680883,
      "loss": 2.0707,
      "step": 10383
    },
    {
      "epoch": 0.3718731534370691,
      "grad_norm": 1.8873536586761475,
      "learning_rate": 0.00014471303608238798,
      "loss": 1.4105,
      "step": 10384
    },
    {
      "epoch": 0.37190896556663744,
      "grad_norm": 2.1229586601257324,
      "learning_rate": 0.00014470266082641095,
      "loss": 1.9201,
      "step": 10385
    },
    {
      "epoch": 0.3719447776962057,
      "grad_norm": 2.1284492015838623,
      "learning_rate": 0.00014469228496901727,
      "loss": 1.5545,
      "step": 10386
    },
    {
      "epoch": 0.371980589825774,
      "grad_norm": 1.978696584701538,
      "learning_rate": 0.00014468190851034656,
      "loss": 1.6813,
      "step": 10387
    },
    {
      "epoch": 0.3720164019553423,
      "grad_norm": 1.5713815689086914,
      "learning_rate": 0.0001446715314505384,
      "loss": 1.6417,
      "step": 10388
    },
    {
      "epoch": 0.37205221408491057,
      "grad_norm": 1.5084996223449707,
      "learning_rate": 0.00014466115378973236,
      "loss": 1.4994,
      "step": 10389
    },
    {
      "epoch": 0.37208802621447884,
      "grad_norm": 1.6603682041168213,
      "learning_rate": 0.00014465077552806813,
      "loss": 1.8155,
      "step": 10390
    },
    {
      "epoch": 0.3721238383440471,
      "grad_norm": 1.6536749601364136,
      "learning_rate": 0.00014464039666568532,
      "loss": 1.1957,
      "step": 10391
    },
    {
      "epoch": 0.37215965047361543,
      "grad_norm": 1.334580659866333,
      "learning_rate": 0.00014463001720272357,
      "loss": 1.4941,
      "step": 10392
    },
    {
      "epoch": 0.3721954626031837,
      "grad_norm": 1.4311217069625854,
      "learning_rate": 0.00014461963713932247,
      "loss": 1.3092,
      "step": 10393
    },
    {
      "epoch": 0.37223127473275197,
      "grad_norm": 1.9430100917816162,
      "learning_rate": 0.00014460925647562174,
      "loss": 1.2872,
      "step": 10394
    },
    {
      "epoch": 0.3722670868623203,
      "grad_norm": 1.8805584907531738,
      "learning_rate": 0.000144598875211761,
      "loss": 1.7865,
      "step": 10395
    },
    {
      "epoch": 0.37230289899188856,
      "grad_norm": 1.2648087739944458,
      "learning_rate": 0.00014458849334787993,
      "loss": 1.4374,
      "step": 10396
    },
    {
      "epoch": 0.37233871112145683,
      "grad_norm": 1.6092779636383057,
      "learning_rate": 0.00014457811088411816,
      "loss": 1.559,
      "step": 10397
    },
    {
      "epoch": 0.3723745232510251,
      "grad_norm": 1.3519525527954102,
      "learning_rate": 0.00014456772782061545,
      "loss": 1.2963,
      "step": 10398
    },
    {
      "epoch": 0.3724103353805934,
      "grad_norm": 2.0180888175964355,
      "learning_rate": 0.00014455734415751143,
      "loss": 1.574,
      "step": 10399
    },
    {
      "epoch": 0.3724461475101617,
      "grad_norm": 1.8862159252166748,
      "learning_rate": 0.00014454695989494582,
      "loss": 1.651,
      "step": 10400
    },
    {
      "epoch": 0.37248195963972996,
      "grad_norm": 1.7211228609085083,
      "learning_rate": 0.00014453657503305832,
      "loss": 1.3319,
      "step": 10401
    },
    {
      "epoch": 0.37251777176929823,
      "grad_norm": 2.443127155303955,
      "learning_rate": 0.00014452618957198866,
      "loss": 1.4359,
      "step": 10402
    },
    {
      "epoch": 0.37255358389886656,
      "grad_norm": 1.24772047996521,
      "learning_rate": 0.00014451580351187656,
      "loss": 1.6141,
      "step": 10403
    },
    {
      "epoch": 0.3725893960284348,
      "grad_norm": 1.9694045782089233,
      "learning_rate": 0.00014450541685286173,
      "loss": 1.6621,
      "step": 10404
    },
    {
      "epoch": 0.3726252081580031,
      "grad_norm": 1.6941626071929932,
      "learning_rate": 0.00014449502959508394,
      "loss": 1.6966,
      "step": 10405
    },
    {
      "epoch": 0.3726610202875714,
      "grad_norm": 1.5425902605056763,
      "learning_rate": 0.00014448464173868293,
      "loss": 1.4486,
      "step": 10406
    },
    {
      "epoch": 0.3726968324171397,
      "grad_norm": 1.4285595417022705,
      "learning_rate": 0.00014447425328379843,
      "loss": 1.0123,
      "step": 10407
    },
    {
      "epoch": 0.37273264454670796,
      "grad_norm": 1.7585430145263672,
      "learning_rate": 0.00014446386423057022,
      "loss": 1.2697,
      "step": 10408
    },
    {
      "epoch": 0.3727684566762762,
      "grad_norm": 2.169542074203491,
      "learning_rate": 0.00014445347457913807,
      "loss": 1.3233,
      "step": 10409
    },
    {
      "epoch": 0.37280426880584455,
      "grad_norm": 1.826337456703186,
      "learning_rate": 0.00014444308432964175,
      "loss": 1.2875,
      "step": 10410
    },
    {
      "epoch": 0.3728400809354128,
      "grad_norm": 2.5263783931732178,
      "learning_rate": 0.00014443269348222109,
      "loss": 1.7066,
      "step": 10411
    },
    {
      "epoch": 0.3728758930649811,
      "grad_norm": 1.9127196073532104,
      "learning_rate": 0.00014442230203701582,
      "loss": 1.4219,
      "step": 10412
    },
    {
      "epoch": 0.3729117051945494,
      "grad_norm": 1.6260210275650024,
      "learning_rate": 0.0001444119099941658,
      "loss": 1.5465,
      "step": 10413
    },
    {
      "epoch": 0.3729475173241177,
      "grad_norm": 1.7325187921524048,
      "learning_rate": 0.0001444015173538108,
      "loss": 1.5095,
      "step": 10414
    },
    {
      "epoch": 0.37298332945368595,
      "grad_norm": 1.852473258972168,
      "learning_rate": 0.00014439112411609065,
      "loss": 1.4091,
      "step": 10415
    },
    {
      "epoch": 0.3730191415832542,
      "grad_norm": 2.1807992458343506,
      "learning_rate": 0.00014438073028114523,
      "loss": 1.3722,
      "step": 10416
    },
    {
      "epoch": 0.37305495371282255,
      "grad_norm": 1.6915746927261353,
      "learning_rate": 0.00014437033584911428,
      "loss": 1.638,
      "step": 10417
    },
    {
      "epoch": 0.3730907658423908,
      "grad_norm": 2.2476296424865723,
      "learning_rate": 0.00014435994082013772,
      "loss": 1.4248,
      "step": 10418
    },
    {
      "epoch": 0.3731265779719591,
      "grad_norm": 1.4829312562942505,
      "learning_rate": 0.00014434954519435537,
      "loss": 1.599,
      "step": 10419
    },
    {
      "epoch": 0.3731623901015274,
      "grad_norm": 2.0151619911193848,
      "learning_rate": 0.0001443391489719071,
      "loss": 1.3809,
      "step": 10420
    },
    {
      "epoch": 0.3731982022310957,
      "grad_norm": 2.2441935539245605,
      "learning_rate": 0.0001443287521529328,
      "loss": 1.5466,
      "step": 10421
    },
    {
      "epoch": 0.37323401436066395,
      "grad_norm": 1.6749489307403564,
      "learning_rate": 0.00014431835473757227,
      "loss": 1.1317,
      "step": 10422
    },
    {
      "epoch": 0.3732698264902322,
      "grad_norm": 1.493473768234253,
      "learning_rate": 0.0001443079567259655,
      "loss": 1.5646,
      "step": 10423
    },
    {
      "epoch": 0.37330563861980054,
      "grad_norm": 2.4875717163085938,
      "learning_rate": 0.00014429755811825226,
      "loss": 1.505,
      "step": 10424
    },
    {
      "epoch": 0.3733414507493688,
      "grad_norm": 1.3609141111373901,
      "learning_rate": 0.00014428715891457255,
      "loss": 1.3917,
      "step": 10425
    },
    {
      "epoch": 0.3733772628789371,
      "grad_norm": 1.8194928169250488,
      "learning_rate": 0.00014427675911506623,
      "loss": 1.3385,
      "step": 10426
    },
    {
      "epoch": 0.3734130750085054,
      "grad_norm": 1.7996721267700195,
      "learning_rate": 0.00014426635871987327,
      "loss": 1.5479,
      "step": 10427
    },
    {
      "epoch": 0.37344888713807367,
      "grad_norm": 2.608707904815674,
      "learning_rate": 0.0001442559577291335,
      "loss": 1.3896,
      "step": 10428
    },
    {
      "epoch": 0.37348469926764194,
      "grad_norm": 1.6226502656936646,
      "learning_rate": 0.00014424555614298693,
      "loss": 1.4873,
      "step": 10429
    },
    {
      "epoch": 0.3735205113972102,
      "grad_norm": 1.5772932767868042,
      "learning_rate": 0.0001442351539615735,
      "loss": 1.2307,
      "step": 10430
    },
    {
      "epoch": 0.37355632352677853,
      "grad_norm": 1.3217772245407104,
      "learning_rate": 0.00014422475118503307,
      "loss": 1.756,
      "step": 10431
    },
    {
      "epoch": 0.3735921356563468,
      "grad_norm": 1.4471333026885986,
      "learning_rate": 0.0001442143478135057,
      "loss": 1.3107,
      "step": 10432
    },
    {
      "epoch": 0.37362794778591507,
      "grad_norm": 1.474479079246521,
      "learning_rate": 0.00014420394384713129,
      "loss": 1.5484,
      "step": 10433
    },
    {
      "epoch": 0.3736637599154834,
      "grad_norm": 1.8751987218856812,
      "learning_rate": 0.00014419353928604988,
      "loss": 1.4065,
      "step": 10434
    },
    {
      "epoch": 0.37369957204505166,
      "grad_norm": 2.282353639602661,
      "learning_rate": 0.00014418313413040138,
      "loss": 1.1441,
      "step": 10435
    },
    {
      "epoch": 0.37373538417461993,
      "grad_norm": 2.0015993118286133,
      "learning_rate": 0.00014417272838032578,
      "loss": 1.5908,
      "step": 10436
    },
    {
      "epoch": 0.3737711963041882,
      "grad_norm": 2.5349526405334473,
      "learning_rate": 0.00014416232203596312,
      "loss": 1.5447,
      "step": 10437
    },
    {
      "epoch": 0.3738070084337565,
      "grad_norm": 1.7229876518249512,
      "learning_rate": 0.00014415191509745338,
      "loss": 1.3676,
      "step": 10438
    },
    {
      "epoch": 0.3738428205633248,
      "grad_norm": 2.226806640625,
      "learning_rate": 0.0001441415075649366,
      "loss": 1.4815,
      "step": 10439
    },
    {
      "epoch": 0.37387863269289306,
      "grad_norm": 3.008434772491455,
      "learning_rate": 0.00014413109943855275,
      "loss": 1.2429,
      "step": 10440
    },
    {
      "epoch": 0.3739144448224614,
      "grad_norm": 1.4525420665740967,
      "learning_rate": 0.00014412069071844186,
      "loss": 1.4308,
      "step": 10441
    },
    {
      "epoch": 0.37395025695202966,
      "grad_norm": 1.8240820169448853,
      "learning_rate": 0.00014411028140474402,
      "loss": 1.6106,
      "step": 10442
    },
    {
      "epoch": 0.3739860690815979,
      "grad_norm": 1.7192800045013428,
      "learning_rate": 0.0001440998714975992,
      "loss": 1.6079,
      "step": 10443
    },
    {
      "epoch": 0.3740218812111662,
      "grad_norm": 1.9115632772445679,
      "learning_rate": 0.00014408946099714754,
      "loss": 1.6781,
      "step": 10444
    },
    {
      "epoch": 0.3740576933407345,
      "grad_norm": 1.8034567832946777,
      "learning_rate": 0.00014407904990352904,
      "loss": 1.4781,
      "step": 10445
    },
    {
      "epoch": 0.3740935054703028,
      "grad_norm": 1.9719936847686768,
      "learning_rate": 0.00014406863821688374,
      "loss": 1.7442,
      "step": 10446
    },
    {
      "epoch": 0.37412931759987106,
      "grad_norm": 1.7410471439361572,
      "learning_rate": 0.00014405822593735183,
      "loss": 1.6857,
      "step": 10447
    },
    {
      "epoch": 0.3741651297294394,
      "grad_norm": 1.7776700258255005,
      "learning_rate": 0.0001440478130650733,
      "loss": 1.1856,
      "step": 10448
    },
    {
      "epoch": 0.37420094185900765,
      "grad_norm": 1.678805947303772,
      "learning_rate": 0.00014403739960018824,
      "loss": 1.3772,
      "step": 10449
    },
    {
      "epoch": 0.3742367539885759,
      "grad_norm": 1.3966861963272095,
      "learning_rate": 0.00014402698554283675,
      "loss": 1.6829,
      "step": 10450
    },
    {
      "epoch": 0.3742725661181442,
      "grad_norm": 2.3683388233184814,
      "learning_rate": 0.00014401657089315904,
      "loss": 1.6944,
      "step": 10451
    },
    {
      "epoch": 0.3743083782477125,
      "grad_norm": 2.4242324829101562,
      "learning_rate": 0.00014400615565129507,
      "loss": 1.4101,
      "step": 10452
    },
    {
      "epoch": 0.3743441903772808,
      "grad_norm": 2.579834461212158,
      "learning_rate": 0.00014399573981738507,
      "loss": 1.3847,
      "step": 10453
    },
    {
      "epoch": 0.37438000250684905,
      "grad_norm": 1.703393578529358,
      "learning_rate": 0.00014398532339156912,
      "loss": 1.4959,
      "step": 10454
    },
    {
      "epoch": 0.3744158146364174,
      "grad_norm": 2.0507161617279053,
      "learning_rate": 0.00014397490637398742,
      "loss": 1.2748,
      "step": 10455
    },
    {
      "epoch": 0.37445162676598565,
      "grad_norm": 1.2460108995437622,
      "learning_rate": 0.00014396448876478007,
      "loss": 1.5361,
      "step": 10456
    },
    {
      "epoch": 0.3744874388955539,
      "grad_norm": 1.6622123718261719,
      "learning_rate": 0.00014395407056408722,
      "loss": 1.5679,
      "step": 10457
    },
    {
      "epoch": 0.3745232510251222,
      "grad_norm": 2.1253249645233154,
      "learning_rate": 0.00014394365177204904,
      "loss": 1.7294,
      "step": 10458
    },
    {
      "epoch": 0.3745590631546905,
      "grad_norm": 2.055772542953491,
      "learning_rate": 0.00014393323238880571,
      "loss": 1.2525,
      "step": 10459
    },
    {
      "epoch": 0.3745948752842588,
      "grad_norm": 1.6590421199798584,
      "learning_rate": 0.00014392281241449743,
      "loss": 1.4193,
      "step": 10460
    },
    {
      "epoch": 0.37463068741382705,
      "grad_norm": 1.8132797479629517,
      "learning_rate": 0.00014391239184926433,
      "loss": 1.4351,
      "step": 10461
    },
    {
      "epoch": 0.37466649954339537,
      "grad_norm": 1.3465622663497925,
      "learning_rate": 0.00014390197069324667,
      "loss": 1.6417,
      "step": 10462
    },
    {
      "epoch": 0.37470231167296364,
      "grad_norm": 1.4198817014694214,
      "learning_rate": 0.0001438915489465846,
      "loss": 1.3433,
      "step": 10463
    },
    {
      "epoch": 0.3747381238025319,
      "grad_norm": 2.0465641021728516,
      "learning_rate": 0.0001438811266094184,
      "loss": 1.8104,
      "step": 10464
    },
    {
      "epoch": 0.3747739359321002,
      "grad_norm": 1.4621689319610596,
      "learning_rate": 0.0001438707036818882,
      "loss": 1.5587,
      "step": 10465
    },
    {
      "epoch": 0.3748097480616685,
      "grad_norm": 1.6371957063674927,
      "learning_rate": 0.00014386028016413426,
      "loss": 1.4263,
      "step": 10466
    },
    {
      "epoch": 0.37484556019123677,
      "grad_norm": 2.203371286392212,
      "learning_rate": 0.00014384985605629685,
      "loss": 1.5286,
      "step": 10467
    },
    {
      "epoch": 0.37488137232080504,
      "grad_norm": 1.7377517223358154,
      "learning_rate": 0.0001438394313585162,
      "loss": 1.3761,
      "step": 10468
    },
    {
      "epoch": 0.37491718445037336,
      "grad_norm": 1.7066103219985962,
      "learning_rate": 0.00014382900607093254,
      "loss": 1.6471,
      "step": 10469
    },
    {
      "epoch": 0.37495299657994163,
      "grad_norm": 3.365816593170166,
      "learning_rate": 0.00014381858019368613,
      "loss": 1.805,
      "step": 10470
    },
    {
      "epoch": 0.3749888087095099,
      "grad_norm": 1.8790528774261475,
      "learning_rate": 0.00014380815372691728,
      "loss": 1.3121,
      "step": 10471
    },
    {
      "epoch": 0.37502462083907817,
      "grad_norm": 1.5260752439498901,
      "learning_rate": 0.00014379772667076618,
      "loss": 1.7436,
      "step": 10472
    },
    {
      "epoch": 0.3750604329686465,
      "grad_norm": 1.6847316026687622,
      "learning_rate": 0.0001437872990253732,
      "loss": 1.5917,
      "step": 10473
    },
    {
      "epoch": 0.37509624509821476,
      "grad_norm": 1.5153547525405884,
      "learning_rate": 0.0001437768707908786,
      "loss": 1.5685,
      "step": 10474
    },
    {
      "epoch": 0.37513205722778303,
      "grad_norm": 1.3384404182434082,
      "learning_rate": 0.00014376644196742263,
      "loss": 1.3586,
      "step": 10475
    },
    {
      "epoch": 0.37516786935735136,
      "grad_norm": 1.3099168539047241,
      "learning_rate": 0.00014375601255514565,
      "loss": 1.1538,
      "step": 10476
    },
    {
      "epoch": 0.3752036814869196,
      "grad_norm": 2.159562110900879,
      "learning_rate": 0.00014374558255418797,
      "loss": 1.3775,
      "step": 10477
    },
    {
      "epoch": 0.3752394936164879,
      "grad_norm": 1.8578730821609497,
      "learning_rate": 0.00014373515196468991,
      "loss": 1.4176,
      "step": 10478
    },
    {
      "epoch": 0.37527530574605616,
      "grad_norm": 1.4514464139938354,
      "learning_rate": 0.00014372472078679177,
      "loss": 1.2886,
      "step": 10479
    },
    {
      "epoch": 0.3753111178756245,
      "grad_norm": 2.7915217876434326,
      "learning_rate": 0.00014371428902063395,
      "loss": 1.6081,
      "step": 10480
    },
    {
      "epoch": 0.37534693000519276,
      "grad_norm": 1.7059495449066162,
      "learning_rate": 0.00014370385666635674,
      "loss": 1.3397,
      "step": 10481
    },
    {
      "epoch": 0.375382742134761,
      "grad_norm": 1.5151960849761963,
      "learning_rate": 0.00014369342372410053,
      "loss": 1.4587,
      "step": 10482
    },
    {
      "epoch": 0.37541855426432935,
      "grad_norm": 1.5763031244277954,
      "learning_rate": 0.00014368299019400563,
      "loss": 1.0639,
      "step": 10483
    },
    {
      "epoch": 0.3754543663938976,
      "grad_norm": 1.905612587928772,
      "learning_rate": 0.0001436725560762125,
      "loss": 1.7506,
      "step": 10484
    },
    {
      "epoch": 0.3754901785234659,
      "grad_norm": 1.5665966272354126,
      "learning_rate": 0.0001436621213708614,
      "loss": 1.4628,
      "step": 10485
    },
    {
      "epoch": 0.37552599065303416,
      "grad_norm": 1.8243300914764404,
      "learning_rate": 0.0001436516860780928,
      "loss": 1.5423,
      "step": 10486
    },
    {
      "epoch": 0.3755618027826025,
      "grad_norm": 1.7763745784759521,
      "learning_rate": 0.00014364125019804708,
      "loss": 1.8964,
      "step": 10487
    },
    {
      "epoch": 0.37559761491217075,
      "grad_norm": 2.380751848220825,
      "learning_rate": 0.00014363081373086462,
      "loss": 1.7111,
      "step": 10488
    },
    {
      "epoch": 0.375633427041739,
      "grad_norm": 2.1618824005126953,
      "learning_rate": 0.00014362037667668584,
      "loss": 1.2592,
      "step": 10489
    },
    {
      "epoch": 0.37566923917130735,
      "grad_norm": 2.174567222595215,
      "learning_rate": 0.00014360993903565116,
      "loss": 1.8956,
      "step": 10490
    },
    {
      "epoch": 0.3757050513008756,
      "grad_norm": 1.3785721063613892,
      "learning_rate": 0.00014359950080790101,
      "loss": 1.5306,
      "step": 10491
    },
    {
      "epoch": 0.3757408634304439,
      "grad_norm": 1.9032706022262573,
      "learning_rate": 0.0001435890619935758,
      "loss": 1.3604,
      "step": 10492
    },
    {
      "epoch": 0.37577667556001215,
      "grad_norm": 1.70711088180542,
      "learning_rate": 0.00014357862259281603,
      "loss": 1.7252,
      "step": 10493
    },
    {
      "epoch": 0.3758124876895805,
      "grad_norm": 1.943627119064331,
      "learning_rate": 0.00014356818260576206,
      "loss": 1.4448,
      "step": 10494
    },
    {
      "epoch": 0.37584829981914875,
      "grad_norm": 1.4574229717254639,
      "learning_rate": 0.0001435577420325544,
      "loss": 1.5815,
      "step": 10495
    },
    {
      "epoch": 0.375884111948717,
      "grad_norm": 1.5903644561767578,
      "learning_rate": 0.0001435473008733335,
      "loss": 1.5568,
      "step": 10496
    },
    {
      "epoch": 0.37591992407828534,
      "grad_norm": 1.8567113876342773,
      "learning_rate": 0.00014353685912823987,
      "loss": 1.2216,
      "step": 10497
    },
    {
      "epoch": 0.3759557362078536,
      "grad_norm": 1.382388949394226,
      "learning_rate": 0.00014352641679741393,
      "loss": 1.6755,
      "step": 10498
    },
    {
      "epoch": 0.3759915483374219,
      "grad_norm": 1.329992651939392,
      "learning_rate": 0.0001435159738809962,
      "loss": 1.2925,
      "step": 10499
    },
    {
      "epoch": 0.37602736046699015,
      "grad_norm": 2.657209873199463,
      "learning_rate": 0.0001435055303791272,
      "loss": 1.673,
      "step": 10500
    },
    {
      "epoch": 0.37606317259655847,
      "grad_norm": 1.9933116436004639,
      "learning_rate": 0.00014349508629194738,
      "loss": 1.4519,
      "step": 10501
    },
    {
      "epoch": 0.37609898472612674,
      "grad_norm": 1.554787516593933,
      "learning_rate": 0.00014348464161959728,
      "loss": 1.6938,
      "step": 10502
    },
    {
      "epoch": 0.376134796855695,
      "grad_norm": 1.7749775648117065,
      "learning_rate": 0.0001434741963622174,
      "loss": 1.3281,
      "step": 10503
    },
    {
      "epoch": 0.37617060898526333,
      "grad_norm": 1.529089331626892,
      "learning_rate": 0.00014346375051994833,
      "loss": 1.3901,
      "step": 10504
    },
    {
      "epoch": 0.3762064211148316,
      "grad_norm": 1.5324797630310059,
      "learning_rate": 0.00014345330409293053,
      "loss": 1.4895,
      "step": 10505
    },
    {
      "epoch": 0.37624223324439987,
      "grad_norm": 1.7260066270828247,
      "learning_rate": 0.0001434428570813046,
      "loss": 1.5439,
      "step": 10506
    },
    {
      "epoch": 0.37627804537396814,
      "grad_norm": 1.5831547975540161,
      "learning_rate": 0.00014343240948521104,
      "loss": 1.4157,
      "step": 10507
    },
    {
      "epoch": 0.37631385750353646,
      "grad_norm": 1.6812795400619507,
      "learning_rate": 0.00014342196130479043,
      "loss": 1.6985,
      "step": 10508
    },
    {
      "epoch": 0.37634966963310473,
      "grad_norm": 1.7851736545562744,
      "learning_rate": 0.0001434115125401834,
      "loss": 1.6871,
      "step": 10509
    },
    {
      "epoch": 0.376385481762673,
      "grad_norm": 1.9314453601837158,
      "learning_rate": 0.00014340106319153038,
      "loss": 1.6787,
      "step": 10510
    },
    {
      "epoch": 0.3764212938922413,
      "grad_norm": 1.592041254043579,
      "learning_rate": 0.0001433906132589721,
      "loss": 1.3062,
      "step": 10511
    },
    {
      "epoch": 0.3764571060218096,
      "grad_norm": 1.572988510131836,
      "learning_rate": 0.00014338016274264905,
      "loss": 1.7478,
      "step": 10512
    },
    {
      "epoch": 0.37649291815137786,
      "grad_norm": 2.4121146202087402,
      "learning_rate": 0.0001433697116427019,
      "loss": 1.6346,
      "step": 10513
    },
    {
      "epoch": 0.37652873028094613,
      "grad_norm": 1.4479660987854004,
      "learning_rate": 0.0001433592599592712,
      "loss": 1.4558,
      "step": 10514
    },
    {
      "epoch": 0.37656454241051446,
      "grad_norm": 1.7201353311538696,
      "learning_rate": 0.00014334880769249758,
      "loss": 1.4518,
      "step": 10515
    },
    {
      "epoch": 0.3766003545400827,
      "grad_norm": 1.6340991258621216,
      "learning_rate": 0.00014333835484252167,
      "loss": 1.6044,
      "step": 10516
    },
    {
      "epoch": 0.376636166669651,
      "grad_norm": 1.3527189493179321,
      "learning_rate": 0.00014332790140948414,
      "loss": 1.4002,
      "step": 10517
    },
    {
      "epoch": 0.3766719787992193,
      "grad_norm": 2.4372007846832275,
      "learning_rate": 0.00014331744739352556,
      "loss": 1.747,
      "step": 10518
    },
    {
      "epoch": 0.3767077909287876,
      "grad_norm": 1.1843258142471313,
      "learning_rate": 0.0001433069927947866,
      "loss": 1.5292,
      "step": 10519
    },
    {
      "epoch": 0.37674360305835586,
      "grad_norm": 2.6395835876464844,
      "learning_rate": 0.0001432965376134079,
      "loss": 1.8212,
      "step": 10520
    },
    {
      "epoch": 0.3767794151879241,
      "grad_norm": 1.5669384002685547,
      "learning_rate": 0.00014328608184953012,
      "loss": 1.3947,
      "step": 10521
    },
    {
      "epoch": 0.37681522731749245,
      "grad_norm": 1.5199379920959473,
      "learning_rate": 0.000143275625503294,
      "loss": 1.4586,
      "step": 10522
    },
    {
      "epoch": 0.3768510394470607,
      "grad_norm": 1.574935793876648,
      "learning_rate": 0.0001432651685748401,
      "loss": 1.4656,
      "step": 10523
    },
    {
      "epoch": 0.376886851576629,
      "grad_norm": 1.8273627758026123,
      "learning_rate": 0.0001432547110643092,
      "loss": 1.596,
      "step": 10524
    },
    {
      "epoch": 0.3769226637061973,
      "grad_norm": 1.7247346639633179,
      "learning_rate": 0.00014324425297184193,
      "loss": 1.555,
      "step": 10525
    },
    {
      "epoch": 0.3769584758357656,
      "grad_norm": 1.646228551864624,
      "learning_rate": 0.00014323379429757906,
      "loss": 1.6124,
      "step": 10526
    },
    {
      "epoch": 0.37699428796533385,
      "grad_norm": 1.649630069732666,
      "learning_rate": 0.00014322333504166124,
      "loss": 1.8379,
      "step": 10527
    },
    {
      "epoch": 0.3770301000949021,
      "grad_norm": 2.7177631855010986,
      "learning_rate": 0.00014321287520422917,
      "loss": 1.3449,
      "step": 10528
    },
    {
      "epoch": 0.37706591222447045,
      "grad_norm": 1.322988748550415,
      "learning_rate": 0.00014320241478542363,
      "loss": 1.4162,
      "step": 10529
    },
    {
      "epoch": 0.3771017243540387,
      "grad_norm": 2.104003429412842,
      "learning_rate": 0.0001431919537853853,
      "loss": 1.7102,
      "step": 10530
    },
    {
      "epoch": 0.377137536483607,
      "grad_norm": 1.8676283359527588,
      "learning_rate": 0.000143181492204255,
      "loss": 1.4098,
      "step": 10531
    },
    {
      "epoch": 0.3771733486131753,
      "grad_norm": 1.5684523582458496,
      "learning_rate": 0.0001431710300421734,
      "loss": 1.5385,
      "step": 10532
    },
    {
      "epoch": 0.3772091607427436,
      "grad_norm": 1.7146109342575073,
      "learning_rate": 0.00014316056729928126,
      "loss": 1.7981,
      "step": 10533
    },
    {
      "epoch": 0.37724497287231185,
      "grad_norm": 1.6780128479003906,
      "learning_rate": 0.00014315010397571937,
      "loss": 1.4944,
      "step": 10534
    },
    {
      "epoch": 0.3772807850018801,
      "grad_norm": 3.0013227462768555,
      "learning_rate": 0.0001431396400716285,
      "loss": 2.0381,
      "step": 10535
    },
    {
      "epoch": 0.37731659713144844,
      "grad_norm": 1.5567467212677002,
      "learning_rate": 0.00014312917558714943,
      "loss": 1.4914,
      "step": 10536
    },
    {
      "epoch": 0.3773524092610167,
      "grad_norm": 1.4302515983581543,
      "learning_rate": 0.00014311871052242293,
      "loss": 1.5742,
      "step": 10537
    },
    {
      "epoch": 0.377388221390585,
      "grad_norm": 2.07220196723938,
      "learning_rate": 0.00014310824487758975,
      "loss": 1.5267,
      "step": 10538
    },
    {
      "epoch": 0.3774240335201533,
      "grad_norm": 1.5602160692214966,
      "learning_rate": 0.00014309777865279078,
      "loss": 1.1705,
      "step": 10539
    },
    {
      "epoch": 0.37745984564972157,
      "grad_norm": 1.5262706279754639,
      "learning_rate": 0.00014308731184816678,
      "loss": 1.6653,
      "step": 10540
    },
    {
      "epoch": 0.37749565777928984,
      "grad_norm": 1.5797264575958252,
      "learning_rate": 0.00014307684446385855,
      "loss": 1.4342,
      "step": 10541
    },
    {
      "epoch": 0.3775314699088581,
      "grad_norm": 1.70090651512146,
      "learning_rate": 0.000143066376500007,
      "loss": 1.7305,
      "step": 10542
    },
    {
      "epoch": 0.37756728203842643,
      "grad_norm": 1.5267583131790161,
      "learning_rate": 0.00014305590795675286,
      "loss": 1.5904,
      "step": 10543
    },
    {
      "epoch": 0.3776030941679947,
      "grad_norm": 1.5385205745697021,
      "learning_rate": 0.00014304543883423708,
      "loss": 1.7465,
      "step": 10544
    },
    {
      "epoch": 0.37763890629756297,
      "grad_norm": 2.824392795562744,
      "learning_rate": 0.0001430349691326004,
      "loss": 1.4748,
      "step": 10545
    },
    {
      "epoch": 0.3776747184271313,
      "grad_norm": 1.8800132274627686,
      "learning_rate": 0.00014302449885198373,
      "loss": 1.4558,
      "step": 10546
    },
    {
      "epoch": 0.37771053055669956,
      "grad_norm": 1.483883023262024,
      "learning_rate": 0.00014301402799252793,
      "loss": 1.384,
      "step": 10547
    },
    {
      "epoch": 0.37774634268626783,
      "grad_norm": 1.6904969215393066,
      "learning_rate": 0.00014300355655437385,
      "loss": 1.6016,
      "step": 10548
    },
    {
      "epoch": 0.3777821548158361,
      "grad_norm": 1.5325121879577637,
      "learning_rate": 0.00014299308453766238,
      "loss": 1.419,
      "step": 10549
    },
    {
      "epoch": 0.3778179669454044,
      "grad_norm": 1.5163652896881104,
      "learning_rate": 0.00014298261194253443,
      "loss": 1.6518,
      "step": 10550
    },
    {
      "epoch": 0.3778537790749727,
      "grad_norm": 1.606885313987732,
      "learning_rate": 0.00014297213876913087,
      "loss": 1.6129,
      "step": 10551
    },
    {
      "epoch": 0.37788959120454096,
      "grad_norm": 2.0452921390533447,
      "learning_rate": 0.00014296166501759263,
      "loss": 1.5371,
      "step": 10552
    },
    {
      "epoch": 0.3779254033341093,
      "grad_norm": 1.4756042957305908,
      "learning_rate": 0.00014295119068806063,
      "loss": 1.4163,
      "step": 10553
    },
    {
      "epoch": 0.37796121546367756,
      "grad_norm": 1.4416539669036865,
      "learning_rate": 0.00014294071578067568,
      "loss": 1.5408,
      "step": 10554
    },
    {
      "epoch": 0.3779970275932458,
      "grad_norm": 1.5078434944152832,
      "learning_rate": 0.00014293024029557886,
      "loss": 1.2568,
      "step": 10555
    },
    {
      "epoch": 0.3780328397228141,
      "grad_norm": 1.6258286237716675,
      "learning_rate": 0.000142919764232911,
      "loss": 1.8241,
      "step": 10556
    },
    {
      "epoch": 0.3780686518523824,
      "grad_norm": 1.5990259647369385,
      "learning_rate": 0.0001429092875928131,
      "loss": 1.4475,
      "step": 10557
    },
    {
      "epoch": 0.3781044639819507,
      "grad_norm": 1.7794666290283203,
      "learning_rate": 0.00014289881037542605,
      "loss": 1.5112,
      "step": 10558
    },
    {
      "epoch": 0.37814027611151896,
      "grad_norm": 1.70427668094635,
      "learning_rate": 0.00014288833258089086,
      "loss": 1.4345,
      "step": 10559
    },
    {
      "epoch": 0.3781760882410873,
      "grad_norm": 1.3057318925857544,
      "learning_rate": 0.00014287785420934846,
      "loss": 1.4043,
      "step": 10560
    },
    {
      "epoch": 0.37821190037065555,
      "grad_norm": 1.8462861776351929,
      "learning_rate": 0.0001428673752609399,
      "loss": 1.131,
      "step": 10561
    },
    {
      "epoch": 0.3782477125002238,
      "grad_norm": 1.4198931455612183,
      "learning_rate": 0.00014285689573580607,
      "loss": 1.5534,
      "step": 10562
    },
    {
      "epoch": 0.3782835246297921,
      "grad_norm": 1.9457460641860962,
      "learning_rate": 0.00014284641563408796,
      "loss": 1.711,
      "step": 10563
    },
    {
      "epoch": 0.3783193367593604,
      "grad_norm": 1.85354483127594,
      "learning_rate": 0.00014283593495592663,
      "loss": 1.6055,
      "step": 10564
    },
    {
      "epoch": 0.3783551488889287,
      "grad_norm": 3.7298476696014404,
      "learning_rate": 0.000142825453701463,
      "loss": 1.5535,
      "step": 10565
    },
    {
      "epoch": 0.37839096101849695,
      "grad_norm": 1.84163498878479,
      "learning_rate": 0.00014281497187083818,
      "loss": 1.3187,
      "step": 10566
    },
    {
      "epoch": 0.3784267731480653,
      "grad_norm": 1.7626897096633911,
      "learning_rate": 0.00014280448946419312,
      "loss": 1.4398,
      "step": 10567
    },
    {
      "epoch": 0.37846258527763355,
      "grad_norm": 1.497876524925232,
      "learning_rate": 0.0001427940064816689,
      "loss": 1.5085,
      "step": 10568
    },
    {
      "epoch": 0.3784983974072018,
      "grad_norm": 2.1282432079315186,
      "learning_rate": 0.00014278352292340646,
      "loss": 1.3299,
      "step": 10569
    },
    {
      "epoch": 0.3785342095367701,
      "grad_norm": 1.8996422290802002,
      "learning_rate": 0.00014277303878954694,
      "loss": 1.7137,
      "step": 10570
    },
    {
      "epoch": 0.3785700216663384,
      "grad_norm": 2.442035675048828,
      "learning_rate": 0.00014276255408023138,
      "loss": 1.4634,
      "step": 10571
    },
    {
      "epoch": 0.3786058337959067,
      "grad_norm": 1.3051122426986694,
      "learning_rate": 0.00014275206879560079,
      "loss": 1.4763,
      "step": 10572
    },
    {
      "epoch": 0.37864164592547495,
      "grad_norm": 1.575643539428711,
      "learning_rate": 0.00014274158293579628,
      "loss": 1.7568,
      "step": 10573
    },
    {
      "epoch": 0.37867745805504327,
      "grad_norm": 1.7613773345947266,
      "learning_rate": 0.00014273109650095886,
      "loss": 1.4811,
      "step": 10574
    },
    {
      "epoch": 0.37871327018461154,
      "grad_norm": 1.501792073249817,
      "learning_rate": 0.0001427206094912297,
      "loss": 1.4062,
      "step": 10575
    },
    {
      "epoch": 0.3787490823141798,
      "grad_norm": 1.534451961517334,
      "learning_rate": 0.00014271012190674983,
      "loss": 1.3444,
      "step": 10576
    },
    {
      "epoch": 0.3787848944437481,
      "grad_norm": 1.2604267597198486,
      "learning_rate": 0.00014269963374766034,
      "loss": 1.2716,
      "step": 10577
    },
    {
      "epoch": 0.3788207065733164,
      "grad_norm": 2.2848007678985596,
      "learning_rate": 0.00014268914501410239,
      "loss": 1.5,
      "step": 10578
    },
    {
      "epoch": 0.37885651870288467,
      "grad_norm": 1.4307903051376343,
      "learning_rate": 0.00014267865570621706,
      "loss": 1.5404,
      "step": 10579
    },
    {
      "epoch": 0.37889233083245294,
      "grad_norm": 2.649573802947998,
      "learning_rate": 0.00014266816582414547,
      "loss": 1.4351,
      "step": 10580
    },
    {
      "epoch": 0.37892814296202126,
      "grad_norm": 2.3270211219787598,
      "learning_rate": 0.00014265767536802873,
      "loss": 1.4689,
      "step": 10581
    },
    {
      "epoch": 0.37896395509158953,
      "grad_norm": 1.3619784116744995,
      "learning_rate": 0.000142647184338008,
      "loss": 1.5399,
      "step": 10582
    },
    {
      "epoch": 0.3789997672211578,
      "grad_norm": 1.9002552032470703,
      "learning_rate": 0.0001426366927342244,
      "loss": 1.6398,
      "step": 10583
    },
    {
      "epoch": 0.37903557935072607,
      "grad_norm": 1.818941593170166,
      "learning_rate": 0.0001426262005568191,
      "loss": 1.3925,
      "step": 10584
    },
    {
      "epoch": 0.3790713914802944,
      "grad_norm": 1.4739636182785034,
      "learning_rate": 0.00014261570780593327,
      "loss": 1.4677,
      "step": 10585
    },
    {
      "epoch": 0.37910720360986266,
      "grad_norm": 1.7951068878173828,
      "learning_rate": 0.00014260521448170805,
      "loss": 1.5913,
      "step": 10586
    },
    {
      "epoch": 0.37914301573943093,
      "grad_norm": 1.6509017944335938,
      "learning_rate": 0.0001425947205842846,
      "loss": 1.3659,
      "step": 10587
    },
    {
      "epoch": 0.37917882786899926,
      "grad_norm": 1.6819849014282227,
      "learning_rate": 0.00014258422611380418,
      "loss": 1.6058,
      "step": 10588
    },
    {
      "epoch": 0.3792146399985675,
      "grad_norm": 2.2275047302246094,
      "learning_rate": 0.0001425737310704079,
      "loss": 1.1905,
      "step": 10589
    },
    {
      "epoch": 0.3792504521281358,
      "grad_norm": 1.8153003454208374,
      "learning_rate": 0.000142563235454237,
      "loss": 1.8863,
      "step": 10590
    },
    {
      "epoch": 0.37928626425770406,
      "grad_norm": 1.4210957288742065,
      "learning_rate": 0.00014255273926543264,
      "loss": 1.4082,
      "step": 10591
    },
    {
      "epoch": 0.3793220763872724,
      "grad_norm": 1.9765852689743042,
      "learning_rate": 0.0001425422425041361,
      "loss": 1.7252,
      "step": 10592
    },
    {
      "epoch": 0.37935788851684066,
      "grad_norm": 1.3808059692382812,
      "learning_rate": 0.00014253174517048854,
      "loss": 1.3727,
      "step": 10593
    },
    {
      "epoch": 0.3793937006464089,
      "grad_norm": 1.5124943256378174,
      "learning_rate": 0.00014252124726463121,
      "loss": 1.456,
      "step": 10594
    },
    {
      "epoch": 0.37942951277597725,
      "grad_norm": 1.4630928039550781,
      "learning_rate": 0.00014251074878670537,
      "loss": 1.6548,
      "step": 10595
    },
    {
      "epoch": 0.3794653249055455,
      "grad_norm": 2.346756935119629,
      "learning_rate": 0.00014250024973685218,
      "loss": 1.6495,
      "step": 10596
    },
    {
      "epoch": 0.3795011370351138,
      "grad_norm": 1.8365200757980347,
      "learning_rate": 0.000142489750115213,
      "loss": 1.4613,
      "step": 10597
    },
    {
      "epoch": 0.37953694916468206,
      "grad_norm": 1.351531982421875,
      "learning_rate": 0.00014247924992192906,
      "loss": 1.497,
      "step": 10598
    },
    {
      "epoch": 0.3795727612942504,
      "grad_norm": 1.3425579071044922,
      "learning_rate": 0.00014246874915714157,
      "loss": 1.0996,
      "step": 10599
    },
    {
      "epoch": 0.37960857342381865,
      "grad_norm": 1.2506022453308105,
      "learning_rate": 0.00014245824782099185,
      "loss": 1.3726,
      "step": 10600
    },
    {
      "epoch": 0.3796443855533869,
      "grad_norm": 1.398263692855835,
      "learning_rate": 0.00014244774591362118,
      "loss": 1.4526,
      "step": 10601
    },
    {
      "epoch": 0.3796801976829552,
      "grad_norm": 1.5792254209518433,
      "learning_rate": 0.00014243724343517082,
      "loss": 1.4306,
      "step": 10602
    },
    {
      "epoch": 0.3797160098125235,
      "grad_norm": 1.728993535041809,
      "learning_rate": 0.0001424267403857821,
      "loss": 1.2494,
      "step": 10603
    },
    {
      "epoch": 0.3797518219420918,
      "grad_norm": 1.8500906229019165,
      "learning_rate": 0.00014241623676559633,
      "loss": 1.2789,
      "step": 10604
    },
    {
      "epoch": 0.37978763407166005,
      "grad_norm": 2.3870582580566406,
      "learning_rate": 0.0001424057325747548,
      "loss": 1.5109,
      "step": 10605
    },
    {
      "epoch": 0.3798234462012284,
      "grad_norm": 2.648805856704712,
      "learning_rate": 0.00014239522781339884,
      "loss": 1.3092,
      "step": 10606
    },
    {
      "epoch": 0.37985925833079665,
      "grad_norm": 1.6133131980895996,
      "learning_rate": 0.00014238472248166977,
      "loss": 1.5906,
      "step": 10607
    },
    {
      "epoch": 0.3798950704603649,
      "grad_norm": 1.9158686399459839,
      "learning_rate": 0.00014237421657970894,
      "loss": 1.7831,
      "step": 10608
    },
    {
      "epoch": 0.3799308825899332,
      "grad_norm": 1.773206353187561,
      "learning_rate": 0.00014236371010765766,
      "loss": 1.4976,
      "step": 10609
    },
    {
      "epoch": 0.3799666947195015,
      "grad_norm": 1.8020765781402588,
      "learning_rate": 0.00014235320306565732,
      "loss": 1.656,
      "step": 10610
    },
    {
      "epoch": 0.3800025068490698,
      "grad_norm": 1.7115094661712646,
      "learning_rate": 0.00014234269545384927,
      "loss": 1.7404,
      "step": 10611
    },
    {
      "epoch": 0.38003831897863805,
      "grad_norm": 1.5301717519760132,
      "learning_rate": 0.00014233218727237489,
      "loss": 1.6062,
      "step": 10612
    },
    {
      "epoch": 0.38007413110820637,
      "grad_norm": 1.6753218173980713,
      "learning_rate": 0.00014232167852137547,
      "loss": 1.6029,
      "step": 10613
    },
    {
      "epoch": 0.38010994323777464,
      "grad_norm": 1.724470853805542,
      "learning_rate": 0.00014231116920099252,
      "loss": 1.4289,
      "step": 10614
    },
    {
      "epoch": 0.3801457553673429,
      "grad_norm": 1.236194133758545,
      "learning_rate": 0.00014230065931136735,
      "loss": 1.4111,
      "step": 10615
    },
    {
      "epoch": 0.3801815674969112,
      "grad_norm": 2.511920213699341,
      "learning_rate": 0.00014229014885264136,
      "loss": 1.4207,
      "step": 10616
    },
    {
      "epoch": 0.3802173796264795,
      "grad_norm": 1.4968230724334717,
      "learning_rate": 0.00014227963782495598,
      "loss": 1.5417,
      "step": 10617
    },
    {
      "epoch": 0.38025319175604777,
      "grad_norm": 1.8402019739151,
      "learning_rate": 0.0001422691262284526,
      "loss": 1.5169,
      "step": 10618
    },
    {
      "epoch": 0.38028900388561604,
      "grad_norm": 1.912214756011963,
      "learning_rate": 0.00014225861406327265,
      "loss": 1.5186,
      "step": 10619
    },
    {
      "epoch": 0.38032481601518436,
      "grad_norm": 1.7003743648529053,
      "learning_rate": 0.00014224810132955755,
      "loss": 1.4935,
      "step": 10620
    },
    {
      "epoch": 0.38036062814475263,
      "grad_norm": 1.291312575340271,
      "learning_rate": 0.00014223758802744878,
      "loss": 1.4836,
      "step": 10621
    },
    {
      "epoch": 0.3803964402743209,
      "grad_norm": 2.281581163406372,
      "learning_rate": 0.0001422270741570877,
      "loss": 1.3742,
      "step": 10622
    },
    {
      "epoch": 0.38043225240388917,
      "grad_norm": 1.739435076713562,
      "learning_rate": 0.00014221655971861582,
      "loss": 1.5536,
      "step": 10623
    },
    {
      "epoch": 0.3804680645334575,
      "grad_norm": 2.075655937194824,
      "learning_rate": 0.0001422060447121746,
      "loss": 1.4077,
      "step": 10624
    },
    {
      "epoch": 0.38050387666302576,
      "grad_norm": 1.8676741123199463,
      "learning_rate": 0.0001421955291379055,
      "loss": 1.7129,
      "step": 10625
    },
    {
      "epoch": 0.38053968879259403,
      "grad_norm": 1.853179693222046,
      "learning_rate": 0.00014218501299594996,
      "loss": 1.369,
      "step": 10626
    },
    {
      "epoch": 0.38057550092216236,
      "grad_norm": 1.8642727136611938,
      "learning_rate": 0.00014217449628644947,
      "loss": 1.3751,
      "step": 10627
    },
    {
      "epoch": 0.3806113130517306,
      "grad_norm": 2.065969467163086,
      "learning_rate": 0.00014216397900954558,
      "loss": 1.4474,
      "step": 10628
    },
    {
      "epoch": 0.3806471251812989,
      "grad_norm": 2.014706611633301,
      "learning_rate": 0.00014215346116537968,
      "loss": 1.7028,
      "step": 10629
    },
    {
      "epoch": 0.38068293731086716,
      "grad_norm": 1.5329726934432983,
      "learning_rate": 0.0001421429427540934,
      "loss": 1.3595,
      "step": 10630
    },
    {
      "epoch": 0.3807187494404355,
      "grad_norm": 1.8910022974014282,
      "learning_rate": 0.00014213242377582815,
      "loss": 1.3986,
      "step": 10631
    },
    {
      "epoch": 0.38075456157000376,
      "grad_norm": 1.6006379127502441,
      "learning_rate": 0.0001421219042307255,
      "loss": 1.6447,
      "step": 10632
    },
    {
      "epoch": 0.380790373699572,
      "grad_norm": 1.9453397989273071,
      "learning_rate": 0.00014211138411892696,
      "loss": 1.5806,
      "step": 10633
    },
    {
      "epoch": 0.38082618582914035,
      "grad_norm": 1.7229535579681396,
      "learning_rate": 0.00014210086344057404,
      "loss": 1.6743,
      "step": 10634
    },
    {
      "epoch": 0.3808619979587086,
      "grad_norm": 1.395990252494812,
      "learning_rate": 0.00014209034219580833,
      "loss": 1.4022,
      "step": 10635
    },
    {
      "epoch": 0.3808978100882769,
      "grad_norm": 1.543500304222107,
      "learning_rate": 0.00014207982038477135,
      "loss": 1.6443,
      "step": 10636
    },
    {
      "epoch": 0.38093362221784516,
      "grad_norm": 1.8862489461898804,
      "learning_rate": 0.00014206929800760466,
      "loss": 1.8299,
      "step": 10637
    },
    {
      "epoch": 0.3809694343474135,
      "grad_norm": 1.6491189002990723,
      "learning_rate": 0.00014205877506444982,
      "loss": 1.5355,
      "step": 10638
    },
    {
      "epoch": 0.38100524647698175,
      "grad_norm": 2.0166943073272705,
      "learning_rate": 0.00014204825155544846,
      "loss": 1.2944,
      "step": 10639
    },
    {
      "epoch": 0.38104105860655,
      "grad_norm": 1.385064959526062,
      "learning_rate": 0.00014203772748074206,
      "loss": 1.5232,
      "step": 10640
    },
    {
      "epoch": 0.38107687073611834,
      "grad_norm": 1.4159349203109741,
      "learning_rate": 0.00014202720284047234,
      "loss": 1.3836,
      "step": 10641
    },
    {
      "epoch": 0.3811126828656866,
      "grad_norm": 1.7959661483764648,
      "learning_rate": 0.00014201667763478074,
      "loss": 1.7063,
      "step": 10642
    },
    {
      "epoch": 0.3811484949952549,
      "grad_norm": 1.8224889039993286,
      "learning_rate": 0.00014200615186380899,
      "loss": 1.6915,
      "step": 10643
    },
    {
      "epoch": 0.38118430712482315,
      "grad_norm": 2.0088465213775635,
      "learning_rate": 0.0001419956255276986,
      "loss": 1.6776,
      "step": 10644
    },
    {
      "epoch": 0.3812201192543915,
      "grad_norm": 1.4970310926437378,
      "learning_rate": 0.00014198509862659129,
      "loss": 1.4258,
      "step": 10645
    },
    {
      "epoch": 0.38125593138395975,
      "grad_norm": 1.4173898696899414,
      "learning_rate": 0.00014197457116062857,
      "loss": 1.7229,
      "step": 10646
    },
    {
      "epoch": 0.381291743513528,
      "grad_norm": 1.68673837184906,
      "learning_rate": 0.0001419640431299522,
      "loss": 1.4912,
      "step": 10647
    },
    {
      "epoch": 0.38132755564309634,
      "grad_norm": 2.059001922607422,
      "learning_rate": 0.00014195351453470374,
      "loss": 1.328,
      "step": 10648
    },
    {
      "epoch": 0.3813633677726646,
      "grad_norm": 1.7006314992904663,
      "learning_rate": 0.00014194298537502487,
      "loss": 1.4513,
      "step": 10649
    },
    {
      "epoch": 0.3813991799022329,
      "grad_norm": 1.4936386346817017,
      "learning_rate": 0.00014193245565105722,
      "loss": 1.6796,
      "step": 10650
    },
    {
      "epoch": 0.38143499203180115,
      "grad_norm": 1.7479816675186157,
      "learning_rate": 0.00014192192536294245,
      "loss": 1.4616,
      "step": 10651
    },
    {
      "epoch": 0.38147080416136947,
      "grad_norm": 1.7160698175430298,
      "learning_rate": 0.00014191139451082228,
      "loss": 1.6427,
      "step": 10652
    },
    {
      "epoch": 0.38150661629093774,
      "grad_norm": 1.7994788885116577,
      "learning_rate": 0.00014190086309483834,
      "loss": 1.4788,
      "step": 10653
    },
    {
      "epoch": 0.381542428420506,
      "grad_norm": 1.9248762130737305,
      "learning_rate": 0.00014189033111513234,
      "loss": 1.6344,
      "step": 10654
    },
    {
      "epoch": 0.38157824055007433,
      "grad_norm": 2.2364611625671387,
      "learning_rate": 0.00014187979857184597,
      "loss": 1.5578,
      "step": 10655
    },
    {
      "epoch": 0.3816140526796426,
      "grad_norm": 1.8766387701034546,
      "learning_rate": 0.00014186926546512095,
      "loss": 1.7108,
      "step": 10656
    },
    {
      "epoch": 0.38164986480921087,
      "grad_norm": 2.0187058448791504,
      "learning_rate": 0.00014185873179509893,
      "loss": 1.3808,
      "step": 10657
    },
    {
      "epoch": 0.38168567693877914,
      "grad_norm": 1.377492070198059,
      "learning_rate": 0.00014184819756192168,
      "loss": 1.2254,
      "step": 10658
    },
    {
      "epoch": 0.38172148906834746,
      "grad_norm": 1.88951575756073,
      "learning_rate": 0.00014183766276573096,
      "loss": 1.5138,
      "step": 10659
    },
    {
      "epoch": 0.38175730119791573,
      "grad_norm": 2.606764554977417,
      "learning_rate": 0.00014182712740666838,
      "loss": 1.9735,
      "step": 10660
    },
    {
      "epoch": 0.381793113327484,
      "grad_norm": 2.150054693222046,
      "learning_rate": 0.00014181659148487582,
      "loss": 1.38,
      "step": 10661
    },
    {
      "epoch": 0.3818289254570523,
      "grad_norm": 1.516271948814392,
      "learning_rate": 0.00014180605500049493,
      "loss": 1.7134,
      "step": 10662
    },
    {
      "epoch": 0.3818647375866206,
      "grad_norm": 1.3602635860443115,
      "learning_rate": 0.0001417955179536675,
      "loss": 1.4687,
      "step": 10663
    },
    {
      "epoch": 0.38190054971618886,
      "grad_norm": 1.6596342325210571,
      "learning_rate": 0.00014178498034453528,
      "loss": 1.5292,
      "step": 10664
    },
    {
      "epoch": 0.38193636184575713,
      "grad_norm": 1.9334921836853027,
      "learning_rate": 0.00014177444217324005,
      "loss": 1.3077,
      "step": 10665
    },
    {
      "epoch": 0.38197217397532546,
      "grad_norm": 1.7454739809036255,
      "learning_rate": 0.00014176390343992358,
      "loss": 1.722,
      "step": 10666
    },
    {
      "epoch": 0.3820079861048937,
      "grad_norm": 1.3706307411193848,
      "learning_rate": 0.0001417533641447277,
      "loss": 1.488,
      "step": 10667
    },
    {
      "epoch": 0.382043798234462,
      "grad_norm": 1.5092109441757202,
      "learning_rate": 0.00014174282428779412,
      "loss": 1.7905,
      "step": 10668
    },
    {
      "epoch": 0.3820796103640303,
      "grad_norm": 1.692136526107788,
      "learning_rate": 0.0001417322838692647,
      "loss": 1.2234,
      "step": 10669
    },
    {
      "epoch": 0.3821154224935986,
      "grad_norm": 1.8451822996139526,
      "learning_rate": 0.00014172174288928124,
      "loss": 1.5694,
      "step": 10670
    },
    {
      "epoch": 0.38215123462316686,
      "grad_norm": 1.4315675497055054,
      "learning_rate": 0.00014171120134798552,
      "loss": 1.4778,
      "step": 10671
    },
    {
      "epoch": 0.3821870467527351,
      "grad_norm": 1.8963932991027832,
      "learning_rate": 0.00014170065924551942,
      "loss": 1.2421,
      "step": 10672
    },
    {
      "epoch": 0.38222285888230345,
      "grad_norm": 1.7373079061508179,
      "learning_rate": 0.00014169011658202472,
      "loss": 1.2913,
      "step": 10673
    },
    {
      "epoch": 0.3822586710118717,
      "grad_norm": 1.6694564819335938,
      "learning_rate": 0.00014167957335764331,
      "loss": 1.6226,
      "step": 10674
    },
    {
      "epoch": 0.38229448314144,
      "grad_norm": 2.1219732761383057,
      "learning_rate": 0.00014166902957251696,
      "loss": 1.6144,
      "step": 10675
    },
    {
      "epoch": 0.3823302952710083,
      "grad_norm": 1.57712721824646,
      "learning_rate": 0.00014165848522678756,
      "loss": 1.1831,
      "step": 10676
    },
    {
      "epoch": 0.3823661074005766,
      "grad_norm": 1.6538848876953125,
      "learning_rate": 0.00014164794032059703,
      "loss": 1.7272,
      "step": 10677
    },
    {
      "epoch": 0.38240191953014485,
      "grad_norm": 1.840171217918396,
      "learning_rate": 0.00014163739485408716,
      "loss": 1.5288,
      "step": 10678
    },
    {
      "epoch": 0.3824377316597131,
      "grad_norm": 1.7260515689849854,
      "learning_rate": 0.00014162684882739984,
      "loss": 1.6227,
      "step": 10679
    },
    {
      "epoch": 0.38247354378928144,
      "grad_norm": 2.425084352493286,
      "learning_rate": 0.00014161630224067694,
      "loss": 1.4853,
      "step": 10680
    },
    {
      "epoch": 0.3825093559188497,
      "grad_norm": 1.8197312355041504,
      "learning_rate": 0.0001416057550940604,
      "loss": 1.797,
      "step": 10681
    },
    {
      "epoch": 0.382545168048418,
      "grad_norm": 1.4861470460891724,
      "learning_rate": 0.00014159520738769212,
      "loss": 1.5128,
      "step": 10682
    },
    {
      "epoch": 0.3825809801779863,
      "grad_norm": 1.9796706438064575,
      "learning_rate": 0.00014158465912171396,
      "loss": 1.574,
      "step": 10683
    },
    {
      "epoch": 0.3826167923075546,
      "grad_norm": 1.8207260370254517,
      "learning_rate": 0.00014157411029626783,
      "loss": 1.3684,
      "step": 10684
    },
    {
      "epoch": 0.38265260443712285,
      "grad_norm": 1.8487191200256348,
      "learning_rate": 0.0001415635609114957,
      "loss": 1.386,
      "step": 10685
    },
    {
      "epoch": 0.3826884165666911,
      "grad_norm": 1.9048634767532349,
      "learning_rate": 0.00014155301096753945,
      "loss": 1.496,
      "step": 10686
    },
    {
      "epoch": 0.38272422869625944,
      "grad_norm": 1.796779751777649,
      "learning_rate": 0.00014154246046454107,
      "loss": 1.4984,
      "step": 10687
    },
    {
      "epoch": 0.3827600408258277,
      "grad_norm": 1.5472294092178345,
      "learning_rate": 0.00014153190940264246,
      "loss": 1.6013,
      "step": 10688
    },
    {
      "epoch": 0.382795852955396,
      "grad_norm": 2.0484673976898193,
      "learning_rate": 0.00014152135778198557,
      "loss": 1.7283,
      "step": 10689
    },
    {
      "epoch": 0.3828316650849643,
      "grad_norm": 1.5459712743759155,
      "learning_rate": 0.00014151080560271235,
      "loss": 1.3491,
      "step": 10690
    },
    {
      "epoch": 0.38286747721453257,
      "grad_norm": 1.379616618156433,
      "learning_rate": 0.00014150025286496483,
      "loss": 1.3149,
      "step": 10691
    },
    {
      "epoch": 0.38290328934410084,
      "grad_norm": 2.17366886138916,
      "learning_rate": 0.0001414896995688849,
      "loss": 1.6789,
      "step": 10692
    },
    {
      "epoch": 0.3829391014736691,
      "grad_norm": 1.3768093585968018,
      "learning_rate": 0.00014147914571461455,
      "loss": 1.5071,
      "step": 10693
    },
    {
      "epoch": 0.38297491360323743,
      "grad_norm": 1.906646728515625,
      "learning_rate": 0.0001414685913022959,
      "loss": 1.6087,
      "step": 10694
    },
    {
      "epoch": 0.3830107257328057,
      "grad_norm": 2.058373212814331,
      "learning_rate": 0.00014145803633207077,
      "loss": 1.7719,
      "step": 10695
    },
    {
      "epoch": 0.38304653786237397,
      "grad_norm": 1.584075689315796,
      "learning_rate": 0.00014144748080408126,
      "loss": 1.4597,
      "step": 10696
    },
    {
      "epoch": 0.3830823499919423,
      "grad_norm": 1.2557836771011353,
      "learning_rate": 0.00014143692471846935,
      "loss": 1.2409,
      "step": 10697
    },
    {
      "epoch": 0.38311816212151056,
      "grad_norm": 1.7716764211654663,
      "learning_rate": 0.0001414263680753771,
      "loss": 1.6804,
      "step": 10698
    },
    {
      "epoch": 0.38315397425107883,
      "grad_norm": 1.3837485313415527,
      "learning_rate": 0.00014141581087494644,
      "loss": 1.5138,
      "step": 10699
    },
    {
      "epoch": 0.3831897863806471,
      "grad_norm": 1.3387659788131714,
      "learning_rate": 0.00014140525311731952,
      "loss": 1.4457,
      "step": 10700
    },
    {
      "epoch": 0.3832255985102154,
      "grad_norm": 1.4828484058380127,
      "learning_rate": 0.00014139469480263828,
      "loss": 1.6133,
      "step": 10701
    },
    {
      "epoch": 0.3832614106397837,
      "grad_norm": 2.200072765350342,
      "learning_rate": 0.00014138413593104486,
      "loss": 1.3238,
      "step": 10702
    },
    {
      "epoch": 0.38329722276935196,
      "grad_norm": 1.9859720468521118,
      "learning_rate": 0.0001413735765026813,
      "loss": 1.7688,
      "step": 10703
    },
    {
      "epoch": 0.3833330348989203,
      "grad_norm": 1.7335363626480103,
      "learning_rate": 0.00014136301651768957,
      "loss": 1.6909,
      "step": 10704
    },
    {
      "epoch": 0.38336884702848856,
      "grad_norm": 1.7980806827545166,
      "learning_rate": 0.00014135245597621184,
      "loss": 1.2562,
      "step": 10705
    },
    {
      "epoch": 0.3834046591580568,
      "grad_norm": 1.8698501586914062,
      "learning_rate": 0.00014134189487839013,
      "loss": 1.6418,
      "step": 10706
    },
    {
      "epoch": 0.3834404712876251,
      "grad_norm": 1.5134303569793701,
      "learning_rate": 0.0001413313332243666,
      "loss": 1.3264,
      "step": 10707
    },
    {
      "epoch": 0.3834762834171934,
      "grad_norm": 1.5431501865386963,
      "learning_rate": 0.00014132077101428324,
      "loss": 1.4467,
      "step": 10708
    },
    {
      "epoch": 0.3835120955467617,
      "grad_norm": 1.60693359375,
      "learning_rate": 0.00014131020824828224,
      "loss": 1.5158,
      "step": 10709
    },
    {
      "epoch": 0.38354790767632996,
      "grad_norm": 2.905691385269165,
      "learning_rate": 0.00014129964492650568,
      "loss": 1.7937,
      "step": 10710
    },
    {
      "epoch": 0.3835837198058983,
      "grad_norm": 1.4310888051986694,
      "learning_rate": 0.00014128908104909567,
      "loss": 1.286,
      "step": 10711
    },
    {
      "epoch": 0.38361953193546655,
      "grad_norm": 1.8025977611541748,
      "learning_rate": 0.00014127851661619432,
      "loss": 1.27,
      "step": 10712
    },
    {
      "epoch": 0.3836553440650348,
      "grad_norm": 1.9007517099380493,
      "learning_rate": 0.00014126795162794378,
      "loss": 1.5953,
      "step": 10713
    },
    {
      "epoch": 0.3836911561946031,
      "grad_norm": 2.110842704772949,
      "learning_rate": 0.00014125738608448618,
      "loss": 1.5767,
      "step": 10714
    },
    {
      "epoch": 0.3837269683241714,
      "grad_norm": 1.5326857566833496,
      "learning_rate": 0.00014124681998596366,
      "loss": 1.6487,
      "step": 10715
    },
    {
      "epoch": 0.3837627804537397,
      "grad_norm": 1.8157939910888672,
      "learning_rate": 0.0001412362533325184,
      "loss": 1.588,
      "step": 10716
    },
    {
      "epoch": 0.38379859258330795,
      "grad_norm": 1.4543850421905518,
      "learning_rate": 0.0001412256861242925,
      "loss": 1.5232,
      "step": 10717
    },
    {
      "epoch": 0.3838344047128763,
      "grad_norm": 1.69603431224823,
      "learning_rate": 0.00014121511836142823,
      "loss": 1.727,
      "step": 10718
    },
    {
      "epoch": 0.38387021684244454,
      "grad_norm": 1.8203978538513184,
      "learning_rate": 0.00014120455004406766,
      "loss": 1.6048,
      "step": 10719
    },
    {
      "epoch": 0.3839060289720128,
      "grad_norm": 1.70366370677948,
      "learning_rate": 0.00014119398117235304,
      "loss": 1.6988,
      "step": 10720
    },
    {
      "epoch": 0.3839418411015811,
      "grad_norm": 1.6187806129455566,
      "learning_rate": 0.00014118341174642653,
      "loss": 1.534,
      "step": 10721
    },
    {
      "epoch": 0.3839776532311494,
      "grad_norm": 1.8127055168151855,
      "learning_rate": 0.00014117284176643033,
      "loss": 1.429,
      "step": 10722
    },
    {
      "epoch": 0.3840134653607177,
      "grad_norm": 2.1849961280822754,
      "learning_rate": 0.00014116227123250668,
      "loss": 1.515,
      "step": 10723
    },
    {
      "epoch": 0.38404927749028595,
      "grad_norm": 1.5022850036621094,
      "learning_rate": 0.00014115170014479775,
      "loss": 1.6862,
      "step": 10724
    },
    {
      "epoch": 0.38408508961985427,
      "grad_norm": 1.709812879562378,
      "learning_rate": 0.0001411411285034458,
      "loss": 1.5168,
      "step": 10725
    },
    {
      "epoch": 0.38412090174942254,
      "grad_norm": 1.5337954759597778,
      "learning_rate": 0.000141130556308593,
      "loss": 1.595,
      "step": 10726
    },
    {
      "epoch": 0.3841567138789908,
      "grad_norm": 1.5661762952804565,
      "learning_rate": 0.00014111998356038162,
      "loss": 1.325,
      "step": 10727
    },
    {
      "epoch": 0.3841925260085591,
      "grad_norm": 2.0021166801452637,
      "learning_rate": 0.00014110941025895392,
      "loss": 1.3808,
      "step": 10728
    },
    {
      "epoch": 0.3842283381381274,
      "grad_norm": 1.3688938617706299,
      "learning_rate": 0.00014109883640445214,
      "loss": 1.366,
      "step": 10729
    },
    {
      "epoch": 0.38426415026769567,
      "grad_norm": 2.30067777633667,
      "learning_rate": 0.00014108826199701852,
      "loss": 1.3969,
      "step": 10730
    },
    {
      "epoch": 0.38429996239726394,
      "grad_norm": 2.1623682975769043,
      "learning_rate": 0.00014107768703679533,
      "loss": 1.4078,
      "step": 10731
    },
    {
      "epoch": 0.38433577452683226,
      "grad_norm": 1.4993696212768555,
      "learning_rate": 0.00014106711152392484,
      "loss": 1.499,
      "step": 10732
    },
    {
      "epoch": 0.38437158665640053,
      "grad_norm": 1.903861403465271,
      "learning_rate": 0.00014105653545854935,
      "loss": 1.5057,
      "step": 10733
    },
    {
      "epoch": 0.3844073987859688,
      "grad_norm": 1.6415305137634277,
      "learning_rate": 0.00014104595884081113,
      "loss": 1.5481,
      "step": 10734
    },
    {
      "epoch": 0.38444321091553707,
      "grad_norm": 1.8244068622589111,
      "learning_rate": 0.00014103538167085247,
      "loss": 1.5552,
      "step": 10735
    },
    {
      "epoch": 0.3844790230451054,
      "grad_norm": 1.8771268129348755,
      "learning_rate": 0.0001410248039488157,
      "loss": 1.4309,
      "step": 10736
    },
    {
      "epoch": 0.38451483517467366,
      "grad_norm": 1.5982030630111694,
      "learning_rate": 0.0001410142256748431,
      "loss": 1.4636,
      "step": 10737
    },
    {
      "epoch": 0.38455064730424193,
      "grad_norm": 1.3975290060043335,
      "learning_rate": 0.00014100364684907702,
      "loss": 1.3491,
      "step": 10738
    },
    {
      "epoch": 0.38458645943381026,
      "grad_norm": 1.9042540788650513,
      "learning_rate": 0.00014099306747165975,
      "loss": 1.2723,
      "step": 10739
    },
    {
      "epoch": 0.3846222715633785,
      "grad_norm": 1.3731106519699097,
      "learning_rate": 0.00014098248754273364,
      "loss": 1.757,
      "step": 10740
    },
    {
      "epoch": 0.3846580836929468,
      "grad_norm": 2.2348291873931885,
      "learning_rate": 0.000140971907062441,
      "loss": 1.2937,
      "step": 10741
    },
    {
      "epoch": 0.38469389582251506,
      "grad_norm": 1.5206588506698608,
      "learning_rate": 0.0001409613260309242,
      "loss": 1.3901,
      "step": 10742
    },
    {
      "epoch": 0.3847297079520834,
      "grad_norm": 1.705182671546936,
      "learning_rate": 0.00014095074444832561,
      "loss": 1.5864,
      "step": 10743
    },
    {
      "epoch": 0.38476552008165166,
      "grad_norm": 1.3100666999816895,
      "learning_rate": 0.0001409401623147876,
      "loss": 1.5685,
      "step": 10744
    },
    {
      "epoch": 0.3848013322112199,
      "grad_norm": 2.2978193759918213,
      "learning_rate": 0.00014092957963045245,
      "loss": 1.895,
      "step": 10745
    },
    {
      "epoch": 0.38483714434078825,
      "grad_norm": 1.4124058485031128,
      "learning_rate": 0.00014091899639546263,
      "loss": 1.0143,
      "step": 10746
    },
    {
      "epoch": 0.3848729564703565,
      "grad_norm": 1.691389799118042,
      "learning_rate": 0.00014090841260996055,
      "loss": 1.596,
      "step": 10747
    },
    {
      "epoch": 0.3849087685999248,
      "grad_norm": 1.6830295324325562,
      "learning_rate": 0.0001408978282740885,
      "loss": 1.3329,
      "step": 10748
    },
    {
      "epoch": 0.38494458072949306,
      "grad_norm": 1.5190984010696411,
      "learning_rate": 0.0001408872433879889,
      "loss": 1.4487,
      "step": 10749
    },
    {
      "epoch": 0.3849803928590614,
      "grad_norm": 1.6178358793258667,
      "learning_rate": 0.00014087665795180422,
      "loss": 1.4554,
      "step": 10750
    },
    {
      "epoch": 0.38501620498862965,
      "grad_norm": 1.344931721687317,
      "learning_rate": 0.00014086607196567682,
      "loss": 1.5279,
      "step": 10751
    },
    {
      "epoch": 0.3850520171181979,
      "grad_norm": 1.3446136713027954,
      "learning_rate": 0.00014085548542974914,
      "loss": 1.4228,
      "step": 10752
    },
    {
      "epoch": 0.38508782924776624,
      "grad_norm": 1.3284231424331665,
      "learning_rate": 0.0001408448983441636,
      "loss": 1.2936,
      "step": 10753
    },
    {
      "epoch": 0.3851236413773345,
      "grad_norm": 1.6570699214935303,
      "learning_rate": 0.00014083431070906262,
      "loss": 1.6028,
      "step": 10754
    },
    {
      "epoch": 0.3851594535069028,
      "grad_norm": 1.6138721704483032,
      "learning_rate": 0.0001408237225245887,
      "loss": 1.4631,
      "step": 10755
    },
    {
      "epoch": 0.38519526563647105,
      "grad_norm": 1.7377655506134033,
      "learning_rate": 0.00014081313379088424,
      "loss": 1.5583,
      "step": 10756
    },
    {
      "epoch": 0.3852310777660394,
      "grad_norm": 1.4762719869613647,
      "learning_rate": 0.0001408025445080917,
      "loss": 1.4366,
      "step": 10757
    },
    {
      "epoch": 0.38526688989560764,
      "grad_norm": 1.4237333536148071,
      "learning_rate": 0.00014079195467635354,
      "loss": 1.3627,
      "step": 10758
    },
    {
      "epoch": 0.3853027020251759,
      "grad_norm": 1.8359794616699219,
      "learning_rate": 0.00014078136429581227,
      "loss": 1.5827,
      "step": 10759
    },
    {
      "epoch": 0.38533851415474424,
      "grad_norm": 1.5892670154571533,
      "learning_rate": 0.00014077077336661036,
      "loss": 1.2771,
      "step": 10760
    },
    {
      "epoch": 0.3853743262843125,
      "grad_norm": 2.054046392440796,
      "learning_rate": 0.00014076018188889026,
      "loss": 1.5361,
      "step": 10761
    },
    {
      "epoch": 0.3854101384138808,
      "grad_norm": 1.899831771850586,
      "learning_rate": 0.0001407495898627945,
      "loss": 1.6762,
      "step": 10762
    },
    {
      "epoch": 0.38544595054344905,
      "grad_norm": 1.9101499319076538,
      "learning_rate": 0.00014073899728846555,
      "loss": 1.5444,
      "step": 10763
    },
    {
      "epoch": 0.38548176267301737,
      "grad_norm": 1.364566445350647,
      "learning_rate": 0.00014072840416604597,
      "loss": 1.4441,
      "step": 10764
    },
    {
      "epoch": 0.38551757480258564,
      "grad_norm": 1.467511773109436,
      "learning_rate": 0.00014071781049567825,
      "loss": 1.5548,
      "step": 10765
    },
    {
      "epoch": 0.3855533869321539,
      "grad_norm": 1.9000247716903687,
      "learning_rate": 0.0001407072162775049,
      "loss": 1.6866,
      "step": 10766
    },
    {
      "epoch": 0.38558919906172223,
      "grad_norm": 1.3294975757598877,
      "learning_rate": 0.00014069662151166846,
      "loss": 1.6307,
      "step": 10767
    },
    {
      "epoch": 0.3856250111912905,
      "grad_norm": 1.628848910331726,
      "learning_rate": 0.00014068602619831148,
      "loss": 1.4484,
      "step": 10768
    },
    {
      "epoch": 0.38566082332085877,
      "grad_norm": 1.6399376392364502,
      "learning_rate": 0.0001406754303375765,
      "loss": 1.7626,
      "step": 10769
    },
    {
      "epoch": 0.38569663545042704,
      "grad_norm": 1.6069952249526978,
      "learning_rate": 0.00014066483392960604,
      "loss": 1.5228,
      "step": 10770
    },
    {
      "epoch": 0.38573244757999536,
      "grad_norm": 2.2001421451568604,
      "learning_rate": 0.00014065423697454273,
      "loss": 1.5175,
      "step": 10771
    },
    {
      "epoch": 0.38576825970956363,
      "grad_norm": 1.3679170608520508,
      "learning_rate": 0.0001406436394725291,
      "loss": 1.4571,
      "step": 10772
    },
    {
      "epoch": 0.3858040718391319,
      "grad_norm": 2.004768133163452,
      "learning_rate": 0.00014063304142370773,
      "loss": 1.1387,
      "step": 10773
    },
    {
      "epoch": 0.3858398839687002,
      "grad_norm": 2.107898235321045,
      "learning_rate": 0.0001406224428282212,
      "loss": 1.7398,
      "step": 10774
    },
    {
      "epoch": 0.3858756960982685,
      "grad_norm": 1.65401029586792,
      "learning_rate": 0.0001406118436862121,
      "loss": 1.4425,
      "step": 10775
    },
    {
      "epoch": 0.38591150822783676,
      "grad_norm": 2.257986307144165,
      "learning_rate": 0.000140601243997823,
      "loss": 1.7269,
      "step": 10776
    },
    {
      "epoch": 0.38594732035740503,
      "grad_norm": 1.4186952114105225,
      "learning_rate": 0.00014059064376319657,
      "loss": 1.647,
      "step": 10777
    },
    {
      "epoch": 0.38598313248697336,
      "grad_norm": 1.5061075687408447,
      "learning_rate": 0.00014058004298247537,
      "loss": 1.4102,
      "step": 10778
    },
    {
      "epoch": 0.3860189446165416,
      "grad_norm": 2.0114200115203857,
      "learning_rate": 0.00014056944165580202,
      "loss": 1.763,
      "step": 10779
    },
    {
      "epoch": 0.3860547567461099,
      "grad_norm": 1.347678780555725,
      "learning_rate": 0.00014055883978331916,
      "loss": 1.5323,
      "step": 10780
    },
    {
      "epoch": 0.3860905688756782,
      "grad_norm": 1.6494736671447754,
      "learning_rate": 0.00014054823736516945,
      "loss": 1.7314,
      "step": 10781
    },
    {
      "epoch": 0.3861263810052465,
      "grad_norm": 1.7377164363861084,
      "learning_rate": 0.00014053763440149552,
      "loss": 1.4244,
      "step": 10782
    },
    {
      "epoch": 0.38616219313481476,
      "grad_norm": 2.1563901901245117,
      "learning_rate": 0.00014052703089244,
      "loss": 1.5343,
      "step": 10783
    },
    {
      "epoch": 0.386198005264383,
      "grad_norm": 1.365647315979004,
      "learning_rate": 0.00014051642683814557,
      "loss": 1.491,
      "step": 10784
    },
    {
      "epoch": 0.38623381739395135,
      "grad_norm": 1.7293301820755005,
      "learning_rate": 0.00014050582223875484,
      "loss": 1.5437,
      "step": 10785
    },
    {
      "epoch": 0.3862696295235196,
      "grad_norm": 2.316777229309082,
      "learning_rate": 0.00014049521709441057,
      "loss": 1.612,
      "step": 10786
    },
    {
      "epoch": 0.3863054416530879,
      "grad_norm": 1.4455130100250244,
      "learning_rate": 0.00014048461140525533,
      "loss": 1.5892,
      "step": 10787
    },
    {
      "epoch": 0.3863412537826562,
      "grad_norm": 1.716500997543335,
      "learning_rate": 0.0001404740051714319,
      "loss": 1.5128,
      "step": 10788
    },
    {
      "epoch": 0.3863770659122245,
      "grad_norm": 1.4911115169525146,
      "learning_rate": 0.00014046339839308294,
      "loss": 1.6088,
      "step": 10789
    },
    {
      "epoch": 0.38641287804179275,
      "grad_norm": 2.029979944229126,
      "learning_rate": 0.00014045279107035116,
      "loss": 1.488,
      "step": 10790
    },
    {
      "epoch": 0.386448690171361,
      "grad_norm": 1.7694584131240845,
      "learning_rate": 0.00014044218320337923,
      "loss": 1.5967,
      "step": 10791
    },
    {
      "epoch": 0.38648450230092934,
      "grad_norm": 1.7229918241500854,
      "learning_rate": 0.00014043157479230988,
      "loss": 1.4873,
      "step": 10792
    },
    {
      "epoch": 0.3865203144304976,
      "grad_norm": 1.647547721862793,
      "learning_rate": 0.00014042096583728587,
      "loss": 1.496,
      "step": 10793
    },
    {
      "epoch": 0.3865561265600659,
      "grad_norm": 1.3134502172470093,
      "learning_rate": 0.0001404103563384499,
      "loss": 1.3876,
      "step": 10794
    },
    {
      "epoch": 0.38659193868963415,
      "grad_norm": 1.5994287729263306,
      "learning_rate": 0.00014039974629594473,
      "loss": 1.7171,
      "step": 10795
    },
    {
      "epoch": 0.3866277508192025,
      "grad_norm": 1.576795220375061,
      "learning_rate": 0.00014038913570991302,
      "loss": 1.3701,
      "step": 10796
    },
    {
      "epoch": 0.38666356294877074,
      "grad_norm": 1.9416791200637817,
      "learning_rate": 0.00014037852458049764,
      "loss": 1.4239,
      "step": 10797
    },
    {
      "epoch": 0.386699375078339,
      "grad_norm": 1.8116300106048584,
      "learning_rate": 0.0001403679129078413,
      "loss": 1.5017,
      "step": 10798
    },
    {
      "epoch": 0.38673518720790734,
      "grad_norm": 1.8027644157409668,
      "learning_rate": 0.00014035730069208676,
      "loss": 1.4818,
      "step": 10799
    },
    {
      "epoch": 0.3867709993374756,
      "grad_norm": 1.8859972953796387,
      "learning_rate": 0.0001403466879333768,
      "loss": 1.5649,
      "step": 10800
    },
    {
      "epoch": 0.3868068114670439,
      "grad_norm": 1.517880916595459,
      "learning_rate": 0.00014033607463185416,
      "loss": 1.2432,
      "step": 10801
    },
    {
      "epoch": 0.38684262359661215,
      "grad_norm": 1.304983377456665,
      "learning_rate": 0.0001403254607876617,
      "loss": 1.5928,
      "step": 10802
    },
    {
      "epoch": 0.38687843572618047,
      "grad_norm": 2.295943021774292,
      "learning_rate": 0.00014031484640094217,
      "loss": 1.9763,
      "step": 10803
    },
    {
      "epoch": 0.38691424785574874,
      "grad_norm": 1.5124098062515259,
      "learning_rate": 0.0001403042314718384,
      "loss": 1.5043,
      "step": 10804
    },
    {
      "epoch": 0.386950059985317,
      "grad_norm": 1.4850661754608154,
      "learning_rate": 0.00014029361600049315,
      "loss": 1.3102,
      "step": 10805
    },
    {
      "epoch": 0.38698587211488533,
      "grad_norm": 1.8223068714141846,
      "learning_rate": 0.0001402829999870493,
      "loss": 1.2721,
      "step": 10806
    },
    {
      "epoch": 0.3870216842444536,
      "grad_norm": 1.4435514211654663,
      "learning_rate": 0.00014027238343164965,
      "loss": 1.4753,
      "step": 10807
    },
    {
      "epoch": 0.38705749637402187,
      "grad_norm": 1.5776134729385376,
      "learning_rate": 0.000140261766334437,
      "loss": 1.7435,
      "step": 10808
    },
    {
      "epoch": 0.38709330850359014,
      "grad_norm": 1.5881763696670532,
      "learning_rate": 0.00014025114869555425,
      "loss": 1.7955,
      "step": 10809
    },
    {
      "epoch": 0.38712912063315846,
      "grad_norm": 1.520105242729187,
      "learning_rate": 0.00014024053051514418,
      "loss": 1.7156,
      "step": 10810
    },
    {
      "epoch": 0.38716493276272673,
      "grad_norm": 2.3292229175567627,
      "learning_rate": 0.00014022991179334971,
      "loss": 1.5198,
      "step": 10811
    },
    {
      "epoch": 0.387200744892295,
      "grad_norm": 1.6294552087783813,
      "learning_rate": 0.00014021929253031366,
      "loss": 1.3763,
      "step": 10812
    },
    {
      "epoch": 0.3872365570218633,
      "grad_norm": 1.838181972503662,
      "learning_rate": 0.0001402086727261789,
      "loss": 1.7643,
      "step": 10813
    },
    {
      "epoch": 0.3872723691514316,
      "grad_norm": 3.064434051513672,
      "learning_rate": 0.0001401980523810883,
      "loss": 1.6755,
      "step": 10814
    },
    {
      "epoch": 0.38730818128099986,
      "grad_norm": 1.7557802200317383,
      "learning_rate": 0.0001401874314951848,
      "loss": 1.4726,
      "step": 10815
    },
    {
      "epoch": 0.38734399341056813,
      "grad_norm": 1.710260033607483,
      "learning_rate": 0.0001401768100686112,
      "loss": 1.6089,
      "step": 10816
    },
    {
      "epoch": 0.38737980554013646,
      "grad_norm": 1.567458152770996,
      "learning_rate": 0.00014016618810151047,
      "loss": 1.5403,
      "step": 10817
    },
    {
      "epoch": 0.3874156176697047,
      "grad_norm": 1.4473252296447754,
      "learning_rate": 0.00014015556559402551,
      "loss": 1.2182,
      "step": 10818
    },
    {
      "epoch": 0.387451429799273,
      "grad_norm": 1.6443681716918945,
      "learning_rate": 0.0001401449425462992,
      "loss": 1.5927,
      "step": 10819
    },
    {
      "epoch": 0.3874872419288413,
      "grad_norm": 1.7293095588684082,
      "learning_rate": 0.00014013431895847447,
      "loss": 1.4499,
      "step": 10820
    },
    {
      "epoch": 0.3875230540584096,
      "grad_norm": 1.4922728538513184,
      "learning_rate": 0.0001401236948306942,
      "loss": 1.5064,
      "step": 10821
    },
    {
      "epoch": 0.38755886618797786,
      "grad_norm": 1.9737058877944946,
      "learning_rate": 0.00014011307016310144,
      "loss": 1.3372,
      "step": 10822
    },
    {
      "epoch": 0.3875946783175461,
      "grad_norm": 3.1489598751068115,
      "learning_rate": 0.00014010244495583901,
      "loss": 1.6959,
      "step": 10823
    },
    {
      "epoch": 0.38763049044711445,
      "grad_norm": 2.1126914024353027,
      "learning_rate": 0.00014009181920904995,
      "loss": 1.5703,
      "step": 10824
    },
    {
      "epoch": 0.3876663025766827,
      "grad_norm": 2.1052439212799072,
      "learning_rate": 0.00014008119292287715,
      "loss": 1.8133,
      "step": 10825
    },
    {
      "epoch": 0.387702114706251,
      "grad_norm": 1.6701992750167847,
      "learning_rate": 0.00014007056609746362,
      "loss": 1.4596,
      "step": 10826
    },
    {
      "epoch": 0.3877379268358193,
      "grad_norm": 1.8624612092971802,
      "learning_rate": 0.00014005993873295234,
      "loss": 1.4561,
      "step": 10827
    },
    {
      "epoch": 0.3877737389653876,
      "grad_norm": 1.4672082662582397,
      "learning_rate": 0.0001400493108294862,
      "loss": 1.5657,
      "step": 10828
    },
    {
      "epoch": 0.38780955109495585,
      "grad_norm": 2.454993486404419,
      "learning_rate": 0.00014003868238720828,
      "loss": 1.7523,
      "step": 10829
    },
    {
      "epoch": 0.3878453632245241,
      "grad_norm": 2.431044578552246,
      "learning_rate": 0.0001400280534062615,
      "loss": 1.798,
      "step": 10830
    },
    {
      "epoch": 0.38788117535409244,
      "grad_norm": 1.4329546689987183,
      "learning_rate": 0.0001400174238867889,
      "loss": 1.3742,
      "step": 10831
    },
    {
      "epoch": 0.3879169874836607,
      "grad_norm": 1.5714000463485718,
      "learning_rate": 0.00014000679382893352,
      "loss": 1.4247,
      "step": 10832
    },
    {
      "epoch": 0.387952799613229,
      "grad_norm": 1.4519602060317993,
      "learning_rate": 0.0001399961632328383,
      "loss": 1.6319,
      "step": 10833
    },
    {
      "epoch": 0.3879886117427973,
      "grad_norm": 1.8219505548477173,
      "learning_rate": 0.00013998553209864628,
      "loss": 1.6325,
      "step": 10834
    },
    {
      "epoch": 0.3880244238723656,
      "grad_norm": 1.3136101961135864,
      "learning_rate": 0.00013997490042650054,
      "loss": 1.2345,
      "step": 10835
    },
    {
      "epoch": 0.38806023600193384,
      "grad_norm": 2.3637969493865967,
      "learning_rate": 0.00013996426821654407,
      "loss": 1.588,
      "step": 10836
    },
    {
      "epoch": 0.3880960481315021,
      "grad_norm": 1.519654631614685,
      "learning_rate": 0.00013995363546891992,
      "loss": 1.6943,
      "step": 10837
    },
    {
      "epoch": 0.38813186026107044,
      "grad_norm": 1.674434781074524,
      "learning_rate": 0.00013994300218377113,
      "loss": 1.9109,
      "step": 10838
    },
    {
      "epoch": 0.3881676723906387,
      "grad_norm": 1.8324869871139526,
      "learning_rate": 0.0001399323683612408,
      "loss": 1.5401,
      "step": 10839
    },
    {
      "epoch": 0.388203484520207,
      "grad_norm": 1.7333762645721436,
      "learning_rate": 0.00013992173400147193,
      "loss": 1.3558,
      "step": 10840
    },
    {
      "epoch": 0.3882392966497753,
      "grad_norm": 1.3128198385238647,
      "learning_rate": 0.00013991109910460763,
      "loss": 1.4118,
      "step": 10841
    },
    {
      "epoch": 0.38827510877934357,
      "grad_norm": 1.262176752090454,
      "learning_rate": 0.00013990046367079098,
      "loss": 1.5413,
      "step": 10842
    },
    {
      "epoch": 0.38831092090891184,
      "grad_norm": 1.3987606763839722,
      "learning_rate": 0.00013988982770016505,
      "loss": 1.4809,
      "step": 10843
    },
    {
      "epoch": 0.3883467330384801,
      "grad_norm": 1.460222601890564,
      "learning_rate": 0.00013987919119287296,
      "loss": 1.6431,
      "step": 10844
    },
    {
      "epoch": 0.38838254516804843,
      "grad_norm": 1.888704776763916,
      "learning_rate": 0.00013986855414905777,
      "loss": 1.3711,
      "step": 10845
    },
    {
      "epoch": 0.3884183572976167,
      "grad_norm": 1.4014217853546143,
      "learning_rate": 0.00013985791656886262,
      "loss": 1.457,
      "step": 10846
    },
    {
      "epoch": 0.38845416942718497,
      "grad_norm": 2.005265951156616,
      "learning_rate": 0.00013984727845243062,
      "loss": 1.5159,
      "step": 10847
    },
    {
      "epoch": 0.3884899815567533,
      "grad_norm": 1.5637860298156738,
      "learning_rate": 0.00013983663979990488,
      "loss": 1.4922,
      "step": 10848
    },
    {
      "epoch": 0.38852579368632156,
      "grad_norm": 2.402989387512207,
      "learning_rate": 0.00013982600061142854,
      "loss": 1.4786,
      "step": 10849
    },
    {
      "epoch": 0.38856160581588983,
      "grad_norm": 1.529707670211792,
      "learning_rate": 0.00013981536088714474,
      "loss": 1.4036,
      "step": 10850
    },
    {
      "epoch": 0.3885974179454581,
      "grad_norm": 1.7621464729309082,
      "learning_rate": 0.0001398047206271966,
      "loss": 1.474,
      "step": 10851
    },
    {
      "epoch": 0.3886332300750264,
      "grad_norm": 1.6757811307907104,
      "learning_rate": 0.00013979407983172733,
      "loss": 1.3865,
      "step": 10852
    },
    {
      "epoch": 0.3886690422045947,
      "grad_norm": 1.592047929763794,
      "learning_rate": 0.00013978343850088002,
      "loss": 1.3851,
      "step": 10853
    },
    {
      "epoch": 0.38870485433416296,
      "grad_norm": 1.6998521089553833,
      "learning_rate": 0.00013977279663479784,
      "loss": 1.3954,
      "step": 10854
    },
    {
      "epoch": 0.3887406664637313,
      "grad_norm": 1.5489436388015747,
      "learning_rate": 0.000139762154233624,
      "loss": 1.4166,
      "step": 10855
    },
    {
      "epoch": 0.38877647859329956,
      "grad_norm": 1.9484127759933472,
      "learning_rate": 0.00013975151129750168,
      "loss": 1.6642,
      "step": 10856
    },
    {
      "epoch": 0.3888122907228678,
      "grad_norm": 1.5500437021255493,
      "learning_rate": 0.00013974086782657404,
      "loss": 1.6083,
      "step": 10857
    },
    {
      "epoch": 0.3888481028524361,
      "grad_norm": 1.78920316696167,
      "learning_rate": 0.00013973022382098428,
      "loss": 1.2267,
      "step": 10858
    },
    {
      "epoch": 0.3888839149820044,
      "grad_norm": 1.2382129430770874,
      "learning_rate": 0.0001397195792808756,
      "loss": 1.5163,
      "step": 10859
    },
    {
      "epoch": 0.3889197271115727,
      "grad_norm": 1.68257737159729,
      "learning_rate": 0.00013970893420639123,
      "loss": 1.6837,
      "step": 10860
    },
    {
      "epoch": 0.38895553924114096,
      "grad_norm": 1.7023179531097412,
      "learning_rate": 0.00013969828859767438,
      "loss": 1.3082,
      "step": 10861
    },
    {
      "epoch": 0.3889913513707093,
      "grad_norm": 1.9222726821899414,
      "learning_rate": 0.00013968764245486824,
      "loss": 1.2721,
      "step": 10862
    },
    {
      "epoch": 0.38902716350027755,
      "grad_norm": 1.7391574382781982,
      "learning_rate": 0.0001396769957781161,
      "loss": 1.5351,
      "step": 10863
    },
    {
      "epoch": 0.3890629756298458,
      "grad_norm": 1.9035283327102661,
      "learning_rate": 0.00013966634856756114,
      "loss": 1.8143,
      "step": 10864
    },
    {
      "epoch": 0.3890987877594141,
      "grad_norm": 1.817996621131897,
      "learning_rate": 0.0001396557008233466,
      "loss": 1.4709,
      "step": 10865
    },
    {
      "epoch": 0.3891345998889824,
      "grad_norm": 1.6776641607284546,
      "learning_rate": 0.0001396450525456158,
      "loss": 1.8253,
      "step": 10866
    },
    {
      "epoch": 0.3891704120185507,
      "grad_norm": 1.5271328687667847,
      "learning_rate": 0.0001396344037345119,
      "loss": 1.4313,
      "step": 10867
    },
    {
      "epoch": 0.38920622414811895,
      "grad_norm": 2.033863067626953,
      "learning_rate": 0.0001396237543901783,
      "loss": 1.3897,
      "step": 10868
    },
    {
      "epoch": 0.3892420362776873,
      "grad_norm": 2.935681104660034,
      "learning_rate": 0.00013961310451275814,
      "loss": 1.6515,
      "step": 10869
    },
    {
      "epoch": 0.38927784840725554,
      "grad_norm": 2.0245561599731445,
      "learning_rate": 0.00013960245410239478,
      "loss": 1.5185,
      "step": 10870
    },
    {
      "epoch": 0.3893136605368238,
      "grad_norm": 1.5656148195266724,
      "learning_rate": 0.00013959180315923148,
      "loss": 1.2929,
      "step": 10871
    },
    {
      "epoch": 0.3893494726663921,
      "grad_norm": 2.131190538406372,
      "learning_rate": 0.00013958115168341155,
      "loss": 1.2898,
      "step": 10872
    },
    {
      "epoch": 0.3893852847959604,
      "grad_norm": 1.8734376430511475,
      "learning_rate": 0.00013957049967507824,
      "loss": 1.292,
      "step": 10873
    },
    {
      "epoch": 0.3894210969255287,
      "grad_norm": 1.5268676280975342,
      "learning_rate": 0.00013955984713437492,
      "loss": 1.6088,
      "step": 10874
    },
    {
      "epoch": 0.38945690905509694,
      "grad_norm": 2.3420557975769043,
      "learning_rate": 0.00013954919406144488,
      "loss": 1.5742,
      "step": 10875
    },
    {
      "epoch": 0.38949272118466527,
      "grad_norm": 2.152472496032715,
      "learning_rate": 0.00013953854045643146,
      "loss": 1.7288,
      "step": 10876
    },
    {
      "epoch": 0.38952853331423354,
      "grad_norm": 1.4512126445770264,
      "learning_rate": 0.00013952788631947798,
      "loss": 1.5296,
      "step": 10877
    },
    {
      "epoch": 0.3895643454438018,
      "grad_norm": 1.953406810760498,
      "learning_rate": 0.00013951723165072776,
      "loss": 1.575,
      "step": 10878
    },
    {
      "epoch": 0.3896001575733701,
      "grad_norm": 1.3942762613296509,
      "learning_rate": 0.00013950657645032418,
      "loss": 1.5451,
      "step": 10879
    },
    {
      "epoch": 0.3896359697029384,
      "grad_norm": 2.523400068283081,
      "learning_rate": 0.0001394959207184106,
      "loss": 1.6235,
      "step": 10880
    },
    {
      "epoch": 0.38967178183250667,
      "grad_norm": 1.6096271276474,
      "learning_rate": 0.00013948526445513033,
      "loss": 1.6995,
      "step": 10881
    },
    {
      "epoch": 0.38970759396207494,
      "grad_norm": 2.4701058864593506,
      "learning_rate": 0.00013947460766062673,
      "loss": 1.5205,
      "step": 10882
    },
    {
      "epoch": 0.38974340609164326,
      "grad_norm": 1.6288639307022095,
      "learning_rate": 0.00013946395033504323,
      "loss": 1.3126,
      "step": 10883
    },
    {
      "epoch": 0.38977921822121153,
      "grad_norm": 1.5390440225601196,
      "learning_rate": 0.00013945329247852317,
      "loss": 1.6212,
      "step": 10884
    },
    {
      "epoch": 0.3898150303507798,
      "grad_norm": 2.1103153228759766,
      "learning_rate": 0.00013944263409120997,
      "loss": 1.2683,
      "step": 10885
    },
    {
      "epoch": 0.38985084248034807,
      "grad_norm": 1.4607430696487427,
      "learning_rate": 0.00013943197517324698,
      "loss": 1.768,
      "step": 10886
    },
    {
      "epoch": 0.3898866546099164,
      "grad_norm": 1.542883276939392,
      "learning_rate": 0.00013942131572477763,
      "loss": 1.5286,
      "step": 10887
    },
    {
      "epoch": 0.38992246673948466,
      "grad_norm": 1.6651703119277954,
      "learning_rate": 0.00013941065574594536,
      "loss": 1.5245,
      "step": 10888
    },
    {
      "epoch": 0.38995827886905293,
      "grad_norm": 2.000406503677368,
      "learning_rate": 0.0001393999952368935,
      "loss": 1.3287,
      "step": 10889
    },
    {
      "epoch": 0.38999409099862126,
      "grad_norm": 1.859923243522644,
      "learning_rate": 0.0001393893341977656,
      "loss": 1.5322,
      "step": 10890
    },
    {
      "epoch": 0.3900299031281895,
      "grad_norm": 1.680471658706665,
      "learning_rate": 0.00013937867262870494,
      "loss": 1.2547,
      "step": 10891
    },
    {
      "epoch": 0.3900657152577578,
      "grad_norm": 1.6399215459823608,
      "learning_rate": 0.00013936801052985508,
      "loss": 1.71,
      "step": 10892
    },
    {
      "epoch": 0.39010152738732606,
      "grad_norm": 1.9132766723632812,
      "learning_rate": 0.0001393573479013594,
      "loss": 1.5958,
      "step": 10893
    },
    {
      "epoch": 0.3901373395168944,
      "grad_norm": 1.6897841691970825,
      "learning_rate": 0.00013934668474336137,
      "loss": 1.7573,
      "step": 10894
    },
    {
      "epoch": 0.39017315164646266,
      "grad_norm": 1.609229564666748,
      "learning_rate": 0.00013933602105600446,
      "loss": 1.3524,
      "step": 10895
    },
    {
      "epoch": 0.3902089637760309,
      "grad_norm": 1.546349287033081,
      "learning_rate": 0.00013932535683943212,
      "loss": 1.3228,
      "step": 10896
    },
    {
      "epoch": 0.39024477590559925,
      "grad_norm": 1.713444709777832,
      "learning_rate": 0.00013931469209378788,
      "loss": 1.5714,
      "step": 10897
    },
    {
      "epoch": 0.3902805880351675,
      "grad_norm": 1.8013064861297607,
      "learning_rate": 0.0001393040268192151,
      "loss": 1.5539,
      "step": 10898
    },
    {
      "epoch": 0.3903164001647358,
      "grad_norm": 1.501839280128479,
      "learning_rate": 0.00013929336101585737,
      "loss": 1.5056,
      "step": 10899
    },
    {
      "epoch": 0.39035221229430406,
      "grad_norm": 1.4470911026000977,
      "learning_rate": 0.00013928269468385814,
      "loss": 1.3623,
      "step": 10900
    },
    {
      "epoch": 0.3903880244238724,
      "grad_norm": 1.782845377922058,
      "learning_rate": 0.00013927202782336093,
      "loss": 1.5582,
      "step": 10901
    },
    {
      "epoch": 0.39042383655344065,
      "grad_norm": 1.5116702318191528,
      "learning_rate": 0.0001392613604345092,
      "loss": 1.2386,
      "step": 10902
    },
    {
      "epoch": 0.3904596486830089,
      "grad_norm": 1.541406512260437,
      "learning_rate": 0.00013925069251744657,
      "loss": 1.5252,
      "step": 10903
    },
    {
      "epoch": 0.39049546081257724,
      "grad_norm": 1.852392315864563,
      "learning_rate": 0.0001392400240723165,
      "loss": 1.5734,
      "step": 10904
    },
    {
      "epoch": 0.3905312729421455,
      "grad_norm": 1.53135085105896,
      "learning_rate": 0.00013922935509926249,
      "loss": 1.5326,
      "step": 10905
    },
    {
      "epoch": 0.3905670850717138,
      "grad_norm": 1.9085524082183838,
      "learning_rate": 0.00013921868559842813,
      "loss": 1.7359,
      "step": 10906
    },
    {
      "epoch": 0.39060289720128205,
      "grad_norm": 1.9585771560668945,
      "learning_rate": 0.00013920801556995693,
      "loss": 1.4947,
      "step": 10907
    },
    {
      "epoch": 0.3906387093308504,
      "grad_norm": 2.3244411945343018,
      "learning_rate": 0.00013919734501399248,
      "loss": 1.4937,
      "step": 10908
    },
    {
      "epoch": 0.39067452146041864,
      "grad_norm": 2.101654052734375,
      "learning_rate": 0.0001391866739306783,
      "loss": 1.5937,
      "step": 10909
    },
    {
      "epoch": 0.3907103335899869,
      "grad_norm": 1.9944405555725098,
      "learning_rate": 0.00013917600232015798,
      "loss": 1.5416,
      "step": 10910
    },
    {
      "epoch": 0.39074614571955524,
      "grad_norm": 1.5313005447387695,
      "learning_rate": 0.00013916533018257506,
      "loss": 1.5824,
      "step": 10911
    },
    {
      "epoch": 0.3907819578491235,
      "grad_norm": 1.490833044052124,
      "learning_rate": 0.00013915465751807314,
      "loss": 1.5661,
      "step": 10912
    },
    {
      "epoch": 0.3908177699786918,
      "grad_norm": 1.355104923248291,
      "learning_rate": 0.00013914398432679582,
      "loss": 1.498,
      "step": 10913
    },
    {
      "epoch": 0.39085358210826004,
      "grad_norm": 1.4459413290023804,
      "learning_rate": 0.00013913331060888667,
      "loss": 1.0748,
      "step": 10914
    },
    {
      "epoch": 0.39088939423782837,
      "grad_norm": 1.934935212135315,
      "learning_rate": 0.00013912263636448936,
      "loss": 1.3628,
      "step": 10915
    },
    {
      "epoch": 0.39092520636739664,
      "grad_norm": 1.31533944606781,
      "learning_rate": 0.00013911196159374737,
      "loss": 1.476,
      "step": 10916
    },
    {
      "epoch": 0.3909610184969649,
      "grad_norm": 1.765876293182373,
      "learning_rate": 0.00013910128629680441,
      "loss": 1.4795,
      "step": 10917
    },
    {
      "epoch": 0.39099683062653323,
      "grad_norm": 1.559696912765503,
      "learning_rate": 0.0001390906104738041,
      "loss": 1.7247,
      "step": 10918
    },
    {
      "epoch": 0.3910326427561015,
      "grad_norm": 1.7019007205963135,
      "learning_rate": 0.00013907993412489003,
      "loss": 1.3482,
      "step": 10919
    },
    {
      "epoch": 0.39106845488566977,
      "grad_norm": 1.4443830251693726,
      "learning_rate": 0.00013906925725020586,
      "loss": 1.5142,
      "step": 10920
    },
    {
      "epoch": 0.39110426701523804,
      "grad_norm": 1.9812666177749634,
      "learning_rate": 0.00013905857984989524,
      "loss": 1.4914,
      "step": 10921
    },
    {
      "epoch": 0.39114007914480636,
      "grad_norm": 1.4861475229263306,
      "learning_rate": 0.00013904790192410178,
      "loss": 1.4392,
      "step": 10922
    },
    {
      "epoch": 0.39117589127437463,
      "grad_norm": 1.292574167251587,
      "learning_rate": 0.0001390372234729692,
      "loss": 1.2898,
      "step": 10923
    },
    {
      "epoch": 0.3912117034039429,
      "grad_norm": 1.654436469078064,
      "learning_rate": 0.00013902654449664115,
      "loss": 1.4982,
      "step": 10924
    },
    {
      "epoch": 0.3912475155335112,
      "grad_norm": 1.3484128713607788,
      "learning_rate": 0.00013901586499526125,
      "loss": 1.6477,
      "step": 10925
    },
    {
      "epoch": 0.3912833276630795,
      "grad_norm": 1.577406883239746,
      "learning_rate": 0.0001390051849689732,
      "loss": 1.5074,
      "step": 10926
    },
    {
      "epoch": 0.39131913979264776,
      "grad_norm": 1.7564611434936523,
      "learning_rate": 0.00013899450441792074,
      "loss": 0.9454,
      "step": 10927
    },
    {
      "epoch": 0.39135495192221603,
      "grad_norm": 1.9713129997253418,
      "learning_rate": 0.00013898382334224748,
      "loss": 1.3936,
      "step": 10928
    },
    {
      "epoch": 0.39139076405178436,
      "grad_norm": 1.8760312795639038,
      "learning_rate": 0.0001389731417420972,
      "loss": 1.5234,
      "step": 10929
    },
    {
      "epoch": 0.3914265761813526,
      "grad_norm": 1.71038019657135,
      "learning_rate": 0.00013896245961761354,
      "loss": 1.4217,
      "step": 10930
    },
    {
      "epoch": 0.3914623883109209,
      "grad_norm": 1.5973619222640991,
      "learning_rate": 0.00013895177696894023,
      "loss": 1.3573,
      "step": 10931
    },
    {
      "epoch": 0.3914982004404892,
      "grad_norm": 1.6818480491638184,
      "learning_rate": 0.00013894109379622104,
      "loss": 1.3301,
      "step": 10932
    },
    {
      "epoch": 0.3915340125700575,
      "grad_norm": 1.7403838634490967,
      "learning_rate": 0.00013893041009959968,
      "loss": 1.5322,
      "step": 10933
    },
    {
      "epoch": 0.39156982469962576,
      "grad_norm": 1.5131548643112183,
      "learning_rate": 0.00013891972587921987,
      "loss": 1.6662,
      "step": 10934
    },
    {
      "epoch": 0.391605636829194,
      "grad_norm": 1.3520056009292603,
      "learning_rate": 0.0001389090411352253,
      "loss": 1.5041,
      "step": 10935
    },
    {
      "epoch": 0.39164144895876235,
      "grad_norm": 1.8540492057800293,
      "learning_rate": 0.0001388983558677598,
      "loss": 1.3033,
      "step": 10936
    },
    {
      "epoch": 0.3916772610883306,
      "grad_norm": 2.1093995571136475,
      "learning_rate": 0.00013888767007696709,
      "loss": 1.2123,
      "step": 10937
    },
    {
      "epoch": 0.3917130732178989,
      "grad_norm": 1.8228912353515625,
      "learning_rate": 0.00013887698376299095,
      "loss": 1.5145,
      "step": 10938
    },
    {
      "epoch": 0.3917488853474672,
      "grad_norm": 2.670003652572632,
      "learning_rate": 0.00013886629692597512,
      "loss": 1.4614,
      "step": 10939
    },
    {
      "epoch": 0.3917846974770355,
      "grad_norm": 1.368762731552124,
      "learning_rate": 0.00013885560956606344,
      "loss": 1.6102,
      "step": 10940
    },
    {
      "epoch": 0.39182050960660375,
      "grad_norm": 1.714065670967102,
      "learning_rate": 0.00013884492168339963,
      "loss": 1.7329,
      "step": 10941
    },
    {
      "epoch": 0.391856321736172,
      "grad_norm": 1.4392451047897339,
      "learning_rate": 0.00013883423327812748,
      "loss": 1.3499,
      "step": 10942
    },
    {
      "epoch": 0.39189213386574034,
      "grad_norm": 1.5470072031021118,
      "learning_rate": 0.00013882354435039085,
      "loss": 1.1483,
      "step": 10943
    },
    {
      "epoch": 0.3919279459953086,
      "grad_norm": 1.6122442483901978,
      "learning_rate": 0.00013881285490033348,
      "loss": 1.3297,
      "step": 10944
    },
    {
      "epoch": 0.3919637581248769,
      "grad_norm": 1.8792885541915894,
      "learning_rate": 0.00013880216492809924,
      "loss": 1.7625,
      "step": 10945
    },
    {
      "epoch": 0.3919995702544452,
      "grad_norm": 1.8728435039520264,
      "learning_rate": 0.00013879147443383188,
      "loss": 1.2987,
      "step": 10946
    },
    {
      "epoch": 0.3920353823840135,
      "grad_norm": 1.4084367752075195,
      "learning_rate": 0.00013878078341767532,
      "loss": 1.3605,
      "step": 10947
    },
    {
      "epoch": 0.39207119451358174,
      "grad_norm": 2.4692366123199463,
      "learning_rate": 0.00013877009187977332,
      "loss": 1.8605,
      "step": 10948
    },
    {
      "epoch": 0.39210700664315,
      "grad_norm": 1.8761765956878662,
      "learning_rate": 0.00013875939982026976,
      "loss": 1.5695,
      "step": 10949
    },
    {
      "epoch": 0.39214281877271834,
      "grad_norm": 1.9340341091156006,
      "learning_rate": 0.00013874870723930847,
      "loss": 1.6881,
      "step": 10950
    },
    {
      "epoch": 0.3921786309022866,
      "grad_norm": 1.415968418121338,
      "learning_rate": 0.00013873801413703327,
      "loss": 1.4739,
      "step": 10951
    },
    {
      "epoch": 0.3922144430318549,
      "grad_norm": 1.7562787532806396,
      "learning_rate": 0.00013872732051358808,
      "loss": 1.7517,
      "step": 10952
    },
    {
      "epoch": 0.3922502551614232,
      "grad_norm": 1.4888852834701538,
      "learning_rate": 0.00013871662636911672,
      "loss": 1.5784,
      "step": 10953
    },
    {
      "epoch": 0.39228606729099147,
      "grad_norm": 1.862896203994751,
      "learning_rate": 0.00013870593170376317,
      "loss": 1.4573,
      "step": 10954
    },
    {
      "epoch": 0.39232187942055974,
      "grad_norm": 1.7131226062774658,
      "learning_rate": 0.00013869523651767116,
      "loss": 1.8789,
      "step": 10955
    },
    {
      "epoch": 0.392357691550128,
      "grad_norm": 1.4080753326416016,
      "learning_rate": 0.0001386845408109847,
      "loss": 1.4324,
      "step": 10956
    },
    {
      "epoch": 0.39239350367969633,
      "grad_norm": 1.5288814306259155,
      "learning_rate": 0.0001386738445838476,
      "loss": 1.4847,
      "step": 10957
    },
    {
      "epoch": 0.3924293158092646,
      "grad_norm": 1.7750880718231201,
      "learning_rate": 0.00013866314783640384,
      "loss": 1.6698,
      "step": 10958
    },
    {
      "epoch": 0.39246512793883287,
      "grad_norm": 1.8281431198120117,
      "learning_rate": 0.0001386524505687973,
      "loss": 1.4908,
      "step": 10959
    },
    {
      "epoch": 0.3925009400684012,
      "grad_norm": 1.838854432106018,
      "learning_rate": 0.00013864175278117187,
      "loss": 1.4854,
      "step": 10960
    },
    {
      "epoch": 0.39253675219796946,
      "grad_norm": 1.9536410570144653,
      "learning_rate": 0.00013863105447367154,
      "loss": 1.6562,
      "step": 10961
    },
    {
      "epoch": 0.39257256432753773,
      "grad_norm": 2.2680394649505615,
      "learning_rate": 0.00013862035564644017,
      "loss": 1.5296,
      "step": 10962
    },
    {
      "epoch": 0.392608376457106,
      "grad_norm": 1.7162909507751465,
      "learning_rate": 0.00013860965629962176,
      "loss": 1.8054,
      "step": 10963
    },
    {
      "epoch": 0.3926441885866743,
      "grad_norm": 1.5176362991333008,
      "learning_rate": 0.0001385989564333602,
      "loss": 1.6443,
      "step": 10964
    },
    {
      "epoch": 0.3926800007162426,
      "grad_norm": 1.4724088907241821,
      "learning_rate": 0.00013858825604779945,
      "loss": 1.7563,
      "step": 10965
    },
    {
      "epoch": 0.39271581284581086,
      "grad_norm": 2.1419851779937744,
      "learning_rate": 0.00013857755514308352,
      "loss": 1.6473,
      "step": 10966
    },
    {
      "epoch": 0.3927516249753792,
      "grad_norm": 1.5697284936904907,
      "learning_rate": 0.00013856685371935637,
      "loss": 1.694,
      "step": 10967
    },
    {
      "epoch": 0.39278743710494746,
      "grad_norm": 1.7596726417541504,
      "learning_rate": 0.00013855615177676191,
      "loss": 1.5056,
      "step": 10968
    },
    {
      "epoch": 0.3928232492345157,
      "grad_norm": 1.5144803524017334,
      "learning_rate": 0.0001385454493154442,
      "loss": 1.589,
      "step": 10969
    },
    {
      "epoch": 0.392859061364084,
      "grad_norm": 2.599090337753296,
      "learning_rate": 0.00013853474633554715,
      "loss": 1.4556,
      "step": 10970
    },
    {
      "epoch": 0.3928948734936523,
      "grad_norm": 1.5587658882141113,
      "learning_rate": 0.00013852404283721482,
      "loss": 1.5985,
      "step": 10971
    },
    {
      "epoch": 0.3929306856232206,
      "grad_norm": 1.9133366346359253,
      "learning_rate": 0.00013851333882059118,
      "loss": 1.2122,
      "step": 10972
    },
    {
      "epoch": 0.39296649775278886,
      "grad_norm": 2.5919742584228516,
      "learning_rate": 0.00013850263428582022,
      "loss": 1.6127,
      "step": 10973
    },
    {
      "epoch": 0.3930023098823572,
      "grad_norm": 1.6070665121078491,
      "learning_rate": 0.000138491929233046,
      "loss": 1.5863,
      "step": 10974
    },
    {
      "epoch": 0.39303812201192545,
      "grad_norm": 1.462077021598816,
      "learning_rate": 0.00013848122366241254,
      "loss": 1.6105,
      "step": 10975
    },
    {
      "epoch": 0.3930739341414937,
      "grad_norm": 2.1995644569396973,
      "learning_rate": 0.00013847051757406384,
      "loss": 1.6227,
      "step": 10976
    },
    {
      "epoch": 0.393109746271062,
      "grad_norm": 2.4004290103912354,
      "learning_rate": 0.00013845981096814397,
      "loss": 1.446,
      "step": 10977
    },
    {
      "epoch": 0.3931455584006303,
      "grad_norm": 1.6153998374938965,
      "learning_rate": 0.00013844910384479693,
      "loss": 1.5331,
      "step": 10978
    },
    {
      "epoch": 0.3931813705301986,
      "grad_norm": 1.8544167280197144,
      "learning_rate": 0.00013843839620416678,
      "loss": 1.8615,
      "step": 10979
    },
    {
      "epoch": 0.39321718265976685,
      "grad_norm": 1.9386411905288696,
      "learning_rate": 0.00013842768804639763,
      "loss": 1.3094,
      "step": 10980
    },
    {
      "epoch": 0.3932529947893352,
      "grad_norm": 1.922755479812622,
      "learning_rate": 0.00013841697937163344,
      "loss": 1.6329,
      "step": 10981
    },
    {
      "epoch": 0.39328880691890344,
      "grad_norm": 1.3943654298782349,
      "learning_rate": 0.0001384062701800184,
      "loss": 1.542,
      "step": 10982
    },
    {
      "epoch": 0.3933246190484717,
      "grad_norm": 1.7033149003982544,
      "learning_rate": 0.00013839556047169654,
      "loss": 1.6458,
      "step": 10983
    },
    {
      "epoch": 0.39336043117804,
      "grad_norm": 1.56367826461792,
      "learning_rate": 0.00013838485024681192,
      "loss": 1.6611,
      "step": 10984
    },
    {
      "epoch": 0.3933962433076083,
      "grad_norm": 1.4666606187820435,
      "learning_rate": 0.00013837413950550865,
      "loss": 1.7409,
      "step": 10985
    },
    {
      "epoch": 0.3934320554371766,
      "grad_norm": 1.5731333494186401,
      "learning_rate": 0.00013836342824793084,
      "loss": 0.9751,
      "step": 10986
    },
    {
      "epoch": 0.39346786756674484,
      "grad_norm": 1.75849449634552,
      "learning_rate": 0.00013835271647422262,
      "loss": 1.4298,
      "step": 10987
    },
    {
      "epoch": 0.39350367969631317,
      "grad_norm": 1.5057103633880615,
      "learning_rate": 0.000138342004184528,
      "loss": 1.4403,
      "step": 10988
    },
    {
      "epoch": 0.39353949182588144,
      "grad_norm": 1.6212468147277832,
      "learning_rate": 0.00013833129137899122,
      "loss": 1.2772,
      "step": 10989
    },
    {
      "epoch": 0.3935753039554497,
      "grad_norm": 3.099400758743286,
      "learning_rate": 0.00013832057805775636,
      "loss": 1.8536,
      "step": 10990
    },
    {
      "epoch": 0.393611116085018,
      "grad_norm": 1.6015698909759521,
      "learning_rate": 0.00013830986422096756,
      "loss": 1.5765,
      "step": 10991
    },
    {
      "epoch": 0.3936469282145863,
      "grad_norm": 1.789380669593811,
      "learning_rate": 0.0001382991498687689,
      "loss": 1.57,
      "step": 10992
    },
    {
      "epoch": 0.39368274034415457,
      "grad_norm": 1.9253126382827759,
      "learning_rate": 0.00013828843500130462,
      "loss": 1.7023,
      "step": 10993
    },
    {
      "epoch": 0.39371855247372284,
      "grad_norm": 1.3724284172058105,
      "learning_rate": 0.00013827771961871885,
      "loss": 1.3947,
      "step": 10994
    },
    {
      "epoch": 0.3937543646032911,
      "grad_norm": 1.8409966230392456,
      "learning_rate": 0.0001382670037211557,
      "loss": 1.3981,
      "step": 10995
    },
    {
      "epoch": 0.39379017673285943,
      "grad_norm": 1.4560490846633911,
      "learning_rate": 0.0001382562873087594,
      "loss": 1.5862,
      "step": 10996
    },
    {
      "epoch": 0.3938259888624277,
      "grad_norm": 2.6296908855438232,
      "learning_rate": 0.00013824557038167408,
      "loss": 1.4751,
      "step": 10997
    },
    {
      "epoch": 0.39386180099199597,
      "grad_norm": 1.4091401100158691,
      "learning_rate": 0.00013823485294004397,
      "loss": 1.2807,
      "step": 10998
    },
    {
      "epoch": 0.3938976131215643,
      "grad_norm": 2.486598253250122,
      "learning_rate": 0.00013822413498401322,
      "loss": 1.4194,
      "step": 10999
    },
    {
      "epoch": 0.39393342525113256,
      "grad_norm": 1.8565677404403687,
      "learning_rate": 0.00013821341651372603,
      "loss": 1.4072,
      "step": 11000
    },
    {
      "epoch": 0.39396923738070083,
      "grad_norm": 1.5582574605941772,
      "learning_rate": 0.00013820269752932662,
      "loss": 1.1239,
      "step": 11001
    },
    {
      "epoch": 0.3940050495102691,
      "grad_norm": 1.714348316192627,
      "learning_rate": 0.0001381919780309592,
      "loss": 1.5673,
      "step": 11002
    },
    {
      "epoch": 0.3940408616398374,
      "grad_norm": 2.184767723083496,
      "learning_rate": 0.000138181258018768,
      "loss": 1.8622,
      "step": 11003
    },
    {
      "epoch": 0.3940766737694057,
      "grad_norm": 1.768456220626831,
      "learning_rate": 0.00013817053749289718,
      "loss": 1.511,
      "step": 11004
    },
    {
      "epoch": 0.39411248589897396,
      "grad_norm": 1.9604367017745972,
      "learning_rate": 0.00013815981645349105,
      "loss": 1.5908,
      "step": 11005
    },
    {
      "epoch": 0.3941482980285423,
      "grad_norm": 1.6243501901626587,
      "learning_rate": 0.00013814909490069378,
      "loss": 1.4674,
      "step": 11006
    },
    {
      "epoch": 0.39418411015811056,
      "grad_norm": 1.532596468925476,
      "learning_rate": 0.00013813837283464968,
      "loss": 1.5244,
      "step": 11007
    },
    {
      "epoch": 0.3942199222876788,
      "grad_norm": 2.020131826400757,
      "learning_rate": 0.00013812765025550294,
      "loss": 1.3743,
      "step": 11008
    },
    {
      "epoch": 0.3942557344172471,
      "grad_norm": 1.766100287437439,
      "learning_rate": 0.00013811692716339785,
      "loss": 1.5782,
      "step": 11009
    },
    {
      "epoch": 0.3942915465468154,
      "grad_norm": 1.5012962818145752,
      "learning_rate": 0.00013810620355847868,
      "loss": 1.5708,
      "step": 11010
    },
    {
      "epoch": 0.3943273586763837,
      "grad_norm": 1.4849494695663452,
      "learning_rate": 0.00013809547944088968,
      "loss": 1.9724,
      "step": 11011
    },
    {
      "epoch": 0.39436317080595196,
      "grad_norm": 2.92488694190979,
      "learning_rate": 0.00013808475481077518,
      "loss": 1.7187,
      "step": 11012
    },
    {
      "epoch": 0.3943989829355203,
      "grad_norm": 1.5572278499603271,
      "learning_rate": 0.00013807402966827944,
      "loss": 1.4533,
      "step": 11013
    },
    {
      "epoch": 0.39443479506508855,
      "grad_norm": 1.4926722049713135,
      "learning_rate": 0.00013806330401354671,
      "loss": 1.3641,
      "step": 11014
    },
    {
      "epoch": 0.3944706071946568,
      "grad_norm": 1.7692962884902954,
      "learning_rate": 0.0001380525778467213,
      "loss": 1.3569,
      "step": 11015
    },
    {
      "epoch": 0.3945064193242251,
      "grad_norm": 1.5108805894851685,
      "learning_rate": 0.00013804185116794755,
      "loss": 1.4749,
      "step": 11016
    },
    {
      "epoch": 0.3945422314537934,
      "grad_norm": 2.093026876449585,
      "learning_rate": 0.00013803112397736976,
      "loss": 1.4546,
      "step": 11017
    },
    {
      "epoch": 0.3945780435833617,
      "grad_norm": 1.4690407514572144,
      "learning_rate": 0.0001380203962751323,
      "loss": 1.2591,
      "step": 11018
    },
    {
      "epoch": 0.39461385571292995,
      "grad_norm": 1.3096941709518433,
      "learning_rate": 0.0001380096680613794,
      "loss": 1.4565,
      "step": 11019
    },
    {
      "epoch": 0.3946496678424983,
      "grad_norm": 1.9637916088104248,
      "learning_rate": 0.00013799893933625547,
      "loss": 1.2479,
      "step": 11020
    },
    {
      "epoch": 0.39468547997206654,
      "grad_norm": 1.3697775602340698,
      "learning_rate": 0.00013798821009990486,
      "loss": 1.457,
      "step": 11021
    },
    {
      "epoch": 0.3947212921016348,
      "grad_norm": 1.6081949472427368,
      "learning_rate": 0.00013797748035247184,
      "loss": 1.509,
      "step": 11022
    },
    {
      "epoch": 0.3947571042312031,
      "grad_norm": 2.8257575035095215,
      "learning_rate": 0.0001379667500941008,
      "loss": 1.5442,
      "step": 11023
    },
    {
      "epoch": 0.3947929163607714,
      "grad_norm": 1.9925357103347778,
      "learning_rate": 0.00013795601932493613,
      "loss": 1.6229,
      "step": 11024
    },
    {
      "epoch": 0.3948287284903397,
      "grad_norm": 1.7090239524841309,
      "learning_rate": 0.0001379452880451222,
      "loss": 1.4409,
      "step": 11025
    },
    {
      "epoch": 0.39486454061990794,
      "grad_norm": 1.6960958242416382,
      "learning_rate": 0.00013793455625480332,
      "loss": 1.4845,
      "step": 11026
    },
    {
      "epoch": 0.39490035274947627,
      "grad_norm": 1.5618412494659424,
      "learning_rate": 0.00013792382395412392,
      "loss": 1.8159,
      "step": 11027
    },
    {
      "epoch": 0.39493616487904454,
      "grad_norm": 1.413428783416748,
      "learning_rate": 0.00013791309114322841,
      "loss": 1.2555,
      "step": 11028
    },
    {
      "epoch": 0.3949719770086128,
      "grad_norm": 2.023017644882202,
      "learning_rate": 0.00013790235782226118,
      "loss": 1.5708,
      "step": 11029
    },
    {
      "epoch": 0.3950077891381811,
      "grad_norm": 1.610162615776062,
      "learning_rate": 0.0001378916239913666,
      "loss": 1.7002,
      "step": 11030
    },
    {
      "epoch": 0.3950436012677494,
      "grad_norm": 1.5207340717315674,
      "learning_rate": 0.0001378808896506891,
      "loss": 1.4454,
      "step": 11031
    },
    {
      "epoch": 0.39507941339731767,
      "grad_norm": 1.6515626907348633,
      "learning_rate": 0.00013787015480037307,
      "loss": 1.5858,
      "step": 11032
    },
    {
      "epoch": 0.39511522552688594,
      "grad_norm": 2.073228359222412,
      "learning_rate": 0.00013785941944056298,
      "loss": 1.5202,
      "step": 11033
    },
    {
      "epoch": 0.39515103765645426,
      "grad_norm": 1.6093931198120117,
      "learning_rate": 0.00013784868357140322,
      "loss": 1.6465,
      "step": 11034
    },
    {
      "epoch": 0.39518684978602253,
      "grad_norm": 2.0961008071899414,
      "learning_rate": 0.00013783794719303825,
      "loss": 1.6145,
      "step": 11035
    },
    {
      "epoch": 0.3952226619155908,
      "grad_norm": 1.5990267992019653,
      "learning_rate": 0.0001378272103056125,
      "loss": 1.6946,
      "step": 11036
    },
    {
      "epoch": 0.39525847404515907,
      "grad_norm": 1.5354348421096802,
      "learning_rate": 0.00013781647290927044,
      "loss": 1.4329,
      "step": 11037
    },
    {
      "epoch": 0.3952942861747274,
      "grad_norm": 1.4117317199707031,
      "learning_rate": 0.00013780573500415654,
      "loss": 1.4226,
      "step": 11038
    },
    {
      "epoch": 0.39533009830429566,
      "grad_norm": 1.3799502849578857,
      "learning_rate": 0.0001377949965904152,
      "loss": 1.3842,
      "step": 11039
    },
    {
      "epoch": 0.39536591043386393,
      "grad_norm": 1.3154661655426025,
      "learning_rate": 0.00013778425766819096,
      "loss": 1.3348,
      "step": 11040
    },
    {
      "epoch": 0.39540172256343226,
      "grad_norm": 2.150657892227173,
      "learning_rate": 0.00013777351823762826,
      "loss": 1.652,
      "step": 11041
    },
    {
      "epoch": 0.3954375346930005,
      "grad_norm": 1.667708396911621,
      "learning_rate": 0.0001377627782988716,
      "loss": 1.3867,
      "step": 11042
    },
    {
      "epoch": 0.3954733468225688,
      "grad_norm": 1.616040825843811,
      "learning_rate": 0.00013775203785206544,
      "loss": 1.2286,
      "step": 11043
    },
    {
      "epoch": 0.39550915895213706,
      "grad_norm": 1.6211915016174316,
      "learning_rate": 0.00013774129689735437,
      "loss": 1.3339,
      "step": 11044
    },
    {
      "epoch": 0.3955449710817054,
      "grad_norm": 2.3818204402923584,
      "learning_rate": 0.00013773055543488276,
      "loss": 1.5299,
      "step": 11045
    },
    {
      "epoch": 0.39558078321127366,
      "grad_norm": 1.4598685503005981,
      "learning_rate": 0.00013771981346479524,
      "loss": 1.5545,
      "step": 11046
    },
    {
      "epoch": 0.3956165953408419,
      "grad_norm": 2.1078193187713623,
      "learning_rate": 0.0001377090709872363,
      "loss": 1.712,
      "step": 11047
    },
    {
      "epoch": 0.39565240747041025,
      "grad_norm": 1.702226161956787,
      "learning_rate": 0.00013769832800235041,
      "loss": 1.4261,
      "step": 11048
    },
    {
      "epoch": 0.3956882195999785,
      "grad_norm": 1.6927379369735718,
      "learning_rate": 0.00013768758451028216,
      "loss": 1.5355,
      "step": 11049
    },
    {
      "epoch": 0.3957240317295468,
      "grad_norm": 1.7418303489685059,
      "learning_rate": 0.00013767684051117605,
      "loss": 1.5623,
      "step": 11050
    },
    {
      "epoch": 0.39575984385911506,
      "grad_norm": 1.6147509813308716,
      "learning_rate": 0.0001376660960051767,
      "loss": 1.7547,
      "step": 11051
    },
    {
      "epoch": 0.3957956559886834,
      "grad_norm": 1.5473684072494507,
      "learning_rate": 0.0001376553509924286,
      "loss": 1.4403,
      "step": 11052
    },
    {
      "epoch": 0.39583146811825165,
      "grad_norm": 1.8391984701156616,
      "learning_rate": 0.00013764460547307632,
      "loss": 1.6435,
      "step": 11053
    },
    {
      "epoch": 0.3958672802478199,
      "grad_norm": 1.5264986753463745,
      "learning_rate": 0.0001376338594472644,
      "loss": 1.4671,
      "step": 11054
    },
    {
      "epoch": 0.39590309237738824,
      "grad_norm": 2.0358102321624756,
      "learning_rate": 0.00013762311291513747,
      "loss": 1.5083,
      "step": 11055
    },
    {
      "epoch": 0.3959389045069565,
      "grad_norm": 1.842419147491455,
      "learning_rate": 0.0001376123658768401,
      "loss": 1.5692,
      "step": 11056
    },
    {
      "epoch": 0.3959747166365248,
      "grad_norm": 1.4179729223251343,
      "learning_rate": 0.00013760161833251683,
      "loss": 1.4895,
      "step": 11057
    },
    {
      "epoch": 0.39601052876609305,
      "grad_norm": 1.3994336128234863,
      "learning_rate": 0.00013759087028231232,
      "loss": 1.3776,
      "step": 11058
    },
    {
      "epoch": 0.3960463408956614,
      "grad_norm": 2.0251388549804688,
      "learning_rate": 0.0001375801217263711,
      "loss": 1.6126,
      "step": 11059
    },
    {
      "epoch": 0.39608215302522964,
      "grad_norm": 1.455640435218811,
      "learning_rate": 0.00013756937266483788,
      "loss": 1.513,
      "step": 11060
    },
    {
      "epoch": 0.3961179651547979,
      "grad_norm": 1.6252317428588867,
      "learning_rate": 0.00013755862309785716,
      "loss": 1.6759,
      "step": 11061
    },
    {
      "epoch": 0.39615377728436624,
      "grad_norm": 2.397956609725952,
      "learning_rate": 0.00013754787302557364,
      "loss": 1.5216,
      "step": 11062
    },
    {
      "epoch": 0.3961895894139345,
      "grad_norm": 1.762626051902771,
      "learning_rate": 0.0001375371224481319,
      "loss": 1.7711,
      "step": 11063
    },
    {
      "epoch": 0.3962254015435028,
      "grad_norm": 1.7544844150543213,
      "learning_rate": 0.0001375263713656766,
      "loss": 1.3992,
      "step": 11064
    },
    {
      "epoch": 0.39626121367307104,
      "grad_norm": 1.8529858589172363,
      "learning_rate": 0.00013751561977835242,
      "loss": 1.9665,
      "step": 11065
    },
    {
      "epoch": 0.39629702580263937,
      "grad_norm": 1.527550220489502,
      "learning_rate": 0.00013750486768630393,
      "loss": 1.6083,
      "step": 11066
    },
    {
      "epoch": 0.39633283793220764,
      "grad_norm": 1.813584327697754,
      "learning_rate": 0.00013749411508967582,
      "loss": 1.5276,
      "step": 11067
    },
    {
      "epoch": 0.3963686500617759,
      "grad_norm": 2.098923921585083,
      "learning_rate": 0.0001374833619886128,
      "loss": 2.1238,
      "step": 11068
    },
    {
      "epoch": 0.39640446219134423,
      "grad_norm": 1.9573220014572144,
      "learning_rate": 0.0001374726083832594,
      "loss": 1.4054,
      "step": 11069
    },
    {
      "epoch": 0.3964402743209125,
      "grad_norm": 1.5510364770889282,
      "learning_rate": 0.00013746185427376047,
      "loss": 1.4817,
      "step": 11070
    },
    {
      "epoch": 0.39647608645048077,
      "grad_norm": 1.4360061883926392,
      "learning_rate": 0.00013745109966026056,
      "loss": 1.533,
      "step": 11071
    },
    {
      "epoch": 0.39651189858004904,
      "grad_norm": 1.4258822202682495,
      "learning_rate": 0.00013744034454290447,
      "loss": 1.6174,
      "step": 11072
    },
    {
      "epoch": 0.39654771070961736,
      "grad_norm": 1.630110740661621,
      "learning_rate": 0.0001374295889218368,
      "loss": 1.4781,
      "step": 11073
    },
    {
      "epoch": 0.39658352283918563,
      "grad_norm": 1.3876277208328247,
      "learning_rate": 0.0001374188327972023,
      "loss": 1.4681,
      "step": 11074
    },
    {
      "epoch": 0.3966193349687539,
      "grad_norm": 1.5342926979064941,
      "learning_rate": 0.0001374080761691457,
      "loss": 1.5694,
      "step": 11075
    },
    {
      "epoch": 0.3966551470983222,
      "grad_norm": 1.8465675115585327,
      "learning_rate": 0.00013739731903781164,
      "loss": 1.4775,
      "step": 11076
    },
    {
      "epoch": 0.3966909592278905,
      "grad_norm": 1.8121492862701416,
      "learning_rate": 0.00013738656140334493,
      "loss": 1.4287,
      "step": 11077
    },
    {
      "epoch": 0.39672677135745876,
      "grad_norm": 1.8412760496139526,
      "learning_rate": 0.00013737580326589024,
      "loss": 1.7312,
      "step": 11078
    },
    {
      "epoch": 0.39676258348702703,
      "grad_norm": 2.1565401554107666,
      "learning_rate": 0.00013736504462559234,
      "loss": 1.4282,
      "step": 11079
    },
    {
      "epoch": 0.39679839561659536,
      "grad_norm": 1.9769231081008911,
      "learning_rate": 0.00013735428548259594,
      "loss": 1.3531,
      "step": 11080
    },
    {
      "epoch": 0.3968342077461636,
      "grad_norm": 2.425816535949707,
      "learning_rate": 0.00013734352583704587,
      "loss": 1.6955,
      "step": 11081
    },
    {
      "epoch": 0.3968700198757319,
      "grad_norm": 1.4811224937438965,
      "learning_rate": 0.00013733276568908678,
      "loss": 1.7023,
      "step": 11082
    },
    {
      "epoch": 0.3969058320053002,
      "grad_norm": 1.5459316968917847,
      "learning_rate": 0.00013732200503886348,
      "loss": 1.3232,
      "step": 11083
    },
    {
      "epoch": 0.3969416441348685,
      "grad_norm": 1.856197476387024,
      "learning_rate": 0.00013731124388652077,
      "loss": 1.5471,
      "step": 11084
    },
    {
      "epoch": 0.39697745626443676,
      "grad_norm": 1.9269754886627197,
      "learning_rate": 0.00013730048223220336,
      "loss": 1.4477,
      "step": 11085
    },
    {
      "epoch": 0.397013268394005,
      "grad_norm": 1.5941188335418701,
      "learning_rate": 0.00013728972007605612,
      "loss": 1.5713,
      "step": 11086
    },
    {
      "epoch": 0.39704908052357335,
      "grad_norm": 1.502655267715454,
      "learning_rate": 0.00013727895741822379,
      "loss": 1.3498,
      "step": 11087
    },
    {
      "epoch": 0.3970848926531416,
      "grad_norm": 1.3070005178451538,
      "learning_rate": 0.00013726819425885115,
      "loss": 1.5596,
      "step": 11088
    },
    {
      "epoch": 0.3971207047827099,
      "grad_norm": 1.96981680393219,
      "learning_rate": 0.00013725743059808305,
      "loss": 1.588,
      "step": 11089
    },
    {
      "epoch": 0.3971565169122782,
      "grad_norm": 2.043673276901245,
      "learning_rate": 0.00013724666643606429,
      "loss": 1.6474,
      "step": 11090
    },
    {
      "epoch": 0.3971923290418465,
      "grad_norm": 1.5892807245254517,
      "learning_rate": 0.00013723590177293967,
      "loss": 1.5802,
      "step": 11091
    },
    {
      "epoch": 0.39722814117141475,
      "grad_norm": 2.098597764968872,
      "learning_rate": 0.000137225136608854,
      "loss": 1.8254,
      "step": 11092
    },
    {
      "epoch": 0.397263953300983,
      "grad_norm": 1.9812650680541992,
      "learning_rate": 0.00013721437094395213,
      "loss": 1.4145,
      "step": 11093
    },
    {
      "epoch": 0.39729976543055134,
      "grad_norm": 2.5424814224243164,
      "learning_rate": 0.00013720360477837893,
      "loss": 1.5099,
      "step": 11094
    },
    {
      "epoch": 0.3973355775601196,
      "grad_norm": 1.5709184408187866,
      "learning_rate": 0.00013719283811227921,
      "loss": 1.6173,
      "step": 11095
    },
    {
      "epoch": 0.3973713896896879,
      "grad_norm": 1.7686853408813477,
      "learning_rate": 0.0001371820709457978,
      "loss": 1.8625,
      "step": 11096
    },
    {
      "epoch": 0.3974072018192562,
      "grad_norm": 2.2390694618225098,
      "learning_rate": 0.00013717130327907964,
      "loss": 1.6054,
      "step": 11097
    },
    {
      "epoch": 0.3974430139488245,
      "grad_norm": 2.229501247406006,
      "learning_rate": 0.00013716053511226949,
      "loss": 1.4254,
      "step": 11098
    },
    {
      "epoch": 0.39747882607839274,
      "grad_norm": 2.011171817779541,
      "learning_rate": 0.00013714976644551232,
      "loss": 1.4815,
      "step": 11099
    },
    {
      "epoch": 0.397514638207961,
      "grad_norm": 1.347294569015503,
      "learning_rate": 0.00013713899727895294,
      "loss": 1.3786,
      "step": 11100
    },
    {
      "epoch": 0.39755045033752934,
      "grad_norm": 1.8997390270233154,
      "learning_rate": 0.00013712822761273625,
      "loss": 1.6241,
      "step": 11101
    },
    {
      "epoch": 0.3975862624670976,
      "grad_norm": 1.4208718538284302,
      "learning_rate": 0.00013711745744700714,
      "loss": 1.5554,
      "step": 11102
    },
    {
      "epoch": 0.3976220745966659,
      "grad_norm": 1.6987639665603638,
      "learning_rate": 0.00013710668678191054,
      "loss": 1.5971,
      "step": 11103
    },
    {
      "epoch": 0.3976578867262342,
      "grad_norm": 2.2498080730438232,
      "learning_rate": 0.00013709591561759135,
      "loss": 1.378,
      "step": 11104
    },
    {
      "epoch": 0.39769369885580247,
      "grad_norm": 1.919611930847168,
      "learning_rate": 0.00013708514395419444,
      "loss": 1.8187,
      "step": 11105
    },
    {
      "epoch": 0.39772951098537074,
      "grad_norm": 1.7510775327682495,
      "learning_rate": 0.00013707437179186476,
      "loss": 1.4297,
      "step": 11106
    },
    {
      "epoch": 0.397765323114939,
      "grad_norm": 1.8031326532363892,
      "learning_rate": 0.0001370635991307472,
      "loss": 1.6006,
      "step": 11107
    },
    {
      "epoch": 0.39780113524450733,
      "grad_norm": 1.6740226745605469,
      "learning_rate": 0.00013705282597098676,
      "loss": 1.2622,
      "step": 11108
    },
    {
      "epoch": 0.3978369473740756,
      "grad_norm": 1.7174181938171387,
      "learning_rate": 0.00013704205231272835,
      "loss": 1.7329,
      "step": 11109
    },
    {
      "epoch": 0.39787275950364387,
      "grad_norm": 1.7932454347610474,
      "learning_rate": 0.0001370312781561169,
      "loss": 1.6485,
      "step": 11110
    },
    {
      "epoch": 0.3979085716332122,
      "grad_norm": 1.698287010192871,
      "learning_rate": 0.00013702050350129734,
      "loss": 1.6169,
      "step": 11111
    },
    {
      "epoch": 0.39794438376278046,
      "grad_norm": 1.7249375581741333,
      "learning_rate": 0.00013700972834841468,
      "loss": 1.9336,
      "step": 11112
    },
    {
      "epoch": 0.39798019589234873,
      "grad_norm": 1.4442729949951172,
      "learning_rate": 0.00013699895269761385,
      "loss": 1.2067,
      "step": 11113
    },
    {
      "epoch": 0.398016008021917,
      "grad_norm": 1.2577317953109741,
      "learning_rate": 0.00013698817654903984,
      "loss": 1.5325,
      "step": 11114
    },
    {
      "epoch": 0.3980518201514853,
      "grad_norm": 1.5931488275527954,
      "learning_rate": 0.00013697739990283764,
      "loss": 1.568,
      "step": 11115
    },
    {
      "epoch": 0.3980876322810536,
      "grad_norm": 1.8899900913238525,
      "learning_rate": 0.00013696662275915222,
      "loss": 1.2821,
      "step": 11116
    },
    {
      "epoch": 0.39812344441062186,
      "grad_norm": 1.9897520542144775,
      "learning_rate": 0.00013695584511812857,
      "loss": 1.5535,
      "step": 11117
    },
    {
      "epoch": 0.3981592565401902,
      "grad_norm": 2.0698153972625732,
      "learning_rate": 0.00013694506697991169,
      "loss": 1.5855,
      "step": 11118
    },
    {
      "epoch": 0.39819506866975846,
      "grad_norm": 2.299964427947998,
      "learning_rate": 0.00013693428834464662,
      "loss": 1.5469,
      "step": 11119
    },
    {
      "epoch": 0.3982308807993267,
      "grad_norm": 1.849847674369812,
      "learning_rate": 0.00013692350921247829,
      "loss": 1.4537,
      "step": 11120
    },
    {
      "epoch": 0.398266692928895,
      "grad_norm": 1.605886459350586,
      "learning_rate": 0.00013691272958355182,
      "loss": 1.6433,
      "step": 11121
    },
    {
      "epoch": 0.3983025050584633,
      "grad_norm": 1.4784764051437378,
      "learning_rate": 0.00013690194945801214,
      "loss": 1.6726,
      "step": 11122
    },
    {
      "epoch": 0.3983383171880316,
      "grad_norm": 2.1290183067321777,
      "learning_rate": 0.00013689116883600436,
      "loss": 1.6535,
      "step": 11123
    },
    {
      "epoch": 0.39837412931759986,
      "grad_norm": 1.8331060409545898,
      "learning_rate": 0.00013688038771767345,
      "loss": 1.5419,
      "step": 11124
    },
    {
      "epoch": 0.3984099414471682,
      "grad_norm": 1.70211660861969,
      "learning_rate": 0.00013686960610316456,
      "loss": 1.2083,
      "step": 11125
    },
    {
      "epoch": 0.39844575357673645,
      "grad_norm": 1.9089553356170654,
      "learning_rate": 0.00013685882399262265,
      "loss": 1.8192,
      "step": 11126
    },
    {
      "epoch": 0.3984815657063047,
      "grad_norm": 1.5048595666885376,
      "learning_rate": 0.0001368480413861928,
      "loss": 1.5921,
      "step": 11127
    },
    {
      "epoch": 0.398517377835873,
      "grad_norm": 1.579541802406311,
      "learning_rate": 0.0001368372582840201,
      "loss": 1.5551,
      "step": 11128
    },
    {
      "epoch": 0.3985531899654413,
      "grad_norm": 1.3262720108032227,
      "learning_rate": 0.00013682647468624958,
      "loss": 1.4928,
      "step": 11129
    },
    {
      "epoch": 0.3985890020950096,
      "grad_norm": 1.7601323127746582,
      "learning_rate": 0.00013681569059302638,
      "loss": 1.5003,
      "step": 11130
    },
    {
      "epoch": 0.39862481422457785,
      "grad_norm": 2.1725378036499023,
      "learning_rate": 0.00013680490600449552,
      "loss": 1.6579,
      "step": 11131
    },
    {
      "epoch": 0.3986606263541462,
      "grad_norm": 1.540235161781311,
      "learning_rate": 0.00013679412092080213,
      "loss": 1.5752,
      "step": 11132
    },
    {
      "epoch": 0.39869643848371444,
      "grad_norm": 1.5734163522720337,
      "learning_rate": 0.0001367833353420913,
      "loss": 1.7024,
      "step": 11133
    },
    {
      "epoch": 0.3987322506132827,
      "grad_norm": 1.3746634721755981,
      "learning_rate": 0.00013677254926850818,
      "loss": 1.53,
      "step": 11134
    },
    {
      "epoch": 0.398768062742851,
      "grad_norm": 1.5181868076324463,
      "learning_rate": 0.0001367617627001978,
      "loss": 1.4935,
      "step": 11135
    },
    {
      "epoch": 0.3988038748724193,
      "grad_norm": 1.9445048570632935,
      "learning_rate": 0.0001367509756373053,
      "loss": 1.541,
      "step": 11136
    },
    {
      "epoch": 0.3988396870019876,
      "grad_norm": 1.680619478225708,
      "learning_rate": 0.0001367401880799759,
      "loss": 1.596,
      "step": 11137
    },
    {
      "epoch": 0.39887549913155584,
      "grad_norm": 1.310683250427246,
      "learning_rate": 0.0001367294000283546,
      "loss": 1.5233,
      "step": 11138
    },
    {
      "epoch": 0.39891131126112417,
      "grad_norm": 1.5829931497573853,
      "learning_rate": 0.00013671861148258665,
      "loss": 1.6821,
      "step": 11139
    },
    {
      "epoch": 0.39894712339069244,
      "grad_norm": 1.4462175369262695,
      "learning_rate": 0.0001367078224428171,
      "loss": 1.5534,
      "step": 11140
    },
    {
      "epoch": 0.3989829355202607,
      "grad_norm": 1.4816615581512451,
      "learning_rate": 0.00013669703290919118,
      "loss": 1.5118,
      "step": 11141
    },
    {
      "epoch": 0.399018747649829,
      "grad_norm": 1.906777262687683,
      "learning_rate": 0.00013668624288185402,
      "loss": 1.8655,
      "step": 11142
    },
    {
      "epoch": 0.3990545597793973,
      "grad_norm": 1.6060969829559326,
      "learning_rate": 0.00013667545236095076,
      "loss": 1.4475,
      "step": 11143
    },
    {
      "epoch": 0.39909037190896557,
      "grad_norm": 1.6323474645614624,
      "learning_rate": 0.00013666466134662662,
      "loss": 1.6315,
      "step": 11144
    },
    {
      "epoch": 0.39912618403853384,
      "grad_norm": 1.5477476119995117,
      "learning_rate": 0.00013665386983902672,
      "loss": 1.7228,
      "step": 11145
    },
    {
      "epoch": 0.39916199616810216,
      "grad_norm": 1.9727189540863037,
      "learning_rate": 0.00013664307783829634,
      "loss": 1.5455,
      "step": 11146
    },
    {
      "epoch": 0.39919780829767043,
      "grad_norm": 1.7289396524429321,
      "learning_rate": 0.00013663228534458054,
      "loss": 1.5659,
      "step": 11147
    },
    {
      "epoch": 0.3992336204272387,
      "grad_norm": 1.9361799955368042,
      "learning_rate": 0.00013662149235802465,
      "loss": 1.4145,
      "step": 11148
    },
    {
      "epoch": 0.39926943255680697,
      "grad_norm": 1.8130748271942139,
      "learning_rate": 0.0001366106988787738,
      "loss": 1.8146,
      "step": 11149
    },
    {
      "epoch": 0.3993052446863753,
      "grad_norm": 1.4281504154205322,
      "learning_rate": 0.00013659990490697322,
      "loss": 1.4696,
      "step": 11150
    },
    {
      "epoch": 0.39934105681594356,
      "grad_norm": 1.5529637336730957,
      "learning_rate": 0.0001365891104427681,
      "loss": 1.6064,
      "step": 11151
    },
    {
      "epoch": 0.39937686894551183,
      "grad_norm": 1.484028697013855,
      "learning_rate": 0.00013657831548630377,
      "loss": 1.5061,
      "step": 11152
    },
    {
      "epoch": 0.39941268107508016,
      "grad_norm": 1.630724310874939,
      "learning_rate": 0.00013656752003772535,
      "loss": 1.5966,
      "step": 11153
    },
    {
      "epoch": 0.3994484932046484,
      "grad_norm": 1.61417818069458,
      "learning_rate": 0.00013655672409717813,
      "loss": 1.4266,
      "step": 11154
    },
    {
      "epoch": 0.3994843053342167,
      "grad_norm": 3.1259992122650146,
      "learning_rate": 0.0001365459276648073,
      "loss": 1.6252,
      "step": 11155
    },
    {
      "epoch": 0.39952011746378496,
      "grad_norm": 1.3071650266647339,
      "learning_rate": 0.00013653513074075816,
      "loss": 1.5813,
      "step": 11156
    },
    {
      "epoch": 0.3995559295933533,
      "grad_norm": 1.3977751731872559,
      "learning_rate": 0.000136524333325176,
      "loss": 1.6765,
      "step": 11157
    },
    {
      "epoch": 0.39959174172292156,
      "grad_norm": 1.4183727502822876,
      "learning_rate": 0.00013651353541820603,
      "loss": 1.299,
      "step": 11158
    },
    {
      "epoch": 0.3996275538524898,
      "grad_norm": 1.6623196601867676,
      "learning_rate": 0.00013650273701999353,
      "loss": 1.4861,
      "step": 11159
    },
    {
      "epoch": 0.39966336598205815,
      "grad_norm": 1.5375479459762573,
      "learning_rate": 0.0001364919381306838,
      "loss": 1.7469,
      "step": 11160
    },
    {
      "epoch": 0.3996991781116264,
      "grad_norm": 1.6787738800048828,
      "learning_rate": 0.00013648113875042213,
      "loss": 1.4347,
      "step": 11161
    },
    {
      "epoch": 0.3997349902411947,
      "grad_norm": 2.2604610919952393,
      "learning_rate": 0.00013647033887935378,
      "loss": 1.3712,
      "step": 11162
    },
    {
      "epoch": 0.39977080237076296,
      "grad_norm": 2.34169340133667,
      "learning_rate": 0.00013645953851762406,
      "loss": 1.1879,
      "step": 11163
    },
    {
      "epoch": 0.3998066145003313,
      "grad_norm": 1.8770477771759033,
      "learning_rate": 0.00013644873766537828,
      "loss": 1.5683,
      "step": 11164
    },
    {
      "epoch": 0.39984242662989955,
      "grad_norm": 1.317833423614502,
      "learning_rate": 0.00013643793632276175,
      "loss": 1.0902,
      "step": 11165
    },
    {
      "epoch": 0.3998782387594678,
      "grad_norm": 1.86030912399292,
      "learning_rate": 0.00013642713448991977,
      "loss": 1.0972,
      "step": 11166
    },
    {
      "epoch": 0.39991405088903614,
      "grad_norm": 3.0373973846435547,
      "learning_rate": 0.0001364163321669977,
      "loss": 1.9666,
      "step": 11167
    },
    {
      "epoch": 0.3999498630186044,
      "grad_norm": 1.878786563873291,
      "learning_rate": 0.00013640552935414085,
      "loss": 1.4069,
      "step": 11168
    },
    {
      "epoch": 0.3999856751481727,
      "grad_norm": 1.554965615272522,
      "learning_rate": 0.00013639472605149456,
      "loss": 1.3437,
      "step": 11169
    },
    {
      "epoch": 0.40002148727774095,
      "grad_norm": 1.9674904346466064,
      "learning_rate": 0.00013638392225920418,
      "loss": 1.7638,
      "step": 11170
    },
    {
      "epoch": 0.4000572994073093,
      "grad_norm": 1.3602914810180664,
      "learning_rate": 0.00013637311797741507,
      "loss": 1.4951,
      "step": 11171
    },
    {
      "epoch": 0.40009311153687754,
      "grad_norm": 2.014244318008423,
      "learning_rate": 0.00013636231320627258,
      "loss": 1.5531,
      "step": 11172
    },
    {
      "epoch": 0.4001289236664458,
      "grad_norm": 1.8055243492126465,
      "learning_rate": 0.00013635150794592205,
      "loss": 1.1671,
      "step": 11173
    },
    {
      "epoch": 0.40016473579601414,
      "grad_norm": 1.5012273788452148,
      "learning_rate": 0.00013634070219650888,
      "loss": 1.5464,
      "step": 11174
    },
    {
      "epoch": 0.4002005479255824,
      "grad_norm": 1.5045928955078125,
      "learning_rate": 0.00013632989595817842,
      "loss": 1.1954,
      "step": 11175
    },
    {
      "epoch": 0.4002363600551507,
      "grad_norm": 1.548518180847168,
      "learning_rate": 0.0001363190892310761,
      "loss": 1.3469,
      "step": 11176
    },
    {
      "epoch": 0.40027217218471894,
      "grad_norm": 1.9656542539596558,
      "learning_rate": 0.00013630828201534727,
      "loss": 1.4379,
      "step": 11177
    },
    {
      "epoch": 0.40030798431428727,
      "grad_norm": 1.4483617544174194,
      "learning_rate": 0.00013629747431113734,
      "loss": 1.7661,
      "step": 11178
    },
    {
      "epoch": 0.40034379644385554,
      "grad_norm": 2.1301279067993164,
      "learning_rate": 0.0001362866661185917,
      "loss": 1.5943,
      "step": 11179
    },
    {
      "epoch": 0.4003796085734238,
      "grad_norm": 1.7540024518966675,
      "learning_rate": 0.00013627585743785582,
      "loss": 1.6543,
      "step": 11180
    },
    {
      "epoch": 0.40041542070299213,
      "grad_norm": 1.4993693828582764,
      "learning_rate": 0.000136265048269075,
      "loss": 1.4884,
      "step": 11181
    },
    {
      "epoch": 0.4004512328325604,
      "grad_norm": 1.55181086063385,
      "learning_rate": 0.0001362542386123948,
      "loss": 1.3658,
      "step": 11182
    },
    {
      "epoch": 0.40048704496212867,
      "grad_norm": 1.8561002016067505,
      "learning_rate": 0.00013624342846796058,
      "loss": 1.321,
      "step": 11183
    },
    {
      "epoch": 0.40052285709169694,
      "grad_norm": 1.5602530241012573,
      "learning_rate": 0.00013623261783591773,
      "loss": 1.6882,
      "step": 11184
    },
    {
      "epoch": 0.40055866922126526,
      "grad_norm": 1.8052551746368408,
      "learning_rate": 0.00013622180671641178,
      "loss": 1.3348,
      "step": 11185
    },
    {
      "epoch": 0.40059448135083353,
      "grad_norm": 1.9859580993652344,
      "learning_rate": 0.0001362109951095881,
      "loss": 1.6761,
      "step": 11186
    },
    {
      "epoch": 0.4006302934804018,
      "grad_norm": 1.7264128923416138,
      "learning_rate": 0.00013620018301559224,
      "loss": 1.5664,
      "step": 11187
    },
    {
      "epoch": 0.4006661056099701,
      "grad_norm": 1.941522240638733,
      "learning_rate": 0.0001361893704345696,
      "loss": 1.5823,
      "step": 11188
    },
    {
      "epoch": 0.4007019177395384,
      "grad_norm": 1.5751817226409912,
      "learning_rate": 0.00013617855736666566,
      "loss": 1.5329,
      "step": 11189
    },
    {
      "epoch": 0.40073772986910666,
      "grad_norm": 2.8278791904449463,
      "learning_rate": 0.00013616774381202591,
      "loss": 1.644,
      "step": 11190
    },
    {
      "epoch": 0.40077354199867493,
      "grad_norm": 1.5788226127624512,
      "learning_rate": 0.00013615692977079577,
      "loss": 1.3641,
      "step": 11191
    },
    {
      "epoch": 0.40080935412824326,
      "grad_norm": 1.462515115737915,
      "learning_rate": 0.00013614611524312084,
      "loss": 1.6041,
      "step": 11192
    },
    {
      "epoch": 0.4008451662578115,
      "grad_norm": 1.6150527000427246,
      "learning_rate": 0.0001361353002291465,
      "loss": 1.4713,
      "step": 11193
    },
    {
      "epoch": 0.4008809783873798,
      "grad_norm": 1.730208158493042,
      "learning_rate": 0.00013612448472901834,
      "loss": 1.5619,
      "step": 11194
    },
    {
      "epoch": 0.40091679051694806,
      "grad_norm": 1.573136329650879,
      "learning_rate": 0.00013611366874288186,
      "loss": 1.8576,
      "step": 11195
    },
    {
      "epoch": 0.4009526026465164,
      "grad_norm": 1.67069411277771,
      "learning_rate": 0.0001361028522708825,
      "loss": 1.6783,
      "step": 11196
    },
    {
      "epoch": 0.40098841477608466,
      "grad_norm": 1.7830244302749634,
      "learning_rate": 0.00013609203531316587,
      "loss": 1.5543,
      "step": 11197
    },
    {
      "epoch": 0.4010242269056529,
      "grad_norm": 1.4327481985092163,
      "learning_rate": 0.0001360812178698774,
      "loss": 1.1594,
      "step": 11198
    },
    {
      "epoch": 0.40106003903522125,
      "grad_norm": 1.6612839698791504,
      "learning_rate": 0.00013607039994116278,
      "loss": 1.4161,
      "step": 11199
    },
    {
      "epoch": 0.4010958511647895,
      "grad_norm": 2.0589075088500977,
      "learning_rate": 0.00013605958152716738,
      "loss": 1.7591,
      "step": 11200
    },
    {
      "epoch": 0.4011316632943578,
      "grad_norm": 1.7634742259979248,
      "learning_rate": 0.00013604876262803686,
      "loss": 1.3869,
      "step": 11201
    },
    {
      "epoch": 0.40116747542392606,
      "grad_norm": 2.3785300254821777,
      "learning_rate": 0.00013603794324391672,
      "loss": 1.3486,
      "step": 11202
    },
    {
      "epoch": 0.4012032875534944,
      "grad_norm": 1.635969877243042,
      "learning_rate": 0.00013602712337495255,
      "loss": 1.3801,
      "step": 11203
    },
    {
      "epoch": 0.40123909968306265,
      "grad_norm": 1.580276370048523,
      "learning_rate": 0.0001360163030212899,
      "loss": 1.6318,
      "step": 11204
    },
    {
      "epoch": 0.4012749118126309,
      "grad_norm": 1.8174611330032349,
      "learning_rate": 0.00013600548218307436,
      "loss": 1.3491,
      "step": 11205
    },
    {
      "epoch": 0.40131072394219924,
      "grad_norm": 1.3981926441192627,
      "learning_rate": 0.0001359946608604515,
      "loss": 1.6881,
      "step": 11206
    },
    {
      "epoch": 0.4013465360717675,
      "grad_norm": 1.802919864654541,
      "learning_rate": 0.00013598383905356692,
      "loss": 1.2794,
      "step": 11207
    },
    {
      "epoch": 0.4013823482013358,
      "grad_norm": 1.6066579818725586,
      "learning_rate": 0.00013597301676256617,
      "loss": 1.356,
      "step": 11208
    },
    {
      "epoch": 0.40141816033090405,
      "grad_norm": 1.9527837038040161,
      "learning_rate": 0.0001359621939875949,
      "loss": 1.5963,
      "step": 11209
    },
    {
      "epoch": 0.4014539724604724,
      "grad_norm": 1.324955701828003,
      "learning_rate": 0.00013595137072879867,
      "loss": 1.2462,
      "step": 11210
    },
    {
      "epoch": 0.40148978459004064,
      "grad_norm": 1.5357877016067505,
      "learning_rate": 0.00013594054698632315,
      "loss": 1.4547,
      "step": 11211
    },
    {
      "epoch": 0.4015255967196089,
      "grad_norm": 1.6919299364089966,
      "learning_rate": 0.00013592972276031394,
      "loss": 1.6643,
      "step": 11212
    },
    {
      "epoch": 0.40156140884917724,
      "grad_norm": 1.8728044033050537,
      "learning_rate": 0.00013591889805091663,
      "loss": 1.6113,
      "step": 11213
    },
    {
      "epoch": 0.4015972209787455,
      "grad_norm": 2.333636999130249,
      "learning_rate": 0.00013590807285827688,
      "loss": 1.7807,
      "step": 11214
    },
    {
      "epoch": 0.4016330331083138,
      "grad_norm": 1.5089396238327026,
      "learning_rate": 0.00013589724718254036,
      "loss": 1.212,
      "step": 11215
    },
    {
      "epoch": 0.40166884523788204,
      "grad_norm": 1.860792636871338,
      "learning_rate": 0.00013588642102385266,
      "loss": 1.782,
      "step": 11216
    },
    {
      "epoch": 0.40170465736745037,
      "grad_norm": 1.6318405866622925,
      "learning_rate": 0.00013587559438235945,
      "loss": 1.5339,
      "step": 11217
    },
    {
      "epoch": 0.40174046949701864,
      "grad_norm": 1.822420358657837,
      "learning_rate": 0.0001358647672582064,
      "loss": 1.453,
      "step": 11218
    },
    {
      "epoch": 0.4017762816265869,
      "grad_norm": 1.4181208610534668,
      "learning_rate": 0.00013585393965153916,
      "loss": 1.3009,
      "step": 11219
    },
    {
      "epoch": 0.40181209375615523,
      "grad_norm": 1.3538849353790283,
      "learning_rate": 0.00013584311156250342,
      "loss": 1.6589,
      "step": 11220
    },
    {
      "epoch": 0.4018479058857235,
      "grad_norm": 1.5220893621444702,
      "learning_rate": 0.00013583228299124484,
      "loss": 1.7311,
      "step": 11221
    },
    {
      "epoch": 0.40188371801529177,
      "grad_norm": 1.5974732637405396,
      "learning_rate": 0.00013582145393790913,
      "loss": 1.4992,
      "step": 11222
    },
    {
      "epoch": 0.40191953014486004,
      "grad_norm": 2.0485281944274902,
      "learning_rate": 0.00013581062440264194,
      "loss": 1.4456,
      "step": 11223
    },
    {
      "epoch": 0.40195534227442836,
      "grad_norm": 1.678962230682373,
      "learning_rate": 0.000135799794385589,
      "loss": 1.5724,
      "step": 11224
    },
    {
      "epoch": 0.40199115440399663,
      "grad_norm": 1.9616618156433105,
      "learning_rate": 0.00013578896388689602,
      "loss": 1.5544,
      "step": 11225
    },
    {
      "epoch": 0.4020269665335649,
      "grad_norm": 1.4487061500549316,
      "learning_rate": 0.00013577813290670867,
      "loss": 1.467,
      "step": 11226
    },
    {
      "epoch": 0.4020627786631332,
      "grad_norm": 2.1156294345855713,
      "learning_rate": 0.00013576730144517271,
      "loss": 1.8479,
      "step": 11227
    },
    {
      "epoch": 0.4020985907927015,
      "grad_norm": 1.5825573205947876,
      "learning_rate": 0.00013575646950243384,
      "loss": 1.7531,
      "step": 11228
    },
    {
      "epoch": 0.40213440292226976,
      "grad_norm": 2.3100428581237793,
      "learning_rate": 0.0001357456370786378,
      "loss": 1.467,
      "step": 11229
    },
    {
      "epoch": 0.40217021505183803,
      "grad_norm": 1.4864134788513184,
      "learning_rate": 0.0001357348041739303,
      "loss": 1.5086,
      "step": 11230
    },
    {
      "epoch": 0.40220602718140636,
      "grad_norm": 1.748326301574707,
      "learning_rate": 0.00013572397078845716,
      "loss": 1.3973,
      "step": 11231
    },
    {
      "epoch": 0.4022418393109746,
      "grad_norm": 1.4479734897613525,
      "learning_rate": 0.00013571313692236405,
      "loss": 1.4109,
      "step": 11232
    },
    {
      "epoch": 0.4022776514405429,
      "grad_norm": 1.8514389991760254,
      "learning_rate": 0.0001357023025757967,
      "loss": 1.5098,
      "step": 11233
    },
    {
      "epoch": 0.4023134635701112,
      "grad_norm": 1.9194141626358032,
      "learning_rate": 0.00013569146774890099,
      "loss": 1.6107,
      "step": 11234
    },
    {
      "epoch": 0.4023492756996795,
      "grad_norm": 1.9210814237594604,
      "learning_rate": 0.00013568063244182257,
      "loss": 1.5859,
      "step": 11235
    },
    {
      "epoch": 0.40238508782924776,
      "grad_norm": 1.5344477891921997,
      "learning_rate": 0.00013566979665470728,
      "loss": 1.4888,
      "step": 11236
    },
    {
      "epoch": 0.402420899958816,
      "grad_norm": 2.001387596130371,
      "learning_rate": 0.0001356589603877009,
      "loss": 1.5566,
      "step": 11237
    },
    {
      "epoch": 0.40245671208838435,
      "grad_norm": 1.8244560956954956,
      "learning_rate": 0.0001356481236409492,
      "loss": 1.2914,
      "step": 11238
    },
    {
      "epoch": 0.4024925242179526,
      "grad_norm": 1.440247654914856,
      "learning_rate": 0.00013563728641459793,
      "loss": 1.5425,
      "step": 11239
    },
    {
      "epoch": 0.4025283363475209,
      "grad_norm": 1.331313133239746,
      "learning_rate": 0.000135626448708793,
      "loss": 1.5918,
      "step": 11240
    },
    {
      "epoch": 0.4025641484770892,
      "grad_norm": 1.4485142230987549,
      "learning_rate": 0.00013561561052368015,
      "loss": 1.5563,
      "step": 11241
    },
    {
      "epoch": 0.4025999606066575,
      "grad_norm": 1.3569890260696411,
      "learning_rate": 0.0001356047718594052,
      "loss": 1.6618,
      "step": 11242
    },
    {
      "epoch": 0.40263577273622575,
      "grad_norm": 1.8978397846221924,
      "learning_rate": 0.00013559393271611397,
      "loss": 1.7617,
      "step": 11243
    },
    {
      "epoch": 0.402671584865794,
      "grad_norm": 1.6500682830810547,
      "learning_rate": 0.00013558309309395224,
      "loss": 1.1864,
      "step": 11244
    },
    {
      "epoch": 0.40270739699536234,
      "grad_norm": 1.9103257656097412,
      "learning_rate": 0.00013557225299306599,
      "loss": 1.4868,
      "step": 11245
    },
    {
      "epoch": 0.4027432091249306,
      "grad_norm": 1.6831821203231812,
      "learning_rate": 0.00013556141241360088,
      "loss": 1.8324,
      "step": 11246
    },
    {
      "epoch": 0.4027790212544989,
      "grad_norm": 1.6163675785064697,
      "learning_rate": 0.00013555057135570286,
      "loss": 1.4925,
      "step": 11247
    },
    {
      "epoch": 0.4028148333840672,
      "grad_norm": 1.781957983970642,
      "learning_rate": 0.00013553972981951776,
      "loss": 1.3588,
      "step": 11248
    },
    {
      "epoch": 0.4028506455136355,
      "grad_norm": 1.2353285551071167,
      "learning_rate": 0.00013552888780519144,
      "loss": 1.4921,
      "step": 11249
    },
    {
      "epoch": 0.40288645764320374,
      "grad_norm": 1.3386553525924683,
      "learning_rate": 0.00013551804531286975,
      "loss": 1.6313,
      "step": 11250
    },
    {
      "epoch": 0.402922269772772,
      "grad_norm": 1.669419288635254,
      "learning_rate": 0.0001355072023426986,
      "loss": 1.432,
      "step": 11251
    },
    {
      "epoch": 0.40295808190234034,
      "grad_norm": 2.4253623485565186,
      "learning_rate": 0.00013549635889482383,
      "loss": 1.4862,
      "step": 11252
    },
    {
      "epoch": 0.4029938940319086,
      "grad_norm": 2.272901773452759,
      "learning_rate": 0.00013548551496939132,
      "loss": 1.4508,
      "step": 11253
    },
    {
      "epoch": 0.4030297061614769,
      "grad_norm": 1.535967230796814,
      "learning_rate": 0.00013547467056654702,
      "loss": 1.576,
      "step": 11254
    },
    {
      "epoch": 0.4030655182910452,
      "grad_norm": 1.4173799753189087,
      "learning_rate": 0.00013546382568643676,
      "loss": 1.7135,
      "step": 11255
    },
    {
      "epoch": 0.40310133042061347,
      "grad_norm": 1.3089243173599243,
      "learning_rate": 0.00013545298032920647,
      "loss": 1.4986,
      "step": 11256
    },
    {
      "epoch": 0.40313714255018174,
      "grad_norm": 1.5522119998931885,
      "learning_rate": 0.00013544213449500204,
      "loss": 1.5136,
      "step": 11257
    },
    {
      "epoch": 0.40317295467975,
      "grad_norm": 1.4867783784866333,
      "learning_rate": 0.00013543128818396946,
      "loss": 1.4835,
      "step": 11258
    },
    {
      "epoch": 0.40320876680931833,
      "grad_norm": 1.8949105739593506,
      "learning_rate": 0.0001354204413962546,
      "loss": 1.5772,
      "step": 11259
    },
    {
      "epoch": 0.4032445789388866,
      "grad_norm": 2.0525848865509033,
      "learning_rate": 0.00013540959413200335,
      "loss": 1.4172,
      "step": 11260
    },
    {
      "epoch": 0.40328039106845487,
      "grad_norm": 1.6664822101593018,
      "learning_rate": 0.0001353987463913617,
      "loss": 1.6019,
      "step": 11261
    },
    {
      "epoch": 0.4033162031980232,
      "grad_norm": 2.6384687423706055,
      "learning_rate": 0.0001353878981744756,
      "loss": 1.633,
      "step": 11262
    },
    {
      "epoch": 0.40335201532759146,
      "grad_norm": 1.2509064674377441,
      "learning_rate": 0.00013537704948149093,
      "loss": 1.5135,
      "step": 11263
    },
    {
      "epoch": 0.40338782745715973,
      "grad_norm": 2.389633893966675,
      "learning_rate": 0.00013536620031255373,
      "loss": 1.7426,
      "step": 11264
    },
    {
      "epoch": 0.403423639586728,
      "grad_norm": 2.0430221557617188,
      "learning_rate": 0.0001353553506678099,
      "loss": 1.5984,
      "step": 11265
    },
    {
      "epoch": 0.4034594517162963,
      "grad_norm": 1.7450594902038574,
      "learning_rate": 0.00013534450054740544,
      "loss": 1.5399,
      "step": 11266
    },
    {
      "epoch": 0.4034952638458646,
      "grad_norm": 1.6870568990707397,
      "learning_rate": 0.0001353336499514863,
      "loss": 1.4738,
      "step": 11267
    },
    {
      "epoch": 0.40353107597543286,
      "grad_norm": 1.8001635074615479,
      "learning_rate": 0.00013532279888019851,
      "loss": 1.2668,
      "step": 11268
    },
    {
      "epoch": 0.4035668881050012,
      "grad_norm": 2.56563663482666,
      "learning_rate": 0.00013531194733368805,
      "loss": 1.8747,
      "step": 11269
    },
    {
      "epoch": 0.40360270023456946,
      "grad_norm": 1.735708236694336,
      "learning_rate": 0.00013530109531210082,
      "loss": 1.5633,
      "step": 11270
    },
    {
      "epoch": 0.4036385123641377,
      "grad_norm": 1.9577866792678833,
      "learning_rate": 0.00013529024281558292,
      "loss": 1.4575,
      "step": 11271
    },
    {
      "epoch": 0.403674324493706,
      "grad_norm": 2.2287938594818115,
      "learning_rate": 0.00013527938984428031,
      "loss": 1.6088,
      "step": 11272
    },
    {
      "epoch": 0.4037101366232743,
      "grad_norm": 1.7325700521469116,
      "learning_rate": 0.00013526853639833904,
      "loss": 1.7316,
      "step": 11273
    },
    {
      "epoch": 0.4037459487528426,
      "grad_norm": 1.6742218732833862,
      "learning_rate": 0.0001352576824779051,
      "loss": 1.3724,
      "step": 11274
    },
    {
      "epoch": 0.40378176088241086,
      "grad_norm": 1.6109325885772705,
      "learning_rate": 0.0001352468280831245,
      "loss": 1.2103,
      "step": 11275
    },
    {
      "epoch": 0.4038175730119792,
      "grad_norm": 1.916214108467102,
      "learning_rate": 0.00013523597321414332,
      "loss": 1.5114,
      "step": 11276
    },
    {
      "epoch": 0.40385338514154745,
      "grad_norm": 2.142789840698242,
      "learning_rate": 0.00013522511787110756,
      "loss": 1.6882,
      "step": 11277
    },
    {
      "epoch": 0.4038891972711157,
      "grad_norm": 1.7598340511322021,
      "learning_rate": 0.00013521426205416326,
      "loss": 1.49,
      "step": 11278
    },
    {
      "epoch": 0.403925009400684,
      "grad_norm": 1.6194044351577759,
      "learning_rate": 0.00013520340576345653,
      "loss": 1.6978,
      "step": 11279
    },
    {
      "epoch": 0.4039608215302523,
      "grad_norm": 1.4875582456588745,
      "learning_rate": 0.00013519254899913333,
      "loss": 1.7141,
      "step": 11280
    },
    {
      "epoch": 0.4039966336598206,
      "grad_norm": 1.9088953733444214,
      "learning_rate": 0.0001351816917613398,
      "loss": 1.4717,
      "step": 11281
    },
    {
      "epoch": 0.40403244578938885,
      "grad_norm": 1.5134812593460083,
      "learning_rate": 0.00013517083405022203,
      "loss": 1.6138,
      "step": 11282
    },
    {
      "epoch": 0.4040682579189572,
      "grad_norm": 2.6607894897460938,
      "learning_rate": 0.000135159975865926,
      "loss": 1.6279,
      "step": 11283
    },
    {
      "epoch": 0.40410407004852544,
      "grad_norm": 2.2413384914398193,
      "learning_rate": 0.00013514911720859785,
      "loss": 1.6775,
      "step": 11284
    },
    {
      "epoch": 0.4041398821780937,
      "grad_norm": 1.3868253231048584,
      "learning_rate": 0.00013513825807838373,
      "loss": 1.425,
      "step": 11285
    },
    {
      "epoch": 0.404175694307662,
      "grad_norm": 1.4734158515930176,
      "learning_rate": 0.0001351273984754296,
      "loss": 1.4511,
      "step": 11286
    },
    {
      "epoch": 0.4042115064372303,
      "grad_norm": 1.6310186386108398,
      "learning_rate": 0.00013511653839988168,
      "loss": 1.5508,
      "step": 11287
    },
    {
      "epoch": 0.4042473185667986,
      "grad_norm": 2.4635744094848633,
      "learning_rate": 0.000135105677851886,
      "loss": 1.4342,
      "step": 11288
    },
    {
      "epoch": 0.40428313069636684,
      "grad_norm": 1.7325843572616577,
      "learning_rate": 0.00013509481683158874,
      "loss": 1.5283,
      "step": 11289
    },
    {
      "epoch": 0.40431894282593517,
      "grad_norm": 1.6843584775924683,
      "learning_rate": 0.00013508395533913593,
      "loss": 1.4873,
      "step": 11290
    },
    {
      "epoch": 0.40435475495550344,
      "grad_norm": 1.6397114992141724,
      "learning_rate": 0.0001350730933746738,
      "loss": 1.3648,
      "step": 11291
    },
    {
      "epoch": 0.4043905670850717,
      "grad_norm": 1.6449841260910034,
      "learning_rate": 0.00013506223093834844,
      "loss": 1.5678,
      "step": 11292
    },
    {
      "epoch": 0.40442637921464,
      "grad_norm": 1.9263081550598145,
      "learning_rate": 0.000135051368030306,
      "loss": 1.6234,
      "step": 11293
    },
    {
      "epoch": 0.4044621913442083,
      "grad_norm": 1.5096728801727295,
      "learning_rate": 0.00013504050465069263,
      "loss": 1.7048,
      "step": 11294
    },
    {
      "epoch": 0.40449800347377657,
      "grad_norm": 1.5539870262145996,
      "learning_rate": 0.0001350296407996544,
      "loss": 1.4478,
      "step": 11295
    },
    {
      "epoch": 0.40453381560334484,
      "grad_norm": 1.7265819311141968,
      "learning_rate": 0.0001350187764773376,
      "loss": 1.5398,
      "step": 11296
    },
    {
      "epoch": 0.40456962773291316,
      "grad_norm": 1.5002291202545166,
      "learning_rate": 0.0001350079116838883,
      "loss": 1.5066,
      "step": 11297
    },
    {
      "epoch": 0.40460543986248143,
      "grad_norm": 1.3665101528167725,
      "learning_rate": 0.0001349970464194527,
      "loss": 1.2355,
      "step": 11298
    },
    {
      "epoch": 0.4046412519920497,
      "grad_norm": 1.7081962823867798,
      "learning_rate": 0.000134986180684177,
      "loss": 1.4824,
      "step": 11299
    },
    {
      "epoch": 0.40467706412161797,
      "grad_norm": 1.3695849180221558,
      "learning_rate": 0.0001349753144782074,
      "loss": 1.5157,
      "step": 11300
    },
    {
      "epoch": 0.4047128762511863,
      "grad_norm": 2.080824613571167,
      "learning_rate": 0.00013496444780169,
      "loss": 1.5212,
      "step": 11301
    },
    {
      "epoch": 0.40474868838075456,
      "grad_norm": 1.3417965173721313,
      "learning_rate": 0.0001349535806547711,
      "loss": 1.2956,
      "step": 11302
    },
    {
      "epoch": 0.40478450051032283,
      "grad_norm": 1.4742838144302368,
      "learning_rate": 0.00013494271303759686,
      "loss": 1.7153,
      "step": 11303
    },
    {
      "epoch": 0.40482031263989116,
      "grad_norm": 1.4015898704528809,
      "learning_rate": 0.0001349318449503135,
      "loss": 1.3194,
      "step": 11304
    },
    {
      "epoch": 0.4048561247694594,
      "grad_norm": 1.504159688949585,
      "learning_rate": 0.00013492097639306716,
      "loss": 1.5072,
      "step": 11305
    },
    {
      "epoch": 0.4048919368990277,
      "grad_norm": 1.7774419784545898,
      "learning_rate": 0.00013491010736600418,
      "loss": 1.7561,
      "step": 11306
    },
    {
      "epoch": 0.40492774902859596,
      "grad_norm": 1.551797866821289,
      "learning_rate": 0.0001348992378692707,
      "loss": 1.5001,
      "step": 11307
    },
    {
      "epoch": 0.4049635611581643,
      "grad_norm": 2.107363224029541,
      "learning_rate": 0.000134888367903013,
      "loss": 1.3526,
      "step": 11308
    },
    {
      "epoch": 0.40499937328773256,
      "grad_norm": 3.087078332901001,
      "learning_rate": 0.00013487749746737734,
      "loss": 1.4878,
      "step": 11309
    },
    {
      "epoch": 0.4050351854173008,
      "grad_norm": 1.5227560997009277,
      "learning_rate": 0.0001348666265625099,
      "loss": 1.4858,
      "step": 11310
    },
    {
      "epoch": 0.40507099754686915,
      "grad_norm": 1.4322707653045654,
      "learning_rate": 0.00013485575518855703,
      "loss": 1.4656,
      "step": 11311
    },
    {
      "epoch": 0.4051068096764374,
      "grad_norm": 2.5870959758758545,
      "learning_rate": 0.00013484488334566488,
      "loss": 1.466,
      "step": 11312
    },
    {
      "epoch": 0.4051426218060057,
      "grad_norm": 1.6336684226989746,
      "learning_rate": 0.00013483401103397982,
      "loss": 1.5328,
      "step": 11313
    },
    {
      "epoch": 0.40517843393557396,
      "grad_norm": 1.972943902015686,
      "learning_rate": 0.00013482313825364804,
      "loss": 1.3578,
      "step": 11314
    },
    {
      "epoch": 0.4052142460651423,
      "grad_norm": 2.440329074859619,
      "learning_rate": 0.00013481226500481588,
      "loss": 1.7942,
      "step": 11315
    },
    {
      "epoch": 0.40525005819471055,
      "grad_norm": 1.815645694732666,
      "learning_rate": 0.00013480139128762956,
      "loss": 1.5568,
      "step": 11316
    },
    {
      "epoch": 0.4052858703242788,
      "grad_norm": 1.4573594331741333,
      "learning_rate": 0.00013479051710223544,
      "loss": 1.5361,
      "step": 11317
    },
    {
      "epoch": 0.40532168245384714,
      "grad_norm": 1.3116105794906616,
      "learning_rate": 0.00013477964244877977,
      "loss": 1.5229,
      "step": 11318
    },
    {
      "epoch": 0.4053574945834154,
      "grad_norm": 1.4079471826553345,
      "learning_rate": 0.0001347687673274089,
      "loss": 1.2103,
      "step": 11319
    },
    {
      "epoch": 0.4053933067129837,
      "grad_norm": 1.2982499599456787,
      "learning_rate": 0.00013475789173826908,
      "loss": 1.5497,
      "step": 11320
    },
    {
      "epoch": 0.40542911884255195,
      "grad_norm": 1.6635154485702515,
      "learning_rate": 0.0001347470156815067,
      "loss": 1.5242,
      "step": 11321
    },
    {
      "epoch": 0.4054649309721203,
      "grad_norm": 1.7815632820129395,
      "learning_rate": 0.000134736139157268,
      "loss": 1.6833,
      "step": 11322
    },
    {
      "epoch": 0.40550074310168854,
      "grad_norm": 1.7860198020935059,
      "learning_rate": 0.0001347252621656994,
      "loss": 1.5752,
      "step": 11323
    },
    {
      "epoch": 0.4055365552312568,
      "grad_norm": 1.5376157760620117,
      "learning_rate": 0.00013471438470694715,
      "loss": 1.5881,
      "step": 11324
    },
    {
      "epoch": 0.40557236736082514,
      "grad_norm": 2.517604112625122,
      "learning_rate": 0.00013470350678115763,
      "loss": 1.3554,
      "step": 11325
    },
    {
      "epoch": 0.4056081794903934,
      "grad_norm": 1.7985392808914185,
      "learning_rate": 0.00013469262838847724,
      "loss": 1.3522,
      "step": 11326
    },
    {
      "epoch": 0.4056439916199617,
      "grad_norm": 1.6951795816421509,
      "learning_rate": 0.00013468174952905223,
      "loss": 1.3,
      "step": 11327
    },
    {
      "epoch": 0.40567980374952994,
      "grad_norm": 1.8560683727264404,
      "learning_rate": 0.00013467087020302906,
      "loss": 1.4333,
      "step": 11328
    },
    {
      "epoch": 0.40571561587909827,
      "grad_norm": 1.9412565231323242,
      "learning_rate": 0.00013465999041055405,
      "loss": 1.4037,
      "step": 11329
    },
    {
      "epoch": 0.40575142800866654,
      "grad_norm": 2.763273000717163,
      "learning_rate": 0.00013464911015177356,
      "loss": 1.3453,
      "step": 11330
    },
    {
      "epoch": 0.4057872401382348,
      "grad_norm": 1.9450551271438599,
      "learning_rate": 0.000134638229426834,
      "loss": 1.6067,
      "step": 11331
    },
    {
      "epoch": 0.40582305226780313,
      "grad_norm": 1.551149606704712,
      "learning_rate": 0.0001346273482358817,
      "loss": 1.5155,
      "step": 11332
    },
    {
      "epoch": 0.4058588643973714,
      "grad_norm": 1.4368470907211304,
      "learning_rate": 0.00013461646657906315,
      "loss": 1.556,
      "step": 11333
    },
    {
      "epoch": 0.40589467652693967,
      "grad_norm": 2.317528486251831,
      "learning_rate": 0.00013460558445652467,
      "loss": 1.9397,
      "step": 11334
    },
    {
      "epoch": 0.40593048865650794,
      "grad_norm": 1.6196165084838867,
      "learning_rate": 0.0001345947018684127,
      "loss": 1.6216,
      "step": 11335
    },
    {
      "epoch": 0.40596630078607626,
      "grad_norm": 1.3833155632019043,
      "learning_rate": 0.00013458381881487362,
      "loss": 1.4274,
      "step": 11336
    },
    {
      "epoch": 0.40600211291564453,
      "grad_norm": 2.2270424365997314,
      "learning_rate": 0.0001345729352960539,
      "loss": 1.3553,
      "step": 11337
    },
    {
      "epoch": 0.4060379250452128,
      "grad_norm": 1.5133336782455444,
      "learning_rate": 0.00013456205131209988,
      "loss": 1.5309,
      "step": 11338
    },
    {
      "epoch": 0.4060737371747811,
      "grad_norm": 1.5618396997451782,
      "learning_rate": 0.0001345511668631581,
      "loss": 1.411,
      "step": 11339
    },
    {
      "epoch": 0.4061095493043494,
      "grad_norm": 1.4137815237045288,
      "learning_rate": 0.0001345402819493749,
      "loss": 1.677,
      "step": 11340
    },
    {
      "epoch": 0.40614536143391766,
      "grad_norm": 1.4480739831924438,
      "learning_rate": 0.00013452939657089677,
      "loss": 1.4027,
      "step": 11341
    },
    {
      "epoch": 0.40618117356348593,
      "grad_norm": 1.5944091081619263,
      "learning_rate": 0.00013451851072787013,
      "loss": 1.2003,
      "step": 11342
    },
    {
      "epoch": 0.40621698569305426,
      "grad_norm": 2.348677635192871,
      "learning_rate": 0.00013450762442044148,
      "loss": 1.3259,
      "step": 11343
    },
    {
      "epoch": 0.4062527978226225,
      "grad_norm": 1.4781785011291504,
      "learning_rate": 0.00013449673764875724,
      "loss": 1.4524,
      "step": 11344
    },
    {
      "epoch": 0.4062886099521908,
      "grad_norm": 1.2914907932281494,
      "learning_rate": 0.00013448585041296392,
      "loss": 1.4464,
      "step": 11345
    },
    {
      "epoch": 0.4063244220817591,
      "grad_norm": 1.4890108108520508,
      "learning_rate": 0.00013447496271320794,
      "loss": 1.5749,
      "step": 11346
    },
    {
      "epoch": 0.4063602342113274,
      "grad_norm": 1.6016771793365479,
      "learning_rate": 0.00013446407454963582,
      "loss": 1.6343,
      "step": 11347
    },
    {
      "epoch": 0.40639604634089566,
      "grad_norm": 1.856371521949768,
      "learning_rate": 0.00013445318592239405,
      "loss": 1.3749,
      "step": 11348
    },
    {
      "epoch": 0.4064318584704639,
      "grad_norm": 1.780447006225586,
      "learning_rate": 0.00013444229683162904,
      "loss": 1.5556,
      "step": 11349
    },
    {
      "epoch": 0.40646767060003225,
      "grad_norm": 2.996824026107788,
      "learning_rate": 0.00013443140727748738,
      "loss": 1.5899,
      "step": 11350
    },
    {
      "epoch": 0.4065034827296005,
      "grad_norm": 1.7526699304580688,
      "learning_rate": 0.0001344205172601156,
      "loss": 1.501,
      "step": 11351
    },
    {
      "epoch": 0.4065392948591688,
      "grad_norm": 3.1774349212646484,
      "learning_rate": 0.00013440962677966012,
      "loss": 1.3241,
      "step": 11352
    },
    {
      "epoch": 0.4065751069887371,
      "grad_norm": 1.467477798461914,
      "learning_rate": 0.0001343987358362675,
      "loss": 1.7437,
      "step": 11353
    },
    {
      "epoch": 0.4066109191183054,
      "grad_norm": 2.0850508213043213,
      "learning_rate": 0.00013438784443008426,
      "loss": 1.3627,
      "step": 11354
    },
    {
      "epoch": 0.40664673124787365,
      "grad_norm": 1.6988506317138672,
      "learning_rate": 0.00013437695256125694,
      "loss": 1.7532,
      "step": 11355
    },
    {
      "epoch": 0.4066825433774419,
      "grad_norm": 2.0369551181793213,
      "learning_rate": 0.00013436606022993207,
      "loss": 1.561,
      "step": 11356
    },
    {
      "epoch": 0.40671835550701024,
      "grad_norm": 1.7097434997558594,
      "learning_rate": 0.00013435516743625617,
      "loss": 1.8608,
      "step": 11357
    },
    {
      "epoch": 0.4067541676365785,
      "grad_norm": 1.62166166305542,
      "learning_rate": 0.0001343442741803758,
      "loss": 1.4686,
      "step": 11358
    },
    {
      "epoch": 0.4067899797661468,
      "grad_norm": 1.4330084323883057,
      "learning_rate": 0.00013433338046243753,
      "loss": 1.5757,
      "step": 11359
    },
    {
      "epoch": 0.4068257918957151,
      "grad_norm": 1.9137028455734253,
      "learning_rate": 0.0001343224862825879,
      "loss": 1.4864,
      "step": 11360
    },
    {
      "epoch": 0.4068616040252834,
      "grad_norm": 1.4365180730819702,
      "learning_rate": 0.00013431159164097354,
      "loss": 1.6628,
      "step": 11361
    },
    {
      "epoch": 0.40689741615485164,
      "grad_norm": 1.4980610609054565,
      "learning_rate": 0.0001343006965377409,
      "loss": 1.1436,
      "step": 11362
    },
    {
      "epoch": 0.4069332282844199,
      "grad_norm": 2.000791549682617,
      "learning_rate": 0.00013428980097303668,
      "loss": 1.6187,
      "step": 11363
    },
    {
      "epoch": 0.40696904041398824,
      "grad_norm": 1.9143515825271606,
      "learning_rate": 0.0001342789049470074,
      "loss": 1.6053,
      "step": 11364
    },
    {
      "epoch": 0.4070048525435565,
      "grad_norm": 1.4883040189743042,
      "learning_rate": 0.0001342680084597997,
      "loss": 1.3271,
      "step": 11365
    },
    {
      "epoch": 0.4070406646731248,
      "grad_norm": 1.6826872825622559,
      "learning_rate": 0.00013425711151156014,
      "loss": 1.7451,
      "step": 11366
    },
    {
      "epoch": 0.4070764768026931,
      "grad_norm": 1.5650379657745361,
      "learning_rate": 0.00013424621410243533,
      "loss": 1.5927,
      "step": 11367
    },
    {
      "epoch": 0.40711228893226137,
      "grad_norm": 2.28118634223938,
      "learning_rate": 0.00013423531623257189,
      "loss": 1.3887,
      "step": 11368
    },
    {
      "epoch": 0.40714810106182964,
      "grad_norm": 1.7174426317214966,
      "learning_rate": 0.0001342244179021164,
      "loss": 1.7299,
      "step": 11369
    },
    {
      "epoch": 0.4071839131913979,
      "grad_norm": 2.240532875061035,
      "learning_rate": 0.00013421351911121554,
      "loss": 1.5032,
      "step": 11370
    },
    {
      "epoch": 0.40721972532096623,
      "grad_norm": 1.6691335439682007,
      "learning_rate": 0.00013420261986001587,
      "loss": 1.6486,
      "step": 11371
    },
    {
      "epoch": 0.4072555374505345,
      "grad_norm": 2.2593131065368652,
      "learning_rate": 0.00013419172014866412,
      "loss": 1.4365,
      "step": 11372
    },
    {
      "epoch": 0.40729134958010277,
      "grad_norm": 2.0483601093292236,
      "learning_rate": 0.00013418081997730686,
      "loss": 1.6662,
      "step": 11373
    },
    {
      "epoch": 0.4073271617096711,
      "grad_norm": 1.5782848596572876,
      "learning_rate": 0.00013416991934609075,
      "loss": 1.5193,
      "step": 11374
    },
    {
      "epoch": 0.40736297383923936,
      "grad_norm": 1.8060847520828247,
      "learning_rate": 0.00013415901825516248,
      "loss": 1.5458,
      "step": 11375
    },
    {
      "epoch": 0.40739878596880763,
      "grad_norm": 1.808779001235962,
      "learning_rate": 0.00013414811670466864,
      "loss": 1.6425,
      "step": 11376
    },
    {
      "epoch": 0.4074345980983759,
      "grad_norm": 1.4207754135131836,
      "learning_rate": 0.00013413721469475597,
      "loss": 1.4269,
      "step": 11377
    },
    {
      "epoch": 0.4074704102279442,
      "grad_norm": 1.7335152626037598,
      "learning_rate": 0.00013412631222557112,
      "loss": 1.5142,
      "step": 11378
    },
    {
      "epoch": 0.4075062223575125,
      "grad_norm": 1.4673773050308228,
      "learning_rate": 0.00013411540929726072,
      "loss": 1.406,
      "step": 11379
    },
    {
      "epoch": 0.40754203448708076,
      "grad_norm": 1.2677370309829712,
      "learning_rate": 0.0001341045059099715,
      "loss": 1.6056,
      "step": 11380
    },
    {
      "epoch": 0.4075778466166491,
      "grad_norm": 2.6489250659942627,
      "learning_rate": 0.00013409360206385017,
      "loss": 1.6785,
      "step": 11381
    },
    {
      "epoch": 0.40761365874621736,
      "grad_norm": 1.6678003072738647,
      "learning_rate": 0.00013408269775904338,
      "loss": 1.64,
      "step": 11382
    },
    {
      "epoch": 0.4076494708757856,
      "grad_norm": 1.7335892915725708,
      "learning_rate": 0.00013407179299569787,
      "loss": 1.405,
      "step": 11383
    },
    {
      "epoch": 0.4076852830053539,
      "grad_norm": 1.8726739883422852,
      "learning_rate": 0.00013406088777396033,
      "loss": 1.3341,
      "step": 11384
    },
    {
      "epoch": 0.4077210951349222,
      "grad_norm": 1.368068814277649,
      "learning_rate": 0.00013404998209397748,
      "loss": 1.5909,
      "step": 11385
    },
    {
      "epoch": 0.4077569072644905,
      "grad_norm": 1.7381190061569214,
      "learning_rate": 0.00013403907595589605,
      "loss": 1.5986,
      "step": 11386
    },
    {
      "epoch": 0.40779271939405876,
      "grad_norm": 1.8664449453353882,
      "learning_rate": 0.0001340281693598627,
      "loss": 1.5861,
      "step": 11387
    },
    {
      "epoch": 0.407828531523627,
      "grad_norm": 1.7823396921157837,
      "learning_rate": 0.0001340172623060243,
      "loss": 1.3708,
      "step": 11388
    },
    {
      "epoch": 0.40786434365319535,
      "grad_norm": 2.0772182941436768,
      "learning_rate": 0.0001340063547945275,
      "loss": 1.5313,
      "step": 11389
    },
    {
      "epoch": 0.4079001557827636,
      "grad_norm": 1.9356260299682617,
      "learning_rate": 0.00013399544682551903,
      "loss": 1.5213,
      "step": 11390
    },
    {
      "epoch": 0.4079359679123319,
      "grad_norm": 1.6405465602874756,
      "learning_rate": 0.00013398453839914574,
      "loss": 1.6985,
      "step": 11391
    },
    {
      "epoch": 0.4079717800419002,
      "grad_norm": 2.4301764965057373,
      "learning_rate": 0.00013397362951555425,
      "loss": 1.3963,
      "step": 11392
    },
    {
      "epoch": 0.4080075921714685,
      "grad_norm": 1.3280025720596313,
      "learning_rate": 0.00013396272017489143,
      "loss": 1.4303,
      "step": 11393
    },
    {
      "epoch": 0.40804340430103675,
      "grad_norm": 2.0638201236724854,
      "learning_rate": 0.000133951810377304,
      "loss": 1.6184,
      "step": 11394
    },
    {
      "epoch": 0.408079216430605,
      "grad_norm": 1.1395586729049683,
      "learning_rate": 0.00013394090012293879,
      "loss": 1.673,
      "step": 11395
    },
    {
      "epoch": 0.40811502856017334,
      "grad_norm": 1.4466770887374878,
      "learning_rate": 0.0001339299894119425,
      "loss": 1.6637,
      "step": 11396
    },
    {
      "epoch": 0.4081508406897416,
      "grad_norm": 2.1084930896759033,
      "learning_rate": 0.00013391907824446202,
      "loss": 1.53,
      "step": 11397
    },
    {
      "epoch": 0.4081866528193099,
      "grad_norm": 1.475360631942749,
      "learning_rate": 0.00013390816662064406,
      "loss": 1.5123,
      "step": 11398
    },
    {
      "epoch": 0.4082224649488782,
      "grad_norm": 1.497314691543579,
      "learning_rate": 0.00013389725454063549,
      "loss": 1.4362,
      "step": 11399
    },
    {
      "epoch": 0.4082582770784465,
      "grad_norm": 2.5741963386535645,
      "learning_rate": 0.00013388634200458305,
      "loss": 1.6763,
      "step": 11400
    },
    {
      "epoch": 0.40829408920801474,
      "grad_norm": 2.4950594902038574,
      "learning_rate": 0.00013387542901263362,
      "loss": 1.5976,
      "step": 11401
    },
    {
      "epoch": 0.408329901337583,
      "grad_norm": 1.6021647453308105,
      "learning_rate": 0.00013386451556493396,
      "loss": 1.8064,
      "step": 11402
    },
    {
      "epoch": 0.40836571346715134,
      "grad_norm": 1.5093995332717896,
      "learning_rate": 0.00013385360166163094,
      "loss": 1.6469,
      "step": 11403
    },
    {
      "epoch": 0.4084015255967196,
      "grad_norm": 1.5861656665802002,
      "learning_rate": 0.00013384268730287136,
      "loss": 1.3995,
      "step": 11404
    },
    {
      "epoch": 0.4084373377262879,
      "grad_norm": 1.5600558519363403,
      "learning_rate": 0.0001338317724888021,
      "loss": 1.1743,
      "step": 11405
    },
    {
      "epoch": 0.4084731498558562,
      "grad_norm": 1.8158670663833618,
      "learning_rate": 0.00013382085721956997,
      "loss": 1.6269,
      "step": 11406
    },
    {
      "epoch": 0.40850896198542447,
      "grad_norm": 1.5254065990447998,
      "learning_rate": 0.00013380994149532181,
      "loss": 1.3899,
      "step": 11407
    },
    {
      "epoch": 0.40854477411499274,
      "grad_norm": 1.4239981174468994,
      "learning_rate": 0.00013379902531620455,
      "loss": 1.6413,
      "step": 11408
    },
    {
      "epoch": 0.408580586244561,
      "grad_norm": 2.2646539211273193,
      "learning_rate": 0.00013378810868236497,
      "loss": 1.7264,
      "step": 11409
    },
    {
      "epoch": 0.40861639837412933,
      "grad_norm": 1.6422070264816284,
      "learning_rate": 0.00013377719159394998,
      "loss": 1.5903,
      "step": 11410
    },
    {
      "epoch": 0.4086522105036976,
      "grad_norm": 1.7455012798309326,
      "learning_rate": 0.00013376627405110644,
      "loss": 1.4329,
      "step": 11411
    },
    {
      "epoch": 0.40868802263326587,
      "grad_norm": 1.596226453781128,
      "learning_rate": 0.00013375535605398127,
      "loss": 1.5022,
      "step": 11412
    },
    {
      "epoch": 0.4087238347628342,
      "grad_norm": 1.9514662027359009,
      "learning_rate": 0.00013374443760272127,
      "loss": 1.3195,
      "step": 11413
    },
    {
      "epoch": 0.40875964689240246,
      "grad_norm": 1.9477596282958984,
      "learning_rate": 0.0001337335186974734,
      "loss": 1.605,
      "step": 11414
    },
    {
      "epoch": 0.40879545902197073,
      "grad_norm": 1.2788459062576294,
      "learning_rate": 0.00013372259933838458,
      "loss": 1.4258,
      "step": 11415
    },
    {
      "epoch": 0.408831271151539,
      "grad_norm": 2.610154628753662,
      "learning_rate": 0.00013371167952560168,
      "loss": 1.2783,
      "step": 11416
    },
    {
      "epoch": 0.4088670832811073,
      "grad_norm": 2.161350727081299,
      "learning_rate": 0.00013370075925927158,
      "loss": 1.4277,
      "step": 11417
    },
    {
      "epoch": 0.4089028954106756,
      "grad_norm": 1.61553156375885,
      "learning_rate": 0.00013368983853954126,
      "loss": 1.5061,
      "step": 11418
    },
    {
      "epoch": 0.40893870754024386,
      "grad_norm": 1.382102370262146,
      "learning_rate": 0.00013367891736655764,
      "loss": 1.4812,
      "step": 11419
    },
    {
      "epoch": 0.4089745196698122,
      "grad_norm": 1.7254366874694824,
      "learning_rate": 0.0001336679957404676,
      "loss": 1.6011,
      "step": 11420
    },
    {
      "epoch": 0.40901033179938046,
      "grad_norm": 1.3416959047317505,
      "learning_rate": 0.00013365707366141814,
      "loss": 1.6771,
      "step": 11421
    },
    {
      "epoch": 0.4090461439289487,
      "grad_norm": 1.6707831621170044,
      "learning_rate": 0.00013364615112955612,
      "loss": 1.5072,
      "step": 11422
    },
    {
      "epoch": 0.409081956058517,
      "grad_norm": 2.198908567428589,
      "learning_rate": 0.0001336352281450286,
      "loss": 1.7351,
      "step": 11423
    },
    {
      "epoch": 0.4091177681880853,
      "grad_norm": 1.4816632270812988,
      "learning_rate": 0.0001336243047079824,
      "loss": 1.5977,
      "step": 11424
    },
    {
      "epoch": 0.4091535803176536,
      "grad_norm": 1.7870755195617676,
      "learning_rate": 0.00013361338081856457,
      "loss": 1.6438,
      "step": 11425
    },
    {
      "epoch": 0.40918939244722186,
      "grad_norm": 1.7458114624023438,
      "learning_rate": 0.0001336024564769221,
      "loss": 1.4006,
      "step": 11426
    },
    {
      "epoch": 0.4092252045767902,
      "grad_norm": 1.7427453994750977,
      "learning_rate": 0.00013359153168320188,
      "loss": 1.3585,
      "step": 11427
    },
    {
      "epoch": 0.40926101670635845,
      "grad_norm": 1.8659039735794067,
      "learning_rate": 0.00013358060643755098,
      "loss": 1.67,
      "step": 11428
    },
    {
      "epoch": 0.4092968288359267,
      "grad_norm": 1.4073402881622314,
      "learning_rate": 0.00013356968074011626,
      "loss": 1.4902,
      "step": 11429
    },
    {
      "epoch": 0.409332640965495,
      "grad_norm": 1.6928961277008057,
      "learning_rate": 0.00013355875459104485,
      "loss": 1.6261,
      "step": 11430
    },
    {
      "epoch": 0.4093684530950633,
      "grad_norm": 1.7611867189407349,
      "learning_rate": 0.00013354782799048366,
      "loss": 1.3424,
      "step": 11431
    },
    {
      "epoch": 0.4094042652246316,
      "grad_norm": 1.9710921049118042,
      "learning_rate": 0.00013353690093857972,
      "loss": 1.4281,
      "step": 11432
    },
    {
      "epoch": 0.40944007735419985,
      "grad_norm": 1.504773497581482,
      "learning_rate": 0.00013352597343548004,
      "loss": 1.4062,
      "step": 11433
    },
    {
      "epoch": 0.4094758894837682,
      "grad_norm": 1.5385361909866333,
      "learning_rate": 0.00013351504548133166,
      "loss": 1.7332,
      "step": 11434
    },
    {
      "epoch": 0.40951170161333644,
      "grad_norm": 2.2584147453308105,
      "learning_rate": 0.00013350411707628153,
      "loss": 1.75,
      "step": 11435
    },
    {
      "epoch": 0.4095475137429047,
      "grad_norm": 1.3944424390792847,
      "learning_rate": 0.00013349318822047674,
      "loss": 1.35,
      "step": 11436
    },
    {
      "epoch": 0.409583325872473,
      "grad_norm": 1.3100533485412598,
      "learning_rate": 0.00013348225891406432,
      "loss": 1.429,
      "step": 11437
    },
    {
      "epoch": 0.4096191380020413,
      "grad_norm": 2.255528688430786,
      "learning_rate": 0.00013347132915719127,
      "loss": 1.749,
      "step": 11438
    },
    {
      "epoch": 0.4096549501316096,
      "grad_norm": 2.328932523727417,
      "learning_rate": 0.0001334603989500047,
      "loss": 1.3934,
      "step": 11439
    },
    {
      "epoch": 0.40969076226117784,
      "grad_norm": 1.5605809688568115,
      "learning_rate": 0.00013344946829265157,
      "loss": 1.5377,
      "step": 11440
    },
    {
      "epoch": 0.40972657439074617,
      "grad_norm": 2.7059600353240967,
      "learning_rate": 0.000133438537185279,
      "loss": 1.8359,
      "step": 11441
    },
    {
      "epoch": 0.40976238652031444,
      "grad_norm": 1.4177329540252686,
      "learning_rate": 0.00013342760562803406,
      "loss": 1.5699,
      "step": 11442
    },
    {
      "epoch": 0.4097981986498827,
      "grad_norm": 2.09751296043396,
      "learning_rate": 0.0001334166736210638,
      "loss": 1.5161,
      "step": 11443
    },
    {
      "epoch": 0.409834010779451,
      "grad_norm": 1.62708580493927,
      "learning_rate": 0.00013340574116451533,
      "loss": 1.3609,
      "step": 11444
    },
    {
      "epoch": 0.4098698229090193,
      "grad_norm": 1.794075608253479,
      "learning_rate": 0.0001333948082585357,
      "loss": 1.4662,
      "step": 11445
    },
    {
      "epoch": 0.40990563503858757,
      "grad_norm": 1.4780937433242798,
      "learning_rate": 0.00013338387490327195,
      "loss": 1.3648,
      "step": 11446
    },
    {
      "epoch": 0.40994144716815584,
      "grad_norm": 1.2816733121871948,
      "learning_rate": 0.00013337294109887123,
      "loss": 1.3649,
      "step": 11447
    },
    {
      "epoch": 0.40997725929772416,
      "grad_norm": 1.4756011962890625,
      "learning_rate": 0.0001333620068454807,
      "loss": 1.4781,
      "step": 11448
    },
    {
      "epoch": 0.41001307142729243,
      "grad_norm": 1.638892412185669,
      "learning_rate": 0.00013335107214324733,
      "loss": 1.7207,
      "step": 11449
    },
    {
      "epoch": 0.4100488835568607,
      "grad_norm": 1.8446863889694214,
      "learning_rate": 0.00013334013699231836,
      "loss": 1.3898,
      "step": 11450
    },
    {
      "epoch": 0.41008469568642897,
      "grad_norm": 1.9317458868026733,
      "learning_rate": 0.0001333292013928408,
      "loss": 1.6734,
      "step": 11451
    },
    {
      "epoch": 0.4101205078159973,
      "grad_norm": 1.8174721002578735,
      "learning_rate": 0.00013331826534496188,
      "loss": 1.6215,
      "step": 11452
    },
    {
      "epoch": 0.41015631994556556,
      "grad_norm": 1.832736849784851,
      "learning_rate": 0.00013330732884882866,
      "loss": 1.4646,
      "step": 11453
    },
    {
      "epoch": 0.41019213207513383,
      "grad_norm": 2.605854034423828,
      "learning_rate": 0.0001332963919045883,
      "loss": 1.5914,
      "step": 11454
    },
    {
      "epoch": 0.41022794420470216,
      "grad_norm": 2.3421616554260254,
      "learning_rate": 0.00013328545451238793,
      "loss": 1.5082,
      "step": 11455
    },
    {
      "epoch": 0.4102637563342704,
      "grad_norm": 1.8743304014205933,
      "learning_rate": 0.00013327451667237468,
      "loss": 1.3477,
      "step": 11456
    },
    {
      "epoch": 0.4102995684638387,
      "grad_norm": 1.307234764099121,
      "learning_rate": 0.00013326357838469574,
      "loss": 1.5954,
      "step": 11457
    },
    {
      "epoch": 0.41033538059340696,
      "grad_norm": 1.5379959344863892,
      "learning_rate": 0.0001332526396494983,
      "loss": 1.4679,
      "step": 11458
    },
    {
      "epoch": 0.4103711927229753,
      "grad_norm": 1.7482596635818481,
      "learning_rate": 0.00013324170046692942,
      "loss": 1.5744,
      "step": 11459
    },
    {
      "epoch": 0.41040700485254356,
      "grad_norm": 1.8453181982040405,
      "learning_rate": 0.00013323076083713637,
      "loss": 1.4618,
      "step": 11460
    },
    {
      "epoch": 0.4104428169821118,
      "grad_norm": 2.0559778213500977,
      "learning_rate": 0.00013321982076026632,
      "loss": 1.632,
      "step": 11461
    },
    {
      "epoch": 0.41047862911168015,
      "grad_norm": 1.8249529600143433,
      "learning_rate": 0.0001332088802364664,
      "loss": 1.6765,
      "step": 11462
    },
    {
      "epoch": 0.4105144412412484,
      "grad_norm": 1.922978162765503,
      "learning_rate": 0.00013319793926588387,
      "loss": 1.1515,
      "step": 11463
    },
    {
      "epoch": 0.4105502533708167,
      "grad_norm": 1.7666670083999634,
      "learning_rate": 0.00013318699784866585,
      "loss": 1.6148,
      "step": 11464
    },
    {
      "epoch": 0.41058606550038496,
      "grad_norm": 1.6551225185394287,
      "learning_rate": 0.0001331760559849596,
      "loss": 1.2441,
      "step": 11465
    },
    {
      "epoch": 0.4106218776299533,
      "grad_norm": 1.6817725896835327,
      "learning_rate": 0.0001331651136749123,
      "loss": 1.3188,
      "step": 11466
    },
    {
      "epoch": 0.41065768975952155,
      "grad_norm": 1.4896317720413208,
      "learning_rate": 0.0001331541709186712,
      "loss": 1.3486,
      "step": 11467
    },
    {
      "epoch": 0.4106935018890898,
      "grad_norm": 1.92238450050354,
      "learning_rate": 0.00013314322771638346,
      "loss": 1.8901,
      "step": 11468
    },
    {
      "epoch": 0.41072931401865814,
      "grad_norm": 1.4980189800262451,
      "learning_rate": 0.00013313228406819637,
      "loss": 1.5268,
      "step": 11469
    },
    {
      "epoch": 0.4107651261482264,
      "grad_norm": 1.5852466821670532,
      "learning_rate": 0.00013312133997425712,
      "loss": 1.697,
      "step": 11470
    },
    {
      "epoch": 0.4108009382777947,
      "grad_norm": 1.6819641590118408,
      "learning_rate": 0.00013311039543471297,
      "loss": 1.4556,
      "step": 11471
    },
    {
      "epoch": 0.41083675040736295,
      "grad_norm": 1.6359261274337769,
      "learning_rate": 0.00013309945044971116,
      "loss": 1.2987,
      "step": 11472
    },
    {
      "epoch": 0.4108725625369313,
      "grad_norm": 1.8399405479431152,
      "learning_rate": 0.00013308850501939892,
      "loss": 1.6601,
      "step": 11473
    },
    {
      "epoch": 0.41090837466649954,
      "grad_norm": 1.4478241205215454,
      "learning_rate": 0.00013307755914392357,
      "loss": 1.2634,
      "step": 11474
    },
    {
      "epoch": 0.4109441867960678,
      "grad_norm": 1.807931661605835,
      "learning_rate": 0.0001330666128234323,
      "loss": 1.7326,
      "step": 11475
    },
    {
      "epoch": 0.41097999892563614,
      "grad_norm": 1.5771223306655884,
      "learning_rate": 0.0001330556660580724,
      "loss": 1.25,
      "step": 11476
    },
    {
      "epoch": 0.4110158110552044,
      "grad_norm": 1.909822702407837,
      "learning_rate": 0.00013304471884799116,
      "loss": 1.7395,
      "step": 11477
    },
    {
      "epoch": 0.4110516231847727,
      "grad_norm": 1.6032434701919556,
      "learning_rate": 0.00013303377119333587,
      "loss": 1.6306,
      "step": 11478
    },
    {
      "epoch": 0.41108743531434094,
      "grad_norm": 1.5883413553237915,
      "learning_rate": 0.0001330228230942538,
      "loss": 1.2959,
      "step": 11479
    },
    {
      "epoch": 0.41112324744390927,
      "grad_norm": 1.9156928062438965,
      "learning_rate": 0.00013301187455089223,
      "loss": 1.936,
      "step": 11480
    },
    {
      "epoch": 0.41115905957347754,
      "grad_norm": 3.1075665950775146,
      "learning_rate": 0.00013300092556339847,
      "loss": 1.8545,
      "step": 11481
    },
    {
      "epoch": 0.4111948717030458,
      "grad_norm": 1.515472412109375,
      "learning_rate": 0.00013298997613191978,
      "loss": 1.5118,
      "step": 11482
    },
    {
      "epoch": 0.41123068383261413,
      "grad_norm": 1.5612744092941284,
      "learning_rate": 0.00013297902625660358,
      "loss": 1.3228,
      "step": 11483
    },
    {
      "epoch": 0.4112664959621824,
      "grad_norm": 2.1935417652130127,
      "learning_rate": 0.00013296807593759708,
      "loss": 1.5941,
      "step": 11484
    },
    {
      "epoch": 0.41130230809175067,
      "grad_norm": 1.5191771984100342,
      "learning_rate": 0.0001329571251750477,
      "loss": 1.5533,
      "step": 11485
    },
    {
      "epoch": 0.41133812022131894,
      "grad_norm": 1.5615873336791992,
      "learning_rate": 0.00013294617396910266,
      "loss": 1.7188,
      "step": 11486
    },
    {
      "epoch": 0.41137393235088726,
      "grad_norm": 1.4643330574035645,
      "learning_rate": 0.00013293522231990935,
      "loss": 1.582,
      "step": 11487
    },
    {
      "epoch": 0.41140974448045553,
      "grad_norm": 2.06705379486084,
      "learning_rate": 0.00013292427022761514,
      "loss": 1.5501,
      "step": 11488
    },
    {
      "epoch": 0.4114455566100238,
      "grad_norm": 1.5073274374008179,
      "learning_rate": 0.0001329133176923673,
      "loss": 1.5135,
      "step": 11489
    },
    {
      "epoch": 0.4114813687395921,
      "grad_norm": 1.912886381149292,
      "learning_rate": 0.00013290236471431326,
      "loss": 1.4496,
      "step": 11490
    },
    {
      "epoch": 0.4115171808691604,
      "grad_norm": 1.8361574411392212,
      "learning_rate": 0.00013289141129360033,
      "loss": 1.6251,
      "step": 11491
    },
    {
      "epoch": 0.41155299299872866,
      "grad_norm": 2.233924388885498,
      "learning_rate": 0.0001328804574303759,
      "loss": 1.8372,
      "step": 11492
    },
    {
      "epoch": 0.41158880512829693,
      "grad_norm": 1.716523289680481,
      "learning_rate": 0.0001328695031247873,
      "loss": 1.6532,
      "step": 11493
    },
    {
      "epoch": 0.41162461725786526,
      "grad_norm": 1.9167834520339966,
      "learning_rate": 0.00013285854837698195,
      "loss": 1.5691,
      "step": 11494
    },
    {
      "epoch": 0.4116604293874335,
      "grad_norm": 1.637646198272705,
      "learning_rate": 0.0001328475931871072,
      "loss": 1.5658,
      "step": 11495
    },
    {
      "epoch": 0.4116962415170018,
      "grad_norm": 1.4613744020462036,
      "learning_rate": 0.0001328366375553105,
      "loss": 1.2251,
      "step": 11496
    },
    {
      "epoch": 0.4117320536465701,
      "grad_norm": 2.3519818782806396,
      "learning_rate": 0.00013282568148173917,
      "loss": 1.4193,
      "step": 11497
    },
    {
      "epoch": 0.4117678657761384,
      "grad_norm": 1.911476731300354,
      "learning_rate": 0.00013281472496654064,
      "loss": 1.3368,
      "step": 11498
    },
    {
      "epoch": 0.41180367790570666,
      "grad_norm": 1.8679819107055664,
      "learning_rate": 0.0001328037680098623,
      "loss": 1.4186,
      "step": 11499
    },
    {
      "epoch": 0.4118394900352749,
      "grad_norm": 1.6194400787353516,
      "learning_rate": 0.00013279281061185158,
      "loss": 1.4609,
      "step": 11500
    },
    {
      "epoch": 0.41187530216484325,
      "grad_norm": 1.498268961906433,
      "learning_rate": 0.0001327818527726559,
      "loss": 1.4629,
      "step": 11501
    },
    {
      "epoch": 0.4119111142944115,
      "grad_norm": 1.735053300857544,
      "learning_rate": 0.00013277089449242267,
      "loss": 1.3692,
      "step": 11502
    },
    {
      "epoch": 0.4119469264239798,
      "grad_norm": 1.634087085723877,
      "learning_rate": 0.00013275993577129932,
      "loss": 1.7921,
      "step": 11503
    },
    {
      "epoch": 0.4119827385535481,
      "grad_norm": 2.1054511070251465,
      "learning_rate": 0.0001327489766094333,
      "loss": 1.3704,
      "step": 11504
    },
    {
      "epoch": 0.4120185506831164,
      "grad_norm": 1.6996413469314575,
      "learning_rate": 0.00013273801700697206,
      "loss": 1.7709,
      "step": 11505
    },
    {
      "epoch": 0.41205436281268465,
      "grad_norm": 1.709455966949463,
      "learning_rate": 0.00013272705696406302,
      "loss": 1.5758,
      "step": 11506
    },
    {
      "epoch": 0.4120901749422529,
      "grad_norm": 1.550470232963562,
      "learning_rate": 0.00013271609648085367,
      "loss": 1.5027,
      "step": 11507
    },
    {
      "epoch": 0.41212598707182124,
      "grad_norm": 1.6597813367843628,
      "learning_rate": 0.0001327051355574914,
      "loss": 1.7267,
      "step": 11508
    },
    {
      "epoch": 0.4121617992013895,
      "grad_norm": 1.2792216539382935,
      "learning_rate": 0.0001326941741941237,
      "loss": 1.3561,
      "step": 11509
    },
    {
      "epoch": 0.4121976113309578,
      "grad_norm": 1.414903163909912,
      "learning_rate": 0.00013268321239089809,
      "loss": 1.5848,
      "step": 11510
    },
    {
      "epoch": 0.4122334234605261,
      "grad_norm": 1.810530185699463,
      "learning_rate": 0.00013267225014796202,
      "loss": 1.4441,
      "step": 11511
    },
    {
      "epoch": 0.4122692355900944,
      "grad_norm": 1.5842233896255493,
      "learning_rate": 0.00013266128746546296,
      "loss": 1.1982,
      "step": 11512
    },
    {
      "epoch": 0.41230504771966264,
      "grad_norm": 1.3252923488616943,
      "learning_rate": 0.0001326503243435484,
      "loss": 1.4864,
      "step": 11513
    },
    {
      "epoch": 0.4123408598492309,
      "grad_norm": 1.435280203819275,
      "learning_rate": 0.00013263936078236586,
      "loss": 1.2947,
      "step": 11514
    },
    {
      "epoch": 0.41237667197879924,
      "grad_norm": 2.3701577186584473,
      "learning_rate": 0.00013262839678206283,
      "loss": 1.7331,
      "step": 11515
    },
    {
      "epoch": 0.4124124841083675,
      "grad_norm": 1.5976884365081787,
      "learning_rate": 0.00013261743234278678,
      "loss": 1.4953,
      "step": 11516
    },
    {
      "epoch": 0.4124482962379358,
      "grad_norm": 1.6659822463989258,
      "learning_rate": 0.00013260646746468527,
      "loss": 1.1783,
      "step": 11517
    },
    {
      "epoch": 0.4124841083675041,
      "grad_norm": 1.7905832529067993,
      "learning_rate": 0.0001325955021479058,
      "loss": 1.5895,
      "step": 11518
    },
    {
      "epoch": 0.41251992049707237,
      "grad_norm": 1.614429235458374,
      "learning_rate": 0.00013258453639259586,
      "loss": 1.3956,
      "step": 11519
    },
    {
      "epoch": 0.41255573262664064,
      "grad_norm": 1.761165738105774,
      "learning_rate": 0.00013257357019890307,
      "loss": 1.3419,
      "step": 11520
    },
    {
      "epoch": 0.4125915447562089,
      "grad_norm": 2.056187868118286,
      "learning_rate": 0.00013256260356697485,
      "loss": 1.4939,
      "step": 11521
    },
    {
      "epoch": 0.41262735688577723,
      "grad_norm": 1.6465550661087036,
      "learning_rate": 0.00013255163649695886,
      "loss": 1.6814,
      "step": 11522
    },
    {
      "epoch": 0.4126631690153455,
      "grad_norm": 2.2140185832977295,
      "learning_rate": 0.00013254066898900257,
      "loss": 1.5524,
      "step": 11523
    },
    {
      "epoch": 0.41269898114491377,
      "grad_norm": 1.8904930353164673,
      "learning_rate": 0.00013252970104325352,
      "loss": 1.4541,
      "step": 11524
    },
    {
      "epoch": 0.4127347932744821,
      "grad_norm": 1.2835383415222168,
      "learning_rate": 0.00013251873265985936,
      "loss": 1.469,
      "step": 11525
    },
    {
      "epoch": 0.41277060540405036,
      "grad_norm": 1.3030658960342407,
      "learning_rate": 0.00013250776383896752,
      "loss": 1.3568,
      "step": 11526
    },
    {
      "epoch": 0.41280641753361863,
      "grad_norm": 1.3432520627975464,
      "learning_rate": 0.00013249679458072572,
      "loss": 1.4884,
      "step": 11527
    },
    {
      "epoch": 0.4128422296631869,
      "grad_norm": 2.3544211387634277,
      "learning_rate": 0.00013248582488528142,
      "loss": 1.2597,
      "step": 11528
    },
    {
      "epoch": 0.4128780417927552,
      "grad_norm": 1.6606460809707642,
      "learning_rate": 0.0001324748547527823,
      "loss": 1.4726,
      "step": 11529
    },
    {
      "epoch": 0.4129138539223235,
      "grad_norm": 1.4075069427490234,
      "learning_rate": 0.00013246388418337586,
      "loss": 1.2732,
      "step": 11530
    },
    {
      "epoch": 0.41294966605189176,
      "grad_norm": 1.6073061227798462,
      "learning_rate": 0.00013245291317720974,
      "loss": 1.3229,
      "step": 11531
    },
    {
      "epoch": 0.4129854781814601,
      "grad_norm": 1.994727373123169,
      "learning_rate": 0.00013244194173443155,
      "loss": 1.3324,
      "step": 11532
    },
    {
      "epoch": 0.41302129031102836,
      "grad_norm": 1.824912190437317,
      "learning_rate": 0.00013243096985518887,
      "loss": 1.3809,
      "step": 11533
    },
    {
      "epoch": 0.4130571024405966,
      "grad_norm": 1.7901192903518677,
      "learning_rate": 0.00013241999753962932,
      "loss": 1.2446,
      "step": 11534
    },
    {
      "epoch": 0.4130929145701649,
      "grad_norm": 1.5490052700042725,
      "learning_rate": 0.00013240902478790052,
      "loss": 1.373,
      "step": 11535
    },
    {
      "epoch": 0.4131287266997332,
      "grad_norm": 1.581001877784729,
      "learning_rate": 0.0001323980516001501,
      "loss": 1.3801,
      "step": 11536
    },
    {
      "epoch": 0.4131645388293015,
      "grad_norm": 2.0129377841949463,
      "learning_rate": 0.00013238707797652569,
      "loss": 1.6714,
      "step": 11537
    },
    {
      "epoch": 0.41320035095886976,
      "grad_norm": 1.7160298824310303,
      "learning_rate": 0.0001323761039171749,
      "loss": 1.2577,
      "step": 11538
    },
    {
      "epoch": 0.4132361630884381,
      "grad_norm": 1.5854284763336182,
      "learning_rate": 0.00013236512942224545,
      "loss": 1.5169,
      "step": 11539
    },
    {
      "epoch": 0.41327197521800635,
      "grad_norm": 2.3109540939331055,
      "learning_rate": 0.0001323541544918849,
      "loss": 1.8822,
      "step": 11540
    },
    {
      "epoch": 0.4133077873475746,
      "grad_norm": 2.023212194442749,
      "learning_rate": 0.00013234317912624093,
      "loss": 1.5002,
      "step": 11541
    },
    {
      "epoch": 0.4133435994771429,
      "grad_norm": 2.173719644546509,
      "learning_rate": 0.0001323322033254612,
      "loss": 1.6894,
      "step": 11542
    },
    {
      "epoch": 0.4133794116067112,
      "grad_norm": 1.548795223236084,
      "learning_rate": 0.00013232122708969337,
      "loss": 1.3606,
      "step": 11543
    },
    {
      "epoch": 0.4134152237362795,
      "grad_norm": 1.749751329421997,
      "learning_rate": 0.00013231025041908514,
      "loss": 1.8365,
      "step": 11544
    },
    {
      "epoch": 0.41345103586584775,
      "grad_norm": 1.4382846355438232,
      "learning_rate": 0.00013229927331378418,
      "loss": 1.3604,
      "step": 11545
    },
    {
      "epoch": 0.4134868479954161,
      "grad_norm": 2.0954902172088623,
      "learning_rate": 0.0001322882957739381,
      "loss": 1.538,
      "step": 11546
    },
    {
      "epoch": 0.41352266012498434,
      "grad_norm": 1.5251587629318237,
      "learning_rate": 0.00013227731779969472,
      "loss": 1.2223,
      "step": 11547
    },
    {
      "epoch": 0.4135584722545526,
      "grad_norm": 2.0528721809387207,
      "learning_rate": 0.00013226633939120164,
      "loss": 1.2464,
      "step": 11548
    },
    {
      "epoch": 0.4135942843841209,
      "grad_norm": 1.6846539974212646,
      "learning_rate": 0.00013225536054860658,
      "loss": 1.3667,
      "step": 11549
    },
    {
      "epoch": 0.4136300965136892,
      "grad_norm": 2.08156418800354,
      "learning_rate": 0.00013224438127205725,
      "loss": 1.2521,
      "step": 11550
    },
    {
      "epoch": 0.4136659086432575,
      "grad_norm": 1.3722084760665894,
      "learning_rate": 0.0001322334015617014,
      "loss": 1.5203,
      "step": 11551
    },
    {
      "epoch": 0.41370172077282574,
      "grad_norm": 1.3709852695465088,
      "learning_rate": 0.00013222242141768664,
      "loss": 0.9214,
      "step": 11552
    },
    {
      "epoch": 0.41373753290239407,
      "grad_norm": 1.7349940538406372,
      "learning_rate": 0.00013221144084016082,
      "loss": 1.5187,
      "step": 11553
    },
    {
      "epoch": 0.41377334503196234,
      "grad_norm": 1.7656437158584595,
      "learning_rate": 0.00013220045982927157,
      "loss": 1.4667,
      "step": 11554
    },
    {
      "epoch": 0.4138091571615306,
      "grad_norm": 1.6982886791229248,
      "learning_rate": 0.00013218947838516672,
      "loss": 1.5724,
      "step": 11555
    },
    {
      "epoch": 0.4138449692910989,
      "grad_norm": 2.2578091621398926,
      "learning_rate": 0.0001321784965079939,
      "loss": 1.9695,
      "step": 11556
    },
    {
      "epoch": 0.4138807814206672,
      "grad_norm": 1.5725754499435425,
      "learning_rate": 0.00013216751419790096,
      "loss": 1.4167,
      "step": 11557
    },
    {
      "epoch": 0.41391659355023547,
      "grad_norm": 1.307800054550171,
      "learning_rate": 0.00013215653145503558,
      "loss": 1.4289,
      "step": 11558
    },
    {
      "epoch": 0.41395240567980374,
      "grad_norm": 1.5263270139694214,
      "learning_rate": 0.00013214554827954556,
      "loss": 1.4657,
      "step": 11559
    },
    {
      "epoch": 0.41398821780937206,
      "grad_norm": 1.9524672031402588,
      "learning_rate": 0.00013213456467157868,
      "loss": 1.4146,
      "step": 11560
    },
    {
      "epoch": 0.41402402993894033,
      "grad_norm": 1.5733195543289185,
      "learning_rate": 0.00013212358063128266,
      "loss": 1.2555,
      "step": 11561
    },
    {
      "epoch": 0.4140598420685086,
      "grad_norm": 1.4676419496536255,
      "learning_rate": 0.0001321125961588053,
      "loss": 1.4726,
      "step": 11562
    },
    {
      "epoch": 0.41409565419807687,
      "grad_norm": 2.0057151317596436,
      "learning_rate": 0.00013210161125429436,
      "loss": 1.6748,
      "step": 11563
    },
    {
      "epoch": 0.4141314663276452,
      "grad_norm": 1.7160898447036743,
      "learning_rate": 0.0001320906259178977,
      "loss": 1.6463,
      "step": 11564
    },
    {
      "epoch": 0.41416727845721346,
      "grad_norm": 1.602569818496704,
      "learning_rate": 0.00013207964014976299,
      "loss": 1.6806,
      "step": 11565
    },
    {
      "epoch": 0.41420309058678173,
      "grad_norm": 1.489131212234497,
      "learning_rate": 0.00013206865395003816,
      "loss": 1.2136,
      "step": 11566
    },
    {
      "epoch": 0.41423890271635005,
      "grad_norm": 1.3555222749710083,
      "learning_rate": 0.00013205766731887094,
      "loss": 1.4043,
      "step": 11567
    },
    {
      "epoch": 0.4142747148459183,
      "grad_norm": 2.4561402797698975,
      "learning_rate": 0.00013204668025640915,
      "loss": 1.5604,
      "step": 11568
    },
    {
      "epoch": 0.4143105269754866,
      "grad_norm": 1.891719102859497,
      "learning_rate": 0.00013203569276280062,
      "loss": 1.7261,
      "step": 11569
    },
    {
      "epoch": 0.41434633910505486,
      "grad_norm": 1.5649393796920776,
      "learning_rate": 0.00013202470483819316,
      "loss": 1.3593,
      "step": 11570
    },
    {
      "epoch": 0.4143821512346232,
      "grad_norm": 1.5750397443771362,
      "learning_rate": 0.00013201371648273463,
      "loss": 1.5991,
      "step": 11571
    },
    {
      "epoch": 0.41441796336419146,
      "grad_norm": 2.288243055343628,
      "learning_rate": 0.00013200272769657283,
      "loss": 1.5515,
      "step": 11572
    },
    {
      "epoch": 0.4144537754937597,
      "grad_norm": 1.4258782863616943,
      "learning_rate": 0.00013199173847985559,
      "loss": 1.3312,
      "step": 11573
    },
    {
      "epoch": 0.41448958762332805,
      "grad_norm": 1.5753511190414429,
      "learning_rate": 0.0001319807488327308,
      "loss": 1.4183,
      "step": 11574
    },
    {
      "epoch": 0.4145253997528963,
      "grad_norm": 1.7554752826690674,
      "learning_rate": 0.00013196975875534624,
      "loss": 1.6035,
      "step": 11575
    },
    {
      "epoch": 0.4145612118824646,
      "grad_norm": 2.1951706409454346,
      "learning_rate": 0.00013195876824784988,
      "loss": 1.4181,
      "step": 11576
    },
    {
      "epoch": 0.41459702401203286,
      "grad_norm": 1.4915632009506226,
      "learning_rate": 0.00013194777731038946,
      "loss": 1.5221,
      "step": 11577
    },
    {
      "epoch": 0.4146328361416012,
      "grad_norm": 1.97896409034729,
      "learning_rate": 0.00013193678594311295,
      "loss": 1.6528,
      "step": 11578
    },
    {
      "epoch": 0.41466864827116945,
      "grad_norm": 1.9763433933258057,
      "learning_rate": 0.00013192579414616815,
      "loss": 1.324,
      "step": 11579
    },
    {
      "epoch": 0.4147044604007377,
      "grad_norm": 1.8413565158843994,
      "learning_rate": 0.000131914801919703,
      "loss": 1.4303,
      "step": 11580
    },
    {
      "epoch": 0.41474027253030604,
      "grad_norm": 1.5440016984939575,
      "learning_rate": 0.0001319038092638653,
      "loss": 1.5037,
      "step": 11581
    },
    {
      "epoch": 0.4147760846598743,
      "grad_norm": 1.447457194328308,
      "learning_rate": 0.00013189281617880308,
      "loss": 1.3649,
      "step": 11582
    },
    {
      "epoch": 0.4148118967894426,
      "grad_norm": 1.756569743156433,
      "learning_rate": 0.0001318818226646641,
      "loss": 1.4381,
      "step": 11583
    },
    {
      "epoch": 0.41484770891901085,
      "grad_norm": 2.2443044185638428,
      "learning_rate": 0.00013187082872159636,
      "loss": 1.5342,
      "step": 11584
    },
    {
      "epoch": 0.4148835210485792,
      "grad_norm": 1.3705558776855469,
      "learning_rate": 0.0001318598343497477,
      "loss": 1.2355,
      "step": 11585
    },
    {
      "epoch": 0.41491933317814744,
      "grad_norm": 2.452343463897705,
      "learning_rate": 0.00013184883954926607,
      "loss": 1.2603,
      "step": 11586
    },
    {
      "epoch": 0.4149551453077157,
      "grad_norm": 1.5595102310180664,
      "learning_rate": 0.0001318378443202994,
      "loss": 1.6226,
      "step": 11587
    },
    {
      "epoch": 0.414990957437284,
      "grad_norm": 2.130498170852661,
      "learning_rate": 0.00013182684866299557,
      "loss": 1.63,
      "step": 11588
    },
    {
      "epoch": 0.4150267695668523,
      "grad_norm": 1.8103166818618774,
      "learning_rate": 0.00013181585257750257,
      "loss": 1.4003,
      "step": 11589
    },
    {
      "epoch": 0.4150625816964206,
      "grad_norm": 1.686850905418396,
      "learning_rate": 0.0001318048560639683,
      "loss": 1.3748,
      "step": 11590
    },
    {
      "epoch": 0.41509839382598884,
      "grad_norm": 1.4466736316680908,
      "learning_rate": 0.00013179385912254072,
      "loss": 1.5537,
      "step": 11591
    },
    {
      "epoch": 0.41513420595555717,
      "grad_norm": 1.7150828838348389,
      "learning_rate": 0.00013178286175336777,
      "loss": 1.1753,
      "step": 11592
    },
    {
      "epoch": 0.41517001808512544,
      "grad_norm": 1.8393244743347168,
      "learning_rate": 0.00013177186395659743,
      "loss": 1.7963,
      "step": 11593
    },
    {
      "epoch": 0.4152058302146937,
      "grad_norm": 1.8172906637191772,
      "learning_rate": 0.00013176086573237766,
      "loss": 1.5713,
      "step": 11594
    },
    {
      "epoch": 0.415241642344262,
      "grad_norm": 1.7263849973678589,
      "learning_rate": 0.0001317498670808564,
      "loss": 1.6143,
      "step": 11595
    },
    {
      "epoch": 0.4152774544738303,
      "grad_norm": 1.7366745471954346,
      "learning_rate": 0.0001317388680021816,
      "loss": 1.6609,
      "step": 11596
    },
    {
      "epoch": 0.41531326660339857,
      "grad_norm": 1.5991458892822266,
      "learning_rate": 0.00013172786849650133,
      "loss": 1.6754,
      "step": 11597
    },
    {
      "epoch": 0.41534907873296684,
      "grad_norm": 1.716823697090149,
      "learning_rate": 0.00013171686856396344,
      "loss": 1.608,
      "step": 11598
    },
    {
      "epoch": 0.41538489086253516,
      "grad_norm": 1.7387841939926147,
      "learning_rate": 0.00013170586820471605,
      "loss": 1.3947,
      "step": 11599
    },
    {
      "epoch": 0.41542070299210343,
      "grad_norm": 1.3563843965530396,
      "learning_rate": 0.00013169486741890706,
      "loss": 1.4469,
      "step": 11600
    },
    {
      "epoch": 0.4154565151216717,
      "grad_norm": 1.6906555891036987,
      "learning_rate": 0.0001316838662066845,
      "loss": 1.3223,
      "step": 11601
    },
    {
      "epoch": 0.41549232725123997,
      "grad_norm": 1.6035621166229248,
      "learning_rate": 0.00013167286456819646,
      "loss": 1.6543,
      "step": 11602
    },
    {
      "epoch": 0.4155281393808083,
      "grad_norm": 1.3639655113220215,
      "learning_rate": 0.00013166186250359086,
      "loss": 1.6864,
      "step": 11603
    },
    {
      "epoch": 0.41556395151037656,
      "grad_norm": 2.101548910140991,
      "learning_rate": 0.00013165086001301575,
      "loss": 1.5005,
      "step": 11604
    },
    {
      "epoch": 0.41559976363994483,
      "grad_norm": 1.2370082139968872,
      "learning_rate": 0.0001316398570966191,
      "loss": 1.4163,
      "step": 11605
    },
    {
      "epoch": 0.41563557576951315,
      "grad_norm": 1.4403676986694336,
      "learning_rate": 0.000131628853754549,
      "loss": 1.7009,
      "step": 11606
    },
    {
      "epoch": 0.4156713878990814,
      "grad_norm": 1.7065383195877075,
      "learning_rate": 0.00013161784998695349,
      "loss": 1.6103,
      "step": 11607
    },
    {
      "epoch": 0.4157072000286497,
      "grad_norm": 3.6323750019073486,
      "learning_rate": 0.00013160684579398057,
      "loss": 1.4806,
      "step": 11608
    },
    {
      "epoch": 0.41574301215821796,
      "grad_norm": 1.4103000164031982,
      "learning_rate": 0.00013159584117577831,
      "loss": 1.3592,
      "step": 11609
    },
    {
      "epoch": 0.4157788242877863,
      "grad_norm": 1.6261008977890015,
      "learning_rate": 0.0001315848361324948,
      "loss": 1.591,
      "step": 11610
    },
    {
      "epoch": 0.41581463641735456,
      "grad_norm": 2.0343966484069824,
      "learning_rate": 0.000131573830664278,
      "loss": 1.6907,
      "step": 11611
    },
    {
      "epoch": 0.4158504485469228,
      "grad_norm": 1.7296793460845947,
      "learning_rate": 0.0001315628247712761,
      "loss": 1.5331,
      "step": 11612
    },
    {
      "epoch": 0.41588626067649115,
      "grad_norm": 1.7771193981170654,
      "learning_rate": 0.0001315518184536371,
      "loss": 1.5747,
      "step": 11613
    },
    {
      "epoch": 0.4159220728060594,
      "grad_norm": 1.993929147720337,
      "learning_rate": 0.00013154081171150902,
      "loss": 1.7395,
      "step": 11614
    },
    {
      "epoch": 0.4159578849356277,
      "grad_norm": 1.6930261850357056,
      "learning_rate": 0.00013152980454504007,
      "loss": 1.7027,
      "step": 11615
    },
    {
      "epoch": 0.41599369706519596,
      "grad_norm": 2.1392862796783447,
      "learning_rate": 0.00013151879695437823,
      "loss": 2.0039,
      "step": 11616
    },
    {
      "epoch": 0.4160295091947643,
      "grad_norm": 1.4961529970169067,
      "learning_rate": 0.00013150778893967165,
      "loss": 1.8718,
      "step": 11617
    },
    {
      "epoch": 0.41606532132433255,
      "grad_norm": 2.090205430984497,
      "learning_rate": 0.0001314967805010684,
      "loss": 1.7351,
      "step": 11618
    },
    {
      "epoch": 0.4161011334539008,
      "grad_norm": 1.9120638370513916,
      "learning_rate": 0.0001314857716387166,
      "loss": 1.6389,
      "step": 11619
    },
    {
      "epoch": 0.41613694558346914,
      "grad_norm": 1.6917821168899536,
      "learning_rate": 0.00013147476235276438,
      "loss": 1.5383,
      "step": 11620
    },
    {
      "epoch": 0.4161727577130374,
      "grad_norm": 1.3282452821731567,
      "learning_rate": 0.00013146375264335978,
      "loss": 1.4817,
      "step": 11621
    },
    {
      "epoch": 0.4162085698426057,
      "grad_norm": 1.9482810497283936,
      "learning_rate": 0.00013145274251065103,
      "loss": 1.5597,
      "step": 11622
    },
    {
      "epoch": 0.41624438197217395,
      "grad_norm": 1.5540615320205688,
      "learning_rate": 0.00013144173195478616,
      "loss": 1.5281,
      "step": 11623
    },
    {
      "epoch": 0.4162801941017423,
      "grad_norm": 1.9348515272140503,
      "learning_rate": 0.0001314307209759134,
      "loss": 1.4764,
      "step": 11624
    },
    {
      "epoch": 0.41631600623131054,
      "grad_norm": 2.240257978439331,
      "learning_rate": 0.00013141970957418074,
      "loss": 1.5697,
      "step": 11625
    },
    {
      "epoch": 0.4163518183608788,
      "grad_norm": 1.4661803245544434,
      "learning_rate": 0.0001314086977497365,
      "loss": 1.2688,
      "step": 11626
    },
    {
      "epoch": 0.41638763049044714,
      "grad_norm": 1.7400275468826294,
      "learning_rate": 0.0001313976855027287,
      "loss": 1.4783,
      "step": 11627
    },
    {
      "epoch": 0.4164234426200154,
      "grad_norm": 1.930822730064392,
      "learning_rate": 0.00013138667283330556,
      "loss": 1.6808,
      "step": 11628
    },
    {
      "epoch": 0.4164592547495837,
      "grad_norm": 1.6102182865142822,
      "learning_rate": 0.00013137565974161524,
      "loss": 1.7887,
      "step": 11629
    },
    {
      "epoch": 0.41649506687915194,
      "grad_norm": 1.7597026824951172,
      "learning_rate": 0.00013136464622780583,
      "loss": 1.6061,
      "step": 11630
    },
    {
      "epoch": 0.41653087900872027,
      "grad_norm": 2.7631938457489014,
      "learning_rate": 0.00013135363229202564,
      "loss": 1.6641,
      "step": 11631
    },
    {
      "epoch": 0.41656669113828854,
      "grad_norm": 1.9850269556045532,
      "learning_rate": 0.0001313426179344227,
      "loss": 1.2111,
      "step": 11632
    },
    {
      "epoch": 0.4166025032678568,
      "grad_norm": 1.4527435302734375,
      "learning_rate": 0.0001313316031551453,
      "loss": 1.5459,
      "step": 11633
    },
    {
      "epoch": 0.41663831539742513,
      "grad_norm": 1.589457631111145,
      "learning_rate": 0.00013132058795434158,
      "loss": 1.6198,
      "step": 11634
    },
    {
      "epoch": 0.4166741275269934,
      "grad_norm": 1.4151225090026855,
      "learning_rate": 0.0001313095723321598,
      "loss": 1.4616,
      "step": 11635
    },
    {
      "epoch": 0.41670993965656167,
      "grad_norm": 1.7117297649383545,
      "learning_rate": 0.00013129855628874805,
      "loss": 1.6629,
      "step": 11636
    },
    {
      "epoch": 0.41674575178612994,
      "grad_norm": 2.363799571990967,
      "learning_rate": 0.0001312875398242546,
      "loss": 1.6905,
      "step": 11637
    },
    {
      "epoch": 0.41678156391569826,
      "grad_norm": 1.919259786605835,
      "learning_rate": 0.0001312765229388277,
      "loss": 1.4872,
      "step": 11638
    },
    {
      "epoch": 0.41681737604526653,
      "grad_norm": 2.0056421756744385,
      "learning_rate": 0.00013126550563261551,
      "loss": 1.5765,
      "step": 11639
    },
    {
      "epoch": 0.4168531881748348,
      "grad_norm": 1.4317076206207275,
      "learning_rate": 0.00013125448790576627,
      "loss": 1.3747,
      "step": 11640
    },
    {
      "epoch": 0.4168890003044031,
      "grad_norm": 1.9106783866882324,
      "learning_rate": 0.00013124346975842822,
      "loss": 1.3107,
      "step": 11641
    },
    {
      "epoch": 0.4169248124339714,
      "grad_norm": 3.9579901695251465,
      "learning_rate": 0.00013123245119074956,
      "loss": 1.5716,
      "step": 11642
    },
    {
      "epoch": 0.41696062456353966,
      "grad_norm": 2.3893094062805176,
      "learning_rate": 0.00013122143220287854,
      "loss": 1.5343,
      "step": 11643
    },
    {
      "epoch": 0.41699643669310793,
      "grad_norm": 1.485538363456726,
      "learning_rate": 0.00013121041279496348,
      "loss": 1.5387,
      "step": 11644
    },
    {
      "epoch": 0.41703224882267625,
      "grad_norm": 1.8348642587661743,
      "learning_rate": 0.00013119939296715253,
      "loss": 1.653,
      "step": 11645
    },
    {
      "epoch": 0.4170680609522445,
      "grad_norm": 1.5645591020584106,
      "learning_rate": 0.00013118837271959403,
      "loss": 1.6355,
      "step": 11646
    },
    {
      "epoch": 0.4171038730818128,
      "grad_norm": 1.9574569463729858,
      "learning_rate": 0.0001311773520524362,
      "loss": 1.6933,
      "step": 11647
    },
    {
      "epoch": 0.4171396852113811,
      "grad_norm": 2.693382740020752,
      "learning_rate": 0.00013116633096582728,
      "loss": 1.2307,
      "step": 11648
    },
    {
      "epoch": 0.4171754973409494,
      "grad_norm": 2.2191860675811768,
      "learning_rate": 0.0001311553094599156,
      "loss": 1.5716,
      "step": 11649
    },
    {
      "epoch": 0.41721130947051766,
      "grad_norm": 1.8328752517700195,
      "learning_rate": 0.00013114428753484942,
      "loss": 1.4303,
      "step": 11650
    },
    {
      "epoch": 0.4172471216000859,
      "grad_norm": 1.64580500125885,
      "learning_rate": 0.00013113326519077702,
      "loss": 1.8983,
      "step": 11651
    },
    {
      "epoch": 0.41728293372965425,
      "grad_norm": 1.5777573585510254,
      "learning_rate": 0.0001311222424278467,
      "loss": 1.3932,
      "step": 11652
    },
    {
      "epoch": 0.4173187458592225,
      "grad_norm": 1.5518149137496948,
      "learning_rate": 0.00013111121924620672,
      "loss": 1.3117,
      "step": 11653
    },
    {
      "epoch": 0.4173545579887908,
      "grad_norm": 1.7711327075958252,
      "learning_rate": 0.00013110019564600546,
      "loss": 1.4527,
      "step": 11654
    },
    {
      "epoch": 0.4173903701183591,
      "grad_norm": 1.4793881177902222,
      "learning_rate": 0.00013108917162739115,
      "loss": 1.7026,
      "step": 11655
    },
    {
      "epoch": 0.4174261822479274,
      "grad_norm": 1.7926698923110962,
      "learning_rate": 0.00013107814719051216,
      "loss": 1.4584,
      "step": 11656
    },
    {
      "epoch": 0.41746199437749565,
      "grad_norm": 1.60378897190094,
      "learning_rate": 0.0001310671223355168,
      "loss": 1.6985,
      "step": 11657
    },
    {
      "epoch": 0.4174978065070639,
      "grad_norm": 1.862740397453308,
      "learning_rate": 0.00013105609706255336,
      "loss": 1.6307,
      "step": 11658
    },
    {
      "epoch": 0.41753361863663224,
      "grad_norm": 1.755902647972107,
      "learning_rate": 0.00013104507137177022,
      "loss": 1.6147,
      "step": 11659
    },
    {
      "epoch": 0.4175694307662005,
      "grad_norm": 1.6584196090698242,
      "learning_rate": 0.00013103404526331564,
      "loss": 1.4459,
      "step": 11660
    },
    {
      "epoch": 0.4176052428957688,
      "grad_norm": 1.5859105587005615,
      "learning_rate": 0.00013102301873733807,
      "loss": 1.5394,
      "step": 11661
    },
    {
      "epoch": 0.4176410550253371,
      "grad_norm": 1.9892284870147705,
      "learning_rate": 0.00013101199179398572,
      "loss": 1.3006,
      "step": 11662
    },
    {
      "epoch": 0.4176768671549054,
      "grad_norm": 1.620436191558838,
      "learning_rate": 0.0001310009644334071,
      "loss": 1.4348,
      "step": 11663
    },
    {
      "epoch": 0.41771267928447364,
      "grad_norm": 1.9087800979614258,
      "learning_rate": 0.00013098993665575047,
      "loss": 1.725,
      "step": 11664
    },
    {
      "epoch": 0.4177484914140419,
      "grad_norm": 1.4620959758758545,
      "learning_rate": 0.0001309789084611642,
      "loss": 1.4238,
      "step": 11665
    },
    {
      "epoch": 0.41778430354361024,
      "grad_norm": 1.6417380571365356,
      "learning_rate": 0.00013096787984979673,
      "loss": 1.9673,
      "step": 11666
    },
    {
      "epoch": 0.4178201156731785,
      "grad_norm": 1.8227124214172363,
      "learning_rate": 0.00013095685082179632,
      "loss": 1.5281,
      "step": 11667
    },
    {
      "epoch": 0.4178559278027468,
      "grad_norm": 1.9912152290344238,
      "learning_rate": 0.00013094582137731145,
      "loss": 1.3199,
      "step": 11668
    },
    {
      "epoch": 0.4178917399323151,
      "grad_norm": 1.9016451835632324,
      "learning_rate": 0.00013093479151649043,
      "loss": 1.5948,
      "step": 11669
    },
    {
      "epoch": 0.41792755206188337,
      "grad_norm": 1.4120560884475708,
      "learning_rate": 0.00013092376123948174,
      "loss": 1.3703,
      "step": 11670
    },
    {
      "epoch": 0.41796336419145164,
      "grad_norm": 1.8982212543487549,
      "learning_rate": 0.0001309127305464337,
      "loss": 1.149,
      "step": 11671
    },
    {
      "epoch": 0.4179991763210199,
      "grad_norm": 1.4889423847198486,
      "learning_rate": 0.00013090169943749476,
      "loss": 1.5313,
      "step": 11672
    },
    {
      "epoch": 0.41803498845058823,
      "grad_norm": 1.54822838306427,
      "learning_rate": 0.00013089066791281332,
      "loss": 1.4798,
      "step": 11673
    },
    {
      "epoch": 0.4180708005801565,
      "grad_norm": 2.2774507999420166,
      "learning_rate": 0.00013087963597253777,
      "loss": 1.3878,
      "step": 11674
    },
    {
      "epoch": 0.41810661270972477,
      "grad_norm": 1.52279794216156,
      "learning_rate": 0.00013086860361681657,
      "loss": 1.3788,
      "step": 11675
    },
    {
      "epoch": 0.4181424248392931,
      "grad_norm": 1.921139121055603,
      "learning_rate": 0.00013085757084579808,
      "loss": 1.6913,
      "step": 11676
    },
    {
      "epoch": 0.41817823696886136,
      "grad_norm": 1.6303210258483887,
      "learning_rate": 0.00013084653765963085,
      "loss": 1.636,
      "step": 11677
    },
    {
      "epoch": 0.41821404909842963,
      "grad_norm": 1.3981053829193115,
      "learning_rate": 0.0001308355040584632,
      "loss": 1.4341,
      "step": 11678
    },
    {
      "epoch": 0.4182498612279979,
      "grad_norm": 1.5886470079421997,
      "learning_rate": 0.0001308244700424436,
      "loss": 1.5647,
      "step": 11679
    },
    {
      "epoch": 0.4182856733575662,
      "grad_norm": 1.4517158269882202,
      "learning_rate": 0.00013081343561172055,
      "loss": 1.4405,
      "step": 11680
    },
    {
      "epoch": 0.4183214854871345,
      "grad_norm": 1.6235915422439575,
      "learning_rate": 0.00013080240076644245,
      "loss": 1.6848,
      "step": 11681
    },
    {
      "epoch": 0.41835729761670276,
      "grad_norm": 1.8105449676513672,
      "learning_rate": 0.0001307913655067578,
      "loss": 1.4255,
      "step": 11682
    },
    {
      "epoch": 0.4183931097462711,
      "grad_norm": 1.8867751359939575,
      "learning_rate": 0.00013078032983281505,
      "loss": 1.8674,
      "step": 11683
    },
    {
      "epoch": 0.41842892187583935,
      "grad_norm": 1.5423702001571655,
      "learning_rate": 0.00013076929374476265,
      "loss": 1.6115,
      "step": 11684
    },
    {
      "epoch": 0.4184647340054076,
      "grad_norm": 1.234898567199707,
      "learning_rate": 0.00013075825724274907,
      "loss": 1.3527,
      "step": 11685
    },
    {
      "epoch": 0.4185005461349759,
      "grad_norm": 1.7368626594543457,
      "learning_rate": 0.00013074722032692287,
      "loss": 1.4389,
      "step": 11686
    },
    {
      "epoch": 0.4185363582645442,
      "grad_norm": 2.3689353466033936,
      "learning_rate": 0.00013073618299743242,
      "loss": 1.4973,
      "step": 11687
    },
    {
      "epoch": 0.4185721703941125,
      "grad_norm": 1.5590380430221558,
      "learning_rate": 0.00013072514525442632,
      "loss": 1.482,
      "step": 11688
    },
    {
      "epoch": 0.41860798252368076,
      "grad_norm": 1.8684630393981934,
      "learning_rate": 0.000130714107098053,
      "loss": 1.4597,
      "step": 11689
    },
    {
      "epoch": 0.4186437946532491,
      "grad_norm": 1.368911623954773,
      "learning_rate": 0.000130703068528461,
      "loss": 1.4001,
      "step": 11690
    },
    {
      "epoch": 0.41867960678281735,
      "grad_norm": 2.009352922439575,
      "learning_rate": 0.00013069202954579882,
      "loss": 1.6073,
      "step": 11691
    },
    {
      "epoch": 0.4187154189123856,
      "grad_norm": 1.8053749799728394,
      "learning_rate": 0.00013068099015021498,
      "loss": 1.2745,
      "step": 11692
    },
    {
      "epoch": 0.4187512310419539,
      "grad_norm": 2.223372459411621,
      "learning_rate": 0.00013066995034185798,
      "loss": 1.7715,
      "step": 11693
    },
    {
      "epoch": 0.4187870431715222,
      "grad_norm": 1.833401083946228,
      "learning_rate": 0.00013065891012087634,
      "loss": 1.7072,
      "step": 11694
    },
    {
      "epoch": 0.4188228553010905,
      "grad_norm": 1.3824347257614136,
      "learning_rate": 0.00013064786948741863,
      "loss": 1.4636,
      "step": 11695
    },
    {
      "epoch": 0.41885866743065875,
      "grad_norm": 1.4842058420181274,
      "learning_rate": 0.00013063682844163338,
      "loss": 1.2638,
      "step": 11696
    },
    {
      "epoch": 0.4188944795602271,
      "grad_norm": 1.8669413328170776,
      "learning_rate": 0.00013062578698366909,
      "loss": 1.5242,
      "step": 11697
    },
    {
      "epoch": 0.41893029168979534,
      "grad_norm": 1.6303125619888306,
      "learning_rate": 0.00013061474511367435,
      "loss": 1.6688,
      "step": 11698
    },
    {
      "epoch": 0.4189661038193636,
      "grad_norm": 1.7029786109924316,
      "learning_rate": 0.00013060370283179772,
      "loss": 1.7938,
      "step": 11699
    },
    {
      "epoch": 0.4190019159489319,
      "grad_norm": 2.1118316650390625,
      "learning_rate": 0.00013059266013818775,
      "loss": 1.5609,
      "step": 11700
    },
    {
      "epoch": 0.4190377280785002,
      "grad_norm": 1.7894207239151,
      "learning_rate": 0.000130581617032993,
      "loss": 1.6328,
      "step": 11701
    },
    {
      "epoch": 0.4190735402080685,
      "grad_norm": 1.6137210130691528,
      "learning_rate": 0.00013057057351636205,
      "loss": 1.1927,
      "step": 11702
    },
    {
      "epoch": 0.41910935233763674,
      "grad_norm": 1.628922700881958,
      "learning_rate": 0.00013055952958844345,
      "loss": 1.7027,
      "step": 11703
    },
    {
      "epoch": 0.41914516446720507,
      "grad_norm": 1.7288146018981934,
      "learning_rate": 0.00013054848524938576,
      "loss": 1.5557,
      "step": 11704
    },
    {
      "epoch": 0.41918097659677334,
      "grad_norm": 1.3177454471588135,
      "learning_rate": 0.00013053744049933765,
      "loss": 1.5625,
      "step": 11705
    },
    {
      "epoch": 0.4192167887263416,
      "grad_norm": 1.693861722946167,
      "learning_rate": 0.00013052639533844766,
      "loss": 1.3449,
      "step": 11706
    },
    {
      "epoch": 0.4192526008559099,
      "grad_norm": 1.604946494102478,
      "learning_rate": 0.0001305153497668644,
      "loss": 1.5387,
      "step": 11707
    },
    {
      "epoch": 0.4192884129854782,
      "grad_norm": 1.9413237571716309,
      "learning_rate": 0.00013050430378473647,
      "loss": 1.275,
      "step": 11708
    },
    {
      "epoch": 0.41932422511504647,
      "grad_norm": 1.5269191265106201,
      "learning_rate": 0.00013049325739221247,
      "loss": 1.6002,
      "step": 11709
    },
    {
      "epoch": 0.41936003724461474,
      "grad_norm": 1.6607357263565063,
      "learning_rate": 0.00013048221058944103,
      "loss": 1.5763,
      "step": 11710
    },
    {
      "epoch": 0.41939584937418306,
      "grad_norm": 1.8496578931808472,
      "learning_rate": 0.00013047116337657077,
      "loss": 1.734,
      "step": 11711
    },
    {
      "epoch": 0.41943166150375133,
      "grad_norm": 1.3404603004455566,
      "learning_rate": 0.00013046011575375033,
      "loss": 1.3846,
      "step": 11712
    },
    {
      "epoch": 0.4194674736333196,
      "grad_norm": 1.9869346618652344,
      "learning_rate": 0.00013044906772112828,
      "loss": 1.4686,
      "step": 11713
    },
    {
      "epoch": 0.41950328576288787,
      "grad_norm": 1.6581941843032837,
      "learning_rate": 0.00013043801927885334,
      "loss": 1.5109,
      "step": 11714
    },
    {
      "epoch": 0.4195390978924562,
      "grad_norm": 2.1327946186065674,
      "learning_rate": 0.00013042697042707407,
      "loss": 1.2431,
      "step": 11715
    },
    {
      "epoch": 0.41957491002202446,
      "grad_norm": 1.4978673458099365,
      "learning_rate": 0.0001304159211659392,
      "loss": 1.5151,
      "step": 11716
    },
    {
      "epoch": 0.41961072215159273,
      "grad_norm": 1.6439182758331299,
      "learning_rate": 0.00013040487149559735,
      "loss": 1.4472,
      "step": 11717
    },
    {
      "epoch": 0.41964653428116105,
      "grad_norm": 1.6311148405075073,
      "learning_rate": 0.00013039382141619713,
      "loss": 1.6517,
      "step": 11718
    },
    {
      "epoch": 0.4196823464107293,
      "grad_norm": 1.6701090335845947,
      "learning_rate": 0.0001303827709278873,
      "loss": 1.8037,
      "step": 11719
    },
    {
      "epoch": 0.4197181585402976,
      "grad_norm": 1.9653955698013306,
      "learning_rate": 0.0001303717200308164,
      "loss": 1.3443,
      "step": 11720
    },
    {
      "epoch": 0.41975397066986586,
      "grad_norm": 1.8693047761917114,
      "learning_rate": 0.00013036066872513326,
      "loss": 1.6978,
      "step": 11721
    },
    {
      "epoch": 0.4197897827994342,
      "grad_norm": 1.5519013404846191,
      "learning_rate": 0.00013034961701098645,
      "loss": 1.355,
      "step": 11722
    },
    {
      "epoch": 0.41982559492900245,
      "grad_norm": 2.0810554027557373,
      "learning_rate": 0.0001303385648885247,
      "loss": 1.8034,
      "step": 11723
    },
    {
      "epoch": 0.4198614070585707,
      "grad_norm": 2.0059664249420166,
      "learning_rate": 0.00013032751235789668,
      "loss": 1.53,
      "step": 11724
    },
    {
      "epoch": 0.41989721918813905,
      "grad_norm": 1.6518661975860596,
      "learning_rate": 0.00013031645941925115,
      "loss": 1.591,
      "step": 11725
    },
    {
      "epoch": 0.4199330313177073,
      "grad_norm": 1.6742372512817383,
      "learning_rate": 0.00013030540607273674,
      "loss": 1.6334,
      "step": 11726
    },
    {
      "epoch": 0.4199688434472756,
      "grad_norm": 1.625395655632019,
      "learning_rate": 0.00013029435231850215,
      "loss": 1.4972,
      "step": 11727
    },
    {
      "epoch": 0.42000465557684385,
      "grad_norm": 1.3360016345977783,
      "learning_rate": 0.00013028329815669616,
      "loss": 1.2917,
      "step": 11728
    },
    {
      "epoch": 0.4200404677064122,
      "grad_norm": 1.6598440408706665,
      "learning_rate": 0.00013027224358746743,
      "loss": 1.4068,
      "step": 11729
    },
    {
      "epoch": 0.42007627983598045,
      "grad_norm": 1.5892927646636963,
      "learning_rate": 0.00013026118861096472,
      "loss": 1.239,
      "step": 11730
    },
    {
      "epoch": 0.4201120919655487,
      "grad_norm": 1.4301259517669678,
      "learning_rate": 0.00013025013322733674,
      "loss": 1.4511,
      "step": 11731
    },
    {
      "epoch": 0.42014790409511704,
      "grad_norm": 1.4316529035568237,
      "learning_rate": 0.00013023907743673228,
      "loss": 1.5084,
      "step": 11732
    },
    {
      "epoch": 0.4201837162246853,
      "grad_norm": 1.458953857421875,
      "learning_rate": 0.0001302280212393,
      "loss": 1.3913,
      "step": 11733
    },
    {
      "epoch": 0.4202195283542536,
      "grad_norm": 2.121302366256714,
      "learning_rate": 0.0001302169646351887,
      "loss": 1.6365,
      "step": 11734
    },
    {
      "epoch": 0.42025534048382185,
      "grad_norm": 1.6328710317611694,
      "learning_rate": 0.00013020590762454713,
      "loss": 1.5309,
      "step": 11735
    },
    {
      "epoch": 0.4202911526133902,
      "grad_norm": 1.7382535934448242,
      "learning_rate": 0.00013019485020752402,
      "loss": 1.28,
      "step": 11736
    },
    {
      "epoch": 0.42032696474295844,
      "grad_norm": 1.6773878335952759,
      "learning_rate": 0.00013018379238426814,
      "loss": 1.7261,
      "step": 11737
    },
    {
      "epoch": 0.4203627768725267,
      "grad_norm": 2.123692750930786,
      "learning_rate": 0.0001301727341549283,
      "loss": 1.6939,
      "step": 11738
    },
    {
      "epoch": 0.42039858900209504,
      "grad_norm": 1.591450810432434,
      "learning_rate": 0.0001301616755196532,
      "loss": 1.6366,
      "step": 11739
    },
    {
      "epoch": 0.4204344011316633,
      "grad_norm": 1.7275581359863281,
      "learning_rate": 0.0001301506164785917,
      "loss": 1.6026,
      "step": 11740
    },
    {
      "epoch": 0.4204702132612316,
      "grad_norm": 1.6297035217285156,
      "learning_rate": 0.00013013955703189252,
      "loss": 1.683,
      "step": 11741
    },
    {
      "epoch": 0.42050602539079984,
      "grad_norm": 1.9007762670516968,
      "learning_rate": 0.00013012849717970447,
      "loss": 1.3047,
      "step": 11742
    },
    {
      "epoch": 0.42054183752036817,
      "grad_norm": 1.8373106718063354,
      "learning_rate": 0.00013011743692217638,
      "loss": 1.6415,
      "step": 11743
    },
    {
      "epoch": 0.42057764964993644,
      "grad_norm": 1.351611852645874,
      "learning_rate": 0.000130106376259457,
      "loss": 1.3333,
      "step": 11744
    },
    {
      "epoch": 0.4206134617795047,
      "grad_norm": 1.4901500940322876,
      "learning_rate": 0.0001300953151916952,
      "loss": 1.6745,
      "step": 11745
    },
    {
      "epoch": 0.42064927390907303,
      "grad_norm": 1.6721500158309937,
      "learning_rate": 0.0001300842537190397,
      "loss": 1.4681,
      "step": 11746
    },
    {
      "epoch": 0.4206850860386413,
      "grad_norm": 2.408515453338623,
      "learning_rate": 0.00013007319184163944,
      "loss": 1.4235,
      "step": 11747
    },
    {
      "epoch": 0.42072089816820957,
      "grad_norm": 2.0116817951202393,
      "learning_rate": 0.00013006212955964311,
      "loss": 1.586,
      "step": 11748
    },
    {
      "epoch": 0.42075671029777784,
      "grad_norm": 1.4370110034942627,
      "learning_rate": 0.00013005106687319966,
      "loss": 1.8267,
      "step": 11749
    },
    {
      "epoch": 0.42079252242734616,
      "grad_norm": 1.985016942024231,
      "learning_rate": 0.00013004000378245782,
      "loss": 1.8785,
      "step": 11750
    },
    {
      "epoch": 0.42082833455691443,
      "grad_norm": 1.489750623703003,
      "learning_rate": 0.00013002894028756653,
      "loss": 1.3115,
      "step": 11751
    },
    {
      "epoch": 0.4208641466864827,
      "grad_norm": 1.9699902534484863,
      "learning_rate": 0.0001300178763886745,
      "loss": 1.4464,
      "step": 11752
    },
    {
      "epoch": 0.420899958816051,
      "grad_norm": 1.4457578659057617,
      "learning_rate": 0.00013000681208593073,
      "loss": 1.6236,
      "step": 11753
    },
    {
      "epoch": 0.4209357709456193,
      "grad_norm": 1.5516526699066162,
      "learning_rate": 0.000129995747379484,
      "loss": 1.5156,
      "step": 11754
    },
    {
      "epoch": 0.42097158307518756,
      "grad_norm": 1.7440998554229736,
      "learning_rate": 0.00012998468226948316,
      "loss": 1.7088,
      "step": 11755
    },
    {
      "epoch": 0.42100739520475583,
      "grad_norm": 1.6341153383255005,
      "learning_rate": 0.00012997361675607714,
      "loss": 1.3348,
      "step": 11756
    },
    {
      "epoch": 0.42104320733432415,
      "grad_norm": 2.47872257232666,
      "learning_rate": 0.0001299625508394147,
      "loss": 1.6474,
      "step": 11757
    },
    {
      "epoch": 0.4210790194638924,
      "grad_norm": 1.4526902437210083,
      "learning_rate": 0.00012995148451964487,
      "loss": 1.5869,
      "step": 11758
    },
    {
      "epoch": 0.4211148315934607,
      "grad_norm": 1.5664465427398682,
      "learning_rate": 0.00012994041779691639,
      "loss": 1.6571,
      "step": 11759
    },
    {
      "epoch": 0.421150643723029,
      "grad_norm": 1.4599276781082153,
      "learning_rate": 0.00012992935067137823,
      "loss": 1.357,
      "step": 11760
    },
    {
      "epoch": 0.4211864558525973,
      "grad_norm": 1.712045669555664,
      "learning_rate": 0.00012991828314317923,
      "loss": 1.531,
      "step": 11761
    },
    {
      "epoch": 0.42122226798216555,
      "grad_norm": 2.1027767658233643,
      "learning_rate": 0.00012990721521246839,
      "loss": 1.6629,
      "step": 11762
    },
    {
      "epoch": 0.4212580801117338,
      "grad_norm": 1.7195019721984863,
      "learning_rate": 0.00012989614687939453,
      "loss": 1.5359,
      "step": 11763
    },
    {
      "epoch": 0.42129389224130215,
      "grad_norm": 1.4466419219970703,
      "learning_rate": 0.00012988507814410652,
      "loss": 1.6942,
      "step": 11764
    },
    {
      "epoch": 0.4213297043708704,
      "grad_norm": 1.2719695568084717,
      "learning_rate": 0.0001298740090067534,
      "loss": 1.5447,
      "step": 11765
    },
    {
      "epoch": 0.4213655165004387,
      "grad_norm": 2.5164668560028076,
      "learning_rate": 0.000129862939467484,
      "loss": 1.5569,
      "step": 11766
    },
    {
      "epoch": 0.421401328630007,
      "grad_norm": 1.691938877105713,
      "learning_rate": 0.00012985186952644724,
      "loss": 1.4605,
      "step": 11767
    },
    {
      "epoch": 0.4214371407595753,
      "grad_norm": 1.8045603036880493,
      "learning_rate": 0.0001298407991837921,
      "loss": 1.3567,
      "step": 11768
    },
    {
      "epoch": 0.42147295288914355,
      "grad_norm": 1.5395952463150024,
      "learning_rate": 0.0001298297284396675,
      "loss": 1.6325,
      "step": 11769
    },
    {
      "epoch": 0.4215087650187118,
      "grad_norm": 1.8432800769805908,
      "learning_rate": 0.0001298186572942224,
      "loss": 1.4202,
      "step": 11770
    },
    {
      "epoch": 0.42154457714828014,
      "grad_norm": 2.240885019302368,
      "learning_rate": 0.00012980758574760573,
      "loss": 1.9024,
      "step": 11771
    },
    {
      "epoch": 0.4215803892778484,
      "grad_norm": 2.1993038654327393,
      "learning_rate": 0.00012979651379996642,
      "loss": 1.5855,
      "step": 11772
    },
    {
      "epoch": 0.4216162014074167,
      "grad_norm": 1.9593286514282227,
      "learning_rate": 0.00012978544145145343,
      "loss": 1.5088,
      "step": 11773
    },
    {
      "epoch": 0.421652013536985,
      "grad_norm": 1.6190402507781982,
      "learning_rate": 0.0001297743687022158,
      "loss": 1.6438,
      "step": 11774
    },
    {
      "epoch": 0.4216878256665533,
      "grad_norm": 1.6543118953704834,
      "learning_rate": 0.0001297632955524024,
      "loss": 1.2093,
      "step": 11775
    },
    {
      "epoch": 0.42172363779612154,
      "grad_norm": 1.6316370964050293,
      "learning_rate": 0.00012975222200216227,
      "loss": 1.3618,
      "step": 11776
    },
    {
      "epoch": 0.4217594499256898,
      "grad_norm": 1.841871738433838,
      "learning_rate": 0.00012974114805164438,
      "loss": 1.6844,
      "step": 11777
    },
    {
      "epoch": 0.42179526205525814,
      "grad_norm": 1.207383632659912,
      "learning_rate": 0.0001297300737009977,
      "loss": 1.4037,
      "step": 11778
    },
    {
      "epoch": 0.4218310741848264,
      "grad_norm": 1.9168089628219604,
      "learning_rate": 0.00012971899895037123,
      "loss": 1.3065,
      "step": 11779
    },
    {
      "epoch": 0.4218668863143947,
      "grad_norm": 1.7665578126907349,
      "learning_rate": 0.00012970792379991396,
      "loss": 1.4799,
      "step": 11780
    },
    {
      "epoch": 0.421902698443963,
      "grad_norm": 1.5356736183166504,
      "learning_rate": 0.00012969684824977492,
      "loss": 1.5496,
      "step": 11781
    },
    {
      "epoch": 0.42193851057353127,
      "grad_norm": 1.6036309003829956,
      "learning_rate": 0.00012968577230010304,
      "loss": 1.6618,
      "step": 11782
    },
    {
      "epoch": 0.42197432270309954,
      "grad_norm": 1.2157410383224487,
      "learning_rate": 0.0001296746959510474,
      "loss": 1.4071,
      "step": 11783
    },
    {
      "epoch": 0.4220101348326678,
      "grad_norm": 1.6793562173843384,
      "learning_rate": 0.00012966361920275702,
      "loss": 1.2486,
      "step": 11784
    },
    {
      "epoch": 0.42204594696223613,
      "grad_norm": 1.6444344520568848,
      "learning_rate": 0.0001296525420553809,
      "loss": 1.5359,
      "step": 11785
    },
    {
      "epoch": 0.4220817590918044,
      "grad_norm": 2.4854736328125,
      "learning_rate": 0.00012964146450906807,
      "loss": 1.4089,
      "step": 11786
    },
    {
      "epoch": 0.42211757122137267,
      "grad_norm": 2.2101008892059326,
      "learning_rate": 0.0001296303865639676,
      "loss": 1.6471,
      "step": 11787
    },
    {
      "epoch": 0.42215338335094094,
      "grad_norm": 1.3930305242538452,
      "learning_rate": 0.00012961930822022848,
      "loss": 1.5859,
      "step": 11788
    },
    {
      "epoch": 0.42218919548050926,
      "grad_norm": 1.6884602308273315,
      "learning_rate": 0.00012960822947799978,
      "loss": 1.4459,
      "step": 11789
    },
    {
      "epoch": 0.42222500761007753,
      "grad_norm": 1.2813154458999634,
      "learning_rate": 0.0001295971503374305,
      "loss": 1.5418,
      "step": 11790
    },
    {
      "epoch": 0.4222608197396458,
      "grad_norm": 1.8060952425003052,
      "learning_rate": 0.0001295860707986698,
      "loss": 1.5921,
      "step": 11791
    },
    {
      "epoch": 0.4222966318692141,
      "grad_norm": 2.59269642829895,
      "learning_rate": 0.00012957499086186665,
      "loss": 1.3835,
      "step": 11792
    },
    {
      "epoch": 0.4223324439987824,
      "grad_norm": 1.729286789894104,
      "learning_rate": 0.00012956391052717017,
      "loss": 1.4255,
      "step": 11793
    },
    {
      "epoch": 0.42236825612835066,
      "grad_norm": 1.4950153827667236,
      "learning_rate": 0.00012955282979472938,
      "loss": 1.1579,
      "step": 11794
    },
    {
      "epoch": 0.42240406825791893,
      "grad_norm": 1.7770169973373413,
      "learning_rate": 0.00012954174866469336,
      "loss": 1.7252,
      "step": 11795
    },
    {
      "epoch": 0.42243988038748725,
      "grad_norm": 1.7633696794509888,
      "learning_rate": 0.0001295306671372113,
      "loss": 1.5152,
      "step": 11796
    },
    {
      "epoch": 0.4224756925170555,
      "grad_norm": 2.1306397914886475,
      "learning_rate": 0.00012951958521243215,
      "loss": 1.6468,
      "step": 11797
    },
    {
      "epoch": 0.4225115046466238,
      "grad_norm": 1.2364559173583984,
      "learning_rate": 0.00012950850289050508,
      "loss": 1.5101,
      "step": 11798
    },
    {
      "epoch": 0.4225473167761921,
      "grad_norm": 1.563754677772522,
      "learning_rate": 0.00012949742017157915,
      "loss": 1.3717,
      "step": 11799
    },
    {
      "epoch": 0.4225831289057604,
      "grad_norm": 1.5410443544387817,
      "learning_rate": 0.0001294863370558035,
      "loss": 1.3353,
      "step": 11800
    },
    {
      "epoch": 0.42261894103532865,
      "grad_norm": 1.564864993095398,
      "learning_rate": 0.0001294752535433272,
      "loss": 1.3936,
      "step": 11801
    },
    {
      "epoch": 0.4226547531648969,
      "grad_norm": 1.5698978900909424,
      "learning_rate": 0.0001294641696342994,
      "loss": 1.5412,
      "step": 11802
    },
    {
      "epoch": 0.42269056529446525,
      "grad_norm": 2.2333059310913086,
      "learning_rate": 0.00012945308532886918,
      "loss": 1.3542,
      "step": 11803
    },
    {
      "epoch": 0.4227263774240335,
      "grad_norm": 1.559870719909668,
      "learning_rate": 0.0001294420006271857,
      "loss": 1.3988,
      "step": 11804
    },
    {
      "epoch": 0.4227621895536018,
      "grad_norm": 1.3145016431808472,
      "learning_rate": 0.00012943091552939807,
      "loss": 1.2073,
      "step": 11805
    },
    {
      "epoch": 0.4227980016831701,
      "grad_norm": 1.6928826570510864,
      "learning_rate": 0.00012941983003565544,
      "loss": 1.376,
      "step": 11806
    },
    {
      "epoch": 0.4228338138127384,
      "grad_norm": 1.3938753604888916,
      "learning_rate": 0.00012940874414610698,
      "loss": 1.5549,
      "step": 11807
    },
    {
      "epoch": 0.42286962594230665,
      "grad_norm": 1.5950511693954468,
      "learning_rate": 0.00012939765786090178,
      "loss": 1.3254,
      "step": 11808
    },
    {
      "epoch": 0.4229054380718749,
      "grad_norm": 1.5573126077651978,
      "learning_rate": 0.000129386571180189,
      "loss": 1.5849,
      "step": 11809
    },
    {
      "epoch": 0.42294125020144324,
      "grad_norm": 2.0934598445892334,
      "learning_rate": 0.00012937548410411778,
      "loss": 1.4638,
      "step": 11810
    },
    {
      "epoch": 0.4229770623310115,
      "grad_norm": 1.5477838516235352,
      "learning_rate": 0.0001293643966328374,
      "loss": 1.4116,
      "step": 11811
    },
    {
      "epoch": 0.4230128744605798,
      "grad_norm": 1.7207345962524414,
      "learning_rate": 0.00012935330876649687,
      "loss": 1.5597,
      "step": 11812
    },
    {
      "epoch": 0.4230486865901481,
      "grad_norm": 1.701752781867981,
      "learning_rate": 0.00012934222050524547,
      "loss": 1.6269,
      "step": 11813
    },
    {
      "epoch": 0.4230844987197164,
      "grad_norm": 1.5868911743164062,
      "learning_rate": 0.00012933113184923232,
      "loss": 1.4728,
      "step": 11814
    },
    {
      "epoch": 0.42312031084928464,
      "grad_norm": 1.414086103439331,
      "learning_rate": 0.00012932004279860663,
      "loss": 1.3683,
      "step": 11815
    },
    {
      "epoch": 0.4231561229788529,
      "grad_norm": 2.1255552768707275,
      "learning_rate": 0.0001293089533535176,
      "loss": 1.8208,
      "step": 11816
    },
    {
      "epoch": 0.42319193510842124,
      "grad_norm": 1.8227784633636475,
      "learning_rate": 0.00012929786351411439,
      "loss": 1.3282,
      "step": 11817
    },
    {
      "epoch": 0.4232277472379895,
      "grad_norm": 1.7852251529693604,
      "learning_rate": 0.00012928677328054623,
      "loss": 1.5736,
      "step": 11818
    },
    {
      "epoch": 0.4232635593675578,
      "grad_norm": 2.8936076164245605,
      "learning_rate": 0.00012927568265296227,
      "loss": 1.5581,
      "step": 11819
    },
    {
      "epoch": 0.4232993714971261,
      "grad_norm": 1.452358365058899,
      "learning_rate": 0.00012926459163151182,
      "loss": 1.6559,
      "step": 11820
    },
    {
      "epoch": 0.42333518362669437,
      "grad_norm": 1.7268511056900024,
      "learning_rate": 0.000129253500216344,
      "loss": 1.604,
      "step": 11821
    },
    {
      "epoch": 0.42337099575626264,
      "grad_norm": 1.3520252704620361,
      "learning_rate": 0.00012924240840760811,
      "loss": 1.4204,
      "step": 11822
    },
    {
      "epoch": 0.4234068078858309,
      "grad_norm": 1.4777060747146606,
      "learning_rate": 0.0001292313162054533,
      "loss": 1.3741,
      "step": 11823
    },
    {
      "epoch": 0.42344262001539923,
      "grad_norm": 1.7785598039627075,
      "learning_rate": 0.00012922022361002886,
      "loss": 1.6714,
      "step": 11824
    },
    {
      "epoch": 0.4234784321449675,
      "grad_norm": 1.92035710811615,
      "learning_rate": 0.00012920913062148398,
      "loss": 1.8324,
      "step": 11825
    },
    {
      "epoch": 0.42351424427453577,
      "grad_norm": 1.8269778490066528,
      "learning_rate": 0.00012919803723996794,
      "loss": 1.7714,
      "step": 11826
    },
    {
      "epoch": 0.4235500564041041,
      "grad_norm": 1.4856898784637451,
      "learning_rate": 0.00012918694346562997,
      "loss": 1.7747,
      "step": 11827
    },
    {
      "epoch": 0.42358586853367236,
      "grad_norm": 1.4505622386932373,
      "learning_rate": 0.0001291758492986193,
      "loss": 1.6209,
      "step": 11828
    },
    {
      "epoch": 0.42362168066324063,
      "grad_norm": 1.4111675024032593,
      "learning_rate": 0.00012916475473908525,
      "loss": 1.2532,
      "step": 11829
    },
    {
      "epoch": 0.4236574927928089,
      "grad_norm": 1.8499622344970703,
      "learning_rate": 0.00012915365978717703,
      "loss": 1.4083,
      "step": 11830
    },
    {
      "epoch": 0.4236933049223772,
      "grad_norm": 1.4217849969863892,
      "learning_rate": 0.0001291425644430439,
      "loss": 1.5601,
      "step": 11831
    },
    {
      "epoch": 0.4237291170519455,
      "grad_norm": 1.7619093656539917,
      "learning_rate": 0.0001291314687068352,
      "loss": 1.3126,
      "step": 11832
    },
    {
      "epoch": 0.42376492918151376,
      "grad_norm": 1.2267673015594482,
      "learning_rate": 0.00012912037257870016,
      "loss": 1.5579,
      "step": 11833
    },
    {
      "epoch": 0.4238007413110821,
      "grad_norm": 1.6975821256637573,
      "learning_rate": 0.00012910927605878803,
      "loss": 1.4415,
      "step": 11834
    },
    {
      "epoch": 0.42383655344065035,
      "grad_norm": 1.650166392326355,
      "learning_rate": 0.0001290981791472482,
      "loss": 1.8445,
      "step": 11835
    },
    {
      "epoch": 0.4238723655702186,
      "grad_norm": 1.423077940940857,
      "learning_rate": 0.00012908708184422983,
      "loss": 1.3797,
      "step": 11836
    },
    {
      "epoch": 0.4239081776997869,
      "grad_norm": 1.5820120573043823,
      "learning_rate": 0.0001290759841498823,
      "loss": 1.6298,
      "step": 11837
    },
    {
      "epoch": 0.4239439898293552,
      "grad_norm": 1.6595290899276733,
      "learning_rate": 0.00012906488606435497,
      "loss": 1.5093,
      "step": 11838
    },
    {
      "epoch": 0.4239798019589235,
      "grad_norm": 2.5189099311828613,
      "learning_rate": 0.00012905378758779702,
      "loss": 1.3822,
      "step": 11839
    },
    {
      "epoch": 0.42401561408849175,
      "grad_norm": 1.7056360244750977,
      "learning_rate": 0.00012904268872035787,
      "loss": 1.253,
      "step": 11840
    },
    {
      "epoch": 0.4240514262180601,
      "grad_norm": 1.8914939165115356,
      "learning_rate": 0.00012903158946218682,
      "loss": 1.4382,
      "step": 11841
    },
    {
      "epoch": 0.42408723834762835,
      "grad_norm": 1.5297006368637085,
      "learning_rate": 0.0001290204898134332,
      "loss": 1.5213,
      "step": 11842
    },
    {
      "epoch": 0.4241230504771966,
      "grad_norm": 1.7533084154129028,
      "learning_rate": 0.00012900938977424627,
      "loss": 1.4682,
      "step": 11843
    },
    {
      "epoch": 0.4241588626067649,
      "grad_norm": 1.830748200416565,
      "learning_rate": 0.00012899828934477545,
      "loss": 1.6938,
      "step": 11844
    },
    {
      "epoch": 0.4241946747363332,
      "grad_norm": 1.4886054992675781,
      "learning_rate": 0.00012898718852517003,
      "loss": 1.4325,
      "step": 11845
    },
    {
      "epoch": 0.4242304868659015,
      "grad_norm": 1.547685146331787,
      "learning_rate": 0.0001289760873155794,
      "loss": 1.6207,
      "step": 11846
    },
    {
      "epoch": 0.42426629899546975,
      "grad_norm": 1.416277289390564,
      "learning_rate": 0.00012896498571615287,
      "loss": 1.5239,
      "step": 11847
    },
    {
      "epoch": 0.4243021111250381,
      "grad_norm": 1.5753027200698853,
      "learning_rate": 0.00012895388372703985,
      "loss": 1.6592,
      "step": 11848
    },
    {
      "epoch": 0.42433792325460634,
      "grad_norm": 1.7635771036148071,
      "learning_rate": 0.00012894278134838963,
      "loss": 1.6179,
      "step": 11849
    },
    {
      "epoch": 0.4243737353841746,
      "grad_norm": 1.644930362701416,
      "learning_rate": 0.00012893167858035168,
      "loss": 1.5943,
      "step": 11850
    },
    {
      "epoch": 0.4244095475137429,
      "grad_norm": 1.333198070526123,
      "learning_rate": 0.00012892057542307527,
      "loss": 1.695,
      "step": 11851
    },
    {
      "epoch": 0.4244453596433112,
      "grad_norm": 2.0017025470733643,
      "learning_rate": 0.00012890947187670982,
      "loss": 1.617,
      "step": 11852
    },
    {
      "epoch": 0.4244811717728795,
      "grad_norm": 1.3186886310577393,
      "learning_rate": 0.00012889836794140473,
      "loss": 1.634,
      "step": 11853
    },
    {
      "epoch": 0.42451698390244774,
      "grad_norm": 1.535506248474121,
      "learning_rate": 0.00012888726361730935,
      "loss": 1.5335,
      "step": 11854
    },
    {
      "epoch": 0.42455279603201607,
      "grad_norm": 1.9215971231460571,
      "learning_rate": 0.00012887615890457314,
      "loss": 1.5718,
      "step": 11855
    },
    {
      "epoch": 0.42458860816158434,
      "grad_norm": 1.4248054027557373,
      "learning_rate": 0.00012886505380334544,
      "loss": 1.4283,
      "step": 11856
    },
    {
      "epoch": 0.4246244202911526,
      "grad_norm": 1.514243483543396,
      "learning_rate": 0.00012885394831377569,
      "loss": 1.6051,
      "step": 11857
    },
    {
      "epoch": 0.4246602324207209,
      "grad_norm": 1.4963092803955078,
      "learning_rate": 0.00012884284243601325,
      "loss": 1.4867,
      "step": 11858
    },
    {
      "epoch": 0.4246960445502892,
      "grad_norm": 1.4643319845199585,
      "learning_rate": 0.0001288317361702076,
      "loss": 1.4199,
      "step": 11859
    },
    {
      "epoch": 0.42473185667985747,
      "grad_norm": 1.4797992706298828,
      "learning_rate": 0.00012882062951650815,
      "loss": 1.4582,
      "step": 11860
    },
    {
      "epoch": 0.42476766880942574,
      "grad_norm": 2.1441917419433594,
      "learning_rate": 0.00012880952247506426,
      "loss": 1.4834,
      "step": 11861
    },
    {
      "epoch": 0.42480348093899406,
      "grad_norm": 1.7394976615905762,
      "learning_rate": 0.00012879841504602544,
      "loss": 1.747,
      "step": 11862
    },
    {
      "epoch": 0.42483929306856233,
      "grad_norm": 2.274009943008423,
      "learning_rate": 0.00012878730722954104,
      "loss": 1.596,
      "step": 11863
    },
    {
      "epoch": 0.4248751051981306,
      "grad_norm": 1.643452525138855,
      "learning_rate": 0.00012877619902576062,
      "loss": 1.6008,
      "step": 11864
    },
    {
      "epoch": 0.42491091732769887,
      "grad_norm": 1.7636380195617676,
      "learning_rate": 0.0001287650904348335,
      "loss": 1.037,
      "step": 11865
    },
    {
      "epoch": 0.4249467294572672,
      "grad_norm": 1.6334855556488037,
      "learning_rate": 0.00012875398145690924,
      "loss": 1.278,
      "step": 11866
    },
    {
      "epoch": 0.42498254158683546,
      "grad_norm": 1.7743029594421387,
      "learning_rate": 0.00012874287209213724,
      "loss": 1.373,
      "step": 11867
    },
    {
      "epoch": 0.42501835371640373,
      "grad_norm": 1.3626136779785156,
      "learning_rate": 0.00012873176234066698,
      "loss": 1.5498,
      "step": 11868
    },
    {
      "epoch": 0.42505416584597205,
      "grad_norm": 1.7889161109924316,
      "learning_rate": 0.0001287206522026479,
      "loss": 1.6937,
      "step": 11869
    },
    {
      "epoch": 0.4250899779755403,
      "grad_norm": 1.5051333904266357,
      "learning_rate": 0.0001287095416782295,
      "loss": 1.4984,
      "step": 11870
    },
    {
      "epoch": 0.4251257901051086,
      "grad_norm": 1.7558956146240234,
      "learning_rate": 0.00012869843076756125,
      "loss": 1.4792,
      "step": 11871
    },
    {
      "epoch": 0.42516160223467686,
      "grad_norm": 1.8319754600524902,
      "learning_rate": 0.0001286873194707926,
      "loss": 1.943,
      "step": 11872
    },
    {
      "epoch": 0.4251974143642452,
      "grad_norm": 1.6281830072402954,
      "learning_rate": 0.0001286762077880731,
      "loss": 1.3749,
      "step": 11873
    },
    {
      "epoch": 0.42523322649381345,
      "grad_norm": 1.5572781562805176,
      "learning_rate": 0.00012866509571955221,
      "loss": 1.7843,
      "step": 11874
    },
    {
      "epoch": 0.4252690386233817,
      "grad_norm": 1.4318057298660278,
      "learning_rate": 0.00012865398326537944,
      "loss": 1.423,
      "step": 11875
    },
    {
      "epoch": 0.42530485075295005,
      "grad_norm": 1.6866233348846436,
      "learning_rate": 0.0001286428704257043,
      "loss": 1.545,
      "step": 11876
    },
    {
      "epoch": 0.4253406628825183,
      "grad_norm": 1.7870521545410156,
      "learning_rate": 0.00012863175720067627,
      "loss": 1.7577,
      "step": 11877
    },
    {
      "epoch": 0.4253764750120866,
      "grad_norm": 1.4595621824264526,
      "learning_rate": 0.00012862064359044485,
      "loss": 1.5902,
      "step": 11878
    },
    {
      "epoch": 0.42541228714165485,
      "grad_norm": 1.636542797088623,
      "learning_rate": 0.00012860952959515962,
      "loss": 1.4817,
      "step": 11879
    },
    {
      "epoch": 0.4254480992712232,
      "grad_norm": 2.6225593090057373,
      "learning_rate": 0.00012859841521497008,
      "loss": 1.208,
      "step": 11880
    },
    {
      "epoch": 0.42548391140079145,
      "grad_norm": 1.9722484350204468,
      "learning_rate": 0.00012858730045002572,
      "loss": 1.7867,
      "step": 11881
    },
    {
      "epoch": 0.4255197235303597,
      "grad_norm": 1.486351490020752,
      "learning_rate": 0.00012857618530047615,
      "loss": 1.5268,
      "step": 11882
    },
    {
      "epoch": 0.42555553565992804,
      "grad_norm": 1.3822557926177979,
      "learning_rate": 0.0001285650697664708,
      "loss": 1.5033,
      "step": 11883
    },
    {
      "epoch": 0.4255913477894963,
      "grad_norm": 1.99789559841156,
      "learning_rate": 0.00012855395384815937,
      "loss": 1.8533,
      "step": 11884
    },
    {
      "epoch": 0.4256271599190646,
      "grad_norm": 1.781887173652649,
      "learning_rate": 0.00012854283754569127,
      "loss": 1.4333,
      "step": 11885
    },
    {
      "epoch": 0.42566297204863285,
      "grad_norm": 1.5997289419174194,
      "learning_rate": 0.00012853172085921613,
      "loss": 1.4675,
      "step": 11886
    },
    {
      "epoch": 0.4256987841782012,
      "grad_norm": 1.4423165321350098,
      "learning_rate": 0.00012852060378888347,
      "loss": 1.6605,
      "step": 11887
    },
    {
      "epoch": 0.42573459630776944,
      "grad_norm": 2.1727519035339355,
      "learning_rate": 0.00012850948633484288,
      "loss": 1.4342,
      "step": 11888
    },
    {
      "epoch": 0.4257704084373377,
      "grad_norm": 1.8773438930511475,
      "learning_rate": 0.00012849836849724392,
      "loss": 1.2294,
      "step": 11889
    },
    {
      "epoch": 0.42580622056690604,
      "grad_norm": 1.5362021923065186,
      "learning_rate": 0.0001284872502762362,
      "loss": 1.6375,
      "step": 11890
    },
    {
      "epoch": 0.4258420326964743,
      "grad_norm": 2.041426658630371,
      "learning_rate": 0.00012847613167196923,
      "loss": 1.4472,
      "step": 11891
    },
    {
      "epoch": 0.4258778448260426,
      "grad_norm": 1.3776212930679321,
      "learning_rate": 0.00012846501268459266,
      "loss": 1.2878,
      "step": 11892
    },
    {
      "epoch": 0.42591365695561084,
      "grad_norm": 1.8141348361968994,
      "learning_rate": 0.00012845389331425606,
      "loss": 1.2643,
      "step": 11893
    },
    {
      "epoch": 0.42594946908517917,
      "grad_norm": 1.4836647510528564,
      "learning_rate": 0.00012844277356110906,
      "loss": 1.6488,
      "step": 11894
    },
    {
      "epoch": 0.42598528121474744,
      "grad_norm": 1.3631209135055542,
      "learning_rate": 0.0001284316534253012,
      "loss": 1.3385,
      "step": 11895
    },
    {
      "epoch": 0.4260210933443157,
      "grad_norm": 2.604870319366455,
      "learning_rate": 0.0001284205329069821,
      "loss": 1.5673,
      "step": 11896
    },
    {
      "epoch": 0.42605690547388403,
      "grad_norm": 1.4967273473739624,
      "learning_rate": 0.00012840941200630143,
      "loss": 1.5662,
      "step": 11897
    },
    {
      "epoch": 0.4260927176034523,
      "grad_norm": 1.547292709350586,
      "learning_rate": 0.00012839829072340875,
      "loss": 1.4669,
      "step": 11898
    },
    {
      "epoch": 0.42612852973302057,
      "grad_norm": 1.751133918762207,
      "learning_rate": 0.0001283871690584537,
      "loss": 1.4935,
      "step": 11899
    },
    {
      "epoch": 0.42616434186258884,
      "grad_norm": 1.6280288696289062,
      "learning_rate": 0.00012837604701158587,
      "loss": 1.7348,
      "step": 11900
    },
    {
      "epoch": 0.42620015399215716,
      "grad_norm": 1.4411042928695679,
      "learning_rate": 0.000128364924582955,
      "loss": 1.5411,
      "step": 11901
    },
    {
      "epoch": 0.42623596612172543,
      "grad_norm": 1.841700553894043,
      "learning_rate": 0.00012835380177271058,
      "loss": 1.4303,
      "step": 11902
    },
    {
      "epoch": 0.4262717782512937,
      "grad_norm": 1.4292374849319458,
      "learning_rate": 0.00012834267858100238,
      "loss": 1.3253,
      "step": 11903
    },
    {
      "epoch": 0.426307590380862,
      "grad_norm": 1.8277291059494019,
      "learning_rate": 0.00012833155500798003,
      "loss": 1.5402,
      "step": 11904
    },
    {
      "epoch": 0.4263434025104303,
      "grad_norm": 1.7723889350891113,
      "learning_rate": 0.0001283204310537931,
      "loss": 1.5183,
      "step": 11905
    },
    {
      "epoch": 0.42637921463999856,
      "grad_norm": 1.554281234741211,
      "learning_rate": 0.00012830930671859132,
      "loss": 1.3464,
      "step": 11906
    },
    {
      "epoch": 0.42641502676956683,
      "grad_norm": 1.9932013750076294,
      "learning_rate": 0.00012829818200252432,
      "loss": 1.584,
      "step": 11907
    },
    {
      "epoch": 0.42645083889913515,
      "grad_norm": 1.3307794332504272,
      "learning_rate": 0.0001282870569057418,
      "loss": 1.5602,
      "step": 11908
    },
    {
      "epoch": 0.4264866510287034,
      "grad_norm": 1.5475578308105469,
      "learning_rate": 0.0001282759314283934,
      "loss": 1.3701,
      "step": 11909
    },
    {
      "epoch": 0.4265224631582717,
      "grad_norm": 1.49881911277771,
      "learning_rate": 0.00012826480557062884,
      "loss": 1.339,
      "step": 11910
    },
    {
      "epoch": 0.42655827528784,
      "grad_norm": 2.00813364982605,
      "learning_rate": 0.00012825367933259774,
      "loss": 1.5767,
      "step": 11911
    },
    {
      "epoch": 0.4265940874174083,
      "grad_norm": 1.6309877634048462,
      "learning_rate": 0.00012824255271444987,
      "loss": 1.5292,
      "step": 11912
    },
    {
      "epoch": 0.42662989954697655,
      "grad_norm": 1.6364531517028809,
      "learning_rate": 0.00012823142571633488,
      "loss": 1.6134,
      "step": 11913
    },
    {
      "epoch": 0.4266657116765448,
      "grad_norm": 1.7823848724365234,
      "learning_rate": 0.00012822029833840245,
      "loss": 1.3748,
      "step": 11914
    },
    {
      "epoch": 0.42670152380611315,
      "grad_norm": 1.6810747385025024,
      "learning_rate": 0.00012820917058080234,
      "loss": 1.6998,
      "step": 11915
    },
    {
      "epoch": 0.4267373359356814,
      "grad_norm": 1.9501549005508423,
      "learning_rate": 0.0001281980424436842,
      "loss": 1.5711,
      "step": 11916
    },
    {
      "epoch": 0.4267731480652497,
      "grad_norm": 1.8748064041137695,
      "learning_rate": 0.00012818691392719778,
      "loss": 1.5153,
      "step": 11917
    },
    {
      "epoch": 0.426808960194818,
      "grad_norm": 1.5978293418884277,
      "learning_rate": 0.00012817578503149276,
      "loss": 1.6448,
      "step": 11918
    },
    {
      "epoch": 0.4268447723243863,
      "grad_norm": 1.8714993000030518,
      "learning_rate": 0.00012816465575671895,
      "loss": 1.4063,
      "step": 11919
    },
    {
      "epoch": 0.42688058445395455,
      "grad_norm": 1.4037456512451172,
      "learning_rate": 0.000128153526103026,
      "loss": 1.5125,
      "step": 11920
    },
    {
      "epoch": 0.4269163965835228,
      "grad_norm": 1.7890756130218506,
      "learning_rate": 0.00012814239607056367,
      "loss": 1.3303,
      "step": 11921
    },
    {
      "epoch": 0.42695220871309114,
      "grad_norm": 2.235222578048706,
      "learning_rate": 0.00012813126565948171,
      "loss": 1.1991,
      "step": 11922
    },
    {
      "epoch": 0.4269880208426594,
      "grad_norm": 2.393725633621216,
      "learning_rate": 0.00012812013486992985,
      "loss": 1.7267,
      "step": 11923
    },
    {
      "epoch": 0.4270238329722277,
      "grad_norm": 1.7477926015853882,
      "learning_rate": 0.00012810900370205784,
      "loss": 1.7094,
      "step": 11924
    },
    {
      "epoch": 0.427059645101796,
      "grad_norm": 1.6497920751571655,
      "learning_rate": 0.00012809787215601546,
      "loss": 1.6286,
      "step": 11925
    },
    {
      "epoch": 0.4270954572313643,
      "grad_norm": 1.7119648456573486,
      "learning_rate": 0.00012808674023195244,
      "loss": 1.5985,
      "step": 11926
    },
    {
      "epoch": 0.42713126936093254,
      "grad_norm": 1.2613149881362915,
      "learning_rate": 0.00012807560793001856,
      "loss": 1.426,
      "step": 11927
    },
    {
      "epoch": 0.4271670814905008,
      "grad_norm": 1.6869969367980957,
      "learning_rate": 0.0001280644752503636,
      "loss": 1.4147,
      "step": 11928
    },
    {
      "epoch": 0.42720289362006914,
      "grad_norm": 1.247328758239746,
      "learning_rate": 0.00012805334219313734,
      "loss": 1.5805,
      "step": 11929
    },
    {
      "epoch": 0.4272387057496374,
      "grad_norm": 2.01601243019104,
      "learning_rate": 0.00012804220875848953,
      "loss": 1.4124,
      "step": 11930
    },
    {
      "epoch": 0.4272745178792057,
      "grad_norm": 2.005664587020874,
      "learning_rate": 0.00012803107494657,
      "loss": 1.4524,
      "step": 11931
    },
    {
      "epoch": 0.427310330008774,
      "grad_norm": 2.120532274246216,
      "learning_rate": 0.0001280199407575285,
      "loss": 1.7616,
      "step": 11932
    },
    {
      "epoch": 0.42734614213834227,
      "grad_norm": 1.7293062210083008,
      "learning_rate": 0.0001280088061915148,
      "loss": 1.3992,
      "step": 11933
    },
    {
      "epoch": 0.42738195426791054,
      "grad_norm": 2.858889579772949,
      "learning_rate": 0.00012799767124867874,
      "loss": 1.418,
      "step": 11934
    },
    {
      "epoch": 0.4274177663974788,
      "grad_norm": 3.2688639163970947,
      "learning_rate": 0.00012798653592917017,
      "loss": 1.47,
      "step": 11935
    },
    {
      "epoch": 0.42745357852704713,
      "grad_norm": 1.553168773651123,
      "learning_rate": 0.00012797540023313882,
      "loss": 1.6188,
      "step": 11936
    },
    {
      "epoch": 0.4274893906566154,
      "grad_norm": 1.8382031917572021,
      "learning_rate": 0.0001279642641607346,
      "loss": 1.5762,
      "step": 11937
    },
    {
      "epoch": 0.42752520278618367,
      "grad_norm": 1.7807074785232544,
      "learning_rate": 0.00012795312771210726,
      "loss": 1.4538,
      "step": 11938
    },
    {
      "epoch": 0.427561014915752,
      "grad_norm": 1.8465766906738281,
      "learning_rate": 0.00012794199088740665,
      "loss": 1.3205,
      "step": 11939
    },
    {
      "epoch": 0.42759682704532026,
      "grad_norm": 1.5674152374267578,
      "learning_rate": 0.00012793085368678254,
      "loss": 1.5163,
      "step": 11940
    },
    {
      "epoch": 0.42763263917488853,
      "grad_norm": 1.7111066579818726,
      "learning_rate": 0.00012791971611038488,
      "loss": 1.2398,
      "step": 11941
    },
    {
      "epoch": 0.4276684513044568,
      "grad_norm": 1.754284381866455,
      "learning_rate": 0.00012790857815836342,
      "loss": 1.4891,
      "step": 11942
    },
    {
      "epoch": 0.4277042634340251,
      "grad_norm": 1.5504387617111206,
      "learning_rate": 0.00012789743983086807,
      "loss": 1.4967,
      "step": 11943
    },
    {
      "epoch": 0.4277400755635934,
      "grad_norm": 1.2893853187561035,
      "learning_rate": 0.00012788630112804862,
      "loss": 1.4526,
      "step": 11944
    },
    {
      "epoch": 0.42777588769316166,
      "grad_norm": 1.9921671152114868,
      "learning_rate": 0.000127875162050055,
      "loss": 1.4574,
      "step": 11945
    },
    {
      "epoch": 0.42781169982273,
      "grad_norm": 1.7197723388671875,
      "learning_rate": 0.00012786402259703697,
      "loss": 1.2273,
      "step": 11946
    },
    {
      "epoch": 0.42784751195229825,
      "grad_norm": 2.255772590637207,
      "learning_rate": 0.0001278528827691445,
      "loss": 1.456,
      "step": 11947
    },
    {
      "epoch": 0.4278833240818665,
      "grad_norm": 2.25209903717041,
      "learning_rate": 0.00012784174256652743,
      "loss": 1.4299,
      "step": 11948
    },
    {
      "epoch": 0.4279191362114348,
      "grad_norm": 1.667028784751892,
      "learning_rate": 0.0001278306019893356,
      "loss": 1.021,
      "step": 11949
    },
    {
      "epoch": 0.4279549483410031,
      "grad_norm": 1.792156457901001,
      "learning_rate": 0.00012781946103771892,
      "loss": 1.347,
      "step": 11950
    },
    {
      "epoch": 0.4279907604705714,
      "grad_norm": 1.6226983070373535,
      "learning_rate": 0.0001278083197118273,
      "loss": 1.6862,
      "step": 11951
    },
    {
      "epoch": 0.42802657260013965,
      "grad_norm": 1.4472019672393799,
      "learning_rate": 0.00012779717801181058,
      "loss": 1.6126,
      "step": 11952
    },
    {
      "epoch": 0.428062384729708,
      "grad_norm": 2.137946844100952,
      "learning_rate": 0.0001277860359378187,
      "loss": 1.5559,
      "step": 11953
    },
    {
      "epoch": 0.42809819685927625,
      "grad_norm": 2.2679343223571777,
      "learning_rate": 0.00012777489349000156,
      "loss": 1.5858,
      "step": 11954
    },
    {
      "epoch": 0.4281340089888445,
      "grad_norm": 1.3759374618530273,
      "learning_rate": 0.00012776375066850902,
      "loss": 1.4777,
      "step": 11955
    },
    {
      "epoch": 0.4281698211184128,
      "grad_norm": 1.530863642692566,
      "learning_rate": 0.00012775260747349107,
      "loss": 1.1968,
      "step": 11956
    },
    {
      "epoch": 0.4282056332479811,
      "grad_norm": 1.4429298639297485,
      "learning_rate": 0.0001277414639050976,
      "loss": 1.6555,
      "step": 11957
    },
    {
      "epoch": 0.4282414453775494,
      "grad_norm": 2.4273345470428467,
      "learning_rate": 0.00012773031996347845,
      "loss": 1.3997,
      "step": 11958
    },
    {
      "epoch": 0.42827725750711765,
      "grad_norm": 1.5408649444580078,
      "learning_rate": 0.00012771917564878367,
      "loss": 1.3479,
      "step": 11959
    },
    {
      "epoch": 0.428313069636686,
      "grad_norm": 1.3045849800109863,
      "learning_rate": 0.00012770803096116308,
      "loss": 1.3514,
      "step": 11960
    },
    {
      "epoch": 0.42834888176625424,
      "grad_norm": 1.5886048078536987,
      "learning_rate": 0.00012769688590076673,
      "loss": 1.3276,
      "step": 11961
    },
    {
      "epoch": 0.4283846938958225,
      "grad_norm": 1.6271336078643799,
      "learning_rate": 0.00012768574046774446,
      "loss": 1.6032,
      "step": 11962
    },
    {
      "epoch": 0.4284205060253908,
      "grad_norm": 1.9809879064559937,
      "learning_rate": 0.00012767459466224632,
      "loss": 1.6735,
      "step": 11963
    },
    {
      "epoch": 0.4284563181549591,
      "grad_norm": 1.7093204259872437,
      "learning_rate": 0.00012766344848442218,
      "loss": 1.7656,
      "step": 11964
    },
    {
      "epoch": 0.4284921302845274,
      "grad_norm": 1.7945045232772827,
      "learning_rate": 0.00012765230193442198,
      "loss": 1.5466,
      "step": 11965
    },
    {
      "epoch": 0.42852794241409564,
      "grad_norm": 2.3185229301452637,
      "learning_rate": 0.0001276411550123958,
      "loss": 1.4034,
      "step": 11966
    },
    {
      "epoch": 0.42856375454366397,
      "grad_norm": 1.5377254486083984,
      "learning_rate": 0.00012763000771849348,
      "loss": 1.7145,
      "step": 11967
    },
    {
      "epoch": 0.42859956667323224,
      "grad_norm": 1.9657632112503052,
      "learning_rate": 0.00012761886005286508,
      "loss": 1.5286,
      "step": 11968
    },
    {
      "epoch": 0.4286353788028005,
      "grad_norm": 1.495625376701355,
      "learning_rate": 0.0001276077120156605,
      "loss": 1.3016,
      "step": 11969
    },
    {
      "epoch": 0.4286711909323688,
      "grad_norm": 1.7579660415649414,
      "learning_rate": 0.0001275965636070298,
      "loss": 1.2155,
      "step": 11970
    },
    {
      "epoch": 0.4287070030619371,
      "grad_norm": 2.0565755367279053,
      "learning_rate": 0.00012758541482712295,
      "loss": 1.3118,
      "step": 11971
    },
    {
      "epoch": 0.42874281519150537,
      "grad_norm": 1.7184526920318604,
      "learning_rate": 0.0001275742656760899,
      "loss": 1.3649,
      "step": 11972
    },
    {
      "epoch": 0.42877862732107364,
      "grad_norm": 1.7197214365005493,
      "learning_rate": 0.00012756311615408068,
      "loss": 1.2731,
      "step": 11973
    },
    {
      "epoch": 0.42881443945064196,
      "grad_norm": 1.3079050779342651,
      "learning_rate": 0.0001275519662612453,
      "loss": 1.3794,
      "step": 11974
    },
    {
      "epoch": 0.42885025158021023,
      "grad_norm": 1.7446621656417847,
      "learning_rate": 0.00012754081599773373,
      "loss": 1.6582,
      "step": 11975
    },
    {
      "epoch": 0.4288860637097785,
      "grad_norm": 1.5403920412063599,
      "learning_rate": 0.00012752966536369603,
      "loss": 1.3348,
      "step": 11976
    },
    {
      "epoch": 0.42892187583934677,
      "grad_norm": 1.4818763732910156,
      "learning_rate": 0.00012751851435928216,
      "loss": 1.5335,
      "step": 11977
    },
    {
      "epoch": 0.4289576879689151,
      "grad_norm": 1.5109602212905884,
      "learning_rate": 0.00012750736298464216,
      "loss": 1.3644,
      "step": 11978
    },
    {
      "epoch": 0.42899350009848336,
      "grad_norm": 2.0222465991973877,
      "learning_rate": 0.00012749621123992613,
      "loss": 1.3843,
      "step": 11979
    },
    {
      "epoch": 0.42902931222805163,
      "grad_norm": 1.4966095685958862,
      "learning_rate": 0.000127485059125284,
      "loss": 1.337,
      "step": 11980
    },
    {
      "epoch": 0.42906512435761995,
      "grad_norm": 1.5781891345977783,
      "learning_rate": 0.0001274739066408659,
      "loss": 1.6262,
      "step": 11981
    },
    {
      "epoch": 0.4291009364871882,
      "grad_norm": 1.5741047859191895,
      "learning_rate": 0.0001274627537868218,
      "loss": 1.5799,
      "step": 11982
    },
    {
      "epoch": 0.4291367486167565,
      "grad_norm": 1.477160096168518,
      "learning_rate": 0.00012745160056330178,
      "loss": 1.658,
      "step": 11983
    },
    {
      "epoch": 0.42917256074632476,
      "grad_norm": 2.298166036605835,
      "learning_rate": 0.00012744044697045586,
      "loss": 1.9077,
      "step": 11984
    },
    {
      "epoch": 0.4292083728758931,
      "grad_norm": 1.8923615217208862,
      "learning_rate": 0.00012742929300843417,
      "loss": 1.634,
      "step": 11985
    },
    {
      "epoch": 0.42924418500546135,
      "grad_norm": 2.1138875484466553,
      "learning_rate": 0.00012741813867738665,
      "loss": 1.8653,
      "step": 11986
    },
    {
      "epoch": 0.4292799971350296,
      "grad_norm": 2.7330331802368164,
      "learning_rate": 0.00012740698397746352,
      "loss": 1.5775,
      "step": 11987
    },
    {
      "epoch": 0.4293158092645979,
      "grad_norm": 1.834524154663086,
      "learning_rate": 0.0001273958289088147,
      "loss": 1.6638,
      "step": 11988
    },
    {
      "epoch": 0.4293516213941662,
      "grad_norm": 1.6845818758010864,
      "learning_rate": 0.0001273846734715904,
      "loss": 1.3616,
      "step": 11989
    },
    {
      "epoch": 0.4293874335237345,
      "grad_norm": 1.5297510623931885,
      "learning_rate": 0.0001273735176659406,
      "loss": 1.2825,
      "step": 11990
    },
    {
      "epoch": 0.42942324565330275,
      "grad_norm": 1.622412085533142,
      "learning_rate": 0.00012736236149201547,
      "loss": 1.5248,
      "step": 11991
    },
    {
      "epoch": 0.4294590577828711,
      "grad_norm": 1.8445333242416382,
      "learning_rate": 0.000127351204949965,
      "loss": 1.6047,
      "step": 11992
    },
    {
      "epoch": 0.42949486991243935,
      "grad_norm": 1.7746164798736572,
      "learning_rate": 0.0001273400480399394,
      "loss": 1.3141,
      "step": 11993
    },
    {
      "epoch": 0.4295306820420076,
      "grad_norm": 1.7072081565856934,
      "learning_rate": 0.00012732889076208872,
      "loss": 1.5455,
      "step": 11994
    },
    {
      "epoch": 0.4295664941715759,
      "grad_norm": 1.5068244934082031,
      "learning_rate": 0.00012731773311656304,
      "loss": 1.6053,
      "step": 11995
    },
    {
      "epoch": 0.4296023063011442,
      "grad_norm": 1.6956735849380493,
      "learning_rate": 0.00012730657510351252,
      "loss": 1.6389,
      "step": 11996
    },
    {
      "epoch": 0.4296381184307125,
      "grad_norm": 1.3766717910766602,
      "learning_rate": 0.00012729541672308722,
      "loss": 1.3844,
      "step": 11997
    },
    {
      "epoch": 0.42967393056028075,
      "grad_norm": 1.9688079357147217,
      "learning_rate": 0.00012728425797543731,
      "loss": 1.5564,
      "step": 11998
    },
    {
      "epoch": 0.4297097426898491,
      "grad_norm": 1.6623727083206177,
      "learning_rate": 0.00012727309886071292,
      "loss": 1.1367,
      "step": 11999
    },
    {
      "epoch": 0.42974555481941734,
      "grad_norm": 1.2293025255203247,
      "learning_rate": 0.00012726193937906416,
      "loss": 1.4147,
      "step": 12000
    },
    {
      "epoch": 0.4297813669489856,
      "grad_norm": 2.1265058517456055,
      "learning_rate": 0.00012725077953064119,
      "loss": 1.5568,
      "step": 12001
    },
    {
      "epoch": 0.4298171790785539,
      "grad_norm": 1.5156117677688599,
      "learning_rate": 0.0001272396193155941,
      "loss": 1.3789,
      "step": 12002
    },
    {
      "epoch": 0.4298529912081222,
      "grad_norm": 2.3674230575561523,
      "learning_rate": 0.0001272284587340731,
      "loss": 1.4465,
      "step": 12003
    },
    {
      "epoch": 0.4298888033376905,
      "grad_norm": 1.604242205619812,
      "learning_rate": 0.00012721729778622826,
      "loss": 1.6652,
      "step": 12004
    },
    {
      "epoch": 0.42992461546725874,
      "grad_norm": 1.3439925909042358,
      "learning_rate": 0.00012720613647220985,
      "loss": 1.2755,
      "step": 12005
    },
    {
      "epoch": 0.42996042759682707,
      "grad_norm": 3.2509360313415527,
      "learning_rate": 0.0001271949747921679,
      "loss": 1.6501,
      "step": 12006
    },
    {
      "epoch": 0.42999623972639534,
      "grad_norm": 1.9098608493804932,
      "learning_rate": 0.0001271838127462527,
      "loss": 1.5589,
      "step": 12007
    },
    {
      "epoch": 0.4300320518559636,
      "grad_norm": 2.2342143058776855,
      "learning_rate": 0.0001271726503346143,
      "loss": 1.221,
      "step": 12008
    },
    {
      "epoch": 0.4300678639855319,
      "grad_norm": 1.2617571353912354,
      "learning_rate": 0.00012716148755740302,
      "loss": 1.5833,
      "step": 12009
    },
    {
      "epoch": 0.4301036761151002,
      "grad_norm": 1.4430612325668335,
      "learning_rate": 0.00012715032441476892,
      "loss": 1.5831,
      "step": 12010
    },
    {
      "epoch": 0.43013948824466847,
      "grad_norm": 1.642476201057434,
      "learning_rate": 0.00012713916090686223,
      "loss": 1.626,
      "step": 12011
    },
    {
      "epoch": 0.43017530037423674,
      "grad_norm": 1.5486873388290405,
      "learning_rate": 0.00012712799703383314,
      "loss": 1.4034,
      "step": 12012
    },
    {
      "epoch": 0.43021111250380506,
      "grad_norm": 2.0201334953308105,
      "learning_rate": 0.00012711683279583181,
      "loss": 1.557,
      "step": 12013
    },
    {
      "epoch": 0.43024692463337333,
      "grad_norm": 1.413825511932373,
      "learning_rate": 0.00012710566819300854,
      "loss": 1.6422,
      "step": 12014
    },
    {
      "epoch": 0.4302827367629416,
      "grad_norm": 1.2569329738616943,
      "learning_rate": 0.00012709450322551338,
      "loss": 1.4474,
      "step": 12015
    },
    {
      "epoch": 0.43031854889250987,
      "grad_norm": 1.874431848526001,
      "learning_rate": 0.00012708333789349671,
      "loss": 1.6731,
      "step": 12016
    },
    {
      "epoch": 0.4303543610220782,
      "grad_norm": 1.6098730564117432,
      "learning_rate": 0.00012707217219710864,
      "loss": 1.416,
      "step": 12017
    },
    {
      "epoch": 0.43039017315164646,
      "grad_norm": 1.6854571104049683,
      "learning_rate": 0.0001270610061364994,
      "loss": 1.4469,
      "step": 12018
    },
    {
      "epoch": 0.43042598528121473,
      "grad_norm": 1.5974562168121338,
      "learning_rate": 0.00012704983971181924,
      "loss": 1.3327,
      "step": 12019
    },
    {
      "epoch": 0.43046179741078305,
      "grad_norm": 1.5737115144729614,
      "learning_rate": 0.00012703867292321837,
      "loss": 1.779,
      "step": 12020
    },
    {
      "epoch": 0.4304976095403513,
      "grad_norm": 1.6570041179656982,
      "learning_rate": 0.00012702750577084704,
      "loss": 1.5277,
      "step": 12021
    },
    {
      "epoch": 0.4305334216699196,
      "grad_norm": 1.3756600618362427,
      "learning_rate": 0.00012701633825485547,
      "loss": 1.616,
      "step": 12022
    },
    {
      "epoch": 0.43056923379948786,
      "grad_norm": 1.5845674276351929,
      "learning_rate": 0.00012700517037539394,
      "loss": 1.7271,
      "step": 12023
    },
    {
      "epoch": 0.4306050459290562,
      "grad_norm": 1.8880497217178345,
      "learning_rate": 0.00012699400213261262,
      "loss": 1.512,
      "step": 12024
    },
    {
      "epoch": 0.43064085805862445,
      "grad_norm": 1.240716576576233,
      "learning_rate": 0.0001269828335266619,
      "loss": 1.258,
      "step": 12025
    },
    {
      "epoch": 0.4306766701881927,
      "grad_norm": 1.8755682706832886,
      "learning_rate": 0.00012697166455769192,
      "loss": 1.6694,
      "step": 12026
    },
    {
      "epoch": 0.43071248231776105,
      "grad_norm": 1.8921705484390259,
      "learning_rate": 0.000126960495225853,
      "loss": 1.3913,
      "step": 12027
    },
    {
      "epoch": 0.4307482944473293,
      "grad_norm": 1.7119035720825195,
      "learning_rate": 0.00012694932553129537,
      "loss": 1.3978,
      "step": 12028
    },
    {
      "epoch": 0.4307841065768976,
      "grad_norm": 1.7253758907318115,
      "learning_rate": 0.00012693815547416934,
      "loss": 1.5819,
      "step": 12029
    },
    {
      "epoch": 0.43081991870646585,
      "grad_norm": 1.5359183549880981,
      "learning_rate": 0.00012692698505462516,
      "loss": 1.3002,
      "step": 12030
    },
    {
      "epoch": 0.4308557308360342,
      "grad_norm": 1.8340015411376953,
      "learning_rate": 0.00012691581427281317,
      "loss": 1.536,
      "step": 12031
    },
    {
      "epoch": 0.43089154296560245,
      "grad_norm": 1.6742093563079834,
      "learning_rate": 0.00012690464312888357,
      "loss": 1.3717,
      "step": 12032
    },
    {
      "epoch": 0.4309273550951707,
      "grad_norm": 2.0416200160980225,
      "learning_rate": 0.0001268934716229867,
      "loss": 1.751,
      "step": 12033
    },
    {
      "epoch": 0.43096316722473904,
      "grad_norm": 1.9123685359954834,
      "learning_rate": 0.0001268822997552729,
      "loss": 1.5788,
      "step": 12034
    },
    {
      "epoch": 0.4309989793543073,
      "grad_norm": 1.4536606073379517,
      "learning_rate": 0.00012687112752589243,
      "loss": 1.5731,
      "step": 12035
    },
    {
      "epoch": 0.4310347914838756,
      "grad_norm": 1.5964137315750122,
      "learning_rate": 0.00012685995493499558,
      "loss": 1.2681,
      "step": 12036
    },
    {
      "epoch": 0.43107060361344385,
      "grad_norm": 1.6218794584274292,
      "learning_rate": 0.00012684878198273268,
      "loss": 1.449,
      "step": 12037
    },
    {
      "epoch": 0.4311064157430122,
      "grad_norm": 1.6466941833496094,
      "learning_rate": 0.00012683760866925408,
      "loss": 1.5735,
      "step": 12038
    },
    {
      "epoch": 0.43114222787258044,
      "grad_norm": 1.5569286346435547,
      "learning_rate": 0.00012682643499471003,
      "loss": 1.4015,
      "step": 12039
    },
    {
      "epoch": 0.4311780400021487,
      "grad_norm": 1.7250436544418335,
      "learning_rate": 0.00012681526095925094,
      "loss": 1.4717,
      "step": 12040
    },
    {
      "epoch": 0.43121385213171703,
      "grad_norm": 1.3829773664474487,
      "learning_rate": 0.00012680408656302707,
      "loss": 1.5221,
      "step": 12041
    },
    {
      "epoch": 0.4312496642612853,
      "grad_norm": 1.6153755187988281,
      "learning_rate": 0.00012679291180618882,
      "loss": 1.6274,
      "step": 12042
    },
    {
      "epoch": 0.4312854763908536,
      "grad_norm": 1.467158555984497,
      "learning_rate": 0.00012678173668888645,
      "loss": 1.4342,
      "step": 12043
    },
    {
      "epoch": 0.43132128852042184,
      "grad_norm": 1.740787148475647,
      "learning_rate": 0.00012677056121127042,
      "loss": 1.855,
      "step": 12044
    },
    {
      "epoch": 0.43135710064999017,
      "grad_norm": 1.7290964126586914,
      "learning_rate": 0.000126759385373491,
      "loss": 1.5766,
      "step": 12045
    },
    {
      "epoch": 0.43139291277955844,
      "grad_norm": 1.2313038110733032,
      "learning_rate": 0.00012674820917569856,
      "loss": 1.0122,
      "step": 12046
    },
    {
      "epoch": 0.4314287249091267,
      "grad_norm": 1.7760591506958008,
      "learning_rate": 0.0001267370326180435,
      "loss": 1.4813,
      "step": 12047
    },
    {
      "epoch": 0.43146453703869503,
      "grad_norm": 2.26284122467041,
      "learning_rate": 0.0001267258557006761,
      "loss": 1.6065,
      "step": 12048
    },
    {
      "epoch": 0.4315003491682633,
      "grad_norm": 1.742746114730835,
      "learning_rate": 0.00012671467842374683,
      "loss": 1.8581,
      "step": 12049
    },
    {
      "epoch": 0.43153616129783157,
      "grad_norm": 2.243811845779419,
      "learning_rate": 0.000126703500787406,
      "loss": 1.4152,
      "step": 12050
    },
    {
      "epoch": 0.43157197342739984,
      "grad_norm": 1.9817683696746826,
      "learning_rate": 0.000126692322791804,
      "loss": 1.4592,
      "step": 12051
    },
    {
      "epoch": 0.43160778555696816,
      "grad_norm": 1.667299509048462,
      "learning_rate": 0.00012668114443709124,
      "loss": 1.8244,
      "step": 12052
    },
    {
      "epoch": 0.43164359768653643,
      "grad_norm": 1.9919838905334473,
      "learning_rate": 0.0001266699657234181,
      "loss": 1.7562,
      "step": 12053
    },
    {
      "epoch": 0.4316794098161047,
      "grad_norm": 1.6336984634399414,
      "learning_rate": 0.000126658786650935,
      "loss": 1.5799,
      "step": 12054
    },
    {
      "epoch": 0.431715221945673,
      "grad_norm": 1.2953890562057495,
      "learning_rate": 0.00012664760721979227,
      "loss": 1.2287,
      "step": 12055
    },
    {
      "epoch": 0.4317510340752413,
      "grad_norm": 1.822380781173706,
      "learning_rate": 0.00012663642743014037,
      "loss": 1.2884,
      "step": 12056
    },
    {
      "epoch": 0.43178684620480956,
      "grad_norm": 2.05786395072937,
      "learning_rate": 0.0001266252472821297,
      "loss": 1.4525,
      "step": 12057
    },
    {
      "epoch": 0.43182265833437783,
      "grad_norm": 1.759456992149353,
      "learning_rate": 0.00012661406677591067,
      "loss": 1.2702,
      "step": 12058
    },
    {
      "epoch": 0.43185847046394615,
      "grad_norm": 1.612596035003662,
      "learning_rate": 0.00012660288591163373,
      "loss": 1.5832,
      "step": 12059
    },
    {
      "epoch": 0.4318942825935144,
      "grad_norm": 1.3318983316421509,
      "learning_rate": 0.00012659170468944924,
      "loss": 1.2931,
      "step": 12060
    },
    {
      "epoch": 0.4319300947230827,
      "grad_norm": 1.8278427124023438,
      "learning_rate": 0.00012658052310950767,
      "loss": 1.6496,
      "step": 12061
    },
    {
      "epoch": 0.431965906852651,
      "grad_norm": 1.4100539684295654,
      "learning_rate": 0.00012656934117195946,
      "loss": 1.6082,
      "step": 12062
    },
    {
      "epoch": 0.4320017189822193,
      "grad_norm": 1.8377265930175781,
      "learning_rate": 0.00012655815887695503,
      "loss": 1.5018,
      "step": 12063
    },
    {
      "epoch": 0.43203753111178755,
      "grad_norm": 1.7149466276168823,
      "learning_rate": 0.00012654697622464483,
      "loss": 1.6423,
      "step": 12064
    },
    {
      "epoch": 0.4320733432413558,
      "grad_norm": 1.707377552986145,
      "learning_rate": 0.0001265357932151793,
      "loss": 1.5497,
      "step": 12065
    },
    {
      "epoch": 0.43210915537092415,
      "grad_norm": 1.3086154460906982,
      "learning_rate": 0.0001265246098487089,
      "loss": 1.3625,
      "step": 12066
    },
    {
      "epoch": 0.4321449675004924,
      "grad_norm": 1.9298018217086792,
      "learning_rate": 0.0001265134261253841,
      "loss": 1.4781,
      "step": 12067
    },
    {
      "epoch": 0.4321807796300607,
      "grad_norm": 1.6140490770339966,
      "learning_rate": 0.00012650224204535535,
      "loss": 1.5533,
      "step": 12068
    },
    {
      "epoch": 0.432216591759629,
      "grad_norm": 1.5894731283187866,
      "learning_rate": 0.00012649105760877312,
      "loss": 1.3304,
      "step": 12069
    },
    {
      "epoch": 0.4322524038891973,
      "grad_norm": 1.940897822380066,
      "learning_rate": 0.00012647987281578789,
      "loss": 1.7295,
      "step": 12070
    },
    {
      "epoch": 0.43228821601876555,
      "grad_norm": 1.6926662921905518,
      "learning_rate": 0.0001264686876665501,
      "loss": 1.2571,
      "step": 12071
    },
    {
      "epoch": 0.4323240281483338,
      "grad_norm": 1.5556963682174683,
      "learning_rate": 0.00012645750216121028,
      "loss": 1.3275,
      "step": 12072
    },
    {
      "epoch": 0.43235984027790214,
      "grad_norm": 1.9698765277862549,
      "learning_rate": 0.0001264463162999189,
      "loss": 1.517,
      "step": 12073
    },
    {
      "epoch": 0.4323956524074704,
      "grad_norm": 1.5589221715927124,
      "learning_rate": 0.00012643513008282645,
      "loss": 1.5866,
      "step": 12074
    },
    {
      "epoch": 0.4324314645370387,
      "grad_norm": 1.5207163095474243,
      "learning_rate": 0.00012642394351008337,
      "loss": 1.5868,
      "step": 12075
    },
    {
      "epoch": 0.432467276666607,
      "grad_norm": 1.5345786809921265,
      "learning_rate": 0.00012641275658184026,
      "loss": 1.5199,
      "step": 12076
    },
    {
      "epoch": 0.4325030887961753,
      "grad_norm": 1.7944084405899048,
      "learning_rate": 0.00012640156929824757,
      "loss": 1.627,
      "step": 12077
    },
    {
      "epoch": 0.43253890092574354,
      "grad_norm": 1.890235185623169,
      "learning_rate": 0.00012639038165945584,
      "loss": 1.4801,
      "step": 12078
    },
    {
      "epoch": 0.4325747130553118,
      "grad_norm": 1.4263070821762085,
      "learning_rate": 0.00012637919366561556,
      "loss": 1.4128,
      "step": 12079
    },
    {
      "epoch": 0.43261052518488013,
      "grad_norm": 1.502324104309082,
      "learning_rate": 0.00012636800531687728,
      "loss": 1.3671,
      "step": 12080
    },
    {
      "epoch": 0.4326463373144484,
      "grad_norm": 2.5059921741485596,
      "learning_rate": 0.00012635681661339146,
      "loss": 1.7858,
      "step": 12081
    },
    {
      "epoch": 0.4326821494440167,
      "grad_norm": 1.9203873872756958,
      "learning_rate": 0.00012634562755530867,
      "loss": 1.7033,
      "step": 12082
    },
    {
      "epoch": 0.432717961573585,
      "grad_norm": 1.613374948501587,
      "learning_rate": 0.00012633443814277946,
      "loss": 1.1675,
      "step": 12083
    },
    {
      "epoch": 0.43275377370315327,
      "grad_norm": 1.6708298921585083,
      "learning_rate": 0.00012632324837595434,
      "loss": 1.5178,
      "step": 12084
    },
    {
      "epoch": 0.43278958583272154,
      "grad_norm": 1.9573644399642944,
      "learning_rate": 0.00012631205825498388,
      "loss": 1.3646,
      "step": 12085
    },
    {
      "epoch": 0.4328253979622898,
      "grad_norm": 1.6646674871444702,
      "learning_rate": 0.0001263008677800186,
      "loss": 1.2774,
      "step": 12086
    },
    {
      "epoch": 0.43286121009185813,
      "grad_norm": 1.70304536819458,
      "learning_rate": 0.0001262896769512091,
      "loss": 1.2577,
      "step": 12087
    },
    {
      "epoch": 0.4328970222214264,
      "grad_norm": 2.205632209777832,
      "learning_rate": 0.0001262784857687059,
      "loss": 1.7386,
      "step": 12088
    },
    {
      "epoch": 0.43293283435099467,
      "grad_norm": 1.4651166200637817,
      "learning_rate": 0.00012626729423265956,
      "loss": 1.5016,
      "step": 12089
    },
    {
      "epoch": 0.432968646480563,
      "grad_norm": 2.221482992172241,
      "learning_rate": 0.00012625610234322064,
      "loss": 1.6606,
      "step": 12090
    },
    {
      "epoch": 0.43300445861013126,
      "grad_norm": 1.4816288948059082,
      "learning_rate": 0.00012624491010053976,
      "loss": 1.5373,
      "step": 12091
    },
    {
      "epoch": 0.43304027073969953,
      "grad_norm": 3.446542978286743,
      "learning_rate": 0.00012623371750476747,
      "loss": 2.0685,
      "step": 12092
    },
    {
      "epoch": 0.4330760828692678,
      "grad_norm": 1.557287573814392,
      "learning_rate": 0.00012622252455605435,
      "loss": 1.4696,
      "step": 12093
    },
    {
      "epoch": 0.4331118949988361,
      "grad_norm": 1.9975833892822266,
      "learning_rate": 0.00012621133125455093,
      "loss": 1.5988,
      "step": 12094
    },
    {
      "epoch": 0.4331477071284044,
      "grad_norm": 1.311399221420288,
      "learning_rate": 0.0001262001376004079,
      "loss": 1.2518,
      "step": 12095
    },
    {
      "epoch": 0.43318351925797266,
      "grad_norm": 1.9268242120742798,
      "learning_rate": 0.00012618894359377585,
      "loss": 1.3004,
      "step": 12096
    },
    {
      "epoch": 0.433219331387541,
      "grad_norm": 1.7908586263656616,
      "learning_rate": 0.0001261777492348053,
      "loss": 1.5952,
      "step": 12097
    },
    {
      "epoch": 0.43325514351710925,
      "grad_norm": 1.4732061624526978,
      "learning_rate": 0.00012616655452364693,
      "loss": 1.6161,
      "step": 12098
    },
    {
      "epoch": 0.4332909556466775,
      "grad_norm": 1.9595246315002441,
      "learning_rate": 0.0001261553594604513,
      "loss": 1.5878,
      "step": 12099
    },
    {
      "epoch": 0.4333267677762458,
      "grad_norm": 1.7076550722122192,
      "learning_rate": 0.00012614416404536905,
      "loss": 1.602,
      "step": 12100
    },
    {
      "epoch": 0.4333625799058141,
      "grad_norm": 1.8889092206954956,
      "learning_rate": 0.00012613296827855078,
      "loss": 1.4688,
      "step": 12101
    },
    {
      "epoch": 0.4333983920353824,
      "grad_norm": 1.7788153886795044,
      "learning_rate": 0.0001261217721601472,
      "loss": 1.7163,
      "step": 12102
    },
    {
      "epoch": 0.43343420416495065,
      "grad_norm": 1.702117919921875,
      "learning_rate": 0.00012611057569030876,
      "loss": 1.3403,
      "step": 12103
    },
    {
      "epoch": 0.433470016294519,
      "grad_norm": 1.5412400960922241,
      "learning_rate": 0.0001260993788691863,
      "loss": 1.5185,
      "step": 12104
    },
    {
      "epoch": 0.43350582842408725,
      "grad_norm": 1.8668831586837769,
      "learning_rate": 0.00012608818169693032,
      "loss": 1.6909,
      "step": 12105
    },
    {
      "epoch": 0.4335416405536555,
      "grad_norm": 2.0278031826019287,
      "learning_rate": 0.00012607698417369152,
      "loss": 1.4961,
      "step": 12106
    },
    {
      "epoch": 0.4335774526832238,
      "grad_norm": 1.955277919769287,
      "learning_rate": 0.00012606578629962054,
      "loss": 1.8627,
      "step": 12107
    },
    {
      "epoch": 0.4336132648127921,
      "grad_norm": 1.6680927276611328,
      "learning_rate": 0.00012605458807486797,
      "loss": 1.3742,
      "step": 12108
    },
    {
      "epoch": 0.4336490769423604,
      "grad_norm": 1.52384352684021,
      "learning_rate": 0.0001260433894995846,
      "loss": 1.182,
      "step": 12109
    },
    {
      "epoch": 0.43368488907192865,
      "grad_norm": 1.390926718711853,
      "learning_rate": 0.00012603219057392097,
      "loss": 1.4092,
      "step": 12110
    },
    {
      "epoch": 0.43372070120149697,
      "grad_norm": 1.52166748046875,
      "learning_rate": 0.0001260209912980278,
      "loss": 1.6895,
      "step": 12111
    },
    {
      "epoch": 0.43375651333106524,
      "grad_norm": 1.4530707597732544,
      "learning_rate": 0.0001260097916720558,
      "loss": 1.0416,
      "step": 12112
    },
    {
      "epoch": 0.4337923254606335,
      "grad_norm": 1.5703519582748413,
      "learning_rate": 0.00012599859169615558,
      "loss": 2.0043,
      "step": 12113
    },
    {
      "epoch": 0.4338281375902018,
      "grad_norm": 1.6512939929962158,
      "learning_rate": 0.00012598739137047784,
      "loss": 1.549,
      "step": 12114
    },
    {
      "epoch": 0.4338639497197701,
      "grad_norm": 2.4029769897460938,
      "learning_rate": 0.00012597619069517328,
      "loss": 1.8357,
      "step": 12115
    },
    {
      "epoch": 0.4338997618493384,
      "grad_norm": 1.5254900455474854,
      "learning_rate": 0.00012596498967039257,
      "loss": 1.5647,
      "step": 12116
    },
    {
      "epoch": 0.43393557397890664,
      "grad_norm": 2.040191411972046,
      "learning_rate": 0.0001259537882962864,
      "loss": 1.4003,
      "step": 12117
    },
    {
      "epoch": 0.43397138610847497,
      "grad_norm": 1.724648118019104,
      "learning_rate": 0.0001259425865730055,
      "loss": 1.3143,
      "step": 12118
    },
    {
      "epoch": 0.43400719823804323,
      "grad_norm": 1.1972649097442627,
      "learning_rate": 0.00012593138450070056,
      "loss": 1.3465,
      "step": 12119
    },
    {
      "epoch": 0.4340430103676115,
      "grad_norm": 1.4744864702224731,
      "learning_rate": 0.0001259201820795223,
      "loss": 1.3086,
      "step": 12120
    },
    {
      "epoch": 0.4340788224971798,
      "grad_norm": 1.8176345825195312,
      "learning_rate": 0.00012590897930962142,
      "loss": 1.3817,
      "step": 12121
    },
    {
      "epoch": 0.4341146346267481,
      "grad_norm": 1.617890477180481,
      "learning_rate": 0.00012589777619114863,
      "loss": 1.7052,
      "step": 12122
    },
    {
      "epoch": 0.43415044675631637,
      "grad_norm": 2.1249704360961914,
      "learning_rate": 0.0001258865727242547,
      "loss": 1.4104,
      "step": 12123
    },
    {
      "epoch": 0.43418625888588464,
      "grad_norm": 1.825442910194397,
      "learning_rate": 0.00012587536890909033,
      "loss": 1.6887,
      "step": 12124
    },
    {
      "epoch": 0.43422207101545296,
      "grad_norm": 1.7071682214736938,
      "learning_rate": 0.0001258641647458062,
      "loss": 1.5086,
      "step": 12125
    },
    {
      "epoch": 0.43425788314502123,
      "grad_norm": 1.7802238464355469,
      "learning_rate": 0.00012585296023455314,
      "loss": 1.6946,
      "step": 12126
    },
    {
      "epoch": 0.4342936952745895,
      "grad_norm": 1.907939076423645,
      "learning_rate": 0.00012584175537548183,
      "loss": 1.3304,
      "step": 12127
    },
    {
      "epoch": 0.43432950740415777,
      "grad_norm": 1.8659141063690186,
      "learning_rate": 0.00012583055016874303,
      "loss": 1.4764,
      "step": 12128
    },
    {
      "epoch": 0.4343653195337261,
      "grad_norm": 1.5259766578674316,
      "learning_rate": 0.00012581934461448747,
      "loss": 1.5803,
      "step": 12129
    },
    {
      "epoch": 0.43440113166329436,
      "grad_norm": 2.2665178775787354,
      "learning_rate": 0.00012580813871286597,
      "loss": 1.5027,
      "step": 12130
    },
    {
      "epoch": 0.43443694379286263,
      "grad_norm": 1.9262688159942627,
      "learning_rate": 0.00012579693246402924,
      "loss": 1.5477,
      "step": 12131
    },
    {
      "epoch": 0.43447275592243095,
      "grad_norm": 2.344991683959961,
      "learning_rate": 0.00012578572586812806,
      "loss": 1.3432,
      "step": 12132
    },
    {
      "epoch": 0.4345085680519992,
      "grad_norm": 2.302216053009033,
      "learning_rate": 0.00012577451892531322,
      "loss": 1.5703,
      "step": 12133
    },
    {
      "epoch": 0.4345443801815675,
      "grad_norm": 1.6896110773086548,
      "learning_rate": 0.00012576331163573548,
      "loss": 1.3081,
      "step": 12134
    },
    {
      "epoch": 0.43458019231113576,
      "grad_norm": 1.8697706460952759,
      "learning_rate": 0.00012575210399954557,
      "loss": 1.3996,
      "step": 12135
    },
    {
      "epoch": 0.4346160044407041,
      "grad_norm": 1.9721256494522095,
      "learning_rate": 0.00012574089601689433,
      "loss": 1.153,
      "step": 12136
    },
    {
      "epoch": 0.43465181657027235,
      "grad_norm": 2.7742340564727783,
      "learning_rate": 0.00012572968768793257,
      "loss": 1.7565,
      "step": 12137
    },
    {
      "epoch": 0.4346876286998406,
      "grad_norm": 2.0509705543518066,
      "learning_rate": 0.00012571847901281103,
      "loss": 1.3774,
      "step": 12138
    },
    {
      "epoch": 0.43472344082940895,
      "grad_norm": 1.5328233242034912,
      "learning_rate": 0.0001257072699916805,
      "loss": 1.3377,
      "step": 12139
    },
    {
      "epoch": 0.4347592529589772,
      "grad_norm": 1.7979052066802979,
      "learning_rate": 0.00012569606062469186,
      "loss": 1.544,
      "step": 12140
    },
    {
      "epoch": 0.4347950650885455,
      "grad_norm": 1.7817195653915405,
      "learning_rate": 0.00012568485091199585,
      "loss": 1.3429,
      "step": 12141
    },
    {
      "epoch": 0.43483087721811375,
      "grad_norm": 1.9444987773895264,
      "learning_rate": 0.0001256736408537433,
      "loss": 1.6757,
      "step": 12142
    },
    {
      "epoch": 0.4348666893476821,
      "grad_norm": 1.5094964504241943,
      "learning_rate": 0.00012566243045008504,
      "loss": 1.6575,
      "step": 12143
    },
    {
      "epoch": 0.43490250147725035,
      "grad_norm": 2.030996322631836,
      "learning_rate": 0.0001256512197011719,
      "loss": 1.636,
      "step": 12144
    },
    {
      "epoch": 0.4349383136068186,
      "grad_norm": 1.5814388990402222,
      "learning_rate": 0.00012564000860715464,
      "loss": 1.5221,
      "step": 12145
    },
    {
      "epoch": 0.43497412573638694,
      "grad_norm": 3.0495545864105225,
      "learning_rate": 0.00012562879716818416,
      "loss": 1.6251,
      "step": 12146
    },
    {
      "epoch": 0.4350099378659552,
      "grad_norm": 1.359506368637085,
      "learning_rate": 0.00012561758538441126,
      "loss": 1.3825,
      "step": 12147
    },
    {
      "epoch": 0.4350457499955235,
      "grad_norm": 1.534612774848938,
      "learning_rate": 0.0001256063732559868,
      "loss": 1.283,
      "step": 12148
    },
    {
      "epoch": 0.43508156212509175,
      "grad_norm": 1.858231544494629,
      "learning_rate": 0.0001255951607830616,
      "loss": 1.363,
      "step": 12149
    },
    {
      "epoch": 0.43511737425466007,
      "grad_norm": 1.4661773443222046,
      "learning_rate": 0.00012558394796578656,
      "loss": 1.3254,
      "step": 12150
    },
    {
      "epoch": 0.43515318638422834,
      "grad_norm": 1.5293601751327515,
      "learning_rate": 0.0001255727348043125,
      "loss": 1.7608,
      "step": 12151
    },
    {
      "epoch": 0.4351889985137966,
      "grad_norm": 1.2543964385986328,
      "learning_rate": 0.00012556152129879027,
      "loss": 1.5267,
      "step": 12152
    },
    {
      "epoch": 0.43522481064336493,
      "grad_norm": 1.9496577978134155,
      "learning_rate": 0.00012555030744937075,
      "loss": 1.2079,
      "step": 12153
    },
    {
      "epoch": 0.4352606227729332,
      "grad_norm": 2.417306661605835,
      "learning_rate": 0.0001255390932562048,
      "loss": 1.3698,
      "step": 12154
    },
    {
      "epoch": 0.4352964349025015,
      "grad_norm": 1.6135632991790771,
      "learning_rate": 0.00012552787871944327,
      "loss": 1.6963,
      "step": 12155
    },
    {
      "epoch": 0.43533224703206974,
      "grad_norm": 2.258427619934082,
      "learning_rate": 0.00012551666383923705,
      "loss": 1.4545,
      "step": 12156
    },
    {
      "epoch": 0.43536805916163807,
      "grad_norm": 1.6863501071929932,
      "learning_rate": 0.00012550544861573707,
      "loss": 1.113,
      "step": 12157
    },
    {
      "epoch": 0.43540387129120633,
      "grad_norm": 1.6821973323822021,
      "learning_rate": 0.0001254942330490942,
      "loss": 1.6309,
      "step": 12158
    },
    {
      "epoch": 0.4354396834207746,
      "grad_norm": 1.7726070880889893,
      "learning_rate": 0.00012548301713945925,
      "loss": 1.5211,
      "step": 12159
    },
    {
      "epoch": 0.43547549555034293,
      "grad_norm": 1.206180453300476,
      "learning_rate": 0.00012547180088698322,
      "loss": 1.4238,
      "step": 12160
    },
    {
      "epoch": 0.4355113076799112,
      "grad_norm": 2.325413227081299,
      "learning_rate": 0.00012546058429181692,
      "loss": 1.8346,
      "step": 12161
    },
    {
      "epoch": 0.43554711980947947,
      "grad_norm": 1.9589569568634033,
      "learning_rate": 0.00012544936735411135,
      "loss": 1.4175,
      "step": 12162
    },
    {
      "epoch": 0.43558293193904773,
      "grad_norm": 2.0499401092529297,
      "learning_rate": 0.00012543815007401733,
      "loss": 1.6312,
      "step": 12163
    },
    {
      "epoch": 0.43561874406861606,
      "grad_norm": 1.615285873413086,
      "learning_rate": 0.00012542693245168584,
      "loss": 1.5656,
      "step": 12164
    },
    {
      "epoch": 0.43565455619818433,
      "grad_norm": 2.342385768890381,
      "learning_rate": 0.00012541571448726775,
      "loss": 1.3668,
      "step": 12165
    },
    {
      "epoch": 0.4356903683277526,
      "grad_norm": 2.1468067169189453,
      "learning_rate": 0.00012540449618091403,
      "loss": 1.5807,
      "step": 12166
    },
    {
      "epoch": 0.4357261804573209,
      "grad_norm": 1.7686083316802979,
      "learning_rate": 0.00012539327753277555,
      "loss": 1.7541,
      "step": 12167
    },
    {
      "epoch": 0.4357619925868892,
      "grad_norm": 1.9072706699371338,
      "learning_rate": 0.00012538205854300334,
      "loss": 1.6316,
      "step": 12168
    },
    {
      "epoch": 0.43579780471645746,
      "grad_norm": 1.3254297971725464,
      "learning_rate": 0.00012537083921174822,
      "loss": 1.516,
      "step": 12169
    },
    {
      "epoch": 0.43583361684602573,
      "grad_norm": 1.3646912574768066,
      "learning_rate": 0.0001253596195391612,
      "loss": 1.3667,
      "step": 12170
    },
    {
      "epoch": 0.43586942897559405,
      "grad_norm": 1.3135740756988525,
      "learning_rate": 0.0001253483995253932,
      "loss": 1.669,
      "step": 12171
    },
    {
      "epoch": 0.4359052411051623,
      "grad_norm": 1.6707199811935425,
      "learning_rate": 0.00012533717917059516,
      "loss": 1.4503,
      "step": 12172
    },
    {
      "epoch": 0.4359410532347306,
      "grad_norm": 1.645582675933838,
      "learning_rate": 0.0001253259584749181,
      "loss": 1.3948,
      "step": 12173
    },
    {
      "epoch": 0.4359768653642989,
      "grad_norm": 1.7216689586639404,
      "learning_rate": 0.0001253147374385129,
      "loss": 1.4025,
      "step": 12174
    },
    {
      "epoch": 0.4360126774938672,
      "grad_norm": 1.6870415210723877,
      "learning_rate": 0.0001253035160615306,
      "loss": 1.7272,
      "step": 12175
    },
    {
      "epoch": 0.43604848962343545,
      "grad_norm": 1.3931629657745361,
      "learning_rate": 0.00012529229434412212,
      "loss": 1.3757,
      "step": 12176
    },
    {
      "epoch": 0.4360843017530037,
      "grad_norm": 1.6886948347091675,
      "learning_rate": 0.00012528107228643843,
      "loss": 1.4498,
      "step": 12177
    },
    {
      "epoch": 0.43612011388257205,
      "grad_norm": 1.4105387926101685,
      "learning_rate": 0.00012526984988863054,
      "loss": 1.5208,
      "step": 12178
    },
    {
      "epoch": 0.4361559260121403,
      "grad_norm": 2.1833765506744385,
      "learning_rate": 0.0001252586271508494,
      "loss": 1.7194,
      "step": 12179
    },
    {
      "epoch": 0.4361917381417086,
      "grad_norm": 1.9593642950057983,
      "learning_rate": 0.000125247404073246,
      "loss": 1.5396,
      "step": 12180
    },
    {
      "epoch": 0.43622755027127685,
      "grad_norm": 2.019223690032959,
      "learning_rate": 0.0001252361806559714,
      "loss": 1.6001,
      "step": 12181
    },
    {
      "epoch": 0.4362633624008452,
      "grad_norm": 2.205815076828003,
      "learning_rate": 0.00012522495689917647,
      "loss": 1.5243,
      "step": 12182
    },
    {
      "epoch": 0.43629917453041345,
      "grad_norm": 1.9048100709915161,
      "learning_rate": 0.00012521373280301233,
      "loss": 1.6801,
      "step": 12183
    },
    {
      "epoch": 0.4363349866599817,
      "grad_norm": 1.4723803997039795,
      "learning_rate": 0.0001252025083676299,
      "loss": 1.1884,
      "step": 12184
    },
    {
      "epoch": 0.43637079878955004,
      "grad_norm": 1.567564845085144,
      "learning_rate": 0.00012519128359318027,
      "loss": 1.4242,
      "step": 12185
    },
    {
      "epoch": 0.4364066109191183,
      "grad_norm": 1.5921801328659058,
      "learning_rate": 0.0001251800584798144,
      "loss": 1.4445,
      "step": 12186
    },
    {
      "epoch": 0.4364424230486866,
      "grad_norm": 2.2769577503204346,
      "learning_rate": 0.0001251688330276833,
      "loss": 1.5012,
      "step": 12187
    },
    {
      "epoch": 0.43647823517825485,
      "grad_norm": 1.4964098930358887,
      "learning_rate": 0.00012515760723693807,
      "loss": 1.7067,
      "step": 12188
    },
    {
      "epoch": 0.43651404730782317,
      "grad_norm": 1.4439140558242798,
      "learning_rate": 0.00012514638110772963,
      "loss": 1.1494,
      "step": 12189
    },
    {
      "epoch": 0.43654985943739144,
      "grad_norm": 1.3597850799560547,
      "learning_rate": 0.0001251351546402091,
      "loss": 1.672,
      "step": 12190
    },
    {
      "epoch": 0.4365856715669597,
      "grad_norm": 1.3259572982788086,
      "learning_rate": 0.00012512392783452746,
      "loss": 1.47,
      "step": 12191
    },
    {
      "epoch": 0.43662148369652803,
      "grad_norm": 1.7703850269317627,
      "learning_rate": 0.00012511270069083582,
      "loss": 1.4195,
      "step": 12192
    },
    {
      "epoch": 0.4366572958260963,
      "grad_norm": 1.2541024684906006,
      "learning_rate": 0.00012510147320928515,
      "loss": 1.3005,
      "step": 12193
    },
    {
      "epoch": 0.4366931079556646,
      "grad_norm": 1.4400556087493896,
      "learning_rate": 0.00012509024539002653,
      "loss": 1.1587,
      "step": 12194
    },
    {
      "epoch": 0.43672892008523284,
      "grad_norm": 1.3740122318267822,
      "learning_rate": 0.00012507901723321106,
      "loss": 1.5497,
      "step": 12195
    },
    {
      "epoch": 0.43676473221480117,
      "grad_norm": 1.596411108970642,
      "learning_rate": 0.00012506778873898976,
      "loss": 1.5982,
      "step": 12196
    },
    {
      "epoch": 0.43680054434436943,
      "grad_norm": 1.8824422359466553,
      "learning_rate": 0.00012505655990751368,
      "loss": 1.9263,
      "step": 12197
    },
    {
      "epoch": 0.4368363564739377,
      "grad_norm": 1.811933159828186,
      "learning_rate": 0.0001250453307389339,
      "loss": 1.6699,
      "step": 12198
    },
    {
      "epoch": 0.43687216860350603,
      "grad_norm": 1.7005141973495483,
      "learning_rate": 0.0001250341012334015,
      "loss": 1.6277,
      "step": 12199
    },
    {
      "epoch": 0.4369079807330743,
      "grad_norm": 1.5186580419540405,
      "learning_rate": 0.00012502287139106756,
      "loss": 1.174,
      "step": 12200
    },
    {
      "epoch": 0.43694379286264257,
      "grad_norm": 1.7548470497131348,
      "learning_rate": 0.0001250116412120832,
      "loss": 1.6404,
      "step": 12201
    },
    {
      "epoch": 0.43697960499221083,
      "grad_norm": 1.803139090538025,
      "learning_rate": 0.00012500041069659943,
      "loss": 1.2575,
      "step": 12202
    },
    {
      "epoch": 0.43701541712177916,
      "grad_norm": 1.5411206483840942,
      "learning_rate": 0.00012498917984476738,
      "loss": 1.5819,
      "step": 12203
    },
    {
      "epoch": 0.43705122925134743,
      "grad_norm": 1.999428391456604,
      "learning_rate": 0.00012497794865673817,
      "loss": 1.2803,
      "step": 12204
    },
    {
      "epoch": 0.4370870413809157,
      "grad_norm": 1.7252352237701416,
      "learning_rate": 0.0001249667171326629,
      "loss": 1.7101,
      "step": 12205
    },
    {
      "epoch": 0.437122853510484,
      "grad_norm": 1.4021079540252686,
      "learning_rate": 0.0001249554852726926,
      "loss": 1.3464,
      "step": 12206
    },
    {
      "epoch": 0.4371586656400523,
      "grad_norm": 1.7048399448394775,
      "learning_rate": 0.00012494425307697847,
      "loss": 1.5214,
      "step": 12207
    },
    {
      "epoch": 0.43719447776962056,
      "grad_norm": 1.2630609273910522,
      "learning_rate": 0.0001249330205456716,
      "loss": 1.5876,
      "step": 12208
    },
    {
      "epoch": 0.43723028989918883,
      "grad_norm": 1.411924958229065,
      "learning_rate": 0.00012492178767892307,
      "loss": 1.4632,
      "step": 12209
    },
    {
      "epoch": 0.43726610202875715,
      "grad_norm": 1.7386451959609985,
      "learning_rate": 0.00012491055447688405,
      "loss": 1.4687,
      "step": 12210
    },
    {
      "epoch": 0.4373019141583254,
      "grad_norm": 1.9168484210968018,
      "learning_rate": 0.00012489932093970568,
      "loss": 1.6312,
      "step": 12211
    },
    {
      "epoch": 0.4373377262878937,
      "grad_norm": 2.0553414821624756,
      "learning_rate": 0.00012488808706753902,
      "loss": 1.7418,
      "step": 12212
    },
    {
      "epoch": 0.437373538417462,
      "grad_norm": 1.310865879058838,
      "learning_rate": 0.00012487685286053526,
      "loss": 1.6169,
      "step": 12213
    },
    {
      "epoch": 0.4374093505470303,
      "grad_norm": 1.4720237255096436,
      "learning_rate": 0.00012486561831884552,
      "loss": 1.11,
      "step": 12214
    },
    {
      "epoch": 0.43744516267659855,
      "grad_norm": 1.5787687301635742,
      "learning_rate": 0.000124854383442621,
      "loss": 1.6203,
      "step": 12215
    },
    {
      "epoch": 0.4374809748061668,
      "grad_norm": 1.404030203819275,
      "learning_rate": 0.00012484314823201276,
      "loss": 1.2607,
      "step": 12216
    },
    {
      "epoch": 0.43751678693573515,
      "grad_norm": 1.9852256774902344,
      "learning_rate": 0.00012483191268717207,
      "loss": 1.6324,
      "step": 12217
    },
    {
      "epoch": 0.4375525990653034,
      "grad_norm": 1.8439569473266602,
      "learning_rate": 0.00012482067680824998,
      "loss": 1.3118,
      "step": 12218
    },
    {
      "epoch": 0.4375884111948717,
      "grad_norm": 1.3990365266799927,
      "learning_rate": 0.0001248094405953977,
      "loss": 1.4231,
      "step": 12219
    },
    {
      "epoch": 0.43762422332444,
      "grad_norm": 1.943037748336792,
      "learning_rate": 0.00012479820404876643,
      "loss": 1.4883,
      "step": 12220
    },
    {
      "epoch": 0.4376600354540083,
      "grad_norm": 1.4832576513290405,
      "learning_rate": 0.0001247869671685073,
      "loss": 1.3203,
      "step": 12221
    },
    {
      "epoch": 0.43769584758357655,
      "grad_norm": 1.6420166492462158,
      "learning_rate": 0.0001247757299547715,
      "loss": 1.254,
      "step": 12222
    },
    {
      "epoch": 0.4377316597131448,
      "grad_norm": 2.4458463191986084,
      "learning_rate": 0.00012476449240771023,
      "loss": 1.5143,
      "step": 12223
    },
    {
      "epoch": 0.43776747184271314,
      "grad_norm": 1.7896454334259033,
      "learning_rate": 0.0001247532545274746,
      "loss": 1.7974,
      "step": 12224
    },
    {
      "epoch": 0.4378032839722814,
      "grad_norm": 1.4857879877090454,
      "learning_rate": 0.00012474201631421588,
      "loss": 1.3927,
      "step": 12225
    },
    {
      "epoch": 0.4378390961018497,
      "grad_norm": 2.298076629638672,
      "learning_rate": 0.00012473077776808527,
      "loss": 1.5126,
      "step": 12226
    },
    {
      "epoch": 0.437874908231418,
      "grad_norm": 1.8825565576553345,
      "learning_rate": 0.00012471953888923393,
      "loss": 1.4028,
      "step": 12227
    },
    {
      "epoch": 0.43791072036098627,
      "grad_norm": 1.3885728120803833,
      "learning_rate": 0.00012470829967781307,
      "loss": 1.3174,
      "step": 12228
    },
    {
      "epoch": 0.43794653249055454,
      "grad_norm": 2.4574766159057617,
      "learning_rate": 0.00012469706013397395,
      "loss": 1.7062,
      "step": 12229
    },
    {
      "epoch": 0.4379823446201228,
      "grad_norm": 1.4123119115829468,
      "learning_rate": 0.00012468582025786774,
      "loss": 1.656,
      "step": 12230
    },
    {
      "epoch": 0.43801815674969113,
      "grad_norm": 1.421113133430481,
      "learning_rate": 0.0001246745800496456,
      "loss": 1.444,
      "step": 12231
    },
    {
      "epoch": 0.4380539688792594,
      "grad_norm": 2.3142082691192627,
      "learning_rate": 0.00012466333950945889,
      "loss": 1.6337,
      "step": 12232
    },
    {
      "epoch": 0.4380897810088277,
      "grad_norm": 2.1670494079589844,
      "learning_rate": 0.0001246520986374587,
      "loss": 1.3216,
      "step": 12233
    },
    {
      "epoch": 0.438125593138396,
      "grad_norm": 1.628150463104248,
      "learning_rate": 0.00012464085743379635,
      "loss": 1.4509,
      "step": 12234
    },
    {
      "epoch": 0.43816140526796427,
      "grad_norm": 1.204681634902954,
      "learning_rate": 0.000124629615898623,
      "loss": 1.4405,
      "step": 12235
    },
    {
      "epoch": 0.43819721739753253,
      "grad_norm": 1.9603196382522583,
      "learning_rate": 0.00012461837403209,
      "loss": 1.7334,
      "step": 12236
    },
    {
      "epoch": 0.4382330295271008,
      "grad_norm": 1.7064441442489624,
      "learning_rate": 0.0001246071318343485,
      "loss": 1.3215,
      "step": 12237
    },
    {
      "epoch": 0.43826884165666913,
      "grad_norm": 2.009902238845825,
      "learning_rate": 0.0001245958893055498,
      "loss": 1.3249,
      "step": 12238
    },
    {
      "epoch": 0.4383046537862374,
      "grad_norm": 1.3667140007019043,
      "learning_rate": 0.00012458464644584516,
      "loss": 1.254,
      "step": 12239
    },
    {
      "epoch": 0.43834046591580567,
      "grad_norm": 1.28581964969635,
      "learning_rate": 0.00012457340325538576,
      "loss": 1.4058,
      "step": 12240
    },
    {
      "epoch": 0.438376278045374,
      "grad_norm": 1.8916394710540771,
      "learning_rate": 0.00012456215973432295,
      "loss": 1.2694,
      "step": 12241
    },
    {
      "epoch": 0.43841209017494226,
      "grad_norm": 1.4369720220565796,
      "learning_rate": 0.00012455091588280793,
      "loss": 1.7439,
      "step": 12242
    },
    {
      "epoch": 0.43844790230451053,
      "grad_norm": 1.723207712173462,
      "learning_rate": 0.00012453967170099204,
      "loss": 1.3808,
      "step": 12243
    },
    {
      "epoch": 0.4384837144340788,
      "grad_norm": 2.0032958984375,
      "learning_rate": 0.00012452842718902647,
      "loss": 1.5483,
      "step": 12244
    },
    {
      "epoch": 0.4385195265636471,
      "grad_norm": 1.5736902952194214,
      "learning_rate": 0.00012451718234706262,
      "loss": 1.3339,
      "step": 12245
    },
    {
      "epoch": 0.4385553386932154,
      "grad_norm": 1.6149911880493164,
      "learning_rate": 0.00012450593717525167,
      "loss": 1.3523,
      "step": 12246
    },
    {
      "epoch": 0.43859115082278366,
      "grad_norm": 2.623713254928589,
      "learning_rate": 0.00012449469167374498,
      "loss": 1.7543,
      "step": 12247
    },
    {
      "epoch": 0.438626962952352,
      "grad_norm": 1.808893084526062,
      "learning_rate": 0.00012448344584269379,
      "loss": 1.5875,
      "step": 12248
    },
    {
      "epoch": 0.43866277508192025,
      "grad_norm": 1.4351035356521606,
      "learning_rate": 0.0001244721996822494,
      "loss": 1.4845,
      "step": 12249
    },
    {
      "epoch": 0.4386985872114885,
      "grad_norm": 1.4466192722320557,
      "learning_rate": 0.00012446095319256314,
      "loss": 1.476,
      "step": 12250
    },
    {
      "epoch": 0.4387343993410568,
      "grad_norm": 1.7350643873214722,
      "learning_rate": 0.00012444970637378631,
      "loss": 1.4816,
      "step": 12251
    },
    {
      "epoch": 0.4387702114706251,
      "grad_norm": 1.5797699689865112,
      "learning_rate": 0.0001244384592260702,
      "loss": 1.4879,
      "step": 12252
    },
    {
      "epoch": 0.4388060236001934,
      "grad_norm": 1.4239426851272583,
      "learning_rate": 0.00012442721174956616,
      "loss": 1.5366,
      "step": 12253
    },
    {
      "epoch": 0.43884183572976165,
      "grad_norm": 1.4414507150650024,
      "learning_rate": 0.0001244159639444255,
      "loss": 1.6121,
      "step": 12254
    },
    {
      "epoch": 0.43887764785933,
      "grad_norm": 1.5636897087097168,
      "learning_rate": 0.00012440471581079952,
      "loss": 1.758,
      "step": 12255
    },
    {
      "epoch": 0.43891345998889825,
      "grad_norm": 1.4484236240386963,
      "learning_rate": 0.0001243934673488396,
      "loss": 1.4626,
      "step": 12256
    },
    {
      "epoch": 0.4389492721184665,
      "grad_norm": 1.4698162078857422,
      "learning_rate": 0.00012438221855869702,
      "loss": 1.4597,
      "step": 12257
    },
    {
      "epoch": 0.4389850842480348,
      "grad_norm": 1.6307086944580078,
      "learning_rate": 0.00012437096944052317,
      "loss": 1.5859,
      "step": 12258
    },
    {
      "epoch": 0.4390208963776031,
      "grad_norm": 1.7705332040786743,
      "learning_rate": 0.0001243597199944693,
      "loss": 1.4369,
      "step": 12259
    },
    {
      "epoch": 0.4390567085071714,
      "grad_norm": 1.8133080005645752,
      "learning_rate": 0.0001243484702206869,
      "loss": 1.2604,
      "step": 12260
    },
    {
      "epoch": 0.43909252063673965,
      "grad_norm": 1.4457899332046509,
      "learning_rate": 0.00012433722011932717,
      "loss": 1.3294,
      "step": 12261
    },
    {
      "epoch": 0.43912833276630797,
      "grad_norm": 1.6493401527404785,
      "learning_rate": 0.00012432596969054157,
      "loss": 1.7194,
      "step": 12262
    },
    {
      "epoch": 0.43916414489587624,
      "grad_norm": 1.8156285285949707,
      "learning_rate": 0.0001243147189344814,
      "loss": 1.4789,
      "step": 12263
    },
    {
      "epoch": 0.4391999570254445,
      "grad_norm": 1.2668172121047974,
      "learning_rate": 0.0001243034678512981,
      "loss": 1.5012,
      "step": 12264
    },
    {
      "epoch": 0.4392357691550128,
      "grad_norm": 1.7122809886932373,
      "learning_rate": 0.00012429221644114294,
      "loss": 1.7531,
      "step": 12265
    },
    {
      "epoch": 0.4392715812845811,
      "grad_norm": 1.553624153137207,
      "learning_rate": 0.00012428096470416738,
      "loss": 1.2313,
      "step": 12266
    },
    {
      "epoch": 0.43930739341414937,
      "grad_norm": 2.247083902359009,
      "learning_rate": 0.00012426971264052275,
      "loss": 1.3864,
      "step": 12267
    },
    {
      "epoch": 0.43934320554371764,
      "grad_norm": 1.7876657247543335,
      "learning_rate": 0.00012425846025036042,
      "loss": 1.4111,
      "step": 12268
    },
    {
      "epoch": 0.43937901767328597,
      "grad_norm": 1.5700833797454834,
      "learning_rate": 0.0001242472075338318,
      "loss": 1.5301,
      "step": 12269
    },
    {
      "epoch": 0.43941482980285423,
      "grad_norm": 1.5357482433319092,
      "learning_rate": 0.0001242359544910883,
      "loss": 1.5518,
      "step": 12270
    },
    {
      "epoch": 0.4394506419324225,
      "grad_norm": 1.4696317911148071,
      "learning_rate": 0.00012422470112228125,
      "loss": 1.6317,
      "step": 12271
    },
    {
      "epoch": 0.43948645406199077,
      "grad_norm": 1.7287973165512085,
      "learning_rate": 0.00012421344742756215,
      "loss": 1.3712,
      "step": 12272
    },
    {
      "epoch": 0.4395222661915591,
      "grad_norm": 3.024156093597412,
      "learning_rate": 0.00012420219340708236,
      "loss": 1.6948,
      "step": 12273
    },
    {
      "epoch": 0.43955807832112737,
      "grad_norm": 1.4425593614578247,
      "learning_rate": 0.00012419093906099323,
      "loss": 1.5203,
      "step": 12274
    },
    {
      "epoch": 0.43959389045069563,
      "grad_norm": 1.4806897640228271,
      "learning_rate": 0.00012417968438944622,
      "loss": 1.3722,
      "step": 12275
    },
    {
      "epoch": 0.43962970258026396,
      "grad_norm": 2.641904354095459,
      "learning_rate": 0.0001241684293925928,
      "loss": 1.5145,
      "step": 12276
    },
    {
      "epoch": 0.43966551470983223,
      "grad_norm": 1.801173448562622,
      "learning_rate": 0.00012415717407058427,
      "loss": 1.5123,
      "step": 12277
    },
    {
      "epoch": 0.4397013268394005,
      "grad_norm": 1.3477219343185425,
      "learning_rate": 0.00012414591842357215,
      "loss": 1.2466,
      "step": 12278
    },
    {
      "epoch": 0.43973713896896877,
      "grad_norm": 1.4272079467773438,
      "learning_rate": 0.00012413466245170783,
      "loss": 1.1956,
      "step": 12279
    },
    {
      "epoch": 0.4397729510985371,
      "grad_norm": 1.5467042922973633,
      "learning_rate": 0.0001241234061551428,
      "loss": 1.3165,
      "step": 12280
    },
    {
      "epoch": 0.43980876322810536,
      "grad_norm": 2.0391323566436768,
      "learning_rate": 0.00012411214953402842,
      "loss": 1.5702,
      "step": 12281
    },
    {
      "epoch": 0.43984457535767363,
      "grad_norm": 1.9930360317230225,
      "learning_rate": 0.00012410089258851618,
      "loss": 1.6248,
      "step": 12282
    },
    {
      "epoch": 0.43988038748724195,
      "grad_norm": 1.4755240678787231,
      "learning_rate": 0.00012408963531875753,
      "loss": 1.6192,
      "step": 12283
    },
    {
      "epoch": 0.4399161996168102,
      "grad_norm": 1.4436578750610352,
      "learning_rate": 0.00012407837772490389,
      "loss": 1.3747,
      "step": 12284
    },
    {
      "epoch": 0.4399520117463785,
      "grad_norm": 1.4381917715072632,
      "learning_rate": 0.00012406711980710676,
      "loss": 1.5223,
      "step": 12285
    },
    {
      "epoch": 0.43998782387594676,
      "grad_norm": 1.4474295377731323,
      "learning_rate": 0.00012405586156551753,
      "loss": 1.4443,
      "step": 12286
    },
    {
      "epoch": 0.4400236360055151,
      "grad_norm": 2.0564377307891846,
      "learning_rate": 0.00012404460300028774,
      "loss": 1.13,
      "step": 12287
    },
    {
      "epoch": 0.44005944813508335,
      "grad_norm": 1.5260390043258667,
      "learning_rate": 0.00012403334411156884,
      "loss": 1.5361,
      "step": 12288
    },
    {
      "epoch": 0.4400952602646516,
      "grad_norm": 1.644852638244629,
      "learning_rate": 0.0001240220848995123,
      "loss": 1.5544,
      "step": 12289
    },
    {
      "epoch": 0.44013107239421995,
      "grad_norm": 1.850511908531189,
      "learning_rate": 0.00012401082536426958,
      "loss": 1.5099,
      "step": 12290
    },
    {
      "epoch": 0.4401668845237882,
      "grad_norm": 1.6762350797653198,
      "learning_rate": 0.00012399956550599218,
      "loss": 1.6942,
      "step": 12291
    },
    {
      "epoch": 0.4402026966533565,
      "grad_norm": 1.6215894222259521,
      "learning_rate": 0.0001239883053248316,
      "loss": 1.5009,
      "step": 12292
    },
    {
      "epoch": 0.44023850878292475,
      "grad_norm": 1.8665066957473755,
      "learning_rate": 0.0001239770448209393,
      "loss": 1.6581,
      "step": 12293
    },
    {
      "epoch": 0.4402743209124931,
      "grad_norm": 2.1635780334472656,
      "learning_rate": 0.00012396578399446678,
      "loss": 1.5384,
      "step": 12294
    },
    {
      "epoch": 0.44031013304206135,
      "grad_norm": 1.7075426578521729,
      "learning_rate": 0.00012395452284556558,
      "loss": 1.7423,
      "step": 12295
    },
    {
      "epoch": 0.4403459451716296,
      "grad_norm": 1.9434218406677246,
      "learning_rate": 0.00012394326137438714,
      "loss": 1.3288,
      "step": 12296
    },
    {
      "epoch": 0.44038175730119794,
      "grad_norm": 1.5657232999801636,
      "learning_rate": 0.000123931999581083,
      "loss": 1.2778,
      "step": 12297
    },
    {
      "epoch": 0.4404175694307662,
      "grad_norm": 1.5658713579177856,
      "learning_rate": 0.00012392073746580472,
      "loss": 1.5705,
      "step": 12298
    },
    {
      "epoch": 0.4404533815603345,
      "grad_norm": 1.3626611232757568,
      "learning_rate": 0.00012390947502870375,
      "loss": 1.286,
      "step": 12299
    },
    {
      "epoch": 0.44048919368990275,
      "grad_norm": 1.8273500204086304,
      "learning_rate": 0.00012389821226993164,
      "loss": 1.304,
      "step": 12300
    },
    {
      "epoch": 0.44052500581947107,
      "grad_norm": 1.7179958820343018,
      "learning_rate": 0.0001238869491896399,
      "loss": 1.674,
      "step": 12301
    },
    {
      "epoch": 0.44056081794903934,
      "grad_norm": 1.5758401155471802,
      "learning_rate": 0.00012387568578798005,
      "loss": 1.2103,
      "step": 12302
    },
    {
      "epoch": 0.4405966300786076,
      "grad_norm": 1.8869274854660034,
      "learning_rate": 0.00012386442206510368,
      "loss": 1.684,
      "step": 12303
    },
    {
      "epoch": 0.44063244220817593,
      "grad_norm": 1.4876755475997925,
      "learning_rate": 0.00012385315802116226,
      "loss": 1.5119,
      "step": 12304
    },
    {
      "epoch": 0.4406682543377442,
      "grad_norm": 2.015285015106201,
      "learning_rate": 0.0001238418936563074,
      "loss": 1.6014,
      "step": 12305
    },
    {
      "epoch": 0.44070406646731247,
      "grad_norm": 1.7106752395629883,
      "learning_rate": 0.0001238306289706906,
      "loss": 1.6048,
      "step": 12306
    },
    {
      "epoch": 0.44073987859688074,
      "grad_norm": 1.6270339488983154,
      "learning_rate": 0.00012381936396446344,
      "loss": 1.2119,
      "step": 12307
    },
    {
      "epoch": 0.44077569072644907,
      "grad_norm": 1.604411005973816,
      "learning_rate": 0.00012380809863777746,
      "loss": 1.4839,
      "step": 12308
    },
    {
      "epoch": 0.44081150285601733,
      "grad_norm": 3.05351185798645,
      "learning_rate": 0.00012379683299078422,
      "loss": 1.8318,
      "step": 12309
    },
    {
      "epoch": 0.4408473149855856,
      "grad_norm": 1.8058035373687744,
      "learning_rate": 0.00012378556702363527,
      "loss": 1.3283,
      "step": 12310
    },
    {
      "epoch": 0.4408831271151539,
      "grad_norm": 1.5172128677368164,
      "learning_rate": 0.00012377430073648218,
      "loss": 1.4016,
      "step": 12311
    },
    {
      "epoch": 0.4409189392447222,
      "grad_norm": 2.027984142303467,
      "learning_rate": 0.0001237630341294766,
      "loss": 1.5973,
      "step": 12312
    },
    {
      "epoch": 0.44095475137429047,
      "grad_norm": 1.950453281402588,
      "learning_rate": 0.00012375176720277002,
      "loss": 1.521,
      "step": 12313
    },
    {
      "epoch": 0.44099056350385873,
      "grad_norm": 2.8115463256835938,
      "learning_rate": 0.00012374049995651405,
      "loss": 1.5522,
      "step": 12314
    },
    {
      "epoch": 0.44102637563342706,
      "grad_norm": 1.677557110786438,
      "learning_rate": 0.00012372923239086024,
      "loss": 1.4244,
      "step": 12315
    },
    {
      "epoch": 0.44106218776299533,
      "grad_norm": 1.6209502220153809,
      "learning_rate": 0.00012371796450596028,
      "loss": 1.2205,
      "step": 12316
    },
    {
      "epoch": 0.4410979998925636,
      "grad_norm": 1.4655624628067017,
      "learning_rate": 0.00012370669630196567,
      "loss": 1.627,
      "step": 12317
    },
    {
      "epoch": 0.4411338120221319,
      "grad_norm": 2.140920639038086,
      "learning_rate": 0.00012369542777902805,
      "loss": 1.5118,
      "step": 12318
    },
    {
      "epoch": 0.4411696241517002,
      "grad_norm": 1.743704915046692,
      "learning_rate": 0.00012368415893729902,
      "loss": 1.6768,
      "step": 12319
    },
    {
      "epoch": 0.44120543628126846,
      "grad_norm": 1.668463945388794,
      "learning_rate": 0.00012367288977693016,
      "loss": 1.5263,
      "step": 12320
    },
    {
      "epoch": 0.44124124841083673,
      "grad_norm": 1.3567086458206177,
      "learning_rate": 0.0001236616202980731,
      "loss": 1.3551,
      "step": 12321
    },
    {
      "epoch": 0.44127706054040505,
      "grad_norm": 1.6387543678283691,
      "learning_rate": 0.0001236503505008795,
      "loss": 1.2526,
      "step": 12322
    },
    {
      "epoch": 0.4413128726699733,
      "grad_norm": 1.8538243770599365,
      "learning_rate": 0.0001236390803855009,
      "loss": 1.745,
      "step": 12323
    },
    {
      "epoch": 0.4413486847995416,
      "grad_norm": 1.6371301412582397,
      "learning_rate": 0.00012362780995208895,
      "loss": 1.5907,
      "step": 12324
    },
    {
      "epoch": 0.4413844969291099,
      "grad_norm": 1.6978520154953003,
      "learning_rate": 0.00012361653920079534,
      "loss": 1.6795,
      "step": 12325
    },
    {
      "epoch": 0.4414203090586782,
      "grad_norm": 1.94414484500885,
      "learning_rate": 0.00012360526813177163,
      "loss": 1.388,
      "step": 12326
    },
    {
      "epoch": 0.44145612118824645,
      "grad_norm": 1.762611746788025,
      "learning_rate": 0.00012359399674516955,
      "loss": 1.4216,
      "step": 12327
    },
    {
      "epoch": 0.4414919333178147,
      "grad_norm": 1.6278012990951538,
      "learning_rate": 0.00012358272504114058,
      "loss": 1.5087,
      "step": 12328
    },
    {
      "epoch": 0.44152774544738305,
      "grad_norm": 1.7553491592407227,
      "learning_rate": 0.00012357145301983651,
      "loss": 1.6467,
      "step": 12329
    },
    {
      "epoch": 0.4415635575769513,
      "grad_norm": 1.4429503679275513,
      "learning_rate": 0.00012356018068140895,
      "loss": 1.6824,
      "step": 12330
    },
    {
      "epoch": 0.4415993697065196,
      "grad_norm": 1.9834673404693604,
      "learning_rate": 0.00012354890802600957,
      "loss": 1.6054,
      "step": 12331
    },
    {
      "epoch": 0.4416351818360879,
      "grad_norm": 1.7594319581985474,
      "learning_rate": 0.00012353763505378997,
      "loss": 1.431,
      "step": 12332
    },
    {
      "epoch": 0.4416709939656562,
      "grad_norm": 2.2970147132873535,
      "learning_rate": 0.00012352636176490186,
      "loss": 2.0729,
      "step": 12333
    },
    {
      "epoch": 0.44170680609522445,
      "grad_norm": 1.2946401834487915,
      "learning_rate": 0.00012351508815949691,
      "loss": 1.4574,
      "step": 12334
    },
    {
      "epoch": 0.4417426182247927,
      "grad_norm": 1.8043230772018433,
      "learning_rate": 0.00012350381423772676,
      "loss": 1.379,
      "step": 12335
    },
    {
      "epoch": 0.44177843035436104,
      "grad_norm": 1.3034635782241821,
      "learning_rate": 0.00012349253999974314,
      "loss": 1.0734,
      "step": 12336
    },
    {
      "epoch": 0.4418142424839293,
      "grad_norm": 1.2619051933288574,
      "learning_rate": 0.00012348126544569767,
      "loss": 1.4902,
      "step": 12337
    },
    {
      "epoch": 0.4418500546134976,
      "grad_norm": 1.3467988967895508,
      "learning_rate": 0.00012346999057574209,
      "loss": 1.6634,
      "step": 12338
    },
    {
      "epoch": 0.4418858667430659,
      "grad_norm": 1.8174192905426025,
      "learning_rate": 0.00012345871539002801,
      "loss": 1.5092,
      "step": 12339
    },
    {
      "epoch": 0.44192167887263417,
      "grad_norm": 1.8187414407730103,
      "learning_rate": 0.00012344743988870722,
      "loss": 1.4353,
      "step": 12340
    },
    {
      "epoch": 0.44195749100220244,
      "grad_norm": 1.7980650663375854,
      "learning_rate": 0.00012343616407193135,
      "loss": 1.5176,
      "step": 12341
    },
    {
      "epoch": 0.4419933031317707,
      "grad_norm": 1.7426297664642334,
      "learning_rate": 0.00012342488793985214,
      "loss": 1.47,
      "step": 12342
    },
    {
      "epoch": 0.44202911526133903,
      "grad_norm": 1.9516719579696655,
      "learning_rate": 0.00012341361149262125,
      "loss": 1.3465,
      "step": 12343
    },
    {
      "epoch": 0.4420649273909073,
      "grad_norm": 2.266122817993164,
      "learning_rate": 0.00012340233473039045,
      "loss": 1.2277,
      "step": 12344
    },
    {
      "epoch": 0.44210073952047557,
      "grad_norm": 2.047231674194336,
      "learning_rate": 0.00012339105765331142,
      "loss": 1.1451,
      "step": 12345
    },
    {
      "epoch": 0.4421365516500439,
      "grad_norm": 2.4801621437072754,
      "learning_rate": 0.00012337978026153587,
      "loss": 1.3022,
      "step": 12346
    },
    {
      "epoch": 0.44217236377961217,
      "grad_norm": 1.6295838356018066,
      "learning_rate": 0.00012336850255521554,
      "loss": 1.4763,
      "step": 12347
    },
    {
      "epoch": 0.44220817590918043,
      "grad_norm": 1.4583467245101929,
      "learning_rate": 0.00012335722453450215,
      "loss": 1.5955,
      "step": 12348
    },
    {
      "epoch": 0.4422439880387487,
      "grad_norm": 1.6887112855911255,
      "learning_rate": 0.00012334594619954742,
      "loss": 1.3007,
      "step": 12349
    },
    {
      "epoch": 0.442279800168317,
      "grad_norm": 2.0861968994140625,
      "learning_rate": 0.0001233346675505031,
      "loss": 1.613,
      "step": 12350
    },
    {
      "epoch": 0.4423156122978853,
      "grad_norm": 2.0231857299804688,
      "learning_rate": 0.00012332338858752094,
      "loss": 1.5615,
      "step": 12351
    },
    {
      "epoch": 0.44235142442745357,
      "grad_norm": 2.4194977283477783,
      "learning_rate": 0.0001233121093107527,
      "loss": 1.5146,
      "step": 12352
    },
    {
      "epoch": 0.4423872365570219,
      "grad_norm": 2.190960645675659,
      "learning_rate": 0.00012330082972035006,
      "loss": 1.3055,
      "step": 12353
    },
    {
      "epoch": 0.44242304868659016,
      "grad_norm": 2.3201444149017334,
      "learning_rate": 0.00012328954981646482,
      "loss": 1.7702,
      "step": 12354
    },
    {
      "epoch": 0.44245886081615843,
      "grad_norm": 1.6395111083984375,
      "learning_rate": 0.0001232782695992487,
      "loss": 1.3856,
      "step": 12355
    },
    {
      "epoch": 0.4424946729457267,
      "grad_norm": 2.251711845397949,
      "learning_rate": 0.00012326698906885353,
      "loss": 1.8577,
      "step": 12356
    },
    {
      "epoch": 0.442530485075295,
      "grad_norm": 1.6219013929367065,
      "learning_rate": 0.00012325570822543103,
      "loss": 1.3516,
      "step": 12357
    },
    {
      "epoch": 0.4425662972048633,
      "grad_norm": 1.8324930667877197,
      "learning_rate": 0.00012324442706913296,
      "loss": 1.3791,
      "step": 12358
    },
    {
      "epoch": 0.44260210933443156,
      "grad_norm": 2.259781837463379,
      "learning_rate": 0.0001232331456001111,
      "loss": 1.6316,
      "step": 12359
    },
    {
      "epoch": 0.4426379214639999,
      "grad_norm": 2.230318307876587,
      "learning_rate": 0.00012322186381851725,
      "loss": 1.7068,
      "step": 12360
    },
    {
      "epoch": 0.44267373359356815,
      "grad_norm": 1.9674252271652222,
      "learning_rate": 0.00012321058172450318,
      "loss": 1.7323,
      "step": 12361
    },
    {
      "epoch": 0.4427095457231364,
      "grad_norm": 1.609484314918518,
      "learning_rate": 0.0001231992993182207,
      "loss": 1.5709,
      "step": 12362
    },
    {
      "epoch": 0.4427453578527047,
      "grad_norm": 1.5630927085876465,
      "learning_rate": 0.00012318801659982152,
      "loss": 1.605,
      "step": 12363
    },
    {
      "epoch": 0.442781169982273,
      "grad_norm": 1.5356788635253906,
      "learning_rate": 0.00012317673356945753,
      "loss": 1.161,
      "step": 12364
    },
    {
      "epoch": 0.4428169821118413,
      "grad_norm": 2.1178667545318604,
      "learning_rate": 0.00012316545022728043,
      "loss": 1.4586,
      "step": 12365
    },
    {
      "epoch": 0.44285279424140955,
      "grad_norm": 3.0236313343048096,
      "learning_rate": 0.00012315416657344213,
      "loss": 1.5712,
      "step": 12366
    },
    {
      "epoch": 0.4428886063709779,
      "grad_norm": 2.9537036418914795,
      "learning_rate": 0.00012314288260809435,
      "loss": 1.5299,
      "step": 12367
    },
    {
      "epoch": 0.44292441850054615,
      "grad_norm": 1.7508946657180786,
      "learning_rate": 0.00012313159833138892,
      "loss": 1.7762,
      "step": 12368
    },
    {
      "epoch": 0.4429602306301144,
      "grad_norm": 1.5886273384094238,
      "learning_rate": 0.00012312031374347773,
      "loss": 1.4263,
      "step": 12369
    },
    {
      "epoch": 0.4429960427596827,
      "grad_norm": 1.5932828187942505,
      "learning_rate": 0.00012310902884451252,
      "loss": 1.6544,
      "step": 12370
    },
    {
      "epoch": 0.443031854889251,
      "grad_norm": 1.7331477403640747,
      "learning_rate": 0.00012309774363464514,
      "loss": 1.6996,
      "step": 12371
    },
    {
      "epoch": 0.4430676670188193,
      "grad_norm": 1.5816339254379272,
      "learning_rate": 0.00012308645811402738,
      "loss": 1.469,
      "step": 12372
    },
    {
      "epoch": 0.44310347914838755,
      "grad_norm": 2.0275886058807373,
      "learning_rate": 0.00012307517228281117,
      "loss": 1.8279,
      "step": 12373
    },
    {
      "epoch": 0.44313929127795587,
      "grad_norm": 1.7863155603408813,
      "learning_rate": 0.00012306388614114822,
      "loss": 1.543,
      "step": 12374
    },
    {
      "epoch": 0.44317510340752414,
      "grad_norm": 1.4801067113876343,
      "learning_rate": 0.00012305259968919046,
      "loss": 1.5689,
      "step": 12375
    },
    {
      "epoch": 0.4432109155370924,
      "grad_norm": 1.4803080558776855,
      "learning_rate": 0.00012304131292708968,
      "loss": 1.3355,
      "step": 12376
    },
    {
      "epoch": 0.4432467276666607,
      "grad_norm": 1.7210575342178345,
      "learning_rate": 0.0001230300258549978,
      "loss": 1.602,
      "step": 12377
    },
    {
      "epoch": 0.443282539796229,
      "grad_norm": 1.4873367547988892,
      "learning_rate": 0.00012301873847306657,
      "loss": 1.3953,
      "step": 12378
    },
    {
      "epoch": 0.44331835192579727,
      "grad_norm": 1.6360687017440796,
      "learning_rate": 0.00012300745078144796,
      "loss": 1.4753,
      "step": 12379
    },
    {
      "epoch": 0.44335416405536554,
      "grad_norm": 1.5548124313354492,
      "learning_rate": 0.00012299616278029375,
      "loss": 1.3875,
      "step": 12380
    },
    {
      "epoch": 0.4433899761849338,
      "grad_norm": 1.4438000917434692,
      "learning_rate": 0.00012298487446975583,
      "loss": 1.5891,
      "step": 12381
    },
    {
      "epoch": 0.44342578831450213,
      "grad_norm": 2.041623830795288,
      "learning_rate": 0.0001229735858499861,
      "loss": 1.3645,
      "step": 12382
    },
    {
      "epoch": 0.4434616004440704,
      "grad_norm": 1.9327969551086426,
      "learning_rate": 0.0001229622969211364,
      "loss": 1.5932,
      "step": 12383
    },
    {
      "epoch": 0.44349741257363867,
      "grad_norm": 1.3270928859710693,
      "learning_rate": 0.00012295100768335858,
      "loss": 1.7006,
      "step": 12384
    },
    {
      "epoch": 0.443533224703207,
      "grad_norm": 1.8910813331604004,
      "learning_rate": 0.00012293971813680458,
      "loss": 1.6462,
      "step": 12385
    },
    {
      "epoch": 0.44356903683277527,
      "grad_norm": 2.8436388969421387,
      "learning_rate": 0.00012292842828162627,
      "loss": 1.5743,
      "step": 12386
    },
    {
      "epoch": 0.44360484896234353,
      "grad_norm": 1.7232835292816162,
      "learning_rate": 0.00012291713811797553,
      "loss": 1.4634,
      "step": 12387
    },
    {
      "epoch": 0.4436406610919118,
      "grad_norm": 1.5384464263916016,
      "learning_rate": 0.00012290584764600425,
      "loss": 1.4929,
      "step": 12388
    },
    {
      "epoch": 0.4436764732214801,
      "grad_norm": 1.418431043624878,
      "learning_rate": 0.00012289455686586434,
      "loss": 1.4444,
      "step": 12389
    },
    {
      "epoch": 0.4437122853510484,
      "grad_norm": 1.4715633392333984,
      "learning_rate": 0.0001228832657777077,
      "loss": 1.5561,
      "step": 12390
    },
    {
      "epoch": 0.44374809748061667,
      "grad_norm": 1.7735953330993652,
      "learning_rate": 0.00012287197438168624,
      "loss": 1.9554,
      "step": 12391
    },
    {
      "epoch": 0.443783909610185,
      "grad_norm": 1.6121642589569092,
      "learning_rate": 0.00012286068267795185,
      "loss": 1.3333,
      "step": 12392
    },
    {
      "epoch": 0.44381972173975326,
      "grad_norm": 1.414981722831726,
      "learning_rate": 0.00012284939066665648,
      "loss": 1.5743,
      "step": 12393
    },
    {
      "epoch": 0.44385553386932153,
      "grad_norm": 1.645015001296997,
      "learning_rate": 0.00012283809834795202,
      "loss": 1.6721,
      "step": 12394
    },
    {
      "epoch": 0.4438913459988898,
      "grad_norm": 1.310796856880188,
      "learning_rate": 0.00012282680572199043,
      "loss": 1.4809,
      "step": 12395
    },
    {
      "epoch": 0.4439271581284581,
      "grad_norm": 1.5144050121307373,
      "learning_rate": 0.00012281551278892357,
      "loss": 1.1173,
      "step": 12396
    },
    {
      "epoch": 0.4439629702580264,
      "grad_norm": 1.5878082513809204,
      "learning_rate": 0.00012280421954890346,
      "loss": 1.4621,
      "step": 12397
    },
    {
      "epoch": 0.44399878238759466,
      "grad_norm": 2.0923068523406982,
      "learning_rate": 0.000122792926002082,
      "loss": 1.4257,
      "step": 12398
    },
    {
      "epoch": 0.444034594517163,
      "grad_norm": 1.6810699701309204,
      "learning_rate": 0.00012278163214861107,
      "loss": 1.3412,
      "step": 12399
    },
    {
      "epoch": 0.44407040664673125,
      "grad_norm": 1.7527968883514404,
      "learning_rate": 0.00012277033798864268,
      "loss": 1.673,
      "step": 12400
    },
    {
      "epoch": 0.4441062187762995,
      "grad_norm": 2.4101831912994385,
      "learning_rate": 0.00012275904352232876,
      "loss": 1.4516,
      "step": 12401
    },
    {
      "epoch": 0.4441420309058678,
      "grad_norm": 1.7955946922302246,
      "learning_rate": 0.00012274774874982132,
      "loss": 1.8107,
      "step": 12402
    },
    {
      "epoch": 0.4441778430354361,
      "grad_norm": 1.6984844207763672,
      "learning_rate": 0.0001227364536712722,
      "loss": 1.5125,
      "step": 12403
    },
    {
      "epoch": 0.4442136551650044,
      "grad_norm": 1.418100357055664,
      "learning_rate": 0.00012272515828683344,
      "loss": 1.487,
      "step": 12404
    },
    {
      "epoch": 0.44424946729457265,
      "grad_norm": 1.629875898361206,
      "learning_rate": 0.00012271386259665701,
      "loss": 1.4305,
      "step": 12405
    },
    {
      "epoch": 0.444285279424141,
      "grad_norm": 2.2954277992248535,
      "learning_rate": 0.00012270256660089484,
      "loss": 1.7927,
      "step": 12406
    },
    {
      "epoch": 0.44432109155370925,
      "grad_norm": 1.8321059942245483,
      "learning_rate": 0.00012269127029969893,
      "loss": 1.5456,
      "step": 12407
    },
    {
      "epoch": 0.4443569036832775,
      "grad_norm": 1.7216614484786987,
      "learning_rate": 0.00012267997369322126,
      "loss": 1.5889,
      "step": 12408
    },
    {
      "epoch": 0.4443927158128458,
      "grad_norm": 1.7237794399261475,
      "learning_rate": 0.00012266867678161375,
      "loss": 1.3024,
      "step": 12409
    },
    {
      "epoch": 0.4444285279424141,
      "grad_norm": 1.995652437210083,
      "learning_rate": 0.00012265737956502847,
      "loss": 1.3176,
      "step": 12410
    },
    {
      "epoch": 0.4444643400719824,
      "grad_norm": 1.5316146612167358,
      "learning_rate": 0.0001226460820436174,
      "loss": 1.627,
      "step": 12411
    },
    {
      "epoch": 0.44450015220155065,
      "grad_norm": 1.9641482830047607,
      "learning_rate": 0.00012263478421753243,
      "loss": 1.3494,
      "step": 12412
    },
    {
      "epoch": 0.44453596433111897,
      "grad_norm": 1.9962435960769653,
      "learning_rate": 0.0001226234860869257,
      "loss": 1.3349,
      "step": 12413
    },
    {
      "epoch": 0.44457177646068724,
      "grad_norm": 1.4365565776824951,
      "learning_rate": 0.00012261218765194913,
      "loss": 1.4867,
      "step": 12414
    },
    {
      "epoch": 0.4446075885902555,
      "grad_norm": 1.7713817358016968,
      "learning_rate": 0.00012260088891275476,
      "loss": 1.5989,
      "step": 12415
    },
    {
      "epoch": 0.4446434007198238,
      "grad_norm": 1.790130376815796,
      "learning_rate": 0.00012258958986949455,
      "loss": 1.5266,
      "step": 12416
    },
    {
      "epoch": 0.4446792128493921,
      "grad_norm": 1.350547432899475,
      "learning_rate": 0.00012257829052232056,
      "loss": 1.8426,
      "step": 12417
    },
    {
      "epoch": 0.44471502497896037,
      "grad_norm": 1.452842116355896,
      "learning_rate": 0.00012256699087138479,
      "loss": 1.4189,
      "step": 12418
    },
    {
      "epoch": 0.44475083710852864,
      "grad_norm": 2.3444418907165527,
      "learning_rate": 0.0001225556909168393,
      "loss": 1.565,
      "step": 12419
    },
    {
      "epoch": 0.44478664923809696,
      "grad_norm": 2.152139902114868,
      "learning_rate": 0.00012254439065883602,
      "loss": 1.4686,
      "step": 12420
    },
    {
      "epoch": 0.44482246136766523,
      "grad_norm": 1.6714543104171753,
      "learning_rate": 0.0001225330900975271,
      "loss": 1.4201,
      "step": 12421
    },
    {
      "epoch": 0.4448582734972335,
      "grad_norm": 1.691991925239563,
      "learning_rate": 0.00012252178923306448,
      "loss": 1.2425,
      "step": 12422
    },
    {
      "epoch": 0.44489408562680177,
      "grad_norm": 1.3791453838348389,
      "learning_rate": 0.00012251048806560027,
      "loss": 1.2884,
      "step": 12423
    },
    {
      "epoch": 0.4449298977563701,
      "grad_norm": 1.6652225255966187,
      "learning_rate": 0.00012249918659528648,
      "loss": 1.4209,
      "step": 12424
    },
    {
      "epoch": 0.44496570988593837,
      "grad_norm": 2.57869291305542,
      "learning_rate": 0.0001224878848222751,
      "loss": 1.6552,
      "step": 12425
    },
    {
      "epoch": 0.44500152201550663,
      "grad_norm": 1.6043140888214111,
      "learning_rate": 0.0001224765827467183,
      "loss": 1.2117,
      "step": 12426
    },
    {
      "epoch": 0.44503733414507496,
      "grad_norm": 1.4723271131515503,
      "learning_rate": 0.00012246528036876807,
      "loss": 1.5539,
      "step": 12427
    },
    {
      "epoch": 0.4450731462746432,
      "grad_norm": 2.1970090866088867,
      "learning_rate": 0.00012245397768857646,
      "loss": 1.4757,
      "step": 12428
    },
    {
      "epoch": 0.4451089584042115,
      "grad_norm": 1.648956537246704,
      "learning_rate": 0.0001224426747062955,
      "loss": 1.4227,
      "step": 12429
    },
    {
      "epoch": 0.44514477053377977,
      "grad_norm": 1.9277909994125366,
      "learning_rate": 0.00012243137142207733,
      "loss": 1.6855,
      "step": 12430
    },
    {
      "epoch": 0.4451805826633481,
      "grad_norm": 1.7373301982879639,
      "learning_rate": 0.000122420067836074,
      "loss": 1.3754,
      "step": 12431
    },
    {
      "epoch": 0.44521639479291636,
      "grad_norm": 1.5616296529769897,
      "learning_rate": 0.0001224087639484376,
      "loss": 1.539,
      "step": 12432
    },
    {
      "epoch": 0.44525220692248463,
      "grad_norm": 2.611417055130005,
      "learning_rate": 0.00012239745975932016,
      "loss": 1.4478,
      "step": 12433
    },
    {
      "epoch": 0.44528801905205295,
      "grad_norm": 3.34085750579834,
      "learning_rate": 0.00012238615526887378,
      "loss": 2.1207,
      "step": 12434
    },
    {
      "epoch": 0.4453238311816212,
      "grad_norm": 1.9090518951416016,
      "learning_rate": 0.00012237485047725057,
      "loss": 1.6396,
      "step": 12435
    },
    {
      "epoch": 0.4453596433111895,
      "grad_norm": 1.521642804145813,
      "learning_rate": 0.00012236354538460259,
      "loss": 1.5772,
      "step": 12436
    },
    {
      "epoch": 0.44539545544075776,
      "grad_norm": 2.2825441360473633,
      "learning_rate": 0.000122352239991082,
      "loss": 1.5948,
      "step": 12437
    },
    {
      "epoch": 0.4454312675703261,
      "grad_norm": 1.408004641532898,
      "learning_rate": 0.0001223409342968408,
      "loss": 1.244,
      "step": 12438
    },
    {
      "epoch": 0.44546707969989435,
      "grad_norm": 1.4252254962921143,
      "learning_rate": 0.00012232962830203116,
      "loss": 1.5186,
      "step": 12439
    },
    {
      "epoch": 0.4455028918294626,
      "grad_norm": 2.0045323371887207,
      "learning_rate": 0.00012231832200680518,
      "loss": 1.4156,
      "step": 12440
    },
    {
      "epoch": 0.44553870395903095,
      "grad_norm": 1.9673701524734497,
      "learning_rate": 0.00012230701541131499,
      "loss": 1.2987,
      "step": 12441
    },
    {
      "epoch": 0.4455745160885992,
      "grad_norm": 1.8658106327056885,
      "learning_rate": 0.00012229570851571265,
      "loss": 1.5869,
      "step": 12442
    },
    {
      "epoch": 0.4456103282181675,
      "grad_norm": 2.748161554336548,
      "learning_rate": 0.00012228440132015033,
      "loss": 1.375,
      "step": 12443
    },
    {
      "epoch": 0.44564614034773575,
      "grad_norm": 2.1242969036102295,
      "learning_rate": 0.0001222730938247801,
      "loss": 1.8089,
      "step": 12444
    },
    {
      "epoch": 0.4456819524773041,
      "grad_norm": 1.5516866445541382,
      "learning_rate": 0.00012226178602975417,
      "loss": 1.4706,
      "step": 12445
    },
    {
      "epoch": 0.44571776460687235,
      "grad_norm": 2.02280855178833,
      "learning_rate": 0.00012225047793522462,
      "loss": 1.583,
      "step": 12446
    },
    {
      "epoch": 0.4457535767364406,
      "grad_norm": 1.7258055210113525,
      "learning_rate": 0.00012223916954134356,
      "loss": 1.6239,
      "step": 12447
    },
    {
      "epoch": 0.44578938886600894,
      "grad_norm": 1.9967955350875854,
      "learning_rate": 0.00012222786084826318,
      "loss": 1.5723,
      "step": 12448
    },
    {
      "epoch": 0.4458252009955772,
      "grad_norm": 1.7219719886779785,
      "learning_rate": 0.00012221655185613557,
      "loss": 1.376,
      "step": 12449
    },
    {
      "epoch": 0.4458610131251455,
      "grad_norm": 1.6591328382492065,
      "learning_rate": 0.00012220524256511297,
      "loss": 1.5033,
      "step": 12450
    },
    {
      "epoch": 0.44589682525471375,
      "grad_norm": 1.6007272005081177,
      "learning_rate": 0.00012219393297534744,
      "loss": 1.0919,
      "step": 12451
    },
    {
      "epoch": 0.44593263738428207,
      "grad_norm": 1.706794261932373,
      "learning_rate": 0.00012218262308699119,
      "loss": 1.319,
      "step": 12452
    },
    {
      "epoch": 0.44596844951385034,
      "grad_norm": 1.502685785293579,
      "learning_rate": 0.00012217131290019633,
      "loss": 1.5424,
      "step": 12453
    },
    {
      "epoch": 0.4460042616434186,
      "grad_norm": 1.5585651397705078,
      "learning_rate": 0.00012216000241511507,
      "loss": 1.5909,
      "step": 12454
    },
    {
      "epoch": 0.44604007377298693,
      "grad_norm": 3.2422244548797607,
      "learning_rate": 0.00012214869163189958,
      "loss": 1.6386,
      "step": 12455
    },
    {
      "epoch": 0.4460758859025552,
      "grad_norm": 1.7532734870910645,
      "learning_rate": 0.00012213738055070195,
      "loss": 1.59,
      "step": 12456
    },
    {
      "epoch": 0.44611169803212347,
      "grad_norm": 1.761208176612854,
      "learning_rate": 0.0001221260691716745,
      "loss": 1.3127,
      "step": 12457
    },
    {
      "epoch": 0.44614751016169174,
      "grad_norm": 1.8905987739562988,
      "learning_rate": 0.0001221147574949693,
      "loss": 1.4267,
      "step": 12458
    },
    {
      "epoch": 0.44618332229126006,
      "grad_norm": 1.5984662771224976,
      "learning_rate": 0.00012210344552073855,
      "loss": 1.8281,
      "step": 12459
    },
    {
      "epoch": 0.44621913442082833,
      "grad_norm": 1.3725858926773071,
      "learning_rate": 0.00012209213324913446,
      "loss": 1.4297,
      "step": 12460
    },
    {
      "epoch": 0.4462549465503966,
      "grad_norm": 2.176337242126465,
      "learning_rate": 0.00012208082068030924,
      "loss": 1.7024,
      "step": 12461
    },
    {
      "epoch": 0.4462907586799649,
      "grad_norm": 1.6479774713516235,
      "learning_rate": 0.00012206950781441502,
      "loss": 1.2077,
      "step": 12462
    },
    {
      "epoch": 0.4463265708095332,
      "grad_norm": 1.9320515394210815,
      "learning_rate": 0.00012205819465160407,
      "loss": 1.8261,
      "step": 12463
    },
    {
      "epoch": 0.44636238293910147,
      "grad_norm": 1.9457646608352661,
      "learning_rate": 0.00012204688119202852,
      "loss": 1.6547,
      "step": 12464
    },
    {
      "epoch": 0.44639819506866973,
      "grad_norm": 1.4203673601150513,
      "learning_rate": 0.00012203556743584063,
      "loss": 1.449,
      "step": 12465
    },
    {
      "epoch": 0.44643400719823806,
      "grad_norm": 1.9975597858428955,
      "learning_rate": 0.00012202425338319265,
      "loss": 1.7199,
      "step": 12466
    },
    {
      "epoch": 0.4464698193278063,
      "grad_norm": 1.3389594554901123,
      "learning_rate": 0.00012201293903423675,
      "loss": 1.3139,
      "step": 12467
    },
    {
      "epoch": 0.4465056314573746,
      "grad_norm": 1.865659475326538,
      "learning_rate": 0.00012200162438912512,
      "loss": 1.7206,
      "step": 12468
    },
    {
      "epoch": 0.4465414435869429,
      "grad_norm": 2.4966471195220947,
      "learning_rate": 0.00012199030944801,
      "loss": 1.8555,
      "step": 12469
    },
    {
      "epoch": 0.4465772557165112,
      "grad_norm": 1.775084137916565,
      "learning_rate": 0.00012197899421104367,
      "loss": 1.52,
      "step": 12470
    },
    {
      "epoch": 0.44661306784607946,
      "grad_norm": 1.4114207029342651,
      "learning_rate": 0.00012196767867837829,
      "loss": 1.531,
      "step": 12471
    },
    {
      "epoch": 0.44664887997564773,
      "grad_norm": 1.6351429224014282,
      "learning_rate": 0.00012195636285016614,
      "loss": 1.5006,
      "step": 12472
    },
    {
      "epoch": 0.44668469210521605,
      "grad_norm": 1.3841909170150757,
      "learning_rate": 0.00012194504672655944,
      "loss": 1.4175,
      "step": 12473
    },
    {
      "epoch": 0.4467205042347843,
      "grad_norm": 1.8590139150619507,
      "learning_rate": 0.00012193373030771046,
      "loss": 1.4761,
      "step": 12474
    },
    {
      "epoch": 0.4467563163643526,
      "grad_norm": 1.685275673866272,
      "learning_rate": 0.00012192241359377143,
      "loss": 1.5694,
      "step": 12475
    },
    {
      "epoch": 0.4467921284939209,
      "grad_norm": 1.7317240238189697,
      "learning_rate": 0.00012191109658489462,
      "loss": 1.3867,
      "step": 12476
    },
    {
      "epoch": 0.4468279406234892,
      "grad_norm": 2.1701266765594482,
      "learning_rate": 0.00012189977928123224,
      "loss": 1.6338,
      "step": 12477
    },
    {
      "epoch": 0.44686375275305745,
      "grad_norm": 1.709526538848877,
      "learning_rate": 0.0001218884616829366,
      "loss": 1.4318,
      "step": 12478
    },
    {
      "epoch": 0.4468995648826257,
      "grad_norm": 2.59258770942688,
      "learning_rate": 0.00012187714379015993,
      "loss": 1.8929,
      "step": 12479
    },
    {
      "epoch": 0.44693537701219405,
      "grad_norm": 1.6843186616897583,
      "learning_rate": 0.00012186582560305448,
      "loss": 1.4508,
      "step": 12480
    },
    {
      "epoch": 0.4469711891417623,
      "grad_norm": 1.3819479942321777,
      "learning_rate": 0.0001218545071217726,
      "loss": 1.5629,
      "step": 12481
    },
    {
      "epoch": 0.4470070012713306,
      "grad_norm": 2.069230079650879,
      "learning_rate": 0.00012184318834646648,
      "loss": 1.745,
      "step": 12482
    },
    {
      "epoch": 0.4470428134008989,
      "grad_norm": 1.6494131088256836,
      "learning_rate": 0.00012183186927728846,
      "loss": 1.4686,
      "step": 12483
    },
    {
      "epoch": 0.4470786255304672,
      "grad_norm": 1.2454302310943604,
      "learning_rate": 0.00012182054991439078,
      "loss": 1.2643,
      "step": 12484
    },
    {
      "epoch": 0.44711443766003545,
      "grad_norm": 1.628246784210205,
      "learning_rate": 0.00012180923025792579,
      "loss": 1.6646,
      "step": 12485
    },
    {
      "epoch": 0.4471502497896037,
      "grad_norm": 1.3497425317764282,
      "learning_rate": 0.00012179791030804573,
      "loss": 1.4701,
      "step": 12486
    },
    {
      "epoch": 0.44718606191917204,
      "grad_norm": 1.5856043100357056,
      "learning_rate": 0.00012178659006490285,
      "loss": 1.625,
      "step": 12487
    },
    {
      "epoch": 0.4472218740487403,
      "grad_norm": 2.0601372718811035,
      "learning_rate": 0.00012177526952864955,
      "loss": 1.6557,
      "step": 12488
    },
    {
      "epoch": 0.4472576861783086,
      "grad_norm": 1.5500800609588623,
      "learning_rate": 0.00012176394869943805,
      "loss": 1.7003,
      "step": 12489
    },
    {
      "epoch": 0.4472934983078769,
      "grad_norm": 1.6595675945281982,
      "learning_rate": 0.00012175262757742074,
      "loss": 1.5211,
      "step": 12490
    },
    {
      "epoch": 0.44732931043744517,
      "grad_norm": 1.8388330936431885,
      "learning_rate": 0.00012174130616274985,
      "loss": 1.5596,
      "step": 12491
    },
    {
      "epoch": 0.44736512256701344,
      "grad_norm": 1.7560348510742188,
      "learning_rate": 0.00012172998445557775,
      "loss": 1.4747,
      "step": 12492
    },
    {
      "epoch": 0.4474009346965817,
      "grad_norm": 2.180821180343628,
      "learning_rate": 0.00012171866245605671,
      "loss": 1.6482,
      "step": 12493
    },
    {
      "epoch": 0.44743674682615003,
      "grad_norm": 1.6616812944412231,
      "learning_rate": 0.00012170734016433914,
      "loss": 1.4381,
      "step": 12494
    },
    {
      "epoch": 0.4474725589557183,
      "grad_norm": 1.351104736328125,
      "learning_rate": 0.00012169601758057727,
      "loss": 1.6061,
      "step": 12495
    },
    {
      "epoch": 0.44750837108528657,
      "grad_norm": 1.8594951629638672,
      "learning_rate": 0.00012168469470492345,
      "loss": 1.5047,
      "step": 12496
    },
    {
      "epoch": 0.4475441832148549,
      "grad_norm": 1.516777515411377,
      "learning_rate": 0.00012167337153753007,
      "loss": 1.9187,
      "step": 12497
    },
    {
      "epoch": 0.44757999534442316,
      "grad_norm": 1.9803366661071777,
      "learning_rate": 0.00012166204807854942,
      "loss": 1.584,
      "step": 12498
    },
    {
      "epoch": 0.44761580747399143,
      "grad_norm": 1.7764346599578857,
      "learning_rate": 0.00012165072432813385,
      "loss": 1.388,
      "step": 12499
    },
    {
      "epoch": 0.4476516196035597,
      "grad_norm": 1.7678972482681274,
      "learning_rate": 0.0001216394002864357,
      "loss": 1.3775,
      "step": 12500
    },
    {
      "epoch": 0.447687431733128,
      "grad_norm": 1.4079352617263794,
      "learning_rate": 0.00012162807595360737,
      "loss": 1.5185,
      "step": 12501
    },
    {
      "epoch": 0.4477232438626963,
      "grad_norm": 1.933819055557251,
      "learning_rate": 0.00012161675132980114,
      "loss": 1.5765,
      "step": 12502
    },
    {
      "epoch": 0.44775905599226457,
      "grad_norm": 2.9456725120544434,
      "learning_rate": 0.00012160542641516945,
      "loss": 1.8331,
      "step": 12503
    },
    {
      "epoch": 0.4477948681218329,
      "grad_norm": 1.9133472442626953,
      "learning_rate": 0.00012159410120986456,
      "loss": 1.6421,
      "step": 12504
    },
    {
      "epoch": 0.44783068025140116,
      "grad_norm": 1.5416887998580933,
      "learning_rate": 0.00012158277571403893,
      "loss": 1.3045,
      "step": 12505
    },
    {
      "epoch": 0.4478664923809694,
      "grad_norm": 2.035646677017212,
      "learning_rate": 0.00012157144992784486,
      "loss": 1.4106,
      "step": 12506
    },
    {
      "epoch": 0.4479023045105377,
      "grad_norm": 1.2931156158447266,
      "learning_rate": 0.00012156012385143479,
      "loss": 1.7832,
      "step": 12507
    },
    {
      "epoch": 0.447938116640106,
      "grad_norm": 1.6115169525146484,
      "learning_rate": 0.00012154879748496104,
      "loss": 1.4598,
      "step": 12508
    },
    {
      "epoch": 0.4479739287696743,
      "grad_norm": 2.0151591300964355,
      "learning_rate": 0.00012153747082857601,
      "loss": 1.7758,
      "step": 12509
    },
    {
      "epoch": 0.44800974089924256,
      "grad_norm": 1.969767451286316,
      "learning_rate": 0.00012152614388243213,
      "loss": 1.692,
      "step": 12510
    },
    {
      "epoch": 0.4480455530288109,
      "grad_norm": 2.38447904586792,
      "learning_rate": 0.00012151481664668175,
      "loss": 1.5807,
      "step": 12511
    },
    {
      "epoch": 0.44808136515837915,
      "grad_norm": 1.623461127281189,
      "learning_rate": 0.00012150348912147723,
      "loss": 1.3877,
      "step": 12512
    },
    {
      "epoch": 0.4481171772879474,
      "grad_norm": 1.4375425577163696,
      "learning_rate": 0.000121492161306971,
      "loss": 1.5076,
      "step": 12513
    },
    {
      "epoch": 0.4481529894175157,
      "grad_norm": 1.580809473991394,
      "learning_rate": 0.00012148083320331549,
      "loss": 1.5652,
      "step": 12514
    },
    {
      "epoch": 0.448188801547084,
      "grad_norm": 2.1061019897460938,
      "learning_rate": 0.00012146950481066304,
      "loss": 1.6577,
      "step": 12515
    },
    {
      "epoch": 0.4482246136766523,
      "grad_norm": 1.5830954313278198,
      "learning_rate": 0.00012145817612916612,
      "loss": 1.5543,
      "step": 12516
    },
    {
      "epoch": 0.44826042580622055,
      "grad_norm": 1.8224624395370483,
      "learning_rate": 0.00012144684715897711,
      "loss": 1.2341,
      "step": 12517
    },
    {
      "epoch": 0.4482962379357889,
      "grad_norm": 1.5227636098861694,
      "learning_rate": 0.00012143551790024848,
      "loss": 1.641,
      "step": 12518
    },
    {
      "epoch": 0.44833205006535715,
      "grad_norm": 1.8081822395324707,
      "learning_rate": 0.00012142418835313254,
      "loss": 1.3633,
      "step": 12519
    },
    {
      "epoch": 0.4483678621949254,
      "grad_norm": 1.5999212265014648,
      "learning_rate": 0.00012141285851778183,
      "loss": 1.444,
      "step": 12520
    },
    {
      "epoch": 0.4484036743244937,
      "grad_norm": 2.473132371902466,
      "learning_rate": 0.00012140152839434873,
      "loss": 1.7155,
      "step": 12521
    },
    {
      "epoch": 0.448439486454062,
      "grad_norm": 2.2659380435943604,
      "learning_rate": 0.00012139019798298563,
      "loss": 1.7137,
      "step": 12522
    },
    {
      "epoch": 0.4484752985836303,
      "grad_norm": 2.1838572025299072,
      "learning_rate": 0.00012137886728384504,
      "loss": 1.5565,
      "step": 12523
    },
    {
      "epoch": 0.44851111071319855,
      "grad_norm": 1.3856747150421143,
      "learning_rate": 0.00012136753629707936,
      "loss": 1.5583,
      "step": 12524
    },
    {
      "epoch": 0.44854692284276687,
      "grad_norm": 1.9057285785675049,
      "learning_rate": 0.00012135620502284104,
      "loss": 1.5314,
      "step": 12525
    },
    {
      "epoch": 0.44858273497233514,
      "grad_norm": 1.7440130710601807,
      "learning_rate": 0.00012134487346128252,
      "loss": 1.6575,
      "step": 12526
    },
    {
      "epoch": 0.4486185471019034,
      "grad_norm": 2.055412769317627,
      "learning_rate": 0.00012133354161255628,
      "loss": 1.4412,
      "step": 12527
    },
    {
      "epoch": 0.4486543592314717,
      "grad_norm": 1.5365710258483887,
      "learning_rate": 0.00012132220947681472,
      "loss": 1.2439,
      "step": 12528
    },
    {
      "epoch": 0.44869017136104,
      "grad_norm": 1.7394073009490967,
      "learning_rate": 0.00012131087705421036,
      "loss": 1.5784,
      "step": 12529
    },
    {
      "epoch": 0.44872598349060827,
      "grad_norm": 1.6387853622436523,
      "learning_rate": 0.00012129954434489566,
      "loss": 1.4735,
      "step": 12530
    },
    {
      "epoch": 0.44876179562017654,
      "grad_norm": 1.3802781105041504,
      "learning_rate": 0.00012128821134902302,
      "loss": 1.4855,
      "step": 12531
    },
    {
      "epoch": 0.44879760774974486,
      "grad_norm": 2.449227809906006,
      "learning_rate": 0.00012127687806674499,
      "loss": 1.7276,
      "step": 12532
    },
    {
      "epoch": 0.44883341987931313,
      "grad_norm": 1.6007570028305054,
      "learning_rate": 0.00012126554449821399,
      "loss": 1.549,
      "step": 12533
    },
    {
      "epoch": 0.4488692320088814,
      "grad_norm": 1.3325066566467285,
      "learning_rate": 0.00012125421064358253,
      "loss": 1.3366,
      "step": 12534
    },
    {
      "epoch": 0.44890504413844967,
      "grad_norm": 1.4373480081558228,
      "learning_rate": 0.00012124287650300307,
      "loss": 1.6836,
      "step": 12535
    },
    {
      "epoch": 0.448940856268018,
      "grad_norm": 2.2037734985351562,
      "learning_rate": 0.00012123154207662815,
      "loss": 1.4235,
      "step": 12536
    },
    {
      "epoch": 0.44897666839758626,
      "grad_norm": 1.5827618837356567,
      "learning_rate": 0.00012122020736461018,
      "loss": 1.5252,
      "step": 12537
    },
    {
      "epoch": 0.44901248052715453,
      "grad_norm": 1.817306399345398,
      "learning_rate": 0.0001212088723671017,
      "loss": 1.6918,
      "step": 12538
    },
    {
      "epoch": 0.44904829265672286,
      "grad_norm": 1.913685917854309,
      "learning_rate": 0.0001211975370842552,
      "loss": 1.6136,
      "step": 12539
    },
    {
      "epoch": 0.4490841047862911,
      "grad_norm": 2.0694994926452637,
      "learning_rate": 0.00012118620151622317,
      "loss": 1.3688,
      "step": 12540
    },
    {
      "epoch": 0.4491199169158594,
      "grad_norm": 1.696552038192749,
      "learning_rate": 0.00012117486566315814,
      "loss": 1.4033,
      "step": 12541
    },
    {
      "epoch": 0.44915572904542767,
      "grad_norm": 1.6482453346252441,
      "learning_rate": 0.0001211635295252126,
      "loss": 1.4401,
      "step": 12542
    },
    {
      "epoch": 0.449191541174996,
      "grad_norm": 1.2823840379714966,
      "learning_rate": 0.0001211521931025391,
      "loss": 1.4209,
      "step": 12543
    },
    {
      "epoch": 0.44922735330456426,
      "grad_norm": 1.704340934753418,
      "learning_rate": 0.00012114085639529007,
      "loss": 1.5546,
      "step": 12544
    },
    {
      "epoch": 0.4492631654341325,
      "grad_norm": 1.6278151273727417,
      "learning_rate": 0.00012112951940361812,
      "loss": 1.6255,
      "step": 12545
    },
    {
      "epoch": 0.44929897756370085,
      "grad_norm": 2.9371583461761475,
      "learning_rate": 0.00012111818212767572,
      "loss": 1.8694,
      "step": 12546
    },
    {
      "epoch": 0.4493347896932691,
      "grad_norm": 1.6199440956115723,
      "learning_rate": 0.00012110684456761547,
      "loss": 1.1853,
      "step": 12547
    },
    {
      "epoch": 0.4493706018228374,
      "grad_norm": 1.508470892906189,
      "learning_rate": 0.0001210955067235898,
      "loss": 1.4939,
      "step": 12548
    },
    {
      "epoch": 0.44940641395240566,
      "grad_norm": 1.4398698806762695,
      "learning_rate": 0.00012108416859575131,
      "loss": 1.5191,
      "step": 12549
    },
    {
      "epoch": 0.449442226081974,
      "grad_norm": 1.856343150138855,
      "learning_rate": 0.00012107283018425256,
      "loss": 1.8702,
      "step": 12550
    },
    {
      "epoch": 0.44947803821154225,
      "grad_norm": 1.3848611116409302,
      "learning_rate": 0.00012106149148924602,
      "loss": 1.6158,
      "step": 12551
    },
    {
      "epoch": 0.4495138503411105,
      "grad_norm": 1.683535099029541,
      "learning_rate": 0.0001210501525108843,
      "loss": 1.8212,
      "step": 12552
    },
    {
      "epoch": 0.44954966247067885,
      "grad_norm": 1.8491222858428955,
      "learning_rate": 0.00012103881324931991,
      "loss": 1.3237,
      "step": 12553
    },
    {
      "epoch": 0.4495854746002471,
      "grad_norm": 1.5038716793060303,
      "learning_rate": 0.00012102747370470546,
      "loss": 1.2653,
      "step": 12554
    },
    {
      "epoch": 0.4496212867298154,
      "grad_norm": 1.5144926309585571,
      "learning_rate": 0.00012101613387719348,
      "loss": 1.6953,
      "step": 12555
    },
    {
      "epoch": 0.44965709885938365,
      "grad_norm": 2.2025644779205322,
      "learning_rate": 0.00012100479376693652,
      "loss": 1.5872,
      "step": 12556
    },
    {
      "epoch": 0.449692910988952,
      "grad_norm": 1.7867844104766846,
      "learning_rate": 0.00012099345337408712,
      "loss": 1.371,
      "step": 12557
    },
    {
      "epoch": 0.44972872311852025,
      "grad_norm": 1.6688547134399414,
      "learning_rate": 0.00012098211269879791,
      "loss": 1.3092,
      "step": 12558
    },
    {
      "epoch": 0.4497645352480885,
      "grad_norm": 1.5913275480270386,
      "learning_rate": 0.00012097077174122143,
      "loss": 1.6201,
      "step": 12559
    },
    {
      "epoch": 0.44980034737765684,
      "grad_norm": 1.4191871881484985,
      "learning_rate": 0.00012095943050151026,
      "loss": 1.5449,
      "step": 12560
    },
    {
      "epoch": 0.4498361595072251,
      "grad_norm": 1.5668357610702515,
      "learning_rate": 0.000120948088979817,
      "loss": 1.2965,
      "step": 12561
    },
    {
      "epoch": 0.4498719716367934,
      "grad_norm": 1.4503635168075562,
      "learning_rate": 0.00012093674717629419,
      "loss": 1.5289,
      "step": 12562
    },
    {
      "epoch": 0.44990778376636165,
      "grad_norm": 1.6835050582885742,
      "learning_rate": 0.00012092540509109451,
      "loss": 1.5644,
      "step": 12563
    },
    {
      "epoch": 0.44994359589592997,
      "grad_norm": 1.895340919494629,
      "learning_rate": 0.00012091406272437049,
      "loss": 1.3566,
      "step": 12564
    },
    {
      "epoch": 0.44997940802549824,
      "grad_norm": 1.6163996458053589,
      "learning_rate": 0.00012090272007627472,
      "loss": 1.5901,
      "step": 12565
    },
    {
      "epoch": 0.4500152201550665,
      "grad_norm": 1.5837438106536865,
      "learning_rate": 0.0001208913771469598,
      "loss": 1.3175,
      "step": 12566
    },
    {
      "epoch": 0.45005103228463483,
      "grad_norm": 1.9061717987060547,
      "learning_rate": 0.00012088003393657837,
      "loss": 1.4182,
      "step": 12567
    },
    {
      "epoch": 0.4500868444142031,
      "grad_norm": 1.5280214548110962,
      "learning_rate": 0.00012086869044528297,
      "loss": 1.5659,
      "step": 12568
    },
    {
      "epoch": 0.45012265654377137,
      "grad_norm": 2.357043743133545,
      "learning_rate": 0.0001208573466732263,
      "loss": 1.5281,
      "step": 12569
    },
    {
      "epoch": 0.45015846867333964,
      "grad_norm": 1.4980483055114746,
      "learning_rate": 0.00012084600262056094,
      "loss": 1.6226,
      "step": 12570
    },
    {
      "epoch": 0.45019428080290796,
      "grad_norm": 1.7290112972259521,
      "learning_rate": 0.0001208346582874395,
      "loss": 1.5311,
      "step": 12571
    },
    {
      "epoch": 0.45023009293247623,
      "grad_norm": 1.7056915760040283,
      "learning_rate": 0.00012082331367401458,
      "loss": 1.532,
      "step": 12572
    },
    {
      "epoch": 0.4502659050620445,
      "grad_norm": 2.0714375972747803,
      "learning_rate": 0.00012081196878043885,
      "loss": 1.2363,
      "step": 12573
    },
    {
      "epoch": 0.4503017171916128,
      "grad_norm": 1.90829598903656,
      "learning_rate": 0.00012080062360686495,
      "loss": 1.5991,
      "step": 12574
    },
    {
      "epoch": 0.4503375293211811,
      "grad_norm": 1.3349385261535645,
      "learning_rate": 0.00012078927815344545,
      "loss": 1.2521,
      "step": 12575
    },
    {
      "epoch": 0.45037334145074936,
      "grad_norm": 1.827406883239746,
      "learning_rate": 0.00012077793242033307,
      "loss": 1.6133,
      "step": 12576
    },
    {
      "epoch": 0.45040915358031763,
      "grad_norm": 1.918264627456665,
      "learning_rate": 0.00012076658640768036,
      "loss": 1.5396,
      "step": 12577
    },
    {
      "epoch": 0.45044496570988596,
      "grad_norm": 1.7360564470291138,
      "learning_rate": 0.00012075524011564005,
      "loss": 1.3884,
      "step": 12578
    },
    {
      "epoch": 0.4504807778394542,
      "grad_norm": 1.4378196001052856,
      "learning_rate": 0.00012074389354436475,
      "loss": 1.5432,
      "step": 12579
    },
    {
      "epoch": 0.4505165899690225,
      "grad_norm": 1.7006964683532715,
      "learning_rate": 0.00012073254669400713,
      "loss": 1.5753,
      "step": 12580
    },
    {
      "epoch": 0.45055240209859077,
      "grad_norm": 1.619866132736206,
      "learning_rate": 0.00012072119956471981,
      "loss": 1.5881,
      "step": 12581
    },
    {
      "epoch": 0.4505882142281591,
      "grad_norm": 1.2958283424377441,
      "learning_rate": 0.00012070985215665551,
      "loss": 1.6152,
      "step": 12582
    },
    {
      "epoch": 0.45062402635772736,
      "grad_norm": 2.0520436763763428,
      "learning_rate": 0.00012069850446996686,
      "loss": 1.7036,
      "step": 12583
    },
    {
      "epoch": 0.4506598384872956,
      "grad_norm": 1.5955454111099243,
      "learning_rate": 0.00012068715650480653,
      "loss": 1.4784,
      "step": 12584
    },
    {
      "epoch": 0.45069565061686395,
      "grad_norm": 2.244309425354004,
      "learning_rate": 0.00012067580826132718,
      "loss": 2.0482,
      "step": 12585
    },
    {
      "epoch": 0.4507314627464322,
      "grad_norm": 1.449869155883789,
      "learning_rate": 0.0001206644597396815,
      "loss": 1.4782,
      "step": 12586
    },
    {
      "epoch": 0.4507672748760005,
      "grad_norm": 1.647153615951538,
      "learning_rate": 0.00012065311094002218,
      "loss": 1.2063,
      "step": 12587
    },
    {
      "epoch": 0.45080308700556876,
      "grad_norm": 1.6689027547836304,
      "learning_rate": 0.00012064176186250189,
      "loss": 1.7451,
      "step": 12588
    },
    {
      "epoch": 0.4508388991351371,
      "grad_norm": 1.4815447330474854,
      "learning_rate": 0.00012063041250727331,
      "loss": 1.3996,
      "step": 12589
    },
    {
      "epoch": 0.45087471126470535,
      "grad_norm": 1.4417043924331665,
      "learning_rate": 0.00012061906287448914,
      "loss": 1.5646,
      "step": 12590
    },
    {
      "epoch": 0.4509105233942736,
      "grad_norm": 2.2002885341644287,
      "learning_rate": 0.00012060771296430209,
      "loss": 1.3061,
      "step": 12591
    },
    {
      "epoch": 0.45094633552384195,
      "grad_norm": 1.4272916316986084,
      "learning_rate": 0.00012059636277686486,
      "loss": 1.0158,
      "step": 12592
    },
    {
      "epoch": 0.4509821476534102,
      "grad_norm": 1.6139081716537476,
      "learning_rate": 0.00012058501231233011,
      "loss": 1.4949,
      "step": 12593
    },
    {
      "epoch": 0.4510179597829785,
      "grad_norm": 1.6155054569244385,
      "learning_rate": 0.00012057366157085058,
      "loss": 1.3464,
      "step": 12594
    },
    {
      "epoch": 0.45105377191254675,
      "grad_norm": 1.561390995979309,
      "learning_rate": 0.00012056231055257896,
      "loss": 1.3473,
      "step": 12595
    },
    {
      "epoch": 0.4510895840421151,
      "grad_norm": 1.356062650680542,
      "learning_rate": 0.000120550959257668,
      "loss": 1.3973,
      "step": 12596
    },
    {
      "epoch": 0.45112539617168335,
      "grad_norm": 1.578811526298523,
      "learning_rate": 0.00012053960768627036,
      "loss": 1.5595,
      "step": 12597
    },
    {
      "epoch": 0.4511612083012516,
      "grad_norm": 2.086986780166626,
      "learning_rate": 0.00012052825583853881,
      "loss": 1.3671,
      "step": 12598
    },
    {
      "epoch": 0.45119702043081994,
      "grad_norm": 2.0918164253234863,
      "learning_rate": 0.00012051690371462608,
      "loss": 1.1172,
      "step": 12599
    },
    {
      "epoch": 0.4512328325603882,
      "grad_norm": 1.4624220132827759,
      "learning_rate": 0.00012050555131468484,
      "loss": 1.4732,
      "step": 12600
    },
    {
      "epoch": 0.4512686446899565,
      "grad_norm": 1.6903061866760254,
      "learning_rate": 0.00012049419863886786,
      "loss": 1.6673,
      "step": 12601
    },
    {
      "epoch": 0.45130445681952475,
      "grad_norm": 1.5074726343154907,
      "learning_rate": 0.00012048284568732791,
      "loss": 2.0319,
      "step": 12602
    },
    {
      "epoch": 0.45134026894909307,
      "grad_norm": 2.3268983364105225,
      "learning_rate": 0.00012047149246021763,
      "loss": 1.5876,
      "step": 12603
    },
    {
      "epoch": 0.45137608107866134,
      "grad_norm": 1.8839945793151855,
      "learning_rate": 0.00012046013895768986,
      "loss": 1.5132,
      "step": 12604
    },
    {
      "epoch": 0.4514118932082296,
      "grad_norm": 1.4591853618621826,
      "learning_rate": 0.0001204487851798973,
      "loss": 1.6994,
      "step": 12605
    },
    {
      "epoch": 0.45144770533779793,
      "grad_norm": 1.9590816497802734,
      "learning_rate": 0.00012043743112699273,
      "loss": 1.2946,
      "step": 12606
    },
    {
      "epoch": 0.4514835174673662,
      "grad_norm": 1.7824480533599854,
      "learning_rate": 0.0001204260767991289,
      "loss": 1.4297,
      "step": 12607
    },
    {
      "epoch": 0.45151932959693447,
      "grad_norm": 2.672687292098999,
      "learning_rate": 0.00012041472219645854,
      "loss": 1.4044,
      "step": 12608
    },
    {
      "epoch": 0.45155514172650274,
      "grad_norm": 2.600595474243164,
      "learning_rate": 0.00012040336731913442,
      "loss": 1.8836,
      "step": 12609
    },
    {
      "epoch": 0.45159095385607106,
      "grad_norm": 1.701601266860962,
      "learning_rate": 0.00012039201216730931,
      "loss": 1.4913,
      "step": 12610
    },
    {
      "epoch": 0.45162676598563933,
      "grad_norm": 2.1332130432128906,
      "learning_rate": 0.00012038065674113598,
      "loss": 1.8227,
      "step": 12611
    },
    {
      "epoch": 0.4516625781152076,
      "grad_norm": 1.9110596179962158,
      "learning_rate": 0.0001203693010407672,
      "loss": 1.6782,
      "step": 12612
    },
    {
      "epoch": 0.4516983902447759,
      "grad_norm": 1.5340908765792847,
      "learning_rate": 0.00012035794506635575,
      "loss": 1.5475,
      "step": 12613
    },
    {
      "epoch": 0.4517342023743442,
      "grad_norm": 1.7282644510269165,
      "learning_rate": 0.0001203465888180544,
      "loss": 1.8253,
      "step": 12614
    },
    {
      "epoch": 0.45177001450391246,
      "grad_norm": 2.7363064289093018,
      "learning_rate": 0.00012033523229601598,
      "loss": 1.4577,
      "step": 12615
    },
    {
      "epoch": 0.45180582663348073,
      "grad_norm": 1.7300324440002441,
      "learning_rate": 0.00012032387550039319,
      "loss": 1.6123,
      "step": 12616
    },
    {
      "epoch": 0.45184163876304906,
      "grad_norm": 1.5794569253921509,
      "learning_rate": 0.00012031251843133891,
      "loss": 1.5446,
      "step": 12617
    },
    {
      "epoch": 0.4518774508926173,
      "grad_norm": 2.11045241355896,
      "learning_rate": 0.00012030116108900589,
      "loss": 1.5784,
      "step": 12618
    },
    {
      "epoch": 0.4519132630221856,
      "grad_norm": 1.8082817792892456,
      "learning_rate": 0.00012028980347354692,
      "loss": 1.3738,
      "step": 12619
    },
    {
      "epoch": 0.4519490751517539,
      "grad_norm": 1.3321770429611206,
      "learning_rate": 0.00012027844558511483,
      "loss": 1.6105,
      "step": 12620
    },
    {
      "epoch": 0.4519848872813222,
      "grad_norm": 1.5995198488235474,
      "learning_rate": 0.00012026708742386239,
      "loss": 1.3222,
      "step": 12621
    },
    {
      "epoch": 0.45202069941089046,
      "grad_norm": 2.029466390609741,
      "learning_rate": 0.00012025572898994246,
      "loss": 1.4356,
      "step": 12622
    },
    {
      "epoch": 0.4520565115404587,
      "grad_norm": 1.4476046562194824,
      "learning_rate": 0.00012024437028350779,
      "loss": 1.5459,
      "step": 12623
    },
    {
      "epoch": 0.45209232367002705,
      "grad_norm": 1.6885541677474976,
      "learning_rate": 0.00012023301130471128,
      "loss": 1.4401,
      "step": 12624
    },
    {
      "epoch": 0.4521281357995953,
      "grad_norm": 1.6791876554489136,
      "learning_rate": 0.00012022165205370565,
      "loss": 1.6886,
      "step": 12625
    },
    {
      "epoch": 0.4521639479291636,
      "grad_norm": 1.5566707849502563,
      "learning_rate": 0.00012021029253064382,
      "loss": 1.6669,
      "step": 12626
    },
    {
      "epoch": 0.4521997600587319,
      "grad_norm": 1.7910056114196777,
      "learning_rate": 0.00012019893273567855,
      "loss": 1.3655,
      "step": 12627
    },
    {
      "epoch": 0.4522355721883002,
      "grad_norm": 1.7929489612579346,
      "learning_rate": 0.00012018757266896267,
      "loss": 1.563,
      "step": 12628
    },
    {
      "epoch": 0.45227138431786845,
      "grad_norm": 2.385453701019287,
      "learning_rate": 0.00012017621233064908,
      "loss": 2.0782,
      "step": 12629
    },
    {
      "epoch": 0.4523071964474367,
      "grad_norm": 1.5448579788208008,
      "learning_rate": 0.00012016485172089056,
      "loss": 1.3892,
      "step": 12630
    },
    {
      "epoch": 0.45234300857700505,
      "grad_norm": 2.077671527862549,
      "learning_rate": 0.00012015349083983998,
      "loss": 1.3285,
      "step": 12631
    },
    {
      "epoch": 0.4523788207065733,
      "grad_norm": 1.3240525722503662,
      "learning_rate": 0.00012014212968765018,
      "loss": 1.5076,
      "step": 12632
    },
    {
      "epoch": 0.4524146328361416,
      "grad_norm": 2.3672759532928467,
      "learning_rate": 0.000120130768264474,
      "loss": 1.4313,
      "step": 12633
    },
    {
      "epoch": 0.4524504449657099,
      "grad_norm": 1.4926321506500244,
      "learning_rate": 0.00012011940657046427,
      "loss": 1.6316,
      "step": 12634
    },
    {
      "epoch": 0.4524862570952782,
      "grad_norm": 1.6162370443344116,
      "learning_rate": 0.00012010804460577395,
      "loss": 1.5586,
      "step": 12635
    },
    {
      "epoch": 0.45252206922484645,
      "grad_norm": 1.8206548690795898,
      "learning_rate": 0.00012009668237055578,
      "loss": 1.8584,
      "step": 12636
    },
    {
      "epoch": 0.4525578813544147,
      "grad_norm": 1.6547601222991943,
      "learning_rate": 0.00012008531986496266,
      "loss": 1.4219,
      "step": 12637
    },
    {
      "epoch": 0.45259369348398304,
      "grad_norm": 1.5482653379440308,
      "learning_rate": 0.0001200739570891475,
      "loss": 1.4394,
      "step": 12638
    },
    {
      "epoch": 0.4526295056135513,
      "grad_norm": 1.3440611362457275,
      "learning_rate": 0.0001200625940432631,
      "loss": 1.211,
      "step": 12639
    },
    {
      "epoch": 0.4526653177431196,
      "grad_norm": 1.2386219501495361,
      "learning_rate": 0.00012005123072746242,
      "loss": 1.4767,
      "step": 12640
    },
    {
      "epoch": 0.4527011298726879,
      "grad_norm": 2.0508835315704346,
      "learning_rate": 0.00012003986714189825,
      "loss": 1.305,
      "step": 12641
    },
    {
      "epoch": 0.45273694200225617,
      "grad_norm": 1.6780390739440918,
      "learning_rate": 0.00012002850328672357,
      "loss": 1.7344,
      "step": 12642
    },
    {
      "epoch": 0.45277275413182444,
      "grad_norm": 1.7541413307189941,
      "learning_rate": 0.00012001713916209117,
      "loss": 1.2988,
      "step": 12643
    },
    {
      "epoch": 0.4528085662613927,
      "grad_norm": 4.736975193023682,
      "learning_rate": 0.00012000577476815402,
      "loss": 1.3954,
      "step": 12644
    },
    {
      "epoch": 0.45284437839096103,
      "grad_norm": 1.9038360118865967,
      "learning_rate": 0.00011999441010506496,
      "loss": 1.5603,
      "step": 12645
    },
    {
      "epoch": 0.4528801905205293,
      "grad_norm": 1.3158032894134521,
      "learning_rate": 0.00011998304517297687,
      "loss": 1.3335,
      "step": 12646
    },
    {
      "epoch": 0.45291600265009757,
      "grad_norm": 1.6054844856262207,
      "learning_rate": 0.00011997167997204272,
      "loss": 1.5157,
      "step": 12647
    },
    {
      "epoch": 0.4529518147796659,
      "grad_norm": 2.2230138778686523,
      "learning_rate": 0.00011996031450241536,
      "loss": 1.4866,
      "step": 12648
    },
    {
      "epoch": 0.45298762690923416,
      "grad_norm": 2.4321999549865723,
      "learning_rate": 0.00011994894876424773,
      "loss": 1.7621,
      "step": 12649
    },
    {
      "epoch": 0.45302343903880243,
      "grad_norm": 1.6048284769058228,
      "learning_rate": 0.00011993758275769273,
      "loss": 1.4613,
      "step": 12650
    },
    {
      "epoch": 0.4530592511683707,
      "grad_norm": 1.7765322923660278,
      "learning_rate": 0.00011992621648290328,
      "loss": 1.4474,
      "step": 12651
    },
    {
      "epoch": 0.453095063297939,
      "grad_norm": 1.6691166162490845,
      "learning_rate": 0.00011991484994003226,
      "loss": 1.5177,
      "step": 12652
    },
    {
      "epoch": 0.4531308754275073,
      "grad_norm": 2.022249698638916,
      "learning_rate": 0.00011990348312923266,
      "loss": 1.697,
      "step": 12653
    },
    {
      "epoch": 0.45316668755707556,
      "grad_norm": 2.6226868629455566,
      "learning_rate": 0.00011989211605065733,
      "loss": 1.6507,
      "step": 12654
    },
    {
      "epoch": 0.4532024996866439,
      "grad_norm": 2.3299243450164795,
      "learning_rate": 0.00011988074870445927,
      "loss": 1.8215,
      "step": 12655
    },
    {
      "epoch": 0.45323831181621216,
      "grad_norm": 1.7379266023635864,
      "learning_rate": 0.00011986938109079133,
      "loss": 1.4798,
      "step": 12656
    },
    {
      "epoch": 0.4532741239457804,
      "grad_norm": 1.6245919466018677,
      "learning_rate": 0.00011985801320980654,
      "loss": 1.6791,
      "step": 12657
    },
    {
      "epoch": 0.4533099360753487,
      "grad_norm": 1.2778525352478027,
      "learning_rate": 0.00011984664506165777,
      "loss": 1.4197,
      "step": 12658
    },
    {
      "epoch": 0.453345748204917,
      "grad_norm": 1.747637152671814,
      "learning_rate": 0.00011983527664649801,
      "loss": 1.6668,
      "step": 12659
    },
    {
      "epoch": 0.4533815603344853,
      "grad_norm": 2.747385025024414,
      "learning_rate": 0.0001198239079644802,
      "loss": 1.4045,
      "step": 12660
    },
    {
      "epoch": 0.45341737246405356,
      "grad_norm": 2.079819917678833,
      "learning_rate": 0.00011981253901575726,
      "loss": 1.73,
      "step": 12661
    },
    {
      "epoch": 0.4534531845936219,
      "grad_norm": 1.518641471862793,
      "learning_rate": 0.00011980116980048217,
      "loss": 1.6875,
      "step": 12662
    },
    {
      "epoch": 0.45348899672319015,
      "grad_norm": 1.9608591794967651,
      "learning_rate": 0.00011978980031880789,
      "loss": 1.85,
      "step": 12663
    },
    {
      "epoch": 0.4535248088527584,
      "grad_norm": 1.6720517873764038,
      "learning_rate": 0.00011977843057088735,
      "loss": 1.7237,
      "step": 12664
    },
    {
      "epoch": 0.4535606209823267,
      "grad_norm": 2.0230205059051514,
      "learning_rate": 0.00011976706055687357,
      "loss": 1.5775,
      "step": 12665
    },
    {
      "epoch": 0.453596433111895,
      "grad_norm": 1.5287965536117554,
      "learning_rate": 0.00011975569027691947,
      "loss": 1.6407,
      "step": 12666
    },
    {
      "epoch": 0.4536322452414633,
      "grad_norm": 1.822375774383545,
      "learning_rate": 0.00011974431973117804,
      "loss": 1.761,
      "step": 12667
    },
    {
      "epoch": 0.45366805737103155,
      "grad_norm": 1.4832029342651367,
      "learning_rate": 0.00011973294891980224,
      "loss": 1.5349,
      "step": 12668
    },
    {
      "epoch": 0.4537038695005999,
      "grad_norm": 1.5550483465194702,
      "learning_rate": 0.00011972157784294508,
      "loss": 1.4935,
      "step": 12669
    },
    {
      "epoch": 0.45373968163016815,
      "grad_norm": 1.5525519847869873,
      "learning_rate": 0.00011971020650075954,
      "loss": 1.2612,
      "step": 12670
    },
    {
      "epoch": 0.4537754937597364,
      "grad_norm": 1.824698805809021,
      "learning_rate": 0.00011969883489339862,
      "loss": 1.3504,
      "step": 12671
    },
    {
      "epoch": 0.4538113058893047,
      "grad_norm": 1.8953741788864136,
      "learning_rate": 0.00011968746302101523,
      "loss": 1.4381,
      "step": 12672
    },
    {
      "epoch": 0.453847118018873,
      "grad_norm": 1.4722163677215576,
      "learning_rate": 0.00011967609088376245,
      "loss": 1.3717,
      "step": 12673
    },
    {
      "epoch": 0.4538829301484413,
      "grad_norm": 1.3879319429397583,
      "learning_rate": 0.00011966471848179324,
      "loss": 1.4582,
      "step": 12674
    },
    {
      "epoch": 0.45391874227800955,
      "grad_norm": 2.1239492893218994,
      "learning_rate": 0.00011965334581526062,
      "loss": 1.3412,
      "step": 12675
    },
    {
      "epoch": 0.45395455440757787,
      "grad_norm": 1.5013140439987183,
      "learning_rate": 0.00011964197288431756,
      "loss": 1.3925,
      "step": 12676
    },
    {
      "epoch": 0.45399036653714614,
      "grad_norm": 1.629763126373291,
      "learning_rate": 0.00011963059968911712,
      "loss": 1.7531,
      "step": 12677
    },
    {
      "epoch": 0.4540261786667144,
      "grad_norm": 2.189053773880005,
      "learning_rate": 0.00011961922622981225,
      "loss": 1.5956,
      "step": 12678
    },
    {
      "epoch": 0.4540619907962827,
      "grad_norm": 1.6427063941955566,
      "learning_rate": 0.00011960785250655604,
      "loss": 1.6511,
      "step": 12679
    },
    {
      "epoch": 0.454097802925851,
      "grad_norm": 1.7626343965530396,
      "learning_rate": 0.00011959647851950145,
      "loss": 1.4323,
      "step": 12680
    },
    {
      "epoch": 0.45413361505541927,
      "grad_norm": 1.5623730421066284,
      "learning_rate": 0.0001195851042688015,
      "loss": 1.7968,
      "step": 12681
    },
    {
      "epoch": 0.45416942718498754,
      "grad_norm": 1.481370210647583,
      "learning_rate": 0.00011957372975460925,
      "loss": 1.5173,
      "step": 12682
    },
    {
      "epoch": 0.45420523931455586,
      "grad_norm": 1.105037808418274,
      "learning_rate": 0.00011956235497707771,
      "loss": 1.403,
      "step": 12683
    },
    {
      "epoch": 0.45424105144412413,
      "grad_norm": 1.2726575136184692,
      "learning_rate": 0.00011955097993635991,
      "loss": 1.0945,
      "step": 12684
    },
    {
      "epoch": 0.4542768635736924,
      "grad_norm": 1.2600315809249878,
      "learning_rate": 0.0001195396046326089,
      "loss": 1.557,
      "step": 12685
    },
    {
      "epoch": 0.45431267570326067,
      "grad_norm": 1.5838819742202759,
      "learning_rate": 0.00011952822906597773,
      "loss": 1.4059,
      "step": 12686
    },
    {
      "epoch": 0.454348487832829,
      "grad_norm": 1.760190725326538,
      "learning_rate": 0.0001195168532366194,
      "loss": 1.3405,
      "step": 12687
    },
    {
      "epoch": 0.45438429996239726,
      "grad_norm": 2.3661205768585205,
      "learning_rate": 0.000119505477144687,
      "loss": 1.6014,
      "step": 12688
    },
    {
      "epoch": 0.45442011209196553,
      "grad_norm": 1.6117066144943237,
      "learning_rate": 0.00011949410079033359,
      "loss": 1.4842,
      "step": 12689
    },
    {
      "epoch": 0.45445592422153386,
      "grad_norm": 1.467039942741394,
      "learning_rate": 0.00011948272417371216,
      "loss": 1.5621,
      "step": 12690
    },
    {
      "epoch": 0.4544917363511021,
      "grad_norm": 1.4559403657913208,
      "learning_rate": 0.00011947134729497583,
      "loss": 1.6804,
      "step": 12691
    },
    {
      "epoch": 0.4545275484806704,
      "grad_norm": 1.5294259786605835,
      "learning_rate": 0.00011945997015427761,
      "loss": 1.5,
      "step": 12692
    },
    {
      "epoch": 0.45456336061023866,
      "grad_norm": 1.8137285709381104,
      "learning_rate": 0.00011944859275177063,
      "loss": 1.5028,
      "step": 12693
    },
    {
      "epoch": 0.454599172739807,
      "grad_norm": 1.7047687768936157,
      "learning_rate": 0.00011943721508760788,
      "loss": 1.5601,
      "step": 12694
    },
    {
      "epoch": 0.45463498486937526,
      "grad_norm": 1.5781142711639404,
      "learning_rate": 0.00011942583716194251,
      "loss": 1.5172,
      "step": 12695
    },
    {
      "epoch": 0.4546707969989435,
      "grad_norm": 1.6754029989242554,
      "learning_rate": 0.00011941445897492755,
      "loss": 1.654,
      "step": 12696
    },
    {
      "epoch": 0.45470660912851185,
      "grad_norm": 1.5918335914611816,
      "learning_rate": 0.0001194030805267161,
      "loss": 1.4808,
      "step": 12697
    },
    {
      "epoch": 0.4547424212580801,
      "grad_norm": 1.8302372694015503,
      "learning_rate": 0.0001193917018174612,
      "loss": 1.2755,
      "step": 12698
    },
    {
      "epoch": 0.4547782333876484,
      "grad_norm": 2.5428664684295654,
      "learning_rate": 0.00011938032284731599,
      "loss": 1.5935,
      "step": 12699
    },
    {
      "epoch": 0.45481404551721666,
      "grad_norm": 1.475474238395691,
      "learning_rate": 0.00011936894361643351,
      "loss": 1.2114,
      "step": 12700
    },
    {
      "epoch": 0.454849857646785,
      "grad_norm": 1.447190284729004,
      "learning_rate": 0.00011935756412496688,
      "loss": 1.4698,
      "step": 12701
    },
    {
      "epoch": 0.45488566977635325,
      "grad_norm": 1.8733444213867188,
      "learning_rate": 0.00011934618437306921,
      "loss": 1.5448,
      "step": 12702
    },
    {
      "epoch": 0.4549214819059215,
      "grad_norm": 1.7487825155258179,
      "learning_rate": 0.00011933480436089357,
      "loss": 1.3573,
      "step": 12703
    },
    {
      "epoch": 0.45495729403548985,
      "grad_norm": 1.2866499423980713,
      "learning_rate": 0.0001193234240885931,
      "loss": 1.2512,
      "step": 12704
    },
    {
      "epoch": 0.4549931061650581,
      "grad_norm": 2.3530335426330566,
      "learning_rate": 0.00011931204355632089,
      "loss": 1.5614,
      "step": 12705
    },
    {
      "epoch": 0.4550289182946264,
      "grad_norm": 2.5091681480407715,
      "learning_rate": 0.00011930066276423003,
      "loss": 1.6465,
      "step": 12706
    },
    {
      "epoch": 0.45506473042419465,
      "grad_norm": 1.8246722221374512,
      "learning_rate": 0.00011928928171247362,
      "loss": 1.2457,
      "step": 12707
    },
    {
      "epoch": 0.455100542553763,
      "grad_norm": 1.3693405389785767,
      "learning_rate": 0.00011927790040120484,
      "loss": 1.4335,
      "step": 12708
    },
    {
      "epoch": 0.45513635468333125,
      "grad_norm": 1.6477317810058594,
      "learning_rate": 0.00011926651883057676,
      "loss": 1.3098,
      "step": 12709
    },
    {
      "epoch": 0.4551721668128995,
      "grad_norm": 1.7943549156188965,
      "learning_rate": 0.00011925513700074253,
      "loss": 1.4057,
      "step": 12710
    },
    {
      "epoch": 0.45520797894246784,
      "grad_norm": 1.7328225374221802,
      "learning_rate": 0.00011924375491185526,
      "loss": 1.3484,
      "step": 12711
    },
    {
      "epoch": 0.4552437910720361,
      "grad_norm": 1.3542251586914062,
      "learning_rate": 0.00011923237256406812,
      "loss": 1.549,
      "step": 12712
    },
    {
      "epoch": 0.4552796032016044,
      "grad_norm": 1.7922602891921997,
      "learning_rate": 0.00011922098995753417,
      "loss": 1.9181,
      "step": 12713
    },
    {
      "epoch": 0.45531541533117265,
      "grad_norm": 1.549342155456543,
      "learning_rate": 0.00011920960709240662,
      "loss": 1.543,
      "step": 12714
    },
    {
      "epoch": 0.45535122746074097,
      "grad_norm": 1.754526972770691,
      "learning_rate": 0.0001191982239688386,
      "loss": 1.2378,
      "step": 12715
    },
    {
      "epoch": 0.45538703959030924,
      "grad_norm": 1.7519073486328125,
      "learning_rate": 0.00011918684058698319,
      "loss": 1.4704,
      "step": 12716
    },
    {
      "epoch": 0.4554228517198775,
      "grad_norm": 1.3553789854049683,
      "learning_rate": 0.00011917545694699365,
      "loss": 1.3812,
      "step": 12717
    },
    {
      "epoch": 0.45545866384944583,
      "grad_norm": 1.7538580894470215,
      "learning_rate": 0.00011916407304902302,
      "loss": 1.7192,
      "step": 12718
    },
    {
      "epoch": 0.4554944759790141,
      "grad_norm": 1.9250893592834473,
      "learning_rate": 0.00011915268889322456,
      "loss": 1.4597,
      "step": 12719
    },
    {
      "epoch": 0.45553028810858237,
      "grad_norm": 1.8414658308029175,
      "learning_rate": 0.00011914130447975131,
      "loss": 1.8114,
      "step": 12720
    },
    {
      "epoch": 0.45556610023815064,
      "grad_norm": 1.3821085691452026,
      "learning_rate": 0.00011912991980875654,
      "loss": 1.4411,
      "step": 12721
    },
    {
      "epoch": 0.45560191236771896,
      "grad_norm": 1.5049561262130737,
      "learning_rate": 0.00011911853488039337,
      "loss": 1.4983,
      "step": 12722
    },
    {
      "epoch": 0.45563772449728723,
      "grad_norm": 1.8283472061157227,
      "learning_rate": 0.00011910714969481498,
      "loss": 1.5525,
      "step": 12723
    },
    {
      "epoch": 0.4556735366268555,
      "grad_norm": 1.6375936269760132,
      "learning_rate": 0.00011909576425217455,
      "loss": 1.8264,
      "step": 12724
    },
    {
      "epoch": 0.4557093487564238,
      "grad_norm": 1.530905842781067,
      "learning_rate": 0.0001190843785526252,
      "loss": 1.5206,
      "step": 12725
    },
    {
      "epoch": 0.4557451608859921,
      "grad_norm": 1.5139286518096924,
      "learning_rate": 0.00011907299259632019,
      "loss": 1.4437,
      "step": 12726
    },
    {
      "epoch": 0.45578097301556036,
      "grad_norm": 1.5392321348190308,
      "learning_rate": 0.00011906160638341264,
      "loss": 1.43,
      "step": 12727
    },
    {
      "epoch": 0.45581678514512863,
      "grad_norm": 2.803849458694458,
      "learning_rate": 0.00011905021991405578,
      "loss": 1.8596,
      "step": 12728
    },
    {
      "epoch": 0.45585259727469696,
      "grad_norm": 1.545616865158081,
      "learning_rate": 0.00011903883318840279,
      "loss": 1.6796,
      "step": 12729
    },
    {
      "epoch": 0.4558884094042652,
      "grad_norm": 1.9524319171905518,
      "learning_rate": 0.00011902744620660686,
      "loss": 1.559,
      "step": 12730
    },
    {
      "epoch": 0.4559242215338335,
      "grad_norm": 2.1568334102630615,
      "learning_rate": 0.00011901605896882116,
      "loss": 1.5606,
      "step": 12731
    },
    {
      "epoch": 0.4559600336634018,
      "grad_norm": 2.381507635116577,
      "learning_rate": 0.00011900467147519893,
      "loss": 1.2422,
      "step": 12732
    },
    {
      "epoch": 0.4559958457929701,
      "grad_norm": 1.6769733428955078,
      "learning_rate": 0.00011899328372589338,
      "loss": 1.4349,
      "step": 12733
    },
    {
      "epoch": 0.45603165792253836,
      "grad_norm": 1.7291309833526611,
      "learning_rate": 0.00011898189572105767,
      "loss": 1.5192,
      "step": 12734
    },
    {
      "epoch": 0.4560674700521066,
      "grad_norm": 2.131802797317505,
      "learning_rate": 0.00011897050746084504,
      "loss": 1.3411,
      "step": 12735
    },
    {
      "epoch": 0.45610328218167495,
      "grad_norm": 1.5254218578338623,
      "learning_rate": 0.0001189591189454087,
      "loss": 1.3057,
      "step": 12736
    },
    {
      "epoch": 0.4561390943112432,
      "grad_norm": 1.46796715259552,
      "learning_rate": 0.00011894773017490189,
      "loss": 1.2683,
      "step": 12737
    },
    {
      "epoch": 0.4561749064408115,
      "grad_norm": 1.5670194625854492,
      "learning_rate": 0.00011893634114947778,
      "loss": 1.3185,
      "step": 12738
    },
    {
      "epoch": 0.4562107185703798,
      "grad_norm": 1.6794108152389526,
      "learning_rate": 0.00011892495186928966,
      "loss": 1.5493,
      "step": 12739
    },
    {
      "epoch": 0.4562465306999481,
      "grad_norm": 1.5749014616012573,
      "learning_rate": 0.00011891356233449069,
      "loss": 1.4679,
      "step": 12740
    },
    {
      "epoch": 0.45628234282951635,
      "grad_norm": 1.948190450668335,
      "learning_rate": 0.0001189021725452342,
      "loss": 1.5716,
      "step": 12741
    },
    {
      "epoch": 0.4563181549590846,
      "grad_norm": 1.7379204034805298,
      "learning_rate": 0.00011889078250167329,
      "loss": 1.4168,
      "step": 12742
    },
    {
      "epoch": 0.45635396708865295,
      "grad_norm": 1.6801133155822754,
      "learning_rate": 0.0001188793922039613,
      "loss": 1.1718,
      "step": 12743
    },
    {
      "epoch": 0.4563897792182212,
      "grad_norm": 2.2052829265594482,
      "learning_rate": 0.00011886800165225143,
      "loss": 1.5625,
      "step": 12744
    },
    {
      "epoch": 0.4564255913477895,
      "grad_norm": 1.8127167224884033,
      "learning_rate": 0.00011885661084669693,
      "loss": 1.4002,
      "step": 12745
    },
    {
      "epoch": 0.4564614034773578,
      "grad_norm": 1.4773932695388794,
      "learning_rate": 0.00011884521978745106,
      "loss": 1.6552,
      "step": 12746
    },
    {
      "epoch": 0.4564972156069261,
      "grad_norm": 1.6315429210662842,
      "learning_rate": 0.00011883382847466706,
      "loss": 1.6935,
      "step": 12747
    },
    {
      "epoch": 0.45653302773649435,
      "grad_norm": 2.3406314849853516,
      "learning_rate": 0.00011882243690849824,
      "loss": 1.343,
      "step": 12748
    },
    {
      "epoch": 0.4565688398660626,
      "grad_norm": 1.2826876640319824,
      "learning_rate": 0.00011881104508909778,
      "loss": 1.4219,
      "step": 12749
    },
    {
      "epoch": 0.45660465199563094,
      "grad_norm": 1.8661410808563232,
      "learning_rate": 0.00011879965301661897,
      "loss": 1.7659,
      "step": 12750
    },
    {
      "epoch": 0.4566404641251992,
      "grad_norm": 1.9449056386947632,
      "learning_rate": 0.00011878826069121505,
      "loss": 1.4855,
      "step": 12751
    },
    {
      "epoch": 0.4566762762547675,
      "grad_norm": 1.3876155614852905,
      "learning_rate": 0.00011877686811303937,
      "loss": 1.3685,
      "step": 12752
    },
    {
      "epoch": 0.4567120883843358,
      "grad_norm": 1.7499020099639893,
      "learning_rate": 0.00011876547528224511,
      "loss": 1.6676,
      "step": 12753
    },
    {
      "epoch": 0.45674790051390407,
      "grad_norm": 1.5707976818084717,
      "learning_rate": 0.00011875408219898561,
      "loss": 1.6834,
      "step": 12754
    },
    {
      "epoch": 0.45678371264347234,
      "grad_norm": 1.6355860233306885,
      "learning_rate": 0.00011874268886341409,
      "loss": 1.6243,
      "step": 12755
    },
    {
      "epoch": 0.4568195247730406,
      "grad_norm": 1.9601993560791016,
      "learning_rate": 0.00011873129527568388,
      "loss": 1.6477,
      "step": 12756
    },
    {
      "epoch": 0.45685533690260893,
      "grad_norm": 1.5781079530715942,
      "learning_rate": 0.00011871990143594827,
      "loss": 1.543,
      "step": 12757
    },
    {
      "epoch": 0.4568911490321772,
      "grad_norm": 2.1427834033966064,
      "learning_rate": 0.00011870850734436054,
      "loss": 1.5517,
      "step": 12758
    },
    {
      "epoch": 0.45692696116174547,
      "grad_norm": 1.4001954793930054,
      "learning_rate": 0.00011869711300107398,
      "loss": 1.3041,
      "step": 12759
    },
    {
      "epoch": 0.4569627732913138,
      "grad_norm": 1.4365400075912476,
      "learning_rate": 0.00011868571840624185,
      "loss": 1.5515,
      "step": 12760
    },
    {
      "epoch": 0.45699858542088206,
      "grad_norm": 1.820007562637329,
      "learning_rate": 0.0001186743235600175,
      "loss": 1.6383,
      "step": 12761
    },
    {
      "epoch": 0.45703439755045033,
      "grad_norm": 2.0829074382781982,
      "learning_rate": 0.0001186629284625542,
      "loss": 1.684,
      "step": 12762
    },
    {
      "epoch": 0.4570702096800186,
      "grad_norm": 2.3919122219085693,
      "learning_rate": 0.00011865153311400529,
      "loss": 1.4495,
      "step": 12763
    },
    {
      "epoch": 0.4571060218095869,
      "grad_norm": 1.607649564743042,
      "learning_rate": 0.00011864013751452405,
      "loss": 1.6239,
      "step": 12764
    },
    {
      "epoch": 0.4571418339391552,
      "grad_norm": 1.874072551727295,
      "learning_rate": 0.00011862874166426381,
      "loss": 1.3917,
      "step": 12765
    },
    {
      "epoch": 0.45717764606872346,
      "grad_norm": 1.8643479347229004,
      "learning_rate": 0.00011861734556337787,
      "loss": 1.5116,
      "step": 12766
    },
    {
      "epoch": 0.4572134581982918,
      "grad_norm": 1.378171682357788,
      "learning_rate": 0.00011860594921201958,
      "loss": 1.5747,
      "step": 12767
    },
    {
      "epoch": 0.45724927032786006,
      "grad_norm": 2.0554890632629395,
      "learning_rate": 0.00011859455261034225,
      "loss": 1.5593,
      "step": 12768
    },
    {
      "epoch": 0.4572850824574283,
      "grad_norm": 1.3280192613601685,
      "learning_rate": 0.00011858315575849914,
      "loss": 1.3724,
      "step": 12769
    },
    {
      "epoch": 0.4573208945869966,
      "grad_norm": 1.340054988861084,
      "learning_rate": 0.00011857175865664372,
      "loss": 1.5089,
      "step": 12770
    },
    {
      "epoch": 0.4573567067165649,
      "grad_norm": 2.0556211471557617,
      "learning_rate": 0.00011856036130492917,
      "loss": 1.8506,
      "step": 12771
    },
    {
      "epoch": 0.4573925188461332,
      "grad_norm": 1.7079956531524658,
      "learning_rate": 0.00011854896370350894,
      "loss": 1.5426,
      "step": 12772
    },
    {
      "epoch": 0.45742833097570146,
      "grad_norm": 1.4690487384796143,
      "learning_rate": 0.00011853756585253633,
      "loss": 1.3461,
      "step": 12773
    },
    {
      "epoch": 0.4574641431052697,
      "grad_norm": 1.5755233764648438,
      "learning_rate": 0.00011852616775216467,
      "loss": 1.6978,
      "step": 12774
    },
    {
      "epoch": 0.45749995523483805,
      "grad_norm": 1.4344228506088257,
      "learning_rate": 0.00011851476940254733,
      "loss": 1.5199,
      "step": 12775
    },
    {
      "epoch": 0.4575357673644063,
      "grad_norm": 1.373246431350708,
      "learning_rate": 0.00011850337080383764,
      "loss": 1.5406,
      "step": 12776
    },
    {
      "epoch": 0.4575715794939746,
      "grad_norm": 1.866769790649414,
      "learning_rate": 0.000118491971956189,
      "loss": 1.3924,
      "step": 12777
    },
    {
      "epoch": 0.4576073916235429,
      "grad_norm": 1.9502559900283813,
      "learning_rate": 0.00011848057285975467,
      "loss": 1.4118,
      "step": 12778
    },
    {
      "epoch": 0.4576432037531112,
      "grad_norm": 1.2907614707946777,
      "learning_rate": 0.00011846917351468811,
      "loss": 1.06,
      "step": 12779
    },
    {
      "epoch": 0.45767901588267945,
      "grad_norm": 1.657185673713684,
      "learning_rate": 0.00011845777392114263,
      "loss": 1.4635,
      "step": 12780
    },
    {
      "epoch": 0.4577148280122477,
      "grad_norm": 2.1318721771240234,
      "learning_rate": 0.00011844637407927161,
      "loss": 1.3132,
      "step": 12781
    },
    {
      "epoch": 0.45775064014181605,
      "grad_norm": 1.5741397142410278,
      "learning_rate": 0.00011843497398922842,
      "loss": 1.3551,
      "step": 12782
    },
    {
      "epoch": 0.4577864522713843,
      "grad_norm": 1.3207534551620483,
      "learning_rate": 0.00011842357365116645,
      "loss": 1.7398,
      "step": 12783
    },
    {
      "epoch": 0.4578222644009526,
      "grad_norm": 1.8568518161773682,
      "learning_rate": 0.00011841217306523904,
      "loss": 1.7068,
      "step": 12784
    },
    {
      "epoch": 0.4578580765305209,
      "grad_norm": 1.3285858631134033,
      "learning_rate": 0.00011840077223159965,
      "loss": 1.676,
      "step": 12785
    },
    {
      "epoch": 0.4578938886600892,
      "grad_norm": 2.2445781230926514,
      "learning_rate": 0.00011838937115040154,
      "loss": 1.5313,
      "step": 12786
    },
    {
      "epoch": 0.45792970078965745,
      "grad_norm": 1.4467315673828125,
      "learning_rate": 0.00011837796982179817,
      "loss": 1.3382,
      "step": 12787
    },
    {
      "epoch": 0.4579655129192257,
      "grad_norm": 1.7218992710113525,
      "learning_rate": 0.00011836656824594295,
      "loss": 1.685,
      "step": 12788
    },
    {
      "epoch": 0.45800132504879404,
      "grad_norm": 1.835891604423523,
      "learning_rate": 0.0001183551664229892,
      "loss": 1.7658,
      "step": 12789
    },
    {
      "epoch": 0.4580371371783623,
      "grad_norm": 1.723070740699768,
      "learning_rate": 0.0001183437643530904,
      "loss": 1.7154,
      "step": 12790
    },
    {
      "epoch": 0.4580729493079306,
      "grad_norm": 2.18641996383667,
      "learning_rate": 0.00011833236203639987,
      "loss": 1.5002,
      "step": 12791
    },
    {
      "epoch": 0.4581087614374989,
      "grad_norm": 1.8997712135314941,
      "learning_rate": 0.00011832095947307111,
      "loss": 1.8594,
      "step": 12792
    },
    {
      "epoch": 0.45814457356706717,
      "grad_norm": 2.286189317703247,
      "learning_rate": 0.00011830955666325748,
      "loss": 1.2922,
      "step": 12793
    },
    {
      "epoch": 0.45818038569663544,
      "grad_norm": 1.3605753183364868,
      "learning_rate": 0.00011829815360711234,
      "loss": 1.4984,
      "step": 12794
    },
    {
      "epoch": 0.4582161978262037,
      "grad_norm": 1.8744243383407593,
      "learning_rate": 0.00011828675030478915,
      "loss": 1.5561,
      "step": 12795
    },
    {
      "epoch": 0.45825200995577203,
      "grad_norm": 1.505820870399475,
      "learning_rate": 0.00011827534675644134,
      "loss": 1.3156,
      "step": 12796
    },
    {
      "epoch": 0.4582878220853403,
      "grad_norm": 1.856209397315979,
      "learning_rate": 0.00011826394296222229,
      "loss": 1.4686,
      "step": 12797
    },
    {
      "epoch": 0.45832363421490857,
      "grad_norm": 1.7676700353622437,
      "learning_rate": 0.00011825253892228547,
      "loss": 1.4413,
      "step": 12798
    },
    {
      "epoch": 0.4583594463444769,
      "grad_norm": 2.2490038871765137,
      "learning_rate": 0.00011824113463678427,
      "loss": 1.6963,
      "step": 12799
    },
    {
      "epoch": 0.45839525847404516,
      "grad_norm": 1.618058204650879,
      "learning_rate": 0.00011822973010587213,
      "loss": 1.6653,
      "step": 12800
    },
    {
      "epoch": 0.45843107060361343,
      "grad_norm": 1.5924299955368042,
      "learning_rate": 0.0001182183253297025,
      "loss": 1.3864,
      "step": 12801
    },
    {
      "epoch": 0.4584668827331817,
      "grad_norm": 2.89251971244812,
      "learning_rate": 0.00011820692030842879,
      "loss": 1.3747,
      "step": 12802
    },
    {
      "epoch": 0.45850269486275,
      "grad_norm": 1.7772977352142334,
      "learning_rate": 0.00011819551504220447,
      "loss": 1.5318,
      "step": 12803
    },
    {
      "epoch": 0.4585385069923183,
      "grad_norm": 1.6167027950286865,
      "learning_rate": 0.00011818410953118296,
      "loss": 1.3861,
      "step": 12804
    },
    {
      "epoch": 0.45857431912188656,
      "grad_norm": 1.2884777784347534,
      "learning_rate": 0.0001181727037755177,
      "loss": 0.8727,
      "step": 12805
    },
    {
      "epoch": 0.4586101312514549,
      "grad_norm": 1.5280265808105469,
      "learning_rate": 0.00011816129777536216,
      "loss": 1.255,
      "step": 12806
    },
    {
      "epoch": 0.45864594338102316,
      "grad_norm": 1.8953804969787598,
      "learning_rate": 0.00011814989153086977,
      "loss": 1.5201,
      "step": 12807
    },
    {
      "epoch": 0.4586817555105914,
      "grad_norm": 1.4978008270263672,
      "learning_rate": 0.00011813848504219403,
      "loss": 1.3398,
      "step": 12808
    },
    {
      "epoch": 0.4587175676401597,
      "grad_norm": 1.7780437469482422,
      "learning_rate": 0.00011812707830948835,
      "loss": 1.6133,
      "step": 12809
    },
    {
      "epoch": 0.458753379769728,
      "grad_norm": 1.5132312774658203,
      "learning_rate": 0.0001181156713329062,
      "loss": 1.4327,
      "step": 12810
    },
    {
      "epoch": 0.4587891918992963,
      "grad_norm": 1.6807321310043335,
      "learning_rate": 0.0001181042641126011,
      "loss": 1.4845,
      "step": 12811
    },
    {
      "epoch": 0.45882500402886456,
      "grad_norm": 1.6463050842285156,
      "learning_rate": 0.00011809285664872645,
      "loss": 1.3477,
      "step": 12812
    },
    {
      "epoch": 0.4588608161584329,
      "grad_norm": 1.670136570930481,
      "learning_rate": 0.00011808144894143575,
      "loss": 1.5761,
      "step": 12813
    },
    {
      "epoch": 0.45889662828800115,
      "grad_norm": 1.6936041116714478,
      "learning_rate": 0.00011807004099088251,
      "loss": 1.4741,
      "step": 12814
    },
    {
      "epoch": 0.4589324404175694,
      "grad_norm": 1.5984362363815308,
      "learning_rate": 0.00011805863279722014,
      "loss": 1.6037,
      "step": 12815
    },
    {
      "epoch": 0.4589682525471377,
      "grad_norm": 1.9859976768493652,
      "learning_rate": 0.00011804722436060218,
      "loss": 1.6921,
      "step": 12816
    },
    {
      "epoch": 0.459004064676706,
      "grad_norm": 1.768145203590393,
      "learning_rate": 0.00011803581568118207,
      "loss": 1.322,
      "step": 12817
    },
    {
      "epoch": 0.4590398768062743,
      "grad_norm": 1.2715734243392944,
      "learning_rate": 0.00011802440675911335,
      "loss": 1.4228,
      "step": 12818
    },
    {
      "epoch": 0.45907568893584255,
      "grad_norm": 1.2998805046081543,
      "learning_rate": 0.00011801299759454947,
      "loss": 1.2947,
      "step": 12819
    },
    {
      "epoch": 0.4591115010654109,
      "grad_norm": 2.153787612915039,
      "learning_rate": 0.00011800158818764395,
      "loss": 1.8239,
      "step": 12820
    },
    {
      "epoch": 0.45914731319497915,
      "grad_norm": 1.4369038343429565,
      "learning_rate": 0.0001179901785385503,
      "loss": 1.4372,
      "step": 12821
    },
    {
      "epoch": 0.4591831253245474,
      "grad_norm": 2.008143901824951,
      "learning_rate": 0.00011797876864742198,
      "loss": 1.6738,
      "step": 12822
    },
    {
      "epoch": 0.4592189374541157,
      "grad_norm": 1.253076434135437,
      "learning_rate": 0.00011796735851441254,
      "loss": 1.2106,
      "step": 12823
    },
    {
      "epoch": 0.459254749583684,
      "grad_norm": 1.9890899658203125,
      "learning_rate": 0.00011795594813967543,
      "loss": 1.865,
      "step": 12824
    },
    {
      "epoch": 0.4592905617132523,
      "grad_norm": 1.4723730087280273,
      "learning_rate": 0.00011794453752336425,
      "loss": 1.3423,
      "step": 12825
    },
    {
      "epoch": 0.45932637384282055,
      "grad_norm": 3.3311033248901367,
      "learning_rate": 0.00011793312666563241,
      "loss": 1.5936,
      "step": 12826
    },
    {
      "epoch": 0.45936218597238887,
      "grad_norm": 1.470004916191101,
      "learning_rate": 0.00011792171556663353,
      "loss": 1.1274,
      "step": 12827
    },
    {
      "epoch": 0.45939799810195714,
      "grad_norm": 1.7341986894607544,
      "learning_rate": 0.00011791030422652105,
      "loss": 1.1953,
      "step": 12828
    },
    {
      "epoch": 0.4594338102315254,
      "grad_norm": 1.5090793371200562,
      "learning_rate": 0.00011789889264544855,
      "loss": 1.3247,
      "step": 12829
    },
    {
      "epoch": 0.4594696223610937,
      "grad_norm": 1.5003623962402344,
      "learning_rate": 0.00011788748082356955,
      "loss": 1.3318,
      "step": 12830
    },
    {
      "epoch": 0.459505434490662,
      "grad_norm": 2.2069389820098877,
      "learning_rate": 0.00011787606876103753,
      "loss": 1.6227,
      "step": 12831
    },
    {
      "epoch": 0.45954124662023027,
      "grad_norm": 1.5146262645721436,
      "learning_rate": 0.00011786465645800609,
      "loss": 1.5597,
      "step": 12832
    },
    {
      "epoch": 0.45957705874979854,
      "grad_norm": 1.5742777585983276,
      "learning_rate": 0.00011785324391462873,
      "loss": 1.4421,
      "step": 12833
    },
    {
      "epoch": 0.45961287087936686,
      "grad_norm": 1.3737620115280151,
      "learning_rate": 0.000117841831131059,
      "loss": 1.6641,
      "step": 12834
    },
    {
      "epoch": 0.45964868300893513,
      "grad_norm": 1.7888473272323608,
      "learning_rate": 0.00011783041810745045,
      "loss": 1.2693,
      "step": 12835
    },
    {
      "epoch": 0.4596844951385034,
      "grad_norm": 1.42727530002594,
      "learning_rate": 0.00011781900484395665,
      "loss": 1.5367,
      "step": 12836
    },
    {
      "epoch": 0.45972030726807167,
      "grad_norm": 1.4414780139923096,
      "learning_rate": 0.00011780759134073107,
      "loss": 2.0235,
      "step": 12837
    },
    {
      "epoch": 0.45975611939764,
      "grad_norm": 1.4907771348953247,
      "learning_rate": 0.00011779617759792738,
      "loss": 1.4951,
      "step": 12838
    },
    {
      "epoch": 0.45979193152720826,
      "grad_norm": 1.888153314590454,
      "learning_rate": 0.00011778476361569903,
      "loss": 1.3646,
      "step": 12839
    },
    {
      "epoch": 0.45982774365677653,
      "grad_norm": 1.6788287162780762,
      "learning_rate": 0.00011777334939419966,
      "loss": 1.3867,
      "step": 12840
    },
    {
      "epoch": 0.45986355578634486,
      "grad_norm": 1.3962904214859009,
      "learning_rate": 0.00011776193493358278,
      "loss": 1.6541,
      "step": 12841
    },
    {
      "epoch": 0.4598993679159131,
      "grad_norm": 1.7222188711166382,
      "learning_rate": 0.00011775052023400197,
      "loss": 1.5073,
      "step": 12842
    },
    {
      "epoch": 0.4599351800454814,
      "grad_norm": 1.299524188041687,
      "learning_rate": 0.0001177391052956108,
      "loss": 1.1971,
      "step": 12843
    },
    {
      "epoch": 0.45997099217504966,
      "grad_norm": 1.7616795301437378,
      "learning_rate": 0.00011772769011856286,
      "loss": 1.499,
      "step": 12844
    },
    {
      "epoch": 0.460006804304618,
      "grad_norm": 1.8444628715515137,
      "learning_rate": 0.00011771627470301174,
      "loss": 1.5539,
      "step": 12845
    },
    {
      "epoch": 0.46004261643418626,
      "grad_norm": 1.4378998279571533,
      "learning_rate": 0.00011770485904911099,
      "loss": 1.5755,
      "step": 12846
    },
    {
      "epoch": 0.4600784285637545,
      "grad_norm": 1.6388529539108276,
      "learning_rate": 0.0001176934431570142,
      "loss": 1.4707,
      "step": 12847
    },
    {
      "epoch": 0.46011424069332285,
      "grad_norm": 1.789455771446228,
      "learning_rate": 0.00011768202702687492,
      "loss": 1.3956,
      "step": 12848
    },
    {
      "epoch": 0.4601500528228911,
      "grad_norm": 1.9952750205993652,
      "learning_rate": 0.00011767061065884682,
      "loss": 1.7959,
      "step": 12849
    },
    {
      "epoch": 0.4601858649524594,
      "grad_norm": 2.108474016189575,
      "learning_rate": 0.00011765919405308341,
      "loss": 1.4997,
      "step": 12850
    },
    {
      "epoch": 0.46022167708202766,
      "grad_norm": 1.9382110834121704,
      "learning_rate": 0.00011764777720973835,
      "loss": 1.4933,
      "step": 12851
    },
    {
      "epoch": 0.460257489211596,
      "grad_norm": 1.7366992235183716,
      "learning_rate": 0.00011763636012896518,
      "loss": 1.6449,
      "step": 12852
    },
    {
      "epoch": 0.46029330134116425,
      "grad_norm": 1.8171309232711792,
      "learning_rate": 0.00011762494281091756,
      "loss": 1.6787,
      "step": 12853
    },
    {
      "epoch": 0.4603291134707325,
      "grad_norm": 1.760102391242981,
      "learning_rate": 0.00011761352525574905,
      "loss": 1.2648,
      "step": 12854
    },
    {
      "epoch": 0.46036492560030084,
      "grad_norm": 2.294532537460327,
      "learning_rate": 0.00011760210746361329,
      "loss": 1.6493,
      "step": 12855
    },
    {
      "epoch": 0.4604007377298691,
      "grad_norm": 1.5895004272460938,
      "learning_rate": 0.00011759068943466389,
      "loss": 1.3155,
      "step": 12856
    },
    {
      "epoch": 0.4604365498594374,
      "grad_norm": 1.6277079582214355,
      "learning_rate": 0.00011757927116905442,
      "loss": 1.4816,
      "step": 12857
    },
    {
      "epoch": 0.46047236198900565,
      "grad_norm": 1.8704755306243896,
      "learning_rate": 0.00011756785266693857,
      "loss": 1.4445,
      "step": 12858
    },
    {
      "epoch": 0.460508174118574,
      "grad_norm": 1.737906813621521,
      "learning_rate": 0.00011755643392846991,
      "loss": 1.7793,
      "step": 12859
    },
    {
      "epoch": 0.46054398624814225,
      "grad_norm": 1.859326720237732,
      "learning_rate": 0.00011754501495380209,
      "loss": 1.6184,
      "step": 12860
    },
    {
      "epoch": 0.4605797983777105,
      "grad_norm": 1.8571624755859375,
      "learning_rate": 0.00011753359574308869,
      "loss": 1.5855,
      "step": 12861
    },
    {
      "epoch": 0.46061561050727884,
      "grad_norm": 1.6223071813583374,
      "learning_rate": 0.00011752217629648341,
      "loss": 1.3289,
      "step": 12862
    },
    {
      "epoch": 0.4606514226368471,
      "grad_norm": 1.7638866901397705,
      "learning_rate": 0.00011751075661413982,
      "loss": 1.5636,
      "step": 12863
    },
    {
      "epoch": 0.4606872347664154,
      "grad_norm": 1.5971314907073975,
      "learning_rate": 0.00011749933669621161,
      "loss": 1.6486,
      "step": 12864
    },
    {
      "epoch": 0.46072304689598365,
      "grad_norm": 1.6933928728103638,
      "learning_rate": 0.0001174879165428524,
      "loss": 1.8318,
      "step": 12865
    },
    {
      "epoch": 0.46075885902555197,
      "grad_norm": 1.8240853548049927,
      "learning_rate": 0.00011747649615421581,
      "loss": 1.5988,
      "step": 12866
    },
    {
      "epoch": 0.46079467115512024,
      "grad_norm": 1.140206217765808,
      "learning_rate": 0.00011746507553045552,
      "loss": 1.3376,
      "step": 12867
    },
    {
      "epoch": 0.4608304832846885,
      "grad_norm": 1.5556854009628296,
      "learning_rate": 0.00011745365467172516,
      "loss": 1.4646,
      "step": 12868
    },
    {
      "epoch": 0.46086629541425683,
      "grad_norm": 2.0136656761169434,
      "learning_rate": 0.00011744223357817841,
      "loss": 1.8497,
      "step": 12869
    },
    {
      "epoch": 0.4609021075438251,
      "grad_norm": 2.2033441066741943,
      "learning_rate": 0.00011743081224996888,
      "loss": 1.5519,
      "step": 12870
    },
    {
      "epoch": 0.46093791967339337,
      "grad_norm": 1.670477032661438,
      "learning_rate": 0.00011741939068725027,
      "loss": 1.5283,
      "step": 12871
    },
    {
      "epoch": 0.46097373180296164,
      "grad_norm": 2.0751805305480957,
      "learning_rate": 0.00011740796889017623,
      "loss": 1.4366,
      "step": 12872
    },
    {
      "epoch": 0.46100954393252996,
      "grad_norm": 1.6487994194030762,
      "learning_rate": 0.0001173965468589004,
      "loss": 1.3776,
      "step": 12873
    },
    {
      "epoch": 0.46104535606209823,
      "grad_norm": 1.5400201082229614,
      "learning_rate": 0.0001173851245935765,
      "loss": 1.4772,
      "step": 12874
    },
    {
      "epoch": 0.4610811681916665,
      "grad_norm": 1.987924337387085,
      "learning_rate": 0.00011737370209435816,
      "loss": 1.3146,
      "step": 12875
    },
    {
      "epoch": 0.4611169803212348,
      "grad_norm": 2.468710422515869,
      "learning_rate": 0.00011736227936139908,
      "loss": 1.2949,
      "step": 12876
    },
    {
      "epoch": 0.4611527924508031,
      "grad_norm": 1.6486579179763794,
      "learning_rate": 0.00011735085639485291,
      "loss": 1.462,
      "step": 12877
    },
    {
      "epoch": 0.46118860458037136,
      "grad_norm": 1.8904602527618408,
      "learning_rate": 0.00011733943319487337,
      "loss": 1.6574,
      "step": 12878
    },
    {
      "epoch": 0.46122441670993963,
      "grad_norm": 1.5434259176254272,
      "learning_rate": 0.00011732800976161408,
      "loss": 1.4993,
      "step": 12879
    },
    {
      "epoch": 0.46126022883950796,
      "grad_norm": 1.8802088499069214,
      "learning_rate": 0.00011731658609522881,
      "loss": 1.2326,
      "step": 12880
    },
    {
      "epoch": 0.4612960409690762,
      "grad_norm": 1.7803751230239868,
      "learning_rate": 0.0001173051621958712,
      "loss": 2.1058,
      "step": 12881
    },
    {
      "epoch": 0.4613318530986445,
      "grad_norm": 1.6606054306030273,
      "learning_rate": 0.00011729373806369499,
      "loss": 1.6868,
      "step": 12882
    },
    {
      "epoch": 0.4613676652282128,
      "grad_norm": 1.609900951385498,
      "learning_rate": 0.0001172823136988538,
      "loss": 1.2621,
      "step": 12883
    },
    {
      "epoch": 0.4614034773577811,
      "grad_norm": 2.8169360160827637,
      "learning_rate": 0.00011727088910150137,
      "loss": 1.2055,
      "step": 12884
    },
    {
      "epoch": 0.46143928948734936,
      "grad_norm": 1.9346868991851807,
      "learning_rate": 0.00011725946427179142,
      "loss": 1.8939,
      "step": 12885
    },
    {
      "epoch": 0.4614751016169176,
      "grad_norm": 1.8201242685317993,
      "learning_rate": 0.00011724803920987761,
      "loss": 1.7242,
      "step": 12886
    },
    {
      "epoch": 0.46151091374648595,
      "grad_norm": 1.8943721055984497,
      "learning_rate": 0.00011723661391591371,
      "loss": 1.7006,
      "step": 12887
    },
    {
      "epoch": 0.4615467258760542,
      "grad_norm": 1.6095993518829346,
      "learning_rate": 0.00011722518839005341,
      "loss": 1.4559,
      "step": 12888
    },
    {
      "epoch": 0.4615825380056225,
      "grad_norm": 1.338904619216919,
      "learning_rate": 0.00011721376263245041,
      "loss": 1.6563,
      "step": 12889
    },
    {
      "epoch": 0.4616183501351908,
      "grad_norm": 1.563049554824829,
      "learning_rate": 0.00011720233664325846,
      "loss": 1.2904,
      "step": 12890
    },
    {
      "epoch": 0.4616541622647591,
      "grad_norm": 1.619492530822754,
      "learning_rate": 0.00011719091042263124,
      "loss": 1.5699,
      "step": 12891
    },
    {
      "epoch": 0.46168997439432735,
      "grad_norm": 1.8848820924758911,
      "learning_rate": 0.00011717948397072246,
      "loss": 1.7522,
      "step": 12892
    },
    {
      "epoch": 0.4617257865238956,
      "grad_norm": 1.626729130744934,
      "learning_rate": 0.00011716805728768593,
      "loss": 1.4691,
      "step": 12893
    },
    {
      "epoch": 0.46176159865346394,
      "grad_norm": 2.1712372303009033,
      "learning_rate": 0.00011715663037367532,
      "loss": 1.5208,
      "step": 12894
    },
    {
      "epoch": 0.4617974107830322,
      "grad_norm": 1.8824517726898193,
      "learning_rate": 0.00011714520322884439,
      "loss": 1.3719,
      "step": 12895
    },
    {
      "epoch": 0.4618332229126005,
      "grad_norm": 2.1174161434173584,
      "learning_rate": 0.00011713377585334684,
      "loss": 1.4144,
      "step": 12896
    },
    {
      "epoch": 0.4618690350421688,
      "grad_norm": 1.8503297567367554,
      "learning_rate": 0.00011712234824733644,
      "loss": 1.5521,
      "step": 12897
    },
    {
      "epoch": 0.4619048471717371,
      "grad_norm": 1.4610905647277832,
      "learning_rate": 0.00011711092041096693,
      "loss": 1.3756,
      "step": 12898
    },
    {
      "epoch": 0.46194065930130535,
      "grad_norm": 2.196420907974243,
      "learning_rate": 0.0001170994923443921,
      "loss": 1.6765,
      "step": 12899
    },
    {
      "epoch": 0.4619764714308736,
      "grad_norm": 1.5925959348678589,
      "learning_rate": 0.00011708806404776563,
      "loss": 1.5467,
      "step": 12900
    },
    {
      "epoch": 0.46201228356044194,
      "grad_norm": 1.4882124662399292,
      "learning_rate": 0.00011707663552124128,
      "loss": 1.7759,
      "step": 12901
    },
    {
      "epoch": 0.4620480956900102,
      "grad_norm": 1.9978431463241577,
      "learning_rate": 0.00011706520676497285,
      "loss": 1.7564,
      "step": 12902
    },
    {
      "epoch": 0.4620839078195785,
      "grad_norm": 1.8712289333343506,
      "learning_rate": 0.00011705377777911406,
      "loss": 1.5388,
      "step": 12903
    },
    {
      "epoch": 0.4621197199491468,
      "grad_norm": 3.0363831520080566,
      "learning_rate": 0.0001170423485638187,
      "loss": 1.5825,
      "step": 12904
    },
    {
      "epoch": 0.46215553207871507,
      "grad_norm": 1.3153165578842163,
      "learning_rate": 0.00011703091911924051,
      "loss": 1.306,
      "step": 12905
    },
    {
      "epoch": 0.46219134420828334,
      "grad_norm": 1.8656954765319824,
      "learning_rate": 0.0001170194894455333,
      "loss": 1.3723,
      "step": 12906
    },
    {
      "epoch": 0.4622271563378516,
      "grad_norm": 1.8745927810668945,
      "learning_rate": 0.0001170080595428508,
      "loss": 1.4533,
      "step": 12907
    },
    {
      "epoch": 0.46226296846741993,
      "grad_norm": 1.440604329109192,
      "learning_rate": 0.00011699662941134679,
      "loss": 1.3826,
      "step": 12908
    },
    {
      "epoch": 0.4622987805969882,
      "grad_norm": 1.6274751424789429,
      "learning_rate": 0.00011698519905117507,
      "loss": 1.3327,
      "step": 12909
    },
    {
      "epoch": 0.46233459272655647,
      "grad_norm": 1.5062453746795654,
      "learning_rate": 0.00011697376846248937,
      "loss": 1.5934,
      "step": 12910
    },
    {
      "epoch": 0.4623704048561248,
      "grad_norm": 1.5800942182540894,
      "learning_rate": 0.00011696233764544353,
      "loss": 1.3237,
      "step": 12911
    },
    {
      "epoch": 0.46240621698569306,
      "grad_norm": 1.7886601686477661,
      "learning_rate": 0.00011695090660019132,
      "loss": 1.512,
      "step": 12912
    },
    {
      "epoch": 0.46244202911526133,
      "grad_norm": 1.6986116170883179,
      "learning_rate": 0.00011693947532688653,
      "loss": 1.5372,
      "step": 12913
    },
    {
      "epoch": 0.4624778412448296,
      "grad_norm": 1.8401761054992676,
      "learning_rate": 0.00011692804382568294,
      "loss": 1.5287,
      "step": 12914
    },
    {
      "epoch": 0.4625136533743979,
      "grad_norm": 1.7199172973632812,
      "learning_rate": 0.00011691661209673437,
      "loss": 1.5185,
      "step": 12915
    },
    {
      "epoch": 0.4625494655039662,
      "grad_norm": 1.6660505533218384,
      "learning_rate": 0.00011690518014019458,
      "loss": 1.3418,
      "step": 12916
    },
    {
      "epoch": 0.46258527763353446,
      "grad_norm": 2.050318956375122,
      "learning_rate": 0.00011689374795621744,
      "loss": 1.4391,
      "step": 12917
    },
    {
      "epoch": 0.4626210897631028,
      "grad_norm": 1.6393921375274658,
      "learning_rate": 0.00011688231554495668,
      "loss": 1.4133,
      "step": 12918
    },
    {
      "epoch": 0.46265690189267106,
      "grad_norm": 1.5814021825790405,
      "learning_rate": 0.00011687088290656613,
      "loss": 1.2826,
      "step": 12919
    },
    {
      "epoch": 0.4626927140222393,
      "grad_norm": 1.5501033067703247,
      "learning_rate": 0.00011685945004119965,
      "loss": 1.6777,
      "step": 12920
    },
    {
      "epoch": 0.4627285261518076,
      "grad_norm": 1.8156019449234009,
      "learning_rate": 0.00011684801694901099,
      "loss": 1.4462,
      "step": 12921
    },
    {
      "epoch": 0.4627643382813759,
      "grad_norm": 1.7225736379623413,
      "learning_rate": 0.00011683658363015402,
      "loss": 1.2303,
      "step": 12922
    },
    {
      "epoch": 0.4628001504109442,
      "grad_norm": 1.728131890296936,
      "learning_rate": 0.0001168251500847825,
      "loss": 1.4446,
      "step": 12923
    },
    {
      "epoch": 0.46283596254051246,
      "grad_norm": 2.56362247467041,
      "learning_rate": 0.00011681371631305032,
      "loss": 1.6334,
      "step": 12924
    },
    {
      "epoch": 0.4628717746700808,
      "grad_norm": 2.016491413116455,
      "learning_rate": 0.00011680228231511123,
      "loss": 1.3181,
      "step": 12925
    },
    {
      "epoch": 0.46290758679964905,
      "grad_norm": 1.6673835515975952,
      "learning_rate": 0.00011679084809111915,
      "loss": 1.51,
      "step": 12926
    },
    {
      "epoch": 0.4629433989292173,
      "grad_norm": 1.5395658016204834,
      "learning_rate": 0.00011677941364122787,
      "loss": 1.3948,
      "step": 12927
    },
    {
      "epoch": 0.4629792110587856,
      "grad_norm": 1.4805071353912354,
      "learning_rate": 0.0001167679789655912,
      "loss": 1.3565,
      "step": 12928
    },
    {
      "epoch": 0.4630150231883539,
      "grad_norm": 1.8518954515457153,
      "learning_rate": 0.00011675654406436301,
      "loss": 1.2927,
      "step": 12929
    },
    {
      "epoch": 0.4630508353179222,
      "grad_norm": 2.3132810592651367,
      "learning_rate": 0.00011674510893769713,
      "loss": 1.5319,
      "step": 12930
    },
    {
      "epoch": 0.46308664744749045,
      "grad_norm": 1.4110107421875,
      "learning_rate": 0.00011673367358574741,
      "loss": 0.9963,
      "step": 12931
    },
    {
      "epoch": 0.4631224595770588,
      "grad_norm": 1.7076191902160645,
      "learning_rate": 0.00011672223800866768,
      "loss": 1.7017,
      "step": 12932
    },
    {
      "epoch": 0.46315827170662704,
      "grad_norm": 2.103776693344116,
      "learning_rate": 0.00011671080220661183,
      "loss": 1.4991,
      "step": 12933
    },
    {
      "epoch": 0.4631940838361953,
      "grad_norm": 1.6993529796600342,
      "learning_rate": 0.00011669936617973367,
      "loss": 1.4374,
      "step": 12934
    },
    {
      "epoch": 0.4632298959657636,
      "grad_norm": 1.5589717626571655,
      "learning_rate": 0.00011668792992818714,
      "loss": 1.3671,
      "step": 12935
    },
    {
      "epoch": 0.4632657080953319,
      "grad_norm": 1.5046939849853516,
      "learning_rate": 0.000116676493452126,
      "loss": 1.6337,
      "step": 12936
    },
    {
      "epoch": 0.4633015202249002,
      "grad_norm": 1.6186277866363525,
      "learning_rate": 0.00011666505675170413,
      "loss": 1.6858,
      "step": 12937
    },
    {
      "epoch": 0.46333733235446845,
      "grad_norm": 2.007828950881958,
      "learning_rate": 0.00011665361982707543,
      "loss": 1.6257,
      "step": 12938
    },
    {
      "epoch": 0.46337314448403677,
      "grad_norm": 1.3552885055541992,
      "learning_rate": 0.00011664218267839375,
      "loss": 1.3457,
      "step": 12939
    },
    {
      "epoch": 0.46340895661360504,
      "grad_norm": 2.314358949661255,
      "learning_rate": 0.000116630745305813,
      "loss": 1.3535,
      "step": 12940
    },
    {
      "epoch": 0.4634447687431733,
      "grad_norm": 1.718798279762268,
      "learning_rate": 0.00011661930770948699,
      "loss": 1.4442,
      "step": 12941
    },
    {
      "epoch": 0.4634805808727416,
      "grad_norm": 1.6392550468444824,
      "learning_rate": 0.00011660786988956964,
      "loss": 1.3206,
      "step": 12942
    },
    {
      "epoch": 0.4635163930023099,
      "grad_norm": 1.5570787191390991,
      "learning_rate": 0.00011659643184621485,
      "loss": 1.3863,
      "step": 12943
    },
    {
      "epoch": 0.46355220513187817,
      "grad_norm": 2.412844181060791,
      "learning_rate": 0.00011658499357957646,
      "loss": 1.6282,
      "step": 12944
    },
    {
      "epoch": 0.46358801726144644,
      "grad_norm": 1.7498209476470947,
      "learning_rate": 0.00011657355508980836,
      "loss": 1.4421,
      "step": 12945
    },
    {
      "epoch": 0.46362382939101476,
      "grad_norm": 1.9625868797302246,
      "learning_rate": 0.00011656211637706449,
      "loss": 1.8111,
      "step": 12946
    },
    {
      "epoch": 0.46365964152058303,
      "grad_norm": 2.1074130535125732,
      "learning_rate": 0.00011655067744149865,
      "loss": 1.4252,
      "step": 12947
    },
    {
      "epoch": 0.4636954536501513,
      "grad_norm": 1.5764514207839966,
      "learning_rate": 0.00011653923828326485,
      "loss": 1.2029,
      "step": 12948
    },
    {
      "epoch": 0.46373126577971957,
      "grad_norm": 1.5708736181259155,
      "learning_rate": 0.0001165277989025169,
      "loss": 1.3832,
      "step": 12949
    },
    {
      "epoch": 0.4637670779092879,
      "grad_norm": 1.6243404150009155,
      "learning_rate": 0.00011651635929940874,
      "loss": 1.2595,
      "step": 12950
    },
    {
      "epoch": 0.46380289003885616,
      "grad_norm": 2.1377639770507812,
      "learning_rate": 0.00011650491947409427,
      "loss": 1.7661,
      "step": 12951
    },
    {
      "epoch": 0.46383870216842443,
      "grad_norm": 1.541272759437561,
      "learning_rate": 0.00011649347942672741,
      "loss": 1.5135,
      "step": 12952
    },
    {
      "epoch": 0.46387451429799276,
      "grad_norm": 2.0266079902648926,
      "learning_rate": 0.00011648203915746208,
      "loss": 1.7173,
      "step": 12953
    },
    {
      "epoch": 0.463910326427561,
      "grad_norm": 1.9679988622665405,
      "learning_rate": 0.00011647059866645213,
      "loss": 1.5395,
      "step": 12954
    },
    {
      "epoch": 0.4639461385571293,
      "grad_norm": 1.322680950164795,
      "learning_rate": 0.00011645915795385154,
      "loss": 1.603,
      "step": 12955
    },
    {
      "epoch": 0.46398195068669756,
      "grad_norm": 1.8947259187698364,
      "learning_rate": 0.0001164477170198142,
      "loss": 1.292,
      "step": 12956
    },
    {
      "epoch": 0.4640177628162659,
      "grad_norm": 1.686653733253479,
      "learning_rate": 0.00011643627586449406,
      "loss": 1.6056,
      "step": 12957
    },
    {
      "epoch": 0.46405357494583416,
      "grad_norm": 2.2820146083831787,
      "learning_rate": 0.000116424834488045,
      "loss": 1.4883,
      "step": 12958
    },
    {
      "epoch": 0.4640893870754024,
      "grad_norm": 2.1820619106292725,
      "learning_rate": 0.00011641339289062101,
      "loss": 1.6662,
      "step": 12959
    },
    {
      "epoch": 0.46412519920497075,
      "grad_norm": 2.1300809383392334,
      "learning_rate": 0.00011640195107237596,
      "loss": 1.3605,
      "step": 12960
    },
    {
      "epoch": 0.464161011334539,
      "grad_norm": 1.3338943719863892,
      "learning_rate": 0.00011639050903346387,
      "loss": 1.4019,
      "step": 12961
    },
    {
      "epoch": 0.4641968234641073,
      "grad_norm": 1.9710988998413086,
      "learning_rate": 0.00011637906677403859,
      "loss": 1.3826,
      "step": 12962
    },
    {
      "epoch": 0.46423263559367556,
      "grad_norm": 1.3079078197479248,
      "learning_rate": 0.00011636762429425407,
      "loss": 1.1247,
      "step": 12963
    },
    {
      "epoch": 0.4642684477232439,
      "grad_norm": 1.8297098875045776,
      "learning_rate": 0.0001163561815942643,
      "loss": 1.5155,
      "step": 12964
    },
    {
      "epoch": 0.46430425985281215,
      "grad_norm": 1.8603323698043823,
      "learning_rate": 0.00011634473867422322,
      "loss": 1.4551,
      "step": 12965
    },
    {
      "epoch": 0.4643400719823804,
      "grad_norm": 1.6557583808898926,
      "learning_rate": 0.00011633329553428476,
      "loss": 1.29,
      "step": 12966
    },
    {
      "epoch": 0.46437588411194874,
      "grad_norm": 1.406653881072998,
      "learning_rate": 0.00011632185217460283,
      "loss": 1.5551,
      "step": 12967
    },
    {
      "epoch": 0.464411696241517,
      "grad_norm": 1.6777414083480835,
      "learning_rate": 0.00011631040859533148,
      "loss": 1.3424,
      "step": 12968
    },
    {
      "epoch": 0.4644475083710853,
      "grad_norm": 2.046297311782837,
      "learning_rate": 0.00011629896479662461,
      "loss": 1.5278,
      "step": 12969
    },
    {
      "epoch": 0.46448332050065355,
      "grad_norm": 1.4424058198928833,
      "learning_rate": 0.0001162875207786362,
      "loss": 1.3149,
      "step": 12970
    },
    {
      "epoch": 0.4645191326302219,
      "grad_norm": 1.1536939144134521,
      "learning_rate": 0.00011627607654152022,
      "loss": 1.4943,
      "step": 12971
    },
    {
      "epoch": 0.46455494475979014,
      "grad_norm": 1.738486409187317,
      "learning_rate": 0.0001162646320854306,
      "loss": 1.3925,
      "step": 12972
    },
    {
      "epoch": 0.4645907568893584,
      "grad_norm": 2.092022657394409,
      "learning_rate": 0.00011625318741052133,
      "loss": 1.7829,
      "step": 12973
    },
    {
      "epoch": 0.4646265690189267,
      "grad_norm": 3.35194993019104,
      "learning_rate": 0.0001162417425169464,
      "loss": 1.3859,
      "step": 12974
    },
    {
      "epoch": 0.464662381148495,
      "grad_norm": 1.7595601081848145,
      "learning_rate": 0.00011623029740485978,
      "loss": 1.4044,
      "step": 12975
    },
    {
      "epoch": 0.4646981932780633,
      "grad_norm": 1.210787057876587,
      "learning_rate": 0.00011621885207441541,
      "loss": 1.5113,
      "step": 12976
    },
    {
      "epoch": 0.46473400540763155,
      "grad_norm": 1.481057047843933,
      "learning_rate": 0.00011620740652576736,
      "loss": 1.1624,
      "step": 12977
    },
    {
      "epoch": 0.46476981753719987,
      "grad_norm": 2.3686330318450928,
      "learning_rate": 0.0001161959607590695,
      "loss": 1.4887,
      "step": 12978
    },
    {
      "epoch": 0.46480562966676814,
      "grad_norm": 2.417525053024292,
      "learning_rate": 0.00011618451477447596,
      "loss": 1.5583,
      "step": 12979
    },
    {
      "epoch": 0.4648414417963364,
      "grad_norm": 2.8242523670196533,
      "learning_rate": 0.00011617306857214059,
      "loss": 1.6658,
      "step": 12980
    },
    {
      "epoch": 0.4648772539259047,
      "grad_norm": 2.044356107711792,
      "learning_rate": 0.00011616162215221744,
      "loss": 1.3755,
      "step": 12981
    },
    {
      "epoch": 0.464913066055473,
      "grad_norm": 1.7669847011566162,
      "learning_rate": 0.00011615017551486054,
      "loss": 1.6859,
      "step": 12982
    },
    {
      "epoch": 0.46494887818504127,
      "grad_norm": 1.8860310316085815,
      "learning_rate": 0.00011613872866022384,
      "loss": 1.4955,
      "step": 12983
    },
    {
      "epoch": 0.46498469031460954,
      "grad_norm": 1.8935774564743042,
      "learning_rate": 0.00011612728158846138,
      "loss": 1.7469,
      "step": 12984
    },
    {
      "epoch": 0.46502050244417786,
      "grad_norm": 1.7466204166412354,
      "learning_rate": 0.00011611583429972715,
      "loss": 1.3623,
      "step": 12985
    },
    {
      "epoch": 0.46505631457374613,
      "grad_norm": 2.1702635288238525,
      "learning_rate": 0.00011610438679417515,
      "loss": 1.4292,
      "step": 12986
    },
    {
      "epoch": 0.4650921267033144,
      "grad_norm": 1.6326416730880737,
      "learning_rate": 0.0001160929390719594,
      "loss": 1.5971,
      "step": 12987
    },
    {
      "epoch": 0.46512793883288267,
      "grad_norm": 3.18381667137146,
      "learning_rate": 0.00011608149113323392,
      "loss": 1.5523,
      "step": 12988
    },
    {
      "epoch": 0.465163750962451,
      "grad_norm": 2.2429869174957275,
      "learning_rate": 0.00011607004297815271,
      "loss": 1.2013,
      "step": 12989
    },
    {
      "epoch": 0.46519956309201926,
      "grad_norm": 1.3546289205551147,
      "learning_rate": 0.00011605859460686981,
      "loss": 1.4236,
      "step": 12990
    },
    {
      "epoch": 0.46523537522158753,
      "grad_norm": 1.5490727424621582,
      "learning_rate": 0.00011604714601953922,
      "loss": 1.567,
      "step": 12991
    },
    {
      "epoch": 0.46527118735115586,
      "grad_norm": 1.2240389585494995,
      "learning_rate": 0.00011603569721631499,
      "loss": 1.0347,
      "step": 12992
    },
    {
      "epoch": 0.4653069994807241,
      "grad_norm": 1.6459600925445557,
      "learning_rate": 0.00011602424819735111,
      "loss": 1.7073,
      "step": 12993
    },
    {
      "epoch": 0.4653428116102924,
      "grad_norm": 1.6901960372924805,
      "learning_rate": 0.00011601279896280167,
      "loss": 1.72,
      "step": 12994
    },
    {
      "epoch": 0.46537862373986066,
      "grad_norm": 1.3724249601364136,
      "learning_rate": 0.00011600134951282067,
      "loss": 1.6381,
      "step": 12995
    },
    {
      "epoch": 0.465414435869429,
      "grad_norm": 2.0271923542022705,
      "learning_rate": 0.00011598989984756216,
      "loss": 1.5456,
      "step": 12996
    },
    {
      "epoch": 0.46545024799899726,
      "grad_norm": 1.7223308086395264,
      "learning_rate": 0.0001159784499671802,
      "loss": 1.9994,
      "step": 12997
    },
    {
      "epoch": 0.4654860601285655,
      "grad_norm": 1.5626134872436523,
      "learning_rate": 0.00011596699987182873,
      "loss": 1.4624,
      "step": 12998
    },
    {
      "epoch": 0.46552187225813385,
      "grad_norm": 1.7725961208343506,
      "learning_rate": 0.00011595554956166195,
      "loss": 1.4648,
      "step": 12999
    },
    {
      "epoch": 0.4655576843877021,
      "grad_norm": 1.5563966035842896,
      "learning_rate": 0.00011594409903683376,
      "loss": 1.4602,
      "step": 13000
    },
    {
      "epoch": 0.4655934965172704,
      "grad_norm": 1.607021450996399,
      "learning_rate": 0.00011593264829749835,
      "loss": 1.4378,
      "step": 13001
    },
    {
      "epoch": 0.46562930864683866,
      "grad_norm": 1.4815516471862793,
      "learning_rate": 0.00011592119734380966,
      "loss": 1.705,
      "step": 13002
    },
    {
      "epoch": 0.465665120776407,
      "grad_norm": 1.9756717681884766,
      "learning_rate": 0.00011590974617592182,
      "loss": 1.6251,
      "step": 13003
    },
    {
      "epoch": 0.46570093290597525,
      "grad_norm": 1.837310791015625,
      "learning_rate": 0.00011589829479398886,
      "loss": 1.8303,
      "step": 13004
    },
    {
      "epoch": 0.4657367450355435,
      "grad_norm": 2.223160982131958,
      "learning_rate": 0.00011588684319816485,
      "loss": 1.3854,
      "step": 13005
    },
    {
      "epoch": 0.46577255716511184,
      "grad_norm": 1.3153750896453857,
      "learning_rate": 0.00011587539138860388,
      "loss": 1.4643,
      "step": 13006
    },
    {
      "epoch": 0.4658083692946801,
      "grad_norm": 1.9978464841842651,
      "learning_rate": 0.00011586393936545995,
      "loss": 1.6035,
      "step": 13007
    },
    {
      "epoch": 0.4658441814242484,
      "grad_norm": 1.5772230625152588,
      "learning_rate": 0.00011585248712888724,
      "loss": 1.7976,
      "step": 13008
    },
    {
      "epoch": 0.46587999355381665,
      "grad_norm": 1.839242935180664,
      "learning_rate": 0.0001158410346790397,
      "loss": 1.4875,
      "step": 13009
    },
    {
      "epoch": 0.465915805683385,
      "grad_norm": 1.7222756147384644,
      "learning_rate": 0.00011582958201607152,
      "loss": 1.562,
      "step": 13010
    },
    {
      "epoch": 0.46595161781295324,
      "grad_norm": 1.490675687789917,
      "learning_rate": 0.0001158181291401367,
      "loss": 1.5901,
      "step": 13011
    },
    {
      "epoch": 0.4659874299425215,
      "grad_norm": 1.3239779472351074,
      "learning_rate": 0.00011580667605138937,
      "loss": 1.3461,
      "step": 13012
    },
    {
      "epoch": 0.46602324207208984,
      "grad_norm": 1.3944867849349976,
      "learning_rate": 0.0001157952227499836,
      "loss": 1.8864,
      "step": 13013
    },
    {
      "epoch": 0.4660590542016581,
      "grad_norm": 1.853308916091919,
      "learning_rate": 0.0001157837692360735,
      "loss": 1.388,
      "step": 13014
    },
    {
      "epoch": 0.4660948663312264,
      "grad_norm": 1.9267886877059937,
      "learning_rate": 0.00011577231550981313,
      "loss": 1.7276,
      "step": 13015
    },
    {
      "epoch": 0.46613067846079465,
      "grad_norm": 1.8433860540390015,
      "learning_rate": 0.00011576086157135659,
      "loss": 1.4372,
      "step": 13016
    },
    {
      "epoch": 0.46616649059036297,
      "grad_norm": 1.6603800058364868,
      "learning_rate": 0.00011574940742085803,
      "loss": 1.5427,
      "step": 13017
    },
    {
      "epoch": 0.46620230271993124,
      "grad_norm": 1.883219838142395,
      "learning_rate": 0.00011573795305847146,
      "loss": 1.6654,
      "step": 13018
    },
    {
      "epoch": 0.4662381148494995,
      "grad_norm": 1.6327214241027832,
      "learning_rate": 0.00011572649848435104,
      "loss": 1.7289,
      "step": 13019
    },
    {
      "epoch": 0.46627392697906783,
      "grad_norm": 1.6758432388305664,
      "learning_rate": 0.00011571504369865087,
      "loss": 1.3519,
      "step": 13020
    },
    {
      "epoch": 0.4663097391086361,
      "grad_norm": 2.6810929775238037,
      "learning_rate": 0.0001157035887015251,
      "loss": 1.2821,
      "step": 13021
    },
    {
      "epoch": 0.46634555123820437,
      "grad_norm": 1.7383793592453003,
      "learning_rate": 0.00011569213349312773,
      "loss": 2.0079,
      "step": 13022
    },
    {
      "epoch": 0.46638136336777264,
      "grad_norm": 2.8930468559265137,
      "learning_rate": 0.000115680678073613,
      "loss": 1.5185,
      "step": 13023
    },
    {
      "epoch": 0.46641717549734096,
      "grad_norm": 1.9949698448181152,
      "learning_rate": 0.00011566922244313496,
      "loss": 1.3717,
      "step": 13024
    },
    {
      "epoch": 0.46645298762690923,
      "grad_norm": 1.4897794723510742,
      "learning_rate": 0.00011565776660184772,
      "loss": 1.3828,
      "step": 13025
    },
    {
      "epoch": 0.4664887997564775,
      "grad_norm": 1.995154619216919,
      "learning_rate": 0.00011564631054990546,
      "loss": 1.618,
      "step": 13026
    },
    {
      "epoch": 0.4665246118860458,
      "grad_norm": 1.3552685976028442,
      "learning_rate": 0.00011563485428746226,
      "loss": 1.4857,
      "step": 13027
    },
    {
      "epoch": 0.4665604240156141,
      "grad_norm": 1.694174885749817,
      "learning_rate": 0.00011562339781467226,
      "loss": 1.5152,
      "step": 13028
    },
    {
      "epoch": 0.46659623614518236,
      "grad_norm": 2.1866040229797363,
      "learning_rate": 0.00011561194113168958,
      "loss": 1.5322,
      "step": 13029
    },
    {
      "epoch": 0.46663204827475063,
      "grad_norm": 1.6240884065628052,
      "learning_rate": 0.0001156004842386684,
      "loss": 1.6096,
      "step": 13030
    },
    {
      "epoch": 0.46666786040431896,
      "grad_norm": 1.6429814100265503,
      "learning_rate": 0.0001155890271357628,
      "loss": 1.136,
      "step": 13031
    },
    {
      "epoch": 0.4667036725338872,
      "grad_norm": 1.7728595733642578,
      "learning_rate": 0.00011557756982312699,
      "loss": 1.627,
      "step": 13032
    },
    {
      "epoch": 0.4667394846634555,
      "grad_norm": 1.5901148319244385,
      "learning_rate": 0.00011556611230091502,
      "loss": 1.7406,
      "step": 13033
    },
    {
      "epoch": 0.4667752967930238,
      "grad_norm": 1.8100167512893677,
      "learning_rate": 0.00011555465456928114,
      "loss": 1.3819,
      "step": 13034
    },
    {
      "epoch": 0.4668111089225921,
      "grad_norm": 1.715928316116333,
      "learning_rate": 0.0001155431966283794,
      "loss": 1.0788,
      "step": 13035
    },
    {
      "epoch": 0.46684692105216036,
      "grad_norm": 1.3283747434616089,
      "learning_rate": 0.00011553173847836403,
      "loss": 1.3998,
      "step": 13036
    },
    {
      "epoch": 0.4668827331817286,
      "grad_norm": 2.5385360717773438,
      "learning_rate": 0.00011552028011938913,
      "loss": 1.2955,
      "step": 13037
    },
    {
      "epoch": 0.46691854531129695,
      "grad_norm": 1.693988561630249,
      "learning_rate": 0.0001155088215516089,
      "loss": 1.2398,
      "step": 13038
    },
    {
      "epoch": 0.4669543574408652,
      "grad_norm": 1.468666911125183,
      "learning_rate": 0.00011549736277517746,
      "loss": 1.5346,
      "step": 13039
    },
    {
      "epoch": 0.4669901695704335,
      "grad_norm": 1.6999561786651611,
      "learning_rate": 0.00011548590379024904,
      "loss": 1.4972,
      "step": 13040
    },
    {
      "epoch": 0.4670259817000018,
      "grad_norm": 2.801893711090088,
      "learning_rate": 0.00011547444459697772,
      "loss": 1.5316,
      "step": 13041
    },
    {
      "epoch": 0.4670617938295701,
      "grad_norm": 1.450239658355713,
      "learning_rate": 0.00011546298519551771,
      "loss": 1.4053,
      "step": 13042
    },
    {
      "epoch": 0.46709760595913835,
      "grad_norm": 1.53555428981781,
      "learning_rate": 0.00011545152558602319,
      "loss": 1.6854,
      "step": 13043
    },
    {
      "epoch": 0.4671334180887066,
      "grad_norm": 1.7407196760177612,
      "learning_rate": 0.00011544006576864832,
      "loss": 1.4238,
      "step": 13044
    },
    {
      "epoch": 0.46716923021827494,
      "grad_norm": 1.4221725463867188,
      "learning_rate": 0.00011542860574354727,
      "loss": 1.6033,
      "step": 13045
    },
    {
      "epoch": 0.4672050423478432,
      "grad_norm": 1.597596526145935,
      "learning_rate": 0.00011541714551087423,
      "loss": 1.5698,
      "step": 13046
    },
    {
      "epoch": 0.4672408544774115,
      "grad_norm": 1.6821784973144531,
      "learning_rate": 0.00011540568507078342,
      "loss": 1.4342,
      "step": 13047
    },
    {
      "epoch": 0.4672766666069798,
      "grad_norm": 1.4949378967285156,
      "learning_rate": 0.00011539422442342895,
      "loss": 1.3515,
      "step": 13048
    },
    {
      "epoch": 0.4673124787365481,
      "grad_norm": 1.5018175840377808,
      "learning_rate": 0.00011538276356896507,
      "loss": 1.4951,
      "step": 13049
    },
    {
      "epoch": 0.46734829086611634,
      "grad_norm": 1.6001633405685425,
      "learning_rate": 0.00011537130250754595,
      "loss": 1.5004,
      "step": 13050
    },
    {
      "epoch": 0.4673841029956846,
      "grad_norm": 1.7056968212127686,
      "learning_rate": 0.00011535984123932578,
      "loss": 1.2113,
      "step": 13051
    },
    {
      "epoch": 0.46741991512525294,
      "grad_norm": 1.4191452264785767,
      "learning_rate": 0.00011534837976445875,
      "loss": 1.1832,
      "step": 13052
    },
    {
      "epoch": 0.4674557272548212,
      "grad_norm": 1.751197338104248,
      "learning_rate": 0.00011533691808309905,
      "loss": 1.8554,
      "step": 13053
    },
    {
      "epoch": 0.4674915393843895,
      "grad_norm": 1.9440951347351074,
      "learning_rate": 0.00011532545619540094,
      "loss": 1.2174,
      "step": 13054
    },
    {
      "epoch": 0.4675273515139578,
      "grad_norm": 1.4406771659851074,
      "learning_rate": 0.00011531399410151855,
      "loss": 1.505,
      "step": 13055
    },
    {
      "epoch": 0.46756316364352607,
      "grad_norm": 1.4604896306991577,
      "learning_rate": 0.00011530253180160614,
      "loss": 1.5097,
      "step": 13056
    },
    {
      "epoch": 0.46759897577309434,
      "grad_norm": 1.3812698125839233,
      "learning_rate": 0.00011529106929581792,
      "loss": 0.9882,
      "step": 13057
    },
    {
      "epoch": 0.4676347879026626,
      "grad_norm": 1.4525638818740845,
      "learning_rate": 0.00011527960658430807,
      "loss": 1.4344,
      "step": 13058
    },
    {
      "epoch": 0.46767060003223093,
      "grad_norm": 1.5971357822418213,
      "learning_rate": 0.00011526814366723084,
      "loss": 1.4928,
      "step": 13059
    },
    {
      "epoch": 0.4677064121617992,
      "grad_norm": 2.0160109996795654,
      "learning_rate": 0.00011525668054474039,
      "loss": 1.2046,
      "step": 13060
    },
    {
      "epoch": 0.46774222429136747,
      "grad_norm": 1.648202657699585,
      "learning_rate": 0.00011524521721699102,
      "loss": 1.3332,
      "step": 13061
    },
    {
      "epoch": 0.4677780364209358,
      "grad_norm": 1.7077761888504028,
      "learning_rate": 0.0001152337536841369,
      "loss": 1.5201,
      "step": 13062
    },
    {
      "epoch": 0.46781384855050406,
      "grad_norm": 2.0169153213500977,
      "learning_rate": 0.00011522228994633229,
      "loss": 1.6287,
      "step": 13063
    },
    {
      "epoch": 0.46784966068007233,
      "grad_norm": 1.86139976978302,
      "learning_rate": 0.00011521082600373136,
      "loss": 1.4691,
      "step": 13064
    },
    {
      "epoch": 0.4678854728096406,
      "grad_norm": 1.6686205863952637,
      "learning_rate": 0.00011519936185648842,
      "loss": 1.3001,
      "step": 13065
    },
    {
      "epoch": 0.4679212849392089,
      "grad_norm": 1.5188180208206177,
      "learning_rate": 0.0001151878975047577,
      "loss": 1.4057,
      "step": 13066
    },
    {
      "epoch": 0.4679570970687772,
      "grad_norm": 1.5588150024414062,
      "learning_rate": 0.00011517643294869339,
      "loss": 1.4067,
      "step": 13067
    },
    {
      "epoch": 0.46799290919834546,
      "grad_norm": 1.6581311225891113,
      "learning_rate": 0.00011516496818844972,
      "loss": 1.6517,
      "step": 13068
    },
    {
      "epoch": 0.4680287213279138,
      "grad_norm": 1.5204647779464722,
      "learning_rate": 0.000115153503224181,
      "loss": 1.6362,
      "step": 13069
    },
    {
      "epoch": 0.46806453345748206,
      "grad_norm": 1.700719952583313,
      "learning_rate": 0.00011514203805604142,
      "loss": 1.2189,
      "step": 13070
    },
    {
      "epoch": 0.4681003455870503,
      "grad_norm": 2.6246724128723145,
      "learning_rate": 0.00011513057268418526,
      "loss": 1.6423,
      "step": 13071
    },
    {
      "epoch": 0.4681361577166186,
      "grad_norm": 1.727502465248108,
      "learning_rate": 0.00011511910710876677,
      "loss": 1.7205,
      "step": 13072
    },
    {
      "epoch": 0.4681719698461869,
      "grad_norm": 1.5376635789871216,
      "learning_rate": 0.00011510764132994016,
      "loss": 1.5014,
      "step": 13073
    },
    {
      "epoch": 0.4682077819757552,
      "grad_norm": 1.3821709156036377,
      "learning_rate": 0.00011509617534785976,
      "loss": 1.5177,
      "step": 13074
    },
    {
      "epoch": 0.46824359410532346,
      "grad_norm": 1.698994755744934,
      "learning_rate": 0.00011508470916267978,
      "loss": 1.4925,
      "step": 13075
    },
    {
      "epoch": 0.4682794062348918,
      "grad_norm": 1.7280097007751465,
      "learning_rate": 0.00011507324277455452,
      "loss": 1.3431,
      "step": 13076
    },
    {
      "epoch": 0.46831521836446005,
      "grad_norm": 1.7975046634674072,
      "learning_rate": 0.00011506177618363818,
      "loss": 1.5999,
      "step": 13077
    },
    {
      "epoch": 0.4683510304940283,
      "grad_norm": 1.7372099161148071,
      "learning_rate": 0.00011505030939008508,
      "loss": 1.3853,
      "step": 13078
    },
    {
      "epoch": 0.4683868426235966,
      "grad_norm": 1.693571925163269,
      "learning_rate": 0.0001150388423940495,
      "loss": 1.3432,
      "step": 13079
    },
    {
      "epoch": 0.4684226547531649,
      "grad_norm": 1.4951120615005493,
      "learning_rate": 0.00011502737519568567,
      "loss": 1.7497,
      "step": 13080
    },
    {
      "epoch": 0.4684584668827332,
      "grad_norm": 1.509081482887268,
      "learning_rate": 0.00011501590779514793,
      "loss": 1.5924,
      "step": 13081
    },
    {
      "epoch": 0.46849427901230145,
      "grad_norm": 2.52064847946167,
      "learning_rate": 0.00011500444019259047,
      "loss": 1.4573,
      "step": 13082
    },
    {
      "epoch": 0.4685300911418698,
      "grad_norm": 1.4621198177337646,
      "learning_rate": 0.00011499297238816767,
      "loss": 1.5936,
      "step": 13083
    },
    {
      "epoch": 0.46856590327143804,
      "grad_norm": 1.402459979057312,
      "learning_rate": 0.00011498150438203373,
      "loss": 1.7086,
      "step": 13084
    },
    {
      "epoch": 0.4686017154010063,
      "grad_norm": 1.5725468397140503,
      "learning_rate": 0.00011497003617434301,
      "loss": 1.4823,
      "step": 13085
    },
    {
      "epoch": 0.4686375275305746,
      "grad_norm": 2.153813123703003,
      "learning_rate": 0.00011495856776524971,
      "loss": 1.2271,
      "step": 13086
    },
    {
      "epoch": 0.4686733396601429,
      "grad_norm": 1.4646168947219849,
      "learning_rate": 0.00011494709915490822,
      "loss": 1.4015,
      "step": 13087
    },
    {
      "epoch": 0.4687091517897112,
      "grad_norm": 1.6754770278930664,
      "learning_rate": 0.00011493563034347277,
      "loss": 1.5845,
      "step": 13088
    },
    {
      "epoch": 0.46874496391927944,
      "grad_norm": 1.6596485376358032,
      "learning_rate": 0.00011492416133109769,
      "loss": 1.3597,
      "step": 13089
    },
    {
      "epoch": 0.46878077604884777,
      "grad_norm": 1.3753113746643066,
      "learning_rate": 0.00011491269211793725,
      "loss": 1.4336,
      "step": 13090
    },
    {
      "epoch": 0.46881658817841604,
      "grad_norm": 1.4884370565414429,
      "learning_rate": 0.00011490122270414578,
      "loss": 1.3863,
      "step": 13091
    },
    {
      "epoch": 0.4688524003079843,
      "grad_norm": 2.0943214893341064,
      "learning_rate": 0.0001148897530898776,
      "loss": 1.1964,
      "step": 13092
    },
    {
      "epoch": 0.4688882124375526,
      "grad_norm": 1.4185950756072998,
      "learning_rate": 0.000114878283275287,
      "loss": 1.3801,
      "step": 13093
    },
    {
      "epoch": 0.4689240245671209,
      "grad_norm": 1.359785795211792,
      "learning_rate": 0.00011486681326052828,
      "loss": 1.501,
      "step": 13094
    },
    {
      "epoch": 0.46895983669668917,
      "grad_norm": 1.7992392778396606,
      "learning_rate": 0.00011485534304575575,
      "loss": 1.6647,
      "step": 13095
    },
    {
      "epoch": 0.46899564882625744,
      "grad_norm": 1.468299150466919,
      "learning_rate": 0.00011484387263112377,
      "loss": 1.3893,
      "step": 13096
    },
    {
      "epoch": 0.46903146095582576,
      "grad_norm": 1.3851277828216553,
      "learning_rate": 0.0001148324020167866,
      "loss": 1.0484,
      "step": 13097
    },
    {
      "epoch": 0.46906727308539403,
      "grad_norm": 2.560601234436035,
      "learning_rate": 0.0001148209312028986,
      "loss": 1.1958,
      "step": 13098
    },
    {
      "epoch": 0.4691030852149623,
      "grad_norm": 1.7320241928100586,
      "learning_rate": 0.0001148094601896141,
      "loss": 1.3779,
      "step": 13099
    },
    {
      "epoch": 0.46913889734453057,
      "grad_norm": 2.7635598182678223,
      "learning_rate": 0.00011479798897708742,
      "loss": 1.9673,
      "step": 13100
    },
    {
      "epoch": 0.4691747094740989,
      "grad_norm": 2.2257347106933594,
      "learning_rate": 0.00011478651756547287,
      "loss": 1.7453,
      "step": 13101
    },
    {
      "epoch": 0.46921052160366716,
      "grad_norm": 1.73111891746521,
      "learning_rate": 0.00011477504595492481,
      "loss": 1.1503,
      "step": 13102
    },
    {
      "epoch": 0.46924633373323543,
      "grad_norm": 1.4754983186721802,
      "learning_rate": 0.00011476357414559757,
      "loss": 1.5525,
      "step": 13103
    },
    {
      "epoch": 0.46928214586280376,
      "grad_norm": 3.5211539268493652,
      "learning_rate": 0.00011475210213764547,
      "loss": 1.7384,
      "step": 13104
    },
    {
      "epoch": 0.469317957992372,
      "grad_norm": 1.7858822345733643,
      "learning_rate": 0.00011474062993122288,
      "loss": 1.4745,
      "step": 13105
    },
    {
      "epoch": 0.4693537701219403,
      "grad_norm": 1.6460684537887573,
      "learning_rate": 0.0001147291575264841,
      "loss": 1.7301,
      "step": 13106
    },
    {
      "epoch": 0.46938958225150856,
      "grad_norm": 1.3619961738586426,
      "learning_rate": 0.00011471768492358354,
      "loss": 1.2949,
      "step": 13107
    },
    {
      "epoch": 0.4694253943810769,
      "grad_norm": 1.6338173151016235,
      "learning_rate": 0.00011470621212267547,
      "loss": 1.3691,
      "step": 13108
    },
    {
      "epoch": 0.46946120651064516,
      "grad_norm": 1.5736229419708252,
      "learning_rate": 0.00011469473912391433,
      "loss": 1.38,
      "step": 13109
    },
    {
      "epoch": 0.4694970186402134,
      "grad_norm": 1.666130542755127,
      "learning_rate": 0.0001146832659274544,
      "loss": 1.5988,
      "step": 13110
    },
    {
      "epoch": 0.46953283076978175,
      "grad_norm": 2.119744062423706,
      "learning_rate": 0.00011467179253345008,
      "loss": 1.6468,
      "step": 13111
    },
    {
      "epoch": 0.46956864289935,
      "grad_norm": 2.0681087970733643,
      "learning_rate": 0.00011466031894205574,
      "loss": 1.404,
      "step": 13112
    },
    {
      "epoch": 0.4696044550289183,
      "grad_norm": 1.5972648859024048,
      "learning_rate": 0.00011464884515342568,
      "loss": 1.4593,
      "step": 13113
    },
    {
      "epoch": 0.46964026715848656,
      "grad_norm": 1.4360387325286865,
      "learning_rate": 0.00011463737116771434,
      "loss": 1.2351,
      "step": 13114
    },
    {
      "epoch": 0.4696760792880549,
      "grad_norm": 1.9295847415924072,
      "learning_rate": 0.00011462589698507603,
      "loss": 1.6692,
      "step": 13115
    },
    {
      "epoch": 0.46971189141762315,
      "grad_norm": 1.6461082696914673,
      "learning_rate": 0.00011461442260566513,
      "loss": 1.2059,
      "step": 13116
    },
    {
      "epoch": 0.4697477035471914,
      "grad_norm": 2.095632791519165,
      "learning_rate": 0.00011460294802963602,
      "loss": 1.2712,
      "step": 13117
    },
    {
      "epoch": 0.46978351567675974,
      "grad_norm": 2.0552680492401123,
      "learning_rate": 0.00011459147325714312,
      "loss": 1.4689,
      "step": 13118
    },
    {
      "epoch": 0.469819327806328,
      "grad_norm": 1.3072434663772583,
      "learning_rate": 0.00011457999828834073,
      "loss": 1.5075,
      "step": 13119
    },
    {
      "epoch": 0.4698551399358963,
      "grad_norm": 2.8187243938446045,
      "learning_rate": 0.00011456852312338331,
      "loss": 1.3692,
      "step": 13120
    },
    {
      "epoch": 0.46989095206546455,
      "grad_norm": 1.7331218719482422,
      "learning_rate": 0.00011455704776242517,
      "loss": 1.6142,
      "step": 13121
    },
    {
      "epoch": 0.4699267641950329,
      "grad_norm": 1.6014031171798706,
      "learning_rate": 0.00011454557220562074,
      "loss": 1.3266,
      "step": 13122
    },
    {
      "epoch": 0.46996257632460114,
      "grad_norm": 1.5360310077667236,
      "learning_rate": 0.0001145340964531244,
      "loss": 1.4116,
      "step": 13123
    },
    {
      "epoch": 0.4699983884541694,
      "grad_norm": 1.7113350629806519,
      "learning_rate": 0.00011452262050509053,
      "loss": 1.3728,
      "step": 13124
    },
    {
      "epoch": 0.47003420058373774,
      "grad_norm": 2.2795071601867676,
      "learning_rate": 0.00011451114436167356,
      "loss": 1.5459,
      "step": 13125
    },
    {
      "epoch": 0.470070012713306,
      "grad_norm": 1.575621247291565,
      "learning_rate": 0.00011449966802302783,
      "loss": 1.7569,
      "step": 13126
    },
    {
      "epoch": 0.4701058248428743,
      "grad_norm": 2.293635129928589,
      "learning_rate": 0.0001144881914893078,
      "loss": 1.1277,
      "step": 13127
    },
    {
      "epoch": 0.47014163697244254,
      "grad_norm": 1.592656135559082,
      "learning_rate": 0.00011447671476066781,
      "loss": 1.4337,
      "step": 13128
    },
    {
      "epoch": 0.47017744910201087,
      "grad_norm": 1.407986044883728,
      "learning_rate": 0.00011446523783726235,
      "loss": 1.4221,
      "step": 13129
    },
    {
      "epoch": 0.47021326123157914,
      "grad_norm": 1.9266597032546997,
      "learning_rate": 0.00011445376071924572,
      "loss": 1.4518,
      "step": 13130
    },
    {
      "epoch": 0.4702490733611474,
      "grad_norm": 1.3258838653564453,
      "learning_rate": 0.00011444228340677241,
      "loss": 1.705,
      "step": 13131
    },
    {
      "epoch": 0.47028488549071573,
      "grad_norm": 1.457871675491333,
      "learning_rate": 0.00011443080589999677,
      "loss": 1.695,
      "step": 13132
    },
    {
      "epoch": 0.470320697620284,
      "grad_norm": 1.8480134010314941,
      "learning_rate": 0.00011441932819907328,
      "loss": 1.5101,
      "step": 13133
    },
    {
      "epoch": 0.47035650974985227,
      "grad_norm": 2.691573143005371,
      "learning_rate": 0.00011440785030415633,
      "loss": 1.2265,
      "step": 13134
    },
    {
      "epoch": 0.47039232187942054,
      "grad_norm": 1.8243192434310913,
      "learning_rate": 0.00011439637221540031,
      "loss": 1.8265,
      "step": 13135
    },
    {
      "epoch": 0.47042813400898886,
      "grad_norm": 1.973116397857666,
      "learning_rate": 0.00011438489393295973,
      "loss": 1.5911,
      "step": 13136
    },
    {
      "epoch": 0.47046394613855713,
      "grad_norm": 1.8532606363296509,
      "learning_rate": 0.00011437341545698892,
      "loss": 1.4446,
      "step": 13137
    },
    {
      "epoch": 0.4704997582681254,
      "grad_norm": 1.6671091318130493,
      "learning_rate": 0.00011436193678764236,
      "loss": 1.3737,
      "step": 13138
    },
    {
      "epoch": 0.4705355703976937,
      "grad_norm": 1.4185718297958374,
      "learning_rate": 0.00011435045792507443,
      "loss": 1.4509,
      "step": 13139
    },
    {
      "epoch": 0.470571382527262,
      "grad_norm": 1.73945152759552,
      "learning_rate": 0.00011433897886943965,
      "loss": 1.8183,
      "step": 13140
    },
    {
      "epoch": 0.47060719465683026,
      "grad_norm": 2.0083441734313965,
      "learning_rate": 0.00011432749962089235,
      "loss": 1.1479,
      "step": 13141
    },
    {
      "epoch": 0.47064300678639853,
      "grad_norm": 2.459001064300537,
      "learning_rate": 0.00011431602017958707,
      "loss": 1.3052,
      "step": 13142
    },
    {
      "epoch": 0.47067881891596686,
      "grad_norm": 2.419342041015625,
      "learning_rate": 0.00011430454054567819,
      "loss": 1.6516,
      "step": 13143
    },
    {
      "epoch": 0.4707146310455351,
      "grad_norm": 1.7359721660614014,
      "learning_rate": 0.00011429306071932018,
      "loss": 1.2583,
      "step": 13144
    },
    {
      "epoch": 0.4707504431751034,
      "grad_norm": 1.5329176187515259,
      "learning_rate": 0.00011428158070066743,
      "loss": 1.6391,
      "step": 13145
    },
    {
      "epoch": 0.4707862553046717,
      "grad_norm": 2.4268431663513184,
      "learning_rate": 0.00011427010048987448,
      "loss": 1.8331,
      "step": 13146
    },
    {
      "epoch": 0.47082206743424,
      "grad_norm": 2.1451423168182373,
      "learning_rate": 0.00011425862008709574,
      "loss": 1.4687,
      "step": 13147
    },
    {
      "epoch": 0.47085787956380826,
      "grad_norm": 1.776679515838623,
      "learning_rate": 0.00011424713949248562,
      "loss": 1.4487,
      "step": 13148
    },
    {
      "epoch": 0.4708936916933765,
      "grad_norm": 1.735393762588501,
      "learning_rate": 0.00011423565870619863,
      "loss": 1.439,
      "step": 13149
    },
    {
      "epoch": 0.47092950382294485,
      "grad_norm": 1.2827190160751343,
      "learning_rate": 0.00011422417772838923,
      "loss": 1.2839,
      "step": 13150
    },
    {
      "epoch": 0.4709653159525131,
      "grad_norm": 1.7964860200881958,
      "learning_rate": 0.00011421269655921185,
      "loss": 1.345,
      "step": 13151
    },
    {
      "epoch": 0.4710011280820814,
      "grad_norm": 1.468343734741211,
      "learning_rate": 0.00011420121519882096,
      "loss": 1.4463,
      "step": 13152
    },
    {
      "epoch": 0.4710369402116497,
      "grad_norm": 1.5348808765411377,
      "learning_rate": 0.00011418973364737107,
      "loss": 1.5609,
      "step": 13153
    },
    {
      "epoch": 0.471072752341218,
      "grad_norm": 1.6379225254058838,
      "learning_rate": 0.00011417825190501658,
      "loss": 1.3981,
      "step": 13154
    },
    {
      "epoch": 0.47110856447078625,
      "grad_norm": 1.5981770753860474,
      "learning_rate": 0.00011416676997191205,
      "loss": 1.5883,
      "step": 13155
    },
    {
      "epoch": 0.4711443766003545,
      "grad_norm": 1.8267465829849243,
      "learning_rate": 0.00011415528784821188,
      "loss": 1.6164,
      "step": 13156
    },
    {
      "epoch": 0.47118018872992284,
      "grad_norm": 1.5526596307754517,
      "learning_rate": 0.00011414380553407055,
      "loss": 1.382,
      "step": 13157
    },
    {
      "epoch": 0.4712160008594911,
      "grad_norm": 2.0587666034698486,
      "learning_rate": 0.00011413232302964258,
      "loss": 1.7454,
      "step": 13158
    },
    {
      "epoch": 0.4712518129890594,
      "grad_norm": 1.841937780380249,
      "learning_rate": 0.00011412084033508242,
      "loss": 1.4733,
      "step": 13159
    },
    {
      "epoch": 0.4712876251186277,
      "grad_norm": 2.0795516967773438,
      "learning_rate": 0.00011410935745054459,
      "loss": 1.391,
      "step": 13160
    },
    {
      "epoch": 0.471323437248196,
      "grad_norm": 1.4707566499710083,
      "learning_rate": 0.00011409787437618353,
      "loss": 1.6325,
      "step": 13161
    },
    {
      "epoch": 0.47135924937776424,
      "grad_norm": 2.5612967014312744,
      "learning_rate": 0.00011408639111215378,
      "loss": 1.6126,
      "step": 13162
    },
    {
      "epoch": 0.4713950615073325,
      "grad_norm": 1.5790444612503052,
      "learning_rate": 0.00011407490765860978,
      "loss": 1.395,
      "step": 13163
    },
    {
      "epoch": 0.47143087363690084,
      "grad_norm": 1.65281081199646,
      "learning_rate": 0.00011406342401570609,
      "loss": 1.5043,
      "step": 13164
    },
    {
      "epoch": 0.4714666857664691,
      "grad_norm": 1.3939710855484009,
      "learning_rate": 0.00011405194018359715,
      "loss": 1.4143,
      "step": 13165
    },
    {
      "epoch": 0.4715024978960374,
      "grad_norm": 1.7264175415039062,
      "learning_rate": 0.00011404045616243745,
      "loss": 1.6513,
      "step": 13166
    },
    {
      "epoch": 0.4715383100256057,
      "grad_norm": 1.5612338781356812,
      "learning_rate": 0.00011402897195238158,
      "loss": 1.8289,
      "step": 13167
    },
    {
      "epoch": 0.47157412215517397,
      "grad_norm": 1.8469725847244263,
      "learning_rate": 0.00011401748755358395,
      "loss": 1.3954,
      "step": 13168
    },
    {
      "epoch": 0.47160993428474224,
      "grad_norm": 1.5684376955032349,
      "learning_rate": 0.00011400600296619912,
      "loss": 1.5721,
      "step": 13169
    },
    {
      "epoch": 0.4716457464143105,
      "grad_norm": 1.6736838817596436,
      "learning_rate": 0.00011399451819038159,
      "loss": 1.3776,
      "step": 13170
    },
    {
      "epoch": 0.47168155854387883,
      "grad_norm": 1.5765982866287231,
      "learning_rate": 0.00011398303322628585,
      "loss": 1.5997,
      "step": 13171
    },
    {
      "epoch": 0.4717173706734471,
      "grad_norm": 1.5754213333129883,
      "learning_rate": 0.00011397154807406645,
      "loss": 1.3318,
      "step": 13172
    },
    {
      "epoch": 0.47175318280301537,
      "grad_norm": 1.5405632257461548,
      "learning_rate": 0.00011396006273387792,
      "loss": 1.2578,
      "step": 13173
    },
    {
      "epoch": 0.47178899493258364,
      "grad_norm": 1.6634725332260132,
      "learning_rate": 0.0001139485772058747,
      "loss": 1.7118,
      "step": 13174
    },
    {
      "epoch": 0.47182480706215196,
      "grad_norm": 1.8009767532348633,
      "learning_rate": 0.0001139370914902114,
      "loss": 1.5235,
      "step": 13175
    },
    {
      "epoch": 0.47186061919172023,
      "grad_norm": 2.3448922634124756,
      "learning_rate": 0.00011392560558704249,
      "loss": 1.4457,
      "step": 13176
    },
    {
      "epoch": 0.4718964313212885,
      "grad_norm": 1.6356074810028076,
      "learning_rate": 0.00011391411949652253,
      "loss": 1.4746,
      "step": 13177
    },
    {
      "epoch": 0.4719322434508568,
      "grad_norm": 2.3725569248199463,
      "learning_rate": 0.00011390263321880605,
      "loss": 1.4069,
      "step": 13178
    },
    {
      "epoch": 0.4719680555804251,
      "grad_norm": 1.9027031660079956,
      "learning_rate": 0.00011389114675404755,
      "loss": 1.6479,
      "step": 13179
    },
    {
      "epoch": 0.47200386770999336,
      "grad_norm": 1.7592540979385376,
      "learning_rate": 0.00011387966010240161,
      "loss": 1.5657,
      "step": 13180
    },
    {
      "epoch": 0.47203967983956163,
      "grad_norm": 3.3529903888702393,
      "learning_rate": 0.00011386817326402273,
      "loss": 1.5566,
      "step": 13181
    },
    {
      "epoch": 0.47207549196912996,
      "grad_norm": 2.228374481201172,
      "learning_rate": 0.00011385668623906551,
      "loss": 1.553,
      "step": 13182
    },
    {
      "epoch": 0.4721113040986982,
      "grad_norm": 1.72072434425354,
      "learning_rate": 0.00011384519902768441,
      "loss": 1.503,
      "step": 13183
    },
    {
      "epoch": 0.4721471162282665,
      "grad_norm": 1.6252473592758179,
      "learning_rate": 0.00011383371163003403,
      "loss": 1.2247,
      "step": 13184
    },
    {
      "epoch": 0.4721829283578348,
      "grad_norm": 2.5225577354431152,
      "learning_rate": 0.00011382222404626888,
      "loss": 1.4804,
      "step": 13185
    },
    {
      "epoch": 0.4722187404874031,
      "grad_norm": 1.921576738357544,
      "learning_rate": 0.00011381073627654357,
      "loss": 1.7643,
      "step": 13186
    },
    {
      "epoch": 0.47225455261697136,
      "grad_norm": 1.2695116996765137,
      "learning_rate": 0.00011379924832101258,
      "loss": 1.6016,
      "step": 13187
    },
    {
      "epoch": 0.4722903647465396,
      "grad_norm": 1.5697298049926758,
      "learning_rate": 0.00011378776017983053,
      "loss": 1.5873,
      "step": 13188
    },
    {
      "epoch": 0.47232617687610795,
      "grad_norm": 1.467513084411621,
      "learning_rate": 0.00011377627185315194,
      "loss": 1.318,
      "step": 13189
    },
    {
      "epoch": 0.4723619890056762,
      "grad_norm": 1.839012861251831,
      "learning_rate": 0.00011376478334113139,
      "loss": 1.3583,
      "step": 13190
    },
    {
      "epoch": 0.4723978011352445,
      "grad_norm": 1.906954050064087,
      "learning_rate": 0.00011375329464392343,
      "loss": 1.7615,
      "step": 13191
    },
    {
      "epoch": 0.4724336132648128,
      "grad_norm": 2.299314022064209,
      "learning_rate": 0.00011374180576168263,
      "loss": 1.3546,
      "step": 13192
    },
    {
      "epoch": 0.4724694253943811,
      "grad_norm": 1.6634674072265625,
      "learning_rate": 0.00011373031669456358,
      "loss": 1.3662,
      "step": 13193
    },
    {
      "epoch": 0.47250523752394935,
      "grad_norm": 1.2625876665115356,
      "learning_rate": 0.0001137188274427208,
      "loss": 1.4021,
      "step": 13194
    },
    {
      "epoch": 0.4725410496535176,
      "grad_norm": 1.9035825729370117,
      "learning_rate": 0.00011370733800630892,
      "loss": 1.4559,
      "step": 13195
    },
    {
      "epoch": 0.47257686178308594,
      "grad_norm": 1.5406285524368286,
      "learning_rate": 0.00011369584838548246,
      "loss": 1.3728,
      "step": 13196
    },
    {
      "epoch": 0.4726126739126542,
      "grad_norm": 1.461430311203003,
      "learning_rate": 0.00011368435858039605,
      "loss": 1.3152,
      "step": 13197
    },
    {
      "epoch": 0.4726484860422225,
      "grad_norm": 1.4157794713974,
      "learning_rate": 0.00011367286859120423,
      "loss": 1.328,
      "step": 13198
    },
    {
      "epoch": 0.4726842981717908,
      "grad_norm": 1.5689760446548462,
      "learning_rate": 0.00011366137841806161,
      "loss": 1.5654,
      "step": 13199
    },
    {
      "epoch": 0.4727201103013591,
      "grad_norm": 1.5379902124404907,
      "learning_rate": 0.00011364988806112278,
      "loss": 1.5379,
      "step": 13200
    },
    {
      "epoch": 0.47275592243092734,
      "grad_norm": 2.011894941329956,
      "learning_rate": 0.00011363839752054228,
      "loss": 1.8056,
      "step": 13201
    },
    {
      "epoch": 0.4727917345604956,
      "grad_norm": 1.3880537748336792,
      "learning_rate": 0.00011362690679647477,
      "loss": 1.5678,
      "step": 13202
    },
    {
      "epoch": 0.47282754669006394,
      "grad_norm": 1.6055556535720825,
      "learning_rate": 0.00011361541588907477,
      "loss": 1.6305,
      "step": 13203
    },
    {
      "epoch": 0.4728633588196322,
      "grad_norm": 1.4604688882827759,
      "learning_rate": 0.00011360392479849693,
      "loss": 1.4214,
      "step": 13204
    },
    {
      "epoch": 0.4728991709492005,
      "grad_norm": 1.9168239831924438,
      "learning_rate": 0.00011359243352489581,
      "loss": 1.7095,
      "step": 13205
    },
    {
      "epoch": 0.4729349830787688,
      "grad_norm": 1.8139801025390625,
      "learning_rate": 0.00011358094206842607,
      "loss": 1.5206,
      "step": 13206
    },
    {
      "epoch": 0.47297079520833707,
      "grad_norm": 1.9469952583312988,
      "learning_rate": 0.00011356945042924223,
      "loss": 1.613,
      "step": 13207
    },
    {
      "epoch": 0.47300660733790534,
      "grad_norm": 1.7959978580474854,
      "learning_rate": 0.00011355795860749899,
      "loss": 1.53,
      "step": 13208
    },
    {
      "epoch": 0.4730424194674736,
      "grad_norm": 1.6030470132827759,
      "learning_rate": 0.00011354646660335086,
      "loss": 1.4548,
      "step": 13209
    },
    {
      "epoch": 0.47307823159704193,
      "grad_norm": 2.472611904144287,
      "learning_rate": 0.00011353497441695251,
      "loss": 1.3722,
      "step": 13210
    },
    {
      "epoch": 0.4731140437266102,
      "grad_norm": 1.9840824604034424,
      "learning_rate": 0.00011352348204845853,
      "loss": 1.2221,
      "step": 13211
    },
    {
      "epoch": 0.47314985585617847,
      "grad_norm": 1.8577934503555298,
      "learning_rate": 0.00011351198949802355,
      "loss": 1.0622,
      "step": 13212
    },
    {
      "epoch": 0.4731856679857468,
      "grad_norm": 1.9593387842178345,
      "learning_rate": 0.0001135004967658022,
      "loss": 1.6045,
      "step": 13213
    },
    {
      "epoch": 0.47322148011531506,
      "grad_norm": 1.7175911664962769,
      "learning_rate": 0.00011348900385194903,
      "loss": 1.5449,
      "step": 13214
    },
    {
      "epoch": 0.47325729224488333,
      "grad_norm": 1.7575346231460571,
      "learning_rate": 0.00011347751075661876,
      "loss": 1.5037,
      "step": 13215
    },
    {
      "epoch": 0.4732931043744516,
      "grad_norm": 2.017416477203369,
      "learning_rate": 0.00011346601747996595,
      "loss": 1.57,
      "step": 13216
    },
    {
      "epoch": 0.4733289165040199,
      "grad_norm": 1.4869771003723145,
      "learning_rate": 0.00011345452402214527,
      "loss": 1.5248,
      "step": 13217
    },
    {
      "epoch": 0.4733647286335882,
      "grad_norm": 1.63876473903656,
      "learning_rate": 0.0001134430303833113,
      "loss": 1.5725,
      "step": 13218
    },
    {
      "epoch": 0.47340054076315646,
      "grad_norm": 1.4624106884002686,
      "learning_rate": 0.00011343153656361867,
      "loss": 1.3269,
      "step": 13219
    },
    {
      "epoch": 0.4734363528927248,
      "grad_norm": 2.3383195400238037,
      "learning_rate": 0.00011342004256322208,
      "loss": 1.415,
      "step": 13220
    },
    {
      "epoch": 0.47347216502229306,
      "grad_norm": 1.4437799453735352,
      "learning_rate": 0.00011340854838227611,
      "loss": 1.6656,
      "step": 13221
    },
    {
      "epoch": 0.4735079771518613,
      "grad_norm": 1.8068561553955078,
      "learning_rate": 0.00011339705402093543,
      "loss": 1.8916,
      "step": 13222
    },
    {
      "epoch": 0.4735437892814296,
      "grad_norm": 2.249796152114868,
      "learning_rate": 0.00011338555947935465,
      "loss": 1.6621,
      "step": 13223
    },
    {
      "epoch": 0.4735796014109979,
      "grad_norm": 1.4828672409057617,
      "learning_rate": 0.00011337406475768846,
      "loss": 1.7506,
      "step": 13224
    },
    {
      "epoch": 0.4736154135405662,
      "grad_norm": 1.364309310913086,
      "learning_rate": 0.00011336256985609144,
      "loss": 1.5618,
      "step": 13225
    },
    {
      "epoch": 0.47365122567013446,
      "grad_norm": 1.287990689277649,
      "learning_rate": 0.00011335107477471834,
      "loss": 1.3136,
      "step": 13226
    },
    {
      "epoch": 0.4736870377997028,
      "grad_norm": 2.1887640953063965,
      "learning_rate": 0.00011333957951372372,
      "loss": 1.2726,
      "step": 13227
    },
    {
      "epoch": 0.47372284992927105,
      "grad_norm": 1.8509095907211304,
      "learning_rate": 0.00011332808407326225,
      "loss": 1.3898,
      "step": 13228
    },
    {
      "epoch": 0.4737586620588393,
      "grad_norm": 1.4751075506210327,
      "learning_rate": 0.0001133165884534886,
      "loss": 1.7061,
      "step": 13229
    },
    {
      "epoch": 0.4737944741884076,
      "grad_norm": 1.6264978647232056,
      "learning_rate": 0.00011330509265455745,
      "loss": 1.5171,
      "step": 13230
    },
    {
      "epoch": 0.4738302863179759,
      "grad_norm": 1.8143638372421265,
      "learning_rate": 0.00011329359667662342,
      "loss": 1.2657,
      "step": 13231
    },
    {
      "epoch": 0.4738660984475442,
      "grad_norm": 1.415449619293213,
      "learning_rate": 0.00011328210051984118,
      "loss": 1.6253,
      "step": 13232
    },
    {
      "epoch": 0.47390191057711245,
      "grad_norm": 1.2469733953475952,
      "learning_rate": 0.00011327060418436545,
      "loss": 1.4911,
      "step": 13233
    },
    {
      "epoch": 0.4739377227066808,
      "grad_norm": 1.2812349796295166,
      "learning_rate": 0.00011325910767035086,
      "loss": 1.2655,
      "step": 13234
    },
    {
      "epoch": 0.47397353483624904,
      "grad_norm": 1.9594454765319824,
      "learning_rate": 0.00011324761097795206,
      "loss": 1.7503,
      "step": 13235
    },
    {
      "epoch": 0.4740093469658173,
      "grad_norm": 1.5524890422821045,
      "learning_rate": 0.00011323611410732375,
      "loss": 1.5084,
      "step": 13236
    },
    {
      "epoch": 0.4740451590953856,
      "grad_norm": 1.7627413272857666,
      "learning_rate": 0.0001132246170586206,
      "loss": 1.5972,
      "step": 13237
    },
    {
      "epoch": 0.4740809712249539,
      "grad_norm": 1.644841194152832,
      "learning_rate": 0.00011321311983199727,
      "loss": 1.7392,
      "step": 13238
    },
    {
      "epoch": 0.4741167833545222,
      "grad_norm": 1.3779560327529907,
      "learning_rate": 0.00011320162242760848,
      "loss": 1.4715,
      "step": 13239
    },
    {
      "epoch": 0.47415259548409044,
      "grad_norm": 1.7711284160614014,
      "learning_rate": 0.00011319012484560885,
      "loss": 1.5618,
      "step": 13240
    },
    {
      "epoch": 0.47418840761365877,
      "grad_norm": 1.8239142894744873,
      "learning_rate": 0.00011317862708615314,
      "loss": 1.6008,
      "step": 13241
    },
    {
      "epoch": 0.47422421974322704,
      "grad_norm": 1.5546090602874756,
      "learning_rate": 0.00011316712914939598,
      "loss": 1.5524,
      "step": 13242
    },
    {
      "epoch": 0.4742600318727953,
      "grad_norm": 1.3858740329742432,
      "learning_rate": 0.00011315563103549211,
      "loss": 1.0793,
      "step": 13243
    },
    {
      "epoch": 0.4742958440023636,
      "grad_norm": 2.5385050773620605,
      "learning_rate": 0.00011314413274459618,
      "loss": 1.5313,
      "step": 13244
    },
    {
      "epoch": 0.4743316561319319,
      "grad_norm": 2.2116689682006836,
      "learning_rate": 0.0001131326342768629,
      "loss": 1.6439,
      "step": 13245
    },
    {
      "epoch": 0.47436746826150017,
      "grad_norm": 1.4536186456680298,
      "learning_rate": 0.00011312113563244695,
      "loss": 1.6257,
      "step": 13246
    },
    {
      "epoch": 0.47440328039106844,
      "grad_norm": 2.8695340156555176,
      "learning_rate": 0.00011310963681150304,
      "loss": 1.7846,
      "step": 13247
    },
    {
      "epoch": 0.47443909252063676,
      "grad_norm": 1.562320590019226,
      "learning_rate": 0.0001130981378141859,
      "loss": 1.8618,
      "step": 13248
    },
    {
      "epoch": 0.47447490465020503,
      "grad_norm": 1.378989815711975,
      "learning_rate": 0.0001130866386406502,
      "loss": 1.4203,
      "step": 13249
    },
    {
      "epoch": 0.4745107167797733,
      "grad_norm": 1.7417534589767456,
      "learning_rate": 0.00011307513929105067,
      "loss": 1.4547,
      "step": 13250
    },
    {
      "epoch": 0.47454652890934157,
      "grad_norm": 2.0014612674713135,
      "learning_rate": 0.000113063639765542,
      "loss": 1.3752,
      "step": 13251
    },
    {
      "epoch": 0.4745823410389099,
      "grad_norm": 1.651104211807251,
      "learning_rate": 0.00011305214006427892,
      "loss": 1.3597,
      "step": 13252
    },
    {
      "epoch": 0.47461815316847816,
      "grad_norm": 2.426689624786377,
      "learning_rate": 0.00011304064018741612,
      "loss": 1.6382,
      "step": 13253
    },
    {
      "epoch": 0.47465396529804643,
      "grad_norm": 1.799432635307312,
      "learning_rate": 0.0001130291401351083,
      "loss": 1.5994,
      "step": 13254
    },
    {
      "epoch": 0.47468977742761476,
      "grad_norm": 1.2933775186538696,
      "learning_rate": 0.00011301763990751025,
      "loss": 1.5075,
      "step": 13255
    },
    {
      "epoch": 0.474725589557183,
      "grad_norm": 1.687794804573059,
      "learning_rate": 0.00011300613950477661,
      "loss": 1.7965,
      "step": 13256
    },
    {
      "epoch": 0.4747614016867513,
      "grad_norm": 1.6241377592086792,
      "learning_rate": 0.00011299463892706217,
      "loss": 1.7507,
      "step": 13257
    },
    {
      "epoch": 0.47479721381631956,
      "grad_norm": 1.474320650100708,
      "learning_rate": 0.0001129831381745216,
      "loss": 1.4066,
      "step": 13258
    },
    {
      "epoch": 0.4748330259458879,
      "grad_norm": 1.6746569871902466,
      "learning_rate": 0.00011297163724730968,
      "loss": 1.5502,
      "step": 13259
    },
    {
      "epoch": 0.47486883807545616,
      "grad_norm": 1.6739968061447144,
      "learning_rate": 0.00011296013614558107,
      "loss": 0.9947,
      "step": 13260
    },
    {
      "epoch": 0.4749046502050244,
      "grad_norm": 1.6017934083938599,
      "learning_rate": 0.00011294863486949059,
      "loss": 1.1272,
      "step": 13261
    },
    {
      "epoch": 0.47494046233459275,
      "grad_norm": 1.7970075607299805,
      "learning_rate": 0.00011293713341919292,
      "loss": 1.5517,
      "step": 13262
    },
    {
      "epoch": 0.474976274464161,
      "grad_norm": 2.006655216217041,
      "learning_rate": 0.0001129256317948428,
      "loss": 1.7006,
      "step": 13263
    },
    {
      "epoch": 0.4750120865937293,
      "grad_norm": 1.580830693244934,
      "learning_rate": 0.00011291412999659499,
      "loss": 1.593,
      "step": 13264
    },
    {
      "epoch": 0.47504789872329756,
      "grad_norm": 1.7830466032028198,
      "learning_rate": 0.00011290262802460419,
      "loss": 1.3509,
      "step": 13265
    },
    {
      "epoch": 0.4750837108528659,
      "grad_norm": 1.5933284759521484,
      "learning_rate": 0.0001128911258790252,
      "loss": 1.3845,
      "step": 13266
    },
    {
      "epoch": 0.47511952298243415,
      "grad_norm": 1.4569121599197388,
      "learning_rate": 0.00011287962356001272,
      "loss": 1.4742,
      "step": 13267
    },
    {
      "epoch": 0.4751553351120024,
      "grad_norm": 1.6202170848846436,
      "learning_rate": 0.00011286812106772153,
      "loss": 1.5486,
      "step": 13268
    },
    {
      "epoch": 0.47519114724157074,
      "grad_norm": 1.7830826044082642,
      "learning_rate": 0.00011285661840230636,
      "loss": 1.2932,
      "step": 13269
    },
    {
      "epoch": 0.475226959371139,
      "grad_norm": 1.6476013660430908,
      "learning_rate": 0.000112845115563922,
      "loss": 1.6752,
      "step": 13270
    },
    {
      "epoch": 0.4752627715007073,
      "grad_norm": 1.4701865911483765,
      "learning_rate": 0.00011283361255272315,
      "loss": 1.2875,
      "step": 13271
    },
    {
      "epoch": 0.47529858363027555,
      "grad_norm": 1.7325506210327148,
      "learning_rate": 0.00011282210936886463,
      "loss": 1.4609,
      "step": 13272
    },
    {
      "epoch": 0.4753343957598439,
      "grad_norm": 1.6176021099090576,
      "learning_rate": 0.00011281060601250113,
      "loss": 1.2128,
      "step": 13273
    },
    {
      "epoch": 0.47537020788941214,
      "grad_norm": 1.9186975955963135,
      "learning_rate": 0.00011279910248378746,
      "loss": 1.6815,
      "step": 13274
    },
    {
      "epoch": 0.4754060200189804,
      "grad_norm": 1.3656400442123413,
      "learning_rate": 0.00011278759878287839,
      "loss": 1.6489,
      "step": 13275
    },
    {
      "epoch": 0.47544183214854874,
      "grad_norm": 1.6302509307861328,
      "learning_rate": 0.00011277609490992866,
      "loss": 1.5046,
      "step": 13276
    },
    {
      "epoch": 0.475477644278117,
      "grad_norm": 1.7330888509750366,
      "learning_rate": 0.00011276459086509305,
      "loss": 1.3758,
      "step": 13277
    },
    {
      "epoch": 0.4755134564076853,
      "grad_norm": 1.6934552192687988,
      "learning_rate": 0.00011275308664852635,
      "loss": 1.5643,
      "step": 13278
    },
    {
      "epoch": 0.47554926853725354,
      "grad_norm": 1.8864017724990845,
      "learning_rate": 0.00011274158226038334,
      "loss": 1.5,
      "step": 13279
    },
    {
      "epoch": 0.47558508066682187,
      "grad_norm": 1.9266653060913086,
      "learning_rate": 0.00011273007770081873,
      "loss": 1.5524,
      "step": 13280
    },
    {
      "epoch": 0.47562089279639014,
      "grad_norm": 1.5648225545883179,
      "learning_rate": 0.00011271857296998737,
      "loss": 1.3658,
      "step": 13281
    },
    {
      "epoch": 0.4756567049259584,
      "grad_norm": 1.9169940948486328,
      "learning_rate": 0.000112707068068044,
      "loss": 1.7119,
      "step": 13282
    },
    {
      "epoch": 0.47569251705552673,
      "grad_norm": 1.5786782503128052,
      "learning_rate": 0.00011269556299514346,
      "loss": 1.2967,
      "step": 13283
    },
    {
      "epoch": 0.475728329185095,
      "grad_norm": 1.858866810798645,
      "learning_rate": 0.00011268405775144044,
      "loss": 1.4823,
      "step": 13284
    },
    {
      "epoch": 0.47576414131466327,
      "grad_norm": 1.7044215202331543,
      "learning_rate": 0.00011267255233708982,
      "loss": 1.3542,
      "step": 13285
    },
    {
      "epoch": 0.47579995344423154,
      "grad_norm": 1.5558875799179077,
      "learning_rate": 0.00011266104675224633,
      "loss": 1.3713,
      "step": 13286
    },
    {
      "epoch": 0.47583576557379986,
      "grad_norm": 1.7449630498886108,
      "learning_rate": 0.00011264954099706481,
      "loss": 1.3809,
      "step": 13287
    },
    {
      "epoch": 0.47587157770336813,
      "grad_norm": 1.7783395051956177,
      "learning_rate": 0.00011263803507170005,
      "loss": 1.4655,
      "step": 13288
    },
    {
      "epoch": 0.4759073898329364,
      "grad_norm": 2.3066070079803467,
      "learning_rate": 0.00011262652897630678,
      "loss": 1.5637,
      "step": 13289
    },
    {
      "epoch": 0.4759432019625047,
      "grad_norm": 1.9367107152938843,
      "learning_rate": 0.0001126150227110399,
      "loss": 1.4018,
      "step": 13290
    },
    {
      "epoch": 0.475979014092073,
      "grad_norm": 1.6617939472198486,
      "learning_rate": 0.00011260351627605413,
      "loss": 1.2903,
      "step": 13291
    },
    {
      "epoch": 0.47601482622164126,
      "grad_norm": 1.5744348764419556,
      "learning_rate": 0.00011259200967150432,
      "loss": 1.2362,
      "step": 13292
    },
    {
      "epoch": 0.47605063835120953,
      "grad_norm": 2.4343421459198,
      "learning_rate": 0.00011258050289754524,
      "loss": 1.8115,
      "step": 13293
    },
    {
      "epoch": 0.47608645048077786,
      "grad_norm": 1.7691670656204224,
      "learning_rate": 0.00011256899595433175,
      "loss": 1.4662,
      "step": 13294
    },
    {
      "epoch": 0.4761222626103461,
      "grad_norm": 1.9748578071594238,
      "learning_rate": 0.0001125574888420186,
      "loss": 1.4279,
      "step": 13295
    },
    {
      "epoch": 0.4761580747399144,
      "grad_norm": 1.6512370109558105,
      "learning_rate": 0.00011254598156076066,
      "loss": 1.7022,
      "step": 13296
    },
    {
      "epoch": 0.4761938868694827,
      "grad_norm": 2.0833606719970703,
      "learning_rate": 0.00011253447411071274,
      "loss": 1.7322,
      "step": 13297
    },
    {
      "epoch": 0.476229698999051,
      "grad_norm": 1.4301164150238037,
      "learning_rate": 0.00011252296649202957,
      "loss": 1.4129,
      "step": 13298
    },
    {
      "epoch": 0.47626551112861926,
      "grad_norm": 1.6318367719650269,
      "learning_rate": 0.00011251145870486612,
      "loss": 1.3278,
      "step": 13299
    },
    {
      "epoch": 0.4763013232581875,
      "grad_norm": 1.5819051265716553,
      "learning_rate": 0.00011249995074937708,
      "loss": 1.6432,
      "step": 13300
    },
    {
      "epoch": 0.47633713538775585,
      "grad_norm": 1.7689194679260254,
      "learning_rate": 0.00011248844262571737,
      "loss": 1.4826,
      "step": 13301
    },
    {
      "epoch": 0.4763729475173241,
      "grad_norm": 1.4098398685455322,
      "learning_rate": 0.00011247693433404172,
      "loss": 1.0356,
      "step": 13302
    },
    {
      "epoch": 0.4764087596468924,
      "grad_norm": 1.6499768495559692,
      "learning_rate": 0.00011246542587450504,
      "loss": 1.6287,
      "step": 13303
    },
    {
      "epoch": 0.4764445717764607,
      "grad_norm": 1.453571081161499,
      "learning_rate": 0.00011245391724726213,
      "loss": 1.3817,
      "step": 13304
    },
    {
      "epoch": 0.476480383906029,
      "grad_norm": 3.2583165168762207,
      "learning_rate": 0.00011244240845246783,
      "loss": 1.4521,
      "step": 13305
    },
    {
      "epoch": 0.47651619603559725,
      "grad_norm": 1.647336721420288,
      "learning_rate": 0.00011243089949027699,
      "loss": 1.613,
      "step": 13306
    },
    {
      "epoch": 0.4765520081651655,
      "grad_norm": 1.7628077268600464,
      "learning_rate": 0.0001124193903608444,
      "loss": 1.5009,
      "step": 13307
    },
    {
      "epoch": 0.47658782029473384,
      "grad_norm": 1.7588131427764893,
      "learning_rate": 0.00011240788106432496,
      "loss": 1.691,
      "step": 13308
    },
    {
      "epoch": 0.4766236324243021,
      "grad_norm": 2.2901549339294434,
      "learning_rate": 0.00011239637160087346,
      "loss": 1.419,
      "step": 13309
    },
    {
      "epoch": 0.4766594445538704,
      "grad_norm": 1.769014596939087,
      "learning_rate": 0.00011238486197064479,
      "loss": 1.4347,
      "step": 13310
    },
    {
      "epoch": 0.4766952566834387,
      "grad_norm": 1.45048987865448,
      "learning_rate": 0.00011237335217379377,
      "loss": 1.4141,
      "step": 13311
    },
    {
      "epoch": 0.476731068813007,
      "grad_norm": 2.432612895965576,
      "learning_rate": 0.00011236184221047526,
      "loss": 1.8489,
      "step": 13312
    },
    {
      "epoch": 0.47676688094257524,
      "grad_norm": 1.4043242931365967,
      "learning_rate": 0.00011235033208084411,
      "loss": 1.5789,
      "step": 13313
    },
    {
      "epoch": 0.4768026930721435,
      "grad_norm": 1.6627109050750732,
      "learning_rate": 0.00011233882178505519,
      "loss": 1.4942,
      "step": 13314
    },
    {
      "epoch": 0.47683850520171184,
      "grad_norm": 2.114083766937256,
      "learning_rate": 0.00011232731132326331,
      "loss": 1.534,
      "step": 13315
    },
    {
      "epoch": 0.4768743173312801,
      "grad_norm": 1.5076711177825928,
      "learning_rate": 0.00011231580069562335,
      "loss": 1.359,
      "step": 13316
    },
    {
      "epoch": 0.4769101294608484,
      "grad_norm": 1.8410205841064453,
      "learning_rate": 0.0001123042899022902,
      "loss": 1.624,
      "step": 13317
    },
    {
      "epoch": 0.4769459415904167,
      "grad_norm": 1.6031441688537598,
      "learning_rate": 0.00011229277894341869,
      "loss": 1.6562,
      "step": 13318
    },
    {
      "epoch": 0.47698175371998497,
      "grad_norm": 1.749197006225586,
      "learning_rate": 0.0001122812678191637,
      "loss": 1.6577,
      "step": 13319
    },
    {
      "epoch": 0.47701756584955324,
      "grad_norm": 2.062668800354004,
      "learning_rate": 0.00011226975652968011,
      "loss": 1.5842,
      "step": 13320
    },
    {
      "epoch": 0.4770533779791215,
      "grad_norm": 2.1983795166015625,
      "learning_rate": 0.00011225824507512275,
      "loss": 1.3001,
      "step": 13321
    },
    {
      "epoch": 0.47708919010868983,
      "grad_norm": 1.686004400253296,
      "learning_rate": 0.00011224673345564651,
      "loss": 1.5864,
      "step": 13322
    },
    {
      "epoch": 0.4771250022382581,
      "grad_norm": 1.8390787839889526,
      "learning_rate": 0.0001122352216714063,
      "loss": 1.6522,
      "step": 13323
    },
    {
      "epoch": 0.47716081436782637,
      "grad_norm": 1.6921519041061401,
      "learning_rate": 0.00011222370972255694,
      "loss": 1.4168,
      "step": 13324
    },
    {
      "epoch": 0.4771966264973947,
      "grad_norm": 1.567881464958191,
      "learning_rate": 0.00011221219760925334,
      "loss": 1.4169,
      "step": 13325
    },
    {
      "epoch": 0.47723243862696296,
      "grad_norm": 1.9966059923171997,
      "learning_rate": 0.00011220068533165036,
      "loss": 1.4439,
      "step": 13326
    },
    {
      "epoch": 0.47726825075653123,
      "grad_norm": 1.6342418193817139,
      "learning_rate": 0.00011218917288990292,
      "loss": 1.4409,
      "step": 13327
    },
    {
      "epoch": 0.4773040628860995,
      "grad_norm": 1.9518680572509766,
      "learning_rate": 0.00011217766028416585,
      "loss": 1.5651,
      "step": 13328
    },
    {
      "epoch": 0.4773398750156678,
      "grad_norm": 1.5526176691055298,
      "learning_rate": 0.00011216614751459408,
      "loss": 1.6035,
      "step": 13329
    },
    {
      "epoch": 0.4773756871452361,
      "grad_norm": 1.5337470769882202,
      "learning_rate": 0.00011215463458134252,
      "loss": 1.6495,
      "step": 13330
    },
    {
      "epoch": 0.47741149927480436,
      "grad_norm": 2.321065902709961,
      "learning_rate": 0.000112143121484566,
      "loss": 1.6215,
      "step": 13331
    },
    {
      "epoch": 0.4774473114043727,
      "grad_norm": 1.355303168296814,
      "learning_rate": 0.00011213160822441948,
      "loss": 1.6907,
      "step": 13332
    },
    {
      "epoch": 0.47748312353394096,
      "grad_norm": 1.7356313467025757,
      "learning_rate": 0.00011212009480105777,
      "loss": 1.5611,
      "step": 13333
    },
    {
      "epoch": 0.4775189356635092,
      "grad_norm": 2.1936237812042236,
      "learning_rate": 0.00011210858121463586,
      "loss": 1.4047,
      "step": 13334
    },
    {
      "epoch": 0.4775547477930775,
      "grad_norm": 1.9643810987472534,
      "learning_rate": 0.00011209706746530858,
      "loss": 1.6057,
      "step": 13335
    },
    {
      "epoch": 0.4775905599226458,
      "grad_norm": 1.537932276725769,
      "learning_rate": 0.00011208555355323088,
      "loss": 1.3793,
      "step": 13336
    },
    {
      "epoch": 0.4776263720522141,
      "grad_norm": 1.4360474348068237,
      "learning_rate": 0.00011207403947855761,
      "loss": 1.2832,
      "step": 13337
    },
    {
      "epoch": 0.47766218418178236,
      "grad_norm": 1.8667532205581665,
      "learning_rate": 0.00011206252524144373,
      "loss": 1.3285,
      "step": 13338
    },
    {
      "epoch": 0.4776979963113507,
      "grad_norm": 1.374731183052063,
      "learning_rate": 0.00011205101084204414,
      "loss": 1.4788,
      "step": 13339
    },
    {
      "epoch": 0.47773380844091895,
      "grad_norm": 1.8810253143310547,
      "learning_rate": 0.00011203949628051376,
      "loss": 1.4788,
      "step": 13340
    },
    {
      "epoch": 0.4777696205704872,
      "grad_norm": 1.5324774980545044,
      "learning_rate": 0.00011202798155700748,
      "loss": 1.472,
      "step": 13341
    },
    {
      "epoch": 0.4778054327000555,
      "grad_norm": 1.8011895418167114,
      "learning_rate": 0.0001120164666716802,
      "loss": 1.3606,
      "step": 13342
    },
    {
      "epoch": 0.4778412448296238,
      "grad_norm": 1.6887016296386719,
      "learning_rate": 0.0001120049516246869,
      "loss": 1.562,
      "step": 13343
    },
    {
      "epoch": 0.4778770569591921,
      "grad_norm": 1.4731894731521606,
      "learning_rate": 0.0001119934364161824,
      "loss": 1.3769,
      "step": 13344
    },
    {
      "epoch": 0.47791286908876035,
      "grad_norm": 1.3754960298538208,
      "learning_rate": 0.00011198192104632174,
      "loss": 1.2356,
      "step": 13345
    },
    {
      "epoch": 0.4779486812183287,
      "grad_norm": 1.9692893028259277,
      "learning_rate": 0.00011197040551525977,
      "loss": 1.4277,
      "step": 13346
    },
    {
      "epoch": 0.47798449334789694,
      "grad_norm": 1.330018401145935,
      "learning_rate": 0.00011195888982315144,
      "loss": 1.4329,
      "step": 13347
    },
    {
      "epoch": 0.4780203054774652,
      "grad_norm": 2.0799808502197266,
      "learning_rate": 0.00011194737397015164,
      "loss": 1.5684,
      "step": 13348
    },
    {
      "epoch": 0.4780561176070335,
      "grad_norm": 1.6616966724395752,
      "learning_rate": 0.00011193585795641539,
      "loss": 1.2649,
      "step": 13349
    },
    {
      "epoch": 0.4780919297366018,
      "grad_norm": 1.8805179595947266,
      "learning_rate": 0.00011192434178209755,
      "loss": 1.3895,
      "step": 13350
    },
    {
      "epoch": 0.4781277418661701,
      "grad_norm": 1.4745457172393799,
      "learning_rate": 0.00011191282544735304,
      "loss": 1.6853,
      "step": 13351
    },
    {
      "epoch": 0.47816355399573834,
      "grad_norm": 1.5154930353164673,
      "learning_rate": 0.00011190130895233686,
      "loss": 1.5363,
      "step": 13352
    },
    {
      "epoch": 0.47819936612530667,
      "grad_norm": 1.273663878440857,
      "learning_rate": 0.00011188979229720389,
      "loss": 1.5082,
      "step": 13353
    },
    {
      "epoch": 0.47823517825487494,
      "grad_norm": 1.5829232931137085,
      "learning_rate": 0.00011187827548210915,
      "loss": 1.4881,
      "step": 13354
    },
    {
      "epoch": 0.4782709903844432,
      "grad_norm": 2.338301658630371,
      "learning_rate": 0.00011186675850720749,
      "loss": 1.3946,
      "step": 13355
    },
    {
      "epoch": 0.4783068025140115,
      "grad_norm": 1.3622664213180542,
      "learning_rate": 0.00011185524137265393,
      "loss": 1.5696,
      "step": 13356
    },
    {
      "epoch": 0.4783426146435798,
      "grad_norm": 2.10280179977417,
      "learning_rate": 0.00011184372407860336,
      "loss": 1.7405,
      "step": 13357
    },
    {
      "epoch": 0.47837842677314807,
      "grad_norm": 1.4048125743865967,
      "learning_rate": 0.00011183220662521079,
      "loss": 1.6725,
      "step": 13358
    },
    {
      "epoch": 0.47841423890271634,
      "grad_norm": 1.3935976028442383,
      "learning_rate": 0.00011182068901263114,
      "loss": 1.3918,
      "step": 13359
    },
    {
      "epoch": 0.47845005103228466,
      "grad_norm": 2.3466100692749023,
      "learning_rate": 0.00011180917124101936,
      "loss": 1.7777,
      "step": 13360
    },
    {
      "epoch": 0.47848586316185293,
      "grad_norm": 1.657720923423767,
      "learning_rate": 0.0001117976533105304,
      "loss": 1.563,
      "step": 13361
    },
    {
      "epoch": 0.4785216752914212,
      "grad_norm": 1.7979260683059692,
      "learning_rate": 0.00011178613522131924,
      "loss": 1.2531,
      "step": 13362
    },
    {
      "epoch": 0.47855748742098947,
      "grad_norm": 1.7003462314605713,
      "learning_rate": 0.00011177461697354084,
      "loss": 1.6431,
      "step": 13363
    },
    {
      "epoch": 0.4785932995505578,
      "grad_norm": 1.7036478519439697,
      "learning_rate": 0.00011176309856735014,
      "loss": 1.3807,
      "step": 13364
    },
    {
      "epoch": 0.47862911168012606,
      "grad_norm": 1.8643072843551636,
      "learning_rate": 0.00011175158000290216,
      "loss": 1.6466,
      "step": 13365
    },
    {
      "epoch": 0.47866492380969433,
      "grad_norm": 1.696761131286621,
      "learning_rate": 0.00011174006128035178,
      "loss": 1.4672,
      "step": 13366
    },
    {
      "epoch": 0.47870073593926266,
      "grad_norm": 1.6161099672317505,
      "learning_rate": 0.00011172854239985409,
      "loss": 1.4775,
      "step": 13367
    },
    {
      "epoch": 0.4787365480688309,
      "grad_norm": 1.5913996696472168,
      "learning_rate": 0.0001117170233615639,
      "loss": 1.2976,
      "step": 13368
    },
    {
      "epoch": 0.4787723601983992,
      "grad_norm": 1.3425949811935425,
      "learning_rate": 0.00011170550416563634,
      "loss": 1.2461,
      "step": 13369
    },
    {
      "epoch": 0.47880817232796746,
      "grad_norm": 1.8969595432281494,
      "learning_rate": 0.0001116939848122263,
      "loss": 1.4931,
      "step": 13370
    },
    {
      "epoch": 0.4788439844575358,
      "grad_norm": 1.8349109888076782,
      "learning_rate": 0.00011168246530148876,
      "loss": 1.6161,
      "step": 13371
    },
    {
      "epoch": 0.47887979658710406,
      "grad_norm": 2.742892265319824,
      "learning_rate": 0.00011167094563357876,
      "loss": 1.6424,
      "step": 13372
    },
    {
      "epoch": 0.4789156087166723,
      "grad_norm": 3.602156162261963,
      "learning_rate": 0.00011165942580865118,
      "loss": 1.5828,
      "step": 13373
    },
    {
      "epoch": 0.4789514208462406,
      "grad_norm": 1.6566842794418335,
      "learning_rate": 0.00011164790582686113,
      "loss": 1.4616,
      "step": 13374
    },
    {
      "epoch": 0.4789872329758089,
      "grad_norm": 1.709389567375183,
      "learning_rate": 0.0001116363856883635,
      "loss": 1.5164,
      "step": 13375
    },
    {
      "epoch": 0.4790230451053772,
      "grad_norm": 2.565922498703003,
      "learning_rate": 0.00011162486539331334,
      "loss": 1.4494,
      "step": 13376
    },
    {
      "epoch": 0.47905885723494546,
      "grad_norm": 1.883772373199463,
      "learning_rate": 0.00011161334494186557,
      "loss": 1.4987,
      "step": 13377
    },
    {
      "epoch": 0.4790946693645138,
      "grad_norm": 1.9089127779006958,
      "learning_rate": 0.00011160182433417525,
      "loss": 1.4315,
      "step": 13378
    },
    {
      "epoch": 0.47913048149408205,
      "grad_norm": 1.619033932685852,
      "learning_rate": 0.00011159030357039733,
      "loss": 1.2727,
      "step": 13379
    },
    {
      "epoch": 0.4791662936236503,
      "grad_norm": 1.3101295232772827,
      "learning_rate": 0.00011157878265068685,
      "loss": 1.4185,
      "step": 13380
    },
    {
      "epoch": 0.4792021057532186,
      "grad_norm": 2.5190060138702393,
      "learning_rate": 0.00011156726157519877,
      "loss": 1.8318,
      "step": 13381
    },
    {
      "epoch": 0.4792379178827869,
      "grad_norm": 1.801878809928894,
      "learning_rate": 0.00011155574034408812,
      "loss": 1.5164,
      "step": 13382
    },
    {
      "epoch": 0.4792737300123552,
      "grad_norm": 1.659848928451538,
      "learning_rate": 0.00011154421895750984,
      "loss": 1.3737,
      "step": 13383
    },
    {
      "epoch": 0.47930954214192345,
      "grad_norm": 1.8164340257644653,
      "learning_rate": 0.00011153269741561905,
      "loss": 1.5673,
      "step": 13384
    },
    {
      "epoch": 0.4793453542714918,
      "grad_norm": 1.5826667547225952,
      "learning_rate": 0.00011152117571857065,
      "loss": 1.0404,
      "step": 13385
    },
    {
      "epoch": 0.47938116640106004,
      "grad_norm": 1.6665927171707153,
      "learning_rate": 0.0001115096538665197,
      "loss": 1.4599,
      "step": 13386
    },
    {
      "epoch": 0.4794169785306283,
      "grad_norm": 2.6693007946014404,
      "learning_rate": 0.00011149813185962124,
      "loss": 1.4425,
      "step": 13387
    },
    {
      "epoch": 0.4794527906601966,
      "grad_norm": 1.5640629529953003,
      "learning_rate": 0.00011148660969803019,
      "loss": 2.0646,
      "step": 13388
    },
    {
      "epoch": 0.4794886027897649,
      "grad_norm": 1.4011385440826416,
      "learning_rate": 0.00011147508738190167,
      "loss": 1.5649,
      "step": 13389
    },
    {
      "epoch": 0.4795244149193332,
      "grad_norm": 1.667718529701233,
      "learning_rate": 0.0001114635649113906,
      "loss": 1.4943,
      "step": 13390
    },
    {
      "epoch": 0.47956022704890144,
      "grad_norm": 1.893916368484497,
      "learning_rate": 0.00011145204228665209,
      "loss": 1.2751,
      "step": 13391
    },
    {
      "epoch": 0.47959603917846977,
      "grad_norm": 1.5027039051055908,
      "learning_rate": 0.00011144051950784111,
      "loss": 1.395,
      "step": 13392
    },
    {
      "epoch": 0.47963185130803804,
      "grad_norm": 1.4640353918075562,
      "learning_rate": 0.00011142899657511272,
      "loss": 1.5298,
      "step": 13393
    },
    {
      "epoch": 0.4796676634376063,
      "grad_norm": 1.6055121421813965,
      "learning_rate": 0.00011141747348862191,
      "loss": 1.5649,
      "step": 13394
    },
    {
      "epoch": 0.4797034755671746,
      "grad_norm": 1.6210955381393433,
      "learning_rate": 0.00011140595024852369,
      "loss": 1.6393,
      "step": 13395
    },
    {
      "epoch": 0.4797392876967429,
      "grad_norm": 1.5539766550064087,
      "learning_rate": 0.00011139442685497317,
      "loss": 1.6614,
      "step": 13396
    },
    {
      "epoch": 0.47977509982631117,
      "grad_norm": 2.0545995235443115,
      "learning_rate": 0.0001113829033081253,
      "loss": 1.3625,
      "step": 13397
    },
    {
      "epoch": 0.47981091195587944,
      "grad_norm": 1.3574188947677612,
      "learning_rate": 0.00011137137960813517,
      "loss": 1.34,
      "step": 13398
    },
    {
      "epoch": 0.47984672408544776,
      "grad_norm": 1.6591427326202393,
      "learning_rate": 0.00011135985575515778,
      "loss": 1.6447,
      "step": 13399
    },
    {
      "epoch": 0.47988253621501603,
      "grad_norm": 1.2985018491744995,
      "learning_rate": 0.0001113483317493482,
      "loss": 1.4672,
      "step": 13400
    },
    {
      "epoch": 0.4799183483445843,
      "grad_norm": 1.8751863241195679,
      "learning_rate": 0.00011133680759086145,
      "loss": 1.725,
      "step": 13401
    },
    {
      "epoch": 0.47995416047415257,
      "grad_norm": 1.4524515867233276,
      "learning_rate": 0.00011132528327985256,
      "loss": 1.6149,
      "step": 13402
    },
    {
      "epoch": 0.4799899726037209,
      "grad_norm": 1.3275116682052612,
      "learning_rate": 0.00011131375881647664,
      "loss": 1.1616,
      "step": 13403
    },
    {
      "epoch": 0.48002578473328916,
      "grad_norm": 1.6403449773788452,
      "learning_rate": 0.00011130223420088864,
      "loss": 1.2983,
      "step": 13404
    },
    {
      "epoch": 0.48006159686285743,
      "grad_norm": 1.282675862312317,
      "learning_rate": 0.00011129070943324366,
      "loss": 1.3949,
      "step": 13405
    },
    {
      "epoch": 0.48009740899242576,
      "grad_norm": 1.388199806213379,
      "learning_rate": 0.00011127918451369676,
      "loss": 1.4264,
      "step": 13406
    },
    {
      "epoch": 0.480133221121994,
      "grad_norm": 1.4312509298324585,
      "learning_rate": 0.00011126765944240298,
      "loss": 1.4627,
      "step": 13407
    },
    {
      "epoch": 0.4801690332515623,
      "grad_norm": 1.5060590505599976,
      "learning_rate": 0.00011125613421951737,
      "loss": 1.3497,
      "step": 13408
    },
    {
      "epoch": 0.48020484538113056,
      "grad_norm": 1.3689676523208618,
      "learning_rate": 0.00011124460884519503,
      "loss": 1.4491,
      "step": 13409
    },
    {
      "epoch": 0.4802406575106989,
      "grad_norm": 1.402147650718689,
      "learning_rate": 0.00011123308331959093,
      "loss": 1.5145,
      "step": 13410
    },
    {
      "epoch": 0.48027646964026716,
      "grad_norm": 1.3701300621032715,
      "learning_rate": 0.00011122155764286024,
      "loss": 1.3315,
      "step": 13411
    },
    {
      "epoch": 0.4803122817698354,
      "grad_norm": 1.9080400466918945,
      "learning_rate": 0.00011121003181515792,
      "loss": 1.461,
      "step": 13412
    },
    {
      "epoch": 0.48034809389940375,
      "grad_norm": 2.7309305667877197,
      "learning_rate": 0.00011119850583663908,
      "loss": 1.8333,
      "step": 13413
    },
    {
      "epoch": 0.480383906028972,
      "grad_norm": 2.27876615524292,
      "learning_rate": 0.00011118697970745881,
      "loss": 1.5516,
      "step": 13414
    },
    {
      "epoch": 0.4804197181585403,
      "grad_norm": 1.6473275423049927,
      "learning_rate": 0.00011117545342777215,
      "loss": 1.3657,
      "step": 13415
    },
    {
      "epoch": 0.48045553028810856,
      "grad_norm": 1.6574909687042236,
      "learning_rate": 0.0001111639269977342,
      "loss": 1.6839,
      "step": 13416
    },
    {
      "epoch": 0.4804913424176769,
      "grad_norm": 1.214024305343628,
      "learning_rate": 0.00011115240041749999,
      "loss": 1.4461,
      "step": 13417
    },
    {
      "epoch": 0.48052715454724515,
      "grad_norm": 2.024338960647583,
      "learning_rate": 0.00011114087368722463,
      "loss": 1.2144,
      "step": 13418
    },
    {
      "epoch": 0.4805629666768134,
      "grad_norm": 1.9191267490386963,
      "learning_rate": 0.00011112934680706317,
      "loss": 1.5104,
      "step": 13419
    },
    {
      "epoch": 0.48059877880638174,
      "grad_norm": 1.7122324705123901,
      "learning_rate": 0.00011111781977717075,
      "loss": 1.2507,
      "step": 13420
    },
    {
      "epoch": 0.48063459093595,
      "grad_norm": 1.6362801790237427,
      "learning_rate": 0.00011110629259770235,
      "loss": 1.6835,
      "step": 13421
    },
    {
      "epoch": 0.4806704030655183,
      "grad_norm": 1.6250897645950317,
      "learning_rate": 0.00011109476526881313,
      "loss": 1.4953,
      "step": 13422
    },
    {
      "epoch": 0.48070621519508655,
      "grad_norm": 1.9805313348770142,
      "learning_rate": 0.00011108323779065814,
      "loss": 1.5814,
      "step": 13423
    },
    {
      "epoch": 0.4807420273246549,
      "grad_norm": 1.6796422004699707,
      "learning_rate": 0.00011107171016339251,
      "loss": 1.6619,
      "step": 13424
    },
    {
      "epoch": 0.48077783945422314,
      "grad_norm": 1.6529179811477661,
      "learning_rate": 0.00011106018238717128,
      "loss": 1.4423,
      "step": 13425
    },
    {
      "epoch": 0.4808136515837914,
      "grad_norm": 1.6704829931259155,
      "learning_rate": 0.00011104865446214957,
      "loss": 1.3773,
      "step": 13426
    },
    {
      "epoch": 0.48084946371335974,
      "grad_norm": 1.3127925395965576,
      "learning_rate": 0.00011103712638848244,
      "loss": 1.5127,
      "step": 13427
    },
    {
      "epoch": 0.480885275842928,
      "grad_norm": 1.6566414833068848,
      "learning_rate": 0.00011102559816632507,
      "loss": 1.6901,
      "step": 13428
    },
    {
      "epoch": 0.4809210879724963,
      "grad_norm": 1.6600404977798462,
      "learning_rate": 0.00011101406979583246,
      "loss": 1.3477,
      "step": 13429
    },
    {
      "epoch": 0.48095690010206454,
      "grad_norm": 2.285459518432617,
      "learning_rate": 0.00011100254127715975,
      "loss": 1.3787,
      "step": 13430
    },
    {
      "epoch": 0.48099271223163287,
      "grad_norm": 1.7669966220855713,
      "learning_rate": 0.00011099101261046205,
      "loss": 1.4803,
      "step": 13431
    },
    {
      "epoch": 0.48102852436120114,
      "grad_norm": 1.491285800933838,
      "learning_rate": 0.00011097948379589444,
      "loss": 1.4013,
      "step": 13432
    },
    {
      "epoch": 0.4810643364907694,
      "grad_norm": 1.9096039533615112,
      "learning_rate": 0.00011096795483361205,
      "loss": 1.3939,
      "step": 13433
    },
    {
      "epoch": 0.48110014862033773,
      "grad_norm": 1.4066028594970703,
      "learning_rate": 0.00011095642572376996,
      "loss": 1.4652,
      "step": 13434
    },
    {
      "epoch": 0.481135960749906,
      "grad_norm": 1.5006831884384155,
      "learning_rate": 0.0001109448964665233,
      "loss": 1.3876,
      "step": 13435
    },
    {
      "epoch": 0.48117177287947427,
      "grad_norm": 1.5165512561798096,
      "learning_rate": 0.00011093336706202717,
      "loss": 1.696,
      "step": 13436
    },
    {
      "epoch": 0.48120758500904254,
      "grad_norm": 1.273987889289856,
      "learning_rate": 0.00011092183751043672,
      "loss": 1.608,
      "step": 13437
    },
    {
      "epoch": 0.48124339713861086,
      "grad_norm": 1.6874688863754272,
      "learning_rate": 0.000110910307811907,
      "loss": 1.301,
      "step": 13438
    },
    {
      "epoch": 0.48127920926817913,
      "grad_norm": 1.5528160333633423,
      "learning_rate": 0.00011089877796659319,
      "loss": 1.3853,
      "step": 13439
    },
    {
      "epoch": 0.4813150213977474,
      "grad_norm": 1.5378245115280151,
      "learning_rate": 0.00011088724797465036,
      "loss": 1.5139,
      "step": 13440
    },
    {
      "epoch": 0.4813508335273157,
      "grad_norm": 2.0010392665863037,
      "learning_rate": 0.00011087571783623365,
      "loss": 1.5756,
      "step": 13441
    },
    {
      "epoch": 0.481386645656884,
      "grad_norm": 1.7677205801010132,
      "learning_rate": 0.0001108641875514982,
      "loss": 1.336,
      "step": 13442
    },
    {
      "epoch": 0.48142245778645226,
      "grad_norm": 2.936492443084717,
      "learning_rate": 0.00011085265712059909,
      "loss": 1.787,
      "step": 13443
    },
    {
      "epoch": 0.48145826991602053,
      "grad_norm": 1.785649061203003,
      "learning_rate": 0.00011084112654369152,
      "loss": 1.2747,
      "step": 13444
    },
    {
      "epoch": 0.48149408204558886,
      "grad_norm": 1.6284798383712769,
      "learning_rate": 0.00011082959582093055,
      "loss": 1.3396,
      "step": 13445
    },
    {
      "epoch": 0.4815298941751571,
      "grad_norm": 1.441011667251587,
      "learning_rate": 0.00011081806495247136,
      "loss": 1.6317,
      "step": 13446
    },
    {
      "epoch": 0.4815657063047254,
      "grad_norm": 1.789101004600525,
      "learning_rate": 0.00011080653393846905,
      "loss": 1.6096,
      "step": 13447
    },
    {
      "epoch": 0.4816015184342937,
      "grad_norm": 2.5847105979919434,
      "learning_rate": 0.00011079500277907875,
      "loss": 1.321,
      "step": 13448
    },
    {
      "epoch": 0.481637330563862,
      "grad_norm": 1.5998847484588623,
      "learning_rate": 0.00011078347147445563,
      "loss": 1.5482,
      "step": 13449
    },
    {
      "epoch": 0.48167314269343026,
      "grad_norm": 1.8561943769454956,
      "learning_rate": 0.0001107719400247548,
      "loss": 1.3777,
      "step": 13450
    },
    {
      "epoch": 0.4817089548229985,
      "grad_norm": 2.0072786808013916,
      "learning_rate": 0.00011076040843013141,
      "loss": 1.149,
      "step": 13451
    },
    {
      "epoch": 0.48174476695256685,
      "grad_norm": 1.9042658805847168,
      "learning_rate": 0.00011074887669074058,
      "loss": 1.3163,
      "step": 13452
    },
    {
      "epoch": 0.4817805790821351,
      "grad_norm": 1.6006115674972534,
      "learning_rate": 0.00011073734480673754,
      "loss": 1.5273,
      "step": 13453
    },
    {
      "epoch": 0.4818163912117034,
      "grad_norm": 1.7646973133087158,
      "learning_rate": 0.00011072581277827732,
      "loss": 1.5288,
      "step": 13454
    },
    {
      "epoch": 0.4818522033412717,
      "grad_norm": 2.283583641052246,
      "learning_rate": 0.00011071428060551517,
      "loss": 1.7976,
      "step": 13455
    },
    {
      "epoch": 0.48188801547084,
      "grad_norm": 1.4994382858276367,
      "learning_rate": 0.00011070274828860618,
      "loss": 1.3215,
      "step": 13456
    },
    {
      "epoch": 0.48192382760040825,
      "grad_norm": 1.39043128490448,
      "learning_rate": 0.0001106912158277055,
      "loss": 1.3991,
      "step": 13457
    },
    {
      "epoch": 0.4819596397299765,
      "grad_norm": 1.6843230724334717,
      "learning_rate": 0.00011067968322296831,
      "loss": 1.6234,
      "step": 13458
    },
    {
      "epoch": 0.48199545185954484,
      "grad_norm": 1.7703015804290771,
      "learning_rate": 0.00011066815047454975,
      "loss": 1.517,
      "step": 13459
    },
    {
      "epoch": 0.4820312639891131,
      "grad_norm": 1.695458173751831,
      "learning_rate": 0.00011065661758260502,
      "loss": 1.6138,
      "step": 13460
    },
    {
      "epoch": 0.4820670761186814,
      "grad_norm": 1.6094825267791748,
      "learning_rate": 0.00011064508454728921,
      "loss": 1.3752,
      "step": 13461
    },
    {
      "epoch": 0.4821028882482497,
      "grad_norm": 1.7632977962493896,
      "learning_rate": 0.00011063355136875753,
      "loss": 1.541,
      "step": 13462
    },
    {
      "epoch": 0.482138700377818,
      "grad_norm": 1.7887697219848633,
      "learning_rate": 0.00011062201804716512,
      "loss": 1.5856,
      "step": 13463
    },
    {
      "epoch": 0.48217451250738624,
      "grad_norm": 1.7318227291107178,
      "learning_rate": 0.0001106104845826672,
      "loss": 1.2289,
      "step": 13464
    },
    {
      "epoch": 0.4822103246369545,
      "grad_norm": 1.6601182222366333,
      "learning_rate": 0.00011059895097541888,
      "loss": 1.875,
      "step": 13465
    },
    {
      "epoch": 0.48224613676652284,
      "grad_norm": 1.9784289598464966,
      "learning_rate": 0.00011058741722557533,
      "loss": 1.3676,
      "step": 13466
    },
    {
      "epoch": 0.4822819488960911,
      "grad_norm": 1.8831995725631714,
      "learning_rate": 0.00011057588333329174,
      "loss": 1.6114,
      "step": 13467
    },
    {
      "epoch": 0.4823177610256594,
      "grad_norm": 1.6999905109405518,
      "learning_rate": 0.00011056434929872325,
      "loss": 1.6761,
      "step": 13468
    },
    {
      "epoch": 0.4823535731552277,
      "grad_norm": 1.5982940196990967,
      "learning_rate": 0.00011055281512202513,
      "loss": 1.4517,
      "step": 13469
    },
    {
      "epoch": 0.48238938528479597,
      "grad_norm": 2.8717644214630127,
      "learning_rate": 0.00011054128080335246,
      "loss": 1.6728,
      "step": 13470
    },
    {
      "epoch": 0.48242519741436424,
      "grad_norm": 1.9627916812896729,
      "learning_rate": 0.00011052974634286046,
      "loss": 1.7458,
      "step": 13471
    },
    {
      "epoch": 0.4824610095439325,
      "grad_norm": 1.698922872543335,
      "learning_rate": 0.00011051821174070429,
      "loss": 1.8436,
      "step": 13472
    },
    {
      "epoch": 0.48249682167350083,
      "grad_norm": 1.2740743160247803,
      "learning_rate": 0.0001105066769970392,
      "loss": 1.5197,
      "step": 13473
    },
    {
      "epoch": 0.4825326338030691,
      "grad_norm": 1.7765341997146606,
      "learning_rate": 0.00011049514211202028,
      "loss": 1.8556,
      "step": 13474
    },
    {
      "epoch": 0.48256844593263737,
      "grad_norm": 1.9646724462509155,
      "learning_rate": 0.00011048360708580279,
      "loss": 1.5785,
      "step": 13475
    },
    {
      "epoch": 0.4826042580622057,
      "grad_norm": 2.2040209770202637,
      "learning_rate": 0.00011047207191854185,
      "loss": 1.5735,
      "step": 13476
    },
    {
      "epoch": 0.48264007019177396,
      "grad_norm": 1.3888829946517944,
      "learning_rate": 0.00011046053661039273,
      "loss": 1.4665,
      "step": 13477
    },
    {
      "epoch": 0.48267588232134223,
      "grad_norm": 1.831881046295166,
      "learning_rate": 0.00011044900116151053,
      "loss": 1.6623,
      "step": 13478
    },
    {
      "epoch": 0.4827116944509105,
      "grad_norm": 1.4265269041061401,
      "learning_rate": 0.00011043746557205054,
      "loss": 1.5404,
      "step": 13479
    },
    {
      "epoch": 0.4827475065804788,
      "grad_norm": 1.828692078590393,
      "learning_rate": 0.0001104259298421679,
      "loss": 1.5943,
      "step": 13480
    },
    {
      "epoch": 0.4827833187100471,
      "grad_norm": 1.9613791704177856,
      "learning_rate": 0.00011041439397201785,
      "loss": 1.6634,
      "step": 13481
    },
    {
      "epoch": 0.48281913083961536,
      "grad_norm": 1.9883949756622314,
      "learning_rate": 0.00011040285796175553,
      "loss": 1.19,
      "step": 13482
    },
    {
      "epoch": 0.4828549429691837,
      "grad_norm": 1.7061982154846191,
      "learning_rate": 0.00011039132181153618,
      "loss": 1.5244,
      "step": 13483
    },
    {
      "epoch": 0.48289075509875196,
      "grad_norm": 1.7107230424880981,
      "learning_rate": 0.00011037978552151502,
      "loss": 1.4301,
      "step": 13484
    },
    {
      "epoch": 0.4829265672283202,
      "grad_norm": 2.234708070755005,
      "learning_rate": 0.0001103682490918472,
      "loss": 1.8736,
      "step": 13485
    },
    {
      "epoch": 0.4829623793578885,
      "grad_norm": 1.9554941654205322,
      "learning_rate": 0.000110356712522688,
      "loss": 1.6488,
      "step": 13486
    },
    {
      "epoch": 0.4829981914874568,
      "grad_norm": 1.7168772220611572,
      "learning_rate": 0.00011034517581419255,
      "loss": 1.7765,
      "step": 13487
    },
    {
      "epoch": 0.4830340036170251,
      "grad_norm": 1.7673360109329224,
      "learning_rate": 0.00011033363896651613,
      "loss": 1.3514,
      "step": 13488
    },
    {
      "epoch": 0.48306981574659336,
      "grad_norm": 1.7775872945785522,
      "learning_rate": 0.00011032210197981392,
      "loss": 1.2557,
      "step": 13489
    },
    {
      "epoch": 0.4831056278761617,
      "grad_norm": 1.549501895904541,
      "learning_rate": 0.00011031056485424116,
      "loss": 1.4594,
      "step": 13490
    },
    {
      "epoch": 0.48314144000572995,
      "grad_norm": 1.6182304620742798,
      "learning_rate": 0.00011029902758995304,
      "loss": 1.4751,
      "step": 13491
    },
    {
      "epoch": 0.4831772521352982,
      "grad_norm": 1.4240484237670898,
      "learning_rate": 0.00011028749018710478,
      "loss": 1.3701,
      "step": 13492
    },
    {
      "epoch": 0.4832130642648665,
      "grad_norm": 1.6473289728164673,
      "learning_rate": 0.00011027595264585162,
      "loss": 1.2206,
      "step": 13493
    },
    {
      "epoch": 0.4832488763944348,
      "grad_norm": 2.2116892337799072,
      "learning_rate": 0.00011026441496634874,
      "loss": 1.6348,
      "step": 13494
    },
    {
      "epoch": 0.4832846885240031,
      "grad_norm": 1.27376127243042,
      "learning_rate": 0.00011025287714875143,
      "loss": 1.3734,
      "step": 13495
    },
    {
      "epoch": 0.48332050065357135,
      "grad_norm": 1.800029993057251,
      "learning_rate": 0.00011024133919321486,
      "loss": 1.5834,
      "step": 13496
    },
    {
      "epoch": 0.4833563127831397,
      "grad_norm": 1.679153561592102,
      "learning_rate": 0.00011022980109989431,
      "loss": 1.4844,
      "step": 13497
    },
    {
      "epoch": 0.48339212491270794,
      "grad_norm": 2.0695579051971436,
      "learning_rate": 0.00011021826286894496,
      "loss": 1.7796,
      "step": 13498
    },
    {
      "epoch": 0.4834279370422762,
      "grad_norm": 1.740222692489624,
      "learning_rate": 0.0001102067245005221,
      "loss": 1.4109,
      "step": 13499
    },
    {
      "epoch": 0.4834637491718445,
      "grad_norm": 1.6901319026947021,
      "learning_rate": 0.0001101951859947809,
      "loss": 1.8769,
      "step": 13500
    },
    {
      "epoch": 0.4834995613014128,
      "grad_norm": 1.6777935028076172,
      "learning_rate": 0.00011018364735187661,
      "loss": 1.4192,
      "step": 13501
    },
    {
      "epoch": 0.4835353734309811,
      "grad_norm": 1.7092137336730957,
      "learning_rate": 0.00011017210857196449,
      "loss": 1.6799,
      "step": 13502
    },
    {
      "epoch": 0.48357118556054934,
      "grad_norm": 1.8037439584732056,
      "learning_rate": 0.00011016056965519979,
      "loss": 1.5957,
      "step": 13503
    },
    {
      "epoch": 0.48360699769011767,
      "grad_norm": 1.3701008558273315,
      "learning_rate": 0.00011014903060173772,
      "loss": 1.6095,
      "step": 13504
    },
    {
      "epoch": 0.48364280981968594,
      "grad_norm": 1.4158246517181396,
      "learning_rate": 0.00011013749141173351,
      "loss": 1.3914,
      "step": 13505
    },
    {
      "epoch": 0.4836786219492542,
      "grad_norm": 1.9191498756408691,
      "learning_rate": 0.00011012595208534247,
      "loss": 1.9305,
      "step": 13506
    },
    {
      "epoch": 0.4837144340788225,
      "grad_norm": 2.0347769260406494,
      "learning_rate": 0.00011011441262271975,
      "loss": 1.3077,
      "step": 13507
    },
    {
      "epoch": 0.4837502462083908,
      "grad_norm": 1.7553139925003052,
      "learning_rate": 0.00011010287302402073,
      "loss": 1.2249,
      "step": 13508
    },
    {
      "epoch": 0.48378605833795907,
      "grad_norm": 1.331820011138916,
      "learning_rate": 0.00011009133328940053,
      "loss": 1.5748,
      "step": 13509
    },
    {
      "epoch": 0.48382187046752734,
      "grad_norm": 1.791427731513977,
      "learning_rate": 0.00011007979341901446,
      "loss": 1.4737,
      "step": 13510
    },
    {
      "epoch": 0.48385768259709566,
      "grad_norm": 1.544029951095581,
      "learning_rate": 0.0001100682534130178,
      "loss": 1.4346,
      "step": 13511
    },
    {
      "epoch": 0.48389349472666393,
      "grad_norm": 1.728143334388733,
      "learning_rate": 0.00011005671327156574,
      "loss": 1.5266,
      "step": 13512
    },
    {
      "epoch": 0.4839293068562322,
      "grad_norm": 1.5404763221740723,
      "learning_rate": 0.00011004517299481363,
      "loss": 1.4122,
      "step": 13513
    },
    {
      "epoch": 0.48396511898580047,
      "grad_norm": 1.9863710403442383,
      "learning_rate": 0.00011003363258291664,
      "loss": 1.6186,
      "step": 13514
    },
    {
      "epoch": 0.4840009311153688,
      "grad_norm": 2.0675201416015625,
      "learning_rate": 0.00011002209203603007,
      "loss": 1.4077,
      "step": 13515
    },
    {
      "epoch": 0.48403674324493706,
      "grad_norm": 1.5609140396118164,
      "learning_rate": 0.00011001055135430916,
      "loss": 1.3322,
      "step": 13516
    },
    {
      "epoch": 0.48407255537450533,
      "grad_norm": 1.3592745065689087,
      "learning_rate": 0.00010999901053790924,
      "loss": 1.5969,
      "step": 13517
    },
    {
      "epoch": 0.48410836750407366,
      "grad_norm": 2.3698625564575195,
      "learning_rate": 0.0001099874695869855,
      "loss": 1.474,
      "step": 13518
    },
    {
      "epoch": 0.4841441796336419,
      "grad_norm": 2.192758798599243,
      "learning_rate": 0.00010997592850169325,
      "loss": 1.759,
      "step": 13519
    },
    {
      "epoch": 0.4841799917632102,
      "grad_norm": 1.3578786849975586,
      "learning_rate": 0.00010996438728218772,
      "loss": 1.4524,
      "step": 13520
    },
    {
      "epoch": 0.48421580389277846,
      "grad_norm": 1.4247230291366577,
      "learning_rate": 0.00010995284592862425,
      "loss": 1.782,
      "step": 13521
    },
    {
      "epoch": 0.4842516160223468,
      "grad_norm": 2.091456890106201,
      "learning_rate": 0.00010994130444115804,
      "loss": 1.5596,
      "step": 13522
    },
    {
      "epoch": 0.48428742815191506,
      "grad_norm": 2.0627150535583496,
      "learning_rate": 0.00010992976281994443,
      "loss": 1.3551,
      "step": 13523
    },
    {
      "epoch": 0.4843232402814833,
      "grad_norm": 1.4531793594360352,
      "learning_rate": 0.00010991822106513867,
      "loss": 1.4569,
      "step": 13524
    },
    {
      "epoch": 0.48435905241105165,
      "grad_norm": 2.3870937824249268,
      "learning_rate": 0.00010990667917689603,
      "loss": 1.7541,
      "step": 13525
    },
    {
      "epoch": 0.4843948645406199,
      "grad_norm": 1.8330572843551636,
      "learning_rate": 0.00010989513715537184,
      "loss": 1.6933,
      "step": 13526
    },
    {
      "epoch": 0.4844306766701882,
      "grad_norm": 1.6797586679458618,
      "learning_rate": 0.00010988359500072128,
      "loss": 1.3094,
      "step": 13527
    },
    {
      "epoch": 0.48446648879975646,
      "grad_norm": 1.7150591611862183,
      "learning_rate": 0.00010987205271309972,
      "loss": 1.5811,
      "step": 13528
    },
    {
      "epoch": 0.4845023009293248,
      "grad_norm": 1.74473237991333,
      "learning_rate": 0.00010986051029266242,
      "loss": 1.6228,
      "step": 13529
    },
    {
      "epoch": 0.48453811305889305,
      "grad_norm": 2.2052595615386963,
      "learning_rate": 0.0001098489677395647,
      "loss": 1.8501,
      "step": 13530
    },
    {
      "epoch": 0.4845739251884613,
      "grad_norm": 1.872136116027832,
      "learning_rate": 0.0001098374250539618,
      "loss": 1.5437,
      "step": 13531
    },
    {
      "epoch": 0.48460973731802964,
      "grad_norm": 2.1946043968200684,
      "learning_rate": 0.00010982588223600905,
      "loss": 1.2338,
      "step": 13532
    },
    {
      "epoch": 0.4846455494475979,
      "grad_norm": 1.7079534530639648,
      "learning_rate": 0.00010981433928586168,
      "loss": 1.8226,
      "step": 13533
    },
    {
      "epoch": 0.4846813615771662,
      "grad_norm": 1.761330008506775,
      "learning_rate": 0.00010980279620367511,
      "loss": 1.4908,
      "step": 13534
    },
    {
      "epoch": 0.48471717370673445,
      "grad_norm": 1.1783751249313354,
      "learning_rate": 0.00010979125298960453,
      "loss": 1.5007,
      "step": 13535
    },
    {
      "epoch": 0.4847529858363028,
      "grad_norm": 1.9154187440872192,
      "learning_rate": 0.00010977970964380526,
      "loss": 1.528,
      "step": 13536
    },
    {
      "epoch": 0.48478879796587104,
      "grad_norm": 1.502217411994934,
      "learning_rate": 0.00010976816616643262,
      "loss": 1.37,
      "step": 13537
    },
    {
      "epoch": 0.4848246100954393,
      "grad_norm": 1.479232907295227,
      "learning_rate": 0.0001097566225576419,
      "loss": 1.4488,
      "step": 13538
    },
    {
      "epoch": 0.48486042222500764,
      "grad_norm": 1.262938380241394,
      "learning_rate": 0.00010974507881758842,
      "loss": 1.6514,
      "step": 13539
    },
    {
      "epoch": 0.4848962343545759,
      "grad_norm": 1.8512773513793945,
      "learning_rate": 0.00010973353494642745,
      "loss": 1.5816,
      "step": 13540
    },
    {
      "epoch": 0.4849320464841442,
      "grad_norm": 1.9465118646621704,
      "learning_rate": 0.00010972199094431435,
      "loss": 1.7751,
      "step": 13541
    },
    {
      "epoch": 0.48496785861371244,
      "grad_norm": 1.3644713163375854,
      "learning_rate": 0.00010971044681140437,
      "loss": 1.7066,
      "step": 13542
    },
    {
      "epoch": 0.48500367074328077,
      "grad_norm": 1.4403965473175049,
      "learning_rate": 0.0001096989025478529,
      "loss": 1.6519,
      "step": 13543
    },
    {
      "epoch": 0.48503948287284904,
      "grad_norm": 1.8270875215530396,
      "learning_rate": 0.0001096873581538152,
      "loss": 1.712,
      "step": 13544
    },
    {
      "epoch": 0.4850752950024173,
      "grad_norm": 1.5152555704116821,
      "learning_rate": 0.00010967581362944654,
      "loss": 1.6795,
      "step": 13545
    },
    {
      "epoch": 0.48511110713198563,
      "grad_norm": 1.2786191701889038,
      "learning_rate": 0.00010966426897490234,
      "loss": 1.5847,
      "step": 13546
    },
    {
      "epoch": 0.4851469192615539,
      "grad_norm": 1.5491951704025269,
      "learning_rate": 0.00010965272419033782,
      "loss": 1.3634,
      "step": 13547
    },
    {
      "epoch": 0.48518273139112217,
      "grad_norm": 2.059805393218994,
      "learning_rate": 0.0001096411792759084,
      "loss": 1.2069,
      "step": 13548
    },
    {
      "epoch": 0.48521854352069044,
      "grad_norm": 1.5590767860412598,
      "learning_rate": 0.0001096296342317693,
      "loss": 1.3519,
      "step": 13549
    },
    {
      "epoch": 0.48525435565025876,
      "grad_norm": 1.7230417728424072,
      "learning_rate": 0.00010961808905807593,
      "loss": 1.4459,
      "step": 13550
    },
    {
      "epoch": 0.48529016777982703,
      "grad_norm": 1.9240474700927734,
      "learning_rate": 0.00010960654375498357,
      "loss": 1.6497,
      "step": 13551
    },
    {
      "epoch": 0.4853259799093953,
      "grad_norm": 1.6086010932922363,
      "learning_rate": 0.00010959499832264754,
      "loss": 1.5178,
      "step": 13552
    },
    {
      "epoch": 0.4853617920389636,
      "grad_norm": 1.5947424173355103,
      "learning_rate": 0.00010958345276122322,
      "loss": 1.203,
      "step": 13553
    },
    {
      "epoch": 0.4853976041685319,
      "grad_norm": 1.5040249824523926,
      "learning_rate": 0.00010957190707086586,
      "loss": 1.3151,
      "step": 13554
    },
    {
      "epoch": 0.48543341629810016,
      "grad_norm": 1.2605984210968018,
      "learning_rate": 0.00010956036125173088,
      "loss": 1.5936,
      "step": 13555
    },
    {
      "epoch": 0.48546922842766843,
      "grad_norm": 1.7355071306228638,
      "learning_rate": 0.00010954881530397352,
      "loss": 1.5246,
      "step": 13556
    },
    {
      "epoch": 0.48550504055723676,
      "grad_norm": 1.455668568611145,
      "learning_rate": 0.0001095372692277492,
      "loss": 1.3897,
      "step": 13557
    },
    {
      "epoch": 0.485540852686805,
      "grad_norm": 1.6682887077331543,
      "learning_rate": 0.00010952572302321322,
      "loss": 1.4213,
      "step": 13558
    },
    {
      "epoch": 0.4855766648163733,
      "grad_norm": 1.7977643013000488,
      "learning_rate": 0.00010951417669052093,
      "loss": 1.573,
      "step": 13559
    },
    {
      "epoch": 0.4856124769459416,
      "grad_norm": 1.5340325832366943,
      "learning_rate": 0.00010950263022982766,
      "loss": 1.6457,
      "step": 13560
    },
    {
      "epoch": 0.4856482890755099,
      "grad_norm": 2.6738224029541016,
      "learning_rate": 0.0001094910836412888,
      "loss": 1.4636,
      "step": 13561
    },
    {
      "epoch": 0.48568410120507816,
      "grad_norm": 1.7948660850524902,
      "learning_rate": 0.00010947953692505959,
      "loss": 1.7762,
      "step": 13562
    },
    {
      "epoch": 0.4857199133346464,
      "grad_norm": 1.6173733472824097,
      "learning_rate": 0.00010946799008129547,
      "loss": 1.169,
      "step": 13563
    },
    {
      "epoch": 0.48575572546421475,
      "grad_norm": 2.0082993507385254,
      "learning_rate": 0.00010945644311015172,
      "loss": 1.8853,
      "step": 13564
    },
    {
      "epoch": 0.485791537593783,
      "grad_norm": 2.996189832687378,
      "learning_rate": 0.00010944489601178373,
      "loss": 1.7372,
      "step": 13565
    },
    {
      "epoch": 0.4858273497233513,
      "grad_norm": 1.859113097190857,
      "learning_rate": 0.0001094333487863469,
      "loss": 1.6007,
      "step": 13566
    },
    {
      "epoch": 0.48586316185291956,
      "grad_norm": 1.7060550451278687,
      "learning_rate": 0.00010942180143399647,
      "loss": 1.445,
      "step": 13567
    },
    {
      "epoch": 0.4858989739824879,
      "grad_norm": 2.118669271469116,
      "learning_rate": 0.0001094102539548879,
      "loss": 1.4383,
      "step": 13568
    },
    {
      "epoch": 0.48593478611205615,
      "grad_norm": 2.2744529247283936,
      "learning_rate": 0.00010939870634917647,
      "loss": 1.8303,
      "step": 13569
    },
    {
      "epoch": 0.4859705982416244,
      "grad_norm": 1.9283888339996338,
      "learning_rate": 0.00010938715861701762,
      "loss": 1.7504,
      "step": 13570
    },
    {
      "epoch": 0.48600641037119274,
      "grad_norm": 1.3578182458877563,
      "learning_rate": 0.00010937561075856662,
      "loss": 1.5591,
      "step": 13571
    },
    {
      "epoch": 0.486042222500761,
      "grad_norm": 1.8386645317077637,
      "learning_rate": 0.00010936406277397888,
      "loss": 1.5983,
      "step": 13572
    },
    {
      "epoch": 0.4860780346303293,
      "grad_norm": 1.6330554485321045,
      "learning_rate": 0.00010935251466340973,
      "loss": 1.4696,
      "step": 13573
    },
    {
      "epoch": 0.48611384675989755,
      "grad_norm": 1.2304561138153076,
      "learning_rate": 0.0001093409664270146,
      "loss": 1.3601,
      "step": 13574
    },
    {
      "epoch": 0.4861496588894659,
      "grad_norm": 1.8582013845443726,
      "learning_rate": 0.00010932941806494876,
      "loss": 1.6357,
      "step": 13575
    },
    {
      "epoch": 0.48618547101903414,
      "grad_norm": 1.4908101558685303,
      "learning_rate": 0.0001093178695773677,
      "loss": 1.2766,
      "step": 13576
    },
    {
      "epoch": 0.4862212831486024,
      "grad_norm": 1.5224658250808716,
      "learning_rate": 0.00010930632096442665,
      "loss": 1.6613,
      "step": 13577
    },
    {
      "epoch": 0.48625709527817074,
      "grad_norm": 1.5885058641433716,
      "learning_rate": 0.00010929477222628113,
      "loss": 1.4425,
      "step": 13578
    },
    {
      "epoch": 0.486292907407739,
      "grad_norm": 1.9274603128433228,
      "learning_rate": 0.00010928322336308641,
      "loss": 1.5363,
      "step": 13579
    },
    {
      "epoch": 0.4863287195373073,
      "grad_norm": 1.4560315608978271,
      "learning_rate": 0.00010927167437499788,
      "loss": 1.6627,
      "step": 13580
    },
    {
      "epoch": 0.48636453166687554,
      "grad_norm": 1.4085999727249146,
      "learning_rate": 0.00010926012526217095,
      "loss": 1.3984,
      "step": 13581
    },
    {
      "epoch": 0.48640034379644387,
      "grad_norm": 1.5395698547363281,
      "learning_rate": 0.00010924857602476095,
      "loss": 1.3167,
      "step": 13582
    },
    {
      "epoch": 0.48643615592601214,
      "grad_norm": 1.8653262853622437,
      "learning_rate": 0.00010923702666292333,
      "loss": 1.6573,
      "step": 13583
    },
    {
      "epoch": 0.4864719680555804,
      "grad_norm": 2.26827073097229,
      "learning_rate": 0.0001092254771768134,
      "loss": 1.6199,
      "step": 13584
    },
    {
      "epoch": 0.48650778018514873,
      "grad_norm": 1.4036325216293335,
      "learning_rate": 0.0001092139275665866,
      "loss": 1.6415,
      "step": 13585
    },
    {
      "epoch": 0.486543592314717,
      "grad_norm": 1.6470149755477905,
      "learning_rate": 0.00010920237783239824,
      "loss": 1.3443,
      "step": 13586
    },
    {
      "epoch": 0.48657940444428527,
      "grad_norm": 1.6262797117233276,
      "learning_rate": 0.0001091908279744038,
      "loss": 1.4092,
      "step": 13587
    },
    {
      "epoch": 0.48661521657385354,
      "grad_norm": 1.7407779693603516,
      "learning_rate": 0.00010917927799275865,
      "loss": 1.4334,
      "step": 13588
    },
    {
      "epoch": 0.48665102870342186,
      "grad_norm": 2.5387730598449707,
      "learning_rate": 0.00010916772788761809,
      "loss": 1.5939,
      "step": 13589
    },
    {
      "epoch": 0.48668684083299013,
      "grad_norm": 2.3920860290527344,
      "learning_rate": 0.00010915617765913761,
      "loss": 1.4822,
      "step": 13590
    },
    {
      "epoch": 0.4867226529625584,
      "grad_norm": 2.5543646812438965,
      "learning_rate": 0.00010914462730747257,
      "loss": 1.477,
      "step": 13591
    },
    {
      "epoch": 0.4867584650921267,
      "grad_norm": 1.4672547578811646,
      "learning_rate": 0.00010913307683277838,
      "loss": 1.4579,
      "step": 13592
    },
    {
      "epoch": 0.486794277221695,
      "grad_norm": 2.104875326156616,
      "learning_rate": 0.0001091215262352104,
      "loss": 1.5106,
      "step": 13593
    },
    {
      "epoch": 0.48683008935126326,
      "grad_norm": 1.5510029792785645,
      "learning_rate": 0.00010910997551492405,
      "loss": 1.319,
      "step": 13594
    },
    {
      "epoch": 0.48686590148083153,
      "grad_norm": 1.9528536796569824,
      "learning_rate": 0.00010909842467207472,
      "loss": 1.5796,
      "step": 13595
    },
    {
      "epoch": 0.48690171361039986,
      "grad_norm": 3.5056748390197754,
      "learning_rate": 0.00010908687370681785,
      "loss": 1.6121,
      "step": 13596
    },
    {
      "epoch": 0.4869375257399681,
      "grad_norm": 2.442275047302246,
      "learning_rate": 0.00010907532261930881,
      "loss": 1.5149,
      "step": 13597
    },
    {
      "epoch": 0.4869733378695364,
      "grad_norm": 2.3697030544281006,
      "learning_rate": 0.00010906377140970301,
      "loss": 1.6489,
      "step": 13598
    },
    {
      "epoch": 0.4870091499991047,
      "grad_norm": 2.148909568786621,
      "learning_rate": 0.00010905222007815585,
      "loss": 1.3923,
      "step": 13599
    },
    {
      "epoch": 0.487044962128673,
      "grad_norm": 2.649132490158081,
      "learning_rate": 0.00010904066862482274,
      "loss": 1.5161,
      "step": 13600
    },
    {
      "epoch": 0.48708077425824126,
      "grad_norm": 1.370690941810608,
      "learning_rate": 0.00010902911704985912,
      "loss": 1.5841,
      "step": 13601
    },
    {
      "epoch": 0.4871165863878095,
      "grad_norm": 2.150310516357422,
      "learning_rate": 0.00010901756535342033,
      "loss": 1.7005,
      "step": 13602
    },
    {
      "epoch": 0.48715239851737785,
      "grad_norm": 1.4429428577423096,
      "learning_rate": 0.00010900601353566188,
      "loss": 1.5266,
      "step": 13603
    },
    {
      "epoch": 0.4871882106469461,
      "grad_norm": 1.5756672620773315,
      "learning_rate": 0.0001089944615967391,
      "loss": 1.3184,
      "step": 13604
    },
    {
      "epoch": 0.4872240227765144,
      "grad_norm": 1.7169619798660278,
      "learning_rate": 0.0001089829095368075,
      "loss": 1.4519,
      "step": 13605
    },
    {
      "epoch": 0.4872598349060827,
      "grad_norm": 1.8188672065734863,
      "learning_rate": 0.00010897135735602238,
      "loss": 1.3889,
      "step": 13606
    },
    {
      "epoch": 0.487295647035651,
      "grad_norm": 1.3726422786712646,
      "learning_rate": 0.00010895980505453924,
      "loss": 1.2956,
      "step": 13607
    },
    {
      "epoch": 0.48733145916521925,
      "grad_norm": 1.986498236656189,
      "learning_rate": 0.00010894825263251345,
      "loss": 1.7822,
      "step": 13608
    },
    {
      "epoch": 0.4873672712947875,
      "grad_norm": 1.6833266019821167,
      "learning_rate": 0.00010893670009010049,
      "loss": 1.4204,
      "step": 13609
    },
    {
      "epoch": 0.48740308342435584,
      "grad_norm": 1.488139271736145,
      "learning_rate": 0.00010892514742745576,
      "loss": 1.1151,
      "step": 13610
    },
    {
      "epoch": 0.4874388955539241,
      "grad_norm": 1.7593934535980225,
      "learning_rate": 0.00010891359464473468,
      "loss": 1.5433,
      "step": 13611
    },
    {
      "epoch": 0.4874747076834924,
      "grad_norm": 1.4429570436477661,
      "learning_rate": 0.00010890204174209269,
      "loss": 1.4472,
      "step": 13612
    },
    {
      "epoch": 0.4875105198130607,
      "grad_norm": 1.4732005596160889,
      "learning_rate": 0.00010889048871968517,
      "loss": 1.4272,
      "step": 13613
    },
    {
      "epoch": 0.487546331942629,
      "grad_norm": 2.193288564682007,
      "learning_rate": 0.00010887893557766766,
      "loss": 1.7031,
      "step": 13614
    },
    {
      "epoch": 0.48758214407219724,
      "grad_norm": 1.877280592918396,
      "learning_rate": 0.00010886738231619549,
      "loss": 1.6894,
      "step": 13615
    },
    {
      "epoch": 0.4876179562017655,
      "grad_norm": 1.6814889907836914,
      "learning_rate": 0.00010885582893542411,
      "loss": 1.5371,
      "step": 13616
    },
    {
      "epoch": 0.48765376833133384,
      "grad_norm": 1.7409570217132568,
      "learning_rate": 0.00010884427543550899,
      "loss": 1.8072,
      "step": 13617
    },
    {
      "epoch": 0.4876895804609021,
      "grad_norm": 1.5650906562805176,
      "learning_rate": 0.00010883272181660558,
      "loss": 1.2065,
      "step": 13618
    },
    {
      "epoch": 0.4877253925904704,
      "grad_norm": 1.6816296577453613,
      "learning_rate": 0.00010882116807886924,
      "loss": 1.491,
      "step": 13619
    },
    {
      "epoch": 0.4877612047200387,
      "grad_norm": 1.757469892501831,
      "learning_rate": 0.0001088096142224555,
      "loss": 1.5318,
      "step": 13620
    },
    {
      "epoch": 0.48779701684960697,
      "grad_norm": 1.3962442874908447,
      "learning_rate": 0.00010879806024751975,
      "loss": 1.3877,
      "step": 13621
    },
    {
      "epoch": 0.48783282897917524,
      "grad_norm": 2.594102144241333,
      "learning_rate": 0.00010878650615421744,
      "loss": 1.4608,
      "step": 13622
    },
    {
      "epoch": 0.4878686411087435,
      "grad_norm": 1.906245231628418,
      "learning_rate": 0.00010877495194270407,
      "loss": 1.5072,
      "step": 13623
    },
    {
      "epoch": 0.48790445323831183,
      "grad_norm": 1.6580541133880615,
      "learning_rate": 0.00010876339761313499,
      "loss": 1.4024,
      "step": 13624
    },
    {
      "epoch": 0.4879402653678801,
      "grad_norm": 1.6919373273849487,
      "learning_rate": 0.00010875184316566571,
      "loss": 1.5576,
      "step": 13625
    },
    {
      "epoch": 0.48797607749744837,
      "grad_norm": 1.9276906251907349,
      "learning_rate": 0.00010874028860045166,
      "loss": 1.4877,
      "step": 13626
    },
    {
      "epoch": 0.4880118896270167,
      "grad_norm": 1.4505412578582764,
      "learning_rate": 0.00010872873391764833,
      "loss": 1.4795,
      "step": 13627
    },
    {
      "epoch": 0.48804770175658496,
      "grad_norm": 1.4646049737930298,
      "learning_rate": 0.00010871717911741113,
      "loss": 1.3365,
      "step": 13628
    },
    {
      "epoch": 0.48808351388615323,
      "grad_norm": 1.593316674232483,
      "learning_rate": 0.00010870562419989552,
      "loss": 1.4228,
      "step": 13629
    },
    {
      "epoch": 0.4881193260157215,
      "grad_norm": 1.555849552154541,
      "learning_rate": 0.00010869406916525698,
      "loss": 1.5343,
      "step": 13630
    },
    {
      "epoch": 0.4881551381452898,
      "grad_norm": 2.0024313926696777,
      "learning_rate": 0.00010868251401365095,
      "loss": 1.4696,
      "step": 13631
    },
    {
      "epoch": 0.4881909502748581,
      "grad_norm": 1.810181736946106,
      "learning_rate": 0.0001086709587452329,
      "loss": 1.3037,
      "step": 13632
    },
    {
      "epoch": 0.48822676240442636,
      "grad_norm": 2.0608158111572266,
      "learning_rate": 0.00010865940336015828,
      "loss": 1.5016,
      "step": 13633
    },
    {
      "epoch": 0.4882625745339947,
      "grad_norm": 1.37624192237854,
      "learning_rate": 0.00010864784785858256,
      "loss": 1.296,
      "step": 13634
    },
    {
      "epoch": 0.48829838666356296,
      "grad_norm": 1.3577290773391724,
      "learning_rate": 0.00010863629224066116,
      "loss": 1.3755,
      "step": 13635
    },
    {
      "epoch": 0.4883341987931312,
      "grad_norm": 1.884279727935791,
      "learning_rate": 0.00010862473650654965,
      "loss": 1.6172,
      "step": 13636
    },
    {
      "epoch": 0.4883700109226995,
      "grad_norm": 1.4926834106445312,
      "learning_rate": 0.00010861318065640338,
      "loss": 1.3022,
      "step": 13637
    },
    {
      "epoch": 0.4884058230522678,
      "grad_norm": 3.462437629699707,
      "learning_rate": 0.00010860162469037792,
      "loss": 1.6107,
      "step": 13638
    },
    {
      "epoch": 0.4884416351818361,
      "grad_norm": 2.0196216106414795,
      "learning_rate": 0.00010859006860862865,
      "loss": 1.5062,
      "step": 13639
    },
    {
      "epoch": 0.48847744731140436,
      "grad_norm": 1.4165526628494263,
      "learning_rate": 0.00010857851241131114,
      "loss": 1.389,
      "step": 13640
    },
    {
      "epoch": 0.4885132594409727,
      "grad_norm": 1.861929178237915,
      "learning_rate": 0.0001085669560985808,
      "loss": 1.6056,
      "step": 13641
    },
    {
      "epoch": 0.48854907157054095,
      "grad_norm": 2.1622304916381836,
      "learning_rate": 0.0001085553996705931,
      "loss": 1.9087,
      "step": 13642
    },
    {
      "epoch": 0.4885848837001092,
      "grad_norm": 1.8672003746032715,
      "learning_rate": 0.00010854384312750354,
      "loss": 1.3609,
      "step": 13643
    },
    {
      "epoch": 0.4886206958296775,
      "grad_norm": 2.506395101547241,
      "learning_rate": 0.00010853228646946758,
      "loss": 1.2844,
      "step": 13644
    },
    {
      "epoch": 0.4886565079592458,
      "grad_norm": 1.754264235496521,
      "learning_rate": 0.00010852072969664073,
      "loss": 1.551,
      "step": 13645
    },
    {
      "epoch": 0.4886923200888141,
      "grad_norm": 1.666245698928833,
      "learning_rate": 0.00010850917280917843,
      "loss": 1.0954,
      "step": 13646
    },
    {
      "epoch": 0.48872813221838235,
      "grad_norm": 1.6731163263320923,
      "learning_rate": 0.0001084976158072362,
      "loss": 1.7465,
      "step": 13647
    },
    {
      "epoch": 0.4887639443479507,
      "grad_norm": 1.6194301843643188,
      "learning_rate": 0.0001084860586909695,
      "loss": 1.1302,
      "step": 13648
    },
    {
      "epoch": 0.48879975647751894,
      "grad_norm": 1.5961239337921143,
      "learning_rate": 0.00010847450146053386,
      "loss": 1.3433,
      "step": 13649
    },
    {
      "epoch": 0.4888355686070872,
      "grad_norm": 1.8925201892852783,
      "learning_rate": 0.0001084629441160847,
      "loss": 1.5822,
      "step": 13650
    },
    {
      "epoch": 0.4888713807366555,
      "grad_norm": 1.876792550086975,
      "learning_rate": 0.00010845138665777754,
      "loss": 1.7623,
      "step": 13651
    },
    {
      "epoch": 0.4889071928662238,
      "grad_norm": 1.422800898551941,
      "learning_rate": 0.0001084398290857679,
      "loss": 1.3395,
      "step": 13652
    },
    {
      "epoch": 0.4889430049957921,
      "grad_norm": 2.462639808654785,
      "learning_rate": 0.00010842827140021121,
      "loss": 1.5845,
      "step": 13653
    },
    {
      "epoch": 0.48897881712536034,
      "grad_norm": 1.4146637916564941,
      "learning_rate": 0.00010841671360126304,
      "loss": 1.4434,
      "step": 13654
    },
    {
      "epoch": 0.48901462925492867,
      "grad_norm": 1.4680838584899902,
      "learning_rate": 0.0001084051556890788,
      "loss": 1.5377,
      "step": 13655
    },
    {
      "epoch": 0.48905044138449694,
      "grad_norm": 1.9159295558929443,
      "learning_rate": 0.0001083935976638141,
      "loss": 1.5379,
      "step": 13656
    },
    {
      "epoch": 0.4890862535140652,
      "grad_norm": 1.6047742366790771,
      "learning_rate": 0.00010838203952562432,
      "loss": 1.4923,
      "step": 13657
    },
    {
      "epoch": 0.4891220656436335,
      "grad_norm": 1.7296887636184692,
      "learning_rate": 0.00010837048127466505,
      "loss": 1.2314,
      "step": 13658
    },
    {
      "epoch": 0.4891578777732018,
      "grad_norm": 1.9508219957351685,
      "learning_rate": 0.00010835892291109169,
      "loss": 1.4367,
      "step": 13659
    },
    {
      "epoch": 0.48919368990277007,
      "grad_norm": 1.9842190742492676,
      "learning_rate": 0.00010834736443505986,
      "loss": 1.613,
      "step": 13660
    },
    {
      "epoch": 0.48922950203233834,
      "grad_norm": 1.6061060428619385,
      "learning_rate": 0.00010833580584672496,
      "loss": 1.2524,
      "step": 13661
    },
    {
      "epoch": 0.48926531416190666,
      "grad_norm": 1.3053823709487915,
      "learning_rate": 0.00010832424714624259,
      "loss": 1.4031,
      "step": 13662
    },
    {
      "epoch": 0.48930112629147493,
      "grad_norm": 1.4019272327423096,
      "learning_rate": 0.00010831268833376817,
      "loss": 1.3219,
      "step": 13663
    },
    {
      "epoch": 0.4893369384210432,
      "grad_norm": 1.8044565916061401,
      "learning_rate": 0.00010830112940945726,
      "loss": 1.5856,
      "step": 13664
    },
    {
      "epoch": 0.48937275055061147,
      "grad_norm": 2.211778163909912,
      "learning_rate": 0.00010828957037346538,
      "loss": 1.3458,
      "step": 13665
    },
    {
      "epoch": 0.4894085626801798,
      "grad_norm": 1.8323140144348145,
      "learning_rate": 0.00010827801122594802,
      "loss": 1.2703,
      "step": 13666
    },
    {
      "epoch": 0.48944437480974806,
      "grad_norm": 1.6370885372161865,
      "learning_rate": 0.00010826645196706074,
      "loss": 1.4509,
      "step": 13667
    },
    {
      "epoch": 0.48948018693931633,
      "grad_norm": 1.6698061227798462,
      "learning_rate": 0.00010825489259695894,
      "loss": 1.5467,
      "step": 13668
    },
    {
      "epoch": 0.48951599906888466,
      "grad_norm": 1.5696372985839844,
      "learning_rate": 0.00010824333311579824,
      "loss": 1.6898,
      "step": 13669
    },
    {
      "epoch": 0.4895518111984529,
      "grad_norm": 2.0624501705169678,
      "learning_rate": 0.00010823177352373412,
      "loss": 1.5877,
      "step": 13670
    },
    {
      "epoch": 0.4895876233280212,
      "grad_norm": 1.8093184232711792,
      "learning_rate": 0.00010822021382092211,
      "loss": 1.575,
      "step": 13671
    },
    {
      "epoch": 0.48962343545758946,
      "grad_norm": 1.586574912071228,
      "learning_rate": 0.00010820865400751772,
      "loss": 1.5977,
      "step": 13672
    },
    {
      "epoch": 0.4896592475871578,
      "grad_norm": 2.6463475227355957,
      "learning_rate": 0.00010819709408367649,
      "loss": 1.5361,
      "step": 13673
    },
    {
      "epoch": 0.48969505971672606,
      "grad_norm": 1.738493800163269,
      "learning_rate": 0.00010818553404955391,
      "loss": 1.6058,
      "step": 13674
    },
    {
      "epoch": 0.4897308718462943,
      "grad_norm": 1.6610356569290161,
      "learning_rate": 0.00010817397390530555,
      "loss": 1.608,
      "step": 13675
    },
    {
      "epoch": 0.48976668397586265,
      "grad_norm": 1.5615880489349365,
      "learning_rate": 0.00010816241365108692,
      "loss": 1.6981,
      "step": 13676
    },
    {
      "epoch": 0.4898024961054309,
      "grad_norm": 1.5565402507781982,
      "learning_rate": 0.00010815085328705352,
      "loss": 1.3005,
      "step": 13677
    },
    {
      "epoch": 0.4898383082349992,
      "grad_norm": 1.6391171216964722,
      "learning_rate": 0.00010813929281336092,
      "loss": 1.4786,
      "step": 13678
    },
    {
      "epoch": 0.48987412036456746,
      "grad_norm": 1.5482856035232544,
      "learning_rate": 0.00010812773223016461,
      "loss": 1.4095,
      "step": 13679
    },
    {
      "epoch": 0.4899099324941358,
      "grad_norm": 1.6968822479248047,
      "learning_rate": 0.00010811617153762017,
      "loss": 1.5978,
      "step": 13680
    },
    {
      "epoch": 0.48994574462370405,
      "grad_norm": 1.4903837442398071,
      "learning_rate": 0.0001081046107358831,
      "loss": 1.3526,
      "step": 13681
    },
    {
      "epoch": 0.4899815567532723,
      "grad_norm": 1.5468765497207642,
      "learning_rate": 0.00010809304982510897,
      "loss": 1.487,
      "step": 13682
    },
    {
      "epoch": 0.49001736888284064,
      "grad_norm": 2.329375743865967,
      "learning_rate": 0.00010808148880545325,
      "loss": 1.346,
      "step": 13683
    },
    {
      "epoch": 0.4900531810124089,
      "grad_norm": 1.809410810470581,
      "learning_rate": 0.00010806992767707155,
      "loss": 1.8127,
      "step": 13684
    },
    {
      "epoch": 0.4900889931419772,
      "grad_norm": 2.0929079055786133,
      "learning_rate": 0.00010805836644011939,
      "loss": 1.697,
      "step": 13685
    },
    {
      "epoch": 0.49012480527154545,
      "grad_norm": 1.9662256240844727,
      "learning_rate": 0.00010804680509475229,
      "loss": 1.3781,
      "step": 13686
    },
    {
      "epoch": 0.4901606174011138,
      "grad_norm": 1.429788589477539,
      "learning_rate": 0.00010803524364112583,
      "loss": 1.3815,
      "step": 13687
    },
    {
      "epoch": 0.49019642953068204,
      "grad_norm": 1.752447247505188,
      "learning_rate": 0.0001080236820793955,
      "loss": 1.3346,
      "step": 13688
    },
    {
      "epoch": 0.4902322416602503,
      "grad_norm": 1.527882695198059,
      "learning_rate": 0.00010801212040971691,
      "loss": 1.5413,
      "step": 13689
    },
    {
      "epoch": 0.49026805378981864,
      "grad_norm": 1.3394904136657715,
      "learning_rate": 0.00010800055863224555,
      "loss": 1.4344,
      "step": 13690
    },
    {
      "epoch": 0.4903038659193869,
      "grad_norm": 1.955166220664978,
      "learning_rate": 0.00010798899674713699,
      "loss": 1.4497,
      "step": 13691
    },
    {
      "epoch": 0.4903396780489552,
      "grad_norm": 1.8119986057281494,
      "learning_rate": 0.00010797743475454678,
      "loss": 1.7094,
      "step": 13692
    },
    {
      "epoch": 0.49037549017852344,
      "grad_norm": 1.3964793682098389,
      "learning_rate": 0.0001079658726546305,
      "loss": 1.4808,
      "step": 13693
    },
    {
      "epoch": 0.49041130230809177,
      "grad_norm": 2.0613391399383545,
      "learning_rate": 0.00010795431044754367,
      "loss": 1.2227,
      "step": 13694
    },
    {
      "epoch": 0.49044711443766004,
      "grad_norm": 1.5445024967193604,
      "learning_rate": 0.00010794274813344185,
      "loss": 1.5322,
      "step": 13695
    },
    {
      "epoch": 0.4904829265672283,
      "grad_norm": 1.8224143981933594,
      "learning_rate": 0.0001079311857124806,
      "loss": 1.3995,
      "step": 13696
    },
    {
      "epoch": 0.49051873869679663,
      "grad_norm": 2.142685890197754,
      "learning_rate": 0.00010791962318481547,
      "loss": 1.7053,
      "step": 13697
    },
    {
      "epoch": 0.4905545508263649,
      "grad_norm": 1.9301700592041016,
      "learning_rate": 0.00010790806055060205,
      "loss": 1.7767,
      "step": 13698
    },
    {
      "epoch": 0.49059036295593317,
      "grad_norm": 2.0079996585845947,
      "learning_rate": 0.00010789649780999585,
      "loss": 1.6479,
      "step": 13699
    },
    {
      "epoch": 0.49062617508550144,
      "grad_norm": 1.3682293891906738,
      "learning_rate": 0.00010788493496315246,
      "loss": 1.2197,
      "step": 13700
    },
    {
      "epoch": 0.49066198721506976,
      "grad_norm": 1.5623469352722168,
      "learning_rate": 0.00010787337201022745,
      "loss": 1.6502,
      "step": 13701
    },
    {
      "epoch": 0.49069779934463803,
      "grad_norm": 1.6920688152313232,
      "learning_rate": 0.00010786180895137639,
      "loss": 1.2194,
      "step": 13702
    },
    {
      "epoch": 0.4907336114742063,
      "grad_norm": 1.5415152311325073,
      "learning_rate": 0.0001078502457867548,
      "loss": 1.4212,
      "step": 13703
    },
    {
      "epoch": 0.4907694236037746,
      "grad_norm": 1.6158078908920288,
      "learning_rate": 0.00010783868251651833,
      "loss": 1.3743,
      "step": 13704
    },
    {
      "epoch": 0.4908052357333429,
      "grad_norm": 1.6433581113815308,
      "learning_rate": 0.00010782711914082242,
      "loss": 1.4155,
      "step": 13705
    },
    {
      "epoch": 0.49084104786291116,
      "grad_norm": 1.4370219707489014,
      "learning_rate": 0.00010781555565982276,
      "loss": 1.429,
      "step": 13706
    },
    {
      "epoch": 0.49087685999247943,
      "grad_norm": 1.5507996082305908,
      "learning_rate": 0.00010780399207367489,
      "loss": 1.1886,
      "step": 13707
    },
    {
      "epoch": 0.49091267212204776,
      "grad_norm": 1.837388515472412,
      "learning_rate": 0.00010779242838253433,
      "loss": 1.5502,
      "step": 13708
    },
    {
      "epoch": 0.490948484251616,
      "grad_norm": 1.5387765169143677,
      "learning_rate": 0.00010778086458655677,
      "loss": 1.5225,
      "step": 13709
    },
    {
      "epoch": 0.4909842963811843,
      "grad_norm": 2.5038530826568604,
      "learning_rate": 0.00010776930068589764,
      "loss": 1.6695,
      "step": 13710
    },
    {
      "epoch": 0.4910201085107526,
      "grad_norm": 2.6884608268737793,
      "learning_rate": 0.00010775773668071265,
      "loss": 1.4632,
      "step": 13711
    },
    {
      "epoch": 0.4910559206403209,
      "grad_norm": 1.9262381792068481,
      "learning_rate": 0.00010774617257115728,
      "loss": 1.0545,
      "step": 13712
    },
    {
      "epoch": 0.49109173276988916,
      "grad_norm": 1.8971209526062012,
      "learning_rate": 0.00010773460835738718,
      "loss": 1.4246,
      "step": 13713
    },
    {
      "epoch": 0.4911275448994574,
      "grad_norm": 1.7927980422973633,
      "learning_rate": 0.00010772304403955789,
      "loss": 1.4387,
      "step": 13714
    },
    {
      "epoch": 0.49116335702902575,
      "grad_norm": 1.5163729190826416,
      "learning_rate": 0.000107711479617825,
      "loss": 1.3497,
      "step": 13715
    },
    {
      "epoch": 0.491199169158594,
      "grad_norm": 1.4729920625686646,
      "learning_rate": 0.00010769991509234408,
      "loss": 1.6522,
      "step": 13716
    },
    {
      "epoch": 0.4912349812881623,
      "grad_norm": 1.7279658317565918,
      "learning_rate": 0.00010768835046327077,
      "loss": 1.1325,
      "step": 13717
    },
    {
      "epoch": 0.4912707934177306,
      "grad_norm": 1.7967742681503296,
      "learning_rate": 0.00010767678573076058,
      "loss": 1.6004,
      "step": 13718
    },
    {
      "epoch": 0.4913066055472989,
      "grad_norm": 2.122061252593994,
      "learning_rate": 0.00010766522089496915,
      "loss": 1.5491,
      "step": 13719
    },
    {
      "epoch": 0.49134241767686715,
      "grad_norm": 1.501920461654663,
      "learning_rate": 0.00010765365595605212,
      "loss": 1.5364,
      "step": 13720
    },
    {
      "epoch": 0.4913782298064354,
      "grad_norm": 1.8357949256896973,
      "learning_rate": 0.00010764209091416497,
      "loss": 1.5825,
      "step": 13721
    },
    {
      "epoch": 0.49141404193600374,
      "grad_norm": 1.3038849830627441,
      "learning_rate": 0.00010763052576946335,
      "loss": 1.5351,
      "step": 13722
    },
    {
      "epoch": 0.491449854065572,
      "grad_norm": 2.2974419593811035,
      "learning_rate": 0.00010761896052210285,
      "loss": 1.3127,
      "step": 13723
    },
    {
      "epoch": 0.4914856661951403,
      "grad_norm": 1.4514243602752686,
      "learning_rate": 0.00010760739517223908,
      "loss": 1.3836,
      "step": 13724
    },
    {
      "epoch": 0.4915214783247086,
      "grad_norm": 2.4426000118255615,
      "learning_rate": 0.00010759582972002758,
      "loss": 1.7502,
      "step": 13725
    },
    {
      "epoch": 0.4915572904542769,
      "grad_norm": 1.43705153465271,
      "learning_rate": 0.00010758426416562402,
      "loss": 1.6914,
      "step": 13726
    },
    {
      "epoch": 0.49159310258384514,
      "grad_norm": 2.030698776245117,
      "learning_rate": 0.00010757269850918394,
      "loss": 1.3523,
      "step": 13727
    },
    {
      "epoch": 0.4916289147134134,
      "grad_norm": 2.493643045425415,
      "learning_rate": 0.00010756113275086302,
      "loss": 1.5557,
      "step": 13728
    },
    {
      "epoch": 0.49166472684298174,
      "grad_norm": 2.235466241836548,
      "learning_rate": 0.00010754956689081678,
      "loss": 1.7038,
      "step": 13729
    },
    {
      "epoch": 0.49170053897255,
      "grad_norm": 1.6229029893875122,
      "learning_rate": 0.00010753800092920086,
      "loss": 1.5441,
      "step": 13730
    },
    {
      "epoch": 0.4917363511021183,
      "grad_norm": 1.7201504707336426,
      "learning_rate": 0.00010752643486617086,
      "loss": 1.6628,
      "step": 13731
    },
    {
      "epoch": 0.4917721632316866,
      "grad_norm": 2.171365737915039,
      "learning_rate": 0.00010751486870188239,
      "loss": 1.4995,
      "step": 13732
    },
    {
      "epoch": 0.49180797536125487,
      "grad_norm": 2.607131004333496,
      "learning_rate": 0.00010750330243649104,
      "loss": 1.6046,
      "step": 13733
    },
    {
      "epoch": 0.49184378749082314,
      "grad_norm": 2.4226737022399902,
      "learning_rate": 0.00010749173607015247,
      "loss": 1.4492,
      "step": 13734
    },
    {
      "epoch": 0.4918795996203914,
      "grad_norm": 1.5893970727920532,
      "learning_rate": 0.00010748016960302223,
      "loss": 1.403,
      "step": 13735
    },
    {
      "epoch": 0.49191541174995973,
      "grad_norm": 1.5359864234924316,
      "learning_rate": 0.00010746860303525595,
      "loss": 1.6285,
      "step": 13736
    },
    {
      "epoch": 0.491951223879528,
      "grad_norm": 2.033862829208374,
      "learning_rate": 0.00010745703636700926,
      "loss": 1.6867,
      "step": 13737
    },
    {
      "epoch": 0.49198703600909627,
      "grad_norm": 1.6129465103149414,
      "learning_rate": 0.00010744546959843777,
      "loss": 1.6712,
      "step": 13738
    },
    {
      "epoch": 0.4920228481386646,
      "grad_norm": 1.5458647012710571,
      "learning_rate": 0.00010743390272969706,
      "loss": 1.4477,
      "step": 13739
    },
    {
      "epoch": 0.49205866026823286,
      "grad_norm": 1.4577420949935913,
      "learning_rate": 0.00010742233576094283,
      "loss": 1.3451,
      "step": 13740
    },
    {
      "epoch": 0.49209447239780113,
      "grad_norm": 2.2439849376678467,
      "learning_rate": 0.0001074107686923306,
      "loss": 1.5242,
      "step": 13741
    },
    {
      "epoch": 0.4921302845273694,
      "grad_norm": 1.4966247081756592,
      "learning_rate": 0.00010739920152401605,
      "loss": 1.8109,
      "step": 13742
    },
    {
      "epoch": 0.4921660966569377,
      "grad_norm": 1.5827927589416504,
      "learning_rate": 0.00010738763425615479,
      "loss": 1.2532,
      "step": 13743
    },
    {
      "epoch": 0.492201908786506,
      "grad_norm": 1.451015591621399,
      "learning_rate": 0.00010737606688890245,
      "loss": 1.5968,
      "step": 13744
    },
    {
      "epoch": 0.49223772091607426,
      "grad_norm": 1.5125499963760376,
      "learning_rate": 0.00010736449942241465,
      "loss": 1.3855,
      "step": 13745
    },
    {
      "epoch": 0.4922735330456426,
      "grad_norm": 1.5985307693481445,
      "learning_rate": 0.000107352931856847,
      "loss": 1.4018,
      "step": 13746
    },
    {
      "epoch": 0.49230934517521086,
      "grad_norm": 1.6459596157073975,
      "learning_rate": 0.00010734136419235512,
      "loss": 1.4337,
      "step": 13747
    },
    {
      "epoch": 0.4923451573047791,
      "grad_norm": 2.634399652481079,
      "learning_rate": 0.00010732979642909466,
      "loss": 1.4944,
      "step": 13748
    },
    {
      "epoch": 0.4923809694343474,
      "grad_norm": 1.453285574913025,
      "learning_rate": 0.00010731822856722127,
      "loss": 1.595,
      "step": 13749
    },
    {
      "epoch": 0.4924167815639157,
      "grad_norm": 2.0285205841064453,
      "learning_rate": 0.00010730666060689053,
      "loss": 1.8837,
      "step": 13750
    },
    {
      "epoch": 0.492452593693484,
      "grad_norm": 1.861598253250122,
      "learning_rate": 0.00010729509254825811,
      "loss": 1.6692,
      "step": 13751
    },
    {
      "epoch": 0.49248840582305226,
      "grad_norm": 2.062983751296997,
      "learning_rate": 0.00010728352439147959,
      "loss": 1.6496,
      "step": 13752
    },
    {
      "epoch": 0.4925242179526206,
      "grad_norm": 2.475240707397461,
      "learning_rate": 0.00010727195613671071,
      "loss": 1.6116,
      "step": 13753
    },
    {
      "epoch": 0.49256003008218885,
      "grad_norm": 1.5981770753860474,
      "learning_rate": 0.00010726038778410699,
      "loss": 1.3383,
      "step": 13754
    },
    {
      "epoch": 0.4925958422117571,
      "grad_norm": 1.3601890802383423,
      "learning_rate": 0.00010724881933382416,
      "loss": 1.55,
      "step": 13755
    },
    {
      "epoch": 0.4926316543413254,
      "grad_norm": 1.8555890321731567,
      "learning_rate": 0.00010723725078601778,
      "loss": 1.701,
      "step": 13756
    },
    {
      "epoch": 0.4926674664708937,
      "grad_norm": 1.4763880968093872,
      "learning_rate": 0.00010722568214084354,
      "loss": 1.7419,
      "step": 13757
    },
    {
      "epoch": 0.492703278600462,
      "grad_norm": 1.3820867538452148,
      "learning_rate": 0.00010721411339845707,
      "loss": 1.4405,
      "step": 13758
    },
    {
      "epoch": 0.49273909073003025,
      "grad_norm": 1.5505809783935547,
      "learning_rate": 0.00010720254455901399,
      "loss": 1.3917,
      "step": 13759
    },
    {
      "epoch": 0.4927749028595986,
      "grad_norm": 3.055396556854248,
      "learning_rate": 0.00010719097562266998,
      "loss": 1.2911,
      "step": 13760
    },
    {
      "epoch": 0.49281071498916684,
      "grad_norm": 1.4654674530029297,
      "learning_rate": 0.00010717940658958066,
      "loss": 1.5803,
      "step": 13761
    },
    {
      "epoch": 0.4928465271187351,
      "grad_norm": 1.8840539455413818,
      "learning_rate": 0.00010716783745990169,
      "loss": 1.2817,
      "step": 13762
    },
    {
      "epoch": 0.4928823392483034,
      "grad_norm": 1.674445390701294,
      "learning_rate": 0.0001071562682337887,
      "loss": 1.5135,
      "step": 13763
    },
    {
      "epoch": 0.4929181513778717,
      "grad_norm": 1.557632327079773,
      "learning_rate": 0.0001071446989113974,
      "loss": 1.4121,
      "step": 13764
    },
    {
      "epoch": 0.49295396350744,
      "grad_norm": 1.402957558631897,
      "learning_rate": 0.00010713312949288334,
      "loss": 1.3646,
      "step": 13765
    },
    {
      "epoch": 0.49298977563700824,
      "grad_norm": 1.951153039932251,
      "learning_rate": 0.00010712155997840225,
      "loss": 1.468,
      "step": 13766
    },
    {
      "epoch": 0.4930255877665765,
      "grad_norm": 1.7936666011810303,
      "learning_rate": 0.00010710999036810975,
      "loss": 1.123,
      "step": 13767
    },
    {
      "epoch": 0.49306139989614484,
      "grad_norm": 1.553545594215393,
      "learning_rate": 0.00010709842066216151,
      "loss": 1.1884,
      "step": 13768
    },
    {
      "epoch": 0.4930972120257131,
      "grad_norm": 1.3447620868682861,
      "learning_rate": 0.00010708685086071316,
      "loss": 1.2112,
      "step": 13769
    },
    {
      "epoch": 0.4931330241552814,
      "grad_norm": 2.356231212615967,
      "learning_rate": 0.00010707528096392038,
      "loss": 1.3566,
      "step": 13770
    },
    {
      "epoch": 0.4931688362848497,
      "grad_norm": 1.819908618927002,
      "learning_rate": 0.00010706371097193881,
      "loss": 1.7402,
      "step": 13771
    },
    {
      "epoch": 0.49320464841441797,
      "grad_norm": 2.1119303703308105,
      "learning_rate": 0.00010705214088492415,
      "loss": 1.3566,
      "step": 13772
    },
    {
      "epoch": 0.49324046054398624,
      "grad_norm": 1.6259195804595947,
      "learning_rate": 0.00010704057070303201,
      "loss": 1.4016,
      "step": 13773
    },
    {
      "epoch": 0.4932762726735545,
      "grad_norm": 1.7888505458831787,
      "learning_rate": 0.00010702900042641806,
      "loss": 1.5483,
      "step": 13774
    },
    {
      "epoch": 0.49331208480312283,
      "grad_norm": 1.7072391510009766,
      "learning_rate": 0.00010701743005523801,
      "loss": 1.4431,
      "step": 13775
    },
    {
      "epoch": 0.4933478969326911,
      "grad_norm": 2.1504554748535156,
      "learning_rate": 0.00010700585958964744,
      "loss": 1.7045,
      "step": 13776
    },
    {
      "epoch": 0.49338370906225937,
      "grad_norm": 2.100933790206909,
      "learning_rate": 0.00010699428902980211,
      "loss": 1.7934,
      "step": 13777
    },
    {
      "epoch": 0.4934195211918277,
      "grad_norm": 1.4316816329956055,
      "learning_rate": 0.00010698271837585762,
      "loss": 1.3737,
      "step": 13778
    },
    {
      "epoch": 0.49345533332139596,
      "grad_norm": 1.4523588418960571,
      "learning_rate": 0.0001069711476279697,
      "loss": 1.533,
      "step": 13779
    },
    {
      "epoch": 0.49349114545096423,
      "grad_norm": 1.4418319463729858,
      "learning_rate": 0.00010695957678629391,
      "loss": 1.3883,
      "step": 13780
    },
    {
      "epoch": 0.4935269575805325,
      "grad_norm": 1.8043620586395264,
      "learning_rate": 0.00010694800585098606,
      "loss": 1.4568,
      "step": 13781
    },
    {
      "epoch": 0.4935627697101008,
      "grad_norm": 1.7246334552764893,
      "learning_rate": 0.00010693643482220173,
      "loss": 1.6841,
      "step": 13782
    },
    {
      "epoch": 0.4935985818396691,
      "grad_norm": 1.7109289169311523,
      "learning_rate": 0.0001069248637000966,
      "loss": 1.3743,
      "step": 13783
    },
    {
      "epoch": 0.49363439396923736,
      "grad_norm": 1.5216609239578247,
      "learning_rate": 0.0001069132924848264,
      "loss": 1.3224,
      "step": 13784
    },
    {
      "epoch": 0.4936702060988057,
      "grad_norm": 1.9945144653320312,
      "learning_rate": 0.00010690172117654672,
      "loss": 1.5012,
      "step": 13785
    },
    {
      "epoch": 0.49370601822837396,
      "grad_norm": 1.4590290784835815,
      "learning_rate": 0.00010689014977541332,
      "loss": 1.408,
      "step": 13786
    },
    {
      "epoch": 0.4937418303579422,
      "grad_norm": 1.3395782709121704,
      "learning_rate": 0.00010687857828158182,
      "loss": 1.5821,
      "step": 13787
    },
    {
      "epoch": 0.4937776424875105,
      "grad_norm": 1.3145579099655151,
      "learning_rate": 0.00010686700669520792,
      "loss": 1.5071,
      "step": 13788
    },
    {
      "epoch": 0.4938134546170788,
      "grad_norm": 1.4483972787857056,
      "learning_rate": 0.00010685543501644732,
      "loss": 1.36,
      "step": 13789
    },
    {
      "epoch": 0.4938492667466471,
      "grad_norm": 5.216784477233887,
      "learning_rate": 0.00010684386324545567,
      "loss": 1.8394,
      "step": 13790
    },
    {
      "epoch": 0.49388507887621536,
      "grad_norm": 1.2942969799041748,
      "learning_rate": 0.0001068322913823887,
      "loss": 1.1399,
      "step": 13791
    },
    {
      "epoch": 0.4939208910057837,
      "grad_norm": 1.4806745052337646,
      "learning_rate": 0.00010682071942740202,
      "loss": 1.2029,
      "step": 13792
    },
    {
      "epoch": 0.49395670313535195,
      "grad_norm": 1.8519744873046875,
      "learning_rate": 0.0001068091473806514,
      "loss": 1.3285,
      "step": 13793
    },
    {
      "epoch": 0.4939925152649202,
      "grad_norm": 1.7038651704788208,
      "learning_rate": 0.00010679757524229244,
      "loss": 1.3596,
      "step": 13794
    },
    {
      "epoch": 0.4940283273944885,
      "grad_norm": 1.638228178024292,
      "learning_rate": 0.0001067860030124809,
      "loss": 1.328,
      "step": 13795
    },
    {
      "epoch": 0.4940641395240568,
      "grad_norm": 1.7471988201141357,
      "learning_rate": 0.00010677443069137242,
      "loss": 1.2233,
      "step": 13796
    },
    {
      "epoch": 0.4940999516536251,
      "grad_norm": 2.5956523418426514,
      "learning_rate": 0.00010676285827912276,
      "loss": 1.2647,
      "step": 13797
    },
    {
      "epoch": 0.49413576378319335,
      "grad_norm": 1.6266028881072998,
      "learning_rate": 0.00010675128577588751,
      "loss": 1.6781,
      "step": 13798
    },
    {
      "epoch": 0.4941715759127617,
      "grad_norm": 1.3634744882583618,
      "learning_rate": 0.00010673971318182247,
      "loss": 1.6182,
      "step": 13799
    },
    {
      "epoch": 0.49420738804232994,
      "grad_norm": 1.7638877630233765,
      "learning_rate": 0.00010672814049708326,
      "loss": 1.5701,
      "step": 13800
    },
    {
      "epoch": 0.4942432001718982,
      "grad_norm": 1.6434555053710938,
      "learning_rate": 0.0001067165677218256,
      "loss": 1.2647,
      "step": 13801
    },
    {
      "epoch": 0.4942790123014665,
      "grad_norm": 1.9135644435882568,
      "learning_rate": 0.00010670499485620517,
      "loss": 1.5567,
      "step": 13802
    },
    {
      "epoch": 0.4943148244310348,
      "grad_norm": 1.8451272249221802,
      "learning_rate": 0.0001066934219003777,
      "loss": 1.2508,
      "step": 13803
    },
    {
      "epoch": 0.4943506365606031,
      "grad_norm": 1.3484454154968262,
      "learning_rate": 0.00010668184885449886,
      "loss": 1.4343,
      "step": 13804
    },
    {
      "epoch": 0.49438644869017134,
      "grad_norm": 1.644727110862732,
      "learning_rate": 0.00010667027571872436,
      "loss": 1.2379,
      "step": 13805
    },
    {
      "epoch": 0.49442226081973967,
      "grad_norm": 1.5474905967712402,
      "learning_rate": 0.00010665870249320993,
      "loss": 1.5323,
      "step": 13806
    },
    {
      "epoch": 0.49445807294930794,
      "grad_norm": 1.9647862911224365,
      "learning_rate": 0.00010664712917811121,
      "loss": 1.7564,
      "step": 13807
    },
    {
      "epoch": 0.4944938850788762,
      "grad_norm": 1.648308515548706,
      "learning_rate": 0.000106635555773584,
      "loss": 1.36,
      "step": 13808
    },
    {
      "epoch": 0.4945296972084445,
      "grad_norm": 1.834714651107788,
      "learning_rate": 0.00010662398227978389,
      "loss": 1.223,
      "step": 13809
    },
    {
      "epoch": 0.4945655093380128,
      "grad_norm": 1.4639192819595337,
      "learning_rate": 0.00010661240869686669,
      "loss": 1.4133,
      "step": 13810
    },
    {
      "epoch": 0.49460132146758107,
      "grad_norm": 1.9548790454864502,
      "learning_rate": 0.00010660083502498801,
      "loss": 1.2963,
      "step": 13811
    },
    {
      "epoch": 0.49463713359714934,
      "grad_norm": 1.7296000719070435,
      "learning_rate": 0.00010658926126430364,
      "loss": 1.6591,
      "step": 13812
    },
    {
      "epoch": 0.49467294572671766,
      "grad_norm": 1.7165069580078125,
      "learning_rate": 0.00010657768741496923,
      "loss": 1.3809,
      "step": 13813
    },
    {
      "epoch": 0.49470875785628593,
      "grad_norm": 1.616808295249939,
      "learning_rate": 0.00010656611347714056,
      "loss": 1.647,
      "step": 13814
    },
    {
      "epoch": 0.4947445699858542,
      "grad_norm": 2.2919797897338867,
      "learning_rate": 0.00010655453945097327,
      "loss": 1.8562,
      "step": 13815
    },
    {
      "epoch": 0.49478038211542247,
      "grad_norm": 2.0228257179260254,
      "learning_rate": 0.0001065429653366231,
      "loss": 1.5348,
      "step": 13816
    },
    {
      "epoch": 0.4948161942449908,
      "grad_norm": 2.164900064468384,
      "learning_rate": 0.00010653139113424581,
      "loss": 1.5874,
      "step": 13817
    },
    {
      "epoch": 0.49485200637455906,
      "grad_norm": 1.9388035535812378,
      "learning_rate": 0.00010651981684399705,
      "loss": 1.3361,
      "step": 13818
    },
    {
      "epoch": 0.49488781850412733,
      "grad_norm": 2.0324113368988037,
      "learning_rate": 0.0001065082424660326,
      "loss": 1.2808,
      "step": 13819
    },
    {
      "epoch": 0.49492363063369565,
      "grad_norm": 1.7220652103424072,
      "learning_rate": 0.00010649666800050808,
      "loss": 1.3884,
      "step": 13820
    },
    {
      "epoch": 0.4949594427632639,
      "grad_norm": 2.1359825134277344,
      "learning_rate": 0.00010648509344757933,
      "loss": 1.4864,
      "step": 13821
    },
    {
      "epoch": 0.4949952548928322,
      "grad_norm": 1.7690446376800537,
      "learning_rate": 0.00010647351880740197,
      "loss": 1.5202,
      "step": 13822
    },
    {
      "epoch": 0.49503106702240046,
      "grad_norm": 1.6136664152145386,
      "learning_rate": 0.00010646194408013179,
      "loss": 1.4969,
      "step": 13823
    },
    {
      "epoch": 0.4950668791519688,
      "grad_norm": 1.6835991144180298,
      "learning_rate": 0.00010645036926592449,
      "loss": 1.3849,
      "step": 13824
    },
    {
      "epoch": 0.49510269128153706,
      "grad_norm": 2.4620182514190674,
      "learning_rate": 0.00010643879436493578,
      "loss": 1.602,
      "step": 13825
    },
    {
      "epoch": 0.4951385034111053,
      "grad_norm": 2.525754928588867,
      "learning_rate": 0.0001064272193773214,
      "loss": 1.4661,
      "step": 13826
    },
    {
      "epoch": 0.49517431554067365,
      "grad_norm": 2.1498541831970215,
      "learning_rate": 0.00010641564430323707,
      "loss": 1.4336,
      "step": 13827
    },
    {
      "epoch": 0.4952101276702419,
      "grad_norm": 2.0926995277404785,
      "learning_rate": 0.00010640406914283854,
      "loss": 1.2882,
      "step": 13828
    },
    {
      "epoch": 0.4952459397998102,
      "grad_norm": 2.03147554397583,
      "learning_rate": 0.00010639249389628149,
      "loss": 1.7856,
      "step": 13829
    },
    {
      "epoch": 0.49528175192937846,
      "grad_norm": 1.6365413665771484,
      "learning_rate": 0.00010638091856372172,
      "loss": 1.3358,
      "step": 13830
    },
    {
      "epoch": 0.4953175640589468,
      "grad_norm": 1.5908212661743164,
      "learning_rate": 0.00010636934314531488,
      "loss": 1.7253,
      "step": 13831
    },
    {
      "epoch": 0.49535337618851505,
      "grad_norm": 1.616167664527893,
      "learning_rate": 0.00010635776764121677,
      "loss": 1.6732,
      "step": 13832
    },
    {
      "epoch": 0.4953891883180833,
      "grad_norm": 1.6805557012557983,
      "learning_rate": 0.00010634619205158307,
      "loss": 1.7188,
      "step": 13833
    },
    {
      "epoch": 0.49542500044765164,
      "grad_norm": 1.613982081413269,
      "learning_rate": 0.00010633461637656958,
      "loss": 1.4378,
      "step": 13834
    },
    {
      "epoch": 0.4954608125772199,
      "grad_norm": 1.827687382698059,
      "learning_rate": 0.00010632304061633199,
      "loss": 1.4883,
      "step": 13835
    },
    {
      "epoch": 0.4954966247067882,
      "grad_norm": 1.3360340595245361,
      "learning_rate": 0.00010631146477102602,
      "loss": 1.2353,
      "step": 13836
    },
    {
      "epoch": 0.49553243683635645,
      "grad_norm": 1.3688260316848755,
      "learning_rate": 0.00010629988884080745,
      "loss": 1.4722,
      "step": 13837
    },
    {
      "epoch": 0.4955682489659248,
      "grad_norm": 2.0016164779663086,
      "learning_rate": 0.00010628831282583201,
      "loss": 1.2342,
      "step": 13838
    },
    {
      "epoch": 0.49560406109549304,
      "grad_norm": 2.039696216583252,
      "learning_rate": 0.00010627673672625542,
      "loss": 1.5256,
      "step": 13839
    },
    {
      "epoch": 0.4956398732250613,
      "grad_norm": 1.414087176322937,
      "learning_rate": 0.00010626516054223341,
      "loss": 1.5765,
      "step": 13840
    },
    {
      "epoch": 0.49567568535462964,
      "grad_norm": 1.4707834720611572,
      "learning_rate": 0.0001062535842739218,
      "loss": 1.3901,
      "step": 13841
    },
    {
      "epoch": 0.4957114974841979,
      "grad_norm": 1.269789695739746,
      "learning_rate": 0.00010624200792147622,
      "loss": 1.707,
      "step": 13842
    },
    {
      "epoch": 0.4957473096137662,
      "grad_norm": 1.7977982759475708,
      "learning_rate": 0.00010623043148505254,
      "loss": 1.3614,
      "step": 13843
    },
    {
      "epoch": 0.49578312174333444,
      "grad_norm": 2.498763084411621,
      "learning_rate": 0.00010621885496480641,
      "loss": 1.4808,
      "step": 13844
    },
    {
      "epoch": 0.49581893387290277,
      "grad_norm": 1.9062105417251587,
      "learning_rate": 0.00010620727836089359,
      "loss": 1.6724,
      "step": 13845
    },
    {
      "epoch": 0.49585474600247104,
      "grad_norm": 1.6607482433319092,
      "learning_rate": 0.00010619570167346987,
      "loss": 1.3214,
      "step": 13846
    },
    {
      "epoch": 0.4958905581320393,
      "grad_norm": 1.4918594360351562,
      "learning_rate": 0.00010618412490269096,
      "loss": 1.4561,
      "step": 13847
    },
    {
      "epoch": 0.49592637026160763,
      "grad_norm": 1.8576520681381226,
      "learning_rate": 0.00010617254804871264,
      "loss": 1.4214,
      "step": 13848
    },
    {
      "epoch": 0.4959621823911759,
      "grad_norm": 1.425275444984436,
      "learning_rate": 0.00010616097111169063,
      "loss": 1.4187,
      "step": 13849
    },
    {
      "epoch": 0.49599799452074417,
      "grad_norm": 1.8478344678878784,
      "learning_rate": 0.00010614939409178072,
      "loss": 1.4093,
      "step": 13850
    },
    {
      "epoch": 0.49603380665031244,
      "grad_norm": 2.1850032806396484,
      "learning_rate": 0.00010613781698913863,
      "loss": 1.2745,
      "step": 13851
    },
    {
      "epoch": 0.49606961877988076,
      "grad_norm": 1.5047380924224854,
      "learning_rate": 0.00010612623980392016,
      "loss": 1.1031,
      "step": 13852
    },
    {
      "epoch": 0.49610543090944903,
      "grad_norm": 1.6822580099105835,
      "learning_rate": 0.00010611466253628101,
      "loss": 1.4585,
      "step": 13853
    },
    {
      "epoch": 0.4961412430390173,
      "grad_norm": 1.5494749546051025,
      "learning_rate": 0.00010610308518637697,
      "loss": 1.3333,
      "step": 13854
    },
    {
      "epoch": 0.4961770551685856,
      "grad_norm": 1.5541067123413086,
      "learning_rate": 0.00010609150775436378,
      "loss": 1.5149,
      "step": 13855
    },
    {
      "epoch": 0.4962128672981539,
      "grad_norm": 1.8944567441940308,
      "learning_rate": 0.00010607993024039722,
      "loss": 1.2425,
      "step": 13856
    },
    {
      "epoch": 0.49624867942772216,
      "grad_norm": 1.1840757131576538,
      "learning_rate": 0.00010606835264463305,
      "loss": 1.4237,
      "step": 13857
    },
    {
      "epoch": 0.49628449155729043,
      "grad_norm": 2.1415929794311523,
      "learning_rate": 0.00010605677496722699,
      "loss": 1.7492,
      "step": 13858
    },
    {
      "epoch": 0.49632030368685875,
      "grad_norm": 1.6626441478729248,
      "learning_rate": 0.00010604519720833486,
      "loss": 1.6241,
      "step": 13859
    },
    {
      "epoch": 0.496356115816427,
      "grad_norm": 1.551418423652649,
      "learning_rate": 0.00010603361936811239,
      "loss": 1.5366,
      "step": 13860
    },
    {
      "epoch": 0.4963919279459953,
      "grad_norm": 2.124976396560669,
      "learning_rate": 0.00010602204144671539,
      "loss": 1.6503,
      "step": 13861
    },
    {
      "epoch": 0.4964277400755636,
      "grad_norm": 2.3942666053771973,
      "learning_rate": 0.00010601046344429955,
      "loss": 1.5967,
      "step": 13862
    },
    {
      "epoch": 0.4964635522051319,
      "grad_norm": 2.203838586807251,
      "learning_rate": 0.0001059988853610207,
      "loss": 1.3865,
      "step": 13863
    },
    {
      "epoch": 0.49649936433470016,
      "grad_norm": 1.3743383884429932,
      "learning_rate": 0.00010598730719703456,
      "loss": 1.78,
      "step": 13864
    },
    {
      "epoch": 0.4965351764642684,
      "grad_norm": 1.5764050483703613,
      "learning_rate": 0.00010597572895249694,
      "loss": 1.1433,
      "step": 13865
    },
    {
      "epoch": 0.49657098859383675,
      "grad_norm": 2.720665454864502,
      "learning_rate": 0.00010596415062756358,
      "loss": 1.4423,
      "step": 13866
    },
    {
      "epoch": 0.496606800723405,
      "grad_norm": 1.9178234338760376,
      "learning_rate": 0.0001059525722223903,
      "loss": 1.5331,
      "step": 13867
    },
    {
      "epoch": 0.4966426128529733,
      "grad_norm": 1.4973068237304688,
      "learning_rate": 0.0001059409937371328,
      "loss": 1.5225,
      "step": 13868
    },
    {
      "epoch": 0.4966784249825416,
      "grad_norm": 1.7021870613098145,
      "learning_rate": 0.00010592941517194692,
      "loss": 1.3905,
      "step": 13869
    },
    {
      "epoch": 0.4967142371121099,
      "grad_norm": 1.600563645362854,
      "learning_rate": 0.00010591783652698841,
      "loss": 1.3413,
      "step": 13870
    },
    {
      "epoch": 0.49675004924167815,
      "grad_norm": 1.8694027662277222,
      "learning_rate": 0.00010590625780241302,
      "loss": 1.4979,
      "step": 13871
    },
    {
      "epoch": 0.4967858613712464,
      "grad_norm": 1.3049858808517456,
      "learning_rate": 0.00010589467899837657,
      "loss": 1.2021,
      "step": 13872
    },
    {
      "epoch": 0.49682167350081474,
      "grad_norm": 1.7500213384628296,
      "learning_rate": 0.0001058831001150348,
      "loss": 1.4798,
      "step": 13873
    },
    {
      "epoch": 0.496857485630383,
      "grad_norm": 1.521600365638733,
      "learning_rate": 0.00010587152115254353,
      "loss": 1.5667,
      "step": 13874
    },
    {
      "epoch": 0.4968932977599513,
      "grad_norm": 1.659149408340454,
      "learning_rate": 0.0001058599421110585,
      "loss": 1.3085,
      "step": 13875
    },
    {
      "epoch": 0.4969291098895196,
      "grad_norm": 1.9441237449645996,
      "learning_rate": 0.0001058483629907355,
      "loss": 1.5502,
      "step": 13876
    },
    {
      "epoch": 0.4969649220190879,
      "grad_norm": 1.8426682949066162,
      "learning_rate": 0.00010583678379173032,
      "loss": 1.7238,
      "step": 13877
    },
    {
      "epoch": 0.49700073414865614,
      "grad_norm": 2.1774513721466064,
      "learning_rate": 0.00010582520451419877,
      "loss": 1.4809,
      "step": 13878
    },
    {
      "epoch": 0.4970365462782244,
      "grad_norm": 1.3162360191345215,
      "learning_rate": 0.0001058136251582966,
      "loss": 1.573,
      "step": 13879
    },
    {
      "epoch": 0.49707235840779274,
      "grad_norm": 1.2653931379318237,
      "learning_rate": 0.00010580204572417957,
      "loss": 1.6874,
      "step": 13880
    },
    {
      "epoch": 0.497108170537361,
      "grad_norm": 2.2073793411254883,
      "learning_rate": 0.00010579046621200355,
      "loss": 1.8626,
      "step": 13881
    },
    {
      "epoch": 0.4971439826669293,
      "grad_norm": 1.9331858158111572,
      "learning_rate": 0.00010577888662192424,
      "loss": 1.6169,
      "step": 13882
    },
    {
      "epoch": 0.4971797947964976,
      "grad_norm": 1.5831600427627563,
      "learning_rate": 0.00010576730695409747,
      "loss": 1.5547,
      "step": 13883
    },
    {
      "epoch": 0.49721560692606587,
      "grad_norm": 1.8316388130187988,
      "learning_rate": 0.00010575572720867901,
      "loss": 1.7315,
      "step": 13884
    },
    {
      "epoch": 0.49725141905563414,
      "grad_norm": 1.7903798818588257,
      "learning_rate": 0.0001057441473858247,
      "loss": 1.4935,
      "step": 13885
    },
    {
      "epoch": 0.4972872311852024,
      "grad_norm": 1.3750848770141602,
      "learning_rate": 0.00010573256748569027,
      "loss": 1.3742,
      "step": 13886
    },
    {
      "epoch": 0.49732304331477073,
      "grad_norm": 2.335555076599121,
      "learning_rate": 0.00010572098750843155,
      "loss": 1.5615,
      "step": 13887
    },
    {
      "epoch": 0.497358855444339,
      "grad_norm": 1.5992259979248047,
      "learning_rate": 0.00010570940745420433,
      "loss": 1.6146,
      "step": 13888
    },
    {
      "epoch": 0.49739466757390727,
      "grad_norm": 1.6702439785003662,
      "learning_rate": 0.00010569782732316438,
      "loss": 1.443,
      "step": 13889
    },
    {
      "epoch": 0.4974304797034756,
      "grad_norm": 1.9406352043151855,
      "learning_rate": 0.00010568624711546752,
      "loss": 1.4274,
      "step": 13890
    },
    {
      "epoch": 0.49746629183304386,
      "grad_norm": 1.4451533555984497,
      "learning_rate": 0.00010567466683126952,
      "loss": 1.5557,
      "step": 13891
    },
    {
      "epoch": 0.49750210396261213,
      "grad_norm": 2.7517757415771484,
      "learning_rate": 0.00010566308647072624,
      "loss": 1.5878,
      "step": 13892
    },
    {
      "epoch": 0.4975379160921804,
      "grad_norm": 1.9673051834106445,
      "learning_rate": 0.0001056515060339934,
      "loss": 1.7578,
      "step": 13893
    },
    {
      "epoch": 0.4975737282217487,
      "grad_norm": 1.653322458267212,
      "learning_rate": 0.00010563992552122686,
      "loss": 1.2221,
      "step": 13894
    },
    {
      "epoch": 0.497609540351317,
      "grad_norm": 1.8169872760772705,
      "learning_rate": 0.00010562834493258237,
      "loss": 1.4192,
      "step": 13895
    },
    {
      "epoch": 0.49764535248088526,
      "grad_norm": 1.7515963315963745,
      "learning_rate": 0.00010561676426821581,
      "loss": 1.4524,
      "step": 13896
    },
    {
      "epoch": 0.4976811646104536,
      "grad_norm": 1.6058905124664307,
      "learning_rate": 0.00010560518352828288,
      "loss": 1.5834,
      "step": 13897
    },
    {
      "epoch": 0.49771697674002185,
      "grad_norm": 2.050147294998169,
      "learning_rate": 0.00010559360271293947,
      "loss": 1.3106,
      "step": 13898
    },
    {
      "epoch": 0.4977527888695901,
      "grad_norm": 1.8379297256469727,
      "learning_rate": 0.00010558202182234132,
      "loss": 1.6386,
      "step": 13899
    },
    {
      "epoch": 0.4977886009991584,
      "grad_norm": 1.9180848598480225,
      "learning_rate": 0.00010557044085664428,
      "loss": 1.4842,
      "step": 13900
    },
    {
      "epoch": 0.4978244131287267,
      "grad_norm": 1.9606422185897827,
      "learning_rate": 0.00010555885981600416,
      "loss": 1.225,
      "step": 13901
    },
    {
      "epoch": 0.497860225258295,
      "grad_norm": 1.8793365955352783,
      "learning_rate": 0.00010554727870057671,
      "loss": 1.6759,
      "step": 13902
    },
    {
      "epoch": 0.49789603738786326,
      "grad_norm": 2.0016098022460938,
      "learning_rate": 0.00010553569751051782,
      "loss": 1.4724,
      "step": 13903
    },
    {
      "epoch": 0.4979318495174316,
      "grad_norm": 2.300502061843872,
      "learning_rate": 0.00010552411624598325,
      "loss": 1.3582,
      "step": 13904
    },
    {
      "epoch": 0.49796766164699985,
      "grad_norm": 2.5776402950286865,
      "learning_rate": 0.00010551253490712882,
      "loss": 1.529,
      "step": 13905
    },
    {
      "epoch": 0.4980034737765681,
      "grad_norm": 1.8135322332382202,
      "learning_rate": 0.00010550095349411033,
      "loss": 1.3956,
      "step": 13906
    },
    {
      "epoch": 0.4980392859061364,
      "grad_norm": 1.6198707818984985,
      "learning_rate": 0.00010548937200708365,
      "loss": 1.348,
      "step": 13907
    },
    {
      "epoch": 0.4980750980357047,
      "grad_norm": 1.9541133642196655,
      "learning_rate": 0.0001054777904462045,
      "loss": 1.4611,
      "step": 13908
    },
    {
      "epoch": 0.498110910165273,
      "grad_norm": 1.7286028861999512,
      "learning_rate": 0.00010546620881162876,
      "loss": 1.1831,
      "step": 13909
    },
    {
      "epoch": 0.49814672229484125,
      "grad_norm": 1.5478944778442383,
      "learning_rate": 0.00010545462710351224,
      "loss": 1.5389,
      "step": 13910
    },
    {
      "epoch": 0.4981825344244096,
      "grad_norm": 1.9012272357940674,
      "learning_rate": 0.00010544304532201075,
      "loss": 1.2058,
      "step": 13911
    },
    {
      "epoch": 0.49821834655397784,
      "grad_norm": 1.586827039718628,
      "learning_rate": 0.0001054314634672801,
      "loss": 1.8115,
      "step": 13912
    },
    {
      "epoch": 0.4982541586835461,
      "grad_norm": 1.4416097402572632,
      "learning_rate": 0.00010541988153947609,
      "loss": 1.4744,
      "step": 13913
    },
    {
      "epoch": 0.4982899708131144,
      "grad_norm": 1.427348256111145,
      "learning_rate": 0.00010540829953875462,
      "loss": 1.3261,
      "step": 13914
    },
    {
      "epoch": 0.4983257829426827,
      "grad_norm": 1.6142998933792114,
      "learning_rate": 0.00010539671746527142,
      "loss": 1.4967,
      "step": 13915
    },
    {
      "epoch": 0.498361595072251,
      "grad_norm": 1.6555075645446777,
      "learning_rate": 0.00010538513531918237,
      "loss": 1.2144,
      "step": 13916
    },
    {
      "epoch": 0.49839740720181924,
      "grad_norm": 1.8272299766540527,
      "learning_rate": 0.00010537355310064323,
      "loss": 1.3989,
      "step": 13917
    },
    {
      "epoch": 0.49843321933138757,
      "grad_norm": 1.2751270532608032,
      "learning_rate": 0.00010536197080980991,
      "loss": 1.6073,
      "step": 13918
    },
    {
      "epoch": 0.49846903146095584,
      "grad_norm": 1.965511679649353,
      "learning_rate": 0.00010535038844683816,
      "loss": 1.6124,
      "step": 13919
    },
    {
      "epoch": 0.4985048435905241,
      "grad_norm": 1.3621838092803955,
      "learning_rate": 0.00010533880601188384,
      "loss": 1.436,
      "step": 13920
    },
    {
      "epoch": 0.4985406557200924,
      "grad_norm": 2.440661668777466,
      "learning_rate": 0.00010532722350510277,
      "loss": 1.3973,
      "step": 13921
    },
    {
      "epoch": 0.4985764678496607,
      "grad_norm": 1.5295976400375366,
      "learning_rate": 0.00010531564092665079,
      "loss": 1.5994,
      "step": 13922
    },
    {
      "epoch": 0.49861227997922897,
      "grad_norm": 2.2189676761627197,
      "learning_rate": 0.00010530405827668372,
      "loss": 1.3265,
      "step": 13923
    },
    {
      "epoch": 0.49864809210879724,
      "grad_norm": 1.4622187614440918,
      "learning_rate": 0.00010529247555535738,
      "loss": 1.7136,
      "step": 13924
    },
    {
      "epoch": 0.49868390423836556,
      "grad_norm": 1.473183274269104,
      "learning_rate": 0.00010528089276282762,
      "loss": 1.6749,
      "step": 13925
    },
    {
      "epoch": 0.49871971636793383,
      "grad_norm": 1.5246021747589111,
      "learning_rate": 0.00010526930989925023,
      "loss": 1.5913,
      "step": 13926
    },
    {
      "epoch": 0.4987555284975021,
      "grad_norm": 1.6022064685821533,
      "learning_rate": 0.0001052577269647811,
      "loss": 1.4599,
      "step": 13927
    },
    {
      "epoch": 0.49879134062707037,
      "grad_norm": 1.3866850137710571,
      "learning_rate": 0.00010524614395957602,
      "loss": 1.3702,
      "step": 13928
    },
    {
      "epoch": 0.4988271527566387,
      "grad_norm": 2.0144548416137695,
      "learning_rate": 0.00010523456088379084,
      "loss": 1.582,
      "step": 13929
    },
    {
      "epoch": 0.49886296488620696,
      "grad_norm": 2.1504111289978027,
      "learning_rate": 0.00010522297773758141,
      "loss": 1.6746,
      "step": 13930
    },
    {
      "epoch": 0.49889877701577523,
      "grad_norm": 1.4238390922546387,
      "learning_rate": 0.00010521139452110354,
      "loss": 1.6415,
      "step": 13931
    },
    {
      "epoch": 0.49893458914534355,
      "grad_norm": 2.0092551708221436,
      "learning_rate": 0.0001051998112345131,
      "loss": 1.1434,
      "step": 13932
    },
    {
      "epoch": 0.4989704012749118,
      "grad_norm": 1.7443785667419434,
      "learning_rate": 0.00010518822787796587,
      "loss": 1.6474,
      "step": 13933
    },
    {
      "epoch": 0.4990062134044801,
      "grad_norm": 1.9198230504989624,
      "learning_rate": 0.00010517664445161775,
      "loss": 1.291,
      "step": 13934
    },
    {
      "epoch": 0.49904202553404836,
      "grad_norm": 2.1047768592834473,
      "learning_rate": 0.00010516506095562455,
      "loss": 1.4204,
      "step": 13935
    },
    {
      "epoch": 0.4990778376636167,
      "grad_norm": 1.772157907485962,
      "learning_rate": 0.00010515347739014212,
      "loss": 1.4815,
      "step": 13936
    },
    {
      "epoch": 0.49911364979318495,
      "grad_norm": 1.4584022760391235,
      "learning_rate": 0.00010514189375532629,
      "loss": 1.1986,
      "step": 13937
    },
    {
      "epoch": 0.4991494619227532,
      "grad_norm": 1.6682506799697876,
      "learning_rate": 0.00010513031005133293,
      "loss": 1.1585,
      "step": 13938
    },
    {
      "epoch": 0.49918527405232155,
      "grad_norm": 2.0032548904418945,
      "learning_rate": 0.00010511872627831785,
      "loss": 1.8689,
      "step": 13939
    },
    {
      "epoch": 0.4992210861818898,
      "grad_norm": 1.799172043800354,
      "learning_rate": 0.00010510714243643693,
      "loss": 1.2141,
      "step": 13940
    },
    {
      "epoch": 0.4992568983114581,
      "grad_norm": 1.4442138671875,
      "learning_rate": 0.00010509555852584598,
      "loss": 1.3636,
      "step": 13941
    },
    {
      "epoch": 0.49929271044102636,
      "grad_norm": 1.4175442457199097,
      "learning_rate": 0.00010508397454670085,
      "loss": 1.4478,
      "step": 13942
    },
    {
      "epoch": 0.4993285225705947,
      "grad_norm": 2.0027573108673096,
      "learning_rate": 0.00010507239049915742,
      "loss": 1.8723,
      "step": 13943
    },
    {
      "epoch": 0.49936433470016295,
      "grad_norm": 1.363532543182373,
      "learning_rate": 0.00010506080638337152,
      "loss": 1.5395,
      "step": 13944
    },
    {
      "epoch": 0.4994001468297312,
      "grad_norm": 2.1520915031433105,
      "learning_rate": 0.000105049222199499,
      "loss": 1.5888,
      "step": 13945
    },
    {
      "epoch": 0.49943595895929954,
      "grad_norm": 1.5983831882476807,
      "learning_rate": 0.0001050376379476957,
      "loss": 1.2995,
      "step": 13946
    },
    {
      "epoch": 0.4994717710888678,
      "grad_norm": 1.9708709716796875,
      "learning_rate": 0.00010502605362811748,
      "loss": 1.3502,
      "step": 13947
    },
    {
      "epoch": 0.4995075832184361,
      "grad_norm": 1.6863350868225098,
      "learning_rate": 0.00010501446924092018,
      "loss": 1.3486,
      "step": 13948
    },
    {
      "epoch": 0.49954339534800435,
      "grad_norm": 1.7589701414108276,
      "learning_rate": 0.0001050028847862597,
      "loss": 1.4596,
      "step": 13949
    },
    {
      "epoch": 0.4995792074775727,
      "grad_norm": 1.4610109329223633,
      "learning_rate": 0.00010499130026429182,
      "loss": 1.5116,
      "step": 13950
    },
    {
      "epoch": 0.49961501960714094,
      "grad_norm": 1.776166558265686,
      "learning_rate": 0.00010497971567517246,
      "loss": 1.3748,
      "step": 13951
    },
    {
      "epoch": 0.4996508317367092,
      "grad_norm": 2.0865702629089355,
      "learning_rate": 0.00010496813101905745,
      "loss": 1.4648,
      "step": 13952
    },
    {
      "epoch": 0.49968664386627754,
      "grad_norm": 1.7952061891555786,
      "learning_rate": 0.00010495654629610264,
      "loss": 1.233,
      "step": 13953
    },
    {
      "epoch": 0.4997224559958458,
      "grad_norm": 2.238560914993286,
      "learning_rate": 0.00010494496150646387,
      "loss": 1.596,
      "step": 13954
    },
    {
      "epoch": 0.4997582681254141,
      "grad_norm": 1.680570125579834,
      "learning_rate": 0.000104933376650297,
      "loss": 1.4756,
      "step": 13955
    },
    {
      "epoch": 0.49979408025498234,
      "grad_norm": 2.210883140563965,
      "learning_rate": 0.00010492179172775797,
      "loss": 1.3497,
      "step": 13956
    },
    {
      "epoch": 0.49982989238455067,
      "grad_norm": 1.6548429727554321,
      "learning_rate": 0.00010491020673900256,
      "loss": 1.4613,
      "step": 13957
    },
    {
      "epoch": 0.49986570451411894,
      "grad_norm": 1.5173407793045044,
      "learning_rate": 0.00010489862168418667,
      "loss": 1.237,
      "step": 13958
    },
    {
      "epoch": 0.4999015166436872,
      "grad_norm": 1.6748205423355103,
      "learning_rate": 0.00010488703656346612,
      "loss": 1.6796,
      "step": 13959
    },
    {
      "epoch": 0.49993732877325553,
      "grad_norm": 2.2903249263763428,
      "learning_rate": 0.00010487545137699682,
      "loss": 1.3784,
      "step": 13960
    },
    {
      "epoch": 0.4999731409028238,
      "grad_norm": 1.832918643951416,
      "learning_rate": 0.00010486386612493458,
      "loss": 1.4155,
      "step": 13961
    },
    {
      "epoch": 0.5000089530323921,
      "grad_norm": 1.8992984294891357,
      "learning_rate": 0.00010485228080743532,
      "loss": 1.3437,
      "step": 13962
    },
    {
      "epoch": 0.5000447651619604,
      "grad_norm": 1.5817441940307617,
      "learning_rate": 0.00010484069542465484,
      "loss": 1.2825,
      "step": 13963
    },
    {
      "epoch": 0.5000805772915287,
      "grad_norm": 1.7901121377944946,
      "learning_rate": 0.00010482910997674911,
      "loss": 1.4525,
      "step": 13964
    },
    {
      "epoch": 0.5001163894210969,
      "grad_norm": 1.9378963708877563,
      "learning_rate": 0.00010481752446387387,
      "loss": 1.5827,
      "step": 13965
    },
    {
      "epoch": 0.5001522015506652,
      "grad_norm": 1.3163588047027588,
      "learning_rate": 0.0001048059388861851,
      "loss": 1.3767,
      "step": 13966
    },
    {
      "epoch": 0.5001880136802335,
      "grad_norm": 2.0047686100006104,
      "learning_rate": 0.00010479435324383861,
      "loss": 1.4476,
      "step": 13967
    },
    {
      "epoch": 0.5002238258098017,
      "grad_norm": 1.7671470642089844,
      "learning_rate": 0.00010478276753699028,
      "loss": 1.523,
      "step": 13968
    },
    {
      "epoch": 0.5002596379393701,
      "grad_norm": 1.4821014404296875,
      "learning_rate": 0.00010477118176579597,
      "loss": 1.4437,
      "step": 13969
    },
    {
      "epoch": 0.5002954500689384,
      "grad_norm": 2.142829656600952,
      "learning_rate": 0.00010475959593041156,
      "loss": 1.5739,
      "step": 13970
    },
    {
      "epoch": 0.5003312621985067,
      "grad_norm": 2.060955047607422,
      "learning_rate": 0.00010474801003099294,
      "loss": 1.5802,
      "step": 13971
    },
    {
      "epoch": 0.5003670743280749,
      "grad_norm": 1.8736474514007568,
      "learning_rate": 0.00010473642406769597,
      "loss": 1.2844,
      "step": 13972
    },
    {
      "epoch": 0.5004028864576432,
      "grad_norm": 1.4875705242156982,
      "learning_rate": 0.00010472483804067652,
      "loss": 1.0674,
      "step": 13973
    },
    {
      "epoch": 0.5004386985872115,
      "grad_norm": 1.404869556427002,
      "learning_rate": 0.00010471325195009047,
      "loss": 1.6174,
      "step": 13974
    },
    {
      "epoch": 0.5004745107167797,
      "grad_norm": 1.461624264717102,
      "learning_rate": 0.00010470166579609371,
      "loss": 1.5879,
      "step": 13975
    },
    {
      "epoch": 0.5005103228463481,
      "grad_norm": 1.3409490585327148,
      "learning_rate": 0.0001046900795788421,
      "loss": 1.4595,
      "step": 13976
    },
    {
      "epoch": 0.5005461349759164,
      "grad_norm": 1.7915490865707397,
      "learning_rate": 0.00010467849329849148,
      "loss": 1.5817,
      "step": 13977
    },
    {
      "epoch": 0.5005819471054846,
      "grad_norm": 1.4532082080841064,
      "learning_rate": 0.00010466690695519781,
      "loss": 1.4423,
      "step": 13978
    },
    {
      "epoch": 0.5006177592350529,
      "grad_norm": 1.5998505353927612,
      "learning_rate": 0.00010465532054911689,
      "loss": 1.564,
      "step": 13979
    },
    {
      "epoch": 0.5006535713646212,
      "grad_norm": 1.4868236780166626,
      "learning_rate": 0.00010464373408040467,
      "loss": 1.6871,
      "step": 13980
    },
    {
      "epoch": 0.5006893834941895,
      "grad_norm": 1.7556103467941284,
      "learning_rate": 0.00010463214754921697,
      "loss": 1.4825,
      "step": 13981
    },
    {
      "epoch": 0.5007251956237577,
      "grad_norm": 2.2433080673217773,
      "learning_rate": 0.00010462056095570974,
      "loss": 1.5892,
      "step": 13982
    },
    {
      "epoch": 0.5007610077533261,
      "grad_norm": 1.4293811321258545,
      "learning_rate": 0.00010460897430003877,
      "loss": 1.4694,
      "step": 13983
    },
    {
      "epoch": 0.5007968198828944,
      "grad_norm": 2.3637685775756836,
      "learning_rate": 0.00010459738758236006,
      "loss": 2.1556,
      "step": 13984
    },
    {
      "epoch": 0.5008326320124626,
      "grad_norm": 1.4137765169143677,
      "learning_rate": 0.00010458580080282938,
      "loss": 1.5547,
      "step": 13985
    },
    {
      "epoch": 0.5008684441420309,
      "grad_norm": 1.7568750381469727,
      "learning_rate": 0.00010457421396160265,
      "loss": 1.4713,
      "step": 13986
    },
    {
      "epoch": 0.5009042562715992,
      "grad_norm": 1.891427755355835,
      "learning_rate": 0.00010456262705883581,
      "loss": 1.0563,
      "step": 13987
    },
    {
      "epoch": 0.5009400684011674,
      "grad_norm": 1.6349992752075195,
      "learning_rate": 0.0001045510400946847,
      "loss": 1.4901,
      "step": 13988
    },
    {
      "epoch": 0.5009758805307357,
      "grad_norm": 1.4608160257339478,
      "learning_rate": 0.00010453945306930521,
      "loss": 1.507,
      "step": 13989
    },
    {
      "epoch": 0.5010116926603041,
      "grad_norm": 1.6028715372085571,
      "learning_rate": 0.00010452786598285323,
      "loss": 1.6952,
      "step": 13990
    },
    {
      "epoch": 0.5010475047898724,
      "grad_norm": 1.4232906103134155,
      "learning_rate": 0.00010451627883548468,
      "loss": 1.4949,
      "step": 13991
    },
    {
      "epoch": 0.5010833169194406,
      "grad_norm": 1.2715191841125488,
      "learning_rate": 0.00010450469162735539,
      "loss": 1.3847,
      "step": 13992
    },
    {
      "epoch": 0.5011191290490089,
      "grad_norm": 1.617362380027771,
      "learning_rate": 0.00010449310435862134,
      "loss": 1.508,
      "step": 13993
    },
    {
      "epoch": 0.5011549411785772,
      "grad_norm": 1.5455505847930908,
      "learning_rate": 0.00010448151702943831,
      "loss": 1.557,
      "step": 13994
    },
    {
      "epoch": 0.5011907533081454,
      "grad_norm": 1.6332513093948364,
      "learning_rate": 0.00010446992963996227,
      "loss": 1.5056,
      "step": 13995
    },
    {
      "epoch": 0.5012265654377137,
      "grad_norm": 1.7719979286193848,
      "learning_rate": 0.00010445834219034909,
      "loss": 1.6498,
      "step": 13996
    },
    {
      "epoch": 0.5012623775672821,
      "grad_norm": 1.3904492855072021,
      "learning_rate": 0.00010444675468075467,
      "loss": 1.4026,
      "step": 13997
    },
    {
      "epoch": 0.5012981896968504,
      "grad_norm": 1.3235598802566528,
      "learning_rate": 0.00010443516711133487,
      "loss": 1.4448,
      "step": 13998
    },
    {
      "epoch": 0.5013340018264186,
      "grad_norm": 1.7896219491958618,
      "learning_rate": 0.00010442357948224564,
      "loss": 1.6782,
      "step": 13999
    },
    {
      "epoch": 0.5013698139559869,
      "grad_norm": 1.3523547649383545,
      "learning_rate": 0.00010441199179364287,
      "loss": 1.0761,
      "step": 14000
    },
    {
      "epoch": 0.5014056260855552,
      "grad_norm": 1.9005093574523926,
      "learning_rate": 0.00010440040404568241,
      "loss": 1.5543,
      "step": 14001
    },
    {
      "epoch": 0.5014414382151234,
      "grad_norm": 1.3568506240844727,
      "learning_rate": 0.00010438881623852026,
      "loss": 1.6062,
      "step": 14002
    },
    {
      "epoch": 0.5014772503446917,
      "grad_norm": 1.6488691568374634,
      "learning_rate": 0.00010437722837231218,
      "loss": 1.7009,
      "step": 14003
    },
    {
      "epoch": 0.5015130624742601,
      "grad_norm": 1.5517204999923706,
      "learning_rate": 0.00010436564044721415,
      "loss": 1.7509,
      "step": 14004
    },
    {
      "epoch": 0.5015488746038284,
      "grad_norm": 1.2834587097167969,
      "learning_rate": 0.00010435405246338205,
      "loss": 1.4208,
      "step": 14005
    },
    {
      "epoch": 0.5015846867333966,
      "grad_norm": 1.3293063640594482,
      "learning_rate": 0.00010434246442097184,
      "loss": 1.4019,
      "step": 14006
    },
    {
      "epoch": 0.5016204988629649,
      "grad_norm": 2.0238394737243652,
      "learning_rate": 0.00010433087632013931,
      "loss": 1.3617,
      "step": 14007
    },
    {
      "epoch": 0.5016563109925332,
      "grad_norm": 1.7734540700912476,
      "learning_rate": 0.00010431928816104048,
      "loss": 1.3808,
      "step": 14008
    },
    {
      "epoch": 0.5016921231221014,
      "grad_norm": 1.458070158958435,
      "learning_rate": 0.00010430769994383116,
      "loss": 1.4744,
      "step": 14009
    },
    {
      "epoch": 0.5017279352516697,
      "grad_norm": 1.3446053266525269,
      "learning_rate": 0.0001042961116686673,
      "loss": 1.4234,
      "step": 14010
    },
    {
      "epoch": 0.5017637473812381,
      "grad_norm": 2.673496723175049,
      "learning_rate": 0.00010428452333570482,
      "loss": 1.9296,
      "step": 14011
    },
    {
      "epoch": 0.5017995595108063,
      "grad_norm": 2.1430649757385254,
      "learning_rate": 0.0001042729349450996,
      "loss": 1.4881,
      "step": 14012
    },
    {
      "epoch": 0.5018353716403746,
      "grad_norm": 1.6296019554138184,
      "learning_rate": 0.00010426134649700754,
      "loss": 1.1715,
      "step": 14013
    },
    {
      "epoch": 0.5018711837699429,
      "grad_norm": 1.853363275527954,
      "learning_rate": 0.00010424975799158456,
      "loss": 1.3376,
      "step": 14014
    },
    {
      "epoch": 0.5019069958995112,
      "grad_norm": 1.6113402843475342,
      "learning_rate": 0.00010423816942898659,
      "loss": 1.6665,
      "step": 14015
    },
    {
      "epoch": 0.5019428080290794,
      "grad_norm": 1.607465147972107,
      "learning_rate": 0.00010422658080936947,
      "loss": 1.5115,
      "step": 14016
    },
    {
      "epoch": 0.5019786201586477,
      "grad_norm": 1.796221375465393,
      "learning_rate": 0.00010421499213288919,
      "loss": 1.5607,
      "step": 14017
    },
    {
      "epoch": 0.5020144322882161,
      "grad_norm": 2.2916972637176514,
      "learning_rate": 0.00010420340339970163,
      "loss": 1.6604,
      "step": 14018
    },
    {
      "epoch": 0.5020502444177843,
      "grad_norm": 1.6994593143463135,
      "learning_rate": 0.0001041918146099627,
      "loss": 1.5249,
      "step": 14019
    },
    {
      "epoch": 0.5020860565473526,
      "grad_norm": 1.3633710145950317,
      "learning_rate": 0.00010418022576382831,
      "loss": 1.2622,
      "step": 14020
    },
    {
      "epoch": 0.5021218686769209,
      "grad_norm": 2.406529426574707,
      "learning_rate": 0.00010416863686145434,
      "loss": 1.3539,
      "step": 14021
    },
    {
      "epoch": 0.5021576808064891,
      "grad_norm": 1.4602991342544556,
      "learning_rate": 0.00010415704790299678,
      "loss": 1.3125,
      "step": 14022
    },
    {
      "epoch": 0.5021934929360574,
      "grad_norm": 1.564888596534729,
      "learning_rate": 0.00010414545888861149,
      "loss": 1.4893,
      "step": 14023
    },
    {
      "epoch": 0.5022293050656257,
      "grad_norm": 1.9032400846481323,
      "learning_rate": 0.0001041338698184544,
      "loss": 1.425,
      "step": 14024
    },
    {
      "epoch": 0.5022651171951941,
      "grad_norm": 2.332868814468384,
      "learning_rate": 0.00010412228069268142,
      "loss": 1.4777,
      "step": 14025
    },
    {
      "epoch": 0.5023009293247623,
      "grad_norm": 1.6929960250854492,
      "learning_rate": 0.00010411069151144848,
      "loss": 1.729,
      "step": 14026
    },
    {
      "epoch": 0.5023367414543306,
      "grad_norm": 2.2395436763763428,
      "learning_rate": 0.00010409910227491146,
      "loss": 1.6846,
      "step": 14027
    },
    {
      "epoch": 0.5023725535838989,
      "grad_norm": 1.671336054801941,
      "learning_rate": 0.00010408751298322634,
      "loss": 1.2541,
      "step": 14028
    },
    {
      "epoch": 0.5024083657134671,
      "grad_norm": 1.7518086433410645,
      "learning_rate": 0.00010407592363654901,
      "loss": 1.3721,
      "step": 14029
    },
    {
      "epoch": 0.5024441778430354,
      "grad_norm": 2.0253491401672363,
      "learning_rate": 0.00010406433423503534,
      "loss": 1.5186,
      "step": 14030
    },
    {
      "epoch": 0.5024799899726037,
      "grad_norm": 1.4864667654037476,
      "learning_rate": 0.00010405274477884135,
      "loss": 1.1936,
      "step": 14031
    },
    {
      "epoch": 0.5025158021021721,
      "grad_norm": 1.1731634140014648,
      "learning_rate": 0.00010404115526812286,
      "loss": 1.4125,
      "step": 14032
    },
    {
      "epoch": 0.5025516142317403,
      "grad_norm": 1.3496315479278564,
      "learning_rate": 0.00010402956570303586,
      "loss": 1.3462,
      "step": 14033
    },
    {
      "epoch": 0.5025874263613086,
      "grad_norm": 1.6473044157028198,
      "learning_rate": 0.00010401797608373625,
      "loss": 1.7143,
      "step": 14034
    },
    {
      "epoch": 0.5026232384908769,
      "grad_norm": 2.9563591480255127,
      "learning_rate": 0.00010400638641037996,
      "loss": 1.505,
      "step": 14035
    },
    {
      "epoch": 0.5026590506204451,
      "grad_norm": 1.4683979749679565,
      "learning_rate": 0.00010399479668312288,
      "loss": 1.5082,
      "step": 14036
    },
    {
      "epoch": 0.5026948627500134,
      "grad_norm": 1.4357550144195557,
      "learning_rate": 0.00010398320690212102,
      "loss": 1.4316,
      "step": 14037
    },
    {
      "epoch": 0.5027306748795817,
      "grad_norm": 1.8764081001281738,
      "learning_rate": 0.00010397161706753021,
      "loss": 1.4092,
      "step": 14038
    },
    {
      "epoch": 0.50276648700915,
      "grad_norm": 1.5990819931030273,
      "learning_rate": 0.00010396002717950644,
      "loss": 1.4917,
      "step": 14039
    },
    {
      "epoch": 0.5028022991387183,
      "grad_norm": 1.571290373802185,
      "learning_rate": 0.00010394843723820558,
      "loss": 1.3678,
      "step": 14040
    },
    {
      "epoch": 0.5028381112682866,
      "grad_norm": 2.3405323028564453,
      "learning_rate": 0.00010393684724378358,
      "loss": 1.4077,
      "step": 14041
    },
    {
      "epoch": 0.5028739233978549,
      "grad_norm": 1.5465885400772095,
      "learning_rate": 0.00010392525719639642,
      "loss": 0.9557,
      "step": 14042
    },
    {
      "epoch": 0.5029097355274231,
      "grad_norm": 1.379290223121643,
      "learning_rate": 0.00010391366709619994,
      "loss": 1.3364,
      "step": 14043
    },
    {
      "epoch": 0.5029455476569914,
      "grad_norm": 1.6196434497833252,
      "learning_rate": 0.00010390207694335017,
      "loss": 1.4637,
      "step": 14044
    },
    {
      "epoch": 0.5029813597865597,
      "grad_norm": 1.7534099817276,
      "learning_rate": 0.00010389048673800294,
      "loss": 1.4994,
      "step": 14045
    },
    {
      "epoch": 0.503017171916128,
      "grad_norm": 2.0215682983398438,
      "learning_rate": 0.00010387889648031428,
      "loss": 1.5993,
      "step": 14046
    },
    {
      "epoch": 0.5030529840456963,
      "grad_norm": 1.6501681804656982,
      "learning_rate": 0.00010386730617044005,
      "loss": 1.3859,
      "step": 14047
    },
    {
      "epoch": 0.5030887961752646,
      "grad_norm": 1.9652694463729858,
      "learning_rate": 0.0001038557158085362,
      "loss": 1.6148,
      "step": 14048
    },
    {
      "epoch": 0.5031246083048329,
      "grad_norm": 1.6333247423171997,
      "learning_rate": 0.00010384412539475865,
      "loss": 1.1812,
      "step": 14049
    },
    {
      "epoch": 0.5031604204344011,
      "grad_norm": 1.772522211074829,
      "learning_rate": 0.00010383253492926339,
      "loss": 1.6932,
      "step": 14050
    },
    {
      "epoch": 0.5031962325639694,
      "grad_norm": 1.6307519674301147,
      "learning_rate": 0.00010382094441220627,
      "loss": 1.4595,
      "step": 14051
    },
    {
      "epoch": 0.5032320446935377,
      "grad_norm": 1.7731724977493286,
      "learning_rate": 0.00010380935384374331,
      "loss": 1.5997,
      "step": 14052
    },
    {
      "epoch": 0.503267856823106,
      "grad_norm": 1.5439388751983643,
      "learning_rate": 0.00010379776322403039,
      "loss": 1.2718,
      "step": 14053
    },
    {
      "epoch": 0.5033036689526743,
      "grad_norm": 2.08516001701355,
      "learning_rate": 0.00010378617255322344,
      "loss": 1.1215,
      "step": 14054
    },
    {
      "epoch": 0.5033394810822426,
      "grad_norm": 1.9739441871643066,
      "learning_rate": 0.00010377458183147848,
      "loss": 1.5623,
      "step": 14055
    },
    {
      "epoch": 0.5033752932118108,
      "grad_norm": 1.473125696182251,
      "learning_rate": 0.00010376299105895135,
      "loss": 1.4087,
      "step": 14056
    },
    {
      "epoch": 0.5034111053413791,
      "grad_norm": 1.925338625907898,
      "learning_rate": 0.00010375140023579805,
      "loss": 1.5258,
      "step": 14057
    },
    {
      "epoch": 0.5034469174709474,
      "grad_norm": 1.9338470697402954,
      "learning_rate": 0.0001037398093621745,
      "loss": 1.7284,
      "step": 14058
    },
    {
      "epoch": 0.5034827296005157,
      "grad_norm": 1.9991499185562134,
      "learning_rate": 0.00010372821843823661,
      "loss": 1.5905,
      "step": 14059
    },
    {
      "epoch": 0.503518541730084,
      "grad_norm": 1.4696353673934937,
      "learning_rate": 0.00010371662746414037,
      "loss": 1.4171,
      "step": 14060
    },
    {
      "epoch": 0.5035543538596523,
      "grad_norm": 1.5892161130905151,
      "learning_rate": 0.00010370503644004171,
      "loss": 1.5635,
      "step": 14061
    },
    {
      "epoch": 0.5035901659892206,
      "grad_norm": 1.718597650527954,
      "learning_rate": 0.00010369344536609653,
      "loss": 1.7008,
      "step": 14062
    },
    {
      "epoch": 0.5036259781187888,
      "grad_norm": 1.6046531200408936,
      "learning_rate": 0.00010368185424246084,
      "loss": 1.7398,
      "step": 14063
    },
    {
      "epoch": 0.5036617902483571,
      "grad_norm": 1.702903389930725,
      "learning_rate": 0.00010367026306929056,
      "loss": 1.6981,
      "step": 14064
    },
    {
      "epoch": 0.5036976023779254,
      "grad_norm": 2.5990512371063232,
      "learning_rate": 0.00010365867184674159,
      "loss": 1.39,
      "step": 14065
    },
    {
      "epoch": 0.5037334145074936,
      "grad_norm": 1.5704472064971924,
      "learning_rate": 0.00010364708057496992,
      "loss": 1.6272,
      "step": 14066
    },
    {
      "epoch": 0.5037692266370619,
      "grad_norm": 1.577022671699524,
      "learning_rate": 0.00010363548925413149,
      "loss": 1.2842,
      "step": 14067
    },
    {
      "epoch": 0.5038050387666303,
      "grad_norm": 2.005908250808716,
      "learning_rate": 0.00010362389788438225,
      "loss": 1.2809,
      "step": 14068
    },
    {
      "epoch": 0.5038408508961986,
      "grad_norm": 2.519007444381714,
      "learning_rate": 0.00010361230646587812,
      "loss": 1.509,
      "step": 14069
    },
    {
      "epoch": 0.5038766630257668,
      "grad_norm": 2.2054390907287598,
      "learning_rate": 0.00010360071499877508,
      "loss": 1.4182,
      "step": 14070
    },
    {
      "epoch": 0.5039124751553351,
      "grad_norm": 1.6459708213806152,
      "learning_rate": 0.00010358912348322904,
      "loss": 1.1879,
      "step": 14071
    },
    {
      "epoch": 0.5039482872849034,
      "grad_norm": 2.3594183921813965,
      "learning_rate": 0.00010357753191939601,
      "loss": 1.7322,
      "step": 14072
    },
    {
      "epoch": 0.5039840994144716,
      "grad_norm": 1.8860596418380737,
      "learning_rate": 0.0001035659403074319,
      "loss": 1.8158,
      "step": 14073
    },
    {
      "epoch": 0.5040199115440399,
      "grad_norm": 1.5959560871124268,
      "learning_rate": 0.00010355434864749262,
      "loss": 1.337,
      "step": 14074
    },
    {
      "epoch": 0.5040557236736083,
      "grad_norm": 1.7443625926971436,
      "learning_rate": 0.0001035427569397342,
      "loss": 1.4158,
      "step": 14075
    },
    {
      "epoch": 0.5040915358031766,
      "grad_norm": 1.536818027496338,
      "learning_rate": 0.00010353116518431254,
      "loss": 1.351,
      "step": 14076
    },
    {
      "epoch": 0.5041273479327448,
      "grad_norm": 1.3087923526763916,
      "learning_rate": 0.00010351957338138363,
      "loss": 1.7281,
      "step": 14077
    },
    {
      "epoch": 0.5041631600623131,
      "grad_norm": 1.8406116962432861,
      "learning_rate": 0.00010350798153110337,
      "loss": 1.114,
      "step": 14078
    },
    {
      "epoch": 0.5041989721918814,
      "grad_norm": 1.7270954847335815,
      "learning_rate": 0.00010349638963362777,
      "loss": 1.6466,
      "step": 14079
    },
    {
      "epoch": 0.5042347843214496,
      "grad_norm": 2.479389190673828,
      "learning_rate": 0.00010348479768911272,
      "loss": 1.2647,
      "step": 14080
    },
    {
      "epoch": 0.5042705964510179,
      "grad_norm": 1.7443976402282715,
      "learning_rate": 0.00010347320569771428,
      "loss": 1.6703,
      "step": 14081
    },
    {
      "epoch": 0.5043064085805863,
      "grad_norm": 1.6824451684951782,
      "learning_rate": 0.00010346161365958829,
      "loss": 1.1959,
      "step": 14082
    },
    {
      "epoch": 0.5043422207101546,
      "grad_norm": 1.8849799633026123,
      "learning_rate": 0.00010345002157489074,
      "loss": 1.3789,
      "step": 14083
    },
    {
      "epoch": 0.5043780328397228,
      "grad_norm": 1.5173338651657104,
      "learning_rate": 0.00010343842944377764,
      "loss": 1.7663,
      "step": 14084
    },
    {
      "epoch": 0.5044138449692911,
      "grad_norm": 1.5596895217895508,
      "learning_rate": 0.00010342683726640487,
      "loss": 1.4884,
      "step": 14085
    },
    {
      "epoch": 0.5044496570988594,
      "grad_norm": 1.4701156616210938,
      "learning_rate": 0.00010341524504292845,
      "loss": 1.5526,
      "step": 14086
    },
    {
      "epoch": 0.5044854692284276,
      "grad_norm": 1.7516635656356812,
      "learning_rate": 0.00010340365277350428,
      "loss": 1.4059,
      "step": 14087
    },
    {
      "epoch": 0.5045212813579959,
      "grad_norm": 1.8188012838363647,
      "learning_rate": 0.0001033920604582884,
      "loss": 1.4001,
      "step": 14088
    },
    {
      "epoch": 0.5045570934875643,
      "grad_norm": 2.2817301750183105,
      "learning_rate": 0.00010338046809743668,
      "loss": 1.162,
      "step": 14089
    },
    {
      "epoch": 0.5045929056171325,
      "grad_norm": 1.7542790174484253,
      "learning_rate": 0.00010336887569110518,
      "loss": 1.505,
      "step": 14090
    },
    {
      "epoch": 0.5046287177467008,
      "grad_norm": 1.8242846727371216,
      "learning_rate": 0.00010335728323944974,
      "loss": 1.319,
      "step": 14091
    },
    {
      "epoch": 0.5046645298762691,
      "grad_norm": 1.5669910907745361,
      "learning_rate": 0.00010334569074262641,
      "loss": 1.4626,
      "step": 14092
    },
    {
      "epoch": 0.5047003420058374,
      "grad_norm": 1.8271517753601074,
      "learning_rate": 0.00010333409820079112,
      "loss": 1.4747,
      "step": 14093
    },
    {
      "epoch": 0.5047361541354056,
      "grad_norm": 1.7885736227035522,
      "learning_rate": 0.00010332250561409986,
      "loss": 1.4699,
      "step": 14094
    },
    {
      "epoch": 0.5047719662649739,
      "grad_norm": 1.3920366764068604,
      "learning_rate": 0.00010331091298270854,
      "loss": 1.4448,
      "step": 14095
    },
    {
      "epoch": 0.5048077783945423,
      "grad_norm": 2.2439980506896973,
      "learning_rate": 0.00010329932030677316,
      "loss": 1.6206,
      "step": 14096
    },
    {
      "epoch": 0.5048435905241105,
      "grad_norm": 1.532996654510498,
      "learning_rate": 0.00010328772758644971,
      "loss": 1.4726,
      "step": 14097
    },
    {
      "epoch": 0.5048794026536788,
      "grad_norm": 1.5202641487121582,
      "learning_rate": 0.00010327613482189409,
      "loss": 1.3134,
      "step": 14098
    },
    {
      "epoch": 0.5049152147832471,
      "grad_norm": 2.1792054176330566,
      "learning_rate": 0.00010326454201326236,
      "loss": 1.3475,
      "step": 14099
    },
    {
      "epoch": 0.5049510269128153,
      "grad_norm": 1.6390247344970703,
      "learning_rate": 0.00010325294916071038,
      "loss": 1.262,
      "step": 14100
    },
    {
      "epoch": 0.5049868390423836,
      "grad_norm": 1.5163406133651733,
      "learning_rate": 0.00010324135626439419,
      "loss": 1.4472,
      "step": 14101
    },
    {
      "epoch": 0.5050226511719519,
      "grad_norm": 1.6749697923660278,
      "learning_rate": 0.0001032297633244697,
      "loss": 1.4647,
      "step": 14102
    },
    {
      "epoch": 0.5050584633015203,
      "grad_norm": 1.6296559572219849,
      "learning_rate": 0.00010321817034109293,
      "loss": 1.1982,
      "step": 14103
    },
    {
      "epoch": 0.5050942754310885,
      "grad_norm": 1.5373493432998657,
      "learning_rate": 0.00010320657731441982,
      "loss": 1.4018,
      "step": 14104
    },
    {
      "epoch": 0.5051300875606568,
      "grad_norm": 1.7386698722839355,
      "learning_rate": 0.00010319498424460636,
      "loss": 1.302,
      "step": 14105
    },
    {
      "epoch": 0.5051658996902251,
      "grad_norm": 2.1452550888061523,
      "learning_rate": 0.0001031833911318085,
      "loss": 1.5013,
      "step": 14106
    },
    {
      "epoch": 0.5052017118197933,
      "grad_norm": 1.7856847047805786,
      "learning_rate": 0.00010317179797618223,
      "loss": 1.5817,
      "step": 14107
    },
    {
      "epoch": 0.5052375239493616,
      "grad_norm": 1.160574197769165,
      "learning_rate": 0.00010316020477788353,
      "loss": 1.6855,
      "step": 14108
    },
    {
      "epoch": 0.5052733360789299,
      "grad_norm": 1.5507274866104126,
      "learning_rate": 0.0001031486115370683,
      "loss": 1.6322,
      "step": 14109
    },
    {
      "epoch": 0.5053091482084983,
      "grad_norm": 1.567002296447754,
      "learning_rate": 0.00010313701825389259,
      "loss": 1.4388,
      "step": 14110
    },
    {
      "epoch": 0.5053449603380665,
      "grad_norm": 1.6256871223449707,
      "learning_rate": 0.00010312542492851234,
      "loss": 1.1,
      "step": 14111
    },
    {
      "epoch": 0.5053807724676348,
      "grad_norm": 1.9150153398513794,
      "learning_rate": 0.00010311383156108354,
      "loss": 1.8181,
      "step": 14112
    },
    {
      "epoch": 0.5054165845972031,
      "grad_norm": 1.7468947172164917,
      "learning_rate": 0.00010310223815176215,
      "loss": 1.3511,
      "step": 14113
    },
    {
      "epoch": 0.5054523967267713,
      "grad_norm": 1.5880811214447021,
      "learning_rate": 0.00010309064470070414,
      "loss": 1.5384,
      "step": 14114
    },
    {
      "epoch": 0.5054882088563396,
      "grad_norm": 1.6013243198394775,
      "learning_rate": 0.00010307905120806549,
      "loss": 1.4457,
      "step": 14115
    },
    {
      "epoch": 0.5055240209859079,
      "grad_norm": 1.7990401983261108,
      "learning_rate": 0.00010306745767400219,
      "loss": 1.4054,
      "step": 14116
    },
    {
      "epoch": 0.5055598331154763,
      "grad_norm": 2.4490861892700195,
      "learning_rate": 0.0001030558640986702,
      "loss": 1.6798,
      "step": 14117
    },
    {
      "epoch": 0.5055956452450445,
      "grad_norm": 1.4762895107269287,
      "learning_rate": 0.0001030442704822255,
      "loss": 1.1657,
      "step": 14118
    },
    {
      "epoch": 0.5056314573746128,
      "grad_norm": 1.4917511940002441,
      "learning_rate": 0.00010303267682482405,
      "loss": 1.4921,
      "step": 14119
    },
    {
      "epoch": 0.5056672695041811,
      "grad_norm": 1.858914852142334,
      "learning_rate": 0.00010302108312662184,
      "loss": 1.2912,
      "step": 14120
    },
    {
      "epoch": 0.5057030816337493,
      "grad_norm": 2.0417349338531494,
      "learning_rate": 0.00010300948938777491,
      "loss": 1.7917,
      "step": 14121
    },
    {
      "epoch": 0.5057388937633176,
      "grad_norm": 1.6435681581497192,
      "learning_rate": 0.00010299789560843911,
      "loss": 1.3145,
      "step": 14122
    },
    {
      "epoch": 0.5057747058928859,
      "grad_norm": 1.7804651260375977,
      "learning_rate": 0.00010298630178877053,
      "loss": 1.4567,
      "step": 14123
    },
    {
      "epoch": 0.5058105180224542,
      "grad_norm": 1.9142791032791138,
      "learning_rate": 0.00010297470792892512,
      "loss": 1.1872,
      "step": 14124
    },
    {
      "epoch": 0.5058463301520225,
      "grad_norm": 2.6280245780944824,
      "learning_rate": 0.00010296311402905884,
      "loss": 1.3906,
      "step": 14125
    },
    {
      "epoch": 0.5058821422815908,
      "grad_norm": 1.2249655723571777,
      "learning_rate": 0.0001029515200893277,
      "loss": 1.4255,
      "step": 14126
    },
    {
      "epoch": 0.505917954411159,
      "grad_norm": 1.620898962020874,
      "learning_rate": 0.00010293992610988763,
      "loss": 1.3322,
      "step": 14127
    },
    {
      "epoch": 0.5059537665407273,
      "grad_norm": 1.8030281066894531,
      "learning_rate": 0.00010292833209089467,
      "loss": 1.2237,
      "step": 14128
    },
    {
      "epoch": 0.5059895786702956,
      "grad_norm": 1.8104889392852783,
      "learning_rate": 0.00010291673803250477,
      "loss": 1.4957,
      "step": 14129
    },
    {
      "epoch": 0.5060253907998639,
      "grad_norm": 1.5003832578659058,
      "learning_rate": 0.00010290514393487391,
      "loss": 1.5665,
      "step": 14130
    },
    {
      "epoch": 0.5060612029294322,
      "grad_norm": 1.9296363592147827,
      "learning_rate": 0.00010289354979815811,
      "loss": 1.5859,
      "step": 14131
    },
    {
      "epoch": 0.5060970150590005,
      "grad_norm": 1.9950153827667236,
      "learning_rate": 0.00010288195562251332,
      "loss": 1.5064,
      "step": 14132
    },
    {
      "epoch": 0.5061328271885688,
      "grad_norm": 2.030654191970825,
      "learning_rate": 0.00010287036140809552,
      "loss": 1.5293,
      "step": 14133
    },
    {
      "epoch": 0.506168639318137,
      "grad_norm": 1.89315664768219,
      "learning_rate": 0.00010285876715506076,
      "loss": 1.3902,
      "step": 14134
    },
    {
      "epoch": 0.5062044514477053,
      "grad_norm": 1.6932226419448853,
      "learning_rate": 0.00010284717286356493,
      "loss": 1.3832,
      "step": 14135
    },
    {
      "epoch": 0.5062402635772736,
      "grad_norm": 1.2903110980987549,
      "learning_rate": 0.00010283557853376408,
      "loss": 1.3911,
      "step": 14136
    },
    {
      "epoch": 0.5062760757068419,
      "grad_norm": 1.9288479089736938,
      "learning_rate": 0.00010282398416581415,
      "loss": 1.7958,
      "step": 14137
    },
    {
      "epoch": 0.5063118878364102,
      "grad_norm": 1.6040328741073608,
      "learning_rate": 0.00010281238975987118,
      "loss": 1.5238,
      "step": 14138
    },
    {
      "epoch": 0.5063476999659785,
      "grad_norm": 2.265596866607666,
      "learning_rate": 0.00010280079531609112,
      "loss": 1.4235,
      "step": 14139
    },
    {
      "epoch": 0.5063835120955468,
      "grad_norm": 1.681878685951233,
      "learning_rate": 0.00010278920083462997,
      "loss": 1.6886,
      "step": 14140
    },
    {
      "epoch": 0.506419324225115,
      "grad_norm": 2.133225917816162,
      "learning_rate": 0.00010277760631564375,
      "loss": 1.2276,
      "step": 14141
    },
    {
      "epoch": 0.5064551363546833,
      "grad_norm": 1.6065014600753784,
      "learning_rate": 0.00010276601175928839,
      "loss": 1.4719,
      "step": 14142
    },
    {
      "epoch": 0.5064909484842516,
      "grad_norm": 1.765504240989685,
      "learning_rate": 0.00010275441716571996,
      "loss": 1.3059,
      "step": 14143
    },
    {
      "epoch": 0.5065267606138198,
      "grad_norm": 1.4708679914474487,
      "learning_rate": 0.00010274282253509436,
      "loss": 1.4785,
      "step": 14144
    },
    {
      "epoch": 0.5065625727433882,
      "grad_norm": 1.4955041408538818,
      "learning_rate": 0.00010273122786756762,
      "loss": 1.5128,
      "step": 14145
    },
    {
      "epoch": 0.5065983848729565,
      "grad_norm": 1.961424469947815,
      "learning_rate": 0.00010271963316329571,
      "loss": 1.4041,
      "step": 14146
    },
    {
      "epoch": 0.5066341970025248,
      "grad_norm": 1.4497519731521606,
      "learning_rate": 0.00010270803842243469,
      "loss": 1.7535,
      "step": 14147
    },
    {
      "epoch": 0.506670009132093,
      "grad_norm": 1.6153457164764404,
      "learning_rate": 0.00010269644364514046,
      "loss": 1.5254,
      "step": 14148
    },
    {
      "epoch": 0.5067058212616613,
      "grad_norm": 1.7884387969970703,
      "learning_rate": 0.0001026848488315691,
      "loss": 1.5087,
      "step": 14149
    },
    {
      "epoch": 0.5067416333912296,
      "grad_norm": 1.7952286005020142,
      "learning_rate": 0.00010267325398187653,
      "loss": 1.4182,
      "step": 14150
    },
    {
      "epoch": 0.5067774455207978,
      "grad_norm": 2.6730382442474365,
      "learning_rate": 0.00010266165909621879,
      "loss": 1.722,
      "step": 14151
    },
    {
      "epoch": 0.5068132576503662,
      "grad_norm": 1.7425639629364014,
      "learning_rate": 0.00010265006417475189,
      "loss": 1.6575,
      "step": 14152
    },
    {
      "epoch": 0.5068490697799345,
      "grad_norm": 1.971174955368042,
      "learning_rate": 0.00010263846921763174,
      "loss": 1.3717,
      "step": 14153
    },
    {
      "epoch": 0.5068848819095028,
      "grad_norm": 1.8181272745132446,
      "learning_rate": 0.00010262687422501442,
      "loss": 1.2995,
      "step": 14154
    },
    {
      "epoch": 0.506920694039071,
      "grad_norm": 1.9725772142410278,
      "learning_rate": 0.00010261527919705589,
      "loss": 1.7611,
      "step": 14155
    },
    {
      "epoch": 0.5069565061686393,
      "grad_norm": 2.269573926925659,
      "learning_rate": 0.00010260368413391217,
      "loss": 1.5659,
      "step": 14156
    },
    {
      "epoch": 0.5069923182982076,
      "grad_norm": 1.5113763809204102,
      "learning_rate": 0.0001025920890357392,
      "loss": 1.4647,
      "step": 14157
    },
    {
      "epoch": 0.5070281304277758,
      "grad_norm": 1.9952131509780884,
      "learning_rate": 0.00010258049390269305,
      "loss": 1.5054,
      "step": 14158
    },
    {
      "epoch": 0.5070639425573442,
      "grad_norm": 1.6506963968276978,
      "learning_rate": 0.00010256889873492966,
      "loss": 1.5954,
      "step": 14159
    },
    {
      "epoch": 0.5070997546869125,
      "grad_norm": 1.6681649684906006,
      "learning_rate": 0.00010255730353260507,
      "loss": 1.2699,
      "step": 14160
    },
    {
      "epoch": 0.5071355668164808,
      "grad_norm": 1.7544012069702148,
      "learning_rate": 0.00010254570829587527,
      "loss": 1.2456,
      "step": 14161
    },
    {
      "epoch": 0.507171378946049,
      "grad_norm": 1.9082049131393433,
      "learning_rate": 0.00010253411302489622,
      "loss": 1.3864,
      "step": 14162
    },
    {
      "epoch": 0.5072071910756173,
      "grad_norm": 5.814779758453369,
      "learning_rate": 0.00010252251771982395,
      "loss": 1.5455,
      "step": 14163
    },
    {
      "epoch": 0.5072430032051856,
      "grad_norm": 1.5114103555679321,
      "learning_rate": 0.00010251092238081446,
      "loss": 1.7425,
      "step": 14164
    },
    {
      "epoch": 0.5072788153347538,
      "grad_norm": 1.5640950202941895,
      "learning_rate": 0.00010249932700802376,
      "loss": 1.5903,
      "step": 14165
    },
    {
      "epoch": 0.5073146274643222,
      "grad_norm": 2.070931911468506,
      "learning_rate": 0.00010248773160160782,
      "loss": 1.5774,
      "step": 14166
    },
    {
      "epoch": 0.5073504395938905,
      "grad_norm": 1.4230797290802002,
      "learning_rate": 0.0001024761361617227,
      "loss": 1.6921,
      "step": 14167
    },
    {
      "epoch": 0.5073862517234587,
      "grad_norm": 1.4877146482467651,
      "learning_rate": 0.00010246454068852431,
      "loss": 1.3322,
      "step": 14168
    },
    {
      "epoch": 0.507422063853027,
      "grad_norm": 2.0262210369110107,
      "learning_rate": 0.00010245294518216875,
      "loss": 1.4121,
      "step": 14169
    },
    {
      "epoch": 0.5074578759825953,
      "grad_norm": 1.7050576210021973,
      "learning_rate": 0.00010244134964281195,
      "loss": 1.336,
      "step": 14170
    },
    {
      "epoch": 0.5074936881121636,
      "grad_norm": 2.0105693340301514,
      "learning_rate": 0.00010242975407060995,
      "loss": 1.2359,
      "step": 14171
    },
    {
      "epoch": 0.5075295002417318,
      "grad_norm": 1.4477938413619995,
      "learning_rate": 0.00010241815846571874,
      "loss": 1.7411,
      "step": 14172
    },
    {
      "epoch": 0.5075653123713002,
      "grad_norm": 1.3273900747299194,
      "learning_rate": 0.00010240656282829433,
      "loss": 1.434,
      "step": 14173
    },
    {
      "epoch": 0.5076011245008685,
      "grad_norm": 1.914575457572937,
      "learning_rate": 0.00010239496715849273,
      "loss": 1.2457,
      "step": 14174
    },
    {
      "epoch": 0.5076369366304367,
      "grad_norm": 1.6995282173156738,
      "learning_rate": 0.0001023833714564699,
      "loss": 1.5573,
      "step": 14175
    },
    {
      "epoch": 0.507672748760005,
      "grad_norm": 1.8010412454605103,
      "learning_rate": 0.00010237177572238192,
      "loss": 1.5153,
      "step": 14176
    },
    {
      "epoch": 0.5077085608895733,
      "grad_norm": 1.5251318216323853,
      "learning_rate": 0.00010236017995638472,
      "loss": 1.5024,
      "step": 14177
    },
    {
      "epoch": 0.5077443730191415,
      "grad_norm": 3.0716559886932373,
      "learning_rate": 0.00010234858415863439,
      "loss": 1.5426,
      "step": 14178
    },
    {
      "epoch": 0.5077801851487098,
      "grad_norm": 2.010403633117676,
      "learning_rate": 0.00010233698832928686,
      "loss": 1.3712,
      "step": 14179
    },
    {
      "epoch": 0.5078159972782782,
      "grad_norm": 1.76226806640625,
      "learning_rate": 0.00010232539246849818,
      "loss": 1.4153,
      "step": 14180
    },
    {
      "epoch": 0.5078518094078465,
      "grad_norm": 1.8289686441421509,
      "learning_rate": 0.00010231379657642432,
      "loss": 1.4113,
      "step": 14181
    },
    {
      "epoch": 0.5078876215374147,
      "grad_norm": 1.5149344205856323,
      "learning_rate": 0.00010230220065322132,
      "loss": 1.4344,
      "step": 14182
    },
    {
      "epoch": 0.507923433666983,
      "grad_norm": 1.530060052871704,
      "learning_rate": 0.00010229060469904519,
      "loss": 1.0206,
      "step": 14183
    },
    {
      "epoch": 0.5079592457965513,
      "grad_norm": 2.5815203189849854,
      "learning_rate": 0.00010227900871405191,
      "loss": 1.6194,
      "step": 14184
    },
    {
      "epoch": 0.5079950579261195,
      "grad_norm": 1.5144827365875244,
      "learning_rate": 0.00010226741269839755,
      "loss": 1.377,
      "step": 14185
    },
    {
      "epoch": 0.5080308700556878,
      "grad_norm": 1.7543262243270874,
      "learning_rate": 0.00010225581665223802,
      "loss": 1.4925,
      "step": 14186
    },
    {
      "epoch": 0.5080666821852562,
      "grad_norm": 1.8062946796417236,
      "learning_rate": 0.00010224422057572947,
      "loss": 1.6961,
      "step": 14187
    },
    {
      "epoch": 0.5081024943148245,
      "grad_norm": 1.5800213813781738,
      "learning_rate": 0.00010223262446902775,
      "loss": 1.6731,
      "step": 14188
    },
    {
      "epoch": 0.5081383064443927,
      "grad_norm": 1.6881344318389893,
      "learning_rate": 0.00010222102833228897,
      "loss": 1.488,
      "step": 14189
    },
    {
      "epoch": 0.508174118573961,
      "grad_norm": 1.4703749418258667,
      "learning_rate": 0.00010220943216566912,
      "loss": 1.2877,
      "step": 14190
    },
    {
      "epoch": 0.5082099307035293,
      "grad_norm": 2.940575361251831,
      "learning_rate": 0.00010219783596932421,
      "loss": 1.3642,
      "step": 14191
    },
    {
      "epoch": 0.5082457428330975,
      "grad_norm": 1.343244194984436,
      "learning_rate": 0.00010218623974341024,
      "loss": 1.5526,
      "step": 14192
    },
    {
      "epoch": 0.5082815549626658,
      "grad_norm": 1.7862319946289062,
      "learning_rate": 0.00010217464348808323,
      "loss": 1.5906,
      "step": 14193
    },
    {
      "epoch": 0.5083173670922342,
      "grad_norm": 2.7882981300354004,
      "learning_rate": 0.00010216304720349922,
      "loss": 1.6598,
      "step": 14194
    },
    {
      "epoch": 0.5083531792218025,
      "grad_norm": 1.9673621654510498,
      "learning_rate": 0.00010215145088981419,
      "loss": 1.6815,
      "step": 14195
    },
    {
      "epoch": 0.5083889913513707,
      "grad_norm": 1.8103598356246948,
      "learning_rate": 0.0001021398545471842,
      "loss": 1.2642,
      "step": 14196
    },
    {
      "epoch": 0.508424803480939,
      "grad_norm": 1.8523528575897217,
      "learning_rate": 0.00010212825817576519,
      "loss": 1.7025,
      "step": 14197
    },
    {
      "epoch": 0.5084606156105073,
      "grad_norm": 1.207170009613037,
      "learning_rate": 0.00010211666177571322,
      "loss": 1.4524,
      "step": 14198
    },
    {
      "epoch": 0.5084964277400755,
      "grad_norm": 1.7149916887283325,
      "learning_rate": 0.00010210506534718427,
      "loss": 1.4284,
      "step": 14199
    },
    {
      "epoch": 0.5085322398696438,
      "grad_norm": 1.2960801124572754,
      "learning_rate": 0.00010209346889033442,
      "loss": 1.4645,
      "step": 14200
    },
    {
      "epoch": 0.5085680519992122,
      "grad_norm": 1.768568754196167,
      "learning_rate": 0.00010208187240531962,
      "loss": 1.2931,
      "step": 14201
    },
    {
      "epoch": 0.5086038641287804,
      "grad_norm": 1.9338597059249878,
      "learning_rate": 0.00010207027589229594,
      "loss": 1.6961,
      "step": 14202
    },
    {
      "epoch": 0.5086396762583487,
      "grad_norm": 1.3252862691879272,
      "learning_rate": 0.00010205867935141933,
      "loss": 1.5093,
      "step": 14203
    },
    {
      "epoch": 0.508675488387917,
      "grad_norm": 1.6431336402893066,
      "learning_rate": 0.00010204708278284587,
      "loss": 1.4769,
      "step": 14204
    },
    {
      "epoch": 0.5087113005174853,
      "grad_norm": 1.6076064109802246,
      "learning_rate": 0.00010203548618673155,
      "loss": 1.5178,
      "step": 14205
    },
    {
      "epoch": 0.5087471126470535,
      "grad_norm": 1.594680905342102,
      "learning_rate": 0.00010202388956323238,
      "loss": 1.4088,
      "step": 14206
    },
    {
      "epoch": 0.5087829247766218,
      "grad_norm": 1.4047375917434692,
      "learning_rate": 0.0001020122929125044,
      "loss": 1.4558,
      "step": 14207
    },
    {
      "epoch": 0.5088187369061902,
      "grad_norm": 1.6259403228759766,
      "learning_rate": 0.00010200069623470358,
      "loss": 1.2941,
      "step": 14208
    },
    {
      "epoch": 0.5088545490357584,
      "grad_norm": 1.942922830581665,
      "learning_rate": 0.00010198909952998603,
      "loss": 1.6474,
      "step": 14209
    },
    {
      "epoch": 0.5088903611653267,
      "grad_norm": 1.5637397766113281,
      "learning_rate": 0.00010197750279850767,
      "loss": 1.4758,
      "step": 14210
    },
    {
      "epoch": 0.508926173294895,
      "grad_norm": 1.5786371231079102,
      "learning_rate": 0.00010196590604042457,
      "loss": 1.3429,
      "step": 14211
    },
    {
      "epoch": 0.5089619854244632,
      "grad_norm": 1.7941700220108032,
      "learning_rate": 0.00010195430925589274,
      "loss": 1.4954,
      "step": 14212
    },
    {
      "epoch": 0.5089977975540315,
      "grad_norm": 1.7135039567947388,
      "learning_rate": 0.00010194271244506821,
      "loss": 1.5166,
      "step": 14213
    },
    {
      "epoch": 0.5090336096835998,
      "grad_norm": 1.4030609130859375,
      "learning_rate": 0.00010193111560810697,
      "loss": 1.5622,
      "step": 14214
    },
    {
      "epoch": 0.5090694218131682,
      "grad_norm": 2.290910005569458,
      "learning_rate": 0.00010191951874516508,
      "loss": 1.1519,
      "step": 14215
    },
    {
      "epoch": 0.5091052339427364,
      "grad_norm": 2.0241501331329346,
      "learning_rate": 0.00010190792185639855,
      "loss": 1.575,
      "step": 14216
    },
    {
      "epoch": 0.5091410460723047,
      "grad_norm": 1.8566899299621582,
      "learning_rate": 0.00010189632494196335,
      "loss": 1.5069,
      "step": 14217
    },
    {
      "epoch": 0.509176858201873,
      "grad_norm": 1.780731201171875,
      "learning_rate": 0.00010188472800201558,
      "loss": 1.4822,
      "step": 14218
    },
    {
      "epoch": 0.5092126703314412,
      "grad_norm": 1.620386004447937,
      "learning_rate": 0.00010187313103671122,
      "loss": 1.4713,
      "step": 14219
    },
    {
      "epoch": 0.5092484824610095,
      "grad_norm": 1.4523818492889404,
      "learning_rate": 0.00010186153404620628,
      "loss": 1.557,
      "step": 14220
    },
    {
      "epoch": 0.5092842945905778,
      "grad_norm": 1.6923600435256958,
      "learning_rate": 0.00010184993703065682,
      "loss": 1.4402,
      "step": 14221
    },
    {
      "epoch": 0.5093201067201462,
      "grad_norm": 1.507537841796875,
      "learning_rate": 0.00010183833999021884,
      "loss": 1.3889,
      "step": 14222
    },
    {
      "epoch": 0.5093559188497144,
      "grad_norm": 1.9163919687271118,
      "learning_rate": 0.00010182674292504837,
      "loss": 1.5374,
      "step": 14223
    },
    {
      "epoch": 0.5093917309792827,
      "grad_norm": 2.0340425968170166,
      "learning_rate": 0.00010181514583530141,
      "loss": 1.4581,
      "step": 14224
    },
    {
      "epoch": 0.509427543108851,
      "grad_norm": 2.15578293800354,
      "learning_rate": 0.00010180354872113403,
      "loss": 1.6459,
      "step": 14225
    },
    {
      "epoch": 0.5094633552384192,
      "grad_norm": 1.6318968534469604,
      "learning_rate": 0.0001017919515827022,
      "loss": 1.5743,
      "step": 14226
    },
    {
      "epoch": 0.5094991673679875,
      "grad_norm": 1.603484034538269,
      "learning_rate": 0.000101780354420162,
      "loss": 1.4433,
      "step": 14227
    },
    {
      "epoch": 0.5095349794975558,
      "grad_norm": 1.2180871963500977,
      "learning_rate": 0.00010176875723366941,
      "loss": 1.3916,
      "step": 14228
    },
    {
      "epoch": 0.5095707916271242,
      "grad_norm": 1.6632503271102905,
      "learning_rate": 0.00010175716002338049,
      "loss": 1.3513,
      "step": 14229
    },
    {
      "epoch": 0.5096066037566924,
      "grad_norm": 1.5550673007965088,
      "learning_rate": 0.00010174556278945123,
      "loss": 1.3926,
      "step": 14230
    },
    {
      "epoch": 0.5096424158862607,
      "grad_norm": 1.7034988403320312,
      "learning_rate": 0.00010173396553203771,
      "loss": 1.4134,
      "step": 14231
    },
    {
      "epoch": 0.509678228015829,
      "grad_norm": 2.4369795322418213,
      "learning_rate": 0.00010172236825129588,
      "loss": 1.4433,
      "step": 14232
    },
    {
      "epoch": 0.5097140401453972,
      "grad_norm": 1.695494532585144,
      "learning_rate": 0.00010171077094738183,
      "loss": 1.6308,
      "step": 14233
    },
    {
      "epoch": 0.5097498522749655,
      "grad_norm": 1.9752106666564941,
      "learning_rate": 0.00010169917362045154,
      "loss": 1.4111,
      "step": 14234
    },
    {
      "epoch": 0.5097856644045338,
      "grad_norm": 1.5427497625350952,
      "learning_rate": 0.00010168757627066105,
      "loss": 1.403,
      "step": 14235
    },
    {
      "epoch": 0.5098214765341021,
      "grad_norm": 1.7571707963943481,
      "learning_rate": 0.00010167597889816644,
      "loss": 1.5686,
      "step": 14236
    },
    {
      "epoch": 0.5098572886636704,
      "grad_norm": 1.598278522491455,
      "learning_rate": 0.00010166438150312367,
      "loss": 1.6431,
      "step": 14237
    },
    {
      "epoch": 0.5098931007932387,
      "grad_norm": 1.8703125715255737,
      "learning_rate": 0.00010165278408568881,
      "loss": 1.3825,
      "step": 14238
    },
    {
      "epoch": 0.509928912922807,
      "grad_norm": 1.4079030752182007,
      "learning_rate": 0.00010164118664601785,
      "loss": 1.5527,
      "step": 14239
    },
    {
      "epoch": 0.5099647250523752,
      "grad_norm": 1.8269530534744263,
      "learning_rate": 0.0001016295891842669,
      "loss": 1.6979,
      "step": 14240
    },
    {
      "epoch": 0.5100005371819435,
      "grad_norm": 1.5141544342041016,
      "learning_rate": 0.00010161799170059187,
      "loss": 1.4384,
      "step": 14241
    },
    {
      "epoch": 0.5100363493115118,
      "grad_norm": 1.7513827085494995,
      "learning_rate": 0.00010160639419514888,
      "loss": 1.094,
      "step": 14242
    },
    {
      "epoch": 0.5100721614410801,
      "grad_norm": 1.3838261365890503,
      "learning_rate": 0.00010159479666809388,
      "loss": 1.3501,
      "step": 14243
    },
    {
      "epoch": 0.5101079735706484,
      "grad_norm": 1.4064701795578003,
      "learning_rate": 0.00010158319911958301,
      "loss": 1.5364,
      "step": 14244
    },
    {
      "epoch": 0.5101437857002167,
      "grad_norm": 1.7723150253295898,
      "learning_rate": 0.00010157160154977219,
      "loss": 1.2483,
      "step": 14245
    },
    {
      "epoch": 0.510179597829785,
      "grad_norm": 1.6301299333572388,
      "learning_rate": 0.00010156000395881752,
      "loss": 1.6533,
      "step": 14246
    },
    {
      "epoch": 0.5102154099593532,
      "grad_norm": 1.759650707244873,
      "learning_rate": 0.000101548406346875,
      "loss": 1.5664,
      "step": 14247
    },
    {
      "epoch": 0.5102512220889215,
      "grad_norm": 1.7729034423828125,
      "learning_rate": 0.00010153680871410065,
      "loss": 1.5183,
      "step": 14248
    },
    {
      "epoch": 0.5102870342184898,
      "grad_norm": 1.9804837703704834,
      "learning_rate": 0.00010152521106065058,
      "loss": 1.4567,
      "step": 14249
    },
    {
      "epoch": 0.5103228463480581,
      "grad_norm": 2.063938856124878,
      "learning_rate": 0.00010151361338668072,
      "loss": 1.6644,
      "step": 14250
    },
    {
      "epoch": 0.5103586584776264,
      "grad_norm": 1.628487467765808,
      "learning_rate": 0.00010150201569234717,
      "loss": 1.1544,
      "step": 14251
    },
    {
      "epoch": 0.5103944706071947,
      "grad_norm": 2.041860580444336,
      "learning_rate": 0.0001014904179778059,
      "loss": 1.6834,
      "step": 14252
    },
    {
      "epoch": 0.5104302827367629,
      "grad_norm": 1.646464467048645,
      "learning_rate": 0.000101478820243213,
      "loss": 1.7831,
      "step": 14253
    },
    {
      "epoch": 0.5104660948663312,
      "grad_norm": 1.8969465494155884,
      "learning_rate": 0.00010146722248872446,
      "loss": 1.4377,
      "step": 14254
    },
    {
      "epoch": 0.5105019069958995,
      "grad_norm": 1.9447615146636963,
      "learning_rate": 0.00010145562471449638,
      "loss": 1.4619,
      "step": 14255
    },
    {
      "epoch": 0.5105377191254677,
      "grad_norm": 1.6894398927688599,
      "learning_rate": 0.00010144402692068472,
      "loss": 1.6591,
      "step": 14256
    },
    {
      "epoch": 0.5105735312550361,
      "grad_norm": 2.469947576522827,
      "learning_rate": 0.00010143242910744555,
      "loss": 1.587,
      "step": 14257
    },
    {
      "epoch": 0.5106093433846044,
      "grad_norm": 1.7453415393829346,
      "learning_rate": 0.00010142083127493489,
      "loss": 1.5956,
      "step": 14258
    },
    {
      "epoch": 0.5106451555141727,
      "grad_norm": 1.6180731058120728,
      "learning_rate": 0.00010140923342330875,
      "loss": 1.5102,
      "step": 14259
    },
    {
      "epoch": 0.5106809676437409,
      "grad_norm": 2.149956226348877,
      "learning_rate": 0.00010139763555272323,
      "loss": 1.8065,
      "step": 14260
    },
    {
      "epoch": 0.5107167797733092,
      "grad_norm": 1.7655017375946045,
      "learning_rate": 0.0001013860376633343,
      "loss": 1.1825,
      "step": 14261
    },
    {
      "epoch": 0.5107525919028775,
      "grad_norm": 1.5514721870422363,
      "learning_rate": 0.00010137443975529804,
      "loss": 1.3279,
      "step": 14262
    },
    {
      "epoch": 0.5107884040324457,
      "grad_norm": 1.7187747955322266,
      "learning_rate": 0.00010136284182877045,
      "loss": 1.4676,
      "step": 14263
    },
    {
      "epoch": 0.5108242161620141,
      "grad_norm": 1.7656067609786987,
      "learning_rate": 0.0001013512438839076,
      "loss": 1.5519,
      "step": 14264
    },
    {
      "epoch": 0.5108600282915824,
      "grad_norm": 1.3366622924804688,
      "learning_rate": 0.00010133964592086547,
      "loss": 1.3903,
      "step": 14265
    },
    {
      "epoch": 0.5108958404211507,
      "grad_norm": 2.655071973800659,
      "learning_rate": 0.00010132804793980018,
      "loss": 1.3475,
      "step": 14266
    },
    {
      "epoch": 0.5109316525507189,
      "grad_norm": 1.4642434120178223,
      "learning_rate": 0.0001013164499408677,
      "loss": 1.736,
      "step": 14267
    },
    {
      "epoch": 0.5109674646802872,
      "grad_norm": 1.510628342628479,
      "learning_rate": 0.00010130485192422408,
      "loss": 1.3315,
      "step": 14268
    },
    {
      "epoch": 0.5110032768098555,
      "grad_norm": 1.7870886325836182,
      "learning_rate": 0.00010129325389002536,
      "loss": 1.4371,
      "step": 14269
    },
    {
      "epoch": 0.5110390889394237,
      "grad_norm": 1.452154517173767,
      "learning_rate": 0.00010128165583842757,
      "loss": 1.5644,
      "step": 14270
    },
    {
      "epoch": 0.5110749010689921,
      "grad_norm": 1.9587507247924805,
      "learning_rate": 0.00010127005776958676,
      "loss": 1.5341,
      "step": 14271
    },
    {
      "epoch": 0.5111107131985604,
      "grad_norm": 1.3689603805541992,
      "learning_rate": 0.00010125845968365895,
      "loss": 1.3541,
      "step": 14272
    },
    {
      "epoch": 0.5111465253281287,
      "grad_norm": 2.153557300567627,
      "learning_rate": 0.00010124686158080021,
      "loss": 1.5126,
      "step": 14273
    },
    {
      "epoch": 0.5111823374576969,
      "grad_norm": 1.3762892484664917,
      "learning_rate": 0.00010123526346116654,
      "loss": 1.5685,
      "step": 14274
    },
    {
      "epoch": 0.5112181495872652,
      "grad_norm": 1.7966989278793335,
      "learning_rate": 0.00010122366532491403,
      "loss": 1.2602,
      "step": 14275
    },
    {
      "epoch": 0.5112539617168335,
      "grad_norm": 1.1151483058929443,
      "learning_rate": 0.00010121206717219865,
      "loss": 1.4299,
      "step": 14276
    },
    {
      "epoch": 0.5112897738464017,
      "grad_norm": 1.83100163936615,
      "learning_rate": 0.00010120046900317646,
      "loss": 1.7361,
      "step": 14277
    },
    {
      "epoch": 0.5113255859759701,
      "grad_norm": 1.368430495262146,
      "learning_rate": 0.00010118887081800352,
      "loss": 1.5495,
      "step": 14278
    },
    {
      "epoch": 0.5113613981055384,
      "grad_norm": 1.5631763935089111,
      "learning_rate": 0.00010117727261683585,
      "loss": 1.3579,
      "step": 14279
    },
    {
      "epoch": 0.5113972102351066,
      "grad_norm": 3.0901801586151123,
      "learning_rate": 0.00010116567439982952,
      "loss": 1.7898,
      "step": 14280
    },
    {
      "epoch": 0.5114330223646749,
      "grad_norm": 1.6391278505325317,
      "learning_rate": 0.0001011540761671405,
      "loss": 1.1463,
      "step": 14281
    },
    {
      "epoch": 0.5114688344942432,
      "grad_norm": 1.5354077816009521,
      "learning_rate": 0.00010114247791892491,
      "loss": 1.4781,
      "step": 14282
    },
    {
      "epoch": 0.5115046466238115,
      "grad_norm": 2.2333121299743652,
      "learning_rate": 0.00010113087965533874,
      "loss": 1.5416,
      "step": 14283
    },
    {
      "epoch": 0.5115404587533797,
      "grad_norm": 1.9875568151474,
      "learning_rate": 0.00010111928137653808,
      "loss": 1.677,
      "step": 14284
    },
    {
      "epoch": 0.5115762708829481,
      "grad_norm": 1.5071085691452026,
      "learning_rate": 0.00010110768308267889,
      "loss": 1.2818,
      "step": 14285
    },
    {
      "epoch": 0.5116120830125164,
      "grad_norm": 1.5551725625991821,
      "learning_rate": 0.00010109608477391725,
      "loss": 1.5007,
      "step": 14286
    },
    {
      "epoch": 0.5116478951420846,
      "grad_norm": 2.123044967651367,
      "learning_rate": 0.00010108448645040919,
      "loss": 1.5999,
      "step": 14287
    },
    {
      "epoch": 0.5116837072716529,
      "grad_norm": 1.681573748588562,
      "learning_rate": 0.00010107288811231081,
      "loss": 1.2607,
      "step": 14288
    },
    {
      "epoch": 0.5117195194012212,
      "grad_norm": 1.9159537553787231,
      "learning_rate": 0.00010106128975977809,
      "loss": 1.5472,
      "step": 14289
    },
    {
      "epoch": 0.5117553315307894,
      "grad_norm": 1.6600178480148315,
      "learning_rate": 0.00010104969139296705,
      "loss": 1.6619,
      "step": 14290
    },
    {
      "epoch": 0.5117911436603577,
      "grad_norm": 1.5200772285461426,
      "learning_rate": 0.00010103809301203382,
      "loss": 1.4235,
      "step": 14291
    },
    {
      "epoch": 0.5118269557899261,
      "grad_norm": 1.593790054321289,
      "learning_rate": 0.00010102649461713434,
      "loss": 1.496,
      "step": 14292
    },
    {
      "epoch": 0.5118627679194944,
      "grad_norm": 1.5583328008651733,
      "learning_rate": 0.00010101489620842475,
      "loss": 1.5407,
      "step": 14293
    },
    {
      "epoch": 0.5118985800490626,
      "grad_norm": 1.877834677696228,
      "learning_rate": 0.00010100329778606101,
      "loss": 1.7199,
      "step": 14294
    },
    {
      "epoch": 0.5119343921786309,
      "grad_norm": 1.555271029472351,
      "learning_rate": 0.0001009916993501992,
      "loss": 1.4016,
      "step": 14295
    },
    {
      "epoch": 0.5119702043081992,
      "grad_norm": 1.6726455688476562,
      "learning_rate": 0.00010098010090099532,
      "loss": 1.306,
      "step": 14296
    },
    {
      "epoch": 0.5120060164377674,
      "grad_norm": 1.5355541706085205,
      "learning_rate": 0.00010096850243860549,
      "loss": 1.3042,
      "step": 14297
    },
    {
      "epoch": 0.5120418285673357,
      "grad_norm": 1.3412790298461914,
      "learning_rate": 0.00010095690396318569,
      "loss": 1.529,
      "step": 14298
    },
    {
      "epoch": 0.5120776406969041,
      "grad_norm": 2.1176607608795166,
      "learning_rate": 0.00010094530547489201,
      "loss": 1.731,
      "step": 14299
    },
    {
      "epoch": 0.5121134528264724,
      "grad_norm": 1.660214900970459,
      "learning_rate": 0.0001009337069738804,
      "loss": 1.5807,
      "step": 14300
    },
    {
      "epoch": 0.5121492649560406,
      "grad_norm": 1.5404260158538818,
      "learning_rate": 0.00010092210846030703,
      "loss": 1.3894,
      "step": 14301
    },
    {
      "epoch": 0.5121850770856089,
      "grad_norm": 1.514736294746399,
      "learning_rate": 0.00010091050993432787,
      "loss": 1.4389,
      "step": 14302
    },
    {
      "epoch": 0.5122208892151772,
      "grad_norm": 2.2876694202423096,
      "learning_rate": 0.00010089891139609895,
      "loss": 1.441,
      "step": 14303
    },
    {
      "epoch": 0.5122567013447454,
      "grad_norm": 1.7587552070617676,
      "learning_rate": 0.00010088731284577636,
      "loss": 1.4818,
      "step": 14304
    },
    {
      "epoch": 0.5122925134743137,
      "grad_norm": 1.566940426826477,
      "learning_rate": 0.0001008757142835161,
      "loss": 1.3247,
      "step": 14305
    },
    {
      "epoch": 0.5123283256038821,
      "grad_norm": 1.3367424011230469,
      "learning_rate": 0.00010086411570947424,
      "loss": 1.6241,
      "step": 14306
    },
    {
      "epoch": 0.5123641377334504,
      "grad_norm": 1.3539397716522217,
      "learning_rate": 0.0001008525171238068,
      "loss": 1.5768,
      "step": 14307
    },
    {
      "epoch": 0.5123999498630186,
      "grad_norm": 1.3885242938995361,
      "learning_rate": 0.00010084091852666988,
      "loss": 1.1827,
      "step": 14308
    },
    {
      "epoch": 0.5124357619925869,
      "grad_norm": 2.392909288406372,
      "learning_rate": 0.00010082931991821945,
      "loss": 1.4581,
      "step": 14309
    },
    {
      "epoch": 0.5124715741221552,
      "grad_norm": 1.3797001838684082,
      "learning_rate": 0.00010081772129861163,
      "loss": 1.6449,
      "step": 14310
    },
    {
      "epoch": 0.5125073862517234,
      "grad_norm": 1.8504879474639893,
      "learning_rate": 0.00010080612266800241,
      "loss": 1.502,
      "step": 14311
    },
    {
      "epoch": 0.5125431983812917,
      "grad_norm": 1.6124776601791382,
      "learning_rate": 0.0001007945240265478,
      "loss": 1.6722,
      "step": 14312
    },
    {
      "epoch": 0.5125790105108601,
      "grad_norm": 1.4799977540969849,
      "learning_rate": 0.00010078292537440397,
      "loss": 2.0444,
      "step": 14313
    },
    {
      "epoch": 0.5126148226404283,
      "grad_norm": 1.6208722591400146,
      "learning_rate": 0.00010077132671172685,
      "loss": 1.4614,
      "step": 14314
    },
    {
      "epoch": 0.5126506347699966,
      "grad_norm": 1.8112127780914307,
      "learning_rate": 0.00010075972803867254,
      "loss": 1.4982,
      "step": 14315
    },
    {
      "epoch": 0.5126864468995649,
      "grad_norm": 1.5253667831420898,
      "learning_rate": 0.00010074812935539703,
      "loss": 1.5199,
      "step": 14316
    },
    {
      "epoch": 0.5127222590291332,
      "grad_norm": 1.505617618560791,
      "learning_rate": 0.00010073653066205644,
      "loss": 1.5824,
      "step": 14317
    },
    {
      "epoch": 0.5127580711587014,
      "grad_norm": 1.8238226175308228,
      "learning_rate": 0.00010072493195880676,
      "loss": 1.239,
      "step": 14318
    },
    {
      "epoch": 0.5127938832882697,
      "grad_norm": 1.3497518301010132,
      "learning_rate": 0.00010071333324580408,
      "loss": 1.4374,
      "step": 14319
    },
    {
      "epoch": 0.5128296954178381,
      "grad_norm": 1.3062442541122437,
      "learning_rate": 0.00010070173452320442,
      "loss": 1.2157,
      "step": 14320
    },
    {
      "epoch": 0.5128655075474063,
      "grad_norm": 1.9949729442596436,
      "learning_rate": 0.0001006901357911638,
      "loss": 1.4122,
      "step": 14321
    },
    {
      "epoch": 0.5129013196769746,
      "grad_norm": 1.8121817111968994,
      "learning_rate": 0.00010067853704983832,
      "loss": 1.277,
      "step": 14322
    },
    {
      "epoch": 0.5129371318065429,
      "grad_norm": 1.786101222038269,
      "learning_rate": 0.00010066693829938398,
      "loss": 1.7341,
      "step": 14323
    },
    {
      "epoch": 0.5129729439361111,
      "grad_norm": 1.6490147113800049,
      "learning_rate": 0.00010065533953995688,
      "loss": 1.5394,
      "step": 14324
    },
    {
      "epoch": 0.5130087560656794,
      "grad_norm": 2.683562755584717,
      "learning_rate": 0.00010064374077171296,
      "loss": 1.633,
      "step": 14325
    },
    {
      "epoch": 0.5130445681952477,
      "grad_norm": 1.9112385511398315,
      "learning_rate": 0.00010063214199480842,
      "loss": 1.6243,
      "step": 14326
    },
    {
      "epoch": 0.5130803803248161,
      "grad_norm": 1.7935097217559814,
      "learning_rate": 0.00010062054320939916,
      "loss": 1.1181,
      "step": 14327
    },
    {
      "epoch": 0.5131161924543843,
      "grad_norm": 1.3657851219177246,
      "learning_rate": 0.00010060894441564135,
      "loss": 1.6137,
      "step": 14328
    },
    {
      "epoch": 0.5131520045839526,
      "grad_norm": 1.3997024297714233,
      "learning_rate": 0.00010059734561369095,
      "loss": 1.4953,
      "step": 14329
    },
    {
      "epoch": 0.5131878167135209,
      "grad_norm": 1.8069899082183838,
      "learning_rate": 0.00010058574680370403,
      "loss": 1.5614,
      "step": 14330
    },
    {
      "epoch": 0.5132236288430891,
      "grad_norm": 1.4796655178070068,
      "learning_rate": 0.00010057414798583664,
      "loss": 1.5437,
      "step": 14331
    },
    {
      "epoch": 0.5132594409726574,
      "grad_norm": 1.5718828439712524,
      "learning_rate": 0.00010056254916024483,
      "loss": 1.4086,
      "step": 14332
    },
    {
      "epoch": 0.5132952531022257,
      "grad_norm": 2.2739932537078857,
      "learning_rate": 0.00010055095032708466,
      "loss": 1.4159,
      "step": 14333
    },
    {
      "epoch": 0.5133310652317941,
      "grad_norm": 1.714881181716919,
      "learning_rate": 0.00010053935148651214,
      "loss": 1.4963,
      "step": 14334
    },
    {
      "epoch": 0.5133668773613623,
      "grad_norm": 1.6868878602981567,
      "learning_rate": 0.00010052775263868337,
      "loss": 1.2638,
      "step": 14335
    },
    {
      "epoch": 0.5134026894909306,
      "grad_norm": 1.6076358556747437,
      "learning_rate": 0.00010051615378375434,
      "loss": 1.7306,
      "step": 14336
    },
    {
      "epoch": 0.5134385016204989,
      "grad_norm": 1.4571819305419922,
      "learning_rate": 0.00010050455492188118,
      "loss": 1.7523,
      "step": 14337
    },
    {
      "epoch": 0.5134743137500671,
      "grad_norm": 1.9559829235076904,
      "learning_rate": 0.00010049295605321984,
      "loss": 1.1767,
      "step": 14338
    },
    {
      "epoch": 0.5135101258796354,
      "grad_norm": 1.8952054977416992,
      "learning_rate": 0.00010048135717792641,
      "loss": 1.3615,
      "step": 14339
    },
    {
      "epoch": 0.5135459380092037,
      "grad_norm": 3.2473466396331787,
      "learning_rate": 0.00010046975829615695,
      "loss": 1.7416,
      "step": 14340
    },
    {
      "epoch": 0.513581750138772,
      "grad_norm": 1.5299546718597412,
      "learning_rate": 0.00010045815940806751,
      "loss": 1.5168,
      "step": 14341
    },
    {
      "epoch": 0.5136175622683403,
      "grad_norm": 1.6583263874053955,
      "learning_rate": 0.00010044656051381411,
      "loss": 1.4021,
      "step": 14342
    },
    {
      "epoch": 0.5136533743979086,
      "grad_norm": 1.6013065576553345,
      "learning_rate": 0.00010043496161355282,
      "loss": 1.6424,
      "step": 14343
    },
    {
      "epoch": 0.5136891865274769,
      "grad_norm": 1.4253227710723877,
      "learning_rate": 0.00010042336270743968,
      "loss": 1.6771,
      "step": 14344
    },
    {
      "epoch": 0.5137249986570451,
      "grad_norm": 2.0314788818359375,
      "learning_rate": 0.00010041176379563073,
      "loss": 1.6002,
      "step": 14345
    },
    {
      "epoch": 0.5137608107866134,
      "grad_norm": 1.651528239250183,
      "learning_rate": 0.00010040016487828208,
      "loss": 1.6224,
      "step": 14346
    },
    {
      "epoch": 0.5137966229161817,
      "grad_norm": 1.8376193046569824,
      "learning_rate": 0.00010038856595554967,
      "loss": 1.4108,
      "step": 14347
    },
    {
      "epoch": 0.51383243504575,
      "grad_norm": 2.1696889400482178,
      "learning_rate": 0.00010037696702758963,
      "loss": 1.3069,
      "step": 14348
    },
    {
      "epoch": 0.5138682471753183,
      "grad_norm": 1.4069865942001343,
      "learning_rate": 0.00010036536809455796,
      "loss": 1.3905,
      "step": 14349
    },
    {
      "epoch": 0.5139040593048866,
      "grad_norm": 2.0680787563323975,
      "learning_rate": 0.00010035376915661076,
      "loss": 1.7688,
      "step": 14350
    },
    {
      "epoch": 0.5139398714344549,
      "grad_norm": 2.34431529045105,
      "learning_rate": 0.00010034217021390404,
      "loss": 1.3928,
      "step": 14351
    },
    {
      "epoch": 0.5139756835640231,
      "grad_norm": 1.8667163848876953,
      "learning_rate": 0.00010033057126659388,
      "loss": 1.6729,
      "step": 14352
    },
    {
      "epoch": 0.5140114956935914,
      "grad_norm": 1.555355429649353,
      "learning_rate": 0.0001003189723148363,
      "loss": 1.4538,
      "step": 14353
    },
    {
      "epoch": 0.5140473078231597,
      "grad_norm": 1.3511971235275269,
      "learning_rate": 0.00010030737335878735,
      "loss": 1.7526,
      "step": 14354
    },
    {
      "epoch": 0.514083119952728,
      "grad_norm": 1.7594633102416992,
      "learning_rate": 0.00010029577439860312,
      "loss": 1.3122,
      "step": 14355
    },
    {
      "epoch": 0.5141189320822963,
      "grad_norm": 1.5887043476104736,
      "learning_rate": 0.00010028417543443958,
      "loss": 1.536,
      "step": 14356
    },
    {
      "epoch": 0.5141547442118646,
      "grad_norm": 1.7294690608978271,
      "learning_rate": 0.00010027257646645285,
      "loss": 1.6294,
      "step": 14357
    },
    {
      "epoch": 0.5141905563414328,
      "grad_norm": 1.778171181678772,
      "learning_rate": 0.00010026097749479895,
      "loss": 1.3429,
      "step": 14358
    },
    {
      "epoch": 0.5142263684710011,
      "grad_norm": 2.0442750453948975,
      "learning_rate": 0.00010024937851963394,
      "loss": 1.1833,
      "step": 14359
    },
    {
      "epoch": 0.5142621806005694,
      "grad_norm": 1.6226485967636108,
      "learning_rate": 0.00010023777954111384,
      "loss": 1.1286,
      "step": 14360
    },
    {
      "epoch": 0.5142979927301377,
      "grad_norm": 1.7150418758392334,
      "learning_rate": 0.00010022618055939477,
      "loss": 1.7438,
      "step": 14361
    },
    {
      "epoch": 0.514333804859706,
      "grad_norm": 1.712880253791809,
      "learning_rate": 0.00010021458157463268,
      "loss": 1.5001,
      "step": 14362
    },
    {
      "epoch": 0.5143696169892743,
      "grad_norm": 2.2642722129821777,
      "learning_rate": 0.0001002029825869837,
      "loss": 1.6687,
      "step": 14363
    },
    {
      "epoch": 0.5144054291188426,
      "grad_norm": 1.4798762798309326,
      "learning_rate": 0.00010019138359660387,
      "loss": 1.6926,
      "step": 14364
    },
    {
      "epoch": 0.5144412412484108,
      "grad_norm": 2.6592345237731934,
      "learning_rate": 0.00010017978460364919,
      "loss": 1.6522,
      "step": 14365
    },
    {
      "epoch": 0.5144770533779791,
      "grad_norm": 1.899743914604187,
      "learning_rate": 0.00010016818560827577,
      "loss": 1.4547,
      "step": 14366
    },
    {
      "epoch": 0.5145128655075474,
      "grad_norm": 1.760870099067688,
      "learning_rate": 0.00010015658661063957,
      "loss": 1.5189,
      "step": 14367
    },
    {
      "epoch": 0.5145486776371156,
      "grad_norm": 1.7101985216140747,
      "learning_rate": 0.00010014498761089677,
      "loss": 1.691,
      "step": 14368
    },
    {
      "epoch": 0.514584489766684,
      "grad_norm": 1.8395957946777344,
      "learning_rate": 0.0001001333886092033,
      "loss": 1.4385,
      "step": 14369
    },
    {
      "epoch": 0.5146203018962523,
      "grad_norm": 1.5016250610351562,
      "learning_rate": 0.00010012178960571527,
      "loss": 1.7028,
      "step": 14370
    },
    {
      "epoch": 0.5146561140258206,
      "grad_norm": 1.3344529867172241,
      "learning_rate": 0.00010011019060058873,
      "loss": 1.5635,
      "step": 14371
    },
    {
      "epoch": 0.5146919261553888,
      "grad_norm": 1.4595482349395752,
      "learning_rate": 0.00010009859159397974,
      "loss": 1.5788,
      "step": 14372
    },
    {
      "epoch": 0.5147277382849571,
      "grad_norm": 1.6308788061141968,
      "learning_rate": 0.00010008699258604429,
      "loss": 1.2402,
      "step": 14373
    },
    {
      "epoch": 0.5147635504145254,
      "grad_norm": 1.6189227104187012,
      "learning_rate": 0.00010007539357693845,
      "loss": 1.4277,
      "step": 14374
    },
    {
      "epoch": 0.5147993625440936,
      "grad_norm": 1.546007513999939,
      "learning_rate": 0.00010006379456681834,
      "loss": 1.3149,
      "step": 14375
    },
    {
      "epoch": 0.514835174673662,
      "grad_norm": 1.400831699371338,
      "learning_rate": 0.00010005219555583991,
      "loss": 1.7555,
      "step": 14376
    },
    {
      "epoch": 0.5148709868032303,
      "grad_norm": 1.3352601528167725,
      "learning_rate": 0.00010004059654415927,
      "loss": 1.6648,
      "step": 14377
    },
    {
      "epoch": 0.5149067989327986,
      "grad_norm": 1.8999031782150269,
      "learning_rate": 0.00010002899753193246,
      "loss": 1.5502,
      "step": 14378
    },
    {
      "epoch": 0.5149426110623668,
      "grad_norm": 2.032066822052002,
      "learning_rate": 0.00010001739851931553,
      "loss": 1.6505,
      "step": 14379
    },
    {
      "epoch": 0.5149784231919351,
      "grad_norm": 1.6035372018814087,
      "learning_rate": 0.00010000579950646452,
      "loss": 1.7394,
      "step": 14380
    },
    {
      "epoch": 0.5150142353215034,
      "grad_norm": 1.6789970397949219,
      "learning_rate": 9.999420049353549e-05,
      "loss": 1.5343,
      "step": 14381
    },
    {
      "epoch": 0.5150500474510716,
      "grad_norm": 2.004620313644409,
      "learning_rate": 9.998260148068449e-05,
      "loss": 1.3025,
      "step": 14382
    },
    {
      "epoch": 0.51508585958064,
      "grad_norm": 2.159876823425293,
      "learning_rate": 9.997100246806755e-05,
      "loss": 1.7772,
      "step": 14383
    },
    {
      "epoch": 0.5151216717102083,
      "grad_norm": 1.748475193977356,
      "learning_rate": 9.995940345584074e-05,
      "loss": 1.7536,
      "step": 14384
    },
    {
      "epoch": 0.5151574838397766,
      "grad_norm": 1.7124178409576416,
      "learning_rate": 9.994780444416013e-05,
      "loss": 1.5581,
      "step": 14385
    },
    {
      "epoch": 0.5151932959693448,
      "grad_norm": 1.6359690427780151,
      "learning_rate": 9.99362054331817e-05,
      "loss": 1.7249,
      "step": 14386
    },
    {
      "epoch": 0.5152291080989131,
      "grad_norm": 1.629350185394287,
      "learning_rate": 9.992460642306156e-05,
      "loss": 1.2294,
      "step": 14387
    },
    {
      "epoch": 0.5152649202284814,
      "grad_norm": 1.9517797231674194,
      "learning_rate": 9.991300741395574e-05,
      "loss": 1.2735,
      "step": 14388
    },
    {
      "epoch": 0.5153007323580496,
      "grad_norm": 1.8916865587234497,
      "learning_rate": 9.99014084060203e-05,
      "loss": 1.5717,
      "step": 14389
    },
    {
      "epoch": 0.515336544487618,
      "grad_norm": 1.816402554512024,
      "learning_rate": 9.988980939941127e-05,
      "loss": 1.5327,
      "step": 14390
    },
    {
      "epoch": 0.5153723566171863,
      "grad_norm": 1.7927991151809692,
      "learning_rate": 9.987821039428474e-05,
      "loss": 1.1213,
      "step": 14391
    },
    {
      "epoch": 0.5154081687467545,
      "grad_norm": 1.3315367698669434,
      "learning_rate": 9.986661139079671e-05,
      "loss": 1.1219,
      "step": 14392
    },
    {
      "epoch": 0.5154439808763228,
      "grad_norm": 1.9263297319412231,
      "learning_rate": 9.985501238910325e-05,
      "loss": 1.585,
      "step": 14393
    },
    {
      "epoch": 0.5154797930058911,
      "grad_norm": 1.5922685861587524,
      "learning_rate": 9.984341338936043e-05,
      "loss": 1.5042,
      "step": 14394
    },
    {
      "epoch": 0.5155156051354594,
      "grad_norm": 2.0485384464263916,
      "learning_rate": 9.983181439172426e-05,
      "loss": 1.5463,
      "step": 14395
    },
    {
      "epoch": 0.5155514172650276,
      "grad_norm": 1.7884843349456787,
      "learning_rate": 9.982021539635084e-05,
      "loss": 1.5589,
      "step": 14396
    },
    {
      "epoch": 0.515587229394596,
      "grad_norm": 1.4187160730361938,
      "learning_rate": 9.980861640339614e-05,
      "loss": 1.1647,
      "step": 14397
    },
    {
      "epoch": 0.5156230415241643,
      "grad_norm": 1.627793312072754,
      "learning_rate": 9.979701741301631e-05,
      "loss": 1.2982,
      "step": 14398
    },
    {
      "epoch": 0.5156588536537325,
      "grad_norm": 1.3230700492858887,
      "learning_rate": 9.978541842536732e-05,
      "loss": 1.4045,
      "step": 14399
    },
    {
      "epoch": 0.5156946657833008,
      "grad_norm": 1.5387710332870483,
      "learning_rate": 9.977381944060525e-05,
      "loss": 1.4987,
      "step": 14400
    },
    {
      "epoch": 0.5157304779128691,
      "grad_norm": 1.577405333518982,
      "learning_rate": 9.976222045888614e-05,
      "loss": 1.5148,
      "step": 14401
    },
    {
      "epoch": 0.5157662900424373,
      "grad_norm": 1.404995322227478,
      "learning_rate": 9.975062148036608e-05,
      "loss": 1.537,
      "step": 14402
    },
    {
      "epoch": 0.5158021021720056,
      "grad_norm": 1.305178165435791,
      "learning_rate": 9.97390225052011e-05,
      "loss": 1.6459,
      "step": 14403
    },
    {
      "epoch": 0.515837914301574,
      "grad_norm": 2.065877676010132,
      "learning_rate": 9.972742353354717e-05,
      "loss": 1.6797,
      "step": 14404
    },
    {
      "epoch": 0.5158737264311423,
      "grad_norm": 1.7629402875900269,
      "learning_rate": 9.971582456556045e-05,
      "loss": 1.4101,
      "step": 14405
    },
    {
      "epoch": 0.5159095385607105,
      "grad_norm": 2.007244348526001,
      "learning_rate": 9.970422560139692e-05,
      "loss": 1.5678,
      "step": 14406
    },
    {
      "epoch": 0.5159453506902788,
      "grad_norm": 1.5562615394592285,
      "learning_rate": 9.969262664121267e-05,
      "loss": 1.6061,
      "step": 14407
    },
    {
      "epoch": 0.5159811628198471,
      "grad_norm": 1.781718134880066,
      "learning_rate": 9.968102768516371e-05,
      "loss": 1.6981,
      "step": 14408
    },
    {
      "epoch": 0.5160169749494153,
      "grad_norm": 1.2723973989486694,
      "learning_rate": 9.966942873340614e-05,
      "loss": 1.3363,
      "step": 14409
    },
    {
      "epoch": 0.5160527870789836,
      "grad_norm": 1.9770539999008179,
      "learning_rate": 9.965782978609595e-05,
      "loss": 1.2804,
      "step": 14410
    },
    {
      "epoch": 0.516088599208552,
      "grad_norm": 1.5593479871749878,
      "learning_rate": 9.964623084338926e-05,
      "loss": 1.5882,
      "step": 14411
    },
    {
      "epoch": 0.5161244113381203,
      "grad_norm": 1.7799978256225586,
      "learning_rate": 9.963463190544208e-05,
      "loss": 1.4457,
      "step": 14412
    },
    {
      "epoch": 0.5161602234676885,
      "grad_norm": 1.773402214050293,
      "learning_rate": 9.96230329724104e-05,
      "loss": 1.6663,
      "step": 14413
    },
    {
      "epoch": 0.5161960355972568,
      "grad_norm": 1.4051496982574463,
      "learning_rate": 9.961143404445038e-05,
      "loss": 1.439,
      "step": 14414
    },
    {
      "epoch": 0.5162318477268251,
      "grad_norm": 1.4480347633361816,
      "learning_rate": 9.959983512171796e-05,
      "loss": 1.4239,
      "step": 14415
    },
    {
      "epoch": 0.5162676598563933,
      "grad_norm": 1.744836449623108,
      "learning_rate": 9.95882362043693e-05,
      "loss": 1.5925,
      "step": 14416
    },
    {
      "epoch": 0.5163034719859616,
      "grad_norm": 2.2484891414642334,
      "learning_rate": 9.957663729256033e-05,
      "loss": 1.4217,
      "step": 14417
    },
    {
      "epoch": 0.51633928411553,
      "grad_norm": 1.41450834274292,
      "learning_rate": 9.956503838644719e-05,
      "loss": 1.4532,
      "step": 14418
    },
    {
      "epoch": 0.5163750962450983,
      "grad_norm": 1.5045841932296753,
      "learning_rate": 9.95534394861859e-05,
      "loss": 1.5254,
      "step": 14419
    },
    {
      "epoch": 0.5164109083746665,
      "grad_norm": 1.430303931236267,
      "learning_rate": 9.954184059193251e-05,
      "loss": 1.5091,
      "step": 14420
    },
    {
      "epoch": 0.5164467205042348,
      "grad_norm": 2.064718008041382,
      "learning_rate": 9.953024170384309e-05,
      "loss": 1.3218,
      "step": 14421
    },
    {
      "epoch": 0.5164825326338031,
      "grad_norm": 1.8702689409255981,
      "learning_rate": 9.95186428220736e-05,
      "loss": 1.3991,
      "step": 14422
    },
    {
      "epoch": 0.5165183447633713,
      "grad_norm": 1.897904634475708,
      "learning_rate": 9.950704394678021e-05,
      "loss": 1.3649,
      "step": 14423
    },
    {
      "epoch": 0.5165541568929396,
      "grad_norm": 1.8715026378631592,
      "learning_rate": 9.949544507811885e-05,
      "loss": 1.4766,
      "step": 14424
    },
    {
      "epoch": 0.516589969022508,
      "grad_norm": 1.6160260438919067,
      "learning_rate": 9.948384621624569e-05,
      "loss": 1.7052,
      "step": 14425
    },
    {
      "epoch": 0.5166257811520762,
      "grad_norm": 1.8741852045059204,
      "learning_rate": 9.947224736131662e-05,
      "loss": 1.4936,
      "step": 14426
    },
    {
      "epoch": 0.5166615932816445,
      "grad_norm": 2.4261343479156494,
      "learning_rate": 9.946064851348788e-05,
      "loss": 1.4065,
      "step": 14427
    },
    {
      "epoch": 0.5166974054112128,
      "grad_norm": 1.7130334377288818,
      "learning_rate": 9.944904967291533e-05,
      "loss": 1.1564,
      "step": 14428
    },
    {
      "epoch": 0.516733217540781,
      "grad_norm": 1.6435831785202026,
      "learning_rate": 9.94374508397552e-05,
      "loss": 1.2979,
      "step": 14429
    },
    {
      "epoch": 0.5167690296703493,
      "grad_norm": 1.4821659326553345,
      "learning_rate": 9.94258520141634e-05,
      "loss": 1.3349,
      "step": 14430
    },
    {
      "epoch": 0.5168048417999176,
      "grad_norm": 1.582342267036438,
      "learning_rate": 9.941425319629598e-05,
      "loss": 1.6496,
      "step": 14431
    },
    {
      "epoch": 0.516840653929486,
      "grad_norm": 1.3501367568969727,
      "learning_rate": 9.94026543863091e-05,
      "loss": 1.2989,
      "step": 14432
    },
    {
      "epoch": 0.5168764660590542,
      "grad_norm": 1.5915173292160034,
      "learning_rate": 9.939105558435866e-05,
      "loss": 1.7494,
      "step": 14433
    },
    {
      "epoch": 0.5169122781886225,
      "grad_norm": 1.774556040763855,
      "learning_rate": 9.937945679060085e-05,
      "loss": 1.3733,
      "step": 14434
    },
    {
      "epoch": 0.5169480903181908,
      "grad_norm": 2.2483110427856445,
      "learning_rate": 9.93678580051916e-05,
      "loss": 1.4633,
      "step": 14435
    },
    {
      "epoch": 0.516983902447759,
      "grad_norm": 1.5181875228881836,
      "learning_rate": 9.935625922828705e-05,
      "loss": 1.6502,
      "step": 14436
    },
    {
      "epoch": 0.5170197145773273,
      "grad_norm": 1.6699066162109375,
      "learning_rate": 9.934466046004313e-05,
      "loss": 1.4195,
      "step": 14437
    },
    {
      "epoch": 0.5170555267068956,
      "grad_norm": 1.4470813274383545,
      "learning_rate": 9.933306170061604e-05,
      "loss": 1.4196,
      "step": 14438
    },
    {
      "epoch": 0.517091338836464,
      "grad_norm": 1.7814797163009644,
      "learning_rate": 9.932146295016172e-05,
      "loss": 1.5409,
      "step": 14439
    },
    {
      "epoch": 0.5171271509660322,
      "grad_norm": 1.530346155166626,
      "learning_rate": 9.930986420883623e-05,
      "loss": 1.4033,
      "step": 14440
    },
    {
      "epoch": 0.5171629630956005,
      "grad_norm": 3.921464443206787,
      "learning_rate": 9.929826547679563e-05,
      "loss": 1.6177,
      "step": 14441
    },
    {
      "epoch": 0.5171987752251688,
      "grad_norm": 1.572906494140625,
      "learning_rate": 9.928666675419595e-05,
      "loss": 1.6209,
      "step": 14442
    },
    {
      "epoch": 0.517234587354737,
      "grad_norm": 2.0591559410095215,
      "learning_rate": 9.927506804119326e-05,
      "loss": 1.1671,
      "step": 14443
    },
    {
      "epoch": 0.5172703994843053,
      "grad_norm": 1.4399479627609253,
      "learning_rate": 9.926346933794357e-05,
      "loss": 1.4316,
      "step": 14444
    },
    {
      "epoch": 0.5173062116138736,
      "grad_norm": 1.3752440214157104,
      "learning_rate": 9.925187064460299e-05,
      "loss": 1.4705,
      "step": 14445
    },
    {
      "epoch": 0.517342023743442,
      "grad_norm": 1.7031986713409424,
      "learning_rate": 9.924027196132747e-05,
      "loss": 1.3291,
      "step": 14446
    },
    {
      "epoch": 0.5173778358730102,
      "grad_norm": 1.779016137123108,
      "learning_rate": 9.922867328827319e-05,
      "loss": 1.2053,
      "step": 14447
    },
    {
      "epoch": 0.5174136480025785,
      "grad_norm": 1.4940849542617798,
      "learning_rate": 9.921707462559608e-05,
      "loss": 1.6245,
      "step": 14448
    },
    {
      "epoch": 0.5174494601321468,
      "grad_norm": 1.6371047496795654,
      "learning_rate": 9.92054759734522e-05,
      "loss": 1.2594,
      "step": 14449
    },
    {
      "epoch": 0.517485272261715,
      "grad_norm": 2.4114179611206055,
      "learning_rate": 9.919387733199764e-05,
      "loss": 1.4802,
      "step": 14450
    },
    {
      "epoch": 0.5175210843912833,
      "grad_norm": 1.8427387475967407,
      "learning_rate": 9.91822787013884e-05,
      "loss": 1.7216,
      "step": 14451
    },
    {
      "epoch": 0.5175568965208516,
      "grad_norm": 1.5938351154327393,
      "learning_rate": 9.917068008178056e-05,
      "loss": 1.3973,
      "step": 14452
    },
    {
      "epoch": 0.51759270865042,
      "grad_norm": 1.515212059020996,
      "learning_rate": 9.915908147333013e-05,
      "loss": 1.3184,
      "step": 14453
    },
    {
      "epoch": 0.5176285207799882,
      "grad_norm": 2.3098485469818115,
      "learning_rate": 9.91474828761932e-05,
      "loss": 1.3732,
      "step": 14454
    },
    {
      "epoch": 0.5176643329095565,
      "grad_norm": 1.29752779006958,
      "learning_rate": 9.913588429052578e-05,
      "loss": 1.3128,
      "step": 14455
    },
    {
      "epoch": 0.5177001450391248,
      "grad_norm": 1.9097155332565308,
      "learning_rate": 9.912428571648393e-05,
      "loss": 1.5393,
      "step": 14456
    },
    {
      "epoch": 0.517735957168693,
      "grad_norm": 1.6922508478164673,
      "learning_rate": 9.91126871542237e-05,
      "loss": 1.4879,
      "step": 14457
    },
    {
      "epoch": 0.5177717692982613,
      "grad_norm": 2.132388114929199,
      "learning_rate": 9.910108860390107e-05,
      "loss": 1.5717,
      "step": 14458
    },
    {
      "epoch": 0.5178075814278296,
      "grad_norm": 2.9997427463531494,
      "learning_rate": 9.908949006567218e-05,
      "loss": 1.6868,
      "step": 14459
    },
    {
      "epoch": 0.5178433935573978,
      "grad_norm": 1.777268409729004,
      "learning_rate": 9.9077891539693e-05,
      "loss": 1.6998,
      "step": 14460
    },
    {
      "epoch": 0.5178792056869662,
      "grad_norm": 2.5405521392822266,
      "learning_rate": 9.906629302611961e-05,
      "loss": 1.8193,
      "step": 14461
    },
    {
      "epoch": 0.5179150178165345,
      "grad_norm": 2.0599591732025146,
      "learning_rate": 9.905469452510803e-05,
      "loss": 1.7768,
      "step": 14462
    },
    {
      "epoch": 0.5179508299461028,
      "grad_norm": 1.894958734512329,
      "learning_rate": 9.904309603681433e-05,
      "loss": 1.4591,
      "step": 14463
    },
    {
      "epoch": 0.517986642075671,
      "grad_norm": 1.3576775789260864,
      "learning_rate": 9.903149756139453e-05,
      "loss": 1.4085,
      "step": 14464
    },
    {
      "epoch": 0.5180224542052393,
      "grad_norm": 1.7294950485229492,
      "learning_rate": 9.901989909900469e-05,
      "loss": 1.7503,
      "step": 14465
    },
    {
      "epoch": 0.5180582663348076,
      "grad_norm": 1.5523263216018677,
      "learning_rate": 9.900830064980084e-05,
      "loss": 1.5677,
      "step": 14466
    },
    {
      "epoch": 0.5180940784643758,
      "grad_norm": 1.6701172590255737,
      "learning_rate": 9.899670221393901e-05,
      "loss": 1.7297,
      "step": 14467
    },
    {
      "epoch": 0.5181298905939442,
      "grad_norm": 1.5365049839019775,
      "learning_rate": 9.89851037915753e-05,
      "loss": 1.4214,
      "step": 14468
    },
    {
      "epoch": 0.5181657027235125,
      "grad_norm": 1.5968941450119019,
      "learning_rate": 9.897350538286566e-05,
      "loss": 1.4887,
      "step": 14469
    },
    {
      "epoch": 0.5182015148530807,
      "grad_norm": 2.073341131210327,
      "learning_rate": 9.896190698796621e-05,
      "loss": 1.4142,
      "step": 14470
    },
    {
      "epoch": 0.518237326982649,
      "grad_norm": 2.2669150829315186,
      "learning_rate": 9.895030860703295e-05,
      "loss": 1.6122,
      "step": 14471
    },
    {
      "epoch": 0.5182731391122173,
      "grad_norm": 1.5452347993850708,
      "learning_rate": 9.893871024022195e-05,
      "loss": 1.2547,
      "step": 14472
    },
    {
      "epoch": 0.5183089512417856,
      "grad_norm": 1.556300401687622,
      "learning_rate": 9.89271118876892e-05,
      "loss": 1.587,
      "step": 14473
    },
    {
      "epoch": 0.5183447633713538,
      "grad_norm": 1.6784199476242065,
      "learning_rate": 9.891551354959082e-05,
      "loss": 1.5504,
      "step": 14474
    },
    {
      "epoch": 0.5183805755009222,
      "grad_norm": 2.0387167930603027,
      "learning_rate": 9.890391522608278e-05,
      "loss": 1.5993,
      "step": 14475
    },
    {
      "epoch": 0.5184163876304905,
      "grad_norm": 1.639019250869751,
      "learning_rate": 9.889231691732115e-05,
      "loss": 1.5219,
      "step": 14476
    },
    {
      "epoch": 0.5184521997600587,
      "grad_norm": 1.59255051612854,
      "learning_rate": 9.888071862346198e-05,
      "loss": 1.5876,
      "step": 14477
    },
    {
      "epoch": 0.518488011889627,
      "grad_norm": 1.663064956665039,
      "learning_rate": 9.886912034466127e-05,
      "loss": 1.0076,
      "step": 14478
    },
    {
      "epoch": 0.5185238240191953,
      "grad_norm": 1.6870861053466797,
      "learning_rate": 9.885752208107511e-05,
      "loss": 1.5689,
      "step": 14479
    },
    {
      "epoch": 0.5185596361487635,
      "grad_norm": 1.2715725898742676,
      "learning_rate": 9.88459238328595e-05,
      "loss": 1.3663,
      "step": 14480
    },
    {
      "epoch": 0.5185954482783318,
      "grad_norm": 2.3270177841186523,
      "learning_rate": 9.883432560017052e-05,
      "loss": 1.6806,
      "step": 14481
    },
    {
      "epoch": 0.5186312604079002,
      "grad_norm": 2.031411647796631,
      "learning_rate": 9.882272738316418e-05,
      "loss": 1.6844,
      "step": 14482
    },
    {
      "epoch": 0.5186670725374685,
      "grad_norm": 1.4794574975967407,
      "learning_rate": 9.88111291819965e-05,
      "loss": 1.5449,
      "step": 14483
    },
    {
      "epoch": 0.5187028846670367,
      "grad_norm": 1.525395154953003,
      "learning_rate": 9.879953099682358e-05,
      "loss": 1.4823,
      "step": 14484
    },
    {
      "epoch": 0.518738696796605,
      "grad_norm": 2.1387875080108643,
      "learning_rate": 9.878793282780137e-05,
      "loss": 1.4335,
      "step": 14485
    },
    {
      "epoch": 0.5187745089261733,
      "grad_norm": 1.8972395658493042,
      "learning_rate": 9.877633467508602e-05,
      "loss": 1.3409,
      "step": 14486
    },
    {
      "epoch": 0.5188103210557415,
      "grad_norm": 1.8318805694580078,
      "learning_rate": 9.876473653883346e-05,
      "loss": 1.3479,
      "step": 14487
    },
    {
      "epoch": 0.5188461331853098,
      "grad_norm": 1.8570977449417114,
      "learning_rate": 9.87531384191998e-05,
      "loss": 1.4737,
      "step": 14488
    },
    {
      "epoch": 0.5188819453148782,
      "grad_norm": 1.4921190738677979,
      "learning_rate": 9.874154031634103e-05,
      "loss": 1.5501,
      "step": 14489
    },
    {
      "epoch": 0.5189177574444465,
      "grad_norm": 2.1454734802246094,
      "learning_rate": 9.872994223041325e-05,
      "loss": 1.7193,
      "step": 14490
    },
    {
      "epoch": 0.5189535695740147,
      "grad_norm": 1.6640002727508545,
      "learning_rate": 9.871834416157246e-05,
      "loss": 1.1278,
      "step": 14491
    },
    {
      "epoch": 0.518989381703583,
      "grad_norm": 1.7006652355194092,
      "learning_rate": 9.870674610997467e-05,
      "loss": 1.6311,
      "step": 14492
    },
    {
      "epoch": 0.5190251938331513,
      "grad_norm": 1.4623268842697144,
      "learning_rate": 9.869514807577595e-05,
      "loss": 1.0958,
      "step": 14493
    },
    {
      "epoch": 0.5190610059627195,
      "grad_norm": 1.8661326169967651,
      "learning_rate": 9.868355005913232e-05,
      "loss": 1.2537,
      "step": 14494
    },
    {
      "epoch": 0.5190968180922878,
      "grad_norm": 1.5580703020095825,
      "learning_rate": 9.867195206019985e-05,
      "loss": 1.6413,
      "step": 14495
    },
    {
      "epoch": 0.5191326302218562,
      "grad_norm": 2.2893712520599365,
      "learning_rate": 9.866035407913452e-05,
      "loss": 1.5394,
      "step": 14496
    },
    {
      "epoch": 0.5191684423514245,
      "grad_norm": 1.4904677867889404,
      "learning_rate": 9.864875611609243e-05,
      "loss": 1.5643,
      "step": 14497
    },
    {
      "epoch": 0.5192042544809927,
      "grad_norm": 1.5410066843032837,
      "learning_rate": 9.863715817122956e-05,
      "loss": 1.3726,
      "step": 14498
    },
    {
      "epoch": 0.519240066610561,
      "grad_norm": 1.7164530754089355,
      "learning_rate": 9.862556024470199e-05,
      "loss": 1.6264,
      "step": 14499
    },
    {
      "epoch": 0.5192758787401293,
      "grad_norm": 1.4902276992797852,
      "learning_rate": 9.861396233666574e-05,
      "loss": 1.5273,
      "step": 14500
    },
    {
      "epoch": 0.5193116908696975,
      "grad_norm": 1.731277346611023,
      "learning_rate": 9.860236444727679e-05,
      "loss": 1.5882,
      "step": 14501
    },
    {
      "epoch": 0.5193475029992658,
      "grad_norm": 1.6587519645690918,
      "learning_rate": 9.859076657669127e-05,
      "loss": 1.3716,
      "step": 14502
    },
    {
      "epoch": 0.5193833151288342,
      "grad_norm": 2.1876940727233887,
      "learning_rate": 9.857916872506513e-05,
      "loss": 1.5538,
      "step": 14503
    },
    {
      "epoch": 0.5194191272584024,
      "grad_norm": 1.7598111629486084,
      "learning_rate": 9.856757089255448e-05,
      "loss": 1.4646,
      "step": 14504
    },
    {
      "epoch": 0.5194549393879707,
      "grad_norm": 1.1007812023162842,
      "learning_rate": 9.85559730793153e-05,
      "loss": 1.2949,
      "step": 14505
    },
    {
      "epoch": 0.519490751517539,
      "grad_norm": 1.786514163017273,
      "learning_rate": 9.854437528550364e-05,
      "loss": 1.4212,
      "step": 14506
    },
    {
      "epoch": 0.5195265636471073,
      "grad_norm": 1.267168641090393,
      "learning_rate": 9.853277751127552e-05,
      "loss": 1.5191,
      "step": 14507
    },
    {
      "epoch": 0.5195623757766755,
      "grad_norm": 1.632846713066101,
      "learning_rate": 9.852117975678701e-05,
      "loss": 1.3769,
      "step": 14508
    },
    {
      "epoch": 0.5195981879062438,
      "grad_norm": 1.708034873008728,
      "learning_rate": 9.850958202219414e-05,
      "loss": 1.4926,
      "step": 14509
    },
    {
      "epoch": 0.5196340000358122,
      "grad_norm": 1.7593141794204712,
      "learning_rate": 9.849798430765286e-05,
      "loss": 1.3824,
      "step": 14510
    },
    {
      "epoch": 0.5196698121653804,
      "grad_norm": 2.2882802486419678,
      "learning_rate": 9.848638661331933e-05,
      "loss": 1.4505,
      "step": 14511
    },
    {
      "epoch": 0.5197056242949487,
      "grad_norm": 1.923661231994629,
      "learning_rate": 9.847478893934944e-05,
      "loss": 1.5545,
      "step": 14512
    },
    {
      "epoch": 0.519741436424517,
      "grad_norm": 2.511514186859131,
      "learning_rate": 9.846319128589936e-05,
      "loss": 1.2665,
      "step": 14513
    },
    {
      "epoch": 0.5197772485540852,
      "grad_norm": 1.4388993978500366,
      "learning_rate": 9.845159365312501e-05,
      "loss": 1.2587,
      "step": 14514
    },
    {
      "epoch": 0.5198130606836535,
      "grad_norm": 1.2576457262039185,
      "learning_rate": 9.84399960411825e-05,
      "loss": 1.1563,
      "step": 14515
    },
    {
      "epoch": 0.5198488728132218,
      "grad_norm": 2.3293657302856445,
      "learning_rate": 9.842839845022781e-05,
      "loss": 1.6719,
      "step": 14516
    },
    {
      "epoch": 0.5198846849427902,
      "grad_norm": 1.6276086568832397,
      "learning_rate": 9.841680088041701e-05,
      "loss": 1.5341,
      "step": 14517
    },
    {
      "epoch": 0.5199204970723584,
      "grad_norm": 1.4863203763961792,
      "learning_rate": 9.840520333190615e-05,
      "loss": 1.3341,
      "step": 14518
    },
    {
      "epoch": 0.5199563092019267,
      "grad_norm": 1.4681768417358398,
      "learning_rate": 9.839360580485115e-05,
      "loss": 1.2057,
      "step": 14519
    },
    {
      "epoch": 0.519992121331495,
      "grad_norm": 2.524519443511963,
      "learning_rate": 9.838200829940818e-05,
      "loss": 1.4854,
      "step": 14520
    },
    {
      "epoch": 0.5200279334610632,
      "grad_norm": 1.424342155456543,
      "learning_rate": 9.837041081573312e-05,
      "loss": 1.1943,
      "step": 14521
    },
    {
      "epoch": 0.5200637455906315,
      "grad_norm": 1.5036619901657104,
      "learning_rate": 9.835881335398216e-05,
      "loss": 1.5235,
      "step": 14522
    },
    {
      "epoch": 0.5200995577201998,
      "grad_norm": 1.508616328239441,
      "learning_rate": 9.834721591431118e-05,
      "loss": 1.6061,
      "step": 14523
    },
    {
      "epoch": 0.5201353698497682,
      "grad_norm": 1.989552617073059,
      "learning_rate": 9.833561849687634e-05,
      "loss": 1.6041,
      "step": 14524
    },
    {
      "epoch": 0.5201711819793364,
      "grad_norm": 1.4946794509887695,
      "learning_rate": 9.832402110183355e-05,
      "loss": 1.4669,
      "step": 14525
    },
    {
      "epoch": 0.5202069941089047,
      "grad_norm": 1.4868971109390259,
      "learning_rate": 9.831242372933896e-05,
      "loss": 1.4832,
      "step": 14526
    },
    {
      "epoch": 0.520242806238473,
      "grad_norm": 1.2756190299987793,
      "learning_rate": 9.830082637954851e-05,
      "loss": 1.2785,
      "step": 14527
    },
    {
      "epoch": 0.5202786183680412,
      "grad_norm": 1.9612704515457153,
      "learning_rate": 9.828922905261819e-05,
      "loss": 1.355,
      "step": 14528
    },
    {
      "epoch": 0.5203144304976095,
      "grad_norm": 2.44775652885437,
      "learning_rate": 9.827763174870417e-05,
      "loss": 1.4828,
      "step": 14529
    },
    {
      "epoch": 0.5203502426271778,
      "grad_norm": 1.8801780939102173,
      "learning_rate": 9.826603446796231e-05,
      "loss": 1.1552,
      "step": 14530
    },
    {
      "epoch": 0.5203860547567462,
      "grad_norm": 2.448315143585205,
      "learning_rate": 9.82544372105488e-05,
      "loss": 1.5563,
      "step": 14531
    },
    {
      "epoch": 0.5204218668863144,
      "grad_norm": 2.337183713912964,
      "learning_rate": 9.824283997661952e-05,
      "loss": 1.6272,
      "step": 14532
    },
    {
      "epoch": 0.5204576790158827,
      "grad_norm": 1.5295284986495972,
      "learning_rate": 9.823124276633061e-05,
      "loss": 1.7039,
      "step": 14533
    },
    {
      "epoch": 0.520493491145451,
      "grad_norm": 2.0706472396850586,
      "learning_rate": 9.821964557983799e-05,
      "loss": 1.4274,
      "step": 14534
    },
    {
      "epoch": 0.5205293032750192,
      "grad_norm": 1.6976041793823242,
      "learning_rate": 9.820804841729782e-05,
      "loss": 1.3637,
      "step": 14535
    },
    {
      "epoch": 0.5205651154045875,
      "grad_norm": 1.4130737781524658,
      "learning_rate": 9.819645127886602e-05,
      "loss": 1.3593,
      "step": 14536
    },
    {
      "epoch": 0.5206009275341558,
      "grad_norm": 1.3637433052062988,
      "learning_rate": 9.818485416469861e-05,
      "loss": 1.4258,
      "step": 14537
    },
    {
      "epoch": 0.5206367396637241,
      "grad_norm": 1.7225490808486938,
      "learning_rate": 9.817325707495167e-05,
      "loss": 1.5675,
      "step": 14538
    },
    {
      "epoch": 0.5206725517932924,
      "grad_norm": 2.0762996673583984,
      "learning_rate": 9.816166000978119e-05,
      "loss": 1.2102,
      "step": 14539
    },
    {
      "epoch": 0.5207083639228607,
      "grad_norm": 1.6375013589859009,
      "learning_rate": 9.815006296934321e-05,
      "loss": 1.5834,
      "step": 14540
    },
    {
      "epoch": 0.520744176052429,
      "grad_norm": 1.525462031364441,
      "learning_rate": 9.813846595379371e-05,
      "loss": 1.6418,
      "step": 14541
    },
    {
      "epoch": 0.5207799881819972,
      "grad_norm": 1.9003756046295166,
      "learning_rate": 9.812686896328882e-05,
      "loss": 1.9051,
      "step": 14542
    },
    {
      "epoch": 0.5208158003115655,
      "grad_norm": 1.4063869714736938,
      "learning_rate": 9.811527199798443e-05,
      "loss": 1.2691,
      "step": 14543
    },
    {
      "epoch": 0.5208516124411338,
      "grad_norm": 2.0352799892425537,
      "learning_rate": 9.810367505803667e-05,
      "loss": 1.5649,
      "step": 14544
    },
    {
      "epoch": 0.5208874245707021,
      "grad_norm": 2.369400978088379,
      "learning_rate": 9.80920781436015e-05,
      "loss": 1.5076,
      "step": 14545
    },
    {
      "epoch": 0.5209232367002704,
      "grad_norm": 1.6463871002197266,
      "learning_rate": 9.808048125483494e-05,
      "loss": 1.2952,
      "step": 14546
    },
    {
      "epoch": 0.5209590488298387,
      "grad_norm": 1.4209014177322388,
      "learning_rate": 9.806888439189306e-05,
      "loss": 1.6418,
      "step": 14547
    },
    {
      "epoch": 0.520994860959407,
      "grad_norm": 1.5328431129455566,
      "learning_rate": 9.805728755493182e-05,
      "loss": 1.1911,
      "step": 14548
    },
    {
      "epoch": 0.5210306730889752,
      "grad_norm": 1.4673058986663818,
      "learning_rate": 9.804569074410729e-05,
      "loss": 1.3006,
      "step": 14549
    },
    {
      "epoch": 0.5210664852185435,
      "grad_norm": 1.8996427059173584,
      "learning_rate": 9.803409395957545e-05,
      "loss": 1.54,
      "step": 14550
    },
    {
      "epoch": 0.5211022973481118,
      "grad_norm": 1.5183322429656982,
      "learning_rate": 9.802249720149236e-05,
      "loss": 1.444,
      "step": 14551
    },
    {
      "epoch": 0.5211381094776801,
      "grad_norm": 2.154158353805542,
      "learning_rate": 9.8010900470014e-05,
      "loss": 1.3751,
      "step": 14552
    },
    {
      "epoch": 0.5211739216072484,
      "grad_norm": 1.3777554035186768,
      "learning_rate": 9.799930376529643e-05,
      "loss": 1.5448,
      "step": 14553
    },
    {
      "epoch": 0.5212097337368167,
      "grad_norm": 1.716951847076416,
      "learning_rate": 9.798770708749563e-05,
      "loss": 1.4537,
      "step": 14554
    },
    {
      "epoch": 0.5212455458663849,
      "grad_norm": 1.4311693906784058,
      "learning_rate": 9.797611043676764e-05,
      "loss": 1.6834,
      "step": 14555
    },
    {
      "epoch": 0.5212813579959532,
      "grad_norm": 1.699777603149414,
      "learning_rate": 9.796451381326849e-05,
      "loss": 1.7289,
      "step": 14556
    },
    {
      "epoch": 0.5213171701255215,
      "grad_norm": 1.704720139503479,
      "learning_rate": 9.795291721715414e-05,
      "loss": 1.7788,
      "step": 14557
    },
    {
      "epoch": 0.5213529822550897,
      "grad_norm": 1.8193825483322144,
      "learning_rate": 9.794132064858069e-05,
      "loss": 1.7059,
      "step": 14558
    },
    {
      "epoch": 0.5213887943846581,
      "grad_norm": 1.9302318096160889,
      "learning_rate": 9.792972410770409e-05,
      "loss": 1.6084,
      "step": 14559
    },
    {
      "epoch": 0.5214246065142264,
      "grad_norm": 1.3157739639282227,
      "learning_rate": 9.791812759468039e-05,
      "loss": 1.2628,
      "step": 14560
    },
    {
      "epoch": 0.5214604186437947,
      "grad_norm": 1.3691009283065796,
      "learning_rate": 9.79065311096656e-05,
      "loss": 1.2926,
      "step": 14561
    },
    {
      "epoch": 0.5214962307733629,
      "grad_norm": 1.2893990278244019,
      "learning_rate": 9.789493465281574e-05,
      "loss": 1.6619,
      "step": 14562
    },
    {
      "epoch": 0.5215320429029312,
      "grad_norm": 1.7624584436416626,
      "learning_rate": 9.788333822428682e-05,
      "loss": 1.5958,
      "step": 14563
    },
    {
      "epoch": 0.5215678550324995,
      "grad_norm": 1.5968478918075562,
      "learning_rate": 9.787174182423484e-05,
      "loss": 1.5239,
      "step": 14564
    },
    {
      "epoch": 0.5216036671620677,
      "grad_norm": 2.377908945083618,
      "learning_rate": 9.786014545281585e-05,
      "loss": 1.5145,
      "step": 14565
    },
    {
      "epoch": 0.5216394792916361,
      "grad_norm": 1.5355181694030762,
      "learning_rate": 9.78485491101858e-05,
      "loss": 1.5049,
      "step": 14566
    },
    {
      "epoch": 0.5216752914212044,
      "grad_norm": 1.638087272644043,
      "learning_rate": 9.783695279650079e-05,
      "loss": 1.3355,
      "step": 14567
    },
    {
      "epoch": 0.5217111035507727,
      "grad_norm": 2.3670177459716797,
      "learning_rate": 9.782535651191676e-05,
      "loss": 1.2759,
      "step": 14568
    },
    {
      "epoch": 0.5217469156803409,
      "grad_norm": 2.8764686584472656,
      "learning_rate": 9.781376025658977e-05,
      "loss": 1.5412,
      "step": 14569
    },
    {
      "epoch": 0.5217827278099092,
      "grad_norm": 1.5233416557312012,
      "learning_rate": 9.78021640306758e-05,
      "loss": 1.1315,
      "step": 14570
    },
    {
      "epoch": 0.5218185399394775,
      "grad_norm": 1.8155995607376099,
      "learning_rate": 9.77905678343309e-05,
      "loss": 1.769,
      "step": 14571
    },
    {
      "epoch": 0.5218543520690457,
      "grad_norm": 1.6024277210235596,
      "learning_rate": 9.777897166771107e-05,
      "loss": 1.2893,
      "step": 14572
    },
    {
      "epoch": 0.5218901641986141,
      "grad_norm": 1.7057386636734009,
      "learning_rate": 9.776737553097227e-05,
      "loss": 1.6424,
      "step": 14573
    },
    {
      "epoch": 0.5219259763281824,
      "grad_norm": 2.125110149383545,
      "learning_rate": 9.775577942427058e-05,
      "loss": 1.3813,
      "step": 14574
    },
    {
      "epoch": 0.5219617884577507,
      "grad_norm": 1.6502152681350708,
      "learning_rate": 9.774418334776196e-05,
      "loss": 1.4526,
      "step": 14575
    },
    {
      "epoch": 0.5219976005873189,
      "grad_norm": 1.7954357862472534,
      "learning_rate": 9.773258730160247e-05,
      "loss": 1.5242,
      "step": 14576
    },
    {
      "epoch": 0.5220334127168872,
      "grad_norm": 1.6591635942459106,
      "learning_rate": 9.772099128594808e-05,
      "loss": 1.3148,
      "step": 14577
    },
    {
      "epoch": 0.5220692248464555,
      "grad_norm": 1.6281665563583374,
      "learning_rate": 9.770939530095482e-05,
      "loss": 1.6924,
      "step": 14578
    },
    {
      "epoch": 0.5221050369760237,
      "grad_norm": 1.7232989072799683,
      "learning_rate": 9.769779934677869e-05,
      "loss": 1.7348,
      "step": 14579
    },
    {
      "epoch": 0.5221408491055921,
      "grad_norm": 2.068788766860962,
      "learning_rate": 9.76862034235757e-05,
      "loss": 1.4131,
      "step": 14580
    },
    {
      "epoch": 0.5221766612351604,
      "grad_norm": 2.2537050247192383,
      "learning_rate": 9.767460753150186e-05,
      "loss": 1.3277,
      "step": 14581
    },
    {
      "epoch": 0.5222124733647286,
      "grad_norm": 2.0896992683410645,
      "learning_rate": 9.766301167071316e-05,
      "loss": 1.5849,
      "step": 14582
    },
    {
      "epoch": 0.5222482854942969,
      "grad_norm": 1.4738152027130127,
      "learning_rate": 9.765141584136565e-05,
      "loss": 1.4946,
      "step": 14583
    },
    {
      "epoch": 0.5222840976238652,
      "grad_norm": 1.8603551387786865,
      "learning_rate": 9.763982004361527e-05,
      "loss": 1.6751,
      "step": 14584
    },
    {
      "epoch": 0.5223199097534335,
      "grad_norm": 1.5763189792633057,
      "learning_rate": 9.76282242776181e-05,
      "loss": 1.2796,
      "step": 14585
    },
    {
      "epoch": 0.5223557218830017,
      "grad_norm": 1.772830843925476,
      "learning_rate": 9.76166285435301e-05,
      "loss": 1.6905,
      "step": 14586
    },
    {
      "epoch": 0.5223915340125701,
      "grad_norm": 1.5136207342147827,
      "learning_rate": 9.76050328415073e-05,
      "loss": 1.3872,
      "step": 14587
    },
    {
      "epoch": 0.5224273461421384,
      "grad_norm": 1.3309680223464966,
      "learning_rate": 9.759343717170571e-05,
      "loss": 1.2488,
      "step": 14588
    },
    {
      "epoch": 0.5224631582717066,
      "grad_norm": 2.110062837600708,
      "learning_rate": 9.758184153428126e-05,
      "loss": 1.5052,
      "step": 14589
    },
    {
      "epoch": 0.5224989704012749,
      "grad_norm": 1.3645905256271362,
      "learning_rate": 9.757024592939008e-05,
      "loss": 1.4249,
      "step": 14590
    },
    {
      "epoch": 0.5225347825308432,
      "grad_norm": 1.6391838788986206,
      "learning_rate": 9.755865035718807e-05,
      "loss": 1.5082,
      "step": 14591
    },
    {
      "epoch": 0.5225705946604114,
      "grad_norm": 1.9833340644836426,
      "learning_rate": 9.754705481783127e-05,
      "loss": 1.6181,
      "step": 14592
    },
    {
      "epoch": 0.5226064067899797,
      "grad_norm": 1.6536145210266113,
      "learning_rate": 9.753545931147569e-05,
      "loss": 1.7407,
      "step": 14593
    },
    {
      "epoch": 0.5226422189195481,
      "grad_norm": 1.3998302221298218,
      "learning_rate": 9.752386383827733e-05,
      "loss": 1.5109,
      "step": 14594
    },
    {
      "epoch": 0.5226780310491164,
      "grad_norm": 2.0293946266174316,
      "learning_rate": 9.751226839839217e-05,
      "loss": 1.5249,
      "step": 14595
    },
    {
      "epoch": 0.5227138431786846,
      "grad_norm": 1.617629885673523,
      "learning_rate": 9.750067299197625e-05,
      "loss": 1.64,
      "step": 14596
    },
    {
      "epoch": 0.5227496553082529,
      "grad_norm": 1.4700734615325928,
      "learning_rate": 9.748907761918558e-05,
      "loss": 1.4105,
      "step": 14597
    },
    {
      "epoch": 0.5227854674378212,
      "grad_norm": 2.149026870727539,
      "learning_rate": 9.747748228017606e-05,
      "loss": 1.4068,
      "step": 14598
    },
    {
      "epoch": 0.5228212795673894,
      "grad_norm": 1.604383111000061,
      "learning_rate": 9.746588697510381e-05,
      "loss": 1.4969,
      "step": 14599
    },
    {
      "epoch": 0.5228570916969577,
      "grad_norm": 2.5129008293151855,
      "learning_rate": 9.745429170412476e-05,
      "loss": 1.5898,
      "step": 14600
    },
    {
      "epoch": 0.5228929038265261,
      "grad_norm": 1.2759195566177368,
      "learning_rate": 9.744269646739494e-05,
      "loss": 1.2335,
      "step": 14601
    },
    {
      "epoch": 0.5229287159560944,
      "grad_norm": 1.4611231088638306,
      "learning_rate": 9.743110126507034e-05,
      "loss": 1.2164,
      "step": 14602
    },
    {
      "epoch": 0.5229645280856626,
      "grad_norm": 1.63002610206604,
      "learning_rate": 9.741950609730696e-05,
      "loss": 1.5584,
      "step": 14603
    },
    {
      "epoch": 0.5230003402152309,
      "grad_norm": 1.3063840866088867,
      "learning_rate": 9.74079109642608e-05,
      "loss": 1.0043,
      "step": 14604
    },
    {
      "epoch": 0.5230361523447992,
      "grad_norm": 1.531385064125061,
      "learning_rate": 9.739631586608786e-05,
      "loss": 1.3484,
      "step": 14605
    },
    {
      "epoch": 0.5230719644743674,
      "grad_norm": 1.7938756942749023,
      "learning_rate": 9.738472080294415e-05,
      "loss": 1.5975,
      "step": 14606
    },
    {
      "epoch": 0.5231077766039357,
      "grad_norm": 1.6742467880249023,
      "learning_rate": 9.737312577498559e-05,
      "loss": 1.5811,
      "step": 14607
    },
    {
      "epoch": 0.5231435887335041,
      "grad_norm": 1.9026448726654053,
      "learning_rate": 9.73615307823683e-05,
      "loss": 1.3136,
      "step": 14608
    },
    {
      "epoch": 0.5231794008630724,
      "grad_norm": 1.5856115818023682,
      "learning_rate": 9.734993582524814e-05,
      "loss": 1.4487,
      "step": 14609
    },
    {
      "epoch": 0.5232152129926406,
      "grad_norm": 1.6417968273162842,
      "learning_rate": 9.733834090378125e-05,
      "loss": 1.3294,
      "step": 14610
    },
    {
      "epoch": 0.5232510251222089,
      "grad_norm": 1.5064736604690552,
      "learning_rate": 9.732674601812347e-05,
      "loss": 1.7278,
      "step": 14611
    },
    {
      "epoch": 0.5232868372517772,
      "grad_norm": 1.5335125923156738,
      "learning_rate": 9.731515116843094e-05,
      "loss": 1.5331,
      "step": 14612
    },
    {
      "epoch": 0.5233226493813454,
      "grad_norm": 1.4457260370254517,
      "learning_rate": 9.730355635485953e-05,
      "loss": 1.3542,
      "step": 14613
    },
    {
      "epoch": 0.5233584615109137,
      "grad_norm": 1.6329476833343506,
      "learning_rate": 9.729196157756534e-05,
      "loss": 1.429,
      "step": 14614
    },
    {
      "epoch": 0.5233942736404821,
      "grad_norm": 2.010593891143799,
      "learning_rate": 9.728036683670433e-05,
      "loss": 1.5153,
      "step": 14615
    },
    {
      "epoch": 0.5234300857700503,
      "grad_norm": 1.8740265369415283,
      "learning_rate": 9.72687721324324e-05,
      "loss": 1.582,
      "step": 14616
    },
    {
      "epoch": 0.5234658978996186,
      "grad_norm": 2.2544100284576416,
      "learning_rate": 9.725717746490571e-05,
      "loss": 1.6876,
      "step": 14617
    },
    {
      "epoch": 0.5235017100291869,
      "grad_norm": 1.692060112953186,
      "learning_rate": 9.724558283428007e-05,
      "loss": 1.5436,
      "step": 14618
    },
    {
      "epoch": 0.5235375221587552,
      "grad_norm": 1.4941781759262085,
      "learning_rate": 9.723398824071164e-05,
      "loss": 1.4658,
      "step": 14619
    },
    {
      "epoch": 0.5235733342883234,
      "grad_norm": 1.7871164083480835,
      "learning_rate": 9.722239368435624e-05,
      "loss": 1.5842,
      "step": 14620
    },
    {
      "epoch": 0.5236091464178917,
      "grad_norm": 2.2636523246765137,
      "learning_rate": 9.721079916537004e-05,
      "loss": 1.709,
      "step": 14621
    },
    {
      "epoch": 0.5236449585474601,
      "grad_norm": 1.7238781452178955,
      "learning_rate": 9.719920468390888e-05,
      "loss": 1.6237,
      "step": 14622
    },
    {
      "epoch": 0.5236807706770283,
      "grad_norm": 2.144887924194336,
      "learning_rate": 9.718761024012886e-05,
      "loss": 1.3904,
      "step": 14623
    },
    {
      "epoch": 0.5237165828065966,
      "grad_norm": 1.6877436637878418,
      "learning_rate": 9.717601583418588e-05,
      "loss": 1.5621,
      "step": 14624
    },
    {
      "epoch": 0.5237523949361649,
      "grad_norm": 1.2671278715133667,
      "learning_rate": 9.716442146623594e-05,
      "loss": 1.2752,
      "step": 14625
    },
    {
      "epoch": 0.5237882070657331,
      "grad_norm": 2.0848145484924316,
      "learning_rate": 9.715282713643512e-05,
      "loss": 1.378,
      "step": 14626
    },
    {
      "epoch": 0.5238240191953014,
      "grad_norm": 2.0249481201171875,
      "learning_rate": 9.714123284493925e-05,
      "loss": 1.227,
      "step": 14627
    },
    {
      "epoch": 0.5238598313248697,
      "grad_norm": 1.9421577453613281,
      "learning_rate": 9.712963859190449e-05,
      "loss": 1.6061,
      "step": 14628
    },
    {
      "epoch": 0.5238956434544381,
      "grad_norm": 1.442251443862915,
      "learning_rate": 9.711804437748669e-05,
      "loss": 1.4937,
      "step": 14629
    },
    {
      "epoch": 0.5239314555840063,
      "grad_norm": 1.8193825483322144,
      "learning_rate": 9.710645020184193e-05,
      "loss": 1.5649,
      "step": 14630
    },
    {
      "epoch": 0.5239672677135746,
      "grad_norm": 1.8133513927459717,
      "learning_rate": 9.709485606512607e-05,
      "loss": 1.6271,
      "step": 14631
    },
    {
      "epoch": 0.5240030798431429,
      "grad_norm": 1.5228266716003418,
      "learning_rate": 9.708326196749527e-05,
      "loss": 1.9068,
      "step": 14632
    },
    {
      "epoch": 0.5240388919727111,
      "grad_norm": 2.925337314605713,
      "learning_rate": 9.707166790910538e-05,
      "loss": 1.6324,
      "step": 14633
    },
    {
      "epoch": 0.5240747041022794,
      "grad_norm": 2.007188081741333,
      "learning_rate": 9.70600738901124e-05,
      "loss": 1.4698,
      "step": 14634
    },
    {
      "epoch": 0.5241105162318477,
      "grad_norm": 1.4450551271438599,
      "learning_rate": 9.704847991067236e-05,
      "loss": 1.4539,
      "step": 14635
    },
    {
      "epoch": 0.5241463283614161,
      "grad_norm": 1.247611165046692,
      "learning_rate": 9.703688597094118e-05,
      "loss": 1.394,
      "step": 14636
    },
    {
      "epoch": 0.5241821404909843,
      "grad_norm": 1.4059821367263794,
      "learning_rate": 9.702529207107491e-05,
      "loss": 1.4542,
      "step": 14637
    },
    {
      "epoch": 0.5242179526205526,
      "grad_norm": 1.3357560634613037,
      "learning_rate": 9.701369821122945e-05,
      "loss": 1.4459,
      "step": 14638
    },
    {
      "epoch": 0.5242537647501209,
      "grad_norm": 1.6569325923919678,
      "learning_rate": 9.70021043915609e-05,
      "loss": 1.3154,
      "step": 14639
    },
    {
      "epoch": 0.5242895768796891,
      "grad_norm": 1.5145827531814575,
      "learning_rate": 9.69905106122251e-05,
      "loss": 1.4301,
      "step": 14640
    },
    {
      "epoch": 0.5243253890092574,
      "grad_norm": 1.7709879875183105,
      "learning_rate": 9.697891687337817e-05,
      "loss": 1.4174,
      "step": 14641
    },
    {
      "epoch": 0.5243612011388257,
      "grad_norm": 1.6160484552383423,
      "learning_rate": 9.696732317517599e-05,
      "loss": 1.5869,
      "step": 14642
    },
    {
      "epoch": 0.524397013268394,
      "grad_norm": 1.8500139713287354,
      "learning_rate": 9.695572951777454e-05,
      "loss": 1.5376,
      "step": 14643
    },
    {
      "epoch": 0.5244328253979623,
      "grad_norm": 2.073073148727417,
      "learning_rate": 9.694413590132985e-05,
      "loss": 1.3357,
      "step": 14644
    },
    {
      "epoch": 0.5244686375275306,
      "grad_norm": 1.8784855604171753,
      "learning_rate": 9.693254232599784e-05,
      "loss": 1.4267,
      "step": 14645
    },
    {
      "epoch": 0.5245044496570989,
      "grad_norm": 1.3910045623779297,
      "learning_rate": 9.692094879193455e-05,
      "loss": 1.5124,
      "step": 14646
    },
    {
      "epoch": 0.5245402617866671,
      "grad_norm": 2.0571978092193604,
      "learning_rate": 9.690935529929587e-05,
      "loss": 1.9757,
      "step": 14647
    },
    {
      "epoch": 0.5245760739162354,
      "grad_norm": 1.7781730890274048,
      "learning_rate": 9.689776184823789e-05,
      "loss": 1.3493,
      "step": 14648
    },
    {
      "epoch": 0.5246118860458037,
      "grad_norm": 1.7598072290420532,
      "learning_rate": 9.688616843891648e-05,
      "loss": 1.1725,
      "step": 14649
    },
    {
      "epoch": 0.524647698175372,
      "grad_norm": 2.3508193492889404,
      "learning_rate": 9.687457507148768e-05,
      "loss": 1.685,
      "step": 14650
    },
    {
      "epoch": 0.5246835103049403,
      "grad_norm": 2.0934391021728516,
      "learning_rate": 9.686298174610745e-05,
      "loss": 1.3282,
      "step": 14651
    },
    {
      "epoch": 0.5247193224345086,
      "grad_norm": 1.3602052927017212,
      "learning_rate": 9.685138846293171e-05,
      "loss": 1.3148,
      "step": 14652
    },
    {
      "epoch": 0.5247551345640769,
      "grad_norm": 1.5019041299819946,
      "learning_rate": 9.683979522211652e-05,
      "loss": 1.4427,
      "step": 14653
    },
    {
      "epoch": 0.5247909466936451,
      "grad_norm": 1.4769161939620972,
      "learning_rate": 9.682820202381779e-05,
      "loss": 1.4798,
      "step": 14654
    },
    {
      "epoch": 0.5248267588232134,
      "grad_norm": 1.3109664916992188,
      "learning_rate": 9.681660886819152e-05,
      "loss": 1.3262,
      "step": 14655
    },
    {
      "epoch": 0.5248625709527817,
      "grad_norm": 1.9010754823684692,
      "learning_rate": 9.680501575539365e-05,
      "loss": 1.4391,
      "step": 14656
    },
    {
      "epoch": 0.52489838308235,
      "grad_norm": 1.923020839691162,
      "learning_rate": 9.679342268558019e-05,
      "loss": 1.4267,
      "step": 14657
    },
    {
      "epoch": 0.5249341952119183,
      "grad_norm": 1.572023630142212,
      "learning_rate": 9.678182965890708e-05,
      "loss": 1.4757,
      "step": 14658
    },
    {
      "epoch": 0.5249700073414866,
      "grad_norm": 2.1957943439483643,
      "learning_rate": 9.677023667553033e-05,
      "loss": 1.4175,
      "step": 14659
    },
    {
      "epoch": 0.5250058194710548,
      "grad_norm": 1.832504153251648,
      "learning_rate": 9.675864373560586e-05,
      "loss": 1.2453,
      "step": 14660
    },
    {
      "epoch": 0.5250416316006231,
      "grad_norm": 2.1098172664642334,
      "learning_rate": 9.674705083928965e-05,
      "loss": 1.6742,
      "step": 14661
    },
    {
      "epoch": 0.5250774437301914,
      "grad_norm": 1.6381540298461914,
      "learning_rate": 9.673545798673769e-05,
      "loss": 1.4837,
      "step": 14662
    },
    {
      "epoch": 0.5251132558597597,
      "grad_norm": 1.4275070428848267,
      "learning_rate": 9.67238651781059e-05,
      "loss": 1.3579,
      "step": 14663
    },
    {
      "epoch": 0.525149067989328,
      "grad_norm": 1.609323501586914,
      "learning_rate": 9.671227241355031e-05,
      "loss": 1.3936,
      "step": 14664
    },
    {
      "epoch": 0.5251848801188963,
      "grad_norm": 1.49817955493927,
      "learning_rate": 9.670067969322684e-05,
      "loss": 1.5228,
      "step": 14665
    },
    {
      "epoch": 0.5252206922484646,
      "grad_norm": 1.4510881900787354,
      "learning_rate": 9.668908701729148e-05,
      "loss": 1.3275,
      "step": 14666
    },
    {
      "epoch": 0.5252565043780328,
      "grad_norm": 1.4491652250289917,
      "learning_rate": 9.667749438590017e-05,
      "loss": 1.4348,
      "step": 14667
    },
    {
      "epoch": 0.5252923165076011,
      "grad_norm": 1.9161545038223267,
      "learning_rate": 9.66659017992089e-05,
      "loss": 1.5006,
      "step": 14668
    },
    {
      "epoch": 0.5253281286371694,
      "grad_norm": 1.586111307144165,
      "learning_rate": 9.665430925737362e-05,
      "loss": 1.5004,
      "step": 14669
    },
    {
      "epoch": 0.5253639407667376,
      "grad_norm": 1.7797801494598389,
      "learning_rate": 9.664271676055027e-05,
      "loss": 1.3378,
      "step": 14670
    },
    {
      "epoch": 0.525399752896306,
      "grad_norm": 1.4467164278030396,
      "learning_rate": 9.663112430889487e-05,
      "loss": 1.249,
      "step": 14671
    },
    {
      "epoch": 0.5254355650258743,
      "grad_norm": 1.5761590003967285,
      "learning_rate": 9.661953190256333e-05,
      "loss": 1.6898,
      "step": 14672
    },
    {
      "epoch": 0.5254713771554426,
      "grad_norm": 1.3600083589553833,
      "learning_rate": 9.660793954171163e-05,
      "loss": 1.6802,
      "step": 14673
    },
    {
      "epoch": 0.5255071892850108,
      "grad_norm": 1.4301543235778809,
      "learning_rate": 9.65963472264957e-05,
      "loss": 1.5224,
      "step": 14674
    },
    {
      "epoch": 0.5255430014145791,
      "grad_norm": 1.6215018033981323,
      "learning_rate": 9.658475495707157e-05,
      "loss": 1.6439,
      "step": 14675
    },
    {
      "epoch": 0.5255788135441474,
      "grad_norm": 1.8243167400360107,
      "learning_rate": 9.657316273359515e-05,
      "loss": 1.2978,
      "step": 14676
    },
    {
      "epoch": 0.5256146256737156,
      "grad_norm": 1.8288040161132812,
      "learning_rate": 9.65615705562224e-05,
      "loss": 1.7493,
      "step": 14677
    },
    {
      "epoch": 0.525650437803284,
      "grad_norm": 1.4830673933029175,
      "learning_rate": 9.654997842510928e-05,
      "loss": 1.687,
      "step": 14678
    },
    {
      "epoch": 0.5256862499328523,
      "grad_norm": 1.6650246381759644,
      "learning_rate": 9.653838634041173e-05,
      "loss": 1.5981,
      "step": 14679
    },
    {
      "epoch": 0.5257220620624206,
      "grad_norm": 1.6408103704452515,
      "learning_rate": 9.652679430228576e-05,
      "loss": 1.7101,
      "step": 14680
    },
    {
      "epoch": 0.5257578741919888,
      "grad_norm": 1.654687523841858,
      "learning_rate": 9.651520231088726e-05,
      "loss": 1.4703,
      "step": 14681
    },
    {
      "epoch": 0.5257936863215571,
      "grad_norm": 1.589911699295044,
      "learning_rate": 9.650361036637225e-05,
      "loss": 1.5057,
      "step": 14682
    },
    {
      "epoch": 0.5258294984511254,
      "grad_norm": 1.8815149068832397,
      "learning_rate": 9.649201846889663e-05,
      "loss": 1.6379,
      "step": 14683
    },
    {
      "epoch": 0.5258653105806936,
      "grad_norm": 1.7435766458511353,
      "learning_rate": 9.64804266186164e-05,
      "loss": 1.4799,
      "step": 14684
    },
    {
      "epoch": 0.525901122710262,
      "grad_norm": 1.667809247970581,
      "learning_rate": 9.646883481568748e-05,
      "loss": 1.5879,
      "step": 14685
    },
    {
      "epoch": 0.5259369348398303,
      "grad_norm": 1.662645936012268,
      "learning_rate": 9.645724306026582e-05,
      "loss": 1.5564,
      "step": 14686
    },
    {
      "epoch": 0.5259727469693986,
      "grad_norm": 1.785279631614685,
      "learning_rate": 9.644565135250739e-05,
      "loss": 1.2906,
      "step": 14687
    },
    {
      "epoch": 0.5260085590989668,
      "grad_norm": 2.513857364654541,
      "learning_rate": 9.643405969256814e-05,
      "loss": 1.3403,
      "step": 14688
    },
    {
      "epoch": 0.5260443712285351,
      "grad_norm": 2.5112111568450928,
      "learning_rate": 9.642246808060401e-05,
      "loss": 1.6178,
      "step": 14689
    },
    {
      "epoch": 0.5260801833581034,
      "grad_norm": 1.8668169975280762,
      "learning_rate": 9.641087651677096e-05,
      "loss": 1.6152,
      "step": 14690
    },
    {
      "epoch": 0.5261159954876716,
      "grad_norm": 1.6442675590515137,
      "learning_rate": 9.639928500122495e-05,
      "loss": 1.8965,
      "step": 14691
    },
    {
      "epoch": 0.52615180761724,
      "grad_norm": 1.996891975402832,
      "learning_rate": 9.638769353412189e-05,
      "loss": 1.8066,
      "step": 14692
    },
    {
      "epoch": 0.5261876197468083,
      "grad_norm": 1.669157862663269,
      "learning_rate": 9.637610211561779e-05,
      "loss": 1.584,
      "step": 14693
    },
    {
      "epoch": 0.5262234318763765,
      "grad_norm": 1.7696603536605835,
      "learning_rate": 9.636451074586856e-05,
      "loss": 1.5034,
      "step": 14694
    },
    {
      "epoch": 0.5262592440059448,
      "grad_norm": 1.5234453678131104,
      "learning_rate": 9.63529194250301e-05,
      "loss": 1.4586,
      "step": 14695
    },
    {
      "epoch": 0.5262950561355131,
      "grad_norm": 1.9366183280944824,
      "learning_rate": 9.634132815325844e-05,
      "loss": 1.4498,
      "step": 14696
    },
    {
      "epoch": 0.5263308682650814,
      "grad_norm": 1.821183443069458,
      "learning_rate": 9.632973693070947e-05,
      "loss": 1.1892,
      "step": 14697
    },
    {
      "epoch": 0.5263666803946496,
      "grad_norm": 1.601720929145813,
      "learning_rate": 9.631814575753918e-05,
      "loss": 1.59,
      "step": 14698
    },
    {
      "epoch": 0.526402492524218,
      "grad_norm": 1.829098105430603,
      "learning_rate": 9.630655463390347e-05,
      "loss": 1.2141,
      "step": 14699
    },
    {
      "epoch": 0.5264383046537863,
      "grad_norm": 1.2035573720932007,
      "learning_rate": 9.629496355995831e-05,
      "loss": 1.3387,
      "step": 14700
    },
    {
      "epoch": 0.5264741167833545,
      "grad_norm": 1.415198802947998,
      "learning_rate": 9.628337253585964e-05,
      "loss": 1.3656,
      "step": 14701
    },
    {
      "epoch": 0.5265099289129228,
      "grad_norm": 1.8165593147277832,
      "learning_rate": 9.62717815617634e-05,
      "loss": 1.5449,
      "step": 14702
    },
    {
      "epoch": 0.5265457410424911,
      "grad_norm": 1.64618718624115,
      "learning_rate": 9.626019063782557e-05,
      "loss": 1.4761,
      "step": 14703
    },
    {
      "epoch": 0.5265815531720593,
      "grad_norm": 1.2743486166000366,
      "learning_rate": 9.624859976420196e-05,
      "loss": 1.4644,
      "step": 14704
    },
    {
      "epoch": 0.5266173653016276,
      "grad_norm": 1.5379201173782349,
      "learning_rate": 9.623700894104869e-05,
      "loss": 1.2605,
      "step": 14705
    },
    {
      "epoch": 0.526653177431196,
      "grad_norm": 1.9712167978286743,
      "learning_rate": 9.622541816852153e-05,
      "loss": 1.3814,
      "step": 14706
    },
    {
      "epoch": 0.5266889895607643,
      "grad_norm": 2.3892998695373535,
      "learning_rate": 9.621382744677658e-05,
      "loss": 1.8143,
      "step": 14707
    },
    {
      "epoch": 0.5267248016903325,
      "grad_norm": 1.2501753568649292,
      "learning_rate": 9.620223677596962e-05,
      "loss": 1.4968,
      "step": 14708
    },
    {
      "epoch": 0.5267606138199008,
      "grad_norm": 1.9475300312042236,
      "learning_rate": 9.619064615625671e-05,
      "loss": 1.5235,
      "step": 14709
    },
    {
      "epoch": 0.5267964259494691,
      "grad_norm": 2.2302019596099854,
      "learning_rate": 9.617905558779373e-05,
      "loss": 1.6564,
      "step": 14710
    },
    {
      "epoch": 0.5268322380790373,
      "grad_norm": 1.5667731761932373,
      "learning_rate": 9.616746507073664e-05,
      "loss": 1.4212,
      "step": 14711
    },
    {
      "epoch": 0.5268680502086056,
      "grad_norm": 2.0810258388519287,
      "learning_rate": 9.61558746052414e-05,
      "loss": 1.8592,
      "step": 14712
    },
    {
      "epoch": 0.526903862338174,
      "grad_norm": 1.1688623428344727,
      "learning_rate": 9.614428419146381e-05,
      "loss": 1.5327,
      "step": 14713
    },
    {
      "epoch": 0.5269396744677423,
      "grad_norm": 2.2579002380371094,
      "learning_rate": 9.613269382956e-05,
      "loss": 1.5626,
      "step": 14714
    },
    {
      "epoch": 0.5269754865973105,
      "grad_norm": 1.4683496952056885,
      "learning_rate": 9.612110351968573e-05,
      "loss": 1.3157,
      "step": 14715
    },
    {
      "epoch": 0.5270112987268788,
      "grad_norm": 1.8713817596435547,
      "learning_rate": 9.610951326199707e-05,
      "loss": 1.4675,
      "step": 14716
    },
    {
      "epoch": 0.5270471108564471,
      "grad_norm": 2.5871567726135254,
      "learning_rate": 9.609792305664984e-05,
      "loss": 1.8185,
      "step": 14717
    },
    {
      "epoch": 0.5270829229860153,
      "grad_norm": 1.4792441129684448,
      "learning_rate": 9.608633290380008e-05,
      "loss": 1.4989,
      "step": 14718
    },
    {
      "epoch": 0.5271187351155836,
      "grad_norm": 1.3156225681304932,
      "learning_rate": 9.60747428036036e-05,
      "loss": 1.3848,
      "step": 14719
    },
    {
      "epoch": 0.527154547245152,
      "grad_norm": 1.3840765953063965,
      "learning_rate": 9.606315275621644e-05,
      "loss": 1.5157,
      "step": 14720
    },
    {
      "epoch": 0.5271903593747203,
      "grad_norm": 1.6863317489624023,
      "learning_rate": 9.605156276179447e-05,
      "loss": 1.5742,
      "step": 14721
    },
    {
      "epoch": 0.5272261715042885,
      "grad_norm": 1.9276484251022339,
      "learning_rate": 9.60399728204936e-05,
      "loss": 1.7787,
      "step": 14722
    },
    {
      "epoch": 0.5272619836338568,
      "grad_norm": 1.3478091955184937,
      "learning_rate": 9.602838293246984e-05,
      "loss": 1.5074,
      "step": 14723
    },
    {
      "epoch": 0.5272977957634251,
      "grad_norm": 1.4894832372665405,
      "learning_rate": 9.6016793097879e-05,
      "loss": 1.3248,
      "step": 14724
    },
    {
      "epoch": 0.5273336078929933,
      "grad_norm": 1.655637502670288,
      "learning_rate": 9.600520331687713e-05,
      "loss": 1.6261,
      "step": 14725
    },
    {
      "epoch": 0.5273694200225616,
      "grad_norm": 1.4634572267532349,
      "learning_rate": 9.599361358962005e-05,
      "loss": 1.3915,
      "step": 14726
    },
    {
      "epoch": 0.52740523215213,
      "grad_norm": 1.8374460935592651,
      "learning_rate": 9.598202391626379e-05,
      "loss": 1.5723,
      "step": 14727
    },
    {
      "epoch": 0.5274410442816982,
      "grad_norm": 1.6577955484390259,
      "learning_rate": 9.597043429696413e-05,
      "loss": 1.5211,
      "step": 14728
    },
    {
      "epoch": 0.5274768564112665,
      "grad_norm": 1.743890404701233,
      "learning_rate": 9.595884473187716e-05,
      "loss": 1.6837,
      "step": 14729
    },
    {
      "epoch": 0.5275126685408348,
      "grad_norm": 1.3714507818222046,
      "learning_rate": 9.594725522115871e-05,
      "loss": 1.5259,
      "step": 14730
    },
    {
      "epoch": 0.527548480670403,
      "grad_norm": 1.3609662055969238,
      "learning_rate": 9.593566576496468e-05,
      "loss": 1.5633,
      "step": 14731
    },
    {
      "epoch": 0.5275842927999713,
      "grad_norm": 1.580083966255188,
      "learning_rate": 9.592407636345104e-05,
      "loss": 1.3267,
      "step": 14732
    },
    {
      "epoch": 0.5276201049295396,
      "grad_norm": 1.6475521326065063,
      "learning_rate": 9.591248701677368e-05,
      "loss": 1.6335,
      "step": 14733
    },
    {
      "epoch": 0.527655917059108,
      "grad_norm": 1.232351541519165,
      "learning_rate": 9.590089772508856e-05,
      "loss": 1.3328,
      "step": 14734
    },
    {
      "epoch": 0.5276917291886762,
      "grad_norm": 1.7131110429763794,
      "learning_rate": 9.588930848855152e-05,
      "loss": 1.8763,
      "step": 14735
    },
    {
      "epoch": 0.5277275413182445,
      "grad_norm": 1.3705060482025146,
      "learning_rate": 9.58777193073186e-05,
      "loss": 1.5272,
      "step": 14736
    },
    {
      "epoch": 0.5277633534478128,
      "grad_norm": 2.4891092777252197,
      "learning_rate": 9.58661301815456e-05,
      "loss": 1.6995,
      "step": 14737
    },
    {
      "epoch": 0.527799165577381,
      "grad_norm": 1.6959772109985352,
      "learning_rate": 9.585454111138853e-05,
      "loss": 1.2758,
      "step": 14738
    },
    {
      "epoch": 0.5278349777069493,
      "grad_norm": 1.470699667930603,
      "learning_rate": 9.584295209700326e-05,
      "loss": 1.6098,
      "step": 14739
    },
    {
      "epoch": 0.5278707898365176,
      "grad_norm": 2.051999092102051,
      "learning_rate": 9.583136313854567e-05,
      "loss": 1.627,
      "step": 14740
    },
    {
      "epoch": 0.527906601966086,
      "grad_norm": 1.461935043334961,
      "learning_rate": 9.581977423617173e-05,
      "loss": 1.0868,
      "step": 14741
    },
    {
      "epoch": 0.5279424140956542,
      "grad_norm": 1.76284921169281,
      "learning_rate": 9.580818539003733e-05,
      "loss": 1.3957,
      "step": 14742
    },
    {
      "epoch": 0.5279782262252225,
      "grad_norm": 1.7833982706069946,
      "learning_rate": 9.579659660029841e-05,
      "loss": 1.5588,
      "step": 14743
    },
    {
      "epoch": 0.5280140383547908,
      "grad_norm": 1.697553038597107,
      "learning_rate": 9.578500786711082e-05,
      "loss": 1.6178,
      "step": 14744
    },
    {
      "epoch": 0.528049850484359,
      "grad_norm": 2.319493055343628,
      "learning_rate": 9.577341919063055e-05,
      "loss": 1.2546,
      "step": 14745
    },
    {
      "epoch": 0.5280856626139273,
      "grad_norm": 1.524808406829834,
      "learning_rate": 9.576183057101345e-05,
      "loss": 1.2757,
      "step": 14746
    },
    {
      "epoch": 0.5281214747434956,
      "grad_norm": 1.9178670644760132,
      "learning_rate": 9.575024200841547e-05,
      "loss": 1.4959,
      "step": 14747
    },
    {
      "epoch": 0.528157286873064,
      "grad_norm": 2.4134249687194824,
      "learning_rate": 9.573865350299251e-05,
      "loss": 1.5957,
      "step": 14748
    },
    {
      "epoch": 0.5281930990026322,
      "grad_norm": 1.6927388906478882,
      "learning_rate": 9.572706505490043e-05,
      "loss": 1.6095,
      "step": 14749
    },
    {
      "epoch": 0.5282289111322005,
      "grad_norm": 1.5404099225997925,
      "learning_rate": 9.571547666429521e-05,
      "loss": 1.338,
      "step": 14750
    },
    {
      "epoch": 0.5282647232617688,
      "grad_norm": 1.3156019449234009,
      "learning_rate": 9.57038883313327e-05,
      "loss": 1.345,
      "step": 14751
    },
    {
      "epoch": 0.528300535391337,
      "grad_norm": 1.3669182062149048,
      "learning_rate": 9.569230005616887e-05,
      "loss": 1.6818,
      "step": 14752
    },
    {
      "epoch": 0.5283363475209053,
      "grad_norm": 1.1214356422424316,
      "learning_rate": 9.568071183895954e-05,
      "loss": 1.3239,
      "step": 14753
    },
    {
      "epoch": 0.5283721596504736,
      "grad_norm": 2.038661479949951,
      "learning_rate": 9.56691236798607e-05,
      "loss": 1.3369,
      "step": 14754
    },
    {
      "epoch": 0.528407971780042,
      "grad_norm": 2.319150686264038,
      "learning_rate": 9.565753557902818e-05,
      "loss": 1.7293,
      "step": 14755
    },
    {
      "epoch": 0.5284437839096102,
      "grad_norm": 1.3403654098510742,
      "learning_rate": 9.564594753661796e-05,
      "loss": 1.5201,
      "step": 14756
    },
    {
      "epoch": 0.5284795960391785,
      "grad_norm": 1.679591178894043,
      "learning_rate": 9.563435955278587e-05,
      "loss": 1.3043,
      "step": 14757
    },
    {
      "epoch": 0.5285154081687468,
      "grad_norm": 1.7903519868850708,
      "learning_rate": 9.562277162768785e-05,
      "loss": 1.5777,
      "step": 14758
    },
    {
      "epoch": 0.528551220298315,
      "grad_norm": 1.7412021160125732,
      "learning_rate": 9.561118376147979e-05,
      "loss": 1.7425,
      "step": 14759
    },
    {
      "epoch": 0.5285870324278833,
      "grad_norm": 1.7483241558074951,
      "learning_rate": 9.559959595431758e-05,
      "loss": 1.6111,
      "step": 14760
    },
    {
      "epoch": 0.5286228445574516,
      "grad_norm": 1.5660126209259033,
      "learning_rate": 9.558800820635715e-05,
      "loss": 1.5506,
      "step": 14761
    },
    {
      "epoch": 0.52865865668702,
      "grad_norm": 1.5264723300933838,
      "learning_rate": 9.557642051775436e-05,
      "loss": 1.4398,
      "step": 14762
    },
    {
      "epoch": 0.5286944688165882,
      "grad_norm": 1.40674889087677,
      "learning_rate": 9.556483288866515e-05,
      "loss": 1.2538,
      "step": 14763
    },
    {
      "epoch": 0.5287302809461565,
      "grad_norm": 1.4413470029830933,
      "learning_rate": 9.555324531924536e-05,
      "loss": 1.5525,
      "step": 14764
    },
    {
      "epoch": 0.5287660930757248,
      "grad_norm": 1.8058099746704102,
      "learning_rate": 9.554165780965095e-05,
      "loss": 1.6483,
      "step": 14765
    },
    {
      "epoch": 0.528801905205293,
      "grad_norm": 1.4641554355621338,
      "learning_rate": 9.553007036003777e-05,
      "loss": 1.4808,
      "step": 14766
    },
    {
      "epoch": 0.5288377173348613,
      "grad_norm": 1.9598422050476074,
      "learning_rate": 9.551848297056171e-05,
      "loss": 1.3693,
      "step": 14767
    },
    {
      "epoch": 0.5288735294644296,
      "grad_norm": 1.3409147262573242,
      "learning_rate": 9.550689564137872e-05,
      "loss": 1.4255,
      "step": 14768
    },
    {
      "epoch": 0.5289093415939979,
      "grad_norm": 1.9200905561447144,
      "learning_rate": 9.54953083726446e-05,
      "loss": 1.692,
      "step": 14769
    },
    {
      "epoch": 0.5289451537235662,
      "grad_norm": 2.1611762046813965,
      "learning_rate": 9.548372116451535e-05,
      "loss": 1.7349,
      "step": 14770
    },
    {
      "epoch": 0.5289809658531345,
      "grad_norm": 1.589970350265503,
      "learning_rate": 9.547213401714677e-05,
      "loss": 1.4199,
      "step": 14771
    },
    {
      "epoch": 0.5290167779827027,
      "grad_norm": 1.5894039869308472,
      "learning_rate": 9.546054693069481e-05,
      "loss": 1.5671,
      "step": 14772
    },
    {
      "epoch": 0.529052590112271,
      "grad_norm": 1.421728253364563,
      "learning_rate": 9.544895990531532e-05,
      "loss": 1.4939,
      "step": 14773
    },
    {
      "epoch": 0.5290884022418393,
      "grad_norm": 1.5457122325897217,
      "learning_rate": 9.54373729411642e-05,
      "loss": 1.363,
      "step": 14774
    },
    {
      "epoch": 0.5291242143714076,
      "grad_norm": 2.2558932304382324,
      "learning_rate": 9.542578603839736e-05,
      "loss": 1.6362,
      "step": 14775
    },
    {
      "epoch": 0.5291600265009759,
      "grad_norm": 2.0703535079956055,
      "learning_rate": 9.541419919717064e-05,
      "loss": 1.2543,
      "step": 14776
    },
    {
      "epoch": 0.5291958386305442,
      "grad_norm": 1.567732810974121,
      "learning_rate": 9.540261241763999e-05,
      "loss": 1.497,
      "step": 14777
    },
    {
      "epoch": 0.5292316507601125,
      "grad_norm": 1.5150243043899536,
      "learning_rate": 9.539102569996124e-05,
      "loss": 1.4928,
      "step": 14778
    },
    {
      "epoch": 0.5292674628896807,
      "grad_norm": 1.4762626886367798,
      "learning_rate": 9.53794390442903e-05,
      "loss": 1.4142,
      "step": 14779
    },
    {
      "epoch": 0.529303275019249,
      "grad_norm": 1.325579285621643,
      "learning_rate": 9.536785245078304e-05,
      "loss": 1.1916,
      "step": 14780
    },
    {
      "epoch": 0.5293390871488173,
      "grad_norm": 1.704073429107666,
      "learning_rate": 9.535626591959536e-05,
      "loss": 1.7114,
      "step": 14781
    },
    {
      "epoch": 0.5293748992783855,
      "grad_norm": 1.6421252489089966,
      "learning_rate": 9.534467945088313e-05,
      "loss": 1.4885,
      "step": 14782
    },
    {
      "epoch": 0.5294107114079539,
      "grad_norm": 1.4458218812942505,
      "learning_rate": 9.533309304480221e-05,
      "loss": 1.5306,
      "step": 14783
    },
    {
      "epoch": 0.5294465235375222,
      "grad_norm": 1.281543254852295,
      "learning_rate": 9.532150670150854e-05,
      "loss": 1.3572,
      "step": 14784
    },
    {
      "epoch": 0.5294823356670905,
      "grad_norm": 2.6501762866973877,
      "learning_rate": 9.530992042115794e-05,
      "loss": 1.604,
      "step": 14785
    },
    {
      "epoch": 0.5295181477966587,
      "grad_norm": 1.6431045532226562,
      "learning_rate": 9.529833420390631e-05,
      "loss": 1.6053,
      "step": 14786
    },
    {
      "epoch": 0.529553959926227,
      "grad_norm": 1.435895323753357,
      "learning_rate": 9.528674804990954e-05,
      "loss": 1.4593,
      "step": 14787
    },
    {
      "epoch": 0.5295897720557953,
      "grad_norm": 1.460228443145752,
      "learning_rate": 9.527516195932349e-05,
      "loss": 1.246,
      "step": 14788
    },
    {
      "epoch": 0.5296255841853635,
      "grad_norm": 1.7602965831756592,
      "learning_rate": 9.526357593230403e-05,
      "loss": 1.6471,
      "step": 14789
    },
    {
      "epoch": 0.5296613963149319,
      "grad_norm": 2.1996421813964844,
      "learning_rate": 9.525198996900707e-05,
      "loss": 1.6656,
      "step": 14790
    },
    {
      "epoch": 0.5296972084445002,
      "grad_norm": 3.2415080070495605,
      "learning_rate": 9.524040406958847e-05,
      "loss": 1.6736,
      "step": 14791
    },
    {
      "epoch": 0.5297330205740685,
      "grad_norm": 1.3910014629364014,
      "learning_rate": 9.522881823420404e-05,
      "loss": 1.4791,
      "step": 14792
    },
    {
      "epoch": 0.5297688327036367,
      "grad_norm": 1.5591599941253662,
      "learning_rate": 9.521723246300977e-05,
      "loss": 1.5115,
      "step": 14793
    },
    {
      "epoch": 0.529804644833205,
      "grad_norm": 1.4825100898742676,
      "learning_rate": 9.520564675616141e-05,
      "loss": 1.675,
      "step": 14794
    },
    {
      "epoch": 0.5298404569627733,
      "grad_norm": 1.701743245124817,
      "learning_rate": 9.519406111381492e-05,
      "loss": 1.409,
      "step": 14795
    },
    {
      "epoch": 0.5298762690923415,
      "grad_norm": 1.5088096857070923,
      "learning_rate": 9.518247553612613e-05,
      "loss": 1.2716,
      "step": 14796
    },
    {
      "epoch": 0.5299120812219099,
      "grad_norm": 1.9061812162399292,
      "learning_rate": 9.517089002325093e-05,
      "loss": 1.8451,
      "step": 14797
    },
    {
      "epoch": 0.5299478933514782,
      "grad_norm": 1.4009358882904053,
      "learning_rate": 9.515930457534514e-05,
      "loss": 1.1994,
      "step": 14798
    },
    {
      "epoch": 0.5299837054810465,
      "grad_norm": 1.6844242811203003,
      "learning_rate": 9.514771919256472e-05,
      "loss": 1.4261,
      "step": 14799
    },
    {
      "epoch": 0.5300195176106147,
      "grad_norm": 1.8280789852142334,
      "learning_rate": 9.513613387506547e-05,
      "loss": 1.5539,
      "step": 14800
    },
    {
      "epoch": 0.530055329740183,
      "grad_norm": 1.3354803323745728,
      "learning_rate": 9.512454862300321e-05,
      "loss": 1.4928,
      "step": 14801
    },
    {
      "epoch": 0.5300911418697513,
      "grad_norm": 1.5250039100646973,
      "learning_rate": 9.511296343653391e-05,
      "loss": 1.094,
      "step": 14802
    },
    {
      "epoch": 0.5301269539993195,
      "grad_norm": 2.810237169265747,
      "learning_rate": 9.510137831581334e-05,
      "loss": 1.7582,
      "step": 14803
    },
    {
      "epoch": 0.5301627661288879,
      "grad_norm": 1.547545075416565,
      "learning_rate": 9.508979326099747e-05,
      "loss": 1.0475,
      "step": 14804
    },
    {
      "epoch": 0.5301985782584562,
      "grad_norm": 2.176013469696045,
      "learning_rate": 9.507820827224202e-05,
      "loss": 1.4733,
      "step": 14805
    },
    {
      "epoch": 0.5302343903880244,
      "grad_norm": 1.2221146821975708,
      "learning_rate": 9.5066623349703e-05,
      "loss": 1.1236,
      "step": 14806
    },
    {
      "epoch": 0.5302702025175927,
      "grad_norm": 1.900315523147583,
      "learning_rate": 9.505503849353613e-05,
      "loss": 1.6962,
      "step": 14807
    },
    {
      "epoch": 0.530306014647161,
      "grad_norm": 1.854783296585083,
      "learning_rate": 9.504345370389739e-05,
      "loss": 1.5795,
      "step": 14808
    },
    {
      "epoch": 0.5303418267767293,
      "grad_norm": 1.4548566341400146,
      "learning_rate": 9.50318689809426e-05,
      "loss": 1.5435,
      "step": 14809
    },
    {
      "epoch": 0.5303776389062975,
      "grad_norm": 1.689694881439209,
      "learning_rate": 9.502028432482755e-05,
      "loss": 1.5173,
      "step": 14810
    },
    {
      "epoch": 0.5304134510358659,
      "grad_norm": 1.620489478111267,
      "learning_rate": 9.50086997357082e-05,
      "loss": 1.2642,
      "step": 14811
    },
    {
      "epoch": 0.5304492631654342,
      "grad_norm": 2.3089258670806885,
      "learning_rate": 9.499711521374031e-05,
      "loss": 1.5737,
      "step": 14812
    },
    {
      "epoch": 0.5304850752950024,
      "grad_norm": 2.3097901344299316,
      "learning_rate": 9.498553075907985e-05,
      "loss": 1.688,
      "step": 14813
    },
    {
      "epoch": 0.5305208874245707,
      "grad_norm": 1.4224505424499512,
      "learning_rate": 9.497394637188251e-05,
      "loss": 1.4562,
      "step": 14814
    },
    {
      "epoch": 0.530556699554139,
      "grad_norm": 1.8305726051330566,
      "learning_rate": 9.496236205230433e-05,
      "loss": 1.4744,
      "step": 14815
    },
    {
      "epoch": 0.5305925116837072,
      "grad_norm": 1.4154525995254517,
      "learning_rate": 9.4950777800501e-05,
      "loss": 1.4215,
      "step": 14816
    },
    {
      "epoch": 0.5306283238132755,
      "grad_norm": 1.8109185695648193,
      "learning_rate": 9.49391936166285e-05,
      "loss": 1.3802,
      "step": 14817
    },
    {
      "epoch": 0.5306641359428439,
      "grad_norm": 1.5938786268234253,
      "learning_rate": 9.492760950084261e-05,
      "loss": 1.4896,
      "step": 14818
    },
    {
      "epoch": 0.5306999480724122,
      "grad_norm": 1.4196759462356567,
      "learning_rate": 9.491602545329916e-05,
      "loss": 1.4166,
      "step": 14819
    },
    {
      "epoch": 0.5307357602019804,
      "grad_norm": 1.5144906044006348,
      "learning_rate": 9.490444147415407e-05,
      "loss": 1.3838,
      "step": 14820
    },
    {
      "epoch": 0.5307715723315487,
      "grad_norm": 2.2972259521484375,
      "learning_rate": 9.489285756356307e-05,
      "loss": 1.5379,
      "step": 14821
    },
    {
      "epoch": 0.530807384461117,
      "grad_norm": 1.4997377395629883,
      "learning_rate": 9.488127372168218e-05,
      "loss": 1.4986,
      "step": 14822
    },
    {
      "epoch": 0.5308431965906852,
      "grad_norm": 2.4222404956817627,
      "learning_rate": 9.486968994866708e-05,
      "loss": 1.652,
      "step": 14823
    },
    {
      "epoch": 0.5308790087202535,
      "grad_norm": 1.8816639184951782,
      "learning_rate": 9.485810624467372e-05,
      "loss": 1.3654,
      "step": 14824
    },
    {
      "epoch": 0.5309148208498219,
      "grad_norm": 1.8047064542770386,
      "learning_rate": 9.484652260985787e-05,
      "loss": 1.4498,
      "step": 14825
    },
    {
      "epoch": 0.5309506329793902,
      "grad_norm": 1.329789638519287,
      "learning_rate": 9.483493904437548e-05,
      "loss": 1.3135,
      "step": 14826
    },
    {
      "epoch": 0.5309864451089584,
      "grad_norm": 3.3496294021606445,
      "learning_rate": 9.482335554838229e-05,
      "loss": 1.1479,
      "step": 14827
    },
    {
      "epoch": 0.5310222572385267,
      "grad_norm": 1.7045812606811523,
      "learning_rate": 9.481177212203415e-05,
      "loss": 1.8932,
      "step": 14828
    },
    {
      "epoch": 0.531058069368095,
      "grad_norm": 1.3624588251113892,
      "learning_rate": 9.480018876548695e-05,
      "loss": 1.4705,
      "step": 14829
    },
    {
      "epoch": 0.5310938814976632,
      "grad_norm": 1.6676069498062134,
      "learning_rate": 9.478860547889647e-05,
      "loss": 1.5551,
      "step": 14830
    },
    {
      "epoch": 0.5311296936272315,
      "grad_norm": 1.7879300117492676,
      "learning_rate": 9.477702226241862e-05,
      "loss": 1.5219,
      "step": 14831
    },
    {
      "epoch": 0.5311655057567999,
      "grad_norm": 1.5257636308670044,
      "learning_rate": 9.476543911620918e-05,
      "loss": 1.5649,
      "step": 14832
    },
    {
      "epoch": 0.5312013178863682,
      "grad_norm": 2.001162052154541,
      "learning_rate": 9.4753856040424e-05,
      "loss": 1.549,
      "step": 14833
    },
    {
      "epoch": 0.5312371300159364,
      "grad_norm": 2.1974189281463623,
      "learning_rate": 9.47422730352189e-05,
      "loss": 1.283,
      "step": 14834
    },
    {
      "epoch": 0.5312729421455047,
      "grad_norm": 1.6734133958816528,
      "learning_rate": 9.47306901007498e-05,
      "loss": 1.4485,
      "step": 14835
    },
    {
      "epoch": 0.531308754275073,
      "grad_norm": 1.9478635787963867,
      "learning_rate": 9.471910723717243e-05,
      "loss": 1.4408,
      "step": 14836
    },
    {
      "epoch": 0.5313445664046412,
      "grad_norm": 2.0421502590179443,
      "learning_rate": 9.470752444464265e-05,
      "loss": 1.6513,
      "step": 14837
    },
    {
      "epoch": 0.5313803785342095,
      "grad_norm": 1.3936954736709595,
      "learning_rate": 9.469594172331631e-05,
      "loss": 1.3181,
      "step": 14838
    },
    {
      "epoch": 0.5314161906637779,
      "grad_norm": 1.9355803728103638,
      "learning_rate": 9.468435907334922e-05,
      "loss": 1.3821,
      "step": 14839
    },
    {
      "epoch": 0.5314520027933461,
      "grad_norm": 1.3044540882110596,
      "learning_rate": 9.467277649489725e-05,
      "loss": 1.381,
      "step": 14840
    },
    {
      "epoch": 0.5314878149229144,
      "grad_norm": 1.856377124786377,
      "learning_rate": 9.466119398811617e-05,
      "loss": 1.5137,
      "step": 14841
    },
    {
      "epoch": 0.5315236270524827,
      "grad_norm": 1.3732273578643799,
      "learning_rate": 9.464961155316187e-05,
      "loss": 1.4637,
      "step": 14842
    },
    {
      "epoch": 0.531559439182051,
      "grad_norm": 1.383651852607727,
      "learning_rate": 9.463802919019011e-05,
      "loss": 1.2581,
      "step": 14843
    },
    {
      "epoch": 0.5315952513116192,
      "grad_norm": 2.1127378940582275,
      "learning_rate": 9.462644689935678e-05,
      "loss": 1.6943,
      "step": 14844
    },
    {
      "epoch": 0.5316310634411875,
      "grad_norm": 1.6620867252349854,
      "learning_rate": 9.461486468081768e-05,
      "loss": 1.5948,
      "step": 14845
    },
    {
      "epoch": 0.5316668755707559,
      "grad_norm": 1.6653188467025757,
      "learning_rate": 9.460328253472859e-05,
      "loss": 1.7102,
      "step": 14846
    },
    {
      "epoch": 0.5317026877003241,
      "grad_norm": 1.5845062732696533,
      "learning_rate": 9.459170046124542e-05,
      "loss": 1.6149,
      "step": 14847
    },
    {
      "epoch": 0.5317384998298924,
      "grad_norm": 1.3799561262130737,
      "learning_rate": 9.458011846052391e-05,
      "loss": 1.5361,
      "step": 14848
    },
    {
      "epoch": 0.5317743119594607,
      "grad_norm": 2.455580949783325,
      "learning_rate": 9.456853653271992e-05,
      "loss": 1.7296,
      "step": 14849
    },
    {
      "epoch": 0.531810124089029,
      "grad_norm": 1.4650472402572632,
      "learning_rate": 9.455695467798927e-05,
      "loss": 1.4852,
      "step": 14850
    },
    {
      "epoch": 0.5318459362185972,
      "grad_norm": 1.7289412021636963,
      "learning_rate": 9.454537289648779e-05,
      "loss": 1.44,
      "step": 14851
    },
    {
      "epoch": 0.5318817483481655,
      "grad_norm": 1.5133506059646606,
      "learning_rate": 9.453379118837125e-05,
      "loss": 1.3471,
      "step": 14852
    },
    {
      "epoch": 0.5319175604777338,
      "grad_norm": 1.7192788124084473,
      "learning_rate": 9.452220955379553e-05,
      "loss": 1.6764,
      "step": 14853
    },
    {
      "epoch": 0.5319533726073021,
      "grad_norm": 2.162766456604004,
      "learning_rate": 9.45106279929164e-05,
      "loss": 1.3744,
      "step": 14854
    },
    {
      "epoch": 0.5319891847368704,
      "grad_norm": 1.7734508514404297,
      "learning_rate": 9.449904650588968e-05,
      "loss": 1.3988,
      "step": 14855
    },
    {
      "epoch": 0.5320249968664387,
      "grad_norm": 1.7432608604431152,
      "learning_rate": 9.44874650928712e-05,
      "loss": 1.2843,
      "step": 14856
    },
    {
      "epoch": 0.5320608089960069,
      "grad_norm": 1.3881850242614746,
      "learning_rate": 9.447588375401676e-05,
      "loss": 1.3026,
      "step": 14857
    },
    {
      "epoch": 0.5320966211255752,
      "grad_norm": 1.451550006866455,
      "learning_rate": 9.44643024894822e-05,
      "loss": 1.2257,
      "step": 14858
    },
    {
      "epoch": 0.5321324332551435,
      "grad_norm": 1.406640648841858,
      "learning_rate": 9.445272129942329e-05,
      "loss": 1.3289,
      "step": 14859
    },
    {
      "epoch": 0.5321682453847117,
      "grad_norm": 1.8054431676864624,
      "learning_rate": 9.444114018399588e-05,
      "loss": 1.4196,
      "step": 14860
    },
    {
      "epoch": 0.5322040575142801,
      "grad_norm": 1.7391663789749146,
      "learning_rate": 9.442955914335573e-05,
      "loss": 1.5092,
      "step": 14861
    },
    {
      "epoch": 0.5322398696438484,
      "grad_norm": 1.8281937837600708,
      "learning_rate": 9.441797817765869e-05,
      "loss": 1.2757,
      "step": 14862
    },
    {
      "epoch": 0.5322756817734167,
      "grad_norm": 1.541002869606018,
      "learning_rate": 9.440639728706058e-05,
      "loss": 1.2199,
      "step": 14863
    },
    {
      "epoch": 0.5323114939029849,
      "grad_norm": 1.8189629316329956,
      "learning_rate": 9.439481647171714e-05,
      "loss": 1.6036,
      "step": 14864
    },
    {
      "epoch": 0.5323473060325532,
      "grad_norm": 1.3375712633132935,
      "learning_rate": 9.438323573178424e-05,
      "loss": 1.4157,
      "step": 14865
    },
    {
      "epoch": 0.5323831181621215,
      "grad_norm": 2.0243418216705322,
      "learning_rate": 9.437165506741764e-05,
      "loss": 1.5781,
      "step": 14866
    },
    {
      "epoch": 0.5324189302916897,
      "grad_norm": 1.824872612953186,
      "learning_rate": 9.436007447877316e-05,
      "loss": 1.1952,
      "step": 14867
    },
    {
      "epoch": 0.5324547424212581,
      "grad_norm": 1.589608073234558,
      "learning_rate": 9.43484939660066e-05,
      "loss": 1.1883,
      "step": 14868
    },
    {
      "epoch": 0.5324905545508264,
      "grad_norm": 1.530319333076477,
      "learning_rate": 9.433691352927378e-05,
      "loss": 1.3574,
      "step": 14869
    },
    {
      "epoch": 0.5325263666803947,
      "grad_norm": 1.7386776208877563,
      "learning_rate": 9.43253331687305e-05,
      "loss": 1.5034,
      "step": 14870
    },
    {
      "epoch": 0.5325621788099629,
      "grad_norm": 1.9396528005599976,
      "learning_rate": 9.43137528845325e-05,
      "loss": 1.5202,
      "step": 14871
    },
    {
      "epoch": 0.5325979909395312,
      "grad_norm": 1.8094663619995117,
      "learning_rate": 9.430217267683566e-05,
      "loss": 1.7822,
      "step": 14872
    },
    {
      "epoch": 0.5326338030690995,
      "grad_norm": 1.1985933780670166,
      "learning_rate": 9.42905925457957e-05,
      "loss": 1.4856,
      "step": 14873
    },
    {
      "epoch": 0.5326696151986677,
      "grad_norm": 1.9986737966537476,
      "learning_rate": 9.427901249156847e-05,
      "loss": 1.5435,
      "step": 14874
    },
    {
      "epoch": 0.5327054273282361,
      "grad_norm": 2.1009483337402344,
      "learning_rate": 9.426743251430974e-05,
      "loss": 1.5721,
      "step": 14875
    },
    {
      "epoch": 0.5327412394578044,
      "grad_norm": 1.6523008346557617,
      "learning_rate": 9.425585261417533e-05,
      "loss": 1.4228,
      "step": 14876
    },
    {
      "epoch": 0.5327770515873727,
      "grad_norm": 1.8854312896728516,
      "learning_rate": 9.424427279132099e-05,
      "loss": 1.4939,
      "step": 14877
    },
    {
      "epoch": 0.5328128637169409,
      "grad_norm": 2.156968355178833,
      "learning_rate": 9.423269304590256e-05,
      "loss": 0.997,
      "step": 14878
    },
    {
      "epoch": 0.5328486758465092,
      "grad_norm": 1.7135142087936401,
      "learning_rate": 9.42211133780758e-05,
      "loss": 1.5855,
      "step": 14879
    },
    {
      "epoch": 0.5328844879760775,
      "grad_norm": 1.341581106185913,
      "learning_rate": 9.420953378799649e-05,
      "loss": 1.3719,
      "step": 14880
    },
    {
      "epoch": 0.5329203001056457,
      "grad_norm": 1.8618860244750977,
      "learning_rate": 9.419795427582044e-05,
      "loss": 1.3623,
      "step": 14881
    },
    {
      "epoch": 0.5329561122352141,
      "grad_norm": 2.2615368366241455,
      "learning_rate": 9.418637484170344e-05,
      "loss": 1.4273,
      "step": 14882
    },
    {
      "epoch": 0.5329919243647824,
      "grad_norm": 1.9147205352783203,
      "learning_rate": 9.417479548580126e-05,
      "loss": 1.5137,
      "step": 14883
    },
    {
      "epoch": 0.5330277364943506,
      "grad_norm": 1.4778187274932861,
      "learning_rate": 9.416321620826968e-05,
      "loss": 1.8574,
      "step": 14884
    },
    {
      "epoch": 0.5330635486239189,
      "grad_norm": 1.726284384727478,
      "learning_rate": 9.415163700926451e-05,
      "loss": 1.362,
      "step": 14885
    },
    {
      "epoch": 0.5330993607534872,
      "grad_norm": 2.3607828617095947,
      "learning_rate": 9.414005788894151e-05,
      "loss": 1.7774,
      "step": 14886
    },
    {
      "epoch": 0.5331351728830555,
      "grad_norm": 1.7209479808807373,
      "learning_rate": 9.41284788474565e-05,
      "loss": 1.3326,
      "step": 14887
    },
    {
      "epoch": 0.5331709850126237,
      "grad_norm": 1.455073356628418,
      "learning_rate": 9.411689988496526e-05,
      "loss": 1.7173,
      "step": 14888
    },
    {
      "epoch": 0.5332067971421921,
      "grad_norm": 1.9332654476165771,
      "learning_rate": 9.410532100162344e-05,
      "loss": 1.4577,
      "step": 14889
    },
    {
      "epoch": 0.5332426092717604,
      "grad_norm": 1.50914466381073,
      "learning_rate": 9.409374219758702e-05,
      "loss": 1.2237,
      "step": 14890
    },
    {
      "epoch": 0.5332784214013286,
      "grad_norm": 1.5229016542434692,
      "learning_rate": 9.408216347301161e-05,
      "loss": 1.5624,
      "step": 14891
    },
    {
      "epoch": 0.5333142335308969,
      "grad_norm": 2.082638740539551,
      "learning_rate": 9.40705848280531e-05,
      "loss": 1.3889,
      "step": 14892
    },
    {
      "epoch": 0.5333500456604652,
      "grad_norm": 1.3961759805679321,
      "learning_rate": 9.40590062628672e-05,
      "loss": 1.5499,
      "step": 14893
    },
    {
      "epoch": 0.5333858577900334,
      "grad_norm": 1.6336990594863892,
      "learning_rate": 9.404742777760974e-05,
      "loss": 1.3981,
      "step": 14894
    },
    {
      "epoch": 0.5334216699196017,
      "grad_norm": 1.8657147884368896,
      "learning_rate": 9.403584937243642e-05,
      "loss": 1.5618,
      "step": 14895
    },
    {
      "epoch": 0.5334574820491701,
      "grad_norm": 2.1878414154052734,
      "learning_rate": 9.402427104750308e-05,
      "loss": 1.4487,
      "step": 14896
    },
    {
      "epoch": 0.5334932941787384,
      "grad_norm": 1.487822413444519,
      "learning_rate": 9.401269280296549e-05,
      "loss": 1.2005,
      "step": 14897
    },
    {
      "epoch": 0.5335291063083066,
      "grad_norm": 1.6004787683486938,
      "learning_rate": 9.400111463897932e-05,
      "loss": 1.5498,
      "step": 14898
    },
    {
      "epoch": 0.5335649184378749,
      "grad_norm": 1.8500304222106934,
      "learning_rate": 9.39895365557005e-05,
      "loss": 1.4665,
      "step": 14899
    },
    {
      "epoch": 0.5336007305674432,
      "grad_norm": 1.6620436906814575,
      "learning_rate": 9.397795855328464e-05,
      "loss": 1.6153,
      "step": 14900
    },
    {
      "epoch": 0.5336365426970114,
      "grad_norm": 1.6935945749282837,
      "learning_rate": 9.396638063188764e-05,
      "loss": 1.6425,
      "step": 14901
    },
    {
      "epoch": 0.5336723548265797,
      "grad_norm": 1.664432168006897,
      "learning_rate": 9.395480279166514e-05,
      "loss": 1.6335,
      "step": 14902
    },
    {
      "epoch": 0.5337081669561481,
      "grad_norm": 2.122021436691284,
      "learning_rate": 9.394322503277305e-05,
      "loss": 1.3633,
      "step": 14903
    },
    {
      "epoch": 0.5337439790857164,
      "grad_norm": 1.5068094730377197,
      "learning_rate": 9.393164735536696e-05,
      "loss": 1.3178,
      "step": 14904
    },
    {
      "epoch": 0.5337797912152846,
      "grad_norm": 1.499644160270691,
      "learning_rate": 9.39200697596028e-05,
      "loss": 1.3545,
      "step": 14905
    },
    {
      "epoch": 0.5338156033448529,
      "grad_norm": 1.8724576234817505,
      "learning_rate": 9.390849224563627e-05,
      "loss": 1.5587,
      "step": 14906
    },
    {
      "epoch": 0.5338514154744212,
      "grad_norm": 2.1973133087158203,
      "learning_rate": 9.389691481362304e-05,
      "loss": 1.4083,
      "step": 14907
    },
    {
      "epoch": 0.5338872276039894,
      "grad_norm": 2.883030414581299,
      "learning_rate": 9.388533746371904e-05,
      "loss": 1.3479,
      "step": 14908
    },
    {
      "epoch": 0.5339230397335577,
      "grad_norm": 2.0845448970794678,
      "learning_rate": 9.387376019607985e-05,
      "loss": 1.248,
      "step": 14909
    },
    {
      "epoch": 0.5339588518631261,
      "grad_norm": 1.4338700771331787,
      "learning_rate": 9.386218301086139e-05,
      "loss": 1.3079,
      "step": 14910
    },
    {
      "epoch": 0.5339946639926944,
      "grad_norm": 2.109936475753784,
      "learning_rate": 9.385060590821929e-05,
      "loss": 1.4467,
      "step": 14911
    },
    {
      "epoch": 0.5340304761222626,
      "grad_norm": 1.9435588121414185,
      "learning_rate": 9.38390288883094e-05,
      "loss": 1.5993,
      "step": 14912
    },
    {
      "epoch": 0.5340662882518309,
      "grad_norm": 1.5473748445510864,
      "learning_rate": 9.382745195128736e-05,
      "loss": 1.3432,
      "step": 14913
    },
    {
      "epoch": 0.5341021003813992,
      "grad_norm": 1.929419994354248,
      "learning_rate": 9.381587509730907e-05,
      "loss": 1.627,
      "step": 14914
    },
    {
      "epoch": 0.5341379125109674,
      "grad_norm": 1.4407709836959839,
      "learning_rate": 9.380429832653017e-05,
      "loss": 1.2589,
      "step": 14915
    },
    {
      "epoch": 0.5341737246405357,
      "grad_norm": 1.594947099685669,
      "learning_rate": 9.379272163910643e-05,
      "loss": 1.3406,
      "step": 14916
    },
    {
      "epoch": 0.5342095367701041,
      "grad_norm": 1.8034594058990479,
      "learning_rate": 9.378114503519364e-05,
      "loss": 1.2251,
      "step": 14917
    },
    {
      "epoch": 0.5342453488996723,
      "grad_norm": 1.5109189748764038,
      "learning_rate": 9.376956851494747e-05,
      "loss": 1.5412,
      "step": 14918
    },
    {
      "epoch": 0.5342811610292406,
      "grad_norm": 1.4823503494262695,
      "learning_rate": 9.375799207852379e-05,
      "loss": 1.5642,
      "step": 14919
    },
    {
      "epoch": 0.5343169731588089,
      "grad_norm": 1.2817720174789429,
      "learning_rate": 9.37464157260782e-05,
      "loss": 1.4638,
      "step": 14920
    },
    {
      "epoch": 0.5343527852883772,
      "grad_norm": 1.9059113264083862,
      "learning_rate": 9.37348394577666e-05,
      "loss": 1.6068,
      "step": 14921
    },
    {
      "epoch": 0.5343885974179454,
      "grad_norm": 1.876328945159912,
      "learning_rate": 9.372326327374459e-05,
      "loss": 1.377,
      "step": 14922
    },
    {
      "epoch": 0.5344244095475137,
      "grad_norm": 2.0040924549102783,
      "learning_rate": 9.371168717416803e-05,
      "loss": 1.3396,
      "step": 14923
    },
    {
      "epoch": 0.5344602216770821,
      "grad_norm": 2.007840156555176,
      "learning_rate": 9.370011115919258e-05,
      "loss": 1.6018,
      "step": 14924
    },
    {
      "epoch": 0.5344960338066503,
      "grad_norm": 1.4200763702392578,
      "learning_rate": 9.368853522897399e-05,
      "loss": 1.435,
      "step": 14925
    },
    {
      "epoch": 0.5345318459362186,
      "grad_norm": 1.4511983394622803,
      "learning_rate": 9.367695938366805e-05,
      "loss": 1.2733,
      "step": 14926
    },
    {
      "epoch": 0.5345676580657869,
      "grad_norm": 1.3073828220367432,
      "learning_rate": 9.366538362343043e-05,
      "loss": 1.504,
      "step": 14927
    },
    {
      "epoch": 0.5346034701953551,
      "grad_norm": 1.3790490627288818,
      "learning_rate": 9.365380794841694e-05,
      "loss": 1.4996,
      "step": 14928
    },
    {
      "epoch": 0.5346392823249234,
      "grad_norm": 2.486955165863037,
      "learning_rate": 9.364223235878324e-05,
      "loss": 1.5752,
      "step": 14929
    },
    {
      "epoch": 0.5346750944544917,
      "grad_norm": 2.4846272468566895,
      "learning_rate": 9.363065685468514e-05,
      "loss": 1.585,
      "step": 14930
    },
    {
      "epoch": 0.5347109065840601,
      "grad_norm": 1.5754536390304565,
      "learning_rate": 9.361908143627829e-05,
      "loss": 1.3481,
      "step": 14931
    },
    {
      "epoch": 0.5347467187136283,
      "grad_norm": 1.701556921005249,
      "learning_rate": 9.360750610371852e-05,
      "loss": 2.0244,
      "step": 14932
    },
    {
      "epoch": 0.5347825308431966,
      "grad_norm": 1.8549612760543823,
      "learning_rate": 9.35959308571615e-05,
      "loss": 1.4935,
      "step": 14933
    },
    {
      "epoch": 0.5348183429727649,
      "grad_norm": 1.695146918296814,
      "learning_rate": 9.358435569676295e-05,
      "loss": 1.3858,
      "step": 14934
    },
    {
      "epoch": 0.5348541551023331,
      "grad_norm": 1.631511926651001,
      "learning_rate": 9.357278062267863e-05,
      "loss": 1.59,
      "step": 14935
    },
    {
      "epoch": 0.5348899672319014,
      "grad_norm": 2.117126703262329,
      "learning_rate": 9.356120563506424e-05,
      "loss": 1.409,
      "step": 14936
    },
    {
      "epoch": 0.5349257793614697,
      "grad_norm": 1.4399936199188232,
      "learning_rate": 9.354963073407555e-05,
      "loss": 1.6277,
      "step": 14937
    },
    {
      "epoch": 0.5349615914910381,
      "grad_norm": 1.9248534440994263,
      "learning_rate": 9.353805591986822e-05,
      "loss": 1.2877,
      "step": 14938
    },
    {
      "epoch": 0.5349974036206063,
      "grad_norm": 1.6802445650100708,
      "learning_rate": 9.352648119259804e-05,
      "loss": 1.3423,
      "step": 14939
    },
    {
      "epoch": 0.5350332157501746,
      "grad_norm": 1.4071083068847656,
      "learning_rate": 9.35149065524207e-05,
      "loss": 1.1726,
      "step": 14940
    },
    {
      "epoch": 0.5350690278797429,
      "grad_norm": 2.1739704608917236,
      "learning_rate": 9.350333199949193e-05,
      "loss": 1.7501,
      "step": 14941
    },
    {
      "epoch": 0.5351048400093111,
      "grad_norm": 2.068229913711548,
      "learning_rate": 9.349175753396746e-05,
      "loss": 1.7553,
      "step": 14942
    },
    {
      "epoch": 0.5351406521388794,
      "grad_norm": 2.328054666519165,
      "learning_rate": 9.348018315600297e-05,
      "loss": 1.5384,
      "step": 14943
    },
    {
      "epoch": 0.5351764642684477,
      "grad_norm": 1.6056578159332275,
      "learning_rate": 9.346860886575422e-05,
      "loss": 1.6585,
      "step": 14944
    },
    {
      "epoch": 0.535212276398016,
      "grad_norm": 1.6351563930511475,
      "learning_rate": 9.34570346633769e-05,
      "loss": 1.5569,
      "step": 14945
    },
    {
      "epoch": 0.5352480885275843,
      "grad_norm": 1.8765473365783691,
      "learning_rate": 9.344546054902677e-05,
      "loss": 1.6506,
      "step": 14946
    },
    {
      "epoch": 0.5352839006571526,
      "grad_norm": 2.3817811012268066,
      "learning_rate": 9.343388652285947e-05,
      "loss": 1.5692,
      "step": 14947
    },
    {
      "epoch": 0.5353197127867209,
      "grad_norm": 2.1171789169311523,
      "learning_rate": 9.342231258503079e-05,
      "loss": 1.9199,
      "step": 14948
    },
    {
      "epoch": 0.5353555249162891,
      "grad_norm": 1.4918116331100464,
      "learning_rate": 9.34107387356964e-05,
      "loss": 1.3828,
      "step": 14949
    },
    {
      "epoch": 0.5353913370458574,
      "grad_norm": 1.5187290906906128,
      "learning_rate": 9.339916497501202e-05,
      "loss": 1.5775,
      "step": 14950
    },
    {
      "epoch": 0.5354271491754257,
      "grad_norm": 1.756986379623413,
      "learning_rate": 9.338759130313338e-05,
      "loss": 1.7123,
      "step": 14951
    },
    {
      "epoch": 0.535462961304994,
      "grad_norm": 1.6845223903656006,
      "learning_rate": 9.337601772021612e-05,
      "loss": 1.1958,
      "step": 14952
    },
    {
      "epoch": 0.5354987734345623,
      "grad_norm": 1.5527392625808716,
      "learning_rate": 9.336444422641605e-05,
      "loss": 1.4844,
      "step": 14953
    },
    {
      "epoch": 0.5355345855641306,
      "grad_norm": 1.50359046459198,
      "learning_rate": 9.335287082188878e-05,
      "loss": 1.5718,
      "step": 14954
    },
    {
      "epoch": 0.5355703976936989,
      "grad_norm": 1.854921579360962,
      "learning_rate": 9.334129750679009e-05,
      "loss": 1.6565,
      "step": 14955
    },
    {
      "epoch": 0.5356062098232671,
      "grad_norm": 2.034636974334717,
      "learning_rate": 9.332972428127563e-05,
      "loss": 1.3642,
      "step": 14956
    },
    {
      "epoch": 0.5356420219528354,
      "grad_norm": 1.7532157897949219,
      "learning_rate": 9.331815114550115e-05,
      "loss": 1.5281,
      "step": 14957
    },
    {
      "epoch": 0.5356778340824037,
      "grad_norm": 1.6303654909133911,
      "learning_rate": 9.330657809962231e-05,
      "loss": 1.2697,
      "step": 14958
    },
    {
      "epoch": 0.535713646211972,
      "grad_norm": 1.6121984720230103,
      "learning_rate": 9.329500514379485e-05,
      "loss": 1.4018,
      "step": 14959
    },
    {
      "epoch": 0.5357494583415403,
      "grad_norm": 1.9481133222579956,
      "learning_rate": 9.328343227817443e-05,
      "loss": 1.8712,
      "step": 14960
    },
    {
      "epoch": 0.5357852704711086,
      "grad_norm": 2.3141913414001465,
      "learning_rate": 9.327185950291676e-05,
      "loss": 1.7397,
      "step": 14961
    },
    {
      "epoch": 0.5358210826006768,
      "grad_norm": 2.181881904602051,
      "learning_rate": 9.326028681817755e-05,
      "loss": 1.4634,
      "step": 14962
    },
    {
      "epoch": 0.5358568947302451,
      "grad_norm": 1.6960126161575317,
      "learning_rate": 9.324871422411248e-05,
      "loss": 1.4148,
      "step": 14963
    },
    {
      "epoch": 0.5358927068598134,
      "grad_norm": 1.1461007595062256,
      "learning_rate": 9.323714172087726e-05,
      "loss": 1.4833,
      "step": 14964
    },
    {
      "epoch": 0.5359285189893817,
      "grad_norm": 1.4815764427185059,
      "learning_rate": 9.322556930862757e-05,
      "loss": 1.3105,
      "step": 14965
    },
    {
      "epoch": 0.53596433111895,
      "grad_norm": 1.6154530048370361,
      "learning_rate": 9.321399698751912e-05,
      "loss": 1.2409,
      "step": 14966
    },
    {
      "epoch": 0.5360001432485183,
      "grad_norm": 1.841649055480957,
      "learning_rate": 9.320242475770756e-05,
      "loss": 1.3561,
      "step": 14967
    },
    {
      "epoch": 0.5360359553780866,
      "grad_norm": 2.050635814666748,
      "learning_rate": 9.319085261934864e-05,
      "loss": 1.5387,
      "step": 14968
    },
    {
      "epoch": 0.5360717675076548,
      "grad_norm": 1.3458319902420044,
      "learning_rate": 9.317928057259799e-05,
      "loss": 1.5461,
      "step": 14969
    },
    {
      "epoch": 0.5361075796372231,
      "grad_norm": 1.581529974937439,
      "learning_rate": 9.316770861761132e-05,
      "loss": 1.6266,
      "step": 14970
    },
    {
      "epoch": 0.5361433917667914,
      "grad_norm": 1.65879487991333,
      "learning_rate": 9.315613675454435e-05,
      "loss": 1.7114,
      "step": 14971
    },
    {
      "epoch": 0.5361792038963596,
      "grad_norm": 1.7967455387115479,
      "learning_rate": 9.314456498355269e-05,
      "loss": 1.2497,
      "step": 14972
    },
    {
      "epoch": 0.536215016025928,
      "grad_norm": 1.6957311630249023,
      "learning_rate": 9.313299330479209e-05,
      "loss": 1.3297,
      "step": 14973
    },
    {
      "epoch": 0.5362508281554963,
      "grad_norm": 1.4996039867401123,
      "learning_rate": 9.31214217184182e-05,
      "loss": 1.5683,
      "step": 14974
    },
    {
      "epoch": 0.5362866402850646,
      "grad_norm": 1.9086555242538452,
      "learning_rate": 9.31098502245867e-05,
      "loss": 1.507,
      "step": 14975
    },
    {
      "epoch": 0.5363224524146328,
      "grad_norm": 1.6246367692947388,
      "learning_rate": 9.30982788234533e-05,
      "loss": 1.4698,
      "step": 14976
    },
    {
      "epoch": 0.5363582645442011,
      "grad_norm": 1.8579511642456055,
      "learning_rate": 9.308670751517363e-05,
      "loss": 1.4766,
      "step": 14977
    },
    {
      "epoch": 0.5363940766737694,
      "grad_norm": 2.250375270843506,
      "learning_rate": 9.307513629990342e-05,
      "loss": 1.3034,
      "step": 14978
    },
    {
      "epoch": 0.5364298888033376,
      "grad_norm": 1.5072667598724365,
      "learning_rate": 9.306356517779828e-05,
      "loss": 1.5313,
      "step": 14979
    },
    {
      "epoch": 0.536465700932906,
      "grad_norm": 3.7206661701202393,
      "learning_rate": 9.305199414901397e-05,
      "loss": 1.8558,
      "step": 14980
    },
    {
      "epoch": 0.5365015130624743,
      "grad_norm": 1.4382824897766113,
      "learning_rate": 9.304042321370607e-05,
      "loss": 1.4454,
      "step": 14981
    },
    {
      "epoch": 0.5365373251920426,
      "grad_norm": 2.4771111011505127,
      "learning_rate": 9.302885237203034e-05,
      "loss": 1.402,
      "step": 14982
    },
    {
      "epoch": 0.5365731373216108,
      "grad_norm": 1.8501386642456055,
      "learning_rate": 9.301728162414238e-05,
      "loss": 1.2341,
      "step": 14983
    },
    {
      "epoch": 0.5366089494511791,
      "grad_norm": 1.4732383489608765,
      "learning_rate": 9.30057109701979e-05,
      "loss": 1.5761,
      "step": 14984
    },
    {
      "epoch": 0.5366447615807474,
      "grad_norm": 2.0961062908172607,
      "learning_rate": 9.299414041035259e-05,
      "loss": 1.7277,
      "step": 14985
    },
    {
      "epoch": 0.5366805737103156,
      "grad_norm": 1.619267463684082,
      "learning_rate": 9.298256994476202e-05,
      "loss": 1.5175,
      "step": 14986
    },
    {
      "epoch": 0.536716385839884,
      "grad_norm": 1.4270780086517334,
      "learning_rate": 9.297099957358199e-05,
      "loss": 1.695,
      "step": 14987
    },
    {
      "epoch": 0.5367521979694523,
      "grad_norm": 1.7265422344207764,
      "learning_rate": 9.295942929696801e-05,
      "loss": 1.2141,
      "step": 14988
    },
    {
      "epoch": 0.5367880100990206,
      "grad_norm": 2.4905295372009277,
      "learning_rate": 9.294785911507589e-05,
      "loss": 1.3067,
      "step": 14989
    },
    {
      "epoch": 0.5368238222285888,
      "grad_norm": 1.9801867008209229,
      "learning_rate": 9.29362890280612e-05,
      "loss": 1.8363,
      "step": 14990
    },
    {
      "epoch": 0.5368596343581571,
      "grad_norm": 1.467184066772461,
      "learning_rate": 9.292471903607964e-05,
      "loss": 1.6563,
      "step": 14991
    },
    {
      "epoch": 0.5368954464877254,
      "grad_norm": 1.6070553064346313,
      "learning_rate": 9.291314913928685e-05,
      "loss": 1.35,
      "step": 14992
    },
    {
      "epoch": 0.5369312586172936,
      "grad_norm": 1.391000747680664,
      "learning_rate": 9.290157933783852e-05,
      "loss": 1.2004,
      "step": 14993
    },
    {
      "epoch": 0.536967070746862,
      "grad_norm": 1.8240994215011597,
      "learning_rate": 9.28900096318903e-05,
      "loss": 1.3666,
      "step": 14994
    },
    {
      "epoch": 0.5370028828764303,
      "grad_norm": 1.497835636138916,
      "learning_rate": 9.287844002159776e-05,
      "loss": 1.73,
      "step": 14995
    },
    {
      "epoch": 0.5370386950059985,
      "grad_norm": 1.8223732709884644,
      "learning_rate": 9.286687050711668e-05,
      "loss": 1.3505,
      "step": 14996
    },
    {
      "epoch": 0.5370745071355668,
      "grad_norm": 1.5214042663574219,
      "learning_rate": 9.285530108860262e-05,
      "loss": 1.5422,
      "step": 14997
    },
    {
      "epoch": 0.5371103192651351,
      "grad_norm": 1.55804443359375,
      "learning_rate": 9.284373176621131e-05,
      "loss": 1.5534,
      "step": 14998
    },
    {
      "epoch": 0.5371461313947034,
      "grad_norm": 2.063025712966919,
      "learning_rate": 9.28321625400983e-05,
      "loss": 1.4235,
      "step": 14999
    },
    {
      "epoch": 0.5371819435242716,
      "grad_norm": 1.5878913402557373,
      "learning_rate": 9.282059341041936e-05,
      "loss": 1.5036,
      "step": 15000
    },
    {
      "epoch": 0.53721775565384,
      "grad_norm": 1.506630778312683,
      "learning_rate": 9.280902437733003e-05,
      "loss": 1.208,
      "step": 15001
    },
    {
      "epoch": 0.5372535677834083,
      "grad_norm": 1.781821846961975,
      "learning_rate": 9.279745544098602e-05,
      "loss": 1.4643,
      "step": 15002
    },
    {
      "epoch": 0.5372893799129765,
      "grad_norm": 1.9320119619369507,
      "learning_rate": 9.278588660154298e-05,
      "loss": 1.6491,
      "step": 15003
    },
    {
      "epoch": 0.5373251920425448,
      "grad_norm": 1.846867322921753,
      "learning_rate": 9.277431785915647e-05,
      "loss": 1.3241,
      "step": 15004
    },
    {
      "epoch": 0.5373610041721131,
      "grad_norm": 1.7389922142028809,
      "learning_rate": 9.276274921398225e-05,
      "loss": 1.5716,
      "step": 15005
    },
    {
      "epoch": 0.5373968163016813,
      "grad_norm": 1.6828988790512085,
      "learning_rate": 9.275118066617585e-05,
      "loss": 1.5801,
      "step": 15006
    },
    {
      "epoch": 0.5374326284312496,
      "grad_norm": 1.8095980882644653,
      "learning_rate": 9.273961221589303e-05,
      "loss": 1.3198,
      "step": 15007
    },
    {
      "epoch": 0.537468440560818,
      "grad_norm": 1.7171796560287476,
      "learning_rate": 9.27280438632893e-05,
      "loss": 1.4818,
      "step": 15008
    },
    {
      "epoch": 0.5375042526903863,
      "grad_norm": 1.8513648509979248,
      "learning_rate": 9.271647560852042e-05,
      "loss": 1.6135,
      "step": 15009
    },
    {
      "epoch": 0.5375400648199545,
      "grad_norm": 1.5115734338760376,
      "learning_rate": 9.27049074517419e-05,
      "loss": 1.6142,
      "step": 15010
    },
    {
      "epoch": 0.5375758769495228,
      "grad_norm": 2.058396100997925,
      "learning_rate": 9.26933393931095e-05,
      "loss": 1.4059,
      "step": 15011
    },
    {
      "epoch": 0.5376116890790911,
      "grad_norm": 1.8577311038970947,
      "learning_rate": 9.268177143277877e-05,
      "loss": 1.5906,
      "step": 15012
    },
    {
      "epoch": 0.5376475012086593,
      "grad_norm": 1.5660582780838013,
      "learning_rate": 9.267020357090535e-05,
      "loss": 1.4381,
      "step": 15013
    },
    {
      "epoch": 0.5376833133382276,
      "grad_norm": 1.3787236213684082,
      "learning_rate": 9.265863580764492e-05,
      "loss": 1.4798,
      "step": 15014
    },
    {
      "epoch": 0.537719125467796,
      "grad_norm": 2.0579323768615723,
      "learning_rate": 9.264706814315302e-05,
      "loss": 1.571,
      "step": 15015
    },
    {
      "epoch": 0.5377549375973643,
      "grad_norm": 1.3962444067001343,
      "learning_rate": 9.263550057758539e-05,
      "loss": 1.59,
      "step": 15016
    },
    {
      "epoch": 0.5377907497269325,
      "grad_norm": 1.7206778526306152,
      "learning_rate": 9.262393311109754e-05,
      "loss": 1.3043,
      "step": 15017
    },
    {
      "epoch": 0.5378265618565008,
      "grad_norm": 1.2327122688293457,
      "learning_rate": 9.261236574384523e-05,
      "loss": 1.4594,
      "step": 15018
    },
    {
      "epoch": 0.5378623739860691,
      "grad_norm": 2.1515252590179443,
      "learning_rate": 9.260079847598393e-05,
      "loss": 1.4466,
      "step": 15019
    },
    {
      "epoch": 0.5378981861156373,
      "grad_norm": 1.5963287353515625,
      "learning_rate": 9.258923130766942e-05,
      "loss": 1.4193,
      "step": 15020
    },
    {
      "epoch": 0.5379339982452056,
      "grad_norm": 1.5895057916641235,
      "learning_rate": 9.257766423905722e-05,
      "loss": 1.724,
      "step": 15021
    },
    {
      "epoch": 0.537969810374774,
      "grad_norm": 1.5199759006500244,
      "learning_rate": 9.256609727030294e-05,
      "loss": 1.645,
      "step": 15022
    },
    {
      "epoch": 0.5380056225043423,
      "grad_norm": 1.6282600164413452,
      "learning_rate": 9.255453040156228e-05,
      "loss": 1.43,
      "step": 15023
    },
    {
      "epoch": 0.5380414346339105,
      "grad_norm": 1.6002341508865356,
      "learning_rate": 9.254296363299077e-05,
      "loss": 1.4412,
      "step": 15024
    },
    {
      "epoch": 0.5380772467634788,
      "grad_norm": 2.265721559524536,
      "learning_rate": 9.253139696474409e-05,
      "loss": 1.5447,
      "step": 15025
    },
    {
      "epoch": 0.5381130588930471,
      "grad_norm": 3.0661771297454834,
      "learning_rate": 9.25198303969778e-05,
      "loss": 1.6547,
      "step": 15026
    },
    {
      "epoch": 0.5381488710226153,
      "grad_norm": 2.4014828205108643,
      "learning_rate": 9.250826392984757e-05,
      "loss": 1.3629,
      "step": 15027
    },
    {
      "epoch": 0.5381846831521836,
      "grad_norm": 2.069535732269287,
      "learning_rate": 9.249669756350894e-05,
      "loss": 1.6487,
      "step": 15028
    },
    {
      "epoch": 0.538220495281752,
      "grad_norm": 1.3042799234390259,
      "learning_rate": 9.248513129811765e-05,
      "loss": 1.197,
      "step": 15029
    },
    {
      "epoch": 0.5382563074113202,
      "grad_norm": 1.619163990020752,
      "learning_rate": 9.247356513382917e-05,
      "loss": 1.5524,
      "step": 15030
    },
    {
      "epoch": 0.5382921195408885,
      "grad_norm": 1.8219351768493652,
      "learning_rate": 9.246199907079916e-05,
      "loss": 1.4046,
      "step": 15031
    },
    {
      "epoch": 0.5383279316704568,
      "grad_norm": 1.6462182998657227,
      "learning_rate": 9.245043310918325e-05,
      "loss": 1.5918,
      "step": 15032
    },
    {
      "epoch": 0.538363743800025,
      "grad_norm": 2.164386034011841,
      "learning_rate": 9.2438867249137e-05,
      "loss": 1.681,
      "step": 15033
    },
    {
      "epoch": 0.5383995559295933,
      "grad_norm": 1.3424277305603027,
      "learning_rate": 9.242730149081606e-05,
      "loss": 1.6123,
      "step": 15034
    },
    {
      "epoch": 0.5384353680591616,
      "grad_norm": 1.6956967115402222,
      "learning_rate": 9.241573583437599e-05,
      "loss": 1.6094,
      "step": 15035
    },
    {
      "epoch": 0.53847118018873,
      "grad_norm": 1.7979881763458252,
      "learning_rate": 9.240417027997243e-05,
      "loss": 1.4714,
      "step": 15036
    },
    {
      "epoch": 0.5385069923182982,
      "grad_norm": 1.8987326622009277,
      "learning_rate": 9.239260482776096e-05,
      "loss": 1.5237,
      "step": 15037
    },
    {
      "epoch": 0.5385428044478665,
      "grad_norm": 1.9217215776443481,
      "learning_rate": 9.238103947789718e-05,
      "loss": 1.458,
      "step": 15038
    },
    {
      "epoch": 0.5385786165774348,
      "grad_norm": 2.3088903427124023,
      "learning_rate": 9.236947423053669e-05,
      "loss": 1.6509,
      "step": 15039
    },
    {
      "epoch": 0.538614428707003,
      "grad_norm": 1.4139968156814575,
      "learning_rate": 9.235790908583506e-05,
      "loss": 1.3816,
      "step": 15040
    },
    {
      "epoch": 0.5386502408365713,
      "grad_norm": 2.1365206241607666,
      "learning_rate": 9.234634404394793e-05,
      "loss": 1.4619,
      "step": 15041
    },
    {
      "epoch": 0.5386860529661396,
      "grad_norm": 2.3481082916259766,
      "learning_rate": 9.233477910503083e-05,
      "loss": 1.5677,
      "step": 15042
    },
    {
      "epoch": 0.538721865095708,
      "grad_norm": 2.0316877365112305,
      "learning_rate": 9.232321426923943e-05,
      "loss": 1.3205,
      "step": 15043
    },
    {
      "epoch": 0.5387576772252762,
      "grad_norm": 1.9657565355300903,
      "learning_rate": 9.231164953672926e-05,
      "loss": 1.6539,
      "step": 15044
    },
    {
      "epoch": 0.5387934893548445,
      "grad_norm": 1.8171610832214355,
      "learning_rate": 9.230008490765593e-05,
      "loss": 1.1466,
      "step": 15045
    },
    {
      "epoch": 0.5388293014844128,
      "grad_norm": 1.1965806484222412,
      "learning_rate": 9.228852038217502e-05,
      "loss": 1.2931,
      "step": 15046
    },
    {
      "epoch": 0.538865113613981,
      "grad_norm": 1.6096956729888916,
      "learning_rate": 9.227695596044215e-05,
      "loss": 1.4272,
      "step": 15047
    },
    {
      "epoch": 0.5389009257435493,
      "grad_norm": 2.070911407470703,
      "learning_rate": 9.226539164261286e-05,
      "loss": 1.3034,
      "step": 15048
    },
    {
      "epoch": 0.5389367378731176,
      "grad_norm": 2.4568874835968018,
      "learning_rate": 9.225382742884273e-05,
      "loss": 1.4975,
      "step": 15049
    },
    {
      "epoch": 0.538972550002686,
      "grad_norm": 1.6348614692687988,
      "learning_rate": 9.224226331928738e-05,
      "loss": 1.5438,
      "step": 15050
    },
    {
      "epoch": 0.5390083621322542,
      "grad_norm": 1.6172714233398438,
      "learning_rate": 9.223069931410236e-05,
      "loss": 1.3941,
      "step": 15051
    },
    {
      "epoch": 0.5390441742618225,
      "grad_norm": 1.6091904640197754,
      "learning_rate": 9.221913541344327e-05,
      "loss": 1.4381,
      "step": 15052
    },
    {
      "epoch": 0.5390799863913908,
      "grad_norm": 1.614936351776123,
      "learning_rate": 9.220757161746566e-05,
      "loss": 1.5637,
      "step": 15053
    },
    {
      "epoch": 0.539115798520959,
      "grad_norm": 2.01474928855896,
      "learning_rate": 9.219600792632513e-05,
      "loss": 1.4325,
      "step": 15054
    },
    {
      "epoch": 0.5391516106505273,
      "grad_norm": 1.680156946182251,
      "learning_rate": 9.218444434017724e-05,
      "loss": 1.6222,
      "step": 15055
    },
    {
      "epoch": 0.5391874227800956,
      "grad_norm": 1.7810890674591064,
      "learning_rate": 9.217288085917759e-05,
      "loss": 1.4141,
      "step": 15056
    },
    {
      "epoch": 0.539223234909664,
      "grad_norm": 1.8147783279418945,
      "learning_rate": 9.216131748348174e-05,
      "loss": 1.6173,
      "step": 15057
    },
    {
      "epoch": 0.5392590470392322,
      "grad_norm": 1.4602004289627075,
      "learning_rate": 9.21497542132452e-05,
      "loss": 1.4772,
      "step": 15058
    },
    {
      "epoch": 0.5392948591688005,
      "grad_norm": 1.867448329925537,
      "learning_rate": 9.213819104862365e-05,
      "loss": 1.3885,
      "step": 15059
    },
    {
      "epoch": 0.5393306712983688,
      "grad_norm": 1.4362049102783203,
      "learning_rate": 9.212662798977256e-05,
      "loss": 1.5253,
      "step": 15060
    },
    {
      "epoch": 0.539366483427937,
      "grad_norm": 1.3346989154815674,
      "learning_rate": 9.211506503684755e-05,
      "loss": 1.2639,
      "step": 15061
    },
    {
      "epoch": 0.5394022955575053,
      "grad_norm": 1.570544958114624,
      "learning_rate": 9.210350219000416e-05,
      "loss": 1.6639,
      "step": 15062
    },
    {
      "epoch": 0.5394381076870736,
      "grad_norm": 1.4797194004058838,
      "learning_rate": 9.209193944939798e-05,
      "loss": 1.4074,
      "step": 15063
    },
    {
      "epoch": 0.539473919816642,
      "grad_norm": 1.966138243675232,
      "learning_rate": 9.208037681518454e-05,
      "loss": 1.7929,
      "step": 15064
    },
    {
      "epoch": 0.5395097319462102,
      "grad_norm": 1.6170921325683594,
      "learning_rate": 9.206881428751941e-05,
      "loss": 1.2353,
      "step": 15065
    },
    {
      "epoch": 0.5395455440757785,
      "grad_norm": 1.8537172079086304,
      "learning_rate": 9.205725186655817e-05,
      "loss": 1.466,
      "step": 15066
    },
    {
      "epoch": 0.5395813562053468,
      "grad_norm": 1.2693291902542114,
      "learning_rate": 9.204568955245634e-05,
      "loss": 1.4357,
      "step": 15067
    },
    {
      "epoch": 0.539617168334915,
      "grad_norm": 1.3635295629501343,
      "learning_rate": 9.203412734536951e-05,
      "loss": 1.4153,
      "step": 15068
    },
    {
      "epoch": 0.5396529804644833,
      "grad_norm": 1.594772219657898,
      "learning_rate": 9.202256524545322e-05,
      "loss": 1.3883,
      "step": 15069
    },
    {
      "epoch": 0.5396887925940516,
      "grad_norm": 1.3356691598892212,
      "learning_rate": 9.201100325286302e-05,
      "loss": 1.4835,
      "step": 15070
    },
    {
      "epoch": 0.5397246047236199,
      "grad_norm": 2.0734238624572754,
      "learning_rate": 9.199944136775446e-05,
      "loss": 1.5278,
      "step": 15071
    },
    {
      "epoch": 0.5397604168531882,
      "grad_norm": 1.533120036125183,
      "learning_rate": 9.198787959028312e-05,
      "loss": 1.536,
      "step": 15072
    },
    {
      "epoch": 0.5397962289827565,
      "grad_norm": 1.554632544517517,
      "learning_rate": 9.197631792060453e-05,
      "loss": 1.365,
      "step": 15073
    },
    {
      "epoch": 0.5398320411123247,
      "grad_norm": 1.5368367433547974,
      "learning_rate": 9.196475635887419e-05,
      "loss": 1.6686,
      "step": 15074
    },
    {
      "epoch": 0.539867853241893,
      "grad_norm": 1.6258572340011597,
      "learning_rate": 9.195319490524772e-05,
      "loss": 1.3955,
      "step": 15075
    },
    {
      "epoch": 0.5399036653714613,
      "grad_norm": 1.443794846534729,
      "learning_rate": 9.194163355988062e-05,
      "loss": 1.3147,
      "step": 15076
    },
    {
      "epoch": 0.5399394775010296,
      "grad_norm": 1.916318655014038,
      "learning_rate": 9.193007232292846e-05,
      "loss": 1.7055,
      "step": 15077
    },
    {
      "epoch": 0.5399752896305979,
      "grad_norm": 1.584694504737854,
      "learning_rate": 9.191851119454675e-05,
      "loss": 1.1801,
      "step": 15078
    },
    {
      "epoch": 0.5400111017601662,
      "grad_norm": 1.3532837629318237,
      "learning_rate": 9.190695017489106e-05,
      "loss": 1.3885,
      "step": 15079
    },
    {
      "epoch": 0.5400469138897345,
      "grad_norm": 2.9349710941314697,
      "learning_rate": 9.18953892641169e-05,
      "loss": 1.4769,
      "step": 15080
    },
    {
      "epoch": 0.5400827260193027,
      "grad_norm": 2.0163779258728027,
      "learning_rate": 9.188382846237984e-05,
      "loss": 1.4724,
      "step": 15081
    },
    {
      "epoch": 0.540118538148871,
      "grad_norm": 1.6981626749038696,
      "learning_rate": 9.187226776983543e-05,
      "loss": 1.676,
      "step": 15082
    },
    {
      "epoch": 0.5401543502784393,
      "grad_norm": 1.884413719177246,
      "learning_rate": 9.18607071866391e-05,
      "loss": 1.7854,
      "step": 15083
    },
    {
      "epoch": 0.5401901624080075,
      "grad_norm": 1.4751265048980713,
      "learning_rate": 9.184914671294653e-05,
      "loss": 1.3642,
      "step": 15084
    },
    {
      "epoch": 0.5402259745375759,
      "grad_norm": 1.6983938217163086,
      "learning_rate": 9.18375863489131e-05,
      "loss": 1.3347,
      "step": 15085
    },
    {
      "epoch": 0.5402617866671442,
      "grad_norm": 1.627390742301941,
      "learning_rate": 9.182602609469448e-05,
      "loss": 1.4281,
      "step": 15086
    },
    {
      "epoch": 0.5402975987967125,
      "grad_norm": 1.8508156538009644,
      "learning_rate": 9.18144659504461e-05,
      "loss": 1.2141,
      "step": 15087
    },
    {
      "epoch": 0.5403334109262807,
      "grad_norm": 1.926315188407898,
      "learning_rate": 9.180290591632354e-05,
      "loss": 1.5036,
      "step": 15088
    },
    {
      "epoch": 0.540369223055849,
      "grad_norm": 1.5384167432785034,
      "learning_rate": 9.179134599248228e-05,
      "loss": 1.3299,
      "step": 15089
    },
    {
      "epoch": 0.5404050351854173,
      "grad_norm": 1.4410181045532227,
      "learning_rate": 9.177978617907791e-05,
      "loss": 1.4095,
      "step": 15090
    },
    {
      "epoch": 0.5404408473149855,
      "grad_norm": 1.6018924713134766,
      "learning_rate": 9.176822647626593e-05,
      "loss": 1.627,
      "step": 15091
    },
    {
      "epoch": 0.5404766594445539,
      "grad_norm": 1.6720951795578003,
      "learning_rate": 9.175666688420177e-05,
      "loss": 1.5403,
      "step": 15092
    },
    {
      "epoch": 0.5405124715741222,
      "grad_norm": 1.8459980487823486,
      "learning_rate": 9.17451074030411e-05,
      "loss": 1.4794,
      "step": 15093
    },
    {
      "epoch": 0.5405482837036905,
      "grad_norm": 1.8089991807937622,
      "learning_rate": 9.17335480329393e-05,
      "loss": 1.6141,
      "step": 15094
    },
    {
      "epoch": 0.5405840958332587,
      "grad_norm": 1.9303104877471924,
      "learning_rate": 9.1721988774052e-05,
      "loss": 1.248,
      "step": 15095
    },
    {
      "epoch": 0.540619907962827,
      "grad_norm": 1.3829426765441895,
      "learning_rate": 9.17104296265346e-05,
      "loss": 1.2953,
      "step": 15096
    },
    {
      "epoch": 0.5406557200923953,
      "grad_norm": 1.5334969758987427,
      "learning_rate": 9.169887059054275e-05,
      "loss": 1.4165,
      "step": 15097
    },
    {
      "epoch": 0.5406915322219635,
      "grad_norm": 2.001737117767334,
      "learning_rate": 9.168731166623182e-05,
      "loss": 1.294,
      "step": 15098
    },
    {
      "epoch": 0.5407273443515319,
      "grad_norm": 3.4213063716888428,
      "learning_rate": 9.167575285375744e-05,
      "loss": 1.5401,
      "step": 15099
    },
    {
      "epoch": 0.5407631564811002,
      "grad_norm": 1.7544281482696533,
      "learning_rate": 9.166419415327508e-05,
      "loss": 1.4561,
      "step": 15100
    },
    {
      "epoch": 0.5407989686106685,
      "grad_norm": 1.7456077337265015,
      "learning_rate": 9.165263556494016e-05,
      "loss": 1.1195,
      "step": 15101
    },
    {
      "epoch": 0.5408347807402367,
      "grad_norm": 1.4219002723693848,
      "learning_rate": 9.164107708890835e-05,
      "loss": 1.1713,
      "step": 15102
    },
    {
      "epoch": 0.540870592869805,
      "grad_norm": 1.255926489830017,
      "learning_rate": 9.162951872533498e-05,
      "loss": 1.3403,
      "step": 15103
    },
    {
      "epoch": 0.5409064049993733,
      "grad_norm": 2.029594898223877,
      "learning_rate": 9.161796047437572e-05,
      "loss": 1.5172,
      "step": 15104
    },
    {
      "epoch": 0.5409422171289415,
      "grad_norm": 1.3873099088668823,
      "learning_rate": 9.160640233618591e-05,
      "loss": 1.314,
      "step": 15105
    },
    {
      "epoch": 0.5409780292585099,
      "grad_norm": 1.565388798713684,
      "learning_rate": 9.15948443109212e-05,
      "loss": 1.3721,
      "step": 15106
    },
    {
      "epoch": 0.5410138413880782,
      "grad_norm": 1.4900970458984375,
      "learning_rate": 9.158328639873695e-05,
      "loss": 1.3958,
      "step": 15107
    },
    {
      "epoch": 0.5410496535176464,
      "grad_norm": 1.506885051727295,
      "learning_rate": 9.15717285997888e-05,
      "loss": 1.3126,
      "step": 15108
    },
    {
      "epoch": 0.5410854656472147,
      "grad_norm": 1.5706779956817627,
      "learning_rate": 9.156017091423215e-05,
      "loss": 1.4397,
      "step": 15109
    },
    {
      "epoch": 0.541121277776783,
      "grad_norm": 1.3363198041915894,
      "learning_rate": 9.154861334222248e-05,
      "loss": 1.3809,
      "step": 15110
    },
    {
      "epoch": 0.5411570899063513,
      "grad_norm": 1.466255784034729,
      "learning_rate": 9.153705588391535e-05,
      "loss": 1.0968,
      "step": 15111
    },
    {
      "epoch": 0.5411929020359195,
      "grad_norm": 1.7009021043777466,
      "learning_rate": 9.152549853946615e-05,
      "loss": 1.1897,
      "step": 15112
    },
    {
      "epoch": 0.5412287141654879,
      "grad_norm": 1.6660683155059814,
      "learning_rate": 9.151394130903052e-05,
      "loss": 1.5026,
      "step": 15113
    },
    {
      "epoch": 0.5412645262950562,
      "grad_norm": 1.4293491840362549,
      "learning_rate": 9.15023841927638e-05,
      "loss": 1.7752,
      "step": 15114
    },
    {
      "epoch": 0.5413003384246244,
      "grad_norm": 1.6292088031768799,
      "learning_rate": 9.14908271908216e-05,
      "loss": 1.4279,
      "step": 15115
    },
    {
      "epoch": 0.5413361505541927,
      "grad_norm": 2.2113747596740723,
      "learning_rate": 9.147927030335928e-05,
      "loss": 1.5957,
      "step": 15116
    },
    {
      "epoch": 0.541371962683761,
      "grad_norm": 2.821272611618042,
      "learning_rate": 9.146771353053245e-05,
      "loss": 1.2521,
      "step": 15117
    },
    {
      "epoch": 0.5414077748133292,
      "grad_norm": 1.4644893407821655,
      "learning_rate": 9.14561568724965e-05,
      "loss": 1.1884,
      "step": 15118
    },
    {
      "epoch": 0.5414435869428975,
      "grad_norm": 1.7300174236297607,
      "learning_rate": 9.144460032940693e-05,
      "loss": 1.5423,
      "step": 15119
    },
    {
      "epoch": 0.5414793990724659,
      "grad_norm": 1.2341820001602173,
      "learning_rate": 9.143304390141925e-05,
      "loss": 1.0844,
      "step": 15120
    },
    {
      "epoch": 0.5415152112020342,
      "grad_norm": 1.7158368825912476,
      "learning_rate": 9.142148758868887e-05,
      "loss": 1.6006,
      "step": 15121
    },
    {
      "epoch": 0.5415510233316024,
      "grad_norm": 2.46889328956604,
      "learning_rate": 9.140993139137135e-05,
      "loss": 1.613,
      "step": 15122
    },
    {
      "epoch": 0.5415868354611707,
      "grad_norm": 1.745641827583313,
      "learning_rate": 9.139837530962209e-05,
      "loss": 1.3685,
      "step": 15123
    },
    {
      "epoch": 0.541622647590739,
      "grad_norm": 1.7009758949279785,
      "learning_rate": 9.138681934359663e-05,
      "loss": 1.5774,
      "step": 15124
    },
    {
      "epoch": 0.5416584597203072,
      "grad_norm": 2.010895252227783,
      "learning_rate": 9.137526349345036e-05,
      "loss": 1.421,
      "step": 15125
    },
    {
      "epoch": 0.5416942718498755,
      "grad_norm": 1.514216423034668,
      "learning_rate": 9.136370775933885e-05,
      "loss": 1.4046,
      "step": 15126
    },
    {
      "epoch": 0.5417300839794439,
      "grad_norm": 1.5812219381332397,
      "learning_rate": 9.135215214141751e-05,
      "loss": 1.4894,
      "step": 15127
    },
    {
      "epoch": 0.5417658961090122,
      "grad_norm": 1.7049273252487183,
      "learning_rate": 9.134059663984176e-05,
      "loss": 1.7633,
      "step": 15128
    },
    {
      "epoch": 0.5418017082385804,
      "grad_norm": 1.498794674873352,
      "learning_rate": 9.132904125476715e-05,
      "loss": 1.3923,
      "step": 15129
    },
    {
      "epoch": 0.5418375203681487,
      "grad_norm": 2.9248688220977783,
      "learning_rate": 9.131748598634907e-05,
      "loss": 1.9405,
      "step": 15130
    },
    {
      "epoch": 0.541873332497717,
      "grad_norm": 1.411406397819519,
      "learning_rate": 9.130593083474305e-05,
      "loss": 1.2329,
      "step": 15131
    },
    {
      "epoch": 0.5419091446272852,
      "grad_norm": 1.7705422639846802,
      "learning_rate": 9.129437580010449e-05,
      "loss": 1.3942,
      "step": 15132
    },
    {
      "epoch": 0.5419449567568535,
      "grad_norm": 2.8008735179901123,
      "learning_rate": 9.12828208825889e-05,
      "loss": 1.4574,
      "step": 15133
    },
    {
      "epoch": 0.5419807688864219,
      "grad_norm": 1.664994478225708,
      "learning_rate": 9.12712660823517e-05,
      "loss": 1.5583,
      "step": 15134
    },
    {
      "epoch": 0.5420165810159902,
      "grad_norm": 2.5544071197509766,
      "learning_rate": 9.125971139954835e-05,
      "loss": 1.6255,
      "step": 15135
    },
    {
      "epoch": 0.5420523931455584,
      "grad_norm": 1.645238995552063,
      "learning_rate": 9.124815683433432e-05,
      "loss": 1.2817,
      "step": 15136
    },
    {
      "epoch": 0.5420882052751267,
      "grad_norm": 1.7806026935577393,
      "learning_rate": 9.123660238686503e-05,
      "loss": 1.6258,
      "step": 15137
    },
    {
      "epoch": 0.542124017404695,
      "grad_norm": 1.9285032749176025,
      "learning_rate": 9.122504805729598e-05,
      "loss": 1.6613,
      "step": 15138
    },
    {
      "epoch": 0.5421598295342632,
      "grad_norm": 1.6295703649520874,
      "learning_rate": 9.121349384578255e-05,
      "loss": 1.3031,
      "step": 15139
    },
    {
      "epoch": 0.5421956416638315,
      "grad_norm": 1.7545650005340576,
      "learning_rate": 9.120193975248027e-05,
      "loss": 1.6079,
      "step": 15140
    },
    {
      "epoch": 0.5422314537933999,
      "grad_norm": 1.3660227060317993,
      "learning_rate": 9.119038577754451e-05,
      "loss": 1.2539,
      "step": 15141
    },
    {
      "epoch": 0.5422672659229681,
      "grad_norm": 1.3754740953445435,
      "learning_rate": 9.117883192113077e-05,
      "loss": 1.3293,
      "step": 15142
    },
    {
      "epoch": 0.5423030780525364,
      "grad_norm": 1.3937410116195679,
      "learning_rate": 9.116727818339444e-05,
      "loss": 1.6224,
      "step": 15143
    },
    {
      "epoch": 0.5423388901821047,
      "grad_norm": 2.4949188232421875,
      "learning_rate": 9.115572456449102e-05,
      "loss": 1.6398,
      "step": 15144
    },
    {
      "epoch": 0.542374702311673,
      "grad_norm": 1.8907426595687866,
      "learning_rate": 9.114417106457591e-05,
      "loss": 1.6803,
      "step": 15145
    },
    {
      "epoch": 0.5424105144412412,
      "grad_norm": 1.436632752418518,
      "learning_rate": 9.113261768380454e-05,
      "loss": 1.4615,
      "step": 15146
    },
    {
      "epoch": 0.5424463265708095,
      "grad_norm": 1.9328235387802124,
      "learning_rate": 9.112106442233237e-05,
      "loss": 1.47,
      "step": 15147
    },
    {
      "epoch": 0.5424821387003779,
      "grad_norm": 1.6240284442901611,
      "learning_rate": 9.110951128031482e-05,
      "loss": 1.2528,
      "step": 15148
    },
    {
      "epoch": 0.5425179508299461,
      "grad_norm": 1.9730523824691772,
      "learning_rate": 9.109795825790735e-05,
      "loss": 1.3464,
      "step": 15149
    },
    {
      "epoch": 0.5425537629595144,
      "grad_norm": 1.679447054862976,
      "learning_rate": 9.108640535526533e-05,
      "loss": 1.6473,
      "step": 15150
    },
    {
      "epoch": 0.5425895750890827,
      "grad_norm": 1.7408350706100464,
      "learning_rate": 9.107485257254426e-05,
      "loss": 1.5971,
      "step": 15151
    },
    {
      "epoch": 0.542625387218651,
      "grad_norm": 1.811719536781311,
      "learning_rate": 9.106329990989952e-05,
      "loss": 1.3407,
      "step": 15152
    },
    {
      "epoch": 0.5426611993482192,
      "grad_norm": 2.72822904586792,
      "learning_rate": 9.105174736748656e-05,
      "loss": 1.5943,
      "step": 15153
    },
    {
      "epoch": 0.5426970114777875,
      "grad_norm": 1.4611486196517944,
      "learning_rate": 9.104019494546081e-05,
      "loss": 1.3498,
      "step": 15154
    },
    {
      "epoch": 0.5427328236073559,
      "grad_norm": 1.499182105064392,
      "learning_rate": 9.102864264397765e-05,
      "loss": 1.4977,
      "step": 15155
    },
    {
      "epoch": 0.5427686357369241,
      "grad_norm": 1.542519211769104,
      "learning_rate": 9.101709046319256e-05,
      "loss": 1.1291,
      "step": 15156
    },
    {
      "epoch": 0.5428044478664924,
      "grad_norm": 1.6112427711486816,
      "learning_rate": 9.10055384032609e-05,
      "loss": 1.3433,
      "step": 15157
    },
    {
      "epoch": 0.5428402599960607,
      "grad_norm": 1.7602654695510864,
      "learning_rate": 9.099398646433814e-05,
      "loss": 1.6902,
      "step": 15158
    },
    {
      "epoch": 0.5428760721256289,
      "grad_norm": 1.9010089635849,
      "learning_rate": 9.098243464657966e-05,
      "loss": 1.7847,
      "step": 15159
    },
    {
      "epoch": 0.5429118842551972,
      "grad_norm": 2.192765235900879,
      "learning_rate": 9.097088295014092e-05,
      "loss": 1.4833,
      "step": 15160
    },
    {
      "epoch": 0.5429476963847655,
      "grad_norm": 2.360835313796997,
      "learning_rate": 9.095933137517727e-05,
      "loss": 1.9125,
      "step": 15161
    },
    {
      "epoch": 0.5429835085143339,
      "grad_norm": 1.5438072681427002,
      "learning_rate": 9.094777992184417e-05,
      "loss": 1.1242,
      "step": 15162
    },
    {
      "epoch": 0.5430193206439021,
      "grad_norm": 1.586403727531433,
      "learning_rate": 9.093622859029701e-05,
      "loss": 1.7543,
      "step": 15163
    },
    {
      "epoch": 0.5430551327734704,
      "grad_norm": 1.6280819177627563,
      "learning_rate": 9.09246773806912e-05,
      "loss": 1.2004,
      "step": 15164
    },
    {
      "epoch": 0.5430909449030387,
      "grad_norm": 2.1167192459106445,
      "learning_rate": 9.091312629318216e-05,
      "loss": 1.7255,
      "step": 15165
    },
    {
      "epoch": 0.5431267570326069,
      "grad_norm": 1.3767898082733154,
      "learning_rate": 9.090157532792526e-05,
      "loss": 1.4296,
      "step": 15166
    },
    {
      "epoch": 0.5431625691621752,
      "grad_norm": 1.5592973232269287,
      "learning_rate": 9.089002448507596e-05,
      "loss": 1.3892,
      "step": 15167
    },
    {
      "epoch": 0.5431983812917435,
      "grad_norm": 2.22155499458313,
      "learning_rate": 9.087847376478961e-05,
      "loss": 1.269,
      "step": 15168
    },
    {
      "epoch": 0.5432341934213119,
      "grad_norm": 1.561850905418396,
      "learning_rate": 9.086692316722166e-05,
      "loss": 1.3457,
      "step": 15169
    },
    {
      "epoch": 0.5432700055508801,
      "grad_norm": 1.6075046062469482,
      "learning_rate": 9.085537269252747e-05,
      "loss": 1.3777,
      "step": 15170
    },
    {
      "epoch": 0.5433058176804484,
      "grad_norm": 1.8987433910369873,
      "learning_rate": 9.08438223408624e-05,
      "loss": 1.2371,
      "step": 15171
    },
    {
      "epoch": 0.5433416298100167,
      "grad_norm": 1.4926401376724243,
      "learning_rate": 9.083227211238192e-05,
      "loss": 1.4493,
      "step": 15172
    },
    {
      "epoch": 0.5433774419395849,
      "grad_norm": 2.8346004486083984,
      "learning_rate": 9.082072200724139e-05,
      "loss": 1.5074,
      "step": 15173
    },
    {
      "epoch": 0.5434132540691532,
      "grad_norm": 1.5992660522460938,
      "learning_rate": 9.08091720255962e-05,
      "loss": 1.1084,
      "step": 15174
    },
    {
      "epoch": 0.5434490661987215,
      "grad_norm": 1.871173620223999,
      "learning_rate": 9.079762216760174e-05,
      "loss": 1.3968,
      "step": 15175
    },
    {
      "epoch": 0.5434848783282898,
      "grad_norm": 2.4197909832000732,
      "learning_rate": 9.078607243341344e-05,
      "loss": 1.8805,
      "step": 15176
    },
    {
      "epoch": 0.5435206904578581,
      "grad_norm": 1.300917387008667,
      "learning_rate": 9.077452282318661e-05,
      "loss": 1.3799,
      "step": 15177
    },
    {
      "epoch": 0.5435565025874264,
      "grad_norm": 1.500741720199585,
      "learning_rate": 9.076297333707669e-05,
      "loss": 1.3054,
      "step": 15178
    },
    {
      "epoch": 0.5435923147169947,
      "grad_norm": 1.7115274667739868,
      "learning_rate": 9.07514239752391e-05,
      "loss": 1.2259,
      "step": 15179
    },
    {
      "epoch": 0.5436281268465629,
      "grad_norm": 1.6599082946777344,
      "learning_rate": 9.073987473782907e-05,
      "loss": 1.7227,
      "step": 15180
    },
    {
      "epoch": 0.5436639389761312,
      "grad_norm": 1.701534390449524,
      "learning_rate": 9.072832562500217e-05,
      "loss": 1.5188,
      "step": 15181
    },
    {
      "epoch": 0.5436997511056995,
      "grad_norm": 1.621097445487976,
      "learning_rate": 9.071677663691361e-05,
      "loss": 1.3214,
      "step": 15182
    },
    {
      "epoch": 0.5437355632352678,
      "grad_norm": 1.8554396629333496,
      "learning_rate": 9.070522777371892e-05,
      "loss": 2.0134,
      "step": 15183
    },
    {
      "epoch": 0.5437713753648361,
      "grad_norm": 1.690596342086792,
      "learning_rate": 9.069367903557333e-05,
      "loss": 1.5976,
      "step": 15184
    },
    {
      "epoch": 0.5438071874944044,
      "grad_norm": 1.5468312501907349,
      "learning_rate": 9.068213042263234e-05,
      "loss": 1.3421,
      "step": 15185
    },
    {
      "epoch": 0.5438429996239726,
      "grad_norm": 1.7078195810317993,
      "learning_rate": 9.067058193505124e-05,
      "loss": 1.5431,
      "step": 15186
    },
    {
      "epoch": 0.5438788117535409,
      "grad_norm": 1.6992605924606323,
      "learning_rate": 9.065903357298544e-05,
      "loss": 1.4592,
      "step": 15187
    },
    {
      "epoch": 0.5439146238831092,
      "grad_norm": 1.696571707725525,
      "learning_rate": 9.064748533659031e-05,
      "loss": 1.644,
      "step": 15188
    },
    {
      "epoch": 0.5439504360126775,
      "grad_norm": 1.737825632095337,
      "learning_rate": 9.063593722602115e-05,
      "loss": 1.2788,
      "step": 15189
    },
    {
      "epoch": 0.5439862481422458,
      "grad_norm": 2.366448402404785,
      "learning_rate": 9.062438924143344e-05,
      "loss": 1.4315,
      "step": 15190
    },
    {
      "epoch": 0.5440220602718141,
      "grad_norm": 2.0418972969055176,
      "learning_rate": 9.06128413829824e-05,
      "loss": 1.4927,
      "step": 15191
    },
    {
      "epoch": 0.5440578724013824,
      "grad_norm": 2.0547597408294678,
      "learning_rate": 9.060129365082354e-05,
      "loss": 1.5544,
      "step": 15192
    },
    {
      "epoch": 0.5440936845309506,
      "grad_norm": 2.0122506618499756,
      "learning_rate": 9.05897460451121e-05,
      "loss": 1.5898,
      "step": 15193
    },
    {
      "epoch": 0.5441294966605189,
      "grad_norm": 2.682504177093506,
      "learning_rate": 9.057819856600355e-05,
      "loss": 1.5509,
      "step": 15194
    },
    {
      "epoch": 0.5441653087900872,
      "grad_norm": 2.28589129447937,
      "learning_rate": 9.056665121365311e-05,
      "loss": 1.5996,
      "step": 15195
    },
    {
      "epoch": 0.5442011209196554,
      "grad_norm": 1.6263935565948486,
      "learning_rate": 9.055510398821627e-05,
      "loss": 1.2338,
      "step": 15196
    },
    {
      "epoch": 0.5442369330492238,
      "grad_norm": 1.4128479957580566,
      "learning_rate": 9.054355688984833e-05,
      "loss": 1.6509,
      "step": 15197
    },
    {
      "epoch": 0.5442727451787921,
      "grad_norm": 1.4908677339553833,
      "learning_rate": 9.053200991870456e-05,
      "loss": 1.0973,
      "step": 15198
    },
    {
      "epoch": 0.5443085573083604,
      "grad_norm": 2.3797690868377686,
      "learning_rate": 9.052046307494046e-05,
      "loss": 1.4909,
      "step": 15199
    },
    {
      "epoch": 0.5443443694379286,
      "grad_norm": 1.5624659061431885,
      "learning_rate": 9.050891635871124e-05,
      "loss": 1.1965,
      "step": 15200
    },
    {
      "epoch": 0.5443801815674969,
      "grad_norm": 1.94329833984375,
      "learning_rate": 9.049736977017236e-05,
      "loss": 1.661,
      "step": 15201
    },
    {
      "epoch": 0.5444159936970652,
      "grad_norm": 1.6274837255477905,
      "learning_rate": 9.048582330947906e-05,
      "loss": 1.4358,
      "step": 15202
    },
    {
      "epoch": 0.5444518058266334,
      "grad_norm": 2.1027121543884277,
      "learning_rate": 9.04742769767868e-05,
      "loss": 1.8225,
      "step": 15203
    },
    {
      "epoch": 0.5444876179562018,
      "grad_norm": 1.6133577823638916,
      "learning_rate": 9.046273077225078e-05,
      "loss": 1.5472,
      "step": 15204
    },
    {
      "epoch": 0.5445234300857701,
      "grad_norm": 1.9349349737167358,
      "learning_rate": 9.045118469602649e-05,
      "loss": 1.4636,
      "step": 15205
    },
    {
      "epoch": 0.5445592422153384,
      "grad_norm": 1.7289633750915527,
      "learning_rate": 9.043963874826917e-05,
      "loss": 1.6001,
      "step": 15206
    },
    {
      "epoch": 0.5445950543449066,
      "grad_norm": 1.6920100450515747,
      "learning_rate": 9.042809292913415e-05,
      "loss": 1.3169,
      "step": 15207
    },
    {
      "epoch": 0.5446308664744749,
      "grad_norm": 1.8096444606781006,
      "learning_rate": 9.041654723877683e-05,
      "loss": 1.2999,
      "step": 15208
    },
    {
      "epoch": 0.5446666786040432,
      "grad_norm": 1.588729977607727,
      "learning_rate": 9.040500167735247e-05,
      "loss": 1.7168,
      "step": 15209
    },
    {
      "epoch": 0.5447024907336114,
      "grad_norm": 1.802148699760437,
      "learning_rate": 9.039345624501646e-05,
      "loss": 1.3229,
      "step": 15210
    },
    {
      "epoch": 0.5447383028631798,
      "grad_norm": 1.531667947769165,
      "learning_rate": 9.038191094192407e-05,
      "loss": 1.3616,
      "step": 15211
    },
    {
      "epoch": 0.5447741149927481,
      "grad_norm": 1.4396610260009766,
      "learning_rate": 9.037036576823072e-05,
      "loss": 0.9985,
      "step": 15212
    },
    {
      "epoch": 0.5448099271223164,
      "grad_norm": 1.5046268701553345,
      "learning_rate": 9.035882072409161e-05,
      "loss": 1.5455,
      "step": 15213
    },
    {
      "epoch": 0.5448457392518846,
      "grad_norm": 2.3657383918762207,
      "learning_rate": 9.034727580966219e-05,
      "loss": 1.8284,
      "step": 15214
    },
    {
      "epoch": 0.5448815513814529,
      "grad_norm": 1.6111987829208374,
      "learning_rate": 9.033573102509771e-05,
      "loss": 1.4184,
      "step": 15215
    },
    {
      "epoch": 0.5449173635110212,
      "grad_norm": 1.466270923614502,
      "learning_rate": 9.032418637055348e-05,
      "loss": 1.4768,
      "step": 15216
    },
    {
      "epoch": 0.5449531756405894,
      "grad_norm": 1.4635989665985107,
      "learning_rate": 9.031264184618487e-05,
      "loss": 1.4865,
      "step": 15217
    },
    {
      "epoch": 0.5449889877701578,
      "grad_norm": 2.0012905597686768,
      "learning_rate": 9.030109745214713e-05,
      "loss": 1.5261,
      "step": 15218
    },
    {
      "epoch": 0.5450247998997261,
      "grad_norm": 1.6931449174880981,
      "learning_rate": 9.028955318859564e-05,
      "loss": 1.3085,
      "step": 15219
    },
    {
      "epoch": 0.5450606120292943,
      "grad_norm": 1.5689709186553955,
      "learning_rate": 9.027800905568568e-05,
      "loss": 1.5905,
      "step": 15220
    },
    {
      "epoch": 0.5450964241588626,
      "grad_norm": 1.997705101966858,
      "learning_rate": 9.026646505357258e-05,
      "loss": 1.835,
      "step": 15221
    },
    {
      "epoch": 0.5451322362884309,
      "grad_norm": 1.988075852394104,
      "learning_rate": 9.025492118241161e-05,
      "loss": 1.92,
      "step": 15222
    },
    {
      "epoch": 0.5451680484179992,
      "grad_norm": 1.7597702741622925,
      "learning_rate": 9.024337744235814e-05,
      "loss": 1.6645,
      "step": 15223
    },
    {
      "epoch": 0.5452038605475674,
      "grad_norm": 1.6029980182647705,
      "learning_rate": 9.023183383356743e-05,
      "loss": 1.5673,
      "step": 15224
    },
    {
      "epoch": 0.5452396726771358,
      "grad_norm": 1.7730108499526978,
      "learning_rate": 9.022029035619478e-05,
      "loss": 1.4413,
      "step": 15225
    },
    {
      "epoch": 0.5452754848067041,
      "grad_norm": 1.6495401859283447,
      "learning_rate": 9.020874701039552e-05,
      "loss": 1.5705,
      "step": 15226
    },
    {
      "epoch": 0.5453112969362723,
      "grad_norm": 2.009526252746582,
      "learning_rate": 9.019720379632493e-05,
      "loss": 1.4947,
      "step": 15227
    },
    {
      "epoch": 0.5453471090658406,
      "grad_norm": 1.3570290803909302,
      "learning_rate": 9.018566071413833e-05,
      "loss": 1.4492,
      "step": 15228
    },
    {
      "epoch": 0.5453829211954089,
      "grad_norm": 1.9762985706329346,
      "learning_rate": 9.017411776399099e-05,
      "loss": 1.8505,
      "step": 15229
    },
    {
      "epoch": 0.5454187333249771,
      "grad_norm": 1.482008695602417,
      "learning_rate": 9.016257494603824e-05,
      "loss": 1.0744,
      "step": 15230
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 1.898165225982666,
      "learning_rate": 9.015103226043533e-05,
      "loss": 1.5189,
      "step": 15231
    },
    {
      "epoch": 0.5454903575841138,
      "grad_norm": 1.4939240217208862,
      "learning_rate": 9.01394897073376e-05,
      "loss": 1.5322,
      "step": 15232
    },
    {
      "epoch": 0.5455261697136821,
      "grad_norm": 1.3671813011169434,
      "learning_rate": 9.012794728690032e-05,
      "loss": 1.5075,
      "step": 15233
    },
    {
      "epoch": 0.5455619818432503,
      "grad_norm": 1.485896348953247,
      "learning_rate": 9.011640499927875e-05,
      "loss": 1.589,
      "step": 15234
    },
    {
      "epoch": 0.5455977939728186,
      "grad_norm": 1.7984567880630493,
      "learning_rate": 9.010486284462823e-05,
      "loss": 1.4442,
      "step": 15235
    },
    {
      "epoch": 0.5456336061023869,
      "grad_norm": 1.7843817472457886,
      "learning_rate": 9.009332082310398e-05,
      "loss": 1.3122,
      "step": 15236
    },
    {
      "epoch": 0.5456694182319551,
      "grad_norm": 1.760123610496521,
      "learning_rate": 9.008177893486136e-05,
      "loss": 1.5041,
      "step": 15237
    },
    {
      "epoch": 0.5457052303615234,
      "grad_norm": 1.4291465282440186,
      "learning_rate": 9.007023718005558e-05,
      "loss": 1.6004,
      "step": 15238
    },
    {
      "epoch": 0.5457410424910918,
      "grad_norm": 2.074390172958374,
      "learning_rate": 9.005869555884197e-05,
      "loss": 1.4731,
      "step": 15239
    },
    {
      "epoch": 0.5457768546206601,
      "grad_norm": 1.5591212511062622,
      "learning_rate": 9.004715407137577e-05,
      "loss": 1.4422,
      "step": 15240
    },
    {
      "epoch": 0.5458126667502283,
      "grad_norm": 1.5369257926940918,
      "learning_rate": 9.003561271781229e-05,
      "loss": 1.4392,
      "step": 15241
    },
    {
      "epoch": 0.5458484788797966,
      "grad_norm": 1.3781731128692627,
      "learning_rate": 9.002407149830679e-05,
      "loss": 1.5801,
      "step": 15242
    },
    {
      "epoch": 0.5458842910093649,
      "grad_norm": 1.7291080951690674,
      "learning_rate": 9.001253041301453e-05,
      "loss": 1.6995,
      "step": 15243
    },
    {
      "epoch": 0.5459201031389331,
      "grad_norm": 1.5849418640136719,
      "learning_rate": 9.00009894620908e-05,
      "loss": 1.5973,
      "step": 15244
    },
    {
      "epoch": 0.5459559152685014,
      "grad_norm": 1.6605151891708374,
      "learning_rate": 8.998944864569084e-05,
      "loss": 1.6434,
      "step": 15245
    },
    {
      "epoch": 0.5459917273980698,
      "grad_norm": 1.9970816373825073,
      "learning_rate": 8.997790796396996e-05,
      "loss": 1.8821,
      "step": 15246
    },
    {
      "epoch": 0.546027539527638,
      "grad_norm": 1.8191704750061035,
      "learning_rate": 8.996636741708337e-05,
      "loss": 1.7131,
      "step": 15247
    },
    {
      "epoch": 0.5460633516572063,
      "grad_norm": 1.318717122077942,
      "learning_rate": 8.995482700518639e-05,
      "loss": 1.3839,
      "step": 15248
    },
    {
      "epoch": 0.5460991637867746,
      "grad_norm": 1.5386637449264526,
      "learning_rate": 8.994328672843424e-05,
      "loss": 1.74,
      "step": 15249
    },
    {
      "epoch": 0.5461349759163429,
      "grad_norm": 1.6829789876937866,
      "learning_rate": 8.993174658698221e-05,
      "loss": 1.3832,
      "step": 15250
    },
    {
      "epoch": 0.5461707880459111,
      "grad_norm": 2.5032663345336914,
      "learning_rate": 8.992020658098555e-05,
      "loss": 1.4875,
      "step": 15251
    },
    {
      "epoch": 0.5462066001754794,
      "grad_norm": 2.0662424564361572,
      "learning_rate": 8.990866671059948e-05,
      "loss": 1.5559,
      "step": 15252
    },
    {
      "epoch": 0.5462424123050477,
      "grad_norm": 1.9819056987762451,
      "learning_rate": 8.98971269759793e-05,
      "loss": 1.7462,
      "step": 15253
    },
    {
      "epoch": 0.546278224434616,
      "grad_norm": 1.755515456199646,
      "learning_rate": 8.988558737728023e-05,
      "loss": 1.4854,
      "step": 15254
    },
    {
      "epoch": 0.5463140365641843,
      "grad_norm": 1.8276987075805664,
      "learning_rate": 8.987404791465757e-05,
      "loss": 1.5511,
      "step": 15255
    },
    {
      "epoch": 0.5463498486937526,
      "grad_norm": 1.7192363739013672,
      "learning_rate": 8.986250858826649e-05,
      "loss": 1.7131,
      "step": 15256
    },
    {
      "epoch": 0.5463856608233209,
      "grad_norm": 1.8038772344589233,
      "learning_rate": 8.985096939826231e-05,
      "loss": 1.4206,
      "step": 15257
    },
    {
      "epoch": 0.5464214729528891,
      "grad_norm": 2.844478130340576,
      "learning_rate": 8.983943034480022e-05,
      "loss": 1.8703,
      "step": 15258
    },
    {
      "epoch": 0.5464572850824574,
      "grad_norm": 1.6079673767089844,
      "learning_rate": 8.982789142803552e-05,
      "loss": 1.0447,
      "step": 15259
    },
    {
      "epoch": 0.5464930972120257,
      "grad_norm": 1.8626993894577026,
      "learning_rate": 8.981635264812341e-05,
      "loss": 1.5425,
      "step": 15260
    },
    {
      "epoch": 0.546528909341594,
      "grad_norm": 1.9732133150100708,
      "learning_rate": 8.980481400521911e-05,
      "loss": 1.7118,
      "step": 15261
    },
    {
      "epoch": 0.5465647214711623,
      "grad_norm": 1.683758020401001,
      "learning_rate": 8.979327549947794e-05,
      "loss": 1.7804,
      "step": 15262
    },
    {
      "epoch": 0.5466005336007306,
      "grad_norm": 2.286837339401245,
      "learning_rate": 8.978173713105503e-05,
      "loss": 1.6872,
      "step": 15263
    },
    {
      "epoch": 0.5466363457302988,
      "grad_norm": 2.094520330429077,
      "learning_rate": 8.977019890010571e-05,
      "loss": 1.5192,
      "step": 15264
    },
    {
      "epoch": 0.5466721578598671,
      "grad_norm": 1.6780959367752075,
      "learning_rate": 8.975866080678512e-05,
      "loss": 1.451,
      "step": 15265
    },
    {
      "epoch": 0.5467079699894354,
      "grad_norm": 1.653809905052185,
      "learning_rate": 8.974712285124858e-05,
      "loss": 1.4063,
      "step": 15266
    },
    {
      "epoch": 0.5467437821190037,
      "grad_norm": 1.4220283031463623,
      "learning_rate": 8.973558503365129e-05,
      "loss": 1.5108,
      "step": 15267
    },
    {
      "epoch": 0.546779594248572,
      "grad_norm": 2.0603954792022705,
      "learning_rate": 8.97240473541484e-05,
      "loss": 1.6627,
      "step": 15268
    },
    {
      "epoch": 0.5468154063781403,
      "grad_norm": 1.338783621788025,
      "learning_rate": 8.971250981289525e-05,
      "loss": 1.4841,
      "step": 15269
    },
    {
      "epoch": 0.5468512185077086,
      "grad_norm": 2.398456573486328,
      "learning_rate": 8.970097241004697e-05,
      "loss": 1.5663,
      "step": 15270
    },
    {
      "epoch": 0.5468870306372768,
      "grad_norm": 1.6194860935211182,
      "learning_rate": 8.968943514575888e-05,
      "loss": 1.6908,
      "step": 15271
    },
    {
      "epoch": 0.5469228427668451,
      "grad_norm": 1.4247815608978271,
      "learning_rate": 8.967789802018607e-05,
      "loss": 1.4542,
      "step": 15272
    },
    {
      "epoch": 0.5469586548964134,
      "grad_norm": 1.889357566833496,
      "learning_rate": 8.966636103348388e-05,
      "loss": 1.7865,
      "step": 15273
    },
    {
      "epoch": 0.5469944670259816,
      "grad_norm": 1.6137316226959229,
      "learning_rate": 8.965482418580746e-05,
      "loss": 1.8113,
      "step": 15274
    },
    {
      "epoch": 0.54703027915555,
      "grad_norm": 1.8220938444137573,
      "learning_rate": 8.964328747731204e-05,
      "loss": 1.4244,
      "step": 15275
    },
    {
      "epoch": 0.5470660912851183,
      "grad_norm": 1.8542633056640625,
      "learning_rate": 8.963175090815285e-05,
      "loss": 1.32,
      "step": 15276
    },
    {
      "epoch": 0.5471019034146866,
      "grad_norm": 1.4733332395553589,
      "learning_rate": 8.9620214478485e-05,
      "loss": 1.2989,
      "step": 15277
    },
    {
      "epoch": 0.5471377155442548,
      "grad_norm": 1.4526832103729248,
      "learning_rate": 8.960867818846386e-05,
      "loss": 1.1323,
      "step": 15278
    },
    {
      "epoch": 0.5471735276738231,
      "grad_norm": 1.3265506029129028,
      "learning_rate": 8.959714203824449e-05,
      "loss": 1.397,
      "step": 15279
    },
    {
      "epoch": 0.5472093398033914,
      "grad_norm": 2.6252152919769287,
      "learning_rate": 8.958560602798221e-05,
      "loss": 1.7199,
      "step": 15280
    },
    {
      "epoch": 0.5472451519329596,
      "grad_norm": 1.9414409399032593,
      "learning_rate": 8.95740701578321e-05,
      "loss": 1.5143,
      "step": 15281
    },
    {
      "epoch": 0.547280964062528,
      "grad_norm": 1.725482702255249,
      "learning_rate": 8.956253442794948e-05,
      "loss": 1.1287,
      "step": 15282
    },
    {
      "epoch": 0.5473167761920963,
      "grad_norm": 1.5736626386642456,
      "learning_rate": 8.955099883848945e-05,
      "loss": 1.4276,
      "step": 15283
    },
    {
      "epoch": 0.5473525883216646,
      "grad_norm": 1.4038923978805542,
      "learning_rate": 8.953946338960731e-05,
      "loss": 1.2686,
      "step": 15284
    },
    {
      "epoch": 0.5473884004512328,
      "grad_norm": 1.9812275171279907,
      "learning_rate": 8.952792808145819e-05,
      "loss": 1.5779,
      "step": 15285
    },
    {
      "epoch": 0.5474242125808011,
      "grad_norm": 1.4937734603881836,
      "learning_rate": 8.951639291419723e-05,
      "loss": 1.4877,
      "step": 15286
    },
    {
      "epoch": 0.5474600247103694,
      "grad_norm": 1.4445714950561523,
      "learning_rate": 8.950485788797976e-05,
      "loss": 1.6559,
      "step": 15287
    },
    {
      "epoch": 0.5474958368399376,
      "grad_norm": 1.952468752861023,
      "learning_rate": 8.949332300296082e-05,
      "loss": 1.5121,
      "step": 15288
    },
    {
      "epoch": 0.547531648969506,
      "grad_norm": 1.7824208736419678,
      "learning_rate": 8.948178825929572e-05,
      "loss": 1.3028,
      "step": 15289
    },
    {
      "epoch": 0.5475674610990743,
      "grad_norm": 1.73703134059906,
      "learning_rate": 8.947025365713953e-05,
      "loss": 1.5888,
      "step": 15290
    },
    {
      "epoch": 0.5476032732286426,
      "grad_norm": 1.6486480236053467,
      "learning_rate": 8.945871919664757e-05,
      "loss": 1.1964,
      "step": 15291
    },
    {
      "epoch": 0.5476390853582108,
      "grad_norm": 1.4871211051940918,
      "learning_rate": 8.944718487797487e-05,
      "loss": 1.2785,
      "step": 15292
    },
    {
      "epoch": 0.5476748974877791,
      "grad_norm": 1.483238697052002,
      "learning_rate": 8.943565070127676e-05,
      "loss": 1.6927,
      "step": 15293
    },
    {
      "epoch": 0.5477107096173474,
      "grad_norm": 1.6315890550613403,
      "learning_rate": 8.94241166667083e-05,
      "loss": 1.4549,
      "step": 15294
    },
    {
      "epoch": 0.5477465217469156,
      "grad_norm": 2.340033769607544,
      "learning_rate": 8.94125827744247e-05,
      "loss": 1.6152,
      "step": 15295
    },
    {
      "epoch": 0.547782333876484,
      "grad_norm": 1.4662576913833618,
      "learning_rate": 8.940104902458117e-05,
      "loss": 1.6437,
      "step": 15296
    },
    {
      "epoch": 0.5478181460060523,
      "grad_norm": 1.7731101512908936,
      "learning_rate": 8.938951541733282e-05,
      "loss": 1.377,
      "step": 15297
    },
    {
      "epoch": 0.5478539581356205,
      "grad_norm": 1.815285563468933,
      "learning_rate": 8.93779819528349e-05,
      "loss": 1.4197,
      "step": 15298
    },
    {
      "epoch": 0.5478897702651888,
      "grad_norm": 2.2570080757141113,
      "learning_rate": 8.936644863124246e-05,
      "loss": 1.6707,
      "step": 15299
    },
    {
      "epoch": 0.5479255823947571,
      "grad_norm": 1.9335801601409912,
      "learning_rate": 8.935491545271081e-05,
      "loss": 1.5724,
      "step": 15300
    },
    {
      "epoch": 0.5479613945243254,
      "grad_norm": 1.3450572490692139,
      "learning_rate": 8.934338241739498e-05,
      "loss": 0.9793,
      "step": 15301
    },
    {
      "epoch": 0.5479972066538936,
      "grad_norm": 2.0288915634155273,
      "learning_rate": 8.933184952545026e-05,
      "loss": 1.5523,
      "step": 15302
    },
    {
      "epoch": 0.548033018783462,
      "grad_norm": 1.486048936843872,
      "learning_rate": 8.932031677703172e-05,
      "loss": 1.1013,
      "step": 15303
    },
    {
      "epoch": 0.5480688309130303,
      "grad_norm": 1.9262478351593018,
      "learning_rate": 8.930878417229453e-05,
      "loss": 1.5825,
      "step": 15304
    },
    {
      "epoch": 0.5481046430425985,
      "grad_norm": 2.0612335205078125,
      "learning_rate": 8.929725171139387e-05,
      "loss": 1.3004,
      "step": 15305
    },
    {
      "epoch": 0.5481404551721668,
      "grad_norm": 2.523592948913574,
      "learning_rate": 8.928571939448486e-05,
      "loss": 1.8698,
      "step": 15306
    },
    {
      "epoch": 0.5481762673017351,
      "grad_norm": 1.6707018613815308,
      "learning_rate": 8.927418722172269e-05,
      "loss": 1.5277,
      "step": 15307
    },
    {
      "epoch": 0.5482120794313033,
      "grad_norm": 1.4799731969833374,
      "learning_rate": 8.926265519326246e-05,
      "loss": 1.1901,
      "step": 15308
    },
    {
      "epoch": 0.5482478915608716,
      "grad_norm": 1.638659119606018,
      "learning_rate": 8.925112330925943e-05,
      "loss": 1.4529,
      "step": 15309
    },
    {
      "epoch": 0.54828370369044,
      "grad_norm": 1.3840322494506836,
      "learning_rate": 8.923959156986859e-05,
      "loss": 1.625,
      "step": 15310
    },
    {
      "epoch": 0.5483195158200083,
      "grad_norm": 1.525613784790039,
      "learning_rate": 8.922805997524524e-05,
      "loss": 1.5994,
      "step": 15311
    },
    {
      "epoch": 0.5483553279495765,
      "grad_norm": 2.075780153274536,
      "learning_rate": 8.921652852554442e-05,
      "loss": 1.9281,
      "step": 15312
    },
    {
      "epoch": 0.5483911400791448,
      "grad_norm": 1.4483722448349,
      "learning_rate": 8.920499722092129e-05,
      "loss": 1.3819,
      "step": 15313
    },
    {
      "epoch": 0.5484269522087131,
      "grad_norm": 2.663997173309326,
      "learning_rate": 8.9193466061531e-05,
      "loss": 1.02,
      "step": 15314
    },
    {
      "epoch": 0.5484627643382813,
      "grad_norm": 1.449901819229126,
      "learning_rate": 8.918193504752868e-05,
      "loss": 1.4184,
      "step": 15315
    },
    {
      "epoch": 0.5484985764678496,
      "grad_norm": 1.5887750387191772,
      "learning_rate": 8.917040417906947e-05,
      "loss": 1.4567,
      "step": 15316
    },
    {
      "epoch": 0.548534388597418,
      "grad_norm": 1.6006042957305908,
      "learning_rate": 8.91588734563085e-05,
      "loss": 1.5301,
      "step": 15317
    },
    {
      "epoch": 0.5485702007269863,
      "grad_norm": 1.9737050533294678,
      "learning_rate": 8.914734287940092e-05,
      "loss": 1.5652,
      "step": 15318
    },
    {
      "epoch": 0.5486060128565545,
      "grad_norm": 1.3489166498184204,
      "learning_rate": 8.913581244850182e-05,
      "loss": 1.2951,
      "step": 15319
    },
    {
      "epoch": 0.5486418249861228,
      "grad_norm": 1.744086503982544,
      "learning_rate": 8.912428216376637e-05,
      "loss": 1.4222,
      "step": 15320
    },
    {
      "epoch": 0.5486776371156911,
      "grad_norm": 1.5901646614074707,
      "learning_rate": 8.911275202534968e-05,
      "loss": 1.3965,
      "step": 15321
    },
    {
      "epoch": 0.5487134492452593,
      "grad_norm": 2.381009578704834,
      "learning_rate": 8.910122203340684e-05,
      "loss": 1.7014,
      "step": 15322
    },
    {
      "epoch": 0.5487492613748276,
      "grad_norm": 1.773105263710022,
      "learning_rate": 8.908969218809302e-05,
      "loss": 1.2225,
      "step": 15323
    },
    {
      "epoch": 0.548785073504396,
      "grad_norm": 1.7156941890716553,
      "learning_rate": 8.907816248956331e-05,
      "loss": 1.5767,
      "step": 15324
    },
    {
      "epoch": 0.5488208856339643,
      "grad_norm": 2.0456743240356445,
      "learning_rate": 8.906663293797284e-05,
      "loss": 1.7365,
      "step": 15325
    },
    {
      "epoch": 0.5488566977635325,
      "grad_norm": 1.5815179347991943,
      "learning_rate": 8.905510353347671e-05,
      "loss": 1.5657,
      "step": 15326
    },
    {
      "epoch": 0.5488925098931008,
      "grad_norm": 1.702483892440796,
      "learning_rate": 8.904357427623007e-05,
      "loss": 1.6139,
      "step": 15327
    },
    {
      "epoch": 0.5489283220226691,
      "grad_norm": 1.5035403966903687,
      "learning_rate": 8.903204516638796e-05,
      "loss": 1.515,
      "step": 15328
    },
    {
      "epoch": 0.5489641341522373,
      "grad_norm": 1.8640391826629639,
      "learning_rate": 8.902051620410558e-05,
      "loss": 1.5362,
      "step": 15329
    },
    {
      "epoch": 0.5489999462818056,
      "grad_norm": 1.583310842514038,
      "learning_rate": 8.9008987389538e-05,
      "loss": 1.5499,
      "step": 15330
    },
    {
      "epoch": 0.549035758411374,
      "grad_norm": 1.6002247333526611,
      "learning_rate": 8.899745872284026e-05,
      "loss": 1.1217,
      "step": 15331
    },
    {
      "epoch": 0.5490715705409422,
      "grad_norm": 1.4390969276428223,
      "learning_rate": 8.898593020416756e-05,
      "loss": 1.6414,
      "step": 15332
    },
    {
      "epoch": 0.5491073826705105,
      "grad_norm": 1.7034553289413452,
      "learning_rate": 8.897440183367496e-05,
      "loss": 1.549,
      "step": 15333
    },
    {
      "epoch": 0.5491431948000788,
      "grad_norm": 1.4598033428192139,
      "learning_rate": 8.896287361151757e-05,
      "loss": 1.4662,
      "step": 15334
    },
    {
      "epoch": 0.549179006929647,
      "grad_norm": 2.3608148097991943,
      "learning_rate": 8.895134553785044e-05,
      "loss": 1.5284,
      "step": 15335
    },
    {
      "epoch": 0.5492148190592153,
      "grad_norm": 1.7964868545532227,
      "learning_rate": 8.893981761282874e-05,
      "loss": 1.5421,
      "step": 15336
    },
    {
      "epoch": 0.5492506311887836,
      "grad_norm": 1.50372314453125,
      "learning_rate": 8.89282898366075e-05,
      "loss": 1.2766,
      "step": 15337
    },
    {
      "epoch": 0.549286443318352,
      "grad_norm": 1.470895767211914,
      "learning_rate": 8.891676220934188e-05,
      "loss": 1.2957,
      "step": 15338
    },
    {
      "epoch": 0.5493222554479202,
      "grad_norm": 1.9627442359924316,
      "learning_rate": 8.89052347311869e-05,
      "loss": 1.5351,
      "step": 15339
    },
    {
      "epoch": 0.5493580675774885,
      "grad_norm": 2.062045097351074,
      "learning_rate": 8.889370740229767e-05,
      "loss": 1.4394,
      "step": 15340
    },
    {
      "epoch": 0.5493938797070568,
      "grad_norm": 1.9214800596237183,
      "learning_rate": 8.88821802228293e-05,
      "loss": 1.3882,
      "step": 15341
    },
    {
      "epoch": 0.549429691836625,
      "grad_norm": 1.5839799642562866,
      "learning_rate": 8.887065319293684e-05,
      "loss": 1.427,
      "step": 15342
    },
    {
      "epoch": 0.5494655039661933,
      "grad_norm": 1.8803417682647705,
      "learning_rate": 8.88591263127754e-05,
      "loss": 1.4825,
      "step": 15343
    },
    {
      "epoch": 0.5495013160957616,
      "grad_norm": 1.5140496492385864,
      "learning_rate": 8.884759958250002e-05,
      "loss": 1.4992,
      "step": 15344
    },
    {
      "epoch": 0.54953712822533,
      "grad_norm": 1.6109691858291626,
      "learning_rate": 8.883607300226581e-05,
      "loss": 1.3751,
      "step": 15345
    },
    {
      "epoch": 0.5495729403548982,
      "grad_norm": 2.227198600769043,
      "learning_rate": 8.882454657222784e-05,
      "loss": 1.6037,
      "step": 15346
    },
    {
      "epoch": 0.5496087524844665,
      "grad_norm": 1.5396227836608887,
      "learning_rate": 8.88130202925412e-05,
      "loss": 1.4925,
      "step": 15347
    },
    {
      "epoch": 0.5496445646140348,
      "grad_norm": 2.18083119392395,
      "learning_rate": 8.880149416336093e-05,
      "loss": 1.7599,
      "step": 15348
    },
    {
      "epoch": 0.549680376743603,
      "grad_norm": 1.8912416696548462,
      "learning_rate": 8.878996818484209e-05,
      "loss": 1.5699,
      "step": 15349
    },
    {
      "epoch": 0.5497161888731713,
      "grad_norm": 1.3284300565719604,
      "learning_rate": 8.87784423571398e-05,
      "loss": 1.4352,
      "step": 15350
    },
    {
      "epoch": 0.5497520010027396,
      "grad_norm": 1.830657720565796,
      "learning_rate": 8.876691668040907e-05,
      "loss": 1.5481,
      "step": 15351
    },
    {
      "epoch": 0.549787813132308,
      "grad_norm": 2.0324718952178955,
      "learning_rate": 8.8755391154805e-05,
      "loss": 1.5041,
      "step": 15352
    },
    {
      "epoch": 0.5498236252618762,
      "grad_norm": 1.8235435485839844,
      "learning_rate": 8.874386578048261e-05,
      "loss": 1.2626,
      "step": 15353
    },
    {
      "epoch": 0.5498594373914445,
      "grad_norm": 1.490576148033142,
      "learning_rate": 8.873234055759703e-05,
      "loss": 1.5272,
      "step": 15354
    },
    {
      "epoch": 0.5498952495210128,
      "grad_norm": 1.8105982542037964,
      "learning_rate": 8.872081548630325e-05,
      "loss": 1.4809,
      "step": 15355
    },
    {
      "epoch": 0.549931061650581,
      "grad_norm": 1.976482629776001,
      "learning_rate": 8.870929056675636e-05,
      "loss": 1.597,
      "step": 15356
    },
    {
      "epoch": 0.5499668737801493,
      "grad_norm": 1.8228706121444702,
      "learning_rate": 8.86977657991114e-05,
      "loss": 1.7323,
      "step": 15357
    },
    {
      "epoch": 0.5500026859097176,
      "grad_norm": 1.821420669555664,
      "learning_rate": 8.86862411835234e-05,
      "loss": 1.3638,
      "step": 15358
    },
    {
      "epoch": 0.550038498039286,
      "grad_norm": 1.6070480346679688,
      "learning_rate": 8.867471672014745e-05,
      "loss": 1.3667,
      "step": 15359
    },
    {
      "epoch": 0.5500743101688542,
      "grad_norm": 1.3694812059402466,
      "learning_rate": 8.866319240913856e-05,
      "loss": 1.2689,
      "step": 15360
    },
    {
      "epoch": 0.5501101222984225,
      "grad_norm": 1.4904698133468628,
      "learning_rate": 8.865166825065182e-05,
      "loss": 1.3611,
      "step": 15361
    },
    {
      "epoch": 0.5501459344279908,
      "grad_norm": 1.2036080360412598,
      "learning_rate": 8.864014424484222e-05,
      "loss": 1.2641,
      "step": 15362
    },
    {
      "epoch": 0.550181746557559,
      "grad_norm": 1.5762604475021362,
      "learning_rate": 8.862862039186485e-05,
      "loss": 1.6896,
      "step": 15363
    },
    {
      "epoch": 0.5502175586871273,
      "grad_norm": 2.1189725399017334,
      "learning_rate": 8.861709669187474e-05,
      "loss": 1.4898,
      "step": 15364
    },
    {
      "epoch": 0.5502533708166956,
      "grad_norm": 1.6944526433944702,
      "learning_rate": 8.860557314502685e-05,
      "loss": 1.5987,
      "step": 15365
    },
    {
      "epoch": 0.5502891829462639,
      "grad_norm": 1.749204158782959,
      "learning_rate": 8.859404975147632e-05,
      "loss": 1.3471,
      "step": 15366
    },
    {
      "epoch": 0.5503249950758322,
      "grad_norm": 1.700831651687622,
      "learning_rate": 8.858252651137812e-05,
      "loss": 1.526,
      "step": 15367
    },
    {
      "epoch": 0.5503608072054005,
      "grad_norm": 1.3145804405212402,
      "learning_rate": 8.857100342488732e-05,
      "loss": 1.4017,
      "step": 15368
    },
    {
      "epoch": 0.5503966193349688,
      "grad_norm": 1.3753552436828613,
      "learning_rate": 8.855948049215888e-05,
      "loss": 1.5874,
      "step": 15369
    },
    {
      "epoch": 0.550432431464537,
      "grad_norm": 1.5876967906951904,
      "learning_rate": 8.854795771334794e-05,
      "loss": 1.537,
      "step": 15370
    },
    {
      "epoch": 0.5504682435941053,
      "grad_norm": 1.6385436058044434,
      "learning_rate": 8.85364350886094e-05,
      "loss": 1.3111,
      "step": 15371
    },
    {
      "epoch": 0.5505040557236736,
      "grad_norm": 2.4700660705566406,
      "learning_rate": 8.852491261809837e-05,
      "loss": 1.7414,
      "step": 15372
    },
    {
      "epoch": 0.5505398678532419,
      "grad_norm": 1.4579800367355347,
      "learning_rate": 8.851339030196986e-05,
      "loss": 1.4192,
      "step": 15373
    },
    {
      "epoch": 0.5505756799828102,
      "grad_norm": 1.7124450206756592,
      "learning_rate": 8.85018681403788e-05,
      "loss": 1.2621,
      "step": 15374
    },
    {
      "epoch": 0.5506114921123785,
      "grad_norm": 1.5826530456542969,
      "learning_rate": 8.849034613348035e-05,
      "loss": 1.3615,
      "step": 15375
    },
    {
      "epoch": 0.5506473042419467,
      "grad_norm": 1.3330745697021484,
      "learning_rate": 8.847882428142936e-05,
      "loss": 1.3821,
      "step": 15376
    },
    {
      "epoch": 0.550683116371515,
      "grad_norm": 1.582822322845459,
      "learning_rate": 8.8467302584381e-05,
      "loss": 1.4889,
      "step": 15377
    },
    {
      "epoch": 0.5507189285010833,
      "grad_norm": 1.2767612934112549,
      "learning_rate": 8.845578104249014e-05,
      "loss": 1.3505,
      "step": 15378
    },
    {
      "epoch": 0.5507547406306516,
      "grad_norm": 1.4788082838058472,
      "learning_rate": 8.844425965591192e-05,
      "loss": 1.5282,
      "step": 15379
    },
    {
      "epoch": 0.5507905527602199,
      "grad_norm": 1.8105095624923706,
      "learning_rate": 8.843273842480124e-05,
      "loss": 1.425,
      "step": 15380
    },
    {
      "epoch": 0.5508263648897882,
      "grad_norm": 1.4543743133544922,
      "learning_rate": 8.842121734931316e-05,
      "loss": 1.6334,
      "step": 15381
    },
    {
      "epoch": 0.5508621770193565,
      "grad_norm": 1.4791871309280396,
      "learning_rate": 8.840969642960271e-05,
      "loss": 1.3785,
      "step": 15382
    },
    {
      "epoch": 0.5508979891489247,
      "grad_norm": 1.55195152759552,
      "learning_rate": 8.839817566582477e-05,
      "loss": 1.3623,
      "step": 15383
    },
    {
      "epoch": 0.550933801278493,
      "grad_norm": 1.8789154291152954,
      "learning_rate": 8.838665505813448e-05,
      "loss": 1.3948,
      "step": 15384
    },
    {
      "epoch": 0.5509696134080613,
      "grad_norm": 1.813826560974121,
      "learning_rate": 8.837513460668668e-05,
      "loss": 1.4754,
      "step": 15385
    },
    {
      "epoch": 0.5510054255376295,
      "grad_norm": 1.9964311122894287,
      "learning_rate": 8.836361431163653e-05,
      "loss": 1.4923,
      "step": 15386
    },
    {
      "epoch": 0.5510412376671979,
      "grad_norm": 1.5974700450897217,
      "learning_rate": 8.835209417313886e-05,
      "loss": 1.6589,
      "step": 15387
    },
    {
      "epoch": 0.5510770497967662,
      "grad_norm": 1.9215779304504395,
      "learning_rate": 8.834057419134883e-05,
      "loss": 1.7583,
      "step": 15388
    },
    {
      "epoch": 0.5511128619263345,
      "grad_norm": 1.61617910861969,
      "learning_rate": 8.832905436642125e-05,
      "loss": 1.4681,
      "step": 15389
    },
    {
      "epoch": 0.5511486740559027,
      "grad_norm": 1.5911004543304443,
      "learning_rate": 8.831753469851126e-05,
      "loss": 1.537,
      "step": 15390
    },
    {
      "epoch": 0.551184486185471,
      "grad_norm": 1.8469878435134888,
      "learning_rate": 8.830601518777375e-05,
      "loss": 1.4466,
      "step": 15391
    },
    {
      "epoch": 0.5512202983150393,
      "grad_norm": 1.4108341932296753,
      "learning_rate": 8.829449583436367e-05,
      "loss": 1.4935,
      "step": 15392
    },
    {
      "epoch": 0.5512561104446075,
      "grad_norm": 1.4014348983764648,
      "learning_rate": 8.828297663843612e-05,
      "loss": 1.5126,
      "step": 15393
    },
    {
      "epoch": 0.5512919225741759,
      "grad_norm": 2.2653427124023438,
      "learning_rate": 8.827145760014595e-05,
      "loss": 1.7555,
      "step": 15394
    },
    {
      "epoch": 0.5513277347037442,
      "grad_norm": 1.639029622077942,
      "learning_rate": 8.825993871964823e-05,
      "loss": 1.6526,
      "step": 15395
    },
    {
      "epoch": 0.5513635468333125,
      "grad_norm": 1.7813209295272827,
      "learning_rate": 8.824841999709785e-05,
      "loss": 1.1952,
      "step": 15396
    },
    {
      "epoch": 0.5513993589628807,
      "grad_norm": 1.8362061977386475,
      "learning_rate": 8.823690143264988e-05,
      "loss": 1.6557,
      "step": 15397
    },
    {
      "epoch": 0.551435171092449,
      "grad_norm": 1.6683216094970703,
      "learning_rate": 8.822538302645916e-05,
      "loss": 1.7202,
      "step": 15398
    },
    {
      "epoch": 0.5514709832220173,
      "grad_norm": 1.9283270835876465,
      "learning_rate": 8.821386477868078e-05,
      "loss": 1.1112,
      "step": 15399
    },
    {
      "epoch": 0.5515067953515855,
      "grad_norm": 1.510189175605774,
      "learning_rate": 8.820234668946963e-05,
      "loss": 1.5949,
      "step": 15400
    },
    {
      "epoch": 0.5515426074811539,
      "grad_norm": 1.425412893295288,
      "learning_rate": 8.819082875898068e-05,
      "loss": 1.4958,
      "step": 15401
    },
    {
      "epoch": 0.5515784196107222,
      "grad_norm": 1.7452600002288818,
      "learning_rate": 8.817931098736891e-05,
      "loss": 1.4762,
      "step": 15402
    },
    {
      "epoch": 0.5516142317402905,
      "grad_norm": 1.7585675716400146,
      "learning_rate": 8.816779337478923e-05,
      "loss": 1.4523,
      "step": 15403
    },
    {
      "epoch": 0.5516500438698587,
      "grad_norm": 2.050502300262451,
      "learning_rate": 8.815627592139665e-05,
      "loss": 1.1354,
      "step": 15404
    },
    {
      "epoch": 0.551685855999427,
      "grad_norm": 1.3366676568984985,
      "learning_rate": 8.814475862734608e-05,
      "loss": 1.568,
      "step": 15405
    },
    {
      "epoch": 0.5517216681289953,
      "grad_norm": 1.8632868528366089,
      "learning_rate": 8.813324149279254e-05,
      "loss": 1.3182,
      "step": 15406
    },
    {
      "epoch": 0.5517574802585635,
      "grad_norm": 1.6237183809280396,
      "learning_rate": 8.812172451789086e-05,
      "loss": 1.2567,
      "step": 15407
    },
    {
      "epoch": 0.5517932923881319,
      "grad_norm": 1.8499741554260254,
      "learning_rate": 8.811020770279612e-05,
      "loss": 1.4372,
      "step": 15408
    },
    {
      "epoch": 0.5518291045177002,
      "grad_norm": 1.5876901149749756,
      "learning_rate": 8.809869104766318e-05,
      "loss": 1.6619,
      "step": 15409
    },
    {
      "epoch": 0.5518649166472684,
      "grad_norm": 1.637397289276123,
      "learning_rate": 8.808717455264698e-05,
      "loss": 1.7995,
      "step": 15410
    },
    {
      "epoch": 0.5519007287768367,
      "grad_norm": 1.551072120666504,
      "learning_rate": 8.80756582179025e-05,
      "loss": 1.5143,
      "step": 15411
    },
    {
      "epoch": 0.551936540906405,
      "grad_norm": 1.4706289768218994,
      "learning_rate": 8.806414204358465e-05,
      "loss": 1.3013,
      "step": 15412
    },
    {
      "epoch": 0.5519723530359733,
      "grad_norm": 1.3856925964355469,
      "learning_rate": 8.805262602984838e-05,
      "loss": 1.4325,
      "step": 15413
    },
    {
      "epoch": 0.5520081651655415,
      "grad_norm": 1.3337805271148682,
      "learning_rate": 8.804111017684858e-05,
      "loss": 1.4605,
      "step": 15414
    },
    {
      "epoch": 0.5520439772951099,
      "grad_norm": 1.699302077293396,
      "learning_rate": 8.802959448474025e-05,
      "loss": 1.3212,
      "step": 15415
    },
    {
      "epoch": 0.5520797894246782,
      "grad_norm": 1.4418692588806152,
      "learning_rate": 8.801807895367827e-05,
      "loss": 1.4426,
      "step": 15416
    },
    {
      "epoch": 0.5521156015542464,
      "grad_norm": 1.6650549173355103,
      "learning_rate": 8.80065635838176e-05,
      "loss": 1.3448,
      "step": 15417
    },
    {
      "epoch": 0.5521514136838147,
      "grad_norm": 1.9519139528274536,
      "learning_rate": 8.799504837531315e-05,
      "loss": 1.7081,
      "step": 15418
    },
    {
      "epoch": 0.552187225813383,
      "grad_norm": 1.3951752185821533,
      "learning_rate": 8.798353332831981e-05,
      "loss": 1.3193,
      "step": 15419
    },
    {
      "epoch": 0.5522230379429512,
      "grad_norm": 1.665259599685669,
      "learning_rate": 8.797201844299257e-05,
      "loss": 1.5382,
      "step": 15420
    },
    {
      "epoch": 0.5522588500725195,
      "grad_norm": 1.426015019416809,
      "learning_rate": 8.796050371948627e-05,
      "loss": 1.5341,
      "step": 15421
    },
    {
      "epoch": 0.5522946622020879,
      "grad_norm": 1.8713806867599487,
      "learning_rate": 8.794898915795588e-05,
      "loss": 1.6166,
      "step": 15422
    },
    {
      "epoch": 0.5523304743316562,
      "grad_norm": 1.9937385320663452,
      "learning_rate": 8.793747475855628e-05,
      "loss": 1.2537,
      "step": 15423
    },
    {
      "epoch": 0.5523662864612244,
      "grad_norm": 1.5066399574279785,
      "learning_rate": 8.792596052144242e-05,
      "loss": 1.7567,
      "step": 15424
    },
    {
      "epoch": 0.5524020985907927,
      "grad_norm": 1.2568379640579224,
      "learning_rate": 8.791444644676916e-05,
      "loss": 1.3674,
      "step": 15425
    },
    {
      "epoch": 0.552437910720361,
      "grad_norm": 1.4235587120056152,
      "learning_rate": 8.790293253469145e-05,
      "loss": 1.661,
      "step": 15426
    },
    {
      "epoch": 0.5524737228499292,
      "grad_norm": 2.1775028705596924,
      "learning_rate": 8.789141878536419e-05,
      "loss": 1.3307,
      "step": 15427
    },
    {
      "epoch": 0.5525095349794975,
      "grad_norm": 1.7875803709030151,
      "learning_rate": 8.787990519894224e-05,
      "loss": 1.4441,
      "step": 15428
    },
    {
      "epoch": 0.5525453471090659,
      "grad_norm": 1.3177114725112915,
      "learning_rate": 8.786839177558057e-05,
      "loss": 1.7458,
      "step": 15429
    },
    {
      "epoch": 0.5525811592386342,
      "grad_norm": 1.7923847436904907,
      "learning_rate": 8.7856878515434e-05,
      "loss": 1.6531,
      "step": 15430
    },
    {
      "epoch": 0.5526169713682024,
      "grad_norm": 1.7845262289047241,
      "learning_rate": 8.784536541865752e-05,
      "loss": 1.382,
      "step": 15431
    },
    {
      "epoch": 0.5526527834977707,
      "grad_norm": 1.7725820541381836,
      "learning_rate": 8.783385248540591e-05,
      "loss": 1.2906,
      "step": 15432
    },
    {
      "epoch": 0.552688595627339,
      "grad_norm": 1.6670875549316406,
      "learning_rate": 8.782233971583416e-05,
      "loss": 1.7273,
      "step": 15433
    },
    {
      "epoch": 0.5527244077569072,
      "grad_norm": 2.1010894775390625,
      "learning_rate": 8.781082711009709e-05,
      "loss": 1.7307,
      "step": 15434
    },
    {
      "epoch": 0.5527602198864755,
      "grad_norm": 1.56618070602417,
      "learning_rate": 8.779931466834965e-05,
      "loss": 1.2643,
      "step": 15435
    },
    {
      "epoch": 0.5527960320160439,
      "grad_norm": 1.57991623878479,
      "learning_rate": 8.778780239074669e-05,
      "loss": 1.3456,
      "step": 15436
    },
    {
      "epoch": 0.5528318441456122,
      "grad_norm": 2.865915060043335,
      "learning_rate": 8.777629027744307e-05,
      "loss": 1.5232,
      "step": 15437
    },
    {
      "epoch": 0.5528676562751804,
      "grad_norm": 1.6558449268341064,
      "learning_rate": 8.776477832859374e-05,
      "loss": 1.4121,
      "step": 15438
    },
    {
      "epoch": 0.5529034684047487,
      "grad_norm": 2.0313730239868164,
      "learning_rate": 8.77532665443535e-05,
      "loss": 1.3529,
      "step": 15439
    },
    {
      "epoch": 0.552939280534317,
      "grad_norm": 1.625325322151184,
      "learning_rate": 8.774175492487728e-05,
      "loss": 1.4956,
      "step": 15440
    },
    {
      "epoch": 0.5529750926638852,
      "grad_norm": 1.4914642572402954,
      "learning_rate": 8.77302434703199e-05,
      "loss": 1.3326,
      "step": 15441
    },
    {
      "epoch": 0.5530109047934535,
      "grad_norm": 2.1764352321624756,
      "learning_rate": 8.771873218083631e-05,
      "loss": 1.5886,
      "step": 15442
    },
    {
      "epoch": 0.5530467169230219,
      "grad_norm": 1.5987873077392578,
      "learning_rate": 8.770722105658132e-05,
      "loss": 1.3776,
      "step": 15443
    },
    {
      "epoch": 0.5530825290525901,
      "grad_norm": 1.513716459274292,
      "learning_rate": 8.769571009770982e-05,
      "loss": 1.3559,
      "step": 15444
    },
    {
      "epoch": 0.5531183411821584,
      "grad_norm": 1.3887488842010498,
      "learning_rate": 8.768419930437667e-05,
      "loss": 1.6609,
      "step": 15445
    },
    {
      "epoch": 0.5531541533117267,
      "grad_norm": 1.3617610931396484,
      "learning_rate": 8.767268867673671e-05,
      "loss": 1.6054,
      "step": 15446
    },
    {
      "epoch": 0.553189965441295,
      "grad_norm": 1.9251405000686646,
      "learning_rate": 8.766117821494485e-05,
      "loss": 1.323,
      "step": 15447
    },
    {
      "epoch": 0.5532257775708632,
      "grad_norm": 2.4319498538970947,
      "learning_rate": 8.76496679191559e-05,
      "loss": 1.7116,
      "step": 15448
    },
    {
      "epoch": 0.5532615897004315,
      "grad_norm": 2.2947161197662354,
      "learning_rate": 8.763815778952475e-05,
      "loss": 1.504,
      "step": 15449
    },
    {
      "epoch": 0.5532974018299999,
      "grad_norm": 1.3940249681472778,
      "learning_rate": 8.762664782620623e-05,
      "loss": 1.6442,
      "step": 15450
    },
    {
      "epoch": 0.5533332139595681,
      "grad_norm": 1.715877890586853,
      "learning_rate": 8.761513802935523e-05,
      "loss": 1.6743,
      "step": 15451
    },
    {
      "epoch": 0.5533690260891364,
      "grad_norm": 1.2846055030822754,
      "learning_rate": 8.760362839912654e-05,
      "loss": 1.4043,
      "step": 15452
    },
    {
      "epoch": 0.5534048382187047,
      "grad_norm": 1.3711576461791992,
      "learning_rate": 8.759211893567505e-05,
      "loss": 1.5188,
      "step": 15453
    },
    {
      "epoch": 0.5534406503482729,
      "grad_norm": 1.988338589668274,
      "learning_rate": 8.758060963915562e-05,
      "loss": 1.5419,
      "step": 15454
    },
    {
      "epoch": 0.5534764624778412,
      "grad_norm": 2.278092384338379,
      "learning_rate": 8.756910050972304e-05,
      "loss": 1.5371,
      "step": 15455
    },
    {
      "epoch": 0.5535122746074095,
      "grad_norm": 1.3961970806121826,
      "learning_rate": 8.755759154753219e-05,
      "loss": 1.3566,
      "step": 15456
    },
    {
      "epoch": 0.5535480867369779,
      "grad_norm": 1.6591577529907227,
      "learning_rate": 8.754608275273788e-05,
      "loss": 1.2977,
      "step": 15457
    },
    {
      "epoch": 0.5535838988665461,
      "grad_norm": 1.8924493789672852,
      "learning_rate": 8.753457412549497e-05,
      "loss": 1.4803,
      "step": 15458
    },
    {
      "epoch": 0.5536197109961144,
      "grad_norm": 1.7182888984680176,
      "learning_rate": 8.752306566595828e-05,
      "loss": 1.3179,
      "step": 15459
    },
    {
      "epoch": 0.5536555231256827,
      "grad_norm": 1.6327793598175049,
      "learning_rate": 8.751155737428267e-05,
      "loss": 1.6394,
      "step": 15460
    },
    {
      "epoch": 0.5536913352552509,
      "grad_norm": 1.3792765140533447,
      "learning_rate": 8.750004925062296e-05,
      "loss": 1.649,
      "step": 15461
    },
    {
      "epoch": 0.5537271473848192,
      "grad_norm": 1.5709487199783325,
      "learning_rate": 8.74885412951339e-05,
      "loss": 1.5722,
      "step": 15462
    },
    {
      "epoch": 0.5537629595143875,
      "grad_norm": 2.683051347732544,
      "learning_rate": 8.747703350797044e-05,
      "loss": 1.532,
      "step": 15463
    },
    {
      "epoch": 0.5537987716439559,
      "grad_norm": 1.412434458732605,
      "learning_rate": 8.74655258892873e-05,
      "loss": 1.4681,
      "step": 15464
    },
    {
      "epoch": 0.5538345837735241,
      "grad_norm": 1.9447873830795288,
      "learning_rate": 8.745401843923936e-05,
      "loss": 1.7162,
      "step": 15465
    },
    {
      "epoch": 0.5538703959030924,
      "grad_norm": 1.498840093612671,
      "learning_rate": 8.74425111579814e-05,
      "loss": 1.3753,
      "step": 15466
    },
    {
      "epoch": 0.5539062080326607,
      "grad_norm": 1.703001618385315,
      "learning_rate": 8.743100404566828e-05,
      "loss": 1.6041,
      "step": 15467
    },
    {
      "epoch": 0.5539420201622289,
      "grad_norm": 1.341660737991333,
      "learning_rate": 8.741949710245476e-05,
      "loss": 1.349,
      "step": 15468
    },
    {
      "epoch": 0.5539778322917972,
      "grad_norm": 2.8604416847229004,
      "learning_rate": 8.740799032849572e-05,
      "loss": 1.397,
      "step": 15469
    },
    {
      "epoch": 0.5540136444213655,
      "grad_norm": 1.7713994979858398,
      "learning_rate": 8.739648372394592e-05,
      "loss": 1.7506,
      "step": 15470
    },
    {
      "epoch": 0.5540494565509338,
      "grad_norm": 1.9339109659194946,
      "learning_rate": 8.738497728896013e-05,
      "loss": 1.3016,
      "step": 15471
    },
    {
      "epoch": 0.5540852686805021,
      "grad_norm": 1.7749813795089722,
      "learning_rate": 8.737347102369325e-05,
      "loss": 1.0844,
      "step": 15472
    },
    {
      "epoch": 0.5541210808100704,
      "grad_norm": 2.379054307937622,
      "learning_rate": 8.736196492829997e-05,
      "loss": 1.71,
      "step": 15473
    },
    {
      "epoch": 0.5541568929396387,
      "grad_norm": 2.4072840213775635,
      "learning_rate": 8.735045900293522e-05,
      "loss": 1.4868,
      "step": 15474
    },
    {
      "epoch": 0.5541927050692069,
      "grad_norm": 1.7160054445266724,
      "learning_rate": 8.733895324775366e-05,
      "loss": 1.4881,
      "step": 15475
    },
    {
      "epoch": 0.5542285171987752,
      "grad_norm": 2.1116483211517334,
      "learning_rate": 8.73274476629102e-05,
      "loss": 1.5777,
      "step": 15476
    },
    {
      "epoch": 0.5542643293283435,
      "grad_norm": 1.7254281044006348,
      "learning_rate": 8.731594224855956e-05,
      "loss": 1.5687,
      "step": 15477
    },
    {
      "epoch": 0.5543001414579118,
      "grad_norm": 1.8794066905975342,
      "learning_rate": 8.730443700485658e-05,
      "loss": 1.4314,
      "step": 15478
    },
    {
      "epoch": 0.5543359535874801,
      "grad_norm": 1.509204387664795,
      "learning_rate": 8.729293193195603e-05,
      "loss": 1.3987,
      "step": 15479
    },
    {
      "epoch": 0.5543717657170484,
      "grad_norm": 1.610112190246582,
      "learning_rate": 8.728142703001264e-05,
      "loss": 1.4781,
      "step": 15480
    },
    {
      "epoch": 0.5544075778466167,
      "grad_norm": 2.039473295211792,
      "learning_rate": 8.72699222991813e-05,
      "loss": 1.499,
      "step": 15481
    },
    {
      "epoch": 0.5544433899761849,
      "grad_norm": 3.141580581665039,
      "learning_rate": 8.725841773961669e-05,
      "loss": 1.7102,
      "step": 15482
    },
    {
      "epoch": 0.5544792021057532,
      "grad_norm": 2.184549570083618,
      "learning_rate": 8.724691335147367e-05,
      "loss": 1.6045,
      "step": 15483
    },
    {
      "epoch": 0.5545150142353215,
      "grad_norm": 1.9326050281524658,
      "learning_rate": 8.723540913490693e-05,
      "loss": 1.5926,
      "step": 15484
    },
    {
      "epoch": 0.5545508263648898,
      "grad_norm": 1.7067333459854126,
      "learning_rate": 8.722390509007137e-05,
      "loss": 1.3034,
      "step": 15485
    },
    {
      "epoch": 0.5545866384944581,
      "grad_norm": 1.8227620124816895,
      "learning_rate": 8.721240121712161e-05,
      "loss": 1.5729,
      "step": 15486
    },
    {
      "epoch": 0.5546224506240264,
      "grad_norm": 1.3171581029891968,
      "learning_rate": 8.720089751621256e-05,
      "loss": 1.4384,
      "step": 15487
    },
    {
      "epoch": 0.5546582627535946,
      "grad_norm": 1.4920912981033325,
      "learning_rate": 8.71893939874989e-05,
      "loss": 1.294,
      "step": 15488
    },
    {
      "epoch": 0.5546940748831629,
      "grad_norm": 1.4924278259277344,
      "learning_rate": 8.717789063113539e-05,
      "loss": 1.5305,
      "step": 15489
    },
    {
      "epoch": 0.5547298870127312,
      "grad_norm": 1.9032200574874878,
      "learning_rate": 8.716638744727687e-05,
      "loss": 1.2737,
      "step": 15490
    },
    {
      "epoch": 0.5547656991422995,
      "grad_norm": 1.732216477394104,
      "learning_rate": 8.7154884436078e-05,
      "loss": 1.2741,
      "step": 15491
    },
    {
      "epoch": 0.5548015112718678,
      "grad_norm": 1.5945426225662231,
      "learning_rate": 8.714338159769366e-05,
      "loss": 1.5329,
      "step": 15492
    },
    {
      "epoch": 0.5548373234014361,
      "grad_norm": 1.8944737911224365,
      "learning_rate": 8.713187893227847e-05,
      "loss": 1.2802,
      "step": 15493
    },
    {
      "epoch": 0.5548731355310044,
      "grad_norm": 1.787889003753662,
      "learning_rate": 8.71203764399873e-05,
      "loss": 1.5427,
      "step": 15494
    },
    {
      "epoch": 0.5549089476605726,
      "grad_norm": 1.3428040742874146,
      "learning_rate": 8.71088741209748e-05,
      "loss": 1.4309,
      "step": 15495
    },
    {
      "epoch": 0.5549447597901409,
      "grad_norm": 1.8700381517410278,
      "learning_rate": 8.709737197539583e-05,
      "loss": 1.6271,
      "step": 15496
    },
    {
      "epoch": 0.5549805719197092,
      "grad_norm": 1.4687343835830688,
      "learning_rate": 8.708587000340506e-05,
      "loss": 1.4383,
      "step": 15497
    },
    {
      "epoch": 0.5550163840492774,
      "grad_norm": 1.5101184844970703,
      "learning_rate": 8.707436820515723e-05,
      "loss": 1.4074,
      "step": 15498
    },
    {
      "epoch": 0.5550521961788458,
      "grad_norm": 1.7550967931747437,
      "learning_rate": 8.706286658080711e-05,
      "loss": 1.6502,
      "step": 15499
    },
    {
      "epoch": 0.5550880083084141,
      "grad_norm": 1.6945569515228271,
      "learning_rate": 8.705136513050944e-05,
      "loss": 1.2085,
      "step": 15500
    },
    {
      "epoch": 0.5551238204379824,
      "grad_norm": 1.832401990890503,
      "learning_rate": 8.703986385441895e-05,
      "loss": 1.4714,
      "step": 15501
    },
    {
      "epoch": 0.5551596325675506,
      "grad_norm": 1.4257477521896362,
      "learning_rate": 8.702836275269033e-05,
      "loss": 1.3904,
      "step": 15502
    },
    {
      "epoch": 0.5551954446971189,
      "grad_norm": 1.3720061779022217,
      "learning_rate": 8.701686182547842e-05,
      "loss": 1.4555,
      "step": 15503
    },
    {
      "epoch": 0.5552312568266872,
      "grad_norm": 1.8269095420837402,
      "learning_rate": 8.700536107293784e-05,
      "loss": 1.4776,
      "step": 15504
    },
    {
      "epoch": 0.5552670689562554,
      "grad_norm": 1.655152440071106,
      "learning_rate": 8.699386049522341e-05,
      "loss": 1.6387,
      "step": 15505
    },
    {
      "epoch": 0.5553028810858238,
      "grad_norm": 1.5963819026947021,
      "learning_rate": 8.69823600924898e-05,
      "loss": 1.5153,
      "step": 15506
    },
    {
      "epoch": 0.5553386932153921,
      "grad_norm": 1.3038275241851807,
      "learning_rate": 8.697085986489172e-05,
      "loss": 1.6874,
      "step": 15507
    },
    {
      "epoch": 0.5553745053449604,
      "grad_norm": 1.676790714263916,
      "learning_rate": 8.695935981258394e-05,
      "loss": 1.2814,
      "step": 15508
    },
    {
      "epoch": 0.5554103174745286,
      "grad_norm": 1.8038157224655151,
      "learning_rate": 8.694785993572112e-05,
      "loss": 1.3005,
      "step": 15509
    },
    {
      "epoch": 0.5554461296040969,
      "grad_norm": 1.5027525424957275,
      "learning_rate": 8.693636023445804e-05,
      "loss": 1.4746,
      "step": 15510
    },
    {
      "epoch": 0.5554819417336652,
      "grad_norm": 1.7654838562011719,
      "learning_rate": 8.692486070894935e-05,
      "loss": 1.6604,
      "step": 15511
    },
    {
      "epoch": 0.5555177538632334,
      "grad_norm": 1.8913116455078125,
      "learning_rate": 8.691336135934982e-05,
      "loss": 1.6467,
      "step": 15512
    },
    {
      "epoch": 0.5555535659928018,
      "grad_norm": 1.5639491081237793,
      "learning_rate": 8.69018621858141e-05,
      "loss": 1.033,
      "step": 15513
    },
    {
      "epoch": 0.5555893781223701,
      "grad_norm": 1.46904718875885,
      "learning_rate": 8.689036318849697e-05,
      "loss": 1.4832,
      "step": 15514
    },
    {
      "epoch": 0.5556251902519384,
      "grad_norm": 1.6029126644134521,
      "learning_rate": 8.68788643675531e-05,
      "loss": 1.5136,
      "step": 15515
    },
    {
      "epoch": 0.5556610023815066,
      "grad_norm": 1.6579877138137817,
      "learning_rate": 8.686736572313714e-05,
      "loss": 1.373,
      "step": 15516
    },
    {
      "epoch": 0.5556968145110749,
      "grad_norm": 1.6767444610595703,
      "learning_rate": 8.685586725540387e-05,
      "loss": 1.3224,
      "step": 15517
    },
    {
      "epoch": 0.5557326266406432,
      "grad_norm": 1.8820594549179077,
      "learning_rate": 8.684436896450791e-05,
      "loss": 1.6569,
      "step": 15518
    },
    {
      "epoch": 0.5557684387702114,
      "grad_norm": 1.8585944175720215,
      "learning_rate": 8.683287085060404e-05,
      "loss": 1.594,
      "step": 15519
    },
    {
      "epoch": 0.5558042508997798,
      "grad_norm": 1.6034525632858276,
      "learning_rate": 8.682137291384687e-05,
      "loss": 1.4458,
      "step": 15520
    },
    {
      "epoch": 0.5558400630293481,
      "grad_norm": 1.3279809951782227,
      "learning_rate": 8.680987515439116e-05,
      "loss": 1.4297,
      "step": 15521
    },
    {
      "epoch": 0.5558758751589163,
      "grad_norm": 1.51862370967865,
      "learning_rate": 8.679837757239156e-05,
      "loss": 1.6286,
      "step": 15522
    },
    {
      "epoch": 0.5559116872884846,
      "grad_norm": 1.6439845561981201,
      "learning_rate": 8.678688016800276e-05,
      "loss": 1.2965,
      "step": 15523
    },
    {
      "epoch": 0.5559474994180529,
      "grad_norm": 1.7435225248336792,
      "learning_rate": 8.677538294137945e-05,
      "loss": 1.6844,
      "step": 15524
    },
    {
      "epoch": 0.5559833115476212,
      "grad_norm": 1.3427046537399292,
      "learning_rate": 8.676388589267628e-05,
      "loss": 1.3433,
      "step": 15525
    },
    {
      "epoch": 0.5560191236771894,
      "grad_norm": 1.8325779438018799,
      "learning_rate": 8.675238902204797e-05,
      "loss": 1.5351,
      "step": 15526
    },
    {
      "epoch": 0.5560549358067578,
      "grad_norm": 1.5658588409423828,
      "learning_rate": 8.674089232964916e-05,
      "loss": 1.7484,
      "step": 15527
    },
    {
      "epoch": 0.5560907479363261,
      "grad_norm": 1.382843017578125,
      "learning_rate": 8.672939581563456e-05,
      "loss": 1.4252,
      "step": 15528
    },
    {
      "epoch": 0.5561265600658943,
      "grad_norm": 2.3734495639801025,
      "learning_rate": 8.67178994801588e-05,
      "loss": 1.3529,
      "step": 15529
    },
    {
      "epoch": 0.5561623721954626,
      "grad_norm": 1.3029054403305054,
      "learning_rate": 8.67064033233766e-05,
      "loss": 1.331,
      "step": 15530
    },
    {
      "epoch": 0.5561981843250309,
      "grad_norm": 1.900862455368042,
      "learning_rate": 8.669490734544256e-05,
      "loss": 1.6135,
      "step": 15531
    },
    {
      "epoch": 0.5562339964545991,
      "grad_norm": 1.9537557363510132,
      "learning_rate": 8.668341154651141e-05,
      "loss": 1.5744,
      "step": 15532
    },
    {
      "epoch": 0.5562698085841674,
      "grad_norm": 1.6260302066802979,
      "learning_rate": 8.667191592673779e-05,
      "loss": 1.5679,
      "step": 15533
    },
    {
      "epoch": 0.5563056207137358,
      "grad_norm": 2.043226957321167,
      "learning_rate": 8.666042048627632e-05,
      "loss": 1.6189,
      "step": 15534
    },
    {
      "epoch": 0.5563414328433041,
      "grad_norm": 1.6133548021316528,
      "learning_rate": 8.66489252252817e-05,
      "loss": 1.6424,
      "step": 15535
    },
    {
      "epoch": 0.5563772449728723,
      "grad_norm": 1.5509310960769653,
      "learning_rate": 8.663743014390855e-05,
      "loss": 1.4819,
      "step": 15536
    },
    {
      "epoch": 0.5564130571024406,
      "grad_norm": 1.5829284191131592,
      "learning_rate": 8.662593524231158e-05,
      "loss": 1.2861,
      "step": 15537
    },
    {
      "epoch": 0.5564488692320089,
      "grad_norm": 1.2532970905303955,
      "learning_rate": 8.661444052064536e-05,
      "loss": 1.128,
      "step": 15538
    },
    {
      "epoch": 0.5564846813615771,
      "grad_norm": 1.8142882585525513,
      "learning_rate": 8.66029459790646e-05,
      "loss": 1.4828,
      "step": 15539
    },
    {
      "epoch": 0.5565204934911454,
      "grad_norm": 1.6689441204071045,
      "learning_rate": 8.65914516177239e-05,
      "loss": 1.5474,
      "step": 15540
    },
    {
      "epoch": 0.5565563056207138,
      "grad_norm": 1.3482906818389893,
      "learning_rate": 8.657995743677793e-05,
      "loss": 1.5433,
      "step": 15541
    },
    {
      "epoch": 0.5565921177502821,
      "grad_norm": 2.2926156520843506,
      "learning_rate": 8.656846343638135e-05,
      "loss": 1.6321,
      "step": 15542
    },
    {
      "epoch": 0.5566279298798503,
      "grad_norm": 1.465704083442688,
      "learning_rate": 8.655696961668873e-05,
      "loss": 1.5127,
      "step": 15543
    },
    {
      "epoch": 0.5566637420094186,
      "grad_norm": 2.068370819091797,
      "learning_rate": 8.654547597785478e-05,
      "loss": 1.6719,
      "step": 15544
    },
    {
      "epoch": 0.5566995541389869,
      "grad_norm": 1.6043318510055542,
      "learning_rate": 8.653398252003406e-05,
      "loss": 1.426,
      "step": 15545
    },
    {
      "epoch": 0.5567353662685551,
      "grad_norm": 1.5261003971099854,
      "learning_rate": 8.652248924338126e-05,
      "loss": 1.507,
      "step": 15546
    },
    {
      "epoch": 0.5567711783981234,
      "grad_norm": 1.734169363975525,
      "learning_rate": 8.651099614805097e-05,
      "loss": 1.516,
      "step": 15547
    },
    {
      "epoch": 0.5568069905276918,
      "grad_norm": 1.581032633781433,
      "learning_rate": 8.649950323419783e-05,
      "loss": 1.2642,
      "step": 15548
    },
    {
      "epoch": 0.55684280265726,
      "grad_norm": 1.9459565877914429,
      "learning_rate": 8.648801050197646e-05,
      "loss": 1.6195,
      "step": 15549
    },
    {
      "epoch": 0.5568786147868283,
      "grad_norm": 1.9734375476837158,
      "learning_rate": 8.647651795154148e-05,
      "loss": 1.366,
      "step": 15550
    },
    {
      "epoch": 0.5569144269163966,
      "grad_norm": 1.4467110633850098,
      "learning_rate": 8.646502558304751e-05,
      "loss": 1.4128,
      "step": 15551
    },
    {
      "epoch": 0.5569502390459649,
      "grad_norm": 2.0393340587615967,
      "learning_rate": 8.645353339664915e-05,
      "loss": 1.4668,
      "step": 15552
    },
    {
      "epoch": 0.5569860511755331,
      "grad_norm": 1.865412950515747,
      "learning_rate": 8.644204139250105e-05,
      "loss": 1.4342,
      "step": 15553
    },
    {
      "epoch": 0.5570218633051014,
      "grad_norm": 1.486951470375061,
      "learning_rate": 8.643054957075776e-05,
      "loss": 1.3204,
      "step": 15554
    },
    {
      "epoch": 0.5570576754346698,
      "grad_norm": 1.633739709854126,
      "learning_rate": 8.641905793157395e-05,
      "loss": 1.3796,
      "step": 15555
    },
    {
      "epoch": 0.557093487564238,
      "grad_norm": 1.7872612476348877,
      "learning_rate": 8.640756647510417e-05,
      "loss": 1.6145,
      "step": 15556
    },
    {
      "epoch": 0.5571292996938063,
      "grad_norm": 1.5471066236495972,
      "learning_rate": 8.639607520150308e-05,
      "loss": 1.4271,
      "step": 15557
    },
    {
      "epoch": 0.5571651118233746,
      "grad_norm": 1.6009304523468018,
      "learning_rate": 8.638458411092527e-05,
      "loss": 1.2723,
      "step": 15558
    },
    {
      "epoch": 0.5572009239529429,
      "grad_norm": 1.389464259147644,
      "learning_rate": 8.637309320352526e-05,
      "loss": 1.6569,
      "step": 15559
    },
    {
      "epoch": 0.5572367360825111,
      "grad_norm": 1.8739726543426514,
      "learning_rate": 8.636160247945774e-05,
      "loss": 1.6658,
      "step": 15560
    },
    {
      "epoch": 0.5572725482120794,
      "grad_norm": 1.4336637258529663,
      "learning_rate": 8.635011193887725e-05,
      "loss": 1.4241,
      "step": 15561
    },
    {
      "epoch": 0.5573083603416478,
      "grad_norm": 1.8983516693115234,
      "learning_rate": 8.633862158193841e-05,
      "loss": 1.4288,
      "step": 15562
    },
    {
      "epoch": 0.557344172471216,
      "grad_norm": 1.7363568544387817,
      "learning_rate": 8.632713140879577e-05,
      "loss": 1.6088,
      "step": 15563
    },
    {
      "epoch": 0.5573799846007843,
      "grad_norm": 1.7875018119812012,
      "learning_rate": 8.631564141960397e-05,
      "loss": 1.4726,
      "step": 15564
    },
    {
      "epoch": 0.5574157967303526,
      "grad_norm": 1.811866044998169,
      "learning_rate": 8.630415161451754e-05,
      "loss": 1.6785,
      "step": 15565
    },
    {
      "epoch": 0.5574516088599208,
      "grad_norm": 1.505098581314087,
      "learning_rate": 8.62926619936911e-05,
      "loss": 1.4012,
      "step": 15566
    },
    {
      "epoch": 0.5574874209894891,
      "grad_norm": 1.4308191537857056,
      "learning_rate": 8.628117255727924e-05,
      "loss": 1.4559,
      "step": 15567
    },
    {
      "epoch": 0.5575232331190574,
      "grad_norm": 1.7727469205856323,
      "learning_rate": 8.626968330543643e-05,
      "loss": 1.4809,
      "step": 15568
    },
    {
      "epoch": 0.5575590452486258,
      "grad_norm": 1.7335854768753052,
      "learning_rate": 8.62581942383174e-05,
      "loss": 1.5017,
      "step": 15569
    },
    {
      "epoch": 0.557594857378194,
      "grad_norm": 1.486517071723938,
      "learning_rate": 8.624670535607658e-05,
      "loss": 1.3723,
      "step": 15570
    },
    {
      "epoch": 0.5576306695077623,
      "grad_norm": 2.2205562591552734,
      "learning_rate": 8.623521665886865e-05,
      "loss": 1.7048,
      "step": 15571
    },
    {
      "epoch": 0.5576664816373306,
      "grad_norm": 1.4245879650115967,
      "learning_rate": 8.622372814684806e-05,
      "loss": 1.4024,
      "step": 15572
    },
    {
      "epoch": 0.5577022937668988,
      "grad_norm": 1.3684970140457153,
      "learning_rate": 8.621223982016948e-05,
      "loss": 1.4418,
      "step": 15573
    },
    {
      "epoch": 0.5577381058964671,
      "grad_norm": 1.8130236864089966,
      "learning_rate": 8.620075167898743e-05,
      "loss": 1.4848,
      "step": 15574
    },
    {
      "epoch": 0.5577739180260354,
      "grad_norm": 1.3739123344421387,
      "learning_rate": 8.618926372345645e-05,
      "loss": 1.583,
      "step": 15575
    },
    {
      "epoch": 0.5578097301556038,
      "grad_norm": 1.366071105003357,
      "learning_rate": 8.617777595373117e-05,
      "loss": 1.4026,
      "step": 15576
    },
    {
      "epoch": 0.557845542285172,
      "grad_norm": 1.5953060388565063,
      "learning_rate": 8.6166288369966e-05,
      "loss": 1.2789,
      "step": 15577
    },
    {
      "epoch": 0.5578813544147403,
      "grad_norm": 1.4939026832580566,
      "learning_rate": 8.615480097231564e-05,
      "loss": 1.1593,
      "step": 15578
    },
    {
      "epoch": 0.5579171665443086,
      "grad_norm": 1.7710267305374146,
      "learning_rate": 8.614331376093452e-05,
      "loss": 1.7925,
      "step": 15579
    },
    {
      "epoch": 0.5579529786738768,
      "grad_norm": 1.9933295249938965,
      "learning_rate": 8.613182673597729e-05,
      "loss": 1.5553,
      "step": 15580
    },
    {
      "epoch": 0.5579887908034451,
      "grad_norm": 2.1700258255004883,
      "learning_rate": 8.612033989759838e-05,
      "loss": 1.4692,
      "step": 15581
    },
    {
      "epoch": 0.5580246029330134,
      "grad_norm": 1.4422943592071533,
      "learning_rate": 8.610885324595249e-05,
      "loss": 1.5315,
      "step": 15582
    },
    {
      "epoch": 0.5580604150625817,
      "grad_norm": 1.459005355834961,
      "learning_rate": 8.609736678119396e-05,
      "loss": 1.4583,
      "step": 15583
    },
    {
      "epoch": 0.55809622719215,
      "grad_norm": 1.8012620210647583,
      "learning_rate": 8.60858805034775e-05,
      "loss": 1.6828,
      "step": 15584
    },
    {
      "epoch": 0.5581320393217183,
      "grad_norm": 1.4135667085647583,
      "learning_rate": 8.607439441295755e-05,
      "loss": 1.4163,
      "step": 15585
    },
    {
      "epoch": 0.5581678514512866,
      "grad_norm": 1.6235884428024292,
      "learning_rate": 8.606290850978862e-05,
      "loss": 1.4135,
      "step": 15586
    },
    {
      "epoch": 0.5582036635808548,
      "grad_norm": 2.004629611968994,
      "learning_rate": 8.605142279412533e-05,
      "loss": 1.7444,
      "step": 15587
    },
    {
      "epoch": 0.5582394757104231,
      "grad_norm": 1.713678002357483,
      "learning_rate": 8.60399372661221e-05,
      "loss": 1.7045,
      "step": 15588
    },
    {
      "epoch": 0.5582752878399914,
      "grad_norm": 1.9293084144592285,
      "learning_rate": 8.602845192593359e-05,
      "loss": 1.1365,
      "step": 15589
    },
    {
      "epoch": 0.5583110999695597,
      "grad_norm": 1.6164830923080444,
      "learning_rate": 8.601696677371414e-05,
      "loss": 1.6059,
      "step": 15590
    },
    {
      "epoch": 0.558346912099128,
      "grad_norm": 2.2314774990081787,
      "learning_rate": 8.600548180961845e-05,
      "loss": 1.342,
      "step": 15591
    },
    {
      "epoch": 0.5583827242286963,
      "grad_norm": 1.5393816232681274,
      "learning_rate": 8.599399703380087e-05,
      "loss": 1.3335,
      "step": 15592
    },
    {
      "epoch": 0.5584185363582646,
      "grad_norm": 1.6473439931869507,
      "learning_rate": 8.598251244641608e-05,
      "loss": 1.4625,
      "step": 15593
    },
    {
      "epoch": 0.5584543484878328,
      "grad_norm": 1.7329156398773193,
      "learning_rate": 8.597102804761846e-05,
      "loss": 1.6533,
      "step": 15594
    },
    {
      "epoch": 0.5584901606174011,
      "grad_norm": 1.7898093461990356,
      "learning_rate": 8.595954383756256e-05,
      "loss": 1.3945,
      "step": 15595
    },
    {
      "epoch": 0.5585259727469694,
      "grad_norm": 1.6420942544937134,
      "learning_rate": 8.594805981640289e-05,
      "loss": 1.5115,
      "step": 15596
    },
    {
      "epoch": 0.5585617848765377,
      "grad_norm": 1.4303438663482666,
      "learning_rate": 8.593657598429395e-05,
      "loss": 1.4199,
      "step": 15597
    },
    {
      "epoch": 0.558597597006106,
      "grad_norm": 1.711222767829895,
      "learning_rate": 8.592509234139023e-05,
      "loss": 1.3911,
      "step": 15598
    },
    {
      "epoch": 0.5586334091356743,
      "grad_norm": 1.6813573837280273,
      "learning_rate": 8.591360888784622e-05,
      "loss": 1.3908,
      "step": 15599
    },
    {
      "epoch": 0.5586692212652425,
      "grad_norm": 1.3513386249542236,
      "learning_rate": 8.590212562381649e-05,
      "loss": 1.5484,
      "step": 15600
    },
    {
      "epoch": 0.5587050333948108,
      "grad_norm": 2.3890459537506104,
      "learning_rate": 8.589064254945541e-05,
      "loss": 1.5111,
      "step": 15601
    },
    {
      "epoch": 0.5587408455243791,
      "grad_norm": 1.4889732599258423,
      "learning_rate": 8.58791596649176e-05,
      "loss": 1.4165,
      "step": 15602
    },
    {
      "epoch": 0.5587766576539474,
      "grad_norm": 2.1360843181610107,
      "learning_rate": 8.586767697035745e-05,
      "loss": 1.5494,
      "step": 15603
    },
    {
      "epoch": 0.5588124697835157,
      "grad_norm": 1.2866389751434326,
      "learning_rate": 8.585619446592947e-05,
      "loss": 1.4871,
      "step": 15604
    },
    {
      "epoch": 0.558848281913084,
      "grad_norm": 1.881407380104065,
      "learning_rate": 8.584471215178817e-05,
      "loss": 1.5327,
      "step": 15605
    },
    {
      "epoch": 0.5588840940426523,
      "grad_norm": 2.2813119888305664,
      "learning_rate": 8.583323002808798e-05,
      "loss": 1.8198,
      "step": 15606
    },
    {
      "epoch": 0.5589199061722205,
      "grad_norm": 1.8343212604522705,
      "learning_rate": 8.582174809498343e-05,
      "loss": 1.5162,
      "step": 15607
    },
    {
      "epoch": 0.5589557183017888,
      "grad_norm": 1.4812381267547607,
      "learning_rate": 8.581026635262894e-05,
      "loss": 1.6672,
      "step": 15608
    },
    {
      "epoch": 0.5589915304313571,
      "grad_norm": 1.795681357383728,
      "learning_rate": 8.579878480117906e-05,
      "loss": 1.361,
      "step": 15609
    },
    {
      "epoch": 0.5590273425609253,
      "grad_norm": 1.547515630722046,
      "learning_rate": 8.578730344078817e-05,
      "loss": 1.1694,
      "step": 15610
    },
    {
      "epoch": 0.5590631546904937,
      "grad_norm": 2.669339656829834,
      "learning_rate": 8.577582227161081e-05,
      "loss": 1.6002,
      "step": 15611
    },
    {
      "epoch": 0.559098966820062,
      "grad_norm": 1.5305811166763306,
      "learning_rate": 8.57643412938014e-05,
      "loss": 1.2915,
      "step": 15612
    },
    {
      "epoch": 0.5591347789496303,
      "grad_norm": 2.1755855083465576,
      "learning_rate": 8.575286050751441e-05,
      "loss": 1.5044,
      "step": 15613
    },
    {
      "epoch": 0.5591705910791985,
      "grad_norm": 1.5070805549621582,
      "learning_rate": 8.574137991290432e-05,
      "loss": 1.6467,
      "step": 15614
    },
    {
      "epoch": 0.5592064032087668,
      "grad_norm": 1.981643795967102,
      "learning_rate": 8.572989951012554e-05,
      "loss": 1.484,
      "step": 15615
    },
    {
      "epoch": 0.5592422153383351,
      "grad_norm": 1.5245190858840942,
      "learning_rate": 8.571841929933258e-05,
      "loss": 1.5356,
      "step": 15616
    },
    {
      "epoch": 0.5592780274679033,
      "grad_norm": 1.464753270149231,
      "learning_rate": 8.570693928067986e-05,
      "loss": 1.2764,
      "step": 15617
    },
    {
      "epoch": 0.5593138395974717,
      "grad_norm": 1.790344476699829,
      "learning_rate": 8.569545945432185e-05,
      "loss": 1.3695,
      "step": 15618
    },
    {
      "epoch": 0.55934965172704,
      "grad_norm": 2.0477633476257324,
      "learning_rate": 8.568397982041294e-05,
      "loss": 1.5747,
      "step": 15619
    },
    {
      "epoch": 0.5593854638566083,
      "grad_norm": 1.3731274604797363,
      "learning_rate": 8.567250037910766e-05,
      "loss": 1.5434,
      "step": 15620
    },
    {
      "epoch": 0.5594212759861765,
      "grad_norm": 2.218919277191162,
      "learning_rate": 8.56610211305604e-05,
      "loss": 1.7308,
      "step": 15621
    },
    {
      "epoch": 0.5594570881157448,
      "grad_norm": 1.6405763626098633,
      "learning_rate": 8.564954207492558e-05,
      "loss": 1.5039,
      "step": 15622
    },
    {
      "epoch": 0.5594929002453131,
      "grad_norm": 1.4550385475158691,
      "learning_rate": 8.56380632123577e-05,
      "loss": 1.303,
      "step": 15623
    },
    {
      "epoch": 0.5595287123748813,
      "grad_norm": 2.2137668132781982,
      "learning_rate": 8.56265845430111e-05,
      "loss": 1.7529,
      "step": 15624
    },
    {
      "epoch": 0.5595645245044497,
      "grad_norm": 1.7952656745910645,
      "learning_rate": 8.561510606704031e-05,
      "loss": 1.4931,
      "step": 15625
    },
    {
      "epoch": 0.559600336634018,
      "grad_norm": 2.030855417251587,
      "learning_rate": 8.560362778459968e-05,
      "loss": 1.302,
      "step": 15626
    },
    {
      "epoch": 0.5596361487635862,
      "grad_norm": 1.3937506675720215,
      "learning_rate": 8.55921496958437e-05,
      "loss": 1.5187,
      "step": 15627
    },
    {
      "epoch": 0.5596719608931545,
      "grad_norm": 2.013484477996826,
      "learning_rate": 8.558067180092673e-05,
      "loss": 1.5022,
      "step": 15628
    },
    {
      "epoch": 0.5597077730227228,
      "grad_norm": 1.5687587261199951,
      "learning_rate": 8.556919410000323e-05,
      "loss": 1.4391,
      "step": 15629
    },
    {
      "epoch": 0.5597435851522911,
      "grad_norm": 1.5793520212173462,
      "learning_rate": 8.555771659322765e-05,
      "loss": 1.5087,
      "step": 15630
    },
    {
      "epoch": 0.5597793972818593,
      "grad_norm": 1.5034009218215942,
      "learning_rate": 8.55462392807543e-05,
      "loss": 1.4148,
      "step": 15631
    },
    {
      "epoch": 0.5598152094114277,
      "grad_norm": 1.5135999917984009,
      "learning_rate": 8.55347621627377e-05,
      "loss": 1.3035,
      "step": 15632
    },
    {
      "epoch": 0.559851021540996,
      "grad_norm": 1.4237425327301025,
      "learning_rate": 8.55232852393322e-05,
      "loss": 1.4377,
      "step": 15633
    },
    {
      "epoch": 0.5598868336705642,
      "grad_norm": 1.6626579761505127,
      "learning_rate": 8.551180851069222e-05,
      "loss": 1.3463,
      "step": 15634
    },
    {
      "epoch": 0.5599226458001325,
      "grad_norm": 1.5885761976242065,
      "learning_rate": 8.550033197697218e-05,
      "loss": 1.3632,
      "step": 15635
    },
    {
      "epoch": 0.5599584579297008,
      "grad_norm": 1.4434884786605835,
      "learning_rate": 8.548885563832646e-05,
      "loss": 1.2362,
      "step": 15636
    },
    {
      "epoch": 0.559994270059269,
      "grad_norm": 1.8405886888504028,
      "learning_rate": 8.547737949490946e-05,
      "loss": 1.6275,
      "step": 15637
    },
    {
      "epoch": 0.5600300821888373,
      "grad_norm": 1.6832880973815918,
      "learning_rate": 8.546590354687562e-05,
      "loss": 1.7815,
      "step": 15638
    },
    {
      "epoch": 0.5600658943184057,
      "grad_norm": 1.6571459770202637,
      "learning_rate": 8.545442779437928e-05,
      "loss": 1.4971,
      "step": 15639
    },
    {
      "epoch": 0.560101706447974,
      "grad_norm": 1.7768093347549438,
      "learning_rate": 8.544295223757484e-05,
      "loss": 1.6878,
      "step": 15640
    },
    {
      "epoch": 0.5601375185775422,
      "grad_norm": 1.4609851837158203,
      "learning_rate": 8.543147687661673e-05,
      "loss": 1.3395,
      "step": 15641
    },
    {
      "epoch": 0.5601733307071105,
      "grad_norm": 1.545523762702942,
      "learning_rate": 8.542000171165928e-05,
      "loss": 1.5403,
      "step": 15642
    },
    {
      "epoch": 0.5602091428366788,
      "grad_norm": 1.7643197774887085,
      "learning_rate": 8.540852674285691e-05,
      "loss": 1.7647,
      "step": 15643
    },
    {
      "epoch": 0.560244954966247,
      "grad_norm": 1.7095211744308472,
      "learning_rate": 8.539705197036398e-05,
      "loss": 1.1087,
      "step": 15644
    },
    {
      "epoch": 0.5602807670958153,
      "grad_norm": 1.9334076642990112,
      "learning_rate": 8.53855773943349e-05,
      "loss": 1.6575,
      "step": 15645
    },
    {
      "epoch": 0.5603165792253836,
      "grad_norm": 1.9181638956069946,
      "learning_rate": 8.537410301492398e-05,
      "loss": 1.4329,
      "step": 15646
    },
    {
      "epoch": 0.560352391354952,
      "grad_norm": 1.4931620359420776,
      "learning_rate": 8.53626288322857e-05,
      "loss": 1.3542,
      "step": 15647
    },
    {
      "epoch": 0.5603882034845202,
      "grad_norm": 1.7490475177764893,
      "learning_rate": 8.535115484657434e-05,
      "loss": 1.2913,
      "step": 15648
    },
    {
      "epoch": 0.5604240156140885,
      "grad_norm": 1.412573218345642,
      "learning_rate": 8.533968105794428e-05,
      "loss": 1.453,
      "step": 15649
    },
    {
      "epoch": 0.5604598277436568,
      "grad_norm": 1.6083836555480957,
      "learning_rate": 8.532820746654993e-05,
      "loss": 1.6949,
      "step": 15650
    },
    {
      "epoch": 0.560495639873225,
      "grad_norm": 1.9700466394424438,
      "learning_rate": 8.53167340725456e-05,
      "loss": 1.4605,
      "step": 15651
    },
    {
      "epoch": 0.5605314520027933,
      "grad_norm": 1.7195794582366943,
      "learning_rate": 8.530526087608569e-05,
      "loss": 1.54,
      "step": 15652
    },
    {
      "epoch": 0.5605672641323616,
      "grad_norm": 1.3336139917373657,
      "learning_rate": 8.529378787732451e-05,
      "loss": 1.5449,
      "step": 15653
    },
    {
      "epoch": 0.56060307626193,
      "grad_norm": 1.7417322397232056,
      "learning_rate": 8.528231507641648e-05,
      "loss": 1.2722,
      "step": 15654
    },
    {
      "epoch": 0.5606388883914982,
      "grad_norm": 1.8747280836105347,
      "learning_rate": 8.527084247351595e-05,
      "loss": 1.2224,
      "step": 15655
    },
    {
      "epoch": 0.5606747005210665,
      "grad_norm": 1.4631984233856201,
      "learning_rate": 8.525937006877714e-05,
      "loss": 1.2657,
      "step": 15656
    },
    {
      "epoch": 0.5607105126506348,
      "grad_norm": 2.2184841632843018,
      "learning_rate": 8.524789786235458e-05,
      "loss": 1.437,
      "step": 15657
    },
    {
      "epoch": 0.560746324780203,
      "grad_norm": 1.5433226823806763,
      "learning_rate": 8.523642585440245e-05,
      "loss": 1.5117,
      "step": 15658
    },
    {
      "epoch": 0.5607821369097713,
      "grad_norm": 1.7362242937088013,
      "learning_rate": 8.522495404507521e-05,
      "loss": 1.7659,
      "step": 15659
    },
    {
      "epoch": 0.5608179490393396,
      "grad_norm": 1.6187702417373657,
      "learning_rate": 8.521348243452714e-05,
      "loss": 1.3126,
      "step": 15660
    },
    {
      "epoch": 0.560853761168908,
      "grad_norm": 1.7105422019958496,
      "learning_rate": 8.52020110229126e-05,
      "loss": 1.468,
      "step": 15661
    },
    {
      "epoch": 0.5608895732984762,
      "grad_norm": 1.5425986051559448,
      "learning_rate": 8.51905398103859e-05,
      "loss": 1.219,
      "step": 15662
    },
    {
      "epoch": 0.5609253854280445,
      "grad_norm": 1.5455344915390015,
      "learning_rate": 8.51790687971014e-05,
      "loss": 1.4256,
      "step": 15663
    },
    {
      "epoch": 0.5609611975576128,
      "grad_norm": 1.4394350051879883,
      "learning_rate": 8.516759798321345e-05,
      "loss": 1.3927,
      "step": 15664
    },
    {
      "epoch": 0.560997009687181,
      "grad_norm": 1.6390577554702759,
      "learning_rate": 8.515612736887627e-05,
      "loss": 1.3861,
      "step": 15665
    },
    {
      "epoch": 0.5610328218167493,
      "grad_norm": 1.270395040512085,
      "learning_rate": 8.51446569542443e-05,
      "loss": 1.0775,
      "step": 15666
    },
    {
      "epoch": 0.5610686339463176,
      "grad_norm": 2.1888267993927,
      "learning_rate": 8.513318673947173e-05,
      "loss": 1.8735,
      "step": 15667
    },
    {
      "epoch": 0.5611044460758859,
      "grad_norm": 1.4443387985229492,
      "learning_rate": 8.512171672471305e-05,
      "loss": 1.3064,
      "step": 15668
    },
    {
      "epoch": 0.5611402582054542,
      "grad_norm": 1.3692225217819214,
      "learning_rate": 8.51102469101224e-05,
      "loss": 1.2717,
      "step": 15669
    },
    {
      "epoch": 0.5611760703350225,
      "grad_norm": 1.8987263441085815,
      "learning_rate": 8.509877729585423e-05,
      "loss": 1.5354,
      "step": 15670
    },
    {
      "epoch": 0.5612118824645907,
      "grad_norm": 1.4163073301315308,
      "learning_rate": 8.508730788206273e-05,
      "loss": 1.1945,
      "step": 15671
    },
    {
      "epoch": 0.561247694594159,
      "grad_norm": 1.583075761795044,
      "learning_rate": 8.507583866890233e-05,
      "loss": 1.669,
      "step": 15672
    },
    {
      "epoch": 0.5612835067237273,
      "grad_norm": 1.6579574346542358,
      "learning_rate": 8.506436965652728e-05,
      "loss": 1.2003,
      "step": 15673
    },
    {
      "epoch": 0.5613193188532956,
      "grad_norm": 3.4547436237335205,
      "learning_rate": 8.50529008450918e-05,
      "loss": 1.6151,
      "step": 15674
    },
    {
      "epoch": 0.5613551309828639,
      "grad_norm": 1.7894209623336792,
      "learning_rate": 8.504143223475031e-05,
      "loss": 1.4897,
      "step": 15675
    },
    {
      "epoch": 0.5613909431124322,
      "grad_norm": 1.423621416091919,
      "learning_rate": 8.502996382565702e-05,
      "loss": 1.6746,
      "step": 15676
    },
    {
      "epoch": 0.5614267552420005,
      "grad_norm": 1.8405758142471313,
      "learning_rate": 8.501849561796631e-05,
      "loss": 1.2419,
      "step": 15677
    },
    {
      "epoch": 0.5614625673715687,
      "grad_norm": 1.9967997074127197,
      "learning_rate": 8.500702761183234e-05,
      "loss": 1.4297,
      "step": 15678
    },
    {
      "epoch": 0.561498379501137,
      "grad_norm": 2.1513876914978027,
      "learning_rate": 8.499555980740956e-05,
      "loss": 1.4807,
      "step": 15679
    },
    {
      "epoch": 0.5615341916307053,
      "grad_norm": 1.7307462692260742,
      "learning_rate": 8.498409220485208e-05,
      "loss": 1.5421,
      "step": 15680
    },
    {
      "epoch": 0.5615700037602736,
      "grad_norm": 1.9727336168289185,
      "learning_rate": 8.497262480431435e-05,
      "loss": 1.393,
      "step": 15681
    },
    {
      "epoch": 0.5616058158898419,
      "grad_norm": 1.4967458248138428,
      "learning_rate": 8.496115760595054e-05,
      "loss": 1.4945,
      "step": 15682
    },
    {
      "epoch": 0.5616416280194102,
      "grad_norm": 1.8036267757415771,
      "learning_rate": 8.494969060991493e-05,
      "loss": 1.5742,
      "step": 15683
    },
    {
      "epoch": 0.5616774401489785,
      "grad_norm": 1.3200416564941406,
      "learning_rate": 8.493822381636185e-05,
      "loss": 1.4853,
      "step": 15684
    },
    {
      "epoch": 0.5617132522785467,
      "grad_norm": 1.4074757099151611,
      "learning_rate": 8.49267572254455e-05,
      "loss": 1.5135,
      "step": 15685
    },
    {
      "epoch": 0.561749064408115,
      "grad_norm": 1.7859296798706055,
      "learning_rate": 8.491529083732025e-05,
      "loss": 1.4796,
      "step": 15686
    },
    {
      "epoch": 0.5617848765376833,
      "grad_norm": 1.585807204246521,
      "learning_rate": 8.490382465214025e-05,
      "loss": 1.5701,
      "step": 15687
    },
    {
      "epoch": 0.5618206886672515,
      "grad_norm": 2.1410164833068848,
      "learning_rate": 8.489235867005985e-05,
      "loss": 1.5223,
      "step": 15688
    },
    {
      "epoch": 0.5618565007968199,
      "grad_norm": 1.5662782192230225,
      "learning_rate": 8.488089289123324e-05,
      "loss": 1.3799,
      "step": 15689
    },
    {
      "epoch": 0.5618923129263882,
      "grad_norm": 1.8063139915466309,
      "learning_rate": 8.486942731581478e-05,
      "loss": 1.5595,
      "step": 15690
    },
    {
      "epoch": 0.5619281250559565,
      "grad_norm": 2.0988364219665527,
      "learning_rate": 8.485796194395862e-05,
      "loss": 1.4958,
      "step": 15691
    },
    {
      "epoch": 0.5619639371855247,
      "grad_norm": 1.6503610610961914,
      "learning_rate": 8.484649677581904e-05,
      "loss": 1.4374,
      "step": 15692
    },
    {
      "epoch": 0.561999749315093,
      "grad_norm": 2.550079584121704,
      "learning_rate": 8.483503181155031e-05,
      "loss": 1.3537,
      "step": 15693
    },
    {
      "epoch": 0.5620355614446613,
      "grad_norm": 1.5846173763275146,
      "learning_rate": 8.482356705130665e-05,
      "loss": 1.4495,
      "step": 15694
    },
    {
      "epoch": 0.5620713735742295,
      "grad_norm": 1.265183687210083,
      "learning_rate": 8.481210249524234e-05,
      "loss": 1.4833,
      "step": 15695
    },
    {
      "epoch": 0.5621071857037979,
      "grad_norm": 1.5761173963546753,
      "learning_rate": 8.480063814351159e-05,
      "loss": 1.3988,
      "step": 15696
    },
    {
      "epoch": 0.5621429978333662,
      "grad_norm": 1.5768393278121948,
      "learning_rate": 8.478917399626865e-05,
      "loss": 1.3578,
      "step": 15697
    },
    {
      "epoch": 0.5621788099629345,
      "grad_norm": 2.0033600330352783,
      "learning_rate": 8.477771005366772e-05,
      "loss": 1.3293,
      "step": 15698
    },
    {
      "epoch": 0.5622146220925027,
      "grad_norm": 2.234950065612793,
      "learning_rate": 8.476624631586313e-05,
      "loss": 1.5952,
      "step": 15699
    },
    {
      "epoch": 0.562250434222071,
      "grad_norm": 1.814936637878418,
      "learning_rate": 8.475478278300902e-05,
      "loss": 1.1814,
      "step": 15700
    },
    {
      "epoch": 0.5622862463516393,
      "grad_norm": 1.5817487239837646,
      "learning_rate": 8.474331945525963e-05,
      "loss": 1.5717,
      "step": 15701
    },
    {
      "epoch": 0.5623220584812075,
      "grad_norm": 1.440826654434204,
      "learning_rate": 8.47318563327692e-05,
      "loss": 1.4432,
      "step": 15702
    },
    {
      "epoch": 0.5623578706107759,
      "grad_norm": 1.247573733329773,
      "learning_rate": 8.472039341569195e-05,
      "loss": 1.5174,
      "step": 15703
    },
    {
      "epoch": 0.5623936827403442,
      "grad_norm": 1.999563217163086,
      "learning_rate": 8.470893070418211e-05,
      "loss": 1.622,
      "step": 15704
    },
    {
      "epoch": 0.5624294948699124,
      "grad_norm": 2.3003087043762207,
      "learning_rate": 8.469746819839387e-05,
      "loss": 1.4743,
      "step": 15705
    },
    {
      "epoch": 0.5624653069994807,
      "grad_norm": 1.6141663789749146,
      "learning_rate": 8.468600589848146e-05,
      "loss": 1.0057,
      "step": 15706
    },
    {
      "epoch": 0.562501119129049,
      "grad_norm": 1.4969419240951538,
      "learning_rate": 8.467454380459907e-05,
      "loss": 1.4611,
      "step": 15707
    },
    {
      "epoch": 0.5625369312586173,
      "grad_norm": 1.8018219470977783,
      "learning_rate": 8.466308191690096e-05,
      "loss": 1.7724,
      "step": 15708
    },
    {
      "epoch": 0.5625727433881855,
      "grad_norm": 1.618523359298706,
      "learning_rate": 8.46516202355413e-05,
      "loss": 1.318,
      "step": 15709
    },
    {
      "epoch": 0.5626085555177539,
      "grad_norm": 1.5267356634140015,
      "learning_rate": 8.464015876067425e-05,
      "loss": 1.5282,
      "step": 15710
    },
    {
      "epoch": 0.5626443676473222,
      "grad_norm": 2.576260566711426,
      "learning_rate": 8.462869749245408e-05,
      "loss": 1.381,
      "step": 15711
    },
    {
      "epoch": 0.5626801797768904,
      "grad_norm": 1.6371855735778809,
      "learning_rate": 8.461723643103494e-05,
      "loss": 1.516,
      "step": 15712
    },
    {
      "epoch": 0.5627159919064587,
      "grad_norm": 2.0176661014556885,
      "learning_rate": 8.460577557657107e-05,
      "loss": 1.5839,
      "step": 15713
    },
    {
      "epoch": 0.562751804036027,
      "grad_norm": 1.6947283744812012,
      "learning_rate": 8.45943149292166e-05,
      "loss": 1.2497,
      "step": 15714
    },
    {
      "epoch": 0.5627876161655953,
      "grad_norm": 1.763598918914795,
      "learning_rate": 8.458285448912578e-05,
      "loss": 1.5134,
      "step": 15715
    },
    {
      "epoch": 0.5628234282951635,
      "grad_norm": 1.442230463027954,
      "learning_rate": 8.457139425645273e-05,
      "loss": 1.3091,
      "step": 15716
    },
    {
      "epoch": 0.5628592404247319,
      "grad_norm": 2.001328706741333,
      "learning_rate": 8.455993423135172e-05,
      "loss": 1.4218,
      "step": 15717
    },
    {
      "epoch": 0.5628950525543002,
      "grad_norm": 1.904796838760376,
      "learning_rate": 8.454847441397684e-05,
      "loss": 1.7912,
      "step": 15718
    },
    {
      "epoch": 0.5629308646838684,
      "grad_norm": 2.0739176273345947,
      "learning_rate": 8.45370148044823e-05,
      "loss": 1.5481,
      "step": 15719
    },
    {
      "epoch": 0.5629666768134367,
      "grad_norm": 2.203639030456543,
      "learning_rate": 8.452555540302231e-05,
      "loss": 1.6342,
      "step": 15720
    },
    {
      "epoch": 0.563002488943005,
      "grad_norm": 1.9044723510742188,
      "learning_rate": 8.451409620975099e-05,
      "loss": 1.491,
      "step": 15721
    },
    {
      "epoch": 0.5630383010725732,
      "grad_norm": 1.4646804332733154,
      "learning_rate": 8.450263722482255e-05,
      "loss": 1.2206,
      "step": 15722
    },
    {
      "epoch": 0.5630741132021415,
      "grad_norm": 2.891911268234253,
      "learning_rate": 8.44911784483911e-05,
      "loss": 1.8076,
      "step": 15723
    },
    {
      "epoch": 0.5631099253317099,
      "grad_norm": 1.5828810930252075,
      "learning_rate": 8.447971988061088e-05,
      "loss": 1.6057,
      "step": 15724
    },
    {
      "epoch": 0.5631457374612782,
      "grad_norm": 1.6931911706924438,
      "learning_rate": 8.446826152163598e-05,
      "loss": 1.6362,
      "step": 15725
    },
    {
      "epoch": 0.5631815495908464,
      "grad_norm": 1.713527798652649,
      "learning_rate": 8.44568033716206e-05,
      "loss": 1.251,
      "step": 15726
    },
    {
      "epoch": 0.5632173617204147,
      "grad_norm": 1.8473023176193237,
      "learning_rate": 8.444534543071891e-05,
      "loss": 1.661,
      "step": 15727
    },
    {
      "epoch": 0.563253173849983,
      "grad_norm": 1.4170857667922974,
      "learning_rate": 8.443388769908498e-05,
      "loss": 1.5997,
      "step": 15728
    },
    {
      "epoch": 0.5632889859795512,
      "grad_norm": 2.1845030784606934,
      "learning_rate": 8.442243017687304e-05,
      "loss": 1.2984,
      "step": 15729
    },
    {
      "epoch": 0.5633247981091195,
      "grad_norm": 2.019676446914673,
      "learning_rate": 8.44109728642372e-05,
      "loss": 1.5803,
      "step": 15730
    },
    {
      "epoch": 0.5633606102386879,
      "grad_norm": 2.300466537475586,
      "learning_rate": 8.439951576133162e-05,
      "loss": 1.8922,
      "step": 15731
    },
    {
      "epoch": 0.5633964223682562,
      "grad_norm": 1.6747887134552002,
      "learning_rate": 8.438805886831042e-05,
      "loss": 1.4242,
      "step": 15732
    },
    {
      "epoch": 0.5634322344978244,
      "grad_norm": 1.5151009559631348,
      "learning_rate": 8.437660218532777e-05,
      "loss": 1.2717,
      "step": 15733
    },
    {
      "epoch": 0.5634680466273927,
      "grad_norm": 1.482073187828064,
      "learning_rate": 8.436514571253775e-05,
      "loss": 1.4709,
      "step": 15734
    },
    {
      "epoch": 0.563503858756961,
      "grad_norm": 1.9614146947860718,
      "learning_rate": 8.435368945009456e-05,
      "loss": 1.7183,
      "step": 15735
    },
    {
      "epoch": 0.5635396708865292,
      "grad_norm": 1.8879051208496094,
      "learning_rate": 8.434223339815229e-05,
      "loss": 1.5138,
      "step": 15736
    },
    {
      "epoch": 0.5635754830160975,
      "grad_norm": 1.917839527130127,
      "learning_rate": 8.433077755686506e-05,
      "loss": 1.5215,
      "step": 15737
    },
    {
      "epoch": 0.5636112951456659,
      "grad_norm": 1.6038079261779785,
      "learning_rate": 8.431932192638703e-05,
      "loss": 1.4635,
      "step": 15738
    },
    {
      "epoch": 0.5636471072752341,
      "grad_norm": 1.5450505018234253,
      "learning_rate": 8.430786650687227e-05,
      "loss": 1.6093,
      "step": 15739
    },
    {
      "epoch": 0.5636829194048024,
      "grad_norm": 1.6022473573684692,
      "learning_rate": 8.429641129847494e-05,
      "loss": 1.2194,
      "step": 15740
    },
    {
      "epoch": 0.5637187315343707,
      "grad_norm": 2.185281991958618,
      "learning_rate": 8.428495630134912e-05,
      "loss": 1.5651,
      "step": 15741
    },
    {
      "epoch": 0.563754543663939,
      "grad_norm": 1.5093828439712524,
      "learning_rate": 8.427350151564897e-05,
      "loss": 1.4803,
      "step": 15742
    },
    {
      "epoch": 0.5637903557935072,
      "grad_norm": 1.753310203552246,
      "learning_rate": 8.426204694152855e-05,
      "loss": 1.3101,
      "step": 15743
    },
    {
      "epoch": 0.5638261679230755,
      "grad_norm": 1.8272547721862793,
      "learning_rate": 8.425059257914201e-05,
      "loss": 1.466,
      "step": 15744
    },
    {
      "epoch": 0.5638619800526439,
      "grad_norm": 1.5649486780166626,
      "learning_rate": 8.423913842864342e-05,
      "loss": 1.433,
      "step": 15745
    },
    {
      "epoch": 0.5638977921822121,
      "grad_norm": 1.796422004699707,
      "learning_rate": 8.422768449018688e-05,
      "loss": 1.2773,
      "step": 15746
    },
    {
      "epoch": 0.5639336043117804,
      "grad_norm": 1.6080162525177002,
      "learning_rate": 8.421623076392652e-05,
      "loss": 1.522,
      "step": 15747
    },
    {
      "epoch": 0.5639694164413487,
      "grad_norm": 1.624840259552002,
      "learning_rate": 8.420477725001639e-05,
      "loss": 1.7583,
      "step": 15748
    },
    {
      "epoch": 0.564005228570917,
      "grad_norm": 1.7605829238891602,
      "learning_rate": 8.419332394861064e-05,
      "loss": 1.283,
      "step": 15749
    },
    {
      "epoch": 0.5640410407004852,
      "grad_norm": 1.488240361213684,
      "learning_rate": 8.418187085986329e-05,
      "loss": 1.3897,
      "step": 15750
    },
    {
      "epoch": 0.5640768528300535,
      "grad_norm": 1.4900915622711182,
      "learning_rate": 8.41704179839285e-05,
      "loss": 1.5596,
      "step": 15751
    },
    {
      "epoch": 0.5641126649596219,
      "grad_norm": 1.8292967081069946,
      "learning_rate": 8.415896532096034e-05,
      "loss": 1.4501,
      "step": 15752
    },
    {
      "epoch": 0.5641484770891901,
      "grad_norm": 1.3246054649353027,
      "learning_rate": 8.41475128711128e-05,
      "loss": 1.0903,
      "step": 15753
    },
    {
      "epoch": 0.5641842892187584,
      "grad_norm": 1.4729738235473633,
      "learning_rate": 8.413606063454008e-05,
      "loss": 1.3507,
      "step": 15754
    },
    {
      "epoch": 0.5642201013483267,
      "grad_norm": 1.7757606506347656,
      "learning_rate": 8.412460861139615e-05,
      "loss": 1.708,
      "step": 15755
    },
    {
      "epoch": 0.5642559134778949,
      "grad_norm": 3.3102619647979736,
      "learning_rate": 8.411315680183517e-05,
      "loss": 1.4701,
      "step": 15756
    },
    {
      "epoch": 0.5642917256074632,
      "grad_norm": 1.4510157108306885,
      "learning_rate": 8.410170520601115e-05,
      "loss": 1.3559,
      "step": 15757
    },
    {
      "epoch": 0.5643275377370315,
      "grad_norm": 1.7538570165634155,
      "learning_rate": 8.40902538240782e-05,
      "loss": 1.2939,
      "step": 15758
    },
    {
      "epoch": 0.5643633498665999,
      "grad_norm": 1.3784433603286743,
      "learning_rate": 8.407880265619035e-05,
      "loss": 1.5924,
      "step": 15759
    },
    {
      "epoch": 0.5643991619961681,
      "grad_norm": 2.172971248626709,
      "learning_rate": 8.406735170250168e-05,
      "loss": 1.3409,
      "step": 15760
    },
    {
      "epoch": 0.5644349741257364,
      "grad_norm": 2.134845733642578,
      "learning_rate": 8.405590096316626e-05,
      "loss": 1.2721,
      "step": 15761
    },
    {
      "epoch": 0.5644707862553047,
      "grad_norm": 1.4905674457550049,
      "learning_rate": 8.404445043833809e-05,
      "loss": 1.4646,
      "step": 15762
    },
    {
      "epoch": 0.5645065983848729,
      "grad_norm": 1.4775800704956055,
      "learning_rate": 8.40330001281713e-05,
      "loss": 1.5261,
      "step": 15763
    },
    {
      "epoch": 0.5645424105144412,
      "grad_norm": 1.478419303894043,
      "learning_rate": 8.402155003281984e-05,
      "loss": 1.569,
      "step": 15764
    },
    {
      "epoch": 0.5645782226440095,
      "grad_norm": 1.6768771409988403,
      "learning_rate": 8.401010015243787e-05,
      "loss": 1.6972,
      "step": 15765
    },
    {
      "epoch": 0.5646140347735779,
      "grad_norm": 1.3239349126815796,
      "learning_rate": 8.399865048717932e-05,
      "loss": 1.4826,
      "step": 15766
    },
    {
      "epoch": 0.5646498469031461,
      "grad_norm": 2.080510377883911,
      "learning_rate": 8.398720103719836e-05,
      "loss": 1.6596,
      "step": 15767
    },
    {
      "epoch": 0.5646856590327144,
      "grad_norm": 1.5443332195281982,
      "learning_rate": 8.397575180264887e-05,
      "loss": 1.2538,
      "step": 15768
    },
    {
      "epoch": 0.5647214711622827,
      "grad_norm": 1.3316525220870972,
      "learning_rate": 8.396430278368503e-05,
      "loss": 1.3754,
      "step": 15769
    },
    {
      "epoch": 0.5647572832918509,
      "grad_norm": 2.1287786960601807,
      "learning_rate": 8.395285398046084e-05,
      "loss": 1.6287,
      "step": 15770
    },
    {
      "epoch": 0.5647930954214192,
      "grad_norm": 1.7575713396072388,
      "learning_rate": 8.394140539313021e-05,
      "loss": 1.4228,
      "step": 15771
    },
    {
      "epoch": 0.5648289075509875,
      "grad_norm": 1.9802348613739014,
      "learning_rate": 8.392995702184734e-05,
      "loss": 1.5655,
      "step": 15772
    },
    {
      "epoch": 0.5648647196805558,
      "grad_norm": 1.6376569271087646,
      "learning_rate": 8.391850886676609e-05,
      "loss": 0.9234,
      "step": 15773
    },
    {
      "epoch": 0.5649005318101241,
      "grad_norm": 1.4419960975646973,
      "learning_rate": 8.390706092804064e-05,
      "loss": 1.4967,
      "step": 15774
    },
    {
      "epoch": 0.5649363439396924,
      "grad_norm": 1.8754823207855225,
      "learning_rate": 8.389561320582486e-05,
      "loss": 1.3997,
      "step": 15775
    },
    {
      "epoch": 0.5649721560692607,
      "grad_norm": 1.3436617851257324,
      "learning_rate": 8.388416570027289e-05,
      "loss": 1.5209,
      "step": 15776
    },
    {
      "epoch": 0.5650079681988289,
      "grad_norm": 1.7056835889816284,
      "learning_rate": 8.38727184115386e-05,
      "loss": 1.1724,
      "step": 15777
    },
    {
      "epoch": 0.5650437803283972,
      "grad_norm": 1.2598458528518677,
      "learning_rate": 8.386127133977617e-05,
      "loss": 1.352,
      "step": 15778
    },
    {
      "epoch": 0.5650795924579655,
      "grad_norm": 1.2932641506195068,
      "learning_rate": 8.384982448513949e-05,
      "loss": 1.3726,
      "step": 15779
    },
    {
      "epoch": 0.5651154045875338,
      "grad_norm": 1.6122196912765503,
      "learning_rate": 8.383837784778257e-05,
      "loss": 1.218,
      "step": 15780
    },
    {
      "epoch": 0.5651512167171021,
      "grad_norm": 1.5172953605651855,
      "learning_rate": 8.382693142785945e-05,
      "loss": 1.5963,
      "step": 15781
    },
    {
      "epoch": 0.5651870288466704,
      "grad_norm": 1.681890606880188,
      "learning_rate": 8.381548522552406e-05,
      "loss": 1.5023,
      "step": 15782
    },
    {
      "epoch": 0.5652228409762386,
      "grad_norm": 2.0911543369293213,
      "learning_rate": 8.38040392409305e-05,
      "loss": 1.6502,
      "step": 15783
    },
    {
      "epoch": 0.5652586531058069,
      "grad_norm": 2.043477773666382,
      "learning_rate": 8.379259347423265e-05,
      "loss": 1.4353,
      "step": 15784
    },
    {
      "epoch": 0.5652944652353752,
      "grad_norm": 1.461133360862732,
      "learning_rate": 8.37811479255846e-05,
      "loss": 1.5957,
      "step": 15785
    },
    {
      "epoch": 0.5653302773649435,
      "grad_norm": 1.580897331237793,
      "learning_rate": 8.376970259514023e-05,
      "loss": 1.1663,
      "step": 15786
    },
    {
      "epoch": 0.5653660894945118,
      "grad_norm": 1.5734902620315552,
      "learning_rate": 8.375825748305364e-05,
      "loss": 1.6092,
      "step": 15787
    },
    {
      "epoch": 0.5654019016240801,
      "grad_norm": 1.7830482721328735,
      "learning_rate": 8.37468125894787e-05,
      "loss": 1.5706,
      "step": 15788
    },
    {
      "epoch": 0.5654377137536484,
      "grad_norm": 2.396804094314575,
      "learning_rate": 8.373536791456944e-05,
      "loss": 1.6365,
      "step": 15789
    },
    {
      "epoch": 0.5654735258832166,
      "grad_norm": 1.630237340927124,
      "learning_rate": 8.372392345847983e-05,
      "loss": 1.693,
      "step": 15790
    },
    {
      "epoch": 0.5655093380127849,
      "grad_norm": 2.5660781860351562,
      "learning_rate": 8.371247922136383e-05,
      "loss": 1.2867,
      "step": 15791
    },
    {
      "epoch": 0.5655451501423532,
      "grad_norm": 1.5164788961410522,
      "learning_rate": 8.370103520337542e-05,
      "loss": 1.3451,
      "step": 15792
    },
    {
      "epoch": 0.5655809622719215,
      "grad_norm": 1.45986807346344,
      "learning_rate": 8.368959140466853e-05,
      "loss": 1.4041,
      "step": 15793
    },
    {
      "epoch": 0.5656167744014898,
      "grad_norm": 1.8049870729446411,
      "learning_rate": 8.367814782539718e-05,
      "loss": 1.7531,
      "step": 15794
    },
    {
      "epoch": 0.5656525865310581,
      "grad_norm": 2.0593514442443848,
      "learning_rate": 8.366670446571525e-05,
      "loss": 1.1339,
      "step": 15795
    },
    {
      "epoch": 0.5656883986606264,
      "grad_norm": 1.956852912902832,
      "learning_rate": 8.365526132577681e-05,
      "loss": 1.4611,
      "step": 15796
    },
    {
      "epoch": 0.5657242107901946,
      "grad_norm": 1.8688738346099854,
      "learning_rate": 8.364381840573573e-05,
      "loss": 1.3412,
      "step": 15797
    },
    {
      "epoch": 0.5657600229197629,
      "grad_norm": 1.790224313735962,
      "learning_rate": 8.363237570574595e-05,
      "loss": 1.1733,
      "step": 15798
    },
    {
      "epoch": 0.5657958350493312,
      "grad_norm": 1.8264939785003662,
      "learning_rate": 8.362093322596145e-05,
      "loss": 1.4683,
      "step": 15799
    },
    {
      "epoch": 0.5658316471788994,
      "grad_norm": 1.8965346813201904,
      "learning_rate": 8.360949096653616e-05,
      "loss": 1.797,
      "step": 15800
    },
    {
      "epoch": 0.5658674593084678,
      "grad_norm": 2.144782066345215,
      "learning_rate": 8.359804892762405e-05,
      "loss": 1.7707,
      "step": 15801
    },
    {
      "epoch": 0.5659032714380361,
      "grad_norm": 3.100724697113037,
      "learning_rate": 8.3586607109379e-05,
      "loss": 1.3111,
      "step": 15802
    },
    {
      "epoch": 0.5659390835676044,
      "grad_norm": 2.0397491455078125,
      "learning_rate": 8.357516551195501e-05,
      "loss": 1.4468,
      "step": 15803
    },
    {
      "epoch": 0.5659748956971726,
      "grad_norm": 1.3980594873428345,
      "learning_rate": 8.356372413550597e-05,
      "loss": 1.3259,
      "step": 15804
    },
    {
      "epoch": 0.5660107078267409,
      "grad_norm": 1.8807727098464966,
      "learning_rate": 8.355228298018582e-05,
      "loss": 1.7511,
      "step": 15805
    },
    {
      "epoch": 0.5660465199563092,
      "grad_norm": 2.131071090698242,
      "learning_rate": 8.354084204614851e-05,
      "loss": 1.2344,
      "step": 15806
    },
    {
      "epoch": 0.5660823320858774,
      "grad_norm": 1.6939911842346191,
      "learning_rate": 8.35294013335479e-05,
      "loss": 1.4542,
      "step": 15807
    },
    {
      "epoch": 0.5661181442154458,
      "grad_norm": 1.5454076528549194,
      "learning_rate": 8.351796084253797e-05,
      "loss": 1.4836,
      "step": 15808
    },
    {
      "epoch": 0.5661539563450141,
      "grad_norm": 1.7581214904785156,
      "learning_rate": 8.350652057327261e-05,
      "loss": 1.3919,
      "step": 15809
    },
    {
      "epoch": 0.5661897684745824,
      "grad_norm": 1.616428017616272,
      "learning_rate": 8.349508052590574e-05,
      "loss": 1.4052,
      "step": 15810
    },
    {
      "epoch": 0.5662255806041506,
      "grad_norm": 1.2832040786743164,
      "learning_rate": 8.348364070059127e-05,
      "loss": 1.1419,
      "step": 15811
    },
    {
      "epoch": 0.5662613927337189,
      "grad_norm": 1.918565034866333,
      "learning_rate": 8.347220109748312e-05,
      "loss": 1.5448,
      "step": 15812
    },
    {
      "epoch": 0.5662972048632872,
      "grad_norm": 2.462134838104248,
      "learning_rate": 8.346076171673518e-05,
      "loss": 1.3867,
      "step": 15813
    },
    {
      "epoch": 0.5663330169928554,
      "grad_norm": 1.435020923614502,
      "learning_rate": 8.344932255850136e-05,
      "loss": 1.2917,
      "step": 15814
    },
    {
      "epoch": 0.5663688291224238,
      "grad_norm": 1.4067466259002686,
      "learning_rate": 8.343788362293556e-05,
      "loss": 1.4601,
      "step": 15815
    },
    {
      "epoch": 0.5664046412519921,
      "grad_norm": 1.6416200399398804,
      "learning_rate": 8.342644491019165e-05,
      "loss": 1.3699,
      "step": 15816
    },
    {
      "epoch": 0.5664404533815603,
      "grad_norm": 1.66130793094635,
      "learning_rate": 8.341500642042359e-05,
      "loss": 1.3382,
      "step": 15817
    },
    {
      "epoch": 0.5664762655111286,
      "grad_norm": 1.7099733352661133,
      "learning_rate": 8.340356815378517e-05,
      "loss": 1.6387,
      "step": 15818
    },
    {
      "epoch": 0.5665120776406969,
      "grad_norm": 1.7127584218978882,
      "learning_rate": 8.339213011043038e-05,
      "loss": 1.3284,
      "step": 15819
    },
    {
      "epoch": 0.5665478897702652,
      "grad_norm": 1.6124918460845947,
      "learning_rate": 8.338069229051302e-05,
      "loss": 1.4032,
      "step": 15820
    },
    {
      "epoch": 0.5665837018998334,
      "grad_norm": 2.4808313846588135,
      "learning_rate": 8.336925469418704e-05,
      "loss": 1.5178,
      "step": 15821
    },
    {
      "epoch": 0.5666195140294018,
      "grad_norm": 1.6432664394378662,
      "learning_rate": 8.335781732160625e-05,
      "loss": 1.5208,
      "step": 15822
    },
    {
      "epoch": 0.5666553261589701,
      "grad_norm": 1.5675996541976929,
      "learning_rate": 8.334638017292459e-05,
      "loss": 1.5968,
      "step": 15823
    },
    {
      "epoch": 0.5666911382885383,
      "grad_norm": 2.1726925373077393,
      "learning_rate": 8.33349432482959e-05,
      "loss": 1.3455,
      "step": 15824
    },
    {
      "epoch": 0.5667269504181066,
      "grad_norm": 1.6445999145507812,
      "learning_rate": 8.332350654787404e-05,
      "loss": 1.6555,
      "step": 15825
    },
    {
      "epoch": 0.5667627625476749,
      "grad_norm": 1.3497015237808228,
      "learning_rate": 8.33120700718129e-05,
      "loss": 1.2937,
      "step": 15826
    },
    {
      "epoch": 0.5667985746772431,
      "grad_norm": 1.9332237243652344,
      "learning_rate": 8.330063382026631e-05,
      "loss": 1.7678,
      "step": 15827
    },
    {
      "epoch": 0.5668343868068114,
      "grad_norm": 2.0545902252197266,
      "learning_rate": 8.328919779338819e-05,
      "loss": 1.0707,
      "step": 15828
    },
    {
      "epoch": 0.5668701989363798,
      "grad_norm": 1.9223384857177734,
      "learning_rate": 8.327776199133232e-05,
      "loss": 1.0993,
      "step": 15829
    },
    {
      "epoch": 0.5669060110659481,
      "grad_norm": 1.645604133605957,
      "learning_rate": 8.326632641425261e-05,
      "loss": 1.661,
      "step": 15830
    },
    {
      "epoch": 0.5669418231955163,
      "grad_norm": 1.6659955978393555,
      "learning_rate": 8.325489106230288e-05,
      "loss": 1.2141,
      "step": 15831
    },
    {
      "epoch": 0.5669776353250846,
      "grad_norm": 1.6164517402648926,
      "learning_rate": 8.324345593563701e-05,
      "loss": 1.312,
      "step": 15832
    },
    {
      "epoch": 0.5670134474546529,
      "grad_norm": 2.98215913772583,
      "learning_rate": 8.323202103440884e-05,
      "loss": 1.5099,
      "step": 15833
    },
    {
      "epoch": 0.5670492595842211,
      "grad_norm": 1.6239739656448364,
      "learning_rate": 8.322058635877216e-05,
      "loss": 1.3561,
      "step": 15834
    },
    {
      "epoch": 0.5670850717137894,
      "grad_norm": 2.104954957962036,
      "learning_rate": 8.320915190888087e-05,
      "loss": 1.7427,
      "step": 15835
    },
    {
      "epoch": 0.5671208838433578,
      "grad_norm": 1.366011619567871,
      "learning_rate": 8.319771768488877e-05,
      "loss": 1.4357,
      "step": 15836
    },
    {
      "epoch": 0.5671566959729261,
      "grad_norm": 1.9174448251724243,
      "learning_rate": 8.318628368694972e-05,
      "loss": 1.3326,
      "step": 15837
    },
    {
      "epoch": 0.5671925081024943,
      "grad_norm": 1.4196544885635376,
      "learning_rate": 8.317484991521751e-05,
      "loss": 1.4263,
      "step": 15838
    },
    {
      "epoch": 0.5672283202320626,
      "grad_norm": 1.5104621648788452,
      "learning_rate": 8.316341636984602e-05,
      "loss": 1.535,
      "step": 15839
    },
    {
      "epoch": 0.5672641323616309,
      "grad_norm": 1.5836697816848755,
      "learning_rate": 8.315198305098902e-05,
      "loss": 1.2274,
      "step": 15840
    },
    {
      "epoch": 0.5672999444911991,
      "grad_norm": 2.0471384525299072,
      "learning_rate": 8.314054995880036e-05,
      "loss": 1.5095,
      "step": 15841
    },
    {
      "epoch": 0.5673357566207674,
      "grad_norm": 1.560638427734375,
      "learning_rate": 8.312911709343388e-05,
      "loss": 1.417,
      "step": 15842
    },
    {
      "epoch": 0.5673715687503358,
      "grad_norm": 1.8646968603134155,
      "learning_rate": 8.311768445504333e-05,
      "loss": 1.5827,
      "step": 15843
    },
    {
      "epoch": 0.567407380879904,
      "grad_norm": 1.4655557870864868,
      "learning_rate": 8.31062520437826e-05,
      "loss": 1.2498,
      "step": 15844
    },
    {
      "epoch": 0.5674431930094723,
      "grad_norm": 2.248730421066284,
      "learning_rate": 8.309481985980541e-05,
      "loss": 1.4068,
      "step": 15845
    },
    {
      "epoch": 0.5674790051390406,
      "grad_norm": 2.3204147815704346,
      "learning_rate": 8.308338790326565e-05,
      "loss": 1.7721,
      "step": 15846
    },
    {
      "epoch": 0.5675148172686089,
      "grad_norm": 1.9696449041366577,
      "learning_rate": 8.307195617431707e-05,
      "loss": 1.5491,
      "step": 15847
    },
    {
      "epoch": 0.5675506293981771,
      "grad_norm": 1.883335828781128,
      "learning_rate": 8.306052467311349e-05,
      "loss": 1.4826,
      "step": 15848
    },
    {
      "epoch": 0.5675864415277454,
      "grad_norm": 2.11018705368042,
      "learning_rate": 8.304909339980873e-05,
      "loss": 1.7215,
      "step": 15849
    },
    {
      "epoch": 0.5676222536573138,
      "grad_norm": 1.548668384552002,
      "learning_rate": 8.303766235455648e-05,
      "loss": 1.3243,
      "step": 15850
    },
    {
      "epoch": 0.567658065786882,
      "grad_norm": 1.4991328716278076,
      "learning_rate": 8.302623153751068e-05,
      "loss": 1.4878,
      "step": 15851
    },
    {
      "epoch": 0.5676938779164503,
      "grad_norm": 1.4898539781570435,
      "learning_rate": 8.301480094882497e-05,
      "loss": 1.2673,
      "step": 15852
    },
    {
      "epoch": 0.5677296900460186,
      "grad_norm": 1.7317248582839966,
      "learning_rate": 8.300337058865323e-05,
      "loss": 1.4307,
      "step": 15853
    },
    {
      "epoch": 0.5677655021755869,
      "grad_norm": 1.6654188632965088,
      "learning_rate": 8.299194045714921e-05,
      "loss": 1.5343,
      "step": 15854
    },
    {
      "epoch": 0.5678013143051551,
      "grad_norm": 1.7130259275436401,
      "learning_rate": 8.298051055446673e-05,
      "loss": 1.544,
      "step": 15855
    },
    {
      "epoch": 0.5678371264347234,
      "grad_norm": 3.522223472595215,
      "learning_rate": 8.296908088075949e-05,
      "loss": 1.9023,
      "step": 15856
    },
    {
      "epoch": 0.5678729385642918,
      "grad_norm": 2.1206347942352295,
      "learning_rate": 8.295765143618131e-05,
      "loss": 1.6537,
      "step": 15857
    },
    {
      "epoch": 0.56790875069386,
      "grad_norm": 2.9285621643066406,
      "learning_rate": 8.294622222088598e-05,
      "loss": 1.6514,
      "step": 15858
    },
    {
      "epoch": 0.5679445628234283,
      "grad_norm": 2.09845232963562,
      "learning_rate": 8.293479323502716e-05,
      "loss": 1.3896,
      "step": 15859
    },
    {
      "epoch": 0.5679803749529966,
      "grad_norm": 1.8847618103027344,
      "learning_rate": 8.292336447875876e-05,
      "loss": 1.6824,
      "step": 15860
    },
    {
      "epoch": 0.5680161870825648,
      "grad_norm": 3.045405149459839,
      "learning_rate": 8.291193595223438e-05,
      "loss": 1.5033,
      "step": 15861
    },
    {
      "epoch": 0.5680519992121331,
      "grad_norm": 1.4553380012512207,
      "learning_rate": 8.290050765560795e-05,
      "loss": 1.4361,
      "step": 15862
    },
    {
      "epoch": 0.5680878113417014,
      "grad_norm": 1.3813594579696655,
      "learning_rate": 8.288907958903305e-05,
      "loss": 1.7692,
      "step": 15863
    },
    {
      "epoch": 0.5681236234712698,
      "grad_norm": 1.9366446733474731,
      "learning_rate": 8.287765175266358e-05,
      "loss": 1.3047,
      "step": 15864
    },
    {
      "epoch": 0.568159435600838,
      "grad_norm": 2.792715549468994,
      "learning_rate": 8.286622414665317e-05,
      "loss": 2.0675,
      "step": 15865
    },
    {
      "epoch": 0.5681952477304063,
      "grad_norm": 1.838062047958374,
      "learning_rate": 8.285479677115563e-05,
      "loss": 1.4573,
      "step": 15866
    },
    {
      "epoch": 0.5682310598599746,
      "grad_norm": 1.3714712858200073,
      "learning_rate": 8.284336962632473e-05,
      "loss": 1.4702,
      "step": 15867
    },
    {
      "epoch": 0.5682668719895428,
      "grad_norm": 1.2245217561721802,
      "learning_rate": 8.283194271231408e-05,
      "loss": 1.5559,
      "step": 15868
    },
    {
      "epoch": 0.5683026841191111,
      "grad_norm": 1.866948127746582,
      "learning_rate": 8.282051602927757e-05,
      "loss": 1.5925,
      "step": 15869
    },
    {
      "epoch": 0.5683384962486794,
      "grad_norm": 1.918073058128357,
      "learning_rate": 8.28090895773688e-05,
      "loss": 1.2218,
      "step": 15870
    },
    {
      "epoch": 0.5683743083782478,
      "grad_norm": 2.7581140995025635,
      "learning_rate": 8.27976633567416e-05,
      "loss": 1.1811,
      "step": 15871
    },
    {
      "epoch": 0.568410120507816,
      "grad_norm": 1.6794708967208862,
      "learning_rate": 8.27862373675496e-05,
      "loss": 1.3733,
      "step": 15872
    },
    {
      "epoch": 0.5684459326373843,
      "grad_norm": 2.179267644882202,
      "learning_rate": 8.277481160994663e-05,
      "loss": 1.3167,
      "step": 15873
    },
    {
      "epoch": 0.5684817447669526,
      "grad_norm": 1.5156314373016357,
      "learning_rate": 8.276338608408627e-05,
      "loss": 1.2546,
      "step": 15874
    },
    {
      "epoch": 0.5685175568965208,
      "grad_norm": 1.3936090469360352,
      "learning_rate": 8.27519607901224e-05,
      "loss": 1.2427,
      "step": 15875
    },
    {
      "epoch": 0.5685533690260891,
      "grad_norm": 2.1718878746032715,
      "learning_rate": 8.274053572820862e-05,
      "loss": 1.238,
      "step": 15876
    },
    {
      "epoch": 0.5685891811556574,
      "grad_norm": 1.6128958463668823,
      "learning_rate": 8.272911089849866e-05,
      "loss": 1.6238,
      "step": 15877
    },
    {
      "epoch": 0.5686249932852258,
      "grad_norm": 1.745404601097107,
      "learning_rate": 8.271768630114624e-05,
      "loss": 1.3826,
      "step": 15878
    },
    {
      "epoch": 0.568660805414794,
      "grad_norm": 1.595452070236206,
      "learning_rate": 8.270626193630503e-05,
      "loss": 1.4595,
      "step": 15879
    },
    {
      "epoch": 0.5686966175443623,
      "grad_norm": 1.5459754467010498,
      "learning_rate": 8.269483780412883e-05,
      "loss": 1.2893,
      "step": 15880
    },
    {
      "epoch": 0.5687324296739306,
      "grad_norm": 1.7879676818847656,
      "learning_rate": 8.268341390477118e-05,
      "loss": 1.8261,
      "step": 15881
    },
    {
      "epoch": 0.5687682418034988,
      "grad_norm": 1.4515905380249023,
      "learning_rate": 8.267199023838593e-05,
      "loss": 1.7552,
      "step": 15882
    },
    {
      "epoch": 0.5688040539330671,
      "grad_norm": 1.9007455110549927,
      "learning_rate": 8.266056680512664e-05,
      "loss": 1.4626,
      "step": 15883
    },
    {
      "epoch": 0.5688398660626354,
      "grad_norm": 1.7271119356155396,
      "learning_rate": 8.26491436051471e-05,
      "loss": 1.5642,
      "step": 15884
    },
    {
      "epoch": 0.5688756781922037,
      "grad_norm": 1.5597244501113892,
      "learning_rate": 8.263772063860096e-05,
      "loss": 1.3979,
      "step": 15885
    },
    {
      "epoch": 0.568911490321772,
      "grad_norm": 1.8739885091781616,
      "learning_rate": 8.262629790564186e-05,
      "loss": 1.6296,
      "step": 15886
    },
    {
      "epoch": 0.5689473024513403,
      "grad_norm": 1.602601408958435,
      "learning_rate": 8.261487540642353e-05,
      "loss": 1.619,
      "step": 15887
    },
    {
      "epoch": 0.5689831145809086,
      "grad_norm": 1.4145638942718506,
      "learning_rate": 8.26034531410996e-05,
      "loss": 1.5039,
      "step": 15888
    },
    {
      "epoch": 0.5690189267104768,
      "grad_norm": 1.6475313901901245,
      "learning_rate": 8.259203110982381e-05,
      "loss": 1.2161,
      "step": 15889
    },
    {
      "epoch": 0.5690547388400451,
      "grad_norm": 1.7352582216262817,
      "learning_rate": 8.258060931274976e-05,
      "loss": 1.41,
      "step": 15890
    },
    {
      "epoch": 0.5690905509696134,
      "grad_norm": 1.9929988384246826,
      "learning_rate": 8.256918775003115e-05,
      "loss": 1.3584,
      "step": 15891
    },
    {
      "epoch": 0.5691263630991817,
      "grad_norm": 1.258603811264038,
      "learning_rate": 8.255776642182159e-05,
      "loss": 1.276,
      "step": 15892
    },
    {
      "epoch": 0.56916217522875,
      "grad_norm": 1.3469874858856201,
      "learning_rate": 8.254634532827487e-05,
      "loss": 1.661,
      "step": 15893
    },
    {
      "epoch": 0.5691979873583183,
      "grad_norm": 1.4749658107757568,
      "learning_rate": 8.253492446954452e-05,
      "loss": 1.7263,
      "step": 15894
    },
    {
      "epoch": 0.5692337994878865,
      "grad_norm": 1.9066383838653564,
      "learning_rate": 8.252350384578421e-05,
      "loss": 1.4895,
      "step": 15895
    },
    {
      "epoch": 0.5692696116174548,
      "grad_norm": 2.15698504447937,
      "learning_rate": 8.251208345714764e-05,
      "loss": 1.444,
      "step": 15896
    },
    {
      "epoch": 0.5693054237470231,
      "grad_norm": 1.5320926904678345,
      "learning_rate": 8.25006633037884e-05,
      "loss": 1.5376,
      "step": 15897
    },
    {
      "epoch": 0.5693412358765914,
      "grad_norm": 1.3792502880096436,
      "learning_rate": 8.24892433858602e-05,
      "loss": 1.6099,
      "step": 15898
    },
    {
      "epoch": 0.5693770480061597,
      "grad_norm": 1.6922613382339478,
      "learning_rate": 8.247782370351663e-05,
      "loss": 1.3959,
      "step": 15899
    },
    {
      "epoch": 0.569412860135728,
      "grad_norm": 2.027216911315918,
      "learning_rate": 8.246640425691133e-05,
      "loss": 1.516,
      "step": 15900
    },
    {
      "epoch": 0.5694486722652963,
      "grad_norm": 1.6750237941741943,
      "learning_rate": 8.245498504619794e-05,
      "loss": 1.1921,
      "step": 15901
    },
    {
      "epoch": 0.5694844843948645,
      "grad_norm": 1.6753225326538086,
      "learning_rate": 8.244356607153011e-05,
      "loss": 1.5097,
      "step": 15902
    },
    {
      "epoch": 0.5695202965244328,
      "grad_norm": 1.3187848329544067,
      "learning_rate": 8.243214733306145e-05,
      "loss": 1.4227,
      "step": 15903
    },
    {
      "epoch": 0.5695561086540011,
      "grad_norm": 1.5426853895187378,
      "learning_rate": 8.242072883094559e-05,
      "loss": 1.5042,
      "step": 15904
    },
    {
      "epoch": 0.5695919207835693,
      "grad_norm": 1.53534996509552,
      "learning_rate": 8.240931056533615e-05,
      "loss": 1.2341,
      "step": 15905
    },
    {
      "epoch": 0.5696277329131377,
      "grad_norm": 1.2395790815353394,
      "learning_rate": 8.239789253638672e-05,
      "loss": 1.2779,
      "step": 15906
    },
    {
      "epoch": 0.569663545042706,
      "grad_norm": 1.5523747205734253,
      "learning_rate": 8.238647474425097e-05,
      "loss": 1.6073,
      "step": 15907
    },
    {
      "epoch": 0.5696993571722743,
      "grad_norm": 1.5684142112731934,
      "learning_rate": 8.237505718908246e-05,
      "loss": 1.5768,
      "step": 15908
    },
    {
      "epoch": 0.5697351693018425,
      "grad_norm": 1.7088567018508911,
      "learning_rate": 8.236363987103483e-05,
      "loss": 1.4598,
      "step": 15909
    },
    {
      "epoch": 0.5697709814314108,
      "grad_norm": 1.5773688554763794,
      "learning_rate": 8.235222279026168e-05,
      "loss": 1.2386,
      "step": 15910
    },
    {
      "epoch": 0.5698067935609791,
      "grad_norm": 1.7507705688476562,
      "learning_rate": 8.234080594691663e-05,
      "loss": 1.5207,
      "step": 15911
    },
    {
      "epoch": 0.5698426056905473,
      "grad_norm": 1.7203186750411987,
      "learning_rate": 8.232938934115323e-05,
      "loss": 1.472,
      "step": 15912
    },
    {
      "epoch": 0.5698784178201157,
      "grad_norm": 1.6175166368484497,
      "learning_rate": 8.231797297312509e-05,
      "loss": 1.6451,
      "step": 15913
    },
    {
      "epoch": 0.569914229949684,
      "grad_norm": 1.395760416984558,
      "learning_rate": 8.230655684298585e-05,
      "loss": 1.4501,
      "step": 15914
    },
    {
      "epoch": 0.5699500420792523,
      "grad_norm": 1.8864084482192993,
      "learning_rate": 8.229514095088903e-05,
      "loss": 1.4636,
      "step": 15915
    },
    {
      "epoch": 0.5699858542088205,
      "grad_norm": 1.705157995223999,
      "learning_rate": 8.228372529698828e-05,
      "loss": 1.5955,
      "step": 15916
    },
    {
      "epoch": 0.5700216663383888,
      "grad_norm": 2.0784878730773926,
      "learning_rate": 8.227230988143712e-05,
      "loss": 1.3895,
      "step": 15917
    },
    {
      "epoch": 0.5700574784679571,
      "grad_norm": 1.4656957387924194,
      "learning_rate": 8.22608947043892e-05,
      "loss": 1.4649,
      "step": 15918
    },
    {
      "epoch": 0.5700932905975253,
      "grad_norm": 1.8664727210998535,
      "learning_rate": 8.224947976599804e-05,
      "loss": 1.6218,
      "step": 15919
    },
    {
      "epoch": 0.5701291027270937,
      "grad_norm": 1.4809681177139282,
      "learning_rate": 8.223806506641724e-05,
      "loss": 1.4457,
      "step": 15920
    },
    {
      "epoch": 0.570164914856662,
      "grad_norm": 1.9111181497573853,
      "learning_rate": 8.222665060580038e-05,
      "loss": 1.413,
      "step": 15921
    },
    {
      "epoch": 0.5702007269862303,
      "grad_norm": 1.56305992603302,
      "learning_rate": 8.221523638430098e-05,
      "loss": 1.2293,
      "step": 15922
    },
    {
      "epoch": 0.5702365391157985,
      "grad_norm": 1.4683891534805298,
      "learning_rate": 8.220382240207266e-05,
      "loss": 1.5771,
      "step": 15923
    },
    {
      "epoch": 0.5702723512453668,
      "grad_norm": 1.7615280151367188,
      "learning_rate": 8.219240865926892e-05,
      "loss": 1.4823,
      "step": 15924
    },
    {
      "epoch": 0.5703081633749351,
      "grad_norm": 2.0023019313812256,
      "learning_rate": 8.218099515604339e-05,
      "loss": 1.4878,
      "step": 15925
    },
    {
      "epoch": 0.5703439755045033,
      "grad_norm": 1.7098430395126343,
      "learning_rate": 8.216958189254956e-05,
      "loss": 1.6951,
      "step": 15926
    },
    {
      "epoch": 0.5703797876340717,
      "grad_norm": 1.6771968603134155,
      "learning_rate": 8.215816886894102e-05,
      "loss": 1.3032,
      "step": 15927
    },
    {
      "epoch": 0.57041559976364,
      "grad_norm": 1.426419973373413,
      "learning_rate": 8.214675608537128e-05,
      "loss": 1.1594,
      "step": 15928
    },
    {
      "epoch": 0.5704514118932082,
      "grad_norm": 2.378345012664795,
      "learning_rate": 8.213534354199392e-05,
      "loss": 1.9314,
      "step": 15929
    },
    {
      "epoch": 0.5704872240227765,
      "grad_norm": 1.3041222095489502,
      "learning_rate": 8.212393123896249e-05,
      "loss": 1.5663,
      "step": 15930
    },
    {
      "epoch": 0.5705230361523448,
      "grad_norm": 1.2427996397018433,
      "learning_rate": 8.211251917643047e-05,
      "loss": 1.4987,
      "step": 15931
    },
    {
      "epoch": 0.5705588482819131,
      "grad_norm": 1.7306779623031616,
      "learning_rate": 8.210110735455147e-05,
      "loss": 1.4224,
      "step": 15932
    },
    {
      "epoch": 0.5705946604114813,
      "grad_norm": 1.6303404569625854,
      "learning_rate": 8.208969577347894e-05,
      "loss": 1.443,
      "step": 15933
    },
    {
      "epoch": 0.5706304725410497,
      "grad_norm": 1.9509263038635254,
      "learning_rate": 8.207828443336649e-05,
      "loss": 1.3303,
      "step": 15934
    },
    {
      "epoch": 0.570666284670618,
      "grad_norm": 2.3029608726501465,
      "learning_rate": 8.206687333436758e-05,
      "loss": 1.4231,
      "step": 15935
    },
    {
      "epoch": 0.5707020968001862,
      "grad_norm": 1.5229395627975464,
      "learning_rate": 8.205546247663578e-05,
      "loss": 1.52,
      "step": 15936
    },
    {
      "epoch": 0.5707379089297545,
      "grad_norm": 2.0228278636932373,
      "learning_rate": 8.204405186032455e-05,
      "loss": 1.7603,
      "step": 15937
    },
    {
      "epoch": 0.5707737210593228,
      "grad_norm": 1.6591273546218872,
      "learning_rate": 8.203264148558749e-05,
      "loss": 1.3386,
      "step": 15938
    },
    {
      "epoch": 0.570809533188891,
      "grad_norm": 1.5935134887695312,
      "learning_rate": 8.202123135257804e-05,
      "loss": 1.514,
      "step": 15939
    },
    {
      "epoch": 0.5708453453184593,
      "grad_norm": 1.514012098312378,
      "learning_rate": 8.20098214614497e-05,
      "loss": 1.6337,
      "step": 15940
    },
    {
      "epoch": 0.5708811574480277,
      "grad_norm": 1.718827724456787,
      "learning_rate": 8.199841181235606e-05,
      "loss": 1.4344,
      "step": 15941
    },
    {
      "epoch": 0.570916969577596,
      "grad_norm": 2.1700141429901123,
      "learning_rate": 8.198700240545053e-05,
      "loss": 1.3989,
      "step": 15942
    },
    {
      "epoch": 0.5709527817071642,
      "grad_norm": 1.856092929840088,
      "learning_rate": 8.197559324088666e-05,
      "loss": 1.6977,
      "step": 15943
    },
    {
      "epoch": 0.5709885938367325,
      "grad_norm": 1.462905764579773,
      "learning_rate": 8.196418431881793e-05,
      "loss": 1.6105,
      "step": 15944
    },
    {
      "epoch": 0.5710244059663008,
      "grad_norm": 2.0627965927124023,
      "learning_rate": 8.195277563939785e-05,
      "loss": 1.6915,
      "step": 15945
    },
    {
      "epoch": 0.571060218095869,
      "grad_norm": 1.770424246788025,
      "learning_rate": 8.194136720277992e-05,
      "loss": 1.4564,
      "step": 15946
    },
    {
      "epoch": 0.5710960302254373,
      "grad_norm": 3.049556255340576,
      "learning_rate": 8.192995900911751e-05,
      "loss": 1.3256,
      "step": 15947
    },
    {
      "epoch": 0.5711318423550057,
      "grad_norm": 1.7785724401474,
      "learning_rate": 8.191855105856428e-05,
      "loss": 1.4055,
      "step": 15948
    },
    {
      "epoch": 0.571167654484574,
      "grad_norm": 1.4011263847351074,
      "learning_rate": 8.190714335127356e-05,
      "loss": 1.401,
      "step": 15949
    },
    {
      "epoch": 0.5712034666141422,
      "grad_norm": 1.9322203397750854,
      "learning_rate": 8.189573588739892e-05,
      "loss": 1.5981,
      "step": 15950
    },
    {
      "epoch": 0.5712392787437105,
      "grad_norm": 2.2392494678497314,
      "learning_rate": 8.188432866709379e-05,
      "loss": 1.5755,
      "step": 15951
    },
    {
      "epoch": 0.5712750908732788,
      "grad_norm": 1.8817596435546875,
      "learning_rate": 8.187292169051168e-05,
      "loss": 1.4319,
      "step": 15952
    },
    {
      "epoch": 0.571310903002847,
      "grad_norm": 1.4532921314239502,
      "learning_rate": 8.186151495780598e-05,
      "loss": 1.2006,
      "step": 15953
    },
    {
      "epoch": 0.5713467151324153,
      "grad_norm": 2.218801975250244,
      "learning_rate": 8.185010846913024e-05,
      "loss": 1.4041,
      "step": 15954
    },
    {
      "epoch": 0.5713825272619837,
      "grad_norm": 1.6853936910629272,
      "learning_rate": 8.183870222463789e-05,
      "loss": 1.5148,
      "step": 15955
    },
    {
      "epoch": 0.571418339391552,
      "grad_norm": 1.7166179418563843,
      "learning_rate": 8.182729622448231e-05,
      "loss": 1.6018,
      "step": 15956
    },
    {
      "epoch": 0.5714541515211202,
      "grad_norm": 1.7702536582946777,
      "learning_rate": 8.181589046881709e-05,
      "loss": 1.4854,
      "step": 15957
    },
    {
      "epoch": 0.5714899636506885,
      "grad_norm": 1.2718796730041504,
      "learning_rate": 8.180448495779554e-05,
      "loss": 1.5774,
      "step": 15958
    },
    {
      "epoch": 0.5715257757802568,
      "grad_norm": 1.6079905033111572,
      "learning_rate": 8.179307969157123e-05,
      "loss": 1.2911,
      "step": 15959
    },
    {
      "epoch": 0.571561587909825,
      "grad_norm": 1.5168037414550781,
      "learning_rate": 8.17816746702975e-05,
      "loss": 1.6445,
      "step": 15960
    },
    {
      "epoch": 0.5715974000393933,
      "grad_norm": 1.877971887588501,
      "learning_rate": 8.177026989412789e-05,
      "loss": 1.2533,
      "step": 15961
    },
    {
      "epoch": 0.5716332121689617,
      "grad_norm": 1.4898743629455566,
      "learning_rate": 8.175886536321574e-05,
      "loss": 1.2705,
      "step": 15962
    },
    {
      "epoch": 0.57166902429853,
      "grad_norm": 1.495103359222412,
      "learning_rate": 8.174746107771454e-05,
      "loss": 1.6275,
      "step": 15963
    },
    {
      "epoch": 0.5717048364280982,
      "grad_norm": 1.7005457878112793,
      "learning_rate": 8.173605703777774e-05,
      "loss": 1.5007,
      "step": 15964
    },
    {
      "epoch": 0.5717406485576665,
      "grad_norm": 1.619920253753662,
      "learning_rate": 8.172465324355868e-05,
      "loss": 1.3175,
      "step": 15965
    },
    {
      "epoch": 0.5717764606872348,
      "grad_norm": 1.668201208114624,
      "learning_rate": 8.171324969521089e-05,
      "loss": 1.1773,
      "step": 15966
    },
    {
      "epoch": 0.571812272816803,
      "grad_norm": 1.5060385465621948,
      "learning_rate": 8.170184639288767e-05,
      "loss": 1.5491,
      "step": 15967
    },
    {
      "epoch": 0.5718480849463713,
      "grad_norm": 1.9399768114089966,
      "learning_rate": 8.169044333674259e-05,
      "loss": 1.7142,
      "step": 15968
    },
    {
      "epoch": 0.5718838970759397,
      "grad_norm": 1.6110990047454834,
      "learning_rate": 8.167904052692889e-05,
      "loss": 1.3316,
      "step": 15969
    },
    {
      "epoch": 0.5719197092055079,
      "grad_norm": 1.6544129848480225,
      "learning_rate": 8.166763796360014e-05,
      "loss": 1.2218,
      "step": 15970
    },
    {
      "epoch": 0.5719555213350762,
      "grad_norm": 1.499854564666748,
      "learning_rate": 8.165623564690961e-05,
      "loss": 1.5932,
      "step": 15971
    },
    {
      "epoch": 0.5719913334646445,
      "grad_norm": 1.706298589706421,
      "learning_rate": 8.164483357701082e-05,
      "loss": 1.4479,
      "step": 15972
    },
    {
      "epoch": 0.5720271455942127,
      "grad_norm": 1.6135319471359253,
      "learning_rate": 8.163343175405712e-05,
      "loss": 1.3372,
      "step": 15973
    },
    {
      "epoch": 0.572062957723781,
      "grad_norm": 1.8496217727661133,
      "learning_rate": 8.162203017820186e-05,
      "loss": 1.4587,
      "step": 15974
    },
    {
      "epoch": 0.5720987698533493,
      "grad_norm": 1.8537300825119019,
      "learning_rate": 8.161062884959852e-05,
      "loss": 1.5282,
      "step": 15975
    },
    {
      "epoch": 0.5721345819829177,
      "grad_norm": 1.6979246139526367,
      "learning_rate": 8.159922776840039e-05,
      "loss": 1.4608,
      "step": 15976
    },
    {
      "epoch": 0.5721703941124859,
      "grad_norm": 1.7887334823608398,
      "learning_rate": 8.158782693476099e-05,
      "loss": 1.6018,
      "step": 15977
    },
    {
      "epoch": 0.5722062062420542,
      "grad_norm": 1.8376210927963257,
      "learning_rate": 8.157642634883355e-05,
      "loss": 1.3104,
      "step": 15978
    },
    {
      "epoch": 0.5722420183716225,
      "grad_norm": 1.831979751586914,
      "learning_rate": 8.156502601077159e-05,
      "loss": 1.363,
      "step": 15979
    },
    {
      "epoch": 0.5722778305011907,
      "grad_norm": 1.3710800409317017,
      "learning_rate": 8.155362592072837e-05,
      "loss": 1.4114,
      "step": 15980
    },
    {
      "epoch": 0.572313642630759,
      "grad_norm": 1.5481157302856445,
      "learning_rate": 8.15422260788574e-05,
      "loss": 1.4933,
      "step": 15981
    },
    {
      "epoch": 0.5723494547603273,
      "grad_norm": 1.3802767992019653,
      "learning_rate": 8.153082648531192e-05,
      "loss": 1.6345,
      "step": 15982
    },
    {
      "epoch": 0.5723852668898957,
      "grad_norm": 1.3365625143051147,
      "learning_rate": 8.151942714024534e-05,
      "loss": 1.3163,
      "step": 15983
    },
    {
      "epoch": 0.5724210790194639,
      "grad_norm": 1.9832862615585327,
      "learning_rate": 8.150802804381105e-05,
      "loss": 1.4807,
      "step": 15984
    },
    {
      "epoch": 0.5724568911490322,
      "grad_norm": 2.2987887859344482,
      "learning_rate": 8.149662919616238e-05,
      "loss": 1.6213,
      "step": 15985
    },
    {
      "epoch": 0.5724927032786005,
      "grad_norm": 1.9295026063919067,
      "learning_rate": 8.14852305974527e-05,
      "loss": 1.5155,
      "step": 15986
    },
    {
      "epoch": 0.5725285154081687,
      "grad_norm": 2.3797919750213623,
      "learning_rate": 8.147383224783534e-05,
      "loss": 1.4002,
      "step": 15987
    },
    {
      "epoch": 0.572564327537737,
      "grad_norm": 1.760378360748291,
      "learning_rate": 8.146243414746371e-05,
      "loss": 1.4426,
      "step": 15988
    },
    {
      "epoch": 0.5726001396673053,
      "grad_norm": 1.804693579673767,
      "learning_rate": 8.145103629649104e-05,
      "loss": 1.3964,
      "step": 15989
    },
    {
      "epoch": 0.5726359517968737,
      "grad_norm": 2.0817153453826904,
      "learning_rate": 8.143963869507085e-05,
      "loss": 1.2892,
      "step": 15990
    },
    {
      "epoch": 0.5726717639264419,
      "grad_norm": 2.2405829429626465,
      "learning_rate": 8.142824134335633e-05,
      "loss": 1.5307,
      "step": 15991
    },
    {
      "epoch": 0.5727075760560102,
      "grad_norm": 1.2104196548461914,
      "learning_rate": 8.141684424150087e-05,
      "loss": 1.6489,
      "step": 15992
    },
    {
      "epoch": 0.5727433881855785,
      "grad_norm": 1.529676079750061,
      "learning_rate": 8.14054473896578e-05,
      "loss": 1.5803,
      "step": 15993
    },
    {
      "epoch": 0.5727792003151467,
      "grad_norm": 1.550414800643921,
      "learning_rate": 8.139405078798044e-05,
      "loss": 1.6136,
      "step": 15994
    },
    {
      "epoch": 0.572815012444715,
      "grad_norm": 1.7423759698867798,
      "learning_rate": 8.138265443662215e-05,
      "loss": 1.6426,
      "step": 15995
    },
    {
      "epoch": 0.5728508245742833,
      "grad_norm": 1.7016830444335938,
      "learning_rate": 8.137125833573622e-05,
      "loss": 1.4305,
      "step": 15996
    },
    {
      "epoch": 0.5728866367038516,
      "grad_norm": 2.0163040161132812,
      "learning_rate": 8.135986248547597e-05,
      "loss": 1.7001,
      "step": 15997
    },
    {
      "epoch": 0.5729224488334199,
      "grad_norm": 1.8785911798477173,
      "learning_rate": 8.134846688599473e-05,
      "loss": 1.4775,
      "step": 15998
    },
    {
      "epoch": 0.5729582609629882,
      "grad_norm": 1.5626881122589111,
      "learning_rate": 8.133707153744582e-05,
      "loss": 1.2578,
      "step": 15999
    },
    {
      "epoch": 0.5729940730925565,
      "grad_norm": 1.6098331212997437,
      "learning_rate": 8.132567643998254e-05,
      "loss": 1.4393,
      "step": 16000
    },
    {
      "epoch": 0.5730298852221247,
      "grad_norm": 1.5904412269592285,
      "learning_rate": 8.131428159375817e-05,
      "loss": 1.4826,
      "step": 16001
    },
    {
      "epoch": 0.573065697351693,
      "grad_norm": 2.667109489440918,
      "learning_rate": 8.130288699892608e-05,
      "loss": 1.6257,
      "step": 16002
    },
    {
      "epoch": 0.5731015094812613,
      "grad_norm": 1.7234574556350708,
      "learning_rate": 8.129149265563947e-05,
      "loss": 1.4741,
      "step": 16003
    },
    {
      "epoch": 0.5731373216108296,
      "grad_norm": 2.2013614177703857,
      "learning_rate": 8.128009856405174e-05,
      "loss": 1.6424,
      "step": 16004
    },
    {
      "epoch": 0.5731731337403979,
      "grad_norm": 1.7291687726974487,
      "learning_rate": 8.126870472431613e-05,
      "loss": 1.4264,
      "step": 16005
    },
    {
      "epoch": 0.5732089458699662,
      "grad_norm": 1.8733566999435425,
      "learning_rate": 8.125731113658594e-05,
      "loss": 1.3412,
      "step": 16006
    },
    {
      "epoch": 0.5732447579995344,
      "grad_norm": 1.6369208097457886,
      "learning_rate": 8.124591780101443e-05,
      "loss": 1.4097,
      "step": 16007
    },
    {
      "epoch": 0.5732805701291027,
      "grad_norm": 1.460530400276184,
      "learning_rate": 8.123452471775493e-05,
      "loss": 1.4468,
      "step": 16008
    },
    {
      "epoch": 0.573316382258671,
      "grad_norm": 2.263429880142212,
      "learning_rate": 8.122313188696068e-05,
      "loss": 1.2833,
      "step": 16009
    },
    {
      "epoch": 0.5733521943882393,
      "grad_norm": 1.4812617301940918,
      "learning_rate": 8.121173930878496e-05,
      "loss": 1.3627,
      "step": 16010
    },
    {
      "epoch": 0.5733880065178076,
      "grad_norm": 1.905060052871704,
      "learning_rate": 8.120034698338108e-05,
      "loss": 1.6278,
      "step": 16011
    },
    {
      "epoch": 0.5734238186473759,
      "grad_norm": 1.6872472763061523,
      "learning_rate": 8.118895491090225e-05,
      "loss": 1.425,
      "step": 16012
    },
    {
      "epoch": 0.5734596307769442,
      "grad_norm": 1.7046935558319092,
      "learning_rate": 8.11775630915018e-05,
      "loss": 1.628,
      "step": 16013
    },
    {
      "epoch": 0.5734954429065124,
      "grad_norm": 1.5307683944702148,
      "learning_rate": 8.116617152533292e-05,
      "loss": 1.6749,
      "step": 16014
    },
    {
      "epoch": 0.5735312550360807,
      "grad_norm": 1.8192733526229858,
      "learning_rate": 8.115478021254895e-05,
      "loss": 1.5034,
      "step": 16015
    },
    {
      "epoch": 0.573567067165649,
      "grad_norm": 1.5006341934204102,
      "learning_rate": 8.114338915330307e-05,
      "loss": 1.0562,
      "step": 16016
    },
    {
      "epoch": 0.5736028792952172,
      "grad_norm": 1.6144181489944458,
      "learning_rate": 8.113199834774858e-05,
      "loss": 1.4468,
      "step": 16017
    },
    {
      "epoch": 0.5736386914247856,
      "grad_norm": 1.690004825592041,
      "learning_rate": 8.112060779603873e-05,
      "loss": 1.3783,
      "step": 16018
    },
    {
      "epoch": 0.5736745035543539,
      "grad_norm": 1.5593258142471313,
      "learning_rate": 8.110921749832672e-05,
      "loss": 1.6404,
      "step": 16019
    },
    {
      "epoch": 0.5737103156839222,
      "grad_norm": 2.3472657203674316,
      "learning_rate": 8.109782745476585e-05,
      "loss": 1.8699,
      "step": 16020
    },
    {
      "epoch": 0.5737461278134904,
      "grad_norm": 1.9007211923599243,
      "learning_rate": 8.108643766550929e-05,
      "loss": 1.3917,
      "step": 16021
    },
    {
      "epoch": 0.5737819399430587,
      "grad_norm": 1.5271756649017334,
      "learning_rate": 8.107504813071036e-05,
      "loss": 1.498,
      "step": 16022
    },
    {
      "epoch": 0.573817752072627,
      "grad_norm": 1.8045490980148315,
      "learning_rate": 8.10636588505222e-05,
      "loss": 1.4419,
      "step": 16023
    },
    {
      "epoch": 0.5738535642021952,
      "grad_norm": 1.7356674671173096,
      "learning_rate": 8.105226982509812e-05,
      "loss": 1.7494,
      "step": 16024
    },
    {
      "epoch": 0.5738893763317636,
      "grad_norm": 1.2682982683181763,
      "learning_rate": 8.10408810545913e-05,
      "loss": 1.3909,
      "step": 16025
    },
    {
      "epoch": 0.5739251884613319,
      "grad_norm": 1.9603110551834106,
      "learning_rate": 8.102949253915497e-05,
      "loss": 1.4457,
      "step": 16026
    },
    {
      "epoch": 0.5739610005909002,
      "grad_norm": 1.7176648378372192,
      "learning_rate": 8.101810427894236e-05,
      "loss": 1.6139,
      "step": 16027
    },
    {
      "epoch": 0.5739968127204684,
      "grad_norm": 1.5585652589797974,
      "learning_rate": 8.100671627410664e-05,
      "loss": 1.4038,
      "step": 16028
    },
    {
      "epoch": 0.5740326248500367,
      "grad_norm": 1.3921308517456055,
      "learning_rate": 8.099532852480108e-05,
      "loss": 1.4396,
      "step": 16029
    },
    {
      "epoch": 0.574068436979605,
      "grad_norm": 1.6771459579467773,
      "learning_rate": 8.098394103117885e-05,
      "loss": 1.4582,
      "step": 16030
    },
    {
      "epoch": 0.5741042491091732,
      "grad_norm": 1.6584618091583252,
      "learning_rate": 8.097255379339317e-05,
      "loss": 1.3684,
      "step": 16031
    },
    {
      "epoch": 0.5741400612387416,
      "grad_norm": 1.7140179872512817,
      "learning_rate": 8.096116681159722e-05,
      "loss": 1.2808,
      "step": 16032
    },
    {
      "epoch": 0.5741758733683099,
      "grad_norm": 1.8170238733291626,
      "learning_rate": 8.094978008594423e-05,
      "loss": 1.5748,
      "step": 16033
    },
    {
      "epoch": 0.5742116854978782,
      "grad_norm": 1.9986485242843628,
      "learning_rate": 8.093839361658735e-05,
      "loss": 1.4221,
      "step": 16034
    },
    {
      "epoch": 0.5742474976274464,
      "grad_norm": 1.4016669988632202,
      "learning_rate": 8.092700740367983e-05,
      "loss": 1.4064,
      "step": 16035
    },
    {
      "epoch": 0.5742833097570147,
      "grad_norm": 1.619900107383728,
      "learning_rate": 8.091562144737481e-05,
      "loss": 1.6672,
      "step": 16036
    },
    {
      "epoch": 0.574319121886583,
      "grad_norm": 1.6384003162384033,
      "learning_rate": 8.090423574782549e-05,
      "loss": 1.5615,
      "step": 16037
    },
    {
      "epoch": 0.5743549340161512,
      "grad_norm": 1.4524556398391724,
      "learning_rate": 8.089285030518504e-05,
      "loss": 1.5234,
      "step": 16038
    },
    {
      "epoch": 0.5743907461457196,
      "grad_norm": 1.6690996885299683,
      "learning_rate": 8.088146511960663e-05,
      "loss": 1.8467,
      "step": 16039
    },
    {
      "epoch": 0.5744265582752879,
      "grad_norm": 1.6018950939178467,
      "learning_rate": 8.087008019124347e-05,
      "loss": 1.1323,
      "step": 16040
    },
    {
      "epoch": 0.5744623704048561,
      "grad_norm": 1.966084599494934,
      "learning_rate": 8.085869552024869e-05,
      "loss": 1.4668,
      "step": 16041
    },
    {
      "epoch": 0.5744981825344244,
      "grad_norm": 1.7654863595962524,
      "learning_rate": 8.084731110677548e-05,
      "loss": 1.4355,
      "step": 16042
    },
    {
      "epoch": 0.5745339946639927,
      "grad_norm": 1.6190152168273926,
      "learning_rate": 8.083592695097702e-05,
      "loss": 1.531,
      "step": 16043
    },
    {
      "epoch": 0.574569806793561,
      "grad_norm": 1.6661125421524048,
      "learning_rate": 8.082454305300637e-05,
      "loss": 1.3279,
      "step": 16044
    },
    {
      "epoch": 0.5746056189231292,
      "grad_norm": 1.7820805311203003,
      "learning_rate": 8.081315941301683e-05,
      "loss": 1.6347,
      "step": 16045
    },
    {
      "epoch": 0.5746414310526975,
      "grad_norm": 1.4383074045181274,
      "learning_rate": 8.080177603116142e-05,
      "loss": 1.7054,
      "step": 16046
    },
    {
      "epoch": 0.5746772431822659,
      "grad_norm": 1.4362151622772217,
      "learning_rate": 8.079039290759341e-05,
      "loss": 1.6129,
      "step": 16047
    },
    {
      "epoch": 0.5747130553118341,
      "grad_norm": 1.533974051475525,
      "learning_rate": 8.077901004246584e-05,
      "loss": 1.7049,
      "step": 16048
    },
    {
      "epoch": 0.5747488674414024,
      "grad_norm": 1.90034019947052,
      "learning_rate": 8.076762743593191e-05,
      "loss": 1.4234,
      "step": 16049
    },
    {
      "epoch": 0.5747846795709707,
      "grad_norm": 1.5940901041030884,
      "learning_rate": 8.075624508814474e-05,
      "loss": 1.4504,
      "step": 16050
    },
    {
      "epoch": 0.574820491700539,
      "grad_norm": 1.7032641172409058,
      "learning_rate": 8.074486299925749e-05,
      "loss": 1.4603,
      "step": 16051
    },
    {
      "epoch": 0.5748563038301072,
      "grad_norm": 1.4532485008239746,
      "learning_rate": 8.073348116942329e-05,
      "loss": 1.7031,
      "step": 16052
    },
    {
      "epoch": 0.5748921159596755,
      "grad_norm": 2.4565727710723877,
      "learning_rate": 8.072209959879517e-05,
      "loss": 1.1419,
      "step": 16053
    },
    {
      "epoch": 0.5749279280892439,
      "grad_norm": 1.4893320798873901,
      "learning_rate": 8.071071828752643e-05,
      "loss": 1.2288,
      "step": 16054
    },
    {
      "epoch": 0.5749637402188121,
      "grad_norm": 1.7563259601593018,
      "learning_rate": 8.069933723577e-05,
      "loss": 1.6753,
      "step": 16055
    },
    {
      "epoch": 0.5749995523483804,
      "grad_norm": 2.3505520820617676,
      "learning_rate": 8.068795644367918e-05,
      "loss": 1.5375,
      "step": 16056
    },
    {
      "epoch": 0.5750353644779487,
      "grad_norm": 1.5410603284835815,
      "learning_rate": 8.06765759114069e-05,
      "loss": 1.742,
      "step": 16057
    },
    {
      "epoch": 0.5750711766075169,
      "grad_norm": 1.543412446975708,
      "learning_rate": 8.066519563910645e-05,
      "loss": 1.3408,
      "step": 16058
    },
    {
      "epoch": 0.5751069887370852,
      "grad_norm": 2.5051183700561523,
      "learning_rate": 8.065381562693078e-05,
      "loss": 1.6021,
      "step": 16059
    },
    {
      "epoch": 0.5751428008666535,
      "grad_norm": 2.775167942047119,
      "learning_rate": 8.064243587503313e-05,
      "loss": 1.4534,
      "step": 16060
    },
    {
      "epoch": 0.5751786129962219,
      "grad_norm": 1.8102688789367676,
      "learning_rate": 8.063105638356654e-05,
      "loss": 1.2111,
      "step": 16061
    },
    {
      "epoch": 0.5752144251257901,
      "grad_norm": 2.6372623443603516,
      "learning_rate": 8.061967715268403e-05,
      "loss": 1.7916,
      "step": 16062
    },
    {
      "epoch": 0.5752502372553584,
      "grad_norm": 1.2923922538757324,
      "learning_rate": 8.060829818253884e-05,
      "loss": 1.3766,
      "step": 16063
    },
    {
      "epoch": 0.5752860493849267,
      "grad_norm": 1.690511703491211,
      "learning_rate": 8.059691947328391e-05,
      "loss": 1.5836,
      "step": 16064
    },
    {
      "epoch": 0.5753218615144949,
      "grad_norm": 1.6254609823226929,
      "learning_rate": 8.058554102507248e-05,
      "loss": 1.6672,
      "step": 16065
    },
    {
      "epoch": 0.5753576736440632,
      "grad_norm": 1.411546230316162,
      "learning_rate": 8.057416283805748e-05,
      "loss": 1.4695,
      "step": 16066
    },
    {
      "epoch": 0.5753934857736315,
      "grad_norm": 1.4763381481170654,
      "learning_rate": 8.056278491239213e-05,
      "loss": 1.1461,
      "step": 16067
    },
    {
      "epoch": 0.5754292979031999,
      "grad_norm": 2.3086869716644287,
      "learning_rate": 8.055140724822938e-05,
      "loss": 1.614,
      "step": 16068
    },
    {
      "epoch": 0.5754651100327681,
      "grad_norm": 1.1650549173355103,
      "learning_rate": 8.054002984572241e-05,
      "loss": 1.2589,
      "step": 16069
    },
    {
      "epoch": 0.5755009221623364,
      "grad_norm": 1.5132914781570435,
      "learning_rate": 8.052865270502422e-05,
      "loss": 1.4143,
      "step": 16070
    },
    {
      "epoch": 0.5755367342919047,
      "grad_norm": 1.7870749235153198,
      "learning_rate": 8.051727582628788e-05,
      "loss": 1.3014,
      "step": 16071
    },
    {
      "epoch": 0.5755725464214729,
      "grad_norm": 1.9012000560760498,
      "learning_rate": 8.050589920966647e-05,
      "loss": 1.5146,
      "step": 16072
    },
    {
      "epoch": 0.5756083585510412,
      "grad_norm": 1.918483853340149,
      "learning_rate": 8.049452285531302e-05,
      "loss": 1.6011,
      "step": 16073
    },
    {
      "epoch": 0.5756441706806095,
      "grad_norm": 1.824968695640564,
      "learning_rate": 8.048314676338062e-05,
      "loss": 1.629,
      "step": 16074
    },
    {
      "epoch": 0.5756799828101778,
      "grad_norm": 1.566021203994751,
      "learning_rate": 8.047177093402228e-05,
      "loss": 1.4753,
      "step": 16075
    },
    {
      "epoch": 0.5757157949397461,
      "grad_norm": 1.6824095249176025,
      "learning_rate": 8.046039536739111e-05,
      "loss": 1.3666,
      "step": 16076
    },
    {
      "epoch": 0.5757516070693144,
      "grad_norm": 2.086169958114624,
      "learning_rate": 8.044902006364008e-05,
      "loss": 1.6941,
      "step": 16077
    },
    {
      "epoch": 0.5757874191988827,
      "grad_norm": 1.4665168523788452,
      "learning_rate": 8.043764502292232e-05,
      "loss": 1.5978,
      "step": 16078
    },
    {
      "epoch": 0.5758232313284509,
      "grad_norm": 1.646644949913025,
      "learning_rate": 8.04262702453908e-05,
      "loss": 1.5653,
      "step": 16079
    },
    {
      "epoch": 0.5758590434580192,
      "grad_norm": 1.9247905015945435,
      "learning_rate": 8.041489573119853e-05,
      "loss": 1.3912,
      "step": 16080
    },
    {
      "epoch": 0.5758948555875875,
      "grad_norm": 1.658804178237915,
      "learning_rate": 8.04035214804986e-05,
      "loss": 1.3743,
      "step": 16081
    },
    {
      "epoch": 0.5759306677171558,
      "grad_norm": 1.8746556043624878,
      "learning_rate": 8.0392147493444e-05,
      "loss": 1.3236,
      "step": 16082
    },
    {
      "epoch": 0.5759664798467241,
      "grad_norm": 1.518371343612671,
      "learning_rate": 8.038077377018776e-05,
      "loss": 1.2282,
      "step": 16083
    },
    {
      "epoch": 0.5760022919762924,
      "grad_norm": 2.3258750438690186,
      "learning_rate": 8.03694003108829e-05,
      "loss": 1.6595,
      "step": 16084
    },
    {
      "epoch": 0.5760381041058606,
      "grad_norm": 2.2161455154418945,
      "learning_rate": 8.035802711568245e-05,
      "loss": 1.6238,
      "step": 16085
    },
    {
      "epoch": 0.5760739162354289,
      "grad_norm": 1.8260993957519531,
      "learning_rate": 8.03466541847394e-05,
      "loss": 1.3711,
      "step": 16086
    },
    {
      "epoch": 0.5761097283649972,
      "grad_norm": 1.9408621788024902,
      "learning_rate": 8.033528151820679e-05,
      "loss": 1.7465,
      "step": 16087
    },
    {
      "epoch": 0.5761455404945655,
      "grad_norm": 1.375427007675171,
      "learning_rate": 8.032390911623758e-05,
      "loss": 1.6076,
      "step": 16088
    },
    {
      "epoch": 0.5761813526241338,
      "grad_norm": 1.5959179401397705,
      "learning_rate": 8.031253697898478e-05,
      "loss": 1.5481,
      "step": 16089
    },
    {
      "epoch": 0.5762171647537021,
      "grad_norm": 1.694019079208374,
      "learning_rate": 8.030116510660143e-05,
      "loss": 1.0399,
      "step": 16090
    },
    {
      "epoch": 0.5762529768832704,
      "grad_norm": 1.893786907196045,
      "learning_rate": 8.028979349924048e-05,
      "loss": 1.616,
      "step": 16091
    },
    {
      "epoch": 0.5762887890128386,
      "grad_norm": 1.5224804878234863,
      "learning_rate": 8.027842215705494e-05,
      "loss": 1.2418,
      "step": 16092
    },
    {
      "epoch": 0.5763246011424069,
      "grad_norm": 1.8478490114212036,
      "learning_rate": 8.026705108019777e-05,
      "loss": 1.4941,
      "step": 16093
    },
    {
      "epoch": 0.5763604132719752,
      "grad_norm": 1.8832570314407349,
      "learning_rate": 8.0255680268822e-05,
      "loss": 1.2081,
      "step": 16094
    },
    {
      "epoch": 0.5763962254015434,
      "grad_norm": 1.4357125759124756,
      "learning_rate": 8.024430972308056e-05,
      "loss": 1.2784,
      "step": 16095
    },
    {
      "epoch": 0.5764320375311118,
      "grad_norm": 1.4434853792190552,
      "learning_rate": 8.023293944312647e-05,
      "loss": 1.605,
      "step": 16096
    },
    {
      "epoch": 0.5764678496606801,
      "grad_norm": 1.8244801759719849,
      "learning_rate": 8.022156942911267e-05,
      "loss": 1.3657,
      "step": 16097
    },
    {
      "epoch": 0.5765036617902484,
      "grad_norm": 1.4665894508361816,
      "learning_rate": 8.021019968119215e-05,
      "loss": 1.6977,
      "step": 16098
    },
    {
      "epoch": 0.5765394739198166,
      "grad_norm": 1.9285708665847778,
      "learning_rate": 8.019883019951786e-05,
      "loss": 1.2284,
      "step": 16099
    },
    {
      "epoch": 0.5765752860493849,
      "grad_norm": 2.212251663208008,
      "learning_rate": 8.018746098424276e-05,
      "loss": 1.763,
      "step": 16100
    },
    {
      "epoch": 0.5766110981789532,
      "grad_norm": 1.7550066709518433,
      "learning_rate": 8.017609203551983e-05,
      "loss": 1.4755,
      "step": 16101
    },
    {
      "epoch": 0.5766469103085214,
      "grad_norm": 1.8005722761154175,
      "learning_rate": 8.0164723353502e-05,
      "loss": 1.434,
      "step": 16102
    },
    {
      "epoch": 0.5766827224380898,
      "grad_norm": 1.9695098400115967,
      "learning_rate": 8.015335493834224e-05,
      "loss": 1.6995,
      "step": 16103
    },
    {
      "epoch": 0.5767185345676581,
      "grad_norm": 1.847440481185913,
      "learning_rate": 8.014198679019348e-05,
      "loss": 1.7125,
      "step": 16104
    },
    {
      "epoch": 0.5767543466972264,
      "grad_norm": 1.2874605655670166,
      "learning_rate": 8.01306189092087e-05,
      "loss": 1.4815,
      "step": 16105
    },
    {
      "epoch": 0.5767901588267946,
      "grad_norm": 1.833154559135437,
      "learning_rate": 8.011925129554078e-05,
      "loss": 1.2951,
      "step": 16106
    },
    {
      "epoch": 0.5768259709563629,
      "grad_norm": 1.9240434169769287,
      "learning_rate": 8.01078839493427e-05,
      "loss": 1.1895,
      "step": 16107
    },
    {
      "epoch": 0.5768617830859312,
      "grad_norm": 1.921027421951294,
      "learning_rate": 8.009651687076739e-05,
      "loss": 1.4075,
      "step": 16108
    },
    {
      "epoch": 0.5768975952154994,
      "grad_norm": 1.5504812002182007,
      "learning_rate": 8.008515005996775e-05,
      "loss": 1.2061,
      "step": 16109
    },
    {
      "epoch": 0.5769334073450678,
      "grad_norm": 1.6226364374160767,
      "learning_rate": 8.007378351709676e-05,
      "loss": 1.6403,
      "step": 16110
    },
    {
      "epoch": 0.5769692194746361,
      "grad_norm": 1.543958067893982,
      "learning_rate": 8.006241724230728e-05,
      "loss": 1.6479,
      "step": 16111
    },
    {
      "epoch": 0.5770050316042044,
      "grad_norm": 1.6312440633773804,
      "learning_rate": 8.005105123575228e-05,
      "loss": 1.4381,
      "step": 16112
    },
    {
      "epoch": 0.5770408437337726,
      "grad_norm": 1.547906517982483,
      "learning_rate": 8.003968549758462e-05,
      "loss": 1.6837,
      "step": 16113
    },
    {
      "epoch": 0.5770766558633409,
      "grad_norm": 2.2383906841278076,
      "learning_rate": 8.002832002795729e-05,
      "loss": 0.9843,
      "step": 16114
    },
    {
      "epoch": 0.5771124679929092,
      "grad_norm": 1.4814233779907227,
      "learning_rate": 8.001695482702314e-05,
      "loss": 1.3434,
      "step": 16115
    },
    {
      "epoch": 0.5771482801224774,
      "grad_norm": 1.837546706199646,
      "learning_rate": 8.000558989493507e-05,
      "loss": 1.544,
      "step": 16116
    },
    {
      "epoch": 0.5771840922520458,
      "grad_norm": 1.2523295879364014,
      "learning_rate": 7.999422523184601e-05,
      "loss": 1.287,
      "step": 16117
    },
    {
      "epoch": 0.5772199043816141,
      "grad_norm": 1.5280488729476929,
      "learning_rate": 7.998286083790883e-05,
      "loss": 1.323,
      "step": 16118
    },
    {
      "epoch": 0.5772557165111823,
      "grad_norm": 1.6794685125350952,
      "learning_rate": 7.997149671327646e-05,
      "loss": 1.4495,
      "step": 16119
    },
    {
      "epoch": 0.5772915286407506,
      "grad_norm": 2.4332382678985596,
      "learning_rate": 7.996013285810173e-05,
      "loss": 1.5304,
      "step": 16120
    },
    {
      "epoch": 0.5773273407703189,
      "grad_norm": 1.4107089042663574,
      "learning_rate": 7.99487692725376e-05,
      "loss": 1.2941,
      "step": 16121
    },
    {
      "epoch": 0.5773631528998872,
      "grad_norm": 1.6227834224700928,
      "learning_rate": 7.993740595673689e-05,
      "loss": 1.3497,
      "step": 16122
    },
    {
      "epoch": 0.5773989650294554,
      "grad_norm": 1.5155694484710693,
      "learning_rate": 7.992604291085253e-05,
      "loss": 1.6335,
      "step": 16123
    },
    {
      "epoch": 0.5774347771590238,
      "grad_norm": 2.1160717010498047,
      "learning_rate": 7.991468013503735e-05,
      "loss": 1.6124,
      "step": 16124
    },
    {
      "epoch": 0.5774705892885921,
      "grad_norm": 1.6966854333877563,
      "learning_rate": 7.990331762944426e-05,
      "loss": 1.4299,
      "step": 16125
    },
    {
      "epoch": 0.5775064014181603,
      "grad_norm": 1.8344290256500244,
      "learning_rate": 7.989195539422609e-05,
      "loss": 1.5318,
      "step": 16126
    },
    {
      "epoch": 0.5775422135477286,
      "grad_norm": 2.1543333530426025,
      "learning_rate": 7.988059342953571e-05,
      "loss": 1.7969,
      "step": 16127
    },
    {
      "epoch": 0.5775780256772969,
      "grad_norm": 1.9290580749511719,
      "learning_rate": 7.986923173552602e-05,
      "loss": 1.7385,
      "step": 16128
    },
    {
      "epoch": 0.5776138378068651,
      "grad_norm": 1.6262422800064087,
      "learning_rate": 7.985787031234983e-05,
      "loss": 1.4024,
      "step": 16129
    },
    {
      "epoch": 0.5776496499364334,
      "grad_norm": 1.9909801483154297,
      "learning_rate": 7.984650916016003e-05,
      "loss": 1.3964,
      "step": 16130
    },
    {
      "epoch": 0.5776854620660018,
      "grad_norm": 1.3821978569030762,
      "learning_rate": 7.983514827910943e-05,
      "loss": 1.3624,
      "step": 16131
    },
    {
      "epoch": 0.5777212741955701,
      "grad_norm": 1.4430720806121826,
      "learning_rate": 7.982378766935092e-05,
      "loss": 1.6197,
      "step": 16132
    },
    {
      "epoch": 0.5777570863251383,
      "grad_norm": 1.3576154708862305,
      "learning_rate": 7.981242733103734e-05,
      "loss": 1.4386,
      "step": 16133
    },
    {
      "epoch": 0.5777928984547066,
      "grad_norm": 2.1250572204589844,
      "learning_rate": 7.980106726432148e-05,
      "loss": 1.2357,
      "step": 16134
    },
    {
      "epoch": 0.5778287105842749,
      "grad_norm": 1.6728894710540771,
      "learning_rate": 7.978970746935621e-05,
      "loss": 1.5269,
      "step": 16135
    },
    {
      "epoch": 0.5778645227138431,
      "grad_norm": 2.4180312156677246,
      "learning_rate": 7.977834794629436e-05,
      "loss": 1.5775,
      "step": 16136
    },
    {
      "epoch": 0.5779003348434114,
      "grad_norm": 1.6585233211517334,
      "learning_rate": 7.976698869528876e-05,
      "loss": 1.5318,
      "step": 16137
    },
    {
      "epoch": 0.5779361469729798,
      "grad_norm": 1.3715888261795044,
      "learning_rate": 7.975562971649219e-05,
      "loss": 1.5977,
      "step": 16138
    },
    {
      "epoch": 0.5779719591025481,
      "grad_norm": 1.8172003030776978,
      "learning_rate": 7.974427101005756e-05,
      "loss": 1.3096,
      "step": 16139
    },
    {
      "epoch": 0.5780077712321163,
      "grad_norm": 1.5655006170272827,
      "learning_rate": 7.973291257613761e-05,
      "loss": 1.4269,
      "step": 16140
    },
    {
      "epoch": 0.5780435833616846,
      "grad_norm": 1.313693881034851,
      "learning_rate": 7.97215544148852e-05,
      "loss": 1.3408,
      "step": 16141
    },
    {
      "epoch": 0.5780793954912529,
      "grad_norm": 1.644073486328125,
      "learning_rate": 7.971019652645313e-05,
      "loss": 1.4267,
      "step": 16142
    },
    {
      "epoch": 0.5781152076208211,
      "grad_norm": 2.4095518589019775,
      "learning_rate": 7.969883891099412e-05,
      "loss": 1.4167,
      "step": 16143
    },
    {
      "epoch": 0.5781510197503894,
      "grad_norm": 2.725630044937134,
      "learning_rate": 7.968748156866113e-05,
      "loss": 1.35,
      "step": 16144
    },
    {
      "epoch": 0.5781868318799578,
      "grad_norm": 2.1629457473754883,
      "learning_rate": 7.967612449960679e-05,
      "loss": 1.6466,
      "step": 16145
    },
    {
      "epoch": 0.578222644009526,
      "grad_norm": 1.4881287813186646,
      "learning_rate": 7.966476770398404e-05,
      "loss": 1.5601,
      "step": 16146
    },
    {
      "epoch": 0.5782584561390943,
      "grad_norm": 1.5154433250427246,
      "learning_rate": 7.965341118194559e-05,
      "loss": 1.3574,
      "step": 16147
    },
    {
      "epoch": 0.5782942682686626,
      "grad_norm": 1.5236823558807373,
      "learning_rate": 7.964205493364426e-05,
      "loss": 1.5902,
      "step": 16148
    },
    {
      "epoch": 0.5783300803982309,
      "grad_norm": 1.974445104598999,
      "learning_rate": 7.963069895923285e-05,
      "loss": 1.5286,
      "step": 16149
    },
    {
      "epoch": 0.5783658925277991,
      "grad_norm": 2.1835594177246094,
      "learning_rate": 7.961934325886404e-05,
      "loss": 1.2116,
      "step": 16150
    },
    {
      "epoch": 0.5784017046573674,
      "grad_norm": 1.4302401542663574,
      "learning_rate": 7.960798783269074e-05,
      "loss": 1.565,
      "step": 16151
    },
    {
      "epoch": 0.5784375167869358,
      "grad_norm": 2.1140241622924805,
      "learning_rate": 7.95966326808656e-05,
      "loss": 1.4891,
      "step": 16152
    },
    {
      "epoch": 0.578473328916504,
      "grad_norm": 1.4253383874893188,
      "learning_rate": 7.958527780354151e-05,
      "loss": 1.386,
      "step": 16153
    },
    {
      "epoch": 0.5785091410460723,
      "grad_norm": 1.2918790578842163,
      "learning_rate": 7.957392320087112e-05,
      "loss": 1.9235,
      "step": 16154
    },
    {
      "epoch": 0.5785449531756406,
      "grad_norm": 1.124372959136963,
      "learning_rate": 7.956256887300729e-05,
      "loss": 1.4101,
      "step": 16155
    },
    {
      "epoch": 0.5785807653052089,
      "grad_norm": 1.958508014678955,
      "learning_rate": 7.955121482010268e-05,
      "loss": 1.273,
      "step": 16156
    },
    {
      "epoch": 0.5786165774347771,
      "grad_norm": 2.251087188720703,
      "learning_rate": 7.953986104231018e-05,
      "loss": 1.4734,
      "step": 16157
    },
    {
      "epoch": 0.5786523895643454,
      "grad_norm": 2.016096353530884,
      "learning_rate": 7.95285075397824e-05,
      "loss": 1.6336,
      "step": 16158
    },
    {
      "epoch": 0.5786882016939138,
      "grad_norm": 1.6927529573440552,
      "learning_rate": 7.951715431267213e-05,
      "loss": 1.6828,
      "step": 16159
    },
    {
      "epoch": 0.578724013823482,
      "grad_norm": 1.4586488008499146,
      "learning_rate": 7.950580136113219e-05,
      "loss": 1.4452,
      "step": 16160
    },
    {
      "epoch": 0.5787598259530503,
      "grad_norm": 1.6537890434265137,
      "learning_rate": 7.949444868531517e-05,
      "loss": 1.5215,
      "step": 16161
    },
    {
      "epoch": 0.5787956380826186,
      "grad_norm": 1.5151437520980835,
      "learning_rate": 7.948309628537399e-05,
      "loss": 1.3912,
      "step": 16162
    },
    {
      "epoch": 0.5788314502121868,
      "grad_norm": 1.7033724784851074,
      "learning_rate": 7.94717441614612e-05,
      "loss": 1.5666,
      "step": 16163
    },
    {
      "epoch": 0.5788672623417551,
      "grad_norm": 1.2881311178207397,
      "learning_rate": 7.946039231372967e-05,
      "loss": 1.5314,
      "step": 16164
    },
    {
      "epoch": 0.5789030744713234,
      "grad_norm": 1.7593480348587036,
      "learning_rate": 7.944904074233201e-05,
      "loss": 1.5596,
      "step": 16165
    },
    {
      "epoch": 0.5789388866008918,
      "grad_norm": 1.4670255184173584,
      "learning_rate": 7.943768944742107e-05,
      "loss": 1.5353,
      "step": 16166
    },
    {
      "epoch": 0.57897469873046,
      "grad_norm": 1.4414554834365845,
      "learning_rate": 7.942633842914946e-05,
      "loss": 1.538,
      "step": 16167
    },
    {
      "epoch": 0.5790105108600283,
      "grad_norm": 1.5942708253860474,
      "learning_rate": 7.941498768766991e-05,
      "loss": 1.3996,
      "step": 16168
    },
    {
      "epoch": 0.5790463229895966,
      "grad_norm": 2.574983596801758,
      "learning_rate": 7.940363722313519e-05,
      "loss": 1.3455,
      "step": 16169
    },
    {
      "epoch": 0.5790821351191648,
      "grad_norm": 1.7376738786697388,
      "learning_rate": 7.939228703569792e-05,
      "loss": 1.5424,
      "step": 16170
    },
    {
      "epoch": 0.5791179472487331,
      "grad_norm": 1.8088690042495728,
      "learning_rate": 7.938093712551087e-05,
      "loss": 1.7215,
      "step": 16171
    },
    {
      "epoch": 0.5791537593783014,
      "grad_norm": 1.8485448360443115,
      "learning_rate": 7.936958749272669e-05,
      "loss": 1.6542,
      "step": 16172
    },
    {
      "epoch": 0.5791895715078698,
      "grad_norm": 1.5748894214630127,
      "learning_rate": 7.935823813749815e-05,
      "loss": 1.5086,
      "step": 16173
    },
    {
      "epoch": 0.579225383637438,
      "grad_norm": 2.4885716438293457,
      "learning_rate": 7.934688905997781e-05,
      "loss": 1.6712,
      "step": 16174
    },
    {
      "epoch": 0.5792611957670063,
      "grad_norm": 1.5027422904968262,
      "learning_rate": 7.933554026031852e-05,
      "loss": 1.2673,
      "step": 16175
    },
    {
      "epoch": 0.5792970078965746,
      "grad_norm": 1.5155824422836304,
      "learning_rate": 7.932419173867286e-05,
      "loss": 1.5036,
      "step": 16176
    },
    {
      "epoch": 0.5793328200261428,
      "grad_norm": 1.7846710681915283,
      "learning_rate": 7.93128434951935e-05,
      "loss": 1.4373,
      "step": 16177
    },
    {
      "epoch": 0.5793686321557111,
      "grad_norm": 1.8113620281219482,
      "learning_rate": 7.930149553003318e-05,
      "loss": 1.6359,
      "step": 16178
    },
    {
      "epoch": 0.5794044442852794,
      "grad_norm": 1.718705654144287,
      "learning_rate": 7.92901478433445e-05,
      "loss": 1.4123,
      "step": 16179
    },
    {
      "epoch": 0.5794402564148478,
      "grad_norm": 1.435090184211731,
      "learning_rate": 7.92788004352802e-05,
      "loss": 1.567,
      "step": 16180
    },
    {
      "epoch": 0.579476068544416,
      "grad_norm": 1.6663739681243896,
      "learning_rate": 7.926745330599289e-05,
      "loss": 1.545,
      "step": 16181
    },
    {
      "epoch": 0.5795118806739843,
      "grad_norm": 1.994544506072998,
      "learning_rate": 7.925610645563527e-05,
      "loss": 1.5623,
      "step": 16182
    },
    {
      "epoch": 0.5795476928035526,
      "grad_norm": 2.0755226612091064,
      "learning_rate": 7.924475988435996e-05,
      "loss": 1.7355,
      "step": 16183
    },
    {
      "epoch": 0.5795835049331208,
      "grad_norm": 1.354204773902893,
      "learning_rate": 7.923341359231965e-05,
      "loss": 1.4404,
      "step": 16184
    },
    {
      "epoch": 0.5796193170626891,
      "grad_norm": 1.4060965776443481,
      "learning_rate": 7.922206757966698e-05,
      "loss": 1.1692,
      "step": 16185
    },
    {
      "epoch": 0.5796551291922574,
      "grad_norm": 1.943017601966858,
      "learning_rate": 7.921072184655457e-05,
      "loss": 1.2975,
      "step": 16186
    },
    {
      "epoch": 0.5796909413218257,
      "grad_norm": 1.4191889762878418,
      "learning_rate": 7.91993763931351e-05,
      "loss": 1.4388,
      "step": 16187
    },
    {
      "epoch": 0.579726753451394,
      "grad_norm": 1.3876757621765137,
      "learning_rate": 7.918803121956117e-05,
      "loss": 1.4685,
      "step": 16188
    },
    {
      "epoch": 0.5797625655809623,
      "grad_norm": 1.7913190126419067,
      "learning_rate": 7.917668632598545e-05,
      "loss": 1.269,
      "step": 16189
    },
    {
      "epoch": 0.5797983777105306,
      "grad_norm": 2.235036611557007,
      "learning_rate": 7.916534171256054e-05,
      "loss": 1.1393,
      "step": 16190
    },
    {
      "epoch": 0.5798341898400988,
      "grad_norm": 1.541192889213562,
      "learning_rate": 7.91539973794391e-05,
      "loss": 1.7477,
      "step": 16191
    },
    {
      "epoch": 0.5798700019696671,
      "grad_norm": 1.6059012413024902,
      "learning_rate": 7.914265332677371e-05,
      "loss": 1.4964,
      "step": 16192
    },
    {
      "epoch": 0.5799058140992354,
      "grad_norm": 1.5511407852172852,
      "learning_rate": 7.913130955471704e-05,
      "loss": 1.7551,
      "step": 16193
    },
    {
      "epoch": 0.5799416262288037,
      "grad_norm": 1.5841376781463623,
      "learning_rate": 7.911996606342168e-05,
      "loss": 1.3235,
      "step": 16194
    },
    {
      "epoch": 0.579977438358372,
      "grad_norm": 1.651566982269287,
      "learning_rate": 7.910862285304022e-05,
      "loss": 1.4021,
      "step": 16195
    },
    {
      "epoch": 0.5800132504879403,
      "grad_norm": 1.7354012727737427,
      "learning_rate": 7.909727992372533e-05,
      "loss": 1.5678,
      "step": 16196
    },
    {
      "epoch": 0.5800490626175085,
      "grad_norm": 1.7143256664276123,
      "learning_rate": 7.908593727562954e-05,
      "loss": 1.3513,
      "step": 16197
    },
    {
      "epoch": 0.5800848747470768,
      "grad_norm": 1.5278571844100952,
      "learning_rate": 7.907459490890551e-05,
      "loss": 1.2315,
      "step": 16198
    },
    {
      "epoch": 0.5801206868766451,
      "grad_norm": 1.3946882486343384,
      "learning_rate": 7.906325282370579e-05,
      "loss": 1.7144,
      "step": 16199
    },
    {
      "epoch": 0.5801564990062134,
      "grad_norm": 1.6457384824752808,
      "learning_rate": 7.905191102018302e-05,
      "loss": 1.3999,
      "step": 16200
    },
    {
      "epoch": 0.5801923111357817,
      "grad_norm": 1.8805018663406372,
      "learning_rate": 7.904056949848975e-05,
      "loss": 1.5579,
      "step": 16201
    },
    {
      "epoch": 0.58022812326535,
      "grad_norm": 2.192815065383911,
      "learning_rate": 7.90292282587786e-05,
      "loss": 1.4462,
      "step": 16202
    },
    {
      "epoch": 0.5802639353949183,
      "grad_norm": 1.707252860069275,
      "learning_rate": 7.901788730120214e-05,
      "loss": 1.1617,
      "step": 16203
    },
    {
      "epoch": 0.5802997475244865,
      "grad_norm": 2.2788679599761963,
      "learning_rate": 7.90065466259129e-05,
      "loss": 1.3855,
      "step": 16204
    },
    {
      "epoch": 0.5803355596540548,
      "grad_norm": 1.7162657976150513,
      "learning_rate": 7.899520623306353e-05,
      "loss": 1.2427,
      "step": 16205
    },
    {
      "epoch": 0.5803713717836231,
      "grad_norm": 2.0036752223968506,
      "learning_rate": 7.898386612280654e-05,
      "loss": 1.4296,
      "step": 16206
    },
    {
      "epoch": 0.5804071839131913,
      "grad_norm": 1.5575006008148193,
      "learning_rate": 7.897252629529455e-05,
      "loss": 1.6481,
      "step": 16207
    },
    {
      "epoch": 0.5804429960427597,
      "grad_norm": 1.8689552545547485,
      "learning_rate": 7.896118675068007e-05,
      "loss": 1.5716,
      "step": 16208
    },
    {
      "epoch": 0.580478808172328,
      "grad_norm": 1.734002947807312,
      "learning_rate": 7.894984748911572e-05,
      "loss": 1.4126,
      "step": 16209
    },
    {
      "epoch": 0.5805146203018963,
      "grad_norm": 1.9795552492141724,
      "learning_rate": 7.893850851075398e-05,
      "loss": 1.3466,
      "step": 16210
    },
    {
      "epoch": 0.5805504324314645,
      "grad_norm": 1.9898791313171387,
      "learning_rate": 7.892716981574747e-05,
      "loss": 1.5883,
      "step": 16211
    },
    {
      "epoch": 0.5805862445610328,
      "grad_norm": 1.4967081546783447,
      "learning_rate": 7.89158314042487e-05,
      "loss": 1.5994,
      "step": 16212
    },
    {
      "epoch": 0.5806220566906011,
      "grad_norm": 1.672903299331665,
      "learning_rate": 7.890449327641021e-05,
      "loss": 1.7382,
      "step": 16213
    },
    {
      "epoch": 0.5806578688201693,
      "grad_norm": 1.7039663791656494,
      "learning_rate": 7.889315543238457e-05,
      "loss": 1.6315,
      "step": 16214
    },
    {
      "epoch": 0.5806936809497377,
      "grad_norm": 1.6308118104934692,
      "learning_rate": 7.888181787232427e-05,
      "loss": 1.5503,
      "step": 16215
    },
    {
      "epoch": 0.580729493079306,
      "grad_norm": 1.8817521333694458,
      "learning_rate": 7.88704805963819e-05,
      "loss": 1.5972,
      "step": 16216
    },
    {
      "epoch": 0.5807653052088743,
      "grad_norm": 1.8613077402114868,
      "learning_rate": 7.885914360470992e-05,
      "loss": 1.6216,
      "step": 16217
    },
    {
      "epoch": 0.5808011173384425,
      "grad_norm": 1.5126614570617676,
      "learning_rate": 7.884780689746094e-05,
      "loss": 1.4662,
      "step": 16218
    },
    {
      "epoch": 0.5808369294680108,
      "grad_norm": 1.5945414304733276,
      "learning_rate": 7.88364704747874e-05,
      "loss": 1.5989,
      "step": 16219
    },
    {
      "epoch": 0.5808727415975791,
      "grad_norm": 1.5594874620437622,
      "learning_rate": 7.882513433684188e-05,
      "loss": 1.5519,
      "step": 16220
    },
    {
      "epoch": 0.5809085537271473,
      "grad_norm": 1.834822177886963,
      "learning_rate": 7.881379848377685e-05,
      "loss": 1.6284,
      "step": 16221
    },
    {
      "epoch": 0.5809443658567157,
      "grad_norm": 2.2757322788238525,
      "learning_rate": 7.880246291574482e-05,
      "loss": 1.3027,
      "step": 16222
    },
    {
      "epoch": 0.580980177986284,
      "grad_norm": 1.6082566976547241,
      "learning_rate": 7.879112763289833e-05,
      "loss": 1.4609,
      "step": 16223
    },
    {
      "epoch": 0.5810159901158523,
      "grad_norm": 1.623914361000061,
      "learning_rate": 7.877979263538983e-05,
      "loss": 1.5868,
      "step": 16224
    },
    {
      "epoch": 0.5810518022454205,
      "grad_norm": 1.4521030187606812,
      "learning_rate": 7.876845792337189e-05,
      "loss": 1.3755,
      "step": 16225
    },
    {
      "epoch": 0.5810876143749888,
      "grad_norm": 1.9787819385528564,
      "learning_rate": 7.875712349699692e-05,
      "loss": 1.0918,
      "step": 16226
    },
    {
      "epoch": 0.5811234265045571,
      "grad_norm": 2.0238037109375,
      "learning_rate": 7.874578935641748e-05,
      "loss": 1.3673,
      "step": 16227
    },
    {
      "epoch": 0.5811592386341253,
      "grad_norm": 1.6591154336929321,
      "learning_rate": 7.873445550178601e-05,
      "loss": 1.4138,
      "step": 16228
    },
    {
      "epoch": 0.5811950507636937,
      "grad_norm": 1.5953693389892578,
      "learning_rate": 7.872312193325502e-05,
      "loss": 1.095,
      "step": 16229
    },
    {
      "epoch": 0.581230862893262,
      "grad_norm": 1.9868221282958984,
      "learning_rate": 7.871178865097699e-05,
      "loss": 1.5134,
      "step": 16230
    },
    {
      "epoch": 0.5812666750228302,
      "grad_norm": 1.6181082725524902,
      "learning_rate": 7.870045565510436e-05,
      "loss": 1.344,
      "step": 16231
    },
    {
      "epoch": 0.5813024871523985,
      "grad_norm": 1.8344509601593018,
      "learning_rate": 7.868912294578965e-05,
      "loss": 1.2748,
      "step": 16232
    },
    {
      "epoch": 0.5813382992819668,
      "grad_norm": 1.7959150075912476,
      "learning_rate": 7.867779052318528e-05,
      "loss": 1.7419,
      "step": 16233
    },
    {
      "epoch": 0.581374111411535,
      "grad_norm": 1.6314499378204346,
      "learning_rate": 7.866645838744375e-05,
      "loss": 1.7027,
      "step": 16234
    },
    {
      "epoch": 0.5814099235411033,
      "grad_norm": 2.4061760902404785,
      "learning_rate": 7.865512653871749e-05,
      "loss": 1.3808,
      "step": 16235
    },
    {
      "epoch": 0.5814457356706717,
      "grad_norm": 1.7618334293365479,
      "learning_rate": 7.864379497715898e-05,
      "loss": 1.5516,
      "step": 16236
    },
    {
      "epoch": 0.58148154780024,
      "grad_norm": 1.2001464366912842,
      "learning_rate": 7.863246370292065e-05,
      "loss": 1.2329,
      "step": 16237
    },
    {
      "epoch": 0.5815173599298082,
      "grad_norm": 1.4679129123687744,
      "learning_rate": 7.862113271615499e-05,
      "loss": 1.3114,
      "step": 16238
    },
    {
      "epoch": 0.5815531720593765,
      "grad_norm": 1.4955120086669922,
      "learning_rate": 7.860980201701441e-05,
      "loss": 1.4209,
      "step": 16239
    },
    {
      "epoch": 0.5815889841889448,
      "grad_norm": 1.7343052625656128,
      "learning_rate": 7.859847160565131e-05,
      "loss": 1.4165,
      "step": 16240
    },
    {
      "epoch": 0.581624796318513,
      "grad_norm": 2.181267499923706,
      "learning_rate": 7.858714148221822e-05,
      "loss": 1.6077,
      "step": 16241
    },
    {
      "epoch": 0.5816606084480813,
      "grad_norm": 1.4039392471313477,
      "learning_rate": 7.857581164686744e-05,
      "loss": 1.7029,
      "step": 16242
    },
    {
      "epoch": 0.5816964205776497,
      "grad_norm": 2.262587070465088,
      "learning_rate": 7.856448209975156e-05,
      "loss": 1.7322,
      "step": 16243
    },
    {
      "epoch": 0.581732232707218,
      "grad_norm": 1.8444851636886597,
      "learning_rate": 7.855315284102288e-05,
      "loss": 1.194,
      "step": 16244
    },
    {
      "epoch": 0.5817680448367862,
      "grad_norm": 1.571876049041748,
      "learning_rate": 7.854182387083389e-05,
      "loss": 1.4038,
      "step": 16245
    },
    {
      "epoch": 0.5818038569663545,
      "grad_norm": 2.275881767272949,
      "learning_rate": 7.8530495189337e-05,
      "loss": 1.6678,
      "step": 16246
    },
    {
      "epoch": 0.5818396690959228,
      "grad_norm": 1.6172714233398438,
      "learning_rate": 7.851916679668454e-05,
      "loss": 1.4166,
      "step": 16247
    },
    {
      "epoch": 0.581875481225491,
      "grad_norm": 2.5062615871429443,
      "learning_rate": 7.850783869302905e-05,
      "loss": 1.5164,
      "step": 16248
    },
    {
      "epoch": 0.5819112933550593,
      "grad_norm": 2.113309144973755,
      "learning_rate": 7.849651087852278e-05,
      "loss": 1.3307,
      "step": 16249
    },
    {
      "epoch": 0.5819471054846277,
      "grad_norm": 1.644601821899414,
      "learning_rate": 7.848518335331832e-05,
      "loss": 1.6786,
      "step": 16250
    },
    {
      "epoch": 0.581982917614196,
      "grad_norm": 1.3407495021820068,
      "learning_rate": 7.847385611756788e-05,
      "loss": 1.343,
      "step": 16251
    },
    {
      "epoch": 0.5820187297437642,
      "grad_norm": 1.6835163831710815,
      "learning_rate": 7.8462529171424e-05,
      "loss": 1.1167,
      "step": 16252
    },
    {
      "epoch": 0.5820545418733325,
      "grad_norm": 1.790061593055725,
      "learning_rate": 7.845120251503896e-05,
      "loss": 1.8415,
      "step": 16253
    },
    {
      "epoch": 0.5820903540029008,
      "grad_norm": 1.642216444015503,
      "learning_rate": 7.843987614856525e-05,
      "loss": 1.3947,
      "step": 16254
    },
    {
      "epoch": 0.582126166132469,
      "grad_norm": 1.661774754524231,
      "learning_rate": 7.842855007215517e-05,
      "loss": 1.1908,
      "step": 16255
    },
    {
      "epoch": 0.5821619782620373,
      "grad_norm": 1.867827296257019,
      "learning_rate": 7.841722428596109e-05,
      "loss": 1.5475,
      "step": 16256
    },
    {
      "epoch": 0.5821977903916057,
      "grad_norm": 1.7840546369552612,
      "learning_rate": 7.840589879013548e-05,
      "loss": 1.6112,
      "step": 16257
    },
    {
      "epoch": 0.582233602521174,
      "grad_norm": 1.458702802658081,
      "learning_rate": 7.839457358483057e-05,
      "loss": 1.3471,
      "step": 16258
    },
    {
      "epoch": 0.5822694146507422,
      "grad_norm": 1.494171142578125,
      "learning_rate": 7.838324867019888e-05,
      "loss": 1.2515,
      "step": 16259
    },
    {
      "epoch": 0.5823052267803105,
      "grad_norm": 1.5532360076904297,
      "learning_rate": 7.837192404639264e-05,
      "loss": 1.4255,
      "step": 16260
    },
    {
      "epoch": 0.5823410389098788,
      "grad_norm": 1.6035205125808716,
      "learning_rate": 7.83605997135643e-05,
      "loss": 1.502,
      "step": 16261
    },
    {
      "epoch": 0.582376851039447,
      "grad_norm": 1.7551525831222534,
      "learning_rate": 7.834927567186614e-05,
      "loss": 1.3939,
      "step": 16262
    },
    {
      "epoch": 0.5824126631690153,
      "grad_norm": 1.5942054986953735,
      "learning_rate": 7.833795192145062e-05,
      "loss": 1.4372,
      "step": 16263
    },
    {
      "epoch": 0.5824484752985837,
      "grad_norm": 1.804587960243225,
      "learning_rate": 7.832662846246997e-05,
      "loss": 1.4715,
      "step": 16264
    },
    {
      "epoch": 0.582484287428152,
      "grad_norm": 1.6673866510391235,
      "learning_rate": 7.831530529507656e-05,
      "loss": 1.4346,
      "step": 16265
    },
    {
      "epoch": 0.5825200995577202,
      "grad_norm": 1.700072169303894,
      "learning_rate": 7.830398241942278e-05,
      "loss": 1.438,
      "step": 16266
    },
    {
      "epoch": 0.5825559116872885,
      "grad_norm": 1.6875628232955933,
      "learning_rate": 7.829265983566088e-05,
      "loss": 1.4232,
      "step": 16267
    },
    {
      "epoch": 0.5825917238168568,
      "grad_norm": 1.3997061252593994,
      "learning_rate": 7.82813375439433e-05,
      "loss": 1.5627,
      "step": 16268
    },
    {
      "epoch": 0.582627535946425,
      "grad_norm": 2.5741705894470215,
      "learning_rate": 7.827001554442224e-05,
      "loss": 1.5578,
      "step": 16269
    },
    {
      "epoch": 0.5826633480759933,
      "grad_norm": 1.8426802158355713,
      "learning_rate": 7.825869383725017e-05,
      "loss": 1.4258,
      "step": 16270
    },
    {
      "epoch": 0.5826991602055617,
      "grad_norm": 1.2935112714767456,
      "learning_rate": 7.824737242257925e-05,
      "loss": 1.4715,
      "step": 16271
    },
    {
      "epoch": 0.5827349723351299,
      "grad_norm": 1.767466425895691,
      "learning_rate": 7.823605130056196e-05,
      "loss": 1.5631,
      "step": 16272
    },
    {
      "epoch": 0.5827707844646982,
      "grad_norm": 2.372025489807129,
      "learning_rate": 7.822473047135048e-05,
      "loss": 1.4195,
      "step": 16273
    },
    {
      "epoch": 0.5828065965942665,
      "grad_norm": 1.9583728313446045,
      "learning_rate": 7.821340993509716e-05,
      "loss": 1.5143,
      "step": 16274
    },
    {
      "epoch": 0.5828424087238347,
      "grad_norm": 1.539113163948059,
      "learning_rate": 7.820208969195432e-05,
      "loss": 1.4506,
      "step": 16275
    },
    {
      "epoch": 0.582878220853403,
      "grad_norm": 1.511946201324463,
      "learning_rate": 7.819076974207425e-05,
      "loss": 1.7248,
      "step": 16276
    },
    {
      "epoch": 0.5829140329829713,
      "grad_norm": 1.7202016115188599,
      "learning_rate": 7.817945008560923e-05,
      "loss": 1.5891,
      "step": 16277
    },
    {
      "epoch": 0.5829498451125397,
      "grad_norm": 1.5539473295211792,
      "learning_rate": 7.816813072271155e-05,
      "loss": 1.2524,
      "step": 16278
    },
    {
      "epoch": 0.5829856572421079,
      "grad_norm": 1.6296128034591675,
      "learning_rate": 7.815681165353353e-05,
      "loss": 1.3194,
      "step": 16279
    },
    {
      "epoch": 0.5830214693716762,
      "grad_norm": 1.5075005292892456,
      "learning_rate": 7.814549287822743e-05,
      "loss": 1.8098,
      "step": 16280
    },
    {
      "epoch": 0.5830572815012445,
      "grad_norm": 2.4536232948303223,
      "learning_rate": 7.813417439694553e-05,
      "loss": 1.7013,
      "step": 16281
    },
    {
      "epoch": 0.5830930936308127,
      "grad_norm": 1.7020354270935059,
      "learning_rate": 7.812285620984012e-05,
      "loss": 1.2331,
      "step": 16282
    },
    {
      "epoch": 0.583128905760381,
      "grad_norm": 1.890274167060852,
      "learning_rate": 7.811153831706344e-05,
      "loss": 1.5311,
      "step": 16283
    },
    {
      "epoch": 0.5831647178899493,
      "grad_norm": 1.4711588621139526,
      "learning_rate": 7.81002207187678e-05,
      "loss": 1.5288,
      "step": 16284
    },
    {
      "epoch": 0.5832005300195177,
      "grad_norm": 1.563541293144226,
      "learning_rate": 7.808890341510542e-05,
      "loss": 1.3513,
      "step": 16285
    },
    {
      "epoch": 0.5832363421490859,
      "grad_norm": 2.1257882118225098,
      "learning_rate": 7.80775864062286e-05,
      "loss": 1.2747,
      "step": 16286
    },
    {
      "epoch": 0.5832721542786542,
      "grad_norm": 1.6666767597198486,
      "learning_rate": 7.806626969228955e-05,
      "loss": 1.256,
      "step": 16287
    },
    {
      "epoch": 0.5833079664082225,
      "grad_norm": 1.7806174755096436,
      "learning_rate": 7.805495327344058e-05,
      "loss": 1.6864,
      "step": 16288
    },
    {
      "epoch": 0.5833437785377907,
      "grad_norm": 1.5467238426208496,
      "learning_rate": 7.804363714983387e-05,
      "loss": 1.2907,
      "step": 16289
    },
    {
      "epoch": 0.583379590667359,
      "grad_norm": 1.4566665887832642,
      "learning_rate": 7.803232132162174e-05,
      "loss": 1.5714,
      "step": 16290
    },
    {
      "epoch": 0.5834154027969273,
      "grad_norm": 1.4355965852737427,
      "learning_rate": 7.802100578895638e-05,
      "loss": 1.507,
      "step": 16291
    },
    {
      "epoch": 0.5834512149264957,
      "grad_norm": 1.7385189533233643,
      "learning_rate": 7.800969055199003e-05,
      "loss": 1.5291,
      "step": 16292
    },
    {
      "epoch": 0.5834870270560639,
      "grad_norm": 1.3698575496673584,
      "learning_rate": 7.799837561087493e-05,
      "loss": 1.3751,
      "step": 16293
    },
    {
      "epoch": 0.5835228391856322,
      "grad_norm": 1.8529399633407593,
      "learning_rate": 7.798706096576329e-05,
      "loss": 1.5273,
      "step": 16294
    },
    {
      "epoch": 0.5835586513152005,
      "grad_norm": 1.3171405792236328,
      "learning_rate": 7.797574661680737e-05,
      "loss": 1.0538,
      "step": 16295
    },
    {
      "epoch": 0.5835944634447687,
      "grad_norm": 1.4912500381469727,
      "learning_rate": 7.796443256415935e-05,
      "loss": 1.6792,
      "step": 16296
    },
    {
      "epoch": 0.583630275574337,
      "grad_norm": 1.7634475231170654,
      "learning_rate": 7.79531188079715e-05,
      "loss": 1.4089,
      "step": 16297
    },
    {
      "epoch": 0.5836660877039053,
      "grad_norm": 2.007218599319458,
      "learning_rate": 7.794180534839597e-05,
      "loss": 1.2288,
      "step": 16298
    },
    {
      "epoch": 0.5837018998334736,
      "grad_norm": 2.622135877609253,
      "learning_rate": 7.793049218558501e-05,
      "loss": 1.4607,
      "step": 16299
    },
    {
      "epoch": 0.5837377119630419,
      "grad_norm": 1.8438994884490967,
      "learning_rate": 7.791917931969082e-05,
      "loss": 1.6507,
      "step": 16300
    },
    {
      "epoch": 0.5837735240926102,
      "grad_norm": 1.5510149002075195,
      "learning_rate": 7.790786675086555e-05,
      "loss": 1.5637,
      "step": 16301
    },
    {
      "epoch": 0.5838093362221785,
      "grad_norm": 1.6385968923568726,
      "learning_rate": 7.789655447926147e-05,
      "loss": 1.5741,
      "step": 16302
    },
    {
      "epoch": 0.5838451483517467,
      "grad_norm": 1.8733488321304321,
      "learning_rate": 7.788524250503072e-05,
      "loss": 1.6931,
      "step": 16303
    },
    {
      "epoch": 0.583880960481315,
      "grad_norm": 1.5182271003723145,
      "learning_rate": 7.787393082832553e-05,
      "loss": 1.5635,
      "step": 16304
    },
    {
      "epoch": 0.5839167726108833,
      "grad_norm": 1.9357662200927734,
      "learning_rate": 7.786261944929803e-05,
      "loss": 1.447,
      "step": 16305
    },
    {
      "epoch": 0.5839525847404516,
      "grad_norm": 3.7043070793151855,
      "learning_rate": 7.785130836810045e-05,
      "loss": 1.4161,
      "step": 16306
    },
    {
      "epoch": 0.5839883968700199,
      "grad_norm": 1.9308794736862183,
      "learning_rate": 7.783999758488492e-05,
      "loss": 1.3324,
      "step": 16307
    },
    {
      "epoch": 0.5840242089995882,
      "grad_norm": 1.5649685859680176,
      "learning_rate": 7.782868709980368e-05,
      "loss": 1.5649,
      "step": 16308
    },
    {
      "epoch": 0.5840600211291564,
      "grad_norm": 1.5657800436019897,
      "learning_rate": 7.781737691300884e-05,
      "loss": 1.4936,
      "step": 16309
    },
    {
      "epoch": 0.5840958332587247,
      "grad_norm": 1.9875231981277466,
      "learning_rate": 7.780606702465256e-05,
      "loss": 1.534,
      "step": 16310
    },
    {
      "epoch": 0.584131645388293,
      "grad_norm": 1.580452799797058,
      "learning_rate": 7.779475743488705e-05,
      "loss": 1.5182,
      "step": 16311
    },
    {
      "epoch": 0.5841674575178613,
      "grad_norm": 1.5643384456634521,
      "learning_rate": 7.778344814386441e-05,
      "loss": 1.3858,
      "step": 16312
    },
    {
      "epoch": 0.5842032696474296,
      "grad_norm": 1.5773082971572876,
      "learning_rate": 7.777213915173685e-05,
      "loss": 1.5475,
      "step": 16313
    },
    {
      "epoch": 0.5842390817769979,
      "grad_norm": 1.6503983736038208,
      "learning_rate": 7.776083045865645e-05,
      "loss": 1.3458,
      "step": 16314
    },
    {
      "epoch": 0.5842748939065662,
      "grad_norm": 2.040325403213501,
      "learning_rate": 7.774952206477542e-05,
      "loss": 1.533,
      "step": 16315
    },
    {
      "epoch": 0.5843107060361344,
      "grad_norm": 1.8997361660003662,
      "learning_rate": 7.773821397024584e-05,
      "loss": 1.6031,
      "step": 16316
    },
    {
      "epoch": 0.5843465181657027,
      "grad_norm": 1.708116054534912,
      "learning_rate": 7.77269061752199e-05,
      "loss": 1.4829,
      "step": 16317
    },
    {
      "epoch": 0.584382330295271,
      "grad_norm": 1.731094241142273,
      "learning_rate": 7.77155986798497e-05,
      "loss": 1.5349,
      "step": 16318
    },
    {
      "epoch": 0.5844181424248392,
      "grad_norm": 1.470812439918518,
      "learning_rate": 7.770429148428736e-05,
      "loss": 1.3997,
      "step": 16319
    },
    {
      "epoch": 0.5844539545544076,
      "grad_norm": 1.660354495048523,
      "learning_rate": 7.769298458868504e-05,
      "loss": 1.2777,
      "step": 16320
    },
    {
      "epoch": 0.5844897666839759,
      "grad_norm": 1.6110458374023438,
      "learning_rate": 7.768167799319481e-05,
      "loss": 1.3919,
      "step": 16321
    },
    {
      "epoch": 0.5845255788135442,
      "grad_norm": 1.8067439794540405,
      "learning_rate": 7.767037169796885e-05,
      "loss": 1.4326,
      "step": 16322
    },
    {
      "epoch": 0.5845613909431124,
      "grad_norm": 1.5376381874084473,
      "learning_rate": 7.76590657031592e-05,
      "loss": 1.5476,
      "step": 16323
    },
    {
      "epoch": 0.5845972030726807,
      "grad_norm": 2.144806385040283,
      "learning_rate": 7.764776000891805e-05,
      "loss": 1.2215,
      "step": 16324
    },
    {
      "epoch": 0.584633015202249,
      "grad_norm": 1.5610839128494263,
      "learning_rate": 7.763645461539741e-05,
      "loss": 1.0606,
      "step": 16325
    },
    {
      "epoch": 0.5846688273318172,
      "grad_norm": 1.6355888843536377,
      "learning_rate": 7.762514952274945e-05,
      "loss": 1.3254,
      "step": 16326
    },
    {
      "epoch": 0.5847046394613856,
      "grad_norm": 2.3415260314941406,
      "learning_rate": 7.761384473112625e-05,
      "loss": 1.5028,
      "step": 16327
    },
    {
      "epoch": 0.5847404515909539,
      "grad_norm": 1.3841601610183716,
      "learning_rate": 7.760254024067986e-05,
      "loss": 1.3328,
      "step": 16328
    },
    {
      "epoch": 0.5847762637205222,
      "grad_norm": 2.1620593070983887,
      "learning_rate": 7.759123605156243e-05,
      "loss": 1.6989,
      "step": 16329
    },
    {
      "epoch": 0.5848120758500904,
      "grad_norm": 1.456923007965088,
      "learning_rate": 7.757993216392599e-05,
      "loss": 1.5152,
      "step": 16330
    },
    {
      "epoch": 0.5848478879796587,
      "grad_norm": 1.7215697765350342,
      "learning_rate": 7.756862857792268e-05,
      "loss": 1.7743,
      "step": 16331
    },
    {
      "epoch": 0.584883700109227,
      "grad_norm": 1.5496634244918823,
      "learning_rate": 7.755732529370449e-05,
      "loss": 1.5552,
      "step": 16332
    },
    {
      "epoch": 0.5849195122387952,
      "grad_norm": 1.8300232887268066,
      "learning_rate": 7.754602231142359e-05,
      "loss": 1.7157,
      "step": 16333
    },
    {
      "epoch": 0.5849553243683636,
      "grad_norm": 1.5172995328903198,
      "learning_rate": 7.753471963123196e-05,
      "loss": 1.148,
      "step": 16334
    },
    {
      "epoch": 0.5849911364979319,
      "grad_norm": 1.6155294179916382,
      "learning_rate": 7.752341725328171e-05,
      "loss": 1.2817,
      "step": 16335
    },
    {
      "epoch": 0.5850269486275002,
      "grad_norm": 1.5804482698440552,
      "learning_rate": 7.751211517772491e-05,
      "loss": 1.3887,
      "step": 16336
    },
    {
      "epoch": 0.5850627607570684,
      "grad_norm": 1.509569525718689,
      "learning_rate": 7.750081340471355e-05,
      "loss": 1.3041,
      "step": 16337
    },
    {
      "epoch": 0.5850985728866367,
      "grad_norm": 1.4579132795333862,
      "learning_rate": 7.748951193439977e-05,
      "loss": 1.5923,
      "step": 16338
    },
    {
      "epoch": 0.585134385016205,
      "grad_norm": 1.461970567703247,
      "learning_rate": 7.747821076693551e-05,
      "loss": 1.3511,
      "step": 16339
    },
    {
      "epoch": 0.5851701971457732,
      "grad_norm": 1.2731819152832031,
      "learning_rate": 7.746690990247291e-05,
      "loss": 1.4716,
      "step": 16340
    },
    {
      "epoch": 0.5852060092753416,
      "grad_norm": 1.532819151878357,
      "learning_rate": 7.745560934116398e-05,
      "loss": 1.3444,
      "step": 16341
    },
    {
      "epoch": 0.5852418214049099,
      "grad_norm": 1.585973858833313,
      "learning_rate": 7.744430908316074e-05,
      "loss": 1.5642,
      "step": 16342
    },
    {
      "epoch": 0.5852776335344781,
      "grad_norm": 1.9086010456085205,
      "learning_rate": 7.743300912861525e-05,
      "loss": 1.3634,
      "step": 16343
    },
    {
      "epoch": 0.5853134456640464,
      "grad_norm": 2.089420795440674,
      "learning_rate": 7.742170947767945e-05,
      "loss": 1.5245,
      "step": 16344
    },
    {
      "epoch": 0.5853492577936147,
      "grad_norm": 1.5326855182647705,
      "learning_rate": 7.741041013050549e-05,
      "loss": 1.4089,
      "step": 16345
    },
    {
      "epoch": 0.585385069923183,
      "grad_norm": 1.6081247329711914,
      "learning_rate": 7.739911108724527e-05,
      "loss": 1.3786,
      "step": 16346
    },
    {
      "epoch": 0.5854208820527512,
      "grad_norm": 1.4763007164001465,
      "learning_rate": 7.73878123480509e-05,
      "loss": 1.3,
      "step": 16347
    },
    {
      "epoch": 0.5854566941823196,
      "grad_norm": 1.4239866733551025,
      "learning_rate": 7.73765139130743e-05,
      "loss": 1.4136,
      "step": 16348
    },
    {
      "epoch": 0.5854925063118879,
      "grad_norm": 1.4767128229141235,
      "learning_rate": 7.736521578246758e-05,
      "loss": 1.6589,
      "step": 16349
    },
    {
      "epoch": 0.5855283184414561,
      "grad_norm": 2.082230806350708,
      "learning_rate": 7.735391795638262e-05,
      "loss": 1.2585,
      "step": 16350
    },
    {
      "epoch": 0.5855641305710244,
      "grad_norm": 1.4269074201583862,
      "learning_rate": 7.734262043497155e-05,
      "loss": 1.7076,
      "step": 16351
    },
    {
      "epoch": 0.5855999427005927,
      "grad_norm": 1.9690130949020386,
      "learning_rate": 7.733132321838628e-05,
      "loss": 1.4339,
      "step": 16352
    },
    {
      "epoch": 0.585635754830161,
      "grad_norm": 1.7755242586135864,
      "learning_rate": 7.732002630677878e-05,
      "loss": 1.3142,
      "step": 16353
    },
    {
      "epoch": 0.5856715669597292,
      "grad_norm": 1.3469429016113281,
      "learning_rate": 7.73087297003011e-05,
      "loss": 1.725,
      "step": 16354
    },
    {
      "epoch": 0.5857073790892976,
      "grad_norm": 1.6812926530838013,
      "learning_rate": 7.729743339910515e-05,
      "loss": 1.1344,
      "step": 16355
    },
    {
      "epoch": 0.5857431912188659,
      "grad_norm": 1.4505376815795898,
      "learning_rate": 7.728613740334304e-05,
      "loss": 1.3823,
      "step": 16356
    },
    {
      "epoch": 0.5857790033484341,
      "grad_norm": 1.8146635293960571,
      "learning_rate": 7.727484171316655e-05,
      "loss": 1.3706,
      "step": 16357
    },
    {
      "epoch": 0.5858148154780024,
      "grad_norm": 1.5328073501586914,
      "learning_rate": 7.726354632872783e-05,
      "loss": 1.5141,
      "step": 16358
    },
    {
      "epoch": 0.5858506276075707,
      "grad_norm": 1.5390815734863281,
      "learning_rate": 7.72522512501787e-05,
      "loss": 1.3987,
      "step": 16359
    },
    {
      "epoch": 0.5858864397371389,
      "grad_norm": 1.7571913003921509,
      "learning_rate": 7.724095647767125e-05,
      "loss": 1.7021,
      "step": 16360
    },
    {
      "epoch": 0.5859222518667072,
      "grad_norm": 1.810835361480713,
      "learning_rate": 7.722966201135736e-05,
      "loss": 1.5146,
      "step": 16361
    },
    {
      "epoch": 0.5859580639962756,
      "grad_norm": 1.467986822128296,
      "learning_rate": 7.721836785138896e-05,
      "loss": 1.0774,
      "step": 16362
    },
    {
      "epoch": 0.5859938761258439,
      "grad_norm": 1.3759533166885376,
      "learning_rate": 7.720707399791807e-05,
      "loss": 1.5041,
      "step": 16363
    },
    {
      "epoch": 0.5860296882554121,
      "grad_norm": 1.5349520444869995,
      "learning_rate": 7.719578045109657e-05,
      "loss": 1.4431,
      "step": 16364
    },
    {
      "epoch": 0.5860655003849804,
      "grad_norm": 1.4117509126663208,
      "learning_rate": 7.718448721107645e-05,
      "loss": 1.1674,
      "step": 16365
    },
    {
      "epoch": 0.5861013125145487,
      "grad_norm": 1.8245066404342651,
      "learning_rate": 7.717319427800957e-05,
      "loss": 1.4755,
      "step": 16366
    },
    {
      "epoch": 0.5861371246441169,
      "grad_norm": 1.6505935192108154,
      "learning_rate": 7.7161901652048e-05,
      "loss": 1.2912,
      "step": 16367
    },
    {
      "epoch": 0.5861729367736852,
      "grad_norm": 1.3770312070846558,
      "learning_rate": 7.715060933334351e-05,
      "loss": 1.3058,
      "step": 16368
    },
    {
      "epoch": 0.5862087489032536,
      "grad_norm": 2.2501673698425293,
      "learning_rate": 7.713931732204816e-05,
      "loss": 1.6156,
      "step": 16369
    },
    {
      "epoch": 0.5862445610328219,
      "grad_norm": 1.9228209257125854,
      "learning_rate": 7.712802561831381e-05,
      "loss": 1.5591,
      "step": 16370
    },
    {
      "epoch": 0.5862803731623901,
      "grad_norm": 1.1532500982284546,
      "learning_rate": 7.711673422229232e-05,
      "loss": 1.4264,
      "step": 16371
    },
    {
      "epoch": 0.5863161852919584,
      "grad_norm": 1.937796711921692,
      "learning_rate": 7.71054431341357e-05,
      "loss": 1.4744,
      "step": 16372
    },
    {
      "epoch": 0.5863519974215267,
      "grad_norm": 1.4498701095581055,
      "learning_rate": 7.709415235399577e-05,
      "loss": 1.4989,
      "step": 16373
    },
    {
      "epoch": 0.5863878095510949,
      "grad_norm": 1.9454270601272583,
      "learning_rate": 7.708286188202451e-05,
      "loss": 1.521,
      "step": 16374
    },
    {
      "epoch": 0.5864236216806632,
      "grad_norm": 1.8580267429351807,
      "learning_rate": 7.707157171837374e-05,
      "loss": 1.3413,
      "step": 16375
    },
    {
      "epoch": 0.5864594338102316,
      "grad_norm": 2.0274364948272705,
      "learning_rate": 7.706028186319543e-05,
      "loss": 1.4014,
      "step": 16376
    },
    {
      "epoch": 0.5864952459397998,
      "grad_norm": 1.5042753219604492,
      "learning_rate": 7.704899231664143e-05,
      "loss": 1.5814,
      "step": 16377
    },
    {
      "epoch": 0.5865310580693681,
      "grad_norm": 1.5244011878967285,
      "learning_rate": 7.703770307886364e-05,
      "loss": 1.1341,
      "step": 16378
    },
    {
      "epoch": 0.5865668701989364,
      "grad_norm": 2.079533576965332,
      "learning_rate": 7.702641415001394e-05,
      "loss": 1.9471,
      "step": 16379
    },
    {
      "epoch": 0.5866026823285047,
      "grad_norm": 1.7344539165496826,
      "learning_rate": 7.701512553024418e-05,
      "loss": 1.3302,
      "step": 16380
    },
    {
      "epoch": 0.5866384944580729,
      "grad_norm": 1.6465486288070679,
      "learning_rate": 7.700383721970628e-05,
      "loss": 1.3835,
      "step": 16381
    },
    {
      "epoch": 0.5866743065876412,
      "grad_norm": 2.1004645824432373,
      "learning_rate": 7.699254921855206e-05,
      "loss": 1.6895,
      "step": 16382
    },
    {
      "epoch": 0.5867101187172096,
      "grad_norm": 1.8558729887008667,
      "learning_rate": 7.698126152693345e-05,
      "loss": 1.4859,
      "step": 16383
    },
    {
      "epoch": 0.5867459308467778,
      "grad_norm": 1.7315419912338257,
      "learning_rate": 7.696997414500223e-05,
      "loss": 1.3328,
      "step": 16384
    },
    {
      "epoch": 0.5867817429763461,
      "grad_norm": 2.36985182762146,
      "learning_rate": 7.695868707291034e-05,
      "loss": 1.4141,
      "step": 16385
    },
    {
      "epoch": 0.5868175551059144,
      "grad_norm": 1.4058644771575928,
      "learning_rate": 7.694740031080957e-05,
      "loss": 1.2073,
      "step": 16386
    },
    {
      "epoch": 0.5868533672354826,
      "grad_norm": 1.6645869016647339,
      "learning_rate": 7.693611385885181e-05,
      "loss": 1.488,
      "step": 16387
    },
    {
      "epoch": 0.5868891793650509,
      "grad_norm": 1.4396991729736328,
      "learning_rate": 7.69248277171889e-05,
      "loss": 1.5952,
      "step": 16388
    },
    {
      "epoch": 0.5869249914946192,
      "grad_norm": 2.9401330947875977,
      "learning_rate": 7.691354188597263e-05,
      "loss": 1.8068,
      "step": 16389
    },
    {
      "epoch": 0.5869608036241876,
      "grad_norm": 1.3495395183563232,
      "learning_rate": 7.69022563653549e-05,
      "loss": 1.2889,
      "step": 16390
    },
    {
      "epoch": 0.5869966157537558,
      "grad_norm": 1.5951523780822754,
      "learning_rate": 7.689097115548751e-05,
      "loss": 1.5938,
      "step": 16391
    },
    {
      "epoch": 0.5870324278833241,
      "grad_norm": 1.477667212486267,
      "learning_rate": 7.68796862565223e-05,
      "loss": 1.5373,
      "step": 16392
    },
    {
      "epoch": 0.5870682400128924,
      "grad_norm": 1.5856584310531616,
      "learning_rate": 7.686840166861106e-05,
      "loss": 1.3693,
      "step": 16393
    },
    {
      "epoch": 0.5871040521424606,
      "grad_norm": 1.5418838262557983,
      "learning_rate": 7.685711739190568e-05,
      "loss": 1.4448,
      "step": 16394
    },
    {
      "epoch": 0.5871398642720289,
      "grad_norm": 1.6027543544769287,
      "learning_rate": 7.684583342655791e-05,
      "loss": 1.5726,
      "step": 16395
    },
    {
      "epoch": 0.5871756764015972,
      "grad_norm": 1.5729893445968628,
      "learning_rate": 7.68345497727196e-05,
      "loss": 1.5488,
      "step": 16396
    },
    {
      "epoch": 0.5872114885311656,
      "grad_norm": 1.2920430898666382,
      "learning_rate": 7.682326643054254e-05,
      "loss": 1.663,
      "step": 16397
    },
    {
      "epoch": 0.5872473006607338,
      "grad_norm": 1.6387059688568115,
      "learning_rate": 7.681198340017852e-05,
      "loss": 1.3222,
      "step": 16398
    },
    {
      "epoch": 0.5872831127903021,
      "grad_norm": 1.7342307567596436,
      "learning_rate": 7.680070068177936e-05,
      "loss": 1.3333,
      "step": 16399
    },
    {
      "epoch": 0.5873189249198704,
      "grad_norm": 1.7177081108093262,
      "learning_rate": 7.678941827549683e-05,
      "loss": 1.5123,
      "step": 16400
    },
    {
      "epoch": 0.5873547370494386,
      "grad_norm": 1.5641347169876099,
      "learning_rate": 7.677813618148276e-05,
      "loss": 1.479,
      "step": 16401
    },
    {
      "epoch": 0.5873905491790069,
      "grad_norm": 1.8655952215194702,
      "learning_rate": 7.67668543998889e-05,
      "loss": 1.6996,
      "step": 16402
    },
    {
      "epoch": 0.5874263613085752,
      "grad_norm": 1.3110804557800293,
      "learning_rate": 7.675557293086706e-05,
      "loss": 1.4526,
      "step": 16403
    },
    {
      "epoch": 0.5874621734381436,
      "grad_norm": 1.8281629085540771,
      "learning_rate": 7.674429177456899e-05,
      "loss": 1.6103,
      "step": 16404
    },
    {
      "epoch": 0.5874979855677118,
      "grad_norm": 1.439719557762146,
      "learning_rate": 7.673301093114649e-05,
      "loss": 1.2931,
      "step": 16405
    },
    {
      "epoch": 0.5875337976972801,
      "grad_norm": 1.3842743635177612,
      "learning_rate": 7.672173040075131e-05,
      "loss": 1.3253,
      "step": 16406
    },
    {
      "epoch": 0.5875696098268484,
      "grad_norm": 1.4752776622772217,
      "learning_rate": 7.671045018353521e-05,
      "loss": 1.2556,
      "step": 16407
    },
    {
      "epoch": 0.5876054219564166,
      "grad_norm": 1.8519983291625977,
      "learning_rate": 7.669917027964998e-05,
      "loss": 1.3942,
      "step": 16408
    },
    {
      "epoch": 0.5876412340859849,
      "grad_norm": 1.96647047996521,
      "learning_rate": 7.668789068924734e-05,
      "loss": 1.4563,
      "step": 16409
    },
    {
      "epoch": 0.5876770462155532,
      "grad_norm": 1.5094881057739258,
      "learning_rate": 7.667661141247907e-05,
      "loss": 1.2547,
      "step": 16410
    },
    {
      "epoch": 0.5877128583451215,
      "grad_norm": 1.5891088247299194,
      "learning_rate": 7.66653324494969e-05,
      "loss": 1.5648,
      "step": 16411
    },
    {
      "epoch": 0.5877486704746898,
      "grad_norm": 1.6747032403945923,
      "learning_rate": 7.665405380045258e-05,
      "loss": 1.2975,
      "step": 16412
    },
    {
      "epoch": 0.5877844826042581,
      "grad_norm": 1.7869306802749634,
      "learning_rate": 7.664277546549786e-05,
      "loss": 1.7527,
      "step": 16413
    },
    {
      "epoch": 0.5878202947338264,
      "grad_norm": 2.474652051925659,
      "learning_rate": 7.663149744478448e-05,
      "loss": 1.7304,
      "step": 16414
    },
    {
      "epoch": 0.5878561068633946,
      "grad_norm": 1.5477724075317383,
      "learning_rate": 7.662021973846415e-05,
      "loss": 1.2809,
      "step": 16415
    },
    {
      "epoch": 0.5878919189929629,
      "grad_norm": 1.887107014656067,
      "learning_rate": 7.660894234668859e-05,
      "loss": 1.5836,
      "step": 16416
    },
    {
      "epoch": 0.5879277311225312,
      "grad_norm": 1.466249942779541,
      "learning_rate": 7.659766526960957e-05,
      "loss": 1.607,
      "step": 16417
    },
    {
      "epoch": 0.5879635432520995,
      "grad_norm": 1.4023653268814087,
      "learning_rate": 7.658638850737874e-05,
      "loss": 1.0977,
      "step": 16418
    },
    {
      "epoch": 0.5879993553816678,
      "grad_norm": 1.5768914222717285,
      "learning_rate": 7.657511206014788e-05,
      "loss": 1.587,
      "step": 16419
    },
    {
      "epoch": 0.5880351675112361,
      "grad_norm": 1.5604252815246582,
      "learning_rate": 7.656383592806865e-05,
      "loss": 1.4809,
      "step": 16420
    },
    {
      "epoch": 0.5880709796408043,
      "grad_norm": 1.8900164365768433,
      "learning_rate": 7.655256011129279e-05,
      "loss": 1.368,
      "step": 16421
    },
    {
      "epoch": 0.5881067917703726,
      "grad_norm": 1.759080171585083,
      "learning_rate": 7.654128460997198e-05,
      "loss": 1.4159,
      "step": 16422
    },
    {
      "epoch": 0.5881426038999409,
      "grad_norm": 2.4119346141815186,
      "learning_rate": 7.653000942425794e-05,
      "loss": 1.3701,
      "step": 16423
    },
    {
      "epoch": 0.5881784160295092,
      "grad_norm": 1.836087703704834,
      "learning_rate": 7.651873455430237e-05,
      "loss": 1.5565,
      "step": 16424
    },
    {
      "epoch": 0.5882142281590775,
      "grad_norm": 1.5751034021377563,
      "learning_rate": 7.650746000025688e-05,
      "loss": 1.6149,
      "step": 16425
    },
    {
      "epoch": 0.5882500402886458,
      "grad_norm": 2.3002188205718994,
      "learning_rate": 7.649618576227325e-05,
      "loss": 1.5969,
      "step": 16426
    },
    {
      "epoch": 0.5882858524182141,
      "grad_norm": 1.5687283277511597,
      "learning_rate": 7.648491184050311e-05,
      "loss": 1.6116,
      "step": 16427
    },
    {
      "epoch": 0.5883216645477823,
      "grad_norm": 1.529663324356079,
      "learning_rate": 7.647363823509815e-05,
      "loss": 1.1357,
      "step": 16428
    },
    {
      "epoch": 0.5883574766773506,
      "grad_norm": 1.6907438039779663,
      "learning_rate": 7.646236494621004e-05,
      "loss": 1.4858,
      "step": 16429
    },
    {
      "epoch": 0.5883932888069189,
      "grad_norm": 1.6002248525619507,
      "learning_rate": 7.645109197399047e-05,
      "loss": 1.3588,
      "step": 16430
    },
    {
      "epoch": 0.5884291009364871,
      "grad_norm": 1.6536250114440918,
      "learning_rate": 7.643981931859104e-05,
      "loss": 1.396,
      "step": 16431
    },
    {
      "epoch": 0.5884649130660555,
      "grad_norm": 1.6143574714660645,
      "learning_rate": 7.642854698016348e-05,
      "loss": 1.3903,
      "step": 16432
    },
    {
      "epoch": 0.5885007251956238,
      "grad_norm": 1.4766873121261597,
      "learning_rate": 7.641727495885944e-05,
      "loss": 1.4478,
      "step": 16433
    },
    {
      "epoch": 0.5885365373251921,
      "grad_norm": 1.8612972497940063,
      "learning_rate": 7.640600325483049e-05,
      "loss": 1.4193,
      "step": 16434
    },
    {
      "epoch": 0.5885723494547603,
      "grad_norm": 1.6533550024032593,
      "learning_rate": 7.639473186822839e-05,
      "loss": 1.6117,
      "step": 16435
    },
    {
      "epoch": 0.5886081615843286,
      "grad_norm": 1.6703141927719116,
      "learning_rate": 7.638346079920466e-05,
      "loss": 1.782,
      "step": 16436
    },
    {
      "epoch": 0.5886439737138969,
      "grad_norm": 1.5920336246490479,
      "learning_rate": 7.637219004791106e-05,
      "loss": 1.3684,
      "step": 16437
    },
    {
      "epoch": 0.5886797858434651,
      "grad_norm": 1.2391998767852783,
      "learning_rate": 7.636091961449911e-05,
      "loss": 1.4898,
      "step": 16438
    },
    {
      "epoch": 0.5887155979730334,
      "grad_norm": 1.9100953340530396,
      "learning_rate": 7.634964949912054e-05,
      "loss": 1.7013,
      "step": 16439
    },
    {
      "epoch": 0.5887514101026018,
      "grad_norm": 2.002978563308716,
      "learning_rate": 7.633837970192694e-05,
      "loss": 1.6605,
      "step": 16440
    },
    {
      "epoch": 0.5887872222321701,
      "grad_norm": 2.9292209148406982,
      "learning_rate": 7.632711022306985e-05,
      "loss": 1.3659,
      "step": 16441
    },
    {
      "epoch": 0.5888230343617383,
      "grad_norm": 2.129926919937134,
      "learning_rate": 7.631584106270103e-05,
      "loss": 1.3669,
      "step": 16442
    },
    {
      "epoch": 0.5888588464913066,
      "grad_norm": 1.524235486984253,
      "learning_rate": 7.630457222097196e-05,
      "loss": 1.3326,
      "step": 16443
    },
    {
      "epoch": 0.5888946586208749,
      "grad_norm": 1.5463957786560059,
      "learning_rate": 7.629330369803435e-05,
      "loss": 1.3502,
      "step": 16444
    },
    {
      "epoch": 0.5889304707504431,
      "grad_norm": 1.6512547731399536,
      "learning_rate": 7.628203549403971e-05,
      "loss": 1.7101,
      "step": 16445
    },
    {
      "epoch": 0.5889662828800114,
      "grad_norm": 2.546053171157837,
      "learning_rate": 7.627076760913976e-05,
      "loss": 1.4441,
      "step": 16446
    },
    {
      "epoch": 0.5890020950095798,
      "grad_norm": 1.6109521389007568,
      "learning_rate": 7.625950004348595e-05,
      "loss": 1.4997,
      "step": 16447
    },
    {
      "epoch": 0.589037907139148,
      "grad_norm": 1.9896047115325928,
      "learning_rate": 7.624823279723001e-05,
      "loss": 1.3661,
      "step": 16448
    },
    {
      "epoch": 0.5890737192687163,
      "grad_norm": 1.843091368675232,
      "learning_rate": 7.623696587052343e-05,
      "loss": 1.5451,
      "step": 16449
    },
    {
      "epoch": 0.5891095313982846,
      "grad_norm": 2.0600764751434326,
      "learning_rate": 7.622569926351781e-05,
      "loss": 1.4128,
      "step": 16450
    },
    {
      "epoch": 0.5891453435278529,
      "grad_norm": 1.6426974534988403,
      "learning_rate": 7.621443297636478e-05,
      "loss": 1.5547,
      "step": 16451
    },
    {
      "epoch": 0.5891811556574211,
      "grad_norm": 1.6177315711975098,
      "learning_rate": 7.62031670092158e-05,
      "loss": 1.3688,
      "step": 16452
    },
    {
      "epoch": 0.5892169677869894,
      "grad_norm": 1.4145938158035278,
      "learning_rate": 7.619190136222259e-05,
      "loss": 1.3171,
      "step": 16453
    },
    {
      "epoch": 0.5892527799165578,
      "grad_norm": 2.2485506534576416,
      "learning_rate": 7.618063603553655e-05,
      "loss": 1.3094,
      "step": 16454
    },
    {
      "epoch": 0.589288592046126,
      "grad_norm": 1.2966976165771484,
      "learning_rate": 7.616937102930942e-05,
      "loss": 1.4065,
      "step": 16455
    },
    {
      "epoch": 0.5893244041756943,
      "grad_norm": 1.8887548446655273,
      "learning_rate": 7.61581063436926e-05,
      "loss": 1.3098,
      "step": 16456
    },
    {
      "epoch": 0.5893602163052626,
      "grad_norm": 1.525266408920288,
      "learning_rate": 7.614684197883775e-05,
      "loss": 1.3332,
      "step": 16457
    },
    {
      "epoch": 0.5893960284348309,
      "grad_norm": 2.2836077213287354,
      "learning_rate": 7.613557793489637e-05,
      "loss": 1.5326,
      "step": 16458
    },
    {
      "epoch": 0.5894318405643991,
      "grad_norm": 1.719794511795044,
      "learning_rate": 7.612431421201996e-05,
      "loss": 1.4174,
      "step": 16459
    },
    {
      "epoch": 0.5894676526939674,
      "grad_norm": 1.5337207317352295,
      "learning_rate": 7.611305081036015e-05,
      "loss": 1.1962,
      "step": 16460
    },
    {
      "epoch": 0.5895034648235358,
      "grad_norm": 1.6725982427597046,
      "learning_rate": 7.61017877300684e-05,
      "loss": 1.4057,
      "step": 16461
    },
    {
      "epoch": 0.589539276953104,
      "grad_norm": 1.944868564605713,
      "learning_rate": 7.609052497129629e-05,
      "loss": 1.4168,
      "step": 16462
    },
    {
      "epoch": 0.5895750890826723,
      "grad_norm": 1.736919641494751,
      "learning_rate": 7.607926253419531e-05,
      "loss": 1.619,
      "step": 16463
    },
    {
      "epoch": 0.5896109012122406,
      "grad_norm": 1.4023191928863525,
      "learning_rate": 7.606800041891701e-05,
      "loss": 1.5406,
      "step": 16464
    },
    {
      "epoch": 0.5896467133418088,
      "grad_norm": 2.086000919342041,
      "learning_rate": 7.605673862561284e-05,
      "loss": 1.5427,
      "step": 16465
    },
    {
      "epoch": 0.5896825254713771,
      "grad_norm": 1.7645149230957031,
      "learning_rate": 7.604547715443445e-05,
      "loss": 1.5254,
      "step": 16466
    },
    {
      "epoch": 0.5897183376009454,
      "grad_norm": 1.2208935022354126,
      "learning_rate": 7.603421600553324e-05,
      "loss": 1.5195,
      "step": 16467
    },
    {
      "epoch": 0.5897541497305138,
      "grad_norm": 1.5771421194076538,
      "learning_rate": 7.602295517906072e-05,
      "loss": 1.5837,
      "step": 16468
    },
    {
      "epoch": 0.589789961860082,
      "grad_norm": 1.4784822463989258,
      "learning_rate": 7.601169467516844e-05,
      "loss": 1.4173,
      "step": 16469
    },
    {
      "epoch": 0.5898257739896503,
      "grad_norm": 1.6251667737960815,
      "learning_rate": 7.600043449400782e-05,
      "loss": 1.6069,
      "step": 16470
    },
    {
      "epoch": 0.5898615861192186,
      "grad_norm": 1.9627031087875366,
      "learning_rate": 7.598917463573044e-05,
      "loss": 1.801,
      "step": 16471
    },
    {
      "epoch": 0.5898973982487868,
      "grad_norm": 2.1740000247955322,
      "learning_rate": 7.59779151004877e-05,
      "loss": 1.352,
      "step": 16472
    },
    {
      "epoch": 0.5899332103783551,
      "grad_norm": 1.7333546876907349,
      "learning_rate": 7.596665588843117e-05,
      "loss": 1.5806,
      "step": 16473
    },
    {
      "epoch": 0.5899690225079234,
      "grad_norm": 1.9666831493377686,
      "learning_rate": 7.595539699971225e-05,
      "loss": 1.4948,
      "step": 16474
    },
    {
      "epoch": 0.5900048346374918,
      "grad_norm": 1.692284345626831,
      "learning_rate": 7.594413843448248e-05,
      "loss": 1.4834,
      "step": 16475
    },
    {
      "epoch": 0.59004064676706,
      "grad_norm": 1.3781365156173706,
      "learning_rate": 7.593288019289329e-05,
      "loss": 1.3872,
      "step": 16476
    },
    {
      "epoch": 0.5900764588966283,
      "grad_norm": 1.7096667289733887,
      "learning_rate": 7.592162227509614e-05,
      "loss": 1.4666,
      "step": 16477
    },
    {
      "epoch": 0.5901122710261966,
      "grad_norm": 1.9078587293624878,
      "learning_rate": 7.591036468124252e-05,
      "loss": 1.5795,
      "step": 16478
    },
    {
      "epoch": 0.5901480831557648,
      "grad_norm": 1.6421924829483032,
      "learning_rate": 7.589910741148384e-05,
      "loss": 1.2502,
      "step": 16479
    },
    {
      "epoch": 0.5901838952853331,
      "grad_norm": 1.5889708995819092,
      "learning_rate": 7.588785046597161e-05,
      "loss": 1.4928,
      "step": 16480
    },
    {
      "epoch": 0.5902197074149014,
      "grad_norm": 2.91912841796875,
      "learning_rate": 7.587659384485723e-05,
      "loss": 1.848,
      "step": 16481
    },
    {
      "epoch": 0.5902555195444698,
      "grad_norm": 1.248953938484192,
      "learning_rate": 7.586533754829218e-05,
      "loss": 1.4012,
      "step": 16482
    },
    {
      "epoch": 0.590291331674038,
      "grad_norm": 1.8123250007629395,
      "learning_rate": 7.585408157642786e-05,
      "loss": 1.2366,
      "step": 16483
    },
    {
      "epoch": 0.5903271438036063,
      "grad_norm": 1.7613189220428467,
      "learning_rate": 7.584282592941574e-05,
      "loss": 1.1722,
      "step": 16484
    },
    {
      "epoch": 0.5903629559331746,
      "grad_norm": 1.6711210012435913,
      "learning_rate": 7.583157060740727e-05,
      "loss": 1.635,
      "step": 16485
    },
    {
      "epoch": 0.5903987680627428,
      "grad_norm": 1.593625783920288,
      "learning_rate": 7.582031561055378e-05,
      "loss": 1.5147,
      "step": 16486
    },
    {
      "epoch": 0.5904345801923111,
      "grad_norm": 1.475412130355835,
      "learning_rate": 7.58090609390068e-05,
      "loss": 1.5504,
      "step": 16487
    },
    {
      "epoch": 0.5904703923218794,
      "grad_norm": 2.0667693614959717,
      "learning_rate": 7.579780659291768e-05,
      "loss": 1.4428,
      "step": 16488
    },
    {
      "epoch": 0.5905062044514477,
      "grad_norm": 1.4998021125793457,
      "learning_rate": 7.578655257243786e-05,
      "loss": 1.5839,
      "step": 16489
    },
    {
      "epoch": 0.590542016581016,
      "grad_norm": 1.578344702720642,
      "learning_rate": 7.577529887771873e-05,
      "loss": 1.4212,
      "step": 16490
    },
    {
      "epoch": 0.5905778287105843,
      "grad_norm": 1.7631887197494507,
      "learning_rate": 7.576404550891172e-05,
      "loss": 1.5553,
      "step": 16491
    },
    {
      "epoch": 0.5906136408401526,
      "grad_norm": 1.7021032571792603,
      "learning_rate": 7.57527924661682e-05,
      "loss": 1.458,
      "step": 16492
    },
    {
      "epoch": 0.5906494529697208,
      "grad_norm": 1.9911140203475952,
      "learning_rate": 7.57415397496396e-05,
      "loss": 1.699,
      "step": 16493
    },
    {
      "epoch": 0.5906852650992891,
      "grad_norm": 1.4853942394256592,
      "learning_rate": 7.57302873594773e-05,
      "loss": 1.3604,
      "step": 16494
    },
    {
      "epoch": 0.5907210772288574,
      "grad_norm": 1.4939459562301636,
      "learning_rate": 7.571903529583265e-05,
      "loss": 1.4661,
      "step": 16495
    },
    {
      "epoch": 0.5907568893584257,
      "grad_norm": 1.3259398937225342,
      "learning_rate": 7.570778355885708e-05,
      "loss": 1.5328,
      "step": 16496
    },
    {
      "epoch": 0.590792701487994,
      "grad_norm": 1.5033601522445679,
      "learning_rate": 7.569653214870192e-05,
      "loss": 1.6475,
      "step": 16497
    },
    {
      "epoch": 0.5908285136175623,
      "grad_norm": 1.707831621170044,
      "learning_rate": 7.568528106551862e-05,
      "loss": 1.1923,
      "step": 16498
    },
    {
      "epoch": 0.5908643257471305,
      "grad_norm": 2.281423807144165,
      "learning_rate": 7.567403030945844e-05,
      "loss": 1.4786,
      "step": 16499
    },
    {
      "epoch": 0.5909001378766988,
      "grad_norm": 1.7174972295761108,
      "learning_rate": 7.566277988067285e-05,
      "loss": 1.4596,
      "step": 16500
    },
    {
      "epoch": 0.5909359500062671,
      "grad_norm": 1.3764506578445435,
      "learning_rate": 7.565152977931314e-05,
      "loss": 1.4684,
      "step": 16501
    },
    {
      "epoch": 0.5909717621358354,
      "grad_norm": 2.216895818710327,
      "learning_rate": 7.56402800055307e-05,
      "loss": 1.5763,
      "step": 16502
    },
    {
      "epoch": 0.5910075742654037,
      "grad_norm": 1.5239827632904053,
      "learning_rate": 7.562903055947688e-05,
      "loss": 1.2643,
      "step": 16503
    },
    {
      "epoch": 0.591043386394972,
      "grad_norm": 1.4618558883666992,
      "learning_rate": 7.561778144130299e-05,
      "loss": 1.4947,
      "step": 16504
    },
    {
      "epoch": 0.5910791985245403,
      "grad_norm": 1.6704903841018677,
      "learning_rate": 7.560653265116042e-05,
      "loss": 1.4775,
      "step": 16505
    },
    {
      "epoch": 0.5911150106541085,
      "grad_norm": 1.5482014417648315,
      "learning_rate": 7.559528418920048e-05,
      "loss": 1.3616,
      "step": 16506
    },
    {
      "epoch": 0.5911508227836768,
      "grad_norm": 1.870289921760559,
      "learning_rate": 7.558403605557453e-05,
      "loss": 1.4348,
      "step": 16507
    },
    {
      "epoch": 0.5911866349132451,
      "grad_norm": 1.4968332052230835,
      "learning_rate": 7.557278825043385e-05,
      "loss": 1.3695,
      "step": 16508
    },
    {
      "epoch": 0.5912224470428133,
      "grad_norm": 1.5539411306381226,
      "learning_rate": 7.556154077392982e-05,
      "loss": 1.3545,
      "step": 16509
    },
    {
      "epoch": 0.5912582591723817,
      "grad_norm": 1.5246566534042358,
      "learning_rate": 7.555029362621371e-05,
      "loss": 1.4596,
      "step": 16510
    },
    {
      "epoch": 0.59129407130195,
      "grad_norm": 1.3691539764404297,
      "learning_rate": 7.553904680743688e-05,
      "loss": 1.4245,
      "step": 16511
    },
    {
      "epoch": 0.5913298834315183,
      "grad_norm": 1.5674803256988525,
      "learning_rate": 7.552780031775064e-05,
      "loss": 1.3345,
      "step": 16512
    },
    {
      "epoch": 0.5913656955610865,
      "grad_norm": 1.5378525257110596,
      "learning_rate": 7.551655415730624e-05,
      "loss": 1.5885,
      "step": 16513
    },
    {
      "epoch": 0.5914015076906548,
      "grad_norm": 1.648198127746582,
      "learning_rate": 7.550530832625505e-05,
      "loss": 1.5619,
      "step": 16514
    },
    {
      "epoch": 0.5914373198202231,
      "grad_norm": 2.732581615447998,
      "learning_rate": 7.549406282474833e-05,
      "loss": 2.0635,
      "step": 16515
    },
    {
      "epoch": 0.5914731319497913,
      "grad_norm": 1.5190062522888184,
      "learning_rate": 7.548281765293739e-05,
      "loss": 1.583,
      "step": 16516
    },
    {
      "epoch": 0.5915089440793597,
      "grad_norm": 1.6483665704727173,
      "learning_rate": 7.54715728109735e-05,
      "loss": 1.2961,
      "step": 16517
    },
    {
      "epoch": 0.591544756208928,
      "grad_norm": 1.692216396331787,
      "learning_rate": 7.546032829900797e-05,
      "loss": 1.451,
      "step": 16518
    },
    {
      "epoch": 0.5915805683384963,
      "grad_norm": 1.4392499923706055,
      "learning_rate": 7.544908411719207e-05,
      "loss": 1.3544,
      "step": 16519
    },
    {
      "epoch": 0.5916163804680645,
      "grad_norm": 1.5235916376113892,
      "learning_rate": 7.543784026567708e-05,
      "loss": 1.4944,
      "step": 16520
    },
    {
      "epoch": 0.5916521925976328,
      "grad_norm": 1.9779636859893799,
      "learning_rate": 7.542659674461429e-05,
      "loss": 1.3904,
      "step": 16521
    },
    {
      "epoch": 0.5916880047272011,
      "grad_norm": 1.3239980936050415,
      "learning_rate": 7.541535355415487e-05,
      "loss": 1.524,
      "step": 16522
    },
    {
      "epoch": 0.5917238168567693,
      "grad_norm": 1.3218424320220947,
      "learning_rate": 7.540411069445021e-05,
      "loss": 1.4584,
      "step": 16523
    },
    {
      "epoch": 0.5917596289863377,
      "grad_norm": 1.6652659177780151,
      "learning_rate": 7.53928681656515e-05,
      "loss": 1.1819,
      "step": 16524
    },
    {
      "epoch": 0.591795441115906,
      "grad_norm": 1.502694010734558,
      "learning_rate": 7.538162596791002e-05,
      "loss": 1.8572,
      "step": 16525
    },
    {
      "epoch": 0.5918312532454743,
      "grad_norm": 1.5385327339172363,
      "learning_rate": 7.537038410137698e-05,
      "loss": 1.2326,
      "step": 16526
    },
    {
      "epoch": 0.5918670653750425,
      "grad_norm": 1.5926456451416016,
      "learning_rate": 7.535914256620368e-05,
      "loss": 1.6148,
      "step": 16527
    },
    {
      "epoch": 0.5919028775046108,
      "grad_norm": 1.8524401187896729,
      "learning_rate": 7.534790136254132e-05,
      "loss": 1.1339,
      "step": 16528
    },
    {
      "epoch": 0.5919386896341791,
      "grad_norm": 1.430527687072754,
      "learning_rate": 7.533666049054115e-05,
      "loss": 1.4224,
      "step": 16529
    },
    {
      "epoch": 0.5919745017637473,
      "grad_norm": 1.5670427083969116,
      "learning_rate": 7.532541995035444e-05,
      "loss": 1.6936,
      "step": 16530
    },
    {
      "epoch": 0.5920103138933157,
      "grad_norm": 1.8079187870025635,
      "learning_rate": 7.53141797421323e-05,
      "loss": 1.2269,
      "step": 16531
    },
    {
      "epoch": 0.592046126022884,
      "grad_norm": 1.840510368347168,
      "learning_rate": 7.53029398660261e-05,
      "loss": 1.5701,
      "step": 16532
    },
    {
      "epoch": 0.5920819381524522,
      "grad_norm": 1.3353474140167236,
      "learning_rate": 7.529170032218691e-05,
      "loss": 1.6003,
      "step": 16533
    },
    {
      "epoch": 0.5921177502820205,
      "grad_norm": 2.0196423530578613,
      "learning_rate": 7.52804611107661e-05,
      "loss": 1.605,
      "step": 16534
    },
    {
      "epoch": 0.5921535624115888,
      "grad_norm": 1.7749592065811157,
      "learning_rate": 7.526922223191473e-05,
      "loss": 1.8357,
      "step": 16535
    },
    {
      "epoch": 0.592189374541157,
      "grad_norm": 1.9717570543289185,
      "learning_rate": 7.525798368578412e-05,
      "loss": 1.4984,
      "step": 16536
    },
    {
      "epoch": 0.5922251866707253,
      "grad_norm": 1.611171841621399,
      "learning_rate": 7.524674547252544e-05,
      "loss": 1.3366,
      "step": 16537
    },
    {
      "epoch": 0.5922609988002937,
      "grad_norm": 1.4326066970825195,
      "learning_rate": 7.523550759228981e-05,
      "loss": 1.5233,
      "step": 16538
    },
    {
      "epoch": 0.592296810929862,
      "grad_norm": 1.420731782913208,
      "learning_rate": 7.522427004522855e-05,
      "loss": 1.4304,
      "step": 16539
    },
    {
      "epoch": 0.5923326230594302,
      "grad_norm": 2.4821457862854004,
      "learning_rate": 7.52130328314927e-05,
      "loss": 1.534,
      "step": 16540
    },
    {
      "epoch": 0.5923684351889985,
      "grad_norm": 1.7809147834777832,
      "learning_rate": 7.52017959512336e-05,
      "loss": 1.3436,
      "step": 16541
    },
    {
      "epoch": 0.5924042473185668,
      "grad_norm": 1.487674593925476,
      "learning_rate": 7.519055940460227e-05,
      "loss": 1.3221,
      "step": 16542
    },
    {
      "epoch": 0.592440059448135,
      "grad_norm": 1.630618929862976,
      "learning_rate": 7.517932319175003e-05,
      "loss": 1.391,
      "step": 16543
    },
    {
      "epoch": 0.5924758715777033,
      "grad_norm": 1.7148510217666626,
      "learning_rate": 7.516808731282793e-05,
      "loss": 1.6317,
      "step": 16544
    },
    {
      "epoch": 0.5925116837072717,
      "grad_norm": 1.6634539365768433,
      "learning_rate": 7.515685176798723e-05,
      "loss": 1.8021,
      "step": 16545
    },
    {
      "epoch": 0.59254749583684,
      "grad_norm": 1.7055913209915161,
      "learning_rate": 7.514561655737904e-05,
      "loss": 1.7112,
      "step": 16546
    },
    {
      "epoch": 0.5925833079664082,
      "grad_norm": 1.6069657802581787,
      "learning_rate": 7.513438168115449e-05,
      "loss": 1.4381,
      "step": 16547
    },
    {
      "epoch": 0.5926191200959765,
      "grad_norm": 1.99888277053833,
      "learning_rate": 7.512314713946478e-05,
      "loss": 1.677,
      "step": 16548
    },
    {
      "epoch": 0.5926549322255448,
      "grad_norm": 1.9482409954071045,
      "learning_rate": 7.5111912932461e-05,
      "loss": 1.0121,
      "step": 16549
    },
    {
      "epoch": 0.592690744355113,
      "grad_norm": 1.8642888069152832,
      "learning_rate": 7.510067906029437e-05,
      "loss": 1.5993,
      "step": 16550
    },
    {
      "epoch": 0.5927265564846813,
      "grad_norm": 1.965591311454773,
      "learning_rate": 7.508944552311594e-05,
      "loss": 1.4873,
      "step": 16551
    },
    {
      "epoch": 0.5927623686142497,
      "grad_norm": 1.6098365783691406,
      "learning_rate": 7.507821232107695e-05,
      "loss": 1.433,
      "step": 16552
    },
    {
      "epoch": 0.592798180743818,
      "grad_norm": 1.4152309894561768,
      "learning_rate": 7.506697945432841e-05,
      "loss": 1.7259,
      "step": 16553
    },
    {
      "epoch": 0.5928339928733862,
      "grad_norm": 1.5540013313293457,
      "learning_rate": 7.505574692302155e-05,
      "loss": 1.408,
      "step": 16554
    },
    {
      "epoch": 0.5928698050029545,
      "grad_norm": 1.8713033199310303,
      "learning_rate": 7.504451472730743e-05,
      "loss": 1.4247,
      "step": 16555
    },
    {
      "epoch": 0.5929056171325228,
      "grad_norm": 1.4902887344360352,
      "learning_rate": 7.503328286733715e-05,
      "loss": 1.4305,
      "step": 16556
    },
    {
      "epoch": 0.592941429262091,
      "grad_norm": 2.143108367919922,
      "learning_rate": 7.502205134326185e-05,
      "loss": 1.4066,
      "step": 16557
    },
    {
      "epoch": 0.5929772413916593,
      "grad_norm": 2.025181770324707,
      "learning_rate": 7.501082015523263e-05,
      "loss": 1.6475,
      "step": 16558
    },
    {
      "epoch": 0.5930130535212277,
      "grad_norm": 2.7611522674560547,
      "learning_rate": 7.499958930340061e-05,
      "loss": 1.6847,
      "step": 16559
    },
    {
      "epoch": 0.593048865650796,
      "grad_norm": 1.5839300155639648,
      "learning_rate": 7.498835878791684e-05,
      "loss": 1.6008,
      "step": 16560
    },
    {
      "epoch": 0.5930846777803642,
      "grad_norm": 1.7940725088119507,
      "learning_rate": 7.497712860893245e-05,
      "loss": 1.2224,
      "step": 16561
    },
    {
      "epoch": 0.5931204899099325,
      "grad_norm": 1.4097790718078613,
      "learning_rate": 7.49658987665985e-05,
      "loss": 1.5644,
      "step": 16562
    },
    {
      "epoch": 0.5931563020395008,
      "grad_norm": 1.7357324361801147,
      "learning_rate": 7.495466926106614e-05,
      "loss": 1.3663,
      "step": 16563
    },
    {
      "epoch": 0.593192114169069,
      "grad_norm": 2.125781774520874,
      "learning_rate": 7.494344009248637e-05,
      "loss": 1.6129,
      "step": 16564
    },
    {
      "epoch": 0.5932279262986373,
      "grad_norm": 1.732550024986267,
      "learning_rate": 7.493221126101028e-05,
      "loss": 1.6356,
      "step": 16565
    },
    {
      "epoch": 0.5932637384282057,
      "grad_norm": 1.9407490491867065,
      "learning_rate": 7.492098276678898e-05,
      "loss": 1.3264,
      "step": 16566
    },
    {
      "epoch": 0.593299550557774,
      "grad_norm": 1.5045716762542725,
      "learning_rate": 7.490975460997348e-05,
      "loss": 1.1289,
      "step": 16567
    },
    {
      "epoch": 0.5933353626873422,
      "grad_norm": 1.425619125366211,
      "learning_rate": 7.489852679071488e-05,
      "loss": 1.4394,
      "step": 16568
    },
    {
      "epoch": 0.5933711748169105,
      "grad_norm": 2.01013445854187,
      "learning_rate": 7.488729930916421e-05,
      "loss": 1.4089,
      "step": 16569
    },
    {
      "epoch": 0.5934069869464788,
      "grad_norm": 1.647908329963684,
      "learning_rate": 7.487607216547255e-05,
      "loss": 1.6508,
      "step": 16570
    },
    {
      "epoch": 0.593442799076047,
      "grad_norm": 1.4094040393829346,
      "learning_rate": 7.486484535979092e-05,
      "loss": 1.5401,
      "step": 16571
    },
    {
      "epoch": 0.5934786112056153,
      "grad_norm": 1.2045484781265259,
      "learning_rate": 7.485361889227038e-05,
      "loss": 1.2202,
      "step": 16572
    },
    {
      "epoch": 0.5935144233351837,
      "grad_norm": 1.9746378660202026,
      "learning_rate": 7.484239276306198e-05,
      "loss": 1.4754,
      "step": 16573
    },
    {
      "epoch": 0.5935502354647519,
      "grad_norm": 1.6177870035171509,
      "learning_rate": 7.483116697231671e-05,
      "loss": 1.2873,
      "step": 16574
    },
    {
      "epoch": 0.5935860475943202,
      "grad_norm": 2.0700132846832275,
      "learning_rate": 7.481994152018563e-05,
      "loss": 1.1141,
      "step": 16575
    },
    {
      "epoch": 0.5936218597238885,
      "grad_norm": 1.6111782789230347,
      "learning_rate": 7.480871640681975e-05,
      "loss": 1.4394,
      "step": 16576
    },
    {
      "epoch": 0.5936576718534567,
      "grad_norm": 1.553205966949463,
      "learning_rate": 7.479749163237012e-05,
      "loss": 1.3981,
      "step": 16577
    },
    {
      "epoch": 0.593693483983025,
      "grad_norm": 1.7631208896636963,
      "learning_rate": 7.47862671969877e-05,
      "loss": 1.7236,
      "step": 16578
    },
    {
      "epoch": 0.5937292961125933,
      "grad_norm": 1.4806108474731445,
      "learning_rate": 7.477504310082354e-05,
      "loss": 1.5397,
      "step": 16579
    },
    {
      "epoch": 0.5937651082421617,
      "grad_norm": 1.9398987293243408,
      "learning_rate": 7.476381934402865e-05,
      "loss": 1.2193,
      "step": 16580
    },
    {
      "epoch": 0.5938009203717299,
      "grad_norm": 1.654647946357727,
      "learning_rate": 7.475259592675402e-05,
      "loss": 1.4312,
      "step": 16581
    },
    {
      "epoch": 0.5938367325012982,
      "grad_norm": 1.9441183805465698,
      "learning_rate": 7.474137284915065e-05,
      "loss": 1.4922,
      "step": 16582
    },
    {
      "epoch": 0.5938725446308665,
      "grad_norm": 2.232104539871216,
      "learning_rate": 7.47301501113695e-05,
      "loss": 1.2633,
      "step": 16583
    },
    {
      "epoch": 0.5939083567604347,
      "grad_norm": 1.7381802797317505,
      "learning_rate": 7.471892771356161e-05,
      "loss": 1.4915,
      "step": 16584
    },
    {
      "epoch": 0.593944168890003,
      "grad_norm": 1.7541508674621582,
      "learning_rate": 7.470770565587792e-05,
      "loss": 1.4452,
      "step": 16585
    },
    {
      "epoch": 0.5939799810195713,
      "grad_norm": 1.4216136932373047,
      "learning_rate": 7.469648393846943e-05,
      "loss": 1.3694,
      "step": 16586
    },
    {
      "epoch": 0.5940157931491397,
      "grad_norm": 1.4462414979934692,
      "learning_rate": 7.46852625614871e-05,
      "loss": 1.5004,
      "step": 16587
    },
    {
      "epoch": 0.5940516052787079,
      "grad_norm": 1.6865832805633545,
      "learning_rate": 7.467404152508193e-05,
      "loss": 1.8021,
      "step": 16588
    },
    {
      "epoch": 0.5940874174082762,
      "grad_norm": 2.2418625354766846,
      "learning_rate": 7.466282082940484e-05,
      "loss": 1.493,
      "step": 16589
    },
    {
      "epoch": 0.5941232295378445,
      "grad_norm": 2.0363099575042725,
      "learning_rate": 7.465160047460685e-05,
      "loss": 1.7929,
      "step": 16590
    },
    {
      "epoch": 0.5941590416674127,
      "grad_norm": 1.8767224550247192,
      "learning_rate": 7.464038046083885e-05,
      "loss": 1.5985,
      "step": 16591
    },
    {
      "epoch": 0.594194853796981,
      "grad_norm": 1.5032994747161865,
      "learning_rate": 7.462916078825182e-05,
      "loss": 1.4272,
      "step": 16592
    },
    {
      "epoch": 0.5942306659265493,
      "grad_norm": 1.4486925601959229,
      "learning_rate": 7.46179414569967e-05,
      "loss": 1.4855,
      "step": 16593
    },
    {
      "epoch": 0.5942664780561177,
      "grad_norm": 1.592706561088562,
      "learning_rate": 7.460672246722444e-05,
      "loss": 1.7044,
      "step": 16594
    },
    {
      "epoch": 0.5943022901856859,
      "grad_norm": 1.9221786260604858,
      "learning_rate": 7.4595503819086e-05,
      "loss": 1.2772,
      "step": 16595
    },
    {
      "epoch": 0.5943381023152542,
      "grad_norm": 1.6944142580032349,
      "learning_rate": 7.458428551273226e-05,
      "loss": 1.4654,
      "step": 16596
    },
    {
      "epoch": 0.5943739144448225,
      "grad_norm": 2.0230796337127686,
      "learning_rate": 7.45730675483142e-05,
      "loss": 1.3047,
      "step": 16597
    },
    {
      "epoch": 0.5944097265743907,
      "grad_norm": 1.6869089603424072,
      "learning_rate": 7.456184992598267e-05,
      "loss": 1.5513,
      "step": 16598
    },
    {
      "epoch": 0.594445538703959,
      "grad_norm": 1.6149792671203613,
      "learning_rate": 7.455063264588869e-05,
      "loss": 1.4627,
      "step": 16599
    },
    {
      "epoch": 0.5944813508335273,
      "grad_norm": 1.7246642112731934,
      "learning_rate": 7.453941570818309e-05,
      "loss": 1.502,
      "step": 16600
    },
    {
      "epoch": 0.5945171629630956,
      "grad_norm": 1.4698948860168457,
      "learning_rate": 7.452819911301681e-05,
      "loss": 1.1747,
      "step": 16601
    },
    {
      "epoch": 0.5945529750926639,
      "grad_norm": 2.224147081375122,
      "learning_rate": 7.451698286054076e-05,
      "loss": 1.7846,
      "step": 16602
    },
    {
      "epoch": 0.5945887872222322,
      "grad_norm": 1.9010647535324097,
      "learning_rate": 7.450576695090583e-05,
      "loss": 1.1994,
      "step": 16603
    },
    {
      "epoch": 0.5946245993518005,
      "grad_norm": 1.8986421823501587,
      "learning_rate": 7.449455138426294e-05,
      "loss": 1.392,
      "step": 16604
    },
    {
      "epoch": 0.5946604114813687,
      "grad_norm": 2.1596598625183105,
      "learning_rate": 7.448333616076293e-05,
      "loss": 1.5052,
      "step": 16605
    },
    {
      "epoch": 0.594696223610937,
      "grad_norm": 2.1113686561584473,
      "learning_rate": 7.447212128055675e-05,
      "loss": 1.2468,
      "step": 16606
    },
    {
      "epoch": 0.5947320357405053,
      "grad_norm": 1.4451698064804077,
      "learning_rate": 7.446090674379522e-05,
      "loss": 1.6236,
      "step": 16607
    },
    {
      "epoch": 0.5947678478700736,
      "grad_norm": 2.0924429893493652,
      "learning_rate": 7.444969255062928e-05,
      "loss": 1.4995,
      "step": 16608
    },
    {
      "epoch": 0.5948036599996419,
      "grad_norm": 1.6229060888290405,
      "learning_rate": 7.443847870120976e-05,
      "loss": 1.452,
      "step": 16609
    },
    {
      "epoch": 0.5948394721292102,
      "grad_norm": 1.5565780401229858,
      "learning_rate": 7.442726519568751e-05,
      "loss": 1.4328,
      "step": 16610
    },
    {
      "epoch": 0.5948752842587784,
      "grad_norm": 1.7966111898422241,
      "learning_rate": 7.441605203421345e-05,
      "loss": 1.5491,
      "step": 16611
    },
    {
      "epoch": 0.5949110963883467,
      "grad_norm": 1.5467952489852905,
      "learning_rate": 7.440483921693839e-05,
      "loss": 1.5181,
      "step": 16612
    },
    {
      "epoch": 0.594946908517915,
      "grad_norm": 1.3843365907669067,
      "learning_rate": 7.439362674401322e-05,
      "loss": 1.3912,
      "step": 16613
    },
    {
      "epoch": 0.5949827206474833,
      "grad_norm": 1.6058661937713623,
      "learning_rate": 7.438241461558875e-05,
      "loss": 1.5003,
      "step": 16614
    },
    {
      "epoch": 0.5950185327770516,
      "grad_norm": 1.2036961317062378,
      "learning_rate": 7.437120283181586e-05,
      "loss": 1.3581,
      "step": 16615
    },
    {
      "epoch": 0.5950543449066199,
      "grad_norm": 1.3170970678329468,
      "learning_rate": 7.435999139284538e-05,
      "loss": 1.4601,
      "step": 16616
    },
    {
      "epoch": 0.5950901570361882,
      "grad_norm": 2.2052175998687744,
      "learning_rate": 7.434878029882814e-05,
      "loss": 1.5695,
      "step": 16617
    },
    {
      "epoch": 0.5951259691657564,
      "grad_norm": 1.6902546882629395,
      "learning_rate": 7.433756954991499e-05,
      "loss": 1.9375,
      "step": 16618
    },
    {
      "epoch": 0.5951617812953247,
      "grad_norm": 1.6723322868347168,
      "learning_rate": 7.43263591462567e-05,
      "loss": 1.5111,
      "step": 16619
    },
    {
      "epoch": 0.595197593424893,
      "grad_norm": 1.820304274559021,
      "learning_rate": 7.431514908800417e-05,
      "loss": 1.4817,
      "step": 16620
    },
    {
      "epoch": 0.5952334055544612,
      "grad_norm": 2.129159927368164,
      "learning_rate": 7.430393937530815e-05,
      "loss": 1.5987,
      "step": 16621
    },
    {
      "epoch": 0.5952692176840296,
      "grad_norm": 1.5790053606033325,
      "learning_rate": 7.429273000831949e-05,
      "loss": 1.5129,
      "step": 16622
    },
    {
      "epoch": 0.5953050298135979,
      "grad_norm": 1.2789185047149658,
      "learning_rate": 7.4281520987189e-05,
      "loss": 1.4375,
      "step": 16623
    },
    {
      "epoch": 0.5953408419431662,
      "grad_norm": 1.4424225091934204,
      "learning_rate": 7.427031231206745e-05,
      "loss": 1.2032,
      "step": 16624
    },
    {
      "epoch": 0.5953766540727344,
      "grad_norm": 2.2218496799468994,
      "learning_rate": 7.425910398310566e-05,
      "loss": 1.3099,
      "step": 16625
    },
    {
      "epoch": 0.5954124662023027,
      "grad_norm": 2.0501346588134766,
      "learning_rate": 7.424789600045444e-05,
      "loss": 1.5333,
      "step": 16626
    },
    {
      "epoch": 0.595448278331871,
      "grad_norm": 1.7061489820480347,
      "learning_rate": 7.423668836426458e-05,
      "loss": 1.6931,
      "step": 16627
    },
    {
      "epoch": 0.5954840904614392,
      "grad_norm": 2.1097772121429443,
      "learning_rate": 7.422548107468679e-05,
      "loss": 1.2726,
      "step": 16628
    },
    {
      "epoch": 0.5955199025910076,
      "grad_norm": 1.6923251152038574,
      "learning_rate": 7.421427413187197e-05,
      "loss": 1.7505,
      "step": 16629
    },
    {
      "epoch": 0.5955557147205759,
      "grad_norm": 1.5730112791061401,
      "learning_rate": 7.420306753597076e-05,
      "loss": 1.4845,
      "step": 16630
    },
    {
      "epoch": 0.5955915268501442,
      "grad_norm": 1.3994970321655273,
      "learning_rate": 7.419186128713407e-05,
      "loss": 1.2497,
      "step": 16631
    },
    {
      "epoch": 0.5956273389797124,
      "grad_norm": 1.8320046663284302,
      "learning_rate": 7.418065538551253e-05,
      "loss": 1.3596,
      "step": 16632
    },
    {
      "epoch": 0.5956631511092807,
      "grad_norm": 3.0585572719573975,
      "learning_rate": 7.4169449831257e-05,
      "loss": 1.7986,
      "step": 16633
    },
    {
      "epoch": 0.595698963238849,
      "grad_norm": 1.4562921524047852,
      "learning_rate": 7.415824462451824e-05,
      "loss": 1.3,
      "step": 16634
    },
    {
      "epoch": 0.5957347753684172,
      "grad_norm": 1.64504075050354,
      "learning_rate": 7.41470397654469e-05,
      "loss": 1.5275,
      "step": 16635
    },
    {
      "epoch": 0.5957705874979856,
      "grad_norm": 1.7294472455978394,
      "learning_rate": 7.413583525419385e-05,
      "loss": 1.3219,
      "step": 16636
    },
    {
      "epoch": 0.5958063996275539,
      "grad_norm": 1.6882243156433105,
      "learning_rate": 7.41246310909097e-05,
      "loss": 1.5301,
      "step": 16637
    },
    {
      "epoch": 0.5958422117571222,
      "grad_norm": 1.7502877712249756,
      "learning_rate": 7.411342727574533e-05,
      "loss": 1.6065,
      "step": 16638
    },
    {
      "epoch": 0.5958780238866904,
      "grad_norm": 1.149178385734558,
      "learning_rate": 7.410222380885135e-05,
      "loss": 1.2456,
      "step": 16639
    },
    {
      "epoch": 0.5959138360162587,
      "grad_norm": 1.7622125148773193,
      "learning_rate": 7.409102069037862e-05,
      "loss": 1.017,
      "step": 16640
    },
    {
      "epoch": 0.595949648145827,
      "grad_norm": 2.7804136276245117,
      "learning_rate": 7.407981792047769e-05,
      "loss": 1.479,
      "step": 16641
    },
    {
      "epoch": 0.5959854602753952,
      "grad_norm": 1.9207881689071655,
      "learning_rate": 7.406861549929946e-05,
      "loss": 1.4618,
      "step": 16642
    },
    {
      "epoch": 0.5960212724049636,
      "grad_norm": 1.6801304817199707,
      "learning_rate": 7.405741342699453e-05,
      "loss": 1.2735,
      "step": 16643
    },
    {
      "epoch": 0.5960570845345319,
      "grad_norm": 2.0418813228607178,
      "learning_rate": 7.404621170371362e-05,
      "loss": 1.4934,
      "step": 16644
    },
    {
      "epoch": 0.5960928966641001,
      "grad_norm": 1.2861592769622803,
      "learning_rate": 7.403501032960748e-05,
      "loss": 1.4293,
      "step": 16645
    },
    {
      "epoch": 0.5961287087936684,
      "grad_norm": 1.7379697561264038,
      "learning_rate": 7.402380930482673e-05,
      "loss": 1.3013,
      "step": 16646
    },
    {
      "epoch": 0.5961645209232367,
      "grad_norm": 1.4268943071365356,
      "learning_rate": 7.40126086295222e-05,
      "loss": 1.5411,
      "step": 16647
    },
    {
      "epoch": 0.596200333052805,
      "grad_norm": 1.8618836402893066,
      "learning_rate": 7.400140830384443e-05,
      "loss": 1.5819,
      "step": 16648
    },
    {
      "epoch": 0.5962361451823732,
      "grad_norm": 2.0353612899780273,
      "learning_rate": 7.399020832794424e-05,
      "loss": 1.4602,
      "step": 16649
    },
    {
      "epoch": 0.5962719573119416,
      "grad_norm": 1.4291691780090332,
      "learning_rate": 7.397900870197216e-05,
      "loss": 1.5755,
      "step": 16650
    },
    {
      "epoch": 0.5963077694415099,
      "grad_norm": 1.5784722566604614,
      "learning_rate": 7.396780942607904e-05,
      "loss": 1.2951,
      "step": 16651
    },
    {
      "epoch": 0.5963435815710781,
      "grad_norm": 2.4634132385253906,
      "learning_rate": 7.395661050041545e-05,
      "loss": 1.3903,
      "step": 16652
    },
    {
      "epoch": 0.5963793937006464,
      "grad_norm": 1.3411083221435547,
      "learning_rate": 7.394541192513202e-05,
      "loss": 1.5628,
      "step": 16653
    },
    {
      "epoch": 0.5964152058302147,
      "grad_norm": 1.625428557395935,
      "learning_rate": 7.393421370037952e-05,
      "loss": 1.4502,
      "step": 16654
    },
    {
      "epoch": 0.596451017959783,
      "grad_norm": 1.308756947517395,
      "learning_rate": 7.392301582630852e-05,
      "loss": 0.9473,
      "step": 16655
    },
    {
      "epoch": 0.5964868300893512,
      "grad_norm": 1.509726881980896,
      "learning_rate": 7.391181830306972e-05,
      "loss": 1.2107,
      "step": 16656
    },
    {
      "epoch": 0.5965226422189196,
      "grad_norm": 1.3004424571990967,
      "learning_rate": 7.390062113081373e-05,
      "loss": 1.6063,
      "step": 16657
    },
    {
      "epoch": 0.5965584543484879,
      "grad_norm": 1.2375189065933228,
      "learning_rate": 7.388942430969123e-05,
      "loss": 1.6043,
      "step": 16658
    },
    {
      "epoch": 0.5965942664780561,
      "grad_norm": 1.9687907695770264,
      "learning_rate": 7.387822783985283e-05,
      "loss": 1.3229,
      "step": 16659
    },
    {
      "epoch": 0.5966300786076244,
      "grad_norm": 2.021730422973633,
      "learning_rate": 7.386703172144921e-05,
      "loss": 1.2441,
      "step": 16660
    },
    {
      "epoch": 0.5966658907371927,
      "grad_norm": 1.6941550970077515,
      "learning_rate": 7.385583595463099e-05,
      "loss": 1.7031,
      "step": 16661
    },
    {
      "epoch": 0.5967017028667609,
      "grad_norm": 1.5119400024414062,
      "learning_rate": 7.384464053954872e-05,
      "loss": 1.5488,
      "step": 16662
    },
    {
      "epoch": 0.5967375149963292,
      "grad_norm": 1.5219899415969849,
      "learning_rate": 7.383344547635311e-05,
      "loss": 1.378,
      "step": 16663
    },
    {
      "epoch": 0.5967733271258976,
      "grad_norm": 1.6702772378921509,
      "learning_rate": 7.382225076519471e-05,
      "loss": 1.1648,
      "step": 16664
    },
    {
      "epoch": 0.5968091392554659,
      "grad_norm": 1.8304907083511353,
      "learning_rate": 7.381105640622419e-05,
      "loss": 1.3808,
      "step": 16665
    },
    {
      "epoch": 0.5968449513850341,
      "grad_norm": 1.6182093620300293,
      "learning_rate": 7.379986239959209e-05,
      "loss": 1.4653,
      "step": 16666
    },
    {
      "epoch": 0.5968807635146024,
      "grad_norm": 1.697573184967041,
      "learning_rate": 7.378866874544908e-05,
      "loss": 1.0452,
      "step": 16667
    },
    {
      "epoch": 0.5969165756441707,
      "grad_norm": 1.5860910415649414,
      "learning_rate": 7.377747544394568e-05,
      "loss": 1.2655,
      "step": 16668
    },
    {
      "epoch": 0.5969523877737389,
      "grad_norm": 1.4699441194534302,
      "learning_rate": 7.376628249523257e-05,
      "loss": 1.4476,
      "step": 16669
    },
    {
      "epoch": 0.5969881999033072,
      "grad_norm": 1.399438738822937,
      "learning_rate": 7.375508989946027e-05,
      "loss": 1.5486,
      "step": 16670
    },
    {
      "epoch": 0.5970240120328756,
      "grad_norm": 1.9118127822875977,
      "learning_rate": 7.374389765677938e-05,
      "loss": 1.3724,
      "step": 16671
    },
    {
      "epoch": 0.5970598241624439,
      "grad_norm": 1.6800737380981445,
      "learning_rate": 7.373270576734048e-05,
      "loss": 1.6724,
      "step": 16672
    },
    {
      "epoch": 0.5970956362920121,
      "grad_norm": 1.4176125526428223,
      "learning_rate": 7.372151423129414e-05,
      "loss": 1.4472,
      "step": 16673
    },
    {
      "epoch": 0.5971314484215804,
      "grad_norm": 1.7875847816467285,
      "learning_rate": 7.371032304879094e-05,
      "loss": 1.4848,
      "step": 16674
    },
    {
      "epoch": 0.5971672605511487,
      "grad_norm": 1.5433893203735352,
      "learning_rate": 7.369913221998141e-05,
      "loss": 1.4387,
      "step": 16675
    },
    {
      "epoch": 0.5972030726807169,
      "grad_norm": 2.274228572845459,
      "learning_rate": 7.368794174501615e-05,
      "loss": 1.5471,
      "step": 16676
    },
    {
      "epoch": 0.5972388848102852,
      "grad_norm": 2.2298521995544434,
      "learning_rate": 7.367675162404567e-05,
      "loss": 1.6403,
      "step": 16677
    },
    {
      "epoch": 0.5972746969398536,
      "grad_norm": 1.9916534423828125,
      "learning_rate": 7.366556185722056e-05,
      "loss": 1.4881,
      "step": 16678
    },
    {
      "epoch": 0.5973105090694218,
      "grad_norm": 2.0193593502044678,
      "learning_rate": 7.365437244469135e-05,
      "loss": 1.4196,
      "step": 16679
    },
    {
      "epoch": 0.5973463211989901,
      "grad_norm": 1.4831794500350952,
      "learning_rate": 7.364318338660858e-05,
      "loss": 1.7932,
      "step": 16680
    },
    {
      "epoch": 0.5973821333285584,
      "grad_norm": 1.533564567565918,
      "learning_rate": 7.363199468312277e-05,
      "loss": 1.289,
      "step": 16681
    },
    {
      "epoch": 0.5974179454581267,
      "grad_norm": 2.7487010955810547,
      "learning_rate": 7.362080633438445e-05,
      "loss": 1.6209,
      "step": 16682
    },
    {
      "epoch": 0.5974537575876949,
      "grad_norm": 1.5415210723876953,
      "learning_rate": 7.360961834054418e-05,
      "loss": 1.7845,
      "step": 16683
    },
    {
      "epoch": 0.5974895697172632,
      "grad_norm": 1.641719937324524,
      "learning_rate": 7.359843070175242e-05,
      "loss": 1.4157,
      "step": 16684
    },
    {
      "epoch": 0.5975253818468316,
      "grad_norm": 1.4241138696670532,
      "learning_rate": 7.358724341815975e-05,
      "loss": 1.3276,
      "step": 16685
    },
    {
      "epoch": 0.5975611939763998,
      "grad_norm": 2.071657419204712,
      "learning_rate": 7.357605648991661e-05,
      "loss": 1.293,
      "step": 16686
    },
    {
      "epoch": 0.5975970061059681,
      "grad_norm": 1.8686021566390991,
      "learning_rate": 7.356486991717359e-05,
      "loss": 1.3658,
      "step": 16687
    },
    {
      "epoch": 0.5976328182355364,
      "grad_norm": 2.0502803325653076,
      "learning_rate": 7.355368370008113e-05,
      "loss": 1.5443,
      "step": 16688
    },
    {
      "epoch": 0.5976686303651046,
      "grad_norm": 1.549391508102417,
      "learning_rate": 7.354249783878973e-05,
      "loss": 1.4303,
      "step": 16689
    },
    {
      "epoch": 0.5977044424946729,
      "grad_norm": 1.3476611375808716,
      "learning_rate": 7.353131233344991e-05,
      "loss": 1.569,
      "step": 16690
    },
    {
      "epoch": 0.5977402546242412,
      "grad_norm": 1.3979164361953735,
      "learning_rate": 7.352012718421212e-05,
      "loss": 1.2161,
      "step": 16691
    },
    {
      "epoch": 0.5977760667538096,
      "grad_norm": 2.414729595184326,
      "learning_rate": 7.350894239122689e-05,
      "loss": 1.8839,
      "step": 16692
    },
    {
      "epoch": 0.5978118788833778,
      "grad_norm": 1.8490568399429321,
      "learning_rate": 7.349775795464466e-05,
      "loss": 1.3075,
      "step": 16693
    },
    {
      "epoch": 0.5978476910129461,
      "grad_norm": 1.7273000478744507,
      "learning_rate": 7.348657387461591e-05,
      "loss": 1.4859,
      "step": 16694
    },
    {
      "epoch": 0.5978835031425144,
      "grad_norm": 1.6168313026428223,
      "learning_rate": 7.34753901512911e-05,
      "loss": 1.6608,
      "step": 16695
    },
    {
      "epoch": 0.5979193152720826,
      "grad_norm": 1.6971771717071533,
      "learning_rate": 7.346420678482071e-05,
      "loss": 1.2613,
      "step": 16696
    },
    {
      "epoch": 0.5979551274016509,
      "grad_norm": 1.7362477779388428,
      "learning_rate": 7.345302377535521e-05,
      "loss": 1.3894,
      "step": 16697
    },
    {
      "epoch": 0.5979909395312192,
      "grad_norm": 2.0205066204071045,
      "learning_rate": 7.3441841123045e-05,
      "loss": 1.6138,
      "step": 16698
    },
    {
      "epoch": 0.5980267516607876,
      "grad_norm": 2.0709989070892334,
      "learning_rate": 7.343065882804056e-05,
      "loss": 1.0796,
      "step": 16699
    },
    {
      "epoch": 0.5980625637903558,
      "grad_norm": 1.900892734527588,
      "learning_rate": 7.341947689049233e-05,
      "loss": 1.3942,
      "step": 16700
    },
    {
      "epoch": 0.5980983759199241,
      "grad_norm": 1.6259151697158813,
      "learning_rate": 7.340829531055078e-05,
      "loss": 1.0872,
      "step": 16701
    },
    {
      "epoch": 0.5981341880494924,
      "grad_norm": 2.2745935916900635,
      "learning_rate": 7.339711408836629e-05,
      "loss": 1.7529,
      "step": 16702
    },
    {
      "epoch": 0.5981700001790606,
      "grad_norm": 1.6044317483901978,
      "learning_rate": 7.338593322408933e-05,
      "loss": 1.6281,
      "step": 16703
    },
    {
      "epoch": 0.5982058123086289,
      "grad_norm": 2.226635217666626,
      "learning_rate": 7.33747527178703e-05,
      "loss": 1.595,
      "step": 16704
    },
    {
      "epoch": 0.5982416244381972,
      "grad_norm": 2.034383535385132,
      "learning_rate": 7.336357256985964e-05,
      "loss": 1.4936,
      "step": 16705
    },
    {
      "epoch": 0.5982774365677656,
      "grad_norm": 1.4976028203964233,
      "learning_rate": 7.335239278020776e-05,
      "loss": 1.532,
      "step": 16706
    },
    {
      "epoch": 0.5983132486973338,
      "grad_norm": 1.4269556999206543,
      "learning_rate": 7.334121334906503e-05,
      "loss": 1.4254,
      "step": 16707
    },
    {
      "epoch": 0.5983490608269021,
      "grad_norm": 1.4683499336242676,
      "learning_rate": 7.333003427658192e-05,
      "loss": 1.6565,
      "step": 16708
    },
    {
      "epoch": 0.5983848729564704,
      "grad_norm": 2.059978485107422,
      "learning_rate": 7.331885556290876e-05,
      "loss": 1.6096,
      "step": 16709
    },
    {
      "epoch": 0.5984206850860386,
      "grad_norm": 1.7053067684173584,
      "learning_rate": 7.330767720819601e-05,
      "loss": 1.6937,
      "step": 16710
    },
    {
      "epoch": 0.5984564972156069,
      "grad_norm": 2.052213191986084,
      "learning_rate": 7.329649921259402e-05,
      "loss": 1.4403,
      "step": 16711
    },
    {
      "epoch": 0.5984923093451752,
      "grad_norm": 1.58275306224823,
      "learning_rate": 7.32853215762532e-05,
      "loss": 1.3232,
      "step": 16712
    },
    {
      "epoch": 0.5985281214747435,
      "grad_norm": 1.7357984781265259,
      "learning_rate": 7.32741442993239e-05,
      "loss": 1.4713,
      "step": 16713
    },
    {
      "epoch": 0.5985639336043118,
      "grad_norm": 1.503201961517334,
      "learning_rate": 7.326296738195654e-05,
      "loss": 1.5623,
      "step": 16714
    },
    {
      "epoch": 0.5985997457338801,
      "grad_norm": 1.5212639570236206,
      "learning_rate": 7.325179082430148e-05,
      "loss": 1.3255,
      "step": 16715
    },
    {
      "epoch": 0.5986355578634484,
      "grad_norm": 1.991188645362854,
      "learning_rate": 7.324061462650901e-05,
      "loss": 1.4,
      "step": 16716
    },
    {
      "epoch": 0.5986713699930166,
      "grad_norm": 2.1019017696380615,
      "learning_rate": 7.32294387887296e-05,
      "loss": 1.3448,
      "step": 16717
    },
    {
      "epoch": 0.5987071821225849,
      "grad_norm": 1.5686511993408203,
      "learning_rate": 7.321826331111353e-05,
      "loss": 1.709,
      "step": 16718
    },
    {
      "epoch": 0.5987429942521532,
      "grad_norm": 1.8359214067459106,
      "learning_rate": 7.320708819381121e-05,
      "loss": 1.534,
      "step": 16719
    },
    {
      "epoch": 0.5987788063817215,
      "grad_norm": 1.9091947078704834,
      "learning_rate": 7.319591343697293e-05,
      "loss": 1.4201,
      "step": 16720
    },
    {
      "epoch": 0.5988146185112898,
      "grad_norm": 1.5145343542099,
      "learning_rate": 7.31847390407491e-05,
      "loss": 1.4507,
      "step": 16721
    },
    {
      "epoch": 0.5988504306408581,
      "grad_norm": 1.864920973777771,
      "learning_rate": 7.317356500528996e-05,
      "loss": 1.43,
      "step": 16722
    },
    {
      "epoch": 0.5988862427704263,
      "grad_norm": 1.6559187173843384,
      "learning_rate": 7.316239133074595e-05,
      "loss": 1.3573,
      "step": 16723
    },
    {
      "epoch": 0.5989220548999946,
      "grad_norm": 1.7337490320205688,
      "learning_rate": 7.315121801726737e-05,
      "loss": 1.2266,
      "step": 16724
    },
    {
      "epoch": 0.5989578670295629,
      "grad_norm": 1.527982473373413,
      "learning_rate": 7.314004506500443e-05,
      "loss": 1.6621,
      "step": 16725
    },
    {
      "epoch": 0.5989936791591312,
      "grad_norm": 1.4964022636413574,
      "learning_rate": 7.312887247410762e-05,
      "loss": 1.3453,
      "step": 16726
    },
    {
      "epoch": 0.5990294912886995,
      "grad_norm": 1.426389455795288,
      "learning_rate": 7.311770024472711e-05,
      "loss": 1.503,
      "step": 16727
    },
    {
      "epoch": 0.5990653034182678,
      "grad_norm": 1.8016306161880493,
      "learning_rate": 7.31065283770133e-05,
      "loss": 1.5566,
      "step": 16728
    },
    {
      "epoch": 0.5991011155478361,
      "grad_norm": 1.6403863430023193,
      "learning_rate": 7.309535687111644e-05,
      "loss": 1.613,
      "step": 16729
    },
    {
      "epoch": 0.5991369276774043,
      "grad_norm": 1.8085709810256958,
      "learning_rate": 7.308418572718687e-05,
      "loss": 1.7136,
      "step": 16730
    },
    {
      "epoch": 0.5991727398069726,
      "grad_norm": 1.6913074254989624,
      "learning_rate": 7.307301494537489e-05,
      "loss": 1.2859,
      "step": 16731
    },
    {
      "epoch": 0.5992085519365409,
      "grad_norm": 1.4169158935546875,
      "learning_rate": 7.306184452583067e-05,
      "loss": 1.6721,
      "step": 16732
    },
    {
      "epoch": 0.5992443640661091,
      "grad_norm": 1.5620194673538208,
      "learning_rate": 7.305067446870468e-05,
      "loss": 1.4362,
      "step": 16733
    },
    {
      "epoch": 0.5992801761956775,
      "grad_norm": 1.7782098054885864,
      "learning_rate": 7.303950477414703e-05,
      "loss": 1.5194,
      "step": 16734
    },
    {
      "epoch": 0.5993159883252458,
      "grad_norm": 1.9278711080551147,
      "learning_rate": 7.302833544230812e-05,
      "loss": 1.2504,
      "step": 16735
    },
    {
      "epoch": 0.5993518004548141,
      "grad_norm": 1.8028457164764404,
      "learning_rate": 7.301716647333812e-05,
      "loss": 1.0734,
      "step": 16736
    },
    {
      "epoch": 0.5993876125843823,
      "grad_norm": 1.6356738805770874,
      "learning_rate": 7.300599786738739e-05,
      "loss": 1.6322,
      "step": 16737
    },
    {
      "epoch": 0.5994234247139506,
      "grad_norm": 1.7443691492080688,
      "learning_rate": 7.299482962460607e-05,
      "loss": 1.4234,
      "step": 16738
    },
    {
      "epoch": 0.5994592368435189,
      "grad_norm": 1.5225340127944946,
      "learning_rate": 7.298366174514456e-05,
      "loss": 1.5641,
      "step": 16739
    },
    {
      "epoch": 0.5994950489730871,
      "grad_norm": 1.35145103931427,
      "learning_rate": 7.297249422915301e-05,
      "loss": 1.4949,
      "step": 16740
    },
    {
      "epoch": 0.5995308611026555,
      "grad_norm": 2.0662713050842285,
      "learning_rate": 7.296132707678166e-05,
      "loss": 1.4635,
      "step": 16741
    },
    {
      "epoch": 0.5995666732322238,
      "grad_norm": 1.6183621883392334,
      "learning_rate": 7.29501602881808e-05,
      "loss": 1.6438,
      "step": 16742
    },
    {
      "epoch": 0.5996024853617921,
      "grad_norm": 2.1140263080596924,
      "learning_rate": 7.29389938635006e-05,
      "loss": 1.778,
      "step": 16743
    },
    {
      "epoch": 0.5996382974913603,
      "grad_norm": 1.8030223846435547,
      "learning_rate": 7.292782780289141e-05,
      "loss": 1.6374,
      "step": 16744
    },
    {
      "epoch": 0.5996741096209286,
      "grad_norm": 1.7146693468093872,
      "learning_rate": 7.291666210650328e-05,
      "loss": 1.3417,
      "step": 16745
    },
    {
      "epoch": 0.5997099217504969,
      "grad_norm": 1.5830776691436768,
      "learning_rate": 7.290549677448661e-05,
      "loss": 1.4825,
      "step": 16746
    },
    {
      "epoch": 0.5997457338800651,
      "grad_norm": 2.0402321815490723,
      "learning_rate": 7.289433180699148e-05,
      "loss": 1.2888,
      "step": 16747
    },
    {
      "epoch": 0.5997815460096335,
      "grad_norm": 1.7229087352752686,
      "learning_rate": 7.28831672041682e-05,
      "loss": 1.4803,
      "step": 16748
    },
    {
      "epoch": 0.5998173581392018,
      "grad_norm": 1.6451315879821777,
      "learning_rate": 7.287200296616689e-05,
      "loss": 1.5543,
      "step": 16749
    },
    {
      "epoch": 0.59985317026877,
      "grad_norm": 2.058530569076538,
      "learning_rate": 7.286083909313779e-05,
      "loss": 1.4938,
      "step": 16750
    },
    {
      "epoch": 0.5998889823983383,
      "grad_norm": 1.7364871501922607,
      "learning_rate": 7.284967558523112e-05,
      "loss": 1.5158,
      "step": 16751
    },
    {
      "epoch": 0.5999247945279066,
      "grad_norm": 1.551857829093933,
      "learning_rate": 7.2838512442597e-05,
      "loss": 1.2004,
      "step": 16752
    },
    {
      "epoch": 0.5999606066574749,
      "grad_norm": 1.3351510763168335,
      "learning_rate": 7.282734966538569e-05,
      "loss": 1.6516,
      "step": 16753
    },
    {
      "epoch": 0.5999964187870431,
      "grad_norm": 1.86201012134552,
      "learning_rate": 7.281618725374733e-05,
      "loss": 1.3853,
      "step": 16754
    },
    {
      "epoch": 0.6000322309166115,
      "grad_norm": 1.8424255847930908,
      "learning_rate": 7.28050252078321e-05,
      "loss": 1.5038,
      "step": 16755
    },
    {
      "epoch": 0.6000680430461798,
      "grad_norm": 1.396933913230896,
      "learning_rate": 7.279386352779016e-05,
      "loss": 1.3897,
      "step": 16756
    },
    {
      "epoch": 0.600103855175748,
      "grad_norm": 1.5650379657745361,
      "learning_rate": 7.278270221377174e-05,
      "loss": 1.2216,
      "step": 16757
    },
    {
      "epoch": 0.6001396673053163,
      "grad_norm": 1.6911360025405884,
      "learning_rate": 7.277154126592695e-05,
      "loss": 1.413,
      "step": 16758
    },
    {
      "epoch": 0.6001754794348846,
      "grad_norm": 1.6357327699661255,
      "learning_rate": 7.276038068440592e-05,
      "loss": 1.1571,
      "step": 16759
    },
    {
      "epoch": 0.6002112915644529,
      "grad_norm": 1.6942464113235474,
      "learning_rate": 7.274922046935885e-05,
      "loss": 1.5146,
      "step": 16760
    },
    {
      "epoch": 0.6002471036940211,
      "grad_norm": 1.8150880336761475,
      "learning_rate": 7.273806062093585e-05,
      "loss": 1.4704,
      "step": 16761
    },
    {
      "epoch": 0.6002829158235895,
      "grad_norm": 1.3599683046340942,
      "learning_rate": 7.27269011392871e-05,
      "loss": 1.3613,
      "step": 16762
    },
    {
      "epoch": 0.6003187279531578,
      "grad_norm": 1.4246058464050293,
      "learning_rate": 7.271574202456268e-05,
      "loss": 1.6298,
      "step": 16763
    },
    {
      "epoch": 0.600354540082726,
      "grad_norm": 1.431622862815857,
      "learning_rate": 7.27045832769128e-05,
      "loss": 1.4792,
      "step": 16764
    },
    {
      "epoch": 0.6003903522122943,
      "grad_norm": 2.5366933345794678,
      "learning_rate": 7.269342489648752e-05,
      "loss": 1.4837,
      "step": 16765
    },
    {
      "epoch": 0.6004261643418626,
      "grad_norm": 1.9744517803192139,
      "learning_rate": 7.268226688343699e-05,
      "loss": 1.2326,
      "step": 16766
    },
    {
      "epoch": 0.6004619764714308,
      "grad_norm": 1.7462183237075806,
      "learning_rate": 7.267110923791133e-05,
      "loss": 1.5414,
      "step": 16767
    },
    {
      "epoch": 0.6004977886009991,
      "grad_norm": 1.5139094591140747,
      "learning_rate": 7.265995196006062e-05,
      "loss": 1.6375,
      "step": 16768
    },
    {
      "epoch": 0.6005336007305675,
      "grad_norm": 1.9999083280563354,
      "learning_rate": 7.264879505003502e-05,
      "loss": 1.3057,
      "step": 16769
    },
    {
      "epoch": 0.6005694128601358,
      "grad_norm": 1.6535240411758423,
      "learning_rate": 7.263763850798458e-05,
      "loss": 1.7807,
      "step": 16770
    },
    {
      "epoch": 0.600605224989704,
      "grad_norm": 1.6899582147598267,
      "learning_rate": 7.262648233405942e-05,
      "loss": 1.2351,
      "step": 16771
    },
    {
      "epoch": 0.6006410371192723,
      "grad_norm": 1.7519525289535522,
      "learning_rate": 7.261532652840964e-05,
      "loss": 1.4308,
      "step": 16772
    },
    {
      "epoch": 0.6006768492488406,
      "grad_norm": 1.5700947046279907,
      "learning_rate": 7.260417109118531e-05,
      "loss": 1.3495,
      "step": 16773
    },
    {
      "epoch": 0.6007126613784088,
      "grad_norm": 1.5926108360290527,
      "learning_rate": 7.259301602253652e-05,
      "loss": 1.6475,
      "step": 16774
    },
    {
      "epoch": 0.6007484735079771,
      "grad_norm": 1.5498440265655518,
      "learning_rate": 7.258186132261336e-05,
      "loss": 1.6338,
      "step": 16775
    },
    {
      "epoch": 0.6007842856375455,
      "grad_norm": 1.9977786540985107,
      "learning_rate": 7.25707069915659e-05,
      "loss": 1.3661,
      "step": 16776
    },
    {
      "epoch": 0.6008200977671138,
      "grad_norm": 1.361795425415039,
      "learning_rate": 7.255955302954416e-05,
      "loss": 1.2156,
      "step": 16777
    },
    {
      "epoch": 0.600855909896682,
      "grad_norm": 1.5042104721069336,
      "learning_rate": 7.254839943669826e-05,
      "loss": 1.3412,
      "step": 16778
    },
    {
      "epoch": 0.6008917220262503,
      "grad_norm": 1.656468391418457,
      "learning_rate": 7.253724621317822e-05,
      "loss": 1.5451,
      "step": 16779
    },
    {
      "epoch": 0.6009275341558186,
      "grad_norm": 1.2738529443740845,
      "learning_rate": 7.252609335913413e-05,
      "loss": 1.284,
      "step": 16780
    },
    {
      "epoch": 0.6009633462853868,
      "grad_norm": 1.751894235610962,
      "learning_rate": 7.251494087471599e-05,
      "loss": 1.5876,
      "step": 16781
    },
    {
      "epoch": 0.6009991584149551,
      "grad_norm": 1.967529058456421,
      "learning_rate": 7.250378876007389e-05,
      "loss": 1.2286,
      "step": 16782
    },
    {
      "epoch": 0.6010349705445235,
      "grad_norm": 1.7119519710540771,
      "learning_rate": 7.249263701535782e-05,
      "loss": 1.6704,
      "step": 16783
    },
    {
      "epoch": 0.6010707826740918,
      "grad_norm": 2.0444655418395996,
      "learning_rate": 7.248148564071787e-05,
      "loss": 1.3567,
      "step": 16784
    },
    {
      "epoch": 0.60110659480366,
      "grad_norm": 2.1970937252044678,
      "learning_rate": 7.247033463630402e-05,
      "loss": 1.665,
      "step": 16785
    },
    {
      "epoch": 0.6011424069332283,
      "grad_norm": 1.688825249671936,
      "learning_rate": 7.24591840022663e-05,
      "loss": 1.3525,
      "step": 16786
    },
    {
      "epoch": 0.6011782190627966,
      "grad_norm": 1.3591644763946533,
      "learning_rate": 7.244803373875475e-05,
      "loss": 1.4548,
      "step": 16787
    },
    {
      "epoch": 0.6012140311923648,
      "grad_norm": 2.0499813556671143,
      "learning_rate": 7.243688384591934e-05,
      "loss": 1.4286,
      "step": 16788
    },
    {
      "epoch": 0.6012498433219331,
      "grad_norm": 1.7680226564407349,
      "learning_rate": 7.242573432391012e-05,
      "loss": 1.5824,
      "step": 16789
    },
    {
      "epoch": 0.6012856554515015,
      "grad_norm": 1.749014139175415,
      "learning_rate": 7.241458517287708e-05,
      "loss": 1.6337,
      "step": 16790
    },
    {
      "epoch": 0.6013214675810697,
      "grad_norm": 1.4293020963668823,
      "learning_rate": 7.24034363929702e-05,
      "loss": 1.6886,
      "step": 16791
    },
    {
      "epoch": 0.601357279710638,
      "grad_norm": 1.5241018533706665,
      "learning_rate": 7.23922879843395e-05,
      "loss": 1.5131,
      "step": 16792
    },
    {
      "epoch": 0.6013930918402063,
      "grad_norm": 2.091395854949951,
      "learning_rate": 7.238113994713495e-05,
      "loss": 1.5334,
      "step": 16793
    },
    {
      "epoch": 0.6014289039697746,
      "grad_norm": 1.7409805059432983,
      "learning_rate": 7.236999228150654e-05,
      "loss": 1.5341,
      "step": 16794
    },
    {
      "epoch": 0.6014647160993428,
      "grad_norm": 1.6296451091766357,
      "learning_rate": 7.235884498760423e-05,
      "loss": 1.4788,
      "step": 16795
    },
    {
      "epoch": 0.6015005282289111,
      "grad_norm": 1.537999153137207,
      "learning_rate": 7.234769806557802e-05,
      "loss": 1.5781,
      "step": 16796
    },
    {
      "epoch": 0.6015363403584795,
      "grad_norm": 1.5315535068511963,
      "learning_rate": 7.233655151557786e-05,
      "loss": 1.5663,
      "step": 16797
    },
    {
      "epoch": 0.6015721524880477,
      "grad_norm": 1.7069329023361206,
      "learning_rate": 7.232540533775371e-05,
      "loss": 1.4498,
      "step": 16798
    },
    {
      "epoch": 0.601607964617616,
      "grad_norm": 1.808366298675537,
      "learning_rate": 7.231425953225552e-05,
      "loss": 1.7831,
      "step": 16799
    },
    {
      "epoch": 0.6016437767471843,
      "grad_norm": 1.2750335931777954,
      "learning_rate": 7.230311409923329e-05,
      "loss": 1.6704,
      "step": 16800
    },
    {
      "epoch": 0.6016795888767525,
      "grad_norm": 2.135979652404785,
      "learning_rate": 7.22919690388369e-05,
      "loss": 1.5089,
      "step": 16801
    },
    {
      "epoch": 0.6017154010063208,
      "grad_norm": 1.8281000852584839,
      "learning_rate": 7.228082435121636e-05,
      "loss": 1.7816,
      "step": 16802
    },
    {
      "epoch": 0.6017512131358891,
      "grad_norm": 1.9501540660858154,
      "learning_rate": 7.226968003652157e-05,
      "loss": 1.4132,
      "step": 16803
    },
    {
      "epoch": 0.6017870252654575,
      "grad_norm": 2.566105604171753,
      "learning_rate": 7.225853609490244e-05,
      "loss": 1.3759,
      "step": 16804
    },
    {
      "epoch": 0.6018228373950257,
      "grad_norm": 1.4011183977127075,
      "learning_rate": 7.224739252650894e-05,
      "loss": 1.3426,
      "step": 16805
    },
    {
      "epoch": 0.601858649524594,
      "grad_norm": 1.7511323690414429,
      "learning_rate": 7.223624933149095e-05,
      "loss": 1.3261,
      "step": 16806
    },
    {
      "epoch": 0.6018944616541623,
      "grad_norm": 2.664832830429077,
      "learning_rate": 7.222510650999845e-05,
      "loss": 1.573,
      "step": 16807
    },
    {
      "epoch": 0.6019302737837305,
      "grad_norm": 1.4400556087493896,
      "learning_rate": 7.221396406218129e-05,
      "loss": 1.2956,
      "step": 16808
    },
    {
      "epoch": 0.6019660859132988,
      "grad_norm": 1.4470281600952148,
      "learning_rate": 7.220282198818941e-05,
      "loss": 1.5883,
      "step": 16809
    },
    {
      "epoch": 0.6020018980428671,
      "grad_norm": 1.4601818323135376,
      "learning_rate": 7.21916802881727e-05,
      "loss": 1.4075,
      "step": 16810
    },
    {
      "epoch": 0.6020377101724355,
      "grad_norm": 1.3011351823806763,
      "learning_rate": 7.218053896228107e-05,
      "loss": 1.2083,
      "step": 16811
    },
    {
      "epoch": 0.6020735223020037,
      "grad_norm": 2.1279044151306152,
      "learning_rate": 7.216939801066444e-05,
      "loss": 1.5229,
      "step": 16812
    },
    {
      "epoch": 0.602109334431572,
      "grad_norm": 1.5820890665054321,
      "learning_rate": 7.215825743347259e-05,
      "loss": 1.2355,
      "step": 16813
    },
    {
      "epoch": 0.6021451465611403,
      "grad_norm": 1.6486124992370605,
      "learning_rate": 7.214711723085553e-05,
      "loss": 1.3337,
      "step": 16814
    },
    {
      "epoch": 0.6021809586907085,
      "grad_norm": 1.8116123676300049,
      "learning_rate": 7.213597740296304e-05,
      "loss": 1.3902,
      "step": 16815
    },
    {
      "epoch": 0.6022167708202768,
      "grad_norm": 1.5074397325515747,
      "learning_rate": 7.212483794994503e-05,
      "loss": 1.1192,
      "step": 16816
    },
    {
      "epoch": 0.6022525829498451,
      "grad_norm": 1.316300868988037,
      "learning_rate": 7.211369887195139e-05,
      "loss": 1.3241,
      "step": 16817
    },
    {
      "epoch": 0.6022883950794135,
      "grad_norm": 2.0315091609954834,
      "learning_rate": 7.210256016913195e-05,
      "loss": 1.5004,
      "step": 16818
    },
    {
      "epoch": 0.6023242072089817,
      "grad_norm": 1.8143523931503296,
      "learning_rate": 7.209142184163657e-05,
      "loss": 1.6522,
      "step": 16819
    },
    {
      "epoch": 0.60236001933855,
      "grad_norm": 1.3392434120178223,
      "learning_rate": 7.208028388961515e-05,
      "loss": 1.2916,
      "step": 16820
    },
    {
      "epoch": 0.6023958314681183,
      "grad_norm": 1.4817171096801758,
      "learning_rate": 7.206914631321749e-05,
      "loss": 1.3744,
      "step": 16821
    },
    {
      "epoch": 0.6024316435976865,
      "grad_norm": 1.1809767484664917,
      "learning_rate": 7.205800911259338e-05,
      "loss": 1.3433,
      "step": 16822
    },
    {
      "epoch": 0.6024674557272548,
      "grad_norm": 1.8400154113769531,
      "learning_rate": 7.204687228789279e-05,
      "loss": 1.5053,
      "step": 16823
    },
    {
      "epoch": 0.6025032678568231,
      "grad_norm": 1.5287377834320068,
      "learning_rate": 7.20357358392654e-05,
      "loss": 1.6242,
      "step": 16824
    },
    {
      "epoch": 0.6025390799863914,
      "grad_norm": 1.6333818435668945,
      "learning_rate": 7.202459976686118e-05,
      "loss": 1.249,
      "step": 16825
    },
    {
      "epoch": 0.6025748921159597,
      "grad_norm": 1.624626636505127,
      "learning_rate": 7.201346407082982e-05,
      "loss": 1.3627,
      "step": 16826
    },
    {
      "epoch": 0.602610704245528,
      "grad_norm": 1.5464283227920532,
      "learning_rate": 7.200232875132127e-05,
      "loss": 1.652,
      "step": 16827
    },
    {
      "epoch": 0.6026465163750963,
      "grad_norm": 1.678821325302124,
      "learning_rate": 7.199119380848525e-05,
      "loss": 1.4797,
      "step": 16828
    },
    {
      "epoch": 0.6026823285046645,
      "grad_norm": 1.8502992391586304,
      "learning_rate": 7.198005924247155e-05,
      "loss": 1.7994,
      "step": 16829
    },
    {
      "epoch": 0.6027181406342328,
      "grad_norm": 1.5439445972442627,
      "learning_rate": 7.196892505343007e-05,
      "loss": 1.1767,
      "step": 16830
    },
    {
      "epoch": 0.6027539527638011,
      "grad_norm": 1.6171842813491821,
      "learning_rate": 7.195779124151048e-05,
      "loss": 1.3557,
      "step": 16831
    },
    {
      "epoch": 0.6027897648933694,
      "grad_norm": 2.662693500518799,
      "learning_rate": 7.19466578068627e-05,
      "loss": 1.3941,
      "step": 16832
    },
    {
      "epoch": 0.6028255770229377,
      "grad_norm": 1.5511195659637451,
      "learning_rate": 7.193552474963638e-05,
      "loss": 1.5937,
      "step": 16833
    },
    {
      "epoch": 0.602861389152506,
      "grad_norm": 1.55390202999115,
      "learning_rate": 7.192439206998146e-05,
      "loss": 1.432,
      "step": 16834
    },
    {
      "epoch": 0.6028972012820742,
      "grad_norm": 1.8486980199813843,
      "learning_rate": 7.191325976804754e-05,
      "loss": 1.3208,
      "step": 16835
    },
    {
      "epoch": 0.6029330134116425,
      "grad_norm": 2.2025701999664307,
      "learning_rate": 7.190212784398458e-05,
      "loss": 1.6723,
      "step": 16836
    },
    {
      "epoch": 0.6029688255412108,
      "grad_norm": 1.3402276039123535,
      "learning_rate": 7.18909962979422e-05,
      "loss": 1.279,
      "step": 16837
    },
    {
      "epoch": 0.603004637670779,
      "grad_norm": 1.858518362045288,
      "learning_rate": 7.187986513007018e-05,
      "loss": 1.645,
      "step": 16838
    },
    {
      "epoch": 0.6030404498003473,
      "grad_norm": 1.7115235328674316,
      "learning_rate": 7.186873434051832e-05,
      "loss": 1.5733,
      "step": 16839
    },
    {
      "epoch": 0.6030762619299157,
      "grad_norm": 1.9213659763336182,
      "learning_rate": 7.185760392943637e-05,
      "loss": 1.2916,
      "step": 16840
    },
    {
      "epoch": 0.603112074059484,
      "grad_norm": 1.2515188455581665,
      "learning_rate": 7.184647389697405e-05,
      "loss": 1.5181,
      "step": 16841
    },
    {
      "epoch": 0.6031478861890522,
      "grad_norm": 1.9560471773147583,
      "learning_rate": 7.183534424328106e-05,
      "loss": 1.5256,
      "step": 16842
    },
    {
      "epoch": 0.6031836983186205,
      "grad_norm": 1.352102518081665,
      "learning_rate": 7.182421496850726e-05,
      "loss": 1.4133,
      "step": 16843
    },
    {
      "epoch": 0.6032195104481888,
      "grad_norm": 1.5485777854919434,
      "learning_rate": 7.181308607280223e-05,
      "loss": 1.5151,
      "step": 16844
    },
    {
      "epoch": 0.603255322577757,
      "grad_norm": 1.3369213342666626,
      "learning_rate": 7.180195755631584e-05,
      "loss": 1.2579,
      "step": 16845
    },
    {
      "epoch": 0.6032911347073253,
      "grad_norm": 2.329124689102173,
      "learning_rate": 7.179082941919773e-05,
      "loss": 1.3283,
      "step": 16846
    },
    {
      "epoch": 0.6033269468368937,
      "grad_norm": 1.8115071058273315,
      "learning_rate": 7.177970166159758e-05,
      "loss": 1.5473,
      "step": 16847
    },
    {
      "epoch": 0.603362758966462,
      "grad_norm": 1.4668792486190796,
      "learning_rate": 7.176857428366517e-05,
      "loss": 1.108,
      "step": 16848
    },
    {
      "epoch": 0.6033985710960302,
      "grad_norm": 1.9151508808135986,
      "learning_rate": 7.175744728555016e-05,
      "loss": 1.3883,
      "step": 16849
    },
    {
      "epoch": 0.6034343832255985,
      "grad_norm": 1.7193862199783325,
      "learning_rate": 7.174632066740227e-05,
      "loss": 1.9886,
      "step": 16850
    },
    {
      "epoch": 0.6034701953551668,
      "grad_norm": 1.6349965333938599,
      "learning_rate": 7.17351944293712e-05,
      "loss": 1.7273,
      "step": 16851
    },
    {
      "epoch": 0.603506007484735,
      "grad_norm": 1.7464351654052734,
      "learning_rate": 7.172406857160662e-05,
      "loss": 1.6175,
      "step": 16852
    },
    {
      "epoch": 0.6035418196143033,
      "grad_norm": 1.512633204460144,
      "learning_rate": 7.171294309425823e-05,
      "loss": 1.5267,
      "step": 16853
    },
    {
      "epoch": 0.6035776317438717,
      "grad_norm": 1.607930064201355,
      "learning_rate": 7.17018179974757e-05,
      "loss": 1.4871,
      "step": 16854
    },
    {
      "epoch": 0.60361344387344,
      "grad_norm": 1.959700345993042,
      "learning_rate": 7.169069328140872e-05,
      "loss": 1.3557,
      "step": 16855
    },
    {
      "epoch": 0.6036492560030082,
      "grad_norm": 1.5639288425445557,
      "learning_rate": 7.167956894620694e-05,
      "loss": 1.1726,
      "step": 16856
    },
    {
      "epoch": 0.6036850681325765,
      "grad_norm": 1.6049681901931763,
      "learning_rate": 7.166844499202002e-05,
      "loss": 1.658,
      "step": 16857
    },
    {
      "epoch": 0.6037208802621448,
      "grad_norm": 1.4732575416564941,
      "learning_rate": 7.165732141899761e-05,
      "loss": 1.4717,
      "step": 16858
    },
    {
      "epoch": 0.603756692391713,
      "grad_norm": 2.0398313999176025,
      "learning_rate": 7.164619822728941e-05,
      "loss": 1.4927,
      "step": 16859
    },
    {
      "epoch": 0.6037925045212813,
      "grad_norm": 1.3296444416046143,
      "learning_rate": 7.163507541704503e-05,
      "loss": 1.5785,
      "step": 16860
    },
    {
      "epoch": 0.6038283166508497,
      "grad_norm": 1.3880574703216553,
      "learning_rate": 7.162395298841414e-05,
      "loss": 1.5667,
      "step": 16861
    },
    {
      "epoch": 0.603864128780418,
      "grad_norm": 1.4414122104644775,
      "learning_rate": 7.161283094154633e-05,
      "loss": 1.5618,
      "step": 16862
    },
    {
      "epoch": 0.6038999409099862,
      "grad_norm": 2.2014966011047363,
      "learning_rate": 7.160170927659128e-05,
      "loss": 1.4817,
      "step": 16863
    },
    {
      "epoch": 0.6039357530395545,
      "grad_norm": 2.323162078857422,
      "learning_rate": 7.159058799369861e-05,
      "loss": 1.346,
      "step": 16864
    },
    {
      "epoch": 0.6039715651691228,
      "grad_norm": 1.2874650955200195,
      "learning_rate": 7.157946709301791e-05,
      "loss": 0.9955,
      "step": 16865
    },
    {
      "epoch": 0.604007377298691,
      "grad_norm": 1.3906731605529785,
      "learning_rate": 7.156834657469885e-05,
      "loss": 1.5495,
      "step": 16866
    },
    {
      "epoch": 0.6040431894282593,
      "grad_norm": 1.6526511907577515,
      "learning_rate": 7.155722643889097e-05,
      "loss": 1.6115,
      "step": 16867
    },
    {
      "epoch": 0.6040790015578277,
      "grad_norm": 1.9570540189743042,
      "learning_rate": 7.154610668574395e-05,
      "loss": 1.0826,
      "step": 16868
    },
    {
      "epoch": 0.604114813687396,
      "grad_norm": 1.2908234596252441,
      "learning_rate": 7.153498731540735e-05,
      "loss": 1.1568,
      "step": 16869
    },
    {
      "epoch": 0.6041506258169642,
      "grad_norm": 2.093600273132324,
      "learning_rate": 7.15238683280308e-05,
      "loss": 1.6383,
      "step": 16870
    },
    {
      "epoch": 0.6041864379465325,
      "grad_norm": 1.7811310291290283,
      "learning_rate": 7.151274972376383e-05,
      "loss": 1.1835,
      "step": 16871
    },
    {
      "epoch": 0.6042222500761008,
      "grad_norm": 1.4339679479599,
      "learning_rate": 7.15016315027561e-05,
      "loss": 1.1261,
      "step": 16872
    },
    {
      "epoch": 0.604258062205669,
      "grad_norm": 1.8503432273864746,
      "learning_rate": 7.149051366515716e-05,
      "loss": 1.7787,
      "step": 16873
    },
    {
      "epoch": 0.6042938743352373,
      "grad_norm": 1.6346187591552734,
      "learning_rate": 7.147939621111655e-05,
      "loss": 1.658,
      "step": 16874
    },
    {
      "epoch": 0.6043296864648057,
      "grad_norm": 1.4566487073898315,
      "learning_rate": 7.146827914078391e-05,
      "loss": 1.5136,
      "step": 16875
    },
    {
      "epoch": 0.6043654985943739,
      "grad_norm": 1.6601160764694214,
      "learning_rate": 7.145716245430876e-05,
      "loss": 1.5001,
      "step": 16876
    },
    {
      "epoch": 0.6044013107239422,
      "grad_norm": 1.377268671989441,
      "learning_rate": 7.144604615184067e-05,
      "loss": 1.6418,
      "step": 16877
    },
    {
      "epoch": 0.6044371228535105,
      "grad_norm": 1.7275559902191162,
      "learning_rate": 7.143493023352918e-05,
      "loss": 1.5162,
      "step": 16878
    },
    {
      "epoch": 0.6044729349830787,
      "grad_norm": 1.9887871742248535,
      "learning_rate": 7.142381469952388e-05,
      "loss": 1.7494,
      "step": 16879
    },
    {
      "epoch": 0.604508747112647,
      "grad_norm": 1.5621000528335571,
      "learning_rate": 7.141269954997428e-05,
      "loss": 1.5148,
      "step": 16880
    },
    {
      "epoch": 0.6045445592422153,
      "grad_norm": 1.5300337076187134,
      "learning_rate": 7.140158478502995e-05,
      "loss": 1.4512,
      "step": 16881
    },
    {
      "epoch": 0.6045803713717837,
      "grad_norm": 1.5173213481903076,
      "learning_rate": 7.13904704048404e-05,
      "loss": 1.2817,
      "step": 16882
    },
    {
      "epoch": 0.6046161835013519,
      "grad_norm": 1.6534255743026733,
      "learning_rate": 7.137935640955516e-05,
      "loss": 1.5485,
      "step": 16883
    },
    {
      "epoch": 0.6046519956309202,
      "grad_norm": 2.1657660007476807,
      "learning_rate": 7.136824279932378e-05,
      "loss": 1.3278,
      "step": 16884
    },
    {
      "epoch": 0.6046878077604885,
      "grad_norm": 1.8624151945114136,
      "learning_rate": 7.135712957429573e-05,
      "loss": 1.2911,
      "step": 16885
    },
    {
      "epoch": 0.6047236198900567,
      "grad_norm": 2.1512229442596436,
      "learning_rate": 7.134601673462058e-05,
      "loss": 1.5735,
      "step": 16886
    },
    {
      "epoch": 0.604759432019625,
      "grad_norm": 1.6311861276626587,
      "learning_rate": 7.133490428044778e-05,
      "loss": 1.2567,
      "step": 16887
    },
    {
      "epoch": 0.6047952441491933,
      "grad_norm": 1.3144361972808838,
      "learning_rate": 7.132379221192691e-05,
      "loss": 1.1432,
      "step": 16888
    },
    {
      "epoch": 0.6048310562787617,
      "grad_norm": 2.04892635345459,
      "learning_rate": 7.131268052920739e-05,
      "loss": 1.3096,
      "step": 16889
    },
    {
      "epoch": 0.6048668684083299,
      "grad_norm": 1.3833351135253906,
      "learning_rate": 7.130156923243879e-05,
      "loss": 1.5811,
      "step": 16890
    },
    {
      "epoch": 0.6049026805378982,
      "grad_norm": 1.4970839023590088,
      "learning_rate": 7.129045832177054e-05,
      "loss": 1.2673,
      "step": 16891
    },
    {
      "epoch": 0.6049384926674665,
      "grad_norm": 1.5738197565078735,
      "learning_rate": 7.127934779735212e-05,
      "loss": 1.552,
      "step": 16892
    },
    {
      "epoch": 0.6049743047970347,
      "grad_norm": 1.8134737014770508,
      "learning_rate": 7.126823765933306e-05,
      "loss": 1.4541,
      "step": 16893
    },
    {
      "epoch": 0.605010116926603,
      "grad_norm": 1.7841168642044067,
      "learning_rate": 7.125712790786277e-05,
      "loss": 1.3122,
      "step": 16894
    },
    {
      "epoch": 0.6050459290561713,
      "grad_norm": 1.7300609350204468,
      "learning_rate": 7.124601854309077e-05,
      "loss": 1.6758,
      "step": 16895
    },
    {
      "epoch": 0.6050817411857397,
      "grad_norm": 1.904471516609192,
      "learning_rate": 7.123490956516649e-05,
      "loss": 1.5818,
      "step": 16896
    },
    {
      "epoch": 0.6051175533153079,
      "grad_norm": 1.6749107837677002,
      "learning_rate": 7.122380097423941e-05,
      "loss": 1.8399,
      "step": 16897
    },
    {
      "epoch": 0.6051533654448762,
      "grad_norm": 1.8210564851760864,
      "learning_rate": 7.121269277045894e-05,
      "loss": 1.3766,
      "step": 16898
    },
    {
      "epoch": 0.6051891775744445,
      "grad_norm": 1.5455830097198486,
      "learning_rate": 7.120158495397459e-05,
      "loss": 1.7764,
      "step": 16899
    },
    {
      "epoch": 0.6052249897040127,
      "grad_norm": 1.4964015483856201,
      "learning_rate": 7.119047752493576e-05,
      "loss": 1.3216,
      "step": 16900
    },
    {
      "epoch": 0.605260801833581,
      "grad_norm": 1.3908185958862305,
      "learning_rate": 7.117937048349188e-05,
      "loss": 1.5999,
      "step": 16901
    },
    {
      "epoch": 0.6052966139631493,
      "grad_norm": 1.8004734516143799,
      "learning_rate": 7.11682638297924e-05,
      "loss": 1.818,
      "step": 16902
    },
    {
      "epoch": 0.6053324260927176,
      "grad_norm": 1.4848984479904175,
      "learning_rate": 7.115715756398674e-05,
      "loss": 1.5299,
      "step": 16903
    },
    {
      "epoch": 0.6053682382222859,
      "grad_norm": 1.5885624885559082,
      "learning_rate": 7.114605168622432e-05,
      "loss": 1.8183,
      "step": 16904
    },
    {
      "epoch": 0.6054040503518542,
      "grad_norm": 1.8533222675323486,
      "learning_rate": 7.113494619665456e-05,
      "loss": 1.5371,
      "step": 16905
    },
    {
      "epoch": 0.6054398624814225,
      "grad_norm": 2.0348994731903076,
      "learning_rate": 7.112384109542687e-05,
      "loss": 1.1959,
      "step": 16906
    },
    {
      "epoch": 0.6054756746109907,
      "grad_norm": 1.8801383972167969,
      "learning_rate": 7.111273638269063e-05,
      "loss": 1.6672,
      "step": 16907
    },
    {
      "epoch": 0.605511486740559,
      "grad_norm": 2.084892749786377,
      "learning_rate": 7.110163205859528e-05,
      "loss": 1.6724,
      "step": 16908
    },
    {
      "epoch": 0.6055472988701273,
      "grad_norm": 1.7891219854354858,
      "learning_rate": 7.109052812329023e-05,
      "loss": 1.126,
      "step": 16909
    },
    {
      "epoch": 0.6055831109996956,
      "grad_norm": 2.057370901107788,
      "learning_rate": 7.107942457692475e-05,
      "loss": 1.6619,
      "step": 16910
    },
    {
      "epoch": 0.6056189231292639,
      "grad_norm": 1.486271858215332,
      "learning_rate": 7.106832141964839e-05,
      "loss": 1.3337,
      "step": 16911
    },
    {
      "epoch": 0.6056547352588322,
      "grad_norm": 1.9159584045410156,
      "learning_rate": 7.105721865161037e-05,
      "loss": 1.5583,
      "step": 16912
    },
    {
      "epoch": 0.6056905473884004,
      "grad_norm": 1.8747551441192627,
      "learning_rate": 7.104611627296018e-05,
      "loss": 1.3585,
      "step": 16913
    },
    {
      "epoch": 0.6057263595179687,
      "grad_norm": 1.439809799194336,
      "learning_rate": 7.103501428384714e-05,
      "loss": 1.4312,
      "step": 16914
    },
    {
      "epoch": 0.605762171647537,
      "grad_norm": 1.6186916828155518,
      "learning_rate": 7.102391268442062e-05,
      "loss": 1.4321,
      "step": 16915
    },
    {
      "epoch": 0.6057979837771053,
      "grad_norm": 1.4521013498306274,
      "learning_rate": 7.101281147482996e-05,
      "loss": 1.4515,
      "step": 16916
    },
    {
      "epoch": 0.6058337959066736,
      "grad_norm": 1.5437028408050537,
      "learning_rate": 7.100171065522457e-05,
      "loss": 1.3871,
      "step": 16917
    },
    {
      "epoch": 0.6058696080362419,
      "grad_norm": 1.9497326612472534,
      "learning_rate": 7.099061022575377e-05,
      "loss": 1.6164,
      "step": 16918
    },
    {
      "epoch": 0.6059054201658102,
      "grad_norm": 1.3493622541427612,
      "learning_rate": 7.097951018656683e-05,
      "loss": 1.4338,
      "step": 16919
    },
    {
      "epoch": 0.6059412322953784,
      "grad_norm": 1.541733741760254,
      "learning_rate": 7.09684105378132e-05,
      "loss": 1.3511,
      "step": 16920
    },
    {
      "epoch": 0.6059770444249467,
      "grad_norm": 1.4317058324813843,
      "learning_rate": 7.095731127964211e-05,
      "loss": 1.3761,
      "step": 16921
    },
    {
      "epoch": 0.606012856554515,
      "grad_norm": 1.4599525928497314,
      "learning_rate": 7.0946212412203e-05,
      "loss": 1.8323,
      "step": 16922
    },
    {
      "epoch": 0.6060486686840832,
      "grad_norm": 1.6733118295669556,
      "learning_rate": 7.093511393564504e-05,
      "loss": 1.6659,
      "step": 16923
    },
    {
      "epoch": 0.6060844808136516,
      "grad_norm": 1.553957462310791,
      "learning_rate": 7.092401585011771e-05,
      "loss": 1.3088,
      "step": 16924
    },
    {
      "epoch": 0.6061202929432199,
      "grad_norm": 1.6866424083709717,
      "learning_rate": 7.091291815577022e-05,
      "loss": 1.6451,
      "step": 16925
    },
    {
      "epoch": 0.6061561050727882,
      "grad_norm": 1.5042775869369507,
      "learning_rate": 7.090182085275185e-05,
      "loss": 1.5038,
      "step": 16926
    },
    {
      "epoch": 0.6061919172023564,
      "grad_norm": 1.7671167850494385,
      "learning_rate": 7.089072394121201e-05,
      "loss": 1.4282,
      "step": 16927
    },
    {
      "epoch": 0.6062277293319247,
      "grad_norm": 1.7589083909988403,
      "learning_rate": 7.087962742129988e-05,
      "loss": 1.4675,
      "step": 16928
    },
    {
      "epoch": 0.606263541461493,
      "grad_norm": 1.4512827396392822,
      "learning_rate": 7.086853129316484e-05,
      "loss": 1.5013,
      "step": 16929
    },
    {
      "epoch": 0.6062993535910612,
      "grad_norm": 1.6962764263153076,
      "learning_rate": 7.085743555695609e-05,
      "loss": 1.1876,
      "step": 16930
    },
    {
      "epoch": 0.6063351657206296,
      "grad_norm": 2.0834460258483887,
      "learning_rate": 7.084634021282301e-05,
      "loss": 1.6448,
      "step": 16931
    },
    {
      "epoch": 0.6063709778501979,
      "grad_norm": 1.53248131275177,
      "learning_rate": 7.083524526091475e-05,
      "loss": 1.4959,
      "step": 16932
    },
    {
      "epoch": 0.6064067899797662,
      "grad_norm": 1.4516412019729614,
      "learning_rate": 7.082415070138071e-05,
      "loss": 1.6379,
      "step": 16933
    },
    {
      "epoch": 0.6064426021093344,
      "grad_norm": 1.47740638256073,
      "learning_rate": 7.081305653437007e-05,
      "loss": 1.4495,
      "step": 16934
    },
    {
      "epoch": 0.6064784142389027,
      "grad_norm": 1.6300405263900757,
      "learning_rate": 7.080196276003209e-05,
      "loss": 1.4695,
      "step": 16935
    },
    {
      "epoch": 0.606514226368471,
      "grad_norm": 2.4109437465667725,
      "learning_rate": 7.079086937851604e-05,
      "loss": 1.3669,
      "step": 16936
    },
    {
      "epoch": 0.6065500384980392,
      "grad_norm": 1.4852944612503052,
      "learning_rate": 7.077977638997117e-05,
      "loss": 1.5768,
      "step": 16937
    },
    {
      "epoch": 0.6065858506276076,
      "grad_norm": 2.6029889583587646,
      "learning_rate": 7.076868379454673e-05,
      "loss": 1.6087,
      "step": 16938
    },
    {
      "epoch": 0.6066216627571759,
      "grad_norm": 1.5269404649734497,
      "learning_rate": 7.07575915923919e-05,
      "loss": 1.405,
      "step": 16939
    },
    {
      "epoch": 0.6066574748867442,
      "grad_norm": 1.8272013664245605,
      "learning_rate": 7.074649978365602e-05,
      "loss": 1.2481,
      "step": 16940
    },
    {
      "epoch": 0.6066932870163124,
      "grad_norm": 1.7521268129348755,
      "learning_rate": 7.073540836848817e-05,
      "loss": 1.5097,
      "step": 16941
    },
    {
      "epoch": 0.6067290991458807,
      "grad_norm": 2.199578046798706,
      "learning_rate": 7.072431734703772e-05,
      "loss": 1.7481,
      "step": 16942
    },
    {
      "epoch": 0.606764911275449,
      "grad_norm": 1.9226924180984497,
      "learning_rate": 7.071322671945382e-05,
      "loss": 1.5111,
      "step": 16943
    },
    {
      "epoch": 0.6068007234050172,
      "grad_norm": 1.5912048816680908,
      "learning_rate": 7.070213648588564e-05,
      "loss": 1.5916,
      "step": 16944
    },
    {
      "epoch": 0.6068365355345856,
      "grad_norm": 1.532690167427063,
      "learning_rate": 7.069104664648244e-05,
      "loss": 1.4422,
      "step": 16945
    },
    {
      "epoch": 0.6068723476641539,
      "grad_norm": 2.0913636684417725,
      "learning_rate": 7.06799572013934e-05,
      "loss": 1.6614,
      "step": 16946
    },
    {
      "epoch": 0.6069081597937221,
      "grad_norm": 1.6115814447402954,
      "learning_rate": 7.066886815076771e-05,
      "loss": 1.3298,
      "step": 16947
    },
    {
      "epoch": 0.6069439719232904,
      "grad_norm": 1.4363234043121338,
      "learning_rate": 7.065777949475456e-05,
      "loss": 1.3889,
      "step": 16948
    },
    {
      "epoch": 0.6069797840528587,
      "grad_norm": 1.3749061822891235,
      "learning_rate": 7.064669123350316e-05,
      "loss": 1.3767,
      "step": 16949
    },
    {
      "epoch": 0.607015596182427,
      "grad_norm": 1.6005500555038452,
      "learning_rate": 7.063560336716263e-05,
      "loss": 1.4588,
      "step": 16950
    },
    {
      "epoch": 0.6070514083119952,
      "grad_norm": 1.9682674407958984,
      "learning_rate": 7.062451589588221e-05,
      "loss": 1.2503,
      "step": 16951
    },
    {
      "epoch": 0.6070872204415636,
      "grad_norm": 1.6852772235870361,
      "learning_rate": 7.061342881981105e-05,
      "loss": 1.4176,
      "step": 16952
    },
    {
      "epoch": 0.6071230325711319,
      "grad_norm": 1.2630378007888794,
      "learning_rate": 7.060234213909826e-05,
      "loss": 1.5105,
      "step": 16953
    },
    {
      "epoch": 0.6071588447007001,
      "grad_norm": 1.6143516302108765,
      "learning_rate": 7.059125585389306e-05,
      "loss": 1.5111,
      "step": 16954
    },
    {
      "epoch": 0.6071946568302684,
      "grad_norm": 1.6906598806381226,
      "learning_rate": 7.058016996434455e-05,
      "loss": 1.713,
      "step": 16955
    },
    {
      "epoch": 0.6072304689598367,
      "grad_norm": 1.9878149032592773,
      "learning_rate": 7.056908447060195e-05,
      "loss": 1.4993,
      "step": 16956
    },
    {
      "epoch": 0.607266281089405,
      "grad_norm": 1.3433419466018677,
      "learning_rate": 7.055799937281432e-05,
      "loss": 1.6973,
      "step": 16957
    },
    {
      "epoch": 0.6073020932189732,
      "grad_norm": 1.7787431478500366,
      "learning_rate": 7.054691467113085e-05,
      "loss": 1.4692,
      "step": 16958
    },
    {
      "epoch": 0.6073379053485416,
      "grad_norm": 1.5732979774475098,
      "learning_rate": 7.053583036570064e-05,
      "loss": 1.3024,
      "step": 16959
    },
    {
      "epoch": 0.6073737174781099,
      "grad_norm": 2.0106754302978516,
      "learning_rate": 7.052474645667283e-05,
      "loss": 1.637,
      "step": 16960
    },
    {
      "epoch": 0.6074095296076781,
      "grad_norm": 1.5223397016525269,
      "learning_rate": 7.051366294419655e-05,
      "loss": 1.3299,
      "step": 16961
    },
    {
      "epoch": 0.6074453417372464,
      "grad_norm": 1.6376421451568604,
      "learning_rate": 7.050257982842088e-05,
      "loss": 1.5557,
      "step": 16962
    },
    {
      "epoch": 0.6074811538668147,
      "grad_norm": 1.5301560163497925,
      "learning_rate": 7.049149710949497e-05,
      "loss": 1.5164,
      "step": 16963
    },
    {
      "epoch": 0.6075169659963829,
      "grad_norm": 1.3848516941070557,
      "learning_rate": 7.048041478756786e-05,
      "loss": 1.5645,
      "step": 16964
    },
    {
      "epoch": 0.6075527781259512,
      "grad_norm": 2.091106653213501,
      "learning_rate": 7.046933286278874e-05,
      "loss": 1.2488,
      "step": 16965
    },
    {
      "epoch": 0.6075885902555196,
      "grad_norm": 1.3417564630508423,
      "learning_rate": 7.04582513353066e-05,
      "loss": 1.3523,
      "step": 16966
    },
    {
      "epoch": 0.6076244023850879,
      "grad_norm": 1.6278449296951294,
      "learning_rate": 7.044717020527065e-05,
      "loss": 1.6019,
      "step": 16967
    },
    {
      "epoch": 0.6076602145146561,
      "grad_norm": 1.618781566619873,
      "learning_rate": 7.043608947282985e-05,
      "loss": 1.5129,
      "step": 16968
    },
    {
      "epoch": 0.6076960266442244,
      "grad_norm": 1.8264074325561523,
      "learning_rate": 7.042500913813337e-05,
      "loss": 1.6284,
      "step": 16969
    },
    {
      "epoch": 0.6077318387737927,
      "grad_norm": 1.528173565864563,
      "learning_rate": 7.041392920133024e-05,
      "loss": 1.384,
      "step": 16970
    },
    {
      "epoch": 0.6077676509033609,
      "grad_norm": 1.6601237058639526,
      "learning_rate": 7.040284966256949e-05,
      "loss": 1.5068,
      "step": 16971
    },
    {
      "epoch": 0.6078034630329292,
      "grad_norm": 1.6839256286621094,
      "learning_rate": 7.039177052200026e-05,
      "loss": 1.4348,
      "step": 16972
    },
    {
      "epoch": 0.6078392751624976,
      "grad_norm": 1.93649160861969,
      "learning_rate": 7.038069177977153e-05,
      "loss": 1.2951,
      "step": 16973
    },
    {
      "epoch": 0.6078750872920659,
      "grad_norm": 2.453674077987671,
      "learning_rate": 7.036961343603243e-05,
      "loss": 1.6723,
      "step": 16974
    },
    {
      "epoch": 0.6079108994216341,
      "grad_norm": 1.814399242401123,
      "learning_rate": 7.035853549093192e-05,
      "loss": 1.5456,
      "step": 16975
    },
    {
      "epoch": 0.6079467115512024,
      "grad_norm": 1.8552604913711548,
      "learning_rate": 7.034745794461912e-05,
      "loss": 1.6286,
      "step": 16976
    },
    {
      "epoch": 0.6079825236807707,
      "grad_norm": 1.676199197769165,
      "learning_rate": 7.033638079724298e-05,
      "loss": 1.4652,
      "step": 16977
    },
    {
      "epoch": 0.6080183358103389,
      "grad_norm": 2.279174566268921,
      "learning_rate": 7.032530404895262e-05,
      "loss": 1.6084,
      "step": 16978
    },
    {
      "epoch": 0.6080541479399072,
      "grad_norm": 1.4861416816711426,
      "learning_rate": 7.0314227699897e-05,
      "loss": 1.3948,
      "step": 16979
    },
    {
      "epoch": 0.6080899600694756,
      "grad_norm": 1.480503797531128,
      "learning_rate": 7.030315175022513e-05,
      "loss": 1.322,
      "step": 16980
    },
    {
      "epoch": 0.6081257721990438,
      "grad_norm": 1.4660117626190186,
      "learning_rate": 7.029207620008606e-05,
      "loss": 1.6761,
      "step": 16981
    },
    {
      "epoch": 0.6081615843286121,
      "grad_norm": 1.395601749420166,
      "learning_rate": 7.028100104962878e-05,
      "loss": 1.3519,
      "step": 16982
    },
    {
      "epoch": 0.6081973964581804,
      "grad_norm": 1.7928545475006104,
      "learning_rate": 7.026992629900232e-05,
      "loss": 1.4069,
      "step": 16983
    },
    {
      "epoch": 0.6082332085877487,
      "grad_norm": 1.6177374124526978,
      "learning_rate": 7.025885194835562e-05,
      "loss": 1.3573,
      "step": 16984
    },
    {
      "epoch": 0.6082690207173169,
      "grad_norm": 2.3778645992279053,
      "learning_rate": 7.024777799783774e-05,
      "loss": 1.6212,
      "step": 16985
    },
    {
      "epoch": 0.6083048328468852,
      "grad_norm": 1.5643190145492554,
      "learning_rate": 7.02367044475976e-05,
      "loss": 1.2325,
      "step": 16986
    },
    {
      "epoch": 0.6083406449764536,
      "grad_norm": 1.7111154794692993,
      "learning_rate": 7.022563129778422e-05,
      "loss": 1.6208,
      "step": 16987
    },
    {
      "epoch": 0.6083764571060218,
      "grad_norm": 2.059918165206909,
      "learning_rate": 7.021455854854657e-05,
      "loss": 1.5891,
      "step": 16988
    },
    {
      "epoch": 0.6084122692355901,
      "grad_norm": 1.2656056880950928,
      "learning_rate": 7.020348620003361e-05,
      "loss": 1.5974,
      "step": 16989
    },
    {
      "epoch": 0.6084480813651584,
      "grad_norm": 1.4218404293060303,
      "learning_rate": 7.019241425239432e-05,
      "loss": 1.4523,
      "step": 16990
    },
    {
      "epoch": 0.6084838934947266,
      "grad_norm": 1.8264952898025513,
      "learning_rate": 7.018134270577761e-05,
      "loss": 1.4599,
      "step": 16991
    },
    {
      "epoch": 0.6085197056242949,
      "grad_norm": 1.4522677659988403,
      "learning_rate": 7.017027156033252e-05,
      "loss": 1.7525,
      "step": 16992
    },
    {
      "epoch": 0.6085555177538632,
      "grad_norm": 1.762906789779663,
      "learning_rate": 7.01592008162079e-05,
      "loss": 1.3414,
      "step": 16993
    },
    {
      "epoch": 0.6085913298834316,
      "grad_norm": 1.9561679363250732,
      "learning_rate": 7.014813047355277e-05,
      "loss": 1.4562,
      "step": 16994
    },
    {
      "epoch": 0.6086271420129998,
      "grad_norm": 2.302292823791504,
      "learning_rate": 7.013706053251603e-05,
      "loss": 1.5998,
      "step": 16995
    },
    {
      "epoch": 0.6086629541425681,
      "grad_norm": 1.6201292276382446,
      "learning_rate": 7.012599099324662e-05,
      "loss": 1.3422,
      "step": 16996
    },
    {
      "epoch": 0.6086987662721364,
      "grad_norm": 1.6970531940460205,
      "learning_rate": 7.011492185589349e-05,
      "loss": 1.5323,
      "step": 16997
    },
    {
      "epoch": 0.6087345784017046,
      "grad_norm": 1.6018362045288086,
      "learning_rate": 7.01038531206055e-05,
      "loss": 1.5924,
      "step": 16998
    },
    {
      "epoch": 0.6087703905312729,
      "grad_norm": 1.6218585968017578,
      "learning_rate": 7.009278478753162e-05,
      "loss": 1.3095,
      "step": 16999
    },
    {
      "epoch": 0.6088062026608412,
      "grad_norm": 1.3571292161941528,
      "learning_rate": 7.008171685682074e-05,
      "loss": 1.1314,
      "step": 17000
    },
    {
      "epoch": 0.6088420147904096,
      "grad_norm": 2.037933588027954,
      "learning_rate": 7.007064932862178e-05,
      "loss": 1.467,
      "step": 17001
    },
    {
      "epoch": 0.6088778269199778,
      "grad_norm": 1.4938690662384033,
      "learning_rate": 7.005958220308362e-05,
      "loss": 1.5368,
      "step": 17002
    },
    {
      "epoch": 0.6089136390495461,
      "grad_norm": 2.2486536502838135,
      "learning_rate": 7.004851548035516e-05,
      "loss": 1.234,
      "step": 17003
    },
    {
      "epoch": 0.6089494511791144,
      "grad_norm": 1.8166847229003906,
      "learning_rate": 7.003744916058528e-05,
      "loss": 1.2638,
      "step": 17004
    },
    {
      "epoch": 0.6089852633086826,
      "grad_norm": 1.850803017616272,
      "learning_rate": 7.00263832439229e-05,
      "loss": 1.3371,
      "step": 17005
    },
    {
      "epoch": 0.6090210754382509,
      "grad_norm": 2.1468968391418457,
      "learning_rate": 7.001531773051688e-05,
      "loss": 1.1834,
      "step": 17006
    },
    {
      "epoch": 0.6090568875678192,
      "grad_norm": 1.7560291290283203,
      "learning_rate": 7.000425262051602e-05,
      "loss": 1.7774,
      "step": 17007
    },
    {
      "epoch": 0.6090926996973876,
      "grad_norm": 1.929193139076233,
      "learning_rate": 6.999318791406931e-05,
      "loss": 1.3495,
      "step": 17008
    },
    {
      "epoch": 0.6091285118269558,
      "grad_norm": 1.4060529470443726,
      "learning_rate": 6.998212361132549e-05,
      "loss": 1.3532,
      "step": 17009
    },
    {
      "epoch": 0.6091643239565241,
      "grad_norm": 1.2726495265960693,
      "learning_rate": 6.997105971243352e-05,
      "loss": 1.5055,
      "step": 17010
    },
    {
      "epoch": 0.6092001360860924,
      "grad_norm": 1.7176103591918945,
      "learning_rate": 6.995999621754219e-05,
      "loss": 1.3575,
      "step": 17011
    },
    {
      "epoch": 0.6092359482156606,
      "grad_norm": 1.3885085582733154,
      "learning_rate": 6.994893312680037e-05,
      "loss": 1.5756,
      "step": 17012
    },
    {
      "epoch": 0.6092717603452289,
      "grad_norm": 1.7695868015289307,
      "learning_rate": 6.99378704403569e-05,
      "loss": 1.5509,
      "step": 17013
    },
    {
      "epoch": 0.6093075724747972,
      "grad_norm": 1.2675243616104126,
      "learning_rate": 6.99268081583606e-05,
      "loss": 1.6999,
      "step": 17014
    },
    {
      "epoch": 0.6093433846043655,
      "grad_norm": 1.3948224782943726,
      "learning_rate": 6.991574628096033e-05,
      "loss": 1.4275,
      "step": 17015
    },
    {
      "epoch": 0.6093791967339338,
      "grad_norm": 2.945615530014038,
      "learning_rate": 6.990468480830482e-05,
      "loss": 1.319,
      "step": 17016
    },
    {
      "epoch": 0.6094150088635021,
      "grad_norm": 2.0076494216918945,
      "learning_rate": 6.989362374054302e-05,
      "loss": 1.4243,
      "step": 17017
    },
    {
      "epoch": 0.6094508209930704,
      "grad_norm": 2.0113306045532227,
      "learning_rate": 6.988256307782363e-05,
      "loss": 1.3713,
      "step": 17018
    },
    {
      "epoch": 0.6094866331226386,
      "grad_norm": 1.3602279424667358,
      "learning_rate": 6.987150282029555e-05,
      "loss": 1.3158,
      "step": 17019
    },
    {
      "epoch": 0.6095224452522069,
      "grad_norm": 2.0649795532226562,
      "learning_rate": 6.986044296810749e-05,
      "loss": 1.5799,
      "step": 17020
    },
    {
      "epoch": 0.6095582573817752,
      "grad_norm": 1.709782600402832,
      "learning_rate": 6.984938352140835e-05,
      "loss": 1.7156,
      "step": 17021
    },
    {
      "epoch": 0.6095940695113435,
      "grad_norm": 1.9718817472457886,
      "learning_rate": 6.983832448034684e-05,
      "loss": 1.3973,
      "step": 17022
    },
    {
      "epoch": 0.6096298816409118,
      "grad_norm": 1.8555179834365845,
      "learning_rate": 6.982726584507173e-05,
      "loss": 1.3785,
      "step": 17023
    },
    {
      "epoch": 0.6096656937704801,
      "grad_norm": 1.518578052520752,
      "learning_rate": 6.981620761573188e-05,
      "loss": 1.5094,
      "step": 17024
    },
    {
      "epoch": 0.6097015059000483,
      "grad_norm": 1.679343342781067,
      "learning_rate": 6.980514979247599e-05,
      "loss": 1.6682,
      "step": 17025
    },
    {
      "epoch": 0.6097373180296166,
      "grad_norm": 1.5894368886947632,
      "learning_rate": 6.979409237545291e-05,
      "loss": 1.3674,
      "step": 17026
    },
    {
      "epoch": 0.6097731301591849,
      "grad_norm": 1.8799184560775757,
      "learning_rate": 6.97830353648113e-05,
      "loss": 1.6178,
      "step": 17027
    },
    {
      "epoch": 0.6098089422887532,
      "grad_norm": 2.117532968521118,
      "learning_rate": 6.977197876070003e-05,
      "loss": 1.3589,
      "step": 17028
    },
    {
      "epoch": 0.6098447544183215,
      "grad_norm": 1.57524573802948,
      "learning_rate": 6.976092256326772e-05,
      "loss": 1.3024,
      "step": 17029
    },
    {
      "epoch": 0.6098805665478898,
      "grad_norm": 1.9387460947036743,
      "learning_rate": 6.974986677266326e-05,
      "loss": 1.4434,
      "step": 17030
    },
    {
      "epoch": 0.6099163786774581,
      "grad_norm": 2.020622730255127,
      "learning_rate": 6.973881138903531e-05,
      "loss": 1.5545,
      "step": 17031
    },
    {
      "epoch": 0.6099521908070263,
      "grad_norm": 1.553113341331482,
      "learning_rate": 6.972775641253259e-05,
      "loss": 1.2865,
      "step": 17032
    },
    {
      "epoch": 0.6099880029365946,
      "grad_norm": 1.6942930221557617,
      "learning_rate": 6.971670184330389e-05,
      "loss": 1.4666,
      "step": 17033
    },
    {
      "epoch": 0.6100238150661629,
      "grad_norm": 1.5093879699707031,
      "learning_rate": 6.970564768149788e-05,
      "loss": 1.3654,
      "step": 17034
    },
    {
      "epoch": 0.6100596271957311,
      "grad_norm": 1.5208278894424438,
      "learning_rate": 6.969459392726331e-05,
      "loss": 1.4862,
      "step": 17035
    },
    {
      "epoch": 0.6100954393252995,
      "grad_norm": 2.0230493545532227,
      "learning_rate": 6.968354058074887e-05,
      "loss": 1.2142,
      "step": 17036
    },
    {
      "epoch": 0.6101312514548678,
      "grad_norm": 1.8557683229446411,
      "learning_rate": 6.967248764210333e-05,
      "loss": 1.2707,
      "step": 17037
    },
    {
      "epoch": 0.6101670635844361,
      "grad_norm": 2.024885654449463,
      "learning_rate": 6.966143511147529e-05,
      "loss": 1.4075,
      "step": 17038
    },
    {
      "epoch": 0.6102028757140043,
      "grad_norm": 1.9817204475402832,
      "learning_rate": 6.965038298901356e-05,
      "loss": 1.39,
      "step": 17039
    },
    {
      "epoch": 0.6102386878435726,
      "grad_norm": 3.7637715339660645,
      "learning_rate": 6.963933127486677e-05,
      "loss": 1.4094,
      "step": 17040
    },
    {
      "epoch": 0.6102744999731409,
      "grad_norm": 1.5731815099716187,
      "learning_rate": 6.96282799691836e-05,
      "loss": 1.4704,
      "step": 17041
    },
    {
      "epoch": 0.6103103121027091,
      "grad_norm": 1.9939146041870117,
      "learning_rate": 6.961722907211277e-05,
      "loss": 1.4589,
      "step": 17042
    },
    {
      "epoch": 0.6103461242322775,
      "grad_norm": 1.8387165069580078,
      "learning_rate": 6.96061785838029e-05,
      "loss": 1.3256,
      "step": 17043
    },
    {
      "epoch": 0.6103819363618458,
      "grad_norm": 1.8265881538391113,
      "learning_rate": 6.95951285044027e-05,
      "loss": 1.5793,
      "step": 17044
    },
    {
      "epoch": 0.6104177484914141,
      "grad_norm": 1.364792823791504,
      "learning_rate": 6.958407883406082e-05,
      "loss": 1.5501,
      "step": 17045
    },
    {
      "epoch": 0.6104535606209823,
      "grad_norm": 1.4304333925247192,
      "learning_rate": 6.957302957292596e-05,
      "loss": 1.3732,
      "step": 17046
    },
    {
      "epoch": 0.6104893727505506,
      "grad_norm": 1.5327463150024414,
      "learning_rate": 6.956198072114669e-05,
      "loss": 1.3424,
      "step": 17047
    },
    {
      "epoch": 0.6105251848801189,
      "grad_norm": 1.5442783832550049,
      "learning_rate": 6.955093227887175e-05,
      "loss": 1.3564,
      "step": 17048
    },
    {
      "epoch": 0.6105609970096871,
      "grad_norm": 1.7592949867248535,
      "learning_rate": 6.953988424624973e-05,
      "loss": 1.5495,
      "step": 17049
    },
    {
      "epoch": 0.6105968091392555,
      "grad_norm": 1.6681785583496094,
      "learning_rate": 6.952883662342926e-05,
      "loss": 1.3294,
      "step": 17050
    },
    {
      "epoch": 0.6106326212688238,
      "grad_norm": 2.178701639175415,
      "learning_rate": 6.9517789410559e-05,
      "loss": 1.6474,
      "step": 17051
    },
    {
      "epoch": 0.610668433398392,
      "grad_norm": 1.7608951330184937,
      "learning_rate": 6.950674260778755e-05,
      "loss": 1.4254,
      "step": 17052
    },
    {
      "epoch": 0.6107042455279603,
      "grad_norm": 1.4168155193328857,
      "learning_rate": 6.949569621526357e-05,
      "loss": 1.4225,
      "step": 17053
    },
    {
      "epoch": 0.6107400576575286,
      "grad_norm": 1.6277146339416504,
      "learning_rate": 6.948465023313562e-05,
      "loss": 1.4284,
      "step": 17054
    },
    {
      "epoch": 0.6107758697870969,
      "grad_norm": 1.4581594467163086,
      "learning_rate": 6.947360466155237e-05,
      "loss": 1.4763,
      "step": 17055
    },
    {
      "epoch": 0.6108116819166651,
      "grad_norm": 2.682927370071411,
      "learning_rate": 6.946255950066236e-05,
      "loss": 1.6454,
      "step": 17056
    },
    {
      "epoch": 0.6108474940462335,
      "grad_norm": 1.8958370685577393,
      "learning_rate": 6.945151475061425e-05,
      "loss": 1.5278,
      "step": 17057
    },
    {
      "epoch": 0.6108833061758018,
      "grad_norm": 2.073317289352417,
      "learning_rate": 6.944047041155662e-05,
      "loss": 1.7897,
      "step": 17058
    },
    {
      "epoch": 0.61091911830537,
      "grad_norm": 1.4627833366394043,
      "learning_rate": 6.9429426483638e-05,
      "loss": 1.3666,
      "step": 17059
    },
    {
      "epoch": 0.6109549304349383,
      "grad_norm": 1.4396252632141113,
      "learning_rate": 6.941838296700703e-05,
      "loss": 1.3275,
      "step": 17060
    },
    {
      "epoch": 0.6109907425645066,
      "grad_norm": 1.7097407579421997,
      "learning_rate": 6.940733986181226e-05,
      "loss": 1.4636,
      "step": 17061
    },
    {
      "epoch": 0.6110265546940749,
      "grad_norm": 2.193366765975952,
      "learning_rate": 6.939629716820229e-05,
      "loss": 1.3695,
      "step": 17062
    },
    {
      "epoch": 0.6110623668236431,
      "grad_norm": 1.5271308422088623,
      "learning_rate": 6.938525488632563e-05,
      "loss": 1.4105,
      "step": 17063
    },
    {
      "epoch": 0.6110981789532115,
      "grad_norm": 1.4365044832229614,
      "learning_rate": 6.937421301633091e-05,
      "loss": 1.2441,
      "step": 17064
    },
    {
      "epoch": 0.6111339910827798,
      "grad_norm": 2.1028079986572266,
      "learning_rate": 6.936317155836664e-05,
      "loss": 1.3424,
      "step": 17065
    },
    {
      "epoch": 0.611169803212348,
      "grad_norm": 1.867521047592163,
      "learning_rate": 6.935213051258138e-05,
      "loss": 1.3437,
      "step": 17066
    },
    {
      "epoch": 0.6112056153419163,
      "grad_norm": 1.709039330482483,
      "learning_rate": 6.934108987912369e-05,
      "loss": 1.5545,
      "step": 17067
    },
    {
      "epoch": 0.6112414274714846,
      "grad_norm": 1.8121676445007324,
      "learning_rate": 6.933004965814205e-05,
      "loss": 1.7838,
      "step": 17068
    },
    {
      "epoch": 0.6112772396010528,
      "grad_norm": 1.734595537185669,
      "learning_rate": 6.931900984978506e-05,
      "loss": 1.3001,
      "step": 17069
    },
    {
      "epoch": 0.6113130517306211,
      "grad_norm": 1.3379672765731812,
      "learning_rate": 6.930797045420119e-05,
      "loss": 1.4297,
      "step": 17070
    },
    {
      "epoch": 0.6113488638601895,
      "grad_norm": 1.8500645160675049,
      "learning_rate": 6.929693147153902e-05,
      "loss": 1.3034,
      "step": 17071
    },
    {
      "epoch": 0.6113846759897578,
      "grad_norm": 1.3984006643295288,
      "learning_rate": 6.9285892901947e-05,
      "loss": 1.4553,
      "step": 17072
    },
    {
      "epoch": 0.611420488119326,
      "grad_norm": 1.7027636766433716,
      "learning_rate": 6.927485474557369e-05,
      "loss": 1.6593,
      "step": 17073
    },
    {
      "epoch": 0.6114563002488943,
      "grad_norm": 1.6550413370132446,
      "learning_rate": 6.926381700256757e-05,
      "loss": 1.5719,
      "step": 17074
    },
    {
      "epoch": 0.6114921123784626,
      "grad_norm": 1.6176135540008545,
      "learning_rate": 6.925277967307717e-05,
      "loss": 1.7674,
      "step": 17075
    },
    {
      "epoch": 0.6115279245080308,
      "grad_norm": 1.5670104026794434,
      "learning_rate": 6.924174275725094e-05,
      "loss": 1.5544,
      "step": 17076
    },
    {
      "epoch": 0.6115637366375991,
      "grad_norm": 1.4483779668807983,
      "learning_rate": 6.923070625523737e-05,
      "loss": 1.6666,
      "step": 17077
    },
    {
      "epoch": 0.6115995487671675,
      "grad_norm": 1.5518686771392822,
      "learning_rate": 6.921967016718499e-05,
      "loss": 1.15,
      "step": 17078
    },
    {
      "epoch": 0.6116353608967358,
      "grad_norm": 1.4891283512115479,
      "learning_rate": 6.920863449324221e-05,
      "loss": 1.4193,
      "step": 17079
    },
    {
      "epoch": 0.611671173026304,
      "grad_norm": 1.8726311922073364,
      "learning_rate": 6.919759923355756e-05,
      "loss": 1.4929,
      "step": 17080
    },
    {
      "epoch": 0.6117069851558723,
      "grad_norm": 2.429152488708496,
      "learning_rate": 6.918656438827946e-05,
      "loss": 1.3696,
      "step": 17081
    },
    {
      "epoch": 0.6117427972854406,
      "grad_norm": 1.599990725517273,
      "learning_rate": 6.917552995755641e-05,
      "loss": 1.4851,
      "step": 17082
    },
    {
      "epoch": 0.6117786094150088,
      "grad_norm": 1.8051741123199463,
      "learning_rate": 6.916449594153682e-05,
      "loss": 1.43,
      "step": 17083
    },
    {
      "epoch": 0.6118144215445771,
      "grad_norm": 1.5804451704025269,
      "learning_rate": 6.915346234036919e-05,
      "loss": 0.9616,
      "step": 17084
    },
    {
      "epoch": 0.6118502336741455,
      "grad_norm": 1.3752074241638184,
      "learning_rate": 6.914242915420193e-05,
      "loss": 1.2154,
      "step": 17085
    },
    {
      "epoch": 0.6118860458037138,
      "grad_norm": 1.573339581489563,
      "learning_rate": 6.913139638318346e-05,
      "loss": 1.3455,
      "step": 17086
    },
    {
      "epoch": 0.611921857933282,
      "grad_norm": 1.6640084981918335,
      "learning_rate": 6.912036402746227e-05,
      "loss": 1.3296,
      "step": 17087
    },
    {
      "epoch": 0.6119576700628503,
      "grad_norm": 1.7510104179382324,
      "learning_rate": 6.910933208718671e-05,
      "loss": 1.3694,
      "step": 17088
    },
    {
      "epoch": 0.6119934821924186,
      "grad_norm": 1.2308443784713745,
      "learning_rate": 6.909830056250527e-05,
      "loss": 1.4283,
      "step": 17089
    },
    {
      "epoch": 0.6120292943219868,
      "grad_norm": 1.7887136936187744,
      "learning_rate": 6.908726945356632e-05,
      "loss": 1.4801,
      "step": 17090
    },
    {
      "epoch": 0.6120651064515551,
      "grad_norm": 1.6139098405838013,
      "learning_rate": 6.90762387605183e-05,
      "loss": 1.6002,
      "step": 17091
    },
    {
      "epoch": 0.6121009185811235,
      "grad_norm": 1.8948484659194946,
      "learning_rate": 6.906520848350957e-05,
      "loss": 1.4782,
      "step": 17092
    },
    {
      "epoch": 0.6121367307106917,
      "grad_norm": 1.8423246145248413,
      "learning_rate": 6.905417862268859e-05,
      "loss": 1.4955,
      "step": 17093
    },
    {
      "epoch": 0.61217254284026,
      "grad_norm": 1.6913235187530518,
      "learning_rate": 6.904314917820371e-05,
      "loss": 1.3755,
      "step": 17094
    },
    {
      "epoch": 0.6122083549698283,
      "grad_norm": 1.9365330934524536,
      "learning_rate": 6.90321201502033e-05,
      "loss": 1.5667,
      "step": 17095
    },
    {
      "epoch": 0.6122441670993966,
      "grad_norm": 2.045860767364502,
      "learning_rate": 6.90210915388358e-05,
      "loss": 1.4949,
      "step": 17096
    },
    {
      "epoch": 0.6122799792289648,
      "grad_norm": 1.4850900173187256,
      "learning_rate": 6.901006334424953e-05,
      "loss": 1.2754,
      "step": 17097
    },
    {
      "epoch": 0.6123157913585331,
      "grad_norm": 1.5501368045806885,
      "learning_rate": 6.89990355665929e-05,
      "loss": 1.3983,
      "step": 17098
    },
    {
      "epoch": 0.6123516034881015,
      "grad_norm": 1.471760630607605,
      "learning_rate": 6.898800820601425e-05,
      "loss": 1.4922,
      "step": 17099
    },
    {
      "epoch": 0.6123874156176697,
      "grad_norm": 1.7545149326324463,
      "learning_rate": 6.897698126266197e-05,
      "loss": 1.83,
      "step": 17100
    },
    {
      "epoch": 0.612423227747238,
      "grad_norm": 1.2826873064041138,
      "learning_rate": 6.896595473668435e-05,
      "loss": 1.589,
      "step": 17101
    },
    {
      "epoch": 0.6124590398768063,
      "grad_norm": 1.2966712713241577,
      "learning_rate": 6.89549286282298e-05,
      "loss": 1.1177,
      "step": 17102
    },
    {
      "epoch": 0.6124948520063745,
      "grad_norm": 1.3103042840957642,
      "learning_rate": 6.894390293744668e-05,
      "loss": 1.4203,
      "step": 17103
    },
    {
      "epoch": 0.6125306641359428,
      "grad_norm": 1.613857388496399,
      "learning_rate": 6.893287766448321e-05,
      "loss": 1.4521,
      "step": 17104
    },
    {
      "epoch": 0.6125664762655111,
      "grad_norm": 1.5288715362548828,
      "learning_rate": 6.892185280948786e-05,
      "loss": 1.3655,
      "step": 17105
    },
    {
      "epoch": 0.6126022883950795,
      "grad_norm": 1.2962150573730469,
      "learning_rate": 6.891082837260885e-05,
      "loss": 1.3534,
      "step": 17106
    },
    {
      "epoch": 0.6126381005246477,
      "grad_norm": 1.786429524421692,
      "learning_rate": 6.889980435399456e-05,
      "loss": 1.8125,
      "step": 17107
    },
    {
      "epoch": 0.612673912654216,
      "grad_norm": 1.44051194190979,
      "learning_rate": 6.888878075379326e-05,
      "loss": 1.4555,
      "step": 17108
    },
    {
      "epoch": 0.6127097247837843,
      "grad_norm": 1.836295485496521,
      "learning_rate": 6.887775757215334e-05,
      "loss": 1.7671,
      "step": 17109
    },
    {
      "epoch": 0.6127455369133525,
      "grad_norm": 1.9834587574005127,
      "learning_rate": 6.886673480922299e-05,
      "loss": 1.5618,
      "step": 17110
    },
    {
      "epoch": 0.6127813490429208,
      "grad_norm": 1.558998942375183,
      "learning_rate": 6.88557124651506e-05,
      "loss": 1.4063,
      "step": 17111
    },
    {
      "epoch": 0.6128171611724891,
      "grad_norm": 1.5703425407409668,
      "learning_rate": 6.884469054008444e-05,
      "loss": 1.3246,
      "step": 17112
    },
    {
      "epoch": 0.6128529733020575,
      "grad_norm": 1.665521264076233,
      "learning_rate": 6.883366903417273e-05,
      "loss": 1.3784,
      "step": 17113
    },
    {
      "epoch": 0.6128887854316257,
      "grad_norm": 1.5847508907318115,
      "learning_rate": 6.882264794756386e-05,
      "loss": 1.3142,
      "step": 17114
    },
    {
      "epoch": 0.612924597561194,
      "grad_norm": 1.8532623052597046,
      "learning_rate": 6.881162728040598e-05,
      "loss": 1.4714,
      "step": 17115
    },
    {
      "epoch": 0.6129604096907623,
      "grad_norm": 1.7318142652511597,
      "learning_rate": 6.880060703284748e-05,
      "loss": 1.6911,
      "step": 17116
    },
    {
      "epoch": 0.6129962218203305,
      "grad_norm": 1.7050365209579468,
      "learning_rate": 6.878958720503652e-05,
      "loss": 1.7391,
      "step": 17117
    },
    {
      "epoch": 0.6130320339498988,
      "grad_norm": 1.766896367073059,
      "learning_rate": 6.877856779712147e-05,
      "loss": 1.4107,
      "step": 17118
    },
    {
      "epoch": 0.6130678460794671,
      "grad_norm": 1.5562337636947632,
      "learning_rate": 6.876754880925049e-05,
      "loss": 1.4708,
      "step": 17119
    },
    {
      "epoch": 0.6131036582090355,
      "grad_norm": 1.6931556463241577,
      "learning_rate": 6.87565302415718e-05,
      "loss": 1.5617,
      "step": 17120
    },
    {
      "epoch": 0.6131394703386037,
      "grad_norm": 1.5268492698669434,
      "learning_rate": 6.874551209423376e-05,
      "loss": 1.5087,
      "step": 17121
    },
    {
      "epoch": 0.613175282468172,
      "grad_norm": 1.6822506189346313,
      "learning_rate": 6.873449436738451e-05,
      "loss": 1.4391,
      "step": 17122
    },
    {
      "epoch": 0.6132110945977403,
      "grad_norm": 1.6287306547164917,
      "learning_rate": 6.872347706117233e-05,
      "loss": 1.4453,
      "step": 17123
    },
    {
      "epoch": 0.6132469067273085,
      "grad_norm": 1.3523920774459839,
      "learning_rate": 6.871246017574537e-05,
      "loss": 1.4051,
      "step": 17124
    },
    {
      "epoch": 0.6132827188568768,
      "grad_norm": 2.1489264965057373,
      "learning_rate": 6.870144371125198e-05,
      "loss": 1.2533,
      "step": 17125
    },
    {
      "epoch": 0.6133185309864451,
      "grad_norm": 1.7597583532333374,
      "learning_rate": 6.869042766784022e-05,
      "loss": 1.439,
      "step": 17126
    },
    {
      "epoch": 0.6133543431160134,
      "grad_norm": 1.3666061162948608,
      "learning_rate": 6.867941204565843e-05,
      "loss": 1.1421,
      "step": 17127
    },
    {
      "epoch": 0.6133901552455817,
      "grad_norm": 1.6297531127929688,
      "learning_rate": 6.866839684485473e-05,
      "loss": 1.4425,
      "step": 17128
    },
    {
      "epoch": 0.61342596737515,
      "grad_norm": 1.778544545173645,
      "learning_rate": 6.865738206557731e-05,
      "loss": 1.3258,
      "step": 17129
    },
    {
      "epoch": 0.6134617795047183,
      "grad_norm": 1.8034090995788574,
      "learning_rate": 6.864636770797441e-05,
      "loss": 1.5363,
      "step": 17130
    },
    {
      "epoch": 0.6134975916342865,
      "grad_norm": 1.9214576482772827,
      "learning_rate": 6.863535377219417e-05,
      "loss": 1.5098,
      "step": 17131
    },
    {
      "epoch": 0.6135334037638548,
      "grad_norm": 1.9046701192855835,
      "learning_rate": 6.862434025838481e-05,
      "loss": 1.6811,
      "step": 17132
    },
    {
      "epoch": 0.6135692158934231,
      "grad_norm": 2.123274087905884,
      "learning_rate": 6.861332716669444e-05,
      "loss": 1.6217,
      "step": 17133
    },
    {
      "epoch": 0.6136050280229914,
      "grad_norm": 1.7774008512496948,
      "learning_rate": 6.860231449727133e-05,
      "loss": 1.5144,
      "step": 17134
    },
    {
      "epoch": 0.6136408401525597,
      "grad_norm": 1.6106023788452148,
      "learning_rate": 6.859130225026351e-05,
      "loss": 1.425,
      "step": 17135
    },
    {
      "epoch": 0.613676652282128,
      "grad_norm": 1.860154628753662,
      "learning_rate": 6.858029042581926e-05,
      "loss": 1.1347,
      "step": 17136
    },
    {
      "epoch": 0.6137124644116962,
      "grad_norm": 1.433030605316162,
      "learning_rate": 6.856927902408666e-05,
      "loss": 1.6064,
      "step": 17137
    },
    {
      "epoch": 0.6137482765412645,
      "grad_norm": 1.4676074981689453,
      "learning_rate": 6.855826804521386e-05,
      "loss": 1.3863,
      "step": 17138
    },
    {
      "epoch": 0.6137840886708328,
      "grad_norm": 1.465326189994812,
      "learning_rate": 6.854725748934901e-05,
      "loss": 1.6177,
      "step": 17139
    },
    {
      "epoch": 0.613819900800401,
      "grad_norm": 2.1251637935638428,
      "learning_rate": 6.853624735664021e-05,
      "loss": 1.0243,
      "step": 17140
    },
    {
      "epoch": 0.6138557129299694,
      "grad_norm": 1.8610551357269287,
      "learning_rate": 6.852523764723566e-05,
      "loss": 1.7238,
      "step": 17141
    },
    {
      "epoch": 0.6138915250595377,
      "grad_norm": 1.2986605167388916,
      "learning_rate": 6.85142283612834e-05,
      "loss": 1.4475,
      "step": 17142
    },
    {
      "epoch": 0.613927337189106,
      "grad_norm": 2.99711537361145,
      "learning_rate": 6.850321949893162e-05,
      "loss": 1.3241,
      "step": 17143
    },
    {
      "epoch": 0.6139631493186742,
      "grad_norm": 1.5854588747024536,
      "learning_rate": 6.849221106032837e-05,
      "loss": 1.308,
      "step": 17144
    },
    {
      "epoch": 0.6139989614482425,
      "grad_norm": 1.4505645036697388,
      "learning_rate": 6.84812030456218e-05,
      "loss": 1.5152,
      "step": 17145
    },
    {
      "epoch": 0.6140347735778108,
      "grad_norm": 1.736111044883728,
      "learning_rate": 6.847019545495998e-05,
      "loss": 1.6171,
      "step": 17146
    },
    {
      "epoch": 0.614070585707379,
      "grad_norm": 1.6995787620544434,
      "learning_rate": 6.845918828849099e-05,
      "loss": 1.0826,
      "step": 17147
    },
    {
      "epoch": 0.6141063978369474,
      "grad_norm": 2.0281624794006348,
      "learning_rate": 6.844818154636295e-05,
      "loss": 1.5205,
      "step": 17148
    },
    {
      "epoch": 0.6141422099665157,
      "grad_norm": 1.5841419696807861,
      "learning_rate": 6.843717522872393e-05,
      "loss": 1.4293,
      "step": 17149
    },
    {
      "epoch": 0.614178022096084,
      "grad_norm": 1.547128677368164,
      "learning_rate": 6.8426169335722e-05,
      "loss": 1.4819,
      "step": 17150
    },
    {
      "epoch": 0.6142138342256522,
      "grad_norm": 1.5373615026474,
      "learning_rate": 6.841516386750523e-05,
      "loss": 1.326,
      "step": 17151
    },
    {
      "epoch": 0.6142496463552205,
      "grad_norm": 1.5892366170883179,
      "learning_rate": 6.84041588242217e-05,
      "loss": 1.772,
      "step": 17152
    },
    {
      "epoch": 0.6142854584847888,
      "grad_norm": 1.3406174182891846,
      "learning_rate": 6.839315420601943e-05,
      "loss": 1.4543,
      "step": 17153
    },
    {
      "epoch": 0.614321270614357,
      "grad_norm": 2.1202826499938965,
      "learning_rate": 6.838215001304654e-05,
      "loss": 1.5323,
      "step": 17154
    },
    {
      "epoch": 0.6143570827439254,
      "grad_norm": 2.2001073360443115,
      "learning_rate": 6.837114624545102e-05,
      "loss": 1.8068,
      "step": 17155
    },
    {
      "epoch": 0.6143928948734937,
      "grad_norm": 1.5223966836929321,
      "learning_rate": 6.836014290338093e-05,
      "loss": 1.5163,
      "step": 17156
    },
    {
      "epoch": 0.614428707003062,
      "grad_norm": 2.0496227741241455,
      "learning_rate": 6.834913998698432e-05,
      "loss": 1.7119,
      "step": 17157
    },
    {
      "epoch": 0.6144645191326302,
      "grad_norm": 1.7774089574813843,
      "learning_rate": 6.833813749640916e-05,
      "loss": 1.1044,
      "step": 17158
    },
    {
      "epoch": 0.6145003312621985,
      "grad_norm": 1.9635884761810303,
      "learning_rate": 6.832713543180356e-05,
      "loss": 1.1691,
      "step": 17159
    },
    {
      "epoch": 0.6145361433917668,
      "grad_norm": 1.539371371269226,
      "learning_rate": 6.831613379331547e-05,
      "loss": 1.3793,
      "step": 17160
    },
    {
      "epoch": 0.614571955521335,
      "grad_norm": 1.989688515663147,
      "learning_rate": 6.830513258109296e-05,
      "loss": 1.8035,
      "step": 17161
    },
    {
      "epoch": 0.6146077676509034,
      "grad_norm": 1.5166738033294678,
      "learning_rate": 6.829413179528398e-05,
      "loss": 1.2237,
      "step": 17162
    },
    {
      "epoch": 0.6146435797804717,
      "grad_norm": 1.495975375175476,
      "learning_rate": 6.828313143603657e-05,
      "loss": 1.1678,
      "step": 17163
    },
    {
      "epoch": 0.61467939191004,
      "grad_norm": 1.3702117204666138,
      "learning_rate": 6.827213150349874e-05,
      "loss": 1.5745,
      "step": 17164
    },
    {
      "epoch": 0.6147152040396082,
      "grad_norm": 1.3687939643859863,
      "learning_rate": 6.826113199781841e-05,
      "loss": 1.2165,
      "step": 17165
    },
    {
      "epoch": 0.6147510161691765,
      "grad_norm": 1.5471529960632324,
      "learning_rate": 6.825013291914363e-05,
      "loss": 1.4653,
      "step": 17166
    },
    {
      "epoch": 0.6147868282987448,
      "grad_norm": 1.3832335472106934,
      "learning_rate": 6.823913426762237e-05,
      "loss": 1.4773,
      "step": 17167
    },
    {
      "epoch": 0.614822640428313,
      "grad_norm": 1.4992239475250244,
      "learning_rate": 6.822813604340257e-05,
      "loss": 1.4386,
      "step": 17168
    },
    {
      "epoch": 0.6148584525578814,
      "grad_norm": 1.5408648252487183,
      "learning_rate": 6.821713824663221e-05,
      "loss": 1.4708,
      "step": 17169
    },
    {
      "epoch": 0.6148942646874497,
      "grad_norm": 1.663112998008728,
      "learning_rate": 6.820614087745929e-05,
      "loss": 1.4123,
      "step": 17170
    },
    {
      "epoch": 0.614930076817018,
      "grad_norm": 1.6703062057495117,
      "learning_rate": 6.81951439360317e-05,
      "loss": 1.288,
      "step": 17171
    },
    {
      "epoch": 0.6149658889465862,
      "grad_norm": 1.5483388900756836,
      "learning_rate": 6.818414742249745e-05,
      "loss": 1.3075,
      "step": 17172
    },
    {
      "epoch": 0.6150017010761545,
      "grad_norm": 1.4388359785079956,
      "learning_rate": 6.817315133700446e-05,
      "loss": 1.5018,
      "step": 17173
    },
    {
      "epoch": 0.6150375132057228,
      "grad_norm": 1.47565758228302,
      "learning_rate": 6.816215567970063e-05,
      "loss": 1.5788,
      "step": 17174
    },
    {
      "epoch": 0.615073325335291,
      "grad_norm": 1.9998646974563599,
      "learning_rate": 6.815116045073396e-05,
      "loss": 1.4184,
      "step": 17175
    },
    {
      "epoch": 0.6151091374648594,
      "grad_norm": 2.213521957397461,
      "learning_rate": 6.814016565025231e-05,
      "loss": 1.6169,
      "step": 17176
    },
    {
      "epoch": 0.6151449495944277,
      "grad_norm": 1.4681707620620728,
      "learning_rate": 6.812917127840368e-05,
      "loss": 1.568,
      "step": 17177
    },
    {
      "epoch": 0.6151807617239959,
      "grad_norm": 1.8381984233856201,
      "learning_rate": 6.81181773353359e-05,
      "loss": 1.657,
      "step": 17178
    },
    {
      "epoch": 0.6152165738535642,
      "grad_norm": 1.8551054000854492,
      "learning_rate": 6.810718382119694e-05,
      "loss": 1.4701,
      "step": 17179
    },
    {
      "epoch": 0.6152523859831325,
      "grad_norm": 1.313694953918457,
      "learning_rate": 6.809619073613467e-05,
      "loss": 1.3976,
      "step": 17180
    },
    {
      "epoch": 0.6152881981127007,
      "grad_norm": 2.456167697906494,
      "learning_rate": 6.808519808029703e-05,
      "loss": 1.3742,
      "step": 17181
    },
    {
      "epoch": 0.615324010242269,
      "grad_norm": 1.4624918699264526,
      "learning_rate": 6.807420585383186e-05,
      "loss": 1.5296,
      "step": 17182
    },
    {
      "epoch": 0.6153598223718374,
      "grad_norm": 1.544368863105774,
      "learning_rate": 6.806321405688707e-05,
      "loss": 1.6921,
      "step": 17183
    },
    {
      "epoch": 0.6153956345014057,
      "grad_norm": 1.5174061059951782,
      "learning_rate": 6.805222268961054e-05,
      "loss": 1.3493,
      "step": 17184
    },
    {
      "epoch": 0.6154314466309739,
      "grad_norm": 1.750654935836792,
      "learning_rate": 6.804123175215014e-05,
      "loss": 1.5409,
      "step": 17185
    },
    {
      "epoch": 0.6154672587605422,
      "grad_norm": 1.7481032609939575,
      "learning_rate": 6.803024124465375e-05,
      "loss": 1.4192,
      "step": 17186
    },
    {
      "epoch": 0.6155030708901105,
      "grad_norm": 1.7678956985473633,
      "learning_rate": 6.801925116726922e-05,
      "loss": 1.3979,
      "step": 17187
    },
    {
      "epoch": 0.6155388830196787,
      "grad_norm": 1.6800671815872192,
      "learning_rate": 6.800826152014442e-05,
      "loss": 1.277,
      "step": 17188
    },
    {
      "epoch": 0.615574695149247,
      "grad_norm": 2.4462714195251465,
      "learning_rate": 6.799727230342718e-05,
      "loss": 1.5407,
      "step": 17189
    },
    {
      "epoch": 0.6156105072788154,
      "grad_norm": 1.6587086915969849,
      "learning_rate": 6.798628351726539e-05,
      "loss": 1.3014,
      "step": 17190
    },
    {
      "epoch": 0.6156463194083837,
      "grad_norm": 1.3719878196716309,
      "learning_rate": 6.797529516180687e-05,
      "loss": 1.2125,
      "step": 17191
    },
    {
      "epoch": 0.6156821315379519,
      "grad_norm": 1.9094762802124023,
      "learning_rate": 6.796430723719939e-05,
      "loss": 1.2202,
      "step": 17192
    },
    {
      "epoch": 0.6157179436675202,
      "grad_norm": 1.774322271347046,
      "learning_rate": 6.795331974359088e-05,
      "loss": 1.7452,
      "step": 17193
    },
    {
      "epoch": 0.6157537557970885,
      "grad_norm": 2.135695695877075,
      "learning_rate": 6.794233268112907e-05,
      "loss": 1.7369,
      "step": 17194
    },
    {
      "epoch": 0.6157895679266567,
      "grad_norm": 1.5609972476959229,
      "learning_rate": 6.793134604996185e-05,
      "loss": 1.347,
      "step": 17195
    },
    {
      "epoch": 0.615825380056225,
      "grad_norm": 1.5753405094146729,
      "learning_rate": 6.7920359850237e-05,
      "loss": 1.3896,
      "step": 17196
    },
    {
      "epoch": 0.6158611921857934,
      "grad_norm": 2.917074203491211,
      "learning_rate": 6.790937408210233e-05,
      "loss": 1.528,
      "step": 17197
    },
    {
      "epoch": 0.6158970043153617,
      "grad_norm": 1.5047675371170044,
      "learning_rate": 6.789838874570565e-05,
      "loss": 1.5584,
      "step": 17198
    },
    {
      "epoch": 0.6159328164449299,
      "grad_norm": 1.545350193977356,
      "learning_rate": 6.788740384119472e-05,
      "loss": 1.5371,
      "step": 17199
    },
    {
      "epoch": 0.6159686285744982,
      "grad_norm": 1.9652788639068604,
      "learning_rate": 6.787641936871739e-05,
      "loss": 1.5017,
      "step": 17200
    },
    {
      "epoch": 0.6160044407040665,
      "grad_norm": 1.3209742307662964,
      "learning_rate": 6.786543532842133e-05,
      "loss": 1.1845,
      "step": 17201
    },
    {
      "epoch": 0.6160402528336347,
      "grad_norm": 2.2090561389923096,
      "learning_rate": 6.785445172045448e-05,
      "loss": 1.6238,
      "step": 17202
    },
    {
      "epoch": 0.616076064963203,
      "grad_norm": 1.6406116485595703,
      "learning_rate": 6.784346854496442e-05,
      "loss": 1.2969,
      "step": 17203
    },
    {
      "epoch": 0.6161118770927714,
      "grad_norm": 1.7043877840042114,
      "learning_rate": 6.78324858020991e-05,
      "loss": 1.5182,
      "step": 17204
    },
    {
      "epoch": 0.6161476892223396,
      "grad_norm": 1.4550445079803467,
      "learning_rate": 6.78215034920061e-05,
      "loss": 1.5474,
      "step": 17205
    },
    {
      "epoch": 0.6161835013519079,
      "grad_norm": 1.4933606386184692,
      "learning_rate": 6.781052161483332e-05,
      "loss": 1.3881,
      "step": 17206
    },
    {
      "epoch": 0.6162193134814762,
      "grad_norm": 1.5615988969802856,
      "learning_rate": 6.779954017072842e-05,
      "loss": 1.5827,
      "step": 17207
    },
    {
      "epoch": 0.6162551256110445,
      "grad_norm": 1.4446730613708496,
      "learning_rate": 6.778855915983921e-05,
      "loss": 1.1332,
      "step": 17208
    },
    {
      "epoch": 0.6162909377406127,
      "grad_norm": 1.8601797819137573,
      "learning_rate": 6.777757858231339e-05,
      "loss": 1.252,
      "step": 17209
    },
    {
      "epoch": 0.616326749870181,
      "grad_norm": 1.5915179252624512,
      "learning_rate": 6.776659843829863e-05,
      "loss": 1.4509,
      "step": 17210
    },
    {
      "epoch": 0.6163625619997494,
      "grad_norm": 1.6909068822860718,
      "learning_rate": 6.775561872794279e-05,
      "loss": 1.477,
      "step": 17211
    },
    {
      "epoch": 0.6163983741293176,
      "grad_norm": 1.5252338647842407,
      "learning_rate": 6.774463945139343e-05,
      "loss": 1.6273,
      "step": 17212
    },
    {
      "epoch": 0.6164341862588859,
      "grad_norm": 1.677097201347351,
      "learning_rate": 6.77336606087984e-05,
      "loss": 1.4723,
      "step": 17213
    },
    {
      "epoch": 0.6164699983884542,
      "grad_norm": 1.8449418544769287,
      "learning_rate": 6.772268220030528e-05,
      "loss": 1.201,
      "step": 17214
    },
    {
      "epoch": 0.6165058105180224,
      "grad_norm": 1.9632333517074585,
      "learning_rate": 6.77117042260619e-05,
      "loss": 1.6116,
      "step": 17215
    },
    {
      "epoch": 0.6165416226475907,
      "grad_norm": 1.7720052003860474,
      "learning_rate": 6.770072668621583e-05,
      "loss": 1.6009,
      "step": 17216
    },
    {
      "epoch": 0.616577434777159,
      "grad_norm": 2.2499513626098633,
      "learning_rate": 6.768974958091488e-05,
      "loss": 1.2514,
      "step": 17217
    },
    {
      "epoch": 0.6166132469067274,
      "grad_norm": 1.5738240480422974,
      "learning_rate": 6.767877291030666e-05,
      "loss": 1.6878,
      "step": 17218
    },
    {
      "epoch": 0.6166490590362956,
      "grad_norm": 1.6386752128601074,
      "learning_rate": 6.766779667453881e-05,
      "loss": 1.4406,
      "step": 17219
    },
    {
      "epoch": 0.6166848711658639,
      "grad_norm": 1.473636507987976,
      "learning_rate": 6.765682087375912e-05,
      "loss": 1.1257,
      "step": 17220
    },
    {
      "epoch": 0.6167206832954322,
      "grad_norm": 1.5793477296829224,
      "learning_rate": 6.764584550811512e-05,
      "loss": 1.3315,
      "step": 17221
    },
    {
      "epoch": 0.6167564954250004,
      "grad_norm": 2.4050142765045166,
      "learning_rate": 6.763487057775459e-05,
      "loss": 1.4098,
      "step": 17222
    },
    {
      "epoch": 0.6167923075545687,
      "grad_norm": 1.66704261302948,
      "learning_rate": 6.762389608282507e-05,
      "loss": 1.2545,
      "step": 17223
    },
    {
      "epoch": 0.616828119684137,
      "grad_norm": 1.397594928741455,
      "learning_rate": 6.761292202347434e-05,
      "loss": 1.5063,
      "step": 17224
    },
    {
      "epoch": 0.6168639318137054,
      "grad_norm": 1.3447819948196411,
      "learning_rate": 6.760194839984994e-05,
      "loss": 1.281,
      "step": 17225
    },
    {
      "epoch": 0.6168997439432736,
      "grad_norm": 1.6767772436141968,
      "learning_rate": 6.75909752120995e-05,
      "loss": 1.8086,
      "step": 17226
    },
    {
      "epoch": 0.6169355560728419,
      "grad_norm": 1.7301512956619263,
      "learning_rate": 6.758000246037072e-05,
      "loss": 1.8668,
      "step": 17227
    },
    {
      "epoch": 0.6169713682024102,
      "grad_norm": 1.7940595149993896,
      "learning_rate": 6.756903014481116e-05,
      "loss": 1.3394,
      "step": 17228
    },
    {
      "epoch": 0.6170071803319784,
      "grad_norm": 1.7685900926589966,
      "learning_rate": 6.75580582655685e-05,
      "loss": 1.7772,
      "step": 17229
    },
    {
      "epoch": 0.6170429924615467,
      "grad_norm": 1.6074085235595703,
      "learning_rate": 6.754708682279027e-05,
      "loss": 1.2769,
      "step": 17230
    },
    {
      "epoch": 0.617078804591115,
      "grad_norm": 2.404395580291748,
      "learning_rate": 6.753611581662418e-05,
      "loss": 1.546,
      "step": 17231
    },
    {
      "epoch": 0.6171146167206832,
      "grad_norm": 1.7004632949829102,
      "learning_rate": 6.752514524721771e-05,
      "loss": 1.705,
      "step": 17232
    },
    {
      "epoch": 0.6171504288502516,
      "grad_norm": 1.2684026956558228,
      "learning_rate": 6.751417511471859e-05,
      "loss": 1.5119,
      "step": 17233
    },
    {
      "epoch": 0.6171862409798199,
      "grad_norm": 1.4887030124664307,
      "learning_rate": 6.750320541927433e-05,
      "loss": 1.4895,
      "step": 17234
    },
    {
      "epoch": 0.6172220531093882,
      "grad_norm": 1.5183050632476807,
      "learning_rate": 6.749223616103249e-05,
      "loss": 1.4796,
      "step": 17235
    },
    {
      "epoch": 0.6172578652389564,
      "grad_norm": 1.4999533891677856,
      "learning_rate": 6.74812673401407e-05,
      "loss": 1.0856,
      "step": 17236
    },
    {
      "epoch": 0.6172936773685247,
      "grad_norm": 1.952275276184082,
      "learning_rate": 6.74702989567465e-05,
      "loss": 1.5083,
      "step": 17237
    },
    {
      "epoch": 0.617329489498093,
      "grad_norm": 1.649373173713684,
      "learning_rate": 6.745933101099748e-05,
      "loss": 1.5587,
      "step": 17238
    },
    {
      "epoch": 0.6173653016276612,
      "grad_norm": 1.4925177097320557,
      "learning_rate": 6.744836350304118e-05,
      "loss": 1.1892,
      "step": 17239
    },
    {
      "epoch": 0.6174011137572296,
      "grad_norm": 1.5347260236740112,
      "learning_rate": 6.743739643302516e-05,
      "loss": 1.4524,
      "step": 17240
    },
    {
      "epoch": 0.6174369258867979,
      "grad_norm": 1.3831895589828491,
      "learning_rate": 6.742642980109696e-05,
      "loss": 1.421,
      "step": 17241
    },
    {
      "epoch": 0.6174727380163662,
      "grad_norm": 2.0352489948272705,
      "learning_rate": 6.741546360740415e-05,
      "loss": 1.8222,
      "step": 17242
    },
    {
      "epoch": 0.6175085501459344,
      "grad_norm": 1.7592264413833618,
      "learning_rate": 6.740449785209425e-05,
      "loss": 1.3654,
      "step": 17243
    },
    {
      "epoch": 0.6175443622755027,
      "grad_norm": 1.4026323556900024,
      "learning_rate": 6.739353253531475e-05,
      "loss": 1.6612,
      "step": 17244
    },
    {
      "epoch": 0.617580174405071,
      "grad_norm": 2.095674991607666,
      "learning_rate": 6.738256765721324e-05,
      "loss": 1.3747,
      "step": 17245
    },
    {
      "epoch": 0.6176159865346392,
      "grad_norm": 1.5699025392532349,
      "learning_rate": 6.73716032179372e-05,
      "loss": 1.3018,
      "step": 17246
    },
    {
      "epoch": 0.6176517986642076,
      "grad_norm": 2.3074796199798584,
      "learning_rate": 6.736063921763415e-05,
      "loss": 1.4714,
      "step": 17247
    },
    {
      "epoch": 0.6176876107937759,
      "grad_norm": 1.499690294265747,
      "learning_rate": 6.73496756564516e-05,
      "loss": 1.3134,
      "step": 17248
    },
    {
      "epoch": 0.6177234229233441,
      "grad_norm": 1.9261491298675537,
      "learning_rate": 6.733871253453707e-05,
      "loss": 1.6609,
      "step": 17249
    },
    {
      "epoch": 0.6177592350529124,
      "grad_norm": 1.6872670650482178,
      "learning_rate": 6.7327749852038e-05,
      "loss": 1.2668,
      "step": 17250
    },
    {
      "epoch": 0.6177950471824807,
      "grad_norm": 1.8014341592788696,
      "learning_rate": 6.731678760910192e-05,
      "loss": 1.5728,
      "step": 17251
    },
    {
      "epoch": 0.617830859312049,
      "grad_norm": 1.621647596359253,
      "learning_rate": 6.730582580587632e-05,
      "loss": 1.3942,
      "step": 17252
    },
    {
      "epoch": 0.6178666714416172,
      "grad_norm": 1.6194264888763428,
      "learning_rate": 6.729486444250863e-05,
      "loss": 1.5561,
      "step": 17253
    },
    {
      "epoch": 0.6179024835711856,
      "grad_norm": 1.5640456676483154,
      "learning_rate": 6.72839035191464e-05,
      "loss": 1.169,
      "step": 17254
    },
    {
      "epoch": 0.6179382957007539,
      "grad_norm": 1.567991018295288,
      "learning_rate": 6.7272943035937e-05,
      "loss": 1.6388,
      "step": 17255
    },
    {
      "epoch": 0.6179741078303221,
      "grad_norm": 1.513309121131897,
      "learning_rate": 6.726198299302796e-05,
      "loss": 1.5298,
      "step": 17256
    },
    {
      "epoch": 0.6180099199598904,
      "grad_norm": 1.6666152477264404,
      "learning_rate": 6.72510233905667e-05,
      "loss": 1.3742,
      "step": 17257
    },
    {
      "epoch": 0.6180457320894587,
      "grad_norm": 1.8331990242004395,
      "learning_rate": 6.724006422870069e-05,
      "loss": 1.3625,
      "step": 17258
    },
    {
      "epoch": 0.618081544219027,
      "grad_norm": 1.3519563674926758,
      "learning_rate": 6.722910550757734e-05,
      "loss": 1.5429,
      "step": 17259
    },
    {
      "epoch": 0.6181173563485952,
      "grad_norm": 1.70087730884552,
      "learning_rate": 6.721814722734412e-05,
      "loss": 1.2871,
      "step": 17260
    },
    {
      "epoch": 0.6181531684781636,
      "grad_norm": 1.6571663618087769,
      "learning_rate": 6.720718938814846e-05,
      "loss": 1.4969,
      "step": 17261
    },
    {
      "epoch": 0.6181889806077319,
      "grad_norm": 1.8243577480316162,
      "learning_rate": 6.719623199013771e-05,
      "loss": 1.343,
      "step": 17262
    },
    {
      "epoch": 0.6182247927373001,
      "grad_norm": 1.3904130458831787,
      "learning_rate": 6.718527503345939e-05,
      "loss": 1.3714,
      "step": 17263
    },
    {
      "epoch": 0.6182606048668684,
      "grad_norm": 1.2243188619613647,
      "learning_rate": 6.717431851826086e-05,
      "loss": 1.0458,
      "step": 17264
    },
    {
      "epoch": 0.6182964169964367,
      "grad_norm": 1.2935394048690796,
      "learning_rate": 6.716336244468954e-05,
      "loss": 1.5686,
      "step": 17265
    },
    {
      "epoch": 0.6183322291260049,
      "grad_norm": 1.9957224130630493,
      "learning_rate": 6.715240681289279e-05,
      "loss": 1.585,
      "step": 17266
    },
    {
      "epoch": 0.6183680412555732,
      "grad_norm": 1.5544694662094116,
      "learning_rate": 6.714145162301808e-05,
      "loss": 1.2735,
      "step": 17267
    },
    {
      "epoch": 0.6184038533851416,
      "grad_norm": 1.6578425168991089,
      "learning_rate": 6.713049687521272e-05,
      "loss": 1.4704,
      "step": 17268
    },
    {
      "epoch": 0.6184396655147099,
      "grad_norm": 1.665738821029663,
      "learning_rate": 6.711954256962414e-05,
      "loss": 1.5401,
      "step": 17269
    },
    {
      "epoch": 0.6184754776442781,
      "grad_norm": 1.3819257020950317,
      "learning_rate": 6.71085887063997e-05,
      "loss": 1.7921,
      "step": 17270
    },
    {
      "epoch": 0.6185112897738464,
      "grad_norm": 1.3936008214950562,
      "learning_rate": 6.709763528568677e-05,
      "loss": 1.2616,
      "step": 17271
    },
    {
      "epoch": 0.6185471019034147,
      "grad_norm": 1.3957215547561646,
      "learning_rate": 6.708668230763272e-05,
      "loss": 1.3208,
      "step": 17272
    },
    {
      "epoch": 0.6185829140329829,
      "grad_norm": 1.659361481666565,
      "learning_rate": 6.707572977238489e-05,
      "loss": 1.3095,
      "step": 17273
    },
    {
      "epoch": 0.6186187261625512,
      "grad_norm": 1.7726492881774902,
      "learning_rate": 6.706477768009067e-05,
      "loss": 1.7466,
      "step": 17274
    },
    {
      "epoch": 0.6186545382921196,
      "grad_norm": 1.3715457916259766,
      "learning_rate": 6.705382603089737e-05,
      "loss": 1.6041,
      "step": 17275
    },
    {
      "epoch": 0.6186903504216879,
      "grad_norm": 1.717397928237915,
      "learning_rate": 6.704287482495233e-05,
      "loss": 1.5324,
      "step": 17276
    },
    {
      "epoch": 0.6187261625512561,
      "grad_norm": 1.2430498600006104,
      "learning_rate": 6.70319240624029e-05,
      "loss": 1.3185,
      "step": 17277
    },
    {
      "epoch": 0.6187619746808244,
      "grad_norm": 1.5718120336532593,
      "learning_rate": 6.702097374339644e-05,
      "loss": 1.1736,
      "step": 17278
    },
    {
      "epoch": 0.6187977868103927,
      "grad_norm": 1.5527795553207397,
      "learning_rate": 6.701002386808021e-05,
      "loss": 1.3052,
      "step": 17279
    },
    {
      "epoch": 0.6188335989399609,
      "grad_norm": 1.8009364604949951,
      "learning_rate": 6.699907443660156e-05,
      "loss": 1.3913,
      "step": 17280
    },
    {
      "epoch": 0.6188694110695292,
      "grad_norm": 3.0369083881378174,
      "learning_rate": 6.698812544910781e-05,
      "loss": 1.4698,
      "step": 17281
    },
    {
      "epoch": 0.6189052231990976,
      "grad_norm": 1.604009747505188,
      "learning_rate": 6.697717690574623e-05,
      "loss": 1.4513,
      "step": 17282
    },
    {
      "epoch": 0.6189410353286658,
      "grad_norm": 1.6419212818145752,
      "learning_rate": 6.696622880666415e-05,
      "loss": 1.2407,
      "step": 17283
    },
    {
      "epoch": 0.6189768474582341,
      "grad_norm": 2.8187286853790283,
      "learning_rate": 6.695528115200883e-05,
      "loss": 1.8745,
      "step": 17284
    },
    {
      "epoch": 0.6190126595878024,
      "grad_norm": 1.8177802562713623,
      "learning_rate": 6.69443339419276e-05,
      "loss": 1.3942,
      "step": 17285
    },
    {
      "epoch": 0.6190484717173707,
      "grad_norm": 1.7879760265350342,
      "learning_rate": 6.69333871765677e-05,
      "loss": 1.7262,
      "step": 17286
    },
    {
      "epoch": 0.6190842838469389,
      "grad_norm": 1.4644464254379272,
      "learning_rate": 6.692244085607644e-05,
      "loss": 1.3389,
      "step": 17287
    },
    {
      "epoch": 0.6191200959765072,
      "grad_norm": 1.9216207265853882,
      "learning_rate": 6.69114949806011e-05,
      "loss": 1.7144,
      "step": 17288
    },
    {
      "epoch": 0.6191559081060756,
      "grad_norm": 1.678239345550537,
      "learning_rate": 6.690054955028885e-05,
      "loss": 1.4053,
      "step": 17289
    },
    {
      "epoch": 0.6191917202356438,
      "grad_norm": 2.289401054382324,
      "learning_rate": 6.688960456528705e-05,
      "loss": 1.5856,
      "step": 17290
    },
    {
      "epoch": 0.6192275323652121,
      "grad_norm": 1.9415135383605957,
      "learning_rate": 6.687866002574289e-05,
      "loss": 1.5077,
      "step": 17291
    },
    {
      "epoch": 0.6192633444947804,
      "grad_norm": 1.520308256149292,
      "learning_rate": 6.686771593180365e-05,
      "loss": 1.2985,
      "step": 17292
    },
    {
      "epoch": 0.6192991566243486,
      "grad_norm": 1.429878830909729,
      "learning_rate": 6.685677228361654e-05,
      "loss": 1.5415,
      "step": 17293
    },
    {
      "epoch": 0.6193349687539169,
      "grad_norm": 1.9562313556671143,
      "learning_rate": 6.684582908132883e-05,
      "loss": 1.6137,
      "step": 17294
    },
    {
      "epoch": 0.6193707808834852,
      "grad_norm": 2.758021831512451,
      "learning_rate": 6.68348863250877e-05,
      "loss": 1.4741,
      "step": 17295
    },
    {
      "epoch": 0.6194065930130536,
      "grad_norm": 1.5317972898483276,
      "learning_rate": 6.682394401504042e-05,
      "loss": 1.7757,
      "step": 17296
    },
    {
      "epoch": 0.6194424051426218,
      "grad_norm": 1.6802877187728882,
      "learning_rate": 6.681300215133419e-05,
      "loss": 1.4877,
      "step": 17297
    },
    {
      "epoch": 0.6194782172721901,
      "grad_norm": 1.4757306575775146,
      "learning_rate": 6.680206073411616e-05,
      "loss": 1.478,
      "step": 17298
    },
    {
      "epoch": 0.6195140294017584,
      "grad_norm": 1.6011658906936646,
      "learning_rate": 6.679111976353362e-05,
      "loss": 1.6643,
      "step": 17299
    },
    {
      "epoch": 0.6195498415313266,
      "grad_norm": 1.3765159845352173,
      "learning_rate": 6.67801792397337e-05,
      "loss": 1.3161,
      "step": 17300
    },
    {
      "epoch": 0.6195856536608949,
      "grad_norm": 1.9646791219711304,
      "learning_rate": 6.676923916286365e-05,
      "loss": 1.4321,
      "step": 17301
    },
    {
      "epoch": 0.6196214657904632,
      "grad_norm": 1.6221250295639038,
      "learning_rate": 6.675829953307057e-05,
      "loss": 1.4556,
      "step": 17302
    },
    {
      "epoch": 0.6196572779200316,
      "grad_norm": 1.8408706188201904,
      "learning_rate": 6.674736035050173e-05,
      "loss": 1.3377,
      "step": 17303
    },
    {
      "epoch": 0.6196930900495998,
      "grad_norm": 1.496396541595459,
      "learning_rate": 6.673642161530424e-05,
      "loss": 1.1605,
      "step": 17304
    },
    {
      "epoch": 0.6197289021791681,
      "grad_norm": 1.4522924423217773,
      "learning_rate": 6.672548332762533e-05,
      "loss": 1.586,
      "step": 17305
    },
    {
      "epoch": 0.6197647143087364,
      "grad_norm": 1.8669017553329468,
      "learning_rate": 6.671454548761212e-05,
      "loss": 1.3713,
      "step": 17306
    },
    {
      "epoch": 0.6198005264383046,
      "grad_norm": 1.4297741651535034,
      "learning_rate": 6.670360809541171e-05,
      "loss": 1.4443,
      "step": 17307
    },
    {
      "epoch": 0.6198363385678729,
      "grad_norm": 1.9161500930786133,
      "learning_rate": 6.669267115117137e-05,
      "loss": 1.2736,
      "step": 17308
    },
    {
      "epoch": 0.6198721506974412,
      "grad_norm": 1.817219614982605,
      "learning_rate": 6.66817346550381e-05,
      "loss": 1.4487,
      "step": 17309
    },
    {
      "epoch": 0.6199079628270096,
      "grad_norm": 1.851157307624817,
      "learning_rate": 6.66707986071592e-05,
      "loss": 1.5179,
      "step": 17310
    },
    {
      "epoch": 0.6199437749565778,
      "grad_norm": 1.622918725013733,
      "learning_rate": 6.665986300768163e-05,
      "loss": 1.6099,
      "step": 17311
    },
    {
      "epoch": 0.6199795870861461,
      "grad_norm": 1.516659140586853,
      "learning_rate": 6.664892785675267e-05,
      "loss": 1.4413,
      "step": 17312
    },
    {
      "epoch": 0.6200153992157144,
      "grad_norm": 1.7160168886184692,
      "learning_rate": 6.663799315451931e-05,
      "loss": 1.743,
      "step": 17313
    },
    {
      "epoch": 0.6200512113452826,
      "grad_norm": 1.5063496828079224,
      "learning_rate": 6.662705890112876e-05,
      "loss": 1.5858,
      "step": 17314
    },
    {
      "epoch": 0.6200870234748509,
      "grad_norm": 1.5053106546401978,
      "learning_rate": 6.661612509672808e-05,
      "loss": 1.6044,
      "step": 17315
    },
    {
      "epoch": 0.6201228356044192,
      "grad_norm": 1.300164818763733,
      "learning_rate": 6.660519174146433e-05,
      "loss": 1.6188,
      "step": 17316
    },
    {
      "epoch": 0.6201586477339875,
      "grad_norm": 3.0625481605529785,
      "learning_rate": 6.659425883548471e-05,
      "loss": 1.6178,
      "step": 17317
    },
    {
      "epoch": 0.6201944598635558,
      "grad_norm": 1.4336274862289429,
      "learning_rate": 6.658332637893619e-05,
      "loss": 1.656,
      "step": 17318
    },
    {
      "epoch": 0.6202302719931241,
      "grad_norm": 2.179532051086426,
      "learning_rate": 6.657239437196596e-05,
      "loss": 1.6422,
      "step": 17319
    },
    {
      "epoch": 0.6202660841226924,
      "grad_norm": 1.5122623443603516,
      "learning_rate": 6.656146281472098e-05,
      "loss": 1.5129,
      "step": 17320
    },
    {
      "epoch": 0.6203018962522606,
      "grad_norm": 1.994327425956726,
      "learning_rate": 6.655053170734846e-05,
      "loss": 1.3558,
      "step": 17321
    },
    {
      "epoch": 0.6203377083818289,
      "grad_norm": 2.061368227005005,
      "learning_rate": 6.653960104999537e-05,
      "loss": 1.2273,
      "step": 17322
    },
    {
      "epoch": 0.6203735205113972,
      "grad_norm": 2.0309760570526123,
      "learning_rate": 6.652867084280876e-05,
      "loss": 1.288,
      "step": 17323
    },
    {
      "epoch": 0.6204093326409655,
      "grad_norm": 1.7593995332717896,
      "learning_rate": 6.651774108593574e-05,
      "loss": 1.3319,
      "step": 17324
    },
    {
      "epoch": 0.6204451447705338,
      "grad_norm": 1.9512978792190552,
      "learning_rate": 6.650681177952328e-05,
      "loss": 1.2813,
      "step": 17325
    },
    {
      "epoch": 0.6204809569001021,
      "grad_norm": 1.8783950805664062,
      "learning_rate": 6.64958829237185e-05,
      "loss": 1.346,
      "step": 17326
    },
    {
      "epoch": 0.6205167690296703,
      "grad_norm": 1.5158321857452393,
      "learning_rate": 6.648495451866838e-05,
      "loss": 1.526,
      "step": 17327
    },
    {
      "epoch": 0.6205525811592386,
      "grad_norm": 1.927087426185608,
      "learning_rate": 6.647402656451998e-05,
      "loss": 1.628,
      "step": 17328
    },
    {
      "epoch": 0.6205883932888069,
      "grad_norm": 1.507460117340088,
      "learning_rate": 6.646309906142027e-05,
      "loss": 1.6315,
      "step": 17329
    },
    {
      "epoch": 0.6206242054183752,
      "grad_norm": 1.6917349100112915,
      "learning_rate": 6.645217200951636e-05,
      "loss": 1.7155,
      "step": 17330
    },
    {
      "epoch": 0.6206600175479435,
      "grad_norm": 1.8586498498916626,
      "learning_rate": 6.644124540895518e-05,
      "loss": 1.39,
      "step": 17331
    },
    {
      "epoch": 0.6206958296775118,
      "grad_norm": 2.0919973850250244,
      "learning_rate": 6.643031925988375e-05,
      "loss": 1.6374,
      "step": 17332
    },
    {
      "epoch": 0.6207316418070801,
      "grad_norm": 1.7580621242523193,
      "learning_rate": 6.641939356244908e-05,
      "loss": 1.2822,
      "step": 17333
    },
    {
      "epoch": 0.6207674539366483,
      "grad_norm": 1.9603501558303833,
      "learning_rate": 6.640846831679815e-05,
      "loss": 1.1557,
      "step": 17334
    },
    {
      "epoch": 0.6208032660662166,
      "grad_norm": 1.6108684539794922,
      "learning_rate": 6.639754352307794e-05,
      "loss": 1.203,
      "step": 17335
    },
    {
      "epoch": 0.6208390781957849,
      "grad_norm": 1.8438420295715332,
      "learning_rate": 6.638661918143542e-05,
      "loss": 1.335,
      "step": 17336
    },
    {
      "epoch": 0.6208748903253531,
      "grad_norm": 1.7761483192443848,
      "learning_rate": 6.637569529201763e-05,
      "loss": 1.3621,
      "step": 17337
    },
    {
      "epoch": 0.6209107024549215,
      "grad_norm": 1.4380193948745728,
      "learning_rate": 6.636477185497145e-05,
      "loss": 1.5014,
      "step": 17338
    },
    {
      "epoch": 0.6209465145844898,
      "grad_norm": 2.0738120079040527,
      "learning_rate": 6.63538488704439e-05,
      "loss": 1.3528,
      "step": 17339
    },
    {
      "epoch": 0.6209823267140581,
      "grad_norm": 1.7343157529830933,
      "learning_rate": 6.634292633858191e-05,
      "loss": 1.5682,
      "step": 17340
    },
    {
      "epoch": 0.6210181388436263,
      "grad_norm": 1.6832574605941772,
      "learning_rate": 6.633200425953241e-05,
      "loss": 1.6061,
      "step": 17341
    },
    {
      "epoch": 0.6210539509731946,
      "grad_norm": 1.3964799642562866,
      "learning_rate": 6.632108263344238e-05,
      "loss": 1.2056,
      "step": 17342
    },
    {
      "epoch": 0.6210897631027629,
      "grad_norm": 1.7358297109603882,
      "learning_rate": 6.631016146045874e-05,
      "loss": 1.2508,
      "step": 17343
    },
    {
      "epoch": 0.6211255752323311,
      "grad_norm": 1.7798261642456055,
      "learning_rate": 6.629924074072844e-05,
      "loss": 1.5458,
      "step": 17344
    },
    {
      "epoch": 0.6211613873618995,
      "grad_norm": 1.7764931917190552,
      "learning_rate": 6.628832047439835e-05,
      "loss": 1.4655,
      "step": 17345
    },
    {
      "epoch": 0.6211971994914678,
      "grad_norm": 2.0410125255584717,
      "learning_rate": 6.627740066161545e-05,
      "loss": 1.3842,
      "step": 17346
    },
    {
      "epoch": 0.6212330116210361,
      "grad_norm": 1.4399596452713013,
      "learning_rate": 6.62664813025266e-05,
      "loss": 1.3467,
      "step": 17347
    },
    {
      "epoch": 0.6212688237506043,
      "grad_norm": 1.5043425559997559,
      "learning_rate": 6.625556239727875e-05,
      "loss": 1.5523,
      "step": 17348
    },
    {
      "epoch": 0.6213046358801726,
      "grad_norm": 1.7739176750183105,
      "learning_rate": 6.624464394601879e-05,
      "loss": 1.1175,
      "step": 17349
    },
    {
      "epoch": 0.6213404480097409,
      "grad_norm": 1.4829490184783936,
      "learning_rate": 6.623372594889358e-05,
      "loss": 1.5191,
      "step": 17350
    },
    {
      "epoch": 0.6213762601393091,
      "grad_norm": 1.4071301221847534,
      "learning_rate": 6.622280840605005e-05,
      "loss": 1.3898,
      "step": 17351
    },
    {
      "epoch": 0.6214120722688775,
      "grad_norm": 3.8605194091796875,
      "learning_rate": 6.621189131763505e-05,
      "loss": 1.3449,
      "step": 17352
    },
    {
      "epoch": 0.6214478843984458,
      "grad_norm": 1.5796082019805908,
      "learning_rate": 6.620097468379548e-05,
      "loss": 1.176,
      "step": 17353
    },
    {
      "epoch": 0.621483696528014,
      "grad_norm": 1.4441759586334229,
      "learning_rate": 6.619005850467818e-05,
      "loss": 1.422,
      "step": 17354
    },
    {
      "epoch": 0.6215195086575823,
      "grad_norm": 1.868899941444397,
      "learning_rate": 6.617914278043005e-05,
      "loss": 1.5502,
      "step": 17355
    },
    {
      "epoch": 0.6215553207871506,
      "grad_norm": 1.4756184816360474,
      "learning_rate": 6.616822751119792e-05,
      "loss": 1.4594,
      "step": 17356
    },
    {
      "epoch": 0.6215911329167189,
      "grad_norm": 1.7463960647583008,
      "learning_rate": 6.615731269712864e-05,
      "loss": 1.4894,
      "step": 17357
    },
    {
      "epoch": 0.6216269450462871,
      "grad_norm": 1.8824423551559448,
      "learning_rate": 6.614639833836908e-05,
      "loss": 1.4013,
      "step": 17358
    },
    {
      "epoch": 0.6216627571758555,
      "grad_norm": 1.6457586288452148,
      "learning_rate": 6.613548443506605e-05,
      "loss": 1.6566,
      "step": 17359
    },
    {
      "epoch": 0.6216985693054238,
      "grad_norm": 2.025904893875122,
      "learning_rate": 6.612457098736642e-05,
      "loss": 1.5478,
      "step": 17360
    },
    {
      "epoch": 0.621734381434992,
      "grad_norm": 1.5645883083343506,
      "learning_rate": 6.611365799541695e-05,
      "loss": 1.5564,
      "step": 17361
    },
    {
      "epoch": 0.6217701935645603,
      "grad_norm": 1.3645058870315552,
      "learning_rate": 6.610274545936455e-05,
      "loss": 1.4919,
      "step": 17362
    },
    {
      "epoch": 0.6218060056941286,
      "grad_norm": 1.9030429124832153,
      "learning_rate": 6.609183337935594e-05,
      "loss": 1.7562,
      "step": 17363
    },
    {
      "epoch": 0.6218418178236969,
      "grad_norm": 1.3259434700012207,
      "learning_rate": 6.6080921755538e-05,
      "loss": 1.4657,
      "step": 17364
    },
    {
      "epoch": 0.6218776299532651,
      "grad_norm": 1.6076582670211792,
      "learning_rate": 6.607001058805749e-05,
      "loss": 1.5636,
      "step": 17365
    },
    {
      "epoch": 0.6219134420828335,
      "grad_norm": 1.4271641969680786,
      "learning_rate": 6.605909987706125e-05,
      "loss": 1.3398,
      "step": 17366
    },
    {
      "epoch": 0.6219492542124018,
      "grad_norm": 1.5322728157043457,
      "learning_rate": 6.604818962269602e-05,
      "loss": 1.1012,
      "step": 17367
    },
    {
      "epoch": 0.62198506634197,
      "grad_norm": 1.7007532119750977,
      "learning_rate": 6.603727982510859e-05,
      "loss": 1.3559,
      "step": 17368
    },
    {
      "epoch": 0.6220208784715383,
      "grad_norm": 1.661720633506775,
      "learning_rate": 6.602637048444578e-05,
      "loss": 1.6362,
      "step": 17369
    },
    {
      "epoch": 0.6220566906011066,
      "grad_norm": 1.6217849254608154,
      "learning_rate": 6.60154616008543e-05,
      "loss": 1.1952,
      "step": 17370
    },
    {
      "epoch": 0.6220925027306748,
      "grad_norm": 1.706960678100586,
      "learning_rate": 6.600455317448098e-05,
      "loss": 1.4172,
      "step": 17371
    },
    {
      "epoch": 0.6221283148602431,
      "grad_norm": 2.1763417720794678,
      "learning_rate": 6.599364520547251e-05,
      "loss": 1.653,
      "step": 17372
    },
    {
      "epoch": 0.6221641269898115,
      "grad_norm": 1.2412214279174805,
      "learning_rate": 6.598273769397572e-05,
      "loss": 1.522,
      "step": 17373
    },
    {
      "epoch": 0.6221999391193798,
      "grad_norm": 2.1275556087493896,
      "learning_rate": 6.597183064013728e-05,
      "loss": 1.3012,
      "step": 17374
    },
    {
      "epoch": 0.622235751248948,
      "grad_norm": 1.7945011854171753,
      "learning_rate": 6.5960924044104e-05,
      "loss": 1.3099,
      "step": 17375
    },
    {
      "epoch": 0.6222715633785163,
      "grad_norm": 1.436598300933838,
      "learning_rate": 6.595001790602255e-05,
      "loss": 1.347,
      "step": 17376
    },
    {
      "epoch": 0.6223073755080846,
      "grad_norm": 1.3291562795639038,
      "learning_rate": 6.593911222603969e-05,
      "loss": 1.2779,
      "step": 17377
    },
    {
      "epoch": 0.6223431876376528,
      "grad_norm": 1.5897414684295654,
      "learning_rate": 6.592820700430215e-05,
      "loss": 1.7034,
      "step": 17378
    },
    {
      "epoch": 0.6223789997672211,
      "grad_norm": 1.8603501319885254,
      "learning_rate": 6.591730224095663e-05,
      "loss": 1.4712,
      "step": 17379
    },
    {
      "epoch": 0.6224148118967895,
      "grad_norm": 2.969459056854248,
      "learning_rate": 6.590639793614985e-05,
      "loss": 1.3636,
      "step": 17380
    },
    {
      "epoch": 0.6224506240263578,
      "grad_norm": 2.0669147968292236,
      "learning_rate": 6.589549409002851e-05,
      "loss": 1.5975,
      "step": 17381
    },
    {
      "epoch": 0.622486436155926,
      "grad_norm": 2.064814329147339,
      "learning_rate": 6.588459070273931e-05,
      "loss": 1.3923,
      "step": 17382
    },
    {
      "epoch": 0.6225222482854943,
      "grad_norm": 1.8036562204360962,
      "learning_rate": 6.58736877744289e-05,
      "loss": 1.8272,
      "step": 17383
    },
    {
      "epoch": 0.6225580604150626,
      "grad_norm": 1.964530348777771,
      "learning_rate": 6.586278530524405e-05,
      "loss": 1.5506,
      "step": 17384
    },
    {
      "epoch": 0.6225938725446308,
      "grad_norm": 1.5007630586624146,
      "learning_rate": 6.58518832953314e-05,
      "loss": 1.519,
      "step": 17385
    },
    {
      "epoch": 0.6226296846741991,
      "grad_norm": 1.749772310256958,
      "learning_rate": 6.584098174483754e-05,
      "loss": 1.2044,
      "step": 17386
    },
    {
      "epoch": 0.6226654968037675,
      "grad_norm": 1.545983910560608,
      "learning_rate": 6.583008065390925e-05,
      "loss": 1.4835,
      "step": 17387
    },
    {
      "epoch": 0.6227013089333358,
      "grad_norm": 1.9022502899169922,
      "learning_rate": 6.581918002269315e-05,
      "loss": 1.415,
      "step": 17388
    },
    {
      "epoch": 0.622737121062904,
      "grad_norm": 1.8049886226654053,
      "learning_rate": 6.58082798513359e-05,
      "loss": 1.5487,
      "step": 17389
    },
    {
      "epoch": 0.6227729331924723,
      "grad_norm": 1.7206758260726929,
      "learning_rate": 6.579738013998411e-05,
      "loss": 1.6095,
      "step": 17390
    },
    {
      "epoch": 0.6228087453220406,
      "grad_norm": 1.4274195432662964,
      "learning_rate": 6.578648088878449e-05,
      "loss": 1.5214,
      "step": 17391
    },
    {
      "epoch": 0.6228445574516088,
      "grad_norm": 2.0439743995666504,
      "learning_rate": 6.577558209788362e-05,
      "loss": 1.6615,
      "step": 17392
    },
    {
      "epoch": 0.6228803695811771,
      "grad_norm": 1.6557294130325317,
      "learning_rate": 6.576468376742815e-05,
      "loss": 1.5092,
      "step": 17393
    },
    {
      "epoch": 0.6229161817107455,
      "grad_norm": 1.3613946437835693,
      "learning_rate": 6.575378589756472e-05,
      "loss": 1.4136,
      "step": 17394
    },
    {
      "epoch": 0.6229519938403137,
      "grad_norm": 1.593880295753479,
      "learning_rate": 6.574288848843988e-05,
      "loss": 1.2501,
      "step": 17395
    },
    {
      "epoch": 0.622987805969882,
      "grad_norm": 2.1387696266174316,
      "learning_rate": 6.573199154020033e-05,
      "loss": 1.563,
      "step": 17396
    },
    {
      "epoch": 0.6230236180994503,
      "grad_norm": 1.517091989517212,
      "learning_rate": 6.57210950529926e-05,
      "loss": 1.4276,
      "step": 17397
    },
    {
      "epoch": 0.6230594302290186,
      "grad_norm": 1.2697229385375977,
      "learning_rate": 6.571019902696335e-05,
      "loss": 1.281,
      "step": 17398
    },
    {
      "epoch": 0.6230952423585868,
      "grad_norm": 1.6039862632751465,
      "learning_rate": 6.569930346225909e-05,
      "loss": 1.5502,
      "step": 17399
    },
    {
      "epoch": 0.6231310544881551,
      "grad_norm": 1.4640159606933594,
      "learning_rate": 6.56884083590265e-05,
      "loss": 1.2747,
      "step": 17400
    },
    {
      "epoch": 0.6231668666177235,
      "grad_norm": 1.5991400480270386,
      "learning_rate": 6.567751371741209e-05,
      "loss": 1.6825,
      "step": 17401
    },
    {
      "epoch": 0.6232026787472917,
      "grad_norm": 1.5831888914108276,
      "learning_rate": 6.566661953756248e-05,
      "loss": 1.3705,
      "step": 17402
    },
    {
      "epoch": 0.62323849087686,
      "grad_norm": 1.7924561500549316,
      "learning_rate": 6.565572581962425e-05,
      "loss": 1.5186,
      "step": 17403
    },
    {
      "epoch": 0.6232743030064283,
      "grad_norm": 2.017235040664673,
      "learning_rate": 6.564483256374386e-05,
      "loss": 1.1964,
      "step": 17404
    },
    {
      "epoch": 0.6233101151359965,
      "grad_norm": 1.7464269399642944,
      "learning_rate": 6.5633939770068e-05,
      "loss": 1.7076,
      "step": 17405
    },
    {
      "epoch": 0.6233459272655648,
      "grad_norm": 1.9727259874343872,
      "learning_rate": 6.562304743874308e-05,
      "loss": 1.4771,
      "step": 17406
    },
    {
      "epoch": 0.6233817393951331,
      "grad_norm": 1.5742576122283936,
      "learning_rate": 6.561215556991578e-05,
      "loss": 1.3384,
      "step": 17407
    },
    {
      "epoch": 0.6234175515247015,
      "grad_norm": 1.320135235786438,
      "learning_rate": 6.56012641637325e-05,
      "loss": 1.4939,
      "step": 17408
    },
    {
      "epoch": 0.6234533636542697,
      "grad_norm": 1.445614218711853,
      "learning_rate": 6.559037322033991e-05,
      "loss": 1.5254,
      "step": 17409
    },
    {
      "epoch": 0.623489175783838,
      "grad_norm": 1.6739253997802734,
      "learning_rate": 6.55794827398844e-05,
      "loss": 1.1231,
      "step": 17410
    },
    {
      "epoch": 0.6235249879134063,
      "grad_norm": 1.5012933015823364,
      "learning_rate": 6.556859272251261e-05,
      "loss": 1.7445,
      "step": 17411
    },
    {
      "epoch": 0.6235608000429745,
      "grad_norm": 1.6745601892471313,
      "learning_rate": 6.555770316837098e-05,
      "loss": 1.4139,
      "step": 17412
    },
    {
      "epoch": 0.6235966121725428,
      "grad_norm": 1.7321993112564087,
      "learning_rate": 6.554681407760598e-05,
      "loss": 1.4242,
      "step": 17413
    },
    {
      "epoch": 0.6236324243021111,
      "grad_norm": 1.5765000581741333,
      "learning_rate": 6.553592545036421e-05,
      "loss": 1.5936,
      "step": 17414
    },
    {
      "epoch": 0.6236682364316795,
      "grad_norm": 2.3662610054016113,
      "learning_rate": 6.552503728679204e-05,
      "loss": 1.3233,
      "step": 17415
    },
    {
      "epoch": 0.6237040485612477,
      "grad_norm": 1.363168478012085,
      "learning_rate": 6.551414958703611e-05,
      "loss": 1.3414,
      "step": 17416
    },
    {
      "epoch": 0.623739860690816,
      "grad_norm": 1.336014986038208,
      "learning_rate": 6.550326235124274e-05,
      "loss": 1.4875,
      "step": 17417
    },
    {
      "epoch": 0.6237756728203843,
      "grad_norm": 1.4121977090835571,
      "learning_rate": 6.549237557955854e-05,
      "loss": 1.8272,
      "step": 17418
    },
    {
      "epoch": 0.6238114849499525,
      "grad_norm": 1.8297215700149536,
      "learning_rate": 6.54814892721299e-05,
      "loss": 1.1333,
      "step": 17419
    },
    {
      "epoch": 0.6238472970795208,
      "grad_norm": 1.8196252584457397,
      "learning_rate": 6.547060342910324e-05,
      "loss": 1.7198,
      "step": 17420
    },
    {
      "epoch": 0.6238831092090891,
      "grad_norm": 2.3765392303466797,
      "learning_rate": 6.545971805062514e-05,
      "loss": 1.7214,
      "step": 17421
    },
    {
      "epoch": 0.6239189213386575,
      "grad_norm": 1.7555404901504517,
      "learning_rate": 6.544883313684193e-05,
      "loss": 1.4639,
      "step": 17422
    },
    {
      "epoch": 0.6239547334682257,
      "grad_norm": 1.5131053924560547,
      "learning_rate": 6.543794868790015e-05,
      "loss": 1.2069,
      "step": 17423
    },
    {
      "epoch": 0.623990545597794,
      "grad_norm": 1.5095545053482056,
      "learning_rate": 6.542706470394614e-05,
      "loss": 1.6183,
      "step": 17424
    },
    {
      "epoch": 0.6240263577273623,
      "grad_norm": 2.3657705783843994,
      "learning_rate": 6.54161811851264e-05,
      "loss": 1.8115,
      "step": 17425
    },
    {
      "epoch": 0.6240621698569305,
      "grad_norm": 1.826514482498169,
      "learning_rate": 6.540529813158732e-05,
      "loss": 1.5298,
      "step": 17426
    },
    {
      "epoch": 0.6240979819864988,
      "grad_norm": 1.7273763418197632,
      "learning_rate": 6.539441554347537e-05,
      "loss": 1.4563,
      "step": 17427
    },
    {
      "epoch": 0.6241337941160671,
      "grad_norm": 2.3398561477661133,
      "learning_rate": 6.538353342093689e-05,
      "loss": 1.6562,
      "step": 17428
    },
    {
      "epoch": 0.6241696062456354,
      "grad_norm": 1.5986614227294922,
      "learning_rate": 6.537265176411831e-05,
      "loss": 1.5047,
      "step": 17429
    },
    {
      "epoch": 0.6242054183752037,
      "grad_norm": 1.5671006441116333,
      "learning_rate": 6.536177057316605e-05,
      "loss": 1.452,
      "step": 17430
    },
    {
      "epoch": 0.624241230504772,
      "grad_norm": 1.5327402353286743,
      "learning_rate": 6.535088984822647e-05,
      "loss": 1.5385,
      "step": 17431
    },
    {
      "epoch": 0.6242770426343403,
      "grad_norm": 1.9305261373519897,
      "learning_rate": 6.5340009589446e-05,
      "loss": 1.5302,
      "step": 17432
    },
    {
      "epoch": 0.6243128547639085,
      "grad_norm": 1.7964725494384766,
      "learning_rate": 6.532912979697095e-05,
      "loss": 1.6203,
      "step": 17433
    },
    {
      "epoch": 0.6243486668934768,
      "grad_norm": 1.4314013719558716,
      "learning_rate": 6.531825047094778e-05,
      "loss": 1.6102,
      "step": 17434
    },
    {
      "epoch": 0.6243844790230451,
      "grad_norm": 1.3836110830307007,
      "learning_rate": 6.530737161152278e-05,
      "loss": 1.6876,
      "step": 17435
    },
    {
      "epoch": 0.6244202911526134,
      "grad_norm": 1.6772618293762207,
      "learning_rate": 6.529649321884237e-05,
      "loss": 1.4962,
      "step": 17436
    },
    {
      "epoch": 0.6244561032821817,
      "grad_norm": 1.6304844617843628,
      "learning_rate": 6.528561529305289e-05,
      "loss": 1.2627,
      "step": 17437
    },
    {
      "epoch": 0.62449191541175,
      "grad_norm": 1.0868332386016846,
      "learning_rate": 6.527473783430064e-05,
      "loss": 1.2082,
      "step": 17438
    },
    {
      "epoch": 0.6245277275413182,
      "grad_norm": 1.9123562574386597,
      "learning_rate": 6.526386084273202e-05,
      "loss": 1.535,
      "step": 17439
    },
    {
      "epoch": 0.6245635396708865,
      "grad_norm": 1.7259448766708374,
      "learning_rate": 6.525298431849334e-05,
      "loss": 1.6374,
      "step": 17440
    },
    {
      "epoch": 0.6245993518004548,
      "grad_norm": 1.6573362350463867,
      "learning_rate": 6.524210826173094e-05,
      "loss": 1.4254,
      "step": 17441
    },
    {
      "epoch": 0.624635163930023,
      "grad_norm": 1.7752695083618164,
      "learning_rate": 6.523123267259113e-05,
      "loss": 1.3835,
      "step": 17442
    },
    {
      "epoch": 0.6246709760595914,
      "grad_norm": 2.4315598011016846,
      "learning_rate": 6.522035755122024e-05,
      "loss": 1.5414,
      "step": 17443
    },
    {
      "epoch": 0.6247067881891597,
      "grad_norm": 1.4300681352615356,
      "learning_rate": 6.520948289776459e-05,
      "loss": 1.7243,
      "step": 17444
    },
    {
      "epoch": 0.624742600318728,
      "grad_norm": 1.8628979921340942,
      "learning_rate": 6.519860871237046e-05,
      "loss": 1.4246,
      "step": 17445
    },
    {
      "epoch": 0.6247784124482962,
      "grad_norm": 2.1132607460021973,
      "learning_rate": 6.518773499518418e-05,
      "loss": 1.5807,
      "step": 17446
    },
    {
      "epoch": 0.6248142245778645,
      "grad_norm": 1.2737668752670288,
      "learning_rate": 6.517686174635198e-05,
      "loss": 1.4673,
      "step": 17447
    },
    {
      "epoch": 0.6248500367074328,
      "grad_norm": 1.5888752937316895,
      "learning_rate": 6.516598896602022e-05,
      "loss": 1.4605,
      "step": 17448
    },
    {
      "epoch": 0.624885848837001,
      "grad_norm": 1.590665578842163,
      "learning_rate": 6.515511665433513e-05,
      "loss": 1.5737,
      "step": 17449
    },
    {
      "epoch": 0.6249216609665694,
      "grad_norm": 2.3191263675689697,
      "learning_rate": 6.514424481144301e-05,
      "loss": 1.3844,
      "step": 17450
    },
    {
      "epoch": 0.6249574730961377,
      "grad_norm": 1.6052703857421875,
      "learning_rate": 6.513337343749008e-05,
      "loss": 1.6712,
      "step": 17451
    },
    {
      "epoch": 0.624993285225706,
      "grad_norm": 1.7228527069091797,
      "learning_rate": 6.512250253262268e-05,
      "loss": 1.4053,
      "step": 17452
    },
    {
      "epoch": 0.6250290973552742,
      "grad_norm": 2.069012403488159,
      "learning_rate": 6.511163209698701e-05,
      "loss": 1.5595,
      "step": 17453
    },
    {
      "epoch": 0.6250649094848425,
      "grad_norm": 1.8847413063049316,
      "learning_rate": 6.510076213072932e-05,
      "loss": 1.4639,
      "step": 17454
    },
    {
      "epoch": 0.6251007216144108,
      "grad_norm": 1.9726908206939697,
      "learning_rate": 6.508989263399588e-05,
      "loss": 1.2424,
      "step": 17455
    },
    {
      "epoch": 0.625136533743979,
      "grad_norm": 1.4060355424880981,
      "learning_rate": 6.507902360693286e-05,
      "loss": 1.4809,
      "step": 17456
    },
    {
      "epoch": 0.6251723458735474,
      "grad_norm": 1.643329381942749,
      "learning_rate": 6.506815504968657e-05,
      "loss": 1.4356,
      "step": 17457
    },
    {
      "epoch": 0.6252081580031157,
      "grad_norm": 1.3762401342391968,
      "learning_rate": 6.505728696240316e-05,
      "loss": 1.3249,
      "step": 17458
    },
    {
      "epoch": 0.625243970132684,
      "grad_norm": 1.967424988746643,
      "learning_rate": 6.504641934522892e-05,
      "loss": 1.3091,
      "step": 17459
    },
    {
      "epoch": 0.6252797822622522,
      "grad_norm": 1.5457974672317505,
      "learning_rate": 6.503555219830999e-05,
      "loss": 1.1949,
      "step": 17460
    },
    {
      "epoch": 0.6253155943918205,
      "grad_norm": 2.106031656265259,
      "learning_rate": 6.502468552179263e-05,
      "loss": 1.6295,
      "step": 17461
    },
    {
      "epoch": 0.6253514065213888,
      "grad_norm": 1.3233288526535034,
      "learning_rate": 6.501381931582297e-05,
      "loss": 1.4302,
      "step": 17462
    },
    {
      "epoch": 0.625387218650957,
      "grad_norm": 2.170650005340576,
      "learning_rate": 6.500295358054729e-05,
      "loss": 1.4235,
      "step": 17463
    },
    {
      "epoch": 0.6254230307805254,
      "grad_norm": 1.5497565269470215,
      "learning_rate": 6.499208831611172e-05,
      "loss": 1.2203,
      "step": 17464
    },
    {
      "epoch": 0.6254588429100937,
      "grad_norm": 2.3494434356689453,
      "learning_rate": 6.498122352266242e-05,
      "loss": 1.5483,
      "step": 17465
    },
    {
      "epoch": 0.625494655039662,
      "grad_norm": 1.4786423444747925,
      "learning_rate": 6.497035920034561e-05,
      "loss": 1.4961,
      "step": 17466
    },
    {
      "epoch": 0.6255304671692302,
      "grad_norm": 1.964049220085144,
      "learning_rate": 6.49594953493074e-05,
      "loss": 1.5353,
      "step": 17467
    },
    {
      "epoch": 0.6255662792987985,
      "grad_norm": 1.518816351890564,
      "learning_rate": 6.494863196969403e-05,
      "loss": 1.7526,
      "step": 17468
    },
    {
      "epoch": 0.6256020914283668,
      "grad_norm": 1.3744655847549438,
      "learning_rate": 6.493776906165155e-05,
      "loss": 1.5259,
      "step": 17469
    },
    {
      "epoch": 0.625637903557935,
      "grad_norm": 1.8796976804733276,
      "learning_rate": 6.49269066253262e-05,
      "loss": 1.6298,
      "step": 17470
    },
    {
      "epoch": 0.6256737156875034,
      "grad_norm": 1.5563455820083618,
      "learning_rate": 6.491604466086405e-05,
      "loss": 1.4925,
      "step": 17471
    },
    {
      "epoch": 0.6257095278170717,
      "grad_norm": 1.5489403009414673,
      "learning_rate": 6.49051831684113e-05,
      "loss": 1.2473,
      "step": 17472
    },
    {
      "epoch": 0.62574533994664,
      "grad_norm": 1.5099996328353882,
      "learning_rate": 6.489432214811403e-05,
      "loss": 1.2018,
      "step": 17473
    },
    {
      "epoch": 0.6257811520762082,
      "grad_norm": 1.9059803485870361,
      "learning_rate": 6.488346160011835e-05,
      "loss": 1.2401,
      "step": 17474
    },
    {
      "epoch": 0.6258169642057765,
      "grad_norm": 1.7727347612380981,
      "learning_rate": 6.487260152457041e-05,
      "loss": 1.5937,
      "step": 17475
    },
    {
      "epoch": 0.6258527763353448,
      "grad_norm": 1.7974789142608643,
      "learning_rate": 6.486174192161632e-05,
      "loss": 1.5348,
      "step": 17476
    },
    {
      "epoch": 0.625888588464913,
      "grad_norm": 1.85459566116333,
      "learning_rate": 6.485088279140214e-05,
      "loss": 1.7568,
      "step": 17477
    },
    {
      "epoch": 0.6259244005944814,
      "grad_norm": 1.5206538438796997,
      "learning_rate": 6.484002413407401e-05,
      "loss": 1.3962,
      "step": 17478
    },
    {
      "epoch": 0.6259602127240497,
      "grad_norm": 1.4857813119888306,
      "learning_rate": 6.4829165949778e-05,
      "loss": 1.5086,
      "step": 17479
    },
    {
      "epoch": 0.6259960248536179,
      "grad_norm": 1.4960447549819946,
      "learning_rate": 6.481830823866018e-05,
      "loss": 1.3399,
      "step": 17480
    },
    {
      "epoch": 0.6260318369831862,
      "grad_norm": 1.2727571725845337,
      "learning_rate": 6.480745100086668e-05,
      "loss": 1.4466,
      "step": 17481
    },
    {
      "epoch": 0.6260676491127545,
      "grad_norm": 1.508490800857544,
      "learning_rate": 6.479659423654352e-05,
      "loss": 1.4005,
      "step": 17482
    },
    {
      "epoch": 0.6261034612423227,
      "grad_norm": 1.6986658573150635,
      "learning_rate": 6.478573794583673e-05,
      "loss": 1.4465,
      "step": 17483
    },
    {
      "epoch": 0.626139273371891,
      "grad_norm": 1.3787263631820679,
      "learning_rate": 6.477488212889246e-05,
      "loss": 1.4511,
      "step": 17484
    },
    {
      "epoch": 0.6261750855014594,
      "grad_norm": 1.7577788829803467,
      "learning_rate": 6.476402678585669e-05,
      "loss": 1.1861,
      "step": 17485
    },
    {
      "epoch": 0.6262108976310277,
      "grad_norm": 1.4000847339630127,
      "learning_rate": 6.47531719168755e-05,
      "loss": 1.5713,
      "step": 17486
    },
    {
      "epoch": 0.6262467097605959,
      "grad_norm": 2.0742688179016113,
      "learning_rate": 6.474231752209492e-05,
      "loss": 1.5913,
      "step": 17487
    },
    {
      "epoch": 0.6262825218901642,
      "grad_norm": 1.7926561832427979,
      "learning_rate": 6.473146360166098e-05,
      "loss": 1.533,
      "step": 17488
    },
    {
      "epoch": 0.6263183340197325,
      "grad_norm": 1.5476453304290771,
      "learning_rate": 6.472061015571968e-05,
      "loss": 1.4394,
      "step": 17489
    },
    {
      "epoch": 0.6263541461493007,
      "grad_norm": 1.240106225013733,
      "learning_rate": 6.47097571844171e-05,
      "loss": 1.4792,
      "step": 17490
    },
    {
      "epoch": 0.626389958278869,
      "grad_norm": 1.3444880247116089,
      "learning_rate": 6.469890468789922e-05,
      "loss": 1.5799,
      "step": 17491
    },
    {
      "epoch": 0.6264257704084374,
      "grad_norm": 1.7917805910110474,
      "learning_rate": 6.468805266631199e-05,
      "loss": 1.5301,
      "step": 17492
    },
    {
      "epoch": 0.6264615825380057,
      "grad_norm": 2.6330838203430176,
      "learning_rate": 6.467720111980151e-05,
      "loss": 1.6501,
      "step": 17493
    },
    {
      "epoch": 0.6264973946675739,
      "grad_norm": 1.631439208984375,
      "learning_rate": 6.466635004851367e-05,
      "loss": 1.368,
      "step": 17494
    },
    {
      "epoch": 0.6265332067971422,
      "grad_norm": 1.8269366025924683,
      "learning_rate": 6.46554994525946e-05,
      "loss": 1.6264,
      "step": 17495
    },
    {
      "epoch": 0.6265690189267105,
      "grad_norm": 1.577916145324707,
      "learning_rate": 6.46446493321901e-05,
      "loss": 1.3685,
      "step": 17496
    },
    {
      "epoch": 0.6266048310562787,
      "grad_norm": 1.450053095817566,
      "learning_rate": 6.46337996874463e-05,
      "loss": 1.572,
      "step": 17497
    },
    {
      "epoch": 0.626640643185847,
      "grad_norm": 1.6219733953475952,
      "learning_rate": 6.462295051850907e-05,
      "loss": 1.4317,
      "step": 17498
    },
    {
      "epoch": 0.6266764553154154,
      "grad_norm": 1.7506242990493774,
      "learning_rate": 6.461210182552444e-05,
      "loss": 1.5448,
      "step": 17499
    },
    {
      "epoch": 0.6267122674449837,
      "grad_norm": 1.4958577156066895,
      "learning_rate": 6.460125360863835e-05,
      "loss": 1.1158,
      "step": 17500
    },
    {
      "epoch": 0.6267480795745519,
      "grad_norm": 1.796212077140808,
      "learning_rate": 6.459040586799666e-05,
      "loss": 1.5875,
      "step": 17501
    },
    {
      "epoch": 0.6267838917041202,
      "grad_norm": 1.9976341724395752,
      "learning_rate": 6.457955860374545e-05,
      "loss": 1.4557,
      "step": 17502
    },
    {
      "epoch": 0.6268197038336885,
      "grad_norm": 1.7986866235733032,
      "learning_rate": 6.456871181603054e-05,
      "loss": 1.7378,
      "step": 17503
    },
    {
      "epoch": 0.6268555159632567,
      "grad_norm": 1.322219967842102,
      "learning_rate": 6.455786550499796e-05,
      "loss": 1.3182,
      "step": 17504
    },
    {
      "epoch": 0.626891328092825,
      "grad_norm": 1.5401942729949951,
      "learning_rate": 6.454701967079354e-05,
      "loss": 1.5181,
      "step": 17505
    },
    {
      "epoch": 0.6269271402223934,
      "grad_norm": 2.3169422149658203,
      "learning_rate": 6.453617431356327e-05,
      "loss": 1.8039,
      "step": 17506
    },
    {
      "epoch": 0.6269629523519616,
      "grad_norm": 1.2413296699523926,
      "learning_rate": 6.452532943345298e-05,
      "loss": 1.6122,
      "step": 17507
    },
    {
      "epoch": 0.6269987644815299,
      "grad_norm": 1.2607512474060059,
      "learning_rate": 6.451448503060868e-05,
      "loss": 1.3618,
      "step": 17508
    },
    {
      "epoch": 0.6270345766110982,
      "grad_norm": 1.9498603343963623,
      "learning_rate": 6.45036411051762e-05,
      "loss": 1.2638,
      "step": 17509
    },
    {
      "epoch": 0.6270703887406665,
      "grad_norm": 1.3453820943832397,
      "learning_rate": 6.449279765730141e-05,
      "loss": 1.6058,
      "step": 17510
    },
    {
      "epoch": 0.6271062008702347,
      "grad_norm": 1.8720459938049316,
      "learning_rate": 6.448195468713028e-05,
      "loss": 1.231,
      "step": 17511
    },
    {
      "epoch": 0.627142012999803,
      "grad_norm": 1.402961254119873,
      "learning_rate": 6.447111219480857e-05,
      "loss": 1.5311,
      "step": 17512
    },
    {
      "epoch": 0.6271778251293714,
      "grad_norm": 1.858162760734558,
      "learning_rate": 6.446027018048228e-05,
      "loss": 1.5916,
      "step": 17513
    },
    {
      "epoch": 0.6272136372589396,
      "grad_norm": 2.5841658115386963,
      "learning_rate": 6.444942864429713e-05,
      "loss": 1.3853,
      "step": 17514
    },
    {
      "epoch": 0.6272494493885079,
      "grad_norm": 1.7156753540039062,
      "learning_rate": 6.443858758639916e-05,
      "loss": 1.5192,
      "step": 17515
    },
    {
      "epoch": 0.6272852615180762,
      "grad_norm": 1.7334517240524292,
      "learning_rate": 6.442774700693408e-05,
      "loss": 1.2638,
      "step": 17516
    },
    {
      "epoch": 0.6273210736476444,
      "grad_norm": 1.4472137689590454,
      "learning_rate": 6.441690690604775e-05,
      "loss": 1.9229,
      "step": 17517
    },
    {
      "epoch": 0.6273568857772127,
      "grad_norm": 1.426712989807129,
      "learning_rate": 6.440606728388607e-05,
      "loss": 1.4067,
      "step": 17518
    },
    {
      "epoch": 0.627392697906781,
      "grad_norm": 1.6888214349746704,
      "learning_rate": 6.439522814059483e-05,
      "loss": 1.5726,
      "step": 17519
    },
    {
      "epoch": 0.6274285100363494,
      "grad_norm": 1.7400039434432983,
      "learning_rate": 6.438438947631989e-05,
      "loss": 1.3974,
      "step": 17520
    },
    {
      "epoch": 0.6274643221659176,
      "grad_norm": 1.7932146787643433,
      "learning_rate": 6.437355129120701e-05,
      "loss": 1.3313,
      "step": 17521
    },
    {
      "epoch": 0.6275001342954859,
      "grad_norm": 1.576096773147583,
      "learning_rate": 6.436271358540206e-05,
      "loss": 1.0929,
      "step": 17522
    },
    {
      "epoch": 0.6275359464250542,
      "grad_norm": 1.7114145755767822,
      "learning_rate": 6.435187635905082e-05,
      "loss": 1.0729,
      "step": 17523
    },
    {
      "epoch": 0.6275717585546224,
      "grad_norm": 1.7404381036758423,
      "learning_rate": 6.434103961229913e-05,
      "loss": 1.6533,
      "step": 17524
    },
    {
      "epoch": 0.6276075706841907,
      "grad_norm": 1.227808952331543,
      "learning_rate": 6.433020334529275e-05,
      "loss": 1.4551,
      "step": 17525
    },
    {
      "epoch": 0.627643382813759,
      "grad_norm": 1.4737696647644043,
      "learning_rate": 6.431936755817746e-05,
      "loss": 1.5414,
      "step": 17526
    },
    {
      "epoch": 0.6276791949433274,
      "grad_norm": 1.8455876111984253,
      "learning_rate": 6.430853225109908e-05,
      "loss": 1.9524,
      "step": 17527
    },
    {
      "epoch": 0.6277150070728956,
      "grad_norm": 1.4521657228469849,
      "learning_rate": 6.42976974242033e-05,
      "loss": 1.5804,
      "step": 17528
    },
    {
      "epoch": 0.6277508192024639,
      "grad_norm": 1.937605381011963,
      "learning_rate": 6.428686307763601e-05,
      "loss": 1.1705,
      "step": 17529
    },
    {
      "epoch": 0.6277866313320322,
      "grad_norm": 1.5884835720062256,
      "learning_rate": 6.427602921154287e-05,
      "loss": 1.5303,
      "step": 17530
    },
    {
      "epoch": 0.6278224434616004,
      "grad_norm": 1.7557952404022217,
      "learning_rate": 6.426519582606971e-05,
      "loss": 1.269,
      "step": 17531
    },
    {
      "epoch": 0.6278582555911687,
      "grad_norm": 1.5802031755447388,
      "learning_rate": 6.42543629213622e-05,
      "loss": 1.5768,
      "step": 17532
    },
    {
      "epoch": 0.627894067720737,
      "grad_norm": 1.7923305034637451,
      "learning_rate": 6.424353049756618e-05,
      "loss": 1.411,
      "step": 17533
    },
    {
      "epoch": 0.6279298798503054,
      "grad_norm": 1.7670484781265259,
      "learning_rate": 6.423269855482732e-05,
      "loss": 1.6283,
      "step": 17534
    },
    {
      "epoch": 0.6279656919798736,
      "grad_norm": 1.5500022172927856,
      "learning_rate": 6.422186709329134e-05,
      "loss": 1.5638,
      "step": 17535
    },
    {
      "epoch": 0.6280015041094419,
      "grad_norm": 1.745823860168457,
      "learning_rate": 6.421103611310402e-05,
      "loss": 1.4258,
      "step": 17536
    },
    {
      "epoch": 0.6280373162390102,
      "grad_norm": 1.6846976280212402,
      "learning_rate": 6.420020561441101e-05,
      "loss": 1.3898,
      "step": 17537
    },
    {
      "epoch": 0.6280731283685784,
      "grad_norm": 1.477970004081726,
      "learning_rate": 6.41893755973581e-05,
      "loss": 1.4848,
      "step": 17538
    },
    {
      "epoch": 0.6281089404981467,
      "grad_norm": 1.6424835920333862,
      "learning_rate": 6.417854606209091e-05,
      "loss": 1.3707,
      "step": 17539
    },
    {
      "epoch": 0.628144752627715,
      "grad_norm": 1.994249939918518,
      "learning_rate": 6.41677170087552e-05,
      "loss": 1.7501,
      "step": 17540
    },
    {
      "epoch": 0.6281805647572833,
      "grad_norm": 2.2836825847625732,
      "learning_rate": 6.41568884374966e-05,
      "loss": 1.3445,
      "step": 17541
    },
    {
      "epoch": 0.6282163768868516,
      "grad_norm": 1.7882944345474243,
      "learning_rate": 6.414606034846087e-05,
      "loss": 1.2719,
      "step": 17542
    },
    {
      "epoch": 0.6282521890164199,
      "grad_norm": 1.532196044921875,
      "learning_rate": 6.413523274179365e-05,
      "loss": 1.3419,
      "step": 17543
    },
    {
      "epoch": 0.6282880011459882,
      "grad_norm": 1.7860722541809082,
      "learning_rate": 6.412440561764059e-05,
      "loss": 1.613,
      "step": 17544
    },
    {
      "epoch": 0.6283238132755564,
      "grad_norm": 1.4192113876342773,
      "learning_rate": 6.411357897614738e-05,
      "loss": 1.2385,
      "step": 17545
    },
    {
      "epoch": 0.6283596254051247,
      "grad_norm": 1.6506589651107788,
      "learning_rate": 6.410275281745967e-05,
      "loss": 1.4374,
      "step": 17546
    },
    {
      "epoch": 0.628395437534693,
      "grad_norm": 1.6427680253982544,
      "learning_rate": 6.409192714172314e-05,
      "loss": 1.4561,
      "step": 17547
    },
    {
      "epoch": 0.6284312496642613,
      "grad_norm": 1.5878175497055054,
      "learning_rate": 6.408110194908338e-05,
      "loss": 1.3458,
      "step": 17548
    },
    {
      "epoch": 0.6284670617938296,
      "grad_norm": 1.747763752937317,
      "learning_rate": 6.407027723968611e-05,
      "loss": 1.3013,
      "step": 17549
    },
    {
      "epoch": 0.6285028739233979,
      "grad_norm": 1.7274367809295654,
      "learning_rate": 6.405945301367687e-05,
      "loss": 1.7029,
      "step": 17550
    },
    {
      "epoch": 0.6285386860529661,
      "grad_norm": 1.8182168006896973,
      "learning_rate": 6.404862927120134e-05,
      "loss": 1.3585,
      "step": 17551
    },
    {
      "epoch": 0.6285744981825344,
      "grad_norm": 1.4165538549423218,
      "learning_rate": 6.403780601240514e-05,
      "loss": 1.2135,
      "step": 17552
    },
    {
      "epoch": 0.6286103103121027,
      "grad_norm": 1.433510661125183,
      "learning_rate": 6.402698323743385e-05,
      "loss": 1.313,
      "step": 17553
    },
    {
      "epoch": 0.628646122441671,
      "grad_norm": 1.786746621131897,
      "learning_rate": 6.401616094643312e-05,
      "loss": 1.6242,
      "step": 17554
    },
    {
      "epoch": 0.6286819345712393,
      "grad_norm": 1.820902705192566,
      "learning_rate": 6.400533913954851e-05,
      "loss": 1.5524,
      "step": 17555
    },
    {
      "epoch": 0.6287177467008076,
      "grad_norm": 1.55777907371521,
      "learning_rate": 6.399451781692567e-05,
      "loss": 1.2882,
      "step": 17556
    },
    {
      "epoch": 0.6287535588303759,
      "grad_norm": 1.7453067302703857,
      "learning_rate": 6.398369697871011e-05,
      "loss": 1.6019,
      "step": 17557
    },
    {
      "epoch": 0.6287893709599441,
      "grad_norm": 1.8225655555725098,
      "learning_rate": 6.397287662504747e-05,
      "loss": 1.365,
      "step": 17558
    },
    {
      "epoch": 0.6288251830895124,
      "grad_norm": 1.931082010269165,
      "learning_rate": 6.39620567560833e-05,
      "loss": 1.4394,
      "step": 17559
    },
    {
      "epoch": 0.6288609952190807,
      "grad_norm": 2.121443271636963,
      "learning_rate": 6.395123737196316e-05,
      "loss": 1.3972,
      "step": 17560
    },
    {
      "epoch": 0.628896807348649,
      "grad_norm": 1.686209797859192,
      "learning_rate": 6.394041847283263e-05,
      "loss": 1.4342,
      "step": 17561
    },
    {
      "epoch": 0.6289326194782173,
      "grad_norm": 1.546787977218628,
      "learning_rate": 6.392960005883726e-05,
      "loss": 1.4053,
      "step": 17562
    },
    {
      "epoch": 0.6289684316077856,
      "grad_norm": 1.6761842966079712,
      "learning_rate": 6.391878213012258e-05,
      "loss": 1.4526,
      "step": 17563
    },
    {
      "epoch": 0.6290042437373539,
      "grad_norm": 2.2791762351989746,
      "learning_rate": 6.390796468683416e-05,
      "loss": 1.5521,
      "step": 17564
    },
    {
      "epoch": 0.6290400558669221,
      "grad_norm": 1.24517822265625,
      "learning_rate": 6.389714772911751e-05,
      "loss": 1.4677,
      "step": 17565
    },
    {
      "epoch": 0.6290758679964904,
      "grad_norm": 1.4214081764221191,
      "learning_rate": 6.388633125711816e-05,
      "loss": 1.5366,
      "step": 17566
    },
    {
      "epoch": 0.6291116801260587,
      "grad_norm": 1.4436924457550049,
      "learning_rate": 6.387551527098165e-05,
      "loss": 1.5554,
      "step": 17567
    },
    {
      "epoch": 0.6291474922556269,
      "grad_norm": 1.6681147813796997,
      "learning_rate": 6.386469977085348e-05,
      "loss": 1.1127,
      "step": 17568
    },
    {
      "epoch": 0.6291833043851953,
      "grad_norm": 2.224066734313965,
      "learning_rate": 6.385388475687918e-05,
      "loss": 1.6578,
      "step": 17569
    },
    {
      "epoch": 0.6292191165147636,
      "grad_norm": 1.7209625244140625,
      "learning_rate": 6.384307022920424e-05,
      "loss": 1.6013,
      "step": 17570
    },
    {
      "epoch": 0.6292549286443319,
      "grad_norm": 1.4986475706100464,
      "learning_rate": 6.383225618797412e-05,
      "loss": 1.4255,
      "step": 17571
    },
    {
      "epoch": 0.6292907407739001,
      "grad_norm": 1.6216480731964111,
      "learning_rate": 6.382144263333436e-05,
      "loss": 1.6899,
      "step": 17572
    },
    {
      "epoch": 0.6293265529034684,
      "grad_norm": 2.151085138320923,
      "learning_rate": 6.381062956543041e-05,
      "loss": 1.7799,
      "step": 17573
    },
    {
      "epoch": 0.6293623650330367,
      "grad_norm": 1.4433550834655762,
      "learning_rate": 6.379981698440778e-05,
      "loss": 1.4163,
      "step": 17574
    },
    {
      "epoch": 0.6293981771626049,
      "grad_norm": 1.6672604084014893,
      "learning_rate": 6.378900489041188e-05,
      "loss": 1.4949,
      "step": 17575
    },
    {
      "epoch": 0.6294339892921733,
      "grad_norm": 2.602228879928589,
      "learning_rate": 6.377819328358826e-05,
      "loss": 1.5507,
      "step": 17576
    },
    {
      "epoch": 0.6294698014217416,
      "grad_norm": 1.5993750095367432,
      "learning_rate": 6.37673821640823e-05,
      "loss": 1.6015,
      "step": 17577
    },
    {
      "epoch": 0.6295056135513099,
      "grad_norm": 1.3872603178024292,
      "learning_rate": 6.375657153203947e-05,
      "loss": 1.6021,
      "step": 17578
    },
    {
      "epoch": 0.6295414256808781,
      "grad_norm": 1.776228904724121,
      "learning_rate": 6.374576138760525e-05,
      "loss": 1.4317,
      "step": 17579
    },
    {
      "epoch": 0.6295772378104464,
      "grad_norm": 1.5714781284332275,
      "learning_rate": 6.3734951730925e-05,
      "loss": 1.1074,
      "step": 17580
    },
    {
      "epoch": 0.6296130499400147,
      "grad_norm": 1.8212718963623047,
      "learning_rate": 6.372414256214423e-05,
      "loss": 1.4209,
      "step": 17581
    },
    {
      "epoch": 0.6296488620695829,
      "grad_norm": 1.9626024961471558,
      "learning_rate": 6.37133338814083e-05,
      "loss": 1.3858,
      "step": 17582
    },
    {
      "epoch": 0.6296846741991513,
      "grad_norm": 1.6976439952850342,
      "learning_rate": 6.370252568886267e-05,
      "loss": 1.4777,
      "step": 17583
    },
    {
      "epoch": 0.6297204863287196,
      "grad_norm": 3.100534677505493,
      "learning_rate": 6.369171798465274e-05,
      "loss": 1.5667,
      "step": 17584
    },
    {
      "epoch": 0.6297562984582878,
      "grad_norm": 1.5115562677383423,
      "learning_rate": 6.368091076892392e-05,
      "loss": 1.5159,
      "step": 17585
    },
    {
      "epoch": 0.6297921105878561,
      "grad_norm": 1.9113214015960693,
      "learning_rate": 6.367010404182158e-05,
      "loss": 1.447,
      "step": 17586
    },
    {
      "epoch": 0.6298279227174244,
      "grad_norm": 1.8984917402267456,
      "learning_rate": 6.365929780349113e-05,
      "loss": 1.4439,
      "step": 17587
    },
    {
      "epoch": 0.6298637348469927,
      "grad_norm": 1.6823961734771729,
      "learning_rate": 6.3648492054078e-05,
      "loss": 1.5414,
      "step": 17588
    },
    {
      "epoch": 0.6298995469765609,
      "grad_norm": 1.8366305828094482,
      "learning_rate": 6.363768679372744e-05,
      "loss": 1.303,
      "step": 17589
    },
    {
      "epoch": 0.6299353591061293,
      "grad_norm": 1.6372846364974976,
      "learning_rate": 6.362688202258496e-05,
      "loss": 1.4375,
      "step": 17590
    },
    {
      "epoch": 0.6299711712356976,
      "grad_norm": 2.158540964126587,
      "learning_rate": 6.361607774079581e-05,
      "loss": 1.5602,
      "step": 17591
    },
    {
      "epoch": 0.6300069833652658,
      "grad_norm": 1.4774373769760132,
      "learning_rate": 6.360527394850547e-05,
      "loss": 1.3032,
      "step": 17592
    },
    {
      "epoch": 0.6300427954948341,
      "grad_norm": 1.7481532096862793,
      "learning_rate": 6.359447064585915e-05,
      "loss": 1.3593,
      "step": 17593
    },
    {
      "epoch": 0.6300786076244024,
      "grad_norm": 1.8624064922332764,
      "learning_rate": 6.358366783300231e-05,
      "loss": 1.2172,
      "step": 17594
    },
    {
      "epoch": 0.6301144197539706,
      "grad_norm": 1.4835082292556763,
      "learning_rate": 6.357286551008024e-05,
      "loss": 1.4949,
      "step": 17595
    },
    {
      "epoch": 0.6301502318835389,
      "grad_norm": 2.123321294784546,
      "learning_rate": 6.356206367723829e-05,
      "loss": 1.5935,
      "step": 17596
    },
    {
      "epoch": 0.6301860440131073,
      "grad_norm": 1.4491807222366333,
      "learning_rate": 6.355126233462179e-05,
      "loss": 1.6078,
      "step": 17597
    },
    {
      "epoch": 0.6302218561426756,
      "grad_norm": 1.65678870677948,
      "learning_rate": 6.354046148237597e-05,
      "loss": 1.5985,
      "step": 17598
    },
    {
      "epoch": 0.6302576682722438,
      "grad_norm": 1.9164150953292847,
      "learning_rate": 6.352966112064627e-05,
      "loss": 1.6386,
      "step": 17599
    },
    {
      "epoch": 0.6302934804018121,
      "grad_norm": 1.5138894319534302,
      "learning_rate": 6.351886124957789e-05,
      "loss": 1.5511,
      "step": 17600
    },
    {
      "epoch": 0.6303292925313804,
      "grad_norm": 1.7643005847930908,
      "learning_rate": 6.350806186931623e-05,
      "loss": 1.5784,
      "step": 17601
    },
    {
      "epoch": 0.6303651046609486,
      "grad_norm": 1.719307780265808,
      "learning_rate": 6.349726298000647e-05,
      "loss": 1.5924,
      "step": 17602
    },
    {
      "epoch": 0.6304009167905169,
      "grad_norm": 1.3820748329162598,
      "learning_rate": 6.3486464581794e-05,
      "loss": 1.4025,
      "step": 17603
    },
    {
      "epoch": 0.6304367289200853,
      "grad_norm": 1.675239086151123,
      "learning_rate": 6.347566667482401e-05,
      "loss": 1.4112,
      "step": 17604
    },
    {
      "epoch": 0.6304725410496536,
      "grad_norm": 1.8763279914855957,
      "learning_rate": 6.346486925924184e-05,
      "loss": 1.6343,
      "step": 17605
    },
    {
      "epoch": 0.6305083531792218,
      "grad_norm": 2.052152633666992,
      "learning_rate": 6.345407233519273e-05,
      "loss": 1.4135,
      "step": 17606
    },
    {
      "epoch": 0.6305441653087901,
      "grad_norm": 1.8708726167678833,
      "learning_rate": 6.344327590282189e-05,
      "loss": 1.5575,
      "step": 17607
    },
    {
      "epoch": 0.6305799774383584,
      "grad_norm": 2.31463360786438,
      "learning_rate": 6.343247996227469e-05,
      "loss": 1.5766,
      "step": 17608
    },
    {
      "epoch": 0.6306157895679266,
      "grad_norm": 1.8864593505859375,
      "learning_rate": 6.342168451369623e-05,
      "loss": 1.2702,
      "step": 17609
    },
    {
      "epoch": 0.6306516016974949,
      "grad_norm": 1.3110343217849731,
      "learning_rate": 6.341088955723189e-05,
      "loss": 1.0155,
      "step": 17610
    },
    {
      "epoch": 0.6306874138270633,
      "grad_norm": 1.729931116104126,
      "learning_rate": 6.340009509302676e-05,
      "loss": 1.5527,
      "step": 17611
    },
    {
      "epoch": 0.6307232259566316,
      "grad_norm": 1.4861717224121094,
      "learning_rate": 6.338930112122622e-05,
      "loss": 1.3048,
      "step": 17612
    },
    {
      "epoch": 0.6307590380861998,
      "grad_norm": 1.8507078886032104,
      "learning_rate": 6.337850764197539e-05,
      "loss": 1.3849,
      "step": 17613
    },
    {
      "epoch": 0.6307948502157681,
      "grad_norm": 1.3783198595046997,
      "learning_rate": 6.336771465541947e-05,
      "loss": 1.2361,
      "step": 17614
    },
    {
      "epoch": 0.6308306623453364,
      "grad_norm": 1.9116185903549194,
      "learning_rate": 6.33569221617037e-05,
      "loss": 1.5638,
      "step": 17615
    },
    {
      "epoch": 0.6308664744749046,
      "grad_norm": 1.7067476511001587,
      "learning_rate": 6.334613016097328e-05,
      "loss": 1.3243,
      "step": 17616
    },
    {
      "epoch": 0.6309022866044729,
      "grad_norm": 1.42518150806427,
      "learning_rate": 6.333533865337343e-05,
      "loss": 1.5246,
      "step": 17617
    },
    {
      "epoch": 0.6309380987340413,
      "grad_norm": 1.7045762538909912,
      "learning_rate": 6.332454763904925e-05,
      "loss": 1.5706,
      "step": 17618
    },
    {
      "epoch": 0.6309739108636095,
      "grad_norm": 1.6341084241867065,
      "learning_rate": 6.3313757118146e-05,
      "loss": 1.2484,
      "step": 17619
    },
    {
      "epoch": 0.6310097229931778,
      "grad_norm": 2.272749423980713,
      "learning_rate": 6.330296709080881e-05,
      "loss": 1.3682,
      "step": 17620
    },
    {
      "epoch": 0.6310455351227461,
      "grad_norm": 2.296658992767334,
      "learning_rate": 6.329217755718291e-05,
      "loss": 1.847,
      "step": 17621
    },
    {
      "epoch": 0.6310813472523144,
      "grad_norm": 1.2976129055023193,
      "learning_rate": 6.328138851741338e-05,
      "loss": 1.4858,
      "step": 17622
    },
    {
      "epoch": 0.6311171593818826,
      "grad_norm": 1.4732706546783447,
      "learning_rate": 6.32705999716454e-05,
      "loss": 1.0149,
      "step": 17623
    },
    {
      "epoch": 0.6311529715114509,
      "grad_norm": 2.097350597381592,
      "learning_rate": 6.325981192002413e-05,
      "loss": 1.7887,
      "step": 17624
    },
    {
      "epoch": 0.6311887836410193,
      "grad_norm": 1.4998894929885864,
      "learning_rate": 6.324902436269469e-05,
      "loss": 1.47,
      "step": 17625
    },
    {
      "epoch": 0.6312245957705875,
      "grad_norm": 2.4181065559387207,
      "learning_rate": 6.323823729980222e-05,
      "loss": 1.8616,
      "step": 17626
    },
    {
      "epoch": 0.6312604079001558,
      "grad_norm": 2.439805507659912,
      "learning_rate": 6.322745073149185e-05,
      "loss": 1.6384,
      "step": 17627
    },
    {
      "epoch": 0.6312962200297241,
      "grad_norm": 1.9213478565216064,
      "learning_rate": 6.321666465790872e-05,
      "loss": 1.6752,
      "step": 17628
    },
    {
      "epoch": 0.6313320321592923,
      "grad_norm": 1.7492610216140747,
      "learning_rate": 6.320587907919788e-05,
      "loss": 1.6131,
      "step": 17629
    },
    {
      "epoch": 0.6313678442888606,
      "grad_norm": 1.5439682006835938,
      "learning_rate": 6.319509399550452e-05,
      "loss": 1.3714,
      "step": 17630
    },
    {
      "epoch": 0.6314036564184289,
      "grad_norm": 1.481459617614746,
      "learning_rate": 6.318430940697367e-05,
      "loss": 1.449,
      "step": 17631
    },
    {
      "epoch": 0.6314394685479972,
      "grad_norm": 1.4452996253967285,
      "learning_rate": 6.317352531375045e-05,
      "loss": 1.2496,
      "step": 17632
    },
    {
      "epoch": 0.6314752806775655,
      "grad_norm": 1.4040240049362183,
      "learning_rate": 6.316274171597995e-05,
      "loss": 1.4423,
      "step": 17633
    },
    {
      "epoch": 0.6315110928071338,
      "grad_norm": 1.6512972116470337,
      "learning_rate": 6.315195861380722e-05,
      "loss": 1.3448,
      "step": 17634
    },
    {
      "epoch": 0.6315469049367021,
      "grad_norm": 1.6966760158538818,
      "learning_rate": 6.314117600737738e-05,
      "loss": 1.5408,
      "step": 17635
    },
    {
      "epoch": 0.6315827170662703,
      "grad_norm": 1.709964394569397,
      "learning_rate": 6.313039389683546e-05,
      "loss": 1.3916,
      "step": 17636
    },
    {
      "epoch": 0.6316185291958386,
      "grad_norm": 1.6564950942993164,
      "learning_rate": 6.311961228232654e-05,
      "loss": 1.6976,
      "step": 17637
    },
    {
      "epoch": 0.6316543413254069,
      "grad_norm": 1.531118631362915,
      "learning_rate": 6.310883116399567e-05,
      "loss": 1.3207,
      "step": 17638
    },
    {
      "epoch": 0.6316901534549751,
      "grad_norm": 2.0177090167999268,
      "learning_rate": 6.309805054198787e-05,
      "loss": 1.6174,
      "step": 17639
    },
    {
      "epoch": 0.6317259655845435,
      "grad_norm": 1.2285786867141724,
      "learning_rate": 6.308727041644824e-05,
      "loss": 1.6658,
      "step": 17640
    },
    {
      "epoch": 0.6317617777141118,
      "grad_norm": 1.6047492027282715,
      "learning_rate": 6.307649078752174e-05,
      "loss": 1.6755,
      "step": 17641
    },
    {
      "epoch": 0.6317975898436801,
      "grad_norm": 2.411289691925049,
      "learning_rate": 6.306571165535343e-05,
      "loss": 1.4401,
      "step": 17642
    },
    {
      "epoch": 0.6318334019732483,
      "grad_norm": 1.399155855178833,
      "learning_rate": 6.305493302008832e-05,
      "loss": 1.7149,
      "step": 17643
    },
    {
      "epoch": 0.6318692141028166,
      "grad_norm": 1.4516044855117798,
      "learning_rate": 6.304415488187145e-05,
      "loss": 1.4904,
      "step": 17644
    },
    {
      "epoch": 0.6319050262323849,
      "grad_norm": 1.6470979452133179,
      "learning_rate": 6.303337724084779e-05,
      "loss": 1.6618,
      "step": 17645
    },
    {
      "epoch": 0.6319408383619531,
      "grad_norm": 1.3379909992218018,
      "learning_rate": 6.302260009716237e-05,
      "loss": 1.3781,
      "step": 17646
    },
    {
      "epoch": 0.6319766504915215,
      "grad_norm": 1.534791111946106,
      "learning_rate": 6.301182345096017e-05,
      "loss": 1.34,
      "step": 17647
    },
    {
      "epoch": 0.6320124626210898,
      "grad_norm": 2.221349000930786,
      "learning_rate": 6.300104730238616e-05,
      "loss": 1.535,
      "step": 17648
    },
    {
      "epoch": 0.6320482747506581,
      "grad_norm": 1.749962329864502,
      "learning_rate": 6.299027165158536e-05,
      "loss": 1.4936,
      "step": 17649
    },
    {
      "epoch": 0.6320840868802263,
      "grad_norm": 1.2860403060913086,
      "learning_rate": 6.297949649870267e-05,
      "loss": 1.0702,
      "step": 17650
    },
    {
      "epoch": 0.6321198990097946,
      "grad_norm": 1.5968939065933228,
      "learning_rate": 6.296872184388315e-05,
      "loss": 1.4381,
      "step": 17651
    },
    {
      "epoch": 0.6321557111393629,
      "grad_norm": 1.518203854560852,
      "learning_rate": 6.295794768727168e-05,
      "loss": 1.2938,
      "step": 17652
    },
    {
      "epoch": 0.6321915232689311,
      "grad_norm": 1.8395830392837524,
      "learning_rate": 6.294717402901325e-05,
      "loss": 1.5632,
      "step": 17653
    },
    {
      "epoch": 0.6322273353984995,
      "grad_norm": 1.8332093954086304,
      "learning_rate": 6.293640086925279e-05,
      "loss": 1.6127,
      "step": 17654
    },
    {
      "epoch": 0.6322631475280678,
      "grad_norm": 1.5718899965286255,
      "learning_rate": 6.292562820813528e-05,
      "loss": 1.4328,
      "step": 17655
    },
    {
      "epoch": 0.632298959657636,
      "grad_norm": 2.0594699382781982,
      "learning_rate": 6.291485604580559e-05,
      "loss": 1.6376,
      "step": 17656
    },
    {
      "epoch": 0.6323347717872043,
      "grad_norm": 1.631384253501892,
      "learning_rate": 6.290408438240869e-05,
      "loss": 1.513,
      "step": 17657
    },
    {
      "epoch": 0.6323705839167726,
      "grad_norm": 1.8818942308425903,
      "learning_rate": 6.289331321808948e-05,
      "loss": 1.2976,
      "step": 17658
    },
    {
      "epoch": 0.6324063960463409,
      "grad_norm": 2.297025680541992,
      "learning_rate": 6.288254255299286e-05,
      "loss": 1.5297,
      "step": 17659
    },
    {
      "epoch": 0.6324422081759091,
      "grad_norm": 1.3660603761672974,
      "learning_rate": 6.287177238726378e-05,
      "loss": 1.504,
      "step": 17660
    },
    {
      "epoch": 0.6324780203054775,
      "grad_norm": 1.9709035158157349,
      "learning_rate": 6.28610027210471e-05,
      "loss": 1.7812,
      "step": 17661
    },
    {
      "epoch": 0.6325138324350458,
      "grad_norm": 1.61428964138031,
      "learning_rate": 6.285023355448772e-05,
      "loss": 1.4382,
      "step": 17662
    },
    {
      "epoch": 0.632549644564614,
      "grad_norm": 2.234609365463257,
      "learning_rate": 6.283946488773051e-05,
      "loss": 1.5856,
      "step": 17663
    },
    {
      "epoch": 0.6325854566941823,
      "grad_norm": 1.6669365167617798,
      "learning_rate": 6.282869672092039e-05,
      "loss": 1.4216,
      "step": 17664
    },
    {
      "epoch": 0.6326212688237506,
      "grad_norm": 1.5173592567443848,
      "learning_rate": 6.281792905420219e-05,
      "loss": 1.4237,
      "step": 17665
    },
    {
      "epoch": 0.6326570809533189,
      "grad_norm": 1.7789220809936523,
      "learning_rate": 6.280716188772082e-05,
      "loss": 1.406,
      "step": 17666
    },
    {
      "epoch": 0.6326928930828871,
      "grad_norm": 2.917492389678955,
      "learning_rate": 6.279639522162111e-05,
      "loss": 1.6137,
      "step": 17667
    },
    {
      "epoch": 0.6327287052124555,
      "grad_norm": 1.5990511178970337,
      "learning_rate": 6.278562905604788e-05,
      "loss": 1.3547,
      "step": 17668
    },
    {
      "epoch": 0.6327645173420238,
      "grad_norm": 2.128406047821045,
      "learning_rate": 6.277486339114605e-05,
      "loss": 1.417,
      "step": 17669
    },
    {
      "epoch": 0.632800329471592,
      "grad_norm": 2.0638668537139893,
      "learning_rate": 6.276409822706038e-05,
      "loss": 1.6075,
      "step": 17670
    },
    {
      "epoch": 0.6328361416011603,
      "grad_norm": 2.864104747772217,
      "learning_rate": 6.275333356393575e-05,
      "loss": 1.4648,
      "step": 17671
    },
    {
      "epoch": 0.6328719537307286,
      "grad_norm": 1.8883485794067383,
      "learning_rate": 6.274256940191696e-05,
      "loss": 1.5711,
      "step": 17672
    },
    {
      "epoch": 0.6329077658602968,
      "grad_norm": 1.6292932033538818,
      "learning_rate": 6.273180574114887e-05,
      "loss": 1.7285,
      "step": 17673
    },
    {
      "epoch": 0.6329435779898651,
      "grad_norm": 1.8495104312896729,
      "learning_rate": 6.272104258177622e-05,
      "loss": 1.1279,
      "step": 17674
    },
    {
      "epoch": 0.6329793901194335,
      "grad_norm": 1.819200038909912,
      "learning_rate": 6.271027992394389e-05,
      "loss": 1.9726,
      "step": 17675
    },
    {
      "epoch": 0.6330152022490018,
      "grad_norm": 1.7758945226669312,
      "learning_rate": 6.269951776779667e-05,
      "loss": 1.2936,
      "step": 17676
    },
    {
      "epoch": 0.63305101437857,
      "grad_norm": 1.482681155204773,
      "learning_rate": 6.268875611347925e-05,
      "loss": 1.4099,
      "step": 17677
    },
    {
      "epoch": 0.6330868265081383,
      "grad_norm": 1.7096611261367798,
      "learning_rate": 6.267799496113656e-05,
      "loss": 1.5464,
      "step": 17678
    },
    {
      "epoch": 0.6331226386377066,
      "grad_norm": 1.5019376277923584,
      "learning_rate": 6.266723431091323e-05,
      "loss": 1.4345,
      "step": 17679
    },
    {
      "epoch": 0.6331584507672748,
      "grad_norm": 2.354271173477173,
      "learning_rate": 6.265647416295417e-05,
      "loss": 1.5104,
      "step": 17680
    },
    {
      "epoch": 0.6331942628968431,
      "grad_norm": 1.3241621255874634,
      "learning_rate": 6.264571451740405e-05,
      "loss": 1.5164,
      "step": 17681
    },
    {
      "epoch": 0.6332300750264115,
      "grad_norm": 1.2999305725097656,
      "learning_rate": 6.263495537440766e-05,
      "loss": 1.5146,
      "step": 17682
    },
    {
      "epoch": 0.6332658871559798,
      "grad_norm": 1.9902713298797607,
      "learning_rate": 6.262419673410976e-05,
      "loss": 1.5923,
      "step": 17683
    },
    {
      "epoch": 0.633301699285548,
      "grad_norm": 1.6341190338134766,
      "learning_rate": 6.261343859665507e-05,
      "loss": 1.5331,
      "step": 17684
    },
    {
      "epoch": 0.6333375114151163,
      "grad_norm": 1.755398154258728,
      "learning_rate": 6.260268096218838e-05,
      "loss": 1.4803,
      "step": 17685
    },
    {
      "epoch": 0.6333733235446846,
      "grad_norm": 1.864999532699585,
      "learning_rate": 6.259192383085432e-05,
      "loss": 1.4069,
      "step": 17686
    },
    {
      "epoch": 0.6334091356742528,
      "grad_norm": 1.3991942405700684,
      "learning_rate": 6.258116720279773e-05,
      "loss": 1.1271,
      "step": 17687
    },
    {
      "epoch": 0.6334449478038211,
      "grad_norm": 1.9088125228881836,
      "learning_rate": 6.257041107816319e-05,
      "loss": 1.548,
      "step": 17688
    },
    {
      "epoch": 0.6334807599333895,
      "grad_norm": 2.0545427799224854,
      "learning_rate": 6.255965545709556e-05,
      "loss": 1.441,
      "step": 17689
    },
    {
      "epoch": 0.6335165720629578,
      "grad_norm": 1.577840805053711,
      "learning_rate": 6.254890033973942e-05,
      "loss": 1.6864,
      "step": 17690
    },
    {
      "epoch": 0.633552384192526,
      "grad_norm": 1.5293114185333252,
      "learning_rate": 6.253814572623958e-05,
      "loss": 1.6347,
      "step": 17691
    },
    {
      "epoch": 0.6335881963220943,
      "grad_norm": 1.3366191387176514,
      "learning_rate": 6.252739161674059e-05,
      "loss": 1.344,
      "step": 17692
    },
    {
      "epoch": 0.6336240084516626,
      "grad_norm": 1.6025129556655884,
      "learning_rate": 6.251663801138725e-05,
      "loss": 1.5063,
      "step": 17693
    },
    {
      "epoch": 0.6336598205812308,
      "grad_norm": 1.4043288230895996,
      "learning_rate": 6.250588491032421e-05,
      "loss": 1.5331,
      "step": 17694
    },
    {
      "epoch": 0.6336956327107991,
      "grad_norm": 1.5216609239578247,
      "learning_rate": 6.249513231369608e-05,
      "loss": 1.4133,
      "step": 17695
    },
    {
      "epoch": 0.6337314448403675,
      "grad_norm": 1.466712236404419,
      "learning_rate": 6.248438022164763e-05,
      "loss": 1.3336,
      "step": 17696
    },
    {
      "epoch": 0.6337672569699357,
      "grad_norm": 2.017925262451172,
      "learning_rate": 6.247362863432337e-05,
      "loss": 1.5305,
      "step": 17697
    },
    {
      "epoch": 0.633803069099504,
      "grad_norm": 1.440697431564331,
      "learning_rate": 6.246287755186813e-05,
      "loss": 1.379,
      "step": 17698
    },
    {
      "epoch": 0.6338388812290723,
      "grad_norm": 1.7415283918380737,
      "learning_rate": 6.245212697442637e-05,
      "loss": 1.4176,
      "step": 17699
    },
    {
      "epoch": 0.6338746933586406,
      "grad_norm": 1.6304659843444824,
      "learning_rate": 6.244137690214287e-05,
      "loss": 1.2807,
      "step": 17700
    },
    {
      "epoch": 0.6339105054882088,
      "grad_norm": 2.0391323566436768,
      "learning_rate": 6.243062733516211e-05,
      "loss": 1.5144,
      "step": 17701
    },
    {
      "epoch": 0.6339463176177771,
      "grad_norm": 1.8402481079101562,
      "learning_rate": 6.24198782736289e-05,
      "loss": 1.3738,
      "step": 17702
    },
    {
      "epoch": 0.6339821297473455,
      "grad_norm": 1.4574590921401978,
      "learning_rate": 6.240912971768771e-05,
      "loss": 1.1814,
      "step": 17703
    },
    {
      "epoch": 0.6340179418769137,
      "grad_norm": 2.0565900802612305,
      "learning_rate": 6.239838166748318e-05,
      "loss": 1.086,
      "step": 17704
    },
    {
      "epoch": 0.634053754006482,
      "grad_norm": 1.5707656145095825,
      "learning_rate": 6.238763412315993e-05,
      "loss": 1.2242,
      "step": 17705
    },
    {
      "epoch": 0.6340895661360503,
      "grad_norm": 2.836914539337158,
      "learning_rate": 6.237688708486252e-05,
      "loss": 1.5508,
      "step": 17706
    },
    {
      "epoch": 0.6341253782656185,
      "grad_norm": 1.3280764818191528,
      "learning_rate": 6.236614055273562e-05,
      "loss": 1.2026,
      "step": 17707
    },
    {
      "epoch": 0.6341611903951868,
      "grad_norm": 2.0606489181518555,
      "learning_rate": 6.23553945269237e-05,
      "loss": 1.6827,
      "step": 17708
    },
    {
      "epoch": 0.6341970025247551,
      "grad_norm": 1.3689779043197632,
      "learning_rate": 6.234464900757144e-05,
      "loss": 1.4438,
      "step": 17709
    },
    {
      "epoch": 0.6342328146543235,
      "grad_norm": 1.415412425994873,
      "learning_rate": 6.233390399482334e-05,
      "loss": 1.3191,
      "step": 17710
    },
    {
      "epoch": 0.6342686267838917,
      "grad_norm": 1.9321517944335938,
      "learning_rate": 6.232315948882394e-05,
      "loss": 1.485,
      "step": 17711
    },
    {
      "epoch": 0.63430443891346,
      "grad_norm": 1.5414435863494873,
      "learning_rate": 6.231241548971788e-05,
      "loss": 1.3746,
      "step": 17712
    },
    {
      "epoch": 0.6343402510430283,
      "grad_norm": 1.7745535373687744,
      "learning_rate": 6.230167199764962e-05,
      "loss": 1.5908,
      "step": 17713
    },
    {
      "epoch": 0.6343760631725965,
      "grad_norm": 1.9899230003356934,
      "learning_rate": 6.229092901276376e-05,
      "loss": 1.6602,
      "step": 17714
    },
    {
      "epoch": 0.6344118753021648,
      "grad_norm": 2.1202757358551025,
      "learning_rate": 6.228018653520477e-05,
      "loss": 1.376,
      "step": 17715
    },
    {
      "epoch": 0.6344476874317331,
      "grad_norm": 1.5492504835128784,
      "learning_rate": 6.226944456511725e-05,
      "loss": 1.4702,
      "step": 17716
    },
    {
      "epoch": 0.6344834995613015,
      "grad_norm": 1.687817931175232,
      "learning_rate": 6.225870310264567e-05,
      "loss": 1.3696,
      "step": 17717
    },
    {
      "epoch": 0.6345193116908697,
      "grad_norm": 2.061065435409546,
      "learning_rate": 6.224796214793458e-05,
      "loss": 1.7091,
      "step": 17718
    },
    {
      "epoch": 0.634555123820438,
      "grad_norm": 1.6443705558776855,
      "learning_rate": 6.223722170112845e-05,
      "loss": 1.4229,
      "step": 17719
    },
    {
      "epoch": 0.6345909359500063,
      "grad_norm": 1.9211294651031494,
      "learning_rate": 6.222648176237179e-05,
      "loss": 1.6184,
      "step": 17720
    },
    {
      "epoch": 0.6346267480795745,
      "grad_norm": 1.7323452234268188,
      "learning_rate": 6.221574233180907e-05,
      "loss": 1.5455,
      "step": 17721
    },
    {
      "epoch": 0.6346625602091428,
      "grad_norm": 1.5966200828552246,
      "learning_rate": 6.220500340958482e-05,
      "loss": 1.3132,
      "step": 17722
    },
    {
      "epoch": 0.6346983723387111,
      "grad_norm": 1.7405643463134766,
      "learning_rate": 6.219426499584351e-05,
      "loss": 1.5249,
      "step": 17723
    },
    {
      "epoch": 0.6347341844682794,
      "grad_norm": 1.7434638738632202,
      "learning_rate": 6.218352709072957e-05,
      "loss": 1.6002,
      "step": 17724
    },
    {
      "epoch": 0.6347699965978477,
      "grad_norm": 1.4588823318481445,
      "learning_rate": 6.21727896943875e-05,
      "loss": 1.2939,
      "step": 17725
    },
    {
      "epoch": 0.634805808727416,
      "grad_norm": 1.711930751800537,
      "learning_rate": 6.216205280696177e-05,
      "loss": 1.2147,
      "step": 17726
    },
    {
      "epoch": 0.6348416208569843,
      "grad_norm": 1.7398806810379028,
      "learning_rate": 6.21513164285968e-05,
      "loss": 1.5812,
      "step": 17727
    },
    {
      "epoch": 0.6348774329865525,
      "grad_norm": 1.6372019052505493,
      "learning_rate": 6.214058055943706e-05,
      "loss": 1.6223,
      "step": 17728
    },
    {
      "epoch": 0.6349132451161208,
      "grad_norm": 1.3632045984268188,
      "learning_rate": 6.212984519962695e-05,
      "loss": 1.4353,
      "step": 17729
    },
    {
      "epoch": 0.6349490572456891,
      "grad_norm": 1.4072738885879517,
      "learning_rate": 6.211911034931094e-05,
      "loss": 1.4157,
      "step": 17730
    },
    {
      "epoch": 0.6349848693752574,
      "grad_norm": 1.8789604902267456,
      "learning_rate": 6.210837600863342e-05,
      "loss": 1.3836,
      "step": 17731
    },
    {
      "epoch": 0.6350206815048257,
      "grad_norm": 1.5280240774154663,
      "learning_rate": 6.209764217773884e-05,
      "loss": 1.4962,
      "step": 17732
    },
    {
      "epoch": 0.635056493634394,
      "grad_norm": 1.6792149543762207,
      "learning_rate": 6.208690885677158e-05,
      "loss": 1.3537,
      "step": 17733
    },
    {
      "epoch": 0.6350923057639623,
      "grad_norm": 2.1413023471832275,
      "learning_rate": 6.207617604587607e-05,
      "loss": 1.4218,
      "step": 17734
    },
    {
      "epoch": 0.6351281178935305,
      "grad_norm": 1.7607442140579224,
      "learning_rate": 6.20654437451967e-05,
      "loss": 1.6476,
      "step": 17735
    },
    {
      "epoch": 0.6351639300230988,
      "grad_norm": 1.53827965259552,
      "learning_rate": 6.205471195487784e-05,
      "loss": 1.5853,
      "step": 17736
    },
    {
      "epoch": 0.6351997421526671,
      "grad_norm": 1.5775997638702393,
      "learning_rate": 6.204398067506389e-05,
      "loss": 1.5806,
      "step": 17737
    },
    {
      "epoch": 0.6352355542822354,
      "grad_norm": 1.4330687522888184,
      "learning_rate": 6.203324990589922e-05,
      "loss": 1.4459,
      "step": 17738
    },
    {
      "epoch": 0.6352713664118037,
      "grad_norm": 1.8977481126785278,
      "learning_rate": 6.20225196475282e-05,
      "loss": 1.4269,
      "step": 17739
    },
    {
      "epoch": 0.635307178541372,
      "grad_norm": 1.7836785316467285,
      "learning_rate": 6.201178990009518e-05,
      "loss": 1.547,
      "step": 17740
    },
    {
      "epoch": 0.6353429906709402,
      "grad_norm": 1.7709898948669434,
      "learning_rate": 6.200106066374454e-05,
      "loss": 1.4268,
      "step": 17741
    },
    {
      "epoch": 0.6353788028005085,
      "grad_norm": 2.1804091930389404,
      "learning_rate": 6.199033193862059e-05,
      "loss": 1.4339,
      "step": 17742
    },
    {
      "epoch": 0.6354146149300768,
      "grad_norm": 1.4796829223632812,
      "learning_rate": 6.197960372486772e-05,
      "loss": 1.3443,
      "step": 17743
    },
    {
      "epoch": 0.635450427059645,
      "grad_norm": 1.9017022848129272,
      "learning_rate": 6.196887602263022e-05,
      "loss": 1.4565,
      "step": 17744
    },
    {
      "epoch": 0.6354862391892134,
      "grad_norm": 1.6325371265411377,
      "learning_rate": 6.195814883205245e-05,
      "loss": 1.6163,
      "step": 17745
    },
    {
      "epoch": 0.6355220513187817,
      "grad_norm": 2.607729196548462,
      "learning_rate": 6.194742215327873e-05,
      "loss": 1.211,
      "step": 17746
    },
    {
      "epoch": 0.63555786344835,
      "grad_norm": 1.6463838815689087,
      "learning_rate": 6.193669598645334e-05,
      "loss": 1.5084,
      "step": 17747
    },
    {
      "epoch": 0.6355936755779182,
      "grad_norm": 1.6740944385528564,
      "learning_rate": 6.19259703317206e-05,
      "loss": 1.3549,
      "step": 17748
    },
    {
      "epoch": 0.6356294877074865,
      "grad_norm": 2.299992561340332,
      "learning_rate": 6.191524518922482e-05,
      "loss": 1.4458,
      "step": 17749
    },
    {
      "epoch": 0.6356652998370548,
      "grad_norm": 1.5784881114959717,
      "learning_rate": 6.190452055911031e-05,
      "loss": 1.5335,
      "step": 17750
    },
    {
      "epoch": 0.635701111966623,
      "grad_norm": 1.283570647239685,
      "learning_rate": 6.189379644152132e-05,
      "loss": 1.3831,
      "step": 17751
    },
    {
      "epoch": 0.6357369240961914,
      "grad_norm": 1.497164249420166,
      "learning_rate": 6.188307283660216e-05,
      "loss": 1.491,
      "step": 17752
    },
    {
      "epoch": 0.6357727362257597,
      "grad_norm": 1.6668617725372314,
      "learning_rate": 6.187234974449707e-05,
      "loss": 1.4718,
      "step": 17753
    },
    {
      "epoch": 0.635808548355328,
      "grad_norm": 1.4830564260482788,
      "learning_rate": 6.186162716535036e-05,
      "loss": 1.3756,
      "step": 17754
    },
    {
      "epoch": 0.6358443604848962,
      "grad_norm": 1.7476052045822144,
      "learning_rate": 6.185090509930624e-05,
      "loss": 1.497,
      "step": 17755
    },
    {
      "epoch": 0.6358801726144645,
      "grad_norm": 1.7061628103256226,
      "learning_rate": 6.184018354650898e-05,
      "loss": 1.3756,
      "step": 17756
    },
    {
      "epoch": 0.6359159847440328,
      "grad_norm": 1.9672883749008179,
      "learning_rate": 6.182946250710284e-05,
      "loss": 1.6373,
      "step": 17757
    },
    {
      "epoch": 0.635951796873601,
      "grad_norm": 1.8433897495269775,
      "learning_rate": 6.181874198123203e-05,
      "loss": 1.8807,
      "step": 17758
    },
    {
      "epoch": 0.6359876090031694,
      "grad_norm": 1.7661159038543701,
      "learning_rate": 6.18080219690408e-05,
      "loss": 1.4735,
      "step": 17759
    },
    {
      "epoch": 0.6360234211327377,
      "grad_norm": 2.0995090007781982,
      "learning_rate": 6.179730247067336e-05,
      "loss": 1.3749,
      "step": 17760
    },
    {
      "epoch": 0.636059233262306,
      "grad_norm": 1.3635008335113525,
      "learning_rate": 6.178658348627398e-05,
      "loss": 1.3731,
      "step": 17761
    },
    {
      "epoch": 0.6360950453918742,
      "grad_norm": 2.132258892059326,
      "learning_rate": 6.177586501598679e-05,
      "loss": 1.3157,
      "step": 17762
    },
    {
      "epoch": 0.6361308575214425,
      "grad_norm": 1.4567443132400513,
      "learning_rate": 6.176514705995604e-05,
      "loss": 1.7088,
      "step": 17763
    },
    {
      "epoch": 0.6361666696510108,
      "grad_norm": 1.368294358253479,
      "learning_rate": 6.175442961832593e-05,
      "loss": 1.4436,
      "step": 17764
    },
    {
      "epoch": 0.636202481780579,
      "grad_norm": 1.7212412357330322,
      "learning_rate": 6.174371269124061e-05,
      "loss": 1.5667,
      "step": 17765
    },
    {
      "epoch": 0.6362382939101474,
      "grad_norm": 1.3152505159378052,
      "learning_rate": 6.173299627884432e-05,
      "loss": 1.3987,
      "step": 17766
    },
    {
      "epoch": 0.6362741060397157,
      "grad_norm": 1.347125768661499,
      "learning_rate": 6.172228038128118e-05,
      "loss": 1.3405,
      "step": 17767
    },
    {
      "epoch": 0.636309918169284,
      "grad_norm": 1.3054901361465454,
      "learning_rate": 6.171156499869539e-05,
      "loss": 1.4982,
      "step": 17768
    },
    {
      "epoch": 0.6363457302988522,
      "grad_norm": 1.41067636013031,
      "learning_rate": 6.17008501312311e-05,
      "loss": 1.1198,
      "step": 17769
    },
    {
      "epoch": 0.6363815424284205,
      "grad_norm": 1.8490687608718872,
      "learning_rate": 6.169013577903248e-05,
      "loss": 1.6673,
      "step": 17770
    },
    {
      "epoch": 0.6364173545579888,
      "grad_norm": 2.312915325164795,
      "learning_rate": 6.167942194224365e-05,
      "loss": 1.4044,
      "step": 17771
    },
    {
      "epoch": 0.636453166687557,
      "grad_norm": 1.7895601987838745,
      "learning_rate": 6.16687086210088e-05,
      "loss": 1.634,
      "step": 17772
    },
    {
      "epoch": 0.6364889788171254,
      "grad_norm": 1.7349082231521606,
      "learning_rate": 6.165799581547203e-05,
      "loss": 1.4799,
      "step": 17773
    },
    {
      "epoch": 0.6365247909466937,
      "grad_norm": 1.9803053140640259,
      "learning_rate": 6.164728352577743e-05,
      "loss": 1.4382,
      "step": 17774
    },
    {
      "epoch": 0.6365606030762619,
      "grad_norm": 2.0456080436706543,
      "learning_rate": 6.16365717520692e-05,
      "loss": 1.2846,
      "step": 17775
    },
    {
      "epoch": 0.6365964152058302,
      "grad_norm": 2.208461046218872,
      "learning_rate": 6.162586049449136e-05,
      "loss": 1.3596,
      "step": 17776
    },
    {
      "epoch": 0.6366322273353985,
      "grad_norm": 1.5197937488555908,
      "learning_rate": 6.161514975318809e-05,
      "loss": 1.5683,
      "step": 17777
    },
    {
      "epoch": 0.6366680394649668,
      "grad_norm": 1.7372349500656128,
      "learning_rate": 6.160443952830347e-05,
      "loss": 1.4085,
      "step": 17778
    },
    {
      "epoch": 0.636703851594535,
      "grad_norm": 1.7026249170303345,
      "learning_rate": 6.159372981998161e-05,
      "loss": 1.5112,
      "step": 17779
    },
    {
      "epoch": 0.6367396637241034,
      "grad_norm": 1.795798897743225,
      "learning_rate": 6.158302062836654e-05,
      "loss": 1.5585,
      "step": 17780
    },
    {
      "epoch": 0.6367754758536717,
      "grad_norm": 1.8992228507995605,
      "learning_rate": 6.157231195360241e-05,
      "loss": 1.6058,
      "step": 17781
    },
    {
      "epoch": 0.6368112879832399,
      "grad_norm": 1.8138514757156372,
      "learning_rate": 6.156160379583325e-05,
      "loss": 1.3325,
      "step": 17782
    },
    {
      "epoch": 0.6368471001128082,
      "grad_norm": 1.941710114479065,
      "learning_rate": 6.155089615520308e-05,
      "loss": 1.449,
      "step": 17783
    },
    {
      "epoch": 0.6368829122423765,
      "grad_norm": 1.922827124595642,
      "learning_rate": 6.154018903185608e-05,
      "loss": 1.1852,
      "step": 17784
    },
    {
      "epoch": 0.6369187243719447,
      "grad_norm": 1.9365390539169312,
      "learning_rate": 6.152948242593615e-05,
      "loss": 1.6456,
      "step": 17785
    },
    {
      "epoch": 0.636954536501513,
      "grad_norm": 1.8548115491867065,
      "learning_rate": 6.15187763375875e-05,
      "loss": 1.6585,
      "step": 17786
    },
    {
      "epoch": 0.6369903486310814,
      "grad_norm": 1.6139572858810425,
      "learning_rate": 6.150807076695399e-05,
      "loss": 1.3493,
      "step": 17787
    },
    {
      "epoch": 0.6370261607606497,
      "grad_norm": 1.368597149848938,
      "learning_rate": 6.149736571417979e-05,
      "loss": 1.3896,
      "step": 17788
    },
    {
      "epoch": 0.6370619728902179,
      "grad_norm": 1.8540292978286743,
      "learning_rate": 6.148666117940882e-05,
      "loss": 1.3981,
      "step": 17789
    },
    {
      "epoch": 0.6370977850197862,
      "grad_norm": 1.3378417491912842,
      "learning_rate": 6.147595716278519e-05,
      "loss": 1.302,
      "step": 17790
    },
    {
      "epoch": 0.6371335971493545,
      "grad_norm": 1.6063618659973145,
      "learning_rate": 6.146525366445288e-05,
      "loss": 1.5243,
      "step": 17791
    },
    {
      "epoch": 0.6371694092789227,
      "grad_norm": 2.179349660873413,
      "learning_rate": 6.145455068455583e-05,
      "loss": 1.31,
      "step": 17792
    },
    {
      "epoch": 0.637205221408491,
      "grad_norm": 1.9033253192901611,
      "learning_rate": 6.144384822323812e-05,
      "loss": 1.2234,
      "step": 17793
    },
    {
      "epoch": 0.6372410335380594,
      "grad_norm": 1.3805716037750244,
      "learning_rate": 6.143314628064365e-05,
      "loss": 1.5376,
      "step": 17794
    },
    {
      "epoch": 0.6372768456676277,
      "grad_norm": 1.3206768035888672,
      "learning_rate": 6.14224448569165e-05,
      "loss": 1.4238,
      "step": 17795
    },
    {
      "epoch": 0.6373126577971959,
      "grad_norm": 1.6055296659469604,
      "learning_rate": 6.141174395220053e-05,
      "loss": 1.451,
      "step": 17796
    },
    {
      "epoch": 0.6373484699267642,
      "grad_norm": 1.5268259048461914,
      "learning_rate": 6.140104356663984e-05,
      "loss": 1.3718,
      "step": 17797
    },
    {
      "epoch": 0.6373842820563325,
      "grad_norm": 1.8566981554031372,
      "learning_rate": 6.139034370037826e-05,
      "loss": 1.4693,
      "step": 17798
    },
    {
      "epoch": 0.6374200941859007,
      "grad_norm": 1.3930617570877075,
      "learning_rate": 6.137964435355984e-05,
      "loss": 1.7598,
      "step": 17799
    },
    {
      "epoch": 0.637455906315469,
      "grad_norm": 2.081630229949951,
      "learning_rate": 6.13689455263285e-05,
      "loss": 1.3816,
      "step": 17800
    },
    {
      "epoch": 0.6374917184450374,
      "grad_norm": 1.8918607234954834,
      "learning_rate": 6.135824721882815e-05,
      "loss": 1.487,
      "step": 17801
    },
    {
      "epoch": 0.6375275305746056,
      "grad_norm": 1.4612147808074951,
      "learning_rate": 6.134754943120273e-05,
      "loss": 1.7173,
      "step": 17802
    },
    {
      "epoch": 0.6375633427041739,
      "grad_norm": 1.6461340188980103,
      "learning_rate": 6.133685216359615e-05,
      "loss": 1.3156,
      "step": 17803
    },
    {
      "epoch": 0.6375991548337422,
      "grad_norm": 2.132502317428589,
      "learning_rate": 6.132615541615242e-05,
      "loss": 1.617,
      "step": 17804
    },
    {
      "epoch": 0.6376349669633105,
      "grad_norm": 1.3095436096191406,
      "learning_rate": 6.131545918901531e-05,
      "loss": 1.4064,
      "step": 17805
    },
    {
      "epoch": 0.6376707790928787,
      "grad_norm": 1.5461561679840088,
      "learning_rate": 6.130476348232887e-05,
      "loss": 1.291,
      "step": 17806
    },
    {
      "epoch": 0.637706591222447,
      "grad_norm": 1.4045078754425049,
      "learning_rate": 6.12940682962369e-05,
      "loss": 1.2808,
      "step": 17807
    },
    {
      "epoch": 0.6377424033520154,
      "grad_norm": 1.6871155500411987,
      "learning_rate": 6.128337363088327e-05,
      "loss": 1.5045,
      "step": 17808
    },
    {
      "epoch": 0.6377782154815836,
      "grad_norm": 1.3531205654144287,
      "learning_rate": 6.127267948641195e-05,
      "loss": 1.3646,
      "step": 17809
    },
    {
      "epoch": 0.6378140276111519,
      "grad_norm": 1.4474282264709473,
      "learning_rate": 6.126198586296676e-05,
      "loss": 1.2597,
      "step": 17810
    },
    {
      "epoch": 0.6378498397407202,
      "grad_norm": 1.5817090272903442,
      "learning_rate": 6.12512927606916e-05,
      "loss": 1.4625,
      "step": 17811
    },
    {
      "epoch": 0.6378856518702885,
      "grad_norm": 1.6557588577270508,
      "learning_rate": 6.124060017973027e-05,
      "loss": 1.8598,
      "step": 17812
    },
    {
      "epoch": 0.6379214639998567,
      "grad_norm": 1.735060453414917,
      "learning_rate": 6.122990812022671e-05,
      "loss": 1.4263,
      "step": 17813
    },
    {
      "epoch": 0.637957276129425,
      "grad_norm": 1.6962659358978271,
      "learning_rate": 6.12192165823247e-05,
      "loss": 1.4953,
      "step": 17814
    },
    {
      "epoch": 0.6379930882589934,
      "grad_norm": 2.5950944423675537,
      "learning_rate": 6.120852556616811e-05,
      "loss": 1.6169,
      "step": 17815
    },
    {
      "epoch": 0.6380289003885616,
      "grad_norm": 1.4274699687957764,
      "learning_rate": 6.11978350719008e-05,
      "loss": 1.2505,
      "step": 17816
    },
    {
      "epoch": 0.6380647125181299,
      "grad_norm": 2.5493216514587402,
      "learning_rate": 6.118714509966654e-05,
      "loss": 1.3164,
      "step": 17817
    },
    {
      "epoch": 0.6381005246476982,
      "grad_norm": 2.081235647201538,
      "learning_rate": 6.117645564960919e-05,
      "loss": 1.1562,
      "step": 17818
    },
    {
      "epoch": 0.6381363367772664,
      "grad_norm": 1.0999113321304321,
      "learning_rate": 6.116576672187254e-05,
      "loss": 1.5064,
      "step": 17819
    },
    {
      "epoch": 0.6381721489068347,
      "grad_norm": 1.29554283618927,
      "learning_rate": 6.115507831660042e-05,
      "loss": 1.4769,
      "step": 17820
    },
    {
      "epoch": 0.638207961036403,
      "grad_norm": 1.3888424634933472,
      "learning_rate": 6.11443904339366e-05,
      "loss": 1.8104,
      "step": 17821
    },
    {
      "epoch": 0.6382437731659714,
      "grad_norm": 1.9146857261657715,
      "learning_rate": 6.11337030740249e-05,
      "loss": 1.4991,
      "step": 17822
    },
    {
      "epoch": 0.6382795852955396,
      "grad_norm": 1.6722118854522705,
      "learning_rate": 6.112301623700907e-05,
      "loss": 1.4583,
      "step": 17823
    },
    {
      "epoch": 0.6383153974251079,
      "grad_norm": 1.50333833694458,
      "learning_rate": 6.111232992303292e-05,
      "loss": 1.4124,
      "step": 17824
    },
    {
      "epoch": 0.6383512095546762,
      "grad_norm": 1.567192792892456,
      "learning_rate": 6.110164413224025e-05,
      "loss": 1.3424,
      "step": 17825
    },
    {
      "epoch": 0.6383870216842444,
      "grad_norm": 2.029167652130127,
      "learning_rate": 6.109095886477472e-05,
      "loss": 1.6334,
      "step": 17826
    },
    {
      "epoch": 0.6384228338138127,
      "grad_norm": 1.5533322095870972,
      "learning_rate": 6.108027412078018e-05,
      "loss": 1.5,
      "step": 17827
    },
    {
      "epoch": 0.638458645943381,
      "grad_norm": 1.8610965013504028,
      "learning_rate": 6.106958990040033e-05,
      "loss": 1.6345,
      "step": 17828
    },
    {
      "epoch": 0.6384944580729494,
      "grad_norm": 1.191743016242981,
      "learning_rate": 6.105890620377897e-05,
      "loss": 1.3893,
      "step": 17829
    },
    {
      "epoch": 0.6385302702025176,
      "grad_norm": 1.9871865510940552,
      "learning_rate": 6.104822303105974e-05,
      "loss": 1.4636,
      "step": 17830
    },
    {
      "epoch": 0.6385660823320859,
      "grad_norm": 1.702633023262024,
      "learning_rate": 6.103754038238648e-05,
      "loss": 1.4101,
      "step": 17831
    },
    {
      "epoch": 0.6386018944616542,
      "grad_norm": 2.4105355739593506,
      "learning_rate": 6.102685825790282e-05,
      "loss": 1.3532,
      "step": 17832
    },
    {
      "epoch": 0.6386377065912224,
      "grad_norm": 1.8922439813613892,
      "learning_rate": 6.1016176657752534e-05,
      "loss": 1.4376,
      "step": 17833
    },
    {
      "epoch": 0.6386735187207907,
      "grad_norm": 2.383431911468506,
      "learning_rate": 6.100549558207931e-05,
      "loss": 1.3371,
      "step": 17834
    },
    {
      "epoch": 0.638709330850359,
      "grad_norm": 2.351727247238159,
      "learning_rate": 6.099481503102682e-05,
      "loss": 1.5323,
      "step": 17835
    },
    {
      "epoch": 0.6387451429799273,
      "grad_norm": 2.596125364303589,
      "learning_rate": 6.0984135004738784e-05,
      "loss": 1.7484,
      "step": 17836
    },
    {
      "epoch": 0.6387809551094956,
      "grad_norm": 1.5746511220932007,
      "learning_rate": 6.097345550335889e-05,
      "loss": 1.2177,
      "step": 17837
    },
    {
      "epoch": 0.6388167672390639,
      "grad_norm": 1.227081298828125,
      "learning_rate": 6.096277652703082e-05,
      "loss": 1.6711,
      "step": 17838
    },
    {
      "epoch": 0.6388525793686322,
      "grad_norm": 2.1082305908203125,
      "learning_rate": 6.0952098075898214e-05,
      "loss": 1.598,
      "step": 17839
    },
    {
      "epoch": 0.6388883914982004,
      "grad_norm": 1.3371353149414062,
      "learning_rate": 6.0941420150104776e-05,
      "loss": 1.3421,
      "step": 17840
    },
    {
      "epoch": 0.6389242036277687,
      "grad_norm": 1.6144399642944336,
      "learning_rate": 6.0930742749794145e-05,
      "loss": 1.352,
      "step": 17841
    },
    {
      "epoch": 0.638960015757337,
      "grad_norm": 1.4856688976287842,
      "learning_rate": 6.0920065875109986e-05,
      "loss": 1.5137,
      "step": 17842
    },
    {
      "epoch": 0.6389958278869053,
      "grad_norm": 2.919574737548828,
      "learning_rate": 6.0909389526195935e-05,
      "loss": 1.4141,
      "step": 17843
    },
    {
      "epoch": 0.6390316400164736,
      "grad_norm": 1.6059921979904175,
      "learning_rate": 6.0898713703195595e-05,
      "loss": 1.6238,
      "step": 17844
    },
    {
      "epoch": 0.6390674521460419,
      "grad_norm": 2.180042028427124,
      "learning_rate": 6.0888038406252656e-05,
      "loss": 1.2913,
      "step": 17845
    },
    {
      "epoch": 0.6391032642756102,
      "grad_norm": 1.7399688959121704,
      "learning_rate": 6.087736363551069e-05,
      "loss": 1.5325,
      "step": 17846
    },
    {
      "epoch": 0.6391390764051784,
      "grad_norm": 2.026592254638672,
      "learning_rate": 6.086668939111333e-05,
      "loss": 1.5609,
      "step": 17847
    },
    {
      "epoch": 0.6391748885347467,
      "grad_norm": 1.4536274671554565,
      "learning_rate": 6.085601567320418e-05,
      "loss": 1.2754,
      "step": 17848
    },
    {
      "epoch": 0.639210700664315,
      "grad_norm": 1.4030091762542725,
      "learning_rate": 6.084534248192688e-05,
      "loss": 1.3692,
      "step": 17849
    },
    {
      "epoch": 0.6392465127938833,
      "grad_norm": 1.6104986667633057,
      "learning_rate": 6.083466981742496e-05,
      "loss": 1.6165,
      "step": 17850
    },
    {
      "epoch": 0.6392823249234516,
      "grad_norm": 1.5701024532318115,
      "learning_rate": 6.082399767984206e-05,
      "loss": 1.5008,
      "step": 17851
    },
    {
      "epoch": 0.6393181370530199,
      "grad_norm": 2.5654051303863525,
      "learning_rate": 6.081332606932173e-05,
      "loss": 1.4486,
      "step": 17852
    },
    {
      "epoch": 0.6393539491825881,
      "grad_norm": 1.540667176246643,
      "learning_rate": 6.0802654986007534e-05,
      "loss": 1.5323,
      "step": 17853
    },
    {
      "epoch": 0.6393897613121564,
      "grad_norm": 1.600110411643982,
      "learning_rate": 6.079198443004308e-05,
      "loss": 1.4927,
      "step": 17854
    },
    {
      "epoch": 0.6394255734417247,
      "grad_norm": 1.3567900657653809,
      "learning_rate": 6.0781314401571875e-05,
      "loss": 1.5003,
      "step": 17855
    },
    {
      "epoch": 0.639461385571293,
      "grad_norm": 3.0247714519500732,
      "learning_rate": 6.077064490073752e-05,
      "loss": 1.5352,
      "step": 17856
    },
    {
      "epoch": 0.6394971977008613,
      "grad_norm": 2.202338695526123,
      "learning_rate": 6.075997592768352e-05,
      "loss": 1.8075,
      "step": 17857
    },
    {
      "epoch": 0.6395330098304296,
      "grad_norm": 1.6735132932662964,
      "learning_rate": 6.074930748255343e-05,
      "loss": 1.3196,
      "step": 17858
    },
    {
      "epoch": 0.6395688219599979,
      "grad_norm": 1.94517982006073,
      "learning_rate": 6.073863956549077e-05,
      "loss": 1.506,
      "step": 17859
    },
    {
      "epoch": 0.6396046340895661,
      "grad_norm": 2.440854787826538,
      "learning_rate": 6.07279721766391e-05,
      "loss": 1.6454,
      "step": 17860
    },
    {
      "epoch": 0.6396404462191344,
      "grad_norm": 1.7555967569351196,
      "learning_rate": 6.071730531614189e-05,
      "loss": 1.4305,
      "step": 17861
    },
    {
      "epoch": 0.6396762583487027,
      "grad_norm": 1.5704050064086914,
      "learning_rate": 6.070663898414266e-05,
      "loss": 1.373,
      "step": 17862
    },
    {
      "epoch": 0.6397120704782709,
      "grad_norm": 2.053732395172119,
      "learning_rate": 6.069597318078493e-05,
      "loss": 1.4501,
      "step": 17863
    },
    {
      "epoch": 0.6397478826078393,
      "grad_norm": 1.6441607475280762,
      "learning_rate": 6.0685307906212163e-05,
      "loss": 1.1916,
      "step": 17864
    },
    {
      "epoch": 0.6397836947374076,
      "grad_norm": 1.6410545110702515,
      "learning_rate": 6.067464316056789e-05,
      "loss": 1.2819,
      "step": 17865
    },
    {
      "epoch": 0.6398195068669759,
      "grad_norm": 1.6242485046386719,
      "learning_rate": 6.066397894399553e-05,
      "loss": 1.5297,
      "step": 17866
    },
    {
      "epoch": 0.6398553189965441,
      "grad_norm": 1.47056245803833,
      "learning_rate": 6.065331525663864e-05,
      "loss": 1.6764,
      "step": 17867
    },
    {
      "epoch": 0.6398911311261124,
      "grad_norm": 1.3741710186004639,
      "learning_rate": 6.064265209864061e-05,
      "loss": 1.3427,
      "step": 17868
    },
    {
      "epoch": 0.6399269432556807,
      "grad_norm": 1.7579123973846436,
      "learning_rate": 6.063198947014495e-05,
      "loss": 1.9434,
      "step": 17869
    },
    {
      "epoch": 0.6399627553852489,
      "grad_norm": 1.678293228149414,
      "learning_rate": 6.06213273712951e-05,
      "loss": 1.2601,
      "step": 17870
    },
    {
      "epoch": 0.6399985675148173,
      "grad_norm": 1.5967705249786377,
      "learning_rate": 6.061066580223445e-05,
      "loss": 1.5347,
      "step": 17871
    },
    {
      "epoch": 0.6400343796443856,
      "grad_norm": 1.7844775915145874,
      "learning_rate": 6.0600004763106524e-05,
      "loss": 1.3676,
      "step": 17872
    },
    {
      "epoch": 0.6400701917739539,
      "grad_norm": 1.7952393293380737,
      "learning_rate": 6.058934425405467e-05,
      "loss": 1.6115,
      "step": 17873
    },
    {
      "epoch": 0.6401060039035221,
      "grad_norm": 1.9247483015060425,
      "learning_rate": 6.0578684275222376e-05,
      "loss": 1.2821,
      "step": 17874
    },
    {
      "epoch": 0.6401418160330904,
      "grad_norm": 1.4412040710449219,
      "learning_rate": 6.056802482675303e-05,
      "loss": 1.4646,
      "step": 17875
    },
    {
      "epoch": 0.6401776281626587,
      "grad_norm": 1.828956127166748,
      "learning_rate": 6.055736590879007e-05,
      "loss": 1.856,
      "step": 17876
    },
    {
      "epoch": 0.6402134402922269,
      "grad_norm": 1.5418649911880493,
      "learning_rate": 6.0546707521476844e-05,
      "loss": 1.4854,
      "step": 17877
    },
    {
      "epoch": 0.6402492524217953,
      "grad_norm": 1.4142812490463257,
      "learning_rate": 6.0536049664956797e-05,
      "loss": 1.3271,
      "step": 17878
    },
    {
      "epoch": 0.6402850645513636,
      "grad_norm": 2.0951547622680664,
      "learning_rate": 6.052539233937331e-05,
      "loss": 1.2959,
      "step": 17879
    },
    {
      "epoch": 0.6403208766809318,
      "grad_norm": 1.931695580482483,
      "learning_rate": 6.0514735544869706e-05,
      "loss": 1.6754,
      "step": 17880
    },
    {
      "epoch": 0.6403566888105001,
      "grad_norm": 2.0580785274505615,
      "learning_rate": 6.0504079281589454e-05,
      "loss": 1.4303,
      "step": 17881
    },
    {
      "epoch": 0.6403925009400684,
      "grad_norm": 1.1910802125930786,
      "learning_rate": 6.049342354967581e-05,
      "loss": 1.6201,
      "step": 17882
    },
    {
      "epoch": 0.6404283130696367,
      "grad_norm": 1.6602959632873535,
      "learning_rate": 6.0482768349272256e-05,
      "loss": 1.3145,
      "step": 17883
    },
    {
      "epoch": 0.6404641251992049,
      "grad_norm": 1.779673457145691,
      "learning_rate": 6.047211368052201e-05,
      "loss": 1.4677,
      "step": 17884
    },
    {
      "epoch": 0.6404999373287733,
      "grad_norm": 1.6596450805664062,
      "learning_rate": 6.0461459543568566e-05,
      "loss": 1.5077,
      "step": 17885
    },
    {
      "epoch": 0.6405357494583416,
      "grad_norm": 3.11470103263855,
      "learning_rate": 6.04508059385551e-05,
      "loss": 1.6707,
      "step": 17886
    },
    {
      "epoch": 0.6405715615879098,
      "grad_norm": 1.251984715461731,
      "learning_rate": 6.0440152865625076e-05,
      "loss": 1.5564,
      "step": 17887
    },
    {
      "epoch": 0.6406073737174781,
      "grad_norm": 2.3035483360290527,
      "learning_rate": 6.042950032492179e-05,
      "loss": 1.3278,
      "step": 17888
    },
    {
      "epoch": 0.6406431858470464,
      "grad_norm": 1.6876386404037476,
      "learning_rate": 6.041884831658848e-05,
      "loss": 1.4541,
      "step": 17889
    },
    {
      "epoch": 0.6406789979766147,
      "grad_norm": 1.6969448328018188,
      "learning_rate": 6.040819684076856e-05,
      "loss": 1.7653,
      "step": 17890
    },
    {
      "epoch": 0.6407148101061829,
      "grad_norm": 1.8200666904449463,
      "learning_rate": 6.039754589760522e-05,
      "loss": 1.4405,
      "step": 17891
    },
    {
      "epoch": 0.6407506222357513,
      "grad_norm": 1.7832893133163452,
      "learning_rate": 6.038689548724189e-05,
      "loss": 1.5234,
      "step": 17892
    },
    {
      "epoch": 0.6407864343653196,
      "grad_norm": 1.9729869365692139,
      "learning_rate": 6.037624560982171e-05,
      "loss": 1.4516,
      "step": 17893
    },
    {
      "epoch": 0.6408222464948878,
      "grad_norm": 1.648978590965271,
      "learning_rate": 6.03655962654881e-05,
      "loss": 1.7351,
      "step": 17894
    },
    {
      "epoch": 0.6408580586244561,
      "grad_norm": 1.5560195446014404,
      "learning_rate": 6.035494745438421e-05,
      "loss": 1.4883,
      "step": 17895
    },
    {
      "epoch": 0.6408938707540244,
      "grad_norm": 1.6403807401657104,
      "learning_rate": 6.034429917665342e-05,
      "loss": 1.7517,
      "step": 17896
    },
    {
      "epoch": 0.6409296828835926,
      "grad_norm": 2.9381628036499023,
      "learning_rate": 6.033365143243891e-05,
      "loss": 1.672,
      "step": 17897
    },
    {
      "epoch": 0.6409654950131609,
      "grad_norm": 1.4989262819290161,
      "learning_rate": 6.0323004221883936e-05,
      "loss": 1.1939,
      "step": 17898
    },
    {
      "epoch": 0.6410013071427293,
      "grad_norm": 1.9374892711639404,
      "learning_rate": 6.031235754513178e-05,
      "loss": 1.4359,
      "step": 17899
    },
    {
      "epoch": 0.6410371192722976,
      "grad_norm": 1.6095060110092163,
      "learning_rate": 6.030171140232562e-05,
      "loss": 1.5939,
      "step": 17900
    },
    {
      "epoch": 0.6410729314018658,
      "grad_norm": 1.5997309684753418,
      "learning_rate": 6.029106579360879e-05,
      "loss": 1.1778,
      "step": 17901
    },
    {
      "epoch": 0.6411087435314341,
      "grad_norm": 2.4153220653533936,
      "learning_rate": 6.028042071912439e-05,
      "loss": 1.6604,
      "step": 17902
    },
    {
      "epoch": 0.6411445556610024,
      "grad_norm": 1.8344680070877075,
      "learning_rate": 6.026977617901575e-05,
      "loss": 1.4529,
      "step": 17903
    },
    {
      "epoch": 0.6411803677905706,
      "grad_norm": 1.8805102109909058,
      "learning_rate": 6.0259132173426006e-05,
      "loss": 1.5929,
      "step": 17904
    },
    {
      "epoch": 0.6412161799201389,
      "grad_norm": 1.684512972831726,
      "learning_rate": 6.0248488702498353e-05,
      "loss": 1.1998,
      "step": 17905
    },
    {
      "epoch": 0.6412519920497073,
      "grad_norm": 2.2713093757629395,
      "learning_rate": 6.0237845766376035e-05,
      "loss": 1.6522,
      "step": 17906
    },
    {
      "epoch": 0.6412878041792756,
      "grad_norm": 1.4675745964050293,
      "learning_rate": 6.022720336520218e-05,
      "loss": 1.632,
      "step": 17907
    },
    {
      "epoch": 0.6413236163088438,
      "grad_norm": 1.731246829032898,
      "learning_rate": 6.021656149912003e-05,
      "loss": 1.7003,
      "step": 17908
    },
    {
      "epoch": 0.6413594284384121,
      "grad_norm": 1.5518698692321777,
      "learning_rate": 6.020592016827271e-05,
      "loss": 1.4585,
      "step": 17909
    },
    {
      "epoch": 0.6413952405679804,
      "grad_norm": 1.5268816947937012,
      "learning_rate": 6.019527937280342e-05,
      "loss": 1.244,
      "step": 17910
    },
    {
      "epoch": 0.6414310526975486,
      "grad_norm": 2.2172024250030518,
      "learning_rate": 6.018463911285528e-05,
      "loss": 1.345,
      "step": 17911
    },
    {
      "epoch": 0.6414668648271169,
      "grad_norm": 2.417259693145752,
      "learning_rate": 6.0173999388571486e-05,
      "loss": 1.4226,
      "step": 17912
    },
    {
      "epoch": 0.6415026769566853,
      "grad_norm": 1.5019891262054443,
      "learning_rate": 6.0163360200095153e-05,
      "loss": 1.4445,
      "step": 17913
    },
    {
      "epoch": 0.6415384890862535,
      "grad_norm": 1.449415683746338,
      "learning_rate": 6.015272154756941e-05,
      "loss": 1.5976,
      "step": 17914
    },
    {
      "epoch": 0.6415743012158218,
      "grad_norm": 1.4184887409210205,
      "learning_rate": 6.014208343113741e-05,
      "loss": 1.3103,
      "step": 17915
    },
    {
      "epoch": 0.6416101133453901,
      "grad_norm": 1.948178768157959,
      "learning_rate": 6.0131445850942256e-05,
      "loss": 1.5137,
      "step": 17916
    },
    {
      "epoch": 0.6416459254749584,
      "grad_norm": 2.6223580837249756,
      "learning_rate": 6.012080880712708e-05,
      "loss": 1.2029,
      "step": 17917
    },
    {
      "epoch": 0.6416817376045266,
      "grad_norm": 1.2477376461029053,
      "learning_rate": 6.011017229983497e-05,
      "loss": 1.3365,
      "step": 17918
    },
    {
      "epoch": 0.6417175497340949,
      "grad_norm": 1.6361243724822998,
      "learning_rate": 6.0099536329209046e-05,
      "loss": 1.5658,
      "step": 17919
    },
    {
      "epoch": 0.6417533618636633,
      "grad_norm": 1.3723866939544678,
      "learning_rate": 6.008890089539239e-05,
      "loss": 1.501,
      "step": 17920
    },
    {
      "epoch": 0.6417891739932315,
      "grad_norm": 1.4834378957748413,
      "learning_rate": 6.0078265998528105e-05,
      "loss": 1.2211,
      "step": 17921
    },
    {
      "epoch": 0.6418249861227998,
      "grad_norm": 1.4249910116195679,
      "learning_rate": 6.006763163875925e-05,
      "loss": 1.2466,
      "step": 17922
    },
    {
      "epoch": 0.6418607982523681,
      "grad_norm": 2.4966237545013428,
      "learning_rate": 6.005699781622889e-05,
      "loss": 1.5329,
      "step": 17923
    },
    {
      "epoch": 0.6418966103819364,
      "grad_norm": 1.3982599973678589,
      "learning_rate": 6.00463645310801e-05,
      "loss": 1.3409,
      "step": 17924
    },
    {
      "epoch": 0.6419324225115046,
      "grad_norm": 1.6006054878234863,
      "learning_rate": 6.003573178345594e-05,
      "loss": 1.4297,
      "step": 17925
    },
    {
      "epoch": 0.6419682346410729,
      "grad_norm": 1.729066014289856,
      "learning_rate": 6.002509957349948e-05,
      "loss": 1.3919,
      "step": 17926
    },
    {
      "epoch": 0.6420040467706413,
      "grad_norm": 1.4634580612182617,
      "learning_rate": 6.001446790135371e-05,
      "loss": 1.2687,
      "step": 17927
    },
    {
      "epoch": 0.6420398589002095,
      "grad_norm": 1.6232527494430542,
      "learning_rate": 6.0003836767161726e-05,
      "loss": 1.2855,
      "step": 17928
    },
    {
      "epoch": 0.6420756710297778,
      "grad_norm": 1.6562719345092773,
      "learning_rate": 5.999320617106649e-05,
      "loss": 1.5466,
      "step": 17929
    },
    {
      "epoch": 0.6421114831593461,
      "grad_norm": 1.4762431383132935,
      "learning_rate": 5.9982576113211095e-05,
      "loss": 1.2822,
      "step": 17930
    },
    {
      "epoch": 0.6421472952889143,
      "grad_norm": 1.5727423429489136,
      "learning_rate": 5.9971946593738525e-05,
      "loss": 1.317,
      "step": 17931
    },
    {
      "epoch": 0.6421831074184826,
      "grad_norm": 1.6262317895889282,
      "learning_rate": 5.996131761279176e-05,
      "loss": 1.701,
      "step": 17932
    },
    {
      "epoch": 0.6422189195480509,
      "grad_norm": 1.5175750255584717,
      "learning_rate": 5.995068917051383e-05,
      "loss": 1.3711,
      "step": 17933
    },
    {
      "epoch": 0.6422547316776193,
      "grad_norm": 1.4356316328048706,
      "learning_rate": 5.9940061267047695e-05,
      "loss": 1.4751,
      "step": 17934
    },
    {
      "epoch": 0.6422905438071875,
      "grad_norm": 2.049543619155884,
      "learning_rate": 5.992943390253639e-05,
      "loss": 1.6956,
      "step": 17935
    },
    {
      "epoch": 0.6423263559367558,
      "grad_norm": 1.311546802520752,
      "learning_rate": 5.991880707712284e-05,
      "loss": 1.445,
      "step": 17936
    },
    {
      "epoch": 0.6423621680663241,
      "grad_norm": 2.0570836067199707,
      "learning_rate": 5.9908180790950064e-05,
      "loss": 1.3155,
      "step": 17937
    },
    {
      "epoch": 0.6423979801958923,
      "grad_norm": 1.6217472553253174,
      "learning_rate": 5.989755504416098e-05,
      "loss": 1.4565,
      "step": 17938
    },
    {
      "epoch": 0.6424337923254606,
      "grad_norm": 1.4959290027618408,
      "learning_rate": 5.988692983689859e-05,
      "loss": 1.7333,
      "step": 17939
    },
    {
      "epoch": 0.6424696044550289,
      "grad_norm": 1.4493623971939087,
      "learning_rate": 5.98763051693058e-05,
      "loss": 1.2832,
      "step": 17940
    },
    {
      "epoch": 0.6425054165845973,
      "grad_norm": 1.8293143510818481,
      "learning_rate": 5.9865681041525566e-05,
      "loss": 1.7435,
      "step": 17941
    },
    {
      "epoch": 0.6425412287141655,
      "grad_norm": 1.8717924356460571,
      "learning_rate": 5.9855057453700836e-05,
      "loss": 1.7222,
      "step": 17942
    },
    {
      "epoch": 0.6425770408437338,
      "grad_norm": 2.5850019454956055,
      "learning_rate": 5.98444344059745e-05,
      "loss": 1.4265,
      "step": 17943
    },
    {
      "epoch": 0.6426128529733021,
      "grad_norm": 1.774579644203186,
      "learning_rate": 5.9833811898489534e-05,
      "loss": 1.478,
      "step": 17944
    },
    {
      "epoch": 0.6426486651028703,
      "grad_norm": 1.7594568729400635,
      "learning_rate": 5.982318993138879e-05,
      "loss": 1.6105,
      "step": 17945
    },
    {
      "epoch": 0.6426844772324386,
      "grad_norm": 1.3756812810897827,
      "learning_rate": 5.981256850481523e-05,
      "loss": 1.362,
      "step": 17946
    },
    {
      "epoch": 0.6427202893620069,
      "grad_norm": 1.6097584962844849,
      "learning_rate": 5.980194761891169e-05,
      "loss": 1.5736,
      "step": 17947
    },
    {
      "epoch": 0.6427561014915752,
      "grad_norm": 1.5040655136108398,
      "learning_rate": 5.9791327273821105e-05,
      "loss": 1.577,
      "step": 17948
    },
    {
      "epoch": 0.6427919136211435,
      "grad_norm": 2.5497679710388184,
      "learning_rate": 5.978070746968637e-05,
      "loss": 1.4751,
      "step": 17949
    },
    {
      "epoch": 0.6428277257507118,
      "grad_norm": 1.8910397291183472,
      "learning_rate": 5.977008820665031e-05,
      "loss": 1.6845,
      "step": 17950
    },
    {
      "epoch": 0.6428635378802801,
      "grad_norm": 1.77099609375,
      "learning_rate": 5.975946948485583e-05,
      "loss": 1.5673,
      "step": 17951
    },
    {
      "epoch": 0.6428993500098483,
      "grad_norm": 2.0042824745178223,
      "learning_rate": 5.974885130444577e-05,
      "loss": 1.5555,
      "step": 17952
    },
    {
      "epoch": 0.6429351621394166,
      "grad_norm": 1.585511565208435,
      "learning_rate": 5.9738233665563017e-05,
      "loss": 1.3588,
      "step": 17953
    },
    {
      "epoch": 0.6429709742689849,
      "grad_norm": 1.3501858711242676,
      "learning_rate": 5.972761656835038e-05,
      "loss": 1.2952,
      "step": 17954
    },
    {
      "epoch": 0.6430067863985532,
      "grad_norm": 1.8139315843582153,
      "learning_rate": 5.971700001295072e-05,
      "loss": 1.4315,
      "step": 17955
    },
    {
      "epoch": 0.6430425985281215,
      "grad_norm": 1.828161597251892,
      "learning_rate": 5.9706383999506855e-05,
      "loss": 1.4066,
      "step": 17956
    },
    {
      "epoch": 0.6430784106576898,
      "grad_norm": 1.7610077857971191,
      "learning_rate": 5.969576852816163e-05,
      "loss": 1.7405,
      "step": 17957
    },
    {
      "epoch": 0.643114222787258,
      "grad_norm": 2.1214122772216797,
      "learning_rate": 5.968515359905785e-05,
      "loss": 1.4642,
      "step": 17958
    },
    {
      "epoch": 0.6431500349168263,
      "grad_norm": 2.081220865249634,
      "learning_rate": 5.967453921233832e-05,
      "loss": 1.465,
      "step": 17959
    },
    {
      "epoch": 0.6431858470463946,
      "grad_norm": 2.047666549682617,
      "learning_rate": 5.966392536814585e-05,
      "loss": 1.4785,
      "step": 17960
    },
    {
      "epoch": 0.6432216591759629,
      "grad_norm": 2.4153099060058594,
      "learning_rate": 5.9653312066623234e-05,
      "loss": 1.7967,
      "step": 17961
    },
    {
      "epoch": 0.6432574713055312,
      "grad_norm": 1.863404393196106,
      "learning_rate": 5.964269930791326e-05,
      "loss": 1.5028,
      "step": 17962
    },
    {
      "epoch": 0.6432932834350995,
      "grad_norm": 1.4706677198410034,
      "learning_rate": 5.963208709215871e-05,
      "loss": 1.3202,
      "step": 17963
    },
    {
      "epoch": 0.6433290955646678,
      "grad_norm": 1.6551233530044556,
      "learning_rate": 5.962147541950236e-05,
      "loss": 1.4942,
      "step": 17964
    },
    {
      "epoch": 0.643364907694236,
      "grad_norm": 2.2377161979675293,
      "learning_rate": 5.961086429008696e-05,
      "loss": 1.3441,
      "step": 17965
    },
    {
      "epoch": 0.6434007198238043,
      "grad_norm": 1.5724660158157349,
      "learning_rate": 5.960025370405531e-05,
      "loss": 1.3756,
      "step": 17966
    },
    {
      "epoch": 0.6434365319533726,
      "grad_norm": 1.5656579732894897,
      "learning_rate": 5.958964366155014e-05,
      "loss": 1.4846,
      "step": 17967
    },
    {
      "epoch": 0.6434723440829409,
      "grad_norm": 2.0878260135650635,
      "learning_rate": 5.957903416271414e-05,
      "loss": 1.6058,
      "step": 17968
    },
    {
      "epoch": 0.6435081562125092,
      "grad_norm": 1.9619213342666626,
      "learning_rate": 5.9568425207690146e-05,
      "loss": 1.5998,
      "step": 17969
    },
    {
      "epoch": 0.6435439683420775,
      "grad_norm": 1.3761613368988037,
      "learning_rate": 5.9557816796620804e-05,
      "loss": 1.6619,
      "step": 17970
    },
    {
      "epoch": 0.6435797804716458,
      "grad_norm": 1.948231816291809,
      "learning_rate": 5.954720892964889e-05,
      "loss": 1.3027,
      "step": 17971
    },
    {
      "epoch": 0.643615592601214,
      "grad_norm": 1.6816844940185547,
      "learning_rate": 5.9536601606917075e-05,
      "loss": 1.4994,
      "step": 17972
    },
    {
      "epoch": 0.6436514047307823,
      "grad_norm": 1.75279700756073,
      "learning_rate": 5.952599482856811e-05,
      "loss": 1.5466,
      "step": 17973
    },
    {
      "epoch": 0.6436872168603506,
      "grad_norm": 1.7188254594802856,
      "learning_rate": 5.951538859474467e-05,
      "loss": 1.5254,
      "step": 17974
    },
    {
      "epoch": 0.6437230289899188,
      "grad_norm": 1.8281207084655762,
      "learning_rate": 5.950478290558947e-05,
      "loss": 1.5956,
      "step": 17975
    },
    {
      "epoch": 0.6437588411194872,
      "grad_norm": 1.9698082208633423,
      "learning_rate": 5.9494177761245194e-05,
      "loss": 1.5074,
      "step": 17976
    },
    {
      "epoch": 0.6437946532490555,
      "grad_norm": 1.3871486186981201,
      "learning_rate": 5.9483573161854464e-05,
      "loss": 1.5655,
      "step": 17977
    },
    {
      "epoch": 0.6438304653786238,
      "grad_norm": 1.6179381608963013,
      "learning_rate": 5.947296910756004e-05,
      "loss": 1.2935,
      "step": 17978
    },
    {
      "epoch": 0.643866277508192,
      "grad_norm": 1.5491410493850708,
      "learning_rate": 5.946236559850449e-05,
      "loss": 1.4044,
      "step": 17979
    },
    {
      "epoch": 0.6439020896377603,
      "grad_norm": 2.000506639480591,
      "learning_rate": 5.945176263483057e-05,
      "loss": 1.536,
      "step": 17980
    },
    {
      "epoch": 0.6439379017673286,
      "grad_norm": 1.5044797658920288,
      "learning_rate": 5.9441160216680826e-05,
      "loss": 1.6398,
      "step": 17981
    },
    {
      "epoch": 0.6439737138968968,
      "grad_norm": 2.1576712131500244,
      "learning_rate": 5.9430558344198016e-05,
      "loss": 1.4296,
      "step": 17982
    },
    {
      "epoch": 0.6440095260264652,
      "grad_norm": 1.4655178785324097,
      "learning_rate": 5.941995701752465e-05,
      "loss": 1.4955,
      "step": 17983
    },
    {
      "epoch": 0.6440453381560335,
      "grad_norm": 1.7146788835525513,
      "learning_rate": 5.9409356236803456e-05,
      "loss": 1.6319,
      "step": 17984
    },
    {
      "epoch": 0.6440811502856018,
      "grad_norm": 2.0093584060668945,
      "learning_rate": 5.9398756002177035e-05,
      "loss": 1.4727,
      "step": 17985
    },
    {
      "epoch": 0.64411696241517,
      "grad_norm": 1.5980204343795776,
      "learning_rate": 5.938815631378794e-05,
      "loss": 1.4705,
      "step": 17986
    },
    {
      "epoch": 0.6441527745447383,
      "grad_norm": 1.7558846473693848,
      "learning_rate": 5.937755717177885e-05,
      "loss": 1.4372,
      "step": 17987
    },
    {
      "epoch": 0.6441885866743066,
      "grad_norm": 2.151200294494629,
      "learning_rate": 5.9366958576292284e-05,
      "loss": 1.5712,
      "step": 17988
    },
    {
      "epoch": 0.6442243988038748,
      "grad_norm": 1.6231688261032104,
      "learning_rate": 5.9356360527470934e-05,
      "loss": 1.5356,
      "step": 17989
    },
    {
      "epoch": 0.6442602109334432,
      "grad_norm": 2.068516969680786,
      "learning_rate": 5.9345763025457266e-05,
      "loss": 1.7413,
      "step": 17990
    },
    {
      "epoch": 0.6442960230630115,
      "grad_norm": 1.3470734357833862,
      "learning_rate": 5.9335166070393975e-05,
      "loss": 1.6185,
      "step": 17991
    },
    {
      "epoch": 0.6443318351925797,
      "grad_norm": 1.7504842281341553,
      "learning_rate": 5.93245696624235e-05,
      "loss": 1.5977,
      "step": 17992
    },
    {
      "epoch": 0.644367647322148,
      "grad_norm": 1.6177539825439453,
      "learning_rate": 5.931397380168855e-05,
      "loss": 1.3695,
      "step": 17993
    },
    {
      "epoch": 0.6444034594517163,
      "grad_norm": 2.0384392738342285,
      "learning_rate": 5.9303378488331576e-05,
      "loss": 1.5366,
      "step": 17994
    },
    {
      "epoch": 0.6444392715812846,
      "grad_norm": 2.010671377182007,
      "learning_rate": 5.9292783722495126e-05,
      "loss": 1.5376,
      "step": 17995
    },
    {
      "epoch": 0.6444750837108528,
      "grad_norm": 1.5663658380508423,
      "learning_rate": 5.928218950432179e-05,
      "loss": 1.3194,
      "step": 17996
    },
    {
      "epoch": 0.6445108958404212,
      "grad_norm": 1.3770331144332886,
      "learning_rate": 5.927159583395403e-05,
      "loss": 1.4241,
      "step": 17997
    },
    {
      "epoch": 0.6445467079699895,
      "grad_norm": 1.6570316553115845,
      "learning_rate": 5.926100271153446e-05,
      "loss": 1.4073,
      "step": 17998
    },
    {
      "epoch": 0.6445825200995577,
      "grad_norm": 1.6265000104904175,
      "learning_rate": 5.9250410137205506e-05,
      "loss": 1.485,
      "step": 17999
    },
    {
      "epoch": 0.644618332229126,
      "grad_norm": 1.9524234533309937,
      "learning_rate": 5.923981811110977e-05,
      "loss": 1.7584,
      "step": 18000
    },
    {
      "epoch": 0.6446541443586943,
      "grad_norm": 2.180847406387329,
      "learning_rate": 5.922922663338969e-05,
      "loss": 1.6183,
      "step": 18001
    },
    {
      "epoch": 0.6446899564882625,
      "grad_norm": 1.5719871520996094,
      "learning_rate": 5.921863570418775e-05,
      "loss": 1.5754,
      "step": 18002
    },
    {
      "epoch": 0.6447257686178308,
      "grad_norm": 1.1843252182006836,
      "learning_rate": 5.9208045323646474e-05,
      "loss": 1.3016,
      "step": 18003
    },
    {
      "epoch": 0.6447615807473992,
      "grad_norm": 2.256765604019165,
      "learning_rate": 5.919745549190834e-05,
      "loss": 1.483,
      "step": 18004
    },
    {
      "epoch": 0.6447973928769675,
      "grad_norm": 1.5203441381454468,
      "learning_rate": 5.91868662091158e-05,
      "loss": 1.4898,
      "step": 18005
    },
    {
      "epoch": 0.6448332050065357,
      "grad_norm": 1.4746381044387817,
      "learning_rate": 5.9176277475411324e-05,
      "loss": 1.4107,
      "step": 18006
    },
    {
      "epoch": 0.644869017136104,
      "grad_norm": 1.7098374366760254,
      "learning_rate": 5.91656892909374e-05,
      "loss": 1.1795,
      "step": 18007
    },
    {
      "epoch": 0.6449048292656723,
      "grad_norm": 1.5304396152496338,
      "learning_rate": 5.915510165583642e-05,
      "loss": 1.2104,
      "step": 18008
    },
    {
      "epoch": 0.6449406413952405,
      "grad_norm": 1.4094974994659424,
      "learning_rate": 5.91445145702509e-05,
      "loss": 1.361,
      "step": 18009
    },
    {
      "epoch": 0.6449764535248088,
      "grad_norm": 1.5532375574111938,
      "learning_rate": 5.9133928034323215e-05,
      "loss": 1.5194,
      "step": 18010
    },
    {
      "epoch": 0.6450122656543772,
      "grad_norm": 2.0995876789093018,
      "learning_rate": 5.912334204819581e-05,
      "loss": 1.6518,
      "step": 18011
    },
    {
      "epoch": 0.6450480777839455,
      "grad_norm": 1.8141247034072876,
      "learning_rate": 5.911275661201112e-05,
      "loss": 1.5584,
      "step": 18012
    },
    {
      "epoch": 0.6450838899135137,
      "grad_norm": 1.501083254814148,
      "learning_rate": 5.910217172591155e-05,
      "loss": 1.437,
      "step": 18013
    },
    {
      "epoch": 0.645119702043082,
      "grad_norm": 1.8988889455795288,
      "learning_rate": 5.90915873900395e-05,
      "loss": 1.2275,
      "step": 18014
    },
    {
      "epoch": 0.6451555141726503,
      "grad_norm": 1.5784447193145752,
      "learning_rate": 5.908100360453737e-05,
      "loss": 1.5859,
      "step": 18015
    },
    {
      "epoch": 0.6451913263022185,
      "grad_norm": 1.3819835186004639,
      "learning_rate": 5.9070420369547564e-05,
      "loss": 1.3364,
      "step": 18016
    },
    {
      "epoch": 0.6452271384317868,
      "grad_norm": 1.9142459630966187,
      "learning_rate": 5.905983768521244e-05,
      "loss": 1.2799,
      "step": 18017
    },
    {
      "epoch": 0.6452629505613552,
      "grad_norm": 1.3652900457382202,
      "learning_rate": 5.904925555167442e-05,
      "loss": 1.206,
      "step": 18018
    },
    {
      "epoch": 0.6452987626909235,
      "grad_norm": 1.5359416007995605,
      "learning_rate": 5.903867396907583e-05,
      "loss": 1.1311,
      "step": 18019
    },
    {
      "epoch": 0.6453345748204917,
      "grad_norm": 1.9005250930786133,
      "learning_rate": 5.9028092937559034e-05,
      "loss": 1.5002,
      "step": 18020
    },
    {
      "epoch": 0.64537038695006,
      "grad_norm": 1.9126530885696411,
      "learning_rate": 5.901751245726641e-05,
      "loss": 1.7099,
      "step": 18021
    },
    {
      "epoch": 0.6454061990796283,
      "grad_norm": 1.833815574645996,
      "learning_rate": 5.9006932528340284e-05,
      "loss": 1.4293,
      "step": 18022
    },
    {
      "epoch": 0.6454420112091965,
      "grad_norm": 3.0067315101623535,
      "learning_rate": 5.899635315092301e-05,
      "loss": 1.5751,
      "step": 18023
    },
    {
      "epoch": 0.6454778233387648,
      "grad_norm": 1.957001805305481,
      "learning_rate": 5.89857743251569e-05,
      "loss": 1.3815,
      "step": 18024
    },
    {
      "epoch": 0.6455136354683331,
      "grad_norm": 1.573081135749817,
      "learning_rate": 5.897519605118431e-05,
      "loss": 1.3973,
      "step": 18025
    },
    {
      "epoch": 0.6455494475979014,
      "grad_norm": 1.9092211723327637,
      "learning_rate": 5.896461832914753e-05,
      "loss": 1.6152,
      "step": 18026
    },
    {
      "epoch": 0.6455852597274697,
      "grad_norm": 1.6830337047576904,
      "learning_rate": 5.8954041159188876e-05,
      "loss": 1.8081,
      "step": 18027
    },
    {
      "epoch": 0.645621071857038,
      "grad_norm": 1.7186933755874634,
      "learning_rate": 5.894346454145068e-05,
      "loss": 1.4044,
      "step": 18028
    },
    {
      "epoch": 0.6456568839866063,
      "grad_norm": 1.6428803205490112,
      "learning_rate": 5.8932888476075166e-05,
      "loss": 1.5604,
      "step": 18029
    },
    {
      "epoch": 0.6456926961161745,
      "grad_norm": 2.214829683303833,
      "learning_rate": 5.89223129632047e-05,
      "loss": 1.3191,
      "step": 18030
    },
    {
      "epoch": 0.6457285082457428,
      "grad_norm": 1.8516603708267212,
      "learning_rate": 5.8911738002981506e-05,
      "loss": 1.5621,
      "step": 18031
    },
    {
      "epoch": 0.6457643203753111,
      "grad_norm": 2.198071002960205,
      "learning_rate": 5.890116359554789e-05,
      "loss": 1.398,
      "step": 18032
    },
    {
      "epoch": 0.6458001325048794,
      "grad_norm": 1.7119433879852295,
      "learning_rate": 5.8890589741046084e-05,
      "loss": 1.6987,
      "step": 18033
    },
    {
      "epoch": 0.6458359446344477,
      "grad_norm": 1.5865453481674194,
      "learning_rate": 5.888001643961839e-05,
      "loss": 1.4178,
      "step": 18034
    },
    {
      "epoch": 0.645871756764016,
      "grad_norm": 1.7404098510742188,
      "learning_rate": 5.886944369140701e-05,
      "loss": 1.8102,
      "step": 18035
    },
    {
      "epoch": 0.6459075688935842,
      "grad_norm": 1.6258031129837036,
      "learning_rate": 5.8858871496554235e-05,
      "loss": 1.4598,
      "step": 18036
    },
    {
      "epoch": 0.6459433810231525,
      "grad_norm": 2.006159543991089,
      "learning_rate": 5.884829985520227e-05,
      "loss": 1.6035,
      "step": 18037
    },
    {
      "epoch": 0.6459791931527208,
      "grad_norm": 1.9466265439987183,
      "learning_rate": 5.883772876749334e-05,
      "loss": 1.3743,
      "step": 18038
    },
    {
      "epoch": 0.6460150052822891,
      "grad_norm": 1.9104527235031128,
      "learning_rate": 5.882715823356968e-05,
      "loss": 1.4299,
      "step": 18039
    },
    {
      "epoch": 0.6460508174118574,
      "grad_norm": 2.058929920196533,
      "learning_rate": 5.881658825357348e-05,
      "loss": 1.7316,
      "step": 18040
    },
    {
      "epoch": 0.6460866295414257,
      "grad_norm": 1.6255000829696655,
      "learning_rate": 5.8806018827646994e-05,
      "loss": 1.4241,
      "step": 18041
    },
    {
      "epoch": 0.646122441670994,
      "grad_norm": 1.745592713356018,
      "learning_rate": 5.879544995593236e-05,
      "loss": 1.6278,
      "step": 18042
    },
    {
      "epoch": 0.6461582538005622,
      "grad_norm": 1.3595131635665894,
      "learning_rate": 5.878488163857181e-05,
      "loss": 1.262,
      "step": 18043
    },
    {
      "epoch": 0.6461940659301305,
      "grad_norm": 1.6638784408569336,
      "learning_rate": 5.87743138757075e-05,
      "loss": 1.5087,
      "step": 18044
    },
    {
      "epoch": 0.6462298780596988,
      "grad_norm": 1.6791967153549194,
      "learning_rate": 5.8763746667481634e-05,
      "loss": 1.3979,
      "step": 18045
    },
    {
      "epoch": 0.646265690189267,
      "grad_norm": 1.6635165214538574,
      "learning_rate": 5.8753180014036377e-05,
      "loss": 1.4787,
      "step": 18046
    },
    {
      "epoch": 0.6463015023188354,
      "grad_norm": 1.7006889581680298,
      "learning_rate": 5.874261391551386e-05,
      "loss": 1.5113,
      "step": 18047
    },
    {
      "epoch": 0.6463373144484037,
      "grad_norm": 1.8248895406723022,
      "learning_rate": 5.873204837205626e-05,
      "loss": 1.518,
      "step": 18048
    },
    {
      "epoch": 0.646373126577972,
      "grad_norm": 1.6426078081130981,
      "learning_rate": 5.8721483383805696e-05,
      "loss": 1.4435,
      "step": 18049
    },
    {
      "epoch": 0.6464089387075402,
      "grad_norm": 1.9970711469650269,
      "learning_rate": 5.871091895090437e-05,
      "loss": 1.2891,
      "step": 18050
    },
    {
      "epoch": 0.6464447508371085,
      "grad_norm": 2.244947910308838,
      "learning_rate": 5.870035507349434e-05,
      "loss": 1.8148,
      "step": 18051
    },
    {
      "epoch": 0.6464805629666768,
      "grad_norm": 1.321719765663147,
      "learning_rate": 5.8689791751717757e-05,
      "loss": 1.286,
      "step": 18052
    },
    {
      "epoch": 0.646516375096245,
      "grad_norm": 1.4628890752792358,
      "learning_rate": 5.867922898571675e-05,
      "loss": 1.6739,
      "step": 18053
    },
    {
      "epoch": 0.6465521872258134,
      "grad_norm": 2.1549174785614014,
      "learning_rate": 5.8668666775633426e-05,
      "loss": 1.8747,
      "step": 18054
    },
    {
      "epoch": 0.6465879993553817,
      "grad_norm": 1.6913148164749146,
      "learning_rate": 5.8658105121609896e-05,
      "loss": 1.4313,
      "step": 18055
    },
    {
      "epoch": 0.64662381148495,
      "grad_norm": 1.9767898321151733,
      "learning_rate": 5.864754402378818e-05,
      "loss": 1.5655,
      "step": 18056
    },
    {
      "epoch": 0.6466596236145182,
      "grad_norm": 1.8027931451797485,
      "learning_rate": 5.863698348231045e-05,
      "loss": 1.4809,
      "step": 18057
    },
    {
      "epoch": 0.6466954357440865,
      "grad_norm": 1.5340521335601807,
      "learning_rate": 5.862642349731874e-05,
      "loss": 1.4937,
      "step": 18058
    },
    {
      "epoch": 0.6467312478736548,
      "grad_norm": 1.65854811668396,
      "learning_rate": 5.861586406895514e-05,
      "loss": 1.2827,
      "step": 18059
    },
    {
      "epoch": 0.646767060003223,
      "grad_norm": 1.2305033206939697,
      "learning_rate": 5.8605305197361705e-05,
      "loss": 1.4615,
      "step": 18060
    },
    {
      "epoch": 0.6468028721327914,
      "grad_norm": 1.3957847356796265,
      "learning_rate": 5.859474688268051e-05,
      "loss": 1.4643,
      "step": 18061
    },
    {
      "epoch": 0.6468386842623597,
      "grad_norm": 1.881250262260437,
      "learning_rate": 5.8584189125053556e-05,
      "loss": 1.2637,
      "step": 18062
    },
    {
      "epoch": 0.646874496391928,
      "grad_norm": 1.5098015069961548,
      "learning_rate": 5.857363192462294e-05,
      "loss": 1.5307,
      "step": 18063
    },
    {
      "epoch": 0.6469103085214962,
      "grad_norm": 1.6502503156661987,
      "learning_rate": 5.8563075281530685e-05,
      "loss": 1.1335,
      "step": 18064
    },
    {
      "epoch": 0.6469461206510645,
      "grad_norm": 1.6741015911102295,
      "learning_rate": 5.855251919591875e-05,
      "loss": 1.5119,
      "step": 18065
    },
    {
      "epoch": 0.6469819327806328,
      "grad_norm": 1.6774035692214966,
      "learning_rate": 5.8541963667929276e-05,
      "loss": 1.4242,
      "step": 18066
    },
    {
      "epoch": 0.647017744910201,
      "grad_norm": 1.69266939163208,
      "learning_rate": 5.8531408697704124e-05,
      "loss": 1.2887,
      "step": 18067
    },
    {
      "epoch": 0.6470535570397694,
      "grad_norm": 1.6085079908370972,
      "learning_rate": 5.852085428538545e-05,
      "loss": 1.5291,
      "step": 18068
    },
    {
      "epoch": 0.6470893691693377,
      "grad_norm": 1.8055675029754639,
      "learning_rate": 5.851030043111512e-05,
      "loss": 1.6574,
      "step": 18069
    },
    {
      "epoch": 0.647125181298906,
      "grad_norm": 1.7611117362976074,
      "learning_rate": 5.849974713503521e-05,
      "loss": 1.5388,
      "step": 18070
    },
    {
      "epoch": 0.6471609934284742,
      "grad_norm": 1.4818743467330933,
      "learning_rate": 5.848919439728765e-05,
      "loss": 1.2733,
      "step": 18071
    },
    {
      "epoch": 0.6471968055580425,
      "grad_norm": 1.5272941589355469,
      "learning_rate": 5.847864221801446e-05,
      "loss": 1.5622,
      "step": 18072
    },
    {
      "epoch": 0.6472326176876108,
      "grad_norm": 1.7738280296325684,
      "learning_rate": 5.8468090597357595e-05,
      "loss": 1.6174,
      "step": 18073
    },
    {
      "epoch": 0.647268429817179,
      "grad_norm": 2.0849111080169678,
      "learning_rate": 5.845753953545894e-05,
      "loss": 1.4549,
      "step": 18074
    },
    {
      "epoch": 0.6473042419467474,
      "grad_norm": 2.5184481143951416,
      "learning_rate": 5.8446989032460574e-05,
      "loss": 1.7102,
      "step": 18075
    },
    {
      "epoch": 0.6473400540763157,
      "grad_norm": 1.650846004486084,
      "learning_rate": 5.84364390885043e-05,
      "loss": 1.2554,
      "step": 18076
    },
    {
      "epoch": 0.6473758662058839,
      "grad_norm": 1.4134989976882935,
      "learning_rate": 5.8425889703732193e-05,
      "loss": 1.3435,
      "step": 18077
    },
    {
      "epoch": 0.6474116783354522,
      "grad_norm": 1.4370696544647217,
      "learning_rate": 5.841534087828604e-05,
      "loss": 1.2767,
      "step": 18078
    },
    {
      "epoch": 0.6474474904650205,
      "grad_norm": 1.943215012550354,
      "learning_rate": 5.840479261230791e-05,
      "loss": 1.6713,
      "step": 18079
    },
    {
      "epoch": 0.6474833025945887,
      "grad_norm": 1.4532264471054077,
      "learning_rate": 5.839424490593957e-05,
      "loss": 1.5316,
      "step": 18080
    },
    {
      "epoch": 0.647519114724157,
      "grad_norm": 2.2967021465301514,
      "learning_rate": 5.8383697759323045e-05,
      "loss": 1.1804,
      "step": 18081
    },
    {
      "epoch": 0.6475549268537254,
      "grad_norm": 1.5080279111862183,
      "learning_rate": 5.8373151172600207e-05,
      "loss": 1.2792,
      "step": 18082
    },
    {
      "epoch": 0.6475907389832937,
      "grad_norm": 1.3563507795333862,
      "learning_rate": 5.836260514591287e-05,
      "loss": 1.6779,
      "step": 18083
    },
    {
      "epoch": 0.6476265511128619,
      "grad_norm": 1.6162803173065186,
      "learning_rate": 5.8352059679402994e-05,
      "loss": 1.4045,
      "step": 18084
    },
    {
      "epoch": 0.6476623632424302,
      "grad_norm": 1.5351731777191162,
      "learning_rate": 5.834151477321242e-05,
      "loss": 1.2639,
      "step": 18085
    },
    {
      "epoch": 0.6476981753719985,
      "grad_norm": 1.8498332500457764,
      "learning_rate": 5.833097042748308e-05,
      "loss": 1.5596,
      "step": 18086
    },
    {
      "epoch": 0.6477339875015667,
      "grad_norm": 1.558227777481079,
      "learning_rate": 5.832042664235673e-05,
      "loss": 1.3214,
      "step": 18087
    },
    {
      "epoch": 0.647769799631135,
      "grad_norm": 1.8643543720245361,
      "learning_rate": 5.8309883417975275e-05,
      "loss": 1.2727,
      "step": 18088
    },
    {
      "epoch": 0.6478056117607034,
      "grad_norm": 1.2156808376312256,
      "learning_rate": 5.829934075448058e-05,
      "loss": 1.1644,
      "step": 18089
    },
    {
      "epoch": 0.6478414238902717,
      "grad_norm": 1.5507395267486572,
      "learning_rate": 5.8288798652014485e-05,
      "loss": 1.4555,
      "step": 18090
    },
    {
      "epoch": 0.6478772360198399,
      "grad_norm": 1.6052091121673584,
      "learning_rate": 5.827825711071877e-05,
      "loss": 1.2833,
      "step": 18091
    },
    {
      "epoch": 0.6479130481494082,
      "grad_norm": 1.7828127145767212,
      "learning_rate": 5.8267716130735295e-05,
      "loss": 1.452,
      "step": 18092
    },
    {
      "epoch": 0.6479488602789765,
      "grad_norm": 2.0876612663269043,
      "learning_rate": 5.82571757122059e-05,
      "loss": 1.4555,
      "step": 18093
    },
    {
      "epoch": 0.6479846724085447,
      "grad_norm": 1.3170816898345947,
      "learning_rate": 5.824663585527232e-05,
      "loss": 1.533,
      "step": 18094
    },
    {
      "epoch": 0.648020484538113,
      "grad_norm": 1.6874192953109741,
      "learning_rate": 5.8236096560076405e-05,
      "loss": 1.3156,
      "step": 18095
    },
    {
      "epoch": 0.6480562966676814,
      "grad_norm": 1.6135790348052979,
      "learning_rate": 5.8225557826759935e-05,
      "loss": 1.5129,
      "step": 18096
    },
    {
      "epoch": 0.6480921087972497,
      "grad_norm": 1.4755977392196655,
      "learning_rate": 5.821501965546474e-05,
      "loss": 1.4064,
      "step": 18097
    },
    {
      "epoch": 0.6481279209268179,
      "grad_norm": 2.032099723815918,
      "learning_rate": 5.820448204633251e-05,
      "loss": 1.5011,
      "step": 18098
    },
    {
      "epoch": 0.6481637330563862,
      "grad_norm": 1.5562313795089722,
      "learning_rate": 5.819394499950508e-05,
      "loss": 1.4597,
      "step": 18099
    },
    {
      "epoch": 0.6481995451859545,
      "grad_norm": 1.6698648929595947,
      "learning_rate": 5.8183408515124216e-05,
      "loss": 1.4674,
      "step": 18100
    },
    {
      "epoch": 0.6482353573155227,
      "grad_norm": 1.8499637842178345,
      "learning_rate": 5.817287259333162e-05,
      "loss": 1.574,
      "step": 18101
    },
    {
      "epoch": 0.648271169445091,
      "grad_norm": 2.0695765018463135,
      "learning_rate": 5.816233723426907e-05,
      "loss": 1.3691,
      "step": 18102
    },
    {
      "epoch": 0.6483069815746594,
      "grad_norm": 1.5621305704116821,
      "learning_rate": 5.81518024380783e-05,
      "loss": 1.9263,
      "step": 18103
    },
    {
      "epoch": 0.6483427937042276,
      "grad_norm": 1.3065502643585205,
      "learning_rate": 5.814126820490109e-05,
      "loss": 1.5336,
      "step": 18104
    },
    {
      "epoch": 0.6483786058337959,
      "grad_norm": 1.9310072660446167,
      "learning_rate": 5.8130734534879075e-05,
      "loss": 1.8141,
      "step": 18105
    },
    {
      "epoch": 0.6484144179633642,
      "grad_norm": 2.5569870471954346,
      "learning_rate": 5.812020142815403e-05,
      "loss": 1.4777,
      "step": 18106
    },
    {
      "epoch": 0.6484502300929325,
      "grad_norm": 1.4359135627746582,
      "learning_rate": 5.810966888486768e-05,
      "loss": 1.0495,
      "step": 18107
    },
    {
      "epoch": 0.6484860422225007,
      "grad_norm": 1.7345796823501587,
      "learning_rate": 5.809913690516169e-05,
      "loss": 1.1905,
      "step": 18108
    },
    {
      "epoch": 0.648521854352069,
      "grad_norm": 1.5571644306182861,
      "learning_rate": 5.808860548917778e-05,
      "loss": 1.4347,
      "step": 18109
    },
    {
      "epoch": 0.6485576664816374,
      "grad_norm": 2.2641749382019043,
      "learning_rate": 5.807807463705754e-05,
      "loss": 1.3108,
      "step": 18110
    },
    {
      "epoch": 0.6485934786112056,
      "grad_norm": 1.5269492864608765,
      "learning_rate": 5.8067544348942825e-05,
      "loss": 1.6782,
      "step": 18111
    },
    {
      "epoch": 0.6486292907407739,
      "grad_norm": 1.2696325778961182,
      "learning_rate": 5.805701462497517e-05,
      "loss": 1.6576,
      "step": 18112
    },
    {
      "epoch": 0.6486651028703422,
      "grad_norm": 2.062124252319336,
      "learning_rate": 5.804648546529627e-05,
      "loss": 1.7734,
      "step": 18113
    },
    {
      "epoch": 0.6487009149999104,
      "grad_norm": 1.5553981065750122,
      "learning_rate": 5.803595687004779e-05,
      "loss": 1.5541,
      "step": 18114
    },
    {
      "epoch": 0.6487367271294787,
      "grad_norm": 2.4918081760406494,
      "learning_rate": 5.802542883937143e-05,
      "loss": 1.7596,
      "step": 18115
    },
    {
      "epoch": 0.648772539259047,
      "grad_norm": 2.098888635635376,
      "learning_rate": 5.801490137340879e-05,
      "loss": 1.3388,
      "step": 18116
    },
    {
      "epoch": 0.6488083513886154,
      "grad_norm": 1.448175311088562,
      "learning_rate": 5.80043744723014e-05,
      "loss": 1.3633,
      "step": 18117
    },
    {
      "epoch": 0.6488441635181836,
      "grad_norm": 1.5029058456420898,
      "learning_rate": 5.7993848136191065e-05,
      "loss": 1.5644,
      "step": 18118
    },
    {
      "epoch": 0.6488799756477519,
      "grad_norm": 2.209278106689453,
      "learning_rate": 5.7983322365219287e-05,
      "loss": 1.5103,
      "step": 18119
    },
    {
      "epoch": 0.6489157877773202,
      "grad_norm": 1.7898609638214111,
      "learning_rate": 5.797279715952774e-05,
      "loss": 1.6017,
      "step": 18120
    },
    {
      "epoch": 0.6489515999068884,
      "grad_norm": 2.122300624847412,
      "learning_rate": 5.796227251925792e-05,
      "loss": 1.7573,
      "step": 18121
    },
    {
      "epoch": 0.6489874120364567,
      "grad_norm": 1.6340723037719727,
      "learning_rate": 5.795174844455157e-05,
      "loss": 1.5623,
      "step": 18122
    },
    {
      "epoch": 0.649023224166025,
      "grad_norm": 1.4408314228057861,
      "learning_rate": 5.7941224935550166e-05,
      "loss": 1.5066,
      "step": 18123
    },
    {
      "epoch": 0.6490590362955934,
      "grad_norm": 1.8224083185195923,
      "learning_rate": 5.793070199239534e-05,
      "loss": 1.2898,
      "step": 18124
    },
    {
      "epoch": 0.6490948484251616,
      "grad_norm": 1.6745611429214478,
      "learning_rate": 5.7920179615228684e-05,
      "loss": 1.5694,
      "step": 18125
    },
    {
      "epoch": 0.6491306605547299,
      "grad_norm": 1.6290042400360107,
      "learning_rate": 5.790965780419171e-05,
      "loss": 1.2513,
      "step": 18126
    },
    {
      "epoch": 0.6491664726842982,
      "grad_norm": 1.885898232460022,
      "learning_rate": 5.7899136559426015e-05,
      "loss": 1.334,
      "step": 18127
    },
    {
      "epoch": 0.6492022848138664,
      "grad_norm": 1.878179669380188,
      "learning_rate": 5.788861588107306e-05,
      "loss": 1.6988,
      "step": 18128
    },
    {
      "epoch": 0.6492380969434347,
      "grad_norm": 2.0154190063476562,
      "learning_rate": 5.787809576927454e-05,
      "loss": 1.3509,
      "step": 18129
    },
    {
      "epoch": 0.649273909073003,
      "grad_norm": 1.7065162658691406,
      "learning_rate": 5.786757622417187e-05,
      "loss": 1.6076,
      "step": 18130
    },
    {
      "epoch": 0.6493097212025714,
      "grad_norm": 1.7891151905059814,
      "learning_rate": 5.7857057245906656e-05,
      "loss": 1.5914,
      "step": 18131
    },
    {
      "epoch": 0.6493455333321396,
      "grad_norm": 1.5112416744232178,
      "learning_rate": 5.784653883462029e-05,
      "loss": 1.3508,
      "step": 18132
    },
    {
      "epoch": 0.6493813454617079,
      "grad_norm": 1.3441530466079712,
      "learning_rate": 5.7836020990454444e-05,
      "loss": 1.4401,
      "step": 18133
    },
    {
      "epoch": 0.6494171575912762,
      "grad_norm": 1.4599716663360596,
      "learning_rate": 5.7825503713550555e-05,
      "loss": 1.3342,
      "step": 18134
    },
    {
      "epoch": 0.6494529697208444,
      "grad_norm": 1.3666131496429443,
      "learning_rate": 5.7814987004050084e-05,
      "loss": 1.2142,
      "step": 18135
    },
    {
      "epoch": 0.6494887818504127,
      "grad_norm": 1.5471409559249878,
      "learning_rate": 5.780447086209453e-05,
      "loss": 1.1893,
      "step": 18136
    },
    {
      "epoch": 0.649524593979981,
      "grad_norm": 1.7935959100723267,
      "learning_rate": 5.779395528782541e-05,
      "loss": 1.6171,
      "step": 18137
    },
    {
      "epoch": 0.6495604061095493,
      "grad_norm": 2.097639799118042,
      "learning_rate": 5.7783440281384205e-05,
      "loss": 1.445,
      "step": 18138
    },
    {
      "epoch": 0.6495962182391176,
      "grad_norm": 1.3056614398956299,
      "learning_rate": 5.777292584291227e-05,
      "loss": 1.4843,
      "step": 18139
    },
    {
      "epoch": 0.6496320303686859,
      "grad_norm": 1.3585635423660278,
      "learning_rate": 5.7762411972551254e-05,
      "loss": 1.2193,
      "step": 18140
    },
    {
      "epoch": 0.6496678424982542,
      "grad_norm": 1.2631720304489136,
      "learning_rate": 5.775189867044244e-05,
      "loss": 1.4512,
      "step": 18141
    },
    {
      "epoch": 0.6497036546278224,
      "grad_norm": 1.6314293146133423,
      "learning_rate": 5.7741385936727375e-05,
      "loss": 1.6188,
      "step": 18142
    },
    {
      "epoch": 0.6497394667573907,
      "grad_norm": 1.9369200468063354,
      "learning_rate": 5.7730873771547423e-05,
      "loss": 1.5405,
      "step": 18143
    },
    {
      "epoch": 0.649775278886959,
      "grad_norm": 1.9945943355560303,
      "learning_rate": 5.772036217504404e-05,
      "loss": 1.4426,
      "step": 18144
    },
    {
      "epoch": 0.6498110910165273,
      "grad_norm": 1.2491214275360107,
      "learning_rate": 5.770985114735868e-05,
      "loss": 1.3435,
      "step": 18145
    },
    {
      "epoch": 0.6498469031460956,
      "grad_norm": 1.999806523323059,
      "learning_rate": 5.76993406886327e-05,
      "loss": 1.4863,
      "step": 18146
    },
    {
      "epoch": 0.6498827152756639,
      "grad_norm": 1.5527349710464478,
      "learning_rate": 5.768883079900751e-05,
      "loss": 1.517,
      "step": 18147
    },
    {
      "epoch": 0.6499185274052321,
      "grad_norm": 1.8263483047485352,
      "learning_rate": 5.767832147862452e-05,
      "loss": 1.562,
      "step": 18148
    },
    {
      "epoch": 0.6499543395348004,
      "grad_norm": 1.532959222793579,
      "learning_rate": 5.7667812727625184e-05,
      "loss": 1.107,
      "step": 18149
    },
    {
      "epoch": 0.6499901516643687,
      "grad_norm": 1.593616008758545,
      "learning_rate": 5.765730454615072e-05,
      "loss": 1.4869,
      "step": 18150
    },
    {
      "epoch": 0.650025963793937,
      "grad_norm": 1.7059237957000732,
      "learning_rate": 5.764679693434269e-05,
      "loss": 1.5631,
      "step": 18151
    },
    {
      "epoch": 0.6500617759235053,
      "grad_norm": 2.210345506668091,
      "learning_rate": 5.763628989234238e-05,
      "loss": 1.4786,
      "step": 18152
    },
    {
      "epoch": 0.6500975880530736,
      "grad_norm": 1.792277455329895,
      "learning_rate": 5.76257834202911e-05,
      "loss": 1.3773,
      "step": 18153
    },
    {
      "epoch": 0.6501334001826419,
      "grad_norm": 1.5056220293045044,
      "learning_rate": 5.761527751833026e-05,
      "loss": 1.4831,
      "step": 18154
    },
    {
      "epoch": 0.6501692123122101,
      "grad_norm": 2.4533936977386475,
      "learning_rate": 5.760477218660119e-05,
      "loss": 1.5878,
      "step": 18155
    },
    {
      "epoch": 0.6502050244417784,
      "grad_norm": 1.424676537513733,
      "learning_rate": 5.759426742524524e-05,
      "loss": 1.2297,
      "step": 18156
    },
    {
      "epoch": 0.6502408365713467,
      "grad_norm": 1.37893807888031,
      "learning_rate": 5.75837632344037e-05,
      "loss": 1.6512,
      "step": 18157
    },
    {
      "epoch": 0.650276648700915,
      "grad_norm": 1.7505850791931152,
      "learning_rate": 5.757325961421791e-05,
      "loss": 1.4344,
      "step": 18158
    },
    {
      "epoch": 0.6503124608304833,
      "grad_norm": 1.5535842180252075,
      "learning_rate": 5.756275656482918e-05,
      "loss": 1.3422,
      "step": 18159
    },
    {
      "epoch": 0.6503482729600516,
      "grad_norm": 1.5393567085266113,
      "learning_rate": 5.7552254086378863e-05,
      "loss": 1.4882,
      "step": 18160
    },
    {
      "epoch": 0.6503840850896199,
      "grad_norm": 1.7456070184707642,
      "learning_rate": 5.754175217900817e-05,
      "loss": 1.766,
      "step": 18161
    },
    {
      "epoch": 0.6504198972191881,
      "grad_norm": 2.0329902172088623,
      "learning_rate": 5.753125084285844e-05,
      "loss": 1.5138,
      "step": 18162
    },
    {
      "epoch": 0.6504557093487564,
      "grad_norm": 1.3546897172927856,
      "learning_rate": 5.752075007807098e-05,
      "loss": 1.5011,
      "step": 18163
    },
    {
      "epoch": 0.6504915214783247,
      "grad_norm": 1.7984650135040283,
      "learning_rate": 5.751024988478701e-05,
      "loss": 1.5396,
      "step": 18164
    },
    {
      "epoch": 0.6505273336078929,
      "grad_norm": 1.3731608390808105,
      "learning_rate": 5.749975026314781e-05,
      "loss": 1.4057,
      "step": 18165
    },
    {
      "epoch": 0.6505631457374613,
      "grad_norm": 2.18913197517395,
      "learning_rate": 5.748925121329465e-05,
      "loss": 1.5947,
      "step": 18166
    },
    {
      "epoch": 0.6505989578670296,
      "grad_norm": 1.4021323919296265,
      "learning_rate": 5.747875273536882e-05,
      "loss": 1.316,
      "step": 18167
    },
    {
      "epoch": 0.6506347699965979,
      "grad_norm": 1.6417192220687866,
      "learning_rate": 5.746825482951148e-05,
      "loss": 1.4414,
      "step": 18168
    },
    {
      "epoch": 0.6506705821261661,
      "grad_norm": 2.237081289291382,
      "learning_rate": 5.7457757495863916e-05,
      "loss": 1.2582,
      "step": 18169
    },
    {
      "epoch": 0.6507063942557344,
      "grad_norm": 1.3216358423233032,
      "learning_rate": 5.744726073456739e-05,
      "loss": 1.7212,
      "step": 18170
    },
    {
      "epoch": 0.6507422063853027,
      "grad_norm": 1.4945186376571655,
      "learning_rate": 5.7436764545763034e-05,
      "loss": 1.6712,
      "step": 18171
    },
    {
      "epoch": 0.6507780185148709,
      "grad_norm": 2.1347923278808594,
      "learning_rate": 5.7426268929592105e-05,
      "loss": 1.6936,
      "step": 18172
    },
    {
      "epoch": 0.6508138306444393,
      "grad_norm": 1.752018690109253,
      "learning_rate": 5.7415773886195834e-05,
      "loss": 1.61,
      "step": 18173
    },
    {
      "epoch": 0.6508496427740076,
      "grad_norm": 1.5680664777755737,
      "learning_rate": 5.740527941571541e-05,
      "loss": 1.3054,
      "step": 18174
    },
    {
      "epoch": 0.6508854549035759,
      "grad_norm": 1.7406561374664307,
      "learning_rate": 5.739478551829198e-05,
      "loss": 1.3754,
      "step": 18175
    },
    {
      "epoch": 0.6509212670331441,
      "grad_norm": 1.6612472534179688,
      "learning_rate": 5.738429219406676e-05,
      "loss": 1.2253,
      "step": 18176
    },
    {
      "epoch": 0.6509570791627124,
      "grad_norm": 1.6799381971359253,
      "learning_rate": 5.7373799443180906e-05,
      "loss": 1.5481,
      "step": 18177
    },
    {
      "epoch": 0.6509928912922807,
      "grad_norm": 2.118349313735962,
      "learning_rate": 5.7363307265775635e-05,
      "loss": 1.381,
      "step": 18178
    },
    {
      "epoch": 0.6510287034218489,
      "grad_norm": 1.8382169008255005,
      "learning_rate": 5.7352815661992046e-05,
      "loss": 1.4768,
      "step": 18179
    },
    {
      "epoch": 0.6510645155514173,
      "grad_norm": 1.4919451475143433,
      "learning_rate": 5.734232463197129e-05,
      "loss": 1.1821,
      "step": 18180
    },
    {
      "epoch": 0.6511003276809856,
      "grad_norm": 1.5245254039764404,
      "learning_rate": 5.7331834175854596e-05,
      "loss": 1.4771,
      "step": 18181
    },
    {
      "epoch": 0.6511361398105538,
      "grad_norm": 1.4875792264938354,
      "learning_rate": 5.732134429378297e-05,
      "loss": 1.3327,
      "step": 18182
    },
    {
      "epoch": 0.6511719519401221,
      "grad_norm": 2.057316303253174,
      "learning_rate": 5.731085498589761e-05,
      "loss": 1.2137,
      "step": 18183
    },
    {
      "epoch": 0.6512077640696904,
      "grad_norm": 1.833107352256775,
      "learning_rate": 5.730036625233963e-05,
      "loss": 2.0461,
      "step": 18184
    },
    {
      "epoch": 0.6512435761992587,
      "grad_norm": 1.508131742477417,
      "learning_rate": 5.728987809325019e-05,
      "loss": 1.6103,
      "step": 18185
    },
    {
      "epoch": 0.6512793883288269,
      "grad_norm": 1.5430577993392944,
      "learning_rate": 5.727939050877031e-05,
      "loss": 1.5925,
      "step": 18186
    },
    {
      "epoch": 0.6513152004583953,
      "grad_norm": 2.0990476608276367,
      "learning_rate": 5.726890349904113e-05,
      "loss": 1.5422,
      "step": 18187
    },
    {
      "epoch": 0.6513510125879636,
      "grad_norm": 2.0384583473205566,
      "learning_rate": 5.725841706420376e-05,
      "loss": 1.45,
      "step": 18188
    },
    {
      "epoch": 0.6513868247175318,
      "grad_norm": 1.6039574146270752,
      "learning_rate": 5.724793120439923e-05,
      "loss": 1.7583,
      "step": 18189
    },
    {
      "epoch": 0.6514226368471001,
      "grad_norm": 1.6907927989959717,
      "learning_rate": 5.723744591976863e-05,
      "loss": 1.2868,
      "step": 18190
    },
    {
      "epoch": 0.6514584489766684,
      "grad_norm": 1.6348273754119873,
      "learning_rate": 5.722696121045303e-05,
      "loss": 1.482,
      "step": 18191
    },
    {
      "epoch": 0.6514942611062366,
      "grad_norm": 1.6397366523742676,
      "learning_rate": 5.7216477076593544e-05,
      "loss": 1.4915,
      "step": 18192
    },
    {
      "epoch": 0.6515300732358049,
      "grad_norm": 1.57917058467865,
      "learning_rate": 5.7205993518331134e-05,
      "loss": 1.2427,
      "step": 18193
    },
    {
      "epoch": 0.6515658853653733,
      "grad_norm": 1.8741562366485596,
      "learning_rate": 5.719551053580687e-05,
      "loss": 1.4965,
      "step": 18194
    },
    {
      "epoch": 0.6516016974949416,
      "grad_norm": 1.567555546760559,
      "learning_rate": 5.718502812916186e-05,
      "loss": 1.158,
      "step": 18195
    },
    {
      "epoch": 0.6516375096245098,
      "grad_norm": 1.4905515909194946,
      "learning_rate": 5.7174546298537005e-05,
      "loss": 1.5231,
      "step": 18196
    },
    {
      "epoch": 0.6516733217540781,
      "grad_norm": 1.853344202041626,
      "learning_rate": 5.71640650440734e-05,
      "loss": 1.7658,
      "step": 18197
    },
    {
      "epoch": 0.6517091338836464,
      "grad_norm": 2.142179250717163,
      "learning_rate": 5.715358436591205e-05,
      "loss": 1.9089,
      "step": 18198
    },
    {
      "epoch": 0.6517449460132146,
      "grad_norm": 2.0253007411956787,
      "learning_rate": 5.7143104264193984e-05,
      "loss": 1.3156,
      "step": 18199
    },
    {
      "epoch": 0.6517807581427829,
      "grad_norm": 1.5804954767227173,
      "learning_rate": 5.7132624739060134e-05,
      "loss": 1.1425,
      "step": 18200
    },
    {
      "epoch": 0.6518165702723513,
      "grad_norm": 1.8263784646987915,
      "learning_rate": 5.712214579065152e-05,
      "loss": 1.3975,
      "step": 18201
    },
    {
      "epoch": 0.6518523824019196,
      "grad_norm": 1.8882803916931152,
      "learning_rate": 5.711166741910912e-05,
      "loss": 1.453,
      "step": 18202
    },
    {
      "epoch": 0.6518881945314878,
      "grad_norm": 1.7997485399246216,
      "learning_rate": 5.710118962457396e-05,
      "loss": 1.3057,
      "step": 18203
    },
    {
      "epoch": 0.6519240066610561,
      "grad_norm": 1.5908443927764893,
      "learning_rate": 5.709071240718695e-05,
      "loss": 1.2405,
      "step": 18204
    },
    {
      "epoch": 0.6519598187906244,
      "grad_norm": 1.749822974205017,
      "learning_rate": 5.7080235767088994e-05,
      "loss": 1.4987,
      "step": 18205
    },
    {
      "epoch": 0.6519956309201926,
      "grad_norm": 1.3174817562103271,
      "learning_rate": 5.706975970442117e-05,
      "loss": 1.3869,
      "step": 18206
    },
    {
      "epoch": 0.6520314430497609,
      "grad_norm": 1.5723850727081299,
      "learning_rate": 5.7059284219324315e-05,
      "loss": 1.6457,
      "step": 18207
    },
    {
      "epoch": 0.6520672551793293,
      "grad_norm": 1.735958218574524,
      "learning_rate": 5.7048809311939446e-05,
      "loss": 1.3457,
      "step": 18208
    },
    {
      "epoch": 0.6521030673088976,
      "grad_norm": 1.7627031803131104,
      "learning_rate": 5.703833498240736e-05,
      "loss": 1.6223,
      "step": 18209
    },
    {
      "epoch": 0.6521388794384658,
      "grad_norm": 2.1572265625,
      "learning_rate": 5.702786123086914e-05,
      "loss": 1.2937,
      "step": 18210
    },
    {
      "epoch": 0.6521746915680341,
      "grad_norm": 1.6451489925384521,
      "learning_rate": 5.701738805746558e-05,
      "loss": 1.5857,
      "step": 18211
    },
    {
      "epoch": 0.6522105036976024,
      "grad_norm": 1.494231104850769,
      "learning_rate": 5.700691546233762e-05,
      "loss": 1.493,
      "step": 18212
    },
    {
      "epoch": 0.6522463158271706,
      "grad_norm": 1.4229929447174072,
      "learning_rate": 5.699644344562619e-05,
      "loss": 1.355,
      "step": 18213
    },
    {
      "epoch": 0.6522821279567389,
      "grad_norm": 1.5671422481536865,
      "learning_rate": 5.698597200747211e-05,
      "loss": 1.2931,
      "step": 18214
    },
    {
      "epoch": 0.6523179400863073,
      "grad_norm": 1.9639018774032593,
      "learning_rate": 5.697550114801633e-05,
      "loss": 1.4782,
      "step": 18215
    },
    {
      "epoch": 0.6523537522158755,
      "grad_norm": 1.4121224880218506,
      "learning_rate": 5.696503086739961e-05,
      "loss": 1.2247,
      "step": 18216
    },
    {
      "epoch": 0.6523895643454438,
      "grad_norm": 2.129319429397583,
      "learning_rate": 5.695456116576296e-05,
      "loss": 1.4432,
      "step": 18217
    },
    {
      "epoch": 0.6524253764750121,
      "grad_norm": 1.39368736743927,
      "learning_rate": 5.6944092043247124e-05,
      "loss": 1.3975,
      "step": 18218
    },
    {
      "epoch": 0.6524611886045804,
      "grad_norm": 1.3888850212097168,
      "learning_rate": 5.693362349999303e-05,
      "loss": 1.5083,
      "step": 18219
    },
    {
      "epoch": 0.6524970007341486,
      "grad_norm": 1.4592702388763428,
      "learning_rate": 5.6923155536141404e-05,
      "loss": 1.449,
      "step": 18220
    },
    {
      "epoch": 0.6525328128637169,
      "grad_norm": 2.164543628692627,
      "learning_rate": 5.691268815183324e-05,
      "loss": 1.2571,
      "step": 18221
    },
    {
      "epoch": 0.6525686249932853,
      "grad_norm": 2.2034432888031006,
      "learning_rate": 5.690222134720927e-05,
      "loss": 1.562,
      "step": 18222
    },
    {
      "epoch": 0.6526044371228535,
      "grad_norm": 1.5978227853775024,
      "learning_rate": 5.6891755122410254e-05,
      "loss": 1.0764,
      "step": 18223
    },
    {
      "epoch": 0.6526402492524218,
      "grad_norm": 1.8049519062042236,
      "learning_rate": 5.688128947757713e-05,
      "loss": 1.4481,
      "step": 18224
    },
    {
      "epoch": 0.6526760613819901,
      "grad_norm": 1.7930026054382324,
      "learning_rate": 5.687082441285061e-05,
      "loss": 1.5977,
      "step": 18225
    },
    {
      "epoch": 0.6527118735115583,
      "grad_norm": 1.4128412008285522,
      "learning_rate": 5.6860359928371546e-05,
      "loss": 1.7345,
      "step": 18226
    },
    {
      "epoch": 0.6527476856411266,
      "grad_norm": 1.6190688610076904,
      "learning_rate": 5.6849896024280614e-05,
      "loss": 1.2488,
      "step": 18227
    },
    {
      "epoch": 0.6527834977706949,
      "grad_norm": 2.397874593734741,
      "learning_rate": 5.6839432700718743e-05,
      "loss": 1.2673,
      "step": 18228
    },
    {
      "epoch": 0.6528193099002633,
      "grad_norm": 2.7797553539276123,
      "learning_rate": 5.682896995782661e-05,
      "loss": 1.5766,
      "step": 18229
    },
    {
      "epoch": 0.6528551220298315,
      "grad_norm": 2.0291695594787598,
      "learning_rate": 5.6818507795745025e-05,
      "loss": 1.2149,
      "step": 18230
    },
    {
      "epoch": 0.6528909341593998,
      "grad_norm": 1.5554500818252563,
      "learning_rate": 5.6808046214614684e-05,
      "loss": 1.534,
      "step": 18231
    },
    {
      "epoch": 0.6529267462889681,
      "grad_norm": 1.7645570039749146,
      "learning_rate": 5.679758521457637e-05,
      "loss": 1.5774,
      "step": 18232
    },
    {
      "epoch": 0.6529625584185363,
      "grad_norm": 1.3924230337142944,
      "learning_rate": 5.678712479577086e-05,
      "loss": 1.1851,
      "step": 18233
    },
    {
      "epoch": 0.6529983705481046,
      "grad_norm": 1.7153980731964111,
      "learning_rate": 5.67766649583388e-05,
      "loss": 1.3035,
      "step": 18234
    },
    {
      "epoch": 0.6530341826776729,
      "grad_norm": 1.4884206056594849,
      "learning_rate": 5.676620570242097e-05,
      "loss": 1.3867,
      "step": 18235
    },
    {
      "epoch": 0.6530699948072413,
      "grad_norm": 2.1972150802612305,
      "learning_rate": 5.675574702815807e-05,
      "loss": 1.7816,
      "step": 18236
    },
    {
      "epoch": 0.6531058069368095,
      "grad_norm": 2.568058967590332,
      "learning_rate": 5.674528893569084e-05,
      "loss": 1.2583,
      "step": 18237
    },
    {
      "epoch": 0.6531416190663778,
      "grad_norm": 1.8737038373947144,
      "learning_rate": 5.673483142515988e-05,
      "loss": 1.2972,
      "step": 18238
    },
    {
      "epoch": 0.6531774311959461,
      "grad_norm": 1.451999545097351,
      "learning_rate": 5.672437449670605e-05,
      "loss": 1.5523,
      "step": 18239
    },
    {
      "epoch": 0.6532132433255143,
      "grad_norm": 1.7182345390319824,
      "learning_rate": 5.6713918150469916e-05,
      "loss": 1.3202,
      "step": 18240
    },
    {
      "epoch": 0.6532490554550826,
      "grad_norm": 2.7142200469970703,
      "learning_rate": 5.6703462386592145e-05,
      "loss": 1.6818,
      "step": 18241
    },
    {
      "epoch": 0.6532848675846509,
      "grad_norm": 1.4753506183624268,
      "learning_rate": 5.6693007205213444e-05,
      "loss": 1.4071,
      "step": 18242
    },
    {
      "epoch": 0.6533206797142193,
      "grad_norm": 1.5137279033660889,
      "learning_rate": 5.668255260647447e-05,
      "loss": 1.6646,
      "step": 18243
    },
    {
      "epoch": 0.6533564918437875,
      "grad_norm": 1.8348652124404907,
      "learning_rate": 5.667209859051592e-05,
      "loss": 1.3507,
      "step": 18244
    },
    {
      "epoch": 0.6533923039733558,
      "grad_norm": 1.683578610420227,
      "learning_rate": 5.6661645157478336e-05,
      "loss": 1.1708,
      "step": 18245
    },
    {
      "epoch": 0.6534281161029241,
      "grad_norm": 1.2713277339935303,
      "learning_rate": 5.665119230750243e-05,
      "loss": 1.412,
      "step": 18246
    },
    {
      "epoch": 0.6534639282324923,
      "grad_norm": 1.712204098701477,
      "learning_rate": 5.664074004072881e-05,
      "loss": 1.2872,
      "step": 18247
    },
    {
      "epoch": 0.6534997403620606,
      "grad_norm": 1.6084846258163452,
      "learning_rate": 5.663028835729815e-05,
      "loss": 1.4954,
      "step": 18248
    },
    {
      "epoch": 0.6535355524916289,
      "grad_norm": 1.7049838304519653,
      "learning_rate": 5.661983725735096e-05,
      "loss": 1.5376,
      "step": 18249
    },
    {
      "epoch": 0.6535713646211972,
      "grad_norm": 2.174365997314453,
      "learning_rate": 5.6609386741027915e-05,
      "loss": 1.7882,
      "step": 18250
    },
    {
      "epoch": 0.6536071767507655,
      "grad_norm": 1.3150783777236938,
      "learning_rate": 5.659893680846965e-05,
      "loss": 1.4351,
      "step": 18251
    },
    {
      "epoch": 0.6536429888803338,
      "grad_norm": 1.477768063545227,
      "learning_rate": 5.658848745981667e-05,
      "loss": 1.6592,
      "step": 18252
    },
    {
      "epoch": 0.653678801009902,
      "grad_norm": 1.5167200565338135,
      "learning_rate": 5.6578038695209566e-05,
      "loss": 1.3087,
      "step": 18253
    },
    {
      "epoch": 0.6537146131394703,
      "grad_norm": 1.7359503507614136,
      "learning_rate": 5.656759051478897e-05,
      "loss": 1.5542,
      "step": 18254
    },
    {
      "epoch": 0.6537504252690386,
      "grad_norm": 1.6979146003723145,
      "learning_rate": 5.655714291869544e-05,
      "loss": 1.6755,
      "step": 18255
    },
    {
      "epoch": 0.6537862373986069,
      "grad_norm": 1.4741054773330688,
      "learning_rate": 5.654669590706948e-05,
      "loss": 1.3516,
      "step": 18256
    },
    {
      "epoch": 0.6538220495281752,
      "grad_norm": 1.841018795967102,
      "learning_rate": 5.653624948005167e-05,
      "loss": 1.3925,
      "step": 18257
    },
    {
      "epoch": 0.6538578616577435,
      "grad_norm": 1.7704675197601318,
      "learning_rate": 5.6525803637782614e-05,
      "loss": 1.8108,
      "step": 18258
    },
    {
      "epoch": 0.6538936737873118,
      "grad_norm": 1.8896334171295166,
      "learning_rate": 5.651535838040275e-05,
      "loss": 1.6515,
      "step": 18259
    },
    {
      "epoch": 0.65392948591688,
      "grad_norm": 1.491990327835083,
      "learning_rate": 5.6504913708052646e-05,
      "loss": 1.5978,
      "step": 18260
    },
    {
      "epoch": 0.6539652980464483,
      "grad_norm": 1.973260521888733,
      "learning_rate": 5.6494469620872814e-05,
      "loss": 1.4268,
      "step": 18261
    },
    {
      "epoch": 0.6540011101760166,
      "grad_norm": 1.601784348487854,
      "learning_rate": 5.648402611900383e-05,
      "loss": 1.2062,
      "step": 18262
    },
    {
      "epoch": 0.6540369223055849,
      "grad_norm": 1.3702232837677002,
      "learning_rate": 5.647358320258609e-05,
      "loss": 1.3752,
      "step": 18263
    },
    {
      "epoch": 0.6540727344351532,
      "grad_norm": 1.3767213821411133,
      "learning_rate": 5.6463140871760144e-05,
      "loss": 1.5021,
      "step": 18264
    },
    {
      "epoch": 0.6541085465647215,
      "grad_norm": 1.3063963651657104,
      "learning_rate": 5.6452699126666486e-05,
      "loss": 1.5594,
      "step": 18265
    },
    {
      "epoch": 0.6541443586942898,
      "grad_norm": 1.5058377981185913,
      "learning_rate": 5.644225796744562e-05,
      "loss": 1.166,
      "step": 18266
    },
    {
      "epoch": 0.654180170823858,
      "grad_norm": 1.676900863647461,
      "learning_rate": 5.6431817394237964e-05,
      "loss": 1.589,
      "step": 18267
    },
    {
      "epoch": 0.6542159829534263,
      "grad_norm": 1.7111408710479736,
      "learning_rate": 5.6421377407183997e-05,
      "loss": 1.5941,
      "step": 18268
    },
    {
      "epoch": 0.6542517950829946,
      "grad_norm": 1.4880335330963135,
      "learning_rate": 5.641093800642423e-05,
      "loss": 1.2594,
      "step": 18269
    },
    {
      "epoch": 0.6542876072125628,
      "grad_norm": 2.1994059085845947,
      "learning_rate": 5.640049919209902e-05,
      "loss": 1.6901,
      "step": 18270
    },
    {
      "epoch": 0.6543234193421312,
      "grad_norm": 1.6333506107330322,
      "learning_rate": 5.6390060964348845e-05,
      "loss": 1.4167,
      "step": 18271
    },
    {
      "epoch": 0.6543592314716995,
      "grad_norm": 1.4062167406082153,
      "learning_rate": 5.637962332331416e-05,
      "loss": 1.2515,
      "step": 18272
    },
    {
      "epoch": 0.6543950436012678,
      "grad_norm": 2.1119511127471924,
      "learning_rate": 5.636918626913541e-05,
      "loss": 1.6651,
      "step": 18273
    },
    {
      "epoch": 0.654430855730836,
      "grad_norm": 1.8894344568252563,
      "learning_rate": 5.6358749801952946e-05,
      "loss": 1.7571,
      "step": 18274
    },
    {
      "epoch": 0.6544666678604043,
      "grad_norm": 1.7292144298553467,
      "learning_rate": 5.63483139219072e-05,
      "loss": 1.4696,
      "step": 18275
    },
    {
      "epoch": 0.6545024799899726,
      "grad_norm": 1.5995243787765503,
      "learning_rate": 5.633787862913864e-05,
      "loss": 1.1873,
      "step": 18276
    },
    {
      "epoch": 0.6545382921195408,
      "grad_norm": 1.5615102052688599,
      "learning_rate": 5.6327443923787546e-05,
      "loss": 1.3503,
      "step": 18277
    },
    {
      "epoch": 0.6545741042491092,
      "grad_norm": 1.3338347673416138,
      "learning_rate": 5.631700980599437e-05,
      "loss": 1.4405,
      "step": 18278
    },
    {
      "epoch": 0.6546099163786775,
      "grad_norm": 1.7192423343658447,
      "learning_rate": 5.630657627589948e-05,
      "loss": 1.7277,
      "step": 18279
    },
    {
      "epoch": 0.6546457285082458,
      "grad_norm": 2.087306499481201,
      "learning_rate": 5.629614333364328e-05,
      "loss": 1.3391,
      "step": 18280
    },
    {
      "epoch": 0.654681540637814,
      "grad_norm": 1.4687906503677368,
      "learning_rate": 5.628571097936606e-05,
      "loss": 1.3238,
      "step": 18281
    },
    {
      "epoch": 0.6547173527673823,
      "grad_norm": 2.546259641647339,
      "learning_rate": 5.627527921320821e-05,
      "loss": 1.4288,
      "step": 18282
    },
    {
      "epoch": 0.6547531648969506,
      "grad_norm": 1.732775092124939,
      "learning_rate": 5.626484803531008e-05,
      "loss": 1.4941,
      "step": 18283
    },
    {
      "epoch": 0.6547889770265188,
      "grad_norm": 1.4935340881347656,
      "learning_rate": 5.625441744581205e-05,
      "loss": 1.154,
      "step": 18284
    },
    {
      "epoch": 0.6548247891560872,
      "grad_norm": 2.029008150100708,
      "learning_rate": 5.624398744485435e-05,
      "loss": 1.4768,
      "step": 18285
    },
    {
      "epoch": 0.6548606012856555,
      "grad_norm": 1.9982153177261353,
      "learning_rate": 5.623355803257737e-05,
      "loss": 1.6078,
      "step": 18286
    },
    {
      "epoch": 0.6548964134152238,
      "grad_norm": 1.5812995433807373,
      "learning_rate": 5.622312920912145e-05,
      "loss": 1.6001,
      "step": 18287
    },
    {
      "epoch": 0.654932225544792,
      "grad_norm": 2.2737843990325928,
      "learning_rate": 5.621270097462682e-05,
      "loss": 1.7967,
      "step": 18288
    },
    {
      "epoch": 0.6549680376743603,
      "grad_norm": 2.59417724609375,
      "learning_rate": 5.620227332923382e-05,
      "loss": 1.2416,
      "step": 18289
    },
    {
      "epoch": 0.6550038498039286,
      "grad_norm": 1.4791409969329834,
      "learning_rate": 5.619184627308273e-05,
      "loss": 1.3457,
      "step": 18290
    },
    {
      "epoch": 0.6550396619334968,
      "grad_norm": 1.6634405851364136,
      "learning_rate": 5.618141980631389e-05,
      "loss": 1.632,
      "step": 18291
    },
    {
      "epoch": 0.6550754740630652,
      "grad_norm": 2.015676975250244,
      "learning_rate": 5.617099392906751e-05,
      "loss": 1.3177,
      "step": 18292
    },
    {
      "epoch": 0.6551112861926335,
      "grad_norm": 1.8615721464157104,
      "learning_rate": 5.61605686414838e-05,
      "loss": 1.6381,
      "step": 18293
    },
    {
      "epoch": 0.6551470983222017,
      "grad_norm": 1.3926013708114624,
      "learning_rate": 5.615014394370317e-05,
      "loss": 0.9642,
      "step": 18294
    },
    {
      "epoch": 0.65518291045177,
      "grad_norm": 1.9081752300262451,
      "learning_rate": 5.6139719835865745e-05,
      "loss": 1.5377,
      "step": 18295
    },
    {
      "epoch": 0.6552187225813383,
      "grad_norm": 1.6380687952041626,
      "learning_rate": 5.612929631811181e-05,
      "loss": 1.3983,
      "step": 18296
    },
    {
      "epoch": 0.6552545347109066,
      "grad_norm": 1.7119892835617065,
      "learning_rate": 5.611887339058162e-05,
      "loss": 1.6677,
      "step": 18297
    },
    {
      "epoch": 0.6552903468404748,
      "grad_norm": 1.7411001920700073,
      "learning_rate": 5.610845105341542e-05,
      "loss": 1.6975,
      "step": 18298
    },
    {
      "epoch": 0.6553261589700432,
      "grad_norm": 2.2733662128448486,
      "learning_rate": 5.609802930675335e-05,
      "loss": 1.448,
      "step": 18299
    },
    {
      "epoch": 0.6553619710996115,
      "grad_norm": 3.091132640838623,
      "learning_rate": 5.608760815073567e-05,
      "loss": 1.3997,
      "step": 18300
    },
    {
      "epoch": 0.6553977832291797,
      "grad_norm": 1.3527662754058838,
      "learning_rate": 5.6077187585502624e-05,
      "loss": 1.2382,
      "step": 18301
    },
    {
      "epoch": 0.655433595358748,
      "grad_norm": 1.2386977672576904,
      "learning_rate": 5.6066767611194316e-05,
      "loss": 1.4487,
      "step": 18302
    },
    {
      "epoch": 0.6554694074883163,
      "grad_norm": 2.7018849849700928,
      "learning_rate": 5.6056348227951025e-05,
      "loss": 1.5491,
      "step": 18303
    },
    {
      "epoch": 0.6555052196178845,
      "grad_norm": 1.7741349935531616,
      "learning_rate": 5.6045929435912805e-05,
      "loss": 1.446,
      "step": 18304
    },
    {
      "epoch": 0.6555410317474528,
      "grad_norm": 1.929344892501831,
      "learning_rate": 5.603551123521997e-05,
      "loss": 1.2201,
      "step": 18305
    },
    {
      "epoch": 0.6555768438770212,
      "grad_norm": 1.3161699771881104,
      "learning_rate": 5.60250936260126e-05,
      "loss": 1.4935,
      "step": 18306
    },
    {
      "epoch": 0.6556126560065895,
      "grad_norm": 1.606744647026062,
      "learning_rate": 5.601467660843087e-05,
      "loss": 1.2209,
      "step": 18307
    },
    {
      "epoch": 0.6556484681361577,
      "grad_norm": 1.6163558959960938,
      "learning_rate": 5.600426018261493e-05,
      "loss": 1.7188,
      "step": 18308
    },
    {
      "epoch": 0.655684280265726,
      "grad_norm": 1.3618152141571045,
      "learning_rate": 5.599384434870496e-05,
      "loss": 1.2453,
      "step": 18309
    },
    {
      "epoch": 0.6557200923952943,
      "grad_norm": 1.6127922534942627,
      "learning_rate": 5.5983429106841046e-05,
      "loss": 1.5509,
      "step": 18310
    },
    {
      "epoch": 0.6557559045248625,
      "grad_norm": 1.6013727188110352,
      "learning_rate": 5.597301445716323e-05,
      "loss": 1.6438,
      "step": 18311
    },
    {
      "epoch": 0.6557917166544308,
      "grad_norm": 1.5582809448242188,
      "learning_rate": 5.59626003998118e-05,
      "loss": 1.4424,
      "step": 18312
    },
    {
      "epoch": 0.6558275287839992,
      "grad_norm": 1.5921964645385742,
      "learning_rate": 5.595218693492674e-05,
      "loss": 1.3418,
      "step": 18313
    },
    {
      "epoch": 0.6558633409135675,
      "grad_norm": 1.9930670261383057,
      "learning_rate": 5.594177406264822e-05,
      "loss": 1.6266,
      "step": 18314
    },
    {
      "epoch": 0.6558991530431357,
      "grad_norm": 1.9355586767196655,
      "learning_rate": 5.593136178311622e-05,
      "loss": 1.26,
      "step": 18315
    },
    {
      "epoch": 0.655934965172704,
      "grad_norm": 2.1823208332061768,
      "learning_rate": 5.592095009647099e-05,
      "loss": 1.57,
      "step": 18316
    },
    {
      "epoch": 0.6559707773022723,
      "grad_norm": 1.3704748153686523,
      "learning_rate": 5.591053900285248e-05,
      "loss": 1.3415,
      "step": 18317
    },
    {
      "epoch": 0.6560065894318405,
      "grad_norm": 1.7008016109466553,
      "learning_rate": 5.590012850240083e-05,
      "loss": 1.5811,
      "step": 18318
    },
    {
      "epoch": 0.6560424015614088,
      "grad_norm": 2.976754665374756,
      "learning_rate": 5.5889718595256026e-05,
      "loss": 1.6042,
      "step": 18319
    },
    {
      "epoch": 0.6560782136909772,
      "grad_norm": 1.7134968042373657,
      "learning_rate": 5.587930928155816e-05,
      "loss": 1.2071,
      "step": 18320
    },
    {
      "epoch": 0.6561140258205455,
      "grad_norm": 1.4464317560195923,
      "learning_rate": 5.586890056144732e-05,
      "loss": 1.6427,
      "step": 18321
    },
    {
      "epoch": 0.6561498379501137,
      "grad_norm": 2.000373125076294,
      "learning_rate": 5.585849243506342e-05,
      "loss": 1.3813,
      "step": 18322
    },
    {
      "epoch": 0.656185650079682,
      "grad_norm": 1.743994951248169,
      "learning_rate": 5.584808490254664e-05,
      "loss": 1.4915,
      "step": 18323
    },
    {
      "epoch": 0.6562214622092503,
      "grad_norm": 1.9050312042236328,
      "learning_rate": 5.5837677964036894e-05,
      "loss": 1.6474,
      "step": 18324
    },
    {
      "epoch": 0.6562572743388185,
      "grad_norm": 1.8708951473236084,
      "learning_rate": 5.582727161967425e-05,
      "loss": 1.3392,
      "step": 18325
    },
    {
      "epoch": 0.6562930864683868,
      "grad_norm": 1.4769537448883057,
      "learning_rate": 5.5816865869598625e-05,
      "loss": 1.4556,
      "step": 18326
    },
    {
      "epoch": 0.6563288985979552,
      "grad_norm": 1.4558557271957397,
      "learning_rate": 5.5806460713950145e-05,
      "loss": 1.4495,
      "step": 18327
    },
    {
      "epoch": 0.6563647107275234,
      "grad_norm": 2.0436086654663086,
      "learning_rate": 5.579605615286874e-05,
      "loss": 1.5254,
      "step": 18328
    },
    {
      "epoch": 0.6564005228570917,
      "grad_norm": 1.4093581438064575,
      "learning_rate": 5.578565218649433e-05,
      "loss": 1.4388,
      "step": 18329
    },
    {
      "epoch": 0.65643633498666,
      "grad_norm": 1.895922064781189,
      "learning_rate": 5.577524881496694e-05,
      "loss": 1.2775,
      "step": 18330
    },
    {
      "epoch": 0.6564721471162283,
      "grad_norm": 2.0171585083007812,
      "learning_rate": 5.5764846038426535e-05,
      "loss": 1.3864,
      "step": 18331
    },
    {
      "epoch": 0.6565079592457965,
      "grad_norm": 2.5566394329071045,
      "learning_rate": 5.57544438570131e-05,
      "loss": 1.6843,
      "step": 18332
    },
    {
      "epoch": 0.6565437713753648,
      "grad_norm": 1.510467290878296,
      "learning_rate": 5.574404227086648e-05,
      "loss": 1.4668,
      "step": 18333
    },
    {
      "epoch": 0.6565795835049332,
      "grad_norm": 1.726326584815979,
      "learning_rate": 5.573364128012677e-05,
      "loss": 1.2047,
      "step": 18334
    },
    {
      "epoch": 0.6566153956345014,
      "grad_norm": 1.3015578985214233,
      "learning_rate": 5.572324088493377e-05,
      "loss": 1.3682,
      "step": 18335
    },
    {
      "epoch": 0.6566512077640697,
      "grad_norm": 2.0535759925842285,
      "learning_rate": 5.571284108542748e-05,
      "loss": 1.4943,
      "step": 18336
    },
    {
      "epoch": 0.656687019893638,
      "grad_norm": 1.3764218091964722,
      "learning_rate": 5.5702441881747755e-05,
      "loss": 1.1779,
      "step": 18337
    },
    {
      "epoch": 0.6567228320232062,
      "grad_norm": 2.7973520755767822,
      "learning_rate": 5.5692043274034544e-05,
      "loss": 1.6795,
      "step": 18338
    },
    {
      "epoch": 0.6567586441527745,
      "grad_norm": 1.9879013299942017,
      "learning_rate": 5.568164526242776e-05,
      "loss": 1.4525,
      "step": 18339
    },
    {
      "epoch": 0.6567944562823428,
      "grad_norm": 1.5268080234527588,
      "learning_rate": 5.5671247847067254e-05,
      "loss": 1.7501,
      "step": 18340
    },
    {
      "epoch": 0.6568302684119112,
      "grad_norm": 1.5404284000396729,
      "learning_rate": 5.566085102809291e-05,
      "loss": 1.3821,
      "step": 18341
    },
    {
      "epoch": 0.6568660805414794,
      "grad_norm": 1.526283860206604,
      "learning_rate": 5.565045480564463e-05,
      "loss": 1.3779,
      "step": 18342
    },
    {
      "epoch": 0.6569018926710477,
      "grad_norm": 1.485095739364624,
      "learning_rate": 5.5640059179862314e-05,
      "loss": 1.3956,
      "step": 18343
    },
    {
      "epoch": 0.656937704800616,
      "grad_norm": 1.6767833232879639,
      "learning_rate": 5.562966415088574e-05,
      "loss": 1.6607,
      "step": 18344
    },
    {
      "epoch": 0.6569735169301842,
      "grad_norm": 1.563358187675476,
      "learning_rate": 5.5619269718854805e-05,
      "loss": 1.8258,
      "step": 18345
    },
    {
      "epoch": 0.6570093290597525,
      "grad_norm": 1.4730910062789917,
      "learning_rate": 5.560887588390938e-05,
      "loss": 1.1704,
      "step": 18346
    },
    {
      "epoch": 0.6570451411893208,
      "grad_norm": 1.5037370920181274,
      "learning_rate": 5.559848264618923e-05,
      "loss": 1.1723,
      "step": 18347
    },
    {
      "epoch": 0.6570809533188892,
      "grad_norm": 1.9317376613616943,
      "learning_rate": 5.5588090005834224e-05,
      "loss": 1.3406,
      "step": 18348
    },
    {
      "epoch": 0.6571167654484574,
      "grad_norm": 1.5945675373077393,
      "learning_rate": 5.5577697962984195e-05,
      "loss": 1.4684,
      "step": 18349
    },
    {
      "epoch": 0.6571525775780257,
      "grad_norm": 1.4103537797927856,
      "learning_rate": 5.556730651777897e-05,
      "loss": 1.3841,
      "step": 18350
    },
    {
      "epoch": 0.657188389707594,
      "grad_norm": 1.6418267488479614,
      "learning_rate": 5.555691567035828e-05,
      "loss": 1.4893,
      "step": 18351
    },
    {
      "epoch": 0.6572242018371622,
      "grad_norm": 1.3882322311401367,
      "learning_rate": 5.554652542086196e-05,
      "loss": 1.4669,
      "step": 18352
    },
    {
      "epoch": 0.6572600139667305,
      "grad_norm": 1.9791117906570435,
      "learning_rate": 5.5536135769429795e-05,
      "loss": 1.3963,
      "step": 18353
    },
    {
      "epoch": 0.6572958260962988,
      "grad_norm": 2.5772440433502197,
      "learning_rate": 5.552574671620161e-05,
      "loss": 1.4643,
      "step": 18354
    },
    {
      "epoch": 0.6573316382258672,
      "grad_norm": 2.2931079864501953,
      "learning_rate": 5.551535826131711e-05,
      "loss": 1.4,
      "step": 18355
    },
    {
      "epoch": 0.6573674503554354,
      "grad_norm": 2.633115768432617,
      "learning_rate": 5.5504970404916066e-05,
      "loss": 1.1993,
      "step": 18356
    },
    {
      "epoch": 0.6574032624850037,
      "grad_norm": 1.4638112783432007,
      "learning_rate": 5.54945831471383e-05,
      "loss": 1.5759,
      "step": 18357
    },
    {
      "epoch": 0.657439074614572,
      "grad_norm": 1.9851884841918945,
      "learning_rate": 5.548419648812346e-05,
      "loss": 1.5184,
      "step": 18358
    },
    {
      "epoch": 0.6574748867441402,
      "grad_norm": 1.440222978591919,
      "learning_rate": 5.547381042801135e-05,
      "loss": 1.5711,
      "step": 18359
    },
    {
      "epoch": 0.6575106988737085,
      "grad_norm": 1.8197391033172607,
      "learning_rate": 5.5463424966941676e-05,
      "loss": 1.0896,
      "step": 18360
    },
    {
      "epoch": 0.6575465110032768,
      "grad_norm": 1.7370043992996216,
      "learning_rate": 5.545304010505421e-05,
      "loss": 1.5494,
      "step": 18361
    },
    {
      "epoch": 0.6575823231328451,
      "grad_norm": 1.4990167617797852,
      "learning_rate": 5.54426558424886e-05,
      "loss": 1.5032,
      "step": 18362
    },
    {
      "epoch": 0.6576181352624134,
      "grad_norm": 2.4379074573516846,
      "learning_rate": 5.543227217938457e-05,
      "loss": 1.5814,
      "step": 18363
    },
    {
      "epoch": 0.6576539473919817,
      "grad_norm": 1.6144640445709229,
      "learning_rate": 5.5421889115881875e-05,
      "loss": 1.3925,
      "step": 18364
    },
    {
      "epoch": 0.65768975952155,
      "grad_norm": 1.5260868072509766,
      "learning_rate": 5.5411506652120115e-05,
      "loss": 1.5798,
      "step": 18365
    },
    {
      "epoch": 0.6577255716511182,
      "grad_norm": 1.313092589378357,
      "learning_rate": 5.540112478823902e-05,
      "loss": 1.3203,
      "step": 18366
    },
    {
      "epoch": 0.6577613837806865,
      "grad_norm": 1.5859086513519287,
      "learning_rate": 5.5390743524378266e-05,
      "loss": 1.5731,
      "step": 18367
    },
    {
      "epoch": 0.6577971959102548,
      "grad_norm": 1.383134365081787,
      "learning_rate": 5.538036286067756e-05,
      "loss": 1.347,
      "step": 18368
    },
    {
      "epoch": 0.6578330080398231,
      "grad_norm": 1.9735016822814941,
      "learning_rate": 5.5369982797276454e-05,
      "loss": 1.6849,
      "step": 18369
    },
    {
      "epoch": 0.6578688201693914,
      "grad_norm": 1.9564305543899536,
      "learning_rate": 5.5359603334314695e-05,
      "loss": 1.2313,
      "step": 18370
    },
    {
      "epoch": 0.6579046322989597,
      "grad_norm": 1.9404182434082031,
      "learning_rate": 5.534922447193187e-05,
      "loss": 1.7799,
      "step": 18371
    },
    {
      "epoch": 0.657940444428528,
      "grad_norm": 1.4757919311523438,
      "learning_rate": 5.533884621026767e-05,
      "loss": 1.3568,
      "step": 18372
    },
    {
      "epoch": 0.6579762565580962,
      "grad_norm": 1.4914617538452148,
      "learning_rate": 5.5328468549461657e-05,
      "loss": 1.5968,
      "step": 18373
    },
    {
      "epoch": 0.6580120686876645,
      "grad_norm": 1.624997615814209,
      "learning_rate": 5.531809148965347e-05,
      "loss": 1.6594,
      "step": 18374
    },
    {
      "epoch": 0.6580478808172328,
      "grad_norm": 1.5839698314666748,
      "learning_rate": 5.530771503098278e-05,
      "loss": 1.5126,
      "step": 18375
    },
    {
      "epoch": 0.6580836929468011,
      "grad_norm": 1.6808476448059082,
      "learning_rate": 5.529733917358908e-05,
      "loss": 1.7064,
      "step": 18376
    },
    {
      "epoch": 0.6581195050763694,
      "grad_norm": 1.3805513381958008,
      "learning_rate": 5.528696391761201e-05,
      "loss": 1.4942,
      "step": 18377
    },
    {
      "epoch": 0.6581553172059377,
      "grad_norm": 1.3921979665756226,
      "learning_rate": 5.527658926319119e-05,
      "loss": 1.5189,
      "step": 18378
    },
    {
      "epoch": 0.6581911293355059,
      "grad_norm": 1.867268681526184,
      "learning_rate": 5.52662152104662e-05,
      "loss": 1.3446,
      "step": 18379
    },
    {
      "epoch": 0.6582269414650742,
      "grad_norm": 1.994785189628601,
      "learning_rate": 5.5255841759576544e-05,
      "loss": 1.253,
      "step": 18380
    },
    {
      "epoch": 0.6582627535946425,
      "grad_norm": 1.367926836013794,
      "learning_rate": 5.524546891066182e-05,
      "loss": 1.4459,
      "step": 18381
    },
    {
      "epoch": 0.6582985657242107,
      "grad_norm": 1.71571946144104,
      "learning_rate": 5.5235096663861617e-05,
      "loss": 1.8809,
      "step": 18382
    },
    {
      "epoch": 0.6583343778537791,
      "grad_norm": 1.36968195438385,
      "learning_rate": 5.5224725019315416e-05,
      "loss": 1.5383,
      "step": 18383
    },
    {
      "epoch": 0.6583701899833474,
      "grad_norm": 2.195829391479492,
      "learning_rate": 5.521435397716278e-05,
      "loss": 1.4893,
      "step": 18384
    },
    {
      "epoch": 0.6584060021129157,
      "grad_norm": 1.5559254884719849,
      "learning_rate": 5.520398353754324e-05,
      "loss": 1.3855,
      "step": 18385
    },
    {
      "epoch": 0.6584418142424839,
      "grad_norm": 1.5635045766830444,
      "learning_rate": 5.519361370059637e-05,
      "loss": 1.3387,
      "step": 18386
    },
    {
      "epoch": 0.6584776263720522,
      "grad_norm": 1.6653882265090942,
      "learning_rate": 5.518324446646157e-05,
      "loss": 1.7117,
      "step": 18387
    },
    {
      "epoch": 0.6585134385016205,
      "grad_norm": 1.4553028345108032,
      "learning_rate": 5.517287583527843e-05,
      "loss": 1.4408,
      "step": 18388
    },
    {
      "epoch": 0.6585492506311887,
      "grad_norm": 1.7703020572662354,
      "learning_rate": 5.51625078071864e-05,
      "loss": 1.4806,
      "step": 18389
    },
    {
      "epoch": 0.6585850627607571,
      "grad_norm": 1.610481858253479,
      "learning_rate": 5.5152140382325044e-05,
      "loss": 1.4433,
      "step": 18390
    },
    {
      "epoch": 0.6586208748903254,
      "grad_norm": 1.5309536457061768,
      "learning_rate": 5.5141773560833756e-05,
      "loss": 1.7228,
      "step": 18391
    },
    {
      "epoch": 0.6586566870198937,
      "grad_norm": 1.8306372165679932,
      "learning_rate": 5.5131407342852026e-05,
      "loss": 1.1958,
      "step": 18392
    },
    {
      "epoch": 0.6586924991494619,
      "grad_norm": 1.701582431793213,
      "learning_rate": 5.5121041728519386e-05,
      "loss": 1.4244,
      "step": 18393
    },
    {
      "epoch": 0.6587283112790302,
      "grad_norm": 1.6949228048324585,
      "learning_rate": 5.5110676717975194e-05,
      "loss": 1.3162,
      "step": 18394
    },
    {
      "epoch": 0.6587641234085985,
      "grad_norm": 1.2450233697891235,
      "learning_rate": 5.510031231135895e-05,
      "loss": 1.482,
      "step": 18395
    },
    {
      "epoch": 0.6587999355381667,
      "grad_norm": 1.6780352592468262,
      "learning_rate": 5.508994850881008e-05,
      "loss": 1.4635,
      "step": 18396
    },
    {
      "epoch": 0.6588357476677351,
      "grad_norm": 2.5002284049987793,
      "learning_rate": 5.507958531046806e-05,
      "loss": 1.4665,
      "step": 18397
    },
    {
      "epoch": 0.6588715597973034,
      "grad_norm": 1.386528730392456,
      "learning_rate": 5.506922271647228e-05,
      "loss": 1.7111,
      "step": 18398
    },
    {
      "epoch": 0.6589073719268717,
      "grad_norm": 1.9130560159683228,
      "learning_rate": 5.505886072696208e-05,
      "loss": 1.3824,
      "step": 18399
    },
    {
      "epoch": 0.6589431840564399,
      "grad_norm": 1.4952514171600342,
      "learning_rate": 5.504849934207701e-05,
      "loss": 1.4765,
      "step": 18400
    },
    {
      "epoch": 0.6589789961860082,
      "grad_norm": 1.4171199798583984,
      "learning_rate": 5.503813856195637e-05,
      "loss": 1.3389,
      "step": 18401
    },
    {
      "epoch": 0.6590148083155765,
      "grad_norm": 1.6343742609024048,
      "learning_rate": 5.5027778386739606e-05,
      "loss": 1.488,
      "step": 18402
    },
    {
      "epoch": 0.6590506204451447,
      "grad_norm": 1.364643931388855,
      "learning_rate": 5.5017418816565994e-05,
      "loss": 1.5663,
      "step": 18403
    },
    {
      "epoch": 0.6590864325747131,
      "grad_norm": 2.7921016216278076,
      "learning_rate": 5.500705985157508e-05,
      "loss": 1.5698,
      "step": 18404
    },
    {
      "epoch": 0.6591222447042814,
      "grad_norm": 1.4503824710845947,
      "learning_rate": 5.499670149190609e-05,
      "loss": 1.3045,
      "step": 18405
    },
    {
      "epoch": 0.6591580568338496,
      "grad_norm": 1.6705472469329834,
      "learning_rate": 5.498634373769843e-05,
      "loss": 1.6779,
      "step": 18406
    },
    {
      "epoch": 0.6591938689634179,
      "grad_norm": 1.5752532482147217,
      "learning_rate": 5.497598658909149e-05,
      "loss": 1.4986,
      "step": 18407
    },
    {
      "epoch": 0.6592296810929862,
      "grad_norm": 1.343363642692566,
      "learning_rate": 5.496563004622455e-05,
      "loss": 1.4247,
      "step": 18408
    },
    {
      "epoch": 0.6592654932225545,
      "grad_norm": 1.5922566652297974,
      "learning_rate": 5.495527410923699e-05,
      "loss": 1.5302,
      "step": 18409
    },
    {
      "epoch": 0.6593013053521227,
      "grad_norm": 1.7536109685897827,
      "learning_rate": 5.494491877826804e-05,
      "loss": 1.1251,
      "step": 18410
    },
    {
      "epoch": 0.6593371174816911,
      "grad_norm": 1.656254768371582,
      "learning_rate": 5.493456405345716e-05,
      "loss": 1.4341,
      "step": 18411
    },
    {
      "epoch": 0.6593729296112594,
      "grad_norm": 1.5095834732055664,
      "learning_rate": 5.492420993494357e-05,
      "loss": 1.4502,
      "step": 18412
    },
    {
      "epoch": 0.6594087417408276,
      "grad_norm": 1.864039421081543,
      "learning_rate": 5.491385642286662e-05,
      "loss": 1.2983,
      "step": 18413
    },
    {
      "epoch": 0.6594445538703959,
      "grad_norm": 1.595781922340393,
      "learning_rate": 5.49035035173655e-05,
      "loss": 1.332,
      "step": 18414
    },
    {
      "epoch": 0.6594803659999642,
      "grad_norm": 1.7632122039794922,
      "learning_rate": 5.4893151218579655e-05,
      "loss": 1.5275,
      "step": 18415
    },
    {
      "epoch": 0.6595161781295324,
      "grad_norm": 1.8122587203979492,
      "learning_rate": 5.488279952664826e-05,
      "loss": 1.4587,
      "step": 18416
    },
    {
      "epoch": 0.6595519902591007,
      "grad_norm": 2.209500789642334,
      "learning_rate": 5.4872448441710536e-05,
      "loss": 1.665,
      "step": 18417
    },
    {
      "epoch": 0.659587802388669,
      "grad_norm": 1.7916383743286133,
      "learning_rate": 5.4862097963905865e-05,
      "loss": 1.6607,
      "step": 18418
    },
    {
      "epoch": 0.6596236145182374,
      "grad_norm": 1.0907057523727417,
      "learning_rate": 5.485174809337342e-05,
      "loss": 1.2834,
      "step": 18419
    },
    {
      "epoch": 0.6596594266478056,
      "grad_norm": 1.6302517652511597,
      "learning_rate": 5.484139883025251e-05,
      "loss": 1.4004,
      "step": 18420
    },
    {
      "epoch": 0.6596952387773739,
      "grad_norm": 1.5323625802993774,
      "learning_rate": 5.4831050174682243e-05,
      "loss": 1.5594,
      "step": 18421
    },
    {
      "epoch": 0.6597310509069422,
      "grad_norm": 1.9435088634490967,
      "learning_rate": 5.482070212680201e-05,
      "loss": 1.5246,
      "step": 18422
    },
    {
      "epoch": 0.6597668630365104,
      "grad_norm": 2.1833503246307373,
      "learning_rate": 5.481035468675092e-05,
      "loss": 1.4532,
      "step": 18423
    },
    {
      "epoch": 0.6598026751660787,
      "grad_norm": 1.534131646156311,
      "learning_rate": 5.4800007854668254e-05,
      "loss": 1.3368,
      "step": 18424
    },
    {
      "epoch": 0.659838487295647,
      "grad_norm": 2.482542037963867,
      "learning_rate": 5.478966163069313e-05,
      "loss": 1.5933,
      "step": 18425
    },
    {
      "epoch": 0.6598742994252154,
      "grad_norm": 1.8399754762649536,
      "learning_rate": 5.47793160149648e-05,
      "loss": 1.5144,
      "step": 18426
    },
    {
      "epoch": 0.6599101115547836,
      "grad_norm": 1.5900001525878906,
      "learning_rate": 5.476897100762248e-05,
      "loss": 1.6037,
      "step": 18427
    },
    {
      "epoch": 0.6599459236843519,
      "grad_norm": 1.6989408731460571,
      "learning_rate": 5.475862660880529e-05,
      "loss": 1.1131,
      "step": 18428
    },
    {
      "epoch": 0.6599817358139202,
      "grad_norm": 1.6499292850494385,
      "learning_rate": 5.4748282818652386e-05,
      "loss": 1.4069,
      "step": 18429
    },
    {
      "epoch": 0.6600175479434884,
      "grad_norm": 1.6214898824691772,
      "learning_rate": 5.473793963730299e-05,
      "loss": 1.3842,
      "step": 18430
    },
    {
      "epoch": 0.6600533600730567,
      "grad_norm": 1.480055332183838,
      "learning_rate": 5.4727597064896276e-05,
      "loss": 1.4681,
      "step": 18431
    },
    {
      "epoch": 0.660089172202625,
      "grad_norm": 1.5914380550384521,
      "learning_rate": 5.4717255101571253e-05,
      "loss": 1.3343,
      "step": 18432
    },
    {
      "epoch": 0.6601249843321934,
      "grad_norm": 1.7542521953582764,
      "learning_rate": 5.470691374746724e-05,
      "loss": 1.5553,
      "step": 18433
    },
    {
      "epoch": 0.6601607964617616,
      "grad_norm": 2.195375442504883,
      "learning_rate": 5.469657300272326e-05,
      "loss": 1.4683,
      "step": 18434
    },
    {
      "epoch": 0.6601966085913299,
      "grad_norm": 1.4942426681518555,
      "learning_rate": 5.468623286747844e-05,
      "loss": 1.2783,
      "step": 18435
    },
    {
      "epoch": 0.6602324207208982,
      "grad_norm": 1.32637619972229,
      "learning_rate": 5.4675893341871886e-05,
      "loss": 1.4105,
      "step": 18436
    },
    {
      "epoch": 0.6602682328504664,
      "grad_norm": 1.3210185766220093,
      "learning_rate": 5.4665554426042734e-05,
      "loss": 1.0158,
      "step": 18437
    },
    {
      "epoch": 0.6603040449800347,
      "grad_norm": 1.8867077827453613,
      "learning_rate": 5.465521612013012e-05,
      "loss": 1.3697,
      "step": 18438
    },
    {
      "epoch": 0.660339857109603,
      "grad_norm": 1.9586296081542969,
      "learning_rate": 5.464487842427302e-05,
      "loss": 1.6983,
      "step": 18439
    },
    {
      "epoch": 0.6603756692391713,
      "grad_norm": 1.508347511291504,
      "learning_rate": 5.463454133861059e-05,
      "loss": 1.2857,
      "step": 18440
    },
    {
      "epoch": 0.6604114813687396,
      "grad_norm": 1.4944887161254883,
      "learning_rate": 5.462420486328188e-05,
      "loss": 1.3681,
      "step": 18441
    },
    {
      "epoch": 0.6604472934983079,
      "grad_norm": 1.4582613706588745,
      "learning_rate": 5.461386899842601e-05,
      "loss": 1.1148,
      "step": 18442
    },
    {
      "epoch": 0.6604831056278762,
      "grad_norm": 1.7177790403366089,
      "learning_rate": 5.460353374418195e-05,
      "loss": 1.6026,
      "step": 18443
    },
    {
      "epoch": 0.6605189177574444,
      "grad_norm": 1.6896547079086304,
      "learning_rate": 5.459319910068879e-05,
      "loss": 1.4609,
      "step": 18444
    },
    {
      "epoch": 0.6605547298870127,
      "grad_norm": 1.8076750040054321,
      "learning_rate": 5.4582865068085585e-05,
      "loss": 1.537,
      "step": 18445
    },
    {
      "epoch": 0.660590542016581,
      "grad_norm": 1.894033670425415,
      "learning_rate": 5.4572531646511325e-05,
      "loss": 1.5241,
      "step": 18446
    },
    {
      "epoch": 0.6606263541461493,
      "grad_norm": 1.7730125188827515,
      "learning_rate": 5.456219883610505e-05,
      "loss": 1.5734,
      "step": 18447
    },
    {
      "epoch": 0.6606621662757176,
      "grad_norm": 1.4446651935577393,
      "learning_rate": 5.455186663700578e-05,
      "loss": 1.6482,
      "step": 18448
    },
    {
      "epoch": 0.6606979784052859,
      "grad_norm": 1.781532883644104,
      "learning_rate": 5.4541535049352566e-05,
      "loss": 1.6462,
      "step": 18449
    },
    {
      "epoch": 0.6607337905348541,
      "grad_norm": 1.8322112560272217,
      "learning_rate": 5.4531204073284316e-05,
      "loss": 1.7429,
      "step": 18450
    },
    {
      "epoch": 0.6607696026644224,
      "grad_norm": 1.6175395250320435,
      "learning_rate": 5.4520873708940056e-05,
      "loss": 1.3589,
      "step": 18451
    },
    {
      "epoch": 0.6608054147939907,
      "grad_norm": 1.7501140832901,
      "learning_rate": 5.451054395645883e-05,
      "loss": 1.6793,
      "step": 18452
    },
    {
      "epoch": 0.660841226923559,
      "grad_norm": 1.5763481855392456,
      "learning_rate": 5.450021481597951e-05,
      "loss": 1.3531,
      "step": 18453
    },
    {
      "epoch": 0.6608770390531273,
      "grad_norm": 1.719218134880066,
      "learning_rate": 5.448988628764111e-05,
      "loss": 1.3094,
      "step": 18454
    },
    {
      "epoch": 0.6609128511826956,
      "grad_norm": 1.532935619354248,
      "learning_rate": 5.4479558371582584e-05,
      "loss": 1.3702,
      "step": 18455
    },
    {
      "epoch": 0.6609486633122639,
      "grad_norm": 1.7714307308197021,
      "learning_rate": 5.446923106794293e-05,
      "loss": 1.5029,
      "step": 18456
    },
    {
      "epoch": 0.6609844754418321,
      "grad_norm": 1.7432639598846436,
      "learning_rate": 5.4458904376860997e-05,
      "loss": 1.5302,
      "step": 18457
    },
    {
      "epoch": 0.6610202875714004,
      "grad_norm": 1.608049750328064,
      "learning_rate": 5.444857829847576e-05,
      "loss": 1.1499,
      "step": 18458
    },
    {
      "epoch": 0.6610560997009687,
      "grad_norm": 1.6713452339172363,
      "learning_rate": 5.443825283292615e-05,
      "loss": 1.4979,
      "step": 18459
    },
    {
      "epoch": 0.661091911830537,
      "grad_norm": 1.5441710948944092,
      "learning_rate": 5.4427927980351124e-05,
      "loss": 1.309,
      "step": 18460
    },
    {
      "epoch": 0.6611277239601053,
      "grad_norm": 1.8041974306106567,
      "learning_rate": 5.441760374088949e-05,
      "loss": 1.2219,
      "step": 18461
    },
    {
      "epoch": 0.6611635360896736,
      "grad_norm": 1.397838830947876,
      "learning_rate": 5.4407280114680206e-05,
      "loss": 1.2665,
      "step": 18462
    },
    {
      "epoch": 0.6611993482192419,
      "grad_norm": 1.60728919506073,
      "learning_rate": 5.439695710186219e-05,
      "loss": 1.5308,
      "step": 18463
    },
    {
      "epoch": 0.6612351603488101,
      "grad_norm": 1.4930665493011475,
      "learning_rate": 5.4386634702574255e-05,
      "loss": 1.5292,
      "step": 18464
    },
    {
      "epoch": 0.6612709724783784,
      "grad_norm": 1.4357012510299683,
      "learning_rate": 5.437631291695533e-05,
      "loss": 1.4713,
      "step": 18465
    },
    {
      "epoch": 0.6613067846079467,
      "grad_norm": 1.4895292520523071,
      "learning_rate": 5.436599174514425e-05,
      "loss": 1.5546,
      "step": 18466
    },
    {
      "epoch": 0.6613425967375149,
      "grad_norm": 1.6815274953842163,
      "learning_rate": 5.435567118727993e-05,
      "loss": 1.2031,
      "step": 18467
    },
    {
      "epoch": 0.6613784088670833,
      "grad_norm": 1.4173117876052856,
      "learning_rate": 5.434535124350113e-05,
      "loss": 1.6375,
      "step": 18468
    },
    {
      "epoch": 0.6614142209966516,
      "grad_norm": 1.7272834777832031,
      "learning_rate": 5.433503191394675e-05,
      "loss": 1.4441,
      "step": 18469
    },
    {
      "epoch": 0.6614500331262199,
      "grad_norm": 1.6564527750015259,
      "learning_rate": 5.432471319875565e-05,
      "loss": 1.4019,
      "step": 18470
    },
    {
      "epoch": 0.6614858452557881,
      "grad_norm": 1.524398922920227,
      "learning_rate": 5.431439509806657e-05,
      "loss": 1.3863,
      "step": 18471
    },
    {
      "epoch": 0.6615216573853564,
      "grad_norm": 1.881738543510437,
      "learning_rate": 5.4304077612018375e-05,
      "loss": 1.5866,
      "step": 18472
    },
    {
      "epoch": 0.6615574695149247,
      "grad_norm": 1.785723328590393,
      "learning_rate": 5.429376074074988e-05,
      "loss": 1.595,
      "step": 18473
    },
    {
      "epoch": 0.6615932816444929,
      "grad_norm": 1.6554456949234009,
      "learning_rate": 5.4283444484399904e-05,
      "loss": 1.2719,
      "step": 18474
    },
    {
      "epoch": 0.6616290937740613,
      "grad_norm": 2.057368516921997,
      "learning_rate": 5.427312884310718e-05,
      "loss": 1.5038,
      "step": 18475
    },
    {
      "epoch": 0.6616649059036296,
      "grad_norm": 1.655664086341858,
      "learning_rate": 5.426281381701053e-05,
      "loss": 1.4807,
      "step": 18476
    },
    {
      "epoch": 0.6617007180331979,
      "grad_norm": 1.4010266065597534,
      "learning_rate": 5.4252499406248724e-05,
      "loss": 1.34,
      "step": 18477
    },
    {
      "epoch": 0.6617365301627661,
      "grad_norm": 1.8840229511260986,
      "learning_rate": 5.424218561096055e-05,
      "loss": 1.7182,
      "step": 18478
    },
    {
      "epoch": 0.6617723422923344,
      "grad_norm": 1.9274563789367676,
      "learning_rate": 5.423187243128472e-05,
      "loss": 1.5018,
      "step": 18479
    },
    {
      "epoch": 0.6618081544219027,
      "grad_norm": 1.7007853984832764,
      "learning_rate": 5.4221559867360014e-05,
      "loss": 1.5183,
      "step": 18480
    },
    {
      "epoch": 0.6618439665514709,
      "grad_norm": 1.4750630855560303,
      "learning_rate": 5.4211247919325206e-05,
      "loss": 1.5261,
      "step": 18481
    },
    {
      "epoch": 0.6618797786810393,
      "grad_norm": 1.4993181228637695,
      "learning_rate": 5.4200936587318954e-05,
      "loss": 1.4829,
      "step": 18482
    },
    {
      "epoch": 0.6619155908106076,
      "grad_norm": 2.4123010635375977,
      "learning_rate": 5.4190625871480016e-05,
      "loss": 1.646,
      "step": 18483
    },
    {
      "epoch": 0.6619514029401758,
      "grad_norm": 1.5104199647903442,
      "learning_rate": 5.4180315771947123e-05,
      "loss": 1.4037,
      "step": 18484
    },
    {
      "epoch": 0.6619872150697441,
      "grad_norm": 1.7275927066802979,
      "learning_rate": 5.417000628885902e-05,
      "loss": 1.6069,
      "step": 18485
    },
    {
      "epoch": 0.6620230271993124,
      "grad_norm": 1.8526519536972046,
      "learning_rate": 5.415969742235432e-05,
      "loss": 1.7501,
      "step": 18486
    },
    {
      "epoch": 0.6620588393288807,
      "grad_norm": 2.1937081813812256,
      "learning_rate": 5.414938917257177e-05,
      "loss": 1.4062,
      "step": 18487
    },
    {
      "epoch": 0.6620946514584489,
      "grad_norm": 1.4393888711929321,
      "learning_rate": 5.4139081539650084e-05,
      "loss": 1.3466,
      "step": 18488
    },
    {
      "epoch": 0.6621304635880173,
      "grad_norm": 1.2877326011657715,
      "learning_rate": 5.412877452372784e-05,
      "loss": 1.4105,
      "step": 18489
    },
    {
      "epoch": 0.6621662757175856,
      "grad_norm": 1.5457091331481934,
      "learning_rate": 5.411846812494379e-05,
      "loss": 1.4512,
      "step": 18490
    },
    {
      "epoch": 0.6622020878471538,
      "grad_norm": 2.318279266357422,
      "learning_rate": 5.410816234343656e-05,
      "loss": 1.4582,
      "step": 18491
    },
    {
      "epoch": 0.6622378999767221,
      "grad_norm": 2.045260429382324,
      "learning_rate": 5.4097857179344846e-05,
      "loss": 1.4889,
      "step": 18492
    },
    {
      "epoch": 0.6622737121062904,
      "grad_norm": 2.0480244159698486,
      "learning_rate": 5.4087552632807225e-05,
      "loss": 1.3724,
      "step": 18493
    },
    {
      "epoch": 0.6623095242358586,
      "grad_norm": 1.5088456869125366,
      "learning_rate": 5.407724870396235e-05,
      "loss": 1.4677,
      "step": 18494
    },
    {
      "epoch": 0.6623453363654269,
      "grad_norm": 1.4615206718444824,
      "learning_rate": 5.4066945392948896e-05,
      "loss": 1.0757,
      "step": 18495
    },
    {
      "epoch": 0.6623811484949953,
      "grad_norm": 1.892917275428772,
      "learning_rate": 5.40566426999054e-05,
      "loss": 1.5292,
      "step": 18496
    },
    {
      "epoch": 0.6624169606245636,
      "grad_norm": 1.6644346714019775,
      "learning_rate": 5.404634062497057e-05,
      "loss": 1.5624,
      "step": 18497
    },
    {
      "epoch": 0.6624527727541318,
      "grad_norm": 2.008518695831299,
      "learning_rate": 5.403603916828286e-05,
      "loss": 1.6561,
      "step": 18498
    },
    {
      "epoch": 0.6624885848837001,
      "grad_norm": 1.7898550033569336,
      "learning_rate": 5.4025738329981035e-05,
      "loss": 1.4068,
      "step": 18499
    },
    {
      "epoch": 0.6625243970132684,
      "grad_norm": 1.7520909309387207,
      "learning_rate": 5.401543811020356e-05,
      "loss": 1.2904,
      "step": 18500
    },
    {
      "epoch": 0.6625602091428366,
      "grad_norm": 1.633162021636963,
      "learning_rate": 5.400513850908905e-05,
      "loss": 1.3964,
      "step": 18501
    },
    {
      "epoch": 0.6625960212724049,
      "grad_norm": 2.3457908630371094,
      "learning_rate": 5.3994839526776065e-05,
      "loss": 1.8168,
      "step": 18502
    },
    {
      "epoch": 0.6626318334019733,
      "grad_norm": 1.3527460098266602,
      "learning_rate": 5.398454116340322e-05,
      "loss": 1.3832,
      "step": 18503
    },
    {
      "epoch": 0.6626676455315416,
      "grad_norm": 1.643717646598816,
      "learning_rate": 5.3974243419109016e-05,
      "loss": 1.1241,
      "step": 18504
    },
    {
      "epoch": 0.6627034576611098,
      "grad_norm": 1.5871790647506714,
      "learning_rate": 5.396394629403192e-05,
      "loss": 1.5845,
      "step": 18505
    },
    {
      "epoch": 0.6627392697906781,
      "grad_norm": 1.3667486906051636,
      "learning_rate": 5.395364978831061e-05,
      "loss": 1.5827,
      "step": 18506
    },
    {
      "epoch": 0.6627750819202464,
      "grad_norm": 1.8814173936843872,
      "learning_rate": 5.394335390208352e-05,
      "loss": 1.1862,
      "step": 18507
    },
    {
      "epoch": 0.6628108940498146,
      "grad_norm": 1.5491501092910767,
      "learning_rate": 5.393305863548924e-05,
      "loss": 1.4095,
      "step": 18508
    },
    {
      "epoch": 0.6628467061793829,
      "grad_norm": 1.667093276977539,
      "learning_rate": 5.392276398866615e-05,
      "loss": 1.4577,
      "step": 18509
    },
    {
      "epoch": 0.6628825183089513,
      "grad_norm": 1.9760993719100952,
      "learning_rate": 5.391246996175291e-05,
      "loss": 1.5,
      "step": 18510
    },
    {
      "epoch": 0.6629183304385196,
      "grad_norm": 2.045569658279419,
      "learning_rate": 5.39021765548879e-05,
      "loss": 1.6401,
      "step": 18511
    },
    {
      "epoch": 0.6629541425680878,
      "grad_norm": 1.5511012077331543,
      "learning_rate": 5.3891883768209686e-05,
      "loss": 1.3945,
      "step": 18512
    },
    {
      "epoch": 0.6629899546976561,
      "grad_norm": 1.8980215787887573,
      "learning_rate": 5.388159160185665e-05,
      "loss": 1.3399,
      "step": 18513
    },
    {
      "epoch": 0.6630257668272244,
      "grad_norm": 1.6040958166122437,
      "learning_rate": 5.387130005596732e-05,
      "loss": 1.5281,
      "step": 18514
    },
    {
      "epoch": 0.6630615789567926,
      "grad_norm": 2.547346830368042,
      "learning_rate": 5.386100913068017e-05,
      "loss": 1.5416,
      "step": 18515
    },
    {
      "epoch": 0.6630973910863609,
      "grad_norm": 1.6553081274032593,
      "learning_rate": 5.385071882613357e-05,
      "loss": 1.2632,
      "step": 18516
    },
    {
      "epoch": 0.6631332032159293,
      "grad_norm": 1.6644037961959839,
      "learning_rate": 5.3840429142466096e-05,
      "loss": 1.786,
      "step": 18517
    },
    {
      "epoch": 0.6631690153454975,
      "grad_norm": 1.7519235610961914,
      "learning_rate": 5.383014007981606e-05,
      "loss": 1.5851,
      "step": 18518
    },
    {
      "epoch": 0.6632048274750658,
      "grad_norm": 1.6565033197402954,
      "learning_rate": 5.381985163832197e-05,
      "loss": 1.5017,
      "step": 18519
    },
    {
      "epoch": 0.6632406396046341,
      "grad_norm": 1.5082019567489624,
      "learning_rate": 5.380956381812213e-05,
      "loss": 1.5402,
      "step": 18520
    },
    {
      "epoch": 0.6632764517342024,
      "grad_norm": 1.5436220169067383,
      "learning_rate": 5.379927661935511e-05,
      "loss": 1.2787,
      "step": 18521
    },
    {
      "epoch": 0.6633122638637706,
      "grad_norm": 1.599420189857483,
      "learning_rate": 5.3788990042159224e-05,
      "loss": 1.3577,
      "step": 18522
    },
    {
      "epoch": 0.6633480759933389,
      "grad_norm": 1.4425898790359497,
      "learning_rate": 5.377870408667285e-05,
      "loss": 1.3236,
      "step": 18523
    },
    {
      "epoch": 0.6633838881229073,
      "grad_norm": 1.4394543170928955,
      "learning_rate": 5.3768418753034375e-05,
      "loss": 1.5358,
      "step": 18524
    },
    {
      "epoch": 0.6634197002524755,
      "grad_norm": 1.8449910879135132,
      "learning_rate": 5.375813404138219e-05,
      "loss": 1.5375,
      "step": 18525
    },
    {
      "epoch": 0.6634555123820438,
      "grad_norm": 1.8355271816253662,
      "learning_rate": 5.37478499518547e-05,
      "loss": 1.309,
      "step": 18526
    },
    {
      "epoch": 0.6634913245116121,
      "grad_norm": 1.466902256011963,
      "learning_rate": 5.3737566484590164e-05,
      "loss": 1.3721,
      "step": 18527
    },
    {
      "epoch": 0.6635271366411803,
      "grad_norm": 1.5819684267044067,
      "learning_rate": 5.372728363972706e-05,
      "loss": 1.4732,
      "step": 18528
    },
    {
      "epoch": 0.6635629487707486,
      "grad_norm": 1.633408546447754,
      "learning_rate": 5.371700141740364e-05,
      "loss": 1.9911,
      "step": 18529
    },
    {
      "epoch": 0.6635987609003169,
      "grad_norm": 1.6528607606887817,
      "learning_rate": 5.3706719817758286e-05,
      "loss": 1.5908,
      "step": 18530
    },
    {
      "epoch": 0.6636345730298853,
      "grad_norm": 2.3122167587280273,
      "learning_rate": 5.3696438840929276e-05,
      "loss": 1.552,
      "step": 18531
    },
    {
      "epoch": 0.6636703851594535,
      "grad_norm": 1.6594113111495972,
      "learning_rate": 5.368615848705496e-05,
      "loss": 1.7639,
      "step": 18532
    },
    {
      "epoch": 0.6637061972890218,
      "grad_norm": 1.49611496925354,
      "learning_rate": 5.367587875627367e-05,
      "loss": 1.2976,
      "step": 18533
    },
    {
      "epoch": 0.6637420094185901,
      "grad_norm": 1.8043385744094849,
      "learning_rate": 5.366559964872364e-05,
      "loss": 1.3638,
      "step": 18534
    },
    {
      "epoch": 0.6637778215481583,
      "grad_norm": 1.6629353761672974,
      "learning_rate": 5.36553211645432e-05,
      "loss": 1.504,
      "step": 18535
    },
    {
      "epoch": 0.6638136336777266,
      "grad_norm": 2.4057998657226562,
      "learning_rate": 5.3645043303870634e-05,
      "loss": 1.3973,
      "step": 18536
    },
    {
      "epoch": 0.6638494458072949,
      "grad_norm": 1.4145482778549194,
      "learning_rate": 5.363476606684425e-05,
      "loss": 1.411,
      "step": 18537
    },
    {
      "epoch": 0.6638852579368633,
      "grad_norm": 1.9292107820510864,
      "learning_rate": 5.3624489453602255e-05,
      "loss": 1.4168,
      "step": 18538
    },
    {
      "epoch": 0.6639210700664315,
      "grad_norm": 2.106179714202881,
      "learning_rate": 5.361421346428294e-05,
      "loss": 1.4649,
      "step": 18539
    },
    {
      "epoch": 0.6639568821959998,
      "grad_norm": 1.8230222463607788,
      "learning_rate": 5.3603938099024576e-05,
      "loss": 1.3864,
      "step": 18540
    },
    {
      "epoch": 0.6639926943255681,
      "grad_norm": 1.3416988849639893,
      "learning_rate": 5.359366335796534e-05,
      "loss": 1.3887,
      "step": 18541
    },
    {
      "epoch": 0.6640285064551363,
      "grad_norm": 1.5865682363510132,
      "learning_rate": 5.35833892412435e-05,
      "loss": 1.6314,
      "step": 18542
    },
    {
      "epoch": 0.6640643185847046,
      "grad_norm": 1.7829359769821167,
      "learning_rate": 5.3573115748997284e-05,
      "loss": 1.3659,
      "step": 18543
    },
    {
      "epoch": 0.6641001307142729,
      "grad_norm": 1.5039947032928467,
      "learning_rate": 5.356284288136496e-05,
      "loss": 1.2948,
      "step": 18544
    },
    {
      "epoch": 0.6641359428438413,
      "grad_norm": 1.5445626974105835,
      "learning_rate": 5.3552570638484644e-05,
      "loss": 1.2314,
      "step": 18545
    },
    {
      "epoch": 0.6641717549734095,
      "grad_norm": 1.75049889087677,
      "learning_rate": 5.3542299020494567e-05,
      "loss": 1.4201,
      "step": 18546
    },
    {
      "epoch": 0.6642075671029778,
      "grad_norm": 1.5456581115722656,
      "learning_rate": 5.3532028027532947e-05,
      "loss": 1.6774,
      "step": 18547
    },
    {
      "epoch": 0.6642433792325461,
      "grad_norm": 1.814924955368042,
      "learning_rate": 5.352175765973797e-05,
      "loss": 1.4873,
      "step": 18548
    },
    {
      "epoch": 0.6642791913621143,
      "grad_norm": 1.8714579343795776,
      "learning_rate": 5.351148791724776e-05,
      "loss": 1.3834,
      "step": 18549
    },
    {
      "epoch": 0.6643150034916826,
      "grad_norm": 1.5379172563552856,
      "learning_rate": 5.3501218800200514e-05,
      "loss": 1.426,
      "step": 18550
    },
    {
      "epoch": 0.6643508156212509,
      "grad_norm": 1.540379524230957,
      "learning_rate": 5.349095030873443e-05,
      "loss": 1.4127,
      "step": 18551
    },
    {
      "epoch": 0.6643866277508192,
      "grad_norm": 1.8711189031600952,
      "learning_rate": 5.348068244298758e-05,
      "loss": 1.3488,
      "step": 18552
    },
    {
      "epoch": 0.6644224398803875,
      "grad_norm": 1.440477728843689,
      "learning_rate": 5.347041520309815e-05,
      "loss": 1.506,
      "step": 18553
    },
    {
      "epoch": 0.6644582520099558,
      "grad_norm": 1.2828096151351929,
      "learning_rate": 5.346014858920425e-05,
      "loss": 1.4859,
      "step": 18554
    },
    {
      "epoch": 0.664494064139524,
      "grad_norm": 1.695265769958496,
      "learning_rate": 5.3449882601444054e-05,
      "loss": 1.2407,
      "step": 18555
    },
    {
      "epoch": 0.6645298762690923,
      "grad_norm": 2.340801239013672,
      "learning_rate": 5.343961723995561e-05,
      "loss": 1.5744,
      "step": 18556
    },
    {
      "epoch": 0.6645656883986606,
      "grad_norm": 1.5414936542510986,
      "learning_rate": 5.342935250487706e-05,
      "loss": 1.3723,
      "step": 18557
    },
    {
      "epoch": 0.6646015005282289,
      "grad_norm": 1.4003174304962158,
      "learning_rate": 5.341908839634654e-05,
      "loss": 1.6129,
      "step": 18558
    },
    {
      "epoch": 0.6646373126577972,
      "grad_norm": 1.514754295349121,
      "learning_rate": 5.340882491450205e-05,
      "loss": 1.3989,
      "step": 18559
    },
    {
      "epoch": 0.6646731247873655,
      "grad_norm": 1.9830070734024048,
      "learning_rate": 5.339856205948175e-05,
      "loss": 1.7887,
      "step": 18560
    },
    {
      "epoch": 0.6647089369169338,
      "grad_norm": 1.8465994596481323,
      "learning_rate": 5.338829983142366e-05,
      "loss": 1.9694,
      "step": 18561
    },
    {
      "epoch": 0.664744749046502,
      "grad_norm": 1.5064421892166138,
      "learning_rate": 5.337803823046592e-05,
      "loss": 1.391,
      "step": 18562
    },
    {
      "epoch": 0.6647805611760703,
      "grad_norm": 1.7226872444152832,
      "learning_rate": 5.33677772567465e-05,
      "loss": 1.4824,
      "step": 18563
    },
    {
      "epoch": 0.6648163733056386,
      "grad_norm": 2.0793983936309814,
      "learning_rate": 5.335751691040348e-05,
      "loss": 1.7713,
      "step": 18564
    },
    {
      "epoch": 0.6648521854352069,
      "grad_norm": 1.8134207725524902,
      "learning_rate": 5.334725719157492e-05,
      "loss": 1.3679,
      "step": 18565
    },
    {
      "epoch": 0.6648879975647752,
      "grad_norm": 2.0199649333953857,
      "learning_rate": 5.333699810039885e-05,
      "loss": 1.5586,
      "step": 18566
    },
    {
      "epoch": 0.6649238096943435,
      "grad_norm": 1.5236722230911255,
      "learning_rate": 5.3326739637013255e-05,
      "loss": 1.5598,
      "step": 18567
    },
    {
      "epoch": 0.6649596218239118,
      "grad_norm": 1.9306199550628662,
      "learning_rate": 5.3316481801556173e-05,
      "loss": 1.6249,
      "step": 18568
    },
    {
      "epoch": 0.66499543395348,
      "grad_norm": 1.7623952627182007,
      "learning_rate": 5.3306224594165654e-05,
      "loss": 1.5419,
      "step": 18569
    },
    {
      "epoch": 0.6650312460830483,
      "grad_norm": 1.5631310939788818,
      "learning_rate": 5.3295968014979613e-05,
      "loss": 1.6085,
      "step": 18570
    },
    {
      "epoch": 0.6650670582126166,
      "grad_norm": 1.6538745164871216,
      "learning_rate": 5.328571206413607e-05,
      "loss": 1.1281,
      "step": 18571
    },
    {
      "epoch": 0.6651028703421848,
      "grad_norm": 1.778361439704895,
      "learning_rate": 5.3275456741773025e-05,
      "loss": 1.6629,
      "step": 18572
    },
    {
      "epoch": 0.6651386824717532,
      "grad_norm": 1.872376561164856,
      "learning_rate": 5.3265202048028474e-05,
      "loss": 1.7646,
      "step": 18573
    },
    {
      "epoch": 0.6651744946013215,
      "grad_norm": 1.703460693359375,
      "learning_rate": 5.32549479830403e-05,
      "loss": 1.2561,
      "step": 18574
    },
    {
      "epoch": 0.6652103067308898,
      "grad_norm": 1.7537646293640137,
      "learning_rate": 5.324469454694651e-05,
      "loss": 1.3299,
      "step": 18575
    },
    {
      "epoch": 0.665246118860458,
      "grad_norm": 1.5586273670196533,
      "learning_rate": 5.323444173988509e-05,
      "loss": 1.397,
      "step": 18576
    },
    {
      "epoch": 0.6652819309900263,
      "grad_norm": 1.939870834350586,
      "learning_rate": 5.3224189561993886e-05,
      "loss": 1.5419,
      "step": 18577
    },
    {
      "epoch": 0.6653177431195946,
      "grad_norm": 1.8316444158554077,
      "learning_rate": 5.321393801341088e-05,
      "loss": 1.1162,
      "step": 18578
    },
    {
      "epoch": 0.6653535552491628,
      "grad_norm": 1.9167686700820923,
      "learning_rate": 5.320368709427399e-05,
      "loss": 1.6643,
      "step": 18579
    },
    {
      "epoch": 0.6653893673787312,
      "grad_norm": 1.688992977142334,
      "learning_rate": 5.3193436804721154e-05,
      "loss": 1.4744,
      "step": 18580
    },
    {
      "epoch": 0.6654251795082995,
      "grad_norm": 1.4735733270645142,
      "learning_rate": 5.318318714489021e-05,
      "loss": 1.5833,
      "step": 18581
    },
    {
      "epoch": 0.6654609916378678,
      "grad_norm": 1.6011989116668701,
      "learning_rate": 5.317293811491911e-05,
      "loss": 1.3692,
      "step": 18582
    },
    {
      "epoch": 0.665496803767436,
      "grad_norm": 1.7354061603546143,
      "learning_rate": 5.316268971494571e-05,
      "loss": 1.3453,
      "step": 18583
    },
    {
      "epoch": 0.6655326158970043,
      "grad_norm": 1.5282679796218872,
      "learning_rate": 5.315244194510795e-05,
      "loss": 1.5145,
      "step": 18584
    },
    {
      "epoch": 0.6655684280265726,
      "grad_norm": 1.8057043552398682,
      "learning_rate": 5.3142194805543625e-05,
      "loss": 1.2936,
      "step": 18585
    },
    {
      "epoch": 0.6656042401561408,
      "grad_norm": 1.4133145809173584,
      "learning_rate": 5.313194829639061e-05,
      "loss": 1.4274,
      "step": 18586
    },
    {
      "epoch": 0.6656400522857092,
      "grad_norm": 1.4651211500167847,
      "learning_rate": 5.312170241778682e-05,
      "loss": 1.5827,
      "step": 18587
    },
    {
      "epoch": 0.6656758644152775,
      "grad_norm": 1.4076982736587524,
      "learning_rate": 5.311145716987003e-05,
      "loss": 1.4423,
      "step": 18588
    },
    {
      "epoch": 0.6657116765448458,
      "grad_norm": 1.5630006790161133,
      "learning_rate": 5.310121255277809e-05,
      "loss": 1.2223,
      "step": 18589
    },
    {
      "epoch": 0.665747488674414,
      "grad_norm": 1.5476254224777222,
      "learning_rate": 5.3090968566648836e-05,
      "loss": 1.3305,
      "step": 18590
    },
    {
      "epoch": 0.6657833008039823,
      "grad_norm": 1.4601191282272339,
      "learning_rate": 5.308072521162013e-05,
      "loss": 1.3458,
      "step": 18591
    },
    {
      "epoch": 0.6658191129335506,
      "grad_norm": 1.768479347229004,
      "learning_rate": 5.307048248782975e-05,
      "loss": 1.4372,
      "step": 18592
    },
    {
      "epoch": 0.6658549250631188,
      "grad_norm": 1.4947692155838013,
      "learning_rate": 5.306024039541542e-05,
      "loss": 1.4677,
      "step": 18593
    },
    {
      "epoch": 0.6658907371926872,
      "grad_norm": 1.4802711009979248,
      "learning_rate": 5.3049998934515076e-05,
      "loss": 1.3701,
      "step": 18594
    },
    {
      "epoch": 0.6659265493222555,
      "grad_norm": 1.391737461090088,
      "learning_rate": 5.30397581052664e-05,
      "loss": 1.2477,
      "step": 18595
    },
    {
      "epoch": 0.6659623614518237,
      "grad_norm": 2.0155186653137207,
      "learning_rate": 5.302951790780725e-05,
      "loss": 1.8189,
      "step": 18596
    },
    {
      "epoch": 0.665998173581392,
      "grad_norm": 1.9427070617675781,
      "learning_rate": 5.3019278342275256e-05,
      "loss": 1.421,
      "step": 18597
    },
    {
      "epoch": 0.6660339857109603,
      "grad_norm": 1.6364455223083496,
      "learning_rate": 5.300903940880837e-05,
      "loss": 1.5302,
      "step": 18598
    },
    {
      "epoch": 0.6660697978405286,
      "grad_norm": 1.3640127182006836,
      "learning_rate": 5.299880110754418e-05,
      "loss": 1.4539,
      "step": 18599
    },
    {
      "epoch": 0.6661056099700968,
      "grad_norm": 1.8149890899658203,
      "learning_rate": 5.298856343862051e-05,
      "loss": 1.3804,
      "step": 18600
    },
    {
      "epoch": 0.6661414220996652,
      "grad_norm": 2.145496368408203,
      "learning_rate": 5.2978326402175125e-05,
      "loss": 1.3423,
      "step": 18601
    },
    {
      "epoch": 0.6661772342292335,
      "grad_norm": 1.472667932510376,
      "learning_rate": 5.296808999834565e-05,
      "loss": 1.0807,
      "step": 18602
    },
    {
      "epoch": 0.6662130463588017,
      "grad_norm": 2.075178861618042,
      "learning_rate": 5.295785422726991e-05,
      "loss": 1.5908,
      "step": 18603
    },
    {
      "epoch": 0.66624885848837,
      "grad_norm": 1.5914913415908813,
      "learning_rate": 5.2947619089085463e-05,
      "loss": 1.2944,
      "step": 18604
    },
    {
      "epoch": 0.6662846706179383,
      "grad_norm": 1.5347884893417358,
      "learning_rate": 5.2937384583930204e-05,
      "loss": 1.5397,
      "step": 18605
    },
    {
      "epoch": 0.6663204827475065,
      "grad_norm": 1.6247977018356323,
      "learning_rate": 5.2927150711941675e-05,
      "loss": 1.1967,
      "step": 18606
    },
    {
      "epoch": 0.6663562948770748,
      "grad_norm": 1.4314088821411133,
      "learning_rate": 5.2916917473257665e-05,
      "loss": 1.5123,
      "step": 18607
    },
    {
      "epoch": 0.6663921070066432,
      "grad_norm": 1.2514408826828003,
      "learning_rate": 5.2906684868015724e-05,
      "loss": 1.664,
      "step": 18608
    },
    {
      "epoch": 0.6664279191362115,
      "grad_norm": 1.1693495512008667,
      "learning_rate": 5.2896452896353656e-05,
      "loss": 1.2389,
      "step": 18609
    },
    {
      "epoch": 0.6664637312657797,
      "grad_norm": 2.151035785675049,
      "learning_rate": 5.2886221558409065e-05,
      "loss": 1.4911,
      "step": 18610
    },
    {
      "epoch": 0.666499543395348,
      "grad_norm": 1.815909743309021,
      "learning_rate": 5.287599085431951e-05,
      "loss": 1.7131,
      "step": 18611
    },
    {
      "epoch": 0.6665353555249163,
      "grad_norm": 1.4415584802627563,
      "learning_rate": 5.2865760784222786e-05,
      "loss": 1.3634,
      "step": 18612
    },
    {
      "epoch": 0.6665711676544845,
      "grad_norm": 2.0365536212921143,
      "learning_rate": 5.2855531348256424e-05,
      "loss": 1.6613,
      "step": 18613
    },
    {
      "epoch": 0.6666069797840528,
      "grad_norm": 1.3029762506484985,
      "learning_rate": 5.2845302546558105e-05,
      "loss": 1.4112,
      "step": 18614
    },
    {
      "epoch": 0.6666427919136212,
      "grad_norm": 2.367351770401001,
      "learning_rate": 5.283507437926534e-05,
      "loss": 1.5832,
      "step": 18615
    },
    {
      "epoch": 0.6666786040431895,
      "grad_norm": 1.4028692245483398,
      "learning_rate": 5.2824846846515886e-05,
      "loss": 1.381,
      "step": 18616
    },
    {
      "epoch": 0.6667144161727577,
      "grad_norm": 1.3128471374511719,
      "learning_rate": 5.281461994844723e-05,
      "loss": 1.2977,
      "step": 18617
    },
    {
      "epoch": 0.666750228302326,
      "grad_norm": 1.9322855472564697,
      "learning_rate": 5.280439368519703e-05,
      "loss": 1.5668,
      "step": 18618
    },
    {
      "epoch": 0.6667860404318943,
      "grad_norm": 2.173131227493286,
      "learning_rate": 5.27941680569028e-05,
      "loss": 1.5765,
      "step": 18619
    },
    {
      "epoch": 0.6668218525614625,
      "grad_norm": 1.7341821193695068,
      "learning_rate": 5.2783943063702155e-05,
      "loss": 1.6393,
      "step": 18620
    },
    {
      "epoch": 0.6668576646910308,
      "grad_norm": 2.0364861488342285,
      "learning_rate": 5.277371870573269e-05,
      "loss": 1.6499,
      "step": 18621
    },
    {
      "epoch": 0.6668934768205992,
      "grad_norm": 1.3587638139724731,
      "learning_rate": 5.276349498313188e-05,
      "loss": 1.3826,
      "step": 18622
    },
    {
      "epoch": 0.6669292889501675,
      "grad_norm": 1.7243382930755615,
      "learning_rate": 5.2753271896037316e-05,
      "loss": 1.3895,
      "step": 18623
    },
    {
      "epoch": 0.6669651010797357,
      "grad_norm": 1.643670916557312,
      "learning_rate": 5.274304944458652e-05,
      "loss": 1.5037,
      "step": 18624
    },
    {
      "epoch": 0.667000913209304,
      "grad_norm": 2.013648509979248,
      "learning_rate": 5.273282762891709e-05,
      "loss": 1.341,
      "step": 18625
    },
    {
      "epoch": 0.6670367253388723,
      "grad_norm": 1.607382893562317,
      "learning_rate": 5.2722606449166426e-05,
      "loss": 1.2063,
      "step": 18626
    },
    {
      "epoch": 0.6670725374684405,
      "grad_norm": 1.6168111562728882,
      "learning_rate": 5.271238590547216e-05,
      "loss": 1.5996,
      "step": 18627
    },
    {
      "epoch": 0.6671083495980088,
      "grad_norm": 1.732462763786316,
      "learning_rate": 5.270216599797176e-05,
      "loss": 1.466,
      "step": 18628
    },
    {
      "epoch": 0.6671441617275772,
      "grad_norm": 1.7212103605270386,
      "learning_rate": 5.269194672680267e-05,
      "loss": 1.5072,
      "step": 18629
    },
    {
      "epoch": 0.6671799738571454,
      "grad_norm": 1.5116984844207764,
      "learning_rate": 5.268172809210241e-05,
      "loss": 1.268,
      "step": 18630
    },
    {
      "epoch": 0.6672157859867137,
      "grad_norm": 1.6461093425750732,
      "learning_rate": 5.267151009400846e-05,
      "loss": 1.543,
      "step": 18631
    },
    {
      "epoch": 0.667251598116282,
      "grad_norm": 1.5249959230422974,
      "learning_rate": 5.266129273265834e-05,
      "loss": 1.5185,
      "step": 18632
    },
    {
      "epoch": 0.6672874102458503,
      "grad_norm": 1.881767749786377,
      "learning_rate": 5.2651076008189415e-05,
      "loss": 1.3663,
      "step": 18633
    },
    {
      "epoch": 0.6673232223754185,
      "grad_norm": 1.7331863641738892,
      "learning_rate": 5.2640859920739194e-05,
      "loss": 1.5388,
      "step": 18634
    },
    {
      "epoch": 0.6673590345049868,
      "grad_norm": 1.5982319116592407,
      "learning_rate": 5.263064447044511e-05,
      "loss": 1.6447,
      "step": 18635
    },
    {
      "epoch": 0.6673948466345552,
      "grad_norm": 1.6819671392440796,
      "learning_rate": 5.262042965744465e-05,
      "loss": 1.4766,
      "step": 18636
    },
    {
      "epoch": 0.6674306587641234,
      "grad_norm": 2.2758350372314453,
      "learning_rate": 5.261021548187515e-05,
      "loss": 1.1752,
      "step": 18637
    },
    {
      "epoch": 0.6674664708936917,
      "grad_norm": 1.5425456762313843,
      "learning_rate": 5.260000194387407e-05,
      "loss": 1.1864,
      "step": 18638
    },
    {
      "epoch": 0.66750228302326,
      "grad_norm": 1.5215123891830444,
      "learning_rate": 5.2589789043578855e-05,
      "loss": 1.2848,
      "step": 18639
    },
    {
      "epoch": 0.6675380951528282,
      "grad_norm": 1.7015901803970337,
      "learning_rate": 5.257957678112684e-05,
      "loss": 1.3467,
      "step": 18640
    },
    {
      "epoch": 0.6675739072823965,
      "grad_norm": 1.6739884614944458,
      "learning_rate": 5.2569365156655446e-05,
      "loss": 1.7775,
      "step": 18641
    },
    {
      "epoch": 0.6676097194119648,
      "grad_norm": 1.2966762781143188,
      "learning_rate": 5.255915417030206e-05,
      "loss": 1.268,
      "step": 18642
    },
    {
      "epoch": 0.6676455315415332,
      "grad_norm": 1.2526237964630127,
      "learning_rate": 5.254894382220412e-05,
      "loss": 1.3321,
      "step": 18643
    },
    {
      "epoch": 0.6676813436711014,
      "grad_norm": 1.7547239065170288,
      "learning_rate": 5.2538734112498876e-05,
      "loss": 1.5343,
      "step": 18644
    },
    {
      "epoch": 0.6677171558006697,
      "grad_norm": 1.8325891494750977,
      "learning_rate": 5.252852504132375e-05,
      "loss": 1.3332,
      "step": 18645
    },
    {
      "epoch": 0.667752967930238,
      "grad_norm": 2.117971658706665,
      "learning_rate": 5.251831660881612e-05,
      "loss": 1.454,
      "step": 18646
    },
    {
      "epoch": 0.6677887800598062,
      "grad_norm": 1.859230875968933,
      "learning_rate": 5.2508108815113264e-05,
      "loss": 1.4038,
      "step": 18647
    },
    {
      "epoch": 0.6678245921893745,
      "grad_norm": 2.172140121459961,
      "learning_rate": 5.249790166035253e-05,
      "loss": 1.1993,
      "step": 18648
    },
    {
      "epoch": 0.6678604043189428,
      "grad_norm": 1.781510353088379,
      "learning_rate": 5.2487695144671264e-05,
      "loss": 1.5604,
      "step": 18649
    },
    {
      "epoch": 0.6678962164485112,
      "grad_norm": 1.547957181930542,
      "learning_rate": 5.247748926820683e-05,
      "loss": 1.305,
      "step": 18650
    },
    {
      "epoch": 0.6679320285780794,
      "grad_norm": 1.5986849069595337,
      "learning_rate": 5.246728403109642e-05,
      "loss": 1.8007,
      "step": 18651
    },
    {
      "epoch": 0.6679678407076477,
      "grad_norm": 1.7270187139511108,
      "learning_rate": 5.245707943347738e-05,
      "loss": 1.1008,
      "step": 18652
    },
    {
      "epoch": 0.668003652837216,
      "grad_norm": 1.8409010171890259,
      "learning_rate": 5.244687547548703e-05,
      "loss": 1.206,
      "step": 18653
    },
    {
      "epoch": 0.6680394649667842,
      "grad_norm": 1.8632875680923462,
      "learning_rate": 5.243667215726267e-05,
      "loss": 1.5249,
      "step": 18654
    },
    {
      "epoch": 0.6680752770963525,
      "grad_norm": 1.8973302841186523,
      "learning_rate": 5.242646947894148e-05,
      "loss": 1.3558,
      "step": 18655
    },
    {
      "epoch": 0.6681110892259208,
      "grad_norm": 1.471661925315857,
      "learning_rate": 5.241626744066079e-05,
      "loss": 1.5711,
      "step": 18656
    },
    {
      "epoch": 0.6681469013554892,
      "grad_norm": 1.3317395448684692,
      "learning_rate": 5.240606604255787e-05,
      "loss": 1.1836,
      "step": 18657
    },
    {
      "epoch": 0.6681827134850574,
      "grad_norm": 1.742253065109253,
      "learning_rate": 5.239586528476992e-05,
      "loss": 1.5158,
      "step": 18658
    },
    {
      "epoch": 0.6682185256146257,
      "grad_norm": 1.6909421682357788,
      "learning_rate": 5.2385665167434175e-05,
      "loss": 1.2893,
      "step": 18659
    },
    {
      "epoch": 0.668254337744194,
      "grad_norm": 1.7493952512741089,
      "learning_rate": 5.2375465690687895e-05,
      "loss": 1.4646,
      "step": 18660
    },
    {
      "epoch": 0.6682901498737622,
      "grad_norm": 1.4725412130355835,
      "learning_rate": 5.236526685466834e-05,
      "loss": 1.5871,
      "step": 18661
    },
    {
      "epoch": 0.6683259620033305,
      "grad_norm": 1.6244035959243774,
      "learning_rate": 5.235506865951263e-05,
      "loss": 1.4979,
      "step": 18662
    },
    {
      "epoch": 0.6683617741328988,
      "grad_norm": 1.7314326763153076,
      "learning_rate": 5.234487110535802e-05,
      "loss": 1.364,
      "step": 18663
    },
    {
      "epoch": 0.6683975862624671,
      "grad_norm": 1.394898772239685,
      "learning_rate": 5.233467419234173e-05,
      "loss": 1.3766,
      "step": 18664
    },
    {
      "epoch": 0.6684333983920354,
      "grad_norm": 1.64047110080719,
      "learning_rate": 5.2324477920600876e-05,
      "loss": 1.3212,
      "step": 18665
    },
    {
      "epoch": 0.6684692105216037,
      "grad_norm": 1.883249044418335,
      "learning_rate": 5.231428229027269e-05,
      "loss": 1.3998,
      "step": 18666
    },
    {
      "epoch": 0.668505022651172,
      "grad_norm": 1.5769582986831665,
      "learning_rate": 5.23040873014943e-05,
      "loss": 1.3623,
      "step": 18667
    },
    {
      "epoch": 0.6685408347807402,
      "grad_norm": 1.6904163360595703,
      "learning_rate": 5.229389295440295e-05,
      "loss": 1.5583,
      "step": 18668
    },
    {
      "epoch": 0.6685766469103085,
      "grad_norm": 1.9609761238098145,
      "learning_rate": 5.228369924913567e-05,
      "loss": 1.2652,
      "step": 18669
    },
    {
      "epoch": 0.6686124590398768,
      "grad_norm": 1.8444284200668335,
      "learning_rate": 5.22735061858297e-05,
      "loss": 1.3636,
      "step": 18670
    },
    {
      "epoch": 0.6686482711694451,
      "grad_norm": 1.3868964910507202,
      "learning_rate": 5.2263313764622124e-05,
      "loss": 1.17,
      "step": 18671
    },
    {
      "epoch": 0.6686840832990134,
      "grad_norm": 1.3549922704696655,
      "learning_rate": 5.225312198565013e-05,
      "loss": 1.4785,
      "step": 18672
    },
    {
      "epoch": 0.6687198954285817,
      "grad_norm": 1.5612128973007202,
      "learning_rate": 5.224293084905074e-05,
      "loss": 1.3645,
      "step": 18673
    },
    {
      "epoch": 0.66875570755815,
      "grad_norm": 1.5936546325683594,
      "learning_rate": 5.223274035496113e-05,
      "loss": 1.5845,
      "step": 18674
    },
    {
      "epoch": 0.6687915196877182,
      "grad_norm": 1.3002891540527344,
      "learning_rate": 5.222255050351841e-05,
      "loss": 1.4866,
      "step": 18675
    },
    {
      "epoch": 0.6688273318172865,
      "grad_norm": 1.6563246250152588,
      "learning_rate": 5.221236129485961e-05,
      "loss": 1.1432,
      "step": 18676
    },
    {
      "epoch": 0.6688631439468548,
      "grad_norm": 1.0846136808395386,
      "learning_rate": 5.2202172729121844e-05,
      "loss": 1.3841,
      "step": 18677
    },
    {
      "epoch": 0.6688989560764231,
      "grad_norm": 1.6632499694824219,
      "learning_rate": 5.219198480644221e-05,
      "loss": 1.7012,
      "step": 18678
    },
    {
      "epoch": 0.6689347682059914,
      "grad_norm": 1.6548495292663574,
      "learning_rate": 5.2181797526957764e-05,
      "loss": 1.4284,
      "step": 18679
    },
    {
      "epoch": 0.6689705803355597,
      "grad_norm": 1.82999849319458,
      "learning_rate": 5.2171610890805524e-05,
      "loss": 1.2064,
      "step": 18680
    },
    {
      "epoch": 0.6690063924651279,
      "grad_norm": 1.543445348739624,
      "learning_rate": 5.216142489812256e-05,
      "loss": 1.4014,
      "step": 18681
    },
    {
      "epoch": 0.6690422045946962,
      "grad_norm": 2.067296028137207,
      "learning_rate": 5.215123954904596e-05,
      "loss": 1.4993,
      "step": 18682
    },
    {
      "epoch": 0.6690780167242645,
      "grad_norm": 1.2235534191131592,
      "learning_rate": 5.2141054843712675e-05,
      "loss": 1.3685,
      "step": 18683
    },
    {
      "epoch": 0.6691138288538327,
      "grad_norm": 1.3292855024337769,
      "learning_rate": 5.213087078225975e-05,
      "loss": 1.5753,
      "step": 18684
    },
    {
      "epoch": 0.6691496409834011,
      "grad_norm": 1.5418928861618042,
      "learning_rate": 5.212068736482423e-05,
      "loss": 1.7125,
      "step": 18685
    },
    {
      "epoch": 0.6691854531129694,
      "grad_norm": 1.34342360496521,
      "learning_rate": 5.211050459154313e-05,
      "loss": 1.6613,
      "step": 18686
    },
    {
      "epoch": 0.6692212652425377,
      "grad_norm": 1.657388687133789,
      "learning_rate": 5.210032246255338e-05,
      "loss": 1.372,
      "step": 18687
    },
    {
      "epoch": 0.6692570773721059,
      "grad_norm": 1.9113507270812988,
      "learning_rate": 5.209014097799201e-05,
      "loss": 1.6323,
      "step": 18688
    },
    {
      "epoch": 0.6692928895016742,
      "grad_norm": 1.7546972036361694,
      "learning_rate": 5.207996013799603e-05,
      "loss": 1.4294,
      "step": 18689
    },
    {
      "epoch": 0.6693287016312425,
      "grad_norm": 2.2239842414855957,
      "learning_rate": 5.206977994270233e-05,
      "loss": 1.5034,
      "step": 18690
    },
    {
      "epoch": 0.6693645137608107,
      "grad_norm": 2.2625253200531006,
      "learning_rate": 5.205960039224795e-05,
      "loss": 1.7117,
      "step": 18691
    },
    {
      "epoch": 0.6694003258903791,
      "grad_norm": 2.014033079147339,
      "learning_rate": 5.2049421486769744e-05,
      "loss": 1.5873,
      "step": 18692
    },
    {
      "epoch": 0.6694361380199474,
      "grad_norm": 1.4991728067398071,
      "learning_rate": 5.203924322640479e-05,
      "loss": 1.2493,
      "step": 18693
    },
    {
      "epoch": 0.6694719501495157,
      "grad_norm": 1.4453134536743164,
      "learning_rate": 5.2029065611289926e-05,
      "loss": 1.5998,
      "step": 18694
    },
    {
      "epoch": 0.6695077622790839,
      "grad_norm": 1.5553264617919922,
      "learning_rate": 5.2018888641562126e-05,
      "loss": 1.2448,
      "step": 18695
    },
    {
      "epoch": 0.6695435744086522,
      "grad_norm": 1.590145230293274,
      "learning_rate": 5.200871231735822e-05,
      "loss": 1.3542,
      "step": 18696
    },
    {
      "epoch": 0.6695793865382205,
      "grad_norm": 1.796265959739685,
      "learning_rate": 5.1998536638815266e-05,
      "loss": 1.4915,
      "step": 18697
    },
    {
      "epoch": 0.6696151986677887,
      "grad_norm": 1.4939812421798706,
      "learning_rate": 5.198836160607008e-05,
      "loss": 1.4749,
      "step": 18698
    },
    {
      "epoch": 0.6696510107973571,
      "grad_norm": 1.3615853786468506,
      "learning_rate": 5.197818721925949e-05,
      "loss": 1.4405,
      "step": 18699
    },
    {
      "epoch": 0.6696868229269254,
      "grad_norm": 1.585461974143982,
      "learning_rate": 5.196801347852051e-05,
      "loss": 1.3092,
      "step": 18700
    },
    {
      "epoch": 0.6697226350564937,
      "grad_norm": 1.6997668743133545,
      "learning_rate": 5.195784038398992e-05,
      "loss": 1.5083,
      "step": 18701
    },
    {
      "epoch": 0.6697584471860619,
      "grad_norm": 1.4121911525726318,
      "learning_rate": 5.194766793580466e-05,
      "loss": 1.1211,
      "step": 18702
    },
    {
      "epoch": 0.6697942593156302,
      "grad_norm": 1.5166962146759033,
      "learning_rate": 5.193749613410146e-05,
      "loss": 1.2174,
      "step": 18703
    },
    {
      "epoch": 0.6698300714451985,
      "grad_norm": 1.7505559921264648,
      "learning_rate": 5.1927324979017335e-05,
      "loss": 1.4222,
      "step": 18704
    },
    {
      "epoch": 0.6698658835747667,
      "grad_norm": 1.718768835067749,
      "learning_rate": 5.191715447068901e-05,
      "loss": 1.2597,
      "step": 18705
    },
    {
      "epoch": 0.6699016957043351,
      "grad_norm": 1.9997526407241821,
      "learning_rate": 5.190698460925338e-05,
      "loss": 1.5697,
      "step": 18706
    },
    {
      "epoch": 0.6699375078339034,
      "grad_norm": 1.2355728149414062,
      "learning_rate": 5.1896815394847195e-05,
      "loss": 1.5054,
      "step": 18707
    },
    {
      "epoch": 0.6699733199634716,
      "grad_norm": 1.4850220680236816,
      "learning_rate": 5.188664682760731e-05,
      "loss": 1.6727,
      "step": 18708
    },
    {
      "epoch": 0.6700091320930399,
      "grad_norm": 1.334753155708313,
      "learning_rate": 5.1876478907670576e-05,
      "loss": 1.4472,
      "step": 18709
    },
    {
      "epoch": 0.6700449442226082,
      "grad_norm": 1.5342482328414917,
      "learning_rate": 5.186631163517367e-05,
      "loss": 1.2759,
      "step": 18710
    },
    {
      "epoch": 0.6700807563521765,
      "grad_norm": 1.6298881769180298,
      "learning_rate": 5.185614501025353e-05,
      "loss": 1.4384,
      "step": 18711
    },
    {
      "epoch": 0.6701165684817447,
      "grad_norm": 1.4282634258270264,
      "learning_rate": 5.184597903304681e-05,
      "loss": 1.4343,
      "step": 18712
    },
    {
      "epoch": 0.6701523806113131,
      "grad_norm": 1.7089579105377197,
      "learning_rate": 5.183581370369037e-05,
      "loss": 1.67,
      "step": 18713
    },
    {
      "epoch": 0.6701881927408814,
      "grad_norm": 1.507480502128601,
      "learning_rate": 5.182564902232086e-05,
      "loss": 1.4983,
      "step": 18714
    },
    {
      "epoch": 0.6702240048704496,
      "grad_norm": 1.9724632501602173,
      "learning_rate": 5.1815484989075157e-05,
      "loss": 1.4372,
      "step": 18715
    },
    {
      "epoch": 0.6702598170000179,
      "grad_norm": 1.533915400505066,
      "learning_rate": 5.1805321604089974e-05,
      "loss": 1.4643,
      "step": 18716
    },
    {
      "epoch": 0.6702956291295862,
      "grad_norm": 1.8036576509475708,
      "learning_rate": 5.1795158867501966e-05,
      "loss": 1.322,
      "step": 18717
    },
    {
      "epoch": 0.6703314412591544,
      "grad_norm": 1.8307304382324219,
      "learning_rate": 5.1784996779447926e-05,
      "loss": 1.4068,
      "step": 18718
    },
    {
      "epoch": 0.6703672533887227,
      "grad_norm": 1.794142484664917,
      "learning_rate": 5.177483534006455e-05,
      "loss": 1.6021,
      "step": 18719
    },
    {
      "epoch": 0.6704030655182911,
      "grad_norm": 1.9921245574951172,
      "learning_rate": 5.1764674549488614e-05,
      "loss": 1.0792,
      "step": 18720
    },
    {
      "epoch": 0.6704388776478594,
      "grad_norm": 1.671963095664978,
      "learning_rate": 5.175451440785671e-05,
      "loss": 1.7145,
      "step": 18721
    },
    {
      "epoch": 0.6704746897774276,
      "grad_norm": 2.1620876789093018,
      "learning_rate": 5.174435491530559e-05,
      "loss": 1.6685,
      "step": 18722
    },
    {
      "epoch": 0.6705105019069959,
      "grad_norm": 1.408326268196106,
      "learning_rate": 5.173419607197193e-05,
      "loss": 1.1387,
      "step": 18723
    },
    {
      "epoch": 0.6705463140365642,
      "grad_norm": 1.5527623891830444,
      "learning_rate": 5.172403787799245e-05,
      "loss": 1.1697,
      "step": 18724
    },
    {
      "epoch": 0.6705821261661324,
      "grad_norm": 1.6615278720855713,
      "learning_rate": 5.1713880333503704e-05,
      "loss": 1.6103,
      "step": 18725
    },
    {
      "epoch": 0.6706179382957007,
      "grad_norm": 2.0983941555023193,
      "learning_rate": 5.1703723438642436e-05,
      "loss": 1.6692,
      "step": 18726
    },
    {
      "epoch": 0.6706537504252691,
      "grad_norm": 1.7213473320007324,
      "learning_rate": 5.16935671935453e-05,
      "loss": 1.4165,
      "step": 18727
    },
    {
      "epoch": 0.6706895625548374,
      "grad_norm": 2.2718751430511475,
      "learning_rate": 5.1683411598348876e-05,
      "loss": 1.4563,
      "step": 18728
    },
    {
      "epoch": 0.6707253746844056,
      "grad_norm": 1.7657798528671265,
      "learning_rate": 5.167325665318983e-05,
      "loss": 1.6676,
      "step": 18729
    },
    {
      "epoch": 0.6707611868139739,
      "grad_norm": 1.8499521017074585,
      "learning_rate": 5.1663102358204754e-05,
      "loss": 1.5885,
      "step": 18730
    },
    {
      "epoch": 0.6707969989435422,
      "grad_norm": 1.6801620721817017,
      "learning_rate": 5.165294871353035e-05,
      "loss": 1.2257,
      "step": 18731
    },
    {
      "epoch": 0.6708328110731104,
      "grad_norm": 1.3649678230285645,
      "learning_rate": 5.16427957193031e-05,
      "loss": 1.4112,
      "step": 18732
    },
    {
      "epoch": 0.6708686232026787,
      "grad_norm": 1.5700057744979858,
      "learning_rate": 5.163264337565967e-05,
      "loss": 1.5946,
      "step": 18733
    },
    {
      "epoch": 0.6709044353322471,
      "grad_norm": 1.8407195806503296,
      "learning_rate": 5.1622491682736675e-05,
      "loss": 1.2455,
      "step": 18734
    },
    {
      "epoch": 0.6709402474618154,
      "grad_norm": 2.2659173011779785,
      "learning_rate": 5.16123406406706e-05,
      "loss": 1.8748,
      "step": 18735
    },
    {
      "epoch": 0.6709760595913836,
      "grad_norm": 2.0498504638671875,
      "learning_rate": 5.160219024959807e-05,
      "loss": 1.9133,
      "step": 18736
    },
    {
      "epoch": 0.6710118717209519,
      "grad_norm": 1.556580662727356,
      "learning_rate": 5.159204050965565e-05,
      "loss": 1.183,
      "step": 18737
    },
    {
      "epoch": 0.6710476838505202,
      "grad_norm": 1.2161188125610352,
      "learning_rate": 5.158189142097991e-05,
      "loss": 1.2662,
      "step": 18738
    },
    {
      "epoch": 0.6710834959800884,
      "grad_norm": 1.29056715965271,
      "learning_rate": 5.157174298370734e-05,
      "loss": 1.2986,
      "step": 18739
    },
    {
      "epoch": 0.6711193081096567,
      "grad_norm": 2.0055019855499268,
      "learning_rate": 5.15615951979745e-05,
      "loss": 1.2297,
      "step": 18740
    },
    {
      "epoch": 0.6711551202392251,
      "grad_norm": 1.7979457378387451,
      "learning_rate": 5.155144806391789e-05,
      "loss": 1.413,
      "step": 18741
    },
    {
      "epoch": 0.6711909323687933,
      "grad_norm": 1.5609806776046753,
      "learning_rate": 5.154130158167412e-05,
      "loss": 1.4996,
      "step": 18742
    },
    {
      "epoch": 0.6712267444983616,
      "grad_norm": 1.425581455230713,
      "learning_rate": 5.153115575137959e-05,
      "loss": 1.2355,
      "step": 18743
    },
    {
      "epoch": 0.6712625566279299,
      "grad_norm": 2.3990538120269775,
      "learning_rate": 5.152101057317082e-05,
      "loss": 1.3091,
      "step": 18744
    },
    {
      "epoch": 0.6712983687574982,
      "grad_norm": 1.8333427906036377,
      "learning_rate": 5.151086604718438e-05,
      "loss": 1.1728,
      "step": 18745
    },
    {
      "epoch": 0.6713341808870664,
      "grad_norm": 1.5617541074752808,
      "learning_rate": 5.150072217355664e-05,
      "loss": 1.1859,
      "step": 18746
    },
    {
      "epoch": 0.6713699930166347,
      "grad_norm": 1.6639366149902344,
      "learning_rate": 5.149057895242412e-05,
      "loss": 1.5773,
      "step": 18747
    },
    {
      "epoch": 0.6714058051462031,
      "grad_norm": 1.8866615295410156,
      "learning_rate": 5.148043638392329e-05,
      "loss": 1.2715,
      "step": 18748
    },
    {
      "epoch": 0.6714416172757713,
      "grad_norm": 1.8568611145019531,
      "learning_rate": 5.147029446819065e-05,
      "loss": 1.4749,
      "step": 18749
    },
    {
      "epoch": 0.6714774294053396,
      "grad_norm": 2.128819704055786,
      "learning_rate": 5.146015320536255e-05,
      "loss": 1.2029,
      "step": 18750
    },
    {
      "epoch": 0.6715132415349079,
      "grad_norm": 1.3638184070587158,
      "learning_rate": 5.145001259557548e-05,
      "loss": 1.3943,
      "step": 18751
    },
    {
      "epoch": 0.6715490536644761,
      "grad_norm": 1.7398433685302734,
      "learning_rate": 5.14398726389659e-05,
      "loss": 1.082,
      "step": 18752
    },
    {
      "epoch": 0.6715848657940444,
      "grad_norm": 2.225938320159912,
      "learning_rate": 5.142973333567016e-05,
      "loss": 1.4355,
      "step": 18753
    },
    {
      "epoch": 0.6716206779236127,
      "grad_norm": 2.0473129749298096,
      "learning_rate": 5.141959468582471e-05,
      "loss": 1.4406,
      "step": 18754
    },
    {
      "epoch": 0.6716564900531811,
      "grad_norm": 1.4689384698867798,
      "learning_rate": 5.140945668956595e-05,
      "loss": 1.4992,
      "step": 18755
    },
    {
      "epoch": 0.6716923021827493,
      "grad_norm": 1.6381616592407227,
      "learning_rate": 5.1399319347030306e-05,
      "loss": 1.6428,
      "step": 18756
    },
    {
      "epoch": 0.6717281143123176,
      "grad_norm": 1.9008187055587769,
      "learning_rate": 5.1389182658354105e-05,
      "loss": 1.8243,
      "step": 18757
    },
    {
      "epoch": 0.6717639264418859,
      "grad_norm": 3.256089448928833,
      "learning_rate": 5.137904662367373e-05,
      "loss": 1.5297,
      "step": 18758
    },
    {
      "epoch": 0.6717997385714541,
      "grad_norm": 1.397004246711731,
      "learning_rate": 5.136891124312557e-05,
      "loss": 1.2896,
      "step": 18759
    },
    {
      "epoch": 0.6718355507010224,
      "grad_norm": 1.8906328678131104,
      "learning_rate": 5.135877651684603e-05,
      "loss": 1.5225,
      "step": 18760
    },
    {
      "epoch": 0.6718713628305907,
      "grad_norm": 1.9930126667022705,
      "learning_rate": 5.1348642444971364e-05,
      "loss": 1.4106,
      "step": 18761
    },
    {
      "epoch": 0.6719071749601591,
      "grad_norm": 1.5230907201766968,
      "learning_rate": 5.133850902763795e-05,
      "loss": 1.4992,
      "step": 18762
    },
    {
      "epoch": 0.6719429870897273,
      "grad_norm": 1.6010066270828247,
      "learning_rate": 5.132837626498217e-05,
      "loss": 1.4002,
      "step": 18763
    },
    {
      "epoch": 0.6719787992192956,
      "grad_norm": 1.5590764284133911,
      "learning_rate": 5.1318244157140285e-05,
      "loss": 1.2998,
      "step": 18764
    },
    {
      "epoch": 0.6720146113488639,
      "grad_norm": 1.6329294443130493,
      "learning_rate": 5.13081127042486e-05,
      "loss": 1.1118,
      "step": 18765
    },
    {
      "epoch": 0.6720504234784321,
      "grad_norm": 1.5088551044464111,
      "learning_rate": 5.129798190644348e-05,
      "loss": 1.2605,
      "step": 18766
    },
    {
      "epoch": 0.6720862356080004,
      "grad_norm": 1.7145344018936157,
      "learning_rate": 5.128785176386122e-05,
      "loss": 1.3524,
      "step": 18767
    },
    {
      "epoch": 0.6721220477375687,
      "grad_norm": 1.9711799621582031,
      "learning_rate": 5.127772227663803e-05,
      "loss": 1.4702,
      "step": 18768
    },
    {
      "epoch": 0.672157859867137,
      "grad_norm": 2.004667043685913,
      "learning_rate": 5.1267593444910254e-05,
      "loss": 1.5904,
      "step": 18769
    },
    {
      "epoch": 0.6721936719967053,
      "grad_norm": 2.5094735622406006,
      "learning_rate": 5.125746526881417e-05,
      "loss": 1.4116,
      "step": 18770
    },
    {
      "epoch": 0.6722294841262736,
      "grad_norm": 1.3970831632614136,
      "learning_rate": 5.1247337748486005e-05,
      "loss": 1.3415,
      "step": 18771
    },
    {
      "epoch": 0.6722652962558419,
      "grad_norm": 1.901026725769043,
      "learning_rate": 5.1237210884061994e-05,
      "loss": 1.382,
      "step": 18772
    },
    {
      "epoch": 0.6723011083854101,
      "grad_norm": 1.8950276374816895,
      "learning_rate": 5.1227084675678425e-05,
      "loss": 1.2152,
      "step": 18773
    },
    {
      "epoch": 0.6723369205149784,
      "grad_norm": 1.730116844177246,
      "learning_rate": 5.121695912347156e-05,
      "loss": 1.4224,
      "step": 18774
    },
    {
      "epoch": 0.6723727326445467,
      "grad_norm": 1.482043743133545,
      "learning_rate": 5.120683422757755e-05,
      "loss": 1.526,
      "step": 18775
    },
    {
      "epoch": 0.672408544774115,
      "grad_norm": 1.451438069343567,
      "learning_rate": 5.119670998813264e-05,
      "loss": 1.4829,
      "step": 18776
    },
    {
      "epoch": 0.6724443569036833,
      "grad_norm": 2.14319109916687,
      "learning_rate": 5.1186586405273055e-05,
      "loss": 1.3499,
      "step": 18777
    },
    {
      "epoch": 0.6724801690332516,
      "grad_norm": 2.263322353363037,
      "learning_rate": 5.117646347913501e-05,
      "loss": 1.144,
      "step": 18778
    },
    {
      "epoch": 0.6725159811628199,
      "grad_norm": 1.8152785301208496,
      "learning_rate": 5.116634120985467e-05,
      "loss": 1.3347,
      "step": 18779
    },
    {
      "epoch": 0.6725517932923881,
      "grad_norm": 1.4670072793960571,
      "learning_rate": 5.115621959756815e-05,
      "loss": 1.3096,
      "step": 18780
    },
    {
      "epoch": 0.6725876054219564,
      "grad_norm": 1.4545962810516357,
      "learning_rate": 5.1146098642411765e-05,
      "loss": 1.4713,
      "step": 18781
    },
    {
      "epoch": 0.6726234175515247,
      "grad_norm": 1.4335076808929443,
      "learning_rate": 5.113597834452157e-05,
      "loss": 1.323,
      "step": 18782
    },
    {
      "epoch": 0.672659229681093,
      "grad_norm": 1.4578428268432617,
      "learning_rate": 5.1125858704033745e-05,
      "loss": 1.6607,
      "step": 18783
    },
    {
      "epoch": 0.6726950418106613,
      "grad_norm": 1.8013838529586792,
      "learning_rate": 5.111573972108446e-05,
      "loss": 1.7419,
      "step": 18784
    },
    {
      "epoch": 0.6727308539402296,
      "grad_norm": 1.3669114112854004,
      "learning_rate": 5.1105621395809875e-05,
      "loss": 1.3873,
      "step": 18785
    },
    {
      "epoch": 0.6727666660697978,
      "grad_norm": 1.4641414880752563,
      "learning_rate": 5.1095503728346095e-05,
      "loss": 1.4673,
      "step": 18786
    },
    {
      "epoch": 0.6728024781993661,
      "grad_norm": 2.017226219177246,
      "learning_rate": 5.108538671882914e-05,
      "loss": 1.6612,
      "step": 18787
    },
    {
      "epoch": 0.6728382903289344,
      "grad_norm": 1.617396593093872,
      "learning_rate": 5.10752703673953e-05,
      "loss": 1.5628,
      "step": 18788
    },
    {
      "epoch": 0.6728741024585027,
      "grad_norm": 1.8963992595672607,
      "learning_rate": 5.106515467418054e-05,
      "loss": 1.2682,
      "step": 18789
    },
    {
      "epoch": 0.672909914588071,
      "grad_norm": 2.118973731994629,
      "learning_rate": 5.1055039639321046e-05,
      "loss": 1.3527,
      "step": 18790
    },
    {
      "epoch": 0.6729457267176393,
      "grad_norm": 1.6391757726669312,
      "learning_rate": 5.104492526295278e-05,
      "loss": 1.6404,
      "step": 18791
    },
    {
      "epoch": 0.6729815388472076,
      "grad_norm": 1.6890900135040283,
      "learning_rate": 5.103481154521197e-05,
      "loss": 1.1516,
      "step": 18792
    },
    {
      "epoch": 0.6730173509767758,
      "grad_norm": 2.0314505100250244,
      "learning_rate": 5.102469848623459e-05,
      "loss": 1.0655,
      "step": 18793
    },
    {
      "epoch": 0.6730531631063441,
      "grad_norm": 1.655391812324524,
      "learning_rate": 5.10145860861567e-05,
      "loss": 1.3118,
      "step": 18794
    },
    {
      "epoch": 0.6730889752359124,
      "grad_norm": 1.847886085510254,
      "learning_rate": 5.1004474345114404e-05,
      "loss": 1.6991,
      "step": 18795
    },
    {
      "epoch": 0.6731247873654806,
      "grad_norm": 1.691873550415039,
      "learning_rate": 5.099436326324367e-05,
      "loss": 1.4669,
      "step": 18796
    },
    {
      "epoch": 0.673160599495049,
      "grad_norm": 1.4754281044006348,
      "learning_rate": 5.098425284068062e-05,
      "loss": 1.2964,
      "step": 18797
    },
    {
      "epoch": 0.6731964116246173,
      "grad_norm": 1.3968569040298462,
      "learning_rate": 5.0974143077561135e-05,
      "loss": 1.4885,
      "step": 18798
    },
    {
      "epoch": 0.6732322237541856,
      "grad_norm": 1.5505691766738892,
      "learning_rate": 5.0964033974021386e-05,
      "loss": 1.4887,
      "step": 18799
    },
    {
      "epoch": 0.6732680358837538,
      "grad_norm": 1.2982192039489746,
      "learning_rate": 5.095392553019728e-05,
      "loss": 1.4633,
      "step": 18800
    },
    {
      "epoch": 0.6733038480133221,
      "grad_norm": 1.4484449625015259,
      "learning_rate": 5.094381774622488e-05,
      "loss": 1.5766,
      "step": 18801
    },
    {
      "epoch": 0.6733396601428904,
      "grad_norm": 2.2219207286834717,
      "learning_rate": 5.0933710622240036e-05,
      "loss": 1.6196,
      "step": 18802
    },
    {
      "epoch": 0.6733754722724586,
      "grad_norm": 1.9814118146896362,
      "learning_rate": 5.0923604158378924e-05,
      "loss": 1.5831,
      "step": 18803
    },
    {
      "epoch": 0.673411284402027,
      "grad_norm": 2.133800983428955,
      "learning_rate": 5.091349835477741e-05,
      "loss": 1.3814,
      "step": 18804
    },
    {
      "epoch": 0.6734470965315953,
      "grad_norm": 1.7097578048706055,
      "learning_rate": 5.0903393211571414e-05,
      "loss": 1.5687,
      "step": 18805
    },
    {
      "epoch": 0.6734829086611636,
      "grad_norm": 1.711245059967041,
      "learning_rate": 5.089328872889694e-05,
      "loss": 1.2619,
      "step": 18806
    },
    {
      "epoch": 0.6735187207907318,
      "grad_norm": 1.5886105298995972,
      "learning_rate": 5.0883184906889924e-05,
      "loss": 1.3635,
      "step": 18807
    },
    {
      "epoch": 0.6735545329203001,
      "grad_norm": 1.8277711868286133,
      "learning_rate": 5.087308174568632e-05,
      "loss": 1.7714,
      "step": 18808
    },
    {
      "epoch": 0.6735903450498684,
      "grad_norm": 1.5488297939300537,
      "learning_rate": 5.086297924542198e-05,
      "loss": 1.2493,
      "step": 18809
    },
    {
      "epoch": 0.6736261571794366,
      "grad_norm": 1.1911660432815552,
      "learning_rate": 5.085287740623292e-05,
      "loss": 1.4918,
      "step": 18810
    },
    {
      "epoch": 0.673661969309005,
      "grad_norm": 2.2458438873291016,
      "learning_rate": 5.0842776228255e-05,
      "loss": 1.3248,
      "step": 18811
    },
    {
      "epoch": 0.6736977814385733,
      "grad_norm": 1.5594806671142578,
      "learning_rate": 5.083267571162412e-05,
      "loss": 1.5582,
      "step": 18812
    },
    {
      "epoch": 0.6737335935681416,
      "grad_norm": 2.4863622188568115,
      "learning_rate": 5.082257585647614e-05,
      "loss": 1.5133,
      "step": 18813
    },
    {
      "epoch": 0.6737694056977098,
      "grad_norm": 2.48881196975708,
      "learning_rate": 5.0812476662946975e-05,
      "loss": 1.3237,
      "step": 18814
    },
    {
      "epoch": 0.6738052178272781,
      "grad_norm": 1.7107995748519897,
      "learning_rate": 5.0802378131172525e-05,
      "loss": 1.8406,
      "step": 18815
    },
    {
      "epoch": 0.6738410299568464,
      "grad_norm": 1.973778247833252,
      "learning_rate": 5.079228026128857e-05,
      "loss": 1.2656,
      "step": 18816
    },
    {
      "epoch": 0.6738768420864146,
      "grad_norm": 1.4093636274337769,
      "learning_rate": 5.078218305343102e-05,
      "loss": 1.5239,
      "step": 18817
    },
    {
      "epoch": 0.6739126542159829,
      "grad_norm": 2.64251446723938,
      "learning_rate": 5.07720865077357e-05,
      "loss": 1.4003,
      "step": 18818
    },
    {
      "epoch": 0.6739484663455513,
      "grad_norm": 1.3757315874099731,
      "learning_rate": 5.0761990624338504e-05,
      "loss": 1.4934,
      "step": 18819
    },
    {
      "epoch": 0.6739842784751195,
      "grad_norm": 1.2844980955123901,
      "learning_rate": 5.075189540337514e-05,
      "loss": 1.5985,
      "step": 18820
    },
    {
      "epoch": 0.6740200906046878,
      "grad_norm": 2.0259010791778564,
      "learning_rate": 5.074180084498157e-05,
      "loss": 1.5542,
      "step": 18821
    },
    {
      "epoch": 0.6740559027342561,
      "grad_norm": 1.7492506504058838,
      "learning_rate": 5.0731706949293525e-05,
      "loss": 1.4885,
      "step": 18822
    },
    {
      "epoch": 0.6740917148638244,
      "grad_norm": 1.6474558115005493,
      "learning_rate": 5.072161371644677e-05,
      "loss": 1.5,
      "step": 18823
    },
    {
      "epoch": 0.6741275269933926,
      "grad_norm": 1.8024877309799194,
      "learning_rate": 5.0711521146577156e-05,
      "loss": 1.3702,
      "step": 18824
    },
    {
      "epoch": 0.6741633391229609,
      "grad_norm": 1.5845979452133179,
      "learning_rate": 5.070142923982043e-05,
      "loss": 1.2484,
      "step": 18825
    },
    {
      "epoch": 0.6741991512525293,
      "grad_norm": 1.974819540977478,
      "learning_rate": 5.069133799631243e-05,
      "loss": 1.4982,
      "step": 18826
    },
    {
      "epoch": 0.6742349633820975,
      "grad_norm": 2.497584819793701,
      "learning_rate": 5.0681247416188826e-05,
      "loss": 1.5294,
      "step": 18827
    },
    {
      "epoch": 0.6742707755116658,
      "grad_norm": 1.8491204977035522,
      "learning_rate": 5.067115749958543e-05,
      "loss": 1.4184,
      "step": 18828
    },
    {
      "epoch": 0.6743065876412341,
      "grad_norm": 1.4940153360366821,
      "learning_rate": 5.066106824663798e-05,
      "loss": 1.4861,
      "step": 18829
    },
    {
      "epoch": 0.6743423997708023,
      "grad_norm": 1.228395938873291,
      "learning_rate": 5.065097965748224e-05,
      "loss": 1.543,
      "step": 18830
    },
    {
      "epoch": 0.6743782119003706,
      "grad_norm": 1.5438652038574219,
      "learning_rate": 5.0640891732253905e-05,
      "loss": 1.3178,
      "step": 18831
    },
    {
      "epoch": 0.6744140240299389,
      "grad_norm": 1.4345051050186157,
      "learning_rate": 5.063080447108868e-05,
      "loss": 1.5065,
      "step": 18832
    },
    {
      "epoch": 0.6744498361595073,
      "grad_norm": 1.7704910039901733,
      "learning_rate": 5.0620717874122336e-05,
      "loss": 1.4677,
      "step": 18833
    },
    {
      "epoch": 0.6744856482890755,
      "grad_norm": 1.8506509065628052,
      "learning_rate": 5.06106319414905e-05,
      "loss": 1.8333,
      "step": 18834
    },
    {
      "epoch": 0.6745214604186438,
      "grad_norm": 1.587844967842102,
      "learning_rate": 5.0600546673328916e-05,
      "loss": 1.3863,
      "step": 18835
    },
    {
      "epoch": 0.6745572725482121,
      "grad_norm": 1.68564772605896,
      "learning_rate": 5.059046206977325e-05,
      "loss": 1.9733,
      "step": 18836
    },
    {
      "epoch": 0.6745930846777803,
      "grad_norm": 1.5647304058074951,
      "learning_rate": 5.0580378130959216e-05,
      "loss": 1.8071,
      "step": 18837
    },
    {
      "epoch": 0.6746288968073486,
      "grad_norm": 1.7359753847122192,
      "learning_rate": 5.05702948570224e-05,
      "loss": 1.1989,
      "step": 18838
    },
    {
      "epoch": 0.6746647089369169,
      "grad_norm": 2.1853439807891846,
      "learning_rate": 5.056021224809853e-05,
      "loss": 1.3432,
      "step": 18839
    },
    {
      "epoch": 0.6747005210664853,
      "grad_norm": 1.6081196069717407,
      "learning_rate": 5.055013030432326e-05,
      "loss": 1.1698,
      "step": 18840
    },
    {
      "epoch": 0.6747363331960535,
      "grad_norm": 1.6497262716293335,
      "learning_rate": 5.054004902583216e-05,
      "loss": 1.4807,
      "step": 18841
    },
    {
      "epoch": 0.6747721453256218,
      "grad_norm": 2.125016689300537,
      "learning_rate": 5.052996841276091e-05,
      "loss": 1.4742,
      "step": 18842
    },
    {
      "epoch": 0.6748079574551901,
      "grad_norm": 1.5704431533813477,
      "learning_rate": 5.0519888465245116e-05,
      "loss": 1.5663,
      "step": 18843
    },
    {
      "epoch": 0.6748437695847583,
      "grad_norm": 2.178602695465088,
      "learning_rate": 5.050980918342043e-05,
      "loss": 1.57,
      "step": 18844
    },
    {
      "epoch": 0.6748795817143266,
      "grad_norm": 1.4927809238433838,
      "learning_rate": 5.04997305674224e-05,
      "loss": 1.3771,
      "step": 18845
    },
    {
      "epoch": 0.6749153938438949,
      "grad_norm": 3.4965755939483643,
      "learning_rate": 5.048965261738664e-05,
      "loss": 1.6496,
      "step": 18846
    },
    {
      "epoch": 0.6749512059734633,
      "grad_norm": 1.5973447561264038,
      "learning_rate": 5.047957533344874e-05,
      "loss": 1.3466,
      "step": 18847
    },
    {
      "epoch": 0.6749870181030315,
      "grad_norm": 1.5947939157485962,
      "learning_rate": 5.0469498715744314e-05,
      "loss": 1.3644,
      "step": 18848
    },
    {
      "epoch": 0.6750228302325998,
      "grad_norm": 1.60691237449646,
      "learning_rate": 5.045942276440885e-05,
      "loss": 1.6925,
      "step": 18849
    },
    {
      "epoch": 0.6750586423621681,
      "grad_norm": 1.4345650672912598,
      "learning_rate": 5.0449347479577946e-05,
      "loss": 1.163,
      "step": 18850
    },
    {
      "epoch": 0.6750944544917363,
      "grad_norm": 2.119398593902588,
      "learning_rate": 5.043927286138721e-05,
      "loss": 1.5569,
      "step": 18851
    },
    {
      "epoch": 0.6751302666213046,
      "grad_norm": 1.8614598512649536,
      "learning_rate": 5.0429198909972086e-05,
      "loss": 1.5067,
      "step": 18852
    },
    {
      "epoch": 0.6751660787508729,
      "grad_norm": 2.991887331008911,
      "learning_rate": 5.041912562546813e-05,
      "loss": 1.673,
      "step": 18853
    },
    {
      "epoch": 0.6752018908804412,
      "grad_norm": 1.687841534614563,
      "learning_rate": 5.040905300801091e-05,
      "loss": 1.235,
      "step": 18854
    },
    {
      "epoch": 0.6752377030100095,
      "grad_norm": 2.048767566680908,
      "learning_rate": 5.039898105773594e-05,
      "loss": 1.5541,
      "step": 18855
    },
    {
      "epoch": 0.6752735151395778,
      "grad_norm": 1.4294754266738892,
      "learning_rate": 5.038890977477866e-05,
      "loss": 1.2713,
      "step": 18856
    },
    {
      "epoch": 0.675309327269146,
      "grad_norm": 2.4173076152801514,
      "learning_rate": 5.037883915927462e-05,
      "loss": 1.4883,
      "step": 18857
    },
    {
      "epoch": 0.6753451393987143,
      "grad_norm": 1.8582412004470825,
      "learning_rate": 5.036876921135931e-05,
      "loss": 1.2395,
      "step": 18858
    },
    {
      "epoch": 0.6753809515282826,
      "grad_norm": 1.710937738418579,
      "learning_rate": 5.035869993116816e-05,
      "loss": 1.5521,
      "step": 18859
    },
    {
      "epoch": 0.6754167636578509,
      "grad_norm": 1.5905033349990845,
      "learning_rate": 5.034863131883667e-05,
      "loss": 1.5004,
      "step": 18860
    },
    {
      "epoch": 0.6754525757874192,
      "grad_norm": 1.7289782762527466,
      "learning_rate": 5.03385633745003e-05,
      "loss": 1.4988,
      "step": 18861
    },
    {
      "epoch": 0.6754883879169875,
      "grad_norm": 1.712811827659607,
      "learning_rate": 5.032849609829454e-05,
      "loss": 1.6214,
      "step": 18862
    },
    {
      "epoch": 0.6755242000465558,
      "grad_norm": 2.12488055229187,
      "learning_rate": 5.0318429490354754e-05,
      "loss": 1.4074,
      "step": 18863
    },
    {
      "epoch": 0.675560012176124,
      "grad_norm": 1.458487629890442,
      "learning_rate": 5.030836355081643e-05,
      "loss": 1.3575,
      "step": 18864
    },
    {
      "epoch": 0.6755958243056923,
      "grad_norm": 2.2130470275878906,
      "learning_rate": 5.0298298279814956e-05,
      "loss": 1.617,
      "step": 18865
    },
    {
      "epoch": 0.6756316364352606,
      "grad_norm": 1.495819091796875,
      "learning_rate": 5.0288233677485806e-05,
      "loss": 1.4228,
      "step": 18866
    },
    {
      "epoch": 0.6756674485648289,
      "grad_norm": 1.9383759498596191,
      "learning_rate": 5.027816974396432e-05,
      "loss": 1.2309,
      "step": 18867
    },
    {
      "epoch": 0.6757032606943972,
      "grad_norm": 1.6474690437316895,
      "learning_rate": 5.0268106479385924e-05,
      "loss": 1.1464,
      "step": 18868
    },
    {
      "epoch": 0.6757390728239655,
      "grad_norm": 1.8009663820266724,
      "learning_rate": 5.025804388388604e-05,
      "loss": 1.7125,
      "step": 18869
    },
    {
      "epoch": 0.6757748849535338,
      "grad_norm": 1.71844482421875,
      "learning_rate": 5.024798195759998e-05,
      "loss": 1.6644,
      "step": 18870
    },
    {
      "epoch": 0.675810697083102,
      "grad_norm": 1.4656023979187012,
      "learning_rate": 5.023792070066313e-05,
      "loss": 1.552,
      "step": 18871
    },
    {
      "epoch": 0.6758465092126703,
      "grad_norm": 1.699833869934082,
      "learning_rate": 5.022786011321089e-05,
      "loss": 1.2069,
      "step": 18872
    },
    {
      "epoch": 0.6758823213422386,
      "grad_norm": 1.6764025688171387,
      "learning_rate": 5.021780019537862e-05,
      "loss": 1.5697,
      "step": 18873
    },
    {
      "epoch": 0.6759181334718068,
      "grad_norm": 1.3379871845245361,
      "learning_rate": 5.02077409473016e-05,
      "loss": 1.5602,
      "step": 18874
    },
    {
      "epoch": 0.6759539456013752,
      "grad_norm": 2.3208539485931396,
      "learning_rate": 5.019768236911519e-05,
      "loss": 1.5169,
      "step": 18875
    },
    {
      "epoch": 0.6759897577309435,
      "grad_norm": 1.4288512468338013,
      "learning_rate": 5.018762446095476e-05,
      "loss": 1.1977,
      "step": 18876
    },
    {
      "epoch": 0.6760255698605118,
      "grad_norm": 1.9357956647872925,
      "learning_rate": 5.017756722295557e-05,
      "loss": 1.1796,
      "step": 18877
    },
    {
      "epoch": 0.67606138199008,
      "grad_norm": 1.7917031049728394,
      "learning_rate": 5.016751065525292e-05,
      "loss": 1.3224,
      "step": 18878
    },
    {
      "epoch": 0.6760971941196483,
      "grad_norm": 1.693039059638977,
      "learning_rate": 5.015745475798215e-05,
      "loss": 1.3998,
      "step": 18879
    },
    {
      "epoch": 0.6761330062492166,
      "grad_norm": 1.3678598403930664,
      "learning_rate": 5.014739953127857e-05,
      "loss": 1.185,
      "step": 18880
    },
    {
      "epoch": 0.6761688183787848,
      "grad_norm": 1.3098324537277222,
      "learning_rate": 5.013734497527739e-05,
      "loss": 1.3323,
      "step": 18881
    },
    {
      "epoch": 0.6762046305083532,
      "grad_norm": 1.487381935119629,
      "learning_rate": 5.0127291090113917e-05,
      "loss": 1.3497,
      "step": 18882
    },
    {
      "epoch": 0.6762404426379215,
      "grad_norm": 2.2838425636291504,
      "learning_rate": 5.011723787592344e-05,
      "loss": 1.5023,
      "step": 18883
    },
    {
      "epoch": 0.6762762547674898,
      "grad_norm": 1.4968982934951782,
      "learning_rate": 5.0107185332841155e-05,
      "loss": 1.221,
      "step": 18884
    },
    {
      "epoch": 0.676312066897058,
      "grad_norm": 2.1041419506073,
      "learning_rate": 5.009713346100235e-05,
      "loss": 2.0279,
      "step": 18885
    },
    {
      "epoch": 0.6763478790266263,
      "grad_norm": 1.8831173181533813,
      "learning_rate": 5.008708226054219e-05,
      "loss": 1.6359,
      "step": 18886
    },
    {
      "epoch": 0.6763836911561946,
      "grad_norm": 1.8215208053588867,
      "learning_rate": 5.007703173159604e-05,
      "loss": 1.1922,
      "step": 18887
    },
    {
      "epoch": 0.6764195032857628,
      "grad_norm": 1.7861186265945435,
      "learning_rate": 5.0066981874298967e-05,
      "loss": 1.351,
      "step": 18888
    },
    {
      "epoch": 0.6764553154153312,
      "grad_norm": 1.681621789932251,
      "learning_rate": 5.0056932688786294e-05,
      "loss": 1.6124,
      "step": 18889
    },
    {
      "epoch": 0.6764911275448995,
      "grad_norm": 1.4928348064422607,
      "learning_rate": 5.00468841751931e-05,
      "loss": 1.384,
      "step": 18890
    },
    {
      "epoch": 0.6765269396744678,
      "grad_norm": 1.405524730682373,
      "learning_rate": 5.0036836333654715e-05,
      "loss": 1.1791,
      "step": 18891
    },
    {
      "epoch": 0.676562751804036,
      "grad_norm": 1.3856693506240845,
      "learning_rate": 5.0026789164306255e-05,
      "loss": 1.5605,
      "step": 18892
    },
    {
      "epoch": 0.6765985639336043,
      "grad_norm": 1.7855710983276367,
      "learning_rate": 5.00167426672828e-05,
      "loss": 1.2509,
      "step": 18893
    },
    {
      "epoch": 0.6766343760631726,
      "grad_norm": 1.4088751077651978,
      "learning_rate": 5.000669684271968e-05,
      "loss": 1.4611,
      "step": 18894
    },
    {
      "epoch": 0.6766701881927408,
      "grad_norm": 1.4906492233276367,
      "learning_rate": 4.999665169075193e-05,
      "loss": 1.2085,
      "step": 18895
    },
    {
      "epoch": 0.6767060003223092,
      "grad_norm": 1.3844395875930786,
      "learning_rate": 4.998660721151476e-05,
      "loss": 1.0361,
      "step": 18896
    },
    {
      "epoch": 0.6767418124518775,
      "grad_norm": 1.425974726676941,
      "learning_rate": 4.997656340514321e-05,
      "loss": 1.4404,
      "step": 18897
    },
    {
      "epoch": 0.6767776245814457,
      "grad_norm": 2.3073105812072754,
      "learning_rate": 4.996652027177255e-05,
      "loss": 1.4989,
      "step": 18898
    },
    {
      "epoch": 0.676813436711014,
      "grad_norm": 2.7531237602233887,
      "learning_rate": 4.995647781153778e-05,
      "loss": 1.6744,
      "step": 18899
    },
    {
      "epoch": 0.6768492488405823,
      "grad_norm": 2.012333869934082,
      "learning_rate": 4.99464360245741e-05,
      "loss": 1.572,
      "step": 18900
    },
    {
      "epoch": 0.6768850609701506,
      "grad_norm": 1.7777117490768433,
      "learning_rate": 4.9936394911016504e-05,
      "loss": 1.5683,
      "step": 18901
    },
    {
      "epoch": 0.6769208730997188,
      "grad_norm": 1.54745614528656,
      "learning_rate": 4.992635447100015e-05,
      "loss": 1.6978,
      "step": 18902
    },
    {
      "epoch": 0.6769566852292872,
      "grad_norm": 2.0074431896209717,
      "learning_rate": 4.9916314704660126e-05,
      "loss": 1.3513,
      "step": 18903
    },
    {
      "epoch": 0.6769924973588555,
      "grad_norm": 1.5793383121490479,
      "learning_rate": 4.9906275612131424e-05,
      "loss": 1.4361,
      "step": 18904
    },
    {
      "epoch": 0.6770283094884237,
      "grad_norm": 2.9052045345306396,
      "learning_rate": 4.9896237193549244e-05,
      "loss": 1.671,
      "step": 18905
    },
    {
      "epoch": 0.677064121617992,
      "grad_norm": 2.1424288749694824,
      "learning_rate": 4.988619944904852e-05,
      "loss": 1.5612,
      "step": 18906
    },
    {
      "epoch": 0.6770999337475603,
      "grad_norm": 1.384803295135498,
      "learning_rate": 4.987616237876438e-05,
      "loss": 1.2562,
      "step": 18907
    },
    {
      "epoch": 0.6771357458771285,
      "grad_norm": 1.6141088008880615,
      "learning_rate": 4.9866125982831745e-05,
      "loss": 1.4878,
      "step": 18908
    },
    {
      "epoch": 0.6771715580066968,
      "grad_norm": 2.246306896209717,
      "learning_rate": 4.9856090261385793e-05,
      "loss": 1.3069,
      "step": 18909
    },
    {
      "epoch": 0.6772073701362652,
      "grad_norm": 1.907278060913086,
      "learning_rate": 4.984605521456146e-05,
      "loss": 1.3377,
      "step": 18910
    },
    {
      "epoch": 0.6772431822658335,
      "grad_norm": 1.491541862487793,
      "learning_rate": 4.983602084249372e-05,
      "loss": 1.7067,
      "step": 18911
    },
    {
      "epoch": 0.6772789943954017,
      "grad_norm": 1.672043800354004,
      "learning_rate": 4.982598714531762e-05,
      "loss": 1.3417,
      "step": 18912
    },
    {
      "epoch": 0.67731480652497,
      "grad_norm": 1.6884132623672485,
      "learning_rate": 4.981595412316815e-05,
      "loss": 1.5376,
      "step": 18913
    },
    {
      "epoch": 0.6773506186545383,
      "grad_norm": 1.6035497188568115,
      "learning_rate": 4.980592177618031e-05,
      "loss": 1.2029,
      "step": 18914
    },
    {
      "epoch": 0.6773864307841065,
      "grad_norm": 1.87796151638031,
      "learning_rate": 4.979589010448902e-05,
      "loss": 1.2361,
      "step": 18915
    },
    {
      "epoch": 0.6774222429136748,
      "grad_norm": 1.6658316850662231,
      "learning_rate": 4.978585910822926e-05,
      "loss": 1.2257,
      "step": 18916
    },
    {
      "epoch": 0.6774580550432432,
      "grad_norm": 1.7389546632766724,
      "learning_rate": 4.977582878753599e-05,
      "loss": 1.6042,
      "step": 18917
    },
    {
      "epoch": 0.6774938671728115,
      "grad_norm": 1.2868759632110596,
      "learning_rate": 4.9765799142544215e-05,
      "loss": 1.479,
      "step": 18918
    },
    {
      "epoch": 0.6775296793023797,
      "grad_norm": 1.504171371459961,
      "learning_rate": 4.975577017338876e-05,
      "loss": 1.2906,
      "step": 18919
    },
    {
      "epoch": 0.677565491431948,
      "grad_norm": 1.5632758140563965,
      "learning_rate": 4.9745741880204613e-05,
      "loss": 1.4864,
      "step": 18920
    },
    {
      "epoch": 0.6776013035615163,
      "grad_norm": 1.4741456508636475,
      "learning_rate": 4.973571426312673e-05,
      "loss": 1.2293,
      "step": 18921
    },
    {
      "epoch": 0.6776371156910845,
      "grad_norm": 1.4424731731414795,
      "learning_rate": 4.9725687322289926e-05,
      "loss": 1.4578,
      "step": 18922
    },
    {
      "epoch": 0.6776729278206528,
      "grad_norm": 1.7455040216445923,
      "learning_rate": 4.971566105782916e-05,
      "loss": 1.6631,
      "step": 18923
    },
    {
      "epoch": 0.6777087399502212,
      "grad_norm": 2.1548445224761963,
      "learning_rate": 4.9705635469879306e-05,
      "loss": 1.4788,
      "step": 18924
    },
    {
      "epoch": 0.6777445520797895,
      "grad_norm": 1.7783231735229492,
      "learning_rate": 4.969561055857529e-05,
      "loss": 1.2761,
      "step": 18925
    },
    {
      "epoch": 0.6777803642093577,
      "grad_norm": 1.5366977453231812,
      "learning_rate": 4.9685586324051915e-05,
      "loss": 1.4405,
      "step": 18926
    },
    {
      "epoch": 0.677816176338926,
      "grad_norm": 1.5162105560302734,
      "learning_rate": 4.967556276644406e-05,
      "loss": 1.2265,
      "step": 18927
    },
    {
      "epoch": 0.6778519884684943,
      "grad_norm": 1.6410696506500244,
      "learning_rate": 4.966553988588665e-05,
      "loss": 1.3447,
      "step": 18928
    },
    {
      "epoch": 0.6778878005980625,
      "grad_norm": 2.0513529777526855,
      "learning_rate": 4.965551768251442e-05,
      "loss": 1.3223,
      "step": 18929
    },
    {
      "epoch": 0.6779236127276308,
      "grad_norm": 1.720237374305725,
      "learning_rate": 4.9645496156462266e-05,
      "loss": 1.5513,
      "step": 18930
    },
    {
      "epoch": 0.6779594248571992,
      "grad_norm": 1.4774359464645386,
      "learning_rate": 4.963547530786501e-05,
      "loss": 1.555,
      "step": 18931
    },
    {
      "epoch": 0.6779952369867674,
      "grad_norm": 1.7184780836105347,
      "learning_rate": 4.962545513685751e-05,
      "loss": 1.3769,
      "step": 18932
    },
    {
      "epoch": 0.6780310491163357,
      "grad_norm": 1.2298649549484253,
      "learning_rate": 4.961543564357449e-05,
      "loss": 1.2371,
      "step": 18933
    },
    {
      "epoch": 0.678066861245904,
      "grad_norm": 1.4467629194259644,
      "learning_rate": 4.9605416828150795e-05,
      "loss": 1.5372,
      "step": 18934
    },
    {
      "epoch": 0.6781026733754723,
      "grad_norm": 1.4205254316329956,
      "learning_rate": 4.959539869072121e-05,
      "loss": 1.6665,
      "step": 18935
    },
    {
      "epoch": 0.6781384855050405,
      "grad_norm": 1.92221999168396,
      "learning_rate": 4.958538123142056e-05,
      "loss": 1.1817,
      "step": 18936
    },
    {
      "epoch": 0.6781742976346088,
      "grad_norm": 1.6638027429580688,
      "learning_rate": 4.957536445038353e-05,
      "loss": 1.3898,
      "step": 18937
    },
    {
      "epoch": 0.6782101097641772,
      "grad_norm": 1.847952961921692,
      "learning_rate": 4.9565348347744934e-05,
      "loss": 1.5683,
      "step": 18938
    },
    {
      "epoch": 0.6782459218937454,
      "grad_norm": 1.6598225831985474,
      "learning_rate": 4.955533292363955e-05,
      "loss": 1.3324,
      "step": 18939
    },
    {
      "epoch": 0.6782817340233137,
      "grad_norm": 1.522696852684021,
      "learning_rate": 4.954531817820206e-05,
      "loss": 1.4984,
      "step": 18940
    },
    {
      "epoch": 0.678317546152882,
      "grad_norm": 1.516022801399231,
      "learning_rate": 4.953530411156724e-05,
      "loss": 1.3997,
      "step": 18941
    },
    {
      "epoch": 0.6783533582824502,
      "grad_norm": 1.4114199876785278,
      "learning_rate": 4.95252907238698e-05,
      "loss": 1.3475,
      "step": 18942
    },
    {
      "epoch": 0.6783891704120185,
      "grad_norm": 1.7230263948440552,
      "learning_rate": 4.95152780152445e-05,
      "loss": 1.5503,
      "step": 18943
    },
    {
      "epoch": 0.6784249825415868,
      "grad_norm": 1.6824709177017212,
      "learning_rate": 4.9505265985825976e-05,
      "loss": 1.6195,
      "step": 18944
    },
    {
      "epoch": 0.6784607946711552,
      "grad_norm": 1.916304588317871,
      "learning_rate": 4.9495254635748975e-05,
      "loss": 1.4234,
      "step": 18945
    },
    {
      "epoch": 0.6784966068007234,
      "grad_norm": 1.438887596130371,
      "learning_rate": 4.948524396514821e-05,
      "loss": 1.351,
      "step": 18946
    },
    {
      "epoch": 0.6785324189302917,
      "grad_norm": 1.4166353940963745,
      "learning_rate": 4.947523397415829e-05,
      "loss": 1.6459,
      "step": 18947
    },
    {
      "epoch": 0.67856823105986,
      "grad_norm": 1.5291413068771362,
      "learning_rate": 4.9465224662913925e-05,
      "loss": 1.2955,
      "step": 18948
    },
    {
      "epoch": 0.6786040431894282,
      "grad_norm": 1.7831827402114868,
      "learning_rate": 4.9455216031549766e-05,
      "loss": 1.5585,
      "step": 18949
    },
    {
      "epoch": 0.6786398553189965,
      "grad_norm": 1.5598150491714478,
      "learning_rate": 4.9445208080200536e-05,
      "loss": 1.3291,
      "step": 18950
    },
    {
      "epoch": 0.6786756674485648,
      "grad_norm": 1.9024678468704224,
      "learning_rate": 4.943520080900076e-05,
      "loss": 1.536,
      "step": 18951
    },
    {
      "epoch": 0.6787114795781332,
      "grad_norm": 1.5680241584777832,
      "learning_rate": 4.9425194218085145e-05,
      "loss": 1.4492,
      "step": 18952
    },
    {
      "epoch": 0.6787472917077014,
      "grad_norm": 1.4497802257537842,
      "learning_rate": 4.94151883075883e-05,
      "loss": 1.5935,
      "step": 18953
    },
    {
      "epoch": 0.6787831038372697,
      "grad_norm": 1.754224419593811,
      "learning_rate": 4.940518307764489e-05,
      "loss": 1.3861,
      "step": 18954
    },
    {
      "epoch": 0.678818915966838,
      "grad_norm": 1.9332152605056763,
      "learning_rate": 4.939517852838944e-05,
      "loss": 1.2178,
      "step": 18955
    },
    {
      "epoch": 0.6788547280964062,
      "grad_norm": 1.623934268951416,
      "learning_rate": 4.938517465995659e-05,
      "loss": 1.5194,
      "step": 18956
    },
    {
      "epoch": 0.6788905402259745,
      "grad_norm": 1.711578369140625,
      "learning_rate": 4.937517147248096e-05,
      "loss": 1.8629,
      "step": 18957
    },
    {
      "epoch": 0.6789263523555428,
      "grad_norm": 2.0465073585510254,
      "learning_rate": 4.936516896609707e-05,
      "loss": 1.6172,
      "step": 18958
    },
    {
      "epoch": 0.6789621644851112,
      "grad_norm": 2.6911144256591797,
      "learning_rate": 4.9355167140939494e-05,
      "loss": 1.484,
      "step": 18959
    },
    {
      "epoch": 0.6789979766146794,
      "grad_norm": 2.0437419414520264,
      "learning_rate": 4.934516599714284e-05,
      "loss": 1.3593,
      "step": 18960
    },
    {
      "epoch": 0.6790337887442477,
      "grad_norm": 1.5595124959945679,
      "learning_rate": 4.933516553484167e-05,
      "loss": 1.9304,
      "step": 18961
    },
    {
      "epoch": 0.679069600873816,
      "grad_norm": 1.714911699295044,
      "learning_rate": 4.9325165754170446e-05,
      "loss": 1.1749,
      "step": 18962
    },
    {
      "epoch": 0.6791054130033842,
      "grad_norm": 1.7814215421676636,
      "learning_rate": 4.931516665526376e-05,
      "loss": 1.5922,
      "step": 18963
    },
    {
      "epoch": 0.6791412251329525,
      "grad_norm": 1.5183871984481812,
      "learning_rate": 4.930516823825616e-05,
      "loss": 1.4136,
      "step": 18964
    },
    {
      "epoch": 0.6791770372625208,
      "grad_norm": 1.6205239295959473,
      "learning_rate": 4.9295170503282095e-05,
      "loss": 1.5224,
      "step": 18965
    },
    {
      "epoch": 0.6792128493920891,
      "grad_norm": 1.9067203998565674,
      "learning_rate": 4.928517345047611e-05,
      "loss": 1.3327,
      "step": 18966
    },
    {
      "epoch": 0.6792486615216574,
      "grad_norm": 1.7289701700210571,
      "learning_rate": 4.927517707997269e-05,
      "loss": 1.3748,
      "step": 18967
    },
    {
      "epoch": 0.6792844736512257,
      "grad_norm": 1.5952321290969849,
      "learning_rate": 4.926518139190638e-05,
      "loss": 1.404,
      "step": 18968
    },
    {
      "epoch": 0.679320285780794,
      "grad_norm": 1.9176799058914185,
      "learning_rate": 4.925518638641157e-05,
      "loss": 1.9199,
      "step": 18969
    },
    {
      "epoch": 0.6793560979103622,
      "grad_norm": 1.8414347171783447,
      "learning_rate": 4.924519206362276e-05,
      "loss": 1.3767,
      "step": 18970
    },
    {
      "epoch": 0.6793919100399305,
      "grad_norm": 1.6279493570327759,
      "learning_rate": 4.9235198423674435e-05,
      "loss": 1.5276,
      "step": 18971
    },
    {
      "epoch": 0.6794277221694988,
      "grad_norm": 1.5713403224945068,
      "learning_rate": 4.9225205466701064e-05,
      "loss": 1.5543,
      "step": 18972
    },
    {
      "epoch": 0.6794635342990671,
      "grad_norm": 1.7283962965011597,
      "learning_rate": 4.9215213192837064e-05,
      "loss": 1.6197,
      "step": 18973
    },
    {
      "epoch": 0.6794993464286354,
      "grad_norm": 1.273970365524292,
      "learning_rate": 4.920522160221679e-05,
      "loss": 1.3018,
      "step": 18974
    },
    {
      "epoch": 0.6795351585582037,
      "grad_norm": 1.8756721019744873,
      "learning_rate": 4.91952306949748e-05,
      "loss": 1.1821,
      "step": 18975
    },
    {
      "epoch": 0.679570970687772,
      "grad_norm": 1.870096206665039,
      "learning_rate": 4.918524047124543e-05,
      "loss": 1.6877,
      "step": 18976
    },
    {
      "epoch": 0.6796067828173402,
      "grad_norm": 1.3298908472061157,
      "learning_rate": 4.9175250931163085e-05,
      "loss": 1.5494,
      "step": 18977
    },
    {
      "epoch": 0.6796425949469085,
      "grad_norm": 1.4009315967559814,
      "learning_rate": 4.916526207486219e-05,
      "loss": 1.4666,
      "step": 18978
    },
    {
      "epoch": 0.6796784070764768,
      "grad_norm": 1.6784093379974365,
      "learning_rate": 4.915527390247716e-05,
      "loss": 1.684,
      "step": 18979
    },
    {
      "epoch": 0.6797142192060451,
      "grad_norm": 4.324677467346191,
      "learning_rate": 4.914528641414233e-05,
      "loss": 1.8545,
      "step": 18980
    },
    {
      "epoch": 0.6797500313356134,
      "grad_norm": 1.4996967315673828,
      "learning_rate": 4.9135299609992004e-05,
      "loss": 1.3009,
      "step": 18981
    },
    {
      "epoch": 0.6797858434651817,
      "grad_norm": 1.5409586429595947,
      "learning_rate": 4.912531349016067e-05,
      "loss": 1.4616,
      "step": 18982
    },
    {
      "epoch": 0.6798216555947499,
      "grad_norm": 1.7398431301116943,
      "learning_rate": 4.911532805478259e-05,
      "loss": 1.2676,
      "step": 18983
    },
    {
      "epoch": 0.6798574677243182,
      "grad_norm": 1.7865056991577148,
      "learning_rate": 4.910534330399219e-05,
      "loss": 1.4283,
      "step": 18984
    },
    {
      "epoch": 0.6798932798538865,
      "grad_norm": 1.7800086736679077,
      "learning_rate": 4.909535923792365e-05,
      "loss": 1.3978,
      "step": 18985
    },
    {
      "epoch": 0.6799290919834547,
      "grad_norm": 1.393025517463684,
      "learning_rate": 4.9085375856711465e-05,
      "loss": 1.5825,
      "step": 18986
    },
    {
      "epoch": 0.6799649041130231,
      "grad_norm": 1.8537535667419434,
      "learning_rate": 4.907539316048985e-05,
      "loss": 1.4,
      "step": 18987
    },
    {
      "epoch": 0.6800007162425914,
      "grad_norm": 2.8984549045562744,
      "learning_rate": 4.906541114939313e-05,
      "loss": 1.4936,
      "step": 18988
    },
    {
      "epoch": 0.6800365283721597,
      "grad_norm": 2.0612361431121826,
      "learning_rate": 4.9055429823555624e-05,
      "loss": 1.5854,
      "step": 18989
    },
    {
      "epoch": 0.6800723405017279,
      "grad_norm": 1.7456159591674805,
      "learning_rate": 4.9045449183111566e-05,
      "loss": 1.1949,
      "step": 18990
    },
    {
      "epoch": 0.6801081526312962,
      "grad_norm": 1.8124316930770874,
      "learning_rate": 4.903546922819531e-05,
      "loss": 1.5129,
      "step": 18991
    },
    {
      "epoch": 0.6801439647608645,
      "grad_norm": 1.6117404699325562,
      "learning_rate": 4.9025489958940985e-05,
      "loss": 1.5364,
      "step": 18992
    },
    {
      "epoch": 0.6801797768904327,
      "grad_norm": 1.6479026079177856,
      "learning_rate": 4.9015511375483026e-05,
      "loss": 1.2438,
      "step": 18993
    },
    {
      "epoch": 0.6802155890200011,
      "grad_norm": 1.5882446765899658,
      "learning_rate": 4.900553347795556e-05,
      "loss": 1.2344,
      "step": 18994
    },
    {
      "epoch": 0.6802514011495694,
      "grad_norm": 1.787760615348816,
      "learning_rate": 4.899555626649289e-05,
      "loss": 1.066,
      "step": 18995
    },
    {
      "epoch": 0.6802872132791377,
      "grad_norm": 2.0816078186035156,
      "learning_rate": 4.898557974122915e-05,
      "loss": 1.528,
      "step": 18996
    },
    {
      "epoch": 0.6803230254087059,
      "grad_norm": 2.4905543327331543,
      "learning_rate": 4.8975603902298704e-05,
      "loss": 1.3881,
      "step": 18997
    },
    {
      "epoch": 0.6803588375382742,
      "grad_norm": 1.876359224319458,
      "learning_rate": 4.896562874983569e-05,
      "loss": 1.1194,
      "step": 18998
    },
    {
      "epoch": 0.6803946496678425,
      "grad_norm": 1.7218693494796753,
      "learning_rate": 4.8955654283974284e-05,
      "loss": 1.4642,
      "step": 18999
    },
    {
      "epoch": 0.6804304617974107,
      "grad_norm": 1.4344909191131592,
      "learning_rate": 4.89456805048487e-05,
      "loss": 1.5587,
      "step": 19000
    },
    {
      "epoch": 0.6804662739269791,
      "grad_norm": 2.262502670288086,
      "learning_rate": 4.893570741259312e-05,
      "loss": 1.7689,
      "step": 19001
    },
    {
      "epoch": 0.6805020860565474,
      "grad_norm": 1.9338393211364746,
      "learning_rate": 4.892573500734179e-05,
      "loss": 1.4208,
      "step": 19002
    },
    {
      "epoch": 0.6805378981861157,
      "grad_norm": 1.5523253679275513,
      "learning_rate": 4.891576328922872e-05,
      "loss": 1.3446,
      "step": 19003
    },
    {
      "epoch": 0.6805737103156839,
      "grad_norm": 2.149492025375366,
      "learning_rate": 4.890579225838824e-05,
      "loss": 1.575,
      "step": 19004
    },
    {
      "epoch": 0.6806095224452522,
      "grad_norm": 1.690125823020935,
      "learning_rate": 4.8895821914954376e-05,
      "loss": 1.3665,
      "step": 19005
    },
    {
      "epoch": 0.6806453345748205,
      "grad_norm": 1.6124178171157837,
      "learning_rate": 4.888585225906136e-05,
      "loss": 1.3594,
      "step": 19006
    },
    {
      "epoch": 0.6806811467043887,
      "grad_norm": 2.0891807079315186,
      "learning_rate": 4.8875883290843214e-05,
      "loss": 1.3593,
      "step": 19007
    },
    {
      "epoch": 0.6807169588339571,
      "grad_norm": 2.1637518405914307,
      "learning_rate": 4.886591501043413e-05,
      "loss": 1.6377,
      "step": 19008
    },
    {
      "epoch": 0.6807527709635254,
      "grad_norm": 1.429513692855835,
      "learning_rate": 4.885594741796823e-05,
      "loss": 1.3786,
      "step": 19009
    },
    {
      "epoch": 0.6807885830930936,
      "grad_norm": 1.508383870124817,
      "learning_rate": 4.884598051357955e-05,
      "loss": 1.5587,
      "step": 19010
    },
    {
      "epoch": 0.6808243952226619,
      "grad_norm": 1.646700143814087,
      "learning_rate": 4.883601429740222e-05,
      "loss": 1.3783,
      "step": 19011
    },
    {
      "epoch": 0.6808602073522302,
      "grad_norm": 1.7303789854049683,
      "learning_rate": 4.882604876957032e-05,
      "loss": 1.667,
      "step": 19012
    },
    {
      "epoch": 0.6808960194817985,
      "grad_norm": 1.8976233005523682,
      "learning_rate": 4.881608393021796e-05,
      "loss": 1.5103,
      "step": 19013
    },
    {
      "epoch": 0.6809318316113667,
      "grad_norm": 1.9167366027832031,
      "learning_rate": 4.880611977947909e-05,
      "loss": 1.4182,
      "step": 19014
    },
    {
      "epoch": 0.6809676437409351,
      "grad_norm": 1.572396993637085,
      "learning_rate": 4.879615631748793e-05,
      "loss": 1.7171,
      "step": 19015
    },
    {
      "epoch": 0.6810034558705034,
      "grad_norm": 1.5460922718048096,
      "learning_rate": 4.8786193544378424e-05,
      "loss": 1.5438,
      "step": 19016
    },
    {
      "epoch": 0.6810392680000716,
      "grad_norm": 1.4947601556777954,
      "learning_rate": 4.8776231460284595e-05,
      "loss": 1.655,
      "step": 19017
    },
    {
      "epoch": 0.6810750801296399,
      "grad_norm": 1.6060175895690918,
      "learning_rate": 4.876627006534049e-05,
      "loss": 1.5582,
      "step": 19018
    },
    {
      "epoch": 0.6811108922592082,
      "grad_norm": 1.5002024173736572,
      "learning_rate": 4.8756309359680145e-05,
      "loss": 1.3706,
      "step": 19019
    },
    {
      "epoch": 0.6811467043887764,
      "grad_norm": 1.6578645706176758,
      "learning_rate": 4.874634934343759e-05,
      "loss": 1.451,
      "step": 19020
    },
    {
      "epoch": 0.6811825165183447,
      "grad_norm": 1.4593560695648193,
      "learning_rate": 4.873639001674676e-05,
      "loss": 1.2581,
      "step": 19021
    },
    {
      "epoch": 0.6812183286479131,
      "grad_norm": 1.5626795291900635,
      "learning_rate": 4.872643137974167e-05,
      "loss": 1.5399,
      "step": 19022
    },
    {
      "epoch": 0.6812541407774814,
      "grad_norm": 1.6952016353607178,
      "learning_rate": 4.87164734325563e-05,
      "loss": 1.4,
      "step": 19023
    },
    {
      "epoch": 0.6812899529070496,
      "grad_norm": 1.9732818603515625,
      "learning_rate": 4.870651617532468e-05,
      "loss": 1.5631,
      "step": 19024
    },
    {
      "epoch": 0.6813257650366179,
      "grad_norm": 1.7756799459457397,
      "learning_rate": 4.869655960818068e-05,
      "loss": 1.2446,
      "step": 19025
    },
    {
      "epoch": 0.6813615771661862,
      "grad_norm": 1.959659457206726,
      "learning_rate": 4.868660373125829e-05,
      "loss": 1.4592,
      "step": 19026
    },
    {
      "epoch": 0.6813973892957544,
      "grad_norm": 2.5035407543182373,
      "learning_rate": 4.8676648544691495e-05,
      "loss": 1.5678,
      "step": 19027
    },
    {
      "epoch": 0.6814332014253227,
      "grad_norm": 1.5704766511917114,
      "learning_rate": 4.866669404861416e-05,
      "loss": 1.5047,
      "step": 19028
    },
    {
      "epoch": 0.6814690135548911,
      "grad_norm": 2.848478078842163,
      "learning_rate": 4.8656740243160236e-05,
      "loss": 1.8893,
      "step": 19029
    },
    {
      "epoch": 0.6815048256844594,
      "grad_norm": 1.4593943357467651,
      "learning_rate": 4.864678712846365e-05,
      "loss": 1.2788,
      "step": 19030
    },
    {
      "epoch": 0.6815406378140276,
      "grad_norm": 1.6539605855941772,
      "learning_rate": 4.863683470465833e-05,
      "loss": 1.2998,
      "step": 19031
    },
    {
      "epoch": 0.6815764499435959,
      "grad_norm": 1.4724090099334717,
      "learning_rate": 4.862688297187812e-05,
      "loss": 1.5732,
      "step": 19032
    },
    {
      "epoch": 0.6816122620731642,
      "grad_norm": 2.0338592529296875,
      "learning_rate": 4.8616931930256926e-05,
      "loss": 1.4792,
      "step": 19033
    },
    {
      "epoch": 0.6816480742027324,
      "grad_norm": 1.6050416231155396,
      "learning_rate": 4.860698157992867e-05,
      "loss": 1.0027,
      "step": 19034
    },
    {
      "epoch": 0.6816838863323007,
      "grad_norm": 1.8929771184921265,
      "learning_rate": 4.859703192102715e-05,
      "loss": 1.3524,
      "step": 19035
    },
    {
      "epoch": 0.6817196984618691,
      "grad_norm": 1.4023357629776,
      "learning_rate": 4.858708295368626e-05,
      "loss": 1.4543,
      "step": 19036
    },
    {
      "epoch": 0.6817555105914374,
      "grad_norm": 1.6556355953216553,
      "learning_rate": 4.857713467803985e-05,
      "loss": 1.6469,
      "step": 19037
    },
    {
      "epoch": 0.6817913227210056,
      "grad_norm": 2.0519371032714844,
      "learning_rate": 4.85671870942218e-05,
      "loss": 1.5793,
      "step": 19038
    },
    {
      "epoch": 0.6818271348505739,
      "grad_norm": 1.6347476243972778,
      "learning_rate": 4.855724020236586e-05,
      "loss": 1.6161,
      "step": 19039
    },
    {
      "epoch": 0.6818629469801422,
      "grad_norm": 1.3786358833312988,
      "learning_rate": 4.854729400260591e-05,
      "loss": 1.5669,
      "step": 19040
    },
    {
      "epoch": 0.6818987591097104,
      "grad_norm": 1.5543736219406128,
      "learning_rate": 4.853734849507574e-05,
      "loss": 1.5422,
      "step": 19041
    },
    {
      "epoch": 0.6819345712392787,
      "grad_norm": 1.4965298175811768,
      "learning_rate": 4.8527403679909214e-05,
      "loss": 1.1551,
      "step": 19042
    },
    {
      "epoch": 0.6819703833688471,
      "grad_norm": 1.4858630895614624,
      "learning_rate": 4.851745955724002e-05,
      "loss": 1.5663,
      "step": 19043
    },
    {
      "epoch": 0.6820061954984153,
      "grad_norm": 1.4483484029769897,
      "learning_rate": 4.8507516127202014e-05,
      "loss": 1.1364,
      "step": 19044
    },
    {
      "epoch": 0.6820420076279836,
      "grad_norm": 1.929543375968933,
      "learning_rate": 4.849757338992898e-05,
      "loss": 1.5264,
      "step": 19045
    },
    {
      "epoch": 0.6820778197575519,
      "grad_norm": 1.2829221487045288,
      "learning_rate": 4.848763134555465e-05,
      "loss": 1.1193,
      "step": 19046
    },
    {
      "epoch": 0.6821136318871202,
      "grad_norm": 1.5956387519836426,
      "learning_rate": 4.847768999421277e-05,
      "loss": 1.2601,
      "step": 19047
    },
    {
      "epoch": 0.6821494440166884,
      "grad_norm": 2.0086724758148193,
      "learning_rate": 4.8467749336037124e-05,
      "loss": 1.1854,
      "step": 19048
    },
    {
      "epoch": 0.6821852561462567,
      "grad_norm": 1.4452983140945435,
      "learning_rate": 4.8457809371161476e-05,
      "loss": 1.3118,
      "step": 19049
    },
    {
      "epoch": 0.6822210682758251,
      "grad_norm": 1.4968582391738892,
      "learning_rate": 4.844787009971949e-05,
      "loss": 1.4051,
      "step": 19050
    },
    {
      "epoch": 0.6822568804053933,
      "grad_norm": 1.3636406660079956,
      "learning_rate": 4.8437931521844894e-05,
      "loss": 1.3358,
      "step": 19051
    },
    {
      "epoch": 0.6822926925349616,
      "grad_norm": 1.9346957206726074,
      "learning_rate": 4.8427993637671474e-05,
      "loss": 1.9677,
      "step": 19052
    },
    {
      "epoch": 0.6823285046645299,
      "grad_norm": 1.4876221418380737,
      "learning_rate": 4.841805644733283e-05,
      "loss": 1.6532,
      "step": 19053
    },
    {
      "epoch": 0.6823643167940981,
      "grad_norm": 1.683264136314392,
      "learning_rate": 4.8408119950962704e-05,
      "loss": 1.6318,
      "step": 19054
    },
    {
      "epoch": 0.6824001289236664,
      "grad_norm": 1.3608731031417847,
      "learning_rate": 4.839818414869477e-05,
      "loss": 1.3656,
      "step": 19055
    },
    {
      "epoch": 0.6824359410532347,
      "grad_norm": 1.9160677194595337,
      "learning_rate": 4.8388249040662744e-05,
      "loss": 1.5936,
      "step": 19056
    },
    {
      "epoch": 0.6824717531828031,
      "grad_norm": 1.7479875087738037,
      "learning_rate": 4.8378314627000224e-05,
      "loss": 1.7994,
      "step": 19057
    },
    {
      "epoch": 0.6825075653123713,
      "grad_norm": 2.653571367263794,
      "learning_rate": 4.836838090784088e-05,
      "loss": 1.5861,
      "step": 19058
    },
    {
      "epoch": 0.6825433774419396,
      "grad_norm": 1.605258584022522,
      "learning_rate": 4.835844788331839e-05,
      "loss": 1.5436,
      "step": 19059
    },
    {
      "epoch": 0.6825791895715079,
      "grad_norm": 1.4455996751785278,
      "learning_rate": 4.8348515553566396e-05,
      "loss": 1.2487,
      "step": 19060
    },
    {
      "epoch": 0.6826150017010761,
      "grad_norm": 1.330987572669983,
      "learning_rate": 4.833858391871846e-05,
      "loss": 1.4588,
      "step": 19061
    },
    {
      "epoch": 0.6826508138306444,
      "grad_norm": 2.3444483280181885,
      "learning_rate": 4.832865297890825e-05,
      "loss": 1.5377,
      "step": 19062
    },
    {
      "epoch": 0.6826866259602127,
      "grad_norm": 1.7140589952468872,
      "learning_rate": 4.83187227342694e-05,
      "loss": 1.2662,
      "step": 19063
    },
    {
      "epoch": 0.6827224380897811,
      "grad_norm": 1.2725976705551147,
      "learning_rate": 4.830879318493542e-05,
      "loss": 1.3755,
      "step": 19064
    },
    {
      "epoch": 0.6827582502193493,
      "grad_norm": 1.4281116724014282,
      "learning_rate": 4.829886433103995e-05,
      "loss": 1.4356,
      "step": 19065
    },
    {
      "epoch": 0.6827940623489176,
      "grad_norm": 1.4100066423416138,
      "learning_rate": 4.828893617271658e-05,
      "loss": 1.3519,
      "step": 19066
    },
    {
      "epoch": 0.6828298744784859,
      "grad_norm": 1.5357187986373901,
      "learning_rate": 4.8279008710098916e-05,
      "loss": 1.6721,
      "step": 19067
    },
    {
      "epoch": 0.6828656866080541,
      "grad_norm": 1.6073108911514282,
      "learning_rate": 4.8269081943320424e-05,
      "loss": 1.4553,
      "step": 19068
    },
    {
      "epoch": 0.6829014987376224,
      "grad_norm": 1.8977751731872559,
      "learning_rate": 4.825915587251472e-05,
      "loss": 1.4387,
      "step": 19069
    },
    {
      "epoch": 0.6829373108671907,
      "grad_norm": 1.570206642150879,
      "learning_rate": 4.824923049781536e-05,
      "loss": 1.2027,
      "step": 19070
    },
    {
      "epoch": 0.682973122996759,
      "grad_norm": 1.6373735666275024,
      "learning_rate": 4.8239305819355805e-05,
      "loss": 1.7076,
      "step": 19071
    },
    {
      "epoch": 0.6830089351263273,
      "grad_norm": 2.1128671169281006,
      "learning_rate": 4.822938183726967e-05,
      "loss": 1.4418,
      "step": 19072
    },
    {
      "epoch": 0.6830447472558956,
      "grad_norm": 1.6894068717956543,
      "learning_rate": 4.821945855169035e-05,
      "loss": 1.7592,
      "step": 19073
    },
    {
      "epoch": 0.6830805593854639,
      "grad_norm": 1.6356310844421387,
      "learning_rate": 4.8209535962751494e-05,
      "loss": 1.4973,
      "step": 19074
    },
    {
      "epoch": 0.6831163715150321,
      "grad_norm": 1.7310285568237305,
      "learning_rate": 4.81996140705865e-05,
      "loss": 1.5372,
      "step": 19075
    },
    {
      "epoch": 0.6831521836446004,
      "grad_norm": 1.616203784942627,
      "learning_rate": 4.8189692875328864e-05,
      "loss": 1.6467,
      "step": 19076
    },
    {
      "epoch": 0.6831879957741687,
      "grad_norm": 1.7261005640029907,
      "learning_rate": 4.817977237711213e-05,
      "loss": 1.5143,
      "step": 19077
    },
    {
      "epoch": 0.683223807903737,
      "grad_norm": 1.4400568008422852,
      "learning_rate": 4.816985257606967e-05,
      "loss": 1.3393,
      "step": 19078
    },
    {
      "epoch": 0.6832596200333053,
      "grad_norm": 2.898552179336548,
      "learning_rate": 4.815993347233503e-05,
      "loss": 1.7106,
      "step": 19079
    },
    {
      "epoch": 0.6832954321628736,
      "grad_norm": 1.6326524019241333,
      "learning_rate": 4.8150015066041545e-05,
      "loss": 1.2989,
      "step": 19080
    },
    {
      "epoch": 0.6833312442924419,
      "grad_norm": 1.612429141998291,
      "learning_rate": 4.814009735732279e-05,
      "loss": 1.4414,
      "step": 19081
    },
    {
      "epoch": 0.6833670564220101,
      "grad_norm": 1.492590308189392,
      "learning_rate": 4.8130180346312105e-05,
      "loss": 1.5981,
      "step": 19082
    },
    {
      "epoch": 0.6834028685515784,
      "grad_norm": 1.8465263843536377,
      "learning_rate": 4.812026403314297e-05,
      "loss": 1.37,
      "step": 19083
    },
    {
      "epoch": 0.6834386806811467,
      "grad_norm": 1.8145172595977783,
      "learning_rate": 4.811034841794868e-05,
      "loss": 1.3823,
      "step": 19084
    },
    {
      "epoch": 0.683474492810715,
      "grad_norm": 1.4729621410369873,
      "learning_rate": 4.8100433500862794e-05,
      "loss": 1.4076,
      "step": 19085
    },
    {
      "epoch": 0.6835103049402833,
      "grad_norm": 2.7632548809051514,
      "learning_rate": 4.809051928201864e-05,
      "loss": 1.2546,
      "step": 19086
    },
    {
      "epoch": 0.6835461170698516,
      "grad_norm": 1.296805739402771,
      "learning_rate": 4.808060576154951e-05,
      "loss": 1.1367,
      "step": 19087
    },
    {
      "epoch": 0.6835819291994198,
      "grad_norm": 1.7383794784545898,
      "learning_rate": 4.8070692939588934e-05,
      "loss": 1.4106,
      "step": 19088
    },
    {
      "epoch": 0.6836177413289881,
      "grad_norm": 2.4695076942443848,
      "learning_rate": 4.8060780816270165e-05,
      "loss": 1.3039,
      "step": 19089
    },
    {
      "epoch": 0.6836535534585564,
      "grad_norm": 1.2985374927520752,
      "learning_rate": 4.805086939172663e-05,
      "loss": 1.5119,
      "step": 19090
    },
    {
      "epoch": 0.6836893655881247,
      "grad_norm": 1.8521593809127808,
      "learning_rate": 4.804095866609156e-05,
      "loss": 1.4941,
      "step": 19091
    },
    {
      "epoch": 0.683725177717693,
      "grad_norm": 1.4810516834259033,
      "learning_rate": 4.803104863949844e-05,
      "loss": 1.5291,
      "step": 19092
    },
    {
      "epoch": 0.6837609898472613,
      "grad_norm": 1.5847200155258179,
      "learning_rate": 4.80211393120805e-05,
      "loss": 1.6587,
      "step": 19093
    },
    {
      "epoch": 0.6837968019768296,
      "grad_norm": 1.4615356922149658,
      "learning_rate": 4.801123068397111e-05,
      "loss": 1.2841,
      "step": 19094
    },
    {
      "epoch": 0.6838326141063978,
      "grad_norm": 2.2419803142547607,
      "learning_rate": 4.800132275530351e-05,
      "loss": 1.2867,
      "step": 19095
    },
    {
      "epoch": 0.6838684262359661,
      "grad_norm": 1.2413926124572754,
      "learning_rate": 4.799141552621105e-05,
      "loss": 1.3555,
      "step": 19096
    },
    {
      "epoch": 0.6839042383655344,
      "grad_norm": 1.478020191192627,
      "learning_rate": 4.798150899682704e-05,
      "loss": 1.4722,
      "step": 19097
    },
    {
      "epoch": 0.6839400504951026,
      "grad_norm": 1.8806427717208862,
      "learning_rate": 4.79716031672847e-05,
      "loss": 1.7017,
      "step": 19098
    },
    {
      "epoch": 0.683975862624671,
      "grad_norm": 1.7698049545288086,
      "learning_rate": 4.7961698037717306e-05,
      "loss": 1.4021,
      "step": 19099
    },
    {
      "epoch": 0.6840116747542393,
      "grad_norm": 1.7766374349594116,
      "learning_rate": 4.795179360825815e-05,
      "loss": 1.5161,
      "step": 19100
    },
    {
      "epoch": 0.6840474868838076,
      "grad_norm": 1.346203088760376,
      "learning_rate": 4.794188987904051e-05,
      "loss": 1.3395,
      "step": 19101
    },
    {
      "epoch": 0.6840832990133758,
      "grad_norm": 1.899402141571045,
      "learning_rate": 4.793198685019753e-05,
      "loss": 1.436,
      "step": 19102
    },
    {
      "epoch": 0.6841191111429441,
      "grad_norm": 1.439682126045227,
      "learning_rate": 4.7922084521862565e-05,
      "loss": 1.3542,
      "step": 19103
    },
    {
      "epoch": 0.6841549232725124,
      "grad_norm": 1.2503219842910767,
      "learning_rate": 4.791218289416879e-05,
      "loss": 1.2213,
      "step": 19104
    },
    {
      "epoch": 0.6841907354020806,
      "grad_norm": 1.6100847721099854,
      "learning_rate": 4.790228196724935e-05,
      "loss": 1.4521,
      "step": 19105
    },
    {
      "epoch": 0.684226547531649,
      "grad_norm": 1.7780996561050415,
      "learning_rate": 4.789238174123751e-05,
      "loss": 1.4513,
      "step": 19106
    },
    {
      "epoch": 0.6842623596612173,
      "grad_norm": 3.03469181060791,
      "learning_rate": 4.788248221626647e-05,
      "loss": 1.3681,
      "step": 19107
    },
    {
      "epoch": 0.6842981717907856,
      "grad_norm": 1.8146162033081055,
      "learning_rate": 4.7872583392469436e-05,
      "loss": 1.3714,
      "step": 19108
    },
    {
      "epoch": 0.6843339839203538,
      "grad_norm": 1.5532610416412354,
      "learning_rate": 4.786268526997951e-05,
      "loss": 1.4951,
      "step": 19109
    },
    {
      "epoch": 0.6843697960499221,
      "grad_norm": 1.6222275495529175,
      "learning_rate": 4.7852787848929916e-05,
      "loss": 1.6106,
      "step": 19110
    },
    {
      "epoch": 0.6844056081794904,
      "grad_norm": 1.9357314109802246,
      "learning_rate": 4.7842891129453784e-05,
      "loss": 1.4879,
      "step": 19111
    },
    {
      "epoch": 0.6844414203090586,
      "grad_norm": 2.5077223777770996,
      "learning_rate": 4.783299511168432e-05,
      "loss": 1.3873,
      "step": 19112
    },
    {
      "epoch": 0.684477232438627,
      "grad_norm": 2.1032609939575195,
      "learning_rate": 4.7823099795754566e-05,
      "loss": 1.42,
      "step": 19113
    },
    {
      "epoch": 0.6845130445681953,
      "grad_norm": 1.413138508796692,
      "learning_rate": 4.781320518179772e-05,
      "loss": 1.5738,
      "step": 19114
    },
    {
      "epoch": 0.6845488566977636,
      "grad_norm": 1.6410788297653198,
      "learning_rate": 4.780331126994691e-05,
      "loss": 1.6905,
      "step": 19115
    },
    {
      "epoch": 0.6845846688273318,
      "grad_norm": 1.455068826675415,
      "learning_rate": 4.779341806033517e-05,
      "loss": 1.2427,
      "step": 19116
    },
    {
      "epoch": 0.6846204809569001,
      "grad_norm": 1.6743617057800293,
      "learning_rate": 4.778352555309565e-05,
      "loss": 1.2889,
      "step": 19117
    },
    {
      "epoch": 0.6846562930864684,
      "grad_norm": 1.5681772232055664,
      "learning_rate": 4.777363374836146e-05,
      "loss": 1.461,
      "step": 19118
    },
    {
      "epoch": 0.6846921052160366,
      "grad_norm": 1.5241618156433105,
      "learning_rate": 4.7763742646265674e-05,
      "loss": 1.6756,
      "step": 19119
    },
    {
      "epoch": 0.684727917345605,
      "grad_norm": 1.7373355627059937,
      "learning_rate": 4.7753852246941335e-05,
      "loss": 1.3074,
      "step": 19120
    },
    {
      "epoch": 0.6847637294751733,
      "grad_norm": 1.4865138530731201,
      "learning_rate": 4.774396255052151e-05,
      "loss": 1.3415,
      "step": 19121
    },
    {
      "epoch": 0.6847995416047415,
      "grad_norm": 1.9230724573135376,
      "learning_rate": 4.773407355713929e-05,
      "loss": 1.6522,
      "step": 19122
    },
    {
      "epoch": 0.6848353537343098,
      "grad_norm": 1.556607961654663,
      "learning_rate": 4.7724185266927666e-05,
      "loss": 1.5193,
      "step": 19123
    },
    {
      "epoch": 0.6848711658638781,
      "grad_norm": 1.8414274454116821,
      "learning_rate": 4.7714297680019704e-05,
      "loss": 1.4167,
      "step": 19124
    },
    {
      "epoch": 0.6849069779934464,
      "grad_norm": 1.8002582788467407,
      "learning_rate": 4.770441079654841e-05,
      "loss": 1.4511,
      "step": 19125
    },
    {
      "epoch": 0.6849427901230146,
      "grad_norm": 1.7834447622299194,
      "learning_rate": 4.7694524616646865e-05,
      "loss": 1.7016,
      "step": 19126
    },
    {
      "epoch": 0.684978602252583,
      "grad_norm": 1.2345691919326782,
      "learning_rate": 4.768463914044797e-05,
      "loss": 1.3703,
      "step": 19127
    },
    {
      "epoch": 0.6850144143821513,
      "grad_norm": 1.528149127960205,
      "learning_rate": 4.767475436808478e-05,
      "loss": 1.4703,
      "step": 19128
    },
    {
      "epoch": 0.6850502265117195,
      "grad_norm": 1.437954306602478,
      "learning_rate": 4.766487029969028e-05,
      "loss": 1.4768,
      "step": 19129
    },
    {
      "epoch": 0.6850860386412878,
      "grad_norm": 1.682350754737854,
      "learning_rate": 4.765498693539747e-05,
      "loss": 1.3122,
      "step": 19130
    },
    {
      "epoch": 0.6851218507708561,
      "grad_norm": 2.119572401046753,
      "learning_rate": 4.764510427533926e-05,
      "loss": 1.4844,
      "step": 19131
    },
    {
      "epoch": 0.6851576629004243,
      "grad_norm": 2.683337926864624,
      "learning_rate": 4.763522231964864e-05,
      "loss": 1.2952,
      "step": 19132
    },
    {
      "epoch": 0.6851934750299926,
      "grad_norm": 1.712811827659607,
      "learning_rate": 4.76253410684586e-05,
      "loss": 1.5528,
      "step": 19133
    },
    {
      "epoch": 0.685229287159561,
      "grad_norm": 1.5293190479278564,
      "learning_rate": 4.761546052190199e-05,
      "loss": 1.2965,
      "step": 19134
    },
    {
      "epoch": 0.6852650992891293,
      "grad_norm": 1.6144920587539673,
      "learning_rate": 4.7605580680111785e-05,
      "loss": 1.313,
      "step": 19135
    },
    {
      "epoch": 0.6853009114186975,
      "grad_norm": 1.2514257431030273,
      "learning_rate": 4.7595701543220916e-05,
      "loss": 1.3082,
      "step": 19136
    },
    {
      "epoch": 0.6853367235482658,
      "grad_norm": 1.5040967464447021,
      "learning_rate": 4.758582311136231e-05,
      "loss": 1.9045,
      "step": 19137
    },
    {
      "epoch": 0.6853725356778341,
      "grad_norm": 1.5086262226104736,
      "learning_rate": 4.757594538466883e-05,
      "loss": 1.5548,
      "step": 19138
    },
    {
      "epoch": 0.6854083478074023,
      "grad_norm": 1.7702431678771973,
      "learning_rate": 4.756606836327337e-05,
      "loss": 1.2219,
      "step": 19139
    },
    {
      "epoch": 0.6854441599369706,
      "grad_norm": 1.6345676183700562,
      "learning_rate": 4.755619204730886e-05,
      "loss": 1.534,
      "step": 19140
    },
    {
      "epoch": 0.685479972066539,
      "grad_norm": 1.7369307279586792,
      "learning_rate": 4.75463164369081e-05,
      "loss": 1.377,
      "step": 19141
    },
    {
      "epoch": 0.6855157841961073,
      "grad_norm": 2.0801753997802734,
      "learning_rate": 4.7536441532204e-05,
      "loss": 1.633,
      "step": 19142
    },
    {
      "epoch": 0.6855515963256755,
      "grad_norm": 1.7809818983078003,
      "learning_rate": 4.752656733332941e-05,
      "loss": 1.6268,
      "step": 19143
    },
    {
      "epoch": 0.6855874084552438,
      "grad_norm": 2.1582162380218506,
      "learning_rate": 4.751669384041719e-05,
      "loss": 1.4487,
      "step": 19144
    },
    {
      "epoch": 0.6856232205848121,
      "grad_norm": 1.6797009706497192,
      "learning_rate": 4.750682105360014e-05,
      "loss": 1.5636,
      "step": 19145
    },
    {
      "epoch": 0.6856590327143803,
      "grad_norm": 2.4573843479156494,
      "learning_rate": 4.749694897301108e-05,
      "loss": 1.4913,
      "step": 19146
    },
    {
      "epoch": 0.6856948448439486,
      "grad_norm": 1.3981280326843262,
      "learning_rate": 4.7487077598782856e-05,
      "loss": 1.392,
      "step": 19147
    },
    {
      "epoch": 0.685730656973517,
      "grad_norm": 1.4135205745697021,
      "learning_rate": 4.747720693104831e-05,
      "loss": 1.4138,
      "step": 19148
    },
    {
      "epoch": 0.6857664691030853,
      "grad_norm": 1.4483755826950073,
      "learning_rate": 4.7467336969940156e-05,
      "loss": 1.1195,
      "step": 19149
    },
    {
      "epoch": 0.6858022812326535,
      "grad_norm": 1.7879372835159302,
      "learning_rate": 4.745746771559122e-05,
      "loss": 1.1896,
      "step": 19150
    },
    {
      "epoch": 0.6858380933622218,
      "grad_norm": 1.420569896697998,
      "learning_rate": 4.744759916813432e-05,
      "loss": 1.5825,
      "step": 19151
    },
    {
      "epoch": 0.6858739054917901,
      "grad_norm": 1.677254319190979,
      "learning_rate": 4.743773132770214e-05,
      "loss": 0.9403,
      "step": 19152
    },
    {
      "epoch": 0.6859097176213583,
      "grad_norm": 1.9434819221496582,
      "learning_rate": 4.7427864194427484e-05,
      "loss": 1.7483,
      "step": 19153
    },
    {
      "epoch": 0.6859455297509266,
      "grad_norm": 1.6031337976455688,
      "learning_rate": 4.74179977684431e-05,
      "loss": 1.3214,
      "step": 19154
    },
    {
      "epoch": 0.685981341880495,
      "grad_norm": 1.5116249322891235,
      "learning_rate": 4.740813204988178e-05,
      "loss": 1.3434,
      "step": 19155
    },
    {
      "epoch": 0.6860171540100632,
      "grad_norm": 2.093729257583618,
      "learning_rate": 4.739826703887616e-05,
      "loss": 1.4053,
      "step": 19156
    },
    {
      "epoch": 0.6860529661396315,
      "grad_norm": 1.526672124862671,
      "learning_rate": 4.7388402735559014e-05,
      "loss": 1.429,
      "step": 19157
    },
    {
      "epoch": 0.6860887782691998,
      "grad_norm": 1.9945696592330933,
      "learning_rate": 4.737853914006307e-05,
      "loss": 1.2103,
      "step": 19158
    },
    {
      "epoch": 0.686124590398768,
      "grad_norm": 1.6721898317337036,
      "learning_rate": 4.736867625252097e-05,
      "loss": 1.1878,
      "step": 19159
    },
    {
      "epoch": 0.6861604025283363,
      "grad_norm": 1.792447805404663,
      "learning_rate": 4.735881407306545e-05,
      "loss": 1.1916,
      "step": 19160
    },
    {
      "epoch": 0.6861962146579046,
      "grad_norm": 3.155874729156494,
      "learning_rate": 4.734895260182918e-05,
      "loss": 1.503,
      "step": 19161
    },
    {
      "epoch": 0.686232026787473,
      "grad_norm": 1.5070594549179077,
      "learning_rate": 4.733909183894487e-05,
      "loss": 1.1776,
      "step": 19162
    },
    {
      "epoch": 0.6862678389170412,
      "grad_norm": 1.7831507921218872,
      "learning_rate": 4.732923178454512e-05,
      "loss": 1.4696,
      "step": 19163
    },
    {
      "epoch": 0.6863036510466095,
      "grad_norm": 1.283743977546692,
      "learning_rate": 4.731937243876262e-05,
      "loss": 1.2818,
      "step": 19164
    },
    {
      "epoch": 0.6863394631761778,
      "grad_norm": 2.2443976402282715,
      "learning_rate": 4.730951380173e-05,
      "loss": 1.6624,
      "step": 19165
    },
    {
      "epoch": 0.686375275305746,
      "grad_norm": 1.4468728303909302,
      "learning_rate": 4.729965587357995e-05,
      "loss": 1.0889,
      "step": 19166
    },
    {
      "epoch": 0.6864110874353143,
      "grad_norm": 1.9218125343322754,
      "learning_rate": 4.728979865444505e-05,
      "loss": 1.3015,
      "step": 19167
    },
    {
      "epoch": 0.6864468995648826,
      "grad_norm": 1.458131194114685,
      "learning_rate": 4.7279942144457847e-05,
      "loss": 1.6985,
      "step": 19168
    },
    {
      "epoch": 0.686482711694451,
      "grad_norm": 1.8317525386810303,
      "learning_rate": 4.7270086343751085e-05,
      "loss": 1.5711,
      "step": 19169
    },
    {
      "epoch": 0.6865185238240192,
      "grad_norm": 2.029881715774536,
      "learning_rate": 4.7260231252457265e-05,
      "loss": 1.7153,
      "step": 19170
    },
    {
      "epoch": 0.6865543359535875,
      "grad_norm": 1.9590765237808228,
      "learning_rate": 4.7250376870709e-05,
      "loss": 1.1425,
      "step": 19171
    },
    {
      "epoch": 0.6865901480831558,
      "grad_norm": 1.7559995651245117,
      "learning_rate": 4.7240523198638875e-05,
      "loss": 1.3486,
      "step": 19172
    },
    {
      "epoch": 0.686625960212724,
      "grad_norm": 2.3266706466674805,
      "learning_rate": 4.723067023637949e-05,
      "loss": 1.6996,
      "step": 19173
    },
    {
      "epoch": 0.6866617723422923,
      "grad_norm": 1.5278242826461792,
      "learning_rate": 4.722081798406337e-05,
      "loss": 1.6506,
      "step": 19174
    },
    {
      "epoch": 0.6866975844718606,
      "grad_norm": 2.3450798988342285,
      "learning_rate": 4.721096644182299e-05,
      "loss": 1.345,
      "step": 19175
    },
    {
      "epoch": 0.686733396601429,
      "grad_norm": 2.0274198055267334,
      "learning_rate": 4.720111560979104e-05,
      "loss": 1.4916,
      "step": 19176
    },
    {
      "epoch": 0.6867692087309972,
      "grad_norm": 1.3534672260284424,
      "learning_rate": 4.719126548809993e-05,
      "loss": 1.4593,
      "step": 19177
    },
    {
      "epoch": 0.6868050208605655,
      "grad_norm": 1.6133003234863281,
      "learning_rate": 4.7181416076882266e-05,
      "loss": 1.4837,
      "step": 19178
    },
    {
      "epoch": 0.6868408329901338,
      "grad_norm": 1.8048996925354004,
      "learning_rate": 4.7171567376270443e-05,
      "loss": 1.5241,
      "step": 19179
    },
    {
      "epoch": 0.686876645119702,
      "grad_norm": 1.4059399366378784,
      "learning_rate": 4.716171938639711e-05,
      "loss": 1.4154,
      "step": 19180
    },
    {
      "epoch": 0.6869124572492703,
      "grad_norm": 1.9644883871078491,
      "learning_rate": 4.715187210739466e-05,
      "loss": 1.2348,
      "step": 19181
    },
    {
      "epoch": 0.6869482693788386,
      "grad_norm": 2.089815378189087,
      "learning_rate": 4.714202553939562e-05,
      "loss": 1.4767,
      "step": 19182
    },
    {
      "epoch": 0.686984081508407,
      "grad_norm": 1.4828203916549683,
      "learning_rate": 4.713217968253242e-05,
      "loss": 1.6034,
      "step": 19183
    },
    {
      "epoch": 0.6870198936379752,
      "grad_norm": 1.8886553049087524,
      "learning_rate": 4.712233453693754e-05,
      "loss": 1.5413,
      "step": 19184
    },
    {
      "epoch": 0.6870557057675435,
      "grad_norm": 2.037841796875,
      "learning_rate": 4.711249010274349e-05,
      "loss": 1.6245,
      "step": 19185
    },
    {
      "epoch": 0.6870915178971118,
      "grad_norm": 1.8798308372497559,
      "learning_rate": 4.710264638008258e-05,
      "loss": 1.6074,
      "step": 19186
    },
    {
      "epoch": 0.68712733002668,
      "grad_norm": 2.2628753185272217,
      "learning_rate": 4.709280336908741e-05,
      "loss": 1.5401,
      "step": 19187
    },
    {
      "epoch": 0.6871631421562483,
      "grad_norm": 1.5651167631149292,
      "learning_rate": 4.7082961069890284e-05,
      "loss": 1.5082,
      "step": 19188
    },
    {
      "epoch": 0.6871989542858166,
      "grad_norm": 1.807701826095581,
      "learning_rate": 4.707311948262371e-05,
      "loss": 1.448,
      "step": 19189
    },
    {
      "epoch": 0.687234766415385,
      "grad_norm": 1.4807206392288208,
      "learning_rate": 4.7063278607419944e-05,
      "loss": 1.4774,
      "step": 19190
    },
    {
      "epoch": 0.6872705785449532,
      "grad_norm": 1.4907841682434082,
      "learning_rate": 4.705343844441158e-05,
      "loss": 1.3822,
      "step": 19191
    },
    {
      "epoch": 0.6873063906745215,
      "grad_norm": 2.6561944484710693,
      "learning_rate": 4.704359899373089e-05,
      "loss": 1.3852,
      "step": 19192
    },
    {
      "epoch": 0.6873422028040898,
      "grad_norm": 1.906999111175537,
      "learning_rate": 4.703376025551023e-05,
      "loss": 1.4777,
      "step": 19193
    },
    {
      "epoch": 0.687378014933658,
      "grad_norm": 1.8173359632492065,
      "learning_rate": 4.7023922229882013e-05,
      "loss": 1.3468,
      "step": 19194
    },
    {
      "epoch": 0.6874138270632263,
      "grad_norm": 1.8602315187454224,
      "learning_rate": 4.701408491697859e-05,
      "loss": 1.5548,
      "step": 19195
    },
    {
      "epoch": 0.6874496391927946,
      "grad_norm": 1.9973033666610718,
      "learning_rate": 4.700424831693233e-05,
      "loss": 1.4175,
      "step": 19196
    },
    {
      "epoch": 0.6874854513223629,
      "grad_norm": 1.5902940034866333,
      "learning_rate": 4.699441242987548e-05,
      "loss": 1.2324,
      "step": 19197
    },
    {
      "epoch": 0.6875212634519312,
      "grad_norm": 1.6441142559051514,
      "learning_rate": 4.698457725594052e-05,
      "loss": 1.7983,
      "step": 19198
    },
    {
      "epoch": 0.6875570755814995,
      "grad_norm": 1.9269689321517944,
      "learning_rate": 4.697474279525964e-05,
      "loss": 1.3264,
      "step": 19199
    },
    {
      "epoch": 0.6875928877110677,
      "grad_norm": 1.7410550117492676,
      "learning_rate": 4.6964909047965246e-05,
      "loss": 1.2364,
      "step": 19200
    },
    {
      "epoch": 0.687628699840636,
      "grad_norm": 1.4461188316345215,
      "learning_rate": 4.6955076014189545e-05,
      "loss": 1.3768,
      "step": 19201
    },
    {
      "epoch": 0.6876645119702043,
      "grad_norm": 1.9367955923080444,
      "learning_rate": 4.694524369406488e-05,
      "loss": 1.7564,
      "step": 19202
    },
    {
      "epoch": 0.6877003240997726,
      "grad_norm": 2.228534460067749,
      "learning_rate": 4.693541208772356e-05,
      "loss": 1.5539,
      "step": 19203
    },
    {
      "epoch": 0.6877361362293409,
      "grad_norm": 1.685943365097046,
      "learning_rate": 4.692558119529778e-05,
      "loss": 1.4786,
      "step": 19204
    },
    {
      "epoch": 0.6877719483589092,
      "grad_norm": 1.4004324674606323,
      "learning_rate": 4.691575101691985e-05,
      "loss": 1.3005,
      "step": 19205
    },
    {
      "epoch": 0.6878077604884775,
      "grad_norm": 1.7593801021575928,
      "learning_rate": 4.6905921552722024e-05,
      "loss": 1.1818,
      "step": 19206
    },
    {
      "epoch": 0.6878435726180457,
      "grad_norm": 1.8382956981658936,
      "learning_rate": 4.6896092802836555e-05,
      "loss": 1.292,
      "step": 19207
    },
    {
      "epoch": 0.687879384747614,
      "grad_norm": 1.941030502319336,
      "learning_rate": 4.6886264767395635e-05,
      "loss": 1.4367,
      "step": 19208
    },
    {
      "epoch": 0.6879151968771823,
      "grad_norm": 2.078139543533325,
      "learning_rate": 4.687643744653151e-05,
      "loss": 1.4142,
      "step": 19209
    },
    {
      "epoch": 0.6879510090067505,
      "grad_norm": 1.504701852798462,
      "learning_rate": 4.6866610840376424e-05,
      "loss": 1.4804,
      "step": 19210
    },
    {
      "epoch": 0.6879868211363188,
      "grad_norm": 1.9218412637710571,
      "learning_rate": 4.6856784949062516e-05,
      "loss": 1.6285,
      "step": 19211
    },
    {
      "epoch": 0.6880226332658872,
      "grad_norm": 1.6151492595672607,
      "learning_rate": 4.6846959772722023e-05,
      "loss": 1.4061,
      "step": 19212
    },
    {
      "epoch": 0.6880584453954555,
      "grad_norm": 1.6147105693817139,
      "learning_rate": 4.6837135311487125e-05,
      "loss": 1.1338,
      "step": 19213
    },
    {
      "epoch": 0.6880942575250237,
      "grad_norm": 1.5472468137741089,
      "learning_rate": 4.6827311565490026e-05,
      "loss": 1.6359,
      "step": 19214
    },
    {
      "epoch": 0.688130069654592,
      "grad_norm": 1.5259733200073242,
      "learning_rate": 4.681748853486283e-05,
      "loss": 1.3292,
      "step": 19215
    },
    {
      "epoch": 0.6881658817841603,
      "grad_norm": 1.8748228549957275,
      "learning_rate": 4.6807666219737724e-05,
      "loss": 1.451,
      "step": 19216
    },
    {
      "epoch": 0.6882016939137285,
      "grad_norm": 1.626900315284729,
      "learning_rate": 4.679784462024686e-05,
      "loss": 1.6105,
      "step": 19217
    },
    {
      "epoch": 0.6882375060432968,
      "grad_norm": 1.2625377178192139,
      "learning_rate": 4.6788023736522405e-05,
      "loss": 1.0924,
      "step": 19218
    },
    {
      "epoch": 0.6882733181728652,
      "grad_norm": 1.3529118299484253,
      "learning_rate": 4.677820356869641e-05,
      "loss": 1.4678,
      "step": 19219
    },
    {
      "epoch": 0.6883091303024335,
      "grad_norm": 2.293600082397461,
      "learning_rate": 4.676838411690103e-05,
      "loss": 1.5526,
      "step": 19220
    },
    {
      "epoch": 0.6883449424320017,
      "grad_norm": 2.429054021835327,
      "learning_rate": 4.675856538126843e-05,
      "loss": 1.4861,
      "step": 19221
    },
    {
      "epoch": 0.68838075456157,
      "grad_norm": 1.7435716390609741,
      "learning_rate": 4.674874736193061e-05,
      "loss": 1.6183,
      "step": 19222
    },
    {
      "epoch": 0.6884165666911383,
      "grad_norm": 1.8326066732406616,
      "learning_rate": 4.67389300590197e-05,
      "loss": 1.3012,
      "step": 19223
    },
    {
      "epoch": 0.6884523788207065,
      "grad_norm": 1.5112701654434204,
      "learning_rate": 4.67291134726678e-05,
      "loss": 1.1489,
      "step": 19224
    },
    {
      "epoch": 0.6884881909502748,
      "grad_norm": 1.7756952047348022,
      "learning_rate": 4.6719297603006994e-05,
      "loss": 1.5206,
      "step": 19225
    },
    {
      "epoch": 0.6885240030798432,
      "grad_norm": 2.031517744064331,
      "learning_rate": 4.6709482450169275e-05,
      "loss": 1.5716,
      "step": 19226
    },
    {
      "epoch": 0.6885598152094115,
      "grad_norm": 1.7005447149276733,
      "learning_rate": 4.6699668014286724e-05,
      "loss": 1.6381,
      "step": 19227
    },
    {
      "epoch": 0.6885956273389797,
      "grad_norm": 1.2920862436294556,
      "learning_rate": 4.668985429549143e-05,
      "loss": 1.6038,
      "step": 19228
    },
    {
      "epoch": 0.688631439468548,
      "grad_norm": 1.2875529527664185,
      "learning_rate": 4.6680041293915336e-05,
      "loss": 1.2111,
      "step": 19229
    },
    {
      "epoch": 0.6886672515981163,
      "grad_norm": 1.2606359720230103,
      "learning_rate": 4.6670229009690516e-05,
      "loss": 1.4477,
      "step": 19230
    },
    {
      "epoch": 0.6887030637276845,
      "grad_norm": 2.3484201431274414,
      "learning_rate": 4.666041744294898e-05,
      "loss": 1.7855,
      "step": 19231
    },
    {
      "epoch": 0.6887388758572528,
      "grad_norm": 1.6336703300476074,
      "learning_rate": 4.665060659382274e-05,
      "loss": 1.1235,
      "step": 19232
    },
    {
      "epoch": 0.6887746879868212,
      "grad_norm": 2.369983196258545,
      "learning_rate": 4.664079646244376e-05,
      "loss": 1.323,
      "step": 19233
    },
    {
      "epoch": 0.6888105001163894,
      "grad_norm": 1.3482604026794434,
      "learning_rate": 4.663098704894402e-05,
      "loss": 1.5943,
      "step": 19234
    },
    {
      "epoch": 0.6888463122459577,
      "grad_norm": 1.468631386756897,
      "learning_rate": 4.662117835345552e-05,
      "loss": 1.4307,
      "step": 19235
    },
    {
      "epoch": 0.688882124375526,
      "grad_norm": 1.9677053689956665,
      "learning_rate": 4.661137037611024e-05,
      "loss": 1.1598,
      "step": 19236
    },
    {
      "epoch": 0.6889179365050943,
      "grad_norm": 1.6794158220291138,
      "learning_rate": 4.660156311704007e-05,
      "loss": 1.7906,
      "step": 19237
    },
    {
      "epoch": 0.6889537486346625,
      "grad_norm": 1.7666298151016235,
      "learning_rate": 4.659175657637699e-05,
      "loss": 1.2263,
      "step": 19238
    },
    {
      "epoch": 0.6889895607642308,
      "grad_norm": 2.3482303619384766,
      "learning_rate": 4.658195075425297e-05,
      "loss": 1.285,
      "step": 19239
    },
    {
      "epoch": 0.6890253728937992,
      "grad_norm": 1.2648680210113525,
      "learning_rate": 4.657214565079986e-05,
      "loss": 1.4305,
      "step": 19240
    },
    {
      "epoch": 0.6890611850233674,
      "grad_norm": 2.5876047611236572,
      "learning_rate": 4.656234126614961e-05,
      "loss": 1.7217,
      "step": 19241
    },
    {
      "epoch": 0.6890969971529357,
      "grad_norm": 1.4185413122177124,
      "learning_rate": 4.655253760043413e-05,
      "loss": 1.5885,
      "step": 19242
    },
    {
      "epoch": 0.689132809282504,
      "grad_norm": 1.3307768106460571,
      "learning_rate": 4.654273465378536e-05,
      "loss": 1.2001,
      "step": 19243
    },
    {
      "epoch": 0.6891686214120722,
      "grad_norm": 1.5933369398117065,
      "learning_rate": 4.65329324263351e-05,
      "loss": 1.2176,
      "step": 19244
    },
    {
      "epoch": 0.6892044335416405,
      "grad_norm": 1.4384766817092896,
      "learning_rate": 4.652313091821526e-05,
      "loss": 1.4176,
      "step": 19245
    },
    {
      "epoch": 0.6892402456712088,
      "grad_norm": 1.2715067863464355,
      "learning_rate": 4.651333012955775e-05,
      "loss": 1.6592,
      "step": 19246
    },
    {
      "epoch": 0.6892760578007772,
      "grad_norm": 1.4462275505065918,
      "learning_rate": 4.650353006049436e-05,
      "loss": 1.4901,
      "step": 19247
    },
    {
      "epoch": 0.6893118699303454,
      "grad_norm": 1.8860747814178467,
      "learning_rate": 4.649373071115697e-05,
      "loss": 1.4205,
      "step": 19248
    },
    {
      "epoch": 0.6893476820599137,
      "grad_norm": 1.4204922914505005,
      "learning_rate": 4.6483932081677407e-05,
      "loss": 1.3774,
      "step": 19249
    },
    {
      "epoch": 0.689383494189482,
      "grad_norm": 1.5421233177185059,
      "learning_rate": 4.647413417218756e-05,
      "loss": 1.0301,
      "step": 19250
    },
    {
      "epoch": 0.6894193063190502,
      "grad_norm": 1.686322808265686,
      "learning_rate": 4.646433698281913e-05,
      "loss": 1.5836,
      "step": 19251
    },
    {
      "epoch": 0.6894551184486185,
      "grad_norm": 1.406529426574707,
      "learning_rate": 4.645454051370401e-05,
      "loss": 1.1692,
      "step": 19252
    },
    {
      "epoch": 0.6894909305781868,
      "grad_norm": 1.3951549530029297,
      "learning_rate": 4.644474476497397e-05,
      "loss": 1.3326,
      "step": 19253
    },
    {
      "epoch": 0.6895267427077552,
      "grad_norm": 1.7773816585540771,
      "learning_rate": 4.6434949736760844e-05,
      "loss": 1.3854,
      "step": 19254
    },
    {
      "epoch": 0.6895625548373234,
      "grad_norm": 2.3042869567871094,
      "learning_rate": 4.642515542919635e-05,
      "loss": 1.5021,
      "step": 19255
    },
    {
      "epoch": 0.6895983669668917,
      "grad_norm": 1.5629322528839111,
      "learning_rate": 4.641536184241228e-05,
      "loss": 1.4215,
      "step": 19256
    },
    {
      "epoch": 0.68963417909646,
      "grad_norm": 1.878865122795105,
      "learning_rate": 4.640556897654042e-05,
      "loss": 1.3999,
      "step": 19257
    },
    {
      "epoch": 0.6896699912260282,
      "grad_norm": 1.6263047456741333,
      "learning_rate": 4.639577683171248e-05,
      "loss": 1.5993,
      "step": 19258
    },
    {
      "epoch": 0.6897058033555965,
      "grad_norm": 1.5195845365524292,
      "learning_rate": 4.638598540806021e-05,
      "loss": 1.4672,
      "step": 19259
    },
    {
      "epoch": 0.6897416154851648,
      "grad_norm": 1.7965352535247803,
      "learning_rate": 4.637619470571535e-05,
      "loss": 1.7661,
      "step": 19260
    },
    {
      "epoch": 0.6897774276147332,
      "grad_norm": 1.6480170488357544,
      "learning_rate": 4.636640472480965e-05,
      "loss": 1.4935,
      "step": 19261
    },
    {
      "epoch": 0.6898132397443014,
      "grad_norm": 1.973185658454895,
      "learning_rate": 4.635661546547476e-05,
      "loss": 1.669,
      "step": 19262
    },
    {
      "epoch": 0.6898490518738697,
      "grad_norm": 1.6498275995254517,
      "learning_rate": 4.634682692784241e-05,
      "loss": 1.2816,
      "step": 19263
    },
    {
      "epoch": 0.689884864003438,
      "grad_norm": 1.2615090608596802,
      "learning_rate": 4.6337039112044346e-05,
      "loss": 1.011,
      "step": 19264
    },
    {
      "epoch": 0.6899206761330062,
      "grad_norm": 1.4488906860351562,
      "learning_rate": 4.632725201821215e-05,
      "loss": 1.014,
      "step": 19265
    },
    {
      "epoch": 0.6899564882625745,
      "grad_norm": 1.690446138381958,
      "learning_rate": 4.6317465646477584e-05,
      "loss": 1.1892,
      "step": 19266
    },
    {
      "epoch": 0.6899923003921428,
      "grad_norm": 1.2341758012771606,
      "learning_rate": 4.6307679996972205e-05,
      "loss": 1.3111,
      "step": 19267
    },
    {
      "epoch": 0.6900281125217111,
      "grad_norm": 1.505085825920105,
      "learning_rate": 4.62978950698278e-05,
      "loss": 1.3969,
      "step": 19268
    },
    {
      "epoch": 0.6900639246512794,
      "grad_norm": 1.8536823987960815,
      "learning_rate": 4.6288110865175914e-05,
      "loss": 1.2291,
      "step": 19269
    },
    {
      "epoch": 0.6900997367808477,
      "grad_norm": 1.2914131879806519,
      "learning_rate": 4.627832738314821e-05,
      "loss": 1.3716,
      "step": 19270
    },
    {
      "epoch": 0.690135548910416,
      "grad_norm": 1.180411458015442,
      "learning_rate": 4.6268544623876364e-05,
      "loss": 1.2842,
      "step": 19271
    },
    {
      "epoch": 0.6901713610399842,
      "grad_norm": 1.6704957485198975,
      "learning_rate": 4.625876258749189e-05,
      "loss": 1.3887,
      "step": 19272
    },
    {
      "epoch": 0.6902071731695525,
      "grad_norm": 1.6580283641815186,
      "learning_rate": 4.624898127412649e-05,
      "loss": 1.4075,
      "step": 19273
    },
    {
      "epoch": 0.6902429852991208,
      "grad_norm": 2.034627914428711,
      "learning_rate": 4.623920068391163e-05,
      "loss": 1.2691,
      "step": 19274
    },
    {
      "epoch": 0.6902787974286891,
      "grad_norm": 1.591990351676941,
      "learning_rate": 4.622942081697906e-05,
      "loss": 1.4743,
      "step": 19275
    },
    {
      "epoch": 0.6903146095582574,
      "grad_norm": 1.417268991470337,
      "learning_rate": 4.6219641673460236e-05,
      "loss": 1.257,
      "step": 19276
    },
    {
      "epoch": 0.6903504216878257,
      "grad_norm": 1.6440649032592773,
      "learning_rate": 4.62098632534868e-05,
      "loss": 1.2822,
      "step": 19277
    },
    {
      "epoch": 0.690386233817394,
      "grad_norm": 1.5523018836975098,
      "learning_rate": 4.620008555719019e-05,
      "loss": 1.3786,
      "step": 19278
    },
    {
      "epoch": 0.6904220459469622,
      "grad_norm": 1.6315157413482666,
      "learning_rate": 4.619030858470211e-05,
      "loss": 1.5197,
      "step": 19279
    },
    {
      "epoch": 0.6904578580765305,
      "grad_norm": 1.3777469396591187,
      "learning_rate": 4.6180532336154014e-05,
      "loss": 1.1425,
      "step": 19280
    },
    {
      "epoch": 0.6904936702060988,
      "grad_norm": 1.4311867952346802,
      "learning_rate": 4.617075681167736e-05,
      "loss": 1.2902,
      "step": 19281
    },
    {
      "epoch": 0.6905294823356671,
      "grad_norm": 1.3356856107711792,
      "learning_rate": 4.616098201140382e-05,
      "loss": 1.4217,
      "step": 19282
    },
    {
      "epoch": 0.6905652944652354,
      "grad_norm": 1.3175742626190186,
      "learning_rate": 4.615120793546478e-05,
      "loss": 1.2804,
      "step": 19283
    },
    {
      "epoch": 0.6906011065948037,
      "grad_norm": 1.5983734130859375,
      "learning_rate": 4.6141434583991803e-05,
      "loss": 1.4278,
      "step": 19284
    },
    {
      "epoch": 0.6906369187243719,
      "grad_norm": 1.3159353733062744,
      "learning_rate": 4.613166195711629e-05,
      "loss": 1.3914,
      "step": 19285
    },
    {
      "epoch": 0.6906727308539402,
      "grad_norm": 1.8552560806274414,
      "learning_rate": 4.612189005496985e-05,
      "loss": 1.2788,
      "step": 19286
    },
    {
      "epoch": 0.6907085429835085,
      "grad_norm": 1.4018633365631104,
      "learning_rate": 4.611211887768384e-05,
      "loss": 1.4291,
      "step": 19287
    },
    {
      "epoch": 0.6907443551130767,
      "grad_norm": 1.557361364364624,
      "learning_rate": 4.6102348425389804e-05,
      "loss": 1.2867,
      "step": 19288
    },
    {
      "epoch": 0.6907801672426451,
      "grad_norm": 1.355259895324707,
      "learning_rate": 4.609257869821911e-05,
      "loss": 1.1086,
      "step": 19289
    },
    {
      "epoch": 0.6908159793722134,
      "grad_norm": 1.498028039932251,
      "learning_rate": 4.608280969630323e-05,
      "loss": 1.7323,
      "step": 19290
    },
    {
      "epoch": 0.6908517915017817,
      "grad_norm": 1.7910292148590088,
      "learning_rate": 4.6073041419773635e-05,
      "loss": 1.2225,
      "step": 19291
    },
    {
      "epoch": 0.6908876036313499,
      "grad_norm": 1.6466988325119019,
      "learning_rate": 4.606327386876167e-05,
      "loss": 1.3962,
      "step": 19292
    },
    {
      "epoch": 0.6909234157609182,
      "grad_norm": 1.5631616115570068,
      "learning_rate": 4.605350704339879e-05,
      "loss": 1.4526,
      "step": 19293
    },
    {
      "epoch": 0.6909592278904865,
      "grad_norm": 2.4088470935821533,
      "learning_rate": 4.604374094381637e-05,
      "loss": 1.8184,
      "step": 19294
    },
    {
      "epoch": 0.6909950400200547,
      "grad_norm": 1.8684895038604736,
      "learning_rate": 4.603397557014587e-05,
      "loss": 1.5012,
      "step": 19295
    },
    {
      "epoch": 0.6910308521496231,
      "grad_norm": 1.6042251586914062,
      "learning_rate": 4.602421092251854e-05,
      "loss": 1.2943,
      "step": 19296
    },
    {
      "epoch": 0.6910666642791914,
      "grad_norm": 2.03073787689209,
      "learning_rate": 4.60144470010659e-05,
      "loss": 1.9343,
      "step": 19297
    },
    {
      "epoch": 0.6911024764087597,
      "grad_norm": 1.6332614421844482,
      "learning_rate": 4.600468380591923e-05,
      "loss": 1.4177,
      "step": 19298
    },
    {
      "epoch": 0.6911382885383279,
      "grad_norm": 1.823080062866211,
      "learning_rate": 4.599492133720986e-05,
      "loss": 1.283,
      "step": 19299
    },
    {
      "epoch": 0.6911741006678962,
      "grad_norm": 2.0920989513397217,
      "learning_rate": 4.598515959506917e-05,
      "loss": 1.3483,
      "step": 19300
    },
    {
      "epoch": 0.6912099127974645,
      "grad_norm": 1.4750959873199463,
      "learning_rate": 4.597539857962848e-05,
      "loss": 1.3174,
      "step": 19301
    },
    {
      "epoch": 0.6912457249270327,
      "grad_norm": 1.3776130676269531,
      "learning_rate": 4.5965638291019145e-05,
      "loss": 1.525,
      "step": 19302
    },
    {
      "epoch": 0.6912815370566011,
      "grad_norm": 1.7445027828216553,
      "learning_rate": 4.595587872937241e-05,
      "loss": 1.5717,
      "step": 19303
    },
    {
      "epoch": 0.6913173491861694,
      "grad_norm": 1.7076929807662964,
      "learning_rate": 4.594611989481963e-05,
      "loss": 1.1559,
      "step": 19304
    },
    {
      "epoch": 0.6913531613157377,
      "grad_norm": 1.4765623807907104,
      "learning_rate": 4.593636178749206e-05,
      "loss": 1.6233,
      "step": 19305
    },
    {
      "epoch": 0.6913889734453059,
      "grad_norm": 1.6138931512832642,
      "learning_rate": 4.592660440752107e-05,
      "loss": 1.3967,
      "step": 19306
    },
    {
      "epoch": 0.6914247855748742,
      "grad_norm": 1.3620359897613525,
      "learning_rate": 4.5916847755037806e-05,
      "loss": 1.349,
      "step": 19307
    },
    {
      "epoch": 0.6914605977044425,
      "grad_norm": 1.8544646501541138,
      "learning_rate": 4.590709183017361e-05,
      "loss": 1.5375,
      "step": 19308
    },
    {
      "epoch": 0.6914964098340107,
      "grad_norm": 1.3989589214324951,
      "learning_rate": 4.5897336633059737e-05,
      "loss": 1.3476,
      "step": 19309
    },
    {
      "epoch": 0.6915322219635791,
      "grad_norm": 2.901787281036377,
      "learning_rate": 4.588758216382739e-05,
      "loss": 1.6608,
      "step": 19310
    },
    {
      "epoch": 0.6915680340931474,
      "grad_norm": 1.9451749324798584,
      "learning_rate": 4.5877828422607824e-05,
      "loss": 1.7291,
      "step": 19311
    },
    {
      "epoch": 0.6916038462227156,
      "grad_norm": 1.8403728008270264,
      "learning_rate": 4.5868075409532265e-05,
      "loss": 1.2411,
      "step": 19312
    },
    {
      "epoch": 0.6916396583522839,
      "grad_norm": 1.5571632385253906,
      "learning_rate": 4.585832312473196e-05,
      "loss": 1.6115,
      "step": 19313
    },
    {
      "epoch": 0.6916754704818522,
      "grad_norm": 2.0901341438293457,
      "learning_rate": 4.584857156833804e-05,
      "loss": 1.6583,
      "step": 19314
    },
    {
      "epoch": 0.6917112826114205,
      "grad_norm": 1.7531592845916748,
      "learning_rate": 4.583882074048174e-05,
      "loss": 1.2701,
      "step": 19315
    },
    {
      "epoch": 0.6917470947409887,
      "grad_norm": 1.3582504987716675,
      "learning_rate": 4.582907064129428e-05,
      "loss": 1.3205,
      "step": 19316
    },
    {
      "epoch": 0.6917829068705571,
      "grad_norm": 1.857133150100708,
      "learning_rate": 4.5819321270906765e-05,
      "loss": 1.5085,
      "step": 19317
    },
    {
      "epoch": 0.6918187190001254,
      "grad_norm": 2.7562215328216553,
      "learning_rate": 4.580957262945039e-05,
      "loss": 1.4333,
      "step": 19318
    },
    {
      "epoch": 0.6918545311296936,
      "grad_norm": 2.172489881515503,
      "learning_rate": 4.5799824717056325e-05,
      "loss": 1.6725,
      "step": 19319
    },
    {
      "epoch": 0.6918903432592619,
      "grad_norm": 1.492525577545166,
      "learning_rate": 4.579007753385573e-05,
      "loss": 1.4247,
      "step": 19320
    },
    {
      "epoch": 0.6919261553888302,
      "grad_norm": 1.6404074430465698,
      "learning_rate": 4.578033107997969e-05,
      "loss": 1.5811,
      "step": 19321
    },
    {
      "epoch": 0.6919619675183984,
      "grad_norm": 1.5491911172866821,
      "learning_rate": 4.577058535555935e-05,
      "loss": 1.5538,
      "step": 19322
    },
    {
      "epoch": 0.6919977796479667,
      "grad_norm": 1.4528319835662842,
      "learning_rate": 4.576084036072584e-05,
      "loss": 1.2544,
      "step": 19323
    },
    {
      "epoch": 0.6920335917775351,
      "grad_norm": 1.7799615859985352,
      "learning_rate": 4.575109609561029e-05,
      "loss": 1.4805,
      "step": 19324
    },
    {
      "epoch": 0.6920694039071034,
      "grad_norm": 1.4052472114562988,
      "learning_rate": 4.5741352560343734e-05,
      "loss": 1.364,
      "step": 19325
    },
    {
      "epoch": 0.6921052160366716,
      "grad_norm": 1.4462934732437134,
      "learning_rate": 4.5731609755057284e-05,
      "loss": 1.5724,
      "step": 19326
    },
    {
      "epoch": 0.6921410281662399,
      "grad_norm": 2.7505440711975098,
      "learning_rate": 4.572186767988206e-05,
      "loss": 1.4352,
      "step": 19327
    },
    {
      "epoch": 0.6921768402958082,
      "grad_norm": 2.1405277252197266,
      "learning_rate": 4.571212633494906e-05,
      "loss": 1.8836,
      "step": 19328
    },
    {
      "epoch": 0.6922126524253764,
      "grad_norm": 1.584587574005127,
      "learning_rate": 4.5702385720389376e-05,
      "loss": 1.2718,
      "step": 19329
    },
    {
      "epoch": 0.6922484645549447,
      "grad_norm": 1.6535595655441284,
      "learning_rate": 4.569264583633405e-05,
      "loss": 1.3507,
      "step": 19330
    },
    {
      "epoch": 0.6922842766845131,
      "grad_norm": 1.9386780261993408,
      "learning_rate": 4.568290668291416e-05,
      "loss": 1.345,
      "step": 19331
    },
    {
      "epoch": 0.6923200888140814,
      "grad_norm": 1.418664813041687,
      "learning_rate": 4.567316826026066e-05,
      "loss": 1.1447,
      "step": 19332
    },
    {
      "epoch": 0.6923559009436496,
      "grad_norm": 1.544927716255188,
      "learning_rate": 4.5663430568504603e-05,
      "loss": 1.5124,
      "step": 19333
    },
    {
      "epoch": 0.6923917130732179,
      "grad_norm": 1.3872967958450317,
      "learning_rate": 4.565369360777704e-05,
      "loss": 1.6032,
      "step": 19334
    },
    {
      "epoch": 0.6924275252027862,
      "grad_norm": 1.791256308555603,
      "learning_rate": 4.564395737820888e-05,
      "loss": 1.7009,
      "step": 19335
    },
    {
      "epoch": 0.6924633373323544,
      "grad_norm": 1.7304942607879639,
      "learning_rate": 4.563422187993117e-05,
      "loss": 1.3953,
      "step": 19336
    },
    {
      "epoch": 0.6924991494619227,
      "grad_norm": 1.745445966720581,
      "learning_rate": 4.5624487113074874e-05,
      "loss": 1.4433,
      "step": 19337
    },
    {
      "epoch": 0.6925349615914911,
      "grad_norm": 1.7737088203430176,
      "learning_rate": 4.5614753077771e-05,
      "loss": 1.6636,
      "step": 19338
    },
    {
      "epoch": 0.6925707737210594,
      "grad_norm": 1.6347934007644653,
      "learning_rate": 4.560501977415044e-05,
      "loss": 1.6366,
      "step": 19339
    },
    {
      "epoch": 0.6926065858506276,
      "grad_norm": 1.919651985168457,
      "learning_rate": 4.5595287202344175e-05,
      "loss": 1.5327,
      "step": 19340
    },
    {
      "epoch": 0.6926423979801959,
      "grad_norm": 1.5812537670135498,
      "learning_rate": 4.558555536248313e-05,
      "loss": 1.6632,
      "step": 19341
    },
    {
      "epoch": 0.6926782101097642,
      "grad_norm": 1.4711294174194336,
      "learning_rate": 4.55758242546983e-05,
      "loss": 1.4774,
      "step": 19342
    },
    {
      "epoch": 0.6927140222393324,
      "grad_norm": 1.3836665153503418,
      "learning_rate": 4.5566093879120505e-05,
      "loss": 1.2167,
      "step": 19343
    },
    {
      "epoch": 0.6927498343689007,
      "grad_norm": 1.8161534070968628,
      "learning_rate": 4.555636423588071e-05,
      "loss": 1.4929,
      "step": 19344
    },
    {
      "epoch": 0.6927856464984691,
      "grad_norm": 1.7006677389144897,
      "learning_rate": 4.5546635325109844e-05,
      "loss": 1.1194,
      "step": 19345
    },
    {
      "epoch": 0.6928214586280373,
      "grad_norm": 1.4193700551986694,
      "learning_rate": 4.553690714693872e-05,
      "loss": 1.4037,
      "step": 19346
    },
    {
      "epoch": 0.6928572707576056,
      "grad_norm": 1.9114856719970703,
      "learning_rate": 4.5527179701498256e-05,
      "loss": 1.4374,
      "step": 19347
    },
    {
      "epoch": 0.6928930828871739,
      "grad_norm": 1.6580359935760498,
      "learning_rate": 4.551745298891933e-05,
      "loss": 1.3423,
      "step": 19348
    },
    {
      "epoch": 0.6929288950167422,
      "grad_norm": 1.9556456804275513,
      "learning_rate": 4.5507727009332824e-05,
      "loss": 1.2934,
      "step": 19349
    },
    {
      "epoch": 0.6929647071463104,
      "grad_norm": 2.0172038078308105,
      "learning_rate": 4.549800176286954e-05,
      "loss": 1.599,
      "step": 19350
    },
    {
      "epoch": 0.6930005192758787,
      "grad_norm": 2.1281747817993164,
      "learning_rate": 4.5488277249660325e-05,
      "loss": 1.6343,
      "step": 19351
    },
    {
      "epoch": 0.6930363314054471,
      "grad_norm": 1.7381844520568848,
      "learning_rate": 4.5478553469836064e-05,
      "loss": 1.2599,
      "step": 19352
    },
    {
      "epoch": 0.6930721435350153,
      "grad_norm": 2.3697657585144043,
      "learning_rate": 4.546883042352751e-05,
      "loss": 1.7014,
      "step": 19353
    },
    {
      "epoch": 0.6931079556645836,
      "grad_norm": 2.5228118896484375,
      "learning_rate": 4.545910811086549e-05,
      "loss": 1.4948,
      "step": 19354
    },
    {
      "epoch": 0.6931437677941519,
      "grad_norm": 1.3983299732208252,
      "learning_rate": 4.544938653198082e-05,
      "loss": 1.4429,
      "step": 19355
    },
    {
      "epoch": 0.6931795799237201,
      "grad_norm": 1.7777349948883057,
      "learning_rate": 4.543966568700433e-05,
      "loss": 1.4876,
      "step": 19356
    },
    {
      "epoch": 0.6932153920532884,
      "grad_norm": 1.5651066303253174,
      "learning_rate": 4.542994557606672e-05,
      "loss": 1.683,
      "step": 19357
    },
    {
      "epoch": 0.6932512041828567,
      "grad_norm": 1.167218565940857,
      "learning_rate": 4.542022619929881e-05,
      "loss": 1.5015,
      "step": 19358
    },
    {
      "epoch": 0.6932870163124251,
      "grad_norm": 1.1746838092803955,
      "learning_rate": 4.541050755683135e-05,
      "loss": 1.4014,
      "step": 19359
    },
    {
      "epoch": 0.6933228284419933,
      "grad_norm": 1.9746357202529907,
      "learning_rate": 4.5400789648795136e-05,
      "loss": 1.5701,
      "step": 19360
    },
    {
      "epoch": 0.6933586405715616,
      "grad_norm": 2.039637804031372,
      "learning_rate": 4.539107247532086e-05,
      "loss": 1.6741,
      "step": 19361
    },
    {
      "epoch": 0.6933944527011299,
      "grad_norm": 1.581687331199646,
      "learning_rate": 4.5381356036539204e-05,
      "loss": 1.2782,
      "step": 19362
    },
    {
      "epoch": 0.6934302648306981,
      "grad_norm": 1.605271577835083,
      "learning_rate": 4.537164033258101e-05,
      "loss": 1.582,
      "step": 19363
    },
    {
      "epoch": 0.6934660769602664,
      "grad_norm": 1.3678474426269531,
      "learning_rate": 4.53619253635769e-05,
      "loss": 1.5372,
      "step": 19364
    },
    {
      "epoch": 0.6935018890898347,
      "grad_norm": 1.3803857564926147,
      "learning_rate": 4.5352211129657596e-05,
      "loss": 1.2848,
      "step": 19365
    },
    {
      "epoch": 0.6935377012194031,
      "grad_norm": 1.4274797439575195,
      "learning_rate": 4.5342497630953806e-05,
      "loss": 1.5293,
      "step": 19366
    },
    {
      "epoch": 0.6935735133489713,
      "grad_norm": 1.4727002382278442,
      "learning_rate": 4.533278486759625e-05,
      "loss": 1.2364,
      "step": 19367
    },
    {
      "epoch": 0.6936093254785396,
      "grad_norm": 1.8014189004898071,
      "learning_rate": 4.5323072839715555e-05,
      "loss": 1.4292,
      "step": 19368
    },
    {
      "epoch": 0.6936451376081079,
      "grad_norm": 1.8191452026367188,
      "learning_rate": 4.531336154744231e-05,
      "loss": 1.4423,
      "step": 19369
    },
    {
      "epoch": 0.6936809497376761,
      "grad_norm": 1.5207409858703613,
      "learning_rate": 4.530365099090732e-05,
      "loss": 1.5638,
      "step": 19370
    },
    {
      "epoch": 0.6937167618672444,
      "grad_norm": 1.5536220073699951,
      "learning_rate": 4.5293941170241116e-05,
      "loss": 1.5002,
      "step": 19371
    },
    {
      "epoch": 0.6937525739968127,
      "grad_norm": 2.320056200027466,
      "learning_rate": 4.528423208557441e-05,
      "loss": 1.3389,
      "step": 19372
    },
    {
      "epoch": 0.693788386126381,
      "grad_norm": 1.6876623630523682,
      "learning_rate": 4.52745237370377e-05,
      "loss": 1.2901,
      "step": 19373
    },
    {
      "epoch": 0.6938241982559493,
      "grad_norm": 1.3508429527282715,
      "learning_rate": 4.526481612476176e-05,
      "loss": 1.3113,
      "step": 19374
    },
    {
      "epoch": 0.6938600103855176,
      "grad_norm": 1.5658550262451172,
      "learning_rate": 4.525510924887707e-05,
      "loss": 1.5774,
      "step": 19375
    },
    {
      "epoch": 0.6938958225150859,
      "grad_norm": 1.841926097869873,
      "learning_rate": 4.524540310951432e-05,
      "loss": 1.3422,
      "step": 19376
    },
    {
      "epoch": 0.6939316346446541,
      "grad_norm": 2.058666467666626,
      "learning_rate": 4.5235697706804e-05,
      "loss": 1.4251,
      "step": 19377
    },
    {
      "epoch": 0.6939674467742224,
      "grad_norm": 1.5321693420410156,
      "learning_rate": 4.5225993040876724e-05,
      "loss": 1.5993,
      "step": 19378
    },
    {
      "epoch": 0.6940032589037907,
      "grad_norm": 1.959938645362854,
      "learning_rate": 4.521628911186311e-05,
      "loss": 1.6419,
      "step": 19379
    },
    {
      "epoch": 0.694039071033359,
      "grad_norm": 1.6654266119003296,
      "learning_rate": 4.5206585919893563e-05,
      "loss": 1.5531,
      "step": 19380
    },
    {
      "epoch": 0.6940748831629273,
      "grad_norm": 2.1616580486297607,
      "learning_rate": 4.519688346509881e-05,
      "loss": 1.935,
      "step": 19381
    },
    {
      "epoch": 0.6941106952924956,
      "grad_norm": 1.5678056478500366,
      "learning_rate": 4.5187181747609265e-05,
      "loss": 1.4482,
      "step": 19382
    },
    {
      "epoch": 0.6941465074220639,
      "grad_norm": 1.2296350002288818,
      "learning_rate": 4.5177480767555525e-05,
      "loss": 1.4928,
      "step": 19383
    },
    {
      "epoch": 0.6941823195516321,
      "grad_norm": 1.5905427932739258,
      "learning_rate": 4.516778052506798e-05,
      "loss": 1.4104,
      "step": 19384
    },
    {
      "epoch": 0.6942181316812004,
      "grad_norm": 2.694613456726074,
      "learning_rate": 4.5158081020277296e-05,
      "loss": 1.6461,
      "step": 19385
    },
    {
      "epoch": 0.6942539438107687,
      "grad_norm": 1.8506790399551392,
      "learning_rate": 4.5148382253313904e-05,
      "loss": 1.4123,
      "step": 19386
    },
    {
      "epoch": 0.694289755940337,
      "grad_norm": 1.9660435914993286,
      "learning_rate": 4.513868422430823e-05,
      "loss": 1.4974,
      "step": 19387
    },
    {
      "epoch": 0.6943255680699053,
      "grad_norm": 1.9675049781799316,
      "learning_rate": 4.5128986933390785e-05,
      "loss": 1.1586,
      "step": 19388
    },
    {
      "epoch": 0.6943613801994736,
      "grad_norm": 2.04257869720459,
      "learning_rate": 4.5119290380692046e-05,
      "loss": 1.4606,
      "step": 19389
    },
    {
      "epoch": 0.6943971923290418,
      "grad_norm": 1.488710641860962,
      "learning_rate": 4.5109594566342515e-05,
      "loss": 1.605,
      "step": 19390
    },
    {
      "epoch": 0.6944330044586101,
      "grad_norm": 1.9784034490585327,
      "learning_rate": 4.50998994904725e-05,
      "loss": 1.1938,
      "step": 19391
    },
    {
      "epoch": 0.6944688165881784,
      "grad_norm": 1.4609148502349854,
      "learning_rate": 4.50902051532126e-05,
      "loss": 1.236,
      "step": 19392
    },
    {
      "epoch": 0.6945046287177467,
      "grad_norm": 4.472887992858887,
      "learning_rate": 4.508051155469312e-05,
      "loss": 1.8098,
      "step": 19393
    },
    {
      "epoch": 0.694540440847315,
      "grad_norm": 1.397859811782837,
      "learning_rate": 4.5070818695044545e-05,
      "loss": 1.4173,
      "step": 19394
    },
    {
      "epoch": 0.6945762529768833,
      "grad_norm": 1.3701924085617065,
      "learning_rate": 4.5061126574397225e-05,
      "loss": 1.1701,
      "step": 19395
    },
    {
      "epoch": 0.6946120651064516,
      "grad_norm": 1.419206976890564,
      "learning_rate": 4.505143519288159e-05,
      "loss": 1.4127,
      "step": 19396
    },
    {
      "epoch": 0.6946478772360198,
      "grad_norm": 1.7711102962493896,
      "learning_rate": 4.504174455062803e-05,
      "loss": 1.4287,
      "step": 19397
    },
    {
      "epoch": 0.6946836893655881,
      "grad_norm": 1.9416470527648926,
      "learning_rate": 4.503205464776689e-05,
      "loss": 1.193,
      "step": 19398
    },
    {
      "epoch": 0.6947195014951564,
      "grad_norm": 1.6379601955413818,
      "learning_rate": 4.502236548442853e-05,
      "loss": 1.1867,
      "step": 19399
    },
    {
      "epoch": 0.6947553136247246,
      "grad_norm": 1.5109336376190186,
      "learning_rate": 4.501267706074335e-05,
      "loss": 1.0608,
      "step": 19400
    },
    {
      "epoch": 0.694791125754293,
      "grad_norm": 1.4156792163848877,
      "learning_rate": 4.5002989376841684e-05,
      "loss": 1.2618,
      "step": 19401
    },
    {
      "epoch": 0.6948269378838613,
      "grad_norm": 1.7646799087524414,
      "learning_rate": 4.499330243285383e-05,
      "loss": 1.5062,
      "step": 19402
    },
    {
      "epoch": 0.6948627500134296,
      "grad_norm": 1.45806884765625,
      "learning_rate": 4.4983616228910144e-05,
      "loss": 1.4862,
      "step": 19403
    },
    {
      "epoch": 0.6948985621429978,
      "grad_norm": 2.211923837661743,
      "learning_rate": 4.497393076514097e-05,
      "loss": 1.3923,
      "step": 19404
    },
    {
      "epoch": 0.6949343742725661,
      "grad_norm": 1.3924918174743652,
      "learning_rate": 4.496424604167654e-05,
      "loss": 1.6712,
      "step": 19405
    },
    {
      "epoch": 0.6949701864021344,
      "grad_norm": 1.4601891040802002,
      "learning_rate": 4.4954562058647187e-05,
      "loss": 1.4383,
      "step": 19406
    },
    {
      "epoch": 0.6950059985317026,
      "grad_norm": 1.7488446235656738,
      "learning_rate": 4.49448788161832e-05,
      "loss": 1.5504,
      "step": 19407
    },
    {
      "epoch": 0.695041810661271,
      "grad_norm": 3.262617349624634,
      "learning_rate": 4.493519631441488e-05,
      "loss": 1.5263,
      "step": 19408
    },
    {
      "epoch": 0.6950776227908393,
      "grad_norm": 1.4852958917617798,
      "learning_rate": 4.492551455347245e-05,
      "loss": 1.5381,
      "step": 19409
    },
    {
      "epoch": 0.6951134349204076,
      "grad_norm": 1.5856446027755737,
      "learning_rate": 4.491583353348616e-05,
      "loss": 1.4643,
      "step": 19410
    },
    {
      "epoch": 0.6951492470499758,
      "grad_norm": 2.515482187271118,
      "learning_rate": 4.490615325458628e-05,
      "loss": 1.7184,
      "step": 19411
    },
    {
      "epoch": 0.6951850591795441,
      "grad_norm": 1.6691640615463257,
      "learning_rate": 4.4896473716903085e-05,
      "loss": 1.5105,
      "step": 19412
    },
    {
      "epoch": 0.6952208713091124,
      "grad_norm": 1.558199405670166,
      "learning_rate": 4.488679492056672e-05,
      "loss": 1.3742,
      "step": 19413
    },
    {
      "epoch": 0.6952566834386806,
      "grad_norm": 1.819074034690857,
      "learning_rate": 4.487711686570744e-05,
      "loss": 1.3235,
      "step": 19414
    },
    {
      "epoch": 0.695292495568249,
      "grad_norm": 2.806718587875366,
      "learning_rate": 4.4867439552455485e-05,
      "loss": 1.4652,
      "step": 19415
    },
    {
      "epoch": 0.6953283076978173,
      "grad_norm": 1.7254616022109985,
      "learning_rate": 4.4857762980940974e-05,
      "loss": 1.3435,
      "step": 19416
    },
    {
      "epoch": 0.6953641198273856,
      "grad_norm": 1.4609277248382568,
      "learning_rate": 4.484808715129414e-05,
      "loss": 1.6299,
      "step": 19417
    },
    {
      "epoch": 0.6953999319569538,
      "grad_norm": 2.2493855953216553,
      "learning_rate": 4.483841206364514e-05,
      "loss": 1.7297,
      "step": 19418
    },
    {
      "epoch": 0.6954357440865221,
      "grad_norm": 1.397451639175415,
      "learning_rate": 4.4828737718124204e-05,
      "loss": 1.3422,
      "step": 19419
    },
    {
      "epoch": 0.6954715562160904,
      "grad_norm": 1.560224175453186,
      "learning_rate": 4.481906411486139e-05,
      "loss": 1.0808,
      "step": 19420
    },
    {
      "epoch": 0.6955073683456586,
      "grad_norm": 1.8856585025787354,
      "learning_rate": 4.480939125398689e-05,
      "loss": 1.556,
      "step": 19421
    },
    {
      "epoch": 0.695543180475227,
      "grad_norm": 1.7038389444351196,
      "learning_rate": 4.479971913563088e-05,
      "loss": 1.2636,
      "step": 19422
    },
    {
      "epoch": 0.6955789926047953,
      "grad_norm": 1.789746642112732,
      "learning_rate": 4.4790047759923406e-05,
      "loss": 1.4515,
      "step": 19423
    },
    {
      "epoch": 0.6956148047343635,
      "grad_norm": 1.6814122200012207,
      "learning_rate": 4.478037712699463e-05,
      "loss": 1.6073,
      "step": 19424
    },
    {
      "epoch": 0.6956506168639318,
      "grad_norm": 1.5029243230819702,
      "learning_rate": 4.477070723697464e-05,
      "loss": 1.3639,
      "step": 19425
    },
    {
      "epoch": 0.6956864289935001,
      "grad_norm": 1.991762638092041,
      "learning_rate": 4.476103808999359e-05,
      "loss": 1.288,
      "step": 19426
    },
    {
      "epoch": 0.6957222411230684,
      "grad_norm": 1.5815709829330444,
      "learning_rate": 4.475136968618149e-05,
      "loss": 1.4307,
      "step": 19427
    },
    {
      "epoch": 0.6957580532526366,
      "grad_norm": 1.5167605876922607,
      "learning_rate": 4.474170202566843e-05,
      "loss": 1.3207,
      "step": 19428
    },
    {
      "epoch": 0.695793865382205,
      "grad_norm": 1.501737356185913,
      "learning_rate": 4.47320351085845e-05,
      "loss": 1.1018,
      "step": 19429
    },
    {
      "epoch": 0.6958296775117733,
      "grad_norm": 1.815674901008606,
      "learning_rate": 4.472236893505978e-05,
      "loss": 1.4576,
      "step": 19430
    },
    {
      "epoch": 0.6958654896413415,
      "grad_norm": 1.786224603652954,
      "learning_rate": 4.471270350522424e-05,
      "loss": 1.2835,
      "step": 19431
    },
    {
      "epoch": 0.6959013017709098,
      "grad_norm": 1.9200912714004517,
      "learning_rate": 4.4703038819207975e-05,
      "loss": 1.4545,
      "step": 19432
    },
    {
      "epoch": 0.6959371139004781,
      "grad_norm": 2.014310359954834,
      "learning_rate": 4.4693374877141015e-05,
      "loss": 1.459,
      "step": 19433
    },
    {
      "epoch": 0.6959729260300463,
      "grad_norm": 1.2802836894989014,
      "learning_rate": 4.4683711679153325e-05,
      "loss": 1.4649,
      "step": 19434
    },
    {
      "epoch": 0.6960087381596146,
      "grad_norm": 1.7526559829711914,
      "learning_rate": 4.467404922537495e-05,
      "loss": 1.3993,
      "step": 19435
    },
    {
      "epoch": 0.696044550289183,
      "grad_norm": 1.5115348100662231,
      "learning_rate": 4.466438751593587e-05,
      "loss": 1.4426,
      "step": 19436
    },
    {
      "epoch": 0.6960803624187513,
      "grad_norm": 2.1051738262176514,
      "learning_rate": 4.465472655096611e-05,
      "loss": 1.542,
      "step": 19437
    },
    {
      "epoch": 0.6961161745483195,
      "grad_norm": 1.828957438468933,
      "learning_rate": 4.464506633059559e-05,
      "loss": 1.3994,
      "step": 19438
    },
    {
      "epoch": 0.6961519866778878,
      "grad_norm": 1.8577128648757935,
      "learning_rate": 4.463540685495429e-05,
      "loss": 1.5155,
      "step": 19439
    },
    {
      "epoch": 0.6961877988074561,
      "grad_norm": 1.6411844491958618,
      "learning_rate": 4.4625748124172204e-05,
      "loss": 1.7316,
      "step": 19440
    },
    {
      "epoch": 0.6962236109370243,
      "grad_norm": 1.6481902599334717,
      "learning_rate": 4.461609013837923e-05,
      "loss": 1.1623,
      "step": 19441
    },
    {
      "epoch": 0.6962594230665926,
      "grad_norm": 1.815705418586731,
      "learning_rate": 4.460643289770532e-05,
      "loss": 1.5606,
      "step": 19442
    },
    {
      "epoch": 0.696295235196161,
      "grad_norm": 2.0571019649505615,
      "learning_rate": 4.4596776402280396e-05,
      "loss": 1.4951,
      "step": 19443
    },
    {
      "epoch": 0.6963310473257293,
      "grad_norm": 1.345685362815857,
      "learning_rate": 4.458712065223442e-05,
      "loss": 1.5385,
      "step": 19444
    },
    {
      "epoch": 0.6963668594552975,
      "grad_norm": 2.47963285446167,
      "learning_rate": 4.4577465647697223e-05,
      "loss": 1.3012,
      "step": 19445
    },
    {
      "epoch": 0.6964026715848658,
      "grad_norm": 1.4152905941009521,
      "learning_rate": 4.456781138879873e-05,
      "loss": 1.0918,
      "step": 19446
    },
    {
      "epoch": 0.6964384837144341,
      "grad_norm": 1.7984778881072998,
      "learning_rate": 4.455815787566884e-05,
      "loss": 1.6954,
      "step": 19447
    },
    {
      "epoch": 0.6964742958440023,
      "grad_norm": 1.517166018486023,
      "learning_rate": 4.454850510843745e-05,
      "loss": 1.5157,
      "step": 19448
    },
    {
      "epoch": 0.6965101079735706,
      "grad_norm": 1.8566745519638062,
      "learning_rate": 4.45388530872344e-05,
      "loss": 1.1924,
      "step": 19449
    },
    {
      "epoch": 0.696545920103139,
      "grad_norm": 1.6976255178451538,
      "learning_rate": 4.452920181218947e-05,
      "loss": 1.5863,
      "step": 19450
    },
    {
      "epoch": 0.6965817322327073,
      "grad_norm": 1.6378042697906494,
      "learning_rate": 4.451955128343266e-05,
      "loss": 1.5929,
      "step": 19451
    },
    {
      "epoch": 0.6966175443622755,
      "grad_norm": 1.7423259019851685,
      "learning_rate": 4.450990150109367e-05,
      "loss": 1.5762,
      "step": 19452
    },
    {
      "epoch": 0.6966533564918438,
      "grad_norm": 1.9417845010757446,
      "learning_rate": 4.4500252465302384e-05,
      "loss": 1.7302,
      "step": 19453
    },
    {
      "epoch": 0.6966891686214121,
      "grad_norm": 1.3983720541000366,
      "learning_rate": 4.449060417618861e-05,
      "loss": 1.582,
      "step": 19454
    },
    {
      "epoch": 0.6967249807509803,
      "grad_norm": 1.4114207029342651,
      "learning_rate": 4.44809566338822e-05,
      "loss": 1.5144,
      "step": 19455
    },
    {
      "epoch": 0.6967607928805486,
      "grad_norm": 1.3205265998840332,
      "learning_rate": 4.447130983851285e-05,
      "loss": 1.6001,
      "step": 19456
    },
    {
      "epoch": 0.696796605010117,
      "grad_norm": 1.3919918537139893,
      "learning_rate": 4.446166379021042e-05,
      "loss": 1.5457,
      "step": 19457
    },
    {
      "epoch": 0.6968324171396852,
      "grad_norm": 1.9750583171844482,
      "learning_rate": 4.4452018489104684e-05,
      "loss": 1.4017,
      "step": 19458
    },
    {
      "epoch": 0.6968682292692535,
      "grad_norm": 1.6426039934158325,
      "learning_rate": 4.4442373935325364e-05,
      "loss": 1.3513,
      "step": 19459
    },
    {
      "epoch": 0.6969040413988218,
      "grad_norm": 1.6933245658874512,
      "learning_rate": 4.4432730129002265e-05,
      "loss": 1.542,
      "step": 19460
    },
    {
      "epoch": 0.69693985352839,
      "grad_norm": 1.6183847188949585,
      "learning_rate": 4.442308707026504e-05,
      "loss": 1.3853,
      "step": 19461
    },
    {
      "epoch": 0.6969756656579583,
      "grad_norm": 1.628896713256836,
      "learning_rate": 4.4413444759243564e-05,
      "loss": 1.4663,
      "step": 19462
    },
    {
      "epoch": 0.6970114777875266,
      "grad_norm": 1.992426872253418,
      "learning_rate": 4.440380319606744e-05,
      "loss": 2.0699,
      "step": 19463
    },
    {
      "epoch": 0.697047289917095,
      "grad_norm": 1.5472890138626099,
      "learning_rate": 4.439416238086643e-05,
      "loss": 1.3658,
      "step": 19464
    },
    {
      "epoch": 0.6970831020466632,
      "grad_norm": 1.2758383750915527,
      "learning_rate": 4.438452231377025e-05,
      "loss": 1.4012,
      "step": 19465
    },
    {
      "epoch": 0.6971189141762315,
      "grad_norm": 1.9764435291290283,
      "learning_rate": 4.4374882994908615e-05,
      "loss": 1.5115,
      "step": 19466
    },
    {
      "epoch": 0.6971547263057998,
      "grad_norm": 1.4551410675048828,
      "learning_rate": 4.436524442441118e-05,
      "loss": 1.1521,
      "step": 19467
    },
    {
      "epoch": 0.697190538435368,
      "grad_norm": 1.883821964263916,
      "learning_rate": 4.435560660240754e-05,
      "loss": 1.1386,
      "step": 19468
    },
    {
      "epoch": 0.6972263505649363,
      "grad_norm": 1.3271210193634033,
      "learning_rate": 4.434596952902752e-05,
      "loss": 1.5645,
      "step": 19469
    },
    {
      "epoch": 0.6972621626945046,
      "grad_norm": 1.4583014249801636,
      "learning_rate": 4.433633320440064e-05,
      "loss": 1.4722,
      "step": 19470
    },
    {
      "epoch": 0.697297974824073,
      "grad_norm": 1.2590141296386719,
      "learning_rate": 4.432669762865664e-05,
      "loss": 1.6447,
      "step": 19471
    },
    {
      "epoch": 0.6973337869536412,
      "grad_norm": 1.6529279947280884,
      "learning_rate": 4.431706280192503e-05,
      "loss": 1.2192,
      "step": 19472
    },
    {
      "epoch": 0.6973695990832095,
      "grad_norm": 1.5791985988616943,
      "learning_rate": 4.4307428724335595e-05,
      "loss": 1.4652,
      "step": 19473
    },
    {
      "epoch": 0.6974054112127778,
      "grad_norm": 1.7167974710464478,
      "learning_rate": 4.429779539601787e-05,
      "loss": 1.4894,
      "step": 19474
    },
    {
      "epoch": 0.697441223342346,
      "grad_norm": 2.156100273132324,
      "learning_rate": 4.428816281710142e-05,
      "loss": 1.4869,
      "step": 19475
    },
    {
      "epoch": 0.6974770354719143,
      "grad_norm": 1.4207909107208252,
      "learning_rate": 4.427853098771587e-05,
      "loss": 1.5364,
      "step": 19476
    },
    {
      "epoch": 0.6975128476014826,
      "grad_norm": 1.4815733432769775,
      "learning_rate": 4.426889990799082e-05,
      "loss": 1.434,
      "step": 19477
    },
    {
      "epoch": 0.697548659731051,
      "grad_norm": 1.5822980403900146,
      "learning_rate": 4.425926957805586e-05,
      "loss": 0.9941,
      "step": 19478
    },
    {
      "epoch": 0.6975844718606192,
      "grad_norm": 1.9064446687698364,
      "learning_rate": 4.424963999804046e-05,
      "loss": 1.3855,
      "step": 19479
    },
    {
      "epoch": 0.6976202839901875,
      "grad_norm": 2.002734422683716,
      "learning_rate": 4.4240011168074315e-05,
      "loss": 1.2231,
      "step": 19480
    },
    {
      "epoch": 0.6976560961197558,
      "grad_norm": 1.4305377006530762,
      "learning_rate": 4.423038308828685e-05,
      "loss": 1.5345,
      "step": 19481
    },
    {
      "epoch": 0.697691908249324,
      "grad_norm": 1.573127031326294,
      "learning_rate": 4.4220755758807695e-05,
      "loss": 1.2691,
      "step": 19482
    },
    {
      "epoch": 0.6977277203788923,
      "grad_norm": 1.9930295944213867,
      "learning_rate": 4.421112917976628e-05,
      "loss": 1.402,
      "step": 19483
    },
    {
      "epoch": 0.6977635325084606,
      "grad_norm": 1.787060022354126,
      "learning_rate": 4.420150335129215e-05,
      "loss": 1.0692,
      "step": 19484
    },
    {
      "epoch": 0.697799344638029,
      "grad_norm": 1.6993510723114014,
      "learning_rate": 4.419187827351485e-05,
      "loss": 1.5472,
      "step": 19485
    },
    {
      "epoch": 0.6978351567675972,
      "grad_norm": 1.5781887769699097,
      "learning_rate": 4.418225394656382e-05,
      "loss": 1.5185,
      "step": 19486
    },
    {
      "epoch": 0.6978709688971655,
      "grad_norm": 2.3409838676452637,
      "learning_rate": 4.417263037056856e-05,
      "loss": 1.5516,
      "step": 19487
    },
    {
      "epoch": 0.6979067810267338,
      "grad_norm": 1.6964595317840576,
      "learning_rate": 4.416300754565854e-05,
      "loss": 1.6612,
      "step": 19488
    },
    {
      "epoch": 0.697942593156302,
      "grad_norm": 1.3110731840133667,
      "learning_rate": 4.415338547196326e-05,
      "loss": 1.2798,
      "step": 19489
    },
    {
      "epoch": 0.6979784052858703,
      "grad_norm": 1.5717277526855469,
      "learning_rate": 4.414376414961208e-05,
      "loss": 1.5264,
      "step": 19490
    },
    {
      "epoch": 0.6980142174154386,
      "grad_norm": 1.8437398672103882,
      "learning_rate": 4.4134143578734576e-05,
      "loss": 1.4216,
      "step": 19491
    },
    {
      "epoch": 0.698050029545007,
      "grad_norm": 1.6376128196716309,
      "learning_rate": 4.41245237594601e-05,
      "loss": 1.3332,
      "step": 19492
    },
    {
      "epoch": 0.6980858416745752,
      "grad_norm": 1.9958080053329468,
      "learning_rate": 4.411490469191806e-05,
      "loss": 1.8033,
      "step": 19493
    },
    {
      "epoch": 0.6981216538041435,
      "grad_norm": 1.52802574634552,
      "learning_rate": 4.4105286376237874e-05,
      "loss": 1.495,
      "step": 19494
    },
    {
      "epoch": 0.6981574659337118,
      "grad_norm": 1.652212142944336,
      "learning_rate": 4.409566881254897e-05,
      "loss": 1.5077,
      "step": 19495
    },
    {
      "epoch": 0.69819327806328,
      "grad_norm": 1.9880157709121704,
      "learning_rate": 4.408605200098077e-05,
      "loss": 1.7005,
      "step": 19496
    },
    {
      "epoch": 0.6982290901928483,
      "grad_norm": 1.7847472429275513,
      "learning_rate": 4.407643594166257e-05,
      "loss": 1.4118,
      "step": 19497
    },
    {
      "epoch": 0.6982649023224166,
      "grad_norm": 1.8568692207336426,
      "learning_rate": 4.4066820634723805e-05,
      "loss": 1.539,
      "step": 19498
    },
    {
      "epoch": 0.6983007144519849,
      "grad_norm": 1.3895604610443115,
      "learning_rate": 4.405720608029381e-05,
      "loss": 1.5283,
      "step": 19499
    },
    {
      "epoch": 0.6983365265815532,
      "grad_norm": 1.667009949684143,
      "learning_rate": 4.404759227850198e-05,
      "loss": 1.3788,
      "step": 19500
    },
    {
      "epoch": 0.6983723387111215,
      "grad_norm": 1.8755829334259033,
      "learning_rate": 4.403797922947759e-05,
      "loss": 1.5928,
      "step": 19501
    },
    {
      "epoch": 0.6984081508406897,
      "grad_norm": 1.7361255884170532,
      "learning_rate": 4.4028366933349996e-05,
      "loss": 1.3978,
      "step": 19502
    },
    {
      "epoch": 0.698443962970258,
      "grad_norm": 1.4138158559799194,
      "learning_rate": 4.4018755390248566e-05,
      "loss": 1.3812,
      "step": 19503
    },
    {
      "epoch": 0.6984797750998263,
      "grad_norm": 1.610290765762329,
      "learning_rate": 4.400914460030254e-05,
      "loss": 1.708,
      "step": 19504
    },
    {
      "epoch": 0.6985155872293946,
      "grad_norm": 1.9601976871490479,
      "learning_rate": 4.3999534563641253e-05,
      "loss": 1.4285,
      "step": 19505
    },
    {
      "epoch": 0.6985513993589629,
      "grad_norm": 1.4728658199310303,
      "learning_rate": 4.3989925280393986e-05,
      "loss": 1.671,
      "step": 19506
    },
    {
      "epoch": 0.6985872114885312,
      "grad_norm": 1.6432338953018188,
      "learning_rate": 4.3980316750690065e-05,
      "loss": 1.3902,
      "step": 19507
    },
    {
      "epoch": 0.6986230236180995,
      "grad_norm": 1.4036003351211548,
      "learning_rate": 4.397070897465869e-05,
      "loss": 1.653,
      "step": 19508
    },
    {
      "epoch": 0.6986588357476677,
      "grad_norm": 1.8093888759613037,
      "learning_rate": 4.396110195242915e-05,
      "loss": 1.6891,
      "step": 19509
    },
    {
      "epoch": 0.698694647877236,
      "grad_norm": 1.9804387092590332,
      "learning_rate": 4.395149568413073e-05,
      "loss": 1.5497,
      "step": 19510
    },
    {
      "epoch": 0.6987304600068043,
      "grad_norm": 1.2982250452041626,
      "learning_rate": 4.394189016989261e-05,
      "loss": 1.3489,
      "step": 19511
    },
    {
      "epoch": 0.6987662721363725,
      "grad_norm": 1.9783587455749512,
      "learning_rate": 4.3932285409844046e-05,
      "loss": 1.4654,
      "step": 19512
    },
    {
      "epoch": 0.6988020842659409,
      "grad_norm": 2.4322969913482666,
      "learning_rate": 4.392268140411425e-05,
      "loss": 1.55,
      "step": 19513
    },
    {
      "epoch": 0.6988378963955092,
      "grad_norm": 2.154332160949707,
      "learning_rate": 4.391307815283249e-05,
      "loss": 1.8838,
      "step": 19514
    },
    {
      "epoch": 0.6988737085250775,
      "grad_norm": 1.7512561082839966,
      "learning_rate": 4.390347565612787e-05,
      "loss": 1.6804,
      "step": 19515
    },
    {
      "epoch": 0.6989095206546457,
      "grad_norm": 1.7126457691192627,
      "learning_rate": 4.3893873914129635e-05,
      "loss": 1.0483,
      "step": 19516
    },
    {
      "epoch": 0.698945332784214,
      "grad_norm": 1.6176780462265015,
      "learning_rate": 4.388427292696695e-05,
      "loss": 1.2061,
      "step": 19517
    },
    {
      "epoch": 0.6989811449137823,
      "grad_norm": 1.496763825416565,
      "learning_rate": 4.387467269476902e-05,
      "loss": 1.4142,
      "step": 19518
    },
    {
      "epoch": 0.6990169570433505,
      "grad_norm": 2.7602691650390625,
      "learning_rate": 4.3865073217664944e-05,
      "loss": 1.5354,
      "step": 19519
    },
    {
      "epoch": 0.6990527691729189,
      "grad_norm": 2.1753737926483154,
      "learning_rate": 4.38554744957839e-05,
      "loss": 1.5896,
      "step": 19520
    },
    {
      "epoch": 0.6990885813024872,
      "grad_norm": 2.356755018234253,
      "learning_rate": 4.384587652925506e-05,
      "loss": 1.5859,
      "step": 19521
    },
    {
      "epoch": 0.6991243934320555,
      "grad_norm": 1.905924677848816,
      "learning_rate": 4.383627931820747e-05,
      "loss": 1.5317,
      "step": 19522
    },
    {
      "epoch": 0.6991602055616237,
      "grad_norm": 2.0932862758636475,
      "learning_rate": 4.382668286277031e-05,
      "loss": 1.7417,
      "step": 19523
    },
    {
      "epoch": 0.699196017691192,
      "grad_norm": 1.5198270082473755,
      "learning_rate": 4.381708716307267e-05,
      "loss": 1.3734,
      "step": 19524
    },
    {
      "epoch": 0.6992318298207603,
      "grad_norm": 1.43095862865448,
      "learning_rate": 4.3807492219243686e-05,
      "loss": 1.4217,
      "step": 19525
    },
    {
      "epoch": 0.6992676419503285,
      "grad_norm": 2.038198947906494,
      "learning_rate": 4.379789803141238e-05,
      "loss": 1.6064,
      "step": 19526
    },
    {
      "epoch": 0.6993034540798969,
      "grad_norm": 1.9979199171066284,
      "learning_rate": 4.378830459970785e-05,
      "loss": 1.5887,
      "step": 19527
    },
    {
      "epoch": 0.6993392662094652,
      "grad_norm": 1.6111292839050293,
      "learning_rate": 4.3778711924259216e-05,
      "loss": 1.5713,
      "step": 19528
    },
    {
      "epoch": 0.6993750783390335,
      "grad_norm": 1.396235704421997,
      "learning_rate": 4.3769120005195465e-05,
      "loss": 1.1932,
      "step": 19529
    },
    {
      "epoch": 0.6994108904686017,
      "grad_norm": 1.9758912324905396,
      "learning_rate": 4.375952884264566e-05,
      "loss": 1.4702,
      "step": 19530
    },
    {
      "epoch": 0.69944670259817,
      "grad_norm": 1.5198663473129272,
      "learning_rate": 4.374993843673886e-05,
      "loss": 1.1155,
      "step": 19531
    },
    {
      "epoch": 0.6994825147277383,
      "grad_norm": 1.6794745922088623,
      "learning_rate": 4.37403487876041e-05,
      "loss": 1.3603,
      "step": 19532
    },
    {
      "epoch": 0.6995183268573065,
      "grad_norm": 1.6363730430603027,
      "learning_rate": 4.373075989537035e-05,
      "loss": 1.4035,
      "step": 19533
    },
    {
      "epoch": 0.6995541389868749,
      "grad_norm": 1.9012279510498047,
      "learning_rate": 4.372117176016665e-05,
      "loss": 1.3656,
      "step": 19534
    },
    {
      "epoch": 0.6995899511164432,
      "grad_norm": 1.7077008485794067,
      "learning_rate": 4.371158438212199e-05,
      "loss": 1.2935,
      "step": 19535
    },
    {
      "epoch": 0.6996257632460114,
      "grad_norm": 1.684689998626709,
      "learning_rate": 4.370199776136538e-05,
      "loss": 1.5655,
      "step": 19536
    },
    {
      "epoch": 0.6996615753755797,
      "grad_norm": 2.581475019454956,
      "learning_rate": 4.3692411898025746e-05,
      "loss": 1.4285,
      "step": 19537
    },
    {
      "epoch": 0.699697387505148,
      "grad_norm": 2.067708730697632,
      "learning_rate": 4.368282679223207e-05,
      "loss": 1.7469,
      "step": 19538
    },
    {
      "epoch": 0.6997331996347163,
      "grad_norm": 1.2996182441711426,
      "learning_rate": 4.367324244411335e-05,
      "loss": 1.0087,
      "step": 19539
    },
    {
      "epoch": 0.6997690117642845,
      "grad_norm": 1.6188222169876099,
      "learning_rate": 4.3663658853798476e-05,
      "loss": 1.4346,
      "step": 19540
    },
    {
      "epoch": 0.6998048238938529,
      "grad_norm": 1.531131386756897,
      "learning_rate": 4.365407602141639e-05,
      "loss": 1.5322,
      "step": 19541
    },
    {
      "epoch": 0.6998406360234212,
      "grad_norm": 2.2801246643066406,
      "learning_rate": 4.364449394709603e-05,
      "loss": 1.7024,
      "step": 19542
    },
    {
      "epoch": 0.6998764481529894,
      "grad_norm": 2.3423590660095215,
      "learning_rate": 4.363491263096635e-05,
      "loss": 1.7841,
      "step": 19543
    },
    {
      "epoch": 0.6999122602825577,
      "grad_norm": 1.4121369123458862,
      "learning_rate": 4.362533207315618e-05,
      "loss": 1.5293,
      "step": 19544
    },
    {
      "epoch": 0.699948072412126,
      "grad_norm": 2.1785571575164795,
      "learning_rate": 4.361575227379444e-05,
      "loss": 1.5964,
      "step": 19545
    },
    {
      "epoch": 0.6999838845416942,
      "grad_norm": 1.5997748374938965,
      "learning_rate": 4.360617323301007e-05,
      "loss": 1.5425,
      "step": 19546
    },
    {
      "epoch": 0.7000196966712625,
      "grad_norm": 2.7810516357421875,
      "learning_rate": 4.359659495093186e-05,
      "loss": 1.6532,
      "step": 19547
    },
    {
      "epoch": 0.7000555088008309,
      "grad_norm": 1.4571328163146973,
      "learning_rate": 4.35870174276887e-05,
      "loss": 1.472,
      "step": 19548
    },
    {
      "epoch": 0.7000913209303992,
      "grad_norm": 2.0664026737213135,
      "learning_rate": 4.357744066340946e-05,
      "loss": 1.59,
      "step": 19549
    },
    {
      "epoch": 0.7001271330599674,
      "grad_norm": 1.6976768970489502,
      "learning_rate": 4.356786465822301e-05,
      "loss": 1.6835,
      "step": 19550
    },
    {
      "epoch": 0.7001629451895357,
      "grad_norm": 1.4122880697250366,
      "learning_rate": 4.3558289412258114e-05,
      "loss": 1.3165,
      "step": 19551
    },
    {
      "epoch": 0.700198757319104,
      "grad_norm": 1.928450584411621,
      "learning_rate": 4.354871492564363e-05,
      "loss": 1.4275,
      "step": 19552
    },
    {
      "epoch": 0.7002345694486722,
      "grad_norm": 1.8557058572769165,
      "learning_rate": 4.353914119850837e-05,
      "loss": 1.0927,
      "step": 19553
    },
    {
      "epoch": 0.7002703815782405,
      "grad_norm": 1.8503695726394653,
      "learning_rate": 4.3529568230981165e-05,
      "loss": 1.1398,
      "step": 19554
    },
    {
      "epoch": 0.7003061937078089,
      "grad_norm": 1.4184296131134033,
      "learning_rate": 4.351999602319079e-05,
      "loss": 1.5463,
      "step": 19555
    },
    {
      "epoch": 0.7003420058373772,
      "grad_norm": 2.062753200531006,
      "learning_rate": 4.351042457526594e-05,
      "loss": 1.2678,
      "step": 19556
    },
    {
      "epoch": 0.7003778179669454,
      "grad_norm": 1.8937575817108154,
      "learning_rate": 4.350085388733553e-05,
      "loss": 1.1757,
      "step": 19557
    },
    {
      "epoch": 0.7004136300965137,
      "grad_norm": 1.4495995044708252,
      "learning_rate": 4.349128395952821e-05,
      "loss": 1.4901,
      "step": 19558
    },
    {
      "epoch": 0.700449442226082,
      "grad_norm": 1.6838757991790771,
      "learning_rate": 4.3481714791972816e-05,
      "loss": 1.8831,
      "step": 19559
    },
    {
      "epoch": 0.7004852543556502,
      "grad_norm": 1.3522682189941406,
      "learning_rate": 4.3472146384797973e-05,
      "loss": 1.381,
      "step": 19560
    },
    {
      "epoch": 0.7005210664852185,
      "grad_norm": 1.5966675281524658,
      "learning_rate": 4.3462578738132557e-05,
      "loss": 1.2488,
      "step": 19561
    },
    {
      "epoch": 0.7005568786147869,
      "grad_norm": 1.5900689363479614,
      "learning_rate": 4.345301185210517e-05,
      "loss": 1.3429,
      "step": 19562
    },
    {
      "epoch": 0.7005926907443552,
      "grad_norm": 2.1359074115753174,
      "learning_rate": 4.344344572684459e-05,
      "loss": 1.5794,
      "step": 19563
    },
    {
      "epoch": 0.7006285028739234,
      "grad_norm": 1.512988567352295,
      "learning_rate": 4.343388036247952e-05,
      "loss": 1.2947,
      "step": 19564
    },
    {
      "epoch": 0.7006643150034917,
      "grad_norm": 1.4352556467056274,
      "learning_rate": 4.342431575913858e-05,
      "loss": 1.2725,
      "step": 19565
    },
    {
      "epoch": 0.70070012713306,
      "grad_norm": 1.7753974199295044,
      "learning_rate": 4.341475191695054e-05,
      "loss": 1.6684,
      "step": 19566
    },
    {
      "epoch": 0.7007359392626282,
      "grad_norm": 1.5111207962036133,
      "learning_rate": 4.340518883604395e-05,
      "loss": 1.3503,
      "step": 19567
    },
    {
      "epoch": 0.7007717513921965,
      "grad_norm": 1.4870178699493408,
      "learning_rate": 4.339562651654761e-05,
      "loss": 1.176,
      "step": 19568
    },
    {
      "epoch": 0.7008075635217649,
      "grad_norm": 1.6527915000915527,
      "learning_rate": 4.338606495859007e-05,
      "loss": 1.715,
      "step": 19569
    },
    {
      "epoch": 0.7008433756513331,
      "grad_norm": 2.162165403366089,
      "learning_rate": 4.3376504162300035e-05,
      "loss": 1.7323,
      "step": 19570
    },
    {
      "epoch": 0.7008791877809014,
      "grad_norm": 1.7663673162460327,
      "learning_rate": 4.336694412780605e-05,
      "loss": 1.3105,
      "step": 19571
    },
    {
      "epoch": 0.7009149999104697,
      "grad_norm": 1.6505602598190308,
      "learning_rate": 4.3357384855236796e-05,
      "loss": 1.3928,
      "step": 19572
    },
    {
      "epoch": 0.700950812040038,
      "grad_norm": 1.5301345586776733,
      "learning_rate": 4.33478263447209e-05,
      "loss": 1.2207,
      "step": 19573
    },
    {
      "epoch": 0.7009866241696062,
      "grad_norm": 1.6930309534072876,
      "learning_rate": 4.333826859638684e-05,
      "loss": 1.603,
      "step": 19574
    },
    {
      "epoch": 0.7010224362991745,
      "grad_norm": 1.439517855644226,
      "learning_rate": 4.332871161036337e-05,
      "loss": 1.0671,
      "step": 19575
    },
    {
      "epoch": 0.7010582484287429,
      "grad_norm": 1.5984947681427002,
      "learning_rate": 4.331915538677894e-05,
      "loss": 1.4029,
      "step": 19576
    },
    {
      "epoch": 0.7010940605583111,
      "grad_norm": 1.5472118854522705,
      "learning_rate": 4.3309599925762214e-05,
      "loss": 1.2994,
      "step": 19577
    },
    {
      "epoch": 0.7011298726878794,
      "grad_norm": 1.2958440780639648,
      "learning_rate": 4.330004522744161e-05,
      "loss": 1.6099,
      "step": 19578
    },
    {
      "epoch": 0.7011656848174477,
      "grad_norm": 1.5507088899612427,
      "learning_rate": 4.329049129194583e-05,
      "loss": 1.4012,
      "step": 19579
    },
    {
      "epoch": 0.701201496947016,
      "grad_norm": 2.030113935470581,
      "learning_rate": 4.3280938119403346e-05,
      "loss": 1.4628,
      "step": 19580
    },
    {
      "epoch": 0.7012373090765842,
      "grad_norm": 1.508935570716858,
      "learning_rate": 4.3271385709942636e-05,
      "loss": 1.3467,
      "step": 19581
    },
    {
      "epoch": 0.7012731212061525,
      "grad_norm": 1.4302729368209839,
      "learning_rate": 4.326183406369226e-05,
      "loss": 1.3545,
      "step": 19582
    },
    {
      "epoch": 0.7013089333357209,
      "grad_norm": 2.1700620651245117,
      "learning_rate": 4.325228318078073e-05,
      "loss": 1.4485,
      "step": 19583
    },
    {
      "epoch": 0.7013447454652891,
      "grad_norm": 1.713890790939331,
      "learning_rate": 4.324273306133655e-05,
      "loss": 1.7539,
      "step": 19584
    },
    {
      "epoch": 0.7013805575948574,
      "grad_norm": 1.60768723487854,
      "learning_rate": 4.3233183705488156e-05,
      "loss": 1.1099,
      "step": 19585
    },
    {
      "epoch": 0.7014163697244257,
      "grad_norm": 2.0066118240356445,
      "learning_rate": 4.322363511336405e-05,
      "loss": 1.4433,
      "step": 19586
    },
    {
      "epoch": 0.7014521818539939,
      "grad_norm": 2.282560110092163,
      "learning_rate": 4.32140872850927e-05,
      "loss": 1.2975,
      "step": 19587
    },
    {
      "epoch": 0.7014879939835622,
      "grad_norm": 1.573010802268982,
      "learning_rate": 4.320454022080259e-05,
      "loss": 1.4643,
      "step": 19588
    },
    {
      "epoch": 0.7015238061131305,
      "grad_norm": 1.5037258863449097,
      "learning_rate": 4.3194993920622095e-05,
      "loss": 1.5382,
      "step": 19589
    },
    {
      "epoch": 0.7015596182426989,
      "grad_norm": 1.8554325103759766,
      "learning_rate": 4.318544838467968e-05,
      "loss": 1.5909,
      "step": 19590
    },
    {
      "epoch": 0.7015954303722671,
      "grad_norm": 2.168818950653076,
      "learning_rate": 4.3175903613103815e-05,
      "loss": 1.6751,
      "step": 19591
    },
    {
      "epoch": 0.7016312425018354,
      "grad_norm": 1.7133216857910156,
      "learning_rate": 4.316635960602283e-05,
      "loss": 1.2098,
      "step": 19592
    },
    {
      "epoch": 0.7016670546314037,
      "grad_norm": 1.6379142999649048,
      "learning_rate": 4.3156816363565166e-05,
      "loss": 1.4609,
      "step": 19593
    },
    {
      "epoch": 0.7017028667609719,
      "grad_norm": 1.4387273788452148,
      "learning_rate": 4.3147273885859215e-05,
      "loss": 1.1765,
      "step": 19594
    },
    {
      "epoch": 0.7017386788905402,
      "grad_norm": 1.2745956182479858,
      "learning_rate": 4.3137732173033394e-05,
      "loss": 1.553,
      "step": 19595
    },
    {
      "epoch": 0.7017744910201085,
      "grad_norm": 1.6397004127502441,
      "learning_rate": 4.3128191225216005e-05,
      "loss": 1.2392,
      "step": 19596
    },
    {
      "epoch": 0.7018103031496769,
      "grad_norm": 1.4887009859085083,
      "learning_rate": 4.3118651042535444e-05,
      "loss": 1.3739,
      "step": 19597
    },
    {
      "epoch": 0.7018461152792451,
      "grad_norm": 1.4074066877365112,
      "learning_rate": 4.31091116251201e-05,
      "loss": 1.4146,
      "step": 19598
    },
    {
      "epoch": 0.7018819274088134,
      "grad_norm": 1.663396954536438,
      "learning_rate": 4.3099572973098236e-05,
      "loss": 1.4598,
      "step": 19599
    },
    {
      "epoch": 0.7019177395383817,
      "grad_norm": 2.8375847339630127,
      "learning_rate": 4.309003508659822e-05,
      "loss": 1.3936,
      "step": 19600
    },
    {
      "epoch": 0.7019535516679499,
      "grad_norm": 1.628321886062622,
      "learning_rate": 4.3080497965748376e-05,
      "loss": 1.7446,
      "step": 19601
    },
    {
      "epoch": 0.7019893637975182,
      "grad_norm": 1.359993815422058,
      "learning_rate": 4.307096161067704e-05,
      "loss": 1.1517,
      "step": 19602
    },
    {
      "epoch": 0.7020251759270865,
      "grad_norm": 1.599370002746582,
      "learning_rate": 4.3061426021512453e-05,
      "loss": 1.2006,
      "step": 19603
    },
    {
      "epoch": 0.7020609880566548,
      "grad_norm": 1.3849345445632935,
      "learning_rate": 4.305189119838293e-05,
      "loss": 1.2563,
      "step": 19604
    },
    {
      "epoch": 0.7020968001862231,
      "grad_norm": 1.7541930675506592,
      "learning_rate": 4.304235714141677e-05,
      "loss": 1.5481,
      "step": 19605
    },
    {
      "epoch": 0.7021326123157914,
      "grad_norm": 1.3335884809494019,
      "learning_rate": 4.303282385074224e-05,
      "loss": 1.4315,
      "step": 19606
    },
    {
      "epoch": 0.7021684244453597,
      "grad_norm": 1.3472892045974731,
      "learning_rate": 4.3023291326487556e-05,
      "loss": 1.4085,
      "step": 19607
    },
    {
      "epoch": 0.7022042365749279,
      "grad_norm": 1.5135165452957153,
      "learning_rate": 4.301375956878099e-05,
      "loss": 1.0491,
      "step": 19608
    },
    {
      "epoch": 0.7022400487044962,
      "grad_norm": 2.763059616088867,
      "learning_rate": 4.300422857775081e-05,
      "loss": 1.5307,
      "step": 19609
    },
    {
      "epoch": 0.7022758608340645,
      "grad_norm": 1.5471773147583008,
      "learning_rate": 4.2994698353525184e-05,
      "loss": 1.5784,
      "step": 19610
    },
    {
      "epoch": 0.7023116729636327,
      "grad_norm": 1.618385910987854,
      "learning_rate": 4.298516889623238e-05,
      "loss": 1.3744,
      "step": 19611
    },
    {
      "epoch": 0.7023474850932011,
      "grad_norm": 2.3037314414978027,
      "learning_rate": 4.297564020600056e-05,
      "loss": 1.6599,
      "step": 19612
    },
    {
      "epoch": 0.7023832972227694,
      "grad_norm": 2.060870885848999,
      "learning_rate": 4.2966112282957985e-05,
      "loss": 1.4654,
      "step": 19613
    },
    {
      "epoch": 0.7024191093523376,
      "grad_norm": 1.506044864654541,
      "learning_rate": 4.295658512723277e-05,
      "loss": 1.1635,
      "step": 19614
    },
    {
      "epoch": 0.7024549214819059,
      "grad_norm": 1.5053657293319702,
      "learning_rate": 4.2947058738953115e-05,
      "loss": 1.5164,
      "step": 19615
    },
    {
      "epoch": 0.7024907336114742,
      "grad_norm": 1.444291591644287,
      "learning_rate": 4.293753311824724e-05,
      "loss": 1.4666,
      "step": 19616
    },
    {
      "epoch": 0.7025265457410425,
      "grad_norm": 1.5131818056106567,
      "learning_rate": 4.2928008265243205e-05,
      "loss": 1.1814,
      "step": 19617
    },
    {
      "epoch": 0.7025623578706107,
      "grad_norm": 1.3681418895721436,
      "learning_rate": 4.2918484180069205e-05,
      "loss": 1.2736,
      "step": 19618
    },
    {
      "epoch": 0.7025981700001791,
      "grad_norm": 1.5410593748092651,
      "learning_rate": 4.290896086285338e-05,
      "loss": 1.0753,
      "step": 19619
    },
    {
      "epoch": 0.7026339821297474,
      "grad_norm": 1.949423909187317,
      "learning_rate": 4.289943831372386e-05,
      "loss": 1.4629,
      "step": 19620
    },
    {
      "epoch": 0.7026697942593156,
      "grad_norm": 1.885960578918457,
      "learning_rate": 4.2889916532808716e-05,
      "loss": 1.6269,
      "step": 19621
    },
    {
      "epoch": 0.7027056063888839,
      "grad_norm": 1.5690264701843262,
      "learning_rate": 4.2880395520236086e-05,
      "loss": 0.934,
      "step": 19622
    },
    {
      "epoch": 0.7027414185184522,
      "grad_norm": 1.42863130569458,
      "learning_rate": 4.287087527613405e-05,
      "loss": 1.5016,
      "step": 19623
    },
    {
      "epoch": 0.7027772306480204,
      "grad_norm": 1.8398011922836304,
      "learning_rate": 4.2861355800630734e-05,
      "loss": 1.6303,
      "step": 19624
    },
    {
      "epoch": 0.7028130427775887,
      "grad_norm": 1.50557541847229,
      "learning_rate": 4.285183709385413e-05,
      "loss": 1.5899,
      "step": 19625
    },
    {
      "epoch": 0.7028488549071571,
      "grad_norm": 2.3311352729797363,
      "learning_rate": 4.284231915593234e-05,
      "loss": 1.3451,
      "step": 19626
    },
    {
      "epoch": 0.7028846670367254,
      "grad_norm": 1.7532374858856201,
      "learning_rate": 4.283280198699346e-05,
      "loss": 1.5832,
      "step": 19627
    },
    {
      "epoch": 0.7029204791662936,
      "grad_norm": 1.79032564163208,
      "learning_rate": 4.2823285587165454e-05,
      "loss": 1.2493,
      "step": 19628
    },
    {
      "epoch": 0.7029562912958619,
      "grad_norm": 2.0720388889312744,
      "learning_rate": 4.281376995657638e-05,
      "loss": 1.8434,
      "step": 19629
    },
    {
      "epoch": 0.7029921034254302,
      "grad_norm": 1.5829167366027832,
      "learning_rate": 4.2804255095354276e-05,
      "loss": 1.6059,
      "step": 19630
    },
    {
      "epoch": 0.7030279155549984,
      "grad_norm": 1.614844799041748,
      "learning_rate": 4.279474100362717e-05,
      "loss": 1.6384,
      "step": 19631
    },
    {
      "epoch": 0.7030637276845667,
      "grad_norm": 1.5533795356750488,
      "learning_rate": 4.278522768152301e-05,
      "loss": 1.5463,
      "step": 19632
    },
    {
      "epoch": 0.7030995398141351,
      "grad_norm": 1.7275530099868774,
      "learning_rate": 4.27757151291698e-05,
      "loss": 1.2183,
      "step": 19633
    },
    {
      "epoch": 0.7031353519437034,
      "grad_norm": 1.5814635753631592,
      "learning_rate": 4.2766203346695565e-05,
      "loss": 1.5156,
      "step": 19634
    },
    {
      "epoch": 0.7031711640732716,
      "grad_norm": 2.4060182571411133,
      "learning_rate": 4.27566923342282e-05,
      "loss": 1.5844,
      "step": 19635
    },
    {
      "epoch": 0.7032069762028399,
      "grad_norm": 2.1266350746154785,
      "learning_rate": 4.274718209189571e-05,
      "loss": 1.4377,
      "step": 19636
    },
    {
      "epoch": 0.7032427883324082,
      "grad_norm": 1.3293077945709229,
      "learning_rate": 4.273767261982603e-05,
      "loss": 1.4234,
      "step": 19637
    },
    {
      "epoch": 0.7032786004619764,
      "grad_norm": 1.726370930671692,
      "learning_rate": 4.272816391814714e-05,
      "loss": 1.6574,
      "step": 19638
    },
    {
      "epoch": 0.7033144125915447,
      "grad_norm": 1.4099525213241577,
      "learning_rate": 4.271865598698689e-05,
      "loss": 1.0711,
      "step": 19639
    },
    {
      "epoch": 0.7033502247211131,
      "grad_norm": 1.8294678926467896,
      "learning_rate": 4.2709148826473234e-05,
      "loss": 1.6613,
      "step": 19640
    },
    {
      "epoch": 0.7033860368506814,
      "grad_norm": 1.2897287607192993,
      "learning_rate": 4.269964243673408e-05,
      "loss": 1.2772,
      "step": 19641
    },
    {
      "epoch": 0.7034218489802496,
      "grad_norm": 1.5318269729614258,
      "learning_rate": 4.2690136817897363e-05,
      "loss": 1.2968,
      "step": 19642
    },
    {
      "epoch": 0.7034576611098179,
      "grad_norm": 1.5041943788528442,
      "learning_rate": 4.2680631970090935e-05,
      "loss": 1.3391,
      "step": 19643
    },
    {
      "epoch": 0.7034934732393862,
      "grad_norm": 1.3494664430618286,
      "learning_rate": 4.2671127893442586e-05,
      "loss": 1.594,
      "step": 19644
    },
    {
      "epoch": 0.7035292853689544,
      "grad_norm": 1.539660930633545,
      "learning_rate": 4.266162458808034e-05,
      "loss": 1.4438,
      "step": 19645
    },
    {
      "epoch": 0.7035650974985227,
      "grad_norm": 1.8535983562469482,
      "learning_rate": 4.2652122054131936e-05,
      "loss": 1.7111,
      "step": 19646
    },
    {
      "epoch": 0.7036009096280911,
      "grad_norm": 1.5027899742126465,
      "learning_rate": 4.264262029172527e-05,
      "loss": 1.62,
      "step": 19647
    },
    {
      "epoch": 0.7036367217576593,
      "grad_norm": 2.0121660232543945,
      "learning_rate": 4.2633119300988146e-05,
      "loss": 1.4439,
      "step": 19648
    },
    {
      "epoch": 0.7036725338872276,
      "grad_norm": 1.5586117506027222,
      "learning_rate": 4.262361908204844e-05,
      "loss": 1.2805,
      "step": 19649
    },
    {
      "epoch": 0.7037083460167959,
      "grad_norm": 1.7597720623016357,
      "learning_rate": 4.26141196350339e-05,
      "loss": 1.2432,
      "step": 19650
    },
    {
      "epoch": 0.7037441581463642,
      "grad_norm": 1.6648684740066528,
      "learning_rate": 4.260462096007235e-05,
      "loss": 1.3918,
      "step": 19651
    },
    {
      "epoch": 0.7037799702759324,
      "grad_norm": 1.5515859127044678,
      "learning_rate": 4.259512305729164e-05,
      "loss": 1.4651,
      "step": 19652
    },
    {
      "epoch": 0.7038157824055007,
      "grad_norm": 1.6097532510757446,
      "learning_rate": 4.258562592681945e-05,
      "loss": 1.4623,
      "step": 19653
    },
    {
      "epoch": 0.7038515945350691,
      "grad_norm": 1.509201169013977,
      "learning_rate": 4.2576129568783654e-05,
      "loss": 1.6167,
      "step": 19654
    },
    {
      "epoch": 0.7038874066646373,
      "grad_norm": 2.127002716064453,
      "learning_rate": 4.2566633983311885e-05,
      "loss": 1.636,
      "step": 19655
    },
    {
      "epoch": 0.7039232187942056,
      "grad_norm": 2.234020948410034,
      "learning_rate": 4.2557139170532045e-05,
      "loss": 1.4049,
      "step": 19656
    },
    {
      "epoch": 0.7039590309237739,
      "grad_norm": 1.5483022928237915,
      "learning_rate": 4.2547645130571764e-05,
      "loss": 1.6323,
      "step": 19657
    },
    {
      "epoch": 0.7039948430533421,
      "grad_norm": 1.733654499053955,
      "learning_rate": 4.253815186355881e-05,
      "loss": 1.5718,
      "step": 19658
    },
    {
      "epoch": 0.7040306551829104,
      "grad_norm": 1.627803921699524,
      "learning_rate": 4.2528659369620905e-05,
      "loss": 1.5466,
      "step": 19659
    },
    {
      "epoch": 0.7040664673124787,
      "grad_norm": 1.8518602848052979,
      "learning_rate": 4.2519167648885785e-05,
      "loss": 1.5608,
      "step": 19660
    },
    {
      "epoch": 0.7041022794420471,
      "grad_norm": 1.3604775667190552,
      "learning_rate": 4.250967670148113e-05,
      "loss": 1.2684,
      "step": 19661
    },
    {
      "epoch": 0.7041380915716153,
      "grad_norm": 1.494215488433838,
      "learning_rate": 4.250018652753454e-05,
      "loss": 1.5892,
      "step": 19662
    },
    {
      "epoch": 0.7041739037011836,
      "grad_norm": 1.7942612171173096,
      "learning_rate": 4.2490697127173826e-05,
      "loss": 1.3718,
      "step": 19663
    },
    {
      "epoch": 0.7042097158307519,
      "grad_norm": 1.7529025077819824,
      "learning_rate": 4.248120850052658e-05,
      "loss": 1.3392,
      "step": 19664
    },
    {
      "epoch": 0.7042455279603201,
      "grad_norm": 1.6615201234817505,
      "learning_rate": 4.247172064772053e-05,
      "loss": 1.4284,
      "step": 19665
    },
    {
      "epoch": 0.7042813400898884,
      "grad_norm": 2.163020133972168,
      "learning_rate": 4.246223356888318e-05,
      "loss": 1.4905,
      "step": 19666
    },
    {
      "epoch": 0.7043171522194567,
      "grad_norm": 1.709355115890503,
      "learning_rate": 4.2452747264142335e-05,
      "loss": 1.2316,
      "step": 19667
    },
    {
      "epoch": 0.7043529643490251,
      "grad_norm": 1.8853095769882202,
      "learning_rate": 4.244326173362555e-05,
      "loss": 1.541,
      "step": 19668
    },
    {
      "epoch": 0.7043887764785933,
      "grad_norm": 2.275879383087158,
      "learning_rate": 4.2433776977460396e-05,
      "loss": 1.6956,
      "step": 19669
    },
    {
      "epoch": 0.7044245886081616,
      "grad_norm": 1.2705893516540527,
      "learning_rate": 4.242429299577452e-05,
      "loss": 1.2138,
      "step": 19670
    },
    {
      "epoch": 0.7044604007377299,
      "grad_norm": 2.3329508304595947,
      "learning_rate": 4.241480978869551e-05,
      "loss": 1.461,
      "step": 19671
    },
    {
      "epoch": 0.7044962128672981,
      "grad_norm": 1.6873449087142944,
      "learning_rate": 4.2405327356351e-05,
      "loss": 1.468,
      "step": 19672
    },
    {
      "epoch": 0.7045320249968664,
      "grad_norm": 1.3374345302581787,
      "learning_rate": 4.239584569886843e-05,
      "loss": 1.4267,
      "step": 19673
    },
    {
      "epoch": 0.7045678371264347,
      "grad_norm": 1.6494673490524292,
      "learning_rate": 4.2386364816375545e-05,
      "loss": 1.5675,
      "step": 19674
    },
    {
      "epoch": 0.704603649256003,
      "grad_norm": 2.307361364364624,
      "learning_rate": 4.2376884708999754e-05,
      "loss": 1.659,
      "step": 19675
    },
    {
      "epoch": 0.7046394613855713,
      "grad_norm": 1.6656765937805176,
      "learning_rate": 4.23674053768687e-05,
      "loss": 1.4741,
      "step": 19676
    },
    {
      "epoch": 0.7046752735151396,
      "grad_norm": 2.520934820175171,
      "learning_rate": 4.2357926820109816e-05,
      "loss": 1.3717,
      "step": 19677
    },
    {
      "epoch": 0.7047110856447079,
      "grad_norm": 1.7585898637771606,
      "learning_rate": 4.234844903885068e-05,
      "loss": 1.3923,
      "step": 19678
    },
    {
      "epoch": 0.7047468977742761,
      "grad_norm": 1.6115695238113403,
      "learning_rate": 4.233897203321883e-05,
      "loss": 1.4965,
      "step": 19679
    },
    {
      "epoch": 0.7047827099038444,
      "grad_norm": 1.3951077461242676,
      "learning_rate": 4.23294958033417e-05,
      "loss": 1.6331,
      "step": 19680
    },
    {
      "epoch": 0.7048185220334127,
      "grad_norm": 1.9594569206237793,
      "learning_rate": 4.232002034934681e-05,
      "loss": 1.4238,
      "step": 19681
    },
    {
      "epoch": 0.704854334162981,
      "grad_norm": 1.7811216115951538,
      "learning_rate": 4.231054567136166e-05,
      "loss": 1.4687,
      "step": 19682
    },
    {
      "epoch": 0.7048901462925493,
      "grad_norm": 1.6424189805984497,
      "learning_rate": 4.230107176951372e-05,
      "loss": 1.3477,
      "step": 19683
    },
    {
      "epoch": 0.7049259584221176,
      "grad_norm": 1.5223708152770996,
      "learning_rate": 4.229159864393037e-05,
      "loss": 1.6541,
      "step": 19684
    },
    {
      "epoch": 0.7049617705516859,
      "grad_norm": 1.9250037670135498,
      "learning_rate": 4.2282126294739186e-05,
      "loss": 1.7329,
      "step": 19685
    },
    {
      "epoch": 0.7049975826812541,
      "grad_norm": 1.6363168954849243,
      "learning_rate": 4.227265472206756e-05,
      "loss": 1.5585,
      "step": 19686
    },
    {
      "epoch": 0.7050333948108224,
      "grad_norm": 1.635489821434021,
      "learning_rate": 4.226318392604285e-05,
      "loss": 1.2106,
      "step": 19687
    },
    {
      "epoch": 0.7050692069403907,
      "grad_norm": 1.4792739152908325,
      "learning_rate": 4.225371390679254e-05,
      "loss": 1.4307,
      "step": 19688
    },
    {
      "epoch": 0.705105019069959,
      "grad_norm": 1.7022123336791992,
      "learning_rate": 4.224424466444401e-05,
      "loss": 1.6674,
      "step": 19689
    },
    {
      "epoch": 0.7051408311995273,
      "grad_norm": 1.5994065999984741,
      "learning_rate": 4.2234776199124705e-05,
      "loss": 1.3563,
      "step": 19690
    },
    {
      "epoch": 0.7051766433290956,
      "grad_norm": 1.511945128440857,
      "learning_rate": 4.222530851096194e-05,
      "loss": 1.5933,
      "step": 19691
    },
    {
      "epoch": 0.7052124554586638,
      "grad_norm": 1.8896557092666626,
      "learning_rate": 4.221584160008313e-05,
      "loss": 1.4839,
      "step": 19692
    },
    {
      "epoch": 0.7052482675882321,
      "grad_norm": 1.4836983680725098,
      "learning_rate": 4.220637546661562e-05,
      "loss": 1.4684,
      "step": 19693
    },
    {
      "epoch": 0.7052840797178004,
      "grad_norm": 1.3161821365356445,
      "learning_rate": 4.2196910110686826e-05,
      "loss": 1.5167,
      "step": 19694
    },
    {
      "epoch": 0.7053198918473687,
      "grad_norm": 1.367013692855835,
      "learning_rate": 4.218744553242402e-05,
      "loss": 1.6204,
      "step": 19695
    },
    {
      "epoch": 0.705355703976937,
      "grad_norm": 1.7846018075942993,
      "learning_rate": 4.217798173195454e-05,
      "loss": 1.3237,
      "step": 19696
    },
    {
      "epoch": 0.7053915161065053,
      "grad_norm": 1.9348431825637817,
      "learning_rate": 4.216851870940578e-05,
      "loss": 1.5916,
      "step": 19697
    },
    {
      "epoch": 0.7054273282360736,
      "grad_norm": 1.376637578010559,
      "learning_rate": 4.2159056464904964e-05,
      "loss": 1.316,
      "step": 19698
    },
    {
      "epoch": 0.7054631403656418,
      "grad_norm": 2.1042535305023193,
      "learning_rate": 4.2149594998579445e-05,
      "loss": 1.1512,
      "step": 19699
    },
    {
      "epoch": 0.7054989524952101,
      "grad_norm": 1.6151759624481201,
      "learning_rate": 4.214013431055649e-05,
      "loss": 1.2476,
      "step": 19700
    },
    {
      "epoch": 0.7055347646247784,
      "grad_norm": 1.6883217096328735,
      "learning_rate": 4.213067440096343e-05,
      "loss": 1.5371,
      "step": 19701
    },
    {
      "epoch": 0.7055705767543466,
      "grad_norm": 1.4629133939743042,
      "learning_rate": 4.212121526992747e-05,
      "loss": 1.6678,
      "step": 19702
    },
    {
      "epoch": 0.705606388883915,
      "grad_norm": 1.3969576358795166,
      "learning_rate": 4.211175691757591e-05,
      "loss": 1.4144,
      "step": 19703
    },
    {
      "epoch": 0.7056422010134833,
      "grad_norm": 1.6211917400360107,
      "learning_rate": 4.2102299344036014e-05,
      "loss": 1.4869,
      "step": 19704
    },
    {
      "epoch": 0.7056780131430516,
      "grad_norm": 1.6717114448547363,
      "learning_rate": 4.2092842549434954e-05,
      "loss": 1.259,
      "step": 19705
    },
    {
      "epoch": 0.7057138252726198,
      "grad_norm": 1.6694384813308716,
      "learning_rate": 4.208338653390002e-05,
      "loss": 1.5622,
      "step": 19706
    },
    {
      "epoch": 0.7057496374021881,
      "grad_norm": 1.3916771411895752,
      "learning_rate": 4.20739312975584e-05,
      "loss": 1.5155,
      "step": 19707
    },
    {
      "epoch": 0.7057854495317564,
      "grad_norm": 1.4139550924301147,
      "learning_rate": 4.206447684053735e-05,
      "loss": 1.3517,
      "step": 19708
    },
    {
      "epoch": 0.7058212616613246,
      "grad_norm": 2.4061508178710938,
      "learning_rate": 4.2055023162964e-05,
      "loss": 1.742,
      "step": 19709
    },
    {
      "epoch": 0.705857073790893,
      "grad_norm": 1.6226320266723633,
      "learning_rate": 4.2045570264965574e-05,
      "loss": 1.6635,
      "step": 19710
    },
    {
      "epoch": 0.7058928859204613,
      "grad_norm": 1.434448003768921,
      "learning_rate": 4.203611814666925e-05,
      "loss": 1.2733,
      "step": 19711
    },
    {
      "epoch": 0.7059286980500296,
      "grad_norm": 1.5434811115264893,
      "learning_rate": 4.202666680820221e-05,
      "loss": 1.2843,
      "step": 19712
    },
    {
      "epoch": 0.7059645101795978,
      "grad_norm": 1.7157692909240723,
      "learning_rate": 4.201721624969156e-05,
      "loss": 1.4567,
      "step": 19713
    },
    {
      "epoch": 0.7060003223091661,
      "grad_norm": 1.388255000114441,
      "learning_rate": 4.200776647126447e-05,
      "loss": 1.4829,
      "step": 19714
    },
    {
      "epoch": 0.7060361344387344,
      "grad_norm": 1.2208225727081299,
      "learning_rate": 4.199831747304811e-05,
      "loss": 1.4678,
      "step": 19715
    },
    {
      "epoch": 0.7060719465683026,
      "grad_norm": 1.6040128469467163,
      "learning_rate": 4.198886925516954e-05,
      "loss": 1.5309,
      "step": 19716
    },
    {
      "epoch": 0.706107758697871,
      "grad_norm": 1.2374918460845947,
      "learning_rate": 4.19794218177559e-05,
      "loss": 1.4965,
      "step": 19717
    },
    {
      "epoch": 0.7061435708274393,
      "grad_norm": 1.8679009675979614,
      "learning_rate": 4.196997516093431e-05,
      "loss": 1.6104,
      "step": 19718
    },
    {
      "epoch": 0.7061793829570076,
      "grad_norm": 1.7728638648986816,
      "learning_rate": 4.196052928483188e-05,
      "loss": 1.5767,
      "step": 19719
    },
    {
      "epoch": 0.7062151950865758,
      "grad_norm": 1.6892356872558594,
      "learning_rate": 4.195108418957563e-05,
      "loss": 1.5712,
      "step": 19720
    },
    {
      "epoch": 0.7062510072161441,
      "grad_norm": 1.1942442655563354,
      "learning_rate": 4.194163987529266e-05,
      "loss": 1.105,
      "step": 19721
    },
    {
      "epoch": 0.7062868193457124,
      "grad_norm": 2.3192408084869385,
      "learning_rate": 4.1932196342110076e-05,
      "loss": 1.4804,
      "step": 19722
    },
    {
      "epoch": 0.7063226314752806,
      "grad_norm": 1.7069861888885498,
      "learning_rate": 4.1922753590154854e-05,
      "loss": 1.5507,
      "step": 19723
    },
    {
      "epoch": 0.706358443604849,
      "grad_norm": 2.246624708175659,
      "learning_rate": 4.1913311619554064e-05,
      "loss": 1.5065,
      "step": 19724
    },
    {
      "epoch": 0.7063942557344173,
      "grad_norm": 1.9762353897094727,
      "learning_rate": 4.1903870430434736e-05,
      "loss": 1.7526,
      "step": 19725
    },
    {
      "epoch": 0.7064300678639855,
      "grad_norm": 1.4072891473770142,
      "learning_rate": 4.189443002292392e-05,
      "loss": 1.5152,
      "step": 19726
    },
    {
      "epoch": 0.7064658799935538,
      "grad_norm": 2.209500312805176,
      "learning_rate": 4.1884990397148584e-05,
      "loss": 1.2394,
      "step": 19727
    },
    {
      "epoch": 0.7065016921231221,
      "grad_norm": 2.366694450378418,
      "learning_rate": 4.187555155323572e-05,
      "loss": 1.7438,
      "step": 19728
    },
    {
      "epoch": 0.7065375042526904,
      "grad_norm": 1.3919899463653564,
      "learning_rate": 4.186611349131235e-05,
      "loss": 1.5614,
      "step": 19729
    },
    {
      "epoch": 0.7065733163822586,
      "grad_norm": 1.4385210275650024,
      "learning_rate": 4.1856676211505465e-05,
      "loss": 1.6848,
      "step": 19730
    },
    {
      "epoch": 0.706609128511827,
      "grad_norm": 2.9990813732147217,
      "learning_rate": 4.184723971394197e-05,
      "loss": 1.3986,
      "step": 19731
    },
    {
      "epoch": 0.7066449406413953,
      "grad_norm": 1.7250571250915527,
      "learning_rate": 4.183780399874885e-05,
      "loss": 1.7292,
      "step": 19732
    },
    {
      "epoch": 0.7066807527709635,
      "grad_norm": 1.7180428504943848,
      "learning_rate": 4.182836906605309e-05,
      "loss": 1.2938,
      "step": 19733
    },
    {
      "epoch": 0.7067165649005318,
      "grad_norm": 1.549298882484436,
      "learning_rate": 4.1818934915981544e-05,
      "loss": 1.4342,
      "step": 19734
    },
    {
      "epoch": 0.7067523770301001,
      "grad_norm": 1.4171665906906128,
      "learning_rate": 4.180950154866119e-05,
      "loss": 1.6761,
      "step": 19735
    },
    {
      "epoch": 0.7067881891596683,
      "grad_norm": 1.4383457899093628,
      "learning_rate": 4.180006896421893e-05,
      "loss": 1.3711,
      "step": 19736
    },
    {
      "epoch": 0.7068240012892366,
      "grad_norm": 1.7060292959213257,
      "learning_rate": 4.179063716278171e-05,
      "loss": 1.7098,
      "step": 19737
    },
    {
      "epoch": 0.706859813418805,
      "grad_norm": 1.4669756889343262,
      "learning_rate": 4.178120614447634e-05,
      "loss": 1.5862,
      "step": 19738
    },
    {
      "epoch": 0.7068956255483733,
      "grad_norm": 1.5930043458938599,
      "learning_rate": 4.177177590942974e-05,
      "loss": 1.7459,
      "step": 19739
    },
    {
      "epoch": 0.7069314376779415,
      "grad_norm": 1.4487543106079102,
      "learning_rate": 4.176234645776883e-05,
      "loss": 1.6866,
      "step": 19740
    },
    {
      "epoch": 0.7069672498075098,
      "grad_norm": 1.9951848983764648,
      "learning_rate": 4.1752917789620395e-05,
      "loss": 1.4115,
      "step": 19741
    },
    {
      "epoch": 0.7070030619370781,
      "grad_norm": 1.5408639907836914,
      "learning_rate": 4.174348990511131e-05,
      "loss": 1.3569,
      "step": 19742
    },
    {
      "epoch": 0.7070388740666463,
      "grad_norm": 1.9540464878082275,
      "learning_rate": 4.1734062804368426e-05,
      "loss": 1.1295,
      "step": 19743
    },
    {
      "epoch": 0.7070746861962146,
      "grad_norm": 1.8576759099960327,
      "learning_rate": 4.17246364875186e-05,
      "loss": 1.4529,
      "step": 19744
    },
    {
      "epoch": 0.707110498325783,
      "grad_norm": 1.4856642484664917,
      "learning_rate": 4.171521095468859e-05,
      "loss": 1.6362,
      "step": 19745
    },
    {
      "epoch": 0.7071463104553513,
      "grad_norm": 1.6320254802703857,
      "learning_rate": 4.1705786206005235e-05,
      "loss": 1.6359,
      "step": 19746
    },
    {
      "epoch": 0.7071821225849195,
      "grad_norm": 1.697995901107788,
      "learning_rate": 4.169636224159533e-05,
      "loss": 1.277,
      "step": 19747
    },
    {
      "epoch": 0.7072179347144878,
      "grad_norm": 1.8926945924758911,
      "learning_rate": 4.1686939061585694e-05,
      "loss": 1.8489,
      "step": 19748
    },
    {
      "epoch": 0.7072537468440561,
      "grad_norm": 1.6354869604110718,
      "learning_rate": 4.167751666610309e-05,
      "loss": 1.7138,
      "step": 19749
    },
    {
      "epoch": 0.7072895589736243,
      "grad_norm": 1.598158597946167,
      "learning_rate": 4.166809505527418e-05,
      "loss": 1.3413,
      "step": 19750
    },
    {
      "epoch": 0.7073253711031926,
      "grad_norm": 1.6339963674545288,
      "learning_rate": 4.165867422922589e-05,
      "loss": 1.4861,
      "step": 19751
    },
    {
      "epoch": 0.707361183232761,
      "grad_norm": 1.544044852256775,
      "learning_rate": 4.1649254188084854e-05,
      "loss": 1.2516,
      "step": 19752
    },
    {
      "epoch": 0.7073969953623293,
      "grad_norm": 1.874853491783142,
      "learning_rate": 4.1639834931977864e-05,
      "loss": 1.657,
      "step": 19753
    },
    {
      "epoch": 0.7074328074918975,
      "grad_norm": 1.7855067253112793,
      "learning_rate": 4.163041646103154e-05,
      "loss": 1.4241,
      "step": 19754
    },
    {
      "epoch": 0.7074686196214658,
      "grad_norm": 2.4784038066864014,
      "learning_rate": 4.162099877537274e-05,
      "loss": 1.1453,
      "step": 19755
    },
    {
      "epoch": 0.7075044317510341,
      "grad_norm": 1.4555649757385254,
      "learning_rate": 4.161158187512808e-05,
      "loss": 1.6522,
      "step": 19756
    },
    {
      "epoch": 0.7075402438806023,
      "grad_norm": 2.270052909851074,
      "learning_rate": 4.160216576042426e-05,
      "loss": 1.7847,
      "step": 19757
    },
    {
      "epoch": 0.7075760560101706,
      "grad_norm": 1.4863038063049316,
      "learning_rate": 4.159275043138801e-05,
      "loss": 1.3766,
      "step": 19758
    },
    {
      "epoch": 0.707611868139739,
      "grad_norm": 1.55003023147583,
      "learning_rate": 4.1583335888145915e-05,
      "loss": 1.733,
      "step": 19759
    },
    {
      "epoch": 0.7076476802693072,
      "grad_norm": 1.7473454475402832,
      "learning_rate": 4.1573922130824725e-05,
      "loss": 1.4507,
      "step": 19760
    },
    {
      "epoch": 0.7076834923988755,
      "grad_norm": 1.6226022243499756,
      "learning_rate": 4.156450915955099e-05,
      "loss": 1.676,
      "step": 19761
    },
    {
      "epoch": 0.7077193045284438,
      "grad_norm": 2.1244826316833496,
      "learning_rate": 4.155509697445147e-05,
      "loss": 1.7523,
      "step": 19762
    },
    {
      "epoch": 0.707755116658012,
      "grad_norm": 1.449433445930481,
      "learning_rate": 4.1545685575652695e-05,
      "loss": 1.6795,
      "step": 19763
    },
    {
      "epoch": 0.7077909287875803,
      "grad_norm": 1.377442717552185,
      "learning_rate": 4.1536274963281355e-05,
      "loss": 1.5041,
      "step": 19764
    },
    {
      "epoch": 0.7078267409171486,
      "grad_norm": 1.4417132139205933,
      "learning_rate": 4.152686513746399e-05,
      "loss": 1.5012,
      "step": 19765
    },
    {
      "epoch": 0.707862553046717,
      "grad_norm": 1.397241473197937,
      "learning_rate": 4.151745609832722e-05,
      "loss": 1.2836,
      "step": 19766
    },
    {
      "epoch": 0.7078983651762852,
      "grad_norm": 1.9713245630264282,
      "learning_rate": 4.150804784599769e-05,
      "loss": 1.3324,
      "step": 19767
    },
    {
      "epoch": 0.7079341773058535,
      "grad_norm": 1.558300256729126,
      "learning_rate": 4.149864038060185e-05,
      "loss": 1.6347,
      "step": 19768
    },
    {
      "epoch": 0.7079699894354218,
      "grad_norm": 1.6256293058395386,
      "learning_rate": 4.148923370226642e-05,
      "loss": 1.4837,
      "step": 19769
    },
    {
      "epoch": 0.70800580156499,
      "grad_norm": 1.6766290664672852,
      "learning_rate": 4.147982781111783e-05,
      "loss": 1.3036,
      "step": 19770
    },
    {
      "epoch": 0.7080416136945583,
      "grad_norm": 1.964269757270813,
      "learning_rate": 4.147042270728272e-05,
      "loss": 1.138,
      "step": 19771
    },
    {
      "epoch": 0.7080774258241266,
      "grad_norm": 1.2855141162872314,
      "learning_rate": 4.146101839088749e-05,
      "loss": 1.3127,
      "step": 19772
    },
    {
      "epoch": 0.708113237953695,
      "grad_norm": 1.4238767623901367,
      "learning_rate": 4.145161486205883e-05,
      "loss": 1.6241,
      "step": 19773
    },
    {
      "epoch": 0.7081490500832632,
      "grad_norm": 1.4786126613616943,
      "learning_rate": 4.144221212092316e-05,
      "loss": 1.4925,
      "step": 19774
    },
    {
      "epoch": 0.7081848622128315,
      "grad_norm": 1.7244206666946411,
      "learning_rate": 4.1432810167606964e-05,
      "loss": 1.5353,
      "step": 19775
    },
    {
      "epoch": 0.7082206743423998,
      "grad_norm": 1.447631597518921,
      "learning_rate": 4.1423409002236755e-05,
      "loss": 1.3844,
      "step": 19776
    },
    {
      "epoch": 0.708256486471968,
      "grad_norm": 1.4471874237060547,
      "learning_rate": 4.141400862493903e-05,
      "loss": 1.3439,
      "step": 19777
    },
    {
      "epoch": 0.7082922986015363,
      "grad_norm": 1.7642822265625,
      "learning_rate": 4.140460903584027e-05,
      "loss": 1.7151,
      "step": 19778
    },
    {
      "epoch": 0.7083281107311046,
      "grad_norm": 1.674778699874878,
      "learning_rate": 4.139521023506688e-05,
      "loss": 1.6022,
      "step": 19779
    },
    {
      "epoch": 0.708363922860673,
      "grad_norm": 1.7055083513259888,
      "learning_rate": 4.1385812222745344e-05,
      "loss": 1.5398,
      "step": 19780
    },
    {
      "epoch": 0.7083997349902412,
      "grad_norm": 2.1311376094818115,
      "learning_rate": 4.13764149990021e-05,
      "loss": 1.2179,
      "step": 19781
    },
    {
      "epoch": 0.7084355471198095,
      "grad_norm": 2.307271957397461,
      "learning_rate": 4.136701856396361e-05,
      "loss": 1.3747,
      "step": 19782
    },
    {
      "epoch": 0.7084713592493778,
      "grad_norm": 1.5620884895324707,
      "learning_rate": 4.135762291775622e-05,
      "loss": 1.3208,
      "step": 19783
    },
    {
      "epoch": 0.708507171378946,
      "grad_norm": 1.4920231103897095,
      "learning_rate": 4.1348228060506364e-05,
      "loss": 1.3738,
      "step": 19784
    },
    {
      "epoch": 0.7085429835085143,
      "grad_norm": 1.612851858139038,
      "learning_rate": 4.133883399234049e-05,
      "loss": 1.3598,
      "step": 19785
    },
    {
      "epoch": 0.7085787956380826,
      "grad_norm": 1.8075392246246338,
      "learning_rate": 4.132944071338489e-05,
      "loss": 1.9048,
      "step": 19786
    },
    {
      "epoch": 0.708614607767651,
      "grad_norm": 1.3424999713897705,
      "learning_rate": 4.132004822376598e-05,
      "loss": 1.2624,
      "step": 19787
    },
    {
      "epoch": 0.7086504198972192,
      "grad_norm": 1.75864577293396,
      "learning_rate": 4.1310656523610144e-05,
      "loss": 1.2026,
      "step": 19788
    },
    {
      "epoch": 0.7086862320267875,
      "grad_norm": 1.7845993041992188,
      "learning_rate": 4.130126561304376e-05,
      "loss": 1.5058,
      "step": 19789
    },
    {
      "epoch": 0.7087220441563558,
      "grad_norm": 2.524739980697632,
      "learning_rate": 4.129187549219308e-05,
      "loss": 1.5992,
      "step": 19790
    },
    {
      "epoch": 0.708757856285924,
      "grad_norm": 1.766992211341858,
      "learning_rate": 4.12824861611845e-05,
      "loss": 1.4969,
      "step": 19791
    },
    {
      "epoch": 0.7087936684154923,
      "grad_norm": 1.8509896993637085,
      "learning_rate": 4.127309762014435e-05,
      "loss": 1.6552,
      "step": 19792
    },
    {
      "epoch": 0.7088294805450606,
      "grad_norm": 1.721527338027954,
      "learning_rate": 4.12637098691989e-05,
      "loss": 1.4734,
      "step": 19793
    },
    {
      "epoch": 0.7088652926746289,
      "grad_norm": 2.155327558517456,
      "learning_rate": 4.125432290847446e-05,
      "loss": 1.1883,
      "step": 19794
    },
    {
      "epoch": 0.7089011048041972,
      "grad_norm": 1.6428849697113037,
      "learning_rate": 4.124493673809733e-05,
      "loss": 1.1671,
      "step": 19795
    },
    {
      "epoch": 0.7089369169337655,
      "grad_norm": 3.388340711593628,
      "learning_rate": 4.123555135819382e-05,
      "loss": 1.6063,
      "step": 19796
    },
    {
      "epoch": 0.7089727290633338,
      "grad_norm": 1.632297396659851,
      "learning_rate": 4.122616676889014e-05,
      "loss": 1.5945,
      "step": 19797
    },
    {
      "epoch": 0.709008541192902,
      "grad_norm": 1.713770866394043,
      "learning_rate": 4.121678297031256e-05,
      "loss": 1.6018,
      "step": 19798
    },
    {
      "epoch": 0.7090443533224703,
      "grad_norm": 2.2215230464935303,
      "learning_rate": 4.1207399962587356e-05,
      "loss": 1.5367,
      "step": 19799
    },
    {
      "epoch": 0.7090801654520386,
      "grad_norm": 1.535365343093872,
      "learning_rate": 4.119801774584077e-05,
      "loss": 1.2133,
      "step": 19800
    },
    {
      "epoch": 0.7091159775816069,
      "grad_norm": 1.8413805961608887,
      "learning_rate": 4.118863632019898e-05,
      "loss": 1.5743,
      "step": 19801
    },
    {
      "epoch": 0.7091517897111752,
      "grad_norm": 2.152158260345459,
      "learning_rate": 4.117925568578822e-05,
      "loss": 1.3583,
      "step": 19802
    },
    {
      "epoch": 0.7091876018407435,
      "grad_norm": 1.7258599996566772,
      "learning_rate": 4.116987584273474e-05,
      "loss": 1.077,
      "step": 19803
    },
    {
      "epoch": 0.7092234139703117,
      "grad_norm": 1.4642314910888672,
      "learning_rate": 4.116049679116466e-05,
      "loss": 1.1684,
      "step": 19804
    },
    {
      "epoch": 0.70925922609988,
      "grad_norm": 1.8263181447982788,
      "learning_rate": 4.11511185312042e-05,
      "loss": 1.695,
      "step": 19805
    },
    {
      "epoch": 0.7092950382294483,
      "grad_norm": 1.8259236812591553,
      "learning_rate": 4.114174106297952e-05,
      "loss": 1.166,
      "step": 19806
    },
    {
      "epoch": 0.7093308503590166,
      "grad_norm": 1.4869673252105713,
      "learning_rate": 4.113236438661684e-05,
      "loss": 1.2245,
      "step": 19807
    },
    {
      "epoch": 0.7093666624885849,
      "grad_norm": 1.906758427619934,
      "learning_rate": 4.112298850224223e-05,
      "loss": 1.4127,
      "step": 19808
    },
    {
      "epoch": 0.7094024746181532,
      "grad_norm": 1.2898838520050049,
      "learning_rate": 4.111361340998186e-05,
      "loss": 1.4856,
      "step": 19809
    },
    {
      "epoch": 0.7094382867477215,
      "grad_norm": 1.67406165599823,
      "learning_rate": 4.11042391099619e-05,
      "loss": 1.3279,
      "step": 19810
    },
    {
      "epoch": 0.7094740988772897,
      "grad_norm": 1.590075969696045,
      "learning_rate": 4.109486560230839e-05,
      "loss": 1.5435,
      "step": 19811
    },
    {
      "epoch": 0.709509911006858,
      "grad_norm": 1.462267518043518,
      "learning_rate": 4.108549288714748e-05,
      "loss": 1.507,
      "step": 19812
    },
    {
      "epoch": 0.7095457231364263,
      "grad_norm": 1.8588240146636963,
      "learning_rate": 4.107612096460528e-05,
      "loss": 1.4161,
      "step": 19813
    },
    {
      "epoch": 0.7095815352659945,
      "grad_norm": 1.5106000900268555,
      "learning_rate": 4.1066749834807895e-05,
      "loss": 1.3721,
      "step": 19814
    },
    {
      "epoch": 0.7096173473955629,
      "grad_norm": 1.4201984405517578,
      "learning_rate": 4.105737949788133e-05,
      "loss": 1.5243,
      "step": 19815
    },
    {
      "epoch": 0.7096531595251312,
      "grad_norm": 1.4662309885025024,
      "learning_rate": 4.10480099539517e-05,
      "loss": 1.6587,
      "step": 19816
    },
    {
      "epoch": 0.7096889716546995,
      "grad_norm": 1.658463954925537,
      "learning_rate": 4.103864120314506e-05,
      "loss": 1.4674,
      "step": 19817
    },
    {
      "epoch": 0.7097247837842677,
      "grad_norm": 1.8113993406295776,
      "learning_rate": 4.1029273245587476e-05,
      "loss": 1.505,
      "step": 19818
    },
    {
      "epoch": 0.709760595913836,
      "grad_norm": 2.001314878463745,
      "learning_rate": 4.101990608140492e-05,
      "loss": 1.4655,
      "step": 19819
    },
    {
      "epoch": 0.7097964080434043,
      "grad_norm": 1.4082540273666382,
      "learning_rate": 4.101053971072345e-05,
      "loss": 1.414,
      "step": 19820
    },
    {
      "epoch": 0.7098322201729725,
      "grad_norm": 1.8123232126235962,
      "learning_rate": 4.1001174133669116e-05,
      "loss": 1.5416,
      "step": 19821
    },
    {
      "epoch": 0.7098680323025409,
      "grad_norm": 1.7902493476867676,
      "learning_rate": 4.099180935036784e-05,
      "loss": 1.4017,
      "step": 19822
    },
    {
      "epoch": 0.7099038444321092,
      "grad_norm": 1.911616563796997,
      "learning_rate": 4.0982445360945654e-05,
      "loss": 1.6244,
      "step": 19823
    },
    {
      "epoch": 0.7099396565616775,
      "grad_norm": 1.7272944450378418,
      "learning_rate": 4.097308216552854e-05,
      "loss": 1.5453,
      "step": 19824
    },
    {
      "epoch": 0.7099754686912457,
      "grad_norm": 1.4212089776992798,
      "learning_rate": 4.0963719764242504e-05,
      "loss": 1.5528,
      "step": 19825
    },
    {
      "epoch": 0.710011280820814,
      "grad_norm": 2.0135669708251953,
      "learning_rate": 4.0954358157213436e-05,
      "loss": 1.2744,
      "step": 19826
    },
    {
      "epoch": 0.7100470929503823,
      "grad_norm": 1.7139232158660889,
      "learning_rate": 4.0944997344567304e-05,
      "loss": 1.3807,
      "step": 19827
    },
    {
      "epoch": 0.7100829050799505,
      "grad_norm": 1.644747018814087,
      "learning_rate": 4.0935637326430095e-05,
      "loss": 1.3795,
      "step": 19828
    },
    {
      "epoch": 0.7101187172095189,
      "grad_norm": 1.47983717918396,
      "learning_rate": 4.092627810292767e-05,
      "loss": 1.5415,
      "step": 19829
    },
    {
      "epoch": 0.7101545293390872,
      "grad_norm": 1.429914951324463,
      "learning_rate": 4.0916919674185974e-05,
      "loss": 1.4195,
      "step": 19830
    },
    {
      "epoch": 0.7101903414686555,
      "grad_norm": 1.6520344018936157,
      "learning_rate": 4.09075620403309e-05,
      "loss": 1.4766,
      "step": 19831
    },
    {
      "epoch": 0.7102261535982237,
      "grad_norm": 1.82496976852417,
      "learning_rate": 4.0898205201488404e-05,
      "loss": 1.6098,
      "step": 19832
    },
    {
      "epoch": 0.710261965727792,
      "grad_norm": 1.775235891342163,
      "learning_rate": 4.088884915778427e-05,
      "loss": 1.431,
      "step": 19833
    },
    {
      "epoch": 0.7102977778573603,
      "grad_norm": 1.3506832122802734,
      "learning_rate": 4.087949390934443e-05,
      "loss": 1.537,
      "step": 19834
    },
    {
      "epoch": 0.7103335899869285,
      "grad_norm": 1.5820268392562866,
      "learning_rate": 4.0870139456294745e-05,
      "loss": 1.2898,
      "step": 19835
    },
    {
      "epoch": 0.7103694021164969,
      "grad_norm": 1.426241397857666,
      "learning_rate": 4.0860785798761094e-05,
      "loss": 1.3136,
      "step": 19836
    },
    {
      "epoch": 0.7104052142460652,
      "grad_norm": 1.4646059274673462,
      "learning_rate": 4.0851432936869296e-05,
      "loss": 1.1997,
      "step": 19837
    },
    {
      "epoch": 0.7104410263756334,
      "grad_norm": 1.9413141012191772,
      "learning_rate": 4.0842080870745084e-05,
      "loss": 1.5866,
      "step": 19838
    },
    {
      "epoch": 0.7104768385052017,
      "grad_norm": 1.6418945789337158,
      "learning_rate": 4.083272960051444e-05,
      "loss": 1.4602,
      "step": 19839
    },
    {
      "epoch": 0.71051265063477,
      "grad_norm": 1.7183191776275635,
      "learning_rate": 4.0823379126303064e-05,
      "loss": 1.2933,
      "step": 19840
    },
    {
      "epoch": 0.7105484627643383,
      "grad_norm": 1.4567809104919434,
      "learning_rate": 4.0814029448236803e-05,
      "loss": 1.3024,
      "step": 19841
    },
    {
      "epoch": 0.7105842748939065,
      "grad_norm": 3.3251304626464844,
      "learning_rate": 4.080468056644141e-05,
      "loss": 1.2703,
      "step": 19842
    },
    {
      "epoch": 0.7106200870234749,
      "grad_norm": 2.0006585121154785,
      "learning_rate": 4.0795332481042736e-05,
      "loss": 1.1981,
      "step": 19843
    },
    {
      "epoch": 0.7106558991530432,
      "grad_norm": 1.7054616212844849,
      "learning_rate": 4.078598519216645e-05,
      "loss": 1.5251,
      "step": 19844
    },
    {
      "epoch": 0.7106917112826114,
      "grad_norm": 2.462589979171753,
      "learning_rate": 4.077663869993835e-05,
      "loss": 1.3953,
      "step": 19845
    },
    {
      "epoch": 0.7107275234121797,
      "grad_norm": 1.619739055633545,
      "learning_rate": 4.076729300448423e-05,
      "loss": 1.3919,
      "step": 19846
    },
    {
      "epoch": 0.710763335541748,
      "grad_norm": 1.6942226886749268,
      "learning_rate": 4.075794810592973e-05,
      "loss": 1.2181,
      "step": 19847
    },
    {
      "epoch": 0.7107991476713162,
      "grad_norm": 1.8538376092910767,
      "learning_rate": 4.074860400440067e-05,
      "loss": 1.4492,
      "step": 19848
    },
    {
      "epoch": 0.7108349598008845,
      "grad_norm": 1.5637342929840088,
      "learning_rate": 4.073926070002264e-05,
      "loss": 1.4442,
      "step": 19849
    },
    {
      "epoch": 0.7108707719304529,
      "grad_norm": 1.4168877601623535,
      "learning_rate": 4.072991819292148e-05,
      "loss": 1.9249,
      "step": 19850
    },
    {
      "epoch": 0.7109065840600212,
      "grad_norm": 1.5848156213760376,
      "learning_rate": 4.0720576483222795e-05,
      "loss": 1.3855,
      "step": 19851
    },
    {
      "epoch": 0.7109423961895894,
      "grad_norm": 1.6019319295883179,
      "learning_rate": 4.0711235571052306e-05,
      "loss": 1.6355,
      "step": 19852
    },
    {
      "epoch": 0.7109782083191577,
      "grad_norm": 1.3284941911697388,
      "learning_rate": 4.070189545653561e-05,
      "loss": 1.214,
      "step": 19853
    },
    {
      "epoch": 0.711014020448726,
      "grad_norm": 1.7176045179367065,
      "learning_rate": 4.069255613979849e-05,
      "loss": 1.523,
      "step": 19854
    },
    {
      "epoch": 0.7110498325782942,
      "grad_norm": 1.7608132362365723,
      "learning_rate": 4.068321762096652e-05,
      "loss": 1.1746,
      "step": 19855
    },
    {
      "epoch": 0.7110856447078625,
      "grad_norm": 1.74787437915802,
      "learning_rate": 4.067387990016528e-05,
      "loss": 1.2927,
      "step": 19856
    },
    {
      "epoch": 0.7111214568374309,
      "grad_norm": 1.6266214847564697,
      "learning_rate": 4.0664542977520526e-05,
      "loss": 1.3775,
      "step": 19857
    },
    {
      "epoch": 0.7111572689669992,
      "grad_norm": 1.6558783054351807,
      "learning_rate": 4.065520685315777e-05,
      "loss": 1.3691,
      "step": 19858
    },
    {
      "epoch": 0.7111930810965674,
      "grad_norm": 1.853967308998108,
      "learning_rate": 4.0645871527202695e-05,
      "loss": 1.3375,
      "step": 19859
    },
    {
      "epoch": 0.7112288932261357,
      "grad_norm": 1.459820032119751,
      "learning_rate": 4.063653699978079e-05,
      "loss": 1.3781,
      "step": 19860
    },
    {
      "epoch": 0.711264705355704,
      "grad_norm": 1.3726294040679932,
      "learning_rate": 4.062720327101778e-05,
      "loss": 1.5798,
      "step": 19861
    },
    {
      "epoch": 0.7113005174852722,
      "grad_norm": 1.7460395097732544,
      "learning_rate": 4.0617870341039155e-05,
      "loss": 1.499,
      "step": 19862
    },
    {
      "epoch": 0.7113363296148405,
      "grad_norm": 1.8195322751998901,
      "learning_rate": 4.060853820997046e-05,
      "loss": 1.3855,
      "step": 19863
    },
    {
      "epoch": 0.7113721417444089,
      "grad_norm": 1.6731899976730347,
      "learning_rate": 4.059920687793727e-05,
      "loss": 1.6307,
      "step": 19864
    },
    {
      "epoch": 0.7114079538739772,
      "grad_norm": 1.573760986328125,
      "learning_rate": 4.058987634506514e-05,
      "loss": 1.4133,
      "step": 19865
    },
    {
      "epoch": 0.7114437660035454,
      "grad_norm": 1.6020008325576782,
      "learning_rate": 4.058054661147961e-05,
      "loss": 1.6497,
      "step": 19866
    },
    {
      "epoch": 0.7114795781331137,
      "grad_norm": 1.3755627870559692,
      "learning_rate": 4.057121767730612e-05,
      "loss": 1.3899,
      "step": 19867
    },
    {
      "epoch": 0.711515390262682,
      "grad_norm": 1.3662739992141724,
      "learning_rate": 4.05618895426703e-05,
      "loss": 1.5769,
      "step": 19868
    },
    {
      "epoch": 0.7115512023922502,
      "grad_norm": 1.8042664527893066,
      "learning_rate": 4.055256220769755e-05,
      "loss": 1.3367,
      "step": 19869
    },
    {
      "epoch": 0.7115870145218185,
      "grad_norm": 1.653998851776123,
      "learning_rate": 4.0543235672513434e-05,
      "loss": 1.4023,
      "step": 19870
    },
    {
      "epoch": 0.7116228266513869,
      "grad_norm": 2.058856964111328,
      "learning_rate": 4.0533909937243365e-05,
      "loss": 1.3852,
      "step": 19871
    },
    {
      "epoch": 0.7116586387809551,
      "grad_norm": 1.6064378023147583,
      "learning_rate": 4.0524585002012815e-05,
      "loss": 1.3659,
      "step": 19872
    },
    {
      "epoch": 0.7116944509105234,
      "grad_norm": 1.5606815814971924,
      "learning_rate": 4.05152608669473e-05,
      "loss": 1.3211,
      "step": 19873
    },
    {
      "epoch": 0.7117302630400917,
      "grad_norm": 1.5185139179229736,
      "learning_rate": 4.0505937532172175e-05,
      "loss": 1.4429,
      "step": 19874
    },
    {
      "epoch": 0.71176607516966,
      "grad_norm": 1.587683081626892,
      "learning_rate": 4.049661499781293e-05,
      "loss": 1.6571,
      "step": 19875
    },
    {
      "epoch": 0.7118018872992282,
      "grad_norm": 1.8505325317382812,
      "learning_rate": 4.048729326399498e-05,
      "loss": 1.396,
      "step": 19876
    },
    {
      "epoch": 0.7118376994287965,
      "grad_norm": 1.4003081321716309,
      "learning_rate": 4.047797233084375e-05,
      "loss": 1.402,
      "step": 19877
    },
    {
      "epoch": 0.7118735115583649,
      "grad_norm": 1.828133225440979,
      "learning_rate": 4.0468652198484603e-05,
      "loss": 1.035,
      "step": 19878
    },
    {
      "epoch": 0.7119093236879331,
      "grad_norm": 1.4773956537246704,
      "learning_rate": 4.045933286704296e-05,
      "loss": 1.5263,
      "step": 19879
    },
    {
      "epoch": 0.7119451358175014,
      "grad_norm": 1.9004855155944824,
      "learning_rate": 4.0450014336644204e-05,
      "loss": 1.337,
      "step": 19880
    },
    {
      "epoch": 0.7119809479470697,
      "grad_norm": 2.385497808456421,
      "learning_rate": 4.0440696607413665e-05,
      "loss": 1.6802,
      "step": 19881
    },
    {
      "epoch": 0.712016760076638,
      "grad_norm": 1.733431339263916,
      "learning_rate": 4.0431379679476735e-05,
      "loss": 1.6468,
      "step": 19882
    },
    {
      "epoch": 0.7120525722062062,
      "grad_norm": 1.4043059349060059,
      "learning_rate": 4.042206355295875e-05,
      "loss": 1.4577,
      "step": 19883
    },
    {
      "epoch": 0.7120883843357745,
      "grad_norm": 1.3792320489883423,
      "learning_rate": 4.0412748227985075e-05,
      "loss": 1.4173,
      "step": 19884
    },
    {
      "epoch": 0.7121241964653429,
      "grad_norm": 1.3899900913238525,
      "learning_rate": 4.040343370468098e-05,
      "loss": 1.3588,
      "step": 19885
    },
    {
      "epoch": 0.7121600085949111,
      "grad_norm": 1.430406928062439,
      "learning_rate": 4.039411998317182e-05,
      "loss": 1.6367,
      "step": 19886
    },
    {
      "epoch": 0.7121958207244794,
      "grad_norm": 1.4435672760009766,
      "learning_rate": 4.038480706358287e-05,
      "loss": 1.3293,
      "step": 19887
    },
    {
      "epoch": 0.7122316328540477,
      "grad_norm": 1.3121012449264526,
      "learning_rate": 4.0375494946039495e-05,
      "loss": 1.0906,
      "step": 19888
    },
    {
      "epoch": 0.7122674449836159,
      "grad_norm": 1.6354104280471802,
      "learning_rate": 4.0366183630666885e-05,
      "loss": 1.4199,
      "step": 19889
    },
    {
      "epoch": 0.7123032571131842,
      "grad_norm": 1.617457389831543,
      "learning_rate": 4.035687311759036e-05,
      "loss": 1.2094,
      "step": 19890
    },
    {
      "epoch": 0.7123390692427525,
      "grad_norm": 2.983320713043213,
      "learning_rate": 4.03475634069352e-05,
      "loss": 1.2032,
      "step": 19891
    },
    {
      "epoch": 0.7123748813723209,
      "grad_norm": 1.9741209745407104,
      "learning_rate": 4.033825449882659e-05,
      "loss": 1.6208,
      "step": 19892
    },
    {
      "epoch": 0.7124106935018891,
      "grad_norm": 1.3680564165115356,
      "learning_rate": 4.032894639338981e-05,
      "loss": 1.7192,
      "step": 19893
    },
    {
      "epoch": 0.7124465056314574,
      "grad_norm": 1.6451799869537354,
      "learning_rate": 4.031963909075009e-05,
      "loss": 1.2906,
      "step": 19894
    },
    {
      "epoch": 0.7124823177610257,
      "grad_norm": 1.493516206741333,
      "learning_rate": 4.0310332591032675e-05,
      "loss": 1.4103,
      "step": 19895
    },
    {
      "epoch": 0.7125181298905939,
      "grad_norm": 1.9640635251998901,
      "learning_rate": 4.030102689436271e-05,
      "loss": 1.3533,
      "step": 19896
    },
    {
      "epoch": 0.7125539420201622,
      "grad_norm": 1.6041368246078491,
      "learning_rate": 4.0291722000865416e-05,
      "loss": 1.2857,
      "step": 19897
    },
    {
      "epoch": 0.7125897541497305,
      "grad_norm": 1.584389090538025,
      "learning_rate": 4.0282417910666025e-05,
      "loss": 1.5981,
      "step": 19898
    },
    {
      "epoch": 0.7126255662792988,
      "grad_norm": 1.9366203546524048,
      "learning_rate": 4.027311462388964e-05,
      "loss": 1.6245,
      "step": 19899
    },
    {
      "epoch": 0.7126613784088671,
      "grad_norm": 1.880853533744812,
      "learning_rate": 4.026381214066145e-05,
      "loss": 1.358,
      "step": 19900
    },
    {
      "epoch": 0.7126971905384354,
      "grad_norm": 1.7722489833831787,
      "learning_rate": 4.025451046110661e-05,
      "loss": 1.5929,
      "step": 19901
    },
    {
      "epoch": 0.7127330026680037,
      "grad_norm": 1.6037089824676514,
      "learning_rate": 4.024520958535031e-05,
      "loss": 1.5791,
      "step": 19902
    },
    {
      "epoch": 0.7127688147975719,
      "grad_norm": 2.314192295074463,
      "learning_rate": 4.023590951351759e-05,
      "loss": 1.5203,
      "step": 19903
    },
    {
      "epoch": 0.7128046269271402,
      "grad_norm": 2.308858633041382,
      "learning_rate": 4.022661024573362e-05,
      "loss": 1.2871,
      "step": 19904
    },
    {
      "epoch": 0.7128404390567085,
      "grad_norm": 1.604375958442688,
      "learning_rate": 4.0217311782123514e-05,
      "loss": 1.7423,
      "step": 19905
    },
    {
      "epoch": 0.7128762511862768,
      "grad_norm": 1.2021867036819458,
      "learning_rate": 4.020801412281239e-05,
      "loss": 1.246,
      "step": 19906
    },
    {
      "epoch": 0.7129120633158451,
      "grad_norm": 2.0452868938446045,
      "learning_rate": 4.019871726792528e-05,
      "loss": 1.5018,
      "step": 19907
    },
    {
      "epoch": 0.7129478754454134,
      "grad_norm": 1.4524165391921997,
      "learning_rate": 4.0189421217587297e-05,
      "loss": 1.4838,
      "step": 19908
    },
    {
      "epoch": 0.7129836875749817,
      "grad_norm": 1.3504445552825928,
      "learning_rate": 4.0180125971923524e-05,
      "loss": 1.5034,
      "step": 19909
    },
    {
      "epoch": 0.7130194997045499,
      "grad_norm": 2.032801389694214,
      "learning_rate": 4.017083153105897e-05,
      "loss": 1.5505,
      "step": 19910
    },
    {
      "epoch": 0.7130553118341182,
      "grad_norm": 1.6039384603500366,
      "learning_rate": 4.0161537895118695e-05,
      "loss": 1.3706,
      "step": 19911
    },
    {
      "epoch": 0.7130911239636865,
      "grad_norm": 1.8873577117919922,
      "learning_rate": 4.0152245064227745e-05,
      "loss": 1.2697,
      "step": 19912
    },
    {
      "epoch": 0.7131269360932548,
      "grad_norm": 1.6298884153366089,
      "learning_rate": 4.0142953038511176e-05,
      "loss": 1.422,
      "step": 19913
    },
    {
      "epoch": 0.7131627482228231,
      "grad_norm": 1.6877508163452148,
      "learning_rate": 4.013366181809393e-05,
      "loss": 1.2018,
      "step": 19914
    },
    {
      "epoch": 0.7131985603523914,
      "grad_norm": 1.8622572422027588,
      "learning_rate": 4.0124371403101034e-05,
      "loss": 1.6352,
      "step": 19915
    },
    {
      "epoch": 0.7132343724819596,
      "grad_norm": 2.3264732360839844,
      "learning_rate": 4.0115081793657525e-05,
      "loss": 1.5532,
      "step": 19916
    },
    {
      "epoch": 0.7132701846115279,
      "grad_norm": 2.014953374862671,
      "learning_rate": 4.010579298988832e-05,
      "loss": 1.7364,
      "step": 19917
    },
    {
      "epoch": 0.7133059967410962,
      "grad_norm": 2.2904231548309326,
      "learning_rate": 4.00965049919184e-05,
      "loss": 1.2316,
      "step": 19918
    },
    {
      "epoch": 0.7133418088706645,
      "grad_norm": 1.9575514793395996,
      "learning_rate": 4.0087217799872746e-05,
      "loss": 1.4247,
      "step": 19919
    },
    {
      "epoch": 0.7133776210002328,
      "grad_norm": 1.7283276319503784,
      "learning_rate": 4.007793141387633e-05,
      "loss": 1.4632,
      "step": 19920
    },
    {
      "epoch": 0.7134134331298011,
      "grad_norm": 1.2572942972183228,
      "learning_rate": 4.0068645834054e-05,
      "loss": 1.251,
      "step": 19921
    },
    {
      "epoch": 0.7134492452593694,
      "grad_norm": 1.8809939622879028,
      "learning_rate": 4.0059361060530755e-05,
      "loss": 1.8018,
      "step": 19922
    },
    {
      "epoch": 0.7134850573889376,
      "grad_norm": 2.7003350257873535,
      "learning_rate": 4.005007709343147e-05,
      "loss": 1.3071,
      "step": 19923
    },
    {
      "epoch": 0.7135208695185059,
      "grad_norm": 1.7158159017562866,
      "learning_rate": 4.004079393288112e-05,
      "loss": 1.5144,
      "step": 19924
    },
    {
      "epoch": 0.7135566816480742,
      "grad_norm": 1.4738690853118896,
      "learning_rate": 4.00315115790045e-05,
      "loss": 1.6998,
      "step": 19925
    },
    {
      "epoch": 0.7135924937776424,
      "grad_norm": 1.3729649782180786,
      "learning_rate": 4.002223003192654e-05,
      "loss": 1.2989,
      "step": 19926
    },
    {
      "epoch": 0.7136283059072108,
      "grad_norm": 1.5292845964431763,
      "learning_rate": 4.001294929177215e-05,
      "loss": 1.683,
      "step": 19927
    },
    {
      "epoch": 0.7136641180367791,
      "grad_norm": 1.4550150632858276,
      "learning_rate": 4.0003669358666106e-05,
      "loss": 1.4341,
      "step": 19928
    },
    {
      "epoch": 0.7136999301663474,
      "grad_norm": 1.1493412256240845,
      "learning_rate": 3.9994390232733304e-05,
      "loss": 1.3642,
      "step": 19929
    },
    {
      "epoch": 0.7137357422959156,
      "grad_norm": 1.5714877843856812,
      "learning_rate": 3.9985111914098585e-05,
      "loss": 1.3609,
      "step": 19930
    },
    {
      "epoch": 0.7137715544254839,
      "grad_norm": 1.6583268642425537,
      "learning_rate": 3.99758344028868e-05,
      "loss": 1.5051,
      "step": 19931
    },
    {
      "epoch": 0.7138073665550522,
      "grad_norm": 2.281008005142212,
      "learning_rate": 3.99665576992227e-05,
      "loss": 1.8181,
      "step": 19932
    },
    {
      "epoch": 0.7138431786846204,
      "grad_norm": 1.5775352716445923,
      "learning_rate": 3.995728180323114e-05,
      "loss": 1.5004,
      "step": 19933
    },
    {
      "epoch": 0.7138789908141888,
      "grad_norm": 1.927957534790039,
      "learning_rate": 3.994800671503694e-05,
      "loss": 1.4526,
      "step": 19934
    },
    {
      "epoch": 0.7139148029437571,
      "grad_norm": 1.4535878896713257,
      "learning_rate": 3.9938732434764805e-05,
      "loss": 1.5883,
      "step": 19935
    },
    {
      "epoch": 0.7139506150733254,
      "grad_norm": 1.65828537940979,
      "learning_rate": 3.992945896253958e-05,
      "loss": 1.6344,
      "step": 19936
    },
    {
      "epoch": 0.7139864272028936,
      "grad_norm": 1.9846526384353638,
      "learning_rate": 3.992018629848594e-05,
      "loss": 1.3487,
      "step": 19937
    },
    {
      "epoch": 0.7140222393324619,
      "grad_norm": 1.7143404483795166,
      "learning_rate": 3.991091444272876e-05,
      "loss": 1.2854,
      "step": 19938
    },
    {
      "epoch": 0.7140580514620302,
      "grad_norm": 1.7557034492492676,
      "learning_rate": 3.9901643395392685e-05,
      "loss": 1.5111,
      "step": 19939
    },
    {
      "epoch": 0.7140938635915984,
      "grad_norm": 1.5067017078399658,
      "learning_rate": 3.989237315660248e-05,
      "loss": 1.4103,
      "step": 19940
    },
    {
      "epoch": 0.7141296757211668,
      "grad_norm": 2.0715649127960205,
      "learning_rate": 3.988310372648285e-05,
      "loss": 1.4079,
      "step": 19941
    },
    {
      "epoch": 0.7141654878507351,
      "grad_norm": 1.3460463285446167,
      "learning_rate": 3.9873835105158564e-05,
      "loss": 1.5776,
      "step": 19942
    },
    {
      "epoch": 0.7142012999803034,
      "grad_norm": 1.6910752058029175,
      "learning_rate": 3.9864567292754266e-05,
      "loss": 1.6837,
      "step": 19943
    },
    {
      "epoch": 0.7142371121098716,
      "grad_norm": 2.667029619216919,
      "learning_rate": 3.985530028939456e-05,
      "loss": 1.3131,
      "step": 19944
    },
    {
      "epoch": 0.7142729242394399,
      "grad_norm": 1.554051160812378,
      "learning_rate": 3.9846034095204285e-05,
      "loss": 1.3344,
      "step": 19945
    },
    {
      "epoch": 0.7143087363690082,
      "grad_norm": 1.4796596765518188,
      "learning_rate": 3.9836768710308e-05,
      "loss": 1.6145,
      "step": 19946
    },
    {
      "epoch": 0.7143445484985764,
      "grad_norm": 1.7634326219558716,
      "learning_rate": 3.982750413483043e-05,
      "loss": 1.2404,
      "step": 19947
    },
    {
      "epoch": 0.7143803606281448,
      "grad_norm": 1.5787911415100098,
      "learning_rate": 3.981824036889609e-05,
      "loss": 1.3975,
      "step": 19948
    },
    {
      "epoch": 0.7144161727577131,
      "grad_norm": 1.527350902557373,
      "learning_rate": 3.9808977412629764e-05,
      "loss": 1.4577,
      "step": 19949
    },
    {
      "epoch": 0.7144519848872813,
      "grad_norm": 1.693781852722168,
      "learning_rate": 3.979971526615598e-05,
      "loss": 1.1326,
      "step": 19950
    },
    {
      "epoch": 0.7144877970168496,
      "grad_norm": 1.5585476160049438,
      "learning_rate": 3.9790453929599384e-05,
      "loss": 1.2612,
      "step": 19951
    },
    {
      "epoch": 0.7145236091464179,
      "grad_norm": 1.7941901683807373,
      "learning_rate": 3.978119340308458e-05,
      "loss": 1.5067,
      "step": 19952
    },
    {
      "epoch": 0.7145594212759862,
      "grad_norm": 1.4558805227279663,
      "learning_rate": 3.977193368673612e-05,
      "loss": 1.2185,
      "step": 19953
    },
    {
      "epoch": 0.7145952334055544,
      "grad_norm": 1.8235039710998535,
      "learning_rate": 3.976267478067863e-05,
      "loss": 1.4494,
      "step": 19954
    },
    {
      "epoch": 0.7146310455351228,
      "grad_norm": 2.5660204887390137,
      "learning_rate": 3.975341668503659e-05,
      "loss": 1.5131,
      "step": 19955
    },
    {
      "epoch": 0.7146668576646911,
      "grad_norm": 2.2466602325439453,
      "learning_rate": 3.9744159399934676e-05,
      "loss": 1.6216,
      "step": 19956
    },
    {
      "epoch": 0.7147026697942593,
      "grad_norm": 1.3325560092926025,
      "learning_rate": 3.973490292549735e-05,
      "loss": 1.478,
      "step": 19957
    },
    {
      "epoch": 0.7147384819238276,
      "grad_norm": 1.7814619541168213,
      "learning_rate": 3.97256472618492e-05,
      "loss": 1.5762,
      "step": 19958
    },
    {
      "epoch": 0.7147742940533959,
      "grad_norm": 1.7509052753448486,
      "learning_rate": 3.971639240911468e-05,
      "loss": 1.2004,
      "step": 19959
    },
    {
      "epoch": 0.7148101061829641,
      "grad_norm": 1.4850753545761108,
      "learning_rate": 3.970713836741834e-05,
      "loss": 1.1819,
      "step": 19960
    },
    {
      "epoch": 0.7148459183125324,
      "grad_norm": 1.3009765148162842,
      "learning_rate": 3.9697885136884716e-05,
      "loss": 1.5121,
      "step": 19961
    },
    {
      "epoch": 0.7148817304421008,
      "grad_norm": 1.357944369316101,
      "learning_rate": 3.968863271763822e-05,
      "loss": 1.4413,
      "step": 19962
    },
    {
      "epoch": 0.7149175425716691,
      "grad_norm": 1.5240039825439453,
      "learning_rate": 3.967938110980338e-05,
      "loss": 1.3346,
      "step": 19963
    },
    {
      "epoch": 0.7149533547012373,
      "grad_norm": 1.7359546422958374,
      "learning_rate": 3.9670130313504675e-05,
      "loss": 1.5019,
      "step": 19964
    },
    {
      "epoch": 0.7149891668308056,
      "grad_norm": 1.9136874675750732,
      "learning_rate": 3.9660880328866556e-05,
      "loss": 1.3177,
      "step": 19965
    },
    {
      "epoch": 0.7150249789603739,
      "grad_norm": 1.4809314012527466,
      "learning_rate": 3.96516311560134e-05,
      "loss": 1.5468,
      "step": 19966
    },
    {
      "epoch": 0.7150607910899421,
      "grad_norm": 1.871216058731079,
      "learning_rate": 3.964238279506979e-05,
      "loss": 1.6134,
      "step": 19967
    },
    {
      "epoch": 0.7150966032195104,
      "grad_norm": 1.5043405294418335,
      "learning_rate": 3.963313524616005e-05,
      "loss": 1.3557,
      "step": 19968
    },
    {
      "epoch": 0.7151324153490788,
      "grad_norm": 1.8438221216201782,
      "learning_rate": 3.962388850940857e-05,
      "loss": 1.5986,
      "step": 19969
    },
    {
      "epoch": 0.7151682274786471,
      "grad_norm": 1.7853418588638306,
      "learning_rate": 3.9614642584939784e-05,
      "loss": 1.4494,
      "step": 19970
    },
    {
      "epoch": 0.7152040396082153,
      "grad_norm": 1.7872045040130615,
      "learning_rate": 3.96053974728781e-05,
      "loss": 1.6464,
      "step": 19971
    },
    {
      "epoch": 0.7152398517377836,
      "grad_norm": 1.82921302318573,
      "learning_rate": 3.9596153173347925e-05,
      "loss": 1.6829,
      "step": 19972
    },
    {
      "epoch": 0.7152756638673519,
      "grad_norm": 1.4460041522979736,
      "learning_rate": 3.958690968647356e-05,
      "loss": 1.5682,
      "step": 19973
    },
    {
      "epoch": 0.7153114759969201,
      "grad_norm": 2.098288059234619,
      "learning_rate": 3.9577667012379395e-05,
      "loss": 1.6599,
      "step": 19974
    },
    {
      "epoch": 0.7153472881264884,
      "grad_norm": 1.5381921529769897,
      "learning_rate": 3.956842515118978e-05,
      "loss": 1.543,
      "step": 19975
    },
    {
      "epoch": 0.7153831002560568,
      "grad_norm": 1.5164440870285034,
      "learning_rate": 3.955918410302909e-05,
      "loss": 1.3406,
      "step": 19976
    },
    {
      "epoch": 0.715418912385625,
      "grad_norm": 1.4968868494033813,
      "learning_rate": 3.954994386802158e-05,
      "loss": 1.4984,
      "step": 19977
    },
    {
      "epoch": 0.7154547245151933,
      "grad_norm": 1.6591699123382568,
      "learning_rate": 3.95407044462916e-05,
      "loss": 1.326,
      "step": 19978
    },
    {
      "epoch": 0.7154905366447616,
      "grad_norm": 1.5657908916473389,
      "learning_rate": 3.953146583796349e-05,
      "loss": 1.2784,
      "step": 19979
    },
    {
      "epoch": 0.7155263487743299,
      "grad_norm": 1.5644398927688599,
      "learning_rate": 3.952222804316148e-05,
      "loss": 1.7664,
      "step": 19980
    },
    {
      "epoch": 0.7155621609038981,
      "grad_norm": 1.6657533645629883,
      "learning_rate": 3.9512991062009874e-05,
      "loss": 1.3903,
      "step": 19981
    },
    {
      "epoch": 0.7155979730334664,
      "grad_norm": 1.300279140472412,
      "learning_rate": 3.9503754894632947e-05,
      "loss": 1.5107,
      "step": 19982
    },
    {
      "epoch": 0.7156337851630348,
      "grad_norm": 1.9345062971115112,
      "learning_rate": 3.949451954115501e-05,
      "loss": 1.638,
      "step": 19983
    },
    {
      "epoch": 0.715669597292603,
      "grad_norm": 1.9588655233383179,
      "learning_rate": 3.948528500170021e-05,
      "loss": 1.377,
      "step": 19984
    },
    {
      "epoch": 0.7157054094221713,
      "grad_norm": 1.5046672821044922,
      "learning_rate": 3.9476051276392853e-05,
      "loss": 1.4962,
      "step": 19985
    },
    {
      "epoch": 0.7157412215517396,
      "grad_norm": 1.284699559211731,
      "learning_rate": 3.946681836535721e-05,
      "loss": 1.0213,
      "step": 19986
    },
    {
      "epoch": 0.7157770336813079,
      "grad_norm": 1.8973195552825928,
      "learning_rate": 3.945758626871738e-05,
      "loss": 1.608,
      "step": 19987
    },
    {
      "epoch": 0.7158128458108761,
      "grad_norm": 1.9553606510162354,
      "learning_rate": 3.9448354986597645e-05,
      "loss": 1.344,
      "step": 19988
    },
    {
      "epoch": 0.7158486579404444,
      "grad_norm": 1.8713107109069824,
      "learning_rate": 3.943912451912219e-05,
      "loss": 1.4795,
      "step": 19989
    },
    {
      "epoch": 0.7158844700700128,
      "grad_norm": 2.1087417602539062,
      "learning_rate": 3.9429894866415226e-05,
      "loss": 1.3713,
      "step": 19990
    },
    {
      "epoch": 0.715920282199581,
      "grad_norm": 1.916988492012024,
      "learning_rate": 3.9420666028600874e-05,
      "loss": 1.7671,
      "step": 19991
    },
    {
      "epoch": 0.7159560943291493,
      "grad_norm": 1.5088728666305542,
      "learning_rate": 3.9411438005803305e-05,
      "loss": 1.2984,
      "step": 19992
    },
    {
      "epoch": 0.7159919064587176,
      "grad_norm": 1.6458569765090942,
      "learning_rate": 3.9402210798146686e-05,
      "loss": 1.4498,
      "step": 19993
    },
    {
      "epoch": 0.7160277185882858,
      "grad_norm": 1.7323598861694336,
      "learning_rate": 3.939298440575519e-05,
      "loss": 1.6207,
      "step": 19994
    },
    {
      "epoch": 0.7160635307178541,
      "grad_norm": 1.744994044303894,
      "learning_rate": 3.9383758828752884e-05,
      "loss": 1.537,
      "step": 19995
    },
    {
      "epoch": 0.7160993428474224,
      "grad_norm": 1.408331036567688,
      "learning_rate": 3.93745340672639e-05,
      "loss": 1.5115,
      "step": 19996
    },
    {
      "epoch": 0.7161351549769908,
      "grad_norm": 1.3587737083435059,
      "learning_rate": 3.936531012141241e-05,
      "loss": 1.3417,
      "step": 19997
    },
    {
      "epoch": 0.716170967106559,
      "grad_norm": 1.5102040767669678,
      "learning_rate": 3.935608699132242e-05,
      "loss": 1.1886,
      "step": 19998
    },
    {
      "epoch": 0.7162067792361273,
      "grad_norm": 1.4365791082382202,
      "learning_rate": 3.9346864677118046e-05,
      "loss": 1.4104,
      "step": 19999
    },
    {
      "epoch": 0.7162425913656956,
      "grad_norm": 1.6736547946929932,
      "learning_rate": 3.9337643178923376e-05,
      "loss": 1.2907,
      "step": 20000
    },
    {
      "epoch": 0.7162784034952638,
      "grad_norm": 1.6852706670761108,
      "learning_rate": 3.932842249686251e-05,
      "loss": 1.5204,
      "step": 20001
    },
    {
      "epoch": 0.7163142156248321,
      "grad_norm": 1.6308528184890747,
      "learning_rate": 3.9319202631059414e-05,
      "loss": 1.3969,
      "step": 20002
    },
    {
      "epoch": 0.7163500277544004,
      "grad_norm": 1.8392781019210815,
      "learning_rate": 3.9309983581638173e-05,
      "loss": 1.5858,
      "step": 20003
    },
    {
      "epoch": 0.7163858398839686,
      "grad_norm": 1.9511610269546509,
      "learning_rate": 3.9300765348722854e-05,
      "loss": 1.1621,
      "step": 20004
    },
    {
      "epoch": 0.716421652013537,
      "grad_norm": 1.9499866962432861,
      "learning_rate": 3.929154793243741e-05,
      "loss": 1.5144,
      "step": 20005
    },
    {
      "epoch": 0.7164574641431053,
      "grad_norm": 1.5823554992675781,
      "learning_rate": 3.928233133290589e-05,
      "loss": 1.5373,
      "step": 20006
    },
    {
      "epoch": 0.7164932762726736,
      "grad_norm": 2.0642759799957275,
      "learning_rate": 3.927311555025227e-05,
      "loss": 1.6982,
      "step": 20007
    },
    {
      "epoch": 0.7165290884022418,
      "grad_norm": 1.6306183338165283,
      "learning_rate": 3.926390058460058e-05,
      "loss": 1.4509,
      "step": 20008
    },
    {
      "epoch": 0.7165649005318101,
      "grad_norm": 1.752936601638794,
      "learning_rate": 3.925468643607473e-05,
      "loss": 1.5816,
      "step": 20009
    },
    {
      "epoch": 0.7166007126613784,
      "grad_norm": 1.5219781398773193,
      "learning_rate": 3.9245473104798726e-05,
      "loss": 1.3056,
      "step": 20010
    },
    {
      "epoch": 0.7166365247909466,
      "grad_norm": 1.620758295059204,
      "learning_rate": 3.923626059089651e-05,
      "loss": 1.2589,
      "step": 20011
    },
    {
      "epoch": 0.716672336920515,
      "grad_norm": 1.6076922416687012,
      "learning_rate": 3.9227048894492055e-05,
      "loss": 1.6003,
      "step": 20012
    },
    {
      "epoch": 0.7167081490500833,
      "grad_norm": 1.5931798219680786,
      "learning_rate": 3.921783801570924e-05,
      "loss": 1.341,
      "step": 20013
    },
    {
      "epoch": 0.7167439611796516,
      "grad_norm": 1.723936676979065,
      "learning_rate": 3.9208627954672014e-05,
      "loss": 1.4163,
      "step": 20014
    },
    {
      "epoch": 0.7167797733092198,
      "grad_norm": 1.620898962020874,
      "learning_rate": 3.9199418711504307e-05,
      "loss": 1.3572,
      "step": 20015
    },
    {
      "epoch": 0.7168155854387881,
      "grad_norm": 1.4487621784210205,
      "learning_rate": 3.919021028632998e-05,
      "loss": 1.5906,
      "step": 20016
    },
    {
      "epoch": 0.7168513975683564,
      "grad_norm": 1.5113482475280762,
      "learning_rate": 3.918100267927292e-05,
      "loss": 1.6878,
      "step": 20017
    },
    {
      "epoch": 0.7168872096979246,
      "grad_norm": 1.350650668144226,
      "learning_rate": 3.917179589045701e-05,
      "loss": 1.3855,
      "step": 20018
    },
    {
      "epoch": 0.716923021827493,
      "grad_norm": 2.0095953941345215,
      "learning_rate": 3.9162589920006164e-05,
      "loss": 1.6719,
      "step": 20019
    },
    {
      "epoch": 0.7169588339570613,
      "grad_norm": 2.0317280292510986,
      "learning_rate": 3.9153384768044163e-05,
      "loss": 1.2714,
      "step": 20020
    },
    {
      "epoch": 0.7169946460866296,
      "grad_norm": 1.5975542068481445,
      "learning_rate": 3.9144180434694885e-05,
      "loss": 1.4929,
      "step": 20021
    },
    {
      "epoch": 0.7170304582161978,
      "grad_norm": 1.5322084426879883,
      "learning_rate": 3.91349769200822e-05,
      "loss": 1.0812,
      "step": 20022
    },
    {
      "epoch": 0.7170662703457661,
      "grad_norm": 1.5115982294082642,
      "learning_rate": 3.9125774224329845e-05,
      "loss": 1.4755,
      "step": 20023
    },
    {
      "epoch": 0.7171020824753344,
      "grad_norm": 1.715003252029419,
      "learning_rate": 3.911657234756169e-05,
      "loss": 1.4666,
      "step": 20024
    },
    {
      "epoch": 0.7171378946049026,
      "grad_norm": 1.540813684463501,
      "learning_rate": 3.9107371289901504e-05,
      "loss": 1.4246,
      "step": 20025
    },
    {
      "epoch": 0.717173706734471,
      "grad_norm": 1.781424880027771,
      "learning_rate": 3.909817105147314e-05,
      "loss": 1.3314,
      "step": 20026
    },
    {
      "epoch": 0.7172095188640393,
      "grad_norm": 1.4089975357055664,
      "learning_rate": 3.9088971632400286e-05,
      "loss": 1.2955,
      "step": 20027
    },
    {
      "epoch": 0.7172453309936075,
      "grad_norm": 1.7052937746047974,
      "learning_rate": 3.907977303280674e-05,
      "loss": 1.6117,
      "step": 20028
    },
    {
      "epoch": 0.7172811431231758,
      "grad_norm": 1.4896774291992188,
      "learning_rate": 3.907057525281628e-05,
      "loss": 1.4082,
      "step": 20029
    },
    {
      "epoch": 0.7173169552527441,
      "grad_norm": 1.4978755712509155,
      "learning_rate": 3.906137829255266e-05,
      "loss": 1.3267,
      "step": 20030
    },
    {
      "epoch": 0.7173527673823124,
      "grad_norm": 1.3901692628860474,
      "learning_rate": 3.90521821521396e-05,
      "loss": 1.1466,
      "step": 20031
    },
    {
      "epoch": 0.7173885795118806,
      "grad_norm": 1.3833082914352417,
      "learning_rate": 3.904298683170074e-05,
      "loss": 1.577,
      "step": 20032
    },
    {
      "epoch": 0.717424391641449,
      "grad_norm": 1.5521522760391235,
      "learning_rate": 3.903379233135994e-05,
      "loss": 1.2189,
      "step": 20033
    },
    {
      "epoch": 0.7174602037710173,
      "grad_norm": 1.3972352743148804,
      "learning_rate": 3.9024598651240774e-05,
      "loss": 1.5434,
      "step": 20034
    },
    {
      "epoch": 0.7174960159005855,
      "grad_norm": 1.4136375188827515,
      "learning_rate": 3.901540579146698e-05,
      "loss": 1.4805,
      "step": 20035
    },
    {
      "epoch": 0.7175318280301538,
      "grad_norm": 1.4638866186141968,
      "learning_rate": 3.900621375216226e-05,
      "loss": 1.5089,
      "step": 20036
    },
    {
      "epoch": 0.7175676401597221,
      "grad_norm": 1.317194938659668,
      "learning_rate": 3.8997022533450264e-05,
      "loss": 1.3669,
      "step": 20037
    },
    {
      "epoch": 0.7176034522892903,
      "grad_norm": 2.1901493072509766,
      "learning_rate": 3.898783213545463e-05,
      "loss": 1.1969,
      "step": 20038
    },
    {
      "epoch": 0.7176392644188586,
      "grad_norm": 1.7815966606140137,
      "learning_rate": 3.8978642558298994e-05,
      "loss": 1.6605,
      "step": 20039
    },
    {
      "epoch": 0.717675076548427,
      "grad_norm": 1.8504796028137207,
      "learning_rate": 3.8969453802107057e-05,
      "loss": 1.2785,
      "step": 20040
    },
    {
      "epoch": 0.7177108886779953,
      "grad_norm": 1.529250144958496,
      "learning_rate": 3.8960265867002364e-05,
      "loss": 1.6428,
      "step": 20041
    },
    {
      "epoch": 0.7177467008075635,
      "grad_norm": 1.3686065673828125,
      "learning_rate": 3.895107875310858e-05,
      "loss": 1.1473,
      "step": 20042
    },
    {
      "epoch": 0.7177825129371318,
      "grad_norm": 2.0113461017608643,
      "learning_rate": 3.894189246054922e-05,
      "loss": 1.4563,
      "step": 20043
    },
    {
      "epoch": 0.7178183250667001,
      "grad_norm": 1.429203748703003,
      "learning_rate": 3.893270698944802e-05,
      "loss": 1.3154,
      "step": 20044
    },
    {
      "epoch": 0.7178541371962683,
      "grad_norm": 2.117223024368286,
      "learning_rate": 3.892352233992843e-05,
      "loss": 1.5677,
      "step": 20045
    },
    {
      "epoch": 0.7178899493258366,
      "grad_norm": 1.57451331615448,
      "learning_rate": 3.89143385121141e-05,
      "loss": 1.3936,
      "step": 20046
    },
    {
      "epoch": 0.717925761455405,
      "grad_norm": 1.4294334650039673,
      "learning_rate": 3.8905155506128476e-05,
      "loss": 1.4473,
      "step": 20047
    },
    {
      "epoch": 0.7179615735849733,
      "grad_norm": 1.9604945182800293,
      "learning_rate": 3.889597332209526e-05,
      "loss": 0.9822,
      "step": 20048
    },
    {
      "epoch": 0.7179973857145415,
      "grad_norm": 1.716846227645874,
      "learning_rate": 3.888679196013789e-05,
      "loss": 1.7099,
      "step": 20049
    },
    {
      "epoch": 0.7180331978441098,
      "grad_norm": 1.586357831954956,
      "learning_rate": 3.887761142037984e-05,
      "loss": 1.4334,
      "step": 20050
    },
    {
      "epoch": 0.7180690099736781,
      "grad_norm": 1.982703685760498,
      "learning_rate": 3.886843170294475e-05,
      "loss": 1.7308,
      "step": 20051
    },
    {
      "epoch": 0.7181048221032463,
      "grad_norm": 2.7019457817077637,
      "learning_rate": 3.8859252807956035e-05,
      "loss": 1.4159,
      "step": 20052
    },
    {
      "epoch": 0.7181406342328146,
      "grad_norm": 1.6056925058364868,
      "learning_rate": 3.885007473553723e-05,
      "loss": 1.2037,
      "step": 20053
    },
    {
      "epoch": 0.718176446362383,
      "grad_norm": 1.4310388565063477,
      "learning_rate": 3.8840897485811737e-05,
      "loss": 1.2534,
      "step": 20054
    },
    {
      "epoch": 0.7182122584919512,
      "grad_norm": 1.3919428586959839,
      "learning_rate": 3.883172105890314e-05,
      "loss": 1.6811,
      "step": 20055
    },
    {
      "epoch": 0.7182480706215195,
      "grad_norm": 1.4286328554153442,
      "learning_rate": 3.8822545454934836e-05,
      "loss": 1.3516,
      "step": 20056
    },
    {
      "epoch": 0.7182838827510878,
      "grad_norm": 1.5007710456848145,
      "learning_rate": 3.881337067403022e-05,
      "loss": 1.6128,
      "step": 20057
    },
    {
      "epoch": 0.7183196948806561,
      "grad_norm": 1.7567851543426514,
      "learning_rate": 3.8804196716312805e-05,
      "loss": 1.6251,
      "step": 20058
    },
    {
      "epoch": 0.7183555070102243,
      "grad_norm": 1.5256688594818115,
      "learning_rate": 3.879502358190596e-05,
      "loss": 1.5159,
      "step": 20059
    },
    {
      "epoch": 0.7183913191397926,
      "grad_norm": 1.9770673513412476,
      "learning_rate": 3.878585127093317e-05,
      "loss": 1.393,
      "step": 20060
    },
    {
      "epoch": 0.718427131269361,
      "grad_norm": 1.4122111797332764,
      "learning_rate": 3.877667978351772e-05,
      "loss": 1.4472,
      "step": 20061
    },
    {
      "epoch": 0.7184629433989292,
      "grad_norm": 1.5489094257354736,
      "learning_rate": 3.876750911978315e-05,
      "loss": 1.3976,
      "step": 20062
    },
    {
      "epoch": 0.7184987555284975,
      "grad_norm": 1.4532663822174072,
      "learning_rate": 3.875833927985272e-05,
      "loss": 1.2297,
      "step": 20063
    },
    {
      "epoch": 0.7185345676580658,
      "grad_norm": 1.3802516460418701,
      "learning_rate": 3.8749170263849865e-05,
      "loss": 1.1974,
      "step": 20064
    },
    {
      "epoch": 0.718570379787634,
      "grad_norm": 1.4403727054595947,
      "learning_rate": 3.874000207189789e-05,
      "loss": 1.4378,
      "step": 20065
    },
    {
      "epoch": 0.7186061919172023,
      "grad_norm": 1.9045497179031372,
      "learning_rate": 3.8730834704120164e-05,
      "loss": 1.7718,
      "step": 20066
    },
    {
      "epoch": 0.7186420040467706,
      "grad_norm": 1.5778992176055908,
      "learning_rate": 3.8721668160640054e-05,
      "loss": 1.6047,
      "step": 20067
    },
    {
      "epoch": 0.718677816176339,
      "grad_norm": 1.8331999778747559,
      "learning_rate": 3.871250244158083e-05,
      "loss": 1.6248,
      "step": 20068
    },
    {
      "epoch": 0.7187136283059072,
      "grad_norm": 1.9123164415359497,
      "learning_rate": 3.870333754706583e-05,
      "loss": 1.5269,
      "step": 20069
    },
    {
      "epoch": 0.7187494404354755,
      "grad_norm": 2.063952922821045,
      "learning_rate": 3.8694173477218355e-05,
      "loss": 1.7413,
      "step": 20070
    },
    {
      "epoch": 0.7187852525650438,
      "grad_norm": 1.6923632621765137,
      "learning_rate": 3.8685010232161736e-05,
      "loss": 1.0536,
      "step": 20071
    },
    {
      "epoch": 0.718821064694612,
      "grad_norm": 1.7924041748046875,
      "learning_rate": 3.8675847812019175e-05,
      "loss": 1.3545,
      "step": 20072
    },
    {
      "epoch": 0.7188568768241803,
      "grad_norm": 1.7178550958633423,
      "learning_rate": 3.866668621691397e-05,
      "loss": 1.3935,
      "step": 20073
    },
    {
      "epoch": 0.7188926889537486,
      "grad_norm": 1.6161495447158813,
      "learning_rate": 3.8657525446969436e-05,
      "loss": 1.3834,
      "step": 20074
    },
    {
      "epoch": 0.718928501083317,
      "grad_norm": 2.147392988204956,
      "learning_rate": 3.864836550230874e-05,
      "loss": 1.6886,
      "step": 20075
    },
    {
      "epoch": 0.7189643132128852,
      "grad_norm": 1.8922128677368164,
      "learning_rate": 3.863920638305512e-05,
      "loss": 1.468,
      "step": 20076
    },
    {
      "epoch": 0.7190001253424535,
      "grad_norm": 1.7388283014297485,
      "learning_rate": 3.863004808933186e-05,
      "loss": 1.554,
      "step": 20077
    },
    {
      "epoch": 0.7190359374720218,
      "grad_norm": 1.6552281379699707,
      "learning_rate": 3.8620890621262164e-05,
      "loss": 1.6504,
      "step": 20078
    },
    {
      "epoch": 0.71907174960159,
      "grad_norm": 1.4679269790649414,
      "learning_rate": 3.8611733978969176e-05,
      "loss": 1.451,
      "step": 20079
    },
    {
      "epoch": 0.7191075617311583,
      "grad_norm": 1.8459839820861816,
      "learning_rate": 3.860257816257612e-05,
      "loss": 1.4404,
      "step": 20080
    },
    {
      "epoch": 0.7191433738607266,
      "grad_norm": 1.5493810176849365,
      "learning_rate": 3.859342317220619e-05,
      "loss": 1.3396,
      "step": 20081
    },
    {
      "epoch": 0.719179185990295,
      "grad_norm": 1.4488211870193481,
      "learning_rate": 3.8584269007982565e-05,
      "loss": 1.5612,
      "step": 20082
    },
    {
      "epoch": 0.7192149981198632,
      "grad_norm": 2.1801438331604004,
      "learning_rate": 3.857511567002835e-05,
      "loss": 1.1745,
      "step": 20083
    },
    {
      "epoch": 0.7192508102494315,
      "grad_norm": 1.371329665184021,
      "learning_rate": 3.8565963158466714e-05,
      "loss": 1.4076,
      "step": 20084
    },
    {
      "epoch": 0.7192866223789998,
      "grad_norm": 1.5765669345855713,
      "learning_rate": 3.855681147342084e-05,
      "loss": 1.4628,
      "step": 20085
    },
    {
      "epoch": 0.719322434508568,
      "grad_norm": 1.3415297269821167,
      "learning_rate": 3.854766061501378e-05,
      "loss": 1.6046,
      "step": 20086
    },
    {
      "epoch": 0.7193582466381363,
      "grad_norm": 1.9021408557891846,
      "learning_rate": 3.853851058336867e-05,
      "loss": 1.4807,
      "step": 20087
    },
    {
      "epoch": 0.7193940587677046,
      "grad_norm": 1.543630599975586,
      "learning_rate": 3.852936137860863e-05,
      "loss": 1.2401,
      "step": 20088
    },
    {
      "epoch": 0.719429870897273,
      "grad_norm": 1.5069166421890259,
      "learning_rate": 3.8520213000856763e-05,
      "loss": 1.6359,
      "step": 20089
    },
    {
      "epoch": 0.7194656830268412,
      "grad_norm": 1.6449099779129028,
      "learning_rate": 3.85110654502361e-05,
      "loss": 1.4342,
      "step": 20090
    },
    {
      "epoch": 0.7195014951564095,
      "grad_norm": 1.4722683429718018,
      "learning_rate": 3.8501918726869744e-05,
      "loss": 1.6538,
      "step": 20091
    },
    {
      "epoch": 0.7195373072859778,
      "grad_norm": 1.6070823669433594,
      "learning_rate": 3.8492772830880776e-05,
      "loss": 1.2967,
      "step": 20092
    },
    {
      "epoch": 0.719573119415546,
      "grad_norm": 1.3607196807861328,
      "learning_rate": 3.848362776239217e-05,
      "loss": 1.559,
      "step": 20093
    },
    {
      "epoch": 0.7196089315451143,
      "grad_norm": 1.7744579315185547,
      "learning_rate": 3.847448352152701e-05,
      "loss": 1.7096,
      "step": 20094
    },
    {
      "epoch": 0.7196447436746826,
      "grad_norm": 1.426999807357788,
      "learning_rate": 3.84653401084083e-05,
      "loss": 1.4129,
      "step": 20095
    },
    {
      "epoch": 0.7196805558042509,
      "grad_norm": 1.5814954042434692,
      "learning_rate": 3.8456197523159096e-05,
      "loss": 1.3991,
      "step": 20096
    },
    {
      "epoch": 0.7197163679338192,
      "grad_norm": 1.6949788331985474,
      "learning_rate": 3.844705576590235e-05,
      "loss": 1.2601,
      "step": 20097
    },
    {
      "epoch": 0.7197521800633875,
      "grad_norm": 1.731933832168579,
      "learning_rate": 3.843791483676107e-05,
      "loss": 1.5995,
      "step": 20098
    },
    {
      "epoch": 0.7197879921929558,
      "grad_norm": 1.780846357345581,
      "learning_rate": 3.842877473585823e-05,
      "loss": 1.2574,
      "step": 20099
    },
    {
      "epoch": 0.719823804322524,
      "grad_norm": 1.3572626113891602,
      "learning_rate": 3.841963546331684e-05,
      "loss": 1.4639,
      "step": 20100
    },
    {
      "epoch": 0.7198596164520923,
      "grad_norm": 1.2794239521026611,
      "learning_rate": 3.841049701925978e-05,
      "loss": 1.2574,
      "step": 20101
    },
    {
      "epoch": 0.7198954285816606,
      "grad_norm": 1.5886186361312866,
      "learning_rate": 3.840135940381006e-05,
      "loss": 1.3844,
      "step": 20102
    },
    {
      "epoch": 0.7199312407112289,
      "grad_norm": 2.106218099594116,
      "learning_rate": 3.839222261709061e-05,
      "loss": 1.3317,
      "step": 20103
    },
    {
      "epoch": 0.7199670528407972,
      "grad_norm": 2.0230941772460938,
      "learning_rate": 3.83830866592243e-05,
      "loss": 1.5562,
      "step": 20104
    },
    {
      "epoch": 0.7200028649703655,
      "grad_norm": 1.5232678651809692,
      "learning_rate": 3.8373951530334086e-05,
      "loss": 1.0045,
      "step": 20105
    },
    {
      "epoch": 0.7200386770999337,
      "grad_norm": 3.936558961868286,
      "learning_rate": 3.836481723054286e-05,
      "loss": 1.4348,
      "step": 20106
    },
    {
      "epoch": 0.720074489229502,
      "grad_norm": 1.2785394191741943,
      "learning_rate": 3.835568375997355e-05,
      "loss": 1.5436,
      "step": 20107
    },
    {
      "epoch": 0.7201103013590703,
      "grad_norm": 2.008934259414673,
      "learning_rate": 3.8346551118748967e-05,
      "loss": 1.3253,
      "step": 20108
    },
    {
      "epoch": 0.7201461134886386,
      "grad_norm": 1.5482290983200073,
      "learning_rate": 3.8337419306992e-05,
      "loss": 1.4049,
      "step": 20109
    },
    {
      "epoch": 0.7201819256182069,
      "grad_norm": 1.702072262763977,
      "learning_rate": 3.8328288324825566e-05,
      "loss": 1.4319,
      "step": 20110
    },
    {
      "epoch": 0.7202177377477752,
      "grad_norm": 2.1846251487731934,
      "learning_rate": 3.831915817237243e-05,
      "loss": 1.3064,
      "step": 20111
    },
    {
      "epoch": 0.7202535498773435,
      "grad_norm": 1.7581915855407715,
      "learning_rate": 3.831002884975544e-05,
      "loss": 1.4488,
      "step": 20112
    },
    {
      "epoch": 0.7202893620069117,
      "grad_norm": 1.5173016786575317,
      "learning_rate": 3.830090035709745e-05,
      "loss": 1.2531,
      "step": 20113
    },
    {
      "epoch": 0.72032517413648,
      "grad_norm": 1.3958326578140259,
      "learning_rate": 3.8291772694521285e-05,
      "loss": 1.5105,
      "step": 20114
    },
    {
      "epoch": 0.7203609862660483,
      "grad_norm": 2.2686309814453125,
      "learning_rate": 3.82826458621497e-05,
      "loss": 1.3909,
      "step": 20115
    },
    {
      "epoch": 0.7203967983956165,
      "grad_norm": 2.0584423542022705,
      "learning_rate": 3.82735198601055e-05,
      "loss": 1.6335,
      "step": 20116
    },
    {
      "epoch": 0.7204326105251849,
      "grad_norm": 1.203965187072754,
      "learning_rate": 3.8264394688511466e-05,
      "loss": 1.2574,
      "step": 20117
    },
    {
      "epoch": 0.7204684226547532,
      "grad_norm": 1.8112740516662598,
      "learning_rate": 3.82552703474904e-05,
      "loss": 1.4215,
      "step": 20118
    },
    {
      "epoch": 0.7205042347843215,
      "grad_norm": 1.7669355869293213,
      "learning_rate": 3.8246146837165e-05,
      "loss": 1.6208,
      "step": 20119
    },
    {
      "epoch": 0.7205400469138897,
      "grad_norm": 1.4590661525726318,
      "learning_rate": 3.823702415765803e-05,
      "loss": 1.3783,
      "step": 20120
    },
    {
      "epoch": 0.720575859043458,
      "grad_norm": 2.1782760620117188,
      "learning_rate": 3.822790230909227e-05,
      "loss": 1.64,
      "step": 20121
    },
    {
      "epoch": 0.7206116711730263,
      "grad_norm": 1.6818681955337524,
      "learning_rate": 3.821878129159037e-05,
      "loss": 1.6294,
      "step": 20122
    },
    {
      "epoch": 0.7206474833025945,
      "grad_norm": 1.7230101823806763,
      "learning_rate": 3.8209661105275077e-05,
      "loss": 1.3779,
      "step": 20123
    },
    {
      "epoch": 0.7206832954321629,
      "grad_norm": 1.3677054643630981,
      "learning_rate": 3.820054175026908e-05,
      "loss": 1.0637,
      "step": 20124
    },
    {
      "epoch": 0.7207191075617312,
      "grad_norm": 1.5686241388320923,
      "learning_rate": 3.8191423226695125e-05,
      "loss": 1.278,
      "step": 20125
    },
    {
      "epoch": 0.7207549196912995,
      "grad_norm": 1.5883556604385376,
      "learning_rate": 3.81823055346758e-05,
      "loss": 1.3986,
      "step": 20126
    },
    {
      "epoch": 0.7207907318208677,
      "grad_norm": 1.4147443771362305,
      "learning_rate": 3.817318867433383e-05,
      "loss": 1.3892,
      "step": 20127
    },
    {
      "epoch": 0.720826543950436,
      "grad_norm": 1.6232571601867676,
      "learning_rate": 3.816407264579187e-05,
      "loss": 1.2004,
      "step": 20128
    },
    {
      "epoch": 0.7208623560800043,
      "grad_norm": 1.8220059871673584,
      "learning_rate": 3.8154957449172524e-05,
      "loss": 1.1985,
      "step": 20129
    },
    {
      "epoch": 0.7208981682095725,
      "grad_norm": 1.627021074295044,
      "learning_rate": 3.814584308459849e-05,
      "loss": 1.3011,
      "step": 20130
    },
    {
      "epoch": 0.7209339803391409,
      "grad_norm": 1.998015284538269,
      "learning_rate": 3.8136729552192274e-05,
      "loss": 1.4227,
      "step": 20131
    },
    {
      "epoch": 0.7209697924687092,
      "grad_norm": 1.4177651405334473,
      "learning_rate": 3.812761685207664e-05,
      "loss": 1.2593,
      "step": 20132
    },
    {
      "epoch": 0.7210056045982774,
      "grad_norm": 1.8777852058410645,
      "learning_rate": 3.811850498437407e-05,
      "loss": 1.3507,
      "step": 20133
    },
    {
      "epoch": 0.7210414167278457,
      "grad_norm": 2.2260215282440186,
      "learning_rate": 3.81093939492072e-05,
      "loss": 1.3991,
      "step": 20134
    },
    {
      "epoch": 0.721077228857414,
      "grad_norm": 1.6670490503311157,
      "learning_rate": 3.810028374669859e-05,
      "loss": 1.2042,
      "step": 20135
    },
    {
      "epoch": 0.7211130409869823,
      "grad_norm": 2.2959189414978027,
      "learning_rate": 3.8091174376970876e-05,
      "loss": 1.3529,
      "step": 20136
    },
    {
      "epoch": 0.7211488531165505,
      "grad_norm": 1.5722131729125977,
      "learning_rate": 3.808206584014653e-05,
      "loss": 1.2084,
      "step": 20137
    },
    {
      "epoch": 0.7211846652461189,
      "grad_norm": 1.5049318075180054,
      "learning_rate": 3.807295813634807e-05,
      "loss": 1.6146,
      "step": 20138
    },
    {
      "epoch": 0.7212204773756872,
      "grad_norm": 2.4031875133514404,
      "learning_rate": 3.8063851265698134e-05,
      "loss": 1.467,
      "step": 20139
    },
    {
      "epoch": 0.7212562895052554,
      "grad_norm": 1.4516150951385498,
      "learning_rate": 3.805474522831916e-05,
      "loss": 1.5926,
      "step": 20140
    },
    {
      "epoch": 0.7212921016348237,
      "grad_norm": 1.6928277015686035,
      "learning_rate": 3.804564002433371e-05,
      "loss": 1.4698,
      "step": 20141
    },
    {
      "epoch": 0.721327913764392,
      "grad_norm": 2.470627546310425,
      "learning_rate": 3.8036535653864193e-05,
      "loss": 1.4727,
      "step": 20142
    },
    {
      "epoch": 0.7213637258939603,
      "grad_norm": 1.5808488130569458,
      "learning_rate": 3.8027432117033237e-05,
      "loss": 1.5674,
      "step": 20143
    },
    {
      "epoch": 0.7213995380235285,
      "grad_norm": 1.833074688911438,
      "learning_rate": 3.80183294139632e-05,
      "loss": 1.4833,
      "step": 20144
    },
    {
      "epoch": 0.7214353501530969,
      "grad_norm": 1.6991946697235107,
      "learning_rate": 3.8009227544776595e-05,
      "loss": 1.6437,
      "step": 20145
    },
    {
      "epoch": 0.7214711622826652,
      "grad_norm": 1.4071147441864014,
      "learning_rate": 3.80001265095959e-05,
      "loss": 0.9343,
      "step": 20146
    },
    {
      "epoch": 0.7215069744122334,
      "grad_norm": 1.4687845706939697,
      "learning_rate": 3.799102630854351e-05,
      "loss": 1.4609,
      "step": 20147
    },
    {
      "epoch": 0.7215427865418017,
      "grad_norm": 1.7575792074203491,
      "learning_rate": 3.79819269417419e-05,
      "loss": 1.5505,
      "step": 20148
    },
    {
      "epoch": 0.72157859867137,
      "grad_norm": 2.31231689453125,
      "learning_rate": 3.797282840931339e-05,
      "loss": 1.5541,
      "step": 20149
    },
    {
      "epoch": 0.7216144108009382,
      "grad_norm": 1.4037084579467773,
      "learning_rate": 3.796373071138054e-05,
      "loss": 1.6367,
      "step": 20150
    },
    {
      "epoch": 0.7216502229305065,
      "grad_norm": 2.1780078411102295,
      "learning_rate": 3.795463384806564e-05,
      "loss": 1.6328,
      "step": 20151
    },
    {
      "epoch": 0.7216860350600749,
      "grad_norm": 1.7256742715835571,
      "learning_rate": 3.794553781949114e-05,
      "loss": 1.3406,
      "step": 20152
    },
    {
      "epoch": 0.7217218471896432,
      "grad_norm": 1.9494801759719849,
      "learning_rate": 3.793644262577934e-05,
      "loss": 1.3391,
      "step": 20153
    },
    {
      "epoch": 0.7217576593192114,
      "grad_norm": 1.6500424146652222,
      "learning_rate": 3.7927348267052666e-05,
      "loss": 1.4641,
      "step": 20154
    },
    {
      "epoch": 0.7217934714487797,
      "grad_norm": 1.6580140590667725,
      "learning_rate": 3.791825474343348e-05,
      "loss": 1.5671,
      "step": 20155
    },
    {
      "epoch": 0.721829283578348,
      "grad_norm": 1.598800778388977,
      "learning_rate": 3.790916205504406e-05,
      "loss": 1.4017,
      "step": 20156
    },
    {
      "epoch": 0.7218650957079162,
      "grad_norm": 1.5164865255355835,
      "learning_rate": 3.7900070202006764e-05,
      "loss": 1.1569,
      "step": 20157
    },
    {
      "epoch": 0.7219009078374845,
      "grad_norm": 1.3159844875335693,
      "learning_rate": 3.789097918444394e-05,
      "loss": 1.4161,
      "step": 20158
    },
    {
      "epoch": 0.7219367199670529,
      "grad_norm": 1.960456132888794,
      "learning_rate": 3.78818890024779e-05,
      "loss": 1.3394,
      "step": 20159
    },
    {
      "epoch": 0.7219725320966212,
      "grad_norm": 1.6618950366973877,
      "learning_rate": 3.787279965623085e-05,
      "loss": 1.719,
      "step": 20160
    },
    {
      "epoch": 0.7220083442261894,
      "grad_norm": 1.2125014066696167,
      "learning_rate": 3.786371114582521e-05,
      "loss": 1.1418,
      "step": 20161
    },
    {
      "epoch": 0.7220441563557577,
      "grad_norm": 1.914084792137146,
      "learning_rate": 3.785462347138319e-05,
      "loss": 1.6195,
      "step": 20162
    },
    {
      "epoch": 0.722079968485326,
      "grad_norm": 1.5807491540908813,
      "learning_rate": 3.784553663302701e-05,
      "loss": 1.5063,
      "step": 20163
    },
    {
      "epoch": 0.7221157806148942,
      "grad_norm": 1.9150186777114868,
      "learning_rate": 3.783645063087896e-05,
      "loss": 1.2985,
      "step": 20164
    },
    {
      "epoch": 0.7221515927444625,
      "grad_norm": 2.09673810005188,
      "learning_rate": 3.782736546506128e-05,
      "loss": 1.5551,
      "step": 20165
    },
    {
      "epoch": 0.7221874048740309,
      "grad_norm": 1.5515776872634888,
      "learning_rate": 3.781828113569624e-05,
      "loss": 1.7333,
      "step": 20166
    },
    {
      "epoch": 0.7222232170035991,
      "grad_norm": 1.499321460723877,
      "learning_rate": 3.780919764290599e-05,
      "loss": 1.4817,
      "step": 20167
    },
    {
      "epoch": 0.7222590291331674,
      "grad_norm": 1.4306893348693848,
      "learning_rate": 3.780011498681276e-05,
      "loss": 1.2433,
      "step": 20168
    },
    {
      "epoch": 0.7222948412627357,
      "grad_norm": 1.8531023263931274,
      "learning_rate": 3.779103316753875e-05,
      "loss": 1.205,
      "step": 20169
    },
    {
      "epoch": 0.722330653392304,
      "grad_norm": 1.7147332429885864,
      "learning_rate": 3.778195218520618e-05,
      "loss": 1.3981,
      "step": 20170
    },
    {
      "epoch": 0.7223664655218722,
      "grad_norm": 1.5731045007705688,
      "learning_rate": 3.777287203993716e-05,
      "loss": 1.5347,
      "step": 20171
    },
    {
      "epoch": 0.7224022776514405,
      "grad_norm": 1.9253267049789429,
      "learning_rate": 3.7763792731853865e-05,
      "loss": 1.7233,
      "step": 20172
    },
    {
      "epoch": 0.7224380897810089,
      "grad_norm": 1.4065598249435425,
      "learning_rate": 3.77547142610785e-05,
      "loss": 1.1908,
      "step": 20173
    },
    {
      "epoch": 0.7224739019105771,
      "grad_norm": 1.5758453607559204,
      "learning_rate": 3.774563662773314e-05,
      "loss": 1.2734,
      "step": 20174
    },
    {
      "epoch": 0.7225097140401454,
      "grad_norm": 1.502901315689087,
      "learning_rate": 3.773655983193992e-05,
      "loss": 1.5944,
      "step": 20175
    },
    {
      "epoch": 0.7225455261697137,
      "grad_norm": 1.7962548732757568,
      "learning_rate": 3.772748387382099e-05,
      "loss": 1.2866,
      "step": 20176
    },
    {
      "epoch": 0.722581338299282,
      "grad_norm": 2.224817991256714,
      "learning_rate": 3.7718408753498456e-05,
      "loss": 1.3286,
      "step": 20177
    },
    {
      "epoch": 0.7226171504288502,
      "grad_norm": 1.7116045951843262,
      "learning_rate": 3.770933447109437e-05,
      "loss": 1.2324,
      "step": 20178
    },
    {
      "epoch": 0.7226529625584185,
      "grad_norm": 1.5392895936965942,
      "learning_rate": 3.7700261026730844e-05,
      "loss": 1.2427,
      "step": 20179
    },
    {
      "epoch": 0.7226887746879869,
      "grad_norm": 1.7307379245758057,
      "learning_rate": 3.7691188420529974e-05,
      "loss": 1.4682,
      "step": 20180
    },
    {
      "epoch": 0.7227245868175551,
      "grad_norm": 1.6550228595733643,
      "learning_rate": 3.768211665261375e-05,
      "loss": 1.5901,
      "step": 20181
    },
    {
      "epoch": 0.7227603989471234,
      "grad_norm": 1.491673469543457,
      "learning_rate": 3.7673045723104275e-05,
      "loss": 1.3454,
      "step": 20182
    },
    {
      "epoch": 0.7227962110766917,
      "grad_norm": 1.3128697872161865,
      "learning_rate": 3.7663975632123574e-05,
      "loss": 1.4408,
      "step": 20183
    },
    {
      "epoch": 0.7228320232062599,
      "grad_norm": 1.577620267868042,
      "learning_rate": 3.76549063797937e-05,
      "loss": 1.1176,
      "step": 20184
    },
    {
      "epoch": 0.7228678353358282,
      "grad_norm": 2.135578155517578,
      "learning_rate": 3.7645837966236605e-05,
      "loss": 1.4422,
      "step": 20185
    },
    {
      "epoch": 0.7229036474653965,
      "grad_norm": 1.4765774011611938,
      "learning_rate": 3.763677039157433e-05,
      "loss": 1.2351,
      "step": 20186
    },
    {
      "epoch": 0.7229394595949649,
      "grad_norm": 1.3106032609939575,
      "learning_rate": 3.762770365592887e-05,
      "loss": 1.3763,
      "step": 20187
    },
    {
      "epoch": 0.7229752717245331,
      "grad_norm": 1.6912516355514526,
      "learning_rate": 3.7618637759422236e-05,
      "loss": 1.4735,
      "step": 20188
    },
    {
      "epoch": 0.7230110838541014,
      "grad_norm": 1.7040541172027588,
      "learning_rate": 3.760957270217633e-05,
      "loss": 1.4154,
      "step": 20189
    },
    {
      "epoch": 0.7230468959836697,
      "grad_norm": 1.61090087890625,
      "learning_rate": 3.7600508484313146e-05,
      "loss": 1.4419,
      "step": 20190
    },
    {
      "epoch": 0.7230827081132379,
      "grad_norm": 1.7219356298446655,
      "learning_rate": 3.759144510595467e-05,
      "loss": 1.5413,
      "step": 20191
    },
    {
      "epoch": 0.7231185202428062,
      "grad_norm": 1.4904463291168213,
      "learning_rate": 3.7582382567222754e-05,
      "loss": 1.0032,
      "step": 20192
    },
    {
      "epoch": 0.7231543323723745,
      "grad_norm": 1.4005389213562012,
      "learning_rate": 3.757332086823937e-05,
      "loss": 1.4532,
      "step": 20193
    },
    {
      "epoch": 0.7231901445019429,
      "grad_norm": 1.4816468954086304,
      "learning_rate": 3.756426000912644e-05,
      "loss": 1.5392,
      "step": 20194
    },
    {
      "epoch": 0.7232259566315111,
      "grad_norm": 1.534940481185913,
      "learning_rate": 3.7555199990005874e-05,
      "loss": 1.1027,
      "step": 20195
    },
    {
      "epoch": 0.7232617687610794,
      "grad_norm": 1.4121631383895874,
      "learning_rate": 3.754614081099952e-05,
      "loss": 1.1582,
      "step": 20196
    },
    {
      "epoch": 0.7232975808906477,
      "grad_norm": 1.200635552406311,
      "learning_rate": 3.753708247222928e-05,
      "loss": 1.3734,
      "step": 20197
    },
    {
      "epoch": 0.7233333930202159,
      "grad_norm": 1.5664600133895874,
      "learning_rate": 3.752802497381706e-05,
      "loss": 1.4627,
      "step": 20198
    },
    {
      "epoch": 0.7233692051497842,
      "grad_norm": 1.6560693979263306,
      "learning_rate": 3.751896831588464e-05,
      "loss": 1.3806,
      "step": 20199
    },
    {
      "epoch": 0.7234050172793525,
      "grad_norm": 1.5157887935638428,
      "learning_rate": 3.7509912498553914e-05,
      "loss": 1.495,
      "step": 20200
    },
    {
      "epoch": 0.7234408294089208,
      "grad_norm": 2.2800793647766113,
      "learning_rate": 3.750085752194671e-05,
      "loss": 1.3048,
      "step": 20201
    },
    {
      "epoch": 0.7234766415384891,
      "grad_norm": 1.5321568250656128,
      "learning_rate": 3.749180338618488e-05,
      "loss": 1.5647,
      "step": 20202
    },
    {
      "epoch": 0.7235124536680574,
      "grad_norm": 1.694242238998413,
      "learning_rate": 3.7482750091390176e-05,
      "loss": 1.4978,
      "step": 20203
    },
    {
      "epoch": 0.7235482657976257,
      "grad_norm": 1.7178481817245483,
      "learning_rate": 3.7473697637684416e-05,
      "loss": 1.3819,
      "step": 20204
    },
    {
      "epoch": 0.7235840779271939,
      "grad_norm": 1.4250034093856812,
      "learning_rate": 3.746464602518941e-05,
      "loss": 1.4255,
      "step": 20205
    },
    {
      "epoch": 0.7236198900567622,
      "grad_norm": 1.6631689071655273,
      "learning_rate": 3.745559525402696e-05,
      "loss": 1.5659,
      "step": 20206
    },
    {
      "epoch": 0.7236557021863305,
      "grad_norm": 1.8999367952346802,
      "learning_rate": 3.744654532431876e-05,
      "loss": 1.5464,
      "step": 20207
    },
    {
      "epoch": 0.7236915143158988,
      "grad_norm": 1.983514428138733,
      "learning_rate": 3.743749623618661e-05,
      "loss": 1.4633,
      "step": 20208
    },
    {
      "epoch": 0.7237273264454671,
      "grad_norm": 2.2431626319885254,
      "learning_rate": 3.742844798975229e-05,
      "loss": 1.4338,
      "step": 20209
    },
    {
      "epoch": 0.7237631385750354,
      "grad_norm": 2.3390116691589355,
      "learning_rate": 3.7419400585137444e-05,
      "loss": 1.6569,
      "step": 20210
    },
    {
      "epoch": 0.7237989507046036,
      "grad_norm": 1.4298285245895386,
      "learning_rate": 3.741035402246385e-05,
      "loss": 1.4772,
      "step": 20211
    },
    {
      "epoch": 0.7238347628341719,
      "grad_norm": 1.4829427003860474,
      "learning_rate": 3.74013083018532e-05,
      "loss": 1.6161,
      "step": 20212
    },
    {
      "epoch": 0.7238705749637402,
      "grad_norm": 1.5598061084747314,
      "learning_rate": 3.7392263423427234e-05,
      "loss": 1.6156,
      "step": 20213
    },
    {
      "epoch": 0.7239063870933085,
      "grad_norm": 1.6008559465408325,
      "learning_rate": 3.738321938730758e-05,
      "loss": 1.3155,
      "step": 20214
    },
    {
      "epoch": 0.7239421992228768,
      "grad_norm": 1.7069321870803833,
      "learning_rate": 3.737417619361593e-05,
      "loss": 1.5145,
      "step": 20215
    },
    {
      "epoch": 0.7239780113524451,
      "grad_norm": 1.5327214002609253,
      "learning_rate": 3.7365133842473995e-05,
      "loss": 1.4894,
      "step": 20216
    },
    {
      "epoch": 0.7240138234820134,
      "grad_norm": 1.2927590608596802,
      "learning_rate": 3.735609233400336e-05,
      "loss": 1.4621,
      "step": 20217
    },
    {
      "epoch": 0.7240496356115816,
      "grad_norm": 1.4662754535675049,
      "learning_rate": 3.734705166832569e-05,
      "loss": 1.1987,
      "step": 20218
    },
    {
      "epoch": 0.7240854477411499,
      "grad_norm": 1.7217975854873657,
      "learning_rate": 3.7338011845562624e-05,
      "loss": 1.4568,
      "step": 20219
    },
    {
      "epoch": 0.7241212598707182,
      "grad_norm": 1.6077433824539185,
      "learning_rate": 3.732897286583582e-05,
      "loss": 1.4954,
      "step": 20220
    },
    {
      "epoch": 0.7241570720002865,
      "grad_norm": 1.8069626092910767,
      "learning_rate": 3.7319934729266814e-05,
      "loss": 1.6723,
      "step": 20221
    },
    {
      "epoch": 0.7241928841298548,
      "grad_norm": 1.7534021139144897,
      "learning_rate": 3.731089743597723e-05,
      "loss": 1.7045,
      "step": 20222
    },
    {
      "epoch": 0.7242286962594231,
      "grad_norm": 1.8184095621109009,
      "learning_rate": 3.7301860986088666e-05,
      "loss": 1.724,
      "step": 20223
    },
    {
      "epoch": 0.7242645083889914,
      "grad_norm": 2.0938992500305176,
      "learning_rate": 3.729282537972272e-05,
      "loss": 1.6204,
      "step": 20224
    },
    {
      "epoch": 0.7243003205185596,
      "grad_norm": 1.575728178024292,
      "learning_rate": 3.728379061700091e-05,
      "loss": 1.3117,
      "step": 20225
    },
    {
      "epoch": 0.7243361326481279,
      "grad_norm": 1.9121419191360474,
      "learning_rate": 3.727475669804474e-05,
      "loss": 1.3417,
      "step": 20226
    },
    {
      "epoch": 0.7243719447776962,
      "grad_norm": 2.2062628269195557,
      "learning_rate": 3.726572362297588e-05,
      "loss": 1.7373,
      "step": 20227
    },
    {
      "epoch": 0.7244077569072644,
      "grad_norm": 1.6102908849716187,
      "learning_rate": 3.725669139191574e-05,
      "loss": 1.3962,
      "step": 20228
    },
    {
      "epoch": 0.7244435690368328,
      "grad_norm": 1.7527639865875244,
      "learning_rate": 3.7247660004985897e-05,
      "loss": 1.5474,
      "step": 20229
    },
    {
      "epoch": 0.7244793811664011,
      "grad_norm": 2.023170232772827,
      "learning_rate": 3.723862946230784e-05,
      "loss": 1.3935,
      "step": 20230
    },
    {
      "epoch": 0.7245151932959694,
      "grad_norm": 1.7724480628967285,
      "learning_rate": 3.7229599764003096e-05,
      "loss": 1.3569,
      "step": 20231
    },
    {
      "epoch": 0.7245510054255376,
      "grad_norm": 1.76382315158844,
      "learning_rate": 3.7220570910193096e-05,
      "loss": 1.5279,
      "step": 20232
    },
    {
      "epoch": 0.7245868175551059,
      "grad_norm": 1.4322667121887207,
      "learning_rate": 3.721154290099933e-05,
      "loss": 1.259,
      "step": 20233
    },
    {
      "epoch": 0.7246226296846742,
      "grad_norm": 1.2567883729934692,
      "learning_rate": 3.7202515736543296e-05,
      "loss": 1.3401,
      "step": 20234
    },
    {
      "epoch": 0.7246584418142424,
      "grad_norm": 1.4475733041763306,
      "learning_rate": 3.7193489416946383e-05,
      "loss": 1.4377,
      "step": 20235
    },
    {
      "epoch": 0.7246942539438108,
      "grad_norm": 1.2449748516082764,
      "learning_rate": 3.718446394233007e-05,
      "loss": 1.5922,
      "step": 20236
    },
    {
      "epoch": 0.7247300660733791,
      "grad_norm": 1.514506220817566,
      "learning_rate": 3.717543931281572e-05,
      "loss": 1.3933,
      "step": 20237
    },
    {
      "epoch": 0.7247658782029474,
      "grad_norm": 1.5675554275512695,
      "learning_rate": 3.7166415528524854e-05,
      "loss": 1.0401,
      "step": 20238
    },
    {
      "epoch": 0.7248016903325156,
      "grad_norm": 1.6422276496887207,
      "learning_rate": 3.715739258957879e-05,
      "loss": 1.411,
      "step": 20239
    },
    {
      "epoch": 0.7248375024620839,
      "grad_norm": 1.1734343767166138,
      "learning_rate": 3.714837049609898e-05,
      "loss": 1.5028,
      "step": 20240
    },
    {
      "epoch": 0.7248733145916522,
      "grad_norm": 1.268044352531433,
      "learning_rate": 3.71393492482067e-05,
      "loss": 1.3508,
      "step": 20241
    },
    {
      "epoch": 0.7249091267212204,
      "grad_norm": 1.7053765058517456,
      "learning_rate": 3.713032884602346e-05,
      "loss": 1.4333,
      "step": 20242
    },
    {
      "epoch": 0.7249449388507888,
      "grad_norm": 2.118558406829834,
      "learning_rate": 3.712130928967056e-05,
      "loss": 1.5188,
      "step": 20243
    },
    {
      "epoch": 0.7249807509803571,
      "grad_norm": 1.3212299346923828,
      "learning_rate": 3.711229057926925e-05,
      "loss": 1.7121,
      "step": 20244
    },
    {
      "epoch": 0.7250165631099253,
      "grad_norm": 1.4363412857055664,
      "learning_rate": 3.710327271494103e-05,
      "loss": 1.4603,
      "step": 20245
    },
    {
      "epoch": 0.7250523752394936,
      "grad_norm": 2.0643067359924316,
      "learning_rate": 3.709425569680711e-05,
      "loss": 1.6963,
      "step": 20246
    },
    {
      "epoch": 0.7250881873690619,
      "grad_norm": 1.556532621383667,
      "learning_rate": 3.708523952498887e-05,
      "loss": 1.3901,
      "step": 20247
    },
    {
      "epoch": 0.7251239994986302,
      "grad_norm": 1.4042514562606812,
      "learning_rate": 3.707622419960751e-05,
      "loss": 1.5076,
      "step": 20248
    },
    {
      "epoch": 0.7251598116281984,
      "grad_norm": 2.3042328357696533,
      "learning_rate": 3.7067209720784456e-05,
      "loss": 1.2477,
      "step": 20249
    },
    {
      "epoch": 0.7251956237577668,
      "grad_norm": 1.4075384140014648,
      "learning_rate": 3.705819608864092e-05,
      "loss": 1.472,
      "step": 20250
    },
    {
      "epoch": 0.7252314358873351,
      "grad_norm": 1.6488524675369263,
      "learning_rate": 3.704918330329813e-05,
      "loss": 1.6962,
      "step": 20251
    },
    {
      "epoch": 0.7252672480169033,
      "grad_norm": 1.7311025857925415,
      "learning_rate": 3.704017136487737e-05,
      "loss": 1.3322,
      "step": 20252
    },
    {
      "epoch": 0.7253030601464716,
      "grad_norm": 1.5187498331069946,
      "learning_rate": 3.70311602734999e-05,
      "loss": 1.6023,
      "step": 20253
    },
    {
      "epoch": 0.7253388722760399,
      "grad_norm": 1.8480957746505737,
      "learning_rate": 3.702215002928699e-05,
      "loss": 1.3335,
      "step": 20254
    },
    {
      "epoch": 0.7253746844056081,
      "grad_norm": 2.199833393096924,
      "learning_rate": 3.701314063235972e-05,
      "loss": 1.33,
      "step": 20255
    },
    {
      "epoch": 0.7254104965351764,
      "grad_norm": 1.5730723142623901,
      "learning_rate": 3.7004132082839485e-05,
      "loss": 1.0013,
      "step": 20256
    },
    {
      "epoch": 0.7254463086647448,
      "grad_norm": 1.6218773126602173,
      "learning_rate": 3.699512438084736e-05,
      "loss": 1.2575,
      "step": 20257
    },
    {
      "epoch": 0.7254821207943131,
      "grad_norm": 2.079622507095337,
      "learning_rate": 3.6986117526504595e-05,
      "loss": 1.5331,
      "step": 20258
    },
    {
      "epoch": 0.7255179329238813,
      "grad_norm": 1.8061600923538208,
      "learning_rate": 3.6977111519932295e-05,
      "loss": 1.3609,
      "step": 20259
    },
    {
      "epoch": 0.7255537450534496,
      "grad_norm": 2.0372745990753174,
      "learning_rate": 3.696810636125168e-05,
      "loss": 1.6058,
      "step": 20260
    },
    {
      "epoch": 0.7255895571830179,
      "grad_norm": 1.9786975383758545,
      "learning_rate": 3.69591020505839e-05,
      "loss": 1.6915,
      "step": 20261
    },
    {
      "epoch": 0.7256253693125861,
      "grad_norm": 1.4099403619766235,
      "learning_rate": 3.6950098588050074e-05,
      "loss": 1.2547,
      "step": 20262
    },
    {
      "epoch": 0.7256611814421544,
      "grad_norm": 1.5284037590026855,
      "learning_rate": 3.6941095973771334e-05,
      "loss": 1.3404,
      "step": 20263
    },
    {
      "epoch": 0.7256969935717228,
      "grad_norm": 1.6371190547943115,
      "learning_rate": 3.6932094207868806e-05,
      "loss": 1.5385,
      "step": 20264
    },
    {
      "epoch": 0.7257328057012911,
      "grad_norm": 2.2500505447387695,
      "learning_rate": 3.692309329046364e-05,
      "loss": 1.6577,
      "step": 20265
    },
    {
      "epoch": 0.7257686178308593,
      "grad_norm": 1.5898199081420898,
      "learning_rate": 3.691409322167685e-05,
      "loss": 1.5561,
      "step": 20266
    },
    {
      "epoch": 0.7258044299604276,
      "grad_norm": 1.3862446546554565,
      "learning_rate": 3.690509400162957e-05,
      "loss": 1.1933,
      "step": 20267
    },
    {
      "epoch": 0.7258402420899959,
      "grad_norm": 2.0691616535186768,
      "learning_rate": 3.689609563044288e-05,
      "loss": 1.6209,
      "step": 20268
    },
    {
      "epoch": 0.7258760542195641,
      "grad_norm": 1.7785332202911377,
      "learning_rate": 3.68870981082378e-05,
      "loss": 1.4089,
      "step": 20269
    },
    {
      "epoch": 0.7259118663491324,
      "grad_norm": 1.5039234161376953,
      "learning_rate": 3.687810143513541e-05,
      "loss": 1.1616,
      "step": 20270
    },
    {
      "epoch": 0.7259476784787008,
      "grad_norm": 1.4419066905975342,
      "learning_rate": 3.686910561125675e-05,
      "loss": 1.4868,
      "step": 20271
    },
    {
      "epoch": 0.725983490608269,
      "grad_norm": 1.712319254875183,
      "learning_rate": 3.6860110636722856e-05,
      "loss": 1.5521,
      "step": 20272
    },
    {
      "epoch": 0.7260193027378373,
      "grad_norm": 1.749627709388733,
      "learning_rate": 3.6851116511654705e-05,
      "loss": 1.3738,
      "step": 20273
    },
    {
      "epoch": 0.7260551148674056,
      "grad_norm": 1.3292651176452637,
      "learning_rate": 3.684212323617333e-05,
      "loss": 1.4005,
      "step": 20274
    },
    {
      "epoch": 0.7260909269969739,
      "grad_norm": 1.6305632591247559,
      "learning_rate": 3.683313081039971e-05,
      "loss": 1.5385,
      "step": 20275
    },
    {
      "epoch": 0.7261267391265421,
      "grad_norm": 1.43117094039917,
      "learning_rate": 3.6824139234454876e-05,
      "loss": 1.5204,
      "step": 20276
    },
    {
      "epoch": 0.7261625512561104,
      "grad_norm": 1.7682982683181763,
      "learning_rate": 3.681514850845972e-05,
      "loss": 1.5069,
      "step": 20277
    },
    {
      "epoch": 0.7261983633856788,
      "grad_norm": 1.9330202341079712,
      "learning_rate": 3.6806158632535235e-05,
      "loss": 1.547,
      "step": 20278
    },
    {
      "epoch": 0.726234175515247,
      "grad_norm": 1.5854822397232056,
      "learning_rate": 3.679716960680242e-05,
      "loss": 1.6898,
      "step": 20279
    },
    {
      "epoch": 0.7262699876448153,
      "grad_norm": 1.5087810754776,
      "learning_rate": 3.6788181431382106e-05,
      "loss": 1.6337,
      "step": 20280
    },
    {
      "epoch": 0.7263057997743836,
      "grad_norm": 1.8749876022338867,
      "learning_rate": 3.6779194106395285e-05,
      "loss": 1.6861,
      "step": 20281
    },
    {
      "epoch": 0.7263416119039519,
      "grad_norm": 1.8120014667510986,
      "learning_rate": 3.677020763196286e-05,
      "loss": 1.673,
      "step": 20282
    },
    {
      "epoch": 0.7263774240335201,
      "grad_norm": 3.601649522781372,
      "learning_rate": 3.676122200820577e-05,
      "loss": 2.1907,
      "step": 20283
    },
    {
      "epoch": 0.7264132361630884,
      "grad_norm": 1.9699167013168335,
      "learning_rate": 3.6752237235244825e-05,
      "loss": 1.4932,
      "step": 20284
    },
    {
      "epoch": 0.7264490482926568,
      "grad_norm": 1.2282700538635254,
      "learning_rate": 3.6743253313200945e-05,
      "loss": 1.2088,
      "step": 20285
    },
    {
      "epoch": 0.726484860422225,
      "grad_norm": 1.4451675415039062,
      "learning_rate": 3.673427024219502e-05,
      "loss": 1.669,
      "step": 20286
    },
    {
      "epoch": 0.7265206725517933,
      "grad_norm": 1.7398184537887573,
      "learning_rate": 3.672528802234786e-05,
      "loss": 1.1963,
      "step": 20287
    },
    {
      "epoch": 0.7265564846813616,
      "grad_norm": 1.5839215517044067,
      "learning_rate": 3.671630665378033e-05,
      "loss": 1.6271,
      "step": 20288
    },
    {
      "epoch": 0.7265922968109298,
      "grad_norm": 1.601536512374878,
      "learning_rate": 3.670732613661326e-05,
      "loss": 1.5271,
      "step": 20289
    },
    {
      "epoch": 0.7266281089404981,
      "grad_norm": 1.5038881301879883,
      "learning_rate": 3.669834647096752e-05,
      "loss": 1.4387,
      "step": 20290
    },
    {
      "epoch": 0.7266639210700664,
      "grad_norm": 3.174227476119995,
      "learning_rate": 3.668936765696383e-05,
      "loss": 1.489,
      "step": 20291
    },
    {
      "epoch": 0.7266997331996348,
      "grad_norm": 1.7805509567260742,
      "learning_rate": 3.6680389694723025e-05,
      "loss": 1.6459,
      "step": 20292
    },
    {
      "epoch": 0.726735545329203,
      "grad_norm": 2.242903232574463,
      "learning_rate": 3.667141258436592e-05,
      "loss": 1.7589,
      "step": 20293
    },
    {
      "epoch": 0.7267713574587713,
      "grad_norm": 1.5864604711532593,
      "learning_rate": 3.666243632601329e-05,
      "loss": 1.6975,
      "step": 20294
    },
    {
      "epoch": 0.7268071695883396,
      "grad_norm": 1.4965282678604126,
      "learning_rate": 3.6653460919785855e-05,
      "loss": 1.4238,
      "step": 20295
    },
    {
      "epoch": 0.7268429817179078,
      "grad_norm": 1.7679928541183472,
      "learning_rate": 3.6644486365804385e-05,
      "loss": 1.2437,
      "step": 20296
    },
    {
      "epoch": 0.7268787938474761,
      "grad_norm": 1.58811616897583,
      "learning_rate": 3.663551266418966e-05,
      "loss": 1.5576,
      "step": 20297
    },
    {
      "epoch": 0.7269146059770444,
      "grad_norm": 1.8119292259216309,
      "learning_rate": 3.662653981506235e-05,
      "loss": 1.4472,
      "step": 20298
    },
    {
      "epoch": 0.7269504181066128,
      "grad_norm": 2.19586443901062,
      "learning_rate": 3.661756781854321e-05,
      "loss": 1.371,
      "step": 20299
    },
    {
      "epoch": 0.726986230236181,
      "grad_norm": 1.741518259048462,
      "learning_rate": 3.660859667475293e-05,
      "loss": 1.3297,
      "step": 20300
    },
    {
      "epoch": 0.7270220423657493,
      "grad_norm": 3.0605978965759277,
      "learning_rate": 3.659962638381224e-05,
      "loss": 1.763,
      "step": 20301
    },
    {
      "epoch": 0.7270578544953176,
      "grad_norm": 1.5931988954544067,
      "learning_rate": 3.6590656945841775e-05,
      "loss": 1.4453,
      "step": 20302
    },
    {
      "epoch": 0.7270936666248858,
      "grad_norm": 1.3028597831726074,
      "learning_rate": 3.6581688360962206e-05,
      "loss": 1.4289,
      "step": 20303
    },
    {
      "epoch": 0.7271294787544541,
      "grad_norm": 1.5821110010147095,
      "learning_rate": 3.6572720629294276e-05,
      "loss": 1.7085,
      "step": 20304
    },
    {
      "epoch": 0.7271652908840224,
      "grad_norm": 1.6137861013412476,
      "learning_rate": 3.656375375095853e-05,
      "loss": 1.2497,
      "step": 20305
    },
    {
      "epoch": 0.7272011030135908,
      "grad_norm": 1.879036545753479,
      "learning_rate": 3.655478772607565e-05,
      "loss": 1.6787,
      "step": 20306
    },
    {
      "epoch": 0.727236915143159,
      "grad_norm": 1.7845239639282227,
      "learning_rate": 3.654582255476626e-05,
      "loss": 1.6337,
      "step": 20307
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 2.2749297618865967,
      "learning_rate": 3.6536858237151015e-05,
      "loss": 1.226,
      "step": 20308
    },
    {
      "epoch": 0.7273085394022956,
      "grad_norm": 1.7021535634994507,
      "learning_rate": 3.652789477335045e-05,
      "loss": 1.4401,
      "step": 20309
    },
    {
      "epoch": 0.7273443515318638,
      "grad_norm": 2.237609624862671,
      "learning_rate": 3.651893216348517e-05,
      "loss": 1.3534,
      "step": 20310
    },
    {
      "epoch": 0.7273801636614321,
      "grad_norm": 1.8463900089263916,
      "learning_rate": 3.65099704076758e-05,
      "loss": 1.2325,
      "step": 20311
    },
    {
      "epoch": 0.7274159757910004,
      "grad_norm": 1.7602739334106445,
      "learning_rate": 3.650100950604289e-05,
      "loss": 1.4707,
      "step": 20312
    },
    {
      "epoch": 0.7274517879205687,
      "grad_norm": 1.4012271165847778,
      "learning_rate": 3.649204945870701e-05,
      "loss": 1.4284,
      "step": 20313
    },
    {
      "epoch": 0.727487600050137,
      "grad_norm": 1.6311163902282715,
      "learning_rate": 3.6483090265788614e-05,
      "loss": 1.2821,
      "step": 20314
    },
    {
      "epoch": 0.7275234121797053,
      "grad_norm": 1.6619857549667358,
      "learning_rate": 3.647413192740836e-05,
      "loss": 1.284,
      "step": 20315
    },
    {
      "epoch": 0.7275592243092736,
      "grad_norm": 1.6856287717819214,
      "learning_rate": 3.64651744436867e-05,
      "loss": 1.4404,
      "step": 20316
    },
    {
      "epoch": 0.7275950364388418,
      "grad_norm": 2.37723708152771,
      "learning_rate": 3.6456217814744165e-05,
      "loss": 1.6775,
      "step": 20317
    },
    {
      "epoch": 0.7276308485684101,
      "grad_norm": 1.4512906074523926,
      "learning_rate": 3.644726204070125e-05,
      "loss": 1.6137,
      "step": 20318
    },
    {
      "epoch": 0.7276666606979784,
      "grad_norm": 1.3987200260162354,
      "learning_rate": 3.643830712167847e-05,
      "loss": 1.5412,
      "step": 20319
    },
    {
      "epoch": 0.7277024728275467,
      "grad_norm": 1.5745872259140015,
      "learning_rate": 3.6429353057796255e-05,
      "loss": 1.3009,
      "step": 20320
    },
    {
      "epoch": 0.727738284957115,
      "grad_norm": 2.594204902648926,
      "learning_rate": 3.642039984917509e-05,
      "loss": 1.5018,
      "step": 20321
    },
    {
      "epoch": 0.7277740970866833,
      "grad_norm": 1.6258982419967651,
      "learning_rate": 3.641144749593548e-05,
      "loss": 1.3628,
      "step": 20322
    },
    {
      "epoch": 0.7278099092162515,
      "grad_norm": 1.546212911605835,
      "learning_rate": 3.640249599819777e-05,
      "loss": 1.53,
      "step": 20323
    },
    {
      "epoch": 0.7278457213458198,
      "grad_norm": 1.531711220741272,
      "learning_rate": 3.639354535608248e-05,
      "loss": 1.4359,
      "step": 20324
    },
    {
      "epoch": 0.7278815334753881,
      "grad_norm": 1.293178677558899,
      "learning_rate": 3.638459556970993e-05,
      "loss": 1.3002,
      "step": 20325
    },
    {
      "epoch": 0.7279173456049564,
      "grad_norm": 1.8886504173278809,
      "learning_rate": 3.637564663920066e-05,
      "loss": 1.5775,
      "step": 20326
    },
    {
      "epoch": 0.7279531577345247,
      "grad_norm": 1.3030343055725098,
      "learning_rate": 3.636669856467495e-05,
      "loss": 1.443,
      "step": 20327
    },
    {
      "epoch": 0.727988969864093,
      "grad_norm": 1.8650761842727661,
      "learning_rate": 3.635775134625323e-05,
      "loss": 1.484,
      "step": 20328
    },
    {
      "epoch": 0.7280247819936613,
      "grad_norm": 1.1256822347640991,
      "learning_rate": 3.634880498405587e-05,
      "loss": 1.3617,
      "step": 20329
    },
    {
      "epoch": 0.7280605941232295,
      "grad_norm": 1.5693796873092651,
      "learning_rate": 3.6339859478203274e-05,
      "loss": 1.2764,
      "step": 20330
    },
    {
      "epoch": 0.7280964062527978,
      "grad_norm": 2.721557378768921,
      "learning_rate": 3.6330914828815755e-05,
      "loss": 1.6122,
      "step": 20331
    },
    {
      "epoch": 0.7281322183823661,
      "grad_norm": 1.8987950086593628,
      "learning_rate": 3.632197103601358e-05,
      "loss": 1.1593,
      "step": 20332
    },
    {
      "epoch": 0.7281680305119343,
      "grad_norm": 1.3564764261245728,
      "learning_rate": 3.6313028099917226e-05,
      "loss": 1.4216,
      "step": 20333
    },
    {
      "epoch": 0.7282038426415027,
      "grad_norm": 1.3423123359680176,
      "learning_rate": 3.6304086020646874e-05,
      "loss": 1.2289,
      "step": 20334
    },
    {
      "epoch": 0.728239654771071,
      "grad_norm": 2.045650005340576,
      "learning_rate": 3.629514479832292e-05,
      "loss": 1.5775,
      "step": 20335
    },
    {
      "epoch": 0.7282754669006393,
      "grad_norm": 1.7621314525604248,
      "learning_rate": 3.628620443306556e-05,
      "loss": 1.4242,
      "step": 20336
    },
    {
      "epoch": 0.7283112790302075,
      "grad_norm": 1.5578246116638184,
      "learning_rate": 3.62772649249952e-05,
      "loss": 1.4387,
      "step": 20337
    },
    {
      "epoch": 0.7283470911597758,
      "grad_norm": 1.3517426252365112,
      "learning_rate": 3.626832627423201e-05,
      "loss": 1.7368,
      "step": 20338
    },
    {
      "epoch": 0.7283829032893441,
      "grad_norm": 1.414923906326294,
      "learning_rate": 3.6259388480896316e-05,
      "loss": 1.4573,
      "step": 20339
    },
    {
      "epoch": 0.7284187154189123,
      "grad_norm": 1.8400875329971313,
      "learning_rate": 3.62504515451083e-05,
      "loss": 1.3403,
      "step": 20340
    },
    {
      "epoch": 0.7284545275484807,
      "grad_norm": 1.4856648445129395,
      "learning_rate": 3.624151546698822e-05,
      "loss": 1.5307,
      "step": 20341
    },
    {
      "epoch": 0.728490339678049,
      "grad_norm": 1.957085371017456,
      "learning_rate": 3.623258024665635e-05,
      "loss": 1.3083,
      "step": 20342
    },
    {
      "epoch": 0.7285261518076173,
      "grad_norm": 1.5131561756134033,
      "learning_rate": 3.6223645884232784e-05,
      "loss": 1.4215,
      "step": 20343
    },
    {
      "epoch": 0.7285619639371855,
      "grad_norm": 1.8135859966278076,
      "learning_rate": 3.621471237983787e-05,
      "loss": 1.5306,
      "step": 20344
    },
    {
      "epoch": 0.7285977760667538,
      "grad_norm": 2.384350538253784,
      "learning_rate": 3.620577973359168e-05,
      "loss": 1.492,
      "step": 20345
    },
    {
      "epoch": 0.7286335881963221,
      "grad_norm": 1.0702521800994873,
      "learning_rate": 3.619684794561448e-05,
      "loss": 0.9597,
      "step": 20346
    },
    {
      "epoch": 0.7286694003258903,
      "grad_norm": 1.8656784296035767,
      "learning_rate": 3.618791701602635e-05,
      "loss": 1.4725,
      "step": 20347
    },
    {
      "epoch": 0.7287052124554587,
      "grad_norm": 1.8539940118789673,
      "learning_rate": 3.617898694494749e-05,
      "loss": 1.6358,
      "step": 20348
    },
    {
      "epoch": 0.728741024585027,
      "grad_norm": 1.284334659576416,
      "learning_rate": 3.6170057732498064e-05,
      "loss": 1.6315,
      "step": 20349
    },
    {
      "epoch": 0.7287768367145953,
      "grad_norm": 1.698467493057251,
      "learning_rate": 3.616112937879814e-05,
      "loss": 1.4993,
      "step": 20350
    },
    {
      "epoch": 0.7288126488441635,
      "grad_norm": 1.6168347597122192,
      "learning_rate": 3.6152201883967885e-05,
      "loss": 1.56,
      "step": 20351
    },
    {
      "epoch": 0.7288484609737318,
      "grad_norm": 1.26214599609375,
      "learning_rate": 3.6143275248127394e-05,
      "loss": 1.4896,
      "step": 20352
    },
    {
      "epoch": 0.7288842731033001,
      "grad_norm": 1.7499330043792725,
      "learning_rate": 3.61343494713968e-05,
      "loss": 1.2003,
      "step": 20353
    },
    {
      "epoch": 0.7289200852328683,
      "grad_norm": 2.1772499084472656,
      "learning_rate": 3.612542455389608e-05,
      "loss": 1.5853,
      "step": 20354
    },
    {
      "epoch": 0.7289558973624367,
      "grad_norm": 1.5749849081039429,
      "learning_rate": 3.611650049574545e-05,
      "loss": 1.5532,
      "step": 20355
    },
    {
      "epoch": 0.728991709492005,
      "grad_norm": 1.753783106803894,
      "learning_rate": 3.61075772970649e-05,
      "loss": 1.4504,
      "step": 20356
    },
    {
      "epoch": 0.7290275216215732,
      "grad_norm": 2.087217330932617,
      "learning_rate": 3.609865495797445e-05,
      "loss": 1.6384,
      "step": 20357
    },
    {
      "epoch": 0.7290633337511415,
      "grad_norm": 1.6963598728179932,
      "learning_rate": 3.608973347859418e-05,
      "loss": 1.388,
      "step": 20358
    },
    {
      "epoch": 0.7290991458807098,
      "grad_norm": 1.416736364364624,
      "learning_rate": 3.6080812859044086e-05,
      "loss": 1.4647,
      "step": 20359
    },
    {
      "epoch": 0.7291349580102781,
      "grad_norm": 2.7827279567718506,
      "learning_rate": 3.607189309944427e-05,
      "loss": 1.3479,
      "step": 20360
    },
    {
      "epoch": 0.7291707701398463,
      "grad_norm": 1.966021180152893,
      "learning_rate": 3.6062974199914615e-05,
      "loss": 1.5521,
      "step": 20361
    },
    {
      "epoch": 0.7292065822694147,
      "grad_norm": 1.5805641412734985,
      "learning_rate": 3.6054056160575164e-05,
      "loss": 1.3172,
      "step": 20362
    },
    {
      "epoch": 0.729242394398983,
      "grad_norm": 1.9374136924743652,
      "learning_rate": 3.6045138981545915e-05,
      "loss": 1.4399,
      "step": 20363
    },
    {
      "epoch": 0.7292782065285512,
      "grad_norm": 2.0594398975372314,
      "learning_rate": 3.603622266294686e-05,
      "loss": 1.1237,
      "step": 20364
    },
    {
      "epoch": 0.7293140186581195,
      "grad_norm": 1.5763599872589111,
      "learning_rate": 3.6027307204897886e-05,
      "loss": 1.3851,
      "step": 20365
    },
    {
      "epoch": 0.7293498307876878,
      "grad_norm": 1.818981409072876,
      "learning_rate": 3.601839260751897e-05,
      "loss": 1.3178,
      "step": 20366
    },
    {
      "epoch": 0.729385642917256,
      "grad_norm": 1.4879289865493774,
      "learning_rate": 3.600947887093009e-05,
      "loss": 1.3811,
      "step": 20367
    },
    {
      "epoch": 0.7294214550468243,
      "grad_norm": 1.7002722024917603,
      "learning_rate": 3.600056599525109e-05,
      "loss": 1.3244,
      "step": 20368
    },
    {
      "epoch": 0.7294572671763927,
      "grad_norm": 1.7715823650360107,
      "learning_rate": 3.5991653980601926e-05,
      "loss": 1.2399,
      "step": 20369
    },
    {
      "epoch": 0.729493079305961,
      "grad_norm": 2.60311222076416,
      "learning_rate": 3.59827428271025e-05,
      "loss": 1.6152,
      "step": 20370
    },
    {
      "epoch": 0.7295288914355292,
      "grad_norm": 1.446850061416626,
      "learning_rate": 3.597383253487272e-05,
      "loss": 1.3049,
      "step": 20371
    },
    {
      "epoch": 0.7295647035650975,
      "grad_norm": 1.3441686630249023,
      "learning_rate": 3.59649231040324e-05,
      "loss": 1.4572,
      "step": 20372
    },
    {
      "epoch": 0.7296005156946658,
      "grad_norm": 1.666536569595337,
      "learning_rate": 3.595601453470143e-05,
      "loss": 1.5172,
      "step": 20373
    },
    {
      "epoch": 0.729636327824234,
      "grad_norm": 1.4410146474838257,
      "learning_rate": 3.594710682699972e-05,
      "loss": 1.6782,
      "step": 20374
    },
    {
      "epoch": 0.7296721399538023,
      "grad_norm": 1.8410130739212036,
      "learning_rate": 3.5938199981047036e-05,
      "loss": 1.5535,
      "step": 20375
    },
    {
      "epoch": 0.7297079520833707,
      "grad_norm": 1.9997683763504028,
      "learning_rate": 3.592929399696323e-05,
      "loss": 1.3053,
      "step": 20376
    },
    {
      "epoch": 0.729743764212939,
      "grad_norm": 1.5407726764678955,
      "learning_rate": 3.592038887486813e-05,
      "loss": 1.4949,
      "step": 20377
    },
    {
      "epoch": 0.7297795763425072,
      "grad_norm": 2.03817081451416,
      "learning_rate": 3.591148461488157e-05,
      "loss": 1.5091,
      "step": 20378
    },
    {
      "epoch": 0.7298153884720755,
      "grad_norm": 1.7712717056274414,
      "learning_rate": 3.590258121712329e-05,
      "loss": 1.8162,
      "step": 20379
    },
    {
      "epoch": 0.7298512006016438,
      "grad_norm": 1.590785264968872,
      "learning_rate": 3.589367868171309e-05,
      "loss": 1.2453,
      "step": 20380
    },
    {
      "epoch": 0.729887012731212,
      "grad_norm": 1.64919114112854,
      "learning_rate": 3.5884777008770765e-05,
      "loss": 1.5673,
      "step": 20381
    },
    {
      "epoch": 0.7299228248607803,
      "grad_norm": 1.4542080163955688,
      "learning_rate": 3.587587619841609e-05,
      "loss": 0.9534,
      "step": 20382
    },
    {
      "epoch": 0.7299586369903487,
      "grad_norm": 1.3254774808883667,
      "learning_rate": 3.586697625076876e-05,
      "loss": 1.4704,
      "step": 20383
    },
    {
      "epoch": 0.729994449119917,
      "grad_norm": 1.9751310348510742,
      "learning_rate": 3.585807716594853e-05,
      "loss": 1.4692,
      "step": 20384
    },
    {
      "epoch": 0.7300302612494852,
      "grad_norm": 1.78018057346344,
      "learning_rate": 3.584917894407517e-05,
      "loss": 1.4879,
      "step": 20385
    },
    {
      "epoch": 0.7300660733790535,
      "grad_norm": 1.4248936176300049,
      "learning_rate": 3.584028158526832e-05,
      "loss": 1.4147,
      "step": 20386
    },
    {
      "epoch": 0.7301018855086218,
      "grad_norm": 2.423626184463501,
      "learning_rate": 3.583138508964773e-05,
      "loss": 1.6582,
      "step": 20387
    },
    {
      "epoch": 0.73013769763819,
      "grad_norm": 1.622358798980713,
      "learning_rate": 3.582248945733307e-05,
      "loss": 1.6244,
      "step": 20388
    },
    {
      "epoch": 0.7301735097677583,
      "grad_norm": 2.3228282928466797,
      "learning_rate": 3.581359468844408e-05,
      "loss": 1.8354,
      "step": 20389
    },
    {
      "epoch": 0.7302093218973267,
      "grad_norm": 1.6981981992721558,
      "learning_rate": 3.580470078310034e-05,
      "loss": 1.3045,
      "step": 20390
    },
    {
      "epoch": 0.730245134026895,
      "grad_norm": 2.001455068588257,
      "learning_rate": 3.579580774142155e-05,
      "loss": 1.2112,
      "step": 20391
    },
    {
      "epoch": 0.7302809461564632,
      "grad_norm": 2.1798324584960938,
      "learning_rate": 3.5786915563527376e-05,
      "loss": 1.155,
      "step": 20392
    },
    {
      "epoch": 0.7303167582860315,
      "grad_norm": 1.86077880859375,
      "learning_rate": 3.577802424953739e-05,
      "loss": 1.7546,
      "step": 20393
    },
    {
      "epoch": 0.7303525704155998,
      "grad_norm": 2.3468453884124756,
      "learning_rate": 3.576913379957125e-05,
      "loss": 1.22,
      "step": 20394
    },
    {
      "epoch": 0.730388382545168,
      "grad_norm": 1.3611379861831665,
      "learning_rate": 3.5760244213748565e-05,
      "loss": 1.3976,
      "step": 20395
    },
    {
      "epoch": 0.7304241946747363,
      "grad_norm": 1.5339411497116089,
      "learning_rate": 3.575135549218895e-05,
      "loss": 1.5949,
      "step": 20396
    },
    {
      "epoch": 0.7304600068043047,
      "grad_norm": 2.1556475162506104,
      "learning_rate": 3.5742467635011956e-05,
      "loss": 1.4444,
      "step": 20397
    },
    {
      "epoch": 0.7304958189338729,
      "grad_norm": 1.9611434936523438,
      "learning_rate": 3.5733580642337174e-05,
      "loss": 1.4917,
      "step": 20398
    },
    {
      "epoch": 0.7305316310634412,
      "grad_norm": 1.5778285264968872,
      "learning_rate": 3.572469451428415e-05,
      "loss": 1.274,
      "step": 20399
    },
    {
      "epoch": 0.7305674431930095,
      "grad_norm": 1.7948662042617798,
      "learning_rate": 3.57158092509725e-05,
      "loss": 1.245,
      "step": 20400
    },
    {
      "epoch": 0.7306032553225777,
      "grad_norm": 1.58558988571167,
      "learning_rate": 3.5706924852521674e-05,
      "loss": 1.4525,
      "step": 20401
    },
    {
      "epoch": 0.730639067452146,
      "grad_norm": 2.173316478729248,
      "learning_rate": 3.5698041319051245e-05,
      "loss": 1.7764,
      "step": 20402
    },
    {
      "epoch": 0.7306748795817143,
      "grad_norm": 1.275415062904358,
      "learning_rate": 3.5689158650680765e-05,
      "loss": 1.5483,
      "step": 20403
    },
    {
      "epoch": 0.7307106917112826,
      "grad_norm": 1.8517347574234009,
      "learning_rate": 3.568027684752966e-05,
      "loss": 1.6182,
      "step": 20404
    },
    {
      "epoch": 0.7307465038408509,
      "grad_norm": 1.9856765270233154,
      "learning_rate": 3.5671395909717477e-05,
      "loss": 1.3489,
      "step": 20405
    },
    {
      "epoch": 0.7307823159704192,
      "grad_norm": 1.7788243293762207,
      "learning_rate": 3.566251583736367e-05,
      "loss": 1.5068,
      "step": 20406
    },
    {
      "epoch": 0.7308181280999875,
      "grad_norm": 1.406541109085083,
      "learning_rate": 3.5653636630587764e-05,
      "loss": 1.3148,
      "step": 20407
    },
    {
      "epoch": 0.7308539402295557,
      "grad_norm": 1.4304836988449097,
      "learning_rate": 3.5644758289509126e-05,
      "loss": 1.1475,
      "step": 20408
    },
    {
      "epoch": 0.730889752359124,
      "grad_norm": 1.7873742580413818,
      "learning_rate": 3.563588081424727e-05,
      "loss": 1.363,
      "step": 20409
    },
    {
      "epoch": 0.7309255644886923,
      "grad_norm": 1.6076723337173462,
      "learning_rate": 3.5627004204921645e-05,
      "loss": 1.7126,
      "step": 20410
    },
    {
      "epoch": 0.7309613766182605,
      "grad_norm": 1.798566460609436,
      "learning_rate": 3.561812846165161e-05,
      "loss": 1.4759,
      "step": 20411
    },
    {
      "epoch": 0.7309971887478289,
      "grad_norm": 2.0032665729522705,
      "learning_rate": 3.56092535845566e-05,
      "loss": 1.3545,
      "step": 20412
    },
    {
      "epoch": 0.7310330008773972,
      "grad_norm": 1.4603948593139648,
      "learning_rate": 3.560037957375604e-05,
      "loss": 1.3149,
      "step": 20413
    },
    {
      "epoch": 0.7310688130069655,
      "grad_norm": 1.8404431343078613,
      "learning_rate": 3.5591506429369325e-05,
      "loss": 1.7072,
      "step": 20414
    },
    {
      "epoch": 0.7311046251365337,
      "grad_norm": 1.4005978107452393,
      "learning_rate": 3.558263415151578e-05,
      "loss": 1.4017,
      "step": 20415
    },
    {
      "epoch": 0.731140437266102,
      "grad_norm": 1.3122055530548096,
      "learning_rate": 3.557376274031481e-05,
      "loss": 1.3193,
      "step": 20416
    },
    {
      "epoch": 0.7311762493956703,
      "grad_norm": 1.732801914215088,
      "learning_rate": 3.556489219588575e-05,
      "loss": 1.8159,
      "step": 20417
    },
    {
      "epoch": 0.7312120615252385,
      "grad_norm": 1.7822617292404175,
      "learning_rate": 3.5556022518347975e-05,
      "loss": 1.529,
      "step": 20418
    },
    {
      "epoch": 0.7312478736548069,
      "grad_norm": 1.499024510383606,
      "learning_rate": 3.55471537078208e-05,
      "loss": 1.2738,
      "step": 20419
    },
    {
      "epoch": 0.7312836857843752,
      "grad_norm": 3.1258182525634766,
      "learning_rate": 3.553828576442346e-05,
      "loss": 2.0017,
      "step": 20420
    },
    {
      "epoch": 0.7313194979139435,
      "grad_norm": 1.561532974243164,
      "learning_rate": 3.552941868827542e-05,
      "loss": 1.5181,
      "step": 20421
    },
    {
      "epoch": 0.7313553100435117,
      "grad_norm": 1.2803188562393188,
      "learning_rate": 3.552055247949584e-05,
      "loss": 1.5402,
      "step": 20422
    },
    {
      "epoch": 0.73139112217308,
      "grad_norm": 2.332247734069824,
      "learning_rate": 3.5511687138204097e-05,
      "loss": 1.2652,
      "step": 20423
    },
    {
      "epoch": 0.7314269343026483,
      "grad_norm": 1.3746566772460938,
      "learning_rate": 3.5502822664519345e-05,
      "loss": 1.5465,
      "step": 20424
    },
    {
      "epoch": 0.7314627464322165,
      "grad_norm": 1.6845834255218506,
      "learning_rate": 3.549395905856099e-05,
      "loss": 1.3739,
      "step": 20425
    },
    {
      "epoch": 0.7314985585617849,
      "grad_norm": 1.6807371377944946,
      "learning_rate": 3.5485096320448176e-05,
      "loss": 1.6331,
      "step": 20426
    },
    {
      "epoch": 0.7315343706913532,
      "grad_norm": 1.6530333757400513,
      "learning_rate": 3.547623445030016e-05,
      "loss": 1.2784,
      "step": 20427
    },
    {
      "epoch": 0.7315701828209215,
      "grad_norm": 1.698762059211731,
      "learning_rate": 3.546737344823623e-05,
      "loss": 1.4305,
      "step": 20428
    },
    {
      "epoch": 0.7316059949504897,
      "grad_norm": 2.0598812103271484,
      "learning_rate": 3.545851331437551e-05,
      "loss": 1.3792,
      "step": 20429
    },
    {
      "epoch": 0.731641807080058,
      "grad_norm": 1.8785374164581299,
      "learning_rate": 3.544965404883728e-05,
      "loss": 1.1282,
      "step": 20430
    },
    {
      "epoch": 0.7316776192096263,
      "grad_norm": 2.2922866344451904,
      "learning_rate": 3.544079565174061e-05,
      "loss": 1.4628,
      "step": 20431
    },
    {
      "epoch": 0.7317134313391945,
      "grad_norm": 2.218919277191162,
      "learning_rate": 3.543193812320483e-05,
      "loss": 1.3955,
      "step": 20432
    },
    {
      "epoch": 0.7317492434687629,
      "grad_norm": 1.631608009338379,
      "learning_rate": 3.542308146334901e-05,
      "loss": 1.6856,
      "step": 20433
    },
    {
      "epoch": 0.7317850555983312,
      "grad_norm": 1.6857116222381592,
      "learning_rate": 3.541422567229235e-05,
      "loss": 1.4069,
      "step": 20434
    },
    {
      "epoch": 0.7318208677278994,
      "grad_norm": 1.6167864799499512,
      "learning_rate": 3.540537075015393e-05,
      "loss": 1.6345,
      "step": 20435
    },
    {
      "epoch": 0.7318566798574677,
      "grad_norm": 1.622429609298706,
      "learning_rate": 3.539651669705297e-05,
      "loss": 1.3154,
      "step": 20436
    },
    {
      "epoch": 0.731892491987036,
      "grad_norm": 1.8493098020553589,
      "learning_rate": 3.538766351310856e-05,
      "loss": 1.664,
      "step": 20437
    },
    {
      "epoch": 0.7319283041166043,
      "grad_norm": 2.0410239696502686,
      "learning_rate": 3.537881119843972e-05,
      "loss": 1.0608,
      "step": 20438
    },
    {
      "epoch": 0.7319641162461725,
      "grad_norm": 1.2448198795318604,
      "learning_rate": 3.5369959753165694e-05,
      "loss": 1.3688,
      "step": 20439
    },
    {
      "epoch": 0.7319999283757409,
      "grad_norm": 1.456103801727295,
      "learning_rate": 3.536110917740545e-05,
      "loss": 1.4517,
      "step": 20440
    },
    {
      "epoch": 0.7320357405053092,
      "grad_norm": 1.7088440656661987,
      "learning_rate": 3.5352259471278146e-05,
      "loss": 1.3145,
      "step": 20441
    },
    {
      "epoch": 0.7320715526348774,
      "grad_norm": 1.8974305391311646,
      "learning_rate": 3.534341063490273e-05,
      "loss": 1.3612,
      "step": 20442
    },
    {
      "epoch": 0.7321073647644457,
      "grad_norm": 1.5721908807754517,
      "learning_rate": 3.533456266839838e-05,
      "loss": 1.4555,
      "step": 20443
    },
    {
      "epoch": 0.732143176894014,
      "grad_norm": 1.6463236808776855,
      "learning_rate": 3.532571557188409e-05,
      "loss": 1.4655,
      "step": 20444
    },
    {
      "epoch": 0.7321789890235822,
      "grad_norm": 2.0902090072631836,
      "learning_rate": 3.531686934547884e-05,
      "loss": 1.2894,
      "step": 20445
    },
    {
      "epoch": 0.7322148011531505,
      "grad_norm": 1.4955464601516724,
      "learning_rate": 3.5308023989301676e-05,
      "loss": 1.286,
      "step": 20446
    },
    {
      "epoch": 0.7322506132827189,
      "grad_norm": 2.225278854370117,
      "learning_rate": 3.529917950347159e-05,
      "loss": 1.6805,
      "step": 20447
    },
    {
      "epoch": 0.7322864254122872,
      "grad_norm": 1.585610270500183,
      "learning_rate": 3.529033588810764e-05,
      "loss": 1.1171,
      "step": 20448
    },
    {
      "epoch": 0.7323222375418554,
      "grad_norm": 1.6811686754226685,
      "learning_rate": 3.52814931433287e-05,
      "loss": 1.5252,
      "step": 20449
    },
    {
      "epoch": 0.7323580496714237,
      "grad_norm": 2.0592916011810303,
      "learning_rate": 3.52726512692538e-05,
      "loss": 1.6063,
      "step": 20450
    },
    {
      "epoch": 0.732393861800992,
      "grad_norm": 1.6226791143417358,
      "learning_rate": 3.526381026600188e-05,
      "loss": 1.3123,
      "step": 20451
    },
    {
      "epoch": 0.7324296739305602,
      "grad_norm": 1.4268947839736938,
      "learning_rate": 3.5254970133691925e-05,
      "loss": 1.3222,
      "step": 20452
    },
    {
      "epoch": 0.7324654860601285,
      "grad_norm": 1.6242847442626953,
      "learning_rate": 3.5246130872442794e-05,
      "loss": 1.3596,
      "step": 20453
    },
    {
      "epoch": 0.7325012981896969,
      "grad_norm": 1.6318475008010864,
      "learning_rate": 3.523729248237345e-05,
      "loss": 1.2142,
      "step": 20454
    },
    {
      "epoch": 0.7325371103192652,
      "grad_norm": 1.932065486907959,
      "learning_rate": 3.522845496360283e-05,
      "loss": 1.6373,
      "step": 20455
    },
    {
      "epoch": 0.7325729224488334,
      "grad_norm": 1.579846978187561,
      "learning_rate": 3.5219618316249766e-05,
      "loss": 1.5628,
      "step": 20456
    },
    {
      "epoch": 0.7326087345784017,
      "grad_norm": 1.3308910131454468,
      "learning_rate": 3.521078254043317e-05,
      "loss": 1.6539,
      "step": 20457
    },
    {
      "epoch": 0.73264454670797,
      "grad_norm": 1.5070046186447144,
      "learning_rate": 3.5201947636271934e-05,
      "loss": 1.1408,
      "step": 20458
    },
    {
      "epoch": 0.7326803588375382,
      "grad_norm": 2.3446309566497803,
      "learning_rate": 3.519311360388494e-05,
      "loss": 1.5686,
      "step": 20459
    },
    {
      "epoch": 0.7327161709671065,
      "grad_norm": 1.3851144313812256,
      "learning_rate": 3.518428044339097e-05,
      "loss": 1.4275,
      "step": 20460
    },
    {
      "epoch": 0.7327519830966749,
      "grad_norm": 1.6674261093139648,
      "learning_rate": 3.5175448154908895e-05,
      "loss": 1.459,
      "step": 20461
    },
    {
      "epoch": 0.7327877952262432,
      "grad_norm": 1.6122231483459473,
      "learning_rate": 3.516661673855759e-05,
      "loss": 1.3823,
      "step": 20462
    },
    {
      "epoch": 0.7328236073558114,
      "grad_norm": 1.8272632360458374,
      "learning_rate": 3.51577861944558e-05,
      "loss": 1.5136,
      "step": 20463
    },
    {
      "epoch": 0.7328594194853797,
      "grad_norm": 2.42698335647583,
      "learning_rate": 3.5148956522722346e-05,
      "loss": 1.2525,
      "step": 20464
    },
    {
      "epoch": 0.732895231614948,
      "grad_norm": 1.7631263732910156,
      "learning_rate": 3.5140127723476034e-05,
      "loss": 1.351,
      "step": 20465
    },
    {
      "epoch": 0.7329310437445162,
      "grad_norm": 1.357838749885559,
      "learning_rate": 3.513129979683567e-05,
      "loss": 1.1883,
      "step": 20466
    },
    {
      "epoch": 0.7329668558740845,
      "grad_norm": 1.483927845954895,
      "learning_rate": 3.5122472742919965e-05,
      "loss": 1.3296,
      "step": 20467
    },
    {
      "epoch": 0.7330026680036529,
      "grad_norm": 2.0464913845062256,
      "learning_rate": 3.51136465618477e-05,
      "loss": 1.4515,
      "step": 20468
    },
    {
      "epoch": 0.7330384801332211,
      "grad_norm": 1.5297338962554932,
      "learning_rate": 3.510482125373762e-05,
      "loss": 1.1064,
      "step": 20469
    },
    {
      "epoch": 0.7330742922627894,
      "grad_norm": 1.6181906461715698,
      "learning_rate": 3.50959968187085e-05,
      "loss": 1.2575,
      "step": 20470
    },
    {
      "epoch": 0.7331101043923577,
      "grad_norm": 1.5216774940490723,
      "learning_rate": 3.508717325687898e-05,
      "loss": 1.234,
      "step": 20471
    },
    {
      "epoch": 0.733145916521926,
      "grad_norm": 2.048452615737915,
      "learning_rate": 3.5078350568367825e-05,
      "loss": 1.4341,
      "step": 20472
    },
    {
      "epoch": 0.7331817286514942,
      "grad_norm": 1.6940174102783203,
      "learning_rate": 3.5069528753293746e-05,
      "loss": 1.5693,
      "step": 20473
    },
    {
      "epoch": 0.7332175407810625,
      "grad_norm": 1.9572473764419556,
      "learning_rate": 3.506070781177537e-05,
      "loss": 1.419,
      "step": 20474
    },
    {
      "epoch": 0.7332533529106309,
      "grad_norm": 1.8093883991241455,
      "learning_rate": 3.505188774393141e-05,
      "loss": 1.4015,
      "step": 20475
    },
    {
      "epoch": 0.7332891650401991,
      "grad_norm": 1.1968109607696533,
      "learning_rate": 3.504306854988052e-05,
      "loss": 1.3426,
      "step": 20476
    },
    {
      "epoch": 0.7333249771697674,
      "grad_norm": 1.6046714782714844,
      "learning_rate": 3.5034250229741384e-05,
      "loss": 1.5721,
      "step": 20477
    },
    {
      "epoch": 0.7333607892993357,
      "grad_norm": 2.118436813354492,
      "learning_rate": 3.5025432783632585e-05,
      "loss": 1.56,
      "step": 20478
    },
    {
      "epoch": 0.733396601428904,
      "grad_norm": 2.097813129425049,
      "learning_rate": 3.501661621167277e-05,
      "loss": 1.2924,
      "step": 20479
    },
    {
      "epoch": 0.7334324135584722,
      "grad_norm": 1.8150711059570312,
      "learning_rate": 3.50078005139806e-05,
      "loss": 1.5607,
      "step": 20480
    },
    {
      "epoch": 0.7334682256880405,
      "grad_norm": 2.765345335006714,
      "learning_rate": 3.49989856906746e-05,
      "loss": 1.6204,
      "step": 20481
    },
    {
      "epoch": 0.7335040378176089,
      "grad_norm": 2.4202771186828613,
      "learning_rate": 3.499017174187341e-05,
      "loss": 1.797,
      "step": 20482
    },
    {
      "epoch": 0.7335398499471771,
      "grad_norm": 1.5121333599090576,
      "learning_rate": 3.498135866769561e-05,
      "loss": 1.4781,
      "step": 20483
    },
    {
      "epoch": 0.7335756620767454,
      "grad_norm": 1.4281476736068726,
      "learning_rate": 3.497254646825978e-05,
      "loss": 1.3851,
      "step": 20484
    },
    {
      "epoch": 0.7336114742063137,
      "grad_norm": 1.7604395151138306,
      "learning_rate": 3.496373514368443e-05,
      "loss": 1.3445,
      "step": 20485
    },
    {
      "epoch": 0.7336472863358819,
      "grad_norm": 1.9605915546417236,
      "learning_rate": 3.495492469408813e-05,
      "loss": 1.0695,
      "step": 20486
    },
    {
      "epoch": 0.7336830984654502,
      "grad_norm": 1.3502087593078613,
      "learning_rate": 3.494611511958942e-05,
      "loss": 1.2486,
      "step": 20487
    },
    {
      "epoch": 0.7337189105950185,
      "grad_norm": 2.571399688720703,
      "learning_rate": 3.493730642030685e-05,
      "loss": 1.5465,
      "step": 20488
    },
    {
      "epoch": 0.7337547227245869,
      "grad_norm": 1.424328327178955,
      "learning_rate": 3.492849859635885e-05,
      "loss": 1.2137,
      "step": 20489
    },
    {
      "epoch": 0.7337905348541551,
      "grad_norm": 1.9703563451766968,
      "learning_rate": 3.4919691647863984e-05,
      "loss": 1.3256,
      "step": 20490
    },
    {
      "epoch": 0.7338263469837234,
      "grad_norm": 1.6080145835876465,
      "learning_rate": 3.491088557494074e-05,
      "loss": 1.4436,
      "step": 20491
    },
    {
      "epoch": 0.7338621591132917,
      "grad_norm": 1.7678338289260864,
      "learning_rate": 3.490208037770755e-05,
      "loss": 1.6468,
      "step": 20492
    },
    {
      "epoch": 0.7338979712428599,
      "grad_norm": 1.807364583015442,
      "learning_rate": 3.4893276056282894e-05,
      "loss": 1.4296,
      "step": 20493
    },
    {
      "epoch": 0.7339337833724282,
      "grad_norm": 1.544736623764038,
      "learning_rate": 3.4884472610785224e-05,
      "loss": 1.3378,
      "step": 20494
    },
    {
      "epoch": 0.7339695955019965,
      "grad_norm": 1.670017123222351,
      "learning_rate": 3.487567004133302e-05,
      "loss": 1.5572,
      "step": 20495
    },
    {
      "epoch": 0.7340054076315649,
      "grad_norm": 1.2803508043289185,
      "learning_rate": 3.4866868348044634e-05,
      "loss": 1.5029,
      "step": 20496
    },
    {
      "epoch": 0.7340412197611331,
      "grad_norm": 1.7097656726837158,
      "learning_rate": 3.485806753103852e-05,
      "loss": 1.4671,
      "step": 20497
    },
    {
      "epoch": 0.7340770318907014,
      "grad_norm": 2.6056883335113525,
      "learning_rate": 3.484926759043311e-05,
      "loss": 1.5693,
      "step": 20498
    },
    {
      "epoch": 0.7341128440202697,
      "grad_norm": 2.137596607208252,
      "learning_rate": 3.484046852634674e-05,
      "loss": 1.4641,
      "step": 20499
    },
    {
      "epoch": 0.7341486561498379,
      "grad_norm": 1.4234815835952759,
      "learning_rate": 3.483167033889781e-05,
      "loss": 1.4199,
      "step": 20500
    },
    {
      "epoch": 0.7341844682794062,
      "grad_norm": 1.8504927158355713,
      "learning_rate": 3.4822873028204694e-05,
      "loss": 1.445,
      "step": 20501
    },
    {
      "epoch": 0.7342202804089745,
      "grad_norm": 1.4451167583465576,
      "learning_rate": 3.481407659438579e-05,
      "loss": 1.3129,
      "step": 20502
    },
    {
      "epoch": 0.7342560925385428,
      "grad_norm": 1.8816039562225342,
      "learning_rate": 3.480528103755937e-05,
      "loss": 1.5152,
      "step": 20503
    },
    {
      "epoch": 0.7342919046681111,
      "grad_norm": 1.3867018222808838,
      "learning_rate": 3.479648635784378e-05,
      "loss": 1.3284,
      "step": 20504
    },
    {
      "epoch": 0.7343277167976794,
      "grad_norm": 2.107089042663574,
      "learning_rate": 3.478769255535738e-05,
      "loss": 1.5423,
      "step": 20505
    },
    {
      "epoch": 0.7343635289272477,
      "grad_norm": 1.6862614154815674,
      "learning_rate": 3.4778899630218483e-05,
      "loss": 1.618,
      "step": 20506
    },
    {
      "epoch": 0.7343993410568159,
      "grad_norm": 2.2158303260803223,
      "learning_rate": 3.4770107582545365e-05,
      "loss": 1.2988,
      "step": 20507
    },
    {
      "epoch": 0.7344351531863842,
      "grad_norm": 1.4314731359481812,
      "learning_rate": 3.4761316412456235e-05,
      "loss": 1.4692,
      "step": 20508
    },
    {
      "epoch": 0.7344709653159525,
      "grad_norm": 1.4693208932876587,
      "learning_rate": 3.4752526120069516e-05,
      "loss": 1.2102,
      "step": 20509
    },
    {
      "epoch": 0.7345067774455208,
      "grad_norm": 1.7098275423049927,
      "learning_rate": 3.474373670550336e-05,
      "loss": 1.4572,
      "step": 20510
    },
    {
      "epoch": 0.7345425895750891,
      "grad_norm": 1.5971198081970215,
      "learning_rate": 3.4734948168876045e-05,
      "loss": 1.6368,
      "step": 20511
    },
    {
      "epoch": 0.7345784017046574,
      "grad_norm": 2.2307322025299072,
      "learning_rate": 3.4726160510305824e-05,
      "loss": 1.5739,
      "step": 20512
    },
    {
      "epoch": 0.7346142138342256,
      "grad_norm": 1.6279468536376953,
      "learning_rate": 3.471737372991095e-05,
      "loss": 1.7299,
      "step": 20513
    },
    {
      "epoch": 0.7346500259637939,
      "grad_norm": 2.171621084213257,
      "learning_rate": 3.470858782780957e-05,
      "loss": 1.7927,
      "step": 20514
    },
    {
      "epoch": 0.7346858380933622,
      "grad_norm": 1.6976094245910645,
      "learning_rate": 3.469980280411992e-05,
      "loss": 1.2166,
      "step": 20515
    },
    {
      "epoch": 0.7347216502229305,
      "grad_norm": 1.5864616632461548,
      "learning_rate": 3.469101865896023e-05,
      "loss": 1.4572,
      "step": 20516
    },
    {
      "epoch": 0.7347574623524988,
      "grad_norm": 1.2989685535430908,
      "learning_rate": 3.468223539244859e-05,
      "loss": 1.2894,
      "step": 20517
    },
    {
      "epoch": 0.7347932744820671,
      "grad_norm": 1.3882758617401123,
      "learning_rate": 3.467345300470327e-05,
      "loss": 1.3966,
      "step": 20518
    },
    {
      "epoch": 0.7348290866116354,
      "grad_norm": 1.4743343591690063,
      "learning_rate": 3.466467149584231e-05,
      "loss": 1.7203,
      "step": 20519
    },
    {
      "epoch": 0.7348648987412036,
      "grad_norm": 2.1220481395721436,
      "learning_rate": 3.4655890865983975e-05,
      "loss": 1.2081,
      "step": 20520
    },
    {
      "epoch": 0.7349007108707719,
      "grad_norm": 1.9602645635604858,
      "learning_rate": 3.464711111524631e-05,
      "loss": 1.5757,
      "step": 20521
    },
    {
      "epoch": 0.7349365230003402,
      "grad_norm": 1.7136316299438477,
      "learning_rate": 3.4638332243747464e-05,
      "loss": 1.1922,
      "step": 20522
    },
    {
      "epoch": 0.7349723351299084,
      "grad_norm": 1.697861671447754,
      "learning_rate": 3.4629554251605545e-05,
      "loss": 1.5317,
      "step": 20523
    },
    {
      "epoch": 0.7350081472594768,
      "grad_norm": 1.6139419078826904,
      "learning_rate": 3.4620777138938695e-05,
      "loss": 1.5648,
      "step": 20524
    },
    {
      "epoch": 0.7350439593890451,
      "grad_norm": 1.9879515171051025,
      "learning_rate": 3.461200090586495e-05,
      "loss": 1.5372,
      "step": 20525
    },
    {
      "epoch": 0.7350797715186134,
      "grad_norm": 1.6399869918823242,
      "learning_rate": 3.4603225552502315e-05,
      "loss": 1.3933,
      "step": 20526
    },
    {
      "epoch": 0.7351155836481816,
      "grad_norm": 1.606967568397522,
      "learning_rate": 3.4594451078969005e-05,
      "loss": 1.2946,
      "step": 20527
    },
    {
      "epoch": 0.7351513957777499,
      "grad_norm": 1.6109073162078857,
      "learning_rate": 3.458567748538295e-05,
      "loss": 1.5296,
      "step": 20528
    },
    {
      "epoch": 0.7351872079073182,
      "grad_norm": 1.4543050527572632,
      "learning_rate": 3.457690477186225e-05,
      "loss": 1.4994,
      "step": 20529
    },
    {
      "epoch": 0.7352230200368864,
      "grad_norm": 1.369817852973938,
      "learning_rate": 3.4568132938524845e-05,
      "loss": 1.2237,
      "step": 20530
    },
    {
      "epoch": 0.7352588321664548,
      "grad_norm": 1.4970976114273071,
      "learning_rate": 3.455936198548888e-05,
      "loss": 1.4621,
      "step": 20531
    },
    {
      "epoch": 0.7352946442960231,
      "grad_norm": 1.6439193487167358,
      "learning_rate": 3.455059191287225e-05,
      "loss": 1.5679,
      "step": 20532
    },
    {
      "epoch": 0.7353304564255914,
      "grad_norm": 1.7179007530212402,
      "learning_rate": 3.454182272079303e-05,
      "loss": 1.494,
      "step": 20533
    },
    {
      "epoch": 0.7353662685551596,
      "grad_norm": 1.4830936193466187,
      "learning_rate": 3.45330544093691e-05,
      "loss": 1.4933,
      "step": 20534
    },
    {
      "epoch": 0.7354020806847279,
      "grad_norm": 1.8476966619491577,
      "learning_rate": 3.4524286978718475e-05,
      "loss": 1.3555,
      "step": 20535
    },
    {
      "epoch": 0.7354378928142962,
      "grad_norm": 1.3938565254211426,
      "learning_rate": 3.451552042895916e-05,
      "loss": 1.3831,
      "step": 20536
    },
    {
      "epoch": 0.7354737049438644,
      "grad_norm": 1.6916418075561523,
      "learning_rate": 3.450675476020897e-05,
      "loss": 1.3601,
      "step": 20537
    },
    {
      "epoch": 0.7355095170734328,
      "grad_norm": 2.049785614013672,
      "learning_rate": 3.449798997258599e-05,
      "loss": 1.4742,
      "step": 20538
    },
    {
      "epoch": 0.7355453292030011,
      "grad_norm": 1.3984603881835938,
      "learning_rate": 3.4489226066208025e-05,
      "loss": 1.2174,
      "step": 20539
    },
    {
      "epoch": 0.7355811413325694,
      "grad_norm": 1.4211353063583374,
      "learning_rate": 3.448046304119306e-05,
      "loss": 1.4096,
      "step": 20540
    },
    {
      "epoch": 0.7356169534621376,
      "grad_norm": 1.744436502456665,
      "learning_rate": 3.44717008976589e-05,
      "loss": 1.4444,
      "step": 20541
    },
    {
      "epoch": 0.7356527655917059,
      "grad_norm": 1.7035998106002808,
      "learning_rate": 3.446293963572349e-05,
      "loss": 1.3967,
      "step": 20542
    },
    {
      "epoch": 0.7356885777212742,
      "grad_norm": 1.3411145210266113,
      "learning_rate": 3.4454179255504726e-05,
      "loss": 1.2792,
      "step": 20543
    },
    {
      "epoch": 0.7357243898508424,
      "grad_norm": 1.5670628547668457,
      "learning_rate": 3.44454197571204e-05,
      "loss": 1.4644,
      "step": 20544
    },
    {
      "epoch": 0.7357602019804108,
      "grad_norm": 1.505676031112671,
      "learning_rate": 3.4436661140688386e-05,
      "loss": 1.36,
      "step": 20545
    },
    {
      "epoch": 0.7357960141099791,
      "grad_norm": 1.3100101947784424,
      "learning_rate": 3.442790340632652e-05,
      "loss": 1.2666,
      "step": 20546
    },
    {
      "epoch": 0.7358318262395473,
      "grad_norm": 1.3918676376342773,
      "learning_rate": 3.441914655415268e-05,
      "loss": 1.5261,
      "step": 20547
    },
    {
      "epoch": 0.7358676383691156,
      "grad_norm": 1.698038935661316,
      "learning_rate": 3.441039058428456e-05,
      "loss": 1.4861,
      "step": 20548
    },
    {
      "epoch": 0.7359034504986839,
      "grad_norm": 1.634504795074463,
      "learning_rate": 3.440163549684009e-05,
      "loss": 1.447,
      "step": 20549
    },
    {
      "epoch": 0.7359392626282522,
      "grad_norm": 1.5148628950119019,
      "learning_rate": 3.4392881291936995e-05,
      "loss": 1.4647,
      "step": 20550
    },
    {
      "epoch": 0.7359750747578204,
      "grad_norm": 1.5409778356552124,
      "learning_rate": 3.438412796969304e-05,
      "loss": 1.6875,
      "step": 20551
    },
    {
      "epoch": 0.7360108868873888,
      "grad_norm": 1.373879075050354,
      "learning_rate": 3.4375375530225984e-05,
      "loss": 1.3546,
      "step": 20552
    },
    {
      "epoch": 0.7360466990169571,
      "grad_norm": 1.4863821268081665,
      "learning_rate": 3.436662397365361e-05,
      "loss": 1.6509,
      "step": 20553
    },
    {
      "epoch": 0.7360825111465253,
      "grad_norm": 2.397731065750122,
      "learning_rate": 3.435787330009369e-05,
      "loss": 1.1656,
      "step": 20554
    },
    {
      "epoch": 0.7361183232760936,
      "grad_norm": 1.484554409980774,
      "learning_rate": 3.4349123509663874e-05,
      "loss": 1.5836,
      "step": 20555
    },
    {
      "epoch": 0.7361541354056619,
      "grad_norm": 1.328442096710205,
      "learning_rate": 3.434037460248191e-05,
      "loss": 1.4112,
      "step": 20556
    },
    {
      "epoch": 0.7361899475352301,
      "grad_norm": 1.321702480316162,
      "learning_rate": 3.433162657866552e-05,
      "loss": 1.3672,
      "step": 20557
    },
    {
      "epoch": 0.7362257596647984,
      "grad_norm": 2.4087703227996826,
      "learning_rate": 3.4322879438332414e-05,
      "loss": 1.1557,
      "step": 20558
    },
    {
      "epoch": 0.7362615717943668,
      "grad_norm": 1.7339824438095093,
      "learning_rate": 3.431413318160022e-05,
      "loss": 1.1772,
      "step": 20559
    },
    {
      "epoch": 0.7362973839239351,
      "grad_norm": 1.9930311441421509,
      "learning_rate": 3.430538780858663e-05,
      "loss": 1.5279,
      "step": 20560
    },
    {
      "epoch": 0.7363331960535033,
      "grad_norm": 1.6439847946166992,
      "learning_rate": 3.429664331940935e-05,
      "loss": 1.3942,
      "step": 20561
    },
    {
      "epoch": 0.7363690081830716,
      "grad_norm": 1.385252594947815,
      "learning_rate": 3.4287899714185944e-05,
      "loss": 1.2338,
      "step": 20562
    },
    {
      "epoch": 0.7364048203126399,
      "grad_norm": 1.9350563287734985,
      "learning_rate": 3.427915699303408e-05,
      "loss": 1.143,
      "step": 20563
    },
    {
      "epoch": 0.7364406324422081,
      "grad_norm": 1.605703353881836,
      "learning_rate": 3.427041515607139e-05,
      "loss": 1.2759,
      "step": 20564
    },
    {
      "epoch": 0.7364764445717764,
      "grad_norm": 1.4933596849441528,
      "learning_rate": 3.426167420341552e-05,
      "loss": 1.2269,
      "step": 20565
    },
    {
      "epoch": 0.7365122567013448,
      "grad_norm": 1.7159199714660645,
      "learning_rate": 3.4252934135183977e-05,
      "loss": 1.2173,
      "step": 20566
    },
    {
      "epoch": 0.7365480688309131,
      "grad_norm": 1.7331993579864502,
      "learning_rate": 3.4244194951494414e-05,
      "loss": 1.4148,
      "step": 20567
    },
    {
      "epoch": 0.7365838809604813,
      "grad_norm": 1.2863825559616089,
      "learning_rate": 3.4235456652464405e-05,
      "loss": 1.3893,
      "step": 20568
    },
    {
      "epoch": 0.7366196930900496,
      "grad_norm": 1.5917906761169434,
      "learning_rate": 3.422671923821148e-05,
      "loss": 1.2366,
      "step": 20569
    },
    {
      "epoch": 0.7366555052196179,
      "grad_norm": 1.5418617725372314,
      "learning_rate": 3.421798270885319e-05,
      "loss": 1.2398,
      "step": 20570
    },
    {
      "epoch": 0.7366913173491861,
      "grad_norm": 1.4444352388381958,
      "learning_rate": 3.420924706450711e-05,
      "loss": 1.5176,
      "step": 20571
    },
    {
      "epoch": 0.7367271294787544,
      "grad_norm": 1.4469548463821411,
      "learning_rate": 3.4200512305290764e-05,
      "loss": 1.4863,
      "step": 20572
    },
    {
      "epoch": 0.7367629416083228,
      "grad_norm": 1.415252923965454,
      "learning_rate": 3.419177843132162e-05,
      "loss": 1.3998,
      "step": 20573
    },
    {
      "epoch": 0.736798753737891,
      "grad_norm": 2.118942975997925,
      "learning_rate": 3.418304544271721e-05,
      "loss": 1.3851,
      "step": 20574
    },
    {
      "epoch": 0.7368345658674593,
      "grad_norm": 1.5836764574050903,
      "learning_rate": 3.417431333959503e-05,
      "loss": 1.4488,
      "step": 20575
    },
    {
      "epoch": 0.7368703779970276,
      "grad_norm": 1.3505635261535645,
      "learning_rate": 3.4165582122072594e-05,
      "loss": 1.4246,
      "step": 20576
    },
    {
      "epoch": 0.7369061901265959,
      "grad_norm": 1.835097312927246,
      "learning_rate": 3.4156851790267283e-05,
      "loss": 1.3717,
      "step": 20577
    },
    {
      "epoch": 0.7369420022561641,
      "grad_norm": 1.7995364665985107,
      "learning_rate": 3.4148122344296605e-05,
      "loss": 1.7411,
      "step": 20578
    },
    {
      "epoch": 0.7369778143857324,
      "grad_norm": 1.40411376953125,
      "learning_rate": 3.413939378427804e-05,
      "loss": 1.227,
      "step": 20579
    },
    {
      "epoch": 0.7370136265153008,
      "grad_norm": 1.34280526638031,
      "learning_rate": 3.413066611032894e-05,
      "loss": 1.7086,
      "step": 20580
    },
    {
      "epoch": 0.737049438644869,
      "grad_norm": 1.728757381439209,
      "learning_rate": 3.412193932256675e-05,
      "loss": 1.5762,
      "step": 20581
    },
    {
      "epoch": 0.7370852507744373,
      "grad_norm": 1.793157935142517,
      "learning_rate": 3.41132134211089e-05,
      "loss": 1.6236,
      "step": 20582
    },
    {
      "epoch": 0.7371210629040056,
      "grad_norm": 1.6262106895446777,
      "learning_rate": 3.410448840607281e-05,
      "loss": 1.2699,
      "step": 20583
    },
    {
      "epoch": 0.7371568750335739,
      "grad_norm": 1.7049390077590942,
      "learning_rate": 3.4095764277575795e-05,
      "loss": 1.3049,
      "step": 20584
    },
    {
      "epoch": 0.7371926871631421,
      "grad_norm": 1.4077553749084473,
      "learning_rate": 3.4087041035735256e-05,
      "loss": 1.4297,
      "step": 20585
    },
    {
      "epoch": 0.7372284992927104,
      "grad_norm": 2.50809383392334,
      "learning_rate": 3.40783186806686e-05,
      "loss": 1.3544,
      "step": 20586
    },
    {
      "epoch": 0.7372643114222788,
      "grad_norm": 1.5117441415786743,
      "learning_rate": 3.406959721249309e-05,
      "loss": 1.3596,
      "step": 20587
    },
    {
      "epoch": 0.737300123551847,
      "grad_norm": 1.937361240386963,
      "learning_rate": 3.406087663132611e-05,
      "loss": 1.3741,
      "step": 20588
    },
    {
      "epoch": 0.7373359356814153,
      "grad_norm": 1.7514429092407227,
      "learning_rate": 3.4052156937284984e-05,
      "loss": 1.0765,
      "step": 20589
    },
    {
      "epoch": 0.7373717478109836,
      "grad_norm": 1.5857789516448975,
      "learning_rate": 3.404343813048705e-05,
      "loss": 1.1952,
      "step": 20590
    },
    {
      "epoch": 0.7374075599405518,
      "grad_norm": 1.6648368835449219,
      "learning_rate": 3.4034720211049544e-05,
      "loss": 1.5579,
      "step": 20591
    },
    {
      "epoch": 0.7374433720701201,
      "grad_norm": 1.51568603515625,
      "learning_rate": 3.402600317908978e-05,
      "loss": 1.0948,
      "step": 20592
    },
    {
      "epoch": 0.7374791841996884,
      "grad_norm": 2.056521415710449,
      "learning_rate": 3.401728703472505e-05,
      "loss": 1.2874,
      "step": 20593
    },
    {
      "epoch": 0.7375149963292568,
      "grad_norm": 1.7449597120285034,
      "learning_rate": 3.400857177807265e-05,
      "loss": 1.0096,
      "step": 20594
    },
    {
      "epoch": 0.737550808458825,
      "grad_norm": 3.0433919429779053,
      "learning_rate": 3.399985740924976e-05,
      "loss": 1.276,
      "step": 20595
    },
    {
      "epoch": 0.7375866205883933,
      "grad_norm": 2.3665289878845215,
      "learning_rate": 3.399114392837365e-05,
      "loss": 1.1514,
      "step": 20596
    },
    {
      "epoch": 0.7376224327179616,
      "grad_norm": 1.5721759796142578,
      "learning_rate": 3.3982431335561596e-05,
      "loss": 1.3701,
      "step": 20597
    },
    {
      "epoch": 0.7376582448475298,
      "grad_norm": 1.6225780248641968,
      "learning_rate": 3.397371963093072e-05,
      "loss": 1.0813,
      "step": 20598
    },
    {
      "epoch": 0.7376940569770981,
      "grad_norm": 1.3954616785049438,
      "learning_rate": 3.39650088145983e-05,
      "loss": 1.326,
      "step": 20599
    },
    {
      "epoch": 0.7377298691066664,
      "grad_norm": 1.427741289138794,
      "learning_rate": 3.3956298886681496e-05,
      "loss": 1.2924,
      "step": 20600
    },
    {
      "epoch": 0.7377656812362348,
      "grad_norm": 2.5152885913848877,
      "learning_rate": 3.3947589847297537e-05,
      "loss": 1.5709,
      "step": 20601
    },
    {
      "epoch": 0.737801493365803,
      "grad_norm": 1.531655192375183,
      "learning_rate": 3.393888169656351e-05,
      "loss": 1.6402,
      "step": 20602
    },
    {
      "epoch": 0.7378373054953713,
      "grad_norm": 1.998115062713623,
      "learning_rate": 3.393017443459663e-05,
      "loss": 1.1815,
      "step": 20603
    },
    {
      "epoch": 0.7378731176249396,
      "grad_norm": 1.5857163667678833,
      "learning_rate": 3.392146806151405e-05,
      "loss": 1.403,
      "step": 20604
    },
    {
      "epoch": 0.7379089297545078,
      "grad_norm": 1.4481664896011353,
      "learning_rate": 3.3912762577432864e-05,
      "loss": 1.3232,
      "step": 20605
    },
    {
      "epoch": 0.7379447418840761,
      "grad_norm": 1.4353240728378296,
      "learning_rate": 3.3904057982470204e-05,
      "loss": 1.5147,
      "step": 20606
    },
    {
      "epoch": 0.7379805540136444,
      "grad_norm": 2.2608537673950195,
      "learning_rate": 3.389535427674318e-05,
      "loss": 1.2307,
      "step": 20607
    },
    {
      "epoch": 0.7380163661432128,
      "grad_norm": 1.633110761642456,
      "learning_rate": 3.3886651460368934e-05,
      "loss": 1.3472,
      "step": 20608
    },
    {
      "epoch": 0.738052178272781,
      "grad_norm": 1.5929673910140991,
      "learning_rate": 3.3877949533464485e-05,
      "loss": 1.7282,
      "step": 20609
    },
    {
      "epoch": 0.7380879904023493,
      "grad_norm": 1.4499571323394775,
      "learning_rate": 3.3869248496146935e-05,
      "loss": 1.5401,
      "step": 20610
    },
    {
      "epoch": 0.7381238025319176,
      "grad_norm": 2.3304126262664795,
      "learning_rate": 3.3860548348533326e-05,
      "loss": 1.6311,
      "step": 20611
    },
    {
      "epoch": 0.7381596146614858,
      "grad_norm": 1.2469449043273926,
      "learning_rate": 3.385184909074077e-05,
      "loss": 1.4746,
      "step": 20612
    },
    {
      "epoch": 0.7381954267910541,
      "grad_norm": 1.6836285591125488,
      "learning_rate": 3.384315072288626e-05,
      "loss": 1.4823,
      "step": 20613
    },
    {
      "epoch": 0.7382312389206224,
      "grad_norm": 2.1215879917144775,
      "learning_rate": 3.383445324508676e-05,
      "loss": 1.3305,
      "step": 20614
    },
    {
      "epoch": 0.7382670510501907,
      "grad_norm": 1.1666394472122192,
      "learning_rate": 3.382575665745941e-05,
      "loss": 0.8717,
      "step": 20615
    },
    {
      "epoch": 0.738302863179759,
      "grad_norm": 1.4922236204147339,
      "learning_rate": 3.3817060960121105e-05,
      "loss": 1.6702,
      "step": 20616
    },
    {
      "epoch": 0.7383386753093273,
      "grad_norm": 1.494726300239563,
      "learning_rate": 3.380836615318891e-05,
      "loss": 1.3197,
      "step": 20617
    },
    {
      "epoch": 0.7383744874388956,
      "grad_norm": 1.798538327217102,
      "learning_rate": 3.37996722367797e-05,
      "loss": 1.4822,
      "step": 20618
    },
    {
      "epoch": 0.7384102995684638,
      "grad_norm": 2.2261242866516113,
      "learning_rate": 3.3790979211010576e-05,
      "loss": 1.676,
      "step": 20619
    },
    {
      "epoch": 0.7384461116980321,
      "grad_norm": 1.9397705793380737,
      "learning_rate": 3.3782287075998386e-05,
      "loss": 1.452,
      "step": 20620
    },
    {
      "epoch": 0.7384819238276004,
      "grad_norm": 2.365359306335449,
      "learning_rate": 3.377359583186012e-05,
      "loss": 1.4546,
      "step": 20621
    },
    {
      "epoch": 0.7385177359571687,
      "grad_norm": 1.9693297147750854,
      "learning_rate": 3.376490547871272e-05,
      "loss": 1.6731,
      "step": 20622
    },
    {
      "epoch": 0.738553548086737,
      "grad_norm": 1.434880018234253,
      "learning_rate": 3.375621601667305e-05,
      "loss": 1.2206,
      "step": 20623
    },
    {
      "epoch": 0.7385893602163053,
      "grad_norm": 1.6470460891723633,
      "learning_rate": 3.3747527445858074e-05,
      "loss": 1.6089,
      "step": 20624
    },
    {
      "epoch": 0.7386251723458735,
      "grad_norm": 1.7691365480422974,
      "learning_rate": 3.373883976638459e-05,
      "loss": 1.1974,
      "step": 20625
    },
    {
      "epoch": 0.7386609844754418,
      "grad_norm": 1.4882787466049194,
      "learning_rate": 3.3730152978369614e-05,
      "loss": 1.5165,
      "step": 20626
    },
    {
      "epoch": 0.7386967966050101,
      "grad_norm": 1.4077999591827393,
      "learning_rate": 3.3721467081929914e-05,
      "loss": 1.0703,
      "step": 20627
    },
    {
      "epoch": 0.7387326087345784,
      "grad_norm": 1.4659085273742676,
      "learning_rate": 3.371278207718241e-05,
      "loss": 1.2775,
      "step": 20628
    },
    {
      "epoch": 0.7387684208641467,
      "grad_norm": 1.8654396533966064,
      "learning_rate": 3.370409796424386e-05,
      "loss": 1.6343,
      "step": 20629
    },
    {
      "epoch": 0.738804232993715,
      "grad_norm": 1.9576849937438965,
      "learning_rate": 3.369541474323122e-05,
      "loss": 1.2882,
      "step": 20630
    },
    {
      "epoch": 0.7388400451232833,
      "grad_norm": 1.4226261377334595,
      "learning_rate": 3.3686732414261254e-05,
      "loss": 1.4776,
      "step": 20631
    },
    {
      "epoch": 0.7388758572528515,
      "grad_norm": 1.3740144968032837,
      "learning_rate": 3.367805097745069e-05,
      "loss": 1.4068,
      "step": 20632
    },
    {
      "epoch": 0.7389116693824198,
      "grad_norm": 1.6114263534545898,
      "learning_rate": 3.366937043291648e-05,
      "loss": 1.3795,
      "step": 20633
    },
    {
      "epoch": 0.7389474815119881,
      "grad_norm": 2.5558178424835205,
      "learning_rate": 3.3660690780775286e-05,
      "loss": 1.7663,
      "step": 20634
    },
    {
      "epoch": 0.7389832936415563,
      "grad_norm": 1.6846691370010376,
      "learning_rate": 3.3652012021143964e-05,
      "loss": 1.6188,
      "step": 20635
    },
    {
      "epoch": 0.7390191057711247,
      "grad_norm": 1.8610543012619019,
      "learning_rate": 3.364333415413917e-05,
      "loss": 1.2382,
      "step": 20636
    },
    {
      "epoch": 0.739054917900693,
      "grad_norm": 1.7064975500106812,
      "learning_rate": 3.363465717987778e-05,
      "loss": 1.6473,
      "step": 20637
    },
    {
      "epoch": 0.7390907300302613,
      "grad_norm": 1.726022720336914,
      "learning_rate": 3.3625981098476444e-05,
      "loss": 1.4555,
      "step": 20638
    },
    {
      "epoch": 0.7391265421598295,
      "grad_norm": 1.8379038572311401,
      "learning_rate": 3.3617305910051956e-05,
      "loss": 1.499,
      "step": 20639
    },
    {
      "epoch": 0.7391623542893978,
      "grad_norm": 1.7083884477615356,
      "learning_rate": 3.3608631614720955e-05,
      "loss": 1.7173,
      "step": 20640
    },
    {
      "epoch": 0.7391981664189661,
      "grad_norm": 1.3180886507034302,
      "learning_rate": 3.359995821260017e-05,
      "loss": 1.41,
      "step": 20641
    },
    {
      "epoch": 0.7392339785485343,
      "grad_norm": 1.614916443824768,
      "learning_rate": 3.359128570380633e-05,
      "loss": 1.2589,
      "step": 20642
    },
    {
      "epoch": 0.7392697906781027,
      "grad_norm": 1.9553498029708862,
      "learning_rate": 3.3582614088456055e-05,
      "loss": 1.4967,
      "step": 20643
    },
    {
      "epoch": 0.739305602807671,
      "grad_norm": 1.4962507486343384,
      "learning_rate": 3.3573943366666026e-05,
      "loss": 1.3142,
      "step": 20644
    },
    {
      "epoch": 0.7393414149372393,
      "grad_norm": 1.409155249595642,
      "learning_rate": 3.356527353855291e-05,
      "loss": 1.4818,
      "step": 20645
    },
    {
      "epoch": 0.7393772270668075,
      "grad_norm": 1.665363073348999,
      "learning_rate": 3.355660460423338e-05,
      "loss": 1.1948,
      "step": 20646
    },
    {
      "epoch": 0.7394130391963758,
      "grad_norm": 1.674485445022583,
      "learning_rate": 3.354793656382399e-05,
      "loss": 1.4648,
      "step": 20647
    },
    {
      "epoch": 0.7394488513259441,
      "grad_norm": 1.315306305885315,
      "learning_rate": 3.35392694174414e-05,
      "loss": 1.4364,
      "step": 20648
    },
    {
      "epoch": 0.7394846634555123,
      "grad_norm": 1.6919763088226318,
      "learning_rate": 3.3530603165202245e-05,
      "loss": 1.3387,
      "step": 20649
    },
    {
      "epoch": 0.7395204755850807,
      "grad_norm": 1.8464726209640503,
      "learning_rate": 3.352193780722306e-05,
      "loss": 1.4225,
      "step": 20650
    },
    {
      "epoch": 0.739556287714649,
      "grad_norm": 1.7736725807189941,
      "learning_rate": 3.351327334362043e-05,
      "loss": 1.352,
      "step": 20651
    },
    {
      "epoch": 0.7395920998442173,
      "grad_norm": 1.6074994802474976,
      "learning_rate": 3.3504609774510964e-05,
      "loss": 1.3402,
      "step": 20652
    },
    {
      "epoch": 0.7396279119737855,
      "grad_norm": 1.3930764198303223,
      "learning_rate": 3.349594710001123e-05,
      "loss": 1.3765,
      "step": 20653
    },
    {
      "epoch": 0.7396637241033538,
      "grad_norm": 1.6473559141159058,
      "learning_rate": 3.3487285320237705e-05,
      "loss": 1.2232,
      "step": 20654
    },
    {
      "epoch": 0.7396995362329221,
      "grad_norm": 1.8392513990402222,
      "learning_rate": 3.347862443530697e-05,
      "loss": 1.4693,
      "step": 20655
    },
    {
      "epoch": 0.7397353483624903,
      "grad_norm": 1.4013627767562866,
      "learning_rate": 3.3469964445335566e-05,
      "loss": 1.1409,
      "step": 20656
    },
    {
      "epoch": 0.7397711604920587,
      "grad_norm": 1.6355818510055542,
      "learning_rate": 3.346130535043993e-05,
      "loss": 1.2311,
      "step": 20657
    },
    {
      "epoch": 0.739806972621627,
      "grad_norm": 1.5286083221435547,
      "learning_rate": 3.3452647150736615e-05,
      "loss": 1.3021,
      "step": 20658
    },
    {
      "epoch": 0.7398427847511952,
      "grad_norm": 1.5414046049118042,
      "learning_rate": 3.3443989846342084e-05,
      "loss": 1.1697,
      "step": 20659
    },
    {
      "epoch": 0.7398785968807635,
      "grad_norm": 1.3940845727920532,
      "learning_rate": 3.3435333437372854e-05,
      "loss": 1.5052,
      "step": 20660
    },
    {
      "epoch": 0.7399144090103318,
      "grad_norm": 1.6988524198532104,
      "learning_rate": 3.3426677923945314e-05,
      "loss": 1.4081,
      "step": 20661
    },
    {
      "epoch": 0.7399502211399,
      "grad_norm": 1.670135498046875,
      "learning_rate": 3.341802330617596e-05,
      "loss": 1.4241,
      "step": 20662
    },
    {
      "epoch": 0.7399860332694683,
      "grad_norm": 1.8642773628234863,
      "learning_rate": 3.3409369584181216e-05,
      "loss": 1.4035,
      "step": 20663
    },
    {
      "epoch": 0.7400218453990367,
      "grad_norm": 1.446311593055725,
      "learning_rate": 3.340071675807753e-05,
      "loss": 1.6849,
      "step": 20664
    },
    {
      "epoch": 0.740057657528605,
      "grad_norm": 1.5131405591964722,
      "learning_rate": 3.3392064827981275e-05,
      "loss": 1.6785,
      "step": 20665
    },
    {
      "epoch": 0.7400934696581732,
      "grad_norm": 2.638115406036377,
      "learning_rate": 3.338341379400885e-05,
      "loss": 1.3795,
      "step": 20666
    },
    {
      "epoch": 0.7401292817877415,
      "grad_norm": 1.5250235795974731,
      "learning_rate": 3.337476365627672e-05,
      "loss": 1.421,
      "step": 20667
    },
    {
      "epoch": 0.7401650939173098,
      "grad_norm": 1.3569045066833496,
      "learning_rate": 3.336611441490115e-05,
      "loss": 1.533,
      "step": 20668
    },
    {
      "epoch": 0.740200906046878,
      "grad_norm": 1.659124493598938,
      "learning_rate": 3.335746606999858e-05,
      "loss": 1.3057,
      "step": 20669
    },
    {
      "epoch": 0.7402367181764463,
      "grad_norm": 1.662147045135498,
      "learning_rate": 3.334881862168532e-05,
      "loss": 1.3973,
      "step": 20670
    },
    {
      "epoch": 0.7402725303060147,
      "grad_norm": 1.852306604385376,
      "learning_rate": 3.334017207007778e-05,
      "loss": 1.5306,
      "step": 20671
    },
    {
      "epoch": 0.740308342435583,
      "grad_norm": 1.557503581047058,
      "learning_rate": 3.33315264152922e-05,
      "loss": 1.2448,
      "step": 20672
    },
    {
      "epoch": 0.7403441545651512,
      "grad_norm": 2.1327314376831055,
      "learning_rate": 3.332288165744494e-05,
      "loss": 1.3542,
      "step": 20673
    },
    {
      "epoch": 0.7403799666947195,
      "grad_norm": 1.6384520530700684,
      "learning_rate": 3.3314237796652324e-05,
      "loss": 1.7015,
      "step": 20674
    },
    {
      "epoch": 0.7404157788242878,
      "grad_norm": 1.7940348386764526,
      "learning_rate": 3.33055948330306e-05,
      "loss": 1.4875,
      "step": 20675
    },
    {
      "epoch": 0.740451590953856,
      "grad_norm": 2.3545427322387695,
      "learning_rate": 3.329695276669605e-05,
      "loss": 1.3333,
      "step": 20676
    },
    {
      "epoch": 0.7404874030834243,
      "grad_norm": 1.4605990648269653,
      "learning_rate": 3.3288311597764976e-05,
      "loss": 1.713,
      "step": 20677
    },
    {
      "epoch": 0.7405232152129927,
      "grad_norm": 2.598592758178711,
      "learning_rate": 3.327967132635364e-05,
      "loss": 1.4987,
      "step": 20678
    },
    {
      "epoch": 0.740559027342561,
      "grad_norm": 1.3932163715362549,
      "learning_rate": 3.3271031952578245e-05,
      "loss": 1.2992,
      "step": 20679
    },
    {
      "epoch": 0.7405948394721292,
      "grad_norm": 2.0070300102233887,
      "learning_rate": 3.326239347655503e-05,
      "loss": 1.5381,
      "step": 20680
    },
    {
      "epoch": 0.7406306516016975,
      "grad_norm": 1.4914416074752808,
      "learning_rate": 3.325375589840023e-05,
      "loss": 1.4634,
      "step": 20681
    },
    {
      "epoch": 0.7406664637312658,
      "grad_norm": 1.81553053855896,
      "learning_rate": 3.3245119218230066e-05,
      "loss": 1.3835,
      "step": 20682
    },
    {
      "epoch": 0.740702275860834,
      "grad_norm": 1.629849910736084,
      "learning_rate": 3.32364834361607e-05,
      "loss": 1.4646,
      "step": 20683
    },
    {
      "epoch": 0.7407380879904023,
      "grad_norm": 1.6208226680755615,
      "learning_rate": 3.3227848552308326e-05,
      "loss": 1.4159,
      "step": 20684
    },
    {
      "epoch": 0.7407739001199707,
      "grad_norm": 1.9516637325286865,
      "learning_rate": 3.321921456678915e-05,
      "loss": 1.3905,
      "step": 20685
    },
    {
      "epoch": 0.740809712249539,
      "grad_norm": 1.5003976821899414,
      "learning_rate": 3.321058147971927e-05,
      "loss": 1.7165,
      "step": 20686
    },
    {
      "epoch": 0.7408455243791072,
      "grad_norm": 1.5799978971481323,
      "learning_rate": 3.320194929121486e-05,
      "loss": 1.6016,
      "step": 20687
    },
    {
      "epoch": 0.7408813365086755,
      "grad_norm": 1.946989893913269,
      "learning_rate": 3.319331800139207e-05,
      "loss": 1.8459,
      "step": 20688
    },
    {
      "epoch": 0.7409171486382438,
      "grad_norm": 1.4369703531265259,
      "learning_rate": 3.318468761036704e-05,
      "loss": 1.5164,
      "step": 20689
    },
    {
      "epoch": 0.740952960767812,
      "grad_norm": 1.370896339416504,
      "learning_rate": 3.3176058118255816e-05,
      "loss": 1.4781,
      "step": 20690
    },
    {
      "epoch": 0.7409887728973803,
      "grad_norm": 1.3151863813400269,
      "learning_rate": 3.316742952517453e-05,
      "loss": 1.3523,
      "step": 20691
    },
    {
      "epoch": 0.7410245850269487,
      "grad_norm": 1.8912273645401,
      "learning_rate": 3.3158801831239314e-05,
      "loss": 1.345,
      "step": 20692
    },
    {
      "epoch": 0.741060397156517,
      "grad_norm": 1.3911374807357788,
      "learning_rate": 3.3150175036566166e-05,
      "loss": 1.6394,
      "step": 20693
    },
    {
      "epoch": 0.7410962092860852,
      "grad_norm": 1.4584274291992188,
      "learning_rate": 3.314154914127118e-05,
      "loss": 1.3457,
      "step": 20694
    },
    {
      "epoch": 0.7411320214156535,
      "grad_norm": 1.4885680675506592,
      "learning_rate": 3.31329241454704e-05,
      "loss": 1.709,
      "step": 20695
    },
    {
      "epoch": 0.7411678335452218,
      "grad_norm": 2.0191259384155273,
      "learning_rate": 3.312430004927992e-05,
      "loss": 1.3875,
      "step": 20696
    },
    {
      "epoch": 0.74120364567479,
      "grad_norm": 2.2317590713500977,
      "learning_rate": 3.311567685281568e-05,
      "loss": 1.6509,
      "step": 20697
    },
    {
      "epoch": 0.7412394578043583,
      "grad_norm": 1.3376753330230713,
      "learning_rate": 3.310705455619374e-05,
      "loss": 1.3804,
      "step": 20698
    },
    {
      "epoch": 0.7412752699339267,
      "grad_norm": 1.5647464990615845,
      "learning_rate": 3.309843315953008e-05,
      "loss": 1.2623,
      "step": 20699
    },
    {
      "epoch": 0.7413110820634949,
      "grad_norm": 1.6733916997909546,
      "learning_rate": 3.3089812662940754e-05,
      "loss": 1.1969,
      "step": 20700
    },
    {
      "epoch": 0.7413468941930632,
      "grad_norm": 1.6514251232147217,
      "learning_rate": 3.308119306654168e-05,
      "loss": 1.2187,
      "step": 20701
    },
    {
      "epoch": 0.7413827063226315,
      "grad_norm": 1.6864891052246094,
      "learning_rate": 3.3072574370448783e-05,
      "loss": 1.5163,
      "step": 20702
    },
    {
      "epoch": 0.7414185184521997,
      "grad_norm": 1.344460129737854,
      "learning_rate": 3.306395657477812e-05,
      "loss": 1.4425,
      "step": 20703
    },
    {
      "epoch": 0.741454330581768,
      "grad_norm": 1.5931432247161865,
      "learning_rate": 3.3055339679645544e-05,
      "loss": 1.5247,
      "step": 20704
    },
    {
      "epoch": 0.7414901427113363,
      "grad_norm": 2.037649631500244,
      "learning_rate": 3.304672368516704e-05,
      "loss": 1.5746,
      "step": 20705
    },
    {
      "epoch": 0.7415259548409047,
      "grad_norm": 1.2448827028274536,
      "learning_rate": 3.303810859145848e-05,
      "loss": 1.4724,
      "step": 20706
    },
    {
      "epoch": 0.7415617669704729,
      "grad_norm": 1.5991346836090088,
      "learning_rate": 3.302949439863584e-05,
      "loss": 1.6548,
      "step": 20707
    },
    {
      "epoch": 0.7415975791000412,
      "grad_norm": 1.6858115196228027,
      "learning_rate": 3.3020881106814936e-05,
      "loss": 1.6506,
      "step": 20708
    },
    {
      "epoch": 0.7416333912296095,
      "grad_norm": 1.7268314361572266,
      "learning_rate": 3.301226871611168e-05,
      "loss": 1.278,
      "step": 20709
    },
    {
      "epoch": 0.7416692033591777,
      "grad_norm": 1.3854961395263672,
      "learning_rate": 3.3003657226641974e-05,
      "loss": 1.3806,
      "step": 20710
    },
    {
      "epoch": 0.741705015488746,
      "grad_norm": 1.8541377782821655,
      "learning_rate": 3.2995046638521595e-05,
      "loss": 1.0329,
      "step": 20711
    },
    {
      "epoch": 0.7417408276183143,
      "grad_norm": 1.6589943170547485,
      "learning_rate": 3.2986436951866486e-05,
      "loss": 1.3059,
      "step": 20712
    },
    {
      "epoch": 0.7417766397478827,
      "grad_norm": 1.4660264253616333,
      "learning_rate": 3.2977828166792345e-05,
      "loss": 1.7456,
      "step": 20713
    },
    {
      "epoch": 0.7418124518774509,
      "grad_norm": 1.5078004598617554,
      "learning_rate": 3.296922028341515e-05,
      "loss": 1.3345,
      "step": 20714
    },
    {
      "epoch": 0.7418482640070192,
      "grad_norm": 1.6876119375228882,
      "learning_rate": 3.29606133018506e-05,
      "loss": 1.5791,
      "step": 20715
    },
    {
      "epoch": 0.7418840761365875,
      "grad_norm": 1.7531828880310059,
      "learning_rate": 3.2952007222214545e-05,
      "loss": 1.1627,
      "step": 20716
    },
    {
      "epoch": 0.7419198882661557,
      "grad_norm": 3.027247667312622,
      "learning_rate": 3.29434020446227e-05,
      "loss": 1.1017,
      "step": 20717
    },
    {
      "epoch": 0.741955700395724,
      "grad_norm": 1.8014774322509766,
      "learning_rate": 3.293479776919093e-05,
      "loss": 1.4673,
      "step": 20718
    },
    {
      "epoch": 0.7419915125252923,
      "grad_norm": 1.9687037467956543,
      "learning_rate": 3.292619439603495e-05,
      "loss": 1.5301,
      "step": 20719
    },
    {
      "epoch": 0.7420273246548607,
      "grad_norm": 1.6570146083831787,
      "learning_rate": 3.291759192527045e-05,
      "loss": 1.5799,
      "step": 20720
    },
    {
      "epoch": 0.7420631367844289,
      "grad_norm": 1.3398667573928833,
      "learning_rate": 3.290899035701328e-05,
      "loss": 1.5945,
      "step": 20721
    },
    {
      "epoch": 0.7420989489139972,
      "grad_norm": 1.6556895971298218,
      "learning_rate": 3.2900389691379074e-05,
      "loss": 1.3598,
      "step": 20722
    },
    {
      "epoch": 0.7421347610435655,
      "grad_norm": 1.4631439447402954,
      "learning_rate": 3.2891789928483594e-05,
      "loss": 1.1738,
      "step": 20723
    },
    {
      "epoch": 0.7421705731731337,
      "grad_norm": 1.3914145231246948,
      "learning_rate": 3.2883191068442464e-05,
      "loss": 1.4977,
      "step": 20724
    },
    {
      "epoch": 0.742206385302702,
      "grad_norm": 1.4877334833145142,
      "learning_rate": 3.287459311137149e-05,
      "loss": 1.7173,
      "step": 20725
    },
    {
      "epoch": 0.7422421974322703,
      "grad_norm": 1.5080418586730957,
      "learning_rate": 3.286599605738624e-05,
      "loss": 1.361,
      "step": 20726
    },
    {
      "epoch": 0.7422780095618386,
      "grad_norm": 1.6488600969314575,
      "learning_rate": 3.285739990660246e-05,
      "loss": 1.0568,
      "step": 20727
    },
    {
      "epoch": 0.7423138216914069,
      "grad_norm": 1.9895234107971191,
      "learning_rate": 3.284880465913571e-05,
      "loss": 1.6501,
      "step": 20728
    },
    {
      "epoch": 0.7423496338209752,
      "grad_norm": 1.7913322448730469,
      "learning_rate": 3.284021031510168e-05,
      "loss": 1.395,
      "step": 20729
    },
    {
      "epoch": 0.7423854459505435,
      "grad_norm": 1.5944781303405762,
      "learning_rate": 3.2831616874616036e-05,
      "loss": 1.5749,
      "step": 20730
    },
    {
      "epoch": 0.7424212580801117,
      "grad_norm": 1.548054814338684,
      "learning_rate": 3.282302433779426e-05,
      "loss": 1.6684,
      "step": 20731
    },
    {
      "epoch": 0.74245707020968,
      "grad_norm": 1.558003306388855,
      "learning_rate": 3.281443270475212e-05,
      "loss": 1.3898,
      "step": 20732
    },
    {
      "epoch": 0.7424928823392483,
      "grad_norm": 1.7061145305633545,
      "learning_rate": 3.280584197560508e-05,
      "loss": 1.7347,
      "step": 20733
    },
    {
      "epoch": 0.7425286944688166,
      "grad_norm": 1.6126638650894165,
      "learning_rate": 3.2797252150468804e-05,
      "loss": 1.5521,
      "step": 20734
    },
    {
      "epoch": 0.7425645065983849,
      "grad_norm": 1.6583861112594604,
      "learning_rate": 3.278866322945874e-05,
      "loss": 1.4871,
      "step": 20735
    },
    {
      "epoch": 0.7426003187279532,
      "grad_norm": 1.5712144374847412,
      "learning_rate": 3.278007521269059e-05,
      "loss": 1.5044,
      "step": 20736
    },
    {
      "epoch": 0.7426361308575214,
      "grad_norm": 2.0407989025115967,
      "learning_rate": 3.2771488100279814e-05,
      "loss": 1.4875,
      "step": 20737
    },
    {
      "epoch": 0.7426719429870897,
      "grad_norm": 1.7771286964416504,
      "learning_rate": 3.2762901892341926e-05,
      "loss": 1.2809,
      "step": 20738
    },
    {
      "epoch": 0.742707755116658,
      "grad_norm": 2.0323855876922607,
      "learning_rate": 3.2754316588992454e-05,
      "loss": 1.4569,
      "step": 20739
    },
    {
      "epoch": 0.7427435672462263,
      "grad_norm": 1.6384506225585938,
      "learning_rate": 3.274573219034691e-05,
      "loss": 1.7623,
      "step": 20740
    },
    {
      "epoch": 0.7427793793757946,
      "grad_norm": 1.7023667097091675,
      "learning_rate": 3.2737148696520824e-05,
      "loss": 1.3912,
      "step": 20741
    },
    {
      "epoch": 0.7428151915053629,
      "grad_norm": 1.7247306108474731,
      "learning_rate": 3.272856610762961e-05,
      "loss": 1.496,
      "step": 20742
    },
    {
      "epoch": 0.7428510036349312,
      "grad_norm": 1.7189574241638184,
      "learning_rate": 3.271998442378875e-05,
      "loss": 1.613,
      "step": 20743
    },
    {
      "epoch": 0.7428868157644994,
      "grad_norm": 2.2422940731048584,
      "learning_rate": 3.271140364511377e-05,
      "loss": 1.7311,
      "step": 20744
    },
    {
      "epoch": 0.7429226278940677,
      "grad_norm": 1.8080083131790161,
      "learning_rate": 3.270282377172001e-05,
      "loss": 1.3932,
      "step": 20745
    },
    {
      "epoch": 0.742958440023636,
      "grad_norm": 1.682898759841919,
      "learning_rate": 3.269424480372295e-05,
      "loss": 1.6761,
      "step": 20746
    },
    {
      "epoch": 0.7429942521532042,
      "grad_norm": 1.3644120693206787,
      "learning_rate": 3.268566674123802e-05,
      "loss": 1.4248,
      "step": 20747
    },
    {
      "epoch": 0.7430300642827726,
      "grad_norm": 2.0592284202575684,
      "learning_rate": 3.267708958438063e-05,
      "loss": 1.3517,
      "step": 20748
    },
    {
      "epoch": 0.7430658764123409,
      "grad_norm": 1.3781828880310059,
      "learning_rate": 3.266851333326614e-05,
      "loss": 1.4983,
      "step": 20749
    },
    {
      "epoch": 0.7431016885419092,
      "grad_norm": 1.7163243293762207,
      "learning_rate": 3.265993798800995e-05,
      "loss": 1.0821,
      "step": 20750
    },
    {
      "epoch": 0.7431375006714774,
      "grad_norm": 1.565303087234497,
      "learning_rate": 3.265136354872742e-05,
      "loss": 1.3198,
      "step": 20751
    },
    {
      "epoch": 0.7431733128010457,
      "grad_norm": 1.516454815864563,
      "learning_rate": 3.2642790015533965e-05,
      "loss": 1.1464,
      "step": 20752
    },
    {
      "epoch": 0.743209124930614,
      "grad_norm": 1.5399000644683838,
      "learning_rate": 3.2634217388544855e-05,
      "loss": 1.6111,
      "step": 20753
    },
    {
      "epoch": 0.7432449370601822,
      "grad_norm": 1.608729600906372,
      "learning_rate": 3.2625645667875434e-05,
      "loss": 1.4164,
      "step": 20754
    },
    {
      "epoch": 0.7432807491897506,
      "grad_norm": 1.7107634544372559,
      "learning_rate": 3.26170748536411e-05,
      "loss": 1.4349,
      "step": 20755
    },
    {
      "epoch": 0.7433165613193189,
      "grad_norm": 1.5795230865478516,
      "learning_rate": 3.260850494595707e-05,
      "loss": 1.3175,
      "step": 20756
    },
    {
      "epoch": 0.7433523734488872,
      "grad_norm": 1.8577889204025269,
      "learning_rate": 3.259993594493866e-05,
      "loss": 1.6718,
      "step": 20757
    },
    {
      "epoch": 0.7433881855784554,
      "grad_norm": 1.8486729860305786,
      "learning_rate": 3.2591367850701194e-05,
      "loss": 1.424,
      "step": 20758
    },
    {
      "epoch": 0.7434239977080237,
      "grad_norm": 1.7383266687393188,
      "learning_rate": 3.2582800663359933e-05,
      "loss": 1.584,
      "step": 20759
    },
    {
      "epoch": 0.743459809837592,
      "grad_norm": 1.9045100212097168,
      "learning_rate": 3.257423438303011e-05,
      "loss": 1.4333,
      "step": 20760
    },
    {
      "epoch": 0.7434956219671602,
      "grad_norm": 1.3774325847625732,
      "learning_rate": 3.256566900982699e-05,
      "loss": 1.4165,
      "step": 20761
    },
    {
      "epoch": 0.7435314340967286,
      "grad_norm": 1.4504196643829346,
      "learning_rate": 3.255710454386585e-05,
      "loss": 1.3648,
      "step": 20762
    },
    {
      "epoch": 0.7435672462262969,
      "grad_norm": 1.3335399627685547,
      "learning_rate": 3.2548540985261824e-05,
      "loss": 1.5259,
      "step": 20763
    },
    {
      "epoch": 0.7436030583558652,
      "grad_norm": 1.8252145051956177,
      "learning_rate": 3.2539978334130174e-05,
      "loss": 1.2774,
      "step": 20764
    },
    {
      "epoch": 0.7436388704854334,
      "grad_norm": 1.58820378780365,
      "learning_rate": 3.253141659058611e-05,
      "loss": 1.4351,
      "step": 20765
    },
    {
      "epoch": 0.7436746826150017,
      "grad_norm": 1.7370541095733643,
      "learning_rate": 3.252285575474483e-05,
      "loss": 1.6168,
      "step": 20766
    },
    {
      "epoch": 0.74371049474457,
      "grad_norm": 2.1864709854125977,
      "learning_rate": 3.251429582672145e-05,
      "loss": 2.0227,
      "step": 20767
    },
    {
      "epoch": 0.7437463068741382,
      "grad_norm": 2.4286386966705322,
      "learning_rate": 3.2505736806631185e-05,
      "loss": 1.8233,
      "step": 20768
    },
    {
      "epoch": 0.7437821190037066,
      "grad_norm": 1.746517300605774,
      "learning_rate": 3.249717869458916e-05,
      "loss": 1.2833,
      "step": 20769
    },
    {
      "epoch": 0.7438179311332749,
      "grad_norm": 1.7618461847305298,
      "learning_rate": 3.248862149071056e-05,
      "loss": 1.5458,
      "step": 20770
    },
    {
      "epoch": 0.7438537432628431,
      "grad_norm": 1.8682754039764404,
      "learning_rate": 3.248006519511043e-05,
      "loss": 1.6408,
      "step": 20771
    },
    {
      "epoch": 0.7438895553924114,
      "grad_norm": 1.6848514080047607,
      "learning_rate": 3.247150980790394e-05,
      "loss": 1.5804,
      "step": 20772
    },
    {
      "epoch": 0.7439253675219797,
      "grad_norm": 1.2802892923355103,
      "learning_rate": 3.2462955329206213e-05,
      "loss": 1.3587,
      "step": 20773
    },
    {
      "epoch": 0.743961179651548,
      "grad_norm": 1.8104581832885742,
      "learning_rate": 3.245440175913227e-05,
      "loss": 1.4791,
      "step": 20774
    },
    {
      "epoch": 0.7439969917811162,
      "grad_norm": 1.7056621313095093,
      "learning_rate": 3.244584909779722e-05,
      "loss": 1.6028,
      "step": 20775
    },
    {
      "epoch": 0.7440328039106846,
      "grad_norm": 2.1532299518585205,
      "learning_rate": 3.243729734531614e-05,
      "loss": 1.3802,
      "step": 20776
    },
    {
      "epoch": 0.7440686160402529,
      "grad_norm": 1.530900239944458,
      "learning_rate": 3.2428746501804106e-05,
      "loss": 1.5308,
      "step": 20777
    },
    {
      "epoch": 0.7441044281698211,
      "grad_norm": 1.56569242477417,
      "learning_rate": 3.2420196567376096e-05,
      "loss": 1.538,
      "step": 20778
    },
    {
      "epoch": 0.7441402402993894,
      "grad_norm": 1.6894400119781494,
      "learning_rate": 3.241164754214716e-05,
      "loss": 1.3194,
      "step": 20779
    },
    {
      "epoch": 0.7441760524289577,
      "grad_norm": 2.2958922386169434,
      "learning_rate": 3.2403099426232365e-05,
      "loss": 1.4457,
      "step": 20780
    },
    {
      "epoch": 0.744211864558526,
      "grad_norm": 1.536910057067871,
      "learning_rate": 3.239455221974663e-05,
      "loss": 0.9383,
      "step": 20781
    },
    {
      "epoch": 0.7442476766880942,
      "grad_norm": 1.4549648761749268,
      "learning_rate": 3.2386005922804996e-05,
      "loss": 1.4439,
      "step": 20782
    },
    {
      "epoch": 0.7442834888176626,
      "grad_norm": 1.574837327003479,
      "learning_rate": 3.237746053552244e-05,
      "loss": 1.3761,
      "step": 20783
    },
    {
      "epoch": 0.7443193009472309,
      "grad_norm": 1.8946826457977295,
      "learning_rate": 3.2368916058013956e-05,
      "loss": 1.4469,
      "step": 20784
    },
    {
      "epoch": 0.7443551130767991,
      "grad_norm": 2.358954906463623,
      "learning_rate": 3.236037249039444e-05,
      "loss": 1.6472,
      "step": 20785
    },
    {
      "epoch": 0.7443909252063674,
      "grad_norm": 1.663940668106079,
      "learning_rate": 3.235182983277886e-05,
      "loss": 1.4214,
      "step": 20786
    },
    {
      "epoch": 0.7444267373359357,
      "grad_norm": 1.7048293352127075,
      "learning_rate": 3.234328808528215e-05,
      "loss": 1.4342,
      "step": 20787
    },
    {
      "epoch": 0.7444625494655039,
      "grad_norm": 1.6530091762542725,
      "learning_rate": 3.233474724801926e-05,
      "loss": 1.5719,
      "step": 20788
    },
    {
      "epoch": 0.7444983615950722,
      "grad_norm": 1.466918706893921,
      "learning_rate": 3.232620732110503e-05,
      "loss": 1.3972,
      "step": 20789
    },
    {
      "epoch": 0.7445341737246406,
      "grad_norm": 1.7518681287765503,
      "learning_rate": 3.231766830465439e-05,
      "loss": 1.5575,
      "step": 20790
    },
    {
      "epoch": 0.7445699858542089,
      "grad_norm": 1.4471641778945923,
      "learning_rate": 3.230913019878224e-05,
      "loss": 1.448,
      "step": 20791
    },
    {
      "epoch": 0.7446057979837771,
      "grad_norm": 1.9745919704437256,
      "learning_rate": 3.230059300360342e-05,
      "loss": 1.4811,
      "step": 20792
    },
    {
      "epoch": 0.7446416101133454,
      "grad_norm": 1.7885055541992188,
      "learning_rate": 3.229205671923278e-05,
      "loss": 1.5635,
      "step": 20793
    },
    {
      "epoch": 0.7446774222429137,
      "grad_norm": 1.5172529220581055,
      "learning_rate": 3.2283521345785176e-05,
      "loss": 1.5204,
      "step": 20794
    },
    {
      "epoch": 0.7447132343724819,
      "grad_norm": 1.5513358116149902,
      "learning_rate": 3.227498688337548e-05,
      "loss": 1.4141,
      "step": 20795
    },
    {
      "epoch": 0.7447490465020502,
      "grad_norm": 1.9045683145523071,
      "learning_rate": 3.226645333211845e-05,
      "loss": 1.4086,
      "step": 20796
    },
    {
      "epoch": 0.7447848586316185,
      "grad_norm": 1.5256716012954712,
      "learning_rate": 3.225792069212892e-05,
      "loss": 1.5432,
      "step": 20797
    },
    {
      "epoch": 0.7448206707611869,
      "grad_norm": 1.4963194131851196,
      "learning_rate": 3.224938896352171e-05,
      "loss": 1.1279,
      "step": 20798
    },
    {
      "epoch": 0.7448564828907551,
      "grad_norm": 1.3626344203948975,
      "learning_rate": 3.2240858146411546e-05,
      "loss": 1.1153,
      "step": 20799
    },
    {
      "epoch": 0.7448922950203234,
      "grad_norm": 1.44807767868042,
      "learning_rate": 3.2232328240913277e-05,
      "loss": 1.3652,
      "step": 20800
    },
    {
      "epoch": 0.7449281071498917,
      "grad_norm": 2.269850492477417,
      "learning_rate": 3.222379924714155e-05,
      "loss": 1.1314,
      "step": 20801
    },
    {
      "epoch": 0.7449639192794599,
      "grad_norm": 1.7084736824035645,
      "learning_rate": 3.221527116521124e-05,
      "loss": 1.5226,
      "step": 20802
    },
    {
      "epoch": 0.7449997314090282,
      "grad_norm": 2.213249444961548,
      "learning_rate": 3.220674399523699e-05,
      "loss": 1.3941,
      "step": 20803
    },
    {
      "epoch": 0.7450355435385965,
      "grad_norm": 1.5036464929580688,
      "learning_rate": 3.219821773733355e-05,
      "loss": 1.5555,
      "step": 20804
    },
    {
      "epoch": 0.7450713556681648,
      "grad_norm": 1.3128011226654053,
      "learning_rate": 3.218969239161563e-05,
      "loss": 1.3888,
      "step": 20805
    },
    {
      "epoch": 0.7451071677977331,
      "grad_norm": 1.5558034181594849,
      "learning_rate": 3.2181167958197964e-05,
      "loss": 1.2282,
      "step": 20806
    },
    {
      "epoch": 0.7451429799273014,
      "grad_norm": 2.2121920585632324,
      "learning_rate": 3.2172644437195207e-05,
      "loss": 1.733,
      "step": 20807
    },
    {
      "epoch": 0.7451787920568697,
      "grad_norm": 1.9016462564468384,
      "learning_rate": 3.216412182872196e-05,
      "loss": 1.4272,
      "step": 20808
    },
    {
      "epoch": 0.7452146041864379,
      "grad_norm": 1.629558801651001,
      "learning_rate": 3.215560013289301e-05,
      "loss": 1.5492,
      "step": 20809
    },
    {
      "epoch": 0.7452504163160062,
      "grad_norm": 3.6835665702819824,
      "learning_rate": 3.2147079349822925e-05,
      "loss": 1.5285,
      "step": 20810
    },
    {
      "epoch": 0.7452862284455745,
      "grad_norm": 1.4725559949874878,
      "learning_rate": 3.2138559479626395e-05,
      "loss": 1.4868,
      "step": 20811
    },
    {
      "epoch": 0.7453220405751428,
      "grad_norm": 1.4919462203979492,
      "learning_rate": 3.2130040522417946e-05,
      "loss": 1.5146,
      "step": 20812
    },
    {
      "epoch": 0.7453578527047111,
      "grad_norm": 1.767691731452942,
      "learning_rate": 3.212152247831233e-05,
      "loss": 1.2523,
      "step": 20813
    },
    {
      "epoch": 0.7453936648342794,
      "grad_norm": 1.4259493350982666,
      "learning_rate": 3.211300534742402e-05,
      "loss": 1.7543,
      "step": 20814
    },
    {
      "epoch": 0.7454294769638476,
      "grad_norm": 1.6164785623550415,
      "learning_rate": 3.210448912986767e-05,
      "loss": 1.5031,
      "step": 20815
    },
    {
      "epoch": 0.7454652890934159,
      "grad_norm": 1.4968827962875366,
      "learning_rate": 3.209597382575786e-05,
      "loss": 1.4044,
      "step": 20816
    },
    {
      "epoch": 0.7455011012229842,
      "grad_norm": 1.9254761934280396,
      "learning_rate": 3.208745943520911e-05,
      "loss": 0.9738,
      "step": 20817
    },
    {
      "epoch": 0.7455369133525525,
      "grad_norm": 1.466422438621521,
      "learning_rate": 3.207894595833603e-05,
      "loss": 1.3917,
      "step": 20818
    },
    {
      "epoch": 0.7455727254821208,
      "grad_norm": 1.6191308498382568,
      "learning_rate": 3.207043339525304e-05,
      "loss": 1.6631,
      "step": 20819
    },
    {
      "epoch": 0.7456085376116891,
      "grad_norm": 1.7039192914962769,
      "learning_rate": 3.206192174607482e-05,
      "loss": 1.4387,
      "step": 20820
    },
    {
      "epoch": 0.7456443497412574,
      "grad_norm": 1.9393994808197021,
      "learning_rate": 3.205341101091578e-05,
      "loss": 1.2903,
      "step": 20821
    },
    {
      "epoch": 0.7456801618708256,
      "grad_norm": 1.7326945066452026,
      "learning_rate": 3.2044901189890473e-05,
      "loss": 1.5565,
      "step": 20822
    },
    {
      "epoch": 0.7457159740003939,
      "grad_norm": 1.9713574647903442,
      "learning_rate": 3.2036392283113304e-05,
      "loss": 1.3975,
      "step": 20823
    },
    {
      "epoch": 0.7457517861299622,
      "grad_norm": 1.6577719449996948,
      "learning_rate": 3.202788429069887e-05,
      "loss": 1.582,
      "step": 20824
    },
    {
      "epoch": 0.7457875982595304,
      "grad_norm": 1.6279882192611694,
      "learning_rate": 3.201937721276159e-05,
      "loss": 1.5023,
      "step": 20825
    },
    {
      "epoch": 0.7458234103890988,
      "grad_norm": 2.0655462741851807,
      "learning_rate": 3.201087104941586e-05,
      "loss": 1.4759,
      "step": 20826
    },
    {
      "epoch": 0.7458592225186671,
      "grad_norm": 1.3762956857681274,
      "learning_rate": 3.2002365800776154e-05,
      "loss": 1.3902,
      "step": 20827
    },
    {
      "epoch": 0.7458950346482354,
      "grad_norm": 1.3251439332962036,
      "learning_rate": 3.199386146695691e-05,
      "loss": 1.4928,
      "step": 20828
    },
    {
      "epoch": 0.7459308467778036,
      "grad_norm": 1.480262041091919,
      "learning_rate": 3.1985358048072574e-05,
      "loss": 1.4624,
      "step": 20829
    },
    {
      "epoch": 0.7459666589073719,
      "grad_norm": 1.2799279689788818,
      "learning_rate": 3.197685554423745e-05,
      "loss": 1.3353,
      "step": 20830
    },
    {
      "epoch": 0.7460024710369402,
      "grad_norm": 1.4563822746276855,
      "learning_rate": 3.1968353955566045e-05,
      "loss": 1.0802,
      "step": 20831
    },
    {
      "epoch": 0.7460382831665084,
      "grad_norm": 1.2952502965927124,
      "learning_rate": 3.195985328217266e-05,
      "loss": 1.1515,
      "step": 20832
    },
    {
      "epoch": 0.7460740952960768,
      "grad_norm": 1.3360506296157837,
      "learning_rate": 3.1951353524171715e-05,
      "loss": 1.5667,
      "step": 20833
    },
    {
      "epoch": 0.7461099074256451,
      "grad_norm": 1.5215046405792236,
      "learning_rate": 3.194285468167749e-05,
      "loss": 1.424,
      "step": 20834
    },
    {
      "epoch": 0.7461457195552134,
      "grad_norm": 1.5855753421783447,
      "learning_rate": 3.1934356754804385e-05,
      "loss": 1.4095,
      "step": 20835
    },
    {
      "epoch": 0.7461815316847816,
      "grad_norm": 1.7662148475646973,
      "learning_rate": 3.192585974366673e-05,
      "loss": 1.4234,
      "step": 20836
    },
    {
      "epoch": 0.7462173438143499,
      "grad_norm": 1.916852593421936,
      "learning_rate": 3.19173636483788e-05,
      "loss": 1.4165,
      "step": 20837
    },
    {
      "epoch": 0.7462531559439182,
      "grad_norm": 1.7372554540634155,
      "learning_rate": 3.190886846905491e-05,
      "loss": 1.3308,
      "step": 20838
    },
    {
      "epoch": 0.7462889680734864,
      "grad_norm": 1.336731195449829,
      "learning_rate": 3.190037420580937e-05,
      "loss": 1.2971,
      "step": 20839
    },
    {
      "epoch": 0.7463247802030548,
      "grad_norm": 1.8577148914337158,
      "learning_rate": 3.1891880858756484e-05,
      "loss": 1.6954,
      "step": 20840
    },
    {
      "epoch": 0.7463605923326231,
      "grad_norm": 2.185137987136841,
      "learning_rate": 3.1883388428010465e-05,
      "loss": 1.6788,
      "step": 20841
    },
    {
      "epoch": 0.7463964044621914,
      "grad_norm": 1.7305773496627808,
      "learning_rate": 3.187489691368558e-05,
      "loss": 1.7874,
      "step": 20842
    },
    {
      "epoch": 0.7464322165917596,
      "grad_norm": 1.6137861013412476,
      "learning_rate": 3.186640631589611e-05,
      "loss": 1.2475,
      "step": 20843
    },
    {
      "epoch": 0.7464680287213279,
      "grad_norm": 1.5062600374221802,
      "learning_rate": 3.1857916634756234e-05,
      "loss": 1.2966,
      "step": 20844
    },
    {
      "epoch": 0.7465038408508962,
      "grad_norm": 1.881579875946045,
      "learning_rate": 3.184942787038019e-05,
      "loss": 1.6763,
      "step": 20845
    },
    {
      "epoch": 0.7465396529804644,
      "grad_norm": 1.7157776355743408,
      "learning_rate": 3.184094002288219e-05,
      "loss": 1.4101,
      "step": 20846
    },
    {
      "epoch": 0.7465754651100328,
      "grad_norm": 1.5331974029541016,
      "learning_rate": 3.1832453092376446e-05,
      "loss": 1.4136,
      "step": 20847
    },
    {
      "epoch": 0.7466112772396011,
      "grad_norm": 2.0889968872070312,
      "learning_rate": 3.182396707897709e-05,
      "loss": 1.2539,
      "step": 20848
    },
    {
      "epoch": 0.7466470893691693,
      "grad_norm": 1.3212164640426636,
      "learning_rate": 3.1815481982798324e-05,
      "loss": 1.3957,
      "step": 20849
    },
    {
      "epoch": 0.7466829014987376,
      "grad_norm": 1.5992352962493896,
      "learning_rate": 3.1806997803954316e-05,
      "loss": 1.2118,
      "step": 20850
    },
    {
      "epoch": 0.7467187136283059,
      "grad_norm": 1.8797727823257446,
      "learning_rate": 3.1798514542559164e-05,
      "loss": 1.2311,
      "step": 20851
    },
    {
      "epoch": 0.7467545257578742,
      "grad_norm": 1.4515434503555298,
      "learning_rate": 3.1790032198727014e-05,
      "loss": 1.4252,
      "step": 20852
    },
    {
      "epoch": 0.7467903378874424,
      "grad_norm": 1.5532227754592896,
      "learning_rate": 3.178155077257201e-05,
      "loss": 1.2506,
      "step": 20853
    },
    {
      "epoch": 0.7468261500170108,
      "grad_norm": 1.7346625328063965,
      "learning_rate": 3.177307026420827e-05,
      "loss": 1.4919,
      "step": 20854
    },
    {
      "epoch": 0.7468619621465791,
      "grad_norm": 1.302531361579895,
      "learning_rate": 3.176459067374984e-05,
      "loss": 1.3361,
      "step": 20855
    },
    {
      "epoch": 0.7468977742761473,
      "grad_norm": 2.094909191131592,
      "learning_rate": 3.175611200131081e-05,
      "loss": 1.3555,
      "step": 20856
    },
    {
      "epoch": 0.7469335864057156,
      "grad_norm": 1.6115424633026123,
      "learning_rate": 3.174763424700528e-05,
      "loss": 1.4763,
      "step": 20857
    },
    {
      "epoch": 0.7469693985352839,
      "grad_norm": 2.035444974899292,
      "learning_rate": 3.1739157410947316e-05,
      "loss": 1.6644,
      "step": 20858
    },
    {
      "epoch": 0.7470052106648521,
      "grad_norm": 1.8379755020141602,
      "learning_rate": 3.173068149325091e-05,
      "loss": 1.6804,
      "step": 20859
    },
    {
      "epoch": 0.7470410227944204,
      "grad_norm": 1.5692671537399292,
      "learning_rate": 3.172220649403011e-05,
      "loss": 1.5543,
      "step": 20860
    },
    {
      "epoch": 0.7470768349239888,
      "grad_norm": 1.5705647468566895,
      "learning_rate": 3.1713732413399e-05,
      "loss": 1.475,
      "step": 20861
    },
    {
      "epoch": 0.7471126470535571,
      "grad_norm": 1.777363896369934,
      "learning_rate": 3.1705259251471496e-05,
      "loss": 1.4334,
      "step": 20862
    },
    {
      "epoch": 0.7471484591831253,
      "grad_norm": 2.5444350242614746,
      "learning_rate": 3.169678700836164e-05,
      "loss": 1.3177,
      "step": 20863
    },
    {
      "epoch": 0.7471842713126936,
      "grad_norm": 2.386817216873169,
      "learning_rate": 3.168831568418341e-05,
      "loss": 1.6777,
      "step": 20864
    },
    {
      "epoch": 0.7472200834422619,
      "grad_norm": 1.4960191249847412,
      "learning_rate": 3.16798452790508e-05,
      "loss": 1.3594,
      "step": 20865
    },
    {
      "epoch": 0.7472558955718301,
      "grad_norm": 1.516856074333191,
      "learning_rate": 3.167137579307773e-05,
      "loss": 1.5066,
      "step": 20866
    },
    {
      "epoch": 0.7472917077013984,
      "grad_norm": 1.7102595567703247,
      "learning_rate": 3.1662907226378145e-05,
      "loss": 1.5464,
      "step": 20867
    },
    {
      "epoch": 0.7473275198309668,
      "grad_norm": 1.462852120399475,
      "learning_rate": 3.165443957906603e-05,
      "loss": 1.2122,
      "step": 20868
    },
    {
      "epoch": 0.7473633319605351,
      "grad_norm": 1.76582670211792,
      "learning_rate": 3.164597285125525e-05,
      "loss": 1.5579,
      "step": 20869
    },
    {
      "epoch": 0.7473991440901033,
      "grad_norm": 2.204495906829834,
      "learning_rate": 3.163750704305972e-05,
      "loss": 1.5112,
      "step": 20870
    },
    {
      "epoch": 0.7474349562196716,
      "grad_norm": 2.0119872093200684,
      "learning_rate": 3.162904215459336e-05,
      "loss": 1.4266,
      "step": 20871
    },
    {
      "epoch": 0.7474707683492399,
      "grad_norm": 1.4584416151046753,
      "learning_rate": 3.1620578185970075e-05,
      "loss": 1.4588,
      "step": 20872
    },
    {
      "epoch": 0.7475065804788081,
      "grad_norm": 1.7966676950454712,
      "learning_rate": 3.161211513730368e-05,
      "loss": 1.4396,
      "step": 20873
    },
    {
      "epoch": 0.7475423926083764,
      "grad_norm": 1.828666090965271,
      "learning_rate": 3.160365300870804e-05,
      "loss": 1.3623,
      "step": 20874
    },
    {
      "epoch": 0.7475782047379448,
      "grad_norm": 1.817426085472107,
      "learning_rate": 3.159519180029705e-05,
      "loss": 1.5559,
      "step": 20875
    },
    {
      "epoch": 0.747614016867513,
      "grad_norm": 1.6124011278152466,
      "learning_rate": 3.1586731512184545e-05,
      "loss": 1.2511,
      "step": 20876
    },
    {
      "epoch": 0.7476498289970813,
      "grad_norm": 1.5697824954986572,
      "learning_rate": 3.157827214448428e-05,
      "loss": 1.4974,
      "step": 20877
    },
    {
      "epoch": 0.7476856411266496,
      "grad_norm": 1.6202102899551392,
      "learning_rate": 3.1569813697310115e-05,
      "loss": 1.536,
      "step": 20878
    },
    {
      "epoch": 0.7477214532562179,
      "grad_norm": 1.5296554565429688,
      "learning_rate": 3.156135617077587e-05,
      "loss": 1.4601,
      "step": 20879
    },
    {
      "epoch": 0.7477572653857861,
      "grad_norm": 1.323926568031311,
      "learning_rate": 3.155289956499525e-05,
      "loss": 1.3919,
      "step": 20880
    },
    {
      "epoch": 0.7477930775153544,
      "grad_norm": 1.4089916944503784,
      "learning_rate": 3.15444438800821e-05,
      "loss": 1.5501,
      "step": 20881
    },
    {
      "epoch": 0.7478288896449228,
      "grad_norm": 1.9115785360336304,
      "learning_rate": 3.1535989116150146e-05,
      "loss": 1.6466,
      "step": 20882
    },
    {
      "epoch": 0.747864701774491,
      "grad_norm": 1.4998923540115356,
      "learning_rate": 3.1527535273313166e-05,
      "loss": 1.3259,
      "step": 20883
    },
    {
      "epoch": 0.7479005139040593,
      "grad_norm": 1.8716665506362915,
      "learning_rate": 3.151908235168486e-05,
      "loss": 1.6371,
      "step": 20884
    },
    {
      "epoch": 0.7479363260336276,
      "grad_norm": 1.64845871925354,
      "learning_rate": 3.151063035137896e-05,
      "loss": 1.4893,
      "step": 20885
    },
    {
      "epoch": 0.7479721381631959,
      "grad_norm": 1.3553320169448853,
      "learning_rate": 3.1502179272509216e-05,
      "loss": 1.3028,
      "step": 20886
    },
    {
      "epoch": 0.7480079502927641,
      "grad_norm": 1.9826897382736206,
      "learning_rate": 3.149372911518926e-05,
      "loss": 1.5283,
      "step": 20887
    },
    {
      "epoch": 0.7480437624223324,
      "grad_norm": 1.5831016302108765,
      "learning_rate": 3.1485279879532826e-05,
      "loss": 1.4323,
      "step": 20888
    },
    {
      "epoch": 0.7480795745519008,
      "grad_norm": 1.9086743593215942,
      "learning_rate": 3.147683156565355e-05,
      "loss": 1.4901,
      "step": 20889
    },
    {
      "epoch": 0.748115386681469,
      "grad_norm": 1.5491160154342651,
      "learning_rate": 3.146838417366517e-05,
      "loss": 1.7092,
      "step": 20890
    },
    {
      "epoch": 0.7481511988110373,
      "grad_norm": 1.6630737781524658,
      "learning_rate": 3.145993770368124e-05,
      "loss": 1.4201,
      "step": 20891
    },
    {
      "epoch": 0.7481870109406056,
      "grad_norm": 1.6699607372283936,
      "learning_rate": 3.1451492155815444e-05,
      "loss": 1.2344,
      "step": 20892
    },
    {
      "epoch": 0.7482228230701738,
      "grad_norm": 1.4572018384933472,
      "learning_rate": 3.1443047530181394e-05,
      "loss": 1.5246,
      "step": 20893
    },
    {
      "epoch": 0.7482586351997421,
      "grad_norm": 1.2610589265823364,
      "learning_rate": 3.143460382689274e-05,
      "loss": 1.3851,
      "step": 20894
    },
    {
      "epoch": 0.7482944473293104,
      "grad_norm": 1.5143929719924927,
      "learning_rate": 3.142616104606304e-05,
      "loss": 1.1789,
      "step": 20895
    },
    {
      "epoch": 0.7483302594588788,
      "grad_norm": 1.5444740056991577,
      "learning_rate": 3.141771918780584e-05,
      "loss": 1.2268,
      "step": 20896
    },
    {
      "epoch": 0.748366071588447,
      "grad_norm": 1.3086482286453247,
      "learning_rate": 3.140927825223482e-05,
      "loss": 1.2681,
      "step": 20897
    },
    {
      "epoch": 0.7484018837180153,
      "grad_norm": 1.645628809928894,
      "learning_rate": 3.140083823946346e-05,
      "loss": 1.5036,
      "step": 20898
    },
    {
      "epoch": 0.7484376958475836,
      "grad_norm": 1.3148621320724487,
      "learning_rate": 3.139239914960532e-05,
      "loss": 1.3868,
      "step": 20899
    },
    {
      "epoch": 0.7484735079771518,
      "grad_norm": 1.466416597366333,
      "learning_rate": 3.138396098277396e-05,
      "loss": 1.4579,
      "step": 20900
    },
    {
      "epoch": 0.7485093201067201,
      "grad_norm": 2.0066518783569336,
      "learning_rate": 3.1375523739082936e-05,
      "loss": 1.3685,
      "step": 20901
    },
    {
      "epoch": 0.7485451322362884,
      "grad_norm": 1.307276964187622,
      "learning_rate": 3.136708741864568e-05,
      "loss": 1.0352,
      "step": 20902
    },
    {
      "epoch": 0.7485809443658568,
      "grad_norm": 1.4867724180221558,
      "learning_rate": 3.135865202157574e-05,
      "loss": 1.5873,
      "step": 20903
    },
    {
      "epoch": 0.748616756495425,
      "grad_norm": 1.7414683103561401,
      "learning_rate": 3.135021754798663e-05,
      "loss": 1.5434,
      "step": 20904
    },
    {
      "epoch": 0.7486525686249933,
      "grad_norm": 1.3864867687225342,
      "learning_rate": 3.134178399799175e-05,
      "loss": 1.3304,
      "step": 20905
    },
    {
      "epoch": 0.7486883807545616,
      "grad_norm": 1.6754690408706665,
      "learning_rate": 3.1333351371704634e-05,
      "loss": 1.5759,
      "step": 20906
    },
    {
      "epoch": 0.7487241928841298,
      "grad_norm": 2.0780746936798096,
      "learning_rate": 3.132491966923864e-05,
      "loss": 1.8282,
      "step": 20907
    },
    {
      "epoch": 0.7487600050136981,
      "grad_norm": 1.4286201000213623,
      "learning_rate": 3.131648889070734e-05,
      "loss": 1.4423,
      "step": 20908
    },
    {
      "epoch": 0.7487958171432664,
      "grad_norm": 1.1952495574951172,
      "learning_rate": 3.130805903622405e-05,
      "loss": 1.5991,
      "step": 20909
    },
    {
      "epoch": 0.7488316292728348,
      "grad_norm": 1.5690759420394897,
      "learning_rate": 3.129963010590224e-05,
      "loss": 1.4049,
      "step": 20910
    },
    {
      "epoch": 0.748867441402403,
      "grad_norm": 1.718424916267395,
      "learning_rate": 3.1291202099855245e-05,
      "loss": 1.5049,
      "step": 20911
    },
    {
      "epoch": 0.7489032535319713,
      "grad_norm": 1.4387127161026,
      "learning_rate": 3.1282775018196554e-05,
      "loss": 1.6321,
      "step": 20912
    },
    {
      "epoch": 0.7489390656615396,
      "grad_norm": 1.6412798166275024,
      "learning_rate": 3.127434886103948e-05,
      "loss": 1.3564,
      "step": 20913
    },
    {
      "epoch": 0.7489748777911078,
      "grad_norm": 1.4113123416900635,
      "learning_rate": 3.1265923628497327e-05,
      "loss": 1.3528,
      "step": 20914
    },
    {
      "epoch": 0.7490106899206761,
      "grad_norm": 1.7293850183486938,
      "learning_rate": 3.125749932068359e-05,
      "loss": 1.3285,
      "step": 20915
    },
    {
      "epoch": 0.7490465020502444,
      "grad_norm": 1.4474704265594482,
      "learning_rate": 3.124907593771148e-05,
      "loss": 1.4824,
      "step": 20916
    },
    {
      "epoch": 0.7490823141798127,
      "grad_norm": 2.540874719619751,
      "learning_rate": 3.1240653479694415e-05,
      "loss": 1.2577,
      "step": 20917
    },
    {
      "epoch": 0.749118126309381,
      "grad_norm": 1.6497268676757812,
      "learning_rate": 3.123223194674559e-05,
      "loss": 1.3947,
      "step": 20918
    },
    {
      "epoch": 0.7491539384389493,
      "grad_norm": 1.7501670122146606,
      "learning_rate": 3.122381133897846e-05,
      "loss": 1.7379,
      "step": 20919
    },
    {
      "epoch": 0.7491897505685176,
      "grad_norm": 1.541893482208252,
      "learning_rate": 3.121539165650619e-05,
      "loss": 1.1808,
      "step": 20920
    },
    {
      "epoch": 0.7492255626980858,
      "grad_norm": 1.4933574199676514,
      "learning_rate": 3.120697289944213e-05,
      "loss": 1.4671,
      "step": 20921
    },
    {
      "epoch": 0.7492613748276541,
      "grad_norm": 1.7959959506988525,
      "learning_rate": 3.119855506789948e-05,
      "loss": 1.4678,
      "step": 20922
    },
    {
      "epoch": 0.7492971869572224,
      "grad_norm": 1.9064847230911255,
      "learning_rate": 3.1190138161991536e-05,
      "loss": 1.616,
      "step": 20923
    },
    {
      "epoch": 0.7493329990867907,
      "grad_norm": 1.8895035982131958,
      "learning_rate": 3.118172218183154e-05,
      "loss": 1.3401,
      "step": 20924
    },
    {
      "epoch": 0.749368811216359,
      "grad_norm": 3.042356252670288,
      "learning_rate": 3.117330712753265e-05,
      "loss": 1.5714,
      "step": 20925
    },
    {
      "epoch": 0.7494046233459273,
      "grad_norm": 1.5778942108154297,
      "learning_rate": 3.11648929992082e-05,
      "loss": 1.4569,
      "step": 20926
    },
    {
      "epoch": 0.7494404354754955,
      "grad_norm": 1.8818434476852417,
      "learning_rate": 3.115647979697128e-05,
      "loss": 1.3493,
      "step": 20927
    },
    {
      "epoch": 0.7494762476050638,
      "grad_norm": 1.2638598680496216,
      "learning_rate": 3.114806752093517e-05,
      "loss": 1.4576,
      "step": 20928
    },
    {
      "epoch": 0.7495120597346321,
      "grad_norm": 1.5679749250411987,
      "learning_rate": 3.113965617121291e-05,
      "loss": 1.2887,
      "step": 20929
    },
    {
      "epoch": 0.7495478718642004,
      "grad_norm": 1.635933518409729,
      "learning_rate": 3.1131245747917835e-05,
      "loss": 1.6161,
      "step": 20930
    },
    {
      "epoch": 0.7495836839937687,
      "grad_norm": 1.6987125873565674,
      "learning_rate": 3.1122836251163014e-05,
      "loss": 1.4761,
      "step": 20931
    },
    {
      "epoch": 0.749619496123337,
      "grad_norm": 1.466423511505127,
      "learning_rate": 3.111442768106155e-05,
      "loss": 1.5627,
      "step": 20932
    },
    {
      "epoch": 0.7496553082529053,
      "grad_norm": 1.3633636236190796,
      "learning_rate": 3.1106020037726615e-05,
      "loss": 1.2434,
      "step": 20933
    },
    {
      "epoch": 0.7496911203824735,
      "grad_norm": 1.5454039573669434,
      "learning_rate": 3.1097613321271304e-05,
      "loss": 1.3154,
      "step": 20934
    },
    {
      "epoch": 0.7497269325120418,
      "grad_norm": 1.5664490461349487,
      "learning_rate": 3.108920753180875e-05,
      "loss": 1.4282,
      "step": 20935
    },
    {
      "epoch": 0.7497627446416101,
      "grad_norm": 2.946145534515381,
      "learning_rate": 3.1080802669452e-05,
      "loss": 1.5775,
      "step": 20936
    },
    {
      "epoch": 0.7497985567711783,
      "grad_norm": 2.2433459758758545,
      "learning_rate": 3.107239873431416e-05,
      "loss": 1.4163,
      "step": 20937
    },
    {
      "epoch": 0.7498343689007467,
      "grad_norm": 1.6156102418899536,
      "learning_rate": 3.1063995726508296e-05,
      "loss": 1.1802,
      "step": 20938
    },
    {
      "epoch": 0.749870181030315,
      "grad_norm": 2.013106346130371,
      "learning_rate": 3.105559364614743e-05,
      "loss": 1.434,
      "step": 20939
    },
    {
      "epoch": 0.7499059931598833,
      "grad_norm": 2.1555824279785156,
      "learning_rate": 3.1047192493344624e-05,
      "loss": 1.3182,
      "step": 20940
    },
    {
      "epoch": 0.7499418052894515,
      "grad_norm": 1.372545599937439,
      "learning_rate": 3.103879226821289e-05,
      "loss": 1.2394,
      "step": 20941
    },
    {
      "epoch": 0.7499776174190198,
      "grad_norm": 2.159738302230835,
      "learning_rate": 3.1030392970865286e-05,
      "loss": 1.2531,
      "step": 20942
    },
    {
      "epoch": 0.7500134295485881,
      "grad_norm": 1.378603219985962,
      "learning_rate": 3.102199460141475e-05,
      "loss": 1.3648,
      "step": 20943
    },
    {
      "epoch": 0.7500492416781563,
      "grad_norm": 1.817529320716858,
      "learning_rate": 3.1013597159974304e-05,
      "loss": 1.4851,
      "step": 20944
    },
    {
      "epoch": 0.7500850538077247,
      "grad_norm": 1.4849520921707153,
      "learning_rate": 3.1005200646656915e-05,
      "loss": 1.3406,
      "step": 20945
    },
    {
      "epoch": 0.750120865937293,
      "grad_norm": 2.093026638031006,
      "learning_rate": 3.09968050615756e-05,
      "loss": 1.2892,
      "step": 20946
    },
    {
      "epoch": 0.7501566780668613,
      "grad_norm": 1.8699207305908203,
      "learning_rate": 3.0988410404843216e-05,
      "loss": 1.7263,
      "step": 20947
    },
    {
      "epoch": 0.7501924901964295,
      "grad_norm": 2.0237114429473877,
      "learning_rate": 3.0980016676572766e-05,
      "loss": 1.4728,
      "step": 20948
    },
    {
      "epoch": 0.7502283023259978,
      "grad_norm": 1.5000897645950317,
      "learning_rate": 3.097162387687719e-05,
      "loss": 1.3496,
      "step": 20949
    },
    {
      "epoch": 0.7502641144555661,
      "grad_norm": 2.2044296264648438,
      "learning_rate": 3.096323200586934e-05,
      "loss": 1.1754,
      "step": 20950
    },
    {
      "epoch": 0.7502999265851343,
      "grad_norm": 1.645936369895935,
      "learning_rate": 3.0954841063662145e-05,
      "loss": 1.5664,
      "step": 20951
    },
    {
      "epoch": 0.7503357387147027,
      "grad_norm": 1.730129599571228,
      "learning_rate": 3.094645105036851e-05,
      "loss": 1.2231,
      "step": 20952
    },
    {
      "epoch": 0.750371550844271,
      "grad_norm": 1.6695406436920166,
      "learning_rate": 3.093806196610134e-05,
      "loss": 1.5443,
      "step": 20953
    },
    {
      "epoch": 0.7504073629738393,
      "grad_norm": 1.7872570753097534,
      "learning_rate": 3.092967381097342e-05,
      "loss": 1.2933,
      "step": 20954
    },
    {
      "epoch": 0.7504431751034075,
      "grad_norm": 1.5018796920776367,
      "learning_rate": 3.092128658509765e-05,
      "loss": 1.4061,
      "step": 20955
    },
    {
      "epoch": 0.7504789872329758,
      "grad_norm": 2.464582681655884,
      "learning_rate": 3.09129002885869e-05,
      "loss": 1.5651,
      "step": 20956
    },
    {
      "epoch": 0.7505147993625441,
      "grad_norm": 1.7600558996200562,
      "learning_rate": 3.090451492155392e-05,
      "loss": 1.6592,
      "step": 20957
    },
    {
      "epoch": 0.7505506114921123,
      "grad_norm": 1.316369652748108,
      "learning_rate": 3.089613048411158e-05,
      "loss": 1.3414,
      "step": 20958
    },
    {
      "epoch": 0.7505864236216807,
      "grad_norm": 1.9160432815551758,
      "learning_rate": 3.088774697637265e-05,
      "loss": 1.3255,
      "step": 20959
    },
    {
      "epoch": 0.750622235751249,
      "grad_norm": 2.339635133743286,
      "learning_rate": 3.087936439844997e-05,
      "loss": 1.357,
      "step": 20960
    },
    {
      "epoch": 0.7506580478808172,
      "grad_norm": 1.4159473180770874,
      "learning_rate": 3.087098275045626e-05,
      "loss": 1.3368,
      "step": 20961
    },
    {
      "epoch": 0.7506938600103855,
      "grad_norm": 2.2639219760894775,
      "learning_rate": 3.08626020325043e-05,
      "loss": 1.565,
      "step": 20962
    },
    {
      "epoch": 0.7507296721399538,
      "grad_norm": 1.6115988492965698,
      "learning_rate": 3.0854222244706857e-05,
      "loss": 1.4387,
      "step": 20963
    },
    {
      "epoch": 0.750765484269522,
      "grad_norm": 1.6219232082366943,
      "learning_rate": 3.0845843387176686e-05,
      "loss": 1.5412,
      "step": 20964
    },
    {
      "epoch": 0.7508012963990903,
      "grad_norm": 1.478438377380371,
      "learning_rate": 3.083746546002646e-05,
      "loss": 1.3816,
      "step": 20965
    },
    {
      "epoch": 0.7508371085286587,
      "grad_norm": 1.358998417854309,
      "learning_rate": 3.082908846336891e-05,
      "loss": 1.3876,
      "step": 20966
    },
    {
      "epoch": 0.750872920658227,
      "grad_norm": 1.556822419166565,
      "learning_rate": 3.082071239731681e-05,
      "loss": 1.4421,
      "step": 20967
    },
    {
      "epoch": 0.7509087327877952,
      "grad_norm": 1.4174234867095947,
      "learning_rate": 3.0812337261982735e-05,
      "loss": 1.4759,
      "step": 20968
    },
    {
      "epoch": 0.7509445449173635,
      "grad_norm": 1.8507611751556396,
      "learning_rate": 3.080396305747942e-05,
      "loss": 1.6876,
      "step": 20969
    },
    {
      "epoch": 0.7509803570469318,
      "grad_norm": 1.3811239004135132,
      "learning_rate": 3.0795589783919543e-05,
      "loss": 1.3176,
      "step": 20970
    },
    {
      "epoch": 0.7510161691765,
      "grad_norm": 1.923411250114441,
      "learning_rate": 3.078721744141575e-05,
      "loss": 1.3883,
      "step": 20971
    },
    {
      "epoch": 0.7510519813060683,
      "grad_norm": 2.107851266860962,
      "learning_rate": 3.0778846030080644e-05,
      "loss": 1.5351,
      "step": 20972
    },
    {
      "epoch": 0.7510877934356367,
      "grad_norm": 1.7084307670593262,
      "learning_rate": 3.077047555002688e-05,
      "loss": 1.539,
      "step": 20973
    },
    {
      "epoch": 0.751123605565205,
      "grad_norm": 2.1994099617004395,
      "learning_rate": 3.0762106001367095e-05,
      "loss": 1.2937,
      "step": 20974
    },
    {
      "epoch": 0.7511594176947732,
      "grad_norm": 1.8254406452178955,
      "learning_rate": 3.075373738421383e-05,
      "loss": 1.2871,
      "step": 20975
    },
    {
      "epoch": 0.7511952298243415,
      "grad_norm": 1.9883519411087036,
      "learning_rate": 3.0745369698679715e-05,
      "loss": 1.3643,
      "step": 20976
    },
    {
      "epoch": 0.7512310419539098,
      "grad_norm": 1.6190694570541382,
      "learning_rate": 3.0737002944877314e-05,
      "loss": 1.2145,
      "step": 20977
    },
    {
      "epoch": 0.751266854083478,
      "grad_norm": 1.9418599605560303,
      "learning_rate": 3.072863712291922e-05,
      "loss": 1.3982,
      "step": 20978
    },
    {
      "epoch": 0.7513026662130463,
      "grad_norm": 1.3245214223861694,
      "learning_rate": 3.0720272232917934e-05,
      "loss": 1.4906,
      "step": 20979
    },
    {
      "epoch": 0.7513384783426147,
      "grad_norm": 1.2552196979522705,
      "learning_rate": 3.071190827498602e-05,
      "loss": 1.4326,
      "step": 20980
    },
    {
      "epoch": 0.751374290472183,
      "grad_norm": 1.699867844581604,
      "learning_rate": 3.070354524923601e-05,
      "loss": 1.4727,
      "step": 20981
    },
    {
      "epoch": 0.7514101026017512,
      "grad_norm": 1.6668874025344849,
      "learning_rate": 3.0695183155780435e-05,
      "loss": 1.3547,
      "step": 20982
    },
    {
      "epoch": 0.7514459147313195,
      "grad_norm": 1.845920443534851,
      "learning_rate": 3.068682199473175e-05,
      "loss": 1.4138,
      "step": 20983
    },
    {
      "epoch": 0.7514817268608878,
      "grad_norm": 1.6310741901397705,
      "learning_rate": 3.067846176620247e-05,
      "loss": 1.2511,
      "step": 20984
    },
    {
      "epoch": 0.751517538990456,
      "grad_norm": 1.9002145528793335,
      "learning_rate": 3.06701024703051e-05,
      "loss": 1.4841,
      "step": 20985
    },
    {
      "epoch": 0.7515533511200243,
      "grad_norm": 2.533083438873291,
      "learning_rate": 3.0661744107152025e-05,
      "loss": 1.6083,
      "step": 20986
    },
    {
      "epoch": 0.7515891632495927,
      "grad_norm": 1.722609519958496,
      "learning_rate": 3.0653386676855756e-05,
      "loss": 1.2962,
      "step": 20987
    },
    {
      "epoch": 0.751624975379161,
      "grad_norm": 2.650226593017578,
      "learning_rate": 3.064503017952871e-05,
      "loss": 1.357,
      "step": 20988
    },
    {
      "epoch": 0.7516607875087292,
      "grad_norm": 2.232440948486328,
      "learning_rate": 3.0636674615283364e-05,
      "loss": 1.5926,
      "step": 20989
    },
    {
      "epoch": 0.7516965996382975,
      "grad_norm": 1.7429866790771484,
      "learning_rate": 3.0628319984232056e-05,
      "loss": 1.547,
      "step": 20990
    },
    {
      "epoch": 0.7517324117678658,
      "grad_norm": 1.6939197778701782,
      "learning_rate": 3.061996628648721e-05,
      "loss": 1.1195,
      "step": 20991
    },
    {
      "epoch": 0.751768223897434,
      "grad_norm": 1.5635935068130493,
      "learning_rate": 3.0611613522161266e-05,
      "loss": 1.1881,
      "step": 20992
    },
    {
      "epoch": 0.7518040360270023,
      "grad_norm": 1.8957972526550293,
      "learning_rate": 3.0603261691366525e-05,
      "loss": 1.3517,
      "step": 20993
    },
    {
      "epoch": 0.7518398481565707,
      "grad_norm": 2.051981210708618,
      "learning_rate": 3.05949107942154e-05,
      "loss": 1.413,
      "step": 20994
    },
    {
      "epoch": 0.751875660286139,
      "grad_norm": 1.7447413206100464,
      "learning_rate": 3.0586560830820174e-05,
      "loss": 1.5783,
      "step": 20995
    },
    {
      "epoch": 0.7519114724157072,
      "grad_norm": 1.535361886024475,
      "learning_rate": 3.05782118012933e-05,
      "loss": 1.3961,
      "step": 20996
    },
    {
      "epoch": 0.7519472845452755,
      "grad_norm": 1.363828420639038,
      "learning_rate": 3.0569863705747004e-05,
      "loss": 1.5911,
      "step": 20997
    },
    {
      "epoch": 0.7519830966748438,
      "grad_norm": 3.876757860183716,
      "learning_rate": 3.0561516544293634e-05,
      "loss": 1.889,
      "step": 20998
    },
    {
      "epoch": 0.752018908804412,
      "grad_norm": 1.9156787395477295,
      "learning_rate": 3.0553170317045485e-05,
      "loss": 1.1729,
      "step": 20999
    },
    {
      "epoch": 0.7520547209339803,
      "grad_norm": 2.1314427852630615,
      "learning_rate": 3.054482502411489e-05,
      "loss": 1.4521,
      "step": 21000
    },
    {
      "epoch": 0.7520905330635487,
      "grad_norm": 1.6527806520462036,
      "learning_rate": 3.0536480665614075e-05,
      "loss": 1.6505,
      "step": 21001
    },
    {
      "epoch": 0.7521263451931169,
      "grad_norm": 1.80655837059021,
      "learning_rate": 3.052813724165525e-05,
      "loss": 1.2674,
      "step": 21002
    },
    {
      "epoch": 0.7521621573226852,
      "grad_norm": 2.413799285888672,
      "learning_rate": 3.051979475235078e-05,
      "loss": 1.1142,
      "step": 21003
    },
    {
      "epoch": 0.7521979694522535,
      "grad_norm": 1.5094581842422485,
      "learning_rate": 3.0511453197812834e-05,
      "loss": 1.3326,
      "step": 21004
    },
    {
      "epoch": 0.7522337815818217,
      "grad_norm": 1.4582942724227905,
      "learning_rate": 3.050311257815368e-05,
      "loss": 1.3724,
      "step": 21005
    },
    {
      "epoch": 0.75226959371139,
      "grad_norm": 1.700408935546875,
      "learning_rate": 3.0494772893485435e-05,
      "loss": 1.1878,
      "step": 21006
    },
    {
      "epoch": 0.7523054058409583,
      "grad_norm": 1.694864273071289,
      "learning_rate": 3.0486434143920428e-05,
      "loss": 1.2196,
      "step": 21007
    },
    {
      "epoch": 0.7523412179705267,
      "grad_norm": 1.9889030456542969,
      "learning_rate": 3.047809632957075e-05,
      "loss": 1.3167,
      "step": 21008
    },
    {
      "epoch": 0.7523770301000949,
      "grad_norm": 1.6860452890396118,
      "learning_rate": 3.0469759450548607e-05,
      "loss": 1.2681,
      "step": 21009
    },
    {
      "epoch": 0.7524128422296632,
      "grad_norm": 1.7383288145065308,
      "learning_rate": 3.0461423506966203e-05,
      "loss": 1.2188,
      "step": 21010
    },
    {
      "epoch": 0.7524486543592315,
      "grad_norm": 1.203225016593933,
      "learning_rate": 3.0453088498935612e-05,
      "loss": 1.459,
      "step": 21011
    },
    {
      "epoch": 0.7524844664887997,
      "grad_norm": 1.667195439338684,
      "learning_rate": 3.0444754426569032e-05,
      "loss": 1.4978,
      "step": 21012
    },
    {
      "epoch": 0.752520278618368,
      "grad_norm": 1.8023768663406372,
      "learning_rate": 3.04364212899785e-05,
      "loss": 1.3717,
      "step": 21013
    },
    {
      "epoch": 0.7525560907479363,
      "grad_norm": 1.4114989042282104,
      "learning_rate": 3.0428089089276257e-05,
      "loss": 1.6978,
      "step": 21014
    },
    {
      "epoch": 0.7525919028775047,
      "grad_norm": 1.4679601192474365,
      "learning_rate": 3.04197578245743e-05,
      "loss": 1.5715,
      "step": 21015
    },
    {
      "epoch": 0.7526277150070729,
      "grad_norm": 1.8693400621414185,
      "learning_rate": 3.041142749598479e-05,
      "loss": 1.6269,
      "step": 21016
    },
    {
      "epoch": 0.7526635271366412,
      "grad_norm": 2.4425885677337646,
      "learning_rate": 3.0403098103619687e-05,
      "loss": 1.4659,
      "step": 21017
    },
    {
      "epoch": 0.7526993392662095,
      "grad_norm": 1.5194597244262695,
      "learning_rate": 3.0394769647591194e-05,
      "loss": 1.5173,
      "step": 21018
    },
    {
      "epoch": 0.7527351513957777,
      "grad_norm": 1.4792680740356445,
      "learning_rate": 3.0386442128011282e-05,
      "loss": 1.0879,
      "step": 21019
    },
    {
      "epoch": 0.752770963525346,
      "grad_norm": 1.689097285270691,
      "learning_rate": 3.037811554499197e-05,
      "loss": 1.0078,
      "step": 21020
    },
    {
      "epoch": 0.7528067756549143,
      "grad_norm": 1.4809821844100952,
      "learning_rate": 3.0369789898645306e-05,
      "loss": 1.4036,
      "step": 21021
    },
    {
      "epoch": 0.7528425877844827,
      "grad_norm": 1.493944764137268,
      "learning_rate": 3.0361465189083305e-05,
      "loss": 1.4354,
      "step": 21022
    },
    {
      "epoch": 0.7528783999140509,
      "grad_norm": 1.486365795135498,
      "learning_rate": 3.0353141416417997e-05,
      "loss": 1.2398,
      "step": 21023
    },
    {
      "epoch": 0.7529142120436192,
      "grad_norm": 1.6679426431655884,
      "learning_rate": 3.034481858076127e-05,
      "loss": 1.3382,
      "step": 21024
    },
    {
      "epoch": 0.7529500241731875,
      "grad_norm": 1.6131479740142822,
      "learning_rate": 3.0336496682225214e-05,
      "loss": 1.4169,
      "step": 21025
    },
    {
      "epoch": 0.7529858363027557,
      "grad_norm": 1.4379775524139404,
      "learning_rate": 3.0328175720921715e-05,
      "loss": 1.2663,
      "step": 21026
    },
    {
      "epoch": 0.753021648432324,
      "grad_norm": 1.8165546655654907,
      "learning_rate": 3.0319855696962762e-05,
      "loss": 1.577,
      "step": 21027
    },
    {
      "epoch": 0.7530574605618923,
      "grad_norm": 1.724393367767334,
      "learning_rate": 3.0311536610460245e-05,
      "loss": 1.671,
      "step": 21028
    },
    {
      "epoch": 0.7530932726914606,
      "grad_norm": 1.8695555925369263,
      "learning_rate": 3.0303218461526116e-05,
      "loss": 1.7049,
      "step": 21029
    },
    {
      "epoch": 0.7531290848210289,
      "grad_norm": 2.045958995819092,
      "learning_rate": 3.02949012502723e-05,
      "loss": 1.5631,
      "step": 21030
    },
    {
      "epoch": 0.7531648969505972,
      "grad_norm": 1.6942603588104248,
      "learning_rate": 3.028658497681065e-05,
      "loss": 1.4666,
      "step": 21031
    },
    {
      "epoch": 0.7532007090801655,
      "grad_norm": 2.18617582321167,
      "learning_rate": 3.0278269641253075e-05,
      "loss": 1.3933,
      "step": 21032
    },
    {
      "epoch": 0.7532365212097337,
      "grad_norm": 2.1976447105407715,
      "learning_rate": 3.0269955243711457e-05,
      "loss": 1.314,
      "step": 21033
    },
    {
      "epoch": 0.753272333339302,
      "grad_norm": 1.413646936416626,
      "learning_rate": 3.0261641784297666e-05,
      "loss": 1.5067,
      "step": 21034
    },
    {
      "epoch": 0.7533081454688703,
      "grad_norm": 1.595048189163208,
      "learning_rate": 3.0253329263123497e-05,
      "loss": 1.3638,
      "step": 21035
    },
    {
      "epoch": 0.7533439575984386,
      "grad_norm": 2.5425102710723877,
      "learning_rate": 3.0245017680300813e-05,
      "loss": 1.1756,
      "step": 21036
    },
    {
      "epoch": 0.7533797697280069,
      "grad_norm": 1.6122390031814575,
      "learning_rate": 3.0236707035941482e-05,
      "loss": 1.591,
      "step": 21037
    },
    {
      "epoch": 0.7534155818575752,
      "grad_norm": 1.5871548652648926,
      "learning_rate": 3.0228397330157233e-05,
      "loss": 1.2486,
      "step": 21038
    },
    {
      "epoch": 0.7534513939871434,
      "grad_norm": 1.390411615371704,
      "learning_rate": 3.022008856305989e-05,
      "loss": 1.6519,
      "step": 21039
    },
    {
      "epoch": 0.7534872061167117,
      "grad_norm": 1.7453995943069458,
      "learning_rate": 3.0211780734761254e-05,
      "loss": 1.1648,
      "step": 21040
    },
    {
      "epoch": 0.75352301824628,
      "grad_norm": 1.4689477682113647,
      "learning_rate": 3.020347384537312e-05,
      "loss": 1.5223,
      "step": 21041
    },
    {
      "epoch": 0.7535588303758483,
      "grad_norm": 1.7363903522491455,
      "learning_rate": 3.019516789500718e-05,
      "loss": 1.3932,
      "step": 21042
    },
    {
      "epoch": 0.7535946425054166,
      "grad_norm": 1.9193975925445557,
      "learning_rate": 3.0186862883775214e-05,
      "loss": 1.2524,
      "step": 21043
    },
    {
      "epoch": 0.7536304546349849,
      "grad_norm": 1.6323972940444946,
      "learning_rate": 3.017855881178899e-05,
      "loss": 1.7168,
      "step": 21044
    },
    {
      "epoch": 0.7536662667645532,
      "grad_norm": 1.584980845451355,
      "learning_rate": 3.0170255679160163e-05,
      "loss": 1.1829,
      "step": 21045
    },
    {
      "epoch": 0.7537020788941214,
      "grad_norm": 1.800627589225769,
      "learning_rate": 3.0161953486000473e-05,
      "loss": 1.3106,
      "step": 21046
    },
    {
      "epoch": 0.7537378910236897,
      "grad_norm": 1.3021795749664307,
      "learning_rate": 3.0153652232421603e-05,
      "loss": 1.4727,
      "step": 21047
    },
    {
      "epoch": 0.753773703153258,
      "grad_norm": 1.9696956872940063,
      "learning_rate": 3.014535191853529e-05,
      "loss": 1.3064,
      "step": 21048
    },
    {
      "epoch": 0.7538095152828262,
      "grad_norm": 1.3597009181976318,
      "learning_rate": 3.0137052544453126e-05,
      "loss": 1.4837,
      "step": 21049
    },
    {
      "epoch": 0.7538453274123946,
      "grad_norm": 1.9204648733139038,
      "learning_rate": 3.0128754110286806e-05,
      "loss": 1.6705,
      "step": 21050
    },
    {
      "epoch": 0.7538811395419629,
      "grad_norm": 1.4181511402130127,
      "learning_rate": 3.012045661614796e-05,
      "loss": 1.4157,
      "step": 21051
    },
    {
      "epoch": 0.7539169516715312,
      "grad_norm": 1.776105523109436,
      "learning_rate": 3.0112160062148274e-05,
      "loss": 1.5494,
      "step": 21052
    },
    {
      "epoch": 0.7539527638010994,
      "grad_norm": 1.4429407119750977,
      "learning_rate": 3.01038644483993e-05,
      "loss": 1.3892,
      "step": 21053
    },
    {
      "epoch": 0.7539885759306677,
      "grad_norm": 1.9590340852737427,
      "learning_rate": 3.0095569775012665e-05,
      "loss": 1.2104,
      "step": 21054
    },
    {
      "epoch": 0.754024388060236,
      "grad_norm": 1.9207379817962646,
      "learning_rate": 3.0087276042099997e-05,
      "loss": 1.5865,
      "step": 21055
    },
    {
      "epoch": 0.7540602001898042,
      "grad_norm": 1.3791786432266235,
      "learning_rate": 3.007898324977282e-05,
      "loss": 1.1746,
      "step": 21056
    },
    {
      "epoch": 0.7540960123193726,
      "grad_norm": 1.5347812175750732,
      "learning_rate": 3.0070691398142726e-05,
      "loss": 1.4336,
      "step": 21057
    },
    {
      "epoch": 0.7541318244489409,
      "grad_norm": 1.8425312042236328,
      "learning_rate": 3.0062400487321286e-05,
      "loss": 1.704,
      "step": 21058
    },
    {
      "epoch": 0.7541676365785092,
      "grad_norm": 1.758529543876648,
      "learning_rate": 3.0054110517420052e-05,
      "loss": 1.6625,
      "step": 21059
    },
    {
      "epoch": 0.7542034487080774,
      "grad_norm": 1.4791284799575806,
      "learning_rate": 3.004582148855052e-05,
      "loss": 1.6378,
      "step": 21060
    },
    {
      "epoch": 0.7542392608376457,
      "grad_norm": 1.824438452720642,
      "learning_rate": 3.0037533400824226e-05,
      "loss": 1.3131,
      "step": 21061
    },
    {
      "epoch": 0.754275072967214,
      "grad_norm": 1.6703860759735107,
      "learning_rate": 3.0029246254352694e-05,
      "loss": 1.6793,
      "step": 21062
    },
    {
      "epoch": 0.7543108850967822,
      "grad_norm": 1.5747116804122925,
      "learning_rate": 3.002096004924737e-05,
      "loss": 1.4264,
      "step": 21063
    },
    {
      "epoch": 0.7543466972263506,
      "grad_norm": 1.7274689674377441,
      "learning_rate": 3.0012674785619766e-05,
      "loss": 1.4875,
      "step": 21064
    },
    {
      "epoch": 0.7543825093559189,
      "grad_norm": 1.6294996738433838,
      "learning_rate": 3.0004390463581345e-05,
      "loss": 1.2883,
      "step": 21065
    },
    {
      "epoch": 0.7544183214854872,
      "grad_norm": 1.555970549583435,
      "learning_rate": 2.9996107083243598e-05,
      "loss": 1.3513,
      "step": 21066
    },
    {
      "epoch": 0.7544541336150554,
      "grad_norm": 1.3730995655059814,
      "learning_rate": 2.9987824644717898e-05,
      "loss": 1.3771,
      "step": 21067
    },
    {
      "epoch": 0.7544899457446237,
      "grad_norm": 1.953616738319397,
      "learning_rate": 2.997954314811571e-05,
      "loss": 1.5383,
      "step": 21068
    },
    {
      "epoch": 0.754525757874192,
      "grad_norm": 1.8902431726455688,
      "learning_rate": 2.9971262593548443e-05,
      "loss": 1.5828,
      "step": 21069
    },
    {
      "epoch": 0.7545615700037602,
      "grad_norm": 1.774874210357666,
      "learning_rate": 2.996298298112754e-05,
      "loss": 1.4622,
      "step": 21070
    },
    {
      "epoch": 0.7545973821333286,
      "grad_norm": 1.683915615081787,
      "learning_rate": 2.9954704310964332e-05,
      "loss": 1.5168,
      "step": 21071
    },
    {
      "epoch": 0.7546331942628969,
      "grad_norm": 1.5608534812927246,
      "learning_rate": 2.9946426583170217e-05,
      "loss": 1.3258,
      "step": 21072
    },
    {
      "epoch": 0.7546690063924651,
      "grad_norm": 1.3177636861801147,
      "learning_rate": 2.9938149797856608e-05,
      "loss": 1.1626,
      "step": 21073
    },
    {
      "epoch": 0.7547048185220334,
      "grad_norm": 1.6210861206054688,
      "learning_rate": 2.992987395513479e-05,
      "loss": 1.6502,
      "step": 21074
    },
    {
      "epoch": 0.7547406306516017,
      "grad_norm": 1.682212471961975,
      "learning_rate": 2.9921599055116135e-05,
      "loss": 1.5752,
      "step": 21075
    },
    {
      "epoch": 0.75477644278117,
      "grad_norm": 1.6862162351608276,
      "learning_rate": 2.991332509791196e-05,
      "loss": 1.6626,
      "step": 21076
    },
    {
      "epoch": 0.7548122549107382,
      "grad_norm": 2.048794746398926,
      "learning_rate": 2.9905052083633632e-05,
      "loss": 1.5058,
      "step": 21077
    },
    {
      "epoch": 0.7548480670403066,
      "grad_norm": 1.955229640007019,
      "learning_rate": 2.9896780012392377e-05,
      "loss": 1.5886,
      "step": 21078
    },
    {
      "epoch": 0.7548838791698749,
      "grad_norm": 1.326578140258789,
      "learning_rate": 2.9888508884299516e-05,
      "loss": 1.5777,
      "step": 21079
    },
    {
      "epoch": 0.7549196912994431,
      "grad_norm": 1.9599168300628662,
      "learning_rate": 2.9880238699466367e-05,
      "loss": 1.6122,
      "step": 21080
    },
    {
      "epoch": 0.7549555034290114,
      "grad_norm": 1.659332036972046,
      "learning_rate": 2.9871969458004135e-05,
      "loss": 1.2969,
      "step": 21081
    },
    {
      "epoch": 0.7549913155585797,
      "grad_norm": 2.504672050476074,
      "learning_rate": 2.9863701160024083e-05,
      "loss": 1.5504,
      "step": 21082
    },
    {
      "epoch": 0.755027127688148,
      "grad_norm": 1.9664294719696045,
      "learning_rate": 2.9855433805637467e-05,
      "loss": 1.3975,
      "step": 21083
    },
    {
      "epoch": 0.7550629398177162,
      "grad_norm": 2.3380939960479736,
      "learning_rate": 2.9847167394955543e-05,
      "loss": 1.267,
      "step": 21084
    },
    {
      "epoch": 0.7550987519472846,
      "grad_norm": 1.6964002847671509,
      "learning_rate": 2.9838901928089456e-05,
      "loss": 1.1031,
      "step": 21085
    },
    {
      "epoch": 0.7551345640768529,
      "grad_norm": 1.679082989692688,
      "learning_rate": 2.983063740515044e-05,
      "loss": 1.5012,
      "step": 21086
    },
    {
      "epoch": 0.7551703762064211,
      "grad_norm": 1.3451464176177979,
      "learning_rate": 2.9822373826249693e-05,
      "loss": 1.3227,
      "step": 21087
    },
    {
      "epoch": 0.7552061883359894,
      "grad_norm": 1.67902672290802,
      "learning_rate": 2.9814111191498405e-05,
      "loss": 1.4921,
      "step": 21088
    },
    {
      "epoch": 0.7552420004655577,
      "grad_norm": 1.4258464574813843,
      "learning_rate": 2.9805849501007733e-05,
      "loss": 1.4449,
      "step": 21089
    },
    {
      "epoch": 0.7552778125951259,
      "grad_norm": 1.4557456970214844,
      "learning_rate": 2.979758875488874e-05,
      "loss": 1.0844,
      "step": 21090
    },
    {
      "epoch": 0.7553136247246942,
      "grad_norm": 1.5208238363265991,
      "learning_rate": 2.9789328953252694e-05,
      "loss": 1.5754,
      "step": 21091
    },
    {
      "epoch": 0.7553494368542626,
      "grad_norm": 1.7373424768447876,
      "learning_rate": 2.9781070096210627e-05,
      "loss": 1.6214,
      "step": 21092
    },
    {
      "epoch": 0.7553852489838309,
      "grad_norm": 1.8888261318206787,
      "learning_rate": 2.9772812183873733e-05,
      "loss": 1.4306,
      "step": 21093
    },
    {
      "epoch": 0.7554210611133991,
      "grad_norm": 1.2451953887939453,
      "learning_rate": 2.9764555216352997e-05,
      "loss": 1.1014,
      "step": 21094
    },
    {
      "epoch": 0.7554568732429674,
      "grad_norm": 2.3028512001037598,
      "learning_rate": 2.975629919375963e-05,
      "loss": 1.2671,
      "step": 21095
    },
    {
      "epoch": 0.7554926853725357,
      "grad_norm": 1.363605260848999,
      "learning_rate": 2.974804411620462e-05,
      "loss": 1.3705,
      "step": 21096
    },
    {
      "epoch": 0.7555284975021039,
      "grad_norm": 1.8250839710235596,
      "learning_rate": 2.973978998379906e-05,
      "loss": 1.4062,
      "step": 21097
    },
    {
      "epoch": 0.7555643096316722,
      "grad_norm": 1.7646355628967285,
      "learning_rate": 2.9731536796654026e-05,
      "loss": 1.4773,
      "step": 21098
    },
    {
      "epoch": 0.7556001217612406,
      "grad_norm": 2.1657111644744873,
      "learning_rate": 2.9723284554880493e-05,
      "loss": 1.1014,
      "step": 21099
    },
    {
      "epoch": 0.7556359338908089,
      "grad_norm": 1.4602758884429932,
      "learning_rate": 2.9715033258589543e-05,
      "loss": 1.4667,
      "step": 21100
    },
    {
      "epoch": 0.7556717460203771,
      "grad_norm": 1.4769114255905151,
      "learning_rate": 2.9706782907892104e-05,
      "loss": 1.1822,
      "step": 21101
    },
    {
      "epoch": 0.7557075581499454,
      "grad_norm": 1.8590151071548462,
      "learning_rate": 2.9698533502899294e-05,
      "loss": 1.5489,
      "step": 21102
    },
    {
      "epoch": 0.7557433702795137,
      "grad_norm": 1.3102487325668335,
      "learning_rate": 2.9690285043722e-05,
      "loss": 1.4329,
      "step": 21103
    },
    {
      "epoch": 0.7557791824090819,
      "grad_norm": 1.4828435182571411,
      "learning_rate": 2.9682037530471252e-05,
      "loss": 1.3056,
      "step": 21104
    },
    {
      "epoch": 0.7558149945386502,
      "grad_norm": 1.2443448305130005,
      "learning_rate": 2.967379096325793e-05,
      "loss": 1.3004,
      "step": 21105
    },
    {
      "epoch": 0.7558508066682186,
      "grad_norm": 1.5931856632232666,
      "learning_rate": 2.966554534219309e-05,
      "loss": 1.6132,
      "step": 21106
    },
    {
      "epoch": 0.7558866187977868,
      "grad_norm": 1.3592056035995483,
      "learning_rate": 2.965730066738762e-05,
      "loss": 1.4226,
      "step": 21107
    },
    {
      "epoch": 0.7559224309273551,
      "grad_norm": 1.710934042930603,
      "learning_rate": 2.964905693895237e-05,
      "loss": 1.6701,
      "step": 21108
    },
    {
      "epoch": 0.7559582430569234,
      "grad_norm": 1.9876086711883545,
      "learning_rate": 2.9640814156998374e-05,
      "loss": 1.5418,
      "step": 21109
    },
    {
      "epoch": 0.7559940551864917,
      "grad_norm": 1.912034511566162,
      "learning_rate": 2.9632572321636443e-05,
      "loss": 1.5359,
      "step": 21110
    },
    {
      "epoch": 0.7560298673160599,
      "grad_norm": 1.493776798248291,
      "learning_rate": 2.9624331432977515e-05,
      "loss": 1.5861,
      "step": 21111
    },
    {
      "epoch": 0.7560656794456282,
      "grad_norm": 2.0025715827941895,
      "learning_rate": 2.9616091491132357e-05,
      "loss": 1.3727,
      "step": 21112
    },
    {
      "epoch": 0.7561014915751966,
      "grad_norm": 1.6120564937591553,
      "learning_rate": 2.9607852496211962e-05,
      "loss": 1.554,
      "step": 21113
    },
    {
      "epoch": 0.7561373037047648,
      "grad_norm": 1.4082764387130737,
      "learning_rate": 2.9599614448327084e-05,
      "loss": 1.3681,
      "step": 21114
    },
    {
      "epoch": 0.7561731158343331,
      "grad_norm": 1.5112791061401367,
      "learning_rate": 2.9591377347588623e-05,
      "loss": 1.4086,
      "step": 21115
    },
    {
      "epoch": 0.7562089279639014,
      "grad_norm": 1.5789004564285278,
      "learning_rate": 2.958314119410732e-05,
      "loss": 1.4716,
      "step": 21116
    },
    {
      "epoch": 0.7562447400934696,
      "grad_norm": 1.2509442567825317,
      "learning_rate": 2.9574905987994016e-05,
      "loss": 1.6457,
      "step": 21117
    },
    {
      "epoch": 0.7562805522230379,
      "grad_norm": 2.162203311920166,
      "learning_rate": 2.9566671729359552e-05,
      "loss": 1.665,
      "step": 21118
    },
    {
      "epoch": 0.7563163643526062,
      "grad_norm": 1.7488106489181519,
      "learning_rate": 2.9558438418314626e-05,
      "loss": 1.1802,
      "step": 21119
    },
    {
      "epoch": 0.7563521764821746,
      "grad_norm": 1.4477161169052124,
      "learning_rate": 2.9550206054970063e-05,
      "loss": 1.3837,
      "step": 21120
    },
    {
      "epoch": 0.7563879886117428,
      "grad_norm": 1.7796725034713745,
      "learning_rate": 2.9541974639436588e-05,
      "loss": 1.5306,
      "step": 21121
    },
    {
      "epoch": 0.7564238007413111,
      "grad_norm": 1.4333518743515015,
      "learning_rate": 2.9533744171824996e-05,
      "loss": 1.5737,
      "step": 21122
    },
    {
      "epoch": 0.7564596128708794,
      "grad_norm": 1.4880192279815674,
      "learning_rate": 2.9525514652245922e-05,
      "loss": 1.3167,
      "step": 21123
    },
    {
      "epoch": 0.7564954250004476,
      "grad_norm": 2.2075390815734863,
      "learning_rate": 2.9517286080810204e-05,
      "loss": 1.6719,
      "step": 21124
    },
    {
      "epoch": 0.7565312371300159,
      "grad_norm": 1.4467953443527222,
      "learning_rate": 2.950905845762849e-05,
      "loss": 1.5902,
      "step": 21125
    },
    {
      "epoch": 0.7565670492595842,
      "grad_norm": 1.8242559432983398,
      "learning_rate": 2.9500831782811433e-05,
      "loss": 1.417,
      "step": 21126
    },
    {
      "epoch": 0.7566028613891526,
      "grad_norm": 1.713553547859192,
      "learning_rate": 2.949260605646974e-05,
      "loss": 1.6277,
      "step": 21127
    },
    {
      "epoch": 0.7566386735187208,
      "grad_norm": 2.017822027206421,
      "learning_rate": 2.948438127871409e-05,
      "loss": 1.0925,
      "step": 21128
    },
    {
      "epoch": 0.7566744856482891,
      "grad_norm": 1.8438435792922974,
      "learning_rate": 2.947615744965516e-05,
      "loss": 1.3623,
      "step": 21129
    },
    {
      "epoch": 0.7567102977778574,
      "grad_norm": 1.515105128288269,
      "learning_rate": 2.9467934569403542e-05,
      "loss": 1.3707,
      "step": 21130
    },
    {
      "epoch": 0.7567461099074256,
      "grad_norm": 1.3222929239273071,
      "learning_rate": 2.945971263806987e-05,
      "loss": 1.3478,
      "step": 21131
    },
    {
      "epoch": 0.7567819220369939,
      "grad_norm": 1.3271212577819824,
      "learning_rate": 2.9451491655764816e-05,
      "loss": 1.2689,
      "step": 21132
    },
    {
      "epoch": 0.7568177341665622,
      "grad_norm": 1.8101156949996948,
      "learning_rate": 2.94432716225989e-05,
      "loss": 1.4378,
      "step": 21133
    },
    {
      "epoch": 0.7568535462961306,
      "grad_norm": 1.6955105066299438,
      "learning_rate": 2.943505253868276e-05,
      "loss": 1.4796,
      "step": 21134
    },
    {
      "epoch": 0.7568893584256988,
      "grad_norm": 1.475974440574646,
      "learning_rate": 2.942683440412697e-05,
      "loss": 1.3474,
      "step": 21135
    },
    {
      "epoch": 0.7569251705552671,
      "grad_norm": 1.774215579032898,
      "learning_rate": 2.941861721904211e-05,
      "loss": 1.1609,
      "step": 21136
    },
    {
      "epoch": 0.7569609826848354,
      "grad_norm": 1.5521552562713623,
      "learning_rate": 2.941040098353869e-05,
      "loss": 1.486,
      "step": 21137
    },
    {
      "epoch": 0.7569967948144036,
      "grad_norm": 1.7704689502716064,
      "learning_rate": 2.940218569772726e-05,
      "loss": 1.498,
      "step": 21138
    },
    {
      "epoch": 0.7570326069439719,
      "grad_norm": 2.514124870300293,
      "learning_rate": 2.9393971361718363e-05,
      "loss": 1.3497,
      "step": 21139
    },
    {
      "epoch": 0.7570684190735402,
      "grad_norm": 1.5722203254699707,
      "learning_rate": 2.9385757975622542e-05,
      "loss": 1.5558,
      "step": 21140
    },
    {
      "epoch": 0.7571042312031085,
      "grad_norm": 1.9936293363571167,
      "learning_rate": 2.937754553955022e-05,
      "loss": 1.5372,
      "step": 21141
    },
    {
      "epoch": 0.7571400433326768,
      "grad_norm": 1.4166237115859985,
      "learning_rate": 2.936933405361194e-05,
      "loss": 1.1757,
      "step": 21142
    },
    {
      "epoch": 0.7571758554622451,
      "grad_norm": 1.6292997598648071,
      "learning_rate": 2.936112351791819e-05,
      "loss": 1.3799,
      "step": 21143
    },
    {
      "epoch": 0.7572116675918134,
      "grad_norm": 2.5143439769744873,
      "learning_rate": 2.9352913932579362e-05,
      "loss": 1.3905,
      "step": 21144
    },
    {
      "epoch": 0.7572474797213816,
      "grad_norm": 1.4618525505065918,
      "learning_rate": 2.9344705297705966e-05,
      "loss": 1.3163,
      "step": 21145
    },
    {
      "epoch": 0.7572832918509499,
      "grad_norm": 1.6585017442703247,
      "learning_rate": 2.933649761340841e-05,
      "loss": 1.3903,
      "step": 21146
    },
    {
      "epoch": 0.7573191039805182,
      "grad_norm": 1.43913733959198,
      "learning_rate": 2.932829087979716e-05,
      "loss": 1.3842,
      "step": 21147
    },
    {
      "epoch": 0.7573549161100865,
      "grad_norm": 1.726843237876892,
      "learning_rate": 2.9320085096982575e-05,
      "loss": 1.6791,
      "step": 21148
    },
    {
      "epoch": 0.7573907282396548,
      "grad_norm": 1.5694811344146729,
      "learning_rate": 2.9311880265075066e-05,
      "loss": 1.3672,
      "step": 21149
    },
    {
      "epoch": 0.7574265403692231,
      "grad_norm": 1.3155571222305298,
      "learning_rate": 2.9303676384185064e-05,
      "loss": 1.2628,
      "step": 21150
    },
    {
      "epoch": 0.7574623524987913,
      "grad_norm": 1.3629982471466064,
      "learning_rate": 2.9295473454422863e-05,
      "loss": 1.1417,
      "step": 21151
    },
    {
      "epoch": 0.7574981646283596,
      "grad_norm": 1.683783769607544,
      "learning_rate": 2.928727147589887e-05,
      "loss": 1.7473,
      "step": 21152
    },
    {
      "epoch": 0.7575339767579279,
      "grad_norm": 1.4711257219314575,
      "learning_rate": 2.9279070448723432e-05,
      "loss": 1.245,
      "step": 21153
    },
    {
      "epoch": 0.7575697888874962,
      "grad_norm": 1.780172348022461,
      "learning_rate": 2.927087037300691e-05,
      "loss": 1.3952,
      "step": 21154
    },
    {
      "epoch": 0.7576056010170645,
      "grad_norm": 1.5945377349853516,
      "learning_rate": 2.926267124885955e-05,
      "loss": 1.3441,
      "step": 21155
    },
    {
      "epoch": 0.7576414131466328,
      "grad_norm": 2.19747257232666,
      "learning_rate": 2.9254473076391708e-05,
      "loss": 1.6498,
      "step": 21156
    },
    {
      "epoch": 0.7576772252762011,
      "grad_norm": 1.6467136144638062,
      "learning_rate": 2.924627585571368e-05,
      "loss": 1.4511,
      "step": 21157
    },
    {
      "epoch": 0.7577130374057693,
      "grad_norm": 1.4846833944320679,
      "learning_rate": 2.9238079586935773e-05,
      "loss": 1.4392,
      "step": 21158
    },
    {
      "epoch": 0.7577488495353376,
      "grad_norm": 1.7187808752059937,
      "learning_rate": 2.9229884270168195e-05,
      "loss": 1.3017,
      "step": 21159
    },
    {
      "epoch": 0.7577846616649059,
      "grad_norm": 1.69410240650177,
      "learning_rate": 2.9221689905521245e-05,
      "loss": 1.4055,
      "step": 21160
    },
    {
      "epoch": 0.7578204737944741,
      "grad_norm": 1.315220832824707,
      "learning_rate": 2.9213496493105187e-05,
      "loss": 1.1048,
      "step": 21161
    },
    {
      "epoch": 0.7578562859240425,
      "grad_norm": 1.9386459589004517,
      "learning_rate": 2.920530403303019e-05,
      "loss": 1.5724,
      "step": 21162
    },
    {
      "epoch": 0.7578920980536108,
      "grad_norm": 1.3954391479492188,
      "learning_rate": 2.9197112525406522e-05,
      "loss": 1.0904,
      "step": 21163
    },
    {
      "epoch": 0.7579279101831791,
      "grad_norm": 1.6443402767181396,
      "learning_rate": 2.918892197034436e-05,
      "loss": 1.4831,
      "step": 21164
    },
    {
      "epoch": 0.7579637223127473,
      "grad_norm": 1.8026351928710938,
      "learning_rate": 2.9180732367953956e-05,
      "loss": 1.7182,
      "step": 21165
    },
    {
      "epoch": 0.7579995344423156,
      "grad_norm": 1.4861021041870117,
      "learning_rate": 2.9172543718345413e-05,
      "loss": 1.3764,
      "step": 21166
    },
    {
      "epoch": 0.7580353465718839,
      "grad_norm": 1.567413568496704,
      "learning_rate": 2.9164356021628923e-05,
      "loss": 1.3016,
      "step": 21167
    },
    {
      "epoch": 0.7580711587014521,
      "grad_norm": 1.7095822095870972,
      "learning_rate": 2.915616927791469e-05,
      "loss": 1.23,
      "step": 21168
    },
    {
      "epoch": 0.7581069708310205,
      "grad_norm": 1.373910903930664,
      "learning_rate": 2.9147983487312793e-05,
      "loss": 1.5581,
      "step": 21169
    },
    {
      "epoch": 0.7581427829605888,
      "grad_norm": 1.7896029949188232,
      "learning_rate": 2.913979864993338e-05,
      "loss": 1.3731,
      "step": 21170
    },
    {
      "epoch": 0.7581785950901571,
      "grad_norm": 1.2884318828582764,
      "learning_rate": 2.9131614765886573e-05,
      "loss": 1.4324,
      "step": 21171
    },
    {
      "epoch": 0.7582144072197253,
      "grad_norm": 1.8694905042648315,
      "learning_rate": 2.912343183528251e-05,
      "loss": 1.6948,
      "step": 21172
    },
    {
      "epoch": 0.7582502193492936,
      "grad_norm": 2.111467123031616,
      "learning_rate": 2.9115249858231207e-05,
      "loss": 1.6019,
      "step": 21173
    },
    {
      "epoch": 0.7582860314788619,
      "grad_norm": 1.7151230573654175,
      "learning_rate": 2.91070688348428e-05,
      "loss": 1.6059,
      "step": 21174
    },
    {
      "epoch": 0.7583218436084301,
      "grad_norm": 1.535046935081482,
      "learning_rate": 2.9098888765227316e-05,
      "loss": 1.5526,
      "step": 21175
    },
    {
      "epoch": 0.7583576557379985,
      "grad_norm": 1.7479262351989746,
      "learning_rate": 2.9090709649494873e-05,
      "loss": 1.4793,
      "step": 21176
    },
    {
      "epoch": 0.7583934678675668,
      "grad_norm": 1.8973602056503296,
      "learning_rate": 2.908253148775546e-05,
      "loss": 1.8296,
      "step": 21177
    },
    {
      "epoch": 0.758429279997135,
      "grad_norm": 1.2485648393630981,
      "learning_rate": 2.9074354280119042e-05,
      "loss": 1.2522,
      "step": 21178
    },
    {
      "epoch": 0.7584650921267033,
      "grad_norm": 1.4664356708526611,
      "learning_rate": 2.9066178026695767e-05,
      "loss": 1.2564,
      "step": 21179
    },
    {
      "epoch": 0.7585009042562716,
      "grad_norm": 1.1946889162063599,
      "learning_rate": 2.9058002727595546e-05,
      "loss": 1.536,
      "step": 21180
    },
    {
      "epoch": 0.7585367163858399,
      "grad_norm": 1.3045967817306519,
      "learning_rate": 2.904982838292838e-05,
      "loss": 1.2439,
      "step": 21181
    },
    {
      "epoch": 0.7585725285154081,
      "grad_norm": 1.5080360174179077,
      "learning_rate": 2.9041654992804256e-05,
      "loss": 1.6046,
      "step": 21182
    },
    {
      "epoch": 0.7586083406449765,
      "grad_norm": 1.595199704170227,
      "learning_rate": 2.9033482557333158e-05,
      "loss": 1.3323,
      "step": 21183
    },
    {
      "epoch": 0.7586441527745448,
      "grad_norm": 1.783447265625,
      "learning_rate": 2.9025311076624994e-05,
      "loss": 1.5985,
      "step": 21184
    },
    {
      "epoch": 0.758679964904113,
      "grad_norm": 1.9217463731765747,
      "learning_rate": 2.9017140550789713e-05,
      "loss": 1.1523,
      "step": 21185
    },
    {
      "epoch": 0.7587157770336813,
      "grad_norm": 1.8757216930389404,
      "learning_rate": 2.9008970979937276e-05,
      "loss": 1.3591,
      "step": 21186
    },
    {
      "epoch": 0.7587515891632496,
      "grad_norm": 1.840748906135559,
      "learning_rate": 2.9000802364177527e-05,
      "loss": 1.3812,
      "step": 21187
    },
    {
      "epoch": 0.7587874012928179,
      "grad_norm": 1.5982416868209839,
      "learning_rate": 2.8992634703620437e-05,
      "loss": 1.5119,
      "step": 21188
    },
    {
      "epoch": 0.7588232134223861,
      "grad_norm": 1.6878429651260376,
      "learning_rate": 2.8984467998375786e-05,
      "loss": 1.0781,
      "step": 21189
    },
    {
      "epoch": 0.7588590255519545,
      "grad_norm": 1.7874064445495605,
      "learning_rate": 2.8976302248553576e-05,
      "loss": 1.2698,
      "step": 21190
    },
    {
      "epoch": 0.7588948376815228,
      "grad_norm": 1.8459925651550293,
      "learning_rate": 2.896813745426359e-05,
      "loss": 1.1503,
      "step": 21191
    },
    {
      "epoch": 0.758930649811091,
      "grad_norm": 1.5828872919082642,
      "learning_rate": 2.8959973615615675e-05,
      "loss": 1.2534,
      "step": 21192
    },
    {
      "epoch": 0.7589664619406593,
      "grad_norm": 2.555840492248535,
      "learning_rate": 2.8951810732719685e-05,
      "loss": 1.5415,
      "step": 21193
    },
    {
      "epoch": 0.7590022740702276,
      "grad_norm": 2.2074623107910156,
      "learning_rate": 2.8943648805685464e-05,
      "loss": 1.4024,
      "step": 21194
    },
    {
      "epoch": 0.7590380861997958,
      "grad_norm": 2.5466811656951904,
      "learning_rate": 2.893548783462279e-05,
      "loss": 1.4226,
      "step": 21195
    },
    {
      "epoch": 0.7590738983293641,
      "grad_norm": 1.3953499794006348,
      "learning_rate": 2.8927327819641403e-05,
      "loss": 1.4408,
      "step": 21196
    },
    {
      "epoch": 0.7591097104589324,
      "grad_norm": 1.9189091920852661,
      "learning_rate": 2.8919168760851202e-05,
      "loss": 1.2927,
      "step": 21197
    },
    {
      "epoch": 0.7591455225885008,
      "grad_norm": 1.6638646125793457,
      "learning_rate": 2.891101065836187e-05,
      "loss": 1.5423,
      "step": 21198
    },
    {
      "epoch": 0.759181334718069,
      "grad_norm": 1.5958912372589111,
      "learning_rate": 2.8902853512283225e-05,
      "loss": 1.3269,
      "step": 21199
    },
    {
      "epoch": 0.7592171468476373,
      "grad_norm": 2.3823351860046387,
      "learning_rate": 2.8894697322724908e-05,
      "loss": 1.2585,
      "step": 21200
    },
    {
      "epoch": 0.7592529589772056,
      "grad_norm": 1.484283208847046,
      "learning_rate": 2.8886542089796785e-05,
      "loss": 1.3847,
      "step": 21201
    },
    {
      "epoch": 0.7592887711067738,
      "grad_norm": 1.6040844917297363,
      "learning_rate": 2.8878387813608477e-05,
      "loss": 1.3717,
      "step": 21202
    },
    {
      "epoch": 0.7593245832363421,
      "grad_norm": 1.6195263862609863,
      "learning_rate": 2.8870234494269756e-05,
      "loss": 1.4155,
      "step": 21203
    },
    {
      "epoch": 0.7593603953659104,
      "grad_norm": 1.518485188484192,
      "learning_rate": 2.8862082131890243e-05,
      "loss": 1.3777,
      "step": 21204
    },
    {
      "epoch": 0.7593962074954788,
      "grad_norm": 2.1455929279327393,
      "learning_rate": 2.885393072657966e-05,
      "loss": 1.7861,
      "step": 21205
    },
    {
      "epoch": 0.759432019625047,
      "grad_norm": 1.6914993524551392,
      "learning_rate": 2.8845780278447688e-05,
      "loss": 1.2118,
      "step": 21206
    },
    {
      "epoch": 0.7594678317546153,
      "grad_norm": 1.3727785348892212,
      "learning_rate": 2.8837630787603908e-05,
      "loss": 1.2499,
      "step": 21207
    },
    {
      "epoch": 0.7595036438841836,
      "grad_norm": 1.734089732170105,
      "learning_rate": 2.882948225415807e-05,
      "loss": 1.5871,
      "step": 21208
    },
    {
      "epoch": 0.7595394560137518,
      "grad_norm": 1.2891333103179932,
      "learning_rate": 2.8821334678219712e-05,
      "loss": 1.2715,
      "step": 21209
    },
    {
      "epoch": 0.7595752681433201,
      "grad_norm": 2.27925443649292,
      "learning_rate": 2.8813188059898512e-05,
      "loss": 1.3765,
      "step": 21210
    },
    {
      "epoch": 0.7596110802728884,
      "grad_norm": 1.3303371667861938,
      "learning_rate": 2.8805042399303984e-05,
      "loss": 1.5651,
      "step": 21211
    },
    {
      "epoch": 0.7596468924024568,
      "grad_norm": 2.1135172843933105,
      "learning_rate": 2.8796897696545832e-05,
      "loss": 1.6908,
      "step": 21212
    },
    {
      "epoch": 0.759682704532025,
      "grad_norm": 1.5996071100234985,
      "learning_rate": 2.878875395173358e-05,
      "loss": 1.4533,
      "step": 21213
    },
    {
      "epoch": 0.7597185166615933,
      "grad_norm": 1.5009126663208008,
      "learning_rate": 2.8780611164976767e-05,
      "loss": 1.2599,
      "step": 21214
    },
    {
      "epoch": 0.7597543287911616,
      "grad_norm": 1.407975196838379,
      "learning_rate": 2.8772469336384954e-05,
      "loss": 1.2601,
      "step": 21215
    },
    {
      "epoch": 0.7597901409207298,
      "grad_norm": 1.3562133312225342,
      "learning_rate": 2.876432846606769e-05,
      "loss": 1.3456,
      "step": 21216
    },
    {
      "epoch": 0.7598259530502981,
      "grad_norm": 1.6135796308517456,
      "learning_rate": 2.8756188554134522e-05,
      "loss": 1.2708,
      "step": 21217
    },
    {
      "epoch": 0.7598617651798664,
      "grad_norm": 1.496204137802124,
      "learning_rate": 2.8748049600694893e-05,
      "loss": 1.3726,
      "step": 21218
    },
    {
      "epoch": 0.7598975773094347,
      "grad_norm": 2.0687203407287598,
      "learning_rate": 2.8739911605858394e-05,
      "loss": 1.3556,
      "step": 21219
    },
    {
      "epoch": 0.759933389439003,
      "grad_norm": 1.4460134506225586,
      "learning_rate": 2.873177456973445e-05,
      "loss": 1.1964,
      "step": 21220
    },
    {
      "epoch": 0.7599692015685713,
      "grad_norm": 1.672918677330017,
      "learning_rate": 2.872363849243257e-05,
      "loss": 1.3602,
      "step": 21221
    },
    {
      "epoch": 0.7600050136981396,
      "grad_norm": 1.2204346656799316,
      "learning_rate": 2.871550337406217e-05,
      "loss": 1.1142,
      "step": 21222
    },
    {
      "epoch": 0.7600408258277078,
      "grad_norm": 1.909233808517456,
      "learning_rate": 2.8707369214732716e-05,
      "loss": 1.4387,
      "step": 21223
    },
    {
      "epoch": 0.7600766379572761,
      "grad_norm": 1.3852407932281494,
      "learning_rate": 2.8699236014553686e-05,
      "loss": 1.5051,
      "step": 21224
    },
    {
      "epoch": 0.7601124500868444,
      "grad_norm": 1.9627448320388794,
      "learning_rate": 2.869110377363443e-05,
      "loss": 1.5291,
      "step": 21225
    },
    {
      "epoch": 0.7601482622164127,
      "grad_norm": 1.5735957622528076,
      "learning_rate": 2.868297249208438e-05,
      "loss": 1.5355,
      "step": 21226
    },
    {
      "epoch": 0.760184074345981,
      "grad_norm": 1.7588822841644287,
      "learning_rate": 2.867484217001296e-05,
      "loss": 1.5529,
      "step": 21227
    },
    {
      "epoch": 0.7602198864755493,
      "grad_norm": 1.5545251369476318,
      "learning_rate": 2.866671280752956e-05,
      "loss": 1.4601,
      "step": 21228
    },
    {
      "epoch": 0.7602556986051175,
      "grad_norm": 1.5367778539657593,
      "learning_rate": 2.8658584404743493e-05,
      "loss": 1.306,
      "step": 21229
    },
    {
      "epoch": 0.7602915107346858,
      "grad_norm": 1.8365488052368164,
      "learning_rate": 2.865045696176415e-05,
      "loss": 1.3618,
      "step": 21230
    },
    {
      "epoch": 0.7603273228642541,
      "grad_norm": 1.5795180797576904,
      "learning_rate": 2.8642330478700908e-05,
      "loss": 1.201,
      "step": 21231
    },
    {
      "epoch": 0.7603631349938224,
      "grad_norm": 2.363741397857666,
      "learning_rate": 2.8634204955663024e-05,
      "loss": 1.4861,
      "step": 21232
    },
    {
      "epoch": 0.7603989471233907,
      "grad_norm": 1.6575523614883423,
      "learning_rate": 2.862608039275987e-05,
      "loss": 1.3186,
      "step": 21233
    },
    {
      "epoch": 0.760434759252959,
      "grad_norm": 1.3435629606246948,
      "learning_rate": 2.861795679010073e-05,
      "loss": 1.7503,
      "step": 21234
    },
    {
      "epoch": 0.7604705713825273,
      "grad_norm": 1.258905291557312,
      "learning_rate": 2.8609834147794945e-05,
      "loss": 1.1639,
      "step": 21235
    },
    {
      "epoch": 0.7605063835120955,
      "grad_norm": 2.6588869094848633,
      "learning_rate": 2.8601712465951713e-05,
      "loss": 1.2825,
      "step": 21236
    },
    {
      "epoch": 0.7605421956416638,
      "grad_norm": 1.9487318992614746,
      "learning_rate": 2.8593591744680348e-05,
      "loss": 1.6677,
      "step": 21237
    },
    {
      "epoch": 0.7605780077712321,
      "grad_norm": 1.674471139907837,
      "learning_rate": 2.858547198409013e-05,
      "loss": 1.2264,
      "step": 21238
    },
    {
      "epoch": 0.7606138199008003,
      "grad_norm": 1.990461826324463,
      "learning_rate": 2.8577353184290236e-05,
      "loss": 1.341,
      "step": 21239
    },
    {
      "epoch": 0.7606496320303687,
      "grad_norm": 2.012136220932007,
      "learning_rate": 2.8569235345389922e-05,
      "loss": 1.583,
      "step": 21240
    },
    {
      "epoch": 0.760685444159937,
      "grad_norm": 1.3574470281600952,
      "learning_rate": 2.8561118467498415e-05,
      "loss": 1.6975,
      "step": 21241
    },
    {
      "epoch": 0.7607212562895053,
      "grad_norm": 1.79666006565094,
      "learning_rate": 2.855300255072494e-05,
      "loss": 1.5932,
      "step": 21242
    },
    {
      "epoch": 0.7607570684190735,
      "grad_norm": 1.7653344869613647,
      "learning_rate": 2.8544887595178616e-05,
      "loss": 1.4088,
      "step": 21243
    },
    {
      "epoch": 0.7607928805486418,
      "grad_norm": 1.3888636827468872,
      "learning_rate": 2.853677360096867e-05,
      "loss": 1.5872,
      "step": 21244
    },
    {
      "epoch": 0.7608286926782101,
      "grad_norm": 1.8041224479675293,
      "learning_rate": 2.8528660568204247e-05,
      "loss": 1.7292,
      "step": 21245
    },
    {
      "epoch": 0.7608645048077783,
      "grad_norm": 1.383401870727539,
      "learning_rate": 2.8520548496994536e-05,
      "loss": 1.2626,
      "step": 21246
    },
    {
      "epoch": 0.7609003169373467,
      "grad_norm": 1.8830630779266357,
      "learning_rate": 2.851243738744862e-05,
      "loss": 1.5501,
      "step": 21247
    },
    {
      "epoch": 0.760936129066915,
      "grad_norm": 2.0602715015411377,
      "learning_rate": 2.8504327239675645e-05,
      "loss": 1.3759,
      "step": 21248
    },
    {
      "epoch": 0.7609719411964833,
      "grad_norm": 1.9153468608856201,
      "learning_rate": 2.849621805378474e-05,
      "loss": 1.3334,
      "step": 21249
    },
    {
      "epoch": 0.7610077533260515,
      "grad_norm": 1.9879709482192993,
      "learning_rate": 2.848810982988497e-05,
      "loss": 1.5984,
      "step": 21250
    },
    {
      "epoch": 0.7610435654556198,
      "grad_norm": 1.8865492343902588,
      "learning_rate": 2.848000256808544e-05,
      "loss": 1.3962,
      "step": 21251
    },
    {
      "epoch": 0.7610793775851881,
      "grad_norm": 2.2912323474884033,
      "learning_rate": 2.8471896268495214e-05,
      "loss": 1.5664,
      "step": 21252
    },
    {
      "epoch": 0.7611151897147563,
      "grad_norm": 1.8442264795303345,
      "learning_rate": 2.84637909312234e-05,
      "loss": 1.3058,
      "step": 21253
    },
    {
      "epoch": 0.7611510018443247,
      "grad_norm": 2.1874215602874756,
      "learning_rate": 2.845568655637896e-05,
      "loss": 1.6448,
      "step": 21254
    },
    {
      "epoch": 0.761186813973893,
      "grad_norm": 1.9503223896026611,
      "learning_rate": 2.844758314407098e-05,
      "loss": 1.1485,
      "step": 21255
    },
    {
      "epoch": 0.7612226261034613,
      "grad_norm": 1.5357619524002075,
      "learning_rate": 2.8439480694408506e-05,
      "loss": 1.5472,
      "step": 21256
    },
    {
      "epoch": 0.7612584382330295,
      "grad_norm": 1.703912377357483,
      "learning_rate": 2.8431379207500476e-05,
      "loss": 1.6161,
      "step": 21257
    },
    {
      "epoch": 0.7612942503625978,
      "grad_norm": 2.0555579662323,
      "learning_rate": 2.8423278683455922e-05,
      "loss": 1.7808,
      "step": 21258
    },
    {
      "epoch": 0.7613300624921661,
      "grad_norm": 1.4263310432434082,
      "learning_rate": 2.8415179122383828e-05,
      "loss": 1.0745,
      "step": 21259
    },
    {
      "epoch": 0.7613658746217343,
      "grad_norm": 2.0803184509277344,
      "learning_rate": 2.840708052439319e-05,
      "loss": 1.5224,
      "step": 21260
    },
    {
      "epoch": 0.7614016867513027,
      "grad_norm": 1.6625678539276123,
      "learning_rate": 2.8398982889592908e-05,
      "loss": 1.2313,
      "step": 21261
    },
    {
      "epoch": 0.761437498880871,
      "grad_norm": 1.6900227069854736,
      "learning_rate": 2.839088621809195e-05,
      "loss": 1.4897,
      "step": 21262
    },
    {
      "epoch": 0.7614733110104392,
      "grad_norm": 1.6262860298156738,
      "learning_rate": 2.8382790509999257e-05,
      "loss": 1.4142,
      "step": 21263
    },
    {
      "epoch": 0.7615091231400075,
      "grad_norm": 1.7828209400177002,
      "learning_rate": 2.8374695765423753e-05,
      "loss": 1.7341,
      "step": 21264
    },
    {
      "epoch": 0.7615449352695758,
      "grad_norm": 2.0078964233398438,
      "learning_rate": 2.8366601984474305e-05,
      "loss": 1.5635,
      "step": 21265
    },
    {
      "epoch": 0.761580747399144,
      "grad_norm": 2.1161463260650635,
      "learning_rate": 2.835850916725983e-05,
      "loss": 1.4108,
      "step": 21266
    },
    {
      "epoch": 0.7616165595287123,
      "grad_norm": 1.6242632865905762,
      "learning_rate": 2.8350417313889233e-05,
      "loss": 1.0424,
      "step": 21267
    },
    {
      "epoch": 0.7616523716582807,
      "grad_norm": 1.9338126182556152,
      "learning_rate": 2.8342326424471323e-05,
      "loss": 1.4046,
      "step": 21268
    },
    {
      "epoch": 0.761688183787849,
      "grad_norm": 1.9795308113098145,
      "learning_rate": 2.8334236499114963e-05,
      "loss": 1.5743,
      "step": 21269
    },
    {
      "epoch": 0.7617239959174172,
      "grad_norm": 1.2497432231903076,
      "learning_rate": 2.8326147537929027e-05,
      "loss": 1.5952,
      "step": 21270
    },
    {
      "epoch": 0.7617598080469855,
      "grad_norm": 1.5189738273620605,
      "learning_rate": 2.8318059541022346e-05,
      "loss": 1.6455,
      "step": 21271
    },
    {
      "epoch": 0.7617956201765538,
      "grad_norm": 1.9544116258621216,
      "learning_rate": 2.830997250850368e-05,
      "loss": 1.438,
      "step": 21272
    },
    {
      "epoch": 0.761831432306122,
      "grad_norm": 1.6780002117156982,
      "learning_rate": 2.8301886440481862e-05,
      "loss": 1.2535,
      "step": 21273
    },
    {
      "epoch": 0.7618672444356903,
      "grad_norm": 2.1269612312316895,
      "learning_rate": 2.8293801337065705e-05,
      "loss": 1.2494,
      "step": 21274
    },
    {
      "epoch": 0.7619030565652587,
      "grad_norm": 2.0308775901794434,
      "learning_rate": 2.8285717198363924e-05,
      "loss": 1.6469,
      "step": 21275
    },
    {
      "epoch": 0.761938868694827,
      "grad_norm": 2.287074565887451,
      "learning_rate": 2.8277634024485322e-05,
      "loss": 1.2689,
      "step": 21276
    },
    {
      "epoch": 0.7619746808243952,
      "grad_norm": 1.8691420555114746,
      "learning_rate": 2.826955181553863e-05,
      "loss": 1.2943,
      "step": 21277
    },
    {
      "epoch": 0.7620104929539635,
      "grad_norm": 2.088134765625,
      "learning_rate": 2.826147057163263e-05,
      "loss": 1.2476,
      "step": 21278
    },
    {
      "epoch": 0.7620463050835318,
      "grad_norm": 2.5623204708099365,
      "learning_rate": 2.8253390292875982e-05,
      "loss": 1.7233,
      "step": 21279
    },
    {
      "epoch": 0.7620821172131,
      "grad_norm": 1.437081217765808,
      "learning_rate": 2.8245310979377416e-05,
      "loss": 1.507,
      "step": 21280
    },
    {
      "epoch": 0.7621179293426683,
      "grad_norm": 1.5275484323501587,
      "learning_rate": 2.8237232631245624e-05,
      "loss": 1.5571,
      "step": 21281
    },
    {
      "epoch": 0.7621537414722367,
      "grad_norm": 2.310119152069092,
      "learning_rate": 2.8229155248589345e-05,
      "loss": 1.687,
      "step": 21282
    },
    {
      "epoch": 0.762189553601805,
      "grad_norm": 1.3541539907455444,
      "learning_rate": 2.822107883151719e-05,
      "loss": 1.5511,
      "step": 21283
    },
    {
      "epoch": 0.7622253657313732,
      "grad_norm": 1.8010655641555786,
      "learning_rate": 2.8213003380137783e-05,
      "loss": 1.6483,
      "step": 21284
    },
    {
      "epoch": 0.7622611778609415,
      "grad_norm": 1.4328798055648804,
      "learning_rate": 2.820492889455987e-05,
      "loss": 1.486,
      "step": 21285
    },
    {
      "epoch": 0.7622969899905098,
      "grad_norm": 1.6243557929992676,
      "learning_rate": 2.8196855374892006e-05,
      "loss": 1.6467,
      "step": 21286
    },
    {
      "epoch": 0.762332802120078,
      "grad_norm": 1.6708301305770874,
      "learning_rate": 2.8188782821242855e-05,
      "loss": 1.6195,
      "step": 21287
    },
    {
      "epoch": 0.7623686142496463,
      "grad_norm": 1.5732182264328003,
      "learning_rate": 2.8180711233720947e-05,
      "loss": 1.3332,
      "step": 21288
    },
    {
      "epoch": 0.7624044263792147,
      "grad_norm": 1.813924789428711,
      "learning_rate": 2.8172640612434987e-05,
      "loss": 1.285,
      "step": 21289
    },
    {
      "epoch": 0.762440238508783,
      "grad_norm": 2.3234944343566895,
      "learning_rate": 2.8164570957493473e-05,
      "loss": 1.3432,
      "step": 21290
    },
    {
      "epoch": 0.7624760506383512,
      "grad_norm": 1.5255062580108643,
      "learning_rate": 2.8156502269004992e-05,
      "loss": 1.472,
      "step": 21291
    },
    {
      "epoch": 0.7625118627679195,
      "grad_norm": 2.036771535873413,
      "learning_rate": 2.814843454707813e-05,
      "loss": 1.3433,
      "step": 21292
    },
    {
      "epoch": 0.7625476748974878,
      "grad_norm": 2.0992271900177,
      "learning_rate": 2.8140367791821363e-05,
      "loss": 1.2408,
      "step": 21293
    },
    {
      "epoch": 0.762583487027056,
      "grad_norm": 1.3696377277374268,
      "learning_rate": 2.813230200334329e-05,
      "loss": 1.1414,
      "step": 21294
    },
    {
      "epoch": 0.7626192991566243,
      "grad_norm": 2.283036231994629,
      "learning_rate": 2.8124237181752334e-05,
      "loss": 1.5427,
      "step": 21295
    },
    {
      "epoch": 0.7626551112861927,
      "grad_norm": 1.7576613426208496,
      "learning_rate": 2.8116173327157114e-05,
      "loss": 1.338,
      "step": 21296
    },
    {
      "epoch": 0.762690923415761,
      "grad_norm": 1.6552729606628418,
      "learning_rate": 2.8108110439666024e-05,
      "loss": 1.3773,
      "step": 21297
    },
    {
      "epoch": 0.7627267355453292,
      "grad_norm": 1.3806850910186768,
      "learning_rate": 2.8100048519387613e-05,
      "loss": 1.3199,
      "step": 21298
    },
    {
      "epoch": 0.7627625476748975,
      "grad_norm": 1.8089537620544434,
      "learning_rate": 2.8091987566430233e-05,
      "loss": 1.2466,
      "step": 21299
    },
    {
      "epoch": 0.7627983598044658,
      "grad_norm": 1.5321838855743408,
      "learning_rate": 2.808392758090247e-05,
      "loss": 1.4227,
      "step": 21300
    },
    {
      "epoch": 0.762834171934034,
      "grad_norm": 1.6883888244628906,
      "learning_rate": 2.80758685629127e-05,
      "loss": 1.4317,
      "step": 21301
    },
    {
      "epoch": 0.7628699840636023,
      "grad_norm": 1.5351834297180176,
      "learning_rate": 2.8067810512569282e-05,
      "loss": 1.1779,
      "step": 21302
    },
    {
      "epoch": 0.7629057961931707,
      "grad_norm": 2.15999174118042,
      "learning_rate": 2.805975342998075e-05,
      "loss": 1.4464,
      "step": 21303
    },
    {
      "epoch": 0.7629416083227389,
      "grad_norm": 1.7024953365325928,
      "learning_rate": 2.80516973152554e-05,
      "loss": 1.3371,
      "step": 21304
    },
    {
      "epoch": 0.7629774204523072,
      "grad_norm": 1.595112919807434,
      "learning_rate": 2.8043642168501692e-05,
      "loss": 1.4669,
      "step": 21305
    },
    {
      "epoch": 0.7630132325818755,
      "grad_norm": 1.7060620784759521,
      "learning_rate": 2.8035587989827904e-05,
      "loss": 1.2248,
      "step": 21306
    },
    {
      "epoch": 0.7630490447114437,
      "grad_norm": 1.508658766746521,
      "learning_rate": 2.802753477934251e-05,
      "loss": 1.4452,
      "step": 21307
    },
    {
      "epoch": 0.763084856841012,
      "grad_norm": 3.327977180480957,
      "learning_rate": 2.8019482537153762e-05,
      "loss": 1.7054,
      "step": 21308
    },
    {
      "epoch": 0.7631206689705803,
      "grad_norm": 1.5800703763961792,
      "learning_rate": 2.801143126337007e-05,
      "loss": 1.508,
      "step": 21309
    },
    {
      "epoch": 0.7631564811001487,
      "grad_norm": 1.645244836807251,
      "learning_rate": 2.8003380958099677e-05,
      "loss": 1.5809,
      "step": 21310
    },
    {
      "epoch": 0.7631922932297169,
      "grad_norm": 1.597456455230713,
      "learning_rate": 2.7995331621450917e-05,
      "loss": 1.2949,
      "step": 21311
    },
    {
      "epoch": 0.7632281053592852,
      "grad_norm": 1.4710382223129272,
      "learning_rate": 2.7987283253532125e-05,
      "loss": 1.1875,
      "step": 21312
    },
    {
      "epoch": 0.7632639174888535,
      "grad_norm": 2.0475046634674072,
      "learning_rate": 2.7979235854451523e-05,
      "loss": 1.7837,
      "step": 21313
    },
    {
      "epoch": 0.7632997296184217,
      "grad_norm": 1.670210599899292,
      "learning_rate": 2.79711894243174e-05,
      "loss": 1.6562,
      "step": 21314
    },
    {
      "epoch": 0.76333554174799,
      "grad_norm": 1.4853568077087402,
      "learning_rate": 2.7963143963238005e-05,
      "loss": 1.443,
      "step": 21315
    },
    {
      "epoch": 0.7633713538775583,
      "grad_norm": 1.7376928329467773,
      "learning_rate": 2.795509947132162e-05,
      "loss": 1.2882,
      "step": 21316
    },
    {
      "epoch": 0.7634071660071267,
      "grad_norm": 1.8859137296676636,
      "learning_rate": 2.7947055948676392e-05,
      "loss": 1.7188,
      "step": 21317
    },
    {
      "epoch": 0.7634429781366949,
      "grad_norm": 1.6725388765335083,
      "learning_rate": 2.793901339541063e-05,
      "loss": 1.4696,
      "step": 21318
    },
    {
      "epoch": 0.7634787902662632,
      "grad_norm": 1.5267573595046997,
      "learning_rate": 2.79309718116325e-05,
      "loss": 1.3887,
      "step": 21319
    },
    {
      "epoch": 0.7635146023958315,
      "grad_norm": 2.08896541595459,
      "learning_rate": 2.792293119745014e-05,
      "loss": 1.4965,
      "step": 21320
    },
    {
      "epoch": 0.7635504145253997,
      "grad_norm": 1.5745618343353271,
      "learning_rate": 2.7914891552971776e-05,
      "loss": 1.5053,
      "step": 21321
    },
    {
      "epoch": 0.763586226654968,
      "grad_norm": 1.7663930654525757,
      "learning_rate": 2.7906852878305567e-05,
      "loss": 1.431,
      "step": 21322
    },
    {
      "epoch": 0.7636220387845363,
      "grad_norm": 1.428335428237915,
      "learning_rate": 2.789881517355969e-05,
      "loss": 1.6087,
      "step": 21323
    },
    {
      "epoch": 0.7636578509141047,
      "grad_norm": 1.546515703201294,
      "learning_rate": 2.7890778438842214e-05,
      "loss": 1.1567,
      "step": 21324
    },
    {
      "epoch": 0.7636936630436729,
      "grad_norm": 2.3970510959625244,
      "learning_rate": 2.7882742674261307e-05,
      "loss": 1.8865,
      "step": 21325
    },
    {
      "epoch": 0.7637294751732412,
      "grad_norm": 1.486477255821228,
      "learning_rate": 2.78747078799251e-05,
      "loss": 1.5473,
      "step": 21326
    },
    {
      "epoch": 0.7637652873028095,
      "grad_norm": 1.832295298576355,
      "learning_rate": 2.786667405594163e-05,
      "loss": 1.3825,
      "step": 21327
    },
    {
      "epoch": 0.7638010994323777,
      "grad_norm": 1.7424432039260864,
      "learning_rate": 2.785864120241901e-05,
      "loss": 1.503,
      "step": 21328
    },
    {
      "epoch": 0.763836911561946,
      "grad_norm": 1.831028938293457,
      "learning_rate": 2.7850609319465325e-05,
      "loss": 1.5769,
      "step": 21329
    },
    {
      "epoch": 0.7638727236915143,
      "grad_norm": 1.4432202577590942,
      "learning_rate": 2.7842578407188656e-05,
      "loss": 1.3591,
      "step": 21330
    },
    {
      "epoch": 0.7639085358210826,
      "grad_norm": 1.3321956396102905,
      "learning_rate": 2.7834548465696987e-05,
      "loss": 1.4144,
      "step": 21331
    },
    {
      "epoch": 0.7639443479506509,
      "grad_norm": 1.7596930265426636,
      "learning_rate": 2.7826519495098378e-05,
      "loss": 1.4145,
      "step": 21332
    },
    {
      "epoch": 0.7639801600802192,
      "grad_norm": 1.5014865398406982,
      "learning_rate": 2.7818491495500864e-05,
      "loss": 1.4459,
      "step": 21333
    },
    {
      "epoch": 0.7640159722097875,
      "grad_norm": 2.288329601287842,
      "learning_rate": 2.7810464467012455e-05,
      "loss": 1.7881,
      "step": 21334
    },
    {
      "epoch": 0.7640517843393557,
      "grad_norm": 1.4313377141952515,
      "learning_rate": 2.7802438409741106e-05,
      "loss": 1.5355,
      "step": 21335
    },
    {
      "epoch": 0.764087596468924,
      "grad_norm": 1.7000243663787842,
      "learning_rate": 2.7794413323794822e-05,
      "loss": 1.3548,
      "step": 21336
    },
    {
      "epoch": 0.7641234085984923,
      "grad_norm": 1.5719764232635498,
      "learning_rate": 2.7786389209281592e-05,
      "loss": 1.3211,
      "step": 21337
    },
    {
      "epoch": 0.7641592207280606,
      "grad_norm": 1.5126636028289795,
      "learning_rate": 2.7778366066309326e-05,
      "loss": 1.3154,
      "step": 21338
    },
    {
      "epoch": 0.7641950328576289,
      "grad_norm": 1.7162196636199951,
      "learning_rate": 2.7770343894985974e-05,
      "loss": 1.4414,
      "step": 21339
    },
    {
      "epoch": 0.7642308449871972,
      "grad_norm": 1.7152540683746338,
      "learning_rate": 2.7762322695419485e-05,
      "loss": 1.9311,
      "step": 21340
    },
    {
      "epoch": 0.7642666571167654,
      "grad_norm": 2.3539340496063232,
      "learning_rate": 2.7754302467717785e-05,
      "loss": 1.6651,
      "step": 21341
    },
    {
      "epoch": 0.7643024692463337,
      "grad_norm": 1.1298686265945435,
      "learning_rate": 2.7746283211988734e-05,
      "loss": 1.4936,
      "step": 21342
    },
    {
      "epoch": 0.764338281375902,
      "grad_norm": 1.6960740089416504,
      "learning_rate": 2.773826492834023e-05,
      "loss": 1.3619,
      "step": 21343
    },
    {
      "epoch": 0.7643740935054703,
      "grad_norm": 1.8697292804718018,
      "learning_rate": 2.77302476168802e-05,
      "loss": 1.4366,
      "step": 21344
    },
    {
      "epoch": 0.7644099056350386,
      "grad_norm": 1.8246264457702637,
      "learning_rate": 2.7722231277716437e-05,
      "loss": 1.3414,
      "step": 21345
    },
    {
      "epoch": 0.7644457177646069,
      "grad_norm": 2.1910572052001953,
      "learning_rate": 2.771421591095682e-05,
      "loss": 1.8065,
      "step": 21346
    },
    {
      "epoch": 0.7644815298941752,
      "grad_norm": 1.7753005027770996,
      "learning_rate": 2.7706201516709175e-05,
      "loss": 1.0744,
      "step": 21347
    },
    {
      "epoch": 0.7645173420237434,
      "grad_norm": 1.61532723903656,
      "learning_rate": 2.769818809508138e-05,
      "loss": 1.5135,
      "step": 21348
    },
    {
      "epoch": 0.7645531541533117,
      "grad_norm": 1.4952168464660645,
      "learning_rate": 2.769017564618117e-05,
      "loss": 1.1487,
      "step": 21349
    },
    {
      "epoch": 0.76458896628288,
      "grad_norm": 1.5021734237670898,
      "learning_rate": 2.7682164170116365e-05,
      "loss": 1.6482,
      "step": 21350
    },
    {
      "epoch": 0.7646247784124482,
      "grad_norm": 1.436759114265442,
      "learning_rate": 2.767415366699476e-05,
      "loss": 1.1454,
      "step": 21351
    },
    {
      "epoch": 0.7646605905420166,
      "grad_norm": 1.3185434341430664,
      "learning_rate": 2.7666144136924166e-05,
      "loss": 1.3639,
      "step": 21352
    },
    {
      "epoch": 0.7646964026715849,
      "grad_norm": 1.667108416557312,
      "learning_rate": 2.7658135580012256e-05,
      "loss": 1.135,
      "step": 21353
    },
    {
      "epoch": 0.7647322148011532,
      "grad_norm": 1.9361541271209717,
      "learning_rate": 2.7650127996366826e-05,
      "loss": 1.2538,
      "step": 21354
    },
    {
      "epoch": 0.7647680269307214,
      "grad_norm": 1.9174108505249023,
      "learning_rate": 2.764212138609562e-05,
      "loss": 1.5039,
      "step": 21355
    },
    {
      "epoch": 0.7648038390602897,
      "grad_norm": 2.271050214767456,
      "learning_rate": 2.7634115749306312e-05,
      "loss": 1.4584,
      "step": 21356
    },
    {
      "epoch": 0.764839651189858,
      "grad_norm": 1.594559907913208,
      "learning_rate": 2.762611108610663e-05,
      "loss": 1.1316,
      "step": 21357
    },
    {
      "epoch": 0.7648754633194262,
      "grad_norm": 1.610710859298706,
      "learning_rate": 2.7618107396604263e-05,
      "loss": 1.5176,
      "step": 21358
    },
    {
      "epoch": 0.7649112754489946,
      "grad_norm": 1.653333306312561,
      "learning_rate": 2.7610104680906933e-05,
      "loss": 1.4896,
      "step": 21359
    },
    {
      "epoch": 0.7649470875785629,
      "grad_norm": 1.548635721206665,
      "learning_rate": 2.760210293912223e-05,
      "loss": 1.7664,
      "step": 21360
    },
    {
      "epoch": 0.7649828997081312,
      "grad_norm": 1.920652985572815,
      "learning_rate": 2.759410217135786e-05,
      "loss": 1.4373,
      "step": 21361
    },
    {
      "epoch": 0.7650187118376994,
      "grad_norm": 1.9298456907272339,
      "learning_rate": 2.7586102377721467e-05,
      "loss": 1.2179,
      "step": 21362
    },
    {
      "epoch": 0.7650545239672677,
      "grad_norm": 1.5186060667037964,
      "learning_rate": 2.7578103558320623e-05,
      "loss": 1.5287,
      "step": 21363
    },
    {
      "epoch": 0.765090336096836,
      "grad_norm": 1.91496741771698,
      "learning_rate": 2.7570105713262995e-05,
      "loss": 1.4081,
      "step": 21364
    },
    {
      "epoch": 0.7651261482264042,
      "grad_norm": 2.4558486938476562,
      "learning_rate": 2.7562108842656152e-05,
      "loss": 1.1442,
      "step": 21365
    },
    {
      "epoch": 0.7651619603559726,
      "grad_norm": 1.984779715538025,
      "learning_rate": 2.7554112946607735e-05,
      "loss": 1.5647,
      "step": 21366
    },
    {
      "epoch": 0.7651977724855409,
      "grad_norm": 1.4279773235321045,
      "learning_rate": 2.7546118025225244e-05,
      "loss": 1.4004,
      "step": 21367
    },
    {
      "epoch": 0.7652335846151092,
      "grad_norm": 1.703660011291504,
      "learning_rate": 2.7538124078616278e-05,
      "loss": 1.4426,
      "step": 21368
    },
    {
      "epoch": 0.7652693967446774,
      "grad_norm": 1.6676280498504639,
      "learning_rate": 2.753013110688839e-05,
      "loss": 1.3846,
      "step": 21369
    },
    {
      "epoch": 0.7653052088742457,
      "grad_norm": 1.7314229011535645,
      "learning_rate": 2.7522139110149125e-05,
      "loss": 1.7917,
      "step": 21370
    },
    {
      "epoch": 0.765341021003814,
      "grad_norm": 1.8637425899505615,
      "learning_rate": 2.7514148088505998e-05,
      "loss": 1.595,
      "step": 21371
    },
    {
      "epoch": 0.7653768331333822,
      "grad_norm": 2.5171310901641846,
      "learning_rate": 2.7506158042066454e-05,
      "loss": 1.4561,
      "step": 21372
    },
    {
      "epoch": 0.7654126452629506,
      "grad_norm": 1.657228708267212,
      "learning_rate": 2.74981689709381e-05,
      "loss": 1.5529,
      "step": 21373
    },
    {
      "epoch": 0.7654484573925189,
      "grad_norm": 1.326586365699768,
      "learning_rate": 2.749018087522832e-05,
      "loss": 1.0958,
      "step": 21374
    },
    {
      "epoch": 0.7654842695220871,
      "grad_norm": 1.817240595817566,
      "learning_rate": 2.7482193755044637e-05,
      "loss": 1.5557,
      "step": 21375
    },
    {
      "epoch": 0.7655200816516554,
      "grad_norm": 1.442419171333313,
      "learning_rate": 2.7474207610494495e-05,
      "loss": 1.2791,
      "step": 21376
    },
    {
      "epoch": 0.7655558937812237,
      "grad_norm": 1.3630636930465698,
      "learning_rate": 2.7466222441685362e-05,
      "loss": 1.1471,
      "step": 21377
    },
    {
      "epoch": 0.765591705910792,
      "grad_norm": 1.8548414707183838,
      "learning_rate": 2.7458238248724623e-05,
      "loss": 1.5049,
      "step": 21378
    },
    {
      "epoch": 0.7656275180403602,
      "grad_norm": 1.5222378969192505,
      "learning_rate": 2.7450255031719707e-05,
      "loss": 1.5707,
      "step": 21379
    },
    {
      "epoch": 0.7656633301699286,
      "grad_norm": 1.5617293119430542,
      "learning_rate": 2.7442272790778057e-05,
      "loss": 1.1615,
      "step": 21380
    },
    {
      "epoch": 0.7656991422994969,
      "grad_norm": 1.7830644845962524,
      "learning_rate": 2.7434291526007004e-05,
      "loss": 1.5264,
      "step": 21381
    },
    {
      "epoch": 0.7657349544290651,
      "grad_norm": 1.7911914587020874,
      "learning_rate": 2.742631123751399e-05,
      "loss": 1.3538,
      "step": 21382
    },
    {
      "epoch": 0.7657707665586334,
      "grad_norm": 1.432540774345398,
      "learning_rate": 2.7418331925406293e-05,
      "loss": 1.458,
      "step": 21383
    },
    {
      "epoch": 0.7658065786882017,
      "grad_norm": 2.0374929904937744,
      "learning_rate": 2.741035358979136e-05,
      "loss": 1.5482,
      "step": 21384
    },
    {
      "epoch": 0.76584239081777,
      "grad_norm": 1.5840460062026978,
      "learning_rate": 2.7402376230776473e-05,
      "loss": 1.3272,
      "step": 21385
    },
    {
      "epoch": 0.7658782029473382,
      "grad_norm": 1.9470570087432861,
      "learning_rate": 2.7394399848468953e-05,
      "loss": 1.4448,
      "step": 21386
    },
    {
      "epoch": 0.7659140150769066,
      "grad_norm": 1.5973190069198608,
      "learning_rate": 2.7386424442976132e-05,
      "loss": 1.3651,
      "step": 21387
    },
    {
      "epoch": 0.7659498272064749,
      "grad_norm": 2.248100996017456,
      "learning_rate": 2.7378450014405342e-05,
      "loss": 1.4034,
      "step": 21388
    },
    {
      "epoch": 0.7659856393360431,
      "grad_norm": 1.8361610174179077,
      "learning_rate": 2.7370476562863835e-05,
      "loss": 1.3061,
      "step": 21389
    },
    {
      "epoch": 0.7660214514656114,
      "grad_norm": 1.7707544565200806,
      "learning_rate": 2.7362504088458807e-05,
      "loss": 1.3279,
      "step": 21390
    },
    {
      "epoch": 0.7660572635951797,
      "grad_norm": 1.2263545989990234,
      "learning_rate": 2.7354532591297666e-05,
      "loss": 1.6046,
      "step": 21391
    },
    {
      "epoch": 0.7660930757247479,
      "grad_norm": 1.5667316913604736,
      "learning_rate": 2.7346562071487537e-05,
      "loss": 1.3985,
      "step": 21392
    },
    {
      "epoch": 0.7661288878543162,
      "grad_norm": 1.6585215330123901,
      "learning_rate": 2.7338592529135744e-05,
      "loss": 1.5769,
      "step": 21393
    },
    {
      "epoch": 0.7661646999838846,
      "grad_norm": 1.4081734418869019,
      "learning_rate": 2.7330623964349387e-05,
      "loss": 1.5353,
      "step": 21394
    },
    {
      "epoch": 0.7662005121134529,
      "grad_norm": 2.058746337890625,
      "learning_rate": 2.732265637723582e-05,
      "loss": 1.4592,
      "step": 21395
    },
    {
      "epoch": 0.7662363242430211,
      "grad_norm": 1.8329404592514038,
      "learning_rate": 2.7314689767902134e-05,
      "loss": 1.3414,
      "step": 21396
    },
    {
      "epoch": 0.7662721363725894,
      "grad_norm": 1.811803936958313,
      "learning_rate": 2.7306724136455564e-05,
      "loss": 1.3135,
      "step": 21397
    },
    {
      "epoch": 0.7663079485021577,
      "grad_norm": 1.5890330076217651,
      "learning_rate": 2.7298759483003223e-05,
      "loss": 1.0912,
      "step": 21398
    },
    {
      "epoch": 0.7663437606317259,
      "grad_norm": 1.5598218441009521,
      "learning_rate": 2.7290795807652305e-05,
      "loss": 1.7041,
      "step": 21399
    },
    {
      "epoch": 0.7663795727612942,
      "grad_norm": 1.9628334045410156,
      "learning_rate": 2.7282833110509952e-05,
      "loss": 1.6035,
      "step": 21400
    },
    {
      "epoch": 0.7664153848908626,
      "grad_norm": 1.3994742631912231,
      "learning_rate": 2.7274871391683243e-05,
      "loss": 1.0589,
      "step": 21401
    },
    {
      "epoch": 0.7664511970204309,
      "grad_norm": 1.670101284980774,
      "learning_rate": 2.7266910651279376e-05,
      "loss": 1.189,
      "step": 21402
    },
    {
      "epoch": 0.7664870091499991,
      "grad_norm": 1.4442775249481201,
      "learning_rate": 2.725895088940539e-05,
      "loss": 1.3845,
      "step": 21403
    },
    {
      "epoch": 0.7665228212795674,
      "grad_norm": 1.8506652116775513,
      "learning_rate": 2.7250992106168406e-05,
      "loss": 1.6786,
      "step": 21404
    },
    {
      "epoch": 0.7665586334091357,
      "grad_norm": 1.296434760093689,
      "learning_rate": 2.724303430167543e-05,
      "loss": 1.384,
      "step": 21405
    },
    {
      "epoch": 0.7665944455387039,
      "grad_norm": 1.8313742876052856,
      "learning_rate": 2.7235077476033645e-05,
      "loss": 1.4286,
      "step": 21406
    },
    {
      "epoch": 0.7666302576682722,
      "grad_norm": 1.7008599042892456,
      "learning_rate": 2.7227121629350016e-05,
      "loss": 1.5361,
      "step": 21407
    },
    {
      "epoch": 0.7666660697978406,
      "grad_norm": 1.674019455909729,
      "learning_rate": 2.7219166761731585e-05,
      "loss": 1.6992,
      "step": 21408
    },
    {
      "epoch": 0.7667018819274088,
      "grad_norm": 1.7322226762771606,
      "learning_rate": 2.7211212873285376e-05,
      "loss": 1.4695,
      "step": 21409
    },
    {
      "epoch": 0.7667376940569771,
      "grad_norm": 1.8530694246292114,
      "learning_rate": 2.72032599641184e-05,
      "loss": 1.3144,
      "step": 21410
    },
    {
      "epoch": 0.7667735061865454,
      "grad_norm": 1.8700464963912964,
      "learning_rate": 2.7195308034337698e-05,
      "loss": 1.2808,
      "step": 21411
    },
    {
      "epoch": 0.7668093183161137,
      "grad_norm": 1.8522310256958008,
      "learning_rate": 2.7187357084050147e-05,
      "loss": 1.2198,
      "step": 21412
    },
    {
      "epoch": 0.7668451304456819,
      "grad_norm": 2.2776641845703125,
      "learning_rate": 2.7179407113362853e-05,
      "loss": 1.6649,
      "step": 21413
    },
    {
      "epoch": 0.7668809425752502,
      "grad_norm": 2.573241949081421,
      "learning_rate": 2.7171458122382675e-05,
      "loss": 1.4936,
      "step": 21414
    },
    {
      "epoch": 0.7669167547048186,
      "grad_norm": 1.872636318206787,
      "learning_rate": 2.7163510111216618e-05,
      "loss": 1.4336,
      "step": 21415
    },
    {
      "epoch": 0.7669525668343868,
      "grad_norm": 2.0823798179626465,
      "learning_rate": 2.7155563079971535e-05,
      "loss": 1.5788,
      "step": 21416
    },
    {
      "epoch": 0.7669883789639551,
      "grad_norm": 1.9190607070922852,
      "learning_rate": 2.71476170287544e-05,
      "loss": 1.3688,
      "step": 21417
    },
    {
      "epoch": 0.7670241910935234,
      "grad_norm": 1.5269092321395874,
      "learning_rate": 2.713967195767214e-05,
      "loss": 1.4576,
      "step": 21418
    },
    {
      "epoch": 0.7670600032230916,
      "grad_norm": 1.5405995845794678,
      "learning_rate": 2.713172786683157e-05,
      "loss": 1.3906,
      "step": 21419
    },
    {
      "epoch": 0.7670958153526599,
      "grad_norm": 1.7105742692947388,
      "learning_rate": 2.712378475633961e-05,
      "loss": 1.6096,
      "step": 21420
    },
    {
      "epoch": 0.7671316274822282,
      "grad_norm": 1.8407995700836182,
      "learning_rate": 2.7115842626303134e-05,
      "loss": 1.4489,
      "step": 21421
    },
    {
      "epoch": 0.7671674396117966,
      "grad_norm": 1.9620721340179443,
      "learning_rate": 2.7107901476829e-05,
      "loss": 1.6845,
      "step": 21422
    },
    {
      "epoch": 0.7672032517413648,
      "grad_norm": 2.0660271644592285,
      "learning_rate": 2.7099961308024004e-05,
      "loss": 1.1152,
      "step": 21423
    },
    {
      "epoch": 0.7672390638709331,
      "grad_norm": 1.7892515659332275,
      "learning_rate": 2.7092022119994988e-05,
      "loss": 1.3035,
      "step": 21424
    },
    {
      "epoch": 0.7672748760005014,
      "grad_norm": 1.1042615175247192,
      "learning_rate": 2.70840839128488e-05,
      "loss": 1.2817,
      "step": 21425
    },
    {
      "epoch": 0.7673106881300696,
      "grad_norm": 1.3992247581481934,
      "learning_rate": 2.7076146686692184e-05,
      "loss": 1.4926,
      "step": 21426
    },
    {
      "epoch": 0.7673465002596379,
      "grad_norm": 2.207401990890503,
      "learning_rate": 2.7068210441631947e-05,
      "loss": 1.4668,
      "step": 21427
    },
    {
      "epoch": 0.7673823123892062,
      "grad_norm": 1.5987458229064941,
      "learning_rate": 2.7060275177774862e-05,
      "loss": 1.3982,
      "step": 21428
    },
    {
      "epoch": 0.7674181245187746,
      "grad_norm": 1.4975227117538452,
      "learning_rate": 2.7052340895227714e-05,
      "loss": 1.7043,
      "step": 21429
    },
    {
      "epoch": 0.7674539366483428,
      "grad_norm": 1.8431814908981323,
      "learning_rate": 2.7044407594097197e-05,
      "loss": 1.618,
      "step": 21430
    },
    {
      "epoch": 0.7674897487779111,
      "grad_norm": 1.6562138795852661,
      "learning_rate": 2.703647527449007e-05,
      "loss": 1.2938,
      "step": 21431
    },
    {
      "epoch": 0.7675255609074794,
      "grad_norm": 1.7366337776184082,
      "learning_rate": 2.7028543936513086e-05,
      "loss": 1.255,
      "step": 21432
    },
    {
      "epoch": 0.7675613730370476,
      "grad_norm": 1.6569148302078247,
      "learning_rate": 2.7020613580272893e-05,
      "loss": 1.4548,
      "step": 21433
    },
    {
      "epoch": 0.7675971851666159,
      "grad_norm": 1.7070738077163696,
      "learning_rate": 2.7012684205876192e-05,
      "loss": 1.226,
      "step": 21434
    },
    {
      "epoch": 0.7676329972961842,
      "grad_norm": 1.5093258619308472,
      "learning_rate": 2.7004755813429683e-05,
      "loss": 1.4913,
      "step": 21435
    },
    {
      "epoch": 0.7676688094257526,
      "grad_norm": 1.564212441444397,
      "learning_rate": 2.6996828403040064e-05,
      "loss": 1.5233,
      "step": 21436
    },
    {
      "epoch": 0.7677046215553208,
      "grad_norm": 1.472865343093872,
      "learning_rate": 2.698890197481392e-05,
      "loss": 1.4532,
      "step": 21437
    },
    {
      "epoch": 0.7677404336848891,
      "grad_norm": 2.028477430343628,
      "learning_rate": 2.6980976528857915e-05,
      "loss": 1.3926,
      "step": 21438
    },
    {
      "epoch": 0.7677762458144574,
      "grad_norm": 1.4569189548492432,
      "learning_rate": 2.697305206527869e-05,
      "loss": 1.2849,
      "step": 21439
    },
    {
      "epoch": 0.7678120579440256,
      "grad_norm": 1.7602864503860474,
      "learning_rate": 2.6965128584182886e-05,
      "loss": 1.3607,
      "step": 21440
    },
    {
      "epoch": 0.7678478700735939,
      "grad_norm": 1.7039145231246948,
      "learning_rate": 2.6957206085677023e-05,
      "loss": 1.6376,
      "step": 21441
    },
    {
      "epoch": 0.7678836822031622,
      "grad_norm": 2.3543314933776855,
      "learning_rate": 2.694928456986775e-05,
      "loss": 1.5835,
      "step": 21442
    },
    {
      "epoch": 0.7679194943327305,
      "grad_norm": 1.356101393699646,
      "learning_rate": 2.6941364036861638e-05,
      "loss": 1.0908,
      "step": 21443
    },
    {
      "epoch": 0.7679553064622988,
      "grad_norm": 1.6495617628097534,
      "learning_rate": 2.6933444486765212e-05,
      "loss": 1.3593,
      "step": 21444
    },
    {
      "epoch": 0.7679911185918671,
      "grad_norm": 1.7570487260818481,
      "learning_rate": 2.6925525919685047e-05,
      "loss": 1.6665,
      "step": 21445
    },
    {
      "epoch": 0.7680269307214354,
      "grad_norm": 1.7626287937164307,
      "learning_rate": 2.6917608335727675e-05,
      "loss": 1.5255,
      "step": 21446
    },
    {
      "epoch": 0.7680627428510036,
      "grad_norm": 1.4852346181869507,
      "learning_rate": 2.6909691734999633e-05,
      "loss": 1.3978,
      "step": 21447
    },
    {
      "epoch": 0.7680985549805719,
      "grad_norm": 1.377875566482544,
      "learning_rate": 2.690177611760738e-05,
      "loss": 1.4612,
      "step": 21448
    },
    {
      "epoch": 0.7681343671101402,
      "grad_norm": 1.5788637399673462,
      "learning_rate": 2.6893861483657436e-05,
      "loss": 1.529,
      "step": 21449
    },
    {
      "epoch": 0.7681701792397085,
      "grad_norm": 1.2312066555023193,
      "learning_rate": 2.688594783325632e-05,
      "loss": 1.4575,
      "step": 21450
    },
    {
      "epoch": 0.7682059913692768,
      "grad_norm": 2.2032153606414795,
      "learning_rate": 2.687803516651044e-05,
      "loss": 1.4401,
      "step": 21451
    },
    {
      "epoch": 0.7682418034988451,
      "grad_norm": 1.4819837808609009,
      "learning_rate": 2.6870123483526276e-05,
      "loss": 1.329,
      "step": 21452
    },
    {
      "epoch": 0.7682776156284133,
      "grad_norm": 1.5559954643249512,
      "learning_rate": 2.6862212784410258e-05,
      "loss": 1.3719,
      "step": 21453
    },
    {
      "epoch": 0.7683134277579816,
      "grad_norm": 1.2472002506256104,
      "learning_rate": 2.685430306926887e-05,
      "loss": 1.3489,
      "step": 21454
    },
    {
      "epoch": 0.7683492398875499,
      "grad_norm": 1.4845582246780396,
      "learning_rate": 2.6846394338208446e-05,
      "loss": 1.4089,
      "step": 21455
    },
    {
      "epoch": 0.7683850520171182,
      "grad_norm": 2.007016181945801,
      "learning_rate": 2.683848659133542e-05,
      "loss": 1.4288,
      "step": 21456
    },
    {
      "epoch": 0.7684208641466865,
      "grad_norm": 1.433982253074646,
      "learning_rate": 2.68305798287562e-05,
      "loss": 1.3565,
      "step": 21457
    },
    {
      "epoch": 0.7684566762762548,
      "grad_norm": 1.5630964040756226,
      "learning_rate": 2.682267405057717e-05,
      "loss": 1.4758,
      "step": 21458
    },
    {
      "epoch": 0.7684924884058231,
      "grad_norm": 2.2693796157836914,
      "learning_rate": 2.6814769256904627e-05,
      "loss": 1.3924,
      "step": 21459
    },
    {
      "epoch": 0.7685283005353913,
      "grad_norm": 1.6751092672348022,
      "learning_rate": 2.6806865447844974e-05,
      "loss": 1.4485,
      "step": 21460
    },
    {
      "epoch": 0.7685641126649596,
      "grad_norm": 1.8391053676605225,
      "learning_rate": 2.6798962623504566e-05,
      "loss": 1.6818,
      "step": 21461
    },
    {
      "epoch": 0.7685999247945279,
      "grad_norm": 1.5685383081436157,
      "learning_rate": 2.6791060783989653e-05,
      "loss": 1.2592,
      "step": 21462
    },
    {
      "epoch": 0.7686357369240961,
      "grad_norm": 1.350655436515808,
      "learning_rate": 2.678315992940659e-05,
      "loss": 1.3298,
      "step": 21463
    },
    {
      "epoch": 0.7686715490536645,
      "grad_norm": 1.6182117462158203,
      "learning_rate": 2.6775260059861673e-05,
      "loss": 1.4552,
      "step": 21464
    },
    {
      "epoch": 0.7687073611832328,
      "grad_norm": 1.374692440032959,
      "learning_rate": 2.6767361175461202e-05,
      "loss": 1.2673,
      "step": 21465
    },
    {
      "epoch": 0.7687431733128011,
      "grad_norm": 1.6027882099151611,
      "learning_rate": 2.6759463276311393e-05,
      "loss": 1.2951,
      "step": 21466
    },
    {
      "epoch": 0.7687789854423693,
      "grad_norm": 1.84087336063385,
      "learning_rate": 2.675156636251853e-05,
      "loss": 1.3431,
      "step": 21467
    },
    {
      "epoch": 0.7688147975719376,
      "grad_norm": 2.3304686546325684,
      "learning_rate": 2.6743670434188893e-05,
      "loss": 1.6684,
      "step": 21468
    },
    {
      "epoch": 0.7688506097015059,
      "grad_norm": 2.1128644943237305,
      "learning_rate": 2.673577549142864e-05,
      "loss": 1.5036,
      "step": 21469
    },
    {
      "epoch": 0.7688864218310741,
      "grad_norm": 1.8388144969940186,
      "learning_rate": 2.6727881534344057e-05,
      "loss": 1.6742,
      "step": 21470
    },
    {
      "epoch": 0.7689222339606425,
      "grad_norm": 1.973982810974121,
      "learning_rate": 2.6719988563041264e-05,
      "loss": 1.4739,
      "step": 21471
    },
    {
      "epoch": 0.7689580460902108,
      "grad_norm": 1.7554025650024414,
      "learning_rate": 2.6712096577626543e-05,
      "loss": 1.4875,
      "step": 21472
    },
    {
      "epoch": 0.7689938582197791,
      "grad_norm": 1.8182271718978882,
      "learning_rate": 2.670420557820601e-05,
      "loss": 1.8165,
      "step": 21473
    },
    {
      "epoch": 0.7690296703493473,
      "grad_norm": 1.4545783996582031,
      "learning_rate": 2.6696315564885844e-05,
      "loss": 1.5724,
      "step": 21474
    },
    {
      "epoch": 0.7690654824789156,
      "grad_norm": 1.3026014566421509,
      "learning_rate": 2.6688426537772194e-05,
      "loss": 1.5714,
      "step": 21475
    },
    {
      "epoch": 0.7691012946084839,
      "grad_norm": 1.7563316822052002,
      "learning_rate": 2.668053849697123e-05,
      "loss": 1.3821,
      "step": 21476
    },
    {
      "epoch": 0.7691371067380521,
      "grad_norm": 1.6058958768844604,
      "learning_rate": 2.6672651442589046e-05,
      "loss": 1.4226,
      "step": 21477
    },
    {
      "epoch": 0.7691729188676205,
      "grad_norm": 1.446184754371643,
      "learning_rate": 2.6664765374731693e-05,
      "loss": 1.1693,
      "step": 21478
    },
    {
      "epoch": 0.7692087309971888,
      "grad_norm": 1.7952556610107422,
      "learning_rate": 2.665688029350538e-05,
      "loss": 1.5924,
      "step": 21479
    },
    {
      "epoch": 0.769244543126757,
      "grad_norm": 1.4149837493896484,
      "learning_rate": 2.6648996199016118e-05,
      "loss": 1.5261,
      "step": 21480
    },
    {
      "epoch": 0.7692803552563253,
      "grad_norm": 2.5518479347229004,
      "learning_rate": 2.6641113091370017e-05,
      "loss": 1.4939,
      "step": 21481
    },
    {
      "epoch": 0.7693161673858936,
      "grad_norm": 2.9114930629730225,
      "learning_rate": 2.6633230970673062e-05,
      "loss": 1.8433,
      "step": 21482
    },
    {
      "epoch": 0.7693519795154619,
      "grad_norm": 1.5813169479370117,
      "learning_rate": 2.66253498370314e-05,
      "loss": 1.3981,
      "step": 21483
    },
    {
      "epoch": 0.7693877916450301,
      "grad_norm": 1.501997947692871,
      "learning_rate": 2.661746969055098e-05,
      "loss": 1.425,
      "step": 21484
    },
    {
      "epoch": 0.7694236037745985,
      "grad_norm": 1.4923934936523438,
      "learning_rate": 2.660959053133786e-05,
      "loss": 1.4173,
      "step": 21485
    },
    {
      "epoch": 0.7694594159041668,
      "grad_norm": 1.368891954421997,
      "learning_rate": 2.6601712359498045e-05,
      "loss": 1.5218,
      "step": 21486
    },
    {
      "epoch": 0.769495228033735,
      "grad_norm": 1.6902717351913452,
      "learning_rate": 2.6593835175137494e-05,
      "loss": 1.2863,
      "step": 21487
    },
    {
      "epoch": 0.7695310401633033,
      "grad_norm": 1.7259552478790283,
      "learning_rate": 2.6585958978362235e-05,
      "loss": 1.4637,
      "step": 21488
    },
    {
      "epoch": 0.7695668522928716,
      "grad_norm": 1.6730120182037354,
      "learning_rate": 2.6578083769278127e-05,
      "loss": 1.3071,
      "step": 21489
    },
    {
      "epoch": 0.7696026644224399,
      "grad_norm": 2.149928331375122,
      "learning_rate": 2.6570209547991265e-05,
      "loss": 1.5697,
      "step": 21490
    },
    {
      "epoch": 0.7696384765520081,
      "grad_norm": 2.8818540573120117,
      "learning_rate": 2.6562336314607484e-05,
      "loss": 1.3422,
      "step": 21491
    },
    {
      "epoch": 0.7696742886815765,
      "grad_norm": 1.4298157691955566,
      "learning_rate": 2.6554464069232776e-05,
      "loss": 1.4364,
      "step": 21492
    },
    {
      "epoch": 0.7697101008111448,
      "grad_norm": 1.7484629154205322,
      "learning_rate": 2.6546592811972948e-05,
      "loss": 1.5057,
      "step": 21493
    },
    {
      "epoch": 0.769745912940713,
      "grad_norm": 2.24288272857666,
      "learning_rate": 2.6538722542934035e-05,
      "loss": 1.6272,
      "step": 21494
    },
    {
      "epoch": 0.7697817250702813,
      "grad_norm": 1.6153521537780762,
      "learning_rate": 2.6530853262221843e-05,
      "loss": 1.3215,
      "step": 21495
    },
    {
      "epoch": 0.7698175371998496,
      "grad_norm": 1.5541167259216309,
      "learning_rate": 2.652298496994222e-05,
      "loss": 1.3765,
      "step": 21496
    },
    {
      "epoch": 0.7698533493294178,
      "grad_norm": 2.0340280532836914,
      "learning_rate": 2.6515117666201062e-05,
      "loss": 1.6024,
      "step": 21497
    },
    {
      "epoch": 0.7698891614589861,
      "grad_norm": 1.580460786819458,
      "learning_rate": 2.6507251351104212e-05,
      "loss": 1.1933,
      "step": 21498
    },
    {
      "epoch": 0.7699249735885545,
      "grad_norm": 1.4655922651290894,
      "learning_rate": 2.649938602475751e-05,
      "loss": 0.9808,
      "step": 21499
    },
    {
      "epoch": 0.7699607857181228,
      "grad_norm": 1.751434326171875,
      "learning_rate": 2.6491521687266717e-05,
      "loss": 1.4647,
      "step": 21500
    },
    {
      "epoch": 0.769996597847691,
      "grad_norm": 1.7684653997421265,
      "learning_rate": 2.6483658338737726e-05,
      "loss": 1.6803,
      "step": 21501
    },
    {
      "epoch": 0.7700324099772593,
      "grad_norm": 1.5846799612045288,
      "learning_rate": 2.6475795979276262e-05,
      "loss": 1.5269,
      "step": 21502
    },
    {
      "epoch": 0.7700682221068276,
      "grad_norm": 1.7030701637268066,
      "learning_rate": 2.6467934608988155e-05,
      "loss": 1.4744,
      "step": 21503
    },
    {
      "epoch": 0.7701040342363958,
      "grad_norm": 2.3186452388763428,
      "learning_rate": 2.6460074227979104e-05,
      "loss": 1.5314,
      "step": 21504
    },
    {
      "epoch": 0.7701398463659641,
      "grad_norm": 1.4595292806625366,
      "learning_rate": 2.6452214836354893e-05,
      "loss": 1.7642,
      "step": 21505
    },
    {
      "epoch": 0.7701756584955325,
      "grad_norm": 1.770524501800537,
      "learning_rate": 2.6444356434221296e-05,
      "loss": 1.5098,
      "step": 21506
    },
    {
      "epoch": 0.7702114706251008,
      "grad_norm": 3.266895055770874,
      "learning_rate": 2.643649902168397e-05,
      "loss": 1.4408,
      "step": 21507
    },
    {
      "epoch": 0.770247282754669,
      "grad_norm": 1.3348801136016846,
      "learning_rate": 2.6428642598848663e-05,
      "loss": 1.5114,
      "step": 21508
    },
    {
      "epoch": 0.7702830948842373,
      "grad_norm": 1.5498926639556885,
      "learning_rate": 2.642078716582107e-05,
      "loss": 1.3847,
      "step": 21509
    },
    {
      "epoch": 0.7703189070138056,
      "grad_norm": 1.818127155303955,
      "learning_rate": 2.6412932722706908e-05,
      "loss": 1.3761,
      "step": 21510
    },
    {
      "epoch": 0.7703547191433738,
      "grad_norm": 1.73329758644104,
      "learning_rate": 2.6405079269611744e-05,
      "loss": 1.3512,
      "step": 21511
    },
    {
      "epoch": 0.7703905312729421,
      "grad_norm": 1.2978670597076416,
      "learning_rate": 2.6397226806641375e-05,
      "loss": 1.4232,
      "step": 21512
    },
    {
      "epoch": 0.7704263434025105,
      "grad_norm": 1.5033830404281616,
      "learning_rate": 2.6389375333901377e-05,
      "loss": 1.3919,
      "step": 21513
    },
    {
      "epoch": 0.7704621555320788,
      "grad_norm": 1.4706625938415527,
      "learning_rate": 2.6381524851497353e-05,
      "loss": 1.4614,
      "step": 21514
    },
    {
      "epoch": 0.770497967661647,
      "grad_norm": 1.6710084676742554,
      "learning_rate": 2.6373675359534955e-05,
      "loss": 1.549,
      "step": 21515
    },
    {
      "epoch": 0.7705337797912153,
      "grad_norm": 1.2906526327133179,
      "learning_rate": 2.636582685811978e-05,
      "loss": 1.4191,
      "step": 21516
    },
    {
      "epoch": 0.7705695919207836,
      "grad_norm": 1.3973788022994995,
      "learning_rate": 2.6357979347357454e-05,
      "loss": 1.2873,
      "step": 21517
    },
    {
      "epoch": 0.7706054040503518,
      "grad_norm": 1.4239815473556519,
      "learning_rate": 2.635013282735349e-05,
      "loss": 1.5933,
      "step": 21518
    },
    {
      "epoch": 0.7706412161799201,
      "grad_norm": 2.066704273223877,
      "learning_rate": 2.63422872982135e-05,
      "loss": 1.3148,
      "step": 21519
    },
    {
      "epoch": 0.7706770283094885,
      "grad_norm": 1.2484495639801025,
      "learning_rate": 2.6334442760043044e-05,
      "loss": 1.3429,
      "step": 21520
    },
    {
      "epoch": 0.7707128404390567,
      "grad_norm": 1.6173884868621826,
      "learning_rate": 2.632659921294761e-05,
      "loss": 1.4515,
      "step": 21521
    },
    {
      "epoch": 0.770748652568625,
      "grad_norm": 1.4764695167541504,
      "learning_rate": 2.631875665703275e-05,
      "loss": 1.4135,
      "step": 21522
    },
    {
      "epoch": 0.7707844646981933,
      "grad_norm": 1.487092137336731,
      "learning_rate": 2.6310915092403976e-05,
      "loss": 1.2509,
      "step": 21523
    },
    {
      "epoch": 0.7708202768277616,
      "grad_norm": 1.7480628490447998,
      "learning_rate": 2.6303074519166827e-05,
      "loss": 1.6969,
      "step": 21524
    },
    {
      "epoch": 0.7708560889573298,
      "grad_norm": 1.725825309753418,
      "learning_rate": 2.6295234937426706e-05,
      "loss": 1.3108,
      "step": 21525
    },
    {
      "epoch": 0.7708919010868981,
      "grad_norm": 1.542431116104126,
      "learning_rate": 2.628739634728914e-05,
      "loss": 1.0812,
      "step": 21526
    },
    {
      "epoch": 0.7709277132164665,
      "grad_norm": 1.9049519300460815,
      "learning_rate": 2.6279558748859555e-05,
      "loss": 1.659,
      "step": 21527
    },
    {
      "epoch": 0.7709635253460347,
      "grad_norm": 1.3463095426559448,
      "learning_rate": 2.627172214224346e-05,
      "loss": 1.5486,
      "step": 21528
    },
    {
      "epoch": 0.770999337475603,
      "grad_norm": 2.2055890560150146,
      "learning_rate": 2.626388652754621e-05,
      "loss": 1.6108,
      "step": 21529
    },
    {
      "epoch": 0.7710351496051713,
      "grad_norm": 1.8153146505355835,
      "learning_rate": 2.6256051904873246e-05,
      "loss": 1.1265,
      "step": 21530
    },
    {
      "epoch": 0.7710709617347395,
      "grad_norm": 2.994877338409424,
      "learning_rate": 2.6248218274330017e-05,
      "loss": 1.6599,
      "step": 21531
    },
    {
      "epoch": 0.7711067738643078,
      "grad_norm": 1.478348731994629,
      "learning_rate": 2.6240385636021847e-05,
      "loss": 1.4522,
      "step": 21532
    },
    {
      "epoch": 0.7711425859938761,
      "grad_norm": 1.8544647693634033,
      "learning_rate": 2.6232553990054144e-05,
      "loss": 1.4124,
      "step": 21533
    },
    {
      "epoch": 0.7711783981234445,
      "grad_norm": 1.606522798538208,
      "learning_rate": 2.6224723336532274e-05,
      "loss": 1.3303,
      "step": 21534
    },
    {
      "epoch": 0.7712142102530127,
      "grad_norm": 1.5226876735687256,
      "learning_rate": 2.6216893675561617e-05,
      "loss": 1.5826,
      "step": 21535
    },
    {
      "epoch": 0.771250022382581,
      "grad_norm": 1.7415231466293335,
      "learning_rate": 2.6209065007247458e-05,
      "loss": 1.3703,
      "step": 21536
    },
    {
      "epoch": 0.7712858345121493,
      "grad_norm": 1.5941494703292847,
      "learning_rate": 2.6201237331695138e-05,
      "loss": 1.5155,
      "step": 21537
    },
    {
      "epoch": 0.7713216466417175,
      "grad_norm": 1.5538673400878906,
      "learning_rate": 2.619341064901001e-05,
      "loss": 1.2863,
      "step": 21538
    },
    {
      "epoch": 0.7713574587712858,
      "grad_norm": 1.6205617189407349,
      "learning_rate": 2.6185584959297303e-05,
      "loss": 1.4426,
      "step": 21539
    },
    {
      "epoch": 0.7713932709008541,
      "grad_norm": 1.6439517736434937,
      "learning_rate": 2.6177760262662345e-05,
      "loss": 1.3722,
      "step": 21540
    },
    {
      "epoch": 0.7714290830304225,
      "grad_norm": 2.5068721771240234,
      "learning_rate": 2.6169936559210396e-05,
      "loss": 1.5905,
      "step": 21541
    },
    {
      "epoch": 0.7714648951599907,
      "grad_norm": 1.3637676239013672,
      "learning_rate": 2.6162113849046745e-05,
      "loss": 1.5606,
      "step": 21542
    },
    {
      "epoch": 0.771500707289559,
      "grad_norm": 1.5745965242385864,
      "learning_rate": 2.615429213227658e-05,
      "loss": 1.7289,
      "step": 21543
    },
    {
      "epoch": 0.7715365194191273,
      "grad_norm": 1.5057892799377441,
      "learning_rate": 2.6146471409005158e-05,
      "loss": 1.6678,
      "step": 21544
    },
    {
      "epoch": 0.7715723315486955,
      "grad_norm": 1.9322911500930786,
      "learning_rate": 2.61386516793377e-05,
      "loss": 1.5913,
      "step": 21545
    },
    {
      "epoch": 0.7716081436782638,
      "grad_norm": 1.3579450845718384,
      "learning_rate": 2.6130832943379447e-05,
      "loss": 1.5061,
      "step": 21546
    },
    {
      "epoch": 0.7716439558078321,
      "grad_norm": 1.32883620262146,
      "learning_rate": 2.612301520123551e-05,
      "loss": 1.0265,
      "step": 21547
    },
    {
      "epoch": 0.7716797679374005,
      "grad_norm": 1.9010030031204224,
      "learning_rate": 2.6115198453011114e-05,
      "loss": 1.4166,
      "step": 21548
    },
    {
      "epoch": 0.7717155800669687,
      "grad_norm": 1.6484206914901733,
      "learning_rate": 2.6107382698811446e-05,
      "loss": 1.4505,
      "step": 21549
    },
    {
      "epoch": 0.771751392196537,
      "grad_norm": 1.9972126483917236,
      "learning_rate": 2.609956793874161e-05,
      "loss": 1.6538,
      "step": 21550
    },
    {
      "epoch": 0.7717872043261053,
      "grad_norm": 2.0636024475097656,
      "learning_rate": 2.6091754172906747e-05,
      "loss": 1.4478,
      "step": 21551
    },
    {
      "epoch": 0.7718230164556735,
      "grad_norm": 1.371565341949463,
      "learning_rate": 2.6083941401412005e-05,
      "loss": 1.4717,
      "step": 21552
    },
    {
      "epoch": 0.7718588285852418,
      "grad_norm": 1.405341386795044,
      "learning_rate": 2.6076129624362512e-05,
      "loss": 1.3875,
      "step": 21553
    },
    {
      "epoch": 0.7718946407148101,
      "grad_norm": 1.5658798217773438,
      "learning_rate": 2.6068318841863314e-05,
      "loss": 1.1604,
      "step": 21554
    },
    {
      "epoch": 0.7719304528443784,
      "grad_norm": 1.636956810951233,
      "learning_rate": 2.6060509054019523e-05,
      "loss": 1.1873,
      "step": 21555
    },
    {
      "epoch": 0.7719662649739467,
      "grad_norm": 1.5121686458587646,
      "learning_rate": 2.6052700260936237e-05,
      "loss": 1.1696,
      "step": 21556
    },
    {
      "epoch": 0.772002077103515,
      "grad_norm": 1.6735605001449585,
      "learning_rate": 2.604489246271845e-05,
      "loss": 1.6659,
      "step": 21557
    },
    {
      "epoch": 0.7720378892330833,
      "grad_norm": 1.52809739112854,
      "learning_rate": 2.6037085659471237e-05,
      "loss": 1.4882,
      "step": 21558
    },
    {
      "epoch": 0.7720737013626515,
      "grad_norm": 1.5789731740951538,
      "learning_rate": 2.6029279851299636e-05,
      "loss": 1.2542,
      "step": 21559
    },
    {
      "epoch": 0.7721095134922198,
      "grad_norm": 1.3522050380706787,
      "learning_rate": 2.6021475038308694e-05,
      "loss": 1.6823,
      "step": 21560
    },
    {
      "epoch": 0.7721453256217881,
      "grad_norm": 1.3396024703979492,
      "learning_rate": 2.6013671220603343e-05,
      "loss": 1.5947,
      "step": 21561
    },
    {
      "epoch": 0.7721811377513564,
      "grad_norm": 1.3040772676467896,
      "learning_rate": 2.6005868398288614e-05,
      "loss": 1.1411,
      "step": 21562
    },
    {
      "epoch": 0.7722169498809247,
      "grad_norm": 2.3642704486846924,
      "learning_rate": 2.5998066571469482e-05,
      "loss": 1.5471,
      "step": 21563
    },
    {
      "epoch": 0.772252762010493,
      "grad_norm": 2.173736333847046,
      "learning_rate": 2.599026574025093e-05,
      "loss": 1.4395,
      "step": 21564
    },
    {
      "epoch": 0.7722885741400612,
      "grad_norm": 1.3441381454467773,
      "learning_rate": 2.5982465904737895e-05,
      "loss": 1.3327,
      "step": 21565
    },
    {
      "epoch": 0.7723243862696295,
      "grad_norm": 1.4685009717941284,
      "learning_rate": 2.597466706503524e-05,
      "loss": 1.6688,
      "step": 21566
    },
    {
      "epoch": 0.7723601983991978,
      "grad_norm": 1.6783415079116821,
      "learning_rate": 2.5966869221248013e-05,
      "loss": 1.5029,
      "step": 21567
    },
    {
      "epoch": 0.772396010528766,
      "grad_norm": 1.6194955110549927,
      "learning_rate": 2.595907237348104e-05,
      "loss": 1.4862,
      "step": 21568
    },
    {
      "epoch": 0.7724318226583344,
      "grad_norm": 1.703182339668274,
      "learning_rate": 2.595127652183924e-05,
      "loss": 1.3713,
      "step": 21569
    },
    {
      "epoch": 0.7724676347879027,
      "grad_norm": 1.670957088470459,
      "learning_rate": 2.5943481666427506e-05,
      "loss": 1.3126,
      "step": 21570
    },
    {
      "epoch": 0.772503446917471,
      "grad_norm": 1.6317272186279297,
      "learning_rate": 2.5935687807350718e-05,
      "loss": 1.6463,
      "step": 21571
    },
    {
      "epoch": 0.7725392590470392,
      "grad_norm": 1.457313895225525,
      "learning_rate": 2.5927894944713695e-05,
      "loss": 1.2296,
      "step": 21572
    },
    {
      "epoch": 0.7725750711766075,
      "grad_norm": 2.3845486640930176,
      "learning_rate": 2.5920103078621294e-05,
      "loss": 1.4234,
      "step": 21573
    },
    {
      "epoch": 0.7726108833061758,
      "grad_norm": 2.011991500854492,
      "learning_rate": 2.591231220917837e-05,
      "loss": 1.5692,
      "step": 21574
    },
    {
      "epoch": 0.772646695435744,
      "grad_norm": 1.949367642402649,
      "learning_rate": 2.59045223364897e-05,
      "loss": 1.5255,
      "step": 21575
    },
    {
      "epoch": 0.7726825075653124,
      "grad_norm": 1.472780704498291,
      "learning_rate": 2.5896733460660138e-05,
      "loss": 1.5582,
      "step": 21576
    },
    {
      "epoch": 0.7727183196948807,
      "grad_norm": 1.6741362810134888,
      "learning_rate": 2.5888945581794377e-05,
      "loss": 1.4447,
      "step": 21577
    },
    {
      "epoch": 0.772754131824449,
      "grad_norm": 1.7349146604537964,
      "learning_rate": 2.5881158699997322e-05,
      "loss": 1.4366,
      "step": 21578
    },
    {
      "epoch": 0.7727899439540172,
      "grad_norm": 1.5743865966796875,
      "learning_rate": 2.5873372815373633e-05,
      "loss": 1.5798,
      "step": 21579
    },
    {
      "epoch": 0.7728257560835855,
      "grad_norm": 1.3788039684295654,
      "learning_rate": 2.5865587928028124e-05,
      "loss": 1.1381,
      "step": 21580
    },
    {
      "epoch": 0.7728615682131538,
      "grad_norm": 1.6148918867111206,
      "learning_rate": 2.5857804038065446e-05,
      "loss": 1.263,
      "step": 21581
    },
    {
      "epoch": 0.772897380342722,
      "grad_norm": 1.67681086063385,
      "learning_rate": 2.585002114559044e-05,
      "loss": 1.6036,
      "step": 21582
    },
    {
      "epoch": 0.7729331924722904,
      "grad_norm": 1.5975325107574463,
      "learning_rate": 2.5842239250707757e-05,
      "loss": 1.2015,
      "step": 21583
    },
    {
      "epoch": 0.7729690046018587,
      "grad_norm": 1.4812545776367188,
      "learning_rate": 2.5834458353522018e-05,
      "loss": 1.3873,
      "step": 21584
    },
    {
      "epoch": 0.773004816731427,
      "grad_norm": 1.8217486143112183,
      "learning_rate": 2.5826678454138044e-05,
      "loss": 1.6502,
      "step": 21585
    },
    {
      "epoch": 0.7730406288609952,
      "grad_norm": 1.686937928199768,
      "learning_rate": 2.5818899552660404e-05,
      "loss": 1.5596,
      "step": 21586
    },
    {
      "epoch": 0.7730764409905635,
      "grad_norm": 1.654982328414917,
      "learning_rate": 2.5811121649193805e-05,
      "loss": 1.3852,
      "step": 21587
    },
    {
      "epoch": 0.7731122531201318,
      "grad_norm": 1.7891792058944702,
      "learning_rate": 2.5803344743842817e-05,
      "loss": 1.3776,
      "step": 21588
    },
    {
      "epoch": 0.7731480652497,
      "grad_norm": 1.1618537902832031,
      "learning_rate": 2.579556883671217e-05,
      "loss": 1.4942,
      "step": 21589
    },
    {
      "epoch": 0.7731838773792683,
      "grad_norm": 2.126289129257202,
      "learning_rate": 2.578779392790641e-05,
      "loss": 1.4412,
      "step": 21590
    },
    {
      "epoch": 0.7732196895088367,
      "grad_norm": 1.313543438911438,
      "learning_rate": 2.5780020017530182e-05,
      "loss": 1.3064,
      "step": 21591
    },
    {
      "epoch": 0.773255501638405,
      "grad_norm": 2.4040563106536865,
      "learning_rate": 2.5772247105688006e-05,
      "loss": 1.2878,
      "step": 21592
    },
    {
      "epoch": 0.7732913137679732,
      "grad_norm": 1.7156052589416504,
      "learning_rate": 2.5764475192484506e-05,
      "loss": 1.463,
      "step": 21593
    },
    {
      "epoch": 0.7733271258975415,
      "grad_norm": 1.546714425086975,
      "learning_rate": 2.5756704278024268e-05,
      "loss": 1.183,
      "step": 21594
    },
    {
      "epoch": 0.7733629380271098,
      "grad_norm": 1.4804030656814575,
      "learning_rate": 2.5748934362411747e-05,
      "loss": 1.3276,
      "step": 21595
    },
    {
      "epoch": 0.773398750156678,
      "grad_norm": 1.903746247291565,
      "learning_rate": 2.574116544575159e-05,
      "loss": 1.526,
      "step": 21596
    },
    {
      "epoch": 0.7734345622862463,
      "grad_norm": 1.246662974357605,
      "learning_rate": 2.573339752814825e-05,
      "loss": 1.1375,
      "step": 21597
    },
    {
      "epoch": 0.7734703744158147,
      "grad_norm": 2.333866834640503,
      "learning_rate": 2.5725630609706264e-05,
      "loss": 1.7398,
      "step": 21598
    },
    {
      "epoch": 0.773506186545383,
      "grad_norm": 1.3863903284072876,
      "learning_rate": 2.571786469053006e-05,
      "loss": 1.7066,
      "step": 21599
    },
    {
      "epoch": 0.7735419986749512,
      "grad_norm": 1.4405169486999512,
      "learning_rate": 2.5710099770724227e-05,
      "loss": 1.6633,
      "step": 21600
    },
    {
      "epoch": 0.7735778108045195,
      "grad_norm": 1.563675880432129,
      "learning_rate": 2.5702335850393166e-05,
      "loss": 1.2315,
      "step": 21601
    },
    {
      "epoch": 0.7736136229340878,
      "grad_norm": 1.655320644378662,
      "learning_rate": 2.5694572929641326e-05,
      "loss": 1.5362,
      "step": 21602
    },
    {
      "epoch": 0.773649435063656,
      "grad_norm": 1.73147714138031,
      "learning_rate": 2.5686811008573142e-05,
      "loss": 1.2198,
      "step": 21603
    },
    {
      "epoch": 0.7736852471932243,
      "grad_norm": 1.8005670309066772,
      "learning_rate": 2.5679050087293067e-05,
      "loss": 1.3593,
      "step": 21604
    },
    {
      "epoch": 0.7737210593227927,
      "grad_norm": 1.3524696826934814,
      "learning_rate": 2.5671290165905537e-05,
      "loss": 1.5948,
      "step": 21605
    },
    {
      "epoch": 0.7737568714523609,
      "grad_norm": 1.4053454399108887,
      "learning_rate": 2.5663531244514892e-05,
      "loss": 1.4189,
      "step": 21606
    },
    {
      "epoch": 0.7737926835819292,
      "grad_norm": 1.6903241872787476,
      "learning_rate": 2.5655773323225552e-05,
      "loss": 1.6855,
      "step": 21607
    },
    {
      "epoch": 0.7738284957114975,
      "grad_norm": 2.3848190307617188,
      "learning_rate": 2.564801640214187e-05,
      "loss": 1.481,
      "step": 21608
    },
    {
      "epoch": 0.7738643078410657,
      "grad_norm": 1.8017464876174927,
      "learning_rate": 2.564026048136826e-05,
      "loss": 1.7375,
      "step": 21609
    },
    {
      "epoch": 0.773900119970634,
      "grad_norm": 1.8137153387069702,
      "learning_rate": 2.5632505561009002e-05,
      "loss": 1.7897,
      "step": 21610
    },
    {
      "epoch": 0.7739359321002023,
      "grad_norm": 1.650579810142517,
      "learning_rate": 2.5624751641168442e-05,
      "loss": 1.4311,
      "step": 21611
    },
    {
      "epoch": 0.7739717442297707,
      "grad_norm": 1.6380149126052856,
      "learning_rate": 2.5616998721950948e-05,
      "loss": 1.357,
      "step": 21612
    },
    {
      "epoch": 0.7740075563593389,
      "grad_norm": 1.4498231410980225,
      "learning_rate": 2.5609246803460764e-05,
      "loss": 1.6064,
      "step": 21613
    },
    {
      "epoch": 0.7740433684889072,
      "grad_norm": 1.459731101989746,
      "learning_rate": 2.5601495885802196e-05,
      "loss": 1.58,
      "step": 21614
    },
    {
      "epoch": 0.7740791806184755,
      "grad_norm": 2.7620160579681396,
      "learning_rate": 2.559374596907954e-05,
      "loss": 1.3054,
      "step": 21615
    },
    {
      "epoch": 0.7741149927480437,
      "grad_norm": 1.4486792087554932,
      "learning_rate": 2.5585997053397083e-05,
      "loss": 1.7824,
      "step": 21616
    },
    {
      "epoch": 0.774150804877612,
      "grad_norm": 1.2154338359832764,
      "learning_rate": 2.5578249138859023e-05,
      "loss": 1.4483,
      "step": 21617
    },
    {
      "epoch": 0.7741866170071803,
      "grad_norm": 1.8012681007385254,
      "learning_rate": 2.5570502225569625e-05,
      "loss": 1.4442,
      "step": 21618
    },
    {
      "epoch": 0.7742224291367487,
      "grad_norm": 2.442791700363159,
      "learning_rate": 2.556275631363314e-05,
      "loss": 1.7195,
      "step": 21619
    },
    {
      "epoch": 0.7742582412663169,
      "grad_norm": 1.6836718320846558,
      "learning_rate": 2.5555011403153715e-05,
      "loss": 1.3161,
      "step": 21620
    },
    {
      "epoch": 0.7742940533958852,
      "grad_norm": 1.6451094150543213,
      "learning_rate": 2.5547267494235595e-05,
      "loss": 1.1316,
      "step": 21621
    },
    {
      "epoch": 0.7743298655254535,
      "grad_norm": 1.3877159357070923,
      "learning_rate": 2.5539524586982944e-05,
      "loss": 1.1024,
      "step": 21622
    },
    {
      "epoch": 0.7743656776550217,
      "grad_norm": 1.4425606727600098,
      "learning_rate": 2.553178268149997e-05,
      "loss": 1.544,
      "step": 21623
    },
    {
      "epoch": 0.77440148978459,
      "grad_norm": 1.7422723770141602,
      "learning_rate": 2.5524041777890783e-05,
      "loss": 1.1543,
      "step": 21624
    },
    {
      "epoch": 0.7744373019141583,
      "grad_norm": 1.4674055576324463,
      "learning_rate": 2.5516301876259542e-05,
      "loss": 1.3218,
      "step": 21625
    },
    {
      "epoch": 0.7744731140437267,
      "grad_norm": 1.6906684637069702,
      "learning_rate": 2.5508562976710416e-05,
      "loss": 1.2574,
      "step": 21626
    },
    {
      "epoch": 0.7745089261732949,
      "grad_norm": 1.5787941217422485,
      "learning_rate": 2.5500825079347458e-05,
      "loss": 1.1405,
      "step": 21627
    },
    {
      "epoch": 0.7745447383028632,
      "grad_norm": 2.0668768882751465,
      "learning_rate": 2.5493088184274795e-05,
      "loss": 1.4762,
      "step": 21628
    },
    {
      "epoch": 0.7745805504324315,
      "grad_norm": 2.3916802406311035,
      "learning_rate": 2.548535229159653e-05,
      "loss": 1.322,
      "step": 21629
    },
    {
      "epoch": 0.7746163625619997,
      "grad_norm": 1.847888708114624,
      "learning_rate": 2.5477617401416765e-05,
      "loss": 1.4714,
      "step": 21630
    },
    {
      "epoch": 0.774652174691568,
      "grad_norm": 1.9050281047821045,
      "learning_rate": 2.5469883513839498e-05,
      "loss": 1.3579,
      "step": 21631
    },
    {
      "epoch": 0.7746879868211363,
      "grad_norm": 1.6340993642807007,
      "learning_rate": 2.5462150628968806e-05,
      "loss": 1.6059,
      "step": 21632
    },
    {
      "epoch": 0.7747237989507046,
      "grad_norm": 1.4270821809768677,
      "learning_rate": 2.5454418746908737e-05,
      "loss": 1.3574,
      "step": 21633
    },
    {
      "epoch": 0.7747596110802729,
      "grad_norm": 2.2271409034729004,
      "learning_rate": 2.544668786776333e-05,
      "loss": 1.3854,
      "step": 21634
    },
    {
      "epoch": 0.7747954232098412,
      "grad_norm": 1.591993808746338,
      "learning_rate": 2.5438957991636546e-05,
      "loss": 1.3642,
      "step": 21635
    },
    {
      "epoch": 0.7748312353394095,
      "grad_norm": 1.3219497203826904,
      "learning_rate": 2.5431229118632406e-05,
      "loss": 1.3877,
      "step": 21636
    },
    {
      "epoch": 0.7748670474689777,
      "grad_norm": 1.488739013671875,
      "learning_rate": 2.542350124885492e-05,
      "loss": 1.291,
      "step": 21637
    },
    {
      "epoch": 0.774902859598546,
      "grad_norm": 1.77037513256073,
      "learning_rate": 2.5415774382407997e-05,
      "loss": 1.5163,
      "step": 21638
    },
    {
      "epoch": 0.7749386717281143,
      "grad_norm": 1.768937587738037,
      "learning_rate": 2.5408048519395622e-05,
      "loss": 1.3945,
      "step": 21639
    },
    {
      "epoch": 0.7749744838576826,
      "grad_norm": 1.5960173606872559,
      "learning_rate": 2.5400323659921744e-05,
      "loss": 1.361,
      "step": 21640
    },
    {
      "epoch": 0.7750102959872509,
      "grad_norm": 2.2278168201446533,
      "learning_rate": 2.539259980409031e-05,
      "loss": 1.6758,
      "step": 21641
    },
    {
      "epoch": 0.7750461081168192,
      "grad_norm": 2.10552978515625,
      "learning_rate": 2.5384876952005177e-05,
      "loss": 1.4251,
      "step": 21642
    },
    {
      "epoch": 0.7750819202463874,
      "grad_norm": 1.4312798976898193,
      "learning_rate": 2.537715510377028e-05,
      "loss": 1.7312,
      "step": 21643
    },
    {
      "epoch": 0.7751177323759557,
      "grad_norm": 1.418057918548584,
      "learning_rate": 2.5369434259489534e-05,
      "loss": 1.3847,
      "step": 21644
    },
    {
      "epoch": 0.775153544505524,
      "grad_norm": 1.6209661960601807,
      "learning_rate": 2.5361714419266757e-05,
      "loss": 1.4997,
      "step": 21645
    },
    {
      "epoch": 0.7751893566350923,
      "grad_norm": 1.4104539155960083,
      "learning_rate": 2.5353995583205824e-05,
      "loss": 1.4841,
      "step": 21646
    },
    {
      "epoch": 0.7752251687646606,
      "grad_norm": 1.7481262683868408,
      "learning_rate": 2.5346277751410607e-05,
      "loss": 1.3121,
      "step": 21647
    },
    {
      "epoch": 0.7752609808942289,
      "grad_norm": 1.7740089893341064,
      "learning_rate": 2.5338560923984954e-05,
      "loss": 1.8563,
      "step": 21648
    },
    {
      "epoch": 0.7752967930237972,
      "grad_norm": 1.5107569694519043,
      "learning_rate": 2.533084510103263e-05,
      "loss": 1.2749,
      "step": 21649
    },
    {
      "epoch": 0.7753326051533654,
      "grad_norm": 1.553844690322876,
      "learning_rate": 2.532313028265746e-05,
      "loss": 1.6047,
      "step": 21650
    },
    {
      "epoch": 0.7753684172829337,
      "grad_norm": 1.7860406637191772,
      "learning_rate": 2.531541646896325e-05,
      "loss": 1.3046,
      "step": 21651
    },
    {
      "epoch": 0.775404229412502,
      "grad_norm": 1.328852653503418,
      "learning_rate": 2.5307703660053805e-05,
      "loss": 1.031,
      "step": 21652
    },
    {
      "epoch": 0.7754400415420702,
      "grad_norm": 1.684956669807434,
      "learning_rate": 2.5299991856032835e-05,
      "loss": 1.131,
      "step": 21653
    },
    {
      "epoch": 0.7754758536716386,
      "grad_norm": 1.5441877841949463,
      "learning_rate": 2.5292281057004108e-05,
      "loss": 1.5972,
      "step": 21654
    },
    {
      "epoch": 0.7755116658012069,
      "grad_norm": 2.099240303039551,
      "learning_rate": 2.528457126307141e-05,
      "loss": 1.5017,
      "step": 21655
    },
    {
      "epoch": 0.7755474779307752,
      "grad_norm": 1.9487109184265137,
      "learning_rate": 2.5276862474338404e-05,
      "loss": 1.5184,
      "step": 21656
    },
    {
      "epoch": 0.7755832900603434,
      "grad_norm": 1.8288156986236572,
      "learning_rate": 2.5269154690908827e-05,
      "loss": 1.393,
      "step": 21657
    },
    {
      "epoch": 0.7756191021899117,
      "grad_norm": 1.4676835536956787,
      "learning_rate": 2.526144791288637e-05,
      "loss": 1.2301,
      "step": 21658
    },
    {
      "epoch": 0.77565491431948,
      "grad_norm": 1.6494859457015991,
      "learning_rate": 2.525374214037476e-05,
      "loss": 1.4331,
      "step": 21659
    },
    {
      "epoch": 0.7756907264490482,
      "grad_norm": 2.2830381393432617,
      "learning_rate": 2.5246037373477606e-05,
      "loss": 1.47,
      "step": 21660
    },
    {
      "epoch": 0.7757265385786166,
      "grad_norm": 1.7719887495040894,
      "learning_rate": 2.523833361229859e-05,
      "loss": 1.7523,
      "step": 21661
    },
    {
      "epoch": 0.7757623507081849,
      "grad_norm": 1.495537519454956,
      "learning_rate": 2.5230630856941394e-05,
      "loss": 1.2791,
      "step": 21662
    },
    {
      "epoch": 0.7757981628377532,
      "grad_norm": 1.52877676486969,
      "learning_rate": 2.5222929107509584e-05,
      "loss": 1.5248,
      "step": 21663
    },
    {
      "epoch": 0.7758339749673214,
      "grad_norm": 1.7707349061965942,
      "learning_rate": 2.5215228364106835e-05,
      "loss": 1.3937,
      "step": 21664
    },
    {
      "epoch": 0.7758697870968897,
      "grad_norm": 1.9539955854415894,
      "learning_rate": 2.5207528626836662e-05,
      "loss": 1.7118,
      "step": 21665
    },
    {
      "epoch": 0.775905599226458,
      "grad_norm": 1.8302944898605347,
      "learning_rate": 2.5199829895802775e-05,
      "loss": 1.5033,
      "step": 21666
    },
    {
      "epoch": 0.7759414113560262,
      "grad_norm": 2.0510201454162598,
      "learning_rate": 2.519213217110866e-05,
      "loss": 1.6254,
      "step": 21667
    },
    {
      "epoch": 0.7759772234855946,
      "grad_norm": 2.2383663654327393,
      "learning_rate": 2.5184435452857913e-05,
      "loss": 1.6945,
      "step": 21668
    },
    {
      "epoch": 0.7760130356151629,
      "grad_norm": 1.3355016708374023,
      "learning_rate": 2.517673974115409e-05,
      "loss": 1.1414,
      "step": 21669
    },
    {
      "epoch": 0.7760488477447312,
      "grad_norm": 1.586307406425476,
      "learning_rate": 2.5169045036100736e-05,
      "loss": 1.6073,
      "step": 21670
    },
    {
      "epoch": 0.7760846598742994,
      "grad_norm": 1.2950676679611206,
      "learning_rate": 2.5161351337801363e-05,
      "loss": 1.399,
      "step": 21671
    },
    {
      "epoch": 0.7761204720038677,
      "grad_norm": 1.7637110948562622,
      "learning_rate": 2.5153658646359412e-05,
      "loss": 1.2684,
      "step": 21672
    },
    {
      "epoch": 0.776156284133436,
      "grad_norm": 1.1106634140014648,
      "learning_rate": 2.51459669618785e-05,
      "loss": 1.3017,
      "step": 21673
    },
    {
      "epoch": 0.7761920962630042,
      "grad_norm": 1.6944717168807983,
      "learning_rate": 2.5138276284462016e-05,
      "loss": 1.1044,
      "step": 21674
    },
    {
      "epoch": 0.7762279083925726,
      "grad_norm": 1.657623291015625,
      "learning_rate": 2.513058661421349e-05,
      "loss": 1.4931,
      "step": 21675
    },
    {
      "epoch": 0.7762637205221409,
      "grad_norm": 1.4893158674240112,
      "learning_rate": 2.512289795123629e-05,
      "loss": 1.48,
      "step": 21676
    },
    {
      "epoch": 0.7762995326517091,
      "grad_norm": 1.7651742696762085,
      "learning_rate": 2.5115210295633974e-05,
      "loss": 1.2103,
      "step": 21677
    },
    {
      "epoch": 0.7763353447812774,
      "grad_norm": 2.091895580291748,
      "learning_rate": 2.5107523647509877e-05,
      "loss": 1.4103,
      "step": 21678
    },
    {
      "epoch": 0.7763711569108457,
      "grad_norm": 1.813442587852478,
      "learning_rate": 2.5099838006967446e-05,
      "loss": 1.3978,
      "step": 21679
    },
    {
      "epoch": 0.776406969040414,
      "grad_norm": 1.8207119703292847,
      "learning_rate": 2.5092153374110107e-05,
      "loss": 1.2788,
      "step": 21680
    },
    {
      "epoch": 0.7764427811699822,
      "grad_norm": 1.47296142578125,
      "learning_rate": 2.5084469749041185e-05,
      "loss": 1.4844,
      "step": 21681
    },
    {
      "epoch": 0.7764785932995506,
      "grad_norm": 1.7010948657989502,
      "learning_rate": 2.5076787131864132e-05,
      "loss": 1.5089,
      "step": 21682
    },
    {
      "epoch": 0.7765144054291189,
      "grad_norm": 2.0861148834228516,
      "learning_rate": 2.506910552268219e-05,
      "loss": 1.2157,
      "step": 21683
    },
    {
      "epoch": 0.7765502175586871,
      "grad_norm": 1.6660261154174805,
      "learning_rate": 2.5061424921598853e-05,
      "loss": 1.4986,
      "step": 21684
    },
    {
      "epoch": 0.7765860296882554,
      "grad_norm": 1.3567243814468384,
      "learning_rate": 2.5053745328717336e-05,
      "loss": 1.2758,
      "step": 21685
    },
    {
      "epoch": 0.7766218418178237,
      "grad_norm": 1.429267406463623,
      "learning_rate": 2.504606674414104e-05,
      "loss": 1.2731,
      "step": 21686
    },
    {
      "epoch": 0.776657653947392,
      "grad_norm": 1.6606378555297852,
      "learning_rate": 2.5038389167973177e-05,
      "loss": 1.4486,
      "step": 21687
    },
    {
      "epoch": 0.7766934660769602,
      "grad_norm": 1.906919002532959,
      "learning_rate": 2.5030712600317143e-05,
      "loss": 1.6785,
      "step": 21688
    },
    {
      "epoch": 0.7767292782065286,
      "grad_norm": 1.29216468334198,
      "learning_rate": 2.5023037041276175e-05,
      "loss": 1.4627,
      "step": 21689
    },
    {
      "epoch": 0.7767650903360969,
      "grad_norm": 2.332552433013916,
      "learning_rate": 2.5015362490953497e-05,
      "loss": 1.4408,
      "step": 21690
    },
    {
      "epoch": 0.7768009024656651,
      "grad_norm": 1.672184705734253,
      "learning_rate": 2.5007688949452402e-05,
      "loss": 1.2021,
      "step": 21691
    },
    {
      "epoch": 0.7768367145952334,
      "grad_norm": 1.5943087339401245,
      "learning_rate": 2.5000016416876103e-05,
      "loss": 1.4337,
      "step": 21692
    },
    {
      "epoch": 0.7768725267248017,
      "grad_norm": 1.5827921628952026,
      "learning_rate": 2.499234489332788e-05,
      "loss": 1.4747,
      "step": 21693
    },
    {
      "epoch": 0.7769083388543699,
      "grad_norm": 1.874254584312439,
      "learning_rate": 2.4984674378910845e-05,
      "loss": 1.3405,
      "step": 21694
    },
    {
      "epoch": 0.7769441509839382,
      "grad_norm": 1.8608965873718262,
      "learning_rate": 2.4977004873728315e-05,
      "loss": 1.196,
      "step": 21695
    },
    {
      "epoch": 0.7769799631135066,
      "grad_norm": 2.278493881225586,
      "learning_rate": 2.496933637788338e-05,
      "loss": 1.4088,
      "step": 21696
    },
    {
      "epoch": 0.7770157752430749,
      "grad_norm": 1.8424022197723389,
      "learning_rate": 2.496166889147926e-05,
      "loss": 1.3471,
      "step": 21697
    },
    {
      "epoch": 0.7770515873726431,
      "grad_norm": 1.493421196937561,
      "learning_rate": 2.495400241461907e-05,
      "loss": 1.5338,
      "step": 21698
    },
    {
      "epoch": 0.7770873995022114,
      "grad_norm": 1.4196568727493286,
      "learning_rate": 2.494633694740598e-05,
      "loss": 1.4372,
      "step": 21699
    },
    {
      "epoch": 0.7771232116317797,
      "grad_norm": 1.808488130569458,
      "learning_rate": 2.4938672489943138e-05,
      "loss": 1.3776,
      "step": 21700
    },
    {
      "epoch": 0.7771590237613479,
      "grad_norm": 1.3471368551254272,
      "learning_rate": 2.493100904233361e-05,
      "loss": 1.3275,
      "step": 21701
    },
    {
      "epoch": 0.7771948358909162,
      "grad_norm": 1.5173574686050415,
      "learning_rate": 2.4923346604680532e-05,
      "loss": 1.1689,
      "step": 21702
    },
    {
      "epoch": 0.7772306480204846,
      "grad_norm": 1.4176759719848633,
      "learning_rate": 2.4915685177086967e-05,
      "loss": 1.2773,
      "step": 21703
    },
    {
      "epoch": 0.7772664601500529,
      "grad_norm": 1.5612668991088867,
      "learning_rate": 2.4908024759656046e-05,
      "loss": 1.5212,
      "step": 21704
    },
    {
      "epoch": 0.7773022722796211,
      "grad_norm": 1.379515528678894,
      "learning_rate": 2.490036535249073e-05,
      "loss": 1.3418,
      "step": 21705
    },
    {
      "epoch": 0.7773380844091894,
      "grad_norm": 1.7329143285751343,
      "learning_rate": 2.489270695569418e-05,
      "loss": 1.4872,
      "step": 21706
    },
    {
      "epoch": 0.7773738965387577,
      "grad_norm": 1.6147024631500244,
      "learning_rate": 2.4885049569369378e-05,
      "loss": 1.492,
      "step": 21707
    },
    {
      "epoch": 0.7774097086683259,
      "grad_norm": 1.3835676908493042,
      "learning_rate": 2.4877393193619315e-05,
      "loss": 1.2585,
      "step": 21708
    },
    {
      "epoch": 0.7774455207978942,
      "grad_norm": 1.4012019634246826,
      "learning_rate": 2.4869737828547024e-05,
      "loss": 1.4471,
      "step": 21709
    },
    {
      "epoch": 0.7774813329274626,
      "grad_norm": 1.916560173034668,
      "learning_rate": 2.4862083474255503e-05,
      "loss": 1.3948,
      "step": 21710
    },
    {
      "epoch": 0.7775171450570308,
      "grad_norm": 2.5519630908966064,
      "learning_rate": 2.485443013084775e-05,
      "loss": 1.78,
      "step": 21711
    },
    {
      "epoch": 0.7775529571865991,
      "grad_norm": 1.7727234363555908,
      "learning_rate": 2.484677779842669e-05,
      "loss": 1.4787,
      "step": 21712
    },
    {
      "epoch": 0.7775887693161674,
      "grad_norm": 1.8907781839370728,
      "learning_rate": 2.4839126477095287e-05,
      "loss": 1.3687,
      "step": 21713
    },
    {
      "epoch": 0.7776245814457357,
      "grad_norm": 1.9135130643844604,
      "learning_rate": 2.4831476166956515e-05,
      "loss": 1.402,
      "step": 21714
    },
    {
      "epoch": 0.7776603935753039,
      "grad_norm": 1.6929961442947388,
      "learning_rate": 2.482382686811324e-05,
      "loss": 1.4129,
      "step": 21715
    },
    {
      "epoch": 0.7776962057048722,
      "grad_norm": 1.5624363422393799,
      "learning_rate": 2.4816178580668415e-05,
      "loss": 1.4647,
      "step": 21716
    },
    {
      "epoch": 0.7777320178344406,
      "grad_norm": 1.6534024477005005,
      "learning_rate": 2.4808531304724913e-05,
      "loss": 1.497,
      "step": 21717
    },
    {
      "epoch": 0.7777678299640088,
      "grad_norm": 1.207729458808899,
      "learning_rate": 2.4800885040385668e-05,
      "loss": 1.4953,
      "step": 21718
    },
    {
      "epoch": 0.7778036420935771,
      "grad_norm": 1.8488044738769531,
      "learning_rate": 2.4793239787753487e-05,
      "loss": 1.4934,
      "step": 21719
    },
    {
      "epoch": 0.7778394542231454,
      "grad_norm": 1.6745057106018066,
      "learning_rate": 2.478559554693125e-05,
      "loss": 1.3223,
      "step": 21720
    },
    {
      "epoch": 0.7778752663527136,
      "grad_norm": 1.441912293434143,
      "learning_rate": 2.4777952318021814e-05,
      "loss": 1.5393,
      "step": 21721
    },
    {
      "epoch": 0.7779110784822819,
      "grad_norm": 1.9352595806121826,
      "learning_rate": 2.4770310101128026e-05,
      "loss": 1.526,
      "step": 21722
    },
    {
      "epoch": 0.7779468906118502,
      "grad_norm": 1.4422829151153564,
      "learning_rate": 2.476266889635265e-05,
      "loss": 1.6112,
      "step": 21723
    },
    {
      "epoch": 0.7779827027414186,
      "grad_norm": 1.7226628065109253,
      "learning_rate": 2.475502870379851e-05,
      "loss": 1.1627,
      "step": 21724
    },
    {
      "epoch": 0.7780185148709868,
      "grad_norm": 2.2136683464050293,
      "learning_rate": 2.474738952356842e-05,
      "loss": 1.2243,
      "step": 21725
    },
    {
      "epoch": 0.7780543270005551,
      "grad_norm": 1.4644486904144287,
      "learning_rate": 2.4739751355765116e-05,
      "loss": 1.4847,
      "step": 21726
    },
    {
      "epoch": 0.7780901391301234,
      "grad_norm": 1.2714347839355469,
      "learning_rate": 2.4732114200491386e-05,
      "loss": 1.321,
      "step": 21727
    },
    {
      "epoch": 0.7781259512596916,
      "grad_norm": 1.5196048021316528,
      "learning_rate": 2.4724478057849965e-05,
      "loss": 1.4103,
      "step": 21728
    },
    {
      "epoch": 0.7781617633892599,
      "grad_norm": 1.713654637336731,
      "learning_rate": 2.4716842927943617e-05,
      "loss": 1.4538,
      "step": 21729
    },
    {
      "epoch": 0.7781975755188282,
      "grad_norm": 1.7022624015808105,
      "learning_rate": 2.4709208810875017e-05,
      "loss": 1.7523,
      "step": 21730
    },
    {
      "epoch": 0.7782333876483966,
      "grad_norm": 1.4414794445037842,
      "learning_rate": 2.4701575706746882e-05,
      "loss": 1.2275,
      "step": 21731
    },
    {
      "epoch": 0.7782691997779648,
      "grad_norm": 1.515198826789856,
      "learning_rate": 2.4693943615661963e-05,
      "loss": 1.3186,
      "step": 21732
    },
    {
      "epoch": 0.7783050119075331,
      "grad_norm": 1.7945395708084106,
      "learning_rate": 2.4686312537722855e-05,
      "loss": 1.4992,
      "step": 21733
    },
    {
      "epoch": 0.7783408240371014,
      "grad_norm": 1.8566440343856812,
      "learning_rate": 2.4678682473032267e-05,
      "loss": 1.4913,
      "step": 21734
    },
    {
      "epoch": 0.7783766361666696,
      "grad_norm": 1.767237901687622,
      "learning_rate": 2.4671053421692845e-05,
      "loss": 1.3864,
      "step": 21735
    },
    {
      "epoch": 0.7784124482962379,
      "grad_norm": 1.450467586517334,
      "learning_rate": 2.466342538380727e-05,
      "loss": 1.5967,
      "step": 21736
    },
    {
      "epoch": 0.7784482604258062,
      "grad_norm": 2.0914173126220703,
      "learning_rate": 2.46557983594781e-05,
      "loss": 1.4615,
      "step": 21737
    },
    {
      "epoch": 0.7784840725553746,
      "grad_norm": 1.6237757205963135,
      "learning_rate": 2.4648172348807963e-05,
      "loss": 1.4614,
      "step": 21738
    },
    {
      "epoch": 0.7785198846849428,
      "grad_norm": 1.7426936626434326,
      "learning_rate": 2.464054735189948e-05,
      "loss": 1.3176,
      "step": 21739
    },
    {
      "epoch": 0.7785556968145111,
      "grad_norm": 1.7049877643585205,
      "learning_rate": 2.4632923368855254e-05,
      "loss": 1.2699,
      "step": 21740
    },
    {
      "epoch": 0.7785915089440794,
      "grad_norm": 1.7951089143753052,
      "learning_rate": 2.4625300399777806e-05,
      "loss": 1.4016,
      "step": 21741
    },
    {
      "epoch": 0.7786273210736476,
      "grad_norm": 1.667818546295166,
      "learning_rate": 2.46176784447697e-05,
      "loss": 1.3115,
      "step": 21742
    },
    {
      "epoch": 0.7786631332032159,
      "grad_norm": 1.8031182289123535,
      "learning_rate": 2.4610057503933537e-05,
      "loss": 1.5707,
      "step": 21743
    },
    {
      "epoch": 0.7786989453327842,
      "grad_norm": 1.7315683364868164,
      "learning_rate": 2.4602437577371763e-05,
      "loss": 1.3129,
      "step": 21744
    },
    {
      "epoch": 0.7787347574623525,
      "grad_norm": 1.3032207489013672,
      "learning_rate": 2.4594818665186937e-05,
      "loss": 1.3991,
      "step": 21745
    },
    {
      "epoch": 0.7787705695919208,
      "grad_norm": 1.8537158966064453,
      "learning_rate": 2.4587200767481565e-05,
      "loss": 1.1288,
      "step": 21746
    },
    {
      "epoch": 0.7788063817214891,
      "grad_norm": 2.1073696613311768,
      "learning_rate": 2.457958388435816e-05,
      "loss": 1.5075,
      "step": 21747
    },
    {
      "epoch": 0.7788421938510574,
      "grad_norm": 1.4380515813827515,
      "learning_rate": 2.4571968015919144e-05,
      "loss": 1.2178,
      "step": 21748
    },
    {
      "epoch": 0.7788780059806256,
      "grad_norm": 1.5372939109802246,
      "learning_rate": 2.4564353162266996e-05,
      "loss": 1.037,
      "step": 21749
    },
    {
      "epoch": 0.7789138181101939,
      "grad_norm": 1.7316334247589111,
      "learning_rate": 2.4556739323504195e-05,
      "loss": 1.3958,
      "step": 21750
    },
    {
      "epoch": 0.7789496302397622,
      "grad_norm": 2.508051633834839,
      "learning_rate": 2.454912649973313e-05,
      "loss": 1.6857,
      "step": 21751
    },
    {
      "epoch": 0.7789854423693305,
      "grad_norm": 1.7578423023223877,
      "learning_rate": 2.4541514691056245e-05,
      "loss": 1.3476,
      "step": 21752
    },
    {
      "epoch": 0.7790212544988988,
      "grad_norm": 2.091949462890625,
      "learning_rate": 2.453390389757595e-05,
      "loss": 1.594,
      "step": 21753
    },
    {
      "epoch": 0.7790570666284671,
      "grad_norm": 1.5613199472427368,
      "learning_rate": 2.4526294119394653e-05,
      "loss": 1.5924,
      "step": 21754
    },
    {
      "epoch": 0.7790928787580353,
      "grad_norm": 1.5921406745910645,
      "learning_rate": 2.451868535661469e-05,
      "loss": 1.3637,
      "step": 21755
    },
    {
      "epoch": 0.7791286908876036,
      "grad_norm": 1.6087360382080078,
      "learning_rate": 2.451107760933845e-05,
      "loss": 1.5249,
      "step": 21756
    },
    {
      "epoch": 0.7791645030171719,
      "grad_norm": 1.8573392629623413,
      "learning_rate": 2.4503470877668287e-05,
      "loss": 1.5981,
      "step": 21757
    },
    {
      "epoch": 0.7792003151467402,
      "grad_norm": 1.5626813173294067,
      "learning_rate": 2.4495865161706567e-05,
      "loss": 1.3329,
      "step": 21758
    },
    {
      "epoch": 0.7792361272763085,
      "grad_norm": 2.4538016319274902,
      "learning_rate": 2.448826046155559e-05,
      "loss": 1.4858,
      "step": 21759
    },
    {
      "epoch": 0.7792719394058768,
      "grad_norm": 1.3803672790527344,
      "learning_rate": 2.4480656777317613e-05,
      "loss": 1.4366,
      "step": 21760
    },
    {
      "epoch": 0.7793077515354451,
      "grad_norm": 1.7788008451461792,
      "learning_rate": 2.447305410909504e-05,
      "loss": 1.3099,
      "step": 21761
    },
    {
      "epoch": 0.7793435636650133,
      "grad_norm": 1.5982377529144287,
      "learning_rate": 2.4465452456990067e-05,
      "loss": 1.4482,
      "step": 21762
    },
    {
      "epoch": 0.7793793757945816,
      "grad_norm": 1.2722399234771729,
      "learning_rate": 2.4457851821105006e-05,
      "loss": 1.6228,
      "step": 21763
    },
    {
      "epoch": 0.7794151879241499,
      "grad_norm": 1.843069314956665,
      "learning_rate": 2.4450252201542102e-05,
      "loss": 1.3791,
      "step": 21764
    },
    {
      "epoch": 0.7794510000537181,
      "grad_norm": 1.3950681686401367,
      "learning_rate": 2.444265359840363e-05,
      "loss": 1.3212,
      "step": 21765
    },
    {
      "epoch": 0.7794868121832865,
      "grad_norm": 1.4858050346374512,
      "learning_rate": 2.4435056011791768e-05,
      "loss": 1.3646,
      "step": 21766
    },
    {
      "epoch": 0.7795226243128548,
      "grad_norm": 1.639343023300171,
      "learning_rate": 2.4427459441808754e-05,
      "loss": 1.739,
      "step": 21767
    },
    {
      "epoch": 0.7795584364424231,
      "grad_norm": 1.5904186964035034,
      "learning_rate": 2.4419863888556815e-05,
      "loss": 1.3763,
      "step": 21768
    },
    {
      "epoch": 0.7795942485719913,
      "grad_norm": 1.7701135873794556,
      "learning_rate": 2.4412269352138097e-05,
      "loss": 1.5168,
      "step": 21769
    },
    {
      "epoch": 0.7796300607015596,
      "grad_norm": 1.9620040655136108,
      "learning_rate": 2.4404675832654812e-05,
      "loss": 1.421,
      "step": 21770
    },
    {
      "epoch": 0.7796658728311279,
      "grad_norm": 1.3738888502120972,
      "learning_rate": 2.4397083330209046e-05,
      "loss": 1.1076,
      "step": 21771
    },
    {
      "epoch": 0.7797016849606961,
      "grad_norm": 1.2282766103744507,
      "learning_rate": 2.438949184490307e-05,
      "loss": 1.3747,
      "step": 21772
    },
    {
      "epoch": 0.7797374970902645,
      "grad_norm": 1.7531875371932983,
      "learning_rate": 2.438190137683891e-05,
      "loss": 1.4378,
      "step": 21773
    },
    {
      "epoch": 0.7797733092198328,
      "grad_norm": 1.9136475324630737,
      "learning_rate": 2.4374311926118765e-05,
      "loss": 1.4021,
      "step": 21774
    },
    {
      "epoch": 0.7798091213494011,
      "grad_norm": 1.4294815063476562,
      "learning_rate": 2.4366723492844644e-05,
      "loss": 1.5454,
      "step": 21775
    },
    {
      "epoch": 0.7798449334789693,
      "grad_norm": 1.2743984460830688,
      "learning_rate": 2.435913607711876e-05,
      "loss": 1.4154,
      "step": 21776
    },
    {
      "epoch": 0.7798807456085376,
      "grad_norm": 1.6209650039672852,
      "learning_rate": 2.4351549679043118e-05,
      "loss": 1.8038,
      "step": 21777
    },
    {
      "epoch": 0.7799165577381059,
      "grad_norm": 2.175903081893921,
      "learning_rate": 2.4343964298719746e-05,
      "loss": 1.2831,
      "step": 21778
    },
    {
      "epoch": 0.7799523698676741,
      "grad_norm": 1.389844298362732,
      "learning_rate": 2.4336379936250808e-05,
      "loss": 1.459,
      "step": 21779
    },
    {
      "epoch": 0.7799881819972425,
      "grad_norm": 1.7427324056625366,
      "learning_rate": 2.4328796591738236e-05,
      "loss": 1.4667,
      "step": 21780
    },
    {
      "epoch": 0.7800239941268108,
      "grad_norm": 2.2379322052001953,
      "learning_rate": 2.432121426528414e-05,
      "loss": 1.25,
      "step": 21781
    },
    {
      "epoch": 0.780059806256379,
      "grad_norm": 1.569846510887146,
      "learning_rate": 2.431363295699042e-05,
      "loss": 1.5934,
      "step": 21782
    },
    {
      "epoch": 0.7800956183859473,
      "grad_norm": 1.7577472925186157,
      "learning_rate": 2.43060526669592e-05,
      "loss": 1.2713,
      "step": 21783
    },
    {
      "epoch": 0.7801314305155156,
      "grad_norm": 1.8201638460159302,
      "learning_rate": 2.4298473395292378e-05,
      "loss": 1.416,
      "step": 21784
    },
    {
      "epoch": 0.7801672426450839,
      "grad_norm": 2.0842766761779785,
      "learning_rate": 2.4290895142091974e-05,
      "loss": 1.3904,
      "step": 21785
    },
    {
      "epoch": 0.7802030547746521,
      "grad_norm": 1.799854040145874,
      "learning_rate": 2.428331790745989e-05,
      "loss": 1.6248,
      "step": 21786
    },
    {
      "epoch": 0.7802388669042205,
      "grad_norm": 1.5186867713928223,
      "learning_rate": 2.42757416914981e-05,
      "loss": 1.6752,
      "step": 21787
    },
    {
      "epoch": 0.7802746790337888,
      "grad_norm": 1.4737837314605713,
      "learning_rate": 2.4268166494308553e-05,
      "loss": 1.4571,
      "step": 21788
    },
    {
      "epoch": 0.780310491163357,
      "grad_norm": 1.4512406587600708,
      "learning_rate": 2.426059231599308e-05,
      "loss": 1.2519,
      "step": 21789
    },
    {
      "epoch": 0.7803463032929253,
      "grad_norm": 1.9012922048568726,
      "learning_rate": 2.42530191566537e-05,
      "loss": 1.3534,
      "step": 21790
    },
    {
      "epoch": 0.7803821154224936,
      "grad_norm": 1.6925349235534668,
      "learning_rate": 2.4245447016392207e-05,
      "loss": 1.4462,
      "step": 21791
    },
    {
      "epoch": 0.7804179275520619,
      "grad_norm": 1.83090078830719,
      "learning_rate": 2.4237875895310548e-05,
      "loss": 1.4723,
      "step": 21792
    },
    {
      "epoch": 0.7804537396816301,
      "grad_norm": 1.7170168161392212,
      "learning_rate": 2.4230305793510478e-05,
      "loss": 1.6161,
      "step": 21793
    },
    {
      "epoch": 0.7804895518111985,
      "grad_norm": 1.244886040687561,
      "learning_rate": 2.4222736711093964e-05,
      "loss": 1.4318,
      "step": 21794
    },
    {
      "epoch": 0.7805253639407668,
      "grad_norm": 1.9996095895767212,
      "learning_rate": 2.4215168648162778e-05,
      "loss": 1.3049,
      "step": 21795
    },
    {
      "epoch": 0.780561176070335,
      "grad_norm": 1.6705350875854492,
      "learning_rate": 2.420760160481872e-05,
      "loss": 1.4626,
      "step": 21796
    },
    {
      "epoch": 0.7805969881999033,
      "grad_norm": 1.3955320119857788,
      "learning_rate": 2.4200035581163614e-05,
      "loss": 1.4416,
      "step": 21797
    },
    {
      "epoch": 0.7806328003294716,
      "grad_norm": 2.089934825897217,
      "learning_rate": 2.4192470577299263e-05,
      "loss": 1.2793,
      "step": 21798
    },
    {
      "epoch": 0.7806686124590398,
      "grad_norm": 1.3678085803985596,
      "learning_rate": 2.418490659332746e-05,
      "loss": 1.2663,
      "step": 21799
    },
    {
      "epoch": 0.7807044245886081,
      "grad_norm": 1.447238564491272,
      "learning_rate": 2.4177343629349912e-05,
      "loss": 1.412,
      "step": 21800
    },
    {
      "epoch": 0.7807402367181765,
      "grad_norm": 2.202078342437744,
      "learning_rate": 2.4169781685468407e-05,
      "loss": 1.4239,
      "step": 21801
    },
    {
      "epoch": 0.7807760488477448,
      "grad_norm": 1.9966267347335815,
      "learning_rate": 2.416222076178467e-05,
      "loss": 1.31,
      "step": 21802
    },
    {
      "epoch": 0.780811860977313,
      "grad_norm": 2.0359063148498535,
      "learning_rate": 2.4154660858400456e-05,
      "loss": 1.5723,
      "step": 21803
    },
    {
      "epoch": 0.7808476731068813,
      "grad_norm": 1.8663406372070312,
      "learning_rate": 2.414710197541743e-05,
      "loss": 1.5869,
      "step": 21804
    },
    {
      "epoch": 0.7808834852364496,
      "grad_norm": 1.834692120552063,
      "learning_rate": 2.4139544112937283e-05,
      "loss": 1.5305,
      "step": 21805
    },
    {
      "epoch": 0.7809192973660178,
      "grad_norm": 1.34640634059906,
      "learning_rate": 2.413198727106176e-05,
      "loss": 1.6293,
      "step": 21806
    },
    {
      "epoch": 0.7809551094955861,
      "grad_norm": 2.504323959350586,
      "learning_rate": 2.412443144989246e-05,
      "loss": 1.2204,
      "step": 21807
    },
    {
      "epoch": 0.7809909216251545,
      "grad_norm": 2.4626822471618652,
      "learning_rate": 2.411687664953106e-05,
      "loss": 1.5398,
      "step": 21808
    },
    {
      "epoch": 0.7810267337547228,
      "grad_norm": 1.4830766916275024,
      "learning_rate": 2.41093228700792e-05,
      "loss": 1.4997,
      "step": 21809
    },
    {
      "epoch": 0.781062545884291,
      "grad_norm": 1.2905964851379395,
      "learning_rate": 2.4101770111638534e-05,
      "loss": 1.2161,
      "step": 21810
    },
    {
      "epoch": 0.7810983580138593,
      "grad_norm": 1.5993380546569824,
      "learning_rate": 2.409421837431063e-05,
      "loss": 1.3006,
      "step": 21811
    },
    {
      "epoch": 0.7811341701434276,
      "grad_norm": 1.530030608177185,
      "learning_rate": 2.4086667658197093e-05,
      "loss": 1.3426,
      "step": 21812
    },
    {
      "epoch": 0.7811699822729958,
      "grad_norm": 1.3062310218811035,
      "learning_rate": 2.4079117963399554e-05,
      "loss": 1.4719,
      "step": 21813
    },
    {
      "epoch": 0.7812057944025641,
      "grad_norm": 1.582737684249878,
      "learning_rate": 2.4071569290019535e-05,
      "loss": 1.2686,
      "step": 21814
    },
    {
      "epoch": 0.7812416065321325,
      "grad_norm": 1.5829455852508545,
      "learning_rate": 2.4064021638158596e-05,
      "loss": 1.7681,
      "step": 21815
    },
    {
      "epoch": 0.7812774186617008,
      "grad_norm": 1.5450665950775146,
      "learning_rate": 2.40564750079183e-05,
      "loss": 1.3582,
      "step": 21816
    },
    {
      "epoch": 0.781313230791269,
      "grad_norm": 1.877021312713623,
      "learning_rate": 2.404892939940021e-05,
      "loss": 1.5996,
      "step": 21817
    },
    {
      "epoch": 0.7813490429208373,
      "grad_norm": 1.5125741958618164,
      "learning_rate": 2.404138481270577e-05,
      "loss": 1.4883,
      "step": 21818
    },
    {
      "epoch": 0.7813848550504056,
      "grad_norm": 1.710242748260498,
      "learning_rate": 2.4033841247936517e-05,
      "loss": 1.5537,
      "step": 21819
    },
    {
      "epoch": 0.7814206671799738,
      "grad_norm": 2.031437397003174,
      "learning_rate": 2.4026298705193972e-05,
      "loss": 1.6633,
      "step": 21820
    },
    {
      "epoch": 0.7814564793095421,
      "grad_norm": 1.5349557399749756,
      "learning_rate": 2.4018757184579545e-05,
      "loss": 1.2571,
      "step": 21821
    },
    {
      "epoch": 0.7814922914391105,
      "grad_norm": 1.6792068481445312,
      "learning_rate": 2.401121668619474e-05,
      "loss": 1.5264,
      "step": 21822
    },
    {
      "epoch": 0.7815281035686787,
      "grad_norm": 1.345013976097107,
      "learning_rate": 2.4003677210140986e-05,
      "loss": 1.1816,
      "step": 21823
    },
    {
      "epoch": 0.781563915698247,
      "grad_norm": 1.3718589544296265,
      "learning_rate": 2.3996138756519758e-05,
      "loss": 1.336,
      "step": 21824
    },
    {
      "epoch": 0.7815997278278153,
      "grad_norm": 1.577532410621643,
      "learning_rate": 2.3988601325432415e-05,
      "loss": 1.72,
      "step": 21825
    },
    {
      "epoch": 0.7816355399573836,
      "grad_norm": 1.7576181888580322,
      "learning_rate": 2.39810649169804e-05,
      "loss": 1.5281,
      "step": 21826
    },
    {
      "epoch": 0.7816713520869518,
      "grad_norm": 1.4686623811721802,
      "learning_rate": 2.3973529531265095e-05,
      "loss": 1.1936,
      "step": 21827
    },
    {
      "epoch": 0.7817071642165201,
      "grad_norm": 1.418623685836792,
      "learning_rate": 2.396599516838791e-05,
      "loss": 1.2775,
      "step": 21828
    },
    {
      "epoch": 0.7817429763460885,
      "grad_norm": 1.3246876001358032,
      "learning_rate": 2.3958461828450164e-05,
      "loss": 1.3822,
      "step": 21829
    },
    {
      "epoch": 0.7817787884756567,
      "grad_norm": 1.7767356634140015,
      "learning_rate": 2.3950929511553223e-05,
      "loss": 1.4197,
      "step": 21830
    },
    {
      "epoch": 0.781814600605225,
      "grad_norm": 1.667604684829712,
      "learning_rate": 2.3943398217798452e-05,
      "loss": 1.3056,
      "step": 21831
    },
    {
      "epoch": 0.7818504127347933,
      "grad_norm": 1.5106874704360962,
      "learning_rate": 2.393586794728713e-05,
      "loss": 1.2763,
      "step": 21832
    },
    {
      "epoch": 0.7818862248643615,
      "grad_norm": 1.9181849956512451,
      "learning_rate": 2.3928338700120578e-05,
      "loss": 1.4003,
      "step": 21833
    },
    {
      "epoch": 0.7819220369939298,
      "grad_norm": 1.5137401819229126,
      "learning_rate": 2.3920810476400112e-05,
      "loss": 1.6359,
      "step": 21834
    },
    {
      "epoch": 0.7819578491234981,
      "grad_norm": 1.4727210998535156,
      "learning_rate": 2.391328327622704e-05,
      "loss": 1.39,
      "step": 21835
    },
    {
      "epoch": 0.7819936612530665,
      "grad_norm": 1.6446338891983032,
      "learning_rate": 2.3905757099702564e-05,
      "loss": 1.6383,
      "step": 21836
    },
    {
      "epoch": 0.7820294733826347,
      "grad_norm": 1.7629797458648682,
      "learning_rate": 2.3898231946927963e-05,
      "loss": 1.5718,
      "step": 21837
    },
    {
      "epoch": 0.782065285512203,
      "grad_norm": 1.1665034294128418,
      "learning_rate": 2.3890707818004522e-05,
      "loss": 1.5076,
      "step": 21838
    },
    {
      "epoch": 0.7821010976417713,
      "grad_norm": 1.5729949474334717,
      "learning_rate": 2.3883184713033414e-05,
      "loss": 1.2453,
      "step": 21839
    },
    {
      "epoch": 0.7821369097713395,
      "grad_norm": 1.3064537048339844,
      "learning_rate": 2.387566263211586e-05,
      "loss": 1.4201,
      "step": 21840
    },
    {
      "epoch": 0.7821727219009078,
      "grad_norm": 1.7350980043411255,
      "learning_rate": 2.3868141575353077e-05,
      "loss": 1.461,
      "step": 21841
    },
    {
      "epoch": 0.7822085340304761,
      "grad_norm": 1.9520295858383179,
      "learning_rate": 2.3860621542846273e-05,
      "loss": 1.5801,
      "step": 21842
    },
    {
      "epoch": 0.7822443461600445,
      "grad_norm": 1.5008251667022705,
      "learning_rate": 2.3853102534696557e-05,
      "loss": 0.9946,
      "step": 21843
    },
    {
      "epoch": 0.7822801582896127,
      "grad_norm": 1.6487131118774414,
      "learning_rate": 2.384558455100514e-05,
      "loss": 1.537,
      "step": 21844
    },
    {
      "epoch": 0.782315970419181,
      "grad_norm": 1.7409266233444214,
      "learning_rate": 2.3838067591873136e-05,
      "loss": 1.2899,
      "step": 21845
    },
    {
      "epoch": 0.7823517825487493,
      "grad_norm": 1.6287227869033813,
      "learning_rate": 2.3830551657401723e-05,
      "loss": 1.2628,
      "step": 21846
    },
    {
      "epoch": 0.7823875946783175,
      "grad_norm": 1.5295032262802124,
      "learning_rate": 2.3823036747691995e-05,
      "loss": 1.081,
      "step": 21847
    },
    {
      "epoch": 0.7824234068078858,
      "grad_norm": 1.78057062625885,
      "learning_rate": 2.3815522862844985e-05,
      "loss": 1.4897,
      "step": 21848
    },
    {
      "epoch": 0.7824592189374541,
      "grad_norm": 1.901196002960205,
      "learning_rate": 2.3808010002961902e-05,
      "loss": 1.394,
      "step": 21849
    },
    {
      "epoch": 0.7824950310670225,
      "grad_norm": 1.6514678001403809,
      "learning_rate": 2.3800498168143726e-05,
      "loss": 1.161,
      "step": 21850
    },
    {
      "epoch": 0.7825308431965907,
      "grad_norm": 1.6724635362625122,
      "learning_rate": 2.379298735849156e-05,
      "loss": 1.4317,
      "step": 21851
    },
    {
      "epoch": 0.782566655326159,
      "grad_norm": 1.8803189992904663,
      "learning_rate": 2.378547757410645e-05,
      "loss": 1.1883,
      "step": 21852
    },
    {
      "epoch": 0.7826024674557273,
      "grad_norm": 1.2810298204421997,
      "learning_rate": 2.377796881508947e-05,
      "loss": 1.2568,
      "step": 21853
    },
    {
      "epoch": 0.7826382795852955,
      "grad_norm": 1.5125658512115479,
      "learning_rate": 2.3770461081541563e-05,
      "loss": 1.3659,
      "step": 21854
    },
    {
      "epoch": 0.7826740917148638,
      "grad_norm": 2.5842549800872803,
      "learning_rate": 2.3762954373563763e-05,
      "loss": 1.5197,
      "step": 21855
    },
    {
      "epoch": 0.7827099038444321,
      "grad_norm": 1.7202882766723633,
      "learning_rate": 2.375544869125711e-05,
      "loss": 1.4987,
      "step": 21856
    },
    {
      "epoch": 0.7827457159740004,
      "grad_norm": 1.7006173133850098,
      "learning_rate": 2.3747944034722524e-05,
      "loss": 1.3801,
      "step": 21857
    },
    {
      "epoch": 0.7827815281035687,
      "grad_norm": 1.8149973154067993,
      "learning_rate": 2.3740440404061015e-05,
      "loss": 1.214,
      "step": 21858
    },
    {
      "epoch": 0.782817340233137,
      "grad_norm": 1.8355730772018433,
      "learning_rate": 2.3732937799373455e-05,
      "loss": 1.286,
      "step": 21859
    },
    {
      "epoch": 0.7828531523627053,
      "grad_norm": 1.700737476348877,
      "learning_rate": 2.37254362207609e-05,
      "loss": 1.4152,
      "step": 21860
    },
    {
      "epoch": 0.7828889644922735,
      "grad_norm": 1.5340129137039185,
      "learning_rate": 2.3717935668324186e-05,
      "loss": 1.4033,
      "step": 21861
    },
    {
      "epoch": 0.7829247766218418,
      "grad_norm": 1.533355474472046,
      "learning_rate": 2.371043614216425e-05,
      "loss": 1.205,
      "step": 21862
    },
    {
      "epoch": 0.7829605887514101,
      "grad_norm": 2.0897789001464844,
      "learning_rate": 2.3702937642381985e-05,
      "loss": 1.3425,
      "step": 21863
    },
    {
      "epoch": 0.7829964008809784,
      "grad_norm": 1.8037362098693848,
      "learning_rate": 2.369544016907831e-05,
      "loss": 1.2512,
      "step": 21864
    },
    {
      "epoch": 0.7830322130105467,
      "grad_norm": 2.2047224044799805,
      "learning_rate": 2.3687943722354056e-05,
      "loss": 1.8162,
      "step": 21865
    },
    {
      "epoch": 0.783068025140115,
      "grad_norm": 1.6913021802902222,
      "learning_rate": 2.3680448302310032e-05,
      "loss": 1.2962,
      "step": 21866
    },
    {
      "epoch": 0.7831038372696832,
      "grad_norm": 1.605500340461731,
      "learning_rate": 2.367295390904719e-05,
      "loss": 1.8098,
      "step": 21867
    },
    {
      "epoch": 0.7831396493992515,
      "grad_norm": 2.315788984298706,
      "learning_rate": 2.3665460542666263e-05,
      "loss": 1.3339,
      "step": 21868
    },
    {
      "epoch": 0.7831754615288198,
      "grad_norm": 2.22713041305542,
      "learning_rate": 2.3657968203268133e-05,
      "loss": 1.512,
      "step": 21869
    },
    {
      "epoch": 0.783211273658388,
      "grad_norm": 2.339245319366455,
      "learning_rate": 2.365047689095351e-05,
      "loss": 1.2251,
      "step": 21870
    },
    {
      "epoch": 0.7832470857879564,
      "grad_norm": 2.6898417472839355,
      "learning_rate": 2.3642986605823292e-05,
      "loss": 1.8021,
      "step": 21871
    },
    {
      "epoch": 0.7832828979175247,
      "grad_norm": 1.8308742046356201,
      "learning_rate": 2.3635497347978176e-05,
      "loss": 1.4888,
      "step": 21872
    },
    {
      "epoch": 0.783318710047093,
      "grad_norm": 2.016420841217041,
      "learning_rate": 2.3628009117518956e-05,
      "loss": 1.4186,
      "step": 21873
    },
    {
      "epoch": 0.7833545221766612,
      "grad_norm": 1.976356863975525,
      "learning_rate": 2.3620521914546334e-05,
      "loss": 1.5075,
      "step": 21874
    },
    {
      "epoch": 0.7833903343062295,
      "grad_norm": 2.4779212474823,
      "learning_rate": 2.361303573916107e-05,
      "loss": 1.5026,
      "step": 21875
    },
    {
      "epoch": 0.7834261464357978,
      "grad_norm": 1.4665014743804932,
      "learning_rate": 2.360555059146391e-05,
      "loss": 1.2472,
      "step": 21876
    },
    {
      "epoch": 0.783461958565366,
      "grad_norm": 1.507570505142212,
      "learning_rate": 2.359806647155547e-05,
      "loss": 1.339,
      "step": 21877
    },
    {
      "epoch": 0.7834977706949344,
      "grad_norm": 1.5394234657287598,
      "learning_rate": 2.3590583379536535e-05,
      "loss": 1.5506,
      "step": 21878
    },
    {
      "epoch": 0.7835335828245027,
      "grad_norm": 1.5413225889205933,
      "learning_rate": 2.358310131550773e-05,
      "loss": 1.5083,
      "step": 21879
    },
    {
      "epoch": 0.783569394954071,
      "grad_norm": 1.6394191980361938,
      "learning_rate": 2.3575620279569743e-05,
      "loss": 1.3997,
      "step": 21880
    },
    {
      "epoch": 0.7836052070836392,
      "grad_norm": 1.2522400617599487,
      "learning_rate": 2.3568140271823147e-05,
      "loss": 1.5342,
      "step": 21881
    },
    {
      "epoch": 0.7836410192132075,
      "grad_norm": 1.663016676902771,
      "learning_rate": 2.3560661292368702e-05,
      "loss": 1.2195,
      "step": 21882
    },
    {
      "epoch": 0.7836768313427758,
      "grad_norm": 1.80072820186615,
      "learning_rate": 2.355318334130695e-05,
      "loss": 1.3811,
      "step": 21883
    },
    {
      "epoch": 0.783712643472344,
      "grad_norm": 1.770574688911438,
      "learning_rate": 2.3545706418738476e-05,
      "loss": 1.476,
      "step": 21884
    },
    {
      "epoch": 0.7837484556019124,
      "grad_norm": 1.4288097620010376,
      "learning_rate": 2.3538230524763914e-05,
      "loss": 1.4681,
      "step": 21885
    },
    {
      "epoch": 0.7837842677314807,
      "grad_norm": 1.5454676151275635,
      "learning_rate": 2.353075565948383e-05,
      "loss": 1.1621,
      "step": 21886
    },
    {
      "epoch": 0.783820079861049,
      "grad_norm": 1.748083472251892,
      "learning_rate": 2.352328182299881e-05,
      "loss": 1.2265,
      "step": 21887
    },
    {
      "epoch": 0.7838558919906172,
      "grad_norm": 1.7457653284072876,
      "learning_rate": 2.351580901540933e-05,
      "loss": 1.3389,
      "step": 21888
    },
    {
      "epoch": 0.7838917041201855,
      "grad_norm": 1.464296817779541,
      "learning_rate": 2.3508337236816047e-05,
      "loss": 1.4847,
      "step": 21889
    },
    {
      "epoch": 0.7839275162497538,
      "grad_norm": 4.318163871765137,
      "learning_rate": 2.3500866487319384e-05,
      "loss": 1.2113,
      "step": 21890
    },
    {
      "epoch": 0.783963328379322,
      "grad_norm": 2.506546974182129,
      "learning_rate": 2.3493396767019915e-05,
      "loss": 1.8466,
      "step": 21891
    },
    {
      "epoch": 0.7839991405088904,
      "grad_norm": 1.6039185523986816,
      "learning_rate": 2.348592807601808e-05,
      "loss": 1.4501,
      "step": 21892
    },
    {
      "epoch": 0.7840349526384587,
      "grad_norm": 1.9799975156784058,
      "learning_rate": 2.3478460414414382e-05,
      "loss": 1.2276,
      "step": 21893
    },
    {
      "epoch": 0.784070764768027,
      "grad_norm": 1.671531319618225,
      "learning_rate": 2.3470993782309324e-05,
      "loss": 1.3684,
      "step": 21894
    },
    {
      "epoch": 0.7841065768975952,
      "grad_norm": 1.4565225839614868,
      "learning_rate": 2.3463528179803305e-05,
      "loss": 1.2772,
      "step": 21895
    },
    {
      "epoch": 0.7841423890271635,
      "grad_norm": 1.440674901008606,
      "learning_rate": 2.3456063606996783e-05,
      "loss": 1.4639,
      "step": 21896
    },
    {
      "epoch": 0.7841782011567318,
      "grad_norm": 3.0526788234710693,
      "learning_rate": 2.34486000639902e-05,
      "loss": 1.2457,
      "step": 21897
    },
    {
      "epoch": 0.7842140132863,
      "grad_norm": 1.7137478590011597,
      "learning_rate": 2.3441137550883974e-05,
      "loss": 1.3624,
      "step": 21898
    },
    {
      "epoch": 0.7842498254158684,
      "grad_norm": 1.5892608165740967,
      "learning_rate": 2.3433676067778465e-05,
      "loss": 1.6998,
      "step": 21899
    },
    {
      "epoch": 0.7842856375454367,
      "grad_norm": 1.6187303066253662,
      "learning_rate": 2.3426215614774094e-05,
      "loss": 1.4802,
      "step": 21900
    },
    {
      "epoch": 0.784321449675005,
      "grad_norm": 1.3965293169021606,
      "learning_rate": 2.3418756191971235e-05,
      "loss": 1.6476,
      "step": 21901
    },
    {
      "epoch": 0.7843572618045732,
      "grad_norm": 1.4413694143295288,
      "learning_rate": 2.34112977994702e-05,
      "loss": 1.5815,
      "step": 21902
    },
    {
      "epoch": 0.7843930739341415,
      "grad_norm": 2.764780282974243,
      "learning_rate": 2.340384043737136e-05,
      "loss": 1.3733,
      "step": 21903
    },
    {
      "epoch": 0.7844288860637098,
      "grad_norm": 1.5479607582092285,
      "learning_rate": 2.339638410577505e-05,
      "loss": 1.4778,
      "step": 21904
    },
    {
      "epoch": 0.784464698193278,
      "grad_norm": 1.5237574577331543,
      "learning_rate": 2.3388928804781608e-05,
      "loss": 1.1754,
      "step": 21905
    },
    {
      "epoch": 0.7845005103228464,
      "grad_norm": 2.09736704826355,
      "learning_rate": 2.3381474534491276e-05,
      "loss": 1.315,
      "step": 21906
    },
    {
      "epoch": 0.7845363224524147,
      "grad_norm": 1.7399797439575195,
      "learning_rate": 2.337402129500438e-05,
      "loss": 1.087,
      "step": 21907
    },
    {
      "epoch": 0.7845721345819829,
      "grad_norm": 2.0457167625427246,
      "learning_rate": 2.3366569086421175e-05,
      "loss": 1.4334,
      "step": 21908
    },
    {
      "epoch": 0.7846079467115512,
      "grad_norm": 1.504894733428955,
      "learning_rate": 2.3359117908841966e-05,
      "loss": 1.5605,
      "step": 21909
    },
    {
      "epoch": 0.7846437588411195,
      "grad_norm": 1.2692909240722656,
      "learning_rate": 2.3351667762366948e-05,
      "loss": 1.4162,
      "step": 21910
    },
    {
      "epoch": 0.7846795709706877,
      "grad_norm": 1.8870021104812622,
      "learning_rate": 2.334421864709636e-05,
      "loss": 1.6566,
      "step": 21911
    },
    {
      "epoch": 0.784715383100256,
      "grad_norm": 2.1324985027313232,
      "learning_rate": 2.3336770563130463e-05,
      "loss": 1.5475,
      "step": 21912
    },
    {
      "epoch": 0.7847511952298244,
      "grad_norm": 2.145512580871582,
      "learning_rate": 2.33293235105694e-05,
      "loss": 1.5966,
      "step": 21913
    },
    {
      "epoch": 0.7847870073593927,
      "grad_norm": 1.569747805595398,
      "learning_rate": 2.332187748951339e-05,
      "loss": 1.2442,
      "step": 21914
    },
    {
      "epoch": 0.7848228194889609,
      "grad_norm": 1.6398723125457764,
      "learning_rate": 2.331443250006261e-05,
      "loss": 1.4996,
      "step": 21915
    },
    {
      "epoch": 0.7848586316185292,
      "grad_norm": 1.8749650716781616,
      "learning_rate": 2.3306988542317255e-05,
      "loss": 1.2954,
      "step": 21916
    },
    {
      "epoch": 0.7848944437480975,
      "grad_norm": 1.3666261434555054,
      "learning_rate": 2.3299545616377415e-05,
      "loss": 1.1951,
      "step": 21917
    },
    {
      "epoch": 0.7849302558776657,
      "grad_norm": 1.1874198913574219,
      "learning_rate": 2.329210372234325e-05,
      "loss": 1.4692,
      "step": 21918
    },
    {
      "epoch": 0.784966068007234,
      "grad_norm": 1.760377049446106,
      "learning_rate": 2.3284662860314922e-05,
      "loss": 1.1574,
      "step": 21919
    },
    {
      "epoch": 0.7850018801368024,
      "grad_norm": 1.4190868139266968,
      "learning_rate": 2.3277223030392458e-05,
      "loss": 1.2522,
      "step": 21920
    },
    {
      "epoch": 0.7850376922663707,
      "grad_norm": 1.8131829500198364,
      "learning_rate": 2.3269784232675995e-05,
      "loss": 1.4429,
      "step": 21921
    },
    {
      "epoch": 0.7850735043959389,
      "grad_norm": 1.7088327407836914,
      "learning_rate": 2.3262346467265605e-05,
      "loss": 1.3675,
      "step": 21922
    },
    {
      "epoch": 0.7851093165255072,
      "grad_norm": 1.543448567390442,
      "learning_rate": 2.3254909734261398e-05,
      "loss": 1.5017,
      "step": 21923
    },
    {
      "epoch": 0.7851451286550755,
      "grad_norm": 2.0293030738830566,
      "learning_rate": 2.324747403376336e-05,
      "loss": 1.5802,
      "step": 21924
    },
    {
      "epoch": 0.7851809407846437,
      "grad_norm": 1.8709771633148193,
      "learning_rate": 2.3240039365871546e-05,
      "loss": 1.7063,
      "step": 21925
    },
    {
      "epoch": 0.785216752914212,
      "grad_norm": 1.7643381357192993,
      "learning_rate": 2.3232605730686018e-05,
      "loss": 1.4052,
      "step": 21926
    },
    {
      "epoch": 0.7852525650437804,
      "grad_norm": 1.706915020942688,
      "learning_rate": 2.3225173128306733e-05,
      "loss": 1.2809,
      "step": 21927
    },
    {
      "epoch": 0.7852883771733487,
      "grad_norm": 1.7989875078201294,
      "learning_rate": 2.3217741558833706e-05,
      "loss": 1.5696,
      "step": 21928
    },
    {
      "epoch": 0.7853241893029169,
      "grad_norm": 1.6285196542739868,
      "learning_rate": 2.321031102236694e-05,
      "loss": 1.1875,
      "step": 21929
    },
    {
      "epoch": 0.7853600014324852,
      "grad_norm": 1.5075069665908813,
      "learning_rate": 2.3202881519006393e-05,
      "loss": 1.3279,
      "step": 21930
    },
    {
      "epoch": 0.7853958135620535,
      "grad_norm": 1.7992472648620605,
      "learning_rate": 2.3195453048852e-05,
      "loss": 1.5811,
      "step": 21931
    },
    {
      "epoch": 0.7854316256916217,
      "grad_norm": 1.7869278192520142,
      "learning_rate": 2.3188025612003718e-05,
      "loss": 1.7683,
      "step": 21932
    },
    {
      "epoch": 0.78546743782119,
      "grad_norm": 1.4537584781646729,
      "learning_rate": 2.318059920856146e-05,
      "loss": 1.3219,
      "step": 21933
    },
    {
      "epoch": 0.7855032499507584,
      "grad_norm": 1.5661792755126953,
      "learning_rate": 2.3173173838625183e-05,
      "loss": 1.4321,
      "step": 21934
    },
    {
      "epoch": 0.7855390620803266,
      "grad_norm": 1.5315093994140625,
      "learning_rate": 2.316574950229472e-05,
      "loss": 1.5277,
      "step": 21935
    },
    {
      "epoch": 0.7855748742098949,
      "grad_norm": 2.292695999145508,
      "learning_rate": 2.315832619967e-05,
      "loss": 1.0601,
      "step": 21936
    },
    {
      "epoch": 0.7856106863394632,
      "grad_norm": 1.7151291370391846,
      "learning_rate": 2.3150903930850896e-05,
      "loss": 1.4839,
      "step": 21937
    },
    {
      "epoch": 0.7856464984690315,
      "grad_norm": 1.6335084438323975,
      "learning_rate": 2.3143482695937235e-05,
      "loss": 1.2555,
      "step": 21938
    },
    {
      "epoch": 0.7856823105985997,
      "grad_norm": 1.5263906717300415,
      "learning_rate": 2.3136062495028876e-05,
      "loss": 1.4354,
      "step": 21939
    },
    {
      "epoch": 0.785718122728168,
      "grad_norm": 1.3875404596328735,
      "learning_rate": 2.312864332822564e-05,
      "loss": 1.4227,
      "step": 21940
    },
    {
      "epoch": 0.7857539348577364,
      "grad_norm": 2.1404190063476562,
      "learning_rate": 2.3121225195627382e-05,
      "loss": 1.6252,
      "step": 21941
    },
    {
      "epoch": 0.7857897469873046,
      "grad_norm": 1.6052498817443848,
      "learning_rate": 2.3113808097333854e-05,
      "loss": 1.5175,
      "step": 21942
    },
    {
      "epoch": 0.7858255591168729,
      "grad_norm": 1.234230637550354,
      "learning_rate": 2.3106392033444856e-05,
      "loss": 1.189,
      "step": 21943
    },
    {
      "epoch": 0.7858613712464412,
      "grad_norm": 1.5555446147918701,
      "learning_rate": 2.3098977004060185e-05,
      "loss": 1.3555,
      "step": 21944
    },
    {
      "epoch": 0.7858971833760094,
      "grad_norm": 1.4354848861694336,
      "learning_rate": 2.309156300927957e-05,
      "loss": 1.4896,
      "step": 21945
    },
    {
      "epoch": 0.7859329955055777,
      "grad_norm": 1.6179566383361816,
      "learning_rate": 2.308415004920277e-05,
      "loss": 1.3944,
      "step": 21946
    },
    {
      "epoch": 0.785968807635146,
      "grad_norm": 2.2030816078186035,
      "learning_rate": 2.307673812392951e-05,
      "loss": 1.605,
      "step": 21947
    },
    {
      "epoch": 0.7860046197647144,
      "grad_norm": 1.3560844659805298,
      "learning_rate": 2.3069327233559533e-05,
      "loss": 1.143,
      "step": 21948
    },
    {
      "epoch": 0.7860404318942826,
      "grad_norm": 1.650636076927185,
      "learning_rate": 2.306191737819251e-05,
      "loss": 1.3462,
      "step": 21949
    },
    {
      "epoch": 0.7860762440238509,
      "grad_norm": 1.7846274375915527,
      "learning_rate": 2.3054508557928144e-05,
      "loss": 1.6821,
      "step": 21950
    },
    {
      "epoch": 0.7861120561534192,
      "grad_norm": 1.6764312982559204,
      "learning_rate": 2.3047100772866114e-05,
      "loss": 1.4222,
      "step": 21951
    },
    {
      "epoch": 0.7861478682829874,
      "grad_norm": 2.348435878753662,
      "learning_rate": 2.3039694023106106e-05,
      "loss": 1.341,
      "step": 21952
    },
    {
      "epoch": 0.7861836804125557,
      "grad_norm": 1.734067678451538,
      "learning_rate": 2.303228830874775e-05,
      "loss": 1.3237,
      "step": 21953
    },
    {
      "epoch": 0.786219492542124,
      "grad_norm": 2.134556770324707,
      "learning_rate": 2.3024883629890604e-05,
      "loss": 1.5639,
      "step": 21954
    },
    {
      "epoch": 0.7862553046716924,
      "grad_norm": 1.831900954246521,
      "learning_rate": 2.3017479986634426e-05,
      "loss": 1.4596,
      "step": 21955
    },
    {
      "epoch": 0.7862911168012606,
      "grad_norm": 2.242368698120117,
      "learning_rate": 2.3010077379078722e-05,
      "loss": 1.2863,
      "step": 21956
    },
    {
      "epoch": 0.7863269289308289,
      "grad_norm": 1.671486496925354,
      "learning_rate": 2.300267580732315e-05,
      "loss": 1.3689,
      "step": 21957
    },
    {
      "epoch": 0.7863627410603972,
      "grad_norm": 1.524246335029602,
      "learning_rate": 2.2995275271467187e-05,
      "loss": 1.4023,
      "step": 21958
    },
    {
      "epoch": 0.7863985531899654,
      "grad_norm": 2.152644157409668,
      "learning_rate": 2.2987875771610534e-05,
      "loss": 1.3216,
      "step": 21959
    },
    {
      "epoch": 0.7864343653195337,
      "grad_norm": 1.521066427230835,
      "learning_rate": 2.2980477307852642e-05,
      "loss": 1.1968,
      "step": 21960
    },
    {
      "epoch": 0.786470177449102,
      "grad_norm": 1.6909211874008179,
      "learning_rate": 2.297307988029308e-05,
      "loss": 1.565,
      "step": 21961
    },
    {
      "epoch": 0.7865059895786704,
      "grad_norm": 1.748152732849121,
      "learning_rate": 2.29656834890314e-05,
      "loss": 1.3081,
      "step": 21962
    },
    {
      "epoch": 0.7865418017082386,
      "grad_norm": 1.6962015628814697,
      "learning_rate": 2.2958288134167048e-05,
      "loss": 1.2831,
      "step": 21963
    },
    {
      "epoch": 0.7865776138378069,
      "grad_norm": 1.4486021995544434,
      "learning_rate": 2.295089381579959e-05,
      "loss": 1.4332,
      "step": 21964
    },
    {
      "epoch": 0.7866134259673752,
      "grad_norm": 1.4903663396835327,
      "learning_rate": 2.2943500534028406e-05,
      "loss": 1.4113,
      "step": 21965
    },
    {
      "epoch": 0.7866492380969434,
      "grad_norm": 1.5759414434432983,
      "learning_rate": 2.2936108288953083e-05,
      "loss": 1.2901,
      "step": 21966
    },
    {
      "epoch": 0.7866850502265117,
      "grad_norm": 1.741335153579712,
      "learning_rate": 2.292871708067299e-05,
      "loss": 1.257,
      "step": 21967
    },
    {
      "epoch": 0.78672086235608,
      "grad_norm": 2.2893245220184326,
      "learning_rate": 2.2921326909287634e-05,
      "loss": 1.5896,
      "step": 21968
    },
    {
      "epoch": 0.7867566744856483,
      "grad_norm": 1.9488850831985474,
      "learning_rate": 2.291393777489632e-05,
      "loss": 1.4638,
      "step": 21969
    },
    {
      "epoch": 0.7867924866152166,
      "grad_norm": 1.3496569395065308,
      "learning_rate": 2.290654967759862e-05,
      "loss": 1.6248,
      "step": 21970
    },
    {
      "epoch": 0.7868282987447849,
      "grad_norm": 2.044699192047119,
      "learning_rate": 2.289916261749383e-05,
      "loss": 1.306,
      "step": 21971
    },
    {
      "epoch": 0.7868641108743532,
      "grad_norm": 1.3044853210449219,
      "learning_rate": 2.2891776594681315e-05,
      "loss": 1.4078,
      "step": 21972
    },
    {
      "epoch": 0.7868999230039214,
      "grad_norm": 1.7480604648590088,
      "learning_rate": 2.2884391609260525e-05,
      "loss": 1.2894,
      "step": 21973
    },
    {
      "epoch": 0.7869357351334897,
      "grad_norm": 1.638363242149353,
      "learning_rate": 2.2877007661330762e-05,
      "loss": 1.638,
      "step": 21974
    },
    {
      "epoch": 0.786971547263058,
      "grad_norm": 1.6407605409622192,
      "learning_rate": 2.2869624750991393e-05,
      "loss": 1.3685,
      "step": 21975
    },
    {
      "epoch": 0.7870073593926263,
      "grad_norm": 1.5192142724990845,
      "learning_rate": 2.2862242878341678e-05,
      "loss": 1.8149,
      "step": 21976
    },
    {
      "epoch": 0.7870431715221946,
      "grad_norm": 1.6428052186965942,
      "learning_rate": 2.285486204348105e-05,
      "loss": 1.4984,
      "step": 21977
    },
    {
      "epoch": 0.7870789836517629,
      "grad_norm": 1.9950889348983765,
      "learning_rate": 2.284748224650871e-05,
      "loss": 1.5231,
      "step": 21978
    },
    {
      "epoch": 0.7871147957813311,
      "grad_norm": 2.0278079509735107,
      "learning_rate": 2.2840103487524e-05,
      "loss": 1.3566,
      "step": 21979
    },
    {
      "epoch": 0.7871506079108994,
      "grad_norm": 1.671902060508728,
      "learning_rate": 2.283272576662615e-05,
      "loss": 1.5062,
      "step": 21980
    },
    {
      "epoch": 0.7871864200404677,
      "grad_norm": 1.5835262537002563,
      "learning_rate": 2.2825349083914426e-05,
      "loss": 1.4686,
      "step": 21981
    },
    {
      "epoch": 0.787222232170036,
      "grad_norm": 1.7632768154144287,
      "learning_rate": 2.2817973439488117e-05,
      "loss": 1.3356,
      "step": 21982
    },
    {
      "epoch": 0.7872580442996042,
      "grad_norm": 1.718269944190979,
      "learning_rate": 2.2810598833446382e-05,
      "loss": 1.306,
      "step": 21983
    },
    {
      "epoch": 0.7872938564291726,
      "grad_norm": 1.9307987689971924,
      "learning_rate": 2.2803225265888484e-05,
      "loss": 1.5905,
      "step": 21984
    },
    {
      "epoch": 0.7873296685587409,
      "grad_norm": 1.7241172790527344,
      "learning_rate": 2.2795852736913604e-05,
      "loss": 1.4503,
      "step": 21985
    },
    {
      "epoch": 0.7873654806883091,
      "grad_norm": 1.7079567909240723,
      "learning_rate": 2.2788481246620973e-05,
      "loss": 1.2879,
      "step": 21986
    },
    {
      "epoch": 0.7874012928178774,
      "grad_norm": 1.3769007921218872,
      "learning_rate": 2.2781110795109674e-05,
      "loss": 1.525,
      "step": 21987
    },
    {
      "epoch": 0.7874371049474457,
      "grad_norm": 1.4252955913543701,
      "learning_rate": 2.2773741382478975e-05,
      "loss": 1.5648,
      "step": 21988
    },
    {
      "epoch": 0.787472917077014,
      "grad_norm": 1.4553250074386597,
      "learning_rate": 2.276637300882797e-05,
      "loss": 1.5559,
      "step": 21989
    },
    {
      "epoch": 0.7875087292065822,
      "grad_norm": 1.6732171773910522,
      "learning_rate": 2.2759005674255774e-05,
      "loss": 1.2551,
      "step": 21990
    },
    {
      "epoch": 0.7875445413361506,
      "grad_norm": 1.594787836074829,
      "learning_rate": 2.275163937886151e-05,
      "loss": 1.4659,
      "step": 21991
    },
    {
      "epoch": 0.7875803534657189,
      "grad_norm": 1.33848237991333,
      "learning_rate": 2.2744274122744304e-05,
      "loss": 1.2285,
      "step": 21992
    },
    {
      "epoch": 0.7876161655952871,
      "grad_norm": 1.8487385511398315,
      "learning_rate": 2.2736909906003266e-05,
      "loss": 1.879,
      "step": 21993
    },
    {
      "epoch": 0.7876519777248554,
      "grad_norm": 1.378639817237854,
      "learning_rate": 2.2729546728737416e-05,
      "loss": 1.6364,
      "step": 21994
    },
    {
      "epoch": 0.7876877898544237,
      "grad_norm": 1.6893432140350342,
      "learning_rate": 2.2722184591045835e-05,
      "loss": 1.5968,
      "step": 21995
    },
    {
      "epoch": 0.7877236019839919,
      "grad_norm": 1.7341252565383911,
      "learning_rate": 2.2714823493027583e-05,
      "loss": 1.5264,
      "step": 21996
    },
    {
      "epoch": 0.7877594141135602,
      "grad_norm": 1.3586167097091675,
      "learning_rate": 2.2707463434781718e-05,
      "loss": 1.3724,
      "step": 21997
    },
    {
      "epoch": 0.7877952262431286,
      "grad_norm": 1.5334831476211548,
      "learning_rate": 2.2700104416407208e-05,
      "loss": 1.736,
      "step": 21998
    },
    {
      "epoch": 0.7878310383726969,
      "grad_norm": 1.5055005550384521,
      "learning_rate": 2.2692746438003078e-05,
      "loss": 1.2053,
      "step": 21999
    },
    {
      "epoch": 0.7878668505022651,
      "grad_norm": 2.348088502883911,
      "learning_rate": 2.2685389499668352e-05,
      "loss": 1.5909,
      "step": 22000
    },
    {
      "epoch": 0.7879026626318334,
      "grad_norm": 1.4939537048339844,
      "learning_rate": 2.2678033601501957e-05,
      "loss": 1.5303,
      "step": 22001
    },
    {
      "epoch": 0.7879384747614017,
      "grad_norm": 1.1772562265396118,
      "learning_rate": 2.2670678743602892e-05,
      "loss": 1.2715,
      "step": 22002
    },
    {
      "epoch": 0.7879742868909699,
      "grad_norm": 1.5099385976791382,
      "learning_rate": 2.2663324926070086e-05,
      "loss": 1.3447,
      "step": 22003
    },
    {
      "epoch": 0.7880100990205382,
      "grad_norm": 1.5559260845184326,
      "learning_rate": 2.2655972149002512e-05,
      "loss": 1.1992,
      "step": 22004
    },
    {
      "epoch": 0.7880459111501066,
      "grad_norm": 2.2137198448181152,
      "learning_rate": 2.2648620412499045e-05,
      "loss": 1.7298,
      "step": 22005
    },
    {
      "epoch": 0.7880817232796749,
      "grad_norm": 1.509599208831787,
      "learning_rate": 2.264126971665861e-05,
      "loss": 1.1957,
      "step": 22006
    },
    {
      "epoch": 0.7881175354092431,
      "grad_norm": 1.8853390216827393,
      "learning_rate": 2.2633920061580127e-05,
      "loss": 1.7451,
      "step": 22007
    },
    {
      "epoch": 0.7881533475388114,
      "grad_norm": 1.300235629081726,
      "learning_rate": 2.262657144736243e-05,
      "loss": 1.2197,
      "step": 22008
    },
    {
      "epoch": 0.7881891596683797,
      "grad_norm": 1.5796751976013184,
      "learning_rate": 2.2619223874104423e-05,
      "loss": 1.5641,
      "step": 22009
    },
    {
      "epoch": 0.7882249717979479,
      "grad_norm": 1.7323791980743408,
      "learning_rate": 2.261187734190493e-05,
      "loss": 1.5017,
      "step": 22010
    },
    {
      "epoch": 0.7882607839275162,
      "grad_norm": 1.4595657587051392,
      "learning_rate": 2.2604531850862832e-05,
      "loss": 1.5183,
      "step": 22011
    },
    {
      "epoch": 0.7882965960570846,
      "grad_norm": 2.023123264312744,
      "learning_rate": 2.2597187401076903e-05,
      "loss": 1.6822,
      "step": 22012
    },
    {
      "epoch": 0.7883324081866528,
      "grad_norm": 1.8436816930770874,
      "learning_rate": 2.2589843992645977e-05,
      "loss": 1.6796,
      "step": 22013
    },
    {
      "epoch": 0.7883682203162211,
      "grad_norm": 1.3676259517669678,
      "learning_rate": 2.258250162566887e-05,
      "loss": 1.4,
      "step": 22014
    },
    {
      "epoch": 0.7884040324457894,
      "grad_norm": 1.615670084953308,
      "learning_rate": 2.2575160300244314e-05,
      "loss": 1.2904,
      "step": 22015
    },
    {
      "epoch": 0.7884398445753577,
      "grad_norm": 1.8132332563400269,
      "learning_rate": 2.2567820016471107e-05,
      "loss": 1.19,
      "step": 22016
    },
    {
      "epoch": 0.7884756567049259,
      "grad_norm": 1.5410048961639404,
      "learning_rate": 2.256048077444801e-05,
      "loss": 1.4224,
      "step": 22017
    },
    {
      "epoch": 0.7885114688344942,
      "grad_norm": 1.866053581237793,
      "learning_rate": 2.2553142574273777e-05,
      "loss": 1.4266,
      "step": 22018
    },
    {
      "epoch": 0.7885472809640626,
      "grad_norm": 1.684851884841919,
      "learning_rate": 2.2545805416047073e-05,
      "loss": 1.4562,
      "step": 22019
    },
    {
      "epoch": 0.7885830930936308,
      "grad_norm": 1.7239848375320435,
      "learning_rate": 2.253846929986666e-05,
      "loss": 1.2378,
      "step": 22020
    },
    {
      "epoch": 0.7886189052231991,
      "grad_norm": 2.0426554679870605,
      "learning_rate": 2.253113422583122e-05,
      "loss": 1.1809,
      "step": 22021
    },
    {
      "epoch": 0.7886547173527674,
      "grad_norm": 1.4321867227554321,
      "learning_rate": 2.252380019403947e-05,
      "loss": 1.1409,
      "step": 22022
    },
    {
      "epoch": 0.7886905294823356,
      "grad_norm": 1.3526005744934082,
      "learning_rate": 2.251646720459003e-05,
      "loss": 1.1834,
      "step": 22023
    },
    {
      "epoch": 0.7887263416119039,
      "grad_norm": 1.6505836248397827,
      "learning_rate": 2.250913525758157e-05,
      "loss": 1.3677,
      "step": 22024
    },
    {
      "epoch": 0.7887621537414722,
      "grad_norm": 1.5732851028442383,
      "learning_rate": 2.2501804353112765e-05,
      "loss": 1.6171,
      "step": 22025
    },
    {
      "epoch": 0.7887979658710406,
      "grad_norm": 1.22661554813385,
      "learning_rate": 2.249447449128219e-05,
      "loss": 1.4751,
      "step": 22026
    },
    {
      "epoch": 0.7888337780006088,
      "grad_norm": 1.3099799156188965,
      "learning_rate": 2.248714567218849e-05,
      "loss": 1.1526,
      "step": 22027
    },
    {
      "epoch": 0.7888695901301771,
      "grad_norm": 2.0355942249298096,
      "learning_rate": 2.2479817895930256e-05,
      "loss": 1.4029,
      "step": 22028
    },
    {
      "epoch": 0.7889054022597454,
      "grad_norm": 1.6288301944732666,
      "learning_rate": 2.247249116260611e-05,
      "loss": 1.5374,
      "step": 22029
    },
    {
      "epoch": 0.7889412143893136,
      "grad_norm": 1.6688481569290161,
      "learning_rate": 2.2465165472314564e-05,
      "loss": 1.449,
      "step": 22030
    },
    {
      "epoch": 0.7889770265188819,
      "grad_norm": 1.6263892650604248,
      "learning_rate": 2.2457840825154198e-05,
      "loss": 1.3263,
      "step": 22031
    },
    {
      "epoch": 0.7890128386484502,
      "grad_norm": 1.6034423112869263,
      "learning_rate": 2.24505172212236e-05,
      "loss": 1.4507,
      "step": 22032
    },
    {
      "epoch": 0.7890486507780186,
      "grad_norm": 2.850820541381836,
      "learning_rate": 2.2443194660621225e-05,
      "loss": 1.6127,
      "step": 22033
    },
    {
      "epoch": 0.7890844629075868,
      "grad_norm": 1.3363112211227417,
      "learning_rate": 2.243587314344563e-05,
      "loss": 1.3583,
      "step": 22034
    },
    {
      "epoch": 0.7891202750371551,
      "grad_norm": 1.8845534324645996,
      "learning_rate": 2.242855266979531e-05,
      "loss": 1.353,
      "step": 22035
    },
    {
      "epoch": 0.7891560871667234,
      "grad_norm": 1.926565170288086,
      "learning_rate": 2.242123323976878e-05,
      "loss": 1.6149,
      "step": 22036
    },
    {
      "epoch": 0.7891918992962916,
      "grad_norm": 1.9906445741653442,
      "learning_rate": 2.2413914853464455e-05,
      "loss": 1.466,
      "step": 22037
    },
    {
      "epoch": 0.7892277114258599,
      "grad_norm": 1.8436861038208008,
      "learning_rate": 2.240659751098083e-05,
      "loss": 1.4323,
      "step": 22038
    },
    {
      "epoch": 0.7892635235554282,
      "grad_norm": 1.6157275438308716,
      "learning_rate": 2.2399281212416346e-05,
      "loss": 1.3719,
      "step": 22039
    },
    {
      "epoch": 0.7892993356849966,
      "grad_norm": 1.502375841140747,
      "learning_rate": 2.2391965957869464e-05,
      "loss": 1.5709,
      "step": 22040
    },
    {
      "epoch": 0.7893351478145648,
      "grad_norm": 3.7131903171539307,
      "learning_rate": 2.2384651747438578e-05,
      "loss": 1.413,
      "step": 22041
    },
    {
      "epoch": 0.7893709599441331,
      "grad_norm": 1.402735948562622,
      "learning_rate": 2.237733858122203e-05,
      "loss": 1.5297,
      "step": 22042
    },
    {
      "epoch": 0.7894067720737014,
      "grad_norm": 1.9113142490386963,
      "learning_rate": 2.2370026459318315e-05,
      "loss": 1.4642,
      "step": 22043
    },
    {
      "epoch": 0.7894425842032696,
      "grad_norm": 2.179842710494995,
      "learning_rate": 2.236271538182574e-05,
      "loss": 1.5229,
      "step": 22044
    },
    {
      "epoch": 0.7894783963328379,
      "grad_norm": 1.7188292741775513,
      "learning_rate": 2.2355405348842672e-05,
      "loss": 1.3425,
      "step": 22045
    },
    {
      "epoch": 0.7895142084624062,
      "grad_norm": 2.006430149078369,
      "learning_rate": 2.2348096360467484e-05,
      "loss": 1.4632,
      "step": 22046
    },
    {
      "epoch": 0.7895500205919745,
      "grad_norm": 1.5962008237838745,
      "learning_rate": 2.2340788416798518e-05,
      "loss": 1.4263,
      "step": 22047
    },
    {
      "epoch": 0.7895858327215428,
      "grad_norm": 1.4921599626541138,
      "learning_rate": 2.233348151793404e-05,
      "loss": 1.2119,
      "step": 22048
    },
    {
      "epoch": 0.7896216448511111,
      "grad_norm": 1.3751320838928223,
      "learning_rate": 2.232617566397238e-05,
      "loss": 1.2617,
      "step": 22049
    },
    {
      "epoch": 0.7896574569806794,
      "grad_norm": 1.339401364326477,
      "learning_rate": 2.2318870855011874e-05,
      "loss": 1.5102,
      "step": 22050
    },
    {
      "epoch": 0.7896932691102476,
      "grad_norm": 1.2918152809143066,
      "learning_rate": 2.231156709115073e-05,
      "loss": 1.358,
      "step": 22051
    },
    {
      "epoch": 0.7897290812398159,
      "grad_norm": 1.8511288166046143,
      "learning_rate": 2.230426437248726e-05,
      "loss": 1.1148,
      "step": 22052
    },
    {
      "epoch": 0.7897648933693842,
      "grad_norm": 1.64247727394104,
      "learning_rate": 2.229696269911965e-05,
      "loss": 1.1974,
      "step": 22053
    },
    {
      "epoch": 0.7898007054989525,
      "grad_norm": 1.3478108644485474,
      "learning_rate": 2.228966207114622e-05,
      "loss": 1.6666,
      "step": 22054
    },
    {
      "epoch": 0.7898365176285208,
      "grad_norm": 1.3576805591583252,
      "learning_rate": 2.228236248866512e-05,
      "loss": 1.3979,
      "step": 22055
    },
    {
      "epoch": 0.7898723297580891,
      "grad_norm": 2.1761605739593506,
      "learning_rate": 2.2275063951774587e-05,
      "loss": 1.5848,
      "step": 22056
    },
    {
      "epoch": 0.7899081418876573,
      "grad_norm": 1.8470492362976074,
      "learning_rate": 2.2267766460572814e-05,
      "loss": 1.3471,
      "step": 22057
    },
    {
      "epoch": 0.7899439540172256,
      "grad_norm": 1.5361154079437256,
      "learning_rate": 2.226047001515801e-05,
      "loss": 1.6824,
      "step": 22058
    },
    {
      "epoch": 0.7899797661467939,
      "grad_norm": 1.5846877098083496,
      "learning_rate": 2.225317461562829e-05,
      "loss": 1.3616,
      "step": 22059
    },
    {
      "epoch": 0.7900155782763622,
      "grad_norm": 1.971751093864441,
      "learning_rate": 2.2245880262081774e-05,
      "loss": 1.6053,
      "step": 22060
    },
    {
      "epoch": 0.7900513904059305,
      "grad_norm": 1.9126818180084229,
      "learning_rate": 2.223858695461669e-05,
      "loss": 1.4891,
      "step": 22061
    },
    {
      "epoch": 0.7900872025354988,
      "grad_norm": 1.4563158750534058,
      "learning_rate": 2.2231294693331096e-05,
      "loss": 1.4378,
      "step": 22062
    },
    {
      "epoch": 0.7901230146650671,
      "grad_norm": 1.2912527322769165,
      "learning_rate": 2.222400347832314e-05,
      "loss": 1.4947,
      "step": 22063
    },
    {
      "epoch": 0.7901588267946353,
      "grad_norm": 1.532296061515808,
      "learning_rate": 2.221671330969084e-05,
      "loss": 1.4986,
      "step": 22064
    },
    {
      "epoch": 0.7901946389242036,
      "grad_norm": 1.5457974672317505,
      "learning_rate": 2.220942418753238e-05,
      "loss": 1.4653,
      "step": 22065
    },
    {
      "epoch": 0.7902304510537719,
      "grad_norm": 1.6285170316696167,
      "learning_rate": 2.220213611194576e-05,
      "loss": 1.5869,
      "step": 22066
    },
    {
      "epoch": 0.7902662631833401,
      "grad_norm": 1.7709981203079224,
      "learning_rate": 2.2194849083029057e-05,
      "loss": 1.2938,
      "step": 22067
    },
    {
      "epoch": 0.7903020753129085,
      "grad_norm": 1.4877737760543823,
      "learning_rate": 2.2187563100880282e-05,
      "loss": 1.5076,
      "step": 22068
    },
    {
      "epoch": 0.7903378874424768,
      "grad_norm": 1.7979744672775269,
      "learning_rate": 2.2180278165597467e-05,
      "loss": 1.6471,
      "step": 22069
    },
    {
      "epoch": 0.7903736995720451,
      "grad_norm": 1.7051743268966675,
      "learning_rate": 2.2172994277278668e-05,
      "loss": 1.4707,
      "step": 22070
    },
    {
      "epoch": 0.7904095117016133,
      "grad_norm": 2.3612325191497803,
      "learning_rate": 2.2165711436021774e-05,
      "loss": 1.3351,
      "step": 22071
    },
    {
      "epoch": 0.7904453238311816,
      "grad_norm": 1.6688679456710815,
      "learning_rate": 2.2158429641924895e-05,
      "loss": 1.2814,
      "step": 22072
    },
    {
      "epoch": 0.7904811359607499,
      "grad_norm": 1.6498329639434814,
      "learning_rate": 2.2151148895085906e-05,
      "loss": 1.6461,
      "step": 22073
    },
    {
      "epoch": 0.7905169480903181,
      "grad_norm": 2.0248377323150635,
      "learning_rate": 2.2143869195602816e-05,
      "loss": 1.5359,
      "step": 22074
    },
    {
      "epoch": 0.7905527602198865,
      "grad_norm": 1.48524010181427,
      "learning_rate": 2.2136590543573497e-05,
      "loss": 1.4948,
      "step": 22075
    },
    {
      "epoch": 0.7905885723494548,
      "grad_norm": 1.3780854940414429,
      "learning_rate": 2.212931293909596e-05,
      "loss": 1.2927,
      "step": 22076
    },
    {
      "epoch": 0.7906243844790231,
      "grad_norm": 1.5393905639648438,
      "learning_rate": 2.2122036382268074e-05,
      "loss": 0.9784,
      "step": 22077
    },
    {
      "epoch": 0.7906601966085913,
      "grad_norm": 1.6730303764343262,
      "learning_rate": 2.21147608731877e-05,
      "loss": 1.6592,
      "step": 22078
    },
    {
      "epoch": 0.7906960087381596,
      "grad_norm": 1.3171392679214478,
      "learning_rate": 2.210748641195276e-05,
      "loss": 1.4827,
      "step": 22079
    },
    {
      "epoch": 0.7907318208677279,
      "grad_norm": 1.4257128238677979,
      "learning_rate": 2.210021299866112e-05,
      "loss": 1.3873,
      "step": 22080
    },
    {
      "epoch": 0.7907676329972961,
      "grad_norm": 1.7365443706512451,
      "learning_rate": 2.209294063341065e-05,
      "loss": 1.4258,
      "step": 22081
    },
    {
      "epoch": 0.7908034451268645,
      "grad_norm": 1.7937761545181274,
      "learning_rate": 2.2085669316299117e-05,
      "loss": 1.8,
      "step": 22082
    },
    {
      "epoch": 0.7908392572564328,
      "grad_norm": 2.080672025680542,
      "learning_rate": 2.207839904742446e-05,
      "loss": 1.4081,
      "step": 22083
    },
    {
      "epoch": 0.790875069386001,
      "grad_norm": 2.0380451679229736,
      "learning_rate": 2.2071129826884397e-05,
      "loss": 1.2983,
      "step": 22084
    },
    {
      "epoch": 0.7909108815155693,
      "grad_norm": 1.4107887744903564,
      "learning_rate": 2.2063861654776798e-05,
      "loss": 1.4714,
      "step": 22085
    },
    {
      "epoch": 0.7909466936451376,
      "grad_norm": 1.3979287147521973,
      "learning_rate": 2.205659453119938e-05,
      "loss": 1.2243,
      "step": 22086
    },
    {
      "epoch": 0.7909825057747059,
      "grad_norm": 1.476107120513916,
      "learning_rate": 2.204932845624994e-05,
      "loss": 1.5036,
      "step": 22087
    },
    {
      "epoch": 0.7910183179042741,
      "grad_norm": 1.4876946210861206,
      "learning_rate": 2.204206343002626e-05,
      "loss": 1.4428,
      "step": 22088
    },
    {
      "epoch": 0.7910541300338425,
      "grad_norm": 1.7648842334747314,
      "learning_rate": 2.2034799452626043e-05,
      "loss": 1.5549,
      "step": 22089
    },
    {
      "epoch": 0.7910899421634108,
      "grad_norm": 1.5976637601852417,
      "learning_rate": 2.2027536524147017e-05,
      "loss": 1.6226,
      "step": 22090
    },
    {
      "epoch": 0.791125754292979,
      "grad_norm": 1.9662413597106934,
      "learning_rate": 2.2020274644686922e-05,
      "loss": 1.2422,
      "step": 22091
    },
    {
      "epoch": 0.7911615664225473,
      "grad_norm": 1.4056212902069092,
      "learning_rate": 2.2013013814343465e-05,
      "loss": 1.2724,
      "step": 22092
    },
    {
      "epoch": 0.7911973785521156,
      "grad_norm": 1.9056984186172485,
      "learning_rate": 2.200575403321429e-05,
      "loss": 1.7059,
      "step": 22093
    },
    {
      "epoch": 0.7912331906816839,
      "grad_norm": 1.6030707359313965,
      "learning_rate": 2.1998495301397083e-05,
      "loss": 1.666,
      "step": 22094
    },
    {
      "epoch": 0.7912690028112521,
      "grad_norm": 1.5489169359207153,
      "learning_rate": 2.1991237618989535e-05,
      "loss": 1.2683,
      "step": 22095
    },
    {
      "epoch": 0.7913048149408205,
      "grad_norm": 1.811343789100647,
      "learning_rate": 2.1983980986089235e-05,
      "loss": 1.1581,
      "step": 22096
    },
    {
      "epoch": 0.7913406270703888,
      "grad_norm": 1.4421796798706055,
      "learning_rate": 2.197672540279384e-05,
      "loss": 1.3113,
      "step": 22097
    },
    {
      "epoch": 0.791376439199957,
      "grad_norm": 2.07533860206604,
      "learning_rate": 2.196947086920096e-05,
      "loss": 1.4348,
      "step": 22098
    },
    {
      "epoch": 0.7914122513295253,
      "grad_norm": 1.6728070974349976,
      "learning_rate": 2.1962217385408225e-05,
      "loss": 1.508,
      "step": 22099
    },
    {
      "epoch": 0.7914480634590936,
      "grad_norm": 1.8283659219741821,
      "learning_rate": 2.1954964951513168e-05,
      "loss": 1.5883,
      "step": 22100
    },
    {
      "epoch": 0.7914838755886618,
      "grad_norm": 1.3747072219848633,
      "learning_rate": 2.1947713567613382e-05,
      "loss": 1.1613,
      "step": 22101
    },
    {
      "epoch": 0.7915196877182301,
      "grad_norm": 1.5672385692596436,
      "learning_rate": 2.194046323380643e-05,
      "loss": 1.4233,
      "step": 22102
    },
    {
      "epoch": 0.7915554998477985,
      "grad_norm": 1.5005253553390503,
      "learning_rate": 2.193321395018989e-05,
      "loss": 1.303,
      "step": 22103
    },
    {
      "epoch": 0.7915913119773668,
      "grad_norm": 1.7410175800323486,
      "learning_rate": 2.192596571686123e-05,
      "loss": 1.4531,
      "step": 22104
    },
    {
      "epoch": 0.791627124106935,
      "grad_norm": 3.041870594024658,
      "learning_rate": 2.1918718533917982e-05,
      "loss": 1.7464,
      "step": 22105
    },
    {
      "epoch": 0.7916629362365033,
      "grad_norm": 1.4313814640045166,
      "learning_rate": 2.1911472401457688e-05,
      "loss": 1.426,
      "step": 22106
    },
    {
      "epoch": 0.7916987483660716,
      "grad_norm": 2.0393707752227783,
      "learning_rate": 2.1904227319577786e-05,
      "loss": 1.4193,
      "step": 22107
    },
    {
      "epoch": 0.7917345604956398,
      "grad_norm": 1.6360301971435547,
      "learning_rate": 2.1896983288375773e-05,
      "loss": 1.0499,
      "step": 22108
    },
    {
      "epoch": 0.7917703726252081,
      "grad_norm": 2.0467658042907715,
      "learning_rate": 2.188974030794909e-05,
      "loss": 1.6207,
      "step": 22109
    },
    {
      "epoch": 0.7918061847547765,
      "grad_norm": 1.2687796354293823,
      "learning_rate": 2.1882498378395232e-05,
      "loss": 1.5234,
      "step": 22110
    },
    {
      "epoch": 0.7918419968843448,
      "grad_norm": 1.6729602813720703,
      "learning_rate": 2.1875257499811563e-05,
      "loss": 1.2955,
      "step": 22111
    },
    {
      "epoch": 0.791877809013913,
      "grad_norm": 1.8214075565338135,
      "learning_rate": 2.1868017672295537e-05,
      "loss": 1.5294,
      "step": 22112
    },
    {
      "epoch": 0.7919136211434813,
      "grad_norm": 1.2164210081100464,
      "learning_rate": 2.1860778895944566e-05,
      "loss": 1.3928,
      "step": 22113
    },
    {
      "epoch": 0.7919494332730496,
      "grad_norm": 1.5118670463562012,
      "learning_rate": 2.1853541170856007e-05,
      "loss": 1.39,
      "step": 22114
    },
    {
      "epoch": 0.7919852454026178,
      "grad_norm": 1.5947706699371338,
      "learning_rate": 2.1846304497127247e-05,
      "loss": 1.4102,
      "step": 22115
    },
    {
      "epoch": 0.7920210575321861,
      "grad_norm": 1.5643930435180664,
      "learning_rate": 2.183906887485565e-05,
      "loss": 1.5654,
      "step": 22116
    },
    {
      "epoch": 0.7920568696617545,
      "grad_norm": 1.5630208253860474,
      "learning_rate": 2.1831834304138587e-05,
      "loss": 1.5264,
      "step": 22117
    },
    {
      "epoch": 0.7920926817913228,
      "grad_norm": 2.1537749767303467,
      "learning_rate": 2.1824600785073335e-05,
      "loss": 1.46,
      "step": 22118
    },
    {
      "epoch": 0.792128493920891,
      "grad_norm": 1.5308321714401245,
      "learning_rate": 2.1817368317757235e-05,
      "loss": 1.3042,
      "step": 22119
    },
    {
      "epoch": 0.7921643060504593,
      "grad_norm": 1.5003904104232788,
      "learning_rate": 2.1810136902287625e-05,
      "loss": 0.9253,
      "step": 22120
    },
    {
      "epoch": 0.7922001181800276,
      "grad_norm": 1.9857511520385742,
      "learning_rate": 2.1802906538761748e-05,
      "loss": 1.3661,
      "step": 22121
    },
    {
      "epoch": 0.7922359303095958,
      "grad_norm": 2.4356906414031982,
      "learning_rate": 2.179567722727689e-05,
      "loss": 1.4112,
      "step": 22122
    },
    {
      "epoch": 0.7922717424391641,
      "grad_norm": 1.3355191946029663,
      "learning_rate": 2.178844896793032e-05,
      "loss": 1.0901,
      "step": 22123
    },
    {
      "epoch": 0.7923075545687325,
      "grad_norm": 1.3789863586425781,
      "learning_rate": 2.1781221760819303e-05,
      "loss": 1.6212,
      "step": 22124
    },
    {
      "epoch": 0.7923433666983007,
      "grad_norm": 1.8660190105438232,
      "learning_rate": 2.1773995606041044e-05,
      "loss": 1.2014,
      "step": 22125
    },
    {
      "epoch": 0.792379178827869,
      "grad_norm": 1.63323175907135,
      "learning_rate": 2.1766770503692748e-05,
      "loss": 1.5686,
      "step": 22126
    },
    {
      "epoch": 0.7924149909574373,
      "grad_norm": 1.8331819772720337,
      "learning_rate": 2.1759546453871647e-05,
      "loss": 1.3657,
      "step": 22127
    },
    {
      "epoch": 0.7924508030870056,
      "grad_norm": 1.5052270889282227,
      "learning_rate": 2.1752323456674962e-05,
      "loss": 1.0748,
      "step": 22128
    },
    {
      "epoch": 0.7924866152165738,
      "grad_norm": 1.5629814863204956,
      "learning_rate": 2.1745101512199806e-05,
      "loss": 1.4885,
      "step": 22129
    },
    {
      "epoch": 0.7925224273461421,
      "grad_norm": 1.492061734199524,
      "learning_rate": 2.173788062054336e-05,
      "loss": 1.3173,
      "step": 22130
    },
    {
      "epoch": 0.7925582394757105,
      "grad_norm": 1.879479169845581,
      "learning_rate": 2.1730660781802804e-05,
      "loss": 1.4771,
      "step": 22131
    },
    {
      "epoch": 0.7925940516052787,
      "grad_norm": 1.4208660125732422,
      "learning_rate": 2.1723441996075223e-05,
      "loss": 1.3431,
      "step": 22132
    },
    {
      "epoch": 0.792629863734847,
      "grad_norm": 1.137665867805481,
      "learning_rate": 2.1716224263457763e-05,
      "loss": 1.3964,
      "step": 22133
    },
    {
      "epoch": 0.7926656758644153,
      "grad_norm": 1.7951220273971558,
      "learning_rate": 2.1709007584047524e-05,
      "loss": 1.682,
      "step": 22134
    },
    {
      "epoch": 0.7927014879939835,
      "grad_norm": 1.7697921991348267,
      "learning_rate": 2.170179195794163e-05,
      "loss": 1.3207,
      "step": 22135
    },
    {
      "epoch": 0.7927373001235518,
      "grad_norm": 1.775586724281311,
      "learning_rate": 2.1694577385237104e-05,
      "loss": 1.291,
      "step": 22136
    },
    {
      "epoch": 0.7927731122531201,
      "grad_norm": 1.5712333917617798,
      "learning_rate": 2.168736386603102e-05,
      "loss": 1.2253,
      "step": 22137
    },
    {
      "epoch": 0.7928089243826885,
      "grad_norm": 1.932370901107788,
      "learning_rate": 2.168015140042048e-05,
      "loss": 1.6061,
      "step": 22138
    },
    {
      "epoch": 0.7928447365122567,
      "grad_norm": 1.4659613370895386,
      "learning_rate": 2.167293998850244e-05,
      "loss": 1.1571,
      "step": 22139
    },
    {
      "epoch": 0.792880548641825,
      "grad_norm": 1.5071487426757812,
      "learning_rate": 2.1665729630373965e-05,
      "loss": 1.5654,
      "step": 22140
    },
    {
      "epoch": 0.7929163607713933,
      "grad_norm": 1.8615208864212036,
      "learning_rate": 2.165852032613205e-05,
      "loss": 1.2381,
      "step": 22141
    },
    {
      "epoch": 0.7929521729009615,
      "grad_norm": 1.6738359928131104,
      "learning_rate": 2.1651312075873718e-05,
      "loss": 1.3059,
      "step": 22142
    },
    {
      "epoch": 0.7929879850305298,
      "grad_norm": 1.4147003889083862,
      "learning_rate": 2.1644104879695892e-05,
      "loss": 1.2178,
      "step": 22143
    },
    {
      "epoch": 0.7930237971600981,
      "grad_norm": 1.8694628477096558,
      "learning_rate": 2.1636898737695567e-05,
      "loss": 1.411,
      "step": 22144
    },
    {
      "epoch": 0.7930596092896665,
      "grad_norm": 1.5314850807189941,
      "learning_rate": 2.1629693649969683e-05,
      "loss": 1.3889,
      "step": 22145
    },
    {
      "epoch": 0.7930954214192347,
      "grad_norm": 1.4080699682235718,
      "learning_rate": 2.1622489616615203e-05,
      "loss": 1.335,
      "step": 22146
    },
    {
      "epoch": 0.793131233548803,
      "grad_norm": 1.9649324417114258,
      "learning_rate": 2.1615286637729037e-05,
      "loss": 1.4941,
      "step": 22147
    },
    {
      "epoch": 0.7931670456783713,
      "grad_norm": 1.2882145643234253,
      "learning_rate": 2.1608084713408018e-05,
      "loss": 1.2824,
      "step": 22148
    },
    {
      "epoch": 0.7932028578079395,
      "grad_norm": 1.4714865684509277,
      "learning_rate": 2.1600883843749165e-05,
      "loss": 1.5583,
      "step": 22149
    },
    {
      "epoch": 0.7932386699375078,
      "grad_norm": 1.5543267726898193,
      "learning_rate": 2.159368402884926e-05,
      "loss": 1.4382,
      "step": 22150
    },
    {
      "epoch": 0.7932744820670761,
      "grad_norm": 1.4382576942443848,
      "learning_rate": 2.1586485268805225e-05,
      "loss": 1.3659,
      "step": 22151
    },
    {
      "epoch": 0.7933102941966444,
      "grad_norm": 1.7490313053131104,
      "learning_rate": 2.1579287563713823e-05,
      "loss": 1.0572,
      "step": 22152
    },
    {
      "epoch": 0.7933461063262127,
      "grad_norm": 1.7107819318771362,
      "learning_rate": 2.1572090913672017e-05,
      "loss": 1.3691,
      "step": 22153
    },
    {
      "epoch": 0.793381918455781,
      "grad_norm": 2.4976677894592285,
      "learning_rate": 2.1564895318776534e-05,
      "loss": 1.4157,
      "step": 22154
    },
    {
      "epoch": 0.7934177305853493,
      "grad_norm": 1.6344050168991089,
      "learning_rate": 2.1557700779124214e-05,
      "loss": 1.4745,
      "step": 22155
    },
    {
      "epoch": 0.7934535427149175,
      "grad_norm": 2.2682435512542725,
      "learning_rate": 2.1550507294811863e-05,
      "loss": 1.4652,
      "step": 22156
    },
    {
      "epoch": 0.7934893548444858,
      "grad_norm": 1.6466939449310303,
      "learning_rate": 2.1543314865936225e-05,
      "loss": 1.5305,
      "step": 22157
    },
    {
      "epoch": 0.7935251669740541,
      "grad_norm": 1.3412882089614868,
      "learning_rate": 2.1536123492594106e-05,
      "loss": 1.2984,
      "step": 22158
    },
    {
      "epoch": 0.7935609791036224,
      "grad_norm": 1.7352933883666992,
      "learning_rate": 2.1528933174882183e-05,
      "loss": 1.6905,
      "step": 22159
    },
    {
      "epoch": 0.7935967912331907,
      "grad_norm": 1.512288212776184,
      "learning_rate": 2.1521743912897296e-05,
      "loss": 1.3645,
      "step": 22160
    },
    {
      "epoch": 0.793632603362759,
      "grad_norm": 1.663425326347351,
      "learning_rate": 2.1514555706736084e-05,
      "loss": 1.2875,
      "step": 22161
    },
    {
      "epoch": 0.7936684154923273,
      "grad_norm": 1.7665363550186157,
      "learning_rate": 2.1507368556495323e-05,
      "loss": 1.4851,
      "step": 22162
    },
    {
      "epoch": 0.7937042276218955,
      "grad_norm": 1.8287420272827148,
      "learning_rate": 2.150018246227161e-05,
      "loss": 1.4071,
      "step": 22163
    },
    {
      "epoch": 0.7937400397514638,
      "grad_norm": 1.557187557220459,
      "learning_rate": 2.1492997424161744e-05,
      "loss": 1.3892,
      "step": 22164
    },
    {
      "epoch": 0.7937758518810321,
      "grad_norm": 1.6745768785476685,
      "learning_rate": 2.1485813442262325e-05,
      "loss": 1.1793,
      "step": 22165
    },
    {
      "epoch": 0.7938116640106004,
      "grad_norm": 2.1988525390625,
      "learning_rate": 2.1478630516669952e-05,
      "loss": 1.5253,
      "step": 22166
    },
    {
      "epoch": 0.7938474761401687,
      "grad_norm": 1.7745083570480347,
      "learning_rate": 2.1471448647481384e-05,
      "loss": 1.0861,
      "step": 22167
    },
    {
      "epoch": 0.793883288269737,
      "grad_norm": 1.9180774688720703,
      "learning_rate": 2.1464267834793152e-05,
      "loss": 1.3288,
      "step": 22168
    },
    {
      "epoch": 0.7939191003993052,
      "grad_norm": 1.5834040641784668,
      "learning_rate": 2.1457088078701916e-05,
      "loss": 1.544,
      "step": 22169
    },
    {
      "epoch": 0.7939549125288735,
      "grad_norm": 1.3772550821304321,
      "learning_rate": 2.144990937930419e-05,
      "loss": 1.4562,
      "step": 22170
    },
    {
      "epoch": 0.7939907246584418,
      "grad_norm": 1.783267855644226,
      "learning_rate": 2.1442731736696666e-05,
      "loss": 1.5635,
      "step": 22171
    },
    {
      "epoch": 0.79402653678801,
      "grad_norm": 1.4774640798568726,
      "learning_rate": 2.143555515097583e-05,
      "loss": 1.3106,
      "step": 22172
    },
    {
      "epoch": 0.7940623489175784,
      "grad_norm": 1.496637225151062,
      "learning_rate": 2.1428379622238283e-05,
      "loss": 1.4957,
      "step": 22173
    },
    {
      "epoch": 0.7940981610471467,
      "grad_norm": 2.0105698108673096,
      "learning_rate": 2.1421205150580514e-05,
      "loss": 1.9181,
      "step": 22174
    },
    {
      "epoch": 0.794133973176715,
      "grad_norm": 1.443631887435913,
      "learning_rate": 2.1414031736099072e-05,
      "loss": 1.3103,
      "step": 22175
    },
    {
      "epoch": 0.7941697853062832,
      "grad_norm": 1.856333613395691,
      "learning_rate": 2.1406859378890486e-05,
      "loss": 1.6582,
      "step": 22176
    },
    {
      "epoch": 0.7942055974358515,
      "grad_norm": 1.6512917280197144,
      "learning_rate": 2.1399688079051205e-05,
      "loss": 1.1163,
      "step": 22177
    },
    {
      "epoch": 0.7942414095654198,
      "grad_norm": 1.48997962474823,
      "learning_rate": 2.1392517836677738e-05,
      "loss": 1.2154,
      "step": 22178
    },
    {
      "epoch": 0.794277221694988,
      "grad_norm": 1.7168711423873901,
      "learning_rate": 2.1385348651866542e-05,
      "loss": 1.384,
      "step": 22179
    },
    {
      "epoch": 0.7943130338245564,
      "grad_norm": 1.7471309900283813,
      "learning_rate": 2.13781805247141e-05,
      "loss": 1.5881,
      "step": 22180
    },
    {
      "epoch": 0.7943488459541247,
      "grad_norm": 1.4693900346755981,
      "learning_rate": 2.137101345531677e-05,
      "loss": 1.5126,
      "step": 22181
    },
    {
      "epoch": 0.794384658083693,
      "grad_norm": 1.2849462032318115,
      "learning_rate": 2.136384744377109e-05,
      "loss": 1.44,
      "step": 22182
    },
    {
      "epoch": 0.7944204702132612,
      "grad_norm": 1.9792394638061523,
      "learning_rate": 2.135668249017341e-05,
      "loss": 1.6862,
      "step": 22183
    },
    {
      "epoch": 0.7944562823428295,
      "grad_norm": 2.0086636543273926,
      "learning_rate": 2.134951859462009e-05,
      "loss": 1.3668,
      "step": 22184
    },
    {
      "epoch": 0.7944920944723978,
      "grad_norm": 1.3767683506011963,
      "learning_rate": 2.1342355757207544e-05,
      "loss": 1.2261,
      "step": 22185
    },
    {
      "epoch": 0.794527906601966,
      "grad_norm": 1.7332990169525146,
      "learning_rate": 2.1335193978032152e-05,
      "loss": 1.233,
      "step": 22186
    },
    {
      "epoch": 0.7945637187315344,
      "grad_norm": 1.5471128225326538,
      "learning_rate": 2.1328033257190272e-05,
      "loss": 1.2935,
      "step": 22187
    },
    {
      "epoch": 0.7945995308611027,
      "grad_norm": 1.7943533658981323,
      "learning_rate": 2.13208735947782e-05,
      "loss": 1.4358,
      "step": 22188
    },
    {
      "epoch": 0.794635342990671,
      "grad_norm": 1.4228322505950928,
      "learning_rate": 2.1313714990892285e-05,
      "loss": 1.4599,
      "step": 22189
    },
    {
      "epoch": 0.7946711551202392,
      "grad_norm": 1.6347566843032837,
      "learning_rate": 2.1306557445628837e-05,
      "loss": 1.4877,
      "step": 22190
    },
    {
      "epoch": 0.7947069672498075,
      "grad_norm": 1.4751968383789062,
      "learning_rate": 2.1299400959084183e-05,
      "loss": 1.5368,
      "step": 22191
    },
    {
      "epoch": 0.7947427793793758,
      "grad_norm": 1.509900689125061,
      "learning_rate": 2.1292245531354538e-05,
      "loss": 1.2897,
      "step": 22192
    },
    {
      "epoch": 0.794778591508944,
      "grad_norm": 1.7221992015838623,
      "learning_rate": 2.128509116253621e-05,
      "loss": 1.5262,
      "step": 22193
    },
    {
      "epoch": 0.7948144036385124,
      "grad_norm": 1.6442131996154785,
      "learning_rate": 2.1277937852725472e-05,
      "loss": 1.1642,
      "step": 22194
    },
    {
      "epoch": 0.7948502157680807,
      "grad_norm": 1.5237714052200317,
      "learning_rate": 2.1270785602018505e-05,
      "loss": 1.2724,
      "step": 22195
    },
    {
      "epoch": 0.794886027897649,
      "grad_norm": 2.1376285552978516,
      "learning_rate": 2.1263634410511567e-05,
      "loss": 1.4343,
      "step": 22196
    },
    {
      "epoch": 0.7949218400272172,
      "grad_norm": 2.232311487197876,
      "learning_rate": 2.125648427830086e-05,
      "loss": 1.6066,
      "step": 22197
    },
    {
      "epoch": 0.7949576521567855,
      "grad_norm": 1.5320496559143066,
      "learning_rate": 2.1249335205482613e-05,
      "loss": 1.4768,
      "step": 22198
    },
    {
      "epoch": 0.7949934642863538,
      "grad_norm": 1.4314378499984741,
      "learning_rate": 2.1242187192152964e-05,
      "loss": 1.3451,
      "step": 22199
    },
    {
      "epoch": 0.795029276415922,
      "grad_norm": 1.5462158918380737,
      "learning_rate": 2.1235040238408087e-05,
      "loss": 1.6463,
      "step": 22200
    },
    {
      "epoch": 0.7950650885454904,
      "grad_norm": 1.7351806163787842,
      "learning_rate": 2.122789434434417e-05,
      "loss": 1.5408,
      "step": 22201
    },
    {
      "epoch": 0.7951009006750587,
      "grad_norm": 1.6399816274642944,
      "learning_rate": 2.1220749510057304e-05,
      "loss": 1.4684,
      "step": 22202
    },
    {
      "epoch": 0.7951367128046269,
      "grad_norm": 2.056469440460205,
      "learning_rate": 2.1213605735643625e-05,
      "loss": 1.8509,
      "step": 22203
    },
    {
      "epoch": 0.7951725249341952,
      "grad_norm": 1.7393684387207031,
      "learning_rate": 2.1206463021199263e-05,
      "loss": 1.3368,
      "step": 22204
    },
    {
      "epoch": 0.7952083370637635,
      "grad_norm": 1.5154002904891968,
      "learning_rate": 2.1199321366820336e-05,
      "loss": 1.4059,
      "step": 22205
    },
    {
      "epoch": 0.7952441491933318,
      "grad_norm": 2.056178569793701,
      "learning_rate": 2.1192180772602867e-05,
      "loss": 1.588,
      "step": 22206
    },
    {
      "epoch": 0.7952799613229,
      "grad_norm": 1.605133295059204,
      "learning_rate": 2.1185041238642934e-05,
      "loss": 1.3137,
      "step": 22207
    },
    {
      "epoch": 0.7953157734524684,
      "grad_norm": 1.9872175455093384,
      "learning_rate": 2.117790276503665e-05,
      "loss": 1.3687,
      "step": 22208
    },
    {
      "epoch": 0.7953515855820367,
      "grad_norm": 1.4879945516586304,
      "learning_rate": 2.1170765351879985e-05,
      "loss": 1.5544,
      "step": 22209
    },
    {
      "epoch": 0.7953873977116049,
      "grad_norm": 1.7017463445663452,
      "learning_rate": 2.116362899926898e-05,
      "loss": 1.3034,
      "step": 22210
    },
    {
      "epoch": 0.7954232098411732,
      "grad_norm": 1.7935137748718262,
      "learning_rate": 2.1156493707299664e-05,
      "loss": 1.1258,
      "step": 22211
    },
    {
      "epoch": 0.7954590219707415,
      "grad_norm": 2.020554780960083,
      "learning_rate": 2.1149359476068043e-05,
      "loss": 1.7252,
      "step": 22212
    },
    {
      "epoch": 0.7954948341003097,
      "grad_norm": 2.046848773956299,
      "learning_rate": 2.1142226305670054e-05,
      "loss": 1.5382,
      "step": 22213
    },
    {
      "epoch": 0.795530646229878,
      "grad_norm": 1.9651658535003662,
      "learning_rate": 2.1135094196201698e-05,
      "loss": 1.5724,
      "step": 22214
    },
    {
      "epoch": 0.7955664583594464,
      "grad_norm": 1.6604115962982178,
      "learning_rate": 2.112796314775892e-05,
      "loss": 1.4821,
      "step": 22215
    },
    {
      "epoch": 0.7956022704890147,
      "grad_norm": 1.4332162141799927,
      "learning_rate": 2.112083316043768e-05,
      "loss": 1.5424,
      "step": 22216
    },
    {
      "epoch": 0.7956380826185829,
      "grad_norm": 1.3632845878601074,
      "learning_rate": 2.1113704234333866e-05,
      "loss": 1.4333,
      "step": 22217
    },
    {
      "epoch": 0.7956738947481512,
      "grad_norm": 1.6020152568817139,
      "learning_rate": 2.1106576369543395e-05,
      "loss": 1.3168,
      "step": 22218
    },
    {
      "epoch": 0.7957097068777195,
      "grad_norm": 1.4405721426010132,
      "learning_rate": 2.109944956616221e-05,
      "loss": 1.3853,
      "step": 22219
    },
    {
      "epoch": 0.7957455190072877,
      "grad_norm": 1.601324439048767,
      "learning_rate": 2.109232382428612e-05,
      "loss": 1.4282,
      "step": 22220
    },
    {
      "epoch": 0.795781331136856,
      "grad_norm": 1.5386035442352295,
      "learning_rate": 2.1085199144011037e-05,
      "loss": 1.5231,
      "step": 22221
    },
    {
      "epoch": 0.7958171432664244,
      "grad_norm": 1.4716813564300537,
      "learning_rate": 2.1078075525432805e-05,
      "loss": 1.3985,
      "step": 22222
    },
    {
      "epoch": 0.7958529553959927,
      "grad_norm": 2.0186898708343506,
      "learning_rate": 2.1070952968647296e-05,
      "loss": 1.3269,
      "step": 22223
    },
    {
      "epoch": 0.7958887675255609,
      "grad_norm": 1.4052791595458984,
      "learning_rate": 2.1063831473750272e-05,
      "loss": 1.1185,
      "step": 22224
    },
    {
      "epoch": 0.7959245796551292,
      "grad_norm": 1.8976056575775146,
      "learning_rate": 2.1056711040837574e-05,
      "loss": 1.6696,
      "step": 22225
    },
    {
      "epoch": 0.7959603917846975,
      "grad_norm": 1.5097455978393555,
      "learning_rate": 2.104959167000503e-05,
      "loss": 1.5236,
      "step": 22226
    },
    {
      "epoch": 0.7959962039142657,
      "grad_norm": 1.326099157333374,
      "learning_rate": 2.104247336134836e-05,
      "loss": 1.2501,
      "step": 22227
    },
    {
      "epoch": 0.796032016043834,
      "grad_norm": 1.6710323095321655,
      "learning_rate": 2.103535611496337e-05,
      "loss": 1.243,
      "step": 22228
    },
    {
      "epoch": 0.7960678281734024,
      "grad_norm": 2.964059829711914,
      "learning_rate": 2.1028239930945794e-05,
      "loss": 1.1575,
      "step": 22229
    },
    {
      "epoch": 0.7961036403029706,
      "grad_norm": 1.9996397495269775,
      "learning_rate": 2.1021124809391423e-05,
      "loss": 1.2896,
      "step": 22230
    },
    {
      "epoch": 0.7961394524325389,
      "grad_norm": 2.0370073318481445,
      "learning_rate": 2.1014010750395907e-05,
      "loss": 1.5759,
      "step": 22231
    },
    {
      "epoch": 0.7961752645621072,
      "grad_norm": 2.4355692863464355,
      "learning_rate": 2.1006897754055e-05,
      "loss": 1.5545,
      "step": 22232
    },
    {
      "epoch": 0.7962110766916755,
      "grad_norm": 1.777365803718567,
      "learning_rate": 2.099978582046438e-05,
      "loss": 1.3352,
      "step": 22233
    },
    {
      "epoch": 0.7962468888212437,
      "grad_norm": 1.5381382703781128,
      "learning_rate": 2.099267494971977e-05,
      "loss": 1.5469,
      "step": 22234
    },
    {
      "epoch": 0.796282700950812,
      "grad_norm": 1.5520857572555542,
      "learning_rate": 2.0985565141916808e-05,
      "loss": 1.0467,
      "step": 22235
    },
    {
      "epoch": 0.7963185130803804,
      "grad_norm": 1.8397648334503174,
      "learning_rate": 2.097845639715109e-05,
      "loss": 1.5229,
      "step": 22236
    },
    {
      "epoch": 0.7963543252099486,
      "grad_norm": 1.6012263298034668,
      "learning_rate": 2.0971348715518368e-05,
      "loss": 1.1753,
      "step": 22237
    },
    {
      "epoch": 0.7963901373395169,
      "grad_norm": 1.5385147333145142,
      "learning_rate": 2.0964242097114184e-05,
      "loss": 1.825,
      "step": 22238
    },
    {
      "epoch": 0.7964259494690852,
      "grad_norm": 1.9621472358703613,
      "learning_rate": 2.0957136542034172e-05,
      "loss": 1.3496,
      "step": 22239
    },
    {
      "epoch": 0.7964617615986535,
      "grad_norm": 1.6801329851150513,
      "learning_rate": 2.0950032050373925e-05,
      "loss": 1.2043,
      "step": 22240
    },
    {
      "epoch": 0.7964975737282217,
      "grad_norm": 2.1597633361816406,
      "learning_rate": 2.0942928622229064e-05,
      "loss": 1.5174,
      "step": 22241
    },
    {
      "epoch": 0.79653338585779,
      "grad_norm": 1.5973130464553833,
      "learning_rate": 2.093582625769509e-05,
      "loss": 1.7511,
      "step": 22242
    },
    {
      "epoch": 0.7965691979873584,
      "grad_norm": 2.0921790599823,
      "learning_rate": 2.0928724956867585e-05,
      "loss": 1.3482,
      "step": 22243
    },
    {
      "epoch": 0.7966050101169266,
      "grad_norm": 2.008749008178711,
      "learning_rate": 2.0921624719842126e-05,
      "loss": 1.5626,
      "step": 22244
    },
    {
      "epoch": 0.7966408222464949,
      "grad_norm": 1.3919188976287842,
      "learning_rate": 2.091452554671417e-05,
      "loss": 1.3621,
      "step": 22245
    },
    {
      "epoch": 0.7966766343760632,
      "grad_norm": 2.2559168338775635,
      "learning_rate": 2.0907427437579287e-05,
      "loss": 1.3592,
      "step": 22246
    },
    {
      "epoch": 0.7967124465056314,
      "grad_norm": 1.4418983459472656,
      "learning_rate": 2.0900330392532895e-05,
      "loss": 1.529,
      "step": 22247
    },
    {
      "epoch": 0.7967482586351997,
      "grad_norm": 1.5940090417861938,
      "learning_rate": 2.089323441167058e-05,
      "loss": 1.6447,
      "step": 22248
    },
    {
      "epoch": 0.796784070764768,
      "grad_norm": 1.591691255569458,
      "learning_rate": 2.088613949508772e-05,
      "loss": 1.244,
      "step": 22249
    },
    {
      "epoch": 0.7968198828943364,
      "grad_norm": 1.6028300523757935,
      "learning_rate": 2.0879045642879814e-05,
      "loss": 1.3213,
      "step": 22250
    },
    {
      "epoch": 0.7968556950239046,
      "grad_norm": 1.5213801860809326,
      "learning_rate": 2.0871952855142286e-05,
      "loss": 1.2998,
      "step": 22251
    },
    {
      "epoch": 0.7968915071534729,
      "grad_norm": 3.0959925651550293,
      "learning_rate": 2.0864861131970594e-05,
      "loss": 1.4941,
      "step": 22252
    },
    {
      "epoch": 0.7969273192830412,
      "grad_norm": 1.3404817581176758,
      "learning_rate": 2.0857770473460115e-05,
      "loss": 1.2366,
      "step": 22253
    },
    {
      "epoch": 0.7969631314126094,
      "grad_norm": 2.0209555625915527,
      "learning_rate": 2.08506808797062e-05,
      "loss": 1.3125,
      "step": 22254
    },
    {
      "epoch": 0.7969989435421777,
      "grad_norm": 1.8329658508300781,
      "learning_rate": 2.084359235080433e-05,
      "loss": 1.4451,
      "step": 22255
    },
    {
      "epoch": 0.797034755671746,
      "grad_norm": 1.6439001560211182,
      "learning_rate": 2.0836504886849785e-05,
      "loss": 1.7872,
      "step": 22256
    },
    {
      "epoch": 0.7970705678013144,
      "grad_norm": 2.4358198642730713,
      "learning_rate": 2.082941848793799e-05,
      "loss": 1.4332,
      "step": 22257
    },
    {
      "epoch": 0.7971063799308826,
      "grad_norm": 1.5745573043823242,
      "learning_rate": 2.0822333154164187e-05,
      "loss": 1.1289,
      "step": 22258
    },
    {
      "epoch": 0.7971421920604509,
      "grad_norm": 1.8613457679748535,
      "learning_rate": 2.0815248885623817e-05,
      "loss": 1.2301,
      "step": 22259
    },
    {
      "epoch": 0.7971780041900192,
      "grad_norm": 1.6447163820266724,
      "learning_rate": 2.0808165682412107e-05,
      "loss": 1.2927,
      "step": 22260
    },
    {
      "epoch": 0.7972138163195874,
      "grad_norm": 2.150489568710327,
      "learning_rate": 2.08010835446244e-05,
      "loss": 1.5737,
      "step": 22261
    },
    {
      "epoch": 0.7972496284491557,
      "grad_norm": 2.0756678581237793,
      "learning_rate": 2.0794002472355933e-05,
      "loss": 1.6523,
      "step": 22262
    },
    {
      "epoch": 0.797285440578724,
      "grad_norm": 1.8237303495407104,
      "learning_rate": 2.0786922465701997e-05,
      "loss": 1.5846,
      "step": 22263
    },
    {
      "epoch": 0.7973212527082923,
      "grad_norm": 1.4604321718215942,
      "learning_rate": 2.0779843524757858e-05,
      "loss": 1.128,
      "step": 22264
    },
    {
      "epoch": 0.7973570648378606,
      "grad_norm": 1.7635831832885742,
      "learning_rate": 2.0772765649618688e-05,
      "loss": 1.3801,
      "step": 22265
    },
    {
      "epoch": 0.7973928769674289,
      "grad_norm": 2.052753448486328,
      "learning_rate": 2.0765688840379816e-05,
      "loss": 1.3685,
      "step": 22266
    },
    {
      "epoch": 0.7974286890969972,
      "grad_norm": 1.3673149347305298,
      "learning_rate": 2.075861309713637e-05,
      "loss": 0.9523,
      "step": 22267
    },
    {
      "epoch": 0.7974645012265654,
      "grad_norm": 2.1413803100585938,
      "learning_rate": 2.0751538419983598e-05,
      "loss": 1.1637,
      "step": 22268
    },
    {
      "epoch": 0.7975003133561337,
      "grad_norm": 1.6249077320098877,
      "learning_rate": 2.0744464809016593e-05,
      "loss": 1.676,
      "step": 22269
    },
    {
      "epoch": 0.797536125485702,
      "grad_norm": 1.6026825904846191,
      "learning_rate": 2.0737392264330635e-05,
      "loss": 1.188,
      "step": 22270
    },
    {
      "epoch": 0.7975719376152703,
      "grad_norm": 1.4662625789642334,
      "learning_rate": 2.073032078602083e-05,
      "loss": 1.4825,
      "step": 22271
    },
    {
      "epoch": 0.7976077497448386,
      "grad_norm": 1.4503960609436035,
      "learning_rate": 2.0723250374182278e-05,
      "loss": 1.6084,
      "step": 22272
    },
    {
      "epoch": 0.7976435618744069,
      "grad_norm": 1.5225666761398315,
      "learning_rate": 2.071618102891013e-05,
      "loss": 1.307,
      "step": 22273
    },
    {
      "epoch": 0.7976793740039752,
      "grad_norm": 1.6680306196212769,
      "learning_rate": 2.070911275029951e-05,
      "loss": 1.533,
      "step": 22274
    },
    {
      "epoch": 0.7977151861335434,
      "grad_norm": 2.219346761703491,
      "learning_rate": 2.0702045538445515e-05,
      "loss": 1.7578,
      "step": 22275
    },
    {
      "epoch": 0.7977509982631117,
      "grad_norm": 1.6483958959579468,
      "learning_rate": 2.069497939344316e-05,
      "loss": 1.3416,
      "step": 22276
    },
    {
      "epoch": 0.79778681039268,
      "grad_norm": 1.3028994798660278,
      "learning_rate": 2.0687914315387613e-05,
      "loss": 1.5492,
      "step": 22277
    },
    {
      "epoch": 0.7978226225222483,
      "grad_norm": 1.6726864576339722,
      "learning_rate": 2.0680850304373843e-05,
      "loss": 1.4142,
      "step": 22278
    },
    {
      "epoch": 0.7978584346518166,
      "grad_norm": 1.879017949104309,
      "learning_rate": 2.0673787360496954e-05,
      "loss": 1.5515,
      "step": 22279
    },
    {
      "epoch": 0.7978942467813849,
      "grad_norm": 1.7818199396133423,
      "learning_rate": 2.06667254838519e-05,
      "loss": 1.4629,
      "step": 22280
    },
    {
      "epoch": 0.7979300589109531,
      "grad_norm": 1.6430115699768066,
      "learning_rate": 2.0659664674533728e-05,
      "loss": 1.6182,
      "step": 22281
    },
    {
      "epoch": 0.7979658710405214,
      "grad_norm": 2.086963176727295,
      "learning_rate": 2.065260493263744e-05,
      "loss": 1.5188,
      "step": 22282
    },
    {
      "epoch": 0.7980016831700897,
      "grad_norm": 1.5534586906433105,
      "learning_rate": 2.0645546258257987e-05,
      "loss": 1.5905,
      "step": 22283
    },
    {
      "epoch": 0.798037495299658,
      "grad_norm": 1.8073506355285645,
      "learning_rate": 2.063848865149035e-05,
      "loss": 1.5429,
      "step": 22284
    },
    {
      "epoch": 0.7980733074292263,
      "grad_norm": 1.6143696308135986,
      "learning_rate": 2.0631432112429473e-05,
      "loss": 1.5414,
      "step": 22285
    },
    {
      "epoch": 0.7981091195587946,
      "grad_norm": 1.9073354005813599,
      "learning_rate": 2.062437664117033e-05,
      "loss": 1.6396,
      "step": 22286
    },
    {
      "epoch": 0.7981449316883629,
      "grad_norm": 1.7292273044586182,
      "learning_rate": 2.061732223780778e-05,
      "loss": 1.3851,
      "step": 22287
    },
    {
      "epoch": 0.7981807438179311,
      "grad_norm": 1.2550760507583618,
      "learning_rate": 2.061026890243677e-05,
      "loss": 1.3337,
      "step": 22288
    },
    {
      "epoch": 0.7982165559474994,
      "grad_norm": 1.5007202625274658,
      "learning_rate": 2.06032166351522e-05,
      "loss": 1.1814,
      "step": 22289
    },
    {
      "epoch": 0.7982523680770677,
      "grad_norm": 1.4860681295394897,
      "learning_rate": 2.059616543604892e-05,
      "loss": 1.4702,
      "step": 22290
    },
    {
      "epoch": 0.7982881802066359,
      "grad_norm": 1.7105454206466675,
      "learning_rate": 2.0589115305221807e-05,
      "loss": 1.556,
      "step": 22291
    },
    {
      "epoch": 0.7983239923362043,
      "grad_norm": 1.458163857460022,
      "learning_rate": 2.058206624276572e-05,
      "loss": 1.5388,
      "step": 22292
    },
    {
      "epoch": 0.7983598044657726,
      "grad_norm": 1.4146854877471924,
      "learning_rate": 2.0575018248775513e-05,
      "loss": 1.6969,
      "step": 22293
    },
    {
      "epoch": 0.7983956165953409,
      "grad_norm": 1.3511170148849487,
      "learning_rate": 2.056797132334596e-05,
      "loss": 1.4063,
      "step": 22294
    },
    {
      "epoch": 0.7984314287249091,
      "grad_norm": 1.6224746704101562,
      "learning_rate": 2.056092546657189e-05,
      "loss": 1.4332,
      "step": 22295
    },
    {
      "epoch": 0.7984672408544774,
      "grad_norm": 1.4703950881958008,
      "learning_rate": 2.0553880678548097e-05,
      "loss": 1.3809,
      "step": 22296
    },
    {
      "epoch": 0.7985030529840457,
      "grad_norm": 1.9702130556106567,
      "learning_rate": 2.0546836959369387e-05,
      "loss": 1.431,
      "step": 22297
    },
    {
      "epoch": 0.7985388651136139,
      "grad_norm": 1.900384783744812,
      "learning_rate": 2.0539794309130478e-05,
      "loss": 1.1501,
      "step": 22298
    },
    {
      "epoch": 0.7985746772431823,
      "grad_norm": 1.9203412532806396,
      "learning_rate": 2.0532752727926142e-05,
      "loss": 1.6926,
      "step": 22299
    },
    {
      "epoch": 0.7986104893727506,
      "grad_norm": 1.5627378225326538,
      "learning_rate": 2.0525712215851132e-05,
      "loss": 1.3586,
      "step": 22300
    },
    {
      "epoch": 0.7986463015023189,
      "grad_norm": 1.3566069602966309,
      "learning_rate": 2.0518672773000124e-05,
      "loss": 1.417,
      "step": 22301
    },
    {
      "epoch": 0.7986821136318871,
      "grad_norm": 2.314469814300537,
      "learning_rate": 2.051163439946786e-05,
      "loss": 1.7983,
      "step": 22302
    },
    {
      "epoch": 0.7987179257614554,
      "grad_norm": 1.9831676483154297,
      "learning_rate": 2.050459709534901e-05,
      "loss": 1.6987,
      "step": 22303
    },
    {
      "epoch": 0.7987537378910237,
      "grad_norm": 2.6953423023223877,
      "learning_rate": 2.0497560860738295e-05,
      "loss": 1.2652,
      "step": 22304
    },
    {
      "epoch": 0.7987895500205919,
      "grad_norm": 1.602420449256897,
      "learning_rate": 2.0490525695730323e-05,
      "loss": 1.4029,
      "step": 22305
    },
    {
      "epoch": 0.7988253621501603,
      "grad_norm": 1.531333565711975,
      "learning_rate": 2.048349160041977e-05,
      "loss": 1.3618,
      "step": 22306
    },
    {
      "epoch": 0.7988611742797286,
      "grad_norm": 1.4855061769485474,
      "learning_rate": 2.0476458574901293e-05,
      "loss": 1.2672,
      "step": 22307
    },
    {
      "epoch": 0.7988969864092968,
      "grad_norm": 2.015868663787842,
      "learning_rate": 2.046942661926946e-05,
      "loss": 1.4248,
      "step": 22308
    },
    {
      "epoch": 0.7989327985388651,
      "grad_norm": 1.3681994676589966,
      "learning_rate": 2.04623957336189e-05,
      "loss": 1.1357,
      "step": 22309
    },
    {
      "epoch": 0.7989686106684334,
      "grad_norm": 2.770206928253174,
      "learning_rate": 2.0455365918044224e-05,
      "loss": 1.257,
      "step": 22310
    },
    {
      "epoch": 0.7990044227980017,
      "grad_norm": 1.8386939764022827,
      "learning_rate": 2.044833717264001e-05,
      "loss": 1.4291,
      "step": 22311
    },
    {
      "epoch": 0.7990402349275699,
      "grad_norm": 1.7308319807052612,
      "learning_rate": 2.044130949750077e-05,
      "loss": 1.3881,
      "step": 22312
    },
    {
      "epoch": 0.7990760470571383,
      "grad_norm": 2.2315123081207275,
      "learning_rate": 2.04342828927211e-05,
      "loss": 1.3908,
      "step": 22313
    },
    {
      "epoch": 0.7991118591867066,
      "grad_norm": 1.8249056339263916,
      "learning_rate": 2.0427257358395546e-05,
      "loss": 1.5637,
      "step": 22314
    },
    {
      "epoch": 0.7991476713162748,
      "grad_norm": 1.5045298337936401,
      "learning_rate": 2.0420232894618573e-05,
      "loss": 1.42,
      "step": 22315
    },
    {
      "epoch": 0.7991834834458431,
      "grad_norm": 1.8735840320587158,
      "learning_rate": 2.041320950148472e-05,
      "loss": 1.581,
      "step": 22316
    },
    {
      "epoch": 0.7992192955754114,
      "grad_norm": 1.6931583881378174,
      "learning_rate": 2.0406187179088477e-05,
      "loss": 1.4165,
      "step": 22317
    },
    {
      "epoch": 0.7992551077049797,
      "grad_norm": 2.089118719100952,
      "learning_rate": 2.0399165927524334e-05,
      "loss": 1.4857,
      "step": 22318
    },
    {
      "epoch": 0.7992909198345479,
      "grad_norm": 1.5633699893951416,
      "learning_rate": 2.0392145746886714e-05,
      "loss": 1.5717,
      "step": 22319
    },
    {
      "epoch": 0.7993267319641163,
      "grad_norm": 1.8018674850463867,
      "learning_rate": 2.038512663727009e-05,
      "loss": 1.4133,
      "step": 22320
    },
    {
      "epoch": 0.7993625440936846,
      "grad_norm": 1.3811540603637695,
      "learning_rate": 2.0378108598768887e-05,
      "loss": 1.2667,
      "step": 22321
    },
    {
      "epoch": 0.7993983562232528,
      "grad_norm": 1.6007475852966309,
      "learning_rate": 2.0371091631477557e-05,
      "loss": 1.2712,
      "step": 22322
    },
    {
      "epoch": 0.7994341683528211,
      "grad_norm": 1.5358997583389282,
      "learning_rate": 2.036407573549044e-05,
      "loss": 1.5853,
      "step": 22323
    },
    {
      "epoch": 0.7994699804823894,
      "grad_norm": 2.150416374206543,
      "learning_rate": 2.035706091090197e-05,
      "loss": 1.3255,
      "step": 22324
    },
    {
      "epoch": 0.7995057926119576,
      "grad_norm": 1.9435093402862549,
      "learning_rate": 2.035004715780654e-05,
      "loss": 1.1444,
      "step": 22325
    },
    {
      "epoch": 0.7995416047415259,
      "grad_norm": 1.7371599674224854,
      "learning_rate": 2.0343034476298452e-05,
      "loss": 1.5803,
      "step": 22326
    },
    {
      "epoch": 0.7995774168710943,
      "grad_norm": 1.5360907316207886,
      "learning_rate": 2.0336022866472092e-05,
      "loss": 1.5967,
      "step": 22327
    },
    {
      "epoch": 0.7996132290006626,
      "grad_norm": 1.229755163192749,
      "learning_rate": 2.0329012328421783e-05,
      "loss": 1.2461,
      "step": 22328
    },
    {
      "epoch": 0.7996490411302308,
      "grad_norm": 1.8806687593460083,
      "learning_rate": 2.0322002862241863e-05,
      "loss": 1.5251,
      "step": 22329
    },
    {
      "epoch": 0.7996848532597991,
      "grad_norm": 1.3057172298431396,
      "learning_rate": 2.0314994468026606e-05,
      "loss": 1.4201,
      "step": 22330
    },
    {
      "epoch": 0.7997206653893674,
      "grad_norm": 1.5392056703567505,
      "learning_rate": 2.03079871458703e-05,
      "loss": 1.6202,
      "step": 22331
    },
    {
      "epoch": 0.7997564775189356,
      "grad_norm": 1.719313621520996,
      "learning_rate": 2.0300980895867263e-05,
      "loss": 1.3479,
      "step": 22332
    },
    {
      "epoch": 0.7997922896485039,
      "grad_norm": 1.619842767715454,
      "learning_rate": 2.029397571811169e-05,
      "loss": 1.5288,
      "step": 22333
    },
    {
      "epoch": 0.7998281017780723,
      "grad_norm": 2.1151888370513916,
      "learning_rate": 2.0286971612697902e-05,
      "loss": 1.6217,
      "step": 22334
    },
    {
      "epoch": 0.7998639139076406,
      "grad_norm": 1.7302162647247314,
      "learning_rate": 2.027996857972002e-05,
      "loss": 1.2988,
      "step": 22335
    },
    {
      "epoch": 0.7998997260372088,
      "grad_norm": 1.6297879219055176,
      "learning_rate": 2.0272966619272392e-05,
      "loss": 1.0922,
      "step": 22336
    },
    {
      "epoch": 0.7999355381667771,
      "grad_norm": 1.6117581129074097,
      "learning_rate": 2.026596573144913e-05,
      "loss": 0.9956,
      "step": 22337
    },
    {
      "epoch": 0.7999713502963454,
      "grad_norm": 1.5663176774978638,
      "learning_rate": 2.025896591634444e-05,
      "loss": 1.3235,
      "step": 22338
    },
    {
      "epoch": 0.8000071624259136,
      "grad_norm": 1.4366532564163208,
      "learning_rate": 2.0251967174052523e-05,
      "loss": 1.0999,
      "step": 22339
    },
    {
      "epoch": 0.8000429745554819,
      "grad_norm": 1.7538988590240479,
      "learning_rate": 2.024496950466753e-05,
      "loss": 1.7191,
      "step": 22340
    },
    {
      "epoch": 0.8000787866850503,
      "grad_norm": 1.5960266590118408,
      "learning_rate": 2.023797290828361e-05,
      "loss": 1.3942,
      "step": 22341
    },
    {
      "epoch": 0.8001145988146185,
      "grad_norm": 1.8154847621917725,
      "learning_rate": 2.0230977384994808e-05,
      "loss": 1.1857,
      "step": 22342
    },
    {
      "epoch": 0.8001504109441868,
      "grad_norm": 1.5289015769958496,
      "learning_rate": 2.022398293489538e-05,
      "loss": 1.6854,
      "step": 22343
    },
    {
      "epoch": 0.8001862230737551,
      "grad_norm": 1.8163173198699951,
      "learning_rate": 2.0216989558079326e-05,
      "loss": 1.5605,
      "step": 22344
    },
    {
      "epoch": 0.8002220352033234,
      "grad_norm": 1.35292387008667,
      "learning_rate": 2.020999725464079e-05,
      "loss": 1.3525,
      "step": 22345
    },
    {
      "epoch": 0.8002578473328916,
      "grad_norm": 1.5341395139694214,
      "learning_rate": 2.0203006024673764e-05,
      "loss": 1.326,
      "step": 22346
    },
    {
      "epoch": 0.8002936594624599,
      "grad_norm": 1.8680927753448486,
      "learning_rate": 2.0196015868272412e-05,
      "loss": 1.5703,
      "step": 22347
    },
    {
      "epoch": 0.8003294715920283,
      "grad_norm": 1.65446937084198,
      "learning_rate": 2.0189026785530705e-05,
      "loss": 1.2258,
      "step": 22348
    },
    {
      "epoch": 0.8003652837215965,
      "grad_norm": 2.1270790100097656,
      "learning_rate": 2.01820387765427e-05,
      "loss": 1.492,
      "step": 22349
    },
    {
      "epoch": 0.8004010958511648,
      "grad_norm": 1.5399971008300781,
      "learning_rate": 2.0175051841402426e-05,
      "loss": 1.5374,
      "step": 22350
    },
    {
      "epoch": 0.8004369079807331,
      "grad_norm": 1.526645302772522,
      "learning_rate": 2.016806598020383e-05,
      "loss": 1.2816,
      "step": 22351
    },
    {
      "epoch": 0.8004727201103014,
      "grad_norm": 1.623569130897522,
      "learning_rate": 2.0161081193040964e-05,
      "loss": 1.244,
      "step": 22352
    },
    {
      "epoch": 0.8005085322398696,
      "grad_norm": 1.273235559463501,
      "learning_rate": 2.0154097480007716e-05,
      "loss": 1.1387,
      "step": 22353
    },
    {
      "epoch": 0.8005443443694379,
      "grad_norm": 1.4888707399368286,
      "learning_rate": 2.0147114841198144e-05,
      "loss": 1.4681,
      "step": 22354
    },
    {
      "epoch": 0.8005801564990063,
      "grad_norm": 2.1212549209594727,
      "learning_rate": 2.014013327670611e-05,
      "loss": 1.2677,
      "step": 22355
    },
    {
      "epoch": 0.8006159686285745,
      "grad_norm": 2.3081042766571045,
      "learning_rate": 2.0133152786625598e-05,
      "loss": 1.5632,
      "step": 22356
    },
    {
      "epoch": 0.8006517807581428,
      "grad_norm": 1.2556418180465698,
      "learning_rate": 2.012617337105044e-05,
      "loss": 1.5076,
      "step": 22357
    },
    {
      "epoch": 0.8006875928877111,
      "grad_norm": 1.5202817916870117,
      "learning_rate": 2.0119195030074645e-05,
      "loss": 1.3706,
      "step": 22358
    },
    {
      "epoch": 0.8007234050172793,
      "grad_norm": 1.4703209400177002,
      "learning_rate": 2.011221776379204e-05,
      "loss": 1.6011,
      "step": 22359
    },
    {
      "epoch": 0.8007592171468476,
      "grad_norm": 2.0287888050079346,
      "learning_rate": 2.0105241572296463e-05,
      "loss": 1.3434,
      "step": 22360
    },
    {
      "epoch": 0.8007950292764159,
      "grad_norm": 1.8298935890197754,
      "learning_rate": 2.0098266455681812e-05,
      "loss": 1.6273,
      "step": 22361
    },
    {
      "epoch": 0.8008308414059843,
      "grad_norm": 1.9720675945281982,
      "learning_rate": 2.009129241404192e-05,
      "loss": 1.537,
      "step": 22362
    },
    {
      "epoch": 0.8008666535355525,
      "grad_norm": 1.583101749420166,
      "learning_rate": 2.0084319447470645e-05,
      "loss": 1.4068,
      "step": 22363
    },
    {
      "epoch": 0.8009024656651208,
      "grad_norm": 2.1829562187194824,
      "learning_rate": 2.007734755606171e-05,
      "loss": 1.4279,
      "step": 22364
    },
    {
      "epoch": 0.8009382777946891,
      "grad_norm": 1.5320464372634888,
      "learning_rate": 2.0070376739909024e-05,
      "loss": 1.4431,
      "step": 22365
    },
    {
      "epoch": 0.8009740899242573,
      "grad_norm": 1.512054443359375,
      "learning_rate": 2.0063406999106293e-05,
      "loss": 1.4185,
      "step": 22366
    },
    {
      "epoch": 0.8010099020538256,
      "grad_norm": 1.5620276927947998,
      "learning_rate": 2.005643833374733e-05,
      "loss": 1.3393,
      "step": 22367
    },
    {
      "epoch": 0.8010457141833939,
      "grad_norm": 1.6267145872116089,
      "learning_rate": 2.0049470743925845e-05,
      "loss": 1.2916,
      "step": 22368
    },
    {
      "epoch": 0.8010815263129623,
      "grad_norm": 1.743434190750122,
      "learning_rate": 2.0042504229735604e-05,
      "loss": 1.4186,
      "step": 22369
    },
    {
      "epoch": 0.8011173384425305,
      "grad_norm": 1.7056375741958618,
      "learning_rate": 2.0035538791270358e-05,
      "loss": 1.3937,
      "step": 22370
    },
    {
      "epoch": 0.8011531505720988,
      "grad_norm": 1.7276453971862793,
      "learning_rate": 2.002857442862377e-05,
      "loss": 1.4292,
      "step": 22371
    },
    {
      "epoch": 0.8011889627016671,
      "grad_norm": 1.376869797706604,
      "learning_rate": 2.002161114188955e-05,
      "loss": 1.5486,
      "step": 22372
    },
    {
      "epoch": 0.8012247748312353,
      "grad_norm": 1.5992333889007568,
      "learning_rate": 2.0014648931161386e-05,
      "loss": 1.3535,
      "step": 22373
    },
    {
      "epoch": 0.8012605869608036,
      "grad_norm": 1.8465687036514282,
      "learning_rate": 2.000768779653298e-05,
      "loss": 1.1694,
      "step": 22374
    },
    {
      "epoch": 0.8012963990903719,
      "grad_norm": 1.4649271965026855,
      "learning_rate": 2.000072773809789e-05,
      "loss": 1.6326,
      "step": 22375
    },
    {
      "epoch": 0.8013322112199402,
      "grad_norm": 1.352648377418518,
      "learning_rate": 1.9993768755949882e-05,
      "loss": 1.5192,
      "step": 22376
    },
    {
      "epoch": 0.8013680233495085,
      "grad_norm": 1.7649281024932861,
      "learning_rate": 1.99868108501825e-05,
      "loss": 1.1998,
      "step": 22377
    },
    {
      "epoch": 0.8014038354790768,
      "grad_norm": 1.4050350189208984,
      "learning_rate": 1.9979854020889356e-05,
      "loss": 1.4443,
      "step": 22378
    },
    {
      "epoch": 0.8014396476086451,
      "grad_norm": 1.8340182304382324,
      "learning_rate": 1.9972898268164052e-05,
      "loss": 1.5995,
      "step": 22379
    },
    {
      "epoch": 0.8014754597382133,
      "grad_norm": 1.7190074920654297,
      "learning_rate": 1.9965943592100166e-05,
      "loss": 1.402,
      "step": 22380
    },
    {
      "epoch": 0.8015112718677816,
      "grad_norm": 2.5135326385498047,
      "learning_rate": 1.995898999279131e-05,
      "loss": 1.2719,
      "step": 22381
    },
    {
      "epoch": 0.8015470839973499,
      "grad_norm": 2.2622857093811035,
      "learning_rate": 1.9952037470330964e-05,
      "loss": 1.7018,
      "step": 22382
    },
    {
      "epoch": 0.8015828961269181,
      "grad_norm": 1.4016941785812378,
      "learning_rate": 1.994508602481271e-05,
      "loss": 1.4652,
      "step": 22383
    },
    {
      "epoch": 0.8016187082564865,
      "grad_norm": 1.9714092016220093,
      "learning_rate": 1.993813565633005e-05,
      "loss": 1.7556,
      "step": 22384
    },
    {
      "epoch": 0.8016545203860548,
      "grad_norm": 1.8598322868347168,
      "learning_rate": 1.993118636497654e-05,
      "loss": 1.633,
      "step": 22385
    },
    {
      "epoch": 0.801690332515623,
      "grad_norm": 1.6280977725982666,
      "learning_rate": 1.99242381508456e-05,
      "loss": 1.2321,
      "step": 22386
    },
    {
      "epoch": 0.8017261446451913,
      "grad_norm": 1.4744327068328857,
      "learning_rate": 1.9917291014030747e-05,
      "loss": 1.1498,
      "step": 22387
    },
    {
      "epoch": 0.8017619567747596,
      "grad_norm": 1.4115815162658691,
      "learning_rate": 1.991034495462547e-05,
      "loss": 1.2791,
      "step": 22388
    },
    {
      "epoch": 0.8017977689043279,
      "grad_norm": 1.562197208404541,
      "learning_rate": 1.990339997272317e-05,
      "loss": 1.2146,
      "step": 22389
    },
    {
      "epoch": 0.8018335810338961,
      "grad_norm": 2.2312514781951904,
      "learning_rate": 1.9896456068417302e-05,
      "loss": 1.4709,
      "step": 22390
    },
    {
      "epoch": 0.8018693931634645,
      "grad_norm": 1.673010230064392,
      "learning_rate": 1.9889513241801295e-05,
      "loss": 1.3413,
      "step": 22391
    },
    {
      "epoch": 0.8019052052930328,
      "grad_norm": 1.657200813293457,
      "learning_rate": 1.988257149296857e-05,
      "loss": 1.7043,
      "step": 22392
    },
    {
      "epoch": 0.801941017422601,
      "grad_norm": 1.4257102012634277,
      "learning_rate": 1.987563082201249e-05,
      "loss": 1.4941,
      "step": 22393
    },
    {
      "epoch": 0.8019768295521693,
      "grad_norm": 1.373645544052124,
      "learning_rate": 1.9868691229026437e-05,
      "loss": 1.5311,
      "step": 22394
    },
    {
      "epoch": 0.8020126416817376,
      "grad_norm": 1.3798229694366455,
      "learning_rate": 1.986175271410381e-05,
      "loss": 1.1315,
      "step": 22395
    },
    {
      "epoch": 0.8020484538113059,
      "grad_norm": 1.657395839691162,
      "learning_rate": 1.9854815277337902e-05,
      "loss": 1.2784,
      "step": 22396
    },
    {
      "epoch": 0.8020842659408741,
      "grad_norm": 2.0209593772888184,
      "learning_rate": 1.9847878918822073e-05,
      "loss": 1.2659,
      "step": 22397
    },
    {
      "epoch": 0.8021200780704425,
      "grad_norm": 1.9298338890075684,
      "learning_rate": 1.9840943638649635e-05,
      "loss": 1.2645,
      "step": 22398
    },
    {
      "epoch": 0.8021558902000108,
      "grad_norm": 1.6281800270080566,
      "learning_rate": 1.9834009436913948e-05,
      "loss": 1.4317,
      "step": 22399
    },
    {
      "epoch": 0.802191702329579,
      "grad_norm": 1.7945187091827393,
      "learning_rate": 1.9827076313708216e-05,
      "loss": 1.671,
      "step": 22400
    },
    {
      "epoch": 0.8022275144591473,
      "grad_norm": 1.9173634052276611,
      "learning_rate": 1.9820144269125763e-05,
      "loss": 1.6113,
      "step": 22401
    },
    {
      "epoch": 0.8022633265887156,
      "grad_norm": 1.6076704263687134,
      "learning_rate": 1.981321330325987e-05,
      "loss": 1.5395,
      "step": 22402
    },
    {
      "epoch": 0.8022991387182838,
      "grad_norm": 1.4223419427871704,
      "learning_rate": 1.980628341620373e-05,
      "loss": 1.3787,
      "step": 22403
    },
    {
      "epoch": 0.8023349508478521,
      "grad_norm": 1.6627603769302368,
      "learning_rate": 1.9799354608050614e-05,
      "loss": 1.4402,
      "step": 22404
    },
    {
      "epoch": 0.8023707629774205,
      "grad_norm": 1.7014139890670776,
      "learning_rate": 1.979242687889372e-05,
      "loss": 1.4458,
      "step": 22405
    },
    {
      "epoch": 0.8024065751069888,
      "grad_norm": 1.6976075172424316,
      "learning_rate": 1.9785500228826292e-05,
      "loss": 1.6349,
      "step": 22406
    },
    {
      "epoch": 0.802442387236557,
      "grad_norm": 1.68596351146698,
      "learning_rate": 1.977857465794146e-05,
      "loss": 1.7244,
      "step": 22407
    },
    {
      "epoch": 0.8024781993661253,
      "grad_norm": 2.4121077060699463,
      "learning_rate": 1.977165016633242e-05,
      "loss": 1.5179,
      "step": 22408
    },
    {
      "epoch": 0.8025140114956936,
      "grad_norm": 2.1352639198303223,
      "learning_rate": 1.9764726754092354e-05,
      "loss": 1.5501,
      "step": 22409
    },
    {
      "epoch": 0.8025498236252618,
      "grad_norm": 1.6876717805862427,
      "learning_rate": 1.975780442131442e-05,
      "loss": 1.5704,
      "step": 22410
    },
    {
      "epoch": 0.8025856357548301,
      "grad_norm": 1.326467752456665,
      "learning_rate": 1.9750883168091684e-05,
      "loss": 1.539,
      "step": 22411
    },
    {
      "epoch": 0.8026214478843985,
      "grad_norm": 1.339929223060608,
      "learning_rate": 1.9743962994517316e-05,
      "loss": 1.4927,
      "step": 22412
    },
    {
      "epoch": 0.8026572600139668,
      "grad_norm": 1.7197191715240479,
      "learning_rate": 1.9737043900684416e-05,
      "loss": 1.3768,
      "step": 22413
    },
    {
      "epoch": 0.802693072143535,
      "grad_norm": 1.7365750074386597,
      "learning_rate": 1.9730125886686033e-05,
      "loss": 1.2812,
      "step": 22414
    },
    {
      "epoch": 0.8027288842731033,
      "grad_norm": 1.6885199546813965,
      "learning_rate": 1.972320895261528e-05,
      "loss": 1.4585,
      "step": 22415
    },
    {
      "epoch": 0.8027646964026716,
      "grad_norm": 1.6332989931106567,
      "learning_rate": 1.9716293098565186e-05,
      "loss": 1.4209,
      "step": 22416
    },
    {
      "epoch": 0.8028005085322398,
      "grad_norm": 1.5932855606079102,
      "learning_rate": 1.9709378324628848e-05,
      "loss": 1.5682,
      "step": 22417
    },
    {
      "epoch": 0.8028363206618081,
      "grad_norm": 1.284571886062622,
      "learning_rate": 1.970246463089922e-05,
      "loss": 1.3149,
      "step": 22418
    },
    {
      "epoch": 0.8028721327913765,
      "grad_norm": 1.5631963014602661,
      "learning_rate": 1.9695552017469364e-05,
      "loss": 1.5476,
      "step": 22419
    },
    {
      "epoch": 0.8029079449209447,
      "grad_norm": 1.6943448781967163,
      "learning_rate": 1.9688640484432287e-05,
      "loss": 1.3432,
      "step": 22420
    },
    {
      "epoch": 0.802943757050513,
      "grad_norm": 1.4390662908554077,
      "learning_rate": 1.968173003188094e-05,
      "loss": 1.2311,
      "step": 22421
    },
    {
      "epoch": 0.8029795691800813,
      "grad_norm": 1.7541425228118896,
      "learning_rate": 1.96748206599083e-05,
      "loss": 1.4092,
      "step": 22422
    },
    {
      "epoch": 0.8030153813096496,
      "grad_norm": 1.399704098701477,
      "learning_rate": 1.9667912368607344e-05,
      "loss": 1.3498,
      "step": 22423
    },
    {
      "epoch": 0.8030511934392178,
      "grad_norm": 2.5454680919647217,
      "learning_rate": 1.9661005158071033e-05,
      "loss": 1.2805,
      "step": 22424
    },
    {
      "epoch": 0.8030870055687861,
      "grad_norm": 2.0977659225463867,
      "learning_rate": 1.965409902839225e-05,
      "loss": 1.7922,
      "step": 22425
    },
    {
      "epoch": 0.8031228176983545,
      "grad_norm": 1.8387500047683716,
      "learning_rate": 1.9647193979663915e-05,
      "loss": 1.3484,
      "step": 22426
    },
    {
      "epoch": 0.8031586298279227,
      "grad_norm": 1.7446051836013794,
      "learning_rate": 1.9640290011978935e-05,
      "loss": 1.3899,
      "step": 22427
    },
    {
      "epoch": 0.803194441957491,
      "grad_norm": 2.0197689533233643,
      "learning_rate": 1.9633387125430226e-05,
      "loss": 1.6515,
      "step": 22428
    },
    {
      "epoch": 0.8032302540870593,
      "grad_norm": 1.3845274448394775,
      "learning_rate": 1.9626485320110632e-05,
      "loss": 1.3537,
      "step": 22429
    },
    {
      "epoch": 0.8032660662166276,
      "grad_norm": 1.8746153116226196,
      "learning_rate": 1.961958459611295e-05,
      "loss": 1.1188,
      "step": 22430
    },
    {
      "epoch": 0.8033018783461958,
      "grad_norm": 1.311545491218567,
      "learning_rate": 1.9612684953530124e-05,
      "loss": 1.3475,
      "step": 22431
    },
    {
      "epoch": 0.8033376904757641,
      "grad_norm": 2.6298389434814453,
      "learning_rate": 1.9605786392454904e-05,
      "loss": 1.6516,
      "step": 22432
    },
    {
      "epoch": 0.8033735026053325,
      "grad_norm": 1.625884771347046,
      "learning_rate": 1.9598888912980117e-05,
      "loss": 1.1758,
      "step": 22433
    },
    {
      "epoch": 0.8034093147349007,
      "grad_norm": 2.10892391204834,
      "learning_rate": 1.9591992515198588e-05,
      "loss": 1.4329,
      "step": 22434
    },
    {
      "epoch": 0.803445126864469,
      "grad_norm": 1.5895978212356567,
      "learning_rate": 1.9585097199203096e-05,
      "loss": 1.466,
      "step": 22435
    },
    {
      "epoch": 0.8034809389940373,
      "grad_norm": 1.5277785062789917,
      "learning_rate": 1.957820296508637e-05,
      "loss": 1.7112,
      "step": 22436
    },
    {
      "epoch": 0.8035167511236055,
      "grad_norm": 2.111467123031616,
      "learning_rate": 1.9571309812941184e-05,
      "loss": 1.1178,
      "step": 22437
    },
    {
      "epoch": 0.8035525632531738,
      "grad_norm": 1.8915823698043823,
      "learning_rate": 1.9564417742860307e-05,
      "loss": 1.5494,
      "step": 22438
    },
    {
      "epoch": 0.8035883753827421,
      "grad_norm": 1.9859658479690552,
      "learning_rate": 1.9557526754936405e-05,
      "loss": 1.6199,
      "step": 22439
    },
    {
      "epoch": 0.8036241875123105,
      "grad_norm": 1.3071140050888062,
      "learning_rate": 1.955063684926225e-05,
      "loss": 1.3199,
      "step": 22440
    },
    {
      "epoch": 0.8036599996418787,
      "grad_norm": 1.539048194885254,
      "learning_rate": 1.9543748025930452e-05,
      "loss": 1.3173,
      "step": 22441
    },
    {
      "epoch": 0.803695811771447,
      "grad_norm": 1.7450897693634033,
      "learning_rate": 1.9536860285033797e-05,
      "loss": 1.5985,
      "step": 22442
    },
    {
      "epoch": 0.8037316239010153,
      "grad_norm": 1.7976315021514893,
      "learning_rate": 1.9529973626664865e-05,
      "loss": 1.4624,
      "step": 22443
    },
    {
      "epoch": 0.8037674360305835,
      "grad_norm": 2.065857172012329,
      "learning_rate": 1.952308805091636e-05,
      "loss": 1.5356,
      "step": 22444
    },
    {
      "epoch": 0.8038032481601518,
      "grad_norm": 2.313326597213745,
      "learning_rate": 1.9516203557880852e-05,
      "loss": 1.5283,
      "step": 22445
    },
    {
      "epoch": 0.8038390602897201,
      "grad_norm": 2.275071859359741,
      "learning_rate": 1.9509320147651068e-05,
      "loss": 1.4142,
      "step": 22446
    },
    {
      "epoch": 0.8038748724192885,
      "grad_norm": 2.0012121200561523,
      "learning_rate": 1.950243782031954e-05,
      "loss": 1.5083,
      "step": 22447
    },
    {
      "epoch": 0.8039106845488567,
      "grad_norm": 1.4849156141281128,
      "learning_rate": 1.9495556575978836e-05,
      "loss": 1.3491,
      "step": 22448
    },
    {
      "epoch": 0.803946496678425,
      "grad_norm": 1.4966169595718384,
      "learning_rate": 1.948867641472163e-05,
      "loss": 1.4678,
      "step": 22449
    },
    {
      "epoch": 0.8039823088079933,
      "grad_norm": 1.564099669456482,
      "learning_rate": 1.9481797336640396e-05,
      "loss": 1.1639,
      "step": 22450
    },
    {
      "epoch": 0.8040181209375615,
      "grad_norm": 1.6799238920211792,
      "learning_rate": 1.9474919341827746e-05,
      "loss": 1.2108,
      "step": 22451
    },
    {
      "epoch": 0.8040539330671298,
      "grad_norm": 1.8090828657150269,
      "learning_rate": 1.946804243037613e-05,
      "loss": 1.3893,
      "step": 22452
    },
    {
      "epoch": 0.8040897451966981,
      "grad_norm": 1.8733340501785278,
      "learning_rate": 1.9461166602378176e-05,
      "loss": 1.6914,
      "step": 22453
    },
    {
      "epoch": 0.8041255573262664,
      "grad_norm": 1.3128604888916016,
      "learning_rate": 1.9454291857926323e-05,
      "loss": 1.5375,
      "step": 22454
    },
    {
      "epoch": 0.8041613694558347,
      "grad_norm": 1.3559684753417969,
      "learning_rate": 1.94474181971131e-05,
      "loss": 1.5298,
      "step": 22455
    },
    {
      "epoch": 0.804197181585403,
      "grad_norm": 1.790544033050537,
      "learning_rate": 1.9440545620030924e-05,
      "loss": 1.469,
      "step": 22456
    },
    {
      "epoch": 0.8042329937149713,
      "grad_norm": 1.5944311618804932,
      "learning_rate": 1.9433674126772306e-05,
      "loss": 1.4043,
      "step": 22457
    },
    {
      "epoch": 0.8042688058445395,
      "grad_norm": 1.7077056169509888,
      "learning_rate": 1.9426803717429696e-05,
      "loss": 1.4089,
      "step": 22458
    },
    {
      "epoch": 0.8043046179741078,
      "grad_norm": 1.622973918914795,
      "learning_rate": 1.941993439209546e-05,
      "loss": 1.3107,
      "step": 22459
    },
    {
      "epoch": 0.8043404301036761,
      "grad_norm": 1.6063894033432007,
      "learning_rate": 1.9413066150862113e-05,
      "loss": 1.2024,
      "step": 22460
    },
    {
      "epoch": 0.8043762422332444,
      "grad_norm": 1.5905756950378418,
      "learning_rate": 1.9406198993822e-05,
      "loss": 1.5908,
      "step": 22461
    },
    {
      "epoch": 0.8044120543628127,
      "grad_norm": 1.9238442182540894,
      "learning_rate": 1.9399332921067537e-05,
      "loss": 2.0081,
      "step": 22462
    },
    {
      "epoch": 0.804447866492381,
      "grad_norm": 1.5469660758972168,
      "learning_rate": 1.939246793269103e-05,
      "loss": 1.6734,
      "step": 22463
    },
    {
      "epoch": 0.8044836786219492,
      "grad_norm": 2.7361080646514893,
      "learning_rate": 1.938560402878494e-05,
      "loss": 1.7897,
      "step": 22464
    },
    {
      "epoch": 0.8045194907515175,
      "grad_norm": 1.9339004755020142,
      "learning_rate": 1.9378741209441565e-05,
      "loss": 1.3695,
      "step": 22465
    },
    {
      "epoch": 0.8045553028810858,
      "grad_norm": 1.4602885246276855,
      "learning_rate": 1.9371879474753208e-05,
      "loss": 1.5427,
      "step": 22466
    },
    {
      "epoch": 0.8045911150106541,
      "grad_norm": 1.5360411405563354,
      "learning_rate": 1.9365018824812208e-05,
      "loss": 1.4808,
      "step": 22467
    },
    {
      "epoch": 0.8046269271402224,
      "grad_norm": 1.5212199687957764,
      "learning_rate": 1.9358159259710874e-05,
      "loss": 1.1769,
      "step": 22468
    },
    {
      "epoch": 0.8046627392697907,
      "grad_norm": 2.0291521549224854,
      "learning_rate": 1.9351300779541503e-05,
      "loss": 1.5694,
      "step": 22469
    },
    {
      "epoch": 0.804698551399359,
      "grad_norm": 1.705421805381775,
      "learning_rate": 1.9344443384396337e-05,
      "loss": 1.6615,
      "step": 22470
    },
    {
      "epoch": 0.8047343635289272,
      "grad_norm": 2.076650381088257,
      "learning_rate": 1.9337587074367637e-05,
      "loss": 1.5295,
      "step": 22471
    },
    {
      "epoch": 0.8047701756584955,
      "grad_norm": 1.3150924444198608,
      "learning_rate": 1.9330731849547655e-05,
      "loss": 1.019,
      "step": 22472
    },
    {
      "epoch": 0.8048059877880638,
      "grad_norm": 1.6172802448272705,
      "learning_rate": 1.9323877710028658e-05,
      "loss": 1.4884,
      "step": 22473
    },
    {
      "epoch": 0.804841799917632,
      "grad_norm": 1.3734067678451538,
      "learning_rate": 1.9317024655902782e-05,
      "loss": 1.3739,
      "step": 22474
    },
    {
      "epoch": 0.8048776120472004,
      "grad_norm": 1.7881556749343872,
      "learning_rate": 1.9310172687262273e-05,
      "loss": 1.3257,
      "step": 22475
    },
    {
      "epoch": 0.8049134241767687,
      "grad_norm": 1.5242328643798828,
      "learning_rate": 1.9303321804199338e-05,
      "loss": 1.2506,
      "step": 22476
    },
    {
      "epoch": 0.804949236306337,
      "grad_norm": 1.535390019416809,
      "learning_rate": 1.9296472006806087e-05,
      "loss": 0.9729,
      "step": 22477
    },
    {
      "epoch": 0.8049850484359052,
      "grad_norm": 1.8442243337631226,
      "learning_rate": 1.9289623295174697e-05,
      "loss": 1.6024,
      "step": 22478
    },
    {
      "epoch": 0.8050208605654735,
      "grad_norm": 1.8470580577850342,
      "learning_rate": 1.9282775669397324e-05,
      "loss": 1.6681,
      "step": 22479
    },
    {
      "epoch": 0.8050566726950418,
      "grad_norm": 1.2782084941864014,
      "learning_rate": 1.9275929129566116e-05,
      "loss": 1.4994,
      "step": 22480
    },
    {
      "epoch": 0.80509248482461,
      "grad_norm": 1.826900839805603,
      "learning_rate": 1.9269083675773126e-05,
      "loss": 1.5691,
      "step": 22481
    },
    {
      "epoch": 0.8051282969541784,
      "grad_norm": 1.380555510520935,
      "learning_rate": 1.9262239308110474e-05,
      "loss": 1.4017,
      "step": 22482
    },
    {
      "epoch": 0.8051641090837467,
      "grad_norm": 1.2143405675888062,
      "learning_rate": 1.925539602667028e-05,
      "loss": 1.4549,
      "step": 22483
    },
    {
      "epoch": 0.805199921213315,
      "grad_norm": 1.643473505973816,
      "learning_rate": 1.924855383154456e-05,
      "loss": 1.7101,
      "step": 22484
    },
    {
      "epoch": 0.8052357333428832,
      "grad_norm": 1.6362136602401733,
      "learning_rate": 1.924171272282538e-05,
      "loss": 1.5971,
      "step": 22485
    },
    {
      "epoch": 0.8052715454724515,
      "grad_norm": 1.5110559463500977,
      "learning_rate": 1.9234872700604777e-05,
      "loss": 1.3248,
      "step": 22486
    },
    {
      "epoch": 0.8053073576020198,
      "grad_norm": 1.5972952842712402,
      "learning_rate": 1.9228033764974818e-05,
      "loss": 1.3044,
      "step": 22487
    },
    {
      "epoch": 0.805343169731588,
      "grad_norm": 1.8326233625411987,
      "learning_rate": 1.9221195916027445e-05,
      "loss": 1.611,
      "step": 22488
    },
    {
      "epoch": 0.8053789818611564,
      "grad_norm": 2.9135138988494873,
      "learning_rate": 1.921435915385469e-05,
      "loss": 1.9834,
      "step": 22489
    },
    {
      "epoch": 0.8054147939907247,
      "grad_norm": 1.453930139541626,
      "learning_rate": 1.9207523478548518e-05,
      "loss": 1.3931,
      "step": 22490
    },
    {
      "epoch": 0.805450606120293,
      "grad_norm": 1.5433127880096436,
      "learning_rate": 1.9200688890200936e-05,
      "loss": 1.0553,
      "step": 22491
    },
    {
      "epoch": 0.8054864182498612,
      "grad_norm": 1.3644789457321167,
      "learning_rate": 1.9193855388903824e-05,
      "loss": 1.187,
      "step": 22492
    },
    {
      "epoch": 0.8055222303794295,
      "grad_norm": 1.4411420822143555,
      "learning_rate": 1.918702297474917e-05,
      "loss": 1.4759,
      "step": 22493
    },
    {
      "epoch": 0.8055580425089978,
      "grad_norm": 1.6154589653015137,
      "learning_rate": 1.9180191647828906e-05,
      "loss": 1.3965,
      "step": 22494
    },
    {
      "epoch": 0.805593854638566,
      "grad_norm": 1.5406982898712158,
      "learning_rate": 1.917336140823488e-05,
      "loss": 1.3631,
      "step": 22495
    },
    {
      "epoch": 0.8056296667681344,
      "grad_norm": 1.679787278175354,
      "learning_rate": 1.916653225605901e-05,
      "loss": 1.577,
      "step": 22496
    },
    {
      "epoch": 0.8056654788977027,
      "grad_norm": 1.5739904642105103,
      "learning_rate": 1.915970419139319e-05,
      "loss": 1.641,
      "step": 22497
    },
    {
      "epoch": 0.805701291027271,
      "grad_norm": 1.5268478393554688,
      "learning_rate": 1.91528772143293e-05,
      "loss": 1.0907,
      "step": 22498
    },
    {
      "epoch": 0.8057371031568392,
      "grad_norm": 1.8088730573654175,
      "learning_rate": 1.9146051324959134e-05,
      "loss": 1.4507,
      "step": 22499
    },
    {
      "epoch": 0.8057729152864075,
      "grad_norm": 1.8089536428451538,
      "learning_rate": 1.9139226523374566e-05,
      "loss": 1.4117,
      "step": 22500
    },
    {
      "epoch": 0.8058087274159758,
      "grad_norm": 1.642994999885559,
      "learning_rate": 1.9132402809667416e-05,
      "loss": 1.4066,
      "step": 22501
    },
    {
      "epoch": 0.805844539545544,
      "grad_norm": 2.0810434818267822,
      "learning_rate": 1.9125580183929448e-05,
      "loss": 1.5808,
      "step": 22502
    },
    {
      "epoch": 0.8058803516751124,
      "grad_norm": 1.3621398210525513,
      "learning_rate": 1.9118758646252477e-05,
      "loss": 1.1892,
      "step": 22503
    },
    {
      "epoch": 0.8059161638046807,
      "grad_norm": 1.652403473854065,
      "learning_rate": 1.9111938196728284e-05,
      "loss": 1.3299,
      "step": 22504
    },
    {
      "epoch": 0.8059519759342489,
      "grad_norm": 1.8718891143798828,
      "learning_rate": 1.9105118835448644e-05,
      "loss": 1.6525,
      "step": 22505
    },
    {
      "epoch": 0.8059877880638172,
      "grad_norm": 1.4959759712219238,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 1.2341,
      "step": 22506
    },
    {
      "epoch": 0.8060236001933855,
      "grad_norm": 1.535392165184021,
      "learning_rate": 1.9091483377989895e-05,
      "loss": 1.4926,
      "step": 22507
    },
    {
      "epoch": 0.8060594123229537,
      "grad_norm": 2.2390332221984863,
      "learning_rate": 1.9084667281994273e-05,
      "loss": 1.5246,
      "step": 22508
    },
    {
      "epoch": 0.806095224452522,
      "grad_norm": 1.3575999736785889,
      "learning_rate": 1.9077852274610055e-05,
      "loss": 1.6471,
      "step": 22509
    },
    {
      "epoch": 0.8061310365820904,
      "grad_norm": 2.580101251602173,
      "learning_rate": 1.9071038355928948e-05,
      "loss": 1.5999,
      "step": 22510
    },
    {
      "epoch": 0.8061668487116587,
      "grad_norm": 1.4733290672302246,
      "learning_rate": 1.9064225526042644e-05,
      "loss": 1.2515,
      "step": 22511
    },
    {
      "epoch": 0.8062026608412269,
      "grad_norm": 1.3597815036773682,
      "learning_rate": 1.90574137850428e-05,
      "loss": 1.4526,
      "step": 22512
    },
    {
      "epoch": 0.8062384729707952,
      "grad_norm": 1.9659523963928223,
      "learning_rate": 1.9050603133021017e-05,
      "loss": 1.6421,
      "step": 22513
    },
    {
      "epoch": 0.8062742851003635,
      "grad_norm": 1.9482402801513672,
      "learning_rate": 1.904379357006896e-05,
      "loss": 1.3654,
      "step": 22514
    },
    {
      "epoch": 0.8063100972299317,
      "grad_norm": 1.8046506643295288,
      "learning_rate": 1.9036985096278227e-05,
      "loss": 1.6794,
      "step": 22515
    },
    {
      "epoch": 0.8063459093595,
      "grad_norm": 2.277416944503784,
      "learning_rate": 1.903017771174046e-05,
      "loss": 1.6218,
      "step": 22516
    },
    {
      "epoch": 0.8063817214890684,
      "grad_norm": 1.3956186771392822,
      "learning_rate": 1.9023371416547177e-05,
      "loss": 1.5363,
      "step": 22517
    },
    {
      "epoch": 0.8064175336186367,
      "grad_norm": 1.5622445344924927,
      "learning_rate": 1.9016566210789977e-05,
      "loss": 1.4348,
      "step": 22518
    },
    {
      "epoch": 0.8064533457482049,
      "grad_norm": 1.2763302326202393,
      "learning_rate": 1.9009762094560446e-05,
      "loss": 1.5905,
      "step": 22519
    },
    {
      "epoch": 0.8064891578777732,
      "grad_norm": 1.6055020093917847,
      "learning_rate": 1.9002959067950066e-05,
      "loss": 1.5085,
      "step": 22520
    },
    {
      "epoch": 0.8065249700073415,
      "grad_norm": 1.4015109539031982,
      "learning_rate": 1.8996157131050395e-05,
      "loss": 1.4095,
      "step": 22521
    },
    {
      "epoch": 0.8065607821369097,
      "grad_norm": 2.2692770957946777,
      "learning_rate": 1.8989356283952943e-05,
      "loss": 1.556,
      "step": 22522
    },
    {
      "epoch": 0.806596594266478,
      "grad_norm": 1.5125993490219116,
      "learning_rate": 1.898255652674924e-05,
      "loss": 1.4581,
      "step": 22523
    },
    {
      "epoch": 0.8066324063960464,
      "grad_norm": 1.379358172416687,
      "learning_rate": 1.8975757859530696e-05,
      "loss": 1.5579,
      "step": 22524
    },
    {
      "epoch": 0.8066682185256147,
      "grad_norm": 1.793256163597107,
      "learning_rate": 1.8968960282388826e-05,
      "loss": 1.6786,
      "step": 22525
    },
    {
      "epoch": 0.8067040306551829,
      "grad_norm": 2.028078317642212,
      "learning_rate": 1.896216379541509e-05,
      "loss": 1.3897,
      "step": 22526
    },
    {
      "epoch": 0.8067398427847512,
      "grad_norm": 1.6242742538452148,
      "learning_rate": 1.895536839870089e-05,
      "loss": 1.2961,
      "step": 22527
    },
    {
      "epoch": 0.8067756549143195,
      "grad_norm": 1.5608689785003662,
      "learning_rate": 1.894857409233769e-05,
      "loss": 1.0402,
      "step": 22528
    },
    {
      "epoch": 0.8068114670438877,
      "grad_norm": 1.6409008502960205,
      "learning_rate": 1.8941780876416826e-05,
      "loss": 1.4753,
      "step": 22529
    },
    {
      "epoch": 0.806847279173456,
      "grad_norm": 1.9912757873535156,
      "learning_rate": 1.893498875102979e-05,
      "loss": 1.7401,
      "step": 22530
    },
    {
      "epoch": 0.8068830913030244,
      "grad_norm": 1.8302267789840698,
      "learning_rate": 1.8928197716267894e-05,
      "loss": 1.5594,
      "step": 22531
    },
    {
      "epoch": 0.8069189034325926,
      "grad_norm": 1.399596929550171,
      "learning_rate": 1.892140777222252e-05,
      "loss": 1.4571,
      "step": 22532
    },
    {
      "epoch": 0.8069547155621609,
      "grad_norm": 1.794324278831482,
      "learning_rate": 1.8914618918985028e-05,
      "loss": 1.2283,
      "step": 22533
    },
    {
      "epoch": 0.8069905276917292,
      "grad_norm": 1.7145918607711792,
      "learning_rate": 1.890783115664676e-05,
      "loss": 1.2539,
      "step": 22534
    },
    {
      "epoch": 0.8070263398212975,
      "grad_norm": 1.4848459959030151,
      "learning_rate": 1.8901044485299034e-05,
      "loss": 1.41,
      "step": 22535
    },
    {
      "epoch": 0.8070621519508657,
      "grad_norm": 1.7278633117675781,
      "learning_rate": 1.889425890503308e-05,
      "loss": 1.5744,
      "step": 22536
    },
    {
      "epoch": 0.807097964080434,
      "grad_norm": 1.4958399534225464,
      "learning_rate": 1.888747441594031e-05,
      "loss": 1.3393,
      "step": 22537
    },
    {
      "epoch": 0.8071337762100024,
      "grad_norm": 1.7114999294281006,
      "learning_rate": 1.888069101811193e-05,
      "loss": 1.1309,
      "step": 22538
    },
    {
      "epoch": 0.8071695883395706,
      "grad_norm": 1.3507667779922485,
      "learning_rate": 1.887390871163922e-05,
      "loss": 1.3367,
      "step": 22539
    },
    {
      "epoch": 0.8072054004691389,
      "grad_norm": 2.3738009929656982,
      "learning_rate": 1.886712749661339e-05,
      "loss": 1.5551,
      "step": 22540
    },
    {
      "epoch": 0.8072412125987072,
      "grad_norm": 1.762641191482544,
      "learning_rate": 1.8860347373125753e-05,
      "loss": 1.4421,
      "step": 22541
    },
    {
      "epoch": 0.8072770247282754,
      "grad_norm": 1.4975340366363525,
      "learning_rate": 1.8853568341267448e-05,
      "loss": 1.3572,
      "step": 22542
    },
    {
      "epoch": 0.8073128368578437,
      "grad_norm": 1.7372536659240723,
      "learning_rate": 1.884679040112971e-05,
      "loss": 1.1709,
      "step": 22543
    },
    {
      "epoch": 0.807348648987412,
      "grad_norm": 2.144090175628662,
      "learning_rate": 1.884001355280376e-05,
      "loss": 1.6421,
      "step": 22544
    },
    {
      "epoch": 0.8073844611169804,
      "grad_norm": 1.7356644868850708,
      "learning_rate": 1.8833237796380708e-05,
      "loss": 1.3723,
      "step": 22545
    },
    {
      "epoch": 0.8074202732465486,
      "grad_norm": 1.4547863006591797,
      "learning_rate": 1.8826463131951767e-05,
      "loss": 1.3705,
      "step": 22546
    },
    {
      "epoch": 0.8074560853761169,
      "grad_norm": 1.4055238962173462,
      "learning_rate": 1.8819689559608012e-05,
      "loss": 1.2637,
      "step": 22547
    },
    {
      "epoch": 0.8074918975056852,
      "grad_norm": 1.7052263021469116,
      "learning_rate": 1.8812917079440673e-05,
      "loss": 1.4931,
      "step": 22548
    },
    {
      "epoch": 0.8075277096352534,
      "grad_norm": 1.6998118162155151,
      "learning_rate": 1.8806145691540777e-05,
      "loss": 1.7741,
      "step": 22549
    },
    {
      "epoch": 0.8075635217648217,
      "grad_norm": 1.4957939386367798,
      "learning_rate": 1.8799375395999487e-05,
      "loss": 1.4767,
      "step": 22550
    },
    {
      "epoch": 0.80759933389439,
      "grad_norm": 1.5346788167953491,
      "learning_rate": 1.8792606192907813e-05,
      "loss": 1.3791,
      "step": 22551
    },
    {
      "epoch": 0.8076351460239584,
      "grad_norm": 1.6711095571517944,
      "learning_rate": 1.878583808235692e-05,
      "loss": 1.3828,
      "step": 22552
    },
    {
      "epoch": 0.8076709581535266,
      "grad_norm": 2.229304313659668,
      "learning_rate": 1.8779071064437813e-05,
      "loss": 1.8454,
      "step": 22553
    },
    {
      "epoch": 0.8077067702830949,
      "grad_norm": 1.6836011409759521,
      "learning_rate": 1.877230513924152e-05,
      "loss": 1.14,
      "step": 22554
    },
    {
      "epoch": 0.8077425824126632,
      "grad_norm": 1.786728858947754,
      "learning_rate": 1.8765540306859076e-05,
      "loss": 1.4788,
      "step": 22555
    },
    {
      "epoch": 0.8077783945422314,
      "grad_norm": 3.058734178543091,
      "learning_rate": 1.8758776567381508e-05,
      "loss": 1.3963,
      "step": 22556
    },
    {
      "epoch": 0.8078142066717997,
      "grad_norm": 1.7298702001571655,
      "learning_rate": 1.8752013920899836e-05,
      "loss": 1.3446,
      "step": 22557
    },
    {
      "epoch": 0.807850018801368,
      "grad_norm": 1.8953697681427002,
      "learning_rate": 1.874525236750495e-05,
      "loss": 1.6026,
      "step": 22558
    },
    {
      "epoch": 0.8078858309309364,
      "grad_norm": 1.685193657875061,
      "learning_rate": 1.8738491907287946e-05,
      "loss": 1.5745,
      "step": 22559
    },
    {
      "epoch": 0.8079216430605046,
      "grad_norm": 1.7449640035629272,
      "learning_rate": 1.8731732540339684e-05,
      "loss": 1.4257,
      "step": 22560
    },
    {
      "epoch": 0.8079574551900729,
      "grad_norm": 1.563020944595337,
      "learning_rate": 1.872497426675116e-05,
      "loss": 1.2951,
      "step": 22561
    },
    {
      "epoch": 0.8079932673196412,
      "grad_norm": 1.595478892326355,
      "learning_rate": 1.8718217086613242e-05,
      "loss": 1.1794,
      "step": 22562
    },
    {
      "epoch": 0.8080290794492094,
      "grad_norm": 1.6257182359695435,
      "learning_rate": 1.871146100001687e-05,
      "loss": 1.3228,
      "step": 22563
    },
    {
      "epoch": 0.8080648915787777,
      "grad_norm": 1.8950823545455933,
      "learning_rate": 1.8704706007052963e-05,
      "loss": 1.4481,
      "step": 22564
    },
    {
      "epoch": 0.808100703708346,
      "grad_norm": 1.5394660234451294,
      "learning_rate": 1.8697952107812344e-05,
      "loss": 1.249,
      "step": 22565
    },
    {
      "epoch": 0.8081365158379143,
      "grad_norm": 1.2871990203857422,
      "learning_rate": 1.86911993023859e-05,
      "loss": 1.3532,
      "step": 22566
    },
    {
      "epoch": 0.8081723279674826,
      "grad_norm": 1.3102869987487793,
      "learning_rate": 1.8684447590864494e-05,
      "loss": 1.4537,
      "step": 22567
    },
    {
      "epoch": 0.8082081400970509,
      "grad_norm": 1.5285694599151611,
      "learning_rate": 1.867769697333899e-05,
      "loss": 1.2677,
      "step": 22568
    },
    {
      "epoch": 0.8082439522266192,
      "grad_norm": 1.455868124961853,
      "learning_rate": 1.8670947449900113e-05,
      "loss": 1.2461,
      "step": 22569
    },
    {
      "epoch": 0.8082797643561874,
      "grad_norm": 1.5081433057785034,
      "learning_rate": 1.8664199020638785e-05,
      "loss": 1.6674,
      "step": 22570
    },
    {
      "epoch": 0.8083155764857557,
      "grad_norm": 1.4699350595474243,
      "learning_rate": 1.8657451685645756e-05,
      "loss": 1.5963,
      "step": 22571
    },
    {
      "epoch": 0.808351388615324,
      "grad_norm": 1.518787145614624,
      "learning_rate": 1.8650705445011752e-05,
      "loss": 1.1539,
      "step": 22572
    },
    {
      "epoch": 0.8083872007448923,
      "grad_norm": 1.761563777923584,
      "learning_rate": 1.8643960298827566e-05,
      "loss": 1.2219,
      "step": 22573
    },
    {
      "epoch": 0.8084230128744606,
      "grad_norm": 1.9616132974624634,
      "learning_rate": 1.8637216247183966e-05,
      "loss": 1.5145,
      "step": 22574
    },
    {
      "epoch": 0.8084588250040289,
      "grad_norm": 1.8252731561660767,
      "learning_rate": 1.8630473290171692e-05,
      "loss": 1.6176,
      "step": 22575
    },
    {
      "epoch": 0.8084946371335971,
      "grad_norm": 1.8796595335006714,
      "learning_rate": 1.8623731427881418e-05,
      "loss": 1.5545,
      "step": 22576
    },
    {
      "epoch": 0.8085304492631654,
      "grad_norm": 1.679461121559143,
      "learning_rate": 1.8616990660403865e-05,
      "loss": 1.4918,
      "step": 22577
    },
    {
      "epoch": 0.8085662613927337,
      "grad_norm": 1.4722779989242554,
      "learning_rate": 1.8610250987829725e-05,
      "loss": 1.3942,
      "step": 22578
    },
    {
      "epoch": 0.808602073522302,
      "grad_norm": 1.8476933240890503,
      "learning_rate": 1.86035124102497e-05,
      "loss": 1.3052,
      "step": 22579
    },
    {
      "epoch": 0.8086378856518703,
      "grad_norm": 1.712782621383667,
      "learning_rate": 1.85967749277544e-05,
      "loss": 1.0278,
      "step": 22580
    },
    {
      "epoch": 0.8086736977814386,
      "grad_norm": 1.7284092903137207,
      "learning_rate": 1.8590038540434485e-05,
      "loss": 1.5429,
      "step": 22581
    },
    {
      "epoch": 0.8087095099110069,
      "grad_norm": 1.3968522548675537,
      "learning_rate": 1.8583303248380625e-05,
      "loss": 1.2947,
      "step": 22582
    },
    {
      "epoch": 0.8087453220405751,
      "grad_norm": 1.7425910234451294,
      "learning_rate": 1.8576569051683368e-05,
      "loss": 1.552,
      "step": 22583
    },
    {
      "epoch": 0.8087811341701434,
      "grad_norm": 1.8411518335342407,
      "learning_rate": 1.8569835950433344e-05,
      "loss": 1.7172,
      "step": 22584
    },
    {
      "epoch": 0.8088169462997117,
      "grad_norm": 1.879120945930481,
      "learning_rate": 1.856310394472114e-05,
      "loss": 1.374,
      "step": 22585
    },
    {
      "epoch": 0.80885275842928,
      "grad_norm": 1.4328593015670776,
      "learning_rate": 1.8556373034637353e-05,
      "loss": 1.5658,
      "step": 22586
    },
    {
      "epoch": 0.8088885705588483,
      "grad_norm": 1.5288828611373901,
      "learning_rate": 1.8549643220272494e-05,
      "loss": 1.4516,
      "step": 22587
    },
    {
      "epoch": 0.8089243826884166,
      "grad_norm": 2.343038558959961,
      "learning_rate": 1.8542914501717113e-05,
      "loss": 1.4335,
      "step": 22588
    },
    {
      "epoch": 0.8089601948179849,
      "grad_norm": 1.6488909721374512,
      "learning_rate": 1.853618687906177e-05,
      "loss": 0.9991,
      "step": 22589
    },
    {
      "epoch": 0.8089960069475531,
      "grad_norm": 1.3302148580551147,
      "learning_rate": 1.852946035239693e-05,
      "loss": 1.339,
      "step": 22590
    },
    {
      "epoch": 0.8090318190771214,
      "grad_norm": 1.6579506397247314,
      "learning_rate": 1.8522734921813113e-05,
      "loss": 1.4514,
      "step": 22591
    },
    {
      "epoch": 0.8090676312066897,
      "grad_norm": 1.6817675828933716,
      "learning_rate": 1.85160105874008e-05,
      "loss": 1.1681,
      "step": 22592
    },
    {
      "epoch": 0.8091034433362579,
      "grad_norm": 1.8572685718536377,
      "learning_rate": 1.8509287349250482e-05,
      "loss": 1.456,
      "step": 22593
    },
    {
      "epoch": 0.8091392554658263,
      "grad_norm": 1.4762178659439087,
      "learning_rate": 1.850256520745256e-05,
      "loss": 1.4556,
      "step": 22594
    },
    {
      "epoch": 0.8091750675953946,
      "grad_norm": 1.8500964641571045,
      "learning_rate": 1.84958441620975e-05,
      "loss": 1.5579,
      "step": 22595
    },
    {
      "epoch": 0.8092108797249629,
      "grad_norm": 1.5180333852767944,
      "learning_rate": 1.8489124213275745e-05,
      "loss": 1.34,
      "step": 22596
    },
    {
      "epoch": 0.8092466918545311,
      "grad_norm": 2.089555501937866,
      "learning_rate": 1.8482405361077658e-05,
      "loss": 1.6457,
      "step": 22597
    },
    {
      "epoch": 0.8092825039840994,
      "grad_norm": 1.4597141742706299,
      "learning_rate": 1.847568760559366e-05,
      "loss": 1.3754,
      "step": 22598
    },
    {
      "epoch": 0.8093183161136677,
      "grad_norm": 1.6133396625518799,
      "learning_rate": 1.8468970946914134e-05,
      "loss": 1.1957,
      "step": 22599
    },
    {
      "epoch": 0.8093541282432359,
      "grad_norm": 1.5091086626052856,
      "learning_rate": 1.8462255385129447e-05,
      "loss": 1.3334,
      "step": 22600
    },
    {
      "epoch": 0.8093899403728043,
      "grad_norm": 1.8339647054672241,
      "learning_rate": 1.8455540920329916e-05,
      "loss": 1.7524,
      "step": 22601
    },
    {
      "epoch": 0.8094257525023726,
      "grad_norm": 2.3774209022521973,
      "learning_rate": 1.8448827552605907e-05,
      "loss": 1.4291,
      "step": 22602
    },
    {
      "epoch": 0.8094615646319409,
      "grad_norm": 1.3259594440460205,
      "learning_rate": 1.8442115282047723e-05,
      "loss": 1.5422,
      "step": 22603
    },
    {
      "epoch": 0.8094973767615091,
      "grad_norm": 1.5684672594070435,
      "learning_rate": 1.8435404108745702e-05,
      "loss": 1.4201,
      "step": 22604
    },
    {
      "epoch": 0.8095331888910774,
      "grad_norm": 1.6213066577911377,
      "learning_rate": 1.8428694032790074e-05,
      "loss": 1.1581,
      "step": 22605
    },
    {
      "epoch": 0.8095690010206457,
      "grad_norm": 1.6540571451187134,
      "learning_rate": 1.8421985054271163e-05,
      "loss": 1.319,
      "step": 22606
    },
    {
      "epoch": 0.8096048131502139,
      "grad_norm": 2.1115972995758057,
      "learning_rate": 1.8415277173279234e-05,
      "loss": 1.3976,
      "step": 22607
    },
    {
      "epoch": 0.8096406252797823,
      "grad_norm": 1.4200903177261353,
      "learning_rate": 1.840857038990449e-05,
      "loss": 1.2876,
      "step": 22608
    },
    {
      "epoch": 0.8096764374093506,
      "grad_norm": 2.16260027885437,
      "learning_rate": 1.840186470423718e-05,
      "loss": 1.3341,
      "step": 22609
    },
    {
      "epoch": 0.8097122495389188,
      "grad_norm": 1.4035898447036743,
      "learning_rate": 1.8395160116367528e-05,
      "loss": 1.276,
      "step": 22610
    },
    {
      "epoch": 0.8097480616684871,
      "grad_norm": 2.398355484008789,
      "learning_rate": 1.8388456626385765e-05,
      "loss": 1.5374,
      "step": 22611
    },
    {
      "epoch": 0.8097838737980554,
      "grad_norm": 1.7936995029449463,
      "learning_rate": 1.838175423438202e-05,
      "loss": 1.4219,
      "step": 22612
    },
    {
      "epoch": 0.8098196859276237,
      "grad_norm": 1.4915958642959595,
      "learning_rate": 1.837505294044649e-05,
      "loss": 1.4166,
      "step": 22613
    },
    {
      "epoch": 0.8098554980571919,
      "grad_norm": 1.7961300611495972,
      "learning_rate": 1.836835274466936e-05,
      "loss": 1.4323,
      "step": 22614
    },
    {
      "epoch": 0.8098913101867603,
      "grad_norm": 2.53918719291687,
      "learning_rate": 1.836165364714072e-05,
      "loss": 1.3317,
      "step": 22615
    },
    {
      "epoch": 0.8099271223163286,
      "grad_norm": 1.416369080543518,
      "learning_rate": 1.835495564795072e-05,
      "loss": 1.3952,
      "step": 22616
    },
    {
      "epoch": 0.8099629344458968,
      "grad_norm": 2.098452091217041,
      "learning_rate": 1.8348258747189484e-05,
      "loss": 1.4207,
      "step": 22617
    },
    {
      "epoch": 0.8099987465754651,
      "grad_norm": 1.722528100013733,
      "learning_rate": 1.8341562944947134e-05,
      "loss": 1.4759,
      "step": 22618
    },
    {
      "epoch": 0.8100345587050334,
      "grad_norm": 1.612025499343872,
      "learning_rate": 1.8334868241313685e-05,
      "loss": 1.5993,
      "step": 22619
    },
    {
      "epoch": 0.8100703708346016,
      "grad_norm": 1.5949660539627075,
      "learning_rate": 1.832817463637925e-05,
      "loss": 1.4045,
      "step": 22620
    },
    {
      "epoch": 0.8101061829641699,
      "grad_norm": 1.33736252784729,
      "learning_rate": 1.832148213023387e-05,
      "loss": 1.2006,
      "step": 22621
    },
    {
      "epoch": 0.8101419950937383,
      "grad_norm": 2.1266708374023438,
      "learning_rate": 1.8314790722967624e-05,
      "loss": 1.522,
      "step": 22622
    },
    {
      "epoch": 0.8101778072233066,
      "grad_norm": 1.595590353012085,
      "learning_rate": 1.8308100414670504e-05,
      "loss": 1.4679,
      "step": 22623
    },
    {
      "epoch": 0.8102136193528748,
      "grad_norm": 1.3091840744018555,
      "learning_rate": 1.830141120543246e-05,
      "loss": 1.2643,
      "step": 22624
    },
    {
      "epoch": 0.8102494314824431,
      "grad_norm": 1.8592644929885864,
      "learning_rate": 1.829472309534359e-05,
      "loss": 1.4972,
      "step": 22625
    },
    {
      "epoch": 0.8102852436120114,
      "grad_norm": 1.5312573909759521,
      "learning_rate": 1.828803608449382e-05,
      "loss": 1.4416,
      "step": 22626
    },
    {
      "epoch": 0.8103210557415796,
      "grad_norm": 1.3308011293411255,
      "learning_rate": 1.828135017297311e-05,
      "loss": 1.3562,
      "step": 22627
    },
    {
      "epoch": 0.8103568678711479,
      "grad_norm": 1.6629422903060913,
      "learning_rate": 1.8274665360871425e-05,
      "loss": 1.2463,
      "step": 22628
    },
    {
      "epoch": 0.8103926800007163,
      "grad_norm": 2.118683099746704,
      "learning_rate": 1.8267981648278733e-05,
      "loss": 1.5011,
      "step": 22629
    },
    {
      "epoch": 0.8104284921302846,
      "grad_norm": 1.60123872756958,
      "learning_rate": 1.8261299035284883e-05,
      "loss": 1.4359,
      "step": 22630
    },
    {
      "epoch": 0.8104643042598528,
      "grad_norm": 1.698397159576416,
      "learning_rate": 1.825461752197983e-05,
      "loss": 1.4867,
      "step": 22631
    },
    {
      "epoch": 0.8105001163894211,
      "grad_norm": 1.6148022413253784,
      "learning_rate": 1.8247937108453482e-05,
      "loss": 1.4307,
      "step": 22632
    },
    {
      "epoch": 0.8105359285189894,
      "grad_norm": 1.858364224433899,
      "learning_rate": 1.8241257794795653e-05,
      "loss": 1.3802,
      "step": 22633
    },
    {
      "epoch": 0.8105717406485576,
      "grad_norm": 1.6542712450027466,
      "learning_rate": 1.8234579581096266e-05,
      "loss": 1.419,
      "step": 22634
    },
    {
      "epoch": 0.8106075527781259,
      "grad_norm": 1.4909992218017578,
      "learning_rate": 1.82279024674451e-05,
      "loss": 1.3149,
      "step": 22635
    },
    {
      "epoch": 0.8106433649076943,
      "grad_norm": 1.499837040901184,
      "learning_rate": 1.8221226453932074e-05,
      "loss": 1.3224,
      "step": 22636
    },
    {
      "epoch": 0.8106791770372626,
      "grad_norm": 2.104693651199341,
      "learning_rate": 1.821455154064693e-05,
      "loss": 1.3378,
      "step": 22637
    },
    {
      "epoch": 0.8107149891668308,
      "grad_norm": 2.1410248279571533,
      "learning_rate": 1.8207877727679523e-05,
      "loss": 1.29,
      "step": 22638
    },
    {
      "epoch": 0.8107508012963991,
      "grad_norm": 1.8281619548797607,
      "learning_rate": 1.820120501511957e-05,
      "loss": 1.3479,
      "step": 22639
    },
    {
      "epoch": 0.8107866134259674,
      "grad_norm": 1.728197693824768,
      "learning_rate": 1.8194533403056935e-05,
      "loss": 1.5324,
      "step": 22640
    },
    {
      "epoch": 0.8108224255555356,
      "grad_norm": 1.576294183731079,
      "learning_rate": 1.8187862891581343e-05,
      "loss": 1.4459,
      "step": 22641
    },
    {
      "epoch": 0.8108582376851039,
      "grad_norm": 1.467061996459961,
      "learning_rate": 1.8181193480782466e-05,
      "loss": 1.1734,
      "step": 22642
    },
    {
      "epoch": 0.8108940498146723,
      "grad_norm": 1.5236502885818481,
      "learning_rate": 1.8174525170750145e-05,
      "loss": 1.0675,
      "step": 22643
    },
    {
      "epoch": 0.8109298619442405,
      "grad_norm": 1.7928649187088013,
      "learning_rate": 1.816785796157402e-05,
      "loss": 1.6023,
      "step": 22644
    },
    {
      "epoch": 0.8109656740738088,
      "grad_norm": 2.2303380966186523,
      "learning_rate": 1.8161191853343827e-05,
      "loss": 1.3182,
      "step": 22645
    },
    {
      "epoch": 0.8110014862033771,
      "grad_norm": 1.2965222597122192,
      "learning_rate": 1.815452684614919e-05,
      "loss": 1.5056,
      "step": 22646
    },
    {
      "epoch": 0.8110372983329454,
      "grad_norm": 2.0349364280700684,
      "learning_rate": 1.8147862940079875e-05,
      "loss": 1.7971,
      "step": 22647
    },
    {
      "epoch": 0.8110731104625136,
      "grad_norm": 1.655045509338379,
      "learning_rate": 1.8141200135225444e-05,
      "loss": 1.6047,
      "step": 22648
    },
    {
      "epoch": 0.8111089225920819,
      "grad_norm": 1.3515609502792358,
      "learning_rate": 1.8134538431675608e-05,
      "loss": 1.519,
      "step": 22649
    },
    {
      "epoch": 0.8111447347216503,
      "grad_norm": 1.5632681846618652,
      "learning_rate": 1.8127877829519935e-05,
      "loss": 1.5632,
      "step": 22650
    },
    {
      "epoch": 0.8111805468512185,
      "grad_norm": 1.6890665292739868,
      "learning_rate": 1.8121218328848054e-05,
      "loss": 1.2321,
      "step": 22651
    },
    {
      "epoch": 0.8112163589807868,
      "grad_norm": 1.3986566066741943,
      "learning_rate": 1.8114559929749586e-05,
      "loss": 1.0644,
      "step": 22652
    },
    {
      "epoch": 0.8112521711103551,
      "grad_norm": 2.0749661922454834,
      "learning_rate": 1.8107902632314044e-05,
      "loss": 1.2105,
      "step": 22653
    },
    {
      "epoch": 0.8112879832399233,
      "grad_norm": 1.7615954875946045,
      "learning_rate": 1.8101246436631093e-05,
      "loss": 1.3472,
      "step": 22654
    },
    {
      "epoch": 0.8113237953694916,
      "grad_norm": 1.7018330097198486,
      "learning_rate": 1.8094591342790202e-05,
      "loss": 1.5256,
      "step": 22655
    },
    {
      "epoch": 0.8113596074990599,
      "grad_norm": 1.8836838006973267,
      "learning_rate": 1.8087937350880957e-05,
      "loss": 1.4031,
      "step": 22656
    },
    {
      "epoch": 0.8113954196286283,
      "grad_norm": 1.560875415802002,
      "learning_rate": 1.8081284460992808e-05,
      "loss": 1.2982,
      "step": 22657
    },
    {
      "epoch": 0.8114312317581965,
      "grad_norm": 1.307560920715332,
      "learning_rate": 1.8074632673215365e-05,
      "loss": 1.4884,
      "step": 22658
    },
    {
      "epoch": 0.8114670438877648,
      "grad_norm": 2.2579801082611084,
      "learning_rate": 1.806798198763805e-05,
      "loss": 1.6971,
      "step": 22659
    },
    {
      "epoch": 0.8115028560173331,
      "grad_norm": 1.286411166191101,
      "learning_rate": 1.806133240435034e-05,
      "loss": 1.2505,
      "step": 22660
    },
    {
      "epoch": 0.8115386681469013,
      "grad_norm": 1.7834134101867676,
      "learning_rate": 1.8054683923441694e-05,
      "loss": 1.4802,
      "step": 22661
    },
    {
      "epoch": 0.8115744802764696,
      "grad_norm": 1.6677281856536865,
      "learning_rate": 1.804803654500159e-05,
      "loss": 1.3463,
      "step": 22662
    },
    {
      "epoch": 0.8116102924060379,
      "grad_norm": 1.974789023399353,
      "learning_rate": 1.8041390269119463e-05,
      "loss": 1.7445,
      "step": 22663
    },
    {
      "epoch": 0.8116461045356063,
      "grad_norm": 2.2561769485473633,
      "learning_rate": 1.8034745095884687e-05,
      "loss": 1.4882,
      "step": 22664
    },
    {
      "epoch": 0.8116819166651745,
      "grad_norm": 1.5701444149017334,
      "learning_rate": 1.802810102538668e-05,
      "loss": 1.6699,
      "step": 22665
    },
    {
      "epoch": 0.8117177287947428,
      "grad_norm": 2.0393314361572266,
      "learning_rate": 1.8021458057714845e-05,
      "loss": 1.4406,
      "step": 22666
    },
    {
      "epoch": 0.8117535409243111,
      "grad_norm": 2.785156488418579,
      "learning_rate": 1.8014816192958574e-05,
      "loss": 1.2964,
      "step": 22667
    },
    {
      "epoch": 0.8117893530538793,
      "grad_norm": 1.4221245050430298,
      "learning_rate": 1.8008175431207173e-05,
      "loss": 1.4925,
      "step": 22668
    },
    {
      "epoch": 0.8118251651834476,
      "grad_norm": 1.7360721826553345,
      "learning_rate": 1.8001535772550006e-05,
      "loss": 1.3764,
      "step": 22669
    },
    {
      "epoch": 0.8118609773130159,
      "grad_norm": 1.4904028177261353,
      "learning_rate": 1.7994897217076423e-05,
      "loss": 1.1439,
      "step": 22670
    },
    {
      "epoch": 0.8118967894425843,
      "grad_norm": 1.3772350549697876,
      "learning_rate": 1.7988259764875705e-05,
      "loss": 1.4313,
      "step": 22671
    },
    {
      "epoch": 0.8119326015721525,
      "grad_norm": 3.0207619667053223,
      "learning_rate": 1.7981623416037163e-05,
      "loss": 1.1208,
      "step": 22672
    },
    {
      "epoch": 0.8119684137017208,
      "grad_norm": 1.3616470098495483,
      "learning_rate": 1.7974988170650075e-05,
      "loss": 1.4309,
      "step": 22673
    },
    {
      "epoch": 0.8120042258312891,
      "grad_norm": 1.674111247062683,
      "learning_rate": 1.7968354028803748e-05,
      "loss": 1.4841,
      "step": 22674
    },
    {
      "epoch": 0.8120400379608573,
      "grad_norm": 1.7297577857971191,
      "learning_rate": 1.7961720990587382e-05,
      "loss": 1.0489,
      "step": 22675
    },
    {
      "epoch": 0.8120758500904256,
      "grad_norm": 1.3222150802612305,
      "learning_rate": 1.7955089056090234e-05,
      "loss": 1.5225,
      "step": 22676
    },
    {
      "epoch": 0.8121116622199939,
      "grad_norm": 1.6286765336990356,
      "learning_rate": 1.7948458225401553e-05,
      "loss": 1.7537,
      "step": 22677
    },
    {
      "epoch": 0.8121474743495622,
      "grad_norm": 1.5624029636383057,
      "learning_rate": 1.7941828498610503e-05,
      "loss": 1.5025,
      "step": 22678
    },
    {
      "epoch": 0.8121832864791305,
      "grad_norm": 1.589890956878662,
      "learning_rate": 1.793519987580631e-05,
      "loss": 1.1414,
      "step": 22679
    },
    {
      "epoch": 0.8122190986086988,
      "grad_norm": 2.1349070072174072,
      "learning_rate": 1.7928572357078143e-05,
      "loss": 1.079,
      "step": 22680
    },
    {
      "epoch": 0.812254910738267,
      "grad_norm": 1.6790281534194946,
      "learning_rate": 1.7921945942515195e-05,
      "loss": 1.5152,
      "step": 22681
    },
    {
      "epoch": 0.8122907228678353,
      "grad_norm": 1.7030316591262817,
      "learning_rate": 1.7915320632206566e-05,
      "loss": 1.0342,
      "step": 22682
    },
    {
      "epoch": 0.8123265349974036,
      "grad_norm": 1.1964973211288452,
      "learning_rate": 1.7908696426241422e-05,
      "loss": 1.2793,
      "step": 22683
    },
    {
      "epoch": 0.8123623471269719,
      "grad_norm": 1.6645698547363281,
      "learning_rate": 1.790207332470887e-05,
      "loss": 1.5247,
      "step": 22684
    },
    {
      "epoch": 0.8123981592565402,
      "grad_norm": 2.757772207260132,
      "learning_rate": 1.7895451327698054e-05,
      "loss": 1.5142,
      "step": 22685
    },
    {
      "epoch": 0.8124339713861085,
      "grad_norm": 1.540874719619751,
      "learning_rate": 1.788883043529801e-05,
      "loss": 1.4304,
      "step": 22686
    },
    {
      "epoch": 0.8124697835156768,
      "grad_norm": 1.5452402830123901,
      "learning_rate": 1.788221064759783e-05,
      "loss": 1.5054,
      "step": 22687
    },
    {
      "epoch": 0.812505595645245,
      "grad_norm": 1.5464129447937012,
      "learning_rate": 1.78755919646866e-05,
      "loss": 1.294,
      "step": 22688
    },
    {
      "epoch": 0.8125414077748133,
      "grad_norm": 2.5941967964172363,
      "learning_rate": 1.7868974386653336e-05,
      "loss": 1.4364,
      "step": 22689
    },
    {
      "epoch": 0.8125772199043816,
      "grad_norm": 1.8890364170074463,
      "learning_rate": 1.786235791358707e-05,
      "loss": 1.3582,
      "step": 22690
    },
    {
      "epoch": 0.8126130320339499,
      "grad_norm": 2.1199588775634766,
      "learning_rate": 1.7855742545576836e-05,
      "loss": 1.8747,
      "step": 22691
    },
    {
      "epoch": 0.8126488441635182,
      "grad_norm": 2.581043004989624,
      "learning_rate": 1.7849128282711647e-05,
      "loss": 1.471,
      "step": 22692
    },
    {
      "epoch": 0.8126846562930865,
      "grad_norm": 1.238418698310852,
      "learning_rate": 1.784251512508045e-05,
      "loss": 1.4502,
      "step": 22693
    },
    {
      "epoch": 0.8127204684226548,
      "grad_norm": 1.6828545331954956,
      "learning_rate": 1.783590307277223e-05,
      "loss": 1.8595,
      "step": 22694
    },
    {
      "epoch": 0.812756280552223,
      "grad_norm": 1.997360348701477,
      "learning_rate": 1.7829292125875984e-05,
      "loss": 1.4948,
      "step": 22695
    },
    {
      "epoch": 0.8127920926817913,
      "grad_norm": 1.4616385698318481,
      "learning_rate": 1.7822682284480585e-05,
      "loss": 1.6081,
      "step": 22696
    },
    {
      "epoch": 0.8128279048113596,
      "grad_norm": 1.9267231225967407,
      "learning_rate": 1.7816073548675004e-05,
      "loss": 1.3649,
      "step": 22697
    },
    {
      "epoch": 0.8128637169409278,
      "grad_norm": 2.3256096839904785,
      "learning_rate": 1.780946591854814e-05,
      "loss": 1.3092,
      "step": 22698
    },
    {
      "epoch": 0.8128995290704962,
      "grad_norm": 1.9210196733474731,
      "learning_rate": 1.780285939418892e-05,
      "loss": 1.3377,
      "step": 22699
    },
    {
      "epoch": 0.8129353412000645,
      "grad_norm": 1.4383782148361206,
      "learning_rate": 1.7796253975686172e-05,
      "loss": 1.3547,
      "step": 22700
    },
    {
      "epoch": 0.8129711533296328,
      "grad_norm": 1.6383898258209229,
      "learning_rate": 1.7789649663128793e-05,
      "loss": 1.1577,
      "step": 22701
    },
    {
      "epoch": 0.813006965459201,
      "grad_norm": 1.306366205215454,
      "learning_rate": 1.7783046456605658e-05,
      "loss": 1.4998,
      "step": 22702
    },
    {
      "epoch": 0.8130427775887693,
      "grad_norm": 1.700914978981018,
      "learning_rate": 1.7776444356205556e-05,
      "loss": 1.4648,
      "step": 22703
    },
    {
      "epoch": 0.8130785897183376,
      "grad_norm": 1.3967289924621582,
      "learning_rate": 1.7769843362017336e-05,
      "loss": 1.6431,
      "step": 22704
    },
    {
      "epoch": 0.8131144018479058,
      "grad_norm": 1.7971333265304565,
      "learning_rate": 1.77632434741298e-05,
      "loss": 1.5475,
      "step": 22705
    },
    {
      "epoch": 0.8131502139774742,
      "grad_norm": 1.5884615182876587,
      "learning_rate": 1.7756644692631773e-05,
      "loss": 1.3453,
      "step": 22706
    },
    {
      "epoch": 0.8131860261070425,
      "grad_norm": 2.3906350135803223,
      "learning_rate": 1.7750047017611983e-05,
      "loss": 1.4263,
      "step": 22707
    },
    {
      "epoch": 0.8132218382366108,
      "grad_norm": 1.615214228630066,
      "learning_rate": 1.7743450449159217e-05,
      "loss": 1.0195,
      "step": 22708
    },
    {
      "epoch": 0.813257650366179,
      "grad_norm": 1.702710509300232,
      "learning_rate": 1.7736854987362217e-05,
      "loss": 1.5819,
      "step": 22709
    },
    {
      "epoch": 0.8132934624957473,
      "grad_norm": 1.4902650117874146,
      "learning_rate": 1.773026063230975e-05,
      "loss": 1.6003,
      "step": 22710
    },
    {
      "epoch": 0.8133292746253156,
      "grad_norm": 1.6640796661376953,
      "learning_rate": 1.7723667384090503e-05,
      "loss": 1.5305,
      "step": 22711
    },
    {
      "epoch": 0.8133650867548838,
      "grad_norm": 1.3970403671264648,
      "learning_rate": 1.7717075242793123e-05,
      "loss": 1.4203,
      "step": 22712
    },
    {
      "epoch": 0.8134008988844522,
      "grad_norm": 1.4592771530151367,
      "learning_rate": 1.771048420850643e-05,
      "loss": 1.5402,
      "step": 22713
    },
    {
      "epoch": 0.8134367110140205,
      "grad_norm": 1.4296540021896362,
      "learning_rate": 1.770389428131899e-05,
      "loss": 1.4899,
      "step": 22714
    },
    {
      "epoch": 0.8134725231435888,
      "grad_norm": 1.440382957458496,
      "learning_rate": 1.769730546131949e-05,
      "loss": 1.4909,
      "step": 22715
    },
    {
      "epoch": 0.813508335273157,
      "grad_norm": 1.3283933401107788,
      "learning_rate": 1.7690717748596585e-05,
      "loss": 1.6033,
      "step": 22716
    },
    {
      "epoch": 0.8135441474027253,
      "grad_norm": 1.8007736206054688,
      "learning_rate": 1.7684131143238937e-05,
      "loss": 1.4637,
      "step": 22717
    },
    {
      "epoch": 0.8135799595322936,
      "grad_norm": 1.7173250913619995,
      "learning_rate": 1.767754564533509e-05,
      "loss": 1.4911,
      "step": 22718
    },
    {
      "epoch": 0.8136157716618618,
      "grad_norm": 1.7633723020553589,
      "learning_rate": 1.7670961254973682e-05,
      "loss": 1.6723,
      "step": 22719
    },
    {
      "epoch": 0.8136515837914302,
      "grad_norm": 2.1998519897460938,
      "learning_rate": 1.766437797224332e-05,
      "loss": 1.2689,
      "step": 22720
    },
    {
      "epoch": 0.8136873959209985,
      "grad_norm": 1.6470333337783813,
      "learning_rate": 1.7657795797232525e-05,
      "loss": 1.4681,
      "step": 22721
    },
    {
      "epoch": 0.8137232080505667,
      "grad_norm": 1.4363399744033813,
      "learning_rate": 1.7651214730029897e-05,
      "loss": 1.5119,
      "step": 22722
    },
    {
      "epoch": 0.813759020180135,
      "grad_norm": 1.6144115924835205,
      "learning_rate": 1.7644634770723888e-05,
      "loss": 1.2808,
      "step": 22723
    },
    {
      "epoch": 0.8137948323097033,
      "grad_norm": 1.3689866065979004,
      "learning_rate": 1.763805591940315e-05,
      "loss": 1.2235,
      "step": 22724
    },
    {
      "epoch": 0.8138306444392716,
      "grad_norm": 1.7141789197921753,
      "learning_rate": 1.7631478176156113e-05,
      "loss": 1.2982,
      "step": 22725
    },
    {
      "epoch": 0.8138664565688398,
      "grad_norm": 1.6982225179672241,
      "learning_rate": 1.762490154107128e-05,
      "loss": 1.4998,
      "step": 22726
    },
    {
      "epoch": 0.8139022686984082,
      "grad_norm": 1.3144981861114502,
      "learning_rate": 1.761832601423714e-05,
      "loss": 1.3089,
      "step": 22727
    },
    {
      "epoch": 0.8139380808279765,
      "grad_norm": 1.2160977125167847,
      "learning_rate": 1.7611751595742188e-05,
      "loss": 1.0489,
      "step": 22728
    },
    {
      "epoch": 0.8139738929575447,
      "grad_norm": 1.6171183586120605,
      "learning_rate": 1.760517828567484e-05,
      "loss": 1.6103,
      "step": 22729
    },
    {
      "epoch": 0.814009705087113,
      "grad_norm": 1.4586262702941895,
      "learning_rate": 1.759860608412349e-05,
      "loss": 1.0743,
      "step": 22730
    },
    {
      "epoch": 0.8140455172166813,
      "grad_norm": 1.4333014488220215,
      "learning_rate": 1.7592034991176654e-05,
      "loss": 1.4654,
      "step": 22731
    },
    {
      "epoch": 0.8140813293462495,
      "grad_norm": 1.6419109106063843,
      "learning_rate": 1.7585465006922662e-05,
      "loss": 1.2001,
      "step": 22732
    },
    {
      "epoch": 0.8141171414758178,
      "grad_norm": 1.2818939685821533,
      "learning_rate": 1.757889613144995e-05,
      "loss": 1.3681,
      "step": 22733
    },
    {
      "epoch": 0.8141529536053862,
      "grad_norm": 1.5382506847381592,
      "learning_rate": 1.7572328364846836e-05,
      "loss": 1.4991,
      "step": 22734
    },
    {
      "epoch": 0.8141887657349545,
      "grad_norm": 1.436592698097229,
      "learning_rate": 1.7565761707201767e-05,
      "loss": 1.2363,
      "step": 22735
    },
    {
      "epoch": 0.8142245778645227,
      "grad_norm": 1.5679230690002441,
      "learning_rate": 1.7559196158603018e-05,
      "loss": 1.3993,
      "step": 22736
    },
    {
      "epoch": 0.814260389994091,
      "grad_norm": 1.6096587181091309,
      "learning_rate": 1.755263171913897e-05,
      "loss": 1.2466,
      "step": 22737
    },
    {
      "epoch": 0.8142962021236593,
      "grad_norm": 1.662288784980774,
      "learning_rate": 1.7546068388897885e-05,
      "loss": 1.4613,
      "step": 22738
    },
    {
      "epoch": 0.8143320142532275,
      "grad_norm": 1.7570534944534302,
      "learning_rate": 1.753950616796809e-05,
      "loss": 1.493,
      "step": 22739
    },
    {
      "epoch": 0.8143678263827958,
      "grad_norm": 1.7655926942825317,
      "learning_rate": 1.75329450564379e-05,
      "loss": 1.3305,
      "step": 22740
    },
    {
      "epoch": 0.8144036385123642,
      "grad_norm": 1.5644776821136475,
      "learning_rate": 1.75263850543955e-05,
      "loss": 1.2155,
      "step": 22741
    },
    {
      "epoch": 0.8144394506419325,
      "grad_norm": 2.489290714263916,
      "learning_rate": 1.7519826161929266e-05,
      "loss": 1.6175,
      "step": 22742
    },
    {
      "epoch": 0.8144752627715007,
      "grad_norm": 1.4218047857284546,
      "learning_rate": 1.7513268379127356e-05,
      "loss": 1.5427,
      "step": 22743
    },
    {
      "epoch": 0.814511074901069,
      "grad_norm": 1.5447837114334106,
      "learning_rate": 1.750671170607804e-05,
      "loss": 1.0916,
      "step": 22744
    },
    {
      "epoch": 0.8145468870306373,
      "grad_norm": 1.4648761749267578,
      "learning_rate": 1.7500156142869462e-05,
      "loss": 1.311,
      "step": 22745
    },
    {
      "epoch": 0.8145826991602055,
      "grad_norm": 1.8207685947418213,
      "learning_rate": 1.7493601689589913e-05,
      "loss": 1.169,
      "step": 22746
    },
    {
      "epoch": 0.8146185112897738,
      "grad_norm": 1.640710711479187,
      "learning_rate": 1.748704834632753e-05,
      "loss": 1.5011,
      "step": 22747
    },
    {
      "epoch": 0.8146543234193422,
      "grad_norm": 1.3525753021240234,
      "learning_rate": 1.748049611317045e-05,
      "loss": 1.5688,
      "step": 22748
    },
    {
      "epoch": 0.8146901355489105,
      "grad_norm": 1.8010128736495972,
      "learning_rate": 1.7473944990206858e-05,
      "loss": 1.1804,
      "step": 22749
    },
    {
      "epoch": 0.8147259476784787,
      "grad_norm": 1.575965404510498,
      "learning_rate": 1.7467394977524876e-05,
      "loss": 1.385,
      "step": 22750
    },
    {
      "epoch": 0.814761759808047,
      "grad_norm": 1.5879836082458496,
      "learning_rate": 1.7460846075212666e-05,
      "loss": 1.0562,
      "step": 22751
    },
    {
      "epoch": 0.8147975719376153,
      "grad_norm": 1.2580584287643433,
      "learning_rate": 1.745429828335826e-05,
      "loss": 1.467,
      "step": 22752
    },
    {
      "epoch": 0.8148333840671835,
      "grad_norm": 1.7832731008529663,
      "learning_rate": 1.7447751602049835e-05,
      "loss": 1.1617,
      "step": 22753
    },
    {
      "epoch": 0.8148691961967518,
      "grad_norm": 1.6994675397872925,
      "learning_rate": 1.74412060313754e-05,
      "loss": 1.5291,
      "step": 22754
    },
    {
      "epoch": 0.8149050083263202,
      "grad_norm": 1.608042597770691,
      "learning_rate": 1.7434661571423084e-05,
      "loss": 1.4662,
      "step": 22755
    },
    {
      "epoch": 0.8149408204558884,
      "grad_norm": 1.4791151285171509,
      "learning_rate": 1.7428118222280855e-05,
      "loss": 1.4369,
      "step": 22756
    },
    {
      "epoch": 0.8149766325854567,
      "grad_norm": 1.5534361600875854,
      "learning_rate": 1.7421575984036797e-05,
      "loss": 1.4834,
      "step": 22757
    },
    {
      "epoch": 0.815012444715025,
      "grad_norm": 1.302477240562439,
      "learning_rate": 1.7415034856778934e-05,
      "loss": 1.4125,
      "step": 22758
    },
    {
      "epoch": 0.8150482568445933,
      "grad_norm": 1.541063666343689,
      "learning_rate": 1.7408494840595224e-05,
      "loss": 1.2171,
      "step": 22759
    },
    {
      "epoch": 0.8150840689741615,
      "grad_norm": 2.021052122116089,
      "learning_rate": 1.7401955935573688e-05,
      "loss": 1.6671,
      "step": 22760
    },
    {
      "epoch": 0.8151198811037298,
      "grad_norm": 2.089958429336548,
      "learning_rate": 1.739541814180228e-05,
      "loss": 1.5129,
      "step": 22761
    },
    {
      "epoch": 0.8151556932332982,
      "grad_norm": 1.526520848274231,
      "learning_rate": 1.7388881459369e-05,
      "loss": 1.4541,
      "step": 22762
    },
    {
      "epoch": 0.8151915053628664,
      "grad_norm": 1.9976831674575806,
      "learning_rate": 1.738234588836174e-05,
      "loss": 1.3426,
      "step": 22763
    },
    {
      "epoch": 0.8152273174924347,
      "grad_norm": 1.4353028535842896,
      "learning_rate": 1.737581142886844e-05,
      "loss": 1.292,
      "step": 22764
    },
    {
      "epoch": 0.815263129622003,
      "grad_norm": 1.6733012199401855,
      "learning_rate": 1.7369278080977037e-05,
      "loss": 1.4501,
      "step": 22765
    },
    {
      "epoch": 0.8152989417515712,
      "grad_norm": 1.6345852613449097,
      "learning_rate": 1.7362745844775396e-05,
      "loss": 1.4787,
      "step": 22766
    },
    {
      "epoch": 0.8153347538811395,
      "grad_norm": 1.6095476150512695,
      "learning_rate": 1.7356214720351414e-05,
      "loss": 1.1239,
      "step": 22767
    },
    {
      "epoch": 0.8153705660107078,
      "grad_norm": 2.355900526046753,
      "learning_rate": 1.7349684707792956e-05,
      "loss": 1.4855,
      "step": 22768
    },
    {
      "epoch": 0.8154063781402762,
      "grad_norm": 1.734176516532898,
      "learning_rate": 1.7343155807187915e-05,
      "loss": 1.3749,
      "step": 22769
    },
    {
      "epoch": 0.8154421902698444,
      "grad_norm": 2.209763765335083,
      "learning_rate": 1.7336628018624058e-05,
      "loss": 1.4435,
      "step": 22770
    },
    {
      "epoch": 0.8154780023994127,
      "grad_norm": 1.8333766460418701,
      "learning_rate": 1.7330101342189254e-05,
      "loss": 1.4678,
      "step": 22771
    },
    {
      "epoch": 0.815513814528981,
      "grad_norm": 2.5825207233428955,
      "learning_rate": 1.732357577797129e-05,
      "loss": 1.0537,
      "step": 22772
    },
    {
      "epoch": 0.8155496266585492,
      "grad_norm": 1.7145079374313354,
      "learning_rate": 1.7317051326057998e-05,
      "loss": 1.4113,
      "step": 22773
    },
    {
      "epoch": 0.8155854387881175,
      "grad_norm": 1.4004257917404175,
      "learning_rate": 1.7310527986537095e-05,
      "loss": 1.4613,
      "step": 22774
    },
    {
      "epoch": 0.8156212509176858,
      "grad_norm": 1.3757426738739014,
      "learning_rate": 1.7304005759496377e-05,
      "loss": 1.5297,
      "step": 22775
    },
    {
      "epoch": 0.815657063047254,
      "grad_norm": 1.3896691799163818,
      "learning_rate": 1.729748464502362e-05,
      "loss": 1.3216,
      "step": 22776
    },
    {
      "epoch": 0.8156928751768224,
      "grad_norm": 1.7896462678909302,
      "learning_rate": 1.729096464320651e-05,
      "loss": 1.4618,
      "step": 22777
    },
    {
      "epoch": 0.8157286873063907,
      "grad_norm": 1.3658167123794556,
      "learning_rate": 1.7284445754132772e-05,
      "loss": 1.6294,
      "step": 22778
    },
    {
      "epoch": 0.815764499435959,
      "grad_norm": 1.7124221324920654,
      "learning_rate": 1.727792797789013e-05,
      "loss": 1.4627,
      "step": 22779
    },
    {
      "epoch": 0.8158003115655272,
      "grad_norm": 1.5317054986953735,
      "learning_rate": 1.7271411314566287e-05,
      "loss": 1.4144,
      "step": 22780
    },
    {
      "epoch": 0.8158361236950955,
      "grad_norm": 1.3445594310760498,
      "learning_rate": 1.7264895764248868e-05,
      "loss": 1.4471,
      "step": 22781
    },
    {
      "epoch": 0.8158719358246638,
      "grad_norm": 1.3334885835647583,
      "learning_rate": 1.7258381327025552e-05,
      "loss": 1.3444,
      "step": 22782
    },
    {
      "epoch": 0.815907747954232,
      "grad_norm": 1.8246225118637085,
      "learning_rate": 1.7251868002984005e-05,
      "loss": 1.5161,
      "step": 22783
    },
    {
      "epoch": 0.8159435600838004,
      "grad_norm": 1.5593852996826172,
      "learning_rate": 1.7245355792211826e-05,
      "loss": 1.6261,
      "step": 22784
    },
    {
      "epoch": 0.8159793722133687,
      "grad_norm": 1.728540301322937,
      "learning_rate": 1.723884469479663e-05,
      "loss": 1.3071,
      "step": 22785
    },
    {
      "epoch": 0.816015184342937,
      "grad_norm": 1.4976931810379028,
      "learning_rate": 1.7232334710826025e-05,
      "loss": 1.4344,
      "step": 22786
    },
    {
      "epoch": 0.8160509964725052,
      "grad_norm": 1.657505989074707,
      "learning_rate": 1.722582584038762e-05,
      "loss": 1.5328,
      "step": 22787
    },
    {
      "epoch": 0.8160868086020735,
      "grad_norm": 1.5885651111602783,
      "learning_rate": 1.7219318083568937e-05,
      "loss": 1.1093,
      "step": 22788
    },
    {
      "epoch": 0.8161226207316418,
      "grad_norm": 1.4087108373641968,
      "learning_rate": 1.7212811440457545e-05,
      "loss": 1.2369,
      "step": 22789
    },
    {
      "epoch": 0.81615843286121,
      "grad_norm": 1.7424745559692383,
      "learning_rate": 1.7206305911141017e-05,
      "loss": 1.5162,
      "step": 22790
    },
    {
      "epoch": 0.8161942449907784,
      "grad_norm": 1.5498261451721191,
      "learning_rate": 1.7199801495706812e-05,
      "loss": 1.087,
      "step": 22791
    },
    {
      "epoch": 0.8162300571203467,
      "grad_norm": 1.4619683027267456,
      "learning_rate": 1.719329819424248e-05,
      "loss": 1.4748,
      "step": 22792
    },
    {
      "epoch": 0.816265869249915,
      "grad_norm": 1.8188879489898682,
      "learning_rate": 1.7186796006835514e-05,
      "loss": 1.5167,
      "step": 22793
    },
    {
      "epoch": 0.8163016813794832,
      "grad_norm": 1.7282185554504395,
      "learning_rate": 1.7180294933573405e-05,
      "loss": 1.5527,
      "step": 22794
    },
    {
      "epoch": 0.8163374935090515,
      "grad_norm": 1.842283844947815,
      "learning_rate": 1.7173794974543568e-05,
      "loss": 1.3138,
      "step": 22795
    },
    {
      "epoch": 0.8163733056386198,
      "grad_norm": 2.1575214862823486,
      "learning_rate": 1.7167296129833488e-05,
      "loss": 1.7364,
      "step": 22796
    },
    {
      "epoch": 0.816409117768188,
      "grad_norm": 1.707844614982605,
      "learning_rate": 1.7160798399530586e-05,
      "loss": 1.282,
      "step": 22797
    },
    {
      "epoch": 0.8164449298977564,
      "grad_norm": 1.5645873546600342,
      "learning_rate": 1.7154301783722315e-05,
      "loss": 1.1909,
      "step": 22798
    },
    {
      "epoch": 0.8164807420273247,
      "grad_norm": 1.1664835214614868,
      "learning_rate": 1.7147806282496027e-05,
      "loss": 1.4765,
      "step": 22799
    },
    {
      "epoch": 0.816516554156893,
      "grad_norm": 1.5875605344772339,
      "learning_rate": 1.7141311895939137e-05,
      "loss": 1.5326,
      "step": 22800
    },
    {
      "epoch": 0.8165523662864612,
      "grad_norm": 1.8210203647613525,
      "learning_rate": 1.7134818624139036e-05,
      "loss": 1.204,
      "step": 22801
    },
    {
      "epoch": 0.8165881784160295,
      "grad_norm": 1.5193594694137573,
      "learning_rate": 1.7128326467183032e-05,
      "loss": 1.2251,
      "step": 22802
    },
    {
      "epoch": 0.8166239905455978,
      "grad_norm": 1.719142198562622,
      "learning_rate": 1.7121835425158506e-05,
      "loss": 1.6163,
      "step": 22803
    },
    {
      "epoch": 0.816659802675166,
      "grad_norm": 1.998476266860962,
      "learning_rate": 1.711534549815278e-05,
      "loss": 1.3592,
      "step": 22804
    },
    {
      "epoch": 0.8166956148047344,
      "grad_norm": 1.435386300086975,
      "learning_rate": 1.7108856686253183e-05,
      "loss": 1.2512,
      "step": 22805
    },
    {
      "epoch": 0.8167314269343027,
      "grad_norm": 1.4732717275619507,
      "learning_rate": 1.710236898954698e-05,
      "loss": 1.3226,
      "step": 22806
    },
    {
      "epoch": 0.8167672390638709,
      "grad_norm": 1.7154697179794312,
      "learning_rate": 1.7095882408121468e-05,
      "loss": 1.2685,
      "step": 22807
    },
    {
      "epoch": 0.8168030511934392,
      "grad_norm": 1.3846417665481567,
      "learning_rate": 1.708939694206395e-05,
      "loss": 1.543,
      "step": 22808
    },
    {
      "epoch": 0.8168388633230075,
      "grad_norm": 1.680985689163208,
      "learning_rate": 1.708291259146162e-05,
      "loss": 1.3678,
      "step": 22809
    },
    {
      "epoch": 0.8168746754525757,
      "grad_norm": 1.3640410900115967,
      "learning_rate": 1.7076429356401748e-05,
      "loss": 1.5972,
      "step": 22810
    },
    {
      "epoch": 0.816910487582144,
      "grad_norm": 1.7066844701766968,
      "learning_rate": 1.706994723697155e-05,
      "loss": 1.1552,
      "step": 22811
    },
    {
      "epoch": 0.8169462997117124,
      "grad_norm": 2.2980430126190186,
      "learning_rate": 1.7063466233258275e-05,
      "loss": 1.3957,
      "step": 22812
    },
    {
      "epoch": 0.8169821118412807,
      "grad_norm": 1.7297414541244507,
      "learning_rate": 1.7056986345349046e-05,
      "loss": 1.5695,
      "step": 22813
    },
    {
      "epoch": 0.8170179239708489,
      "grad_norm": 2.15028977394104,
      "learning_rate": 1.7050507573331077e-05,
      "loss": 1.2113,
      "step": 22814
    },
    {
      "epoch": 0.8170537361004172,
      "grad_norm": 1.3350656032562256,
      "learning_rate": 1.7044029917291536e-05,
      "loss": 1.1719,
      "step": 22815
    },
    {
      "epoch": 0.8170895482299855,
      "grad_norm": 1.7217737436294556,
      "learning_rate": 1.7037553377317595e-05,
      "loss": 1.2585,
      "step": 22816
    },
    {
      "epoch": 0.8171253603595537,
      "grad_norm": 1.6362788677215576,
      "learning_rate": 1.7031077953496356e-05,
      "loss": 1.433,
      "step": 22817
    },
    {
      "epoch": 0.817161172489122,
      "grad_norm": 1.7198362350463867,
      "learning_rate": 1.7024603645914896e-05,
      "loss": 1.3631,
      "step": 22818
    },
    {
      "epoch": 0.8171969846186904,
      "grad_norm": 1.4463917016983032,
      "learning_rate": 1.7018130454660395e-05,
      "loss": 1.486,
      "step": 22819
    },
    {
      "epoch": 0.8172327967482587,
      "grad_norm": 2.732374906539917,
      "learning_rate": 1.7011658379819904e-05,
      "loss": 1.2624,
      "step": 22820
    },
    {
      "epoch": 0.8172686088778269,
      "grad_norm": 1.4441410303115845,
      "learning_rate": 1.7005187421480517e-05,
      "loss": 1.2175,
      "step": 22821
    },
    {
      "epoch": 0.8173044210073952,
      "grad_norm": 1.7178325653076172,
      "learning_rate": 1.699871757972924e-05,
      "loss": 1.5518,
      "step": 22822
    },
    {
      "epoch": 0.8173402331369635,
      "grad_norm": 2.081376075744629,
      "learning_rate": 1.6992248854653192e-05,
      "loss": 1.4921,
      "step": 22823
    },
    {
      "epoch": 0.8173760452665317,
      "grad_norm": 2.20284104347229,
      "learning_rate": 1.698578124633934e-05,
      "loss": 1.6761,
      "step": 22824
    },
    {
      "epoch": 0.8174118573961,
      "grad_norm": 1.9613070487976074,
      "learning_rate": 1.6979314754874733e-05,
      "loss": 1.6807,
      "step": 22825
    },
    {
      "epoch": 0.8174476695256684,
      "grad_norm": 1.3908569812774658,
      "learning_rate": 1.6972849380346367e-05,
      "loss": 1.3088,
      "step": 22826
    },
    {
      "epoch": 0.8174834816552367,
      "grad_norm": 1.7931389808654785,
      "learning_rate": 1.696638512284119e-05,
      "loss": 1.5472,
      "step": 22827
    },
    {
      "epoch": 0.8175192937848049,
      "grad_norm": 2.0211594104766846,
      "learning_rate": 1.6959921982446225e-05,
      "loss": 1.6009,
      "step": 22828
    },
    {
      "epoch": 0.8175551059143732,
      "grad_norm": 1.730381965637207,
      "learning_rate": 1.6953459959248354e-05,
      "loss": 1.2824,
      "step": 22829
    },
    {
      "epoch": 0.8175909180439415,
      "grad_norm": 1.6810767650604248,
      "learning_rate": 1.69469990533346e-05,
      "loss": 1.3071,
      "step": 22830
    },
    {
      "epoch": 0.8176267301735097,
      "grad_norm": 1.639445185661316,
      "learning_rate": 1.694053926479181e-05,
      "loss": 1.2268,
      "step": 22831
    },
    {
      "epoch": 0.817662542303078,
      "grad_norm": 1.8805431127548218,
      "learning_rate": 1.6934080593706958e-05,
      "loss": 1.2137,
      "step": 22832
    },
    {
      "epoch": 0.8176983544326464,
      "grad_norm": 1.8979594707489014,
      "learning_rate": 1.692762304016685e-05,
      "loss": 1.4345,
      "step": 22833
    },
    {
      "epoch": 0.8177341665622146,
      "grad_norm": 1.65707528591156,
      "learning_rate": 1.6921166604258475e-05,
      "loss": 1.3987,
      "step": 22834
    },
    {
      "epoch": 0.8177699786917829,
      "grad_norm": 1.6859840154647827,
      "learning_rate": 1.691471128606864e-05,
      "loss": 1.6425,
      "step": 22835
    },
    {
      "epoch": 0.8178057908213512,
      "grad_norm": 1.899539589881897,
      "learning_rate": 1.6908257085684143e-05,
      "loss": 1.5508,
      "step": 22836
    },
    {
      "epoch": 0.8178416029509195,
      "grad_norm": 1.8419172763824463,
      "learning_rate": 1.6901804003191914e-05,
      "loss": 1.3642,
      "step": 22837
    },
    {
      "epoch": 0.8178774150804877,
      "grad_norm": 1.5693961381912231,
      "learning_rate": 1.6895352038678692e-05,
      "loss": 1.6069,
      "step": 22838
    },
    {
      "epoch": 0.817913227210056,
      "grad_norm": 1.3967636823654175,
      "learning_rate": 1.6888901192231342e-05,
      "loss": 1.4029,
      "step": 22839
    },
    {
      "epoch": 0.8179490393396244,
      "grad_norm": 1.2297767400741577,
      "learning_rate": 1.6882451463936566e-05,
      "loss": 1.2852,
      "step": 22840
    },
    {
      "epoch": 0.8179848514691926,
      "grad_norm": 1.8269104957580566,
      "learning_rate": 1.6876002853881244e-05,
      "loss": 1.4348,
      "step": 22841
    },
    {
      "epoch": 0.8180206635987609,
      "grad_norm": 1.6397745609283447,
      "learning_rate": 1.6869555362152056e-05,
      "loss": 1.2093,
      "step": 22842
    },
    {
      "epoch": 0.8180564757283292,
      "grad_norm": 1.4336014986038208,
      "learning_rate": 1.6863108988835797e-05,
      "loss": 1.1987,
      "step": 22843
    },
    {
      "epoch": 0.8180922878578974,
      "grad_norm": 2.0080535411834717,
      "learning_rate": 1.685666373401914e-05,
      "loss": 1.517,
      "step": 22844
    },
    {
      "epoch": 0.8181280999874657,
      "grad_norm": 1.9427084922790527,
      "learning_rate": 1.685021959778883e-05,
      "loss": 1.4457,
      "step": 22845
    },
    {
      "epoch": 0.818163912117034,
      "grad_norm": 1.9075568914413452,
      "learning_rate": 1.6843776580231586e-05,
      "loss": 1.2082,
      "step": 22846
    },
    {
      "epoch": 0.8181997242466024,
      "grad_norm": 1.3768006563186646,
      "learning_rate": 1.6837334681434037e-05,
      "loss": 1.3797,
      "step": 22847
    },
    {
      "epoch": 0.8182355363761706,
      "grad_norm": 2.21150541305542,
      "learning_rate": 1.683089390148287e-05,
      "loss": 1.6664,
      "step": 22848
    },
    {
      "epoch": 0.8182713485057389,
      "grad_norm": 2.0442562103271484,
      "learning_rate": 1.6824454240464748e-05,
      "loss": 1.4715,
      "step": 22849
    },
    {
      "epoch": 0.8183071606353072,
      "grad_norm": 1.8075312376022339,
      "learning_rate": 1.6818015698466338e-05,
      "loss": 1.2415,
      "step": 22850
    },
    {
      "epoch": 0.8183429727648754,
      "grad_norm": 1.6639543771743774,
      "learning_rate": 1.681157827557418e-05,
      "loss": 1.3192,
      "step": 22851
    },
    {
      "epoch": 0.8183787848944437,
      "grad_norm": 1.579306960105896,
      "learning_rate": 1.680514197187497e-05,
      "loss": 1.5554,
      "step": 22852
    },
    {
      "epoch": 0.818414597024012,
      "grad_norm": 1.3775279521942139,
      "learning_rate": 1.6798706787455264e-05,
      "loss": 1.4963,
      "step": 22853
    },
    {
      "epoch": 0.8184504091535804,
      "grad_norm": 2.1027517318725586,
      "learning_rate": 1.6792272722401626e-05,
      "loss": 1.2141,
      "step": 22854
    },
    {
      "epoch": 0.8184862212831486,
      "grad_norm": 1.735244870185852,
      "learning_rate": 1.6785839776800615e-05,
      "loss": 1.4396,
      "step": 22855
    },
    {
      "epoch": 0.8185220334127169,
      "grad_norm": 1.4876716136932373,
      "learning_rate": 1.677940795073879e-05,
      "loss": 1.7742,
      "step": 22856
    },
    {
      "epoch": 0.8185578455422852,
      "grad_norm": 1.8125195503234863,
      "learning_rate": 1.6772977244302714e-05,
      "loss": 1.6015,
      "step": 22857
    },
    {
      "epoch": 0.8185936576718534,
      "grad_norm": 1.8210275173187256,
      "learning_rate": 1.6766547657578844e-05,
      "loss": 1.4319,
      "step": 22858
    },
    {
      "epoch": 0.8186294698014217,
      "grad_norm": 1.593743085861206,
      "learning_rate": 1.6760119190653724e-05,
      "loss": 1.2788,
      "step": 22859
    },
    {
      "epoch": 0.81866528193099,
      "grad_norm": 1.4638431072235107,
      "learning_rate": 1.6753691843613818e-05,
      "loss": 1.3581,
      "step": 22860
    },
    {
      "epoch": 0.8187010940605584,
      "grad_norm": 1.6058062314987183,
      "learning_rate": 1.6747265616545625e-05,
      "loss": 1.0587,
      "step": 22861
    },
    {
      "epoch": 0.8187369061901266,
      "grad_norm": 1.6514182090759277,
      "learning_rate": 1.674084050953557e-05,
      "loss": 1.288,
      "step": 22862
    },
    {
      "epoch": 0.8187727183196949,
      "grad_norm": 1.6830849647521973,
      "learning_rate": 1.6734416522670114e-05,
      "loss": 1.466,
      "step": 22863
    },
    {
      "epoch": 0.8188085304492632,
      "grad_norm": 1.5081751346588135,
      "learning_rate": 1.6727993656035702e-05,
      "loss": 1.1591,
      "step": 22864
    },
    {
      "epoch": 0.8188443425788314,
      "grad_norm": 1.801693081855774,
      "learning_rate": 1.672157190971869e-05,
      "loss": 1.4321,
      "step": 22865
    },
    {
      "epoch": 0.8188801547083997,
      "grad_norm": 1.6656297445297241,
      "learning_rate": 1.671515128380551e-05,
      "loss": 1.4196,
      "step": 22866
    },
    {
      "epoch": 0.818915966837968,
      "grad_norm": 1.7084009647369385,
      "learning_rate": 1.6708731778382546e-05,
      "loss": 1.4801,
      "step": 22867
    },
    {
      "epoch": 0.8189517789675363,
      "grad_norm": 1.5573643445968628,
      "learning_rate": 1.6702313393536173e-05,
      "loss": 1.4912,
      "step": 22868
    },
    {
      "epoch": 0.8189875910971046,
      "grad_norm": 1.7359768152236938,
      "learning_rate": 1.6695896129352705e-05,
      "loss": 1.6653,
      "step": 22869
    },
    {
      "epoch": 0.8190234032266729,
      "grad_norm": 1.4117817878723145,
      "learning_rate": 1.66894799859185e-05,
      "loss": 1.2851,
      "step": 22870
    },
    {
      "epoch": 0.8190592153562412,
      "grad_norm": 1.3178173303604126,
      "learning_rate": 1.6683064963319906e-05,
      "loss": 0.9814,
      "step": 22871
    },
    {
      "epoch": 0.8190950274858094,
      "grad_norm": 1.5659393072128296,
      "learning_rate": 1.6676651061643177e-05,
      "loss": 1.4266,
      "step": 22872
    },
    {
      "epoch": 0.8191308396153777,
      "grad_norm": 1.5842785835266113,
      "learning_rate": 1.6670238280974627e-05,
      "loss": 1.5485,
      "step": 22873
    },
    {
      "epoch": 0.819166651744946,
      "grad_norm": 1.5451146364212036,
      "learning_rate": 1.6663826621400537e-05,
      "loss": 1.4251,
      "step": 22874
    },
    {
      "epoch": 0.8192024638745143,
      "grad_norm": 1.8988233804702759,
      "learning_rate": 1.6657416083007184e-05,
      "loss": 1.5944,
      "step": 22875
    },
    {
      "epoch": 0.8192382760040826,
      "grad_norm": 1.5368101596832275,
      "learning_rate": 1.6651006665880776e-05,
      "loss": 1.3794,
      "step": 22876
    },
    {
      "epoch": 0.8192740881336509,
      "grad_norm": 3.667703628540039,
      "learning_rate": 1.6644598370107554e-05,
      "loss": 1.618,
      "step": 22877
    },
    {
      "epoch": 0.8193099002632191,
      "grad_norm": 1.5384092330932617,
      "learning_rate": 1.6638191195773744e-05,
      "loss": 1.3134,
      "step": 22878
    },
    {
      "epoch": 0.8193457123927874,
      "grad_norm": 1.6481189727783203,
      "learning_rate": 1.6631785142965563e-05,
      "loss": 1.5524,
      "step": 22879
    },
    {
      "epoch": 0.8193815245223557,
      "grad_norm": 1.4624632596969604,
      "learning_rate": 1.6625380211769147e-05,
      "loss": 1.3622,
      "step": 22880
    },
    {
      "epoch": 0.819417336651924,
      "grad_norm": 1.7683665752410889,
      "learning_rate": 1.6618976402270704e-05,
      "loss": 1.4516,
      "step": 22881
    },
    {
      "epoch": 0.8194531487814923,
      "grad_norm": 1.6138049364089966,
      "learning_rate": 1.66125737145564e-05,
      "loss": 1.5001,
      "step": 22882
    },
    {
      "epoch": 0.8194889609110606,
      "grad_norm": 1.509700059890747,
      "learning_rate": 1.6606172148712328e-05,
      "loss": 1.4476,
      "step": 22883
    },
    {
      "epoch": 0.8195247730406289,
      "grad_norm": 1.956424593925476,
      "learning_rate": 1.659977170482464e-05,
      "loss": 1.2311,
      "step": 22884
    },
    {
      "epoch": 0.8195605851701971,
      "grad_norm": 1.6657973527908325,
      "learning_rate": 1.6593372382979455e-05,
      "loss": 1.3247,
      "step": 22885
    },
    {
      "epoch": 0.8195963972997654,
      "grad_norm": 1.5629335641860962,
      "learning_rate": 1.658697418326287e-05,
      "loss": 1.169,
      "step": 22886
    },
    {
      "epoch": 0.8196322094293337,
      "grad_norm": 1.9260749816894531,
      "learning_rate": 1.658057710576093e-05,
      "loss": 1.5238,
      "step": 22887
    },
    {
      "epoch": 0.819668021558902,
      "grad_norm": 1.5545670986175537,
      "learning_rate": 1.657418115055973e-05,
      "loss": 1.4566,
      "step": 22888
    },
    {
      "epoch": 0.8197038336884703,
      "grad_norm": 2.068680763244629,
      "learning_rate": 1.6567786317745327e-05,
      "loss": 1.4548,
      "step": 22889
    },
    {
      "epoch": 0.8197396458180386,
      "grad_norm": 1.982734203338623,
      "learning_rate": 1.6561392607403713e-05,
      "loss": 1.3612,
      "step": 22890
    },
    {
      "epoch": 0.8197754579476069,
      "grad_norm": 1.426444172859192,
      "learning_rate": 1.655500001962095e-05,
      "loss": 1.4532,
      "step": 22891
    },
    {
      "epoch": 0.8198112700771751,
      "grad_norm": 1.4528121948242188,
      "learning_rate": 1.6548608554483e-05,
      "loss": 1.4581,
      "step": 22892
    },
    {
      "epoch": 0.8198470822067434,
      "grad_norm": 1.5887699127197266,
      "learning_rate": 1.6542218212075923e-05,
      "loss": 1.633,
      "step": 22893
    },
    {
      "epoch": 0.8198828943363117,
      "grad_norm": 1.5253174304962158,
      "learning_rate": 1.6535828992485613e-05,
      "loss": 1.286,
      "step": 22894
    },
    {
      "epoch": 0.8199187064658799,
      "grad_norm": 1.6747887134552002,
      "learning_rate": 1.6529440895798065e-05,
      "loss": 1.3259,
      "step": 22895
    },
    {
      "epoch": 0.8199545185954483,
      "grad_norm": 1.4363501071929932,
      "learning_rate": 1.6523053922099242e-05,
      "loss": 1.3707,
      "step": 22896
    },
    {
      "epoch": 0.8199903307250166,
      "grad_norm": 1.5821083784103394,
      "learning_rate": 1.651666807147503e-05,
      "loss": 1.4022,
      "step": 22897
    },
    {
      "epoch": 0.8200261428545849,
      "grad_norm": 1.7014988660812378,
      "learning_rate": 1.651028334401137e-05,
      "loss": 1.6497,
      "step": 22898
    },
    {
      "epoch": 0.8200619549841531,
      "grad_norm": 1.6083884239196777,
      "learning_rate": 1.6503899739794138e-05,
      "loss": 1.8573,
      "step": 22899
    },
    {
      "epoch": 0.8200977671137214,
      "grad_norm": 1.4160634279251099,
      "learning_rate": 1.6497517258909267e-05,
      "loss": 1.4247,
      "step": 22900
    },
    {
      "epoch": 0.8201335792432897,
      "grad_norm": 1.8023004531860352,
      "learning_rate": 1.6491135901442567e-05,
      "loss": 1.4982,
      "step": 22901
    },
    {
      "epoch": 0.8201693913728579,
      "grad_norm": 1.3858096599578857,
      "learning_rate": 1.648475566747991e-05,
      "loss": 1.4687,
      "step": 22902
    },
    {
      "epoch": 0.8202052035024263,
      "grad_norm": 3.3740923404693604,
      "learning_rate": 1.6478376557107145e-05,
      "loss": 1.4911,
      "step": 22903
    },
    {
      "epoch": 0.8202410156319946,
      "grad_norm": 1.7053208351135254,
      "learning_rate": 1.647199857041011e-05,
      "loss": 1.6013,
      "step": 22904
    },
    {
      "epoch": 0.8202768277615629,
      "grad_norm": 1.5999023914337158,
      "learning_rate": 1.6465621707474587e-05,
      "loss": 1.5928,
      "step": 22905
    },
    {
      "epoch": 0.8203126398911311,
      "grad_norm": 2.348151206970215,
      "learning_rate": 1.6459245968386327e-05,
      "loss": 1.6773,
      "step": 22906
    },
    {
      "epoch": 0.8203484520206994,
      "grad_norm": 1.8074836730957031,
      "learning_rate": 1.64528713532312e-05,
      "loss": 1.5876,
      "step": 22907
    },
    {
      "epoch": 0.8203842641502677,
      "grad_norm": 1.8491315841674805,
      "learning_rate": 1.64464978620949e-05,
      "loss": 1.4896,
      "step": 22908
    },
    {
      "epoch": 0.8204200762798359,
      "grad_norm": 1.720956563949585,
      "learning_rate": 1.6440125495063185e-05,
      "loss": 1.3682,
      "step": 22909
    },
    {
      "epoch": 0.8204558884094043,
      "grad_norm": 1.3919970989227295,
      "learning_rate": 1.643375425222181e-05,
      "loss": 1.6958,
      "step": 22910
    },
    {
      "epoch": 0.8204917005389726,
      "grad_norm": 1.6565746068954468,
      "learning_rate": 1.6427384133656498e-05,
      "loss": 1.1266,
      "step": 22911
    },
    {
      "epoch": 0.8205275126685408,
      "grad_norm": 1.7313103675842285,
      "learning_rate": 1.64210151394529e-05,
      "loss": 1.4588,
      "step": 22912
    },
    {
      "epoch": 0.8205633247981091,
      "grad_norm": 1.5546603202819824,
      "learning_rate": 1.641464726969675e-05,
      "loss": 1.3061,
      "step": 22913
    },
    {
      "epoch": 0.8205991369276774,
      "grad_norm": 1.9309124946594238,
      "learning_rate": 1.6408280524473706e-05,
      "loss": 1.6808,
      "step": 22914
    },
    {
      "epoch": 0.8206349490572457,
      "grad_norm": 1.3096654415130615,
      "learning_rate": 1.640191490386942e-05,
      "loss": 1.4659,
      "step": 22915
    },
    {
      "epoch": 0.8206707611868139,
      "grad_norm": 1.6897329092025757,
      "learning_rate": 1.6395550407969552e-05,
      "loss": 1.2159,
      "step": 22916
    },
    {
      "epoch": 0.8207065733163823,
      "grad_norm": 1.5496948957443237,
      "learning_rate": 1.6389187036859655e-05,
      "loss": 1.4833,
      "step": 22917
    },
    {
      "epoch": 0.8207423854459506,
      "grad_norm": 2.1769845485687256,
      "learning_rate": 1.638282479062545e-05,
      "loss": 1.7208,
      "step": 22918
    },
    {
      "epoch": 0.8207781975755188,
      "grad_norm": 1.4986789226531982,
      "learning_rate": 1.637646366935246e-05,
      "loss": 1.3027,
      "step": 22919
    },
    {
      "epoch": 0.8208140097050871,
      "grad_norm": 1.4664111137390137,
      "learning_rate": 1.6370103673126267e-05,
      "loss": 1.5531,
      "step": 22920
    },
    {
      "epoch": 0.8208498218346554,
      "grad_norm": 1.5579311847686768,
      "learning_rate": 1.6363744802032476e-05,
      "loss": 1.5757,
      "step": 22921
    },
    {
      "epoch": 0.8208856339642236,
      "grad_norm": 1.549301266670227,
      "learning_rate": 1.6357387056156626e-05,
      "loss": 1.5303,
      "step": 22922
    },
    {
      "epoch": 0.8209214460937919,
      "grad_norm": 1.373755931854248,
      "learning_rate": 1.6351030435584245e-05,
      "loss": 1.5735,
      "step": 22923
    },
    {
      "epoch": 0.8209572582233603,
      "grad_norm": 1.8269177675247192,
      "learning_rate": 1.6344674940400805e-05,
      "loss": 1.607,
      "step": 22924
    },
    {
      "epoch": 0.8209930703529286,
      "grad_norm": 1.6167367696762085,
      "learning_rate": 1.633832057069191e-05,
      "loss": 1.2946,
      "step": 22925
    },
    {
      "epoch": 0.8210288824824968,
      "grad_norm": 1.9242609739303589,
      "learning_rate": 1.6331967326542963e-05,
      "loss": 1.5833,
      "step": 22926
    },
    {
      "epoch": 0.8210646946120651,
      "grad_norm": 1.805993676185608,
      "learning_rate": 1.63256152080395e-05,
      "loss": 1.2707,
      "step": 22927
    },
    {
      "epoch": 0.8211005067416334,
      "grad_norm": 1.4096330404281616,
      "learning_rate": 1.6319264215266894e-05,
      "loss": 1.2395,
      "step": 22928
    },
    {
      "epoch": 0.8211363188712016,
      "grad_norm": 1.5715415477752686,
      "learning_rate": 1.6312914348310704e-05,
      "loss": 1.3947,
      "step": 22929
    },
    {
      "epoch": 0.8211721310007699,
      "grad_norm": 1.2842967510223389,
      "learning_rate": 1.6306565607256285e-05,
      "loss": 1.0019,
      "step": 22930
    },
    {
      "epoch": 0.8212079431303383,
      "grad_norm": 1.9133511781692505,
      "learning_rate": 1.6300217992189082e-05,
      "loss": 1.8043,
      "step": 22931
    },
    {
      "epoch": 0.8212437552599066,
      "grad_norm": 1.7129391431808472,
      "learning_rate": 1.6293871503194458e-05,
      "loss": 1.3144,
      "step": 22932
    },
    {
      "epoch": 0.8212795673894748,
      "grad_norm": 1.822322130203247,
      "learning_rate": 1.6287526140357822e-05,
      "loss": 1.4964,
      "step": 22933
    },
    {
      "epoch": 0.8213153795190431,
      "grad_norm": 1.658325433731079,
      "learning_rate": 1.6281181903764565e-05,
      "loss": 1.5703,
      "step": 22934
    },
    {
      "epoch": 0.8213511916486114,
      "grad_norm": 1.6791235208511353,
      "learning_rate": 1.627483879349997e-05,
      "loss": 1.4307,
      "step": 22935
    },
    {
      "epoch": 0.8213870037781796,
      "grad_norm": 1.326132893562317,
      "learning_rate": 1.626849680964947e-05,
      "loss": 1.3838,
      "step": 22936
    },
    {
      "epoch": 0.8214228159077479,
      "grad_norm": 1.37123441696167,
      "learning_rate": 1.6262155952298307e-05,
      "loss": 1.1731,
      "step": 22937
    },
    {
      "epoch": 0.8214586280373163,
      "grad_norm": 1.4056603908538818,
      "learning_rate": 1.625581622153186e-05,
      "loss": 1.4067,
      "step": 22938
    },
    {
      "epoch": 0.8214944401668846,
      "grad_norm": 1.6344373226165771,
      "learning_rate": 1.6249477617435327e-05,
      "loss": 1.4917,
      "step": 22939
    },
    {
      "epoch": 0.8215302522964528,
      "grad_norm": 1.7621803283691406,
      "learning_rate": 1.6243140140094093e-05,
      "loss": 1.5295,
      "step": 22940
    },
    {
      "epoch": 0.8215660644260211,
      "grad_norm": 1.3223881721496582,
      "learning_rate": 1.6236803789593368e-05,
      "loss": 1.452,
      "step": 22941
    },
    {
      "epoch": 0.8216018765555894,
      "grad_norm": 1.669700026512146,
      "learning_rate": 1.6230468566018375e-05,
      "loss": 1.4759,
      "step": 22942
    },
    {
      "epoch": 0.8216376886851576,
      "grad_norm": 1.5705947875976562,
      "learning_rate": 1.6224134469454366e-05,
      "loss": 1.4296,
      "step": 22943
    },
    {
      "epoch": 0.8216735008147259,
      "grad_norm": 1.8033430576324463,
      "learning_rate": 1.6217801499986573e-05,
      "loss": 1.5341,
      "step": 22944
    },
    {
      "epoch": 0.8217093129442943,
      "grad_norm": 1.3500990867614746,
      "learning_rate": 1.6211469657700217e-05,
      "loss": 1.4331,
      "step": 22945
    },
    {
      "epoch": 0.8217451250738625,
      "grad_norm": 1.8312416076660156,
      "learning_rate": 1.6205138942680408e-05,
      "loss": 1.6042,
      "step": 22946
    },
    {
      "epoch": 0.8217809372034308,
      "grad_norm": 2.076237201690674,
      "learning_rate": 1.6198809355012412e-05,
      "loss": 1.4416,
      "step": 22947
    },
    {
      "epoch": 0.8218167493329991,
      "grad_norm": 2.3092427253723145,
      "learning_rate": 1.6192480894781316e-05,
      "loss": 1.4371,
      "step": 22948
    },
    {
      "epoch": 0.8218525614625674,
      "grad_norm": 1.5285634994506836,
      "learning_rate": 1.6186153562072316e-05,
      "loss": 1.6155,
      "step": 22949
    },
    {
      "epoch": 0.8218883735921356,
      "grad_norm": 1.3765846490859985,
      "learning_rate": 1.617982735697048e-05,
      "loss": 1.0938,
      "step": 22950
    },
    {
      "epoch": 0.8219241857217039,
      "grad_norm": 1.4085685014724731,
      "learning_rate": 1.6173502279560936e-05,
      "loss": 1.4514,
      "step": 22951
    },
    {
      "epoch": 0.8219599978512723,
      "grad_norm": 1.2525451183319092,
      "learning_rate": 1.6167178329928823e-05,
      "loss": 1.2433,
      "step": 22952
    },
    {
      "epoch": 0.8219958099808405,
      "grad_norm": 1.4547854661941528,
      "learning_rate": 1.6160855508159168e-05,
      "loss": 1.2984,
      "step": 22953
    },
    {
      "epoch": 0.8220316221104088,
      "grad_norm": 1.432128667831421,
      "learning_rate": 1.6154533814337058e-05,
      "loss": 1.5026,
      "step": 22954
    },
    {
      "epoch": 0.8220674342399771,
      "grad_norm": 1.9972702264785767,
      "learning_rate": 1.614821324854754e-05,
      "loss": 1.4845,
      "step": 22955
    },
    {
      "epoch": 0.8221032463695453,
      "grad_norm": 1.6977585554122925,
      "learning_rate": 1.6141893810875675e-05,
      "loss": 1.6666,
      "step": 22956
    },
    {
      "epoch": 0.8221390584991136,
      "grad_norm": 1.9989867210388184,
      "learning_rate": 1.6135575501406432e-05,
      "loss": 1.7976,
      "step": 22957
    },
    {
      "epoch": 0.8221748706286819,
      "grad_norm": 1.8499654531478882,
      "learning_rate": 1.6129258320224848e-05,
      "loss": 1.3219,
      "step": 22958
    },
    {
      "epoch": 0.8222106827582503,
      "grad_norm": 2.332317352294922,
      "learning_rate": 1.612294226741593e-05,
      "loss": 1.4394,
      "step": 22959
    },
    {
      "epoch": 0.8222464948878185,
      "grad_norm": 2.4020214080810547,
      "learning_rate": 1.6116627343064605e-05,
      "loss": 1.7889,
      "step": 22960
    },
    {
      "epoch": 0.8222823070173868,
      "grad_norm": 2.036348342895508,
      "learning_rate": 1.611031354725586e-05,
      "loss": 1.3781,
      "step": 22961
    },
    {
      "epoch": 0.8223181191469551,
      "grad_norm": 2.3071060180664062,
      "learning_rate": 1.6104000880074642e-05,
      "loss": 1.5629,
      "step": 22962
    },
    {
      "epoch": 0.8223539312765233,
      "grad_norm": 1.9471147060394287,
      "learning_rate": 1.6097689341605894e-05,
      "loss": 1.464,
      "step": 22963
    },
    {
      "epoch": 0.8223897434060916,
      "grad_norm": 1.335502028465271,
      "learning_rate": 1.6091378931934474e-05,
      "loss": 1.3216,
      "step": 22964
    },
    {
      "epoch": 0.8224255555356599,
      "grad_norm": 1.5332695245742798,
      "learning_rate": 1.6085069651145334e-05,
      "loss": 1.3925,
      "step": 22965
    },
    {
      "epoch": 0.8224613676652283,
      "grad_norm": 1.4483823776245117,
      "learning_rate": 1.6078761499323326e-05,
      "loss": 1.5104,
      "step": 22966
    },
    {
      "epoch": 0.8224971797947965,
      "grad_norm": 1.384883999824524,
      "learning_rate": 1.6072454476553357e-05,
      "loss": 1.522,
      "step": 22967
    },
    {
      "epoch": 0.8225329919243648,
      "grad_norm": 1.8154127597808838,
      "learning_rate": 1.6066148582920237e-05,
      "loss": 1.4938,
      "step": 22968
    },
    {
      "epoch": 0.8225688040539331,
      "grad_norm": 1.559505581855774,
      "learning_rate": 1.6059843818508814e-05,
      "loss": 1.2616,
      "step": 22969
    },
    {
      "epoch": 0.8226046161835013,
      "grad_norm": 1.9433817863464355,
      "learning_rate": 1.605354018340395e-05,
      "loss": 1.1542,
      "step": 22970
    },
    {
      "epoch": 0.8226404283130696,
      "grad_norm": 1.6815561056137085,
      "learning_rate": 1.6047237677690386e-05,
      "loss": 1.3772,
      "step": 22971
    },
    {
      "epoch": 0.8226762404426379,
      "grad_norm": 1.572665810585022,
      "learning_rate": 1.6040936301452957e-05,
      "loss": 1.7138,
      "step": 22972
    },
    {
      "epoch": 0.8227120525722063,
      "grad_norm": 1.9919036626815796,
      "learning_rate": 1.603463605477643e-05,
      "loss": 1.4052,
      "step": 22973
    },
    {
      "epoch": 0.8227478647017745,
      "grad_norm": 1.5342936515808105,
      "learning_rate": 1.602833693774558e-05,
      "loss": 1.0727,
      "step": 22974
    },
    {
      "epoch": 0.8227836768313428,
      "grad_norm": 1.8794121742248535,
      "learning_rate": 1.6022038950445127e-05,
      "loss": 1.6559,
      "step": 22975
    },
    {
      "epoch": 0.8228194889609111,
      "grad_norm": 1.6604645252227783,
      "learning_rate": 1.6015742092959818e-05,
      "loss": 1.4897,
      "step": 22976
    },
    {
      "epoch": 0.8228553010904793,
      "grad_norm": 1.3487586975097656,
      "learning_rate": 1.6009446365374383e-05,
      "loss": 1.3302,
      "step": 22977
    },
    {
      "epoch": 0.8228911132200476,
      "grad_norm": 1.6160697937011719,
      "learning_rate": 1.6003151767773485e-05,
      "loss": 1.1905,
      "step": 22978
    },
    {
      "epoch": 0.8229269253496159,
      "grad_norm": 2.0315003395080566,
      "learning_rate": 1.5996858300241834e-05,
      "loss": 1.4471,
      "step": 22979
    },
    {
      "epoch": 0.8229627374791842,
      "grad_norm": 1.3102805614471436,
      "learning_rate": 1.5990565962864103e-05,
      "loss": 1.4466,
      "step": 22980
    },
    {
      "epoch": 0.8229985496087525,
      "grad_norm": 1.7477620840072632,
      "learning_rate": 1.5984274755724958e-05,
      "loss": 1.5908,
      "step": 22981
    },
    {
      "epoch": 0.8230343617383208,
      "grad_norm": 1.6393771171569824,
      "learning_rate": 1.5977984678909008e-05,
      "loss": 1.3496,
      "step": 22982
    },
    {
      "epoch": 0.823070173867889,
      "grad_norm": 2.298346519470215,
      "learning_rate": 1.597169573250089e-05,
      "loss": 1.623,
      "step": 22983
    },
    {
      "epoch": 0.8231059859974573,
      "grad_norm": 1.93317711353302,
      "learning_rate": 1.5965407916585208e-05,
      "loss": 1.3362,
      "step": 22984
    },
    {
      "epoch": 0.8231417981270256,
      "grad_norm": 1.3797252178192139,
      "learning_rate": 1.59591212312466e-05,
      "loss": 1.1094,
      "step": 22985
    },
    {
      "epoch": 0.8231776102565939,
      "grad_norm": 1.945043921470642,
      "learning_rate": 1.595283567656959e-05,
      "loss": 1.5189,
      "step": 22986
    },
    {
      "epoch": 0.8232134223861622,
      "grad_norm": 1.6652638912200928,
      "learning_rate": 1.5946551252638754e-05,
      "loss": 1.728,
      "step": 22987
    },
    {
      "epoch": 0.8232492345157305,
      "grad_norm": 1.7131400108337402,
      "learning_rate": 1.594026795953868e-05,
      "loss": 1.1332,
      "step": 22988
    },
    {
      "epoch": 0.8232850466452988,
      "grad_norm": 1.8453240394592285,
      "learning_rate": 1.5933985797353844e-05,
      "loss": 1.5197,
      "step": 22989
    },
    {
      "epoch": 0.823320858774867,
      "grad_norm": 2.4557993412017822,
      "learning_rate": 1.5927704766168793e-05,
      "loss": 1.5135,
      "step": 22990
    },
    {
      "epoch": 0.8233566709044353,
      "grad_norm": 1.5093895196914673,
      "learning_rate": 1.5921424866068026e-05,
      "loss": 1.7071,
      "step": 22991
    },
    {
      "epoch": 0.8233924830340036,
      "grad_norm": 1.661433219909668,
      "learning_rate": 1.5915146097136056e-05,
      "loss": 1.4739,
      "step": 22992
    },
    {
      "epoch": 0.8234282951635719,
      "grad_norm": 1.4338335990905762,
      "learning_rate": 1.5908868459457317e-05,
      "loss": 1.4,
      "step": 22993
    },
    {
      "epoch": 0.8234641072931402,
      "grad_norm": 1.3473552465438843,
      "learning_rate": 1.5902591953116287e-05,
      "loss": 0.9404,
      "step": 22994
    },
    {
      "epoch": 0.8234999194227085,
      "grad_norm": 1.7865482568740845,
      "learning_rate": 1.589631657819741e-05,
      "loss": 1.5638,
      "step": 22995
    },
    {
      "epoch": 0.8235357315522768,
      "grad_norm": 1.594167709350586,
      "learning_rate": 1.5890042334785104e-05,
      "loss": 1.2829,
      "step": 22996
    },
    {
      "epoch": 0.823571543681845,
      "grad_norm": 1.8517452478408813,
      "learning_rate": 1.5883769222963775e-05,
      "loss": 1.4047,
      "step": 22997
    },
    {
      "epoch": 0.8236073558114133,
      "grad_norm": 1.841001033782959,
      "learning_rate": 1.587749724281783e-05,
      "loss": 1.166,
      "step": 22998
    },
    {
      "epoch": 0.8236431679409816,
      "grad_norm": 1.69795823097229,
      "learning_rate": 1.5871226394431672e-05,
      "loss": 1.6688,
      "step": 22999
    },
    {
      "epoch": 0.8236789800705498,
      "grad_norm": 1.5626368522644043,
      "learning_rate": 1.586495667788962e-05,
      "loss": 1.3707,
      "step": 23000
    },
    {
      "epoch": 0.8237147922001182,
      "grad_norm": 1.3981742858886719,
      "learning_rate": 1.5858688093276042e-05,
      "loss": 1.0959,
      "step": 23001
    },
    {
      "epoch": 0.8237506043296865,
      "grad_norm": 2.0931923389434814,
      "learning_rate": 1.5852420640675313e-05,
      "loss": 1.35,
      "step": 23002
    },
    {
      "epoch": 0.8237864164592548,
      "grad_norm": 2.05471134185791,
      "learning_rate": 1.5846154320171703e-05,
      "loss": 1.7002,
      "step": 23003
    },
    {
      "epoch": 0.823822228588823,
      "grad_norm": 1.5984715223312378,
      "learning_rate": 1.583988913184953e-05,
      "loss": 1.4799,
      "step": 23004
    },
    {
      "epoch": 0.8238580407183913,
      "grad_norm": 1.48750638961792,
      "learning_rate": 1.583362507579309e-05,
      "loss": 1.209,
      "step": 23005
    },
    {
      "epoch": 0.8238938528479596,
      "grad_norm": 1.6805229187011719,
      "learning_rate": 1.582736215208669e-05,
      "loss": 1.3504,
      "step": 23006
    },
    {
      "epoch": 0.8239296649775278,
      "grad_norm": 1.4237467050552368,
      "learning_rate": 1.582110036081452e-05,
      "loss": 1.0545,
      "step": 23007
    },
    {
      "epoch": 0.8239654771070962,
      "grad_norm": 1.6374319791793823,
      "learning_rate": 1.581483970206087e-05,
      "loss": 1.4251,
      "step": 23008
    },
    {
      "epoch": 0.8240012892366645,
      "grad_norm": 1.6855441331863403,
      "learning_rate": 1.580858017590996e-05,
      "loss": 1.3527,
      "step": 23009
    },
    {
      "epoch": 0.8240371013662328,
      "grad_norm": 1.4074074029922485,
      "learning_rate": 1.5802321782446028e-05,
      "loss": 1.3838,
      "step": 23010
    },
    {
      "epoch": 0.824072913495801,
      "grad_norm": 1.5578721761703491,
      "learning_rate": 1.5796064521753252e-05,
      "loss": 1.4217,
      "step": 23011
    },
    {
      "epoch": 0.8241087256253693,
      "grad_norm": 1.4159759283065796,
      "learning_rate": 1.5789808393915763e-05,
      "loss": 1.5785,
      "step": 23012
    },
    {
      "epoch": 0.8241445377549376,
      "grad_norm": 1.645815372467041,
      "learning_rate": 1.5783553399017825e-05,
      "loss": 1.3073,
      "step": 23013
    },
    {
      "epoch": 0.8241803498845058,
      "grad_norm": 1.8680726289749146,
      "learning_rate": 1.577729953714352e-05,
      "loss": 1.4468,
      "step": 23014
    },
    {
      "epoch": 0.8242161620140742,
      "grad_norm": 1.9843326807022095,
      "learning_rate": 1.577104680837703e-05,
      "loss": 1.527,
      "step": 23015
    },
    {
      "epoch": 0.8242519741436425,
      "grad_norm": 1.7631583213806152,
      "learning_rate": 1.576479521280242e-05,
      "loss": 1.6603,
      "step": 23016
    },
    {
      "epoch": 0.8242877862732108,
      "grad_norm": 1.3905339241027832,
      "learning_rate": 1.575854475050388e-05,
      "loss": 1.0263,
      "step": 23017
    },
    {
      "epoch": 0.824323598402779,
      "grad_norm": 1.588399887084961,
      "learning_rate": 1.5752295421565423e-05,
      "loss": 1.5257,
      "step": 23018
    },
    {
      "epoch": 0.8243594105323473,
      "grad_norm": 2.075728178024292,
      "learning_rate": 1.574604722607117e-05,
      "loss": 1.4954,
      "step": 23019
    },
    {
      "epoch": 0.8243952226619156,
      "grad_norm": 1.729669213294983,
      "learning_rate": 1.573980016410519e-05,
      "loss": 1.5427,
      "step": 23020
    },
    {
      "epoch": 0.8244310347914838,
      "grad_norm": 1.457468867301941,
      "learning_rate": 1.573355423575149e-05,
      "loss": 1.5744,
      "step": 23021
    },
    {
      "epoch": 0.8244668469210522,
      "grad_norm": 2.133328676223755,
      "learning_rate": 1.572730944109415e-05,
      "loss": 1.3736,
      "step": 23022
    },
    {
      "epoch": 0.8245026590506205,
      "grad_norm": 1.4043382406234741,
      "learning_rate": 1.5721065780217103e-05,
      "loss": 1.4783,
      "step": 23023
    },
    {
      "epoch": 0.8245384711801887,
      "grad_norm": 2.212798833847046,
      "learning_rate": 1.5714823253204447e-05,
      "loss": 1.6248,
      "step": 23024
    },
    {
      "epoch": 0.824574283309757,
      "grad_norm": 1.4132270812988281,
      "learning_rate": 1.5708581860140113e-05,
      "loss": 1.2569,
      "step": 23025
    },
    {
      "epoch": 0.8246100954393253,
      "grad_norm": 1.8781002759933472,
      "learning_rate": 1.5702341601108094e-05,
      "loss": 1.5612,
      "step": 23026
    },
    {
      "epoch": 0.8246459075688936,
      "grad_norm": 1.9271550178527832,
      "learning_rate": 1.56961024761923e-05,
      "loss": 1.4968,
      "step": 23027
    },
    {
      "epoch": 0.8246817196984618,
      "grad_norm": 1.5880156755447388,
      "learning_rate": 1.5689864485476736e-05,
      "loss": 1.3802,
      "step": 23028
    },
    {
      "epoch": 0.8247175318280302,
      "grad_norm": 1.7311402559280396,
      "learning_rate": 1.5683627629045295e-05,
      "loss": 1.5533,
      "step": 23029
    },
    {
      "epoch": 0.8247533439575985,
      "grad_norm": 1.6801574230194092,
      "learning_rate": 1.5677391906981842e-05,
      "loss": 1.268,
      "step": 23030
    },
    {
      "epoch": 0.8247891560871667,
      "grad_norm": 1.9536669254302979,
      "learning_rate": 1.5671157319370357e-05,
      "loss": 1.4332,
      "step": 23031
    },
    {
      "epoch": 0.824824968216735,
      "grad_norm": 1.3697175979614258,
      "learning_rate": 1.5664923866294655e-05,
      "loss": 1.2412,
      "step": 23032
    },
    {
      "epoch": 0.8248607803463033,
      "grad_norm": 1.6064835786819458,
      "learning_rate": 1.565869154783863e-05,
      "loss": 1.5359,
      "step": 23033
    },
    {
      "epoch": 0.8248965924758715,
      "grad_norm": 1.9790090322494507,
      "learning_rate": 1.5652460364086084e-05,
      "loss": 1.4676,
      "step": 23034
    },
    {
      "epoch": 0.8249324046054398,
      "grad_norm": 1.8181332349777222,
      "learning_rate": 1.5646230315120923e-05,
      "loss": 1.3089,
      "step": 23035
    },
    {
      "epoch": 0.8249682167350082,
      "grad_norm": 1.756899356842041,
      "learning_rate": 1.5640001401026904e-05,
      "loss": 1.4869,
      "step": 23036
    },
    {
      "epoch": 0.8250040288645765,
      "grad_norm": 1.8076976537704468,
      "learning_rate": 1.5633773621887872e-05,
      "loss": 1.1611,
      "step": 23037
    },
    {
      "epoch": 0.8250398409941447,
      "grad_norm": 1.5319172143936157,
      "learning_rate": 1.5627546977787565e-05,
      "loss": 1.4351,
      "step": 23038
    },
    {
      "epoch": 0.825075653123713,
      "grad_norm": 1.5525784492492676,
      "learning_rate": 1.5621321468809778e-05,
      "loss": 1.4875,
      "step": 23039
    },
    {
      "epoch": 0.8251114652532813,
      "grad_norm": 2.0940845012664795,
      "learning_rate": 1.56150970950383e-05,
      "loss": 1.1187,
      "step": 23040
    },
    {
      "epoch": 0.8251472773828495,
      "grad_norm": 1.537346601486206,
      "learning_rate": 1.5608873856556828e-05,
      "loss": 1.3419,
      "step": 23041
    },
    {
      "epoch": 0.8251830895124178,
      "grad_norm": 1.681191325187683,
      "learning_rate": 1.5602651753449083e-05,
      "loss": 1.2589,
      "step": 23042
    },
    {
      "epoch": 0.8252189016419862,
      "grad_norm": 1.7197188138961792,
      "learning_rate": 1.5596430785798798e-05,
      "loss": 1.5691,
      "step": 23043
    },
    {
      "epoch": 0.8252547137715545,
      "grad_norm": 1.6822034120559692,
      "learning_rate": 1.55902109536897e-05,
      "loss": 1.2887,
      "step": 23044
    },
    {
      "epoch": 0.8252905259011227,
      "grad_norm": 2.1965887546539307,
      "learning_rate": 1.558399225720537e-05,
      "loss": 1.1151,
      "step": 23045
    },
    {
      "epoch": 0.825326338030691,
      "grad_norm": 1.4301897287368774,
      "learning_rate": 1.5577774696429592e-05,
      "loss": 1.5266,
      "step": 23046
    },
    {
      "epoch": 0.8253621501602593,
      "grad_norm": 1.6066406965255737,
      "learning_rate": 1.5571558271445952e-05,
      "loss": 1.5269,
      "step": 23047
    },
    {
      "epoch": 0.8253979622898275,
      "grad_norm": 1.7901389598846436,
      "learning_rate": 1.556534298233807e-05,
      "loss": 1.2883,
      "step": 23048
    },
    {
      "epoch": 0.8254337744193958,
      "grad_norm": 1.3142569065093994,
      "learning_rate": 1.5559128829189597e-05,
      "loss": 1.6844,
      "step": 23049
    },
    {
      "epoch": 0.8254695865489642,
      "grad_norm": 1.5939184427261353,
      "learning_rate": 1.5552915812084113e-05,
      "loss": 1.3867,
      "step": 23050
    },
    {
      "epoch": 0.8255053986785325,
      "grad_norm": 2.42671275138855,
      "learning_rate": 1.5546703931105233e-05,
      "loss": 1.5441,
      "step": 23051
    },
    {
      "epoch": 0.8255412108081007,
      "grad_norm": 2.1185123920440674,
      "learning_rate": 1.5540493186336503e-05,
      "loss": 1.5006,
      "step": 23052
    },
    {
      "epoch": 0.825577022937669,
      "grad_norm": 1.4602375030517578,
      "learning_rate": 1.5534283577861497e-05,
      "loss": 1.3002,
      "step": 23053
    },
    {
      "epoch": 0.8256128350672373,
      "grad_norm": 1.9092152118682861,
      "learning_rate": 1.552807510576374e-05,
      "loss": 1.5135,
      "step": 23054
    },
    {
      "epoch": 0.8256486471968055,
      "grad_norm": 2.400681734085083,
      "learning_rate": 1.5521867770126795e-05,
      "loss": 1.466,
      "step": 23055
    },
    {
      "epoch": 0.8256844593263738,
      "grad_norm": 1.5862021446228027,
      "learning_rate": 1.5515661571034134e-05,
      "loss": 1.2344,
      "step": 23056
    },
    {
      "epoch": 0.8257202714559422,
      "grad_norm": 2.3739986419677734,
      "learning_rate": 1.5509456508569275e-05,
      "loss": 1.4641,
      "step": 23057
    },
    {
      "epoch": 0.8257560835855104,
      "grad_norm": 1.941270351409912,
      "learning_rate": 1.5503252582815707e-05,
      "loss": 1.5392,
      "step": 23058
    },
    {
      "epoch": 0.8257918957150787,
      "grad_norm": 1.816595435142517,
      "learning_rate": 1.5497049793856868e-05,
      "loss": 1.1223,
      "step": 23059
    },
    {
      "epoch": 0.825827707844647,
      "grad_norm": 1.351844072341919,
      "learning_rate": 1.5490848141776214e-05,
      "loss": 1.1367,
      "step": 23060
    },
    {
      "epoch": 0.8258635199742153,
      "grad_norm": 1.4327374696731567,
      "learning_rate": 1.548464762665719e-05,
      "loss": 1.3438,
      "step": 23061
    },
    {
      "epoch": 0.8258993321037835,
      "grad_norm": 1.515933632850647,
      "learning_rate": 1.5478448248583244e-05,
      "loss": 1.2234,
      "step": 23062
    },
    {
      "epoch": 0.8259351442333518,
      "grad_norm": 1.2966159582138062,
      "learning_rate": 1.5472250007637724e-05,
      "loss": 1.4408,
      "step": 23063
    },
    {
      "epoch": 0.8259709563629202,
      "grad_norm": 1.9407875537872314,
      "learning_rate": 1.546605290390405e-05,
      "loss": 1.5882,
      "step": 23064
    },
    {
      "epoch": 0.8260067684924884,
      "grad_norm": 2.346935749053955,
      "learning_rate": 1.545985693746561e-05,
      "loss": 1.293,
      "step": 23065
    },
    {
      "epoch": 0.8260425806220567,
      "grad_norm": 1.5052082538604736,
      "learning_rate": 1.545366210840573e-05,
      "loss": 1.3619,
      "step": 23066
    },
    {
      "epoch": 0.826078392751625,
      "grad_norm": 1.547210693359375,
      "learning_rate": 1.5447468416807766e-05,
      "loss": 1.5157,
      "step": 23067
    },
    {
      "epoch": 0.8261142048811932,
      "grad_norm": 1.6457045078277588,
      "learning_rate": 1.5441275862755043e-05,
      "loss": 1.1963,
      "step": 23068
    },
    {
      "epoch": 0.8261500170107615,
      "grad_norm": 1.628400206565857,
      "learning_rate": 1.5435084446330917e-05,
      "loss": 1.6531,
      "step": 23069
    },
    {
      "epoch": 0.8261858291403298,
      "grad_norm": 1.6187325716018677,
      "learning_rate": 1.5428894167618622e-05,
      "loss": 1.6925,
      "step": 23070
    },
    {
      "epoch": 0.8262216412698982,
      "grad_norm": 1.649355411529541,
      "learning_rate": 1.5422705026701468e-05,
      "loss": 1.3264,
      "step": 23071
    },
    {
      "epoch": 0.8262574533994664,
      "grad_norm": 1.924761176109314,
      "learning_rate": 1.5416517023662713e-05,
      "loss": 1.6002,
      "step": 23072
    },
    {
      "epoch": 0.8262932655290347,
      "grad_norm": 1.5182862281799316,
      "learning_rate": 1.541033015858565e-05,
      "loss": 1.3946,
      "step": 23073
    },
    {
      "epoch": 0.826329077658603,
      "grad_norm": 1.9584239721298218,
      "learning_rate": 1.540414443155345e-05,
      "loss": 1.4373,
      "step": 23074
    },
    {
      "epoch": 0.8263648897881712,
      "grad_norm": 1.620587706565857,
      "learning_rate": 1.5397959842649367e-05,
      "loss": 1.5348,
      "step": 23075
    },
    {
      "epoch": 0.8264007019177395,
      "grad_norm": 1.8111913204193115,
      "learning_rate": 1.5391776391956638e-05,
      "loss": 1.8022,
      "step": 23076
    },
    {
      "epoch": 0.8264365140473078,
      "grad_norm": 1.7288938760757446,
      "learning_rate": 1.5385594079558387e-05,
      "loss": 1.4836,
      "step": 23077
    },
    {
      "epoch": 0.8264723261768762,
      "grad_norm": 1.5004984140396118,
      "learning_rate": 1.5379412905537828e-05,
      "loss": 1.4875,
      "step": 23078
    },
    {
      "epoch": 0.8265081383064444,
      "grad_norm": 2.3542582988739014,
      "learning_rate": 1.5373232869978116e-05,
      "loss": 1.5091,
      "step": 23079
    },
    {
      "epoch": 0.8265439504360127,
      "grad_norm": 1.4314318895339966,
      "learning_rate": 1.5367053972962408e-05,
      "loss": 1.3066,
      "step": 23080
    },
    {
      "epoch": 0.826579762565581,
      "grad_norm": 1.7067185640335083,
      "learning_rate": 1.5360876214573806e-05,
      "loss": 1.3752,
      "step": 23081
    },
    {
      "epoch": 0.8266155746951492,
      "grad_norm": 1.7711745500564575,
      "learning_rate": 1.5354699594895438e-05,
      "loss": 1.4336,
      "step": 23082
    },
    {
      "epoch": 0.8266513868247175,
      "grad_norm": 2.183694839477539,
      "learning_rate": 1.534852411401043e-05,
      "loss": 1.2257,
      "step": 23083
    },
    {
      "epoch": 0.8266871989542858,
      "grad_norm": 1.6529567241668701,
      "learning_rate": 1.5342349772001808e-05,
      "loss": 1.4126,
      "step": 23084
    },
    {
      "epoch": 0.8267230110838542,
      "grad_norm": 2.0304243564605713,
      "learning_rate": 1.5336176568952666e-05,
      "loss": 1.3902,
      "step": 23085
    },
    {
      "epoch": 0.8267588232134224,
      "grad_norm": 1.541917324066162,
      "learning_rate": 1.5330004504946072e-05,
      "loss": 1.6367,
      "step": 23086
    },
    {
      "epoch": 0.8267946353429907,
      "grad_norm": 1.8716808557510376,
      "learning_rate": 1.532383358006506e-05,
      "loss": 1.4428,
      "step": 23087
    },
    {
      "epoch": 0.826830447472559,
      "grad_norm": 1.713232398033142,
      "learning_rate": 1.5317663794392634e-05,
      "loss": 1.2549,
      "step": 23088
    },
    {
      "epoch": 0.8268662596021272,
      "grad_norm": 1.4402745962142944,
      "learning_rate": 1.53114951480118e-05,
      "loss": 1.3368,
      "step": 23089
    },
    {
      "epoch": 0.8269020717316955,
      "grad_norm": 2.111379384994507,
      "learning_rate": 1.5305327641005584e-05,
      "loss": 1.5311,
      "step": 23090
    },
    {
      "epoch": 0.8269378838612638,
      "grad_norm": 1.7146985530853271,
      "learning_rate": 1.5299161273456907e-05,
      "loss": 1.5118,
      "step": 23091
    },
    {
      "epoch": 0.8269736959908321,
      "grad_norm": 1.5040332078933716,
      "learning_rate": 1.529299604544876e-05,
      "loss": 1.0758,
      "step": 23092
    },
    {
      "epoch": 0.8270095081204004,
      "grad_norm": 1.687289834022522,
      "learning_rate": 1.5286831957064095e-05,
      "loss": 1.294,
      "step": 23093
    },
    {
      "epoch": 0.8270453202499687,
      "grad_norm": 1.5885672569274902,
      "learning_rate": 1.528066900838585e-05,
      "loss": 1.3912,
      "step": 23094
    },
    {
      "epoch": 0.827081132379537,
      "grad_norm": 1.3895875215530396,
      "learning_rate": 1.5274507199496913e-05,
      "loss": 1.2929,
      "step": 23095
    },
    {
      "epoch": 0.8271169445091052,
      "grad_norm": 1.6085909605026245,
      "learning_rate": 1.526834653048018e-05,
      "loss": 1.3754,
      "step": 23096
    },
    {
      "epoch": 0.8271527566386735,
      "grad_norm": 1.6051825284957886,
      "learning_rate": 1.526218700141855e-05,
      "loss": 1.41,
      "step": 23097
    },
    {
      "epoch": 0.8271885687682418,
      "grad_norm": 1.4668638706207275,
      "learning_rate": 1.5256028612394913e-05,
      "loss": 1.2962,
      "step": 23098
    },
    {
      "epoch": 0.8272243808978101,
      "grad_norm": 1.2707617282867432,
      "learning_rate": 1.5249871363492107e-05,
      "loss": 1.3112,
      "step": 23099
    },
    {
      "epoch": 0.8272601930273784,
      "grad_norm": 1.6298561096191406,
      "learning_rate": 1.5243715254792912e-05,
      "loss": 1.605,
      "step": 23100
    },
    {
      "epoch": 0.8272960051569467,
      "grad_norm": 1.4971948862075806,
      "learning_rate": 1.5237560286380247e-05,
      "loss": 1.1962,
      "step": 23101
    },
    {
      "epoch": 0.827331817286515,
      "grad_norm": 1.3633439540863037,
      "learning_rate": 1.5231406458336839e-05,
      "loss": 1.4394,
      "step": 23102
    },
    {
      "epoch": 0.8273676294160832,
      "grad_norm": 1.8583279848098755,
      "learning_rate": 1.5225253770745529e-05,
      "loss": 1.6863,
      "step": 23103
    },
    {
      "epoch": 0.8274034415456515,
      "grad_norm": 1.6401985883712769,
      "learning_rate": 1.5219102223689074e-05,
      "loss": 1.5851,
      "step": 23104
    },
    {
      "epoch": 0.8274392536752198,
      "grad_norm": 1.8377301692962646,
      "learning_rate": 1.5212951817250253e-05,
      "loss": 1.2312,
      "step": 23105
    },
    {
      "epoch": 0.8274750658047881,
      "grad_norm": 1.7044795751571655,
      "learning_rate": 1.5206802551511778e-05,
      "loss": 1.8088,
      "step": 23106
    },
    {
      "epoch": 0.8275108779343564,
      "grad_norm": 1.5256201028823853,
      "learning_rate": 1.5200654426556405e-05,
      "loss": 1.4694,
      "step": 23107
    },
    {
      "epoch": 0.8275466900639247,
      "grad_norm": 1.6948037147521973,
      "learning_rate": 1.5194507442466865e-05,
      "loss": 1.3478,
      "step": 23108
    },
    {
      "epoch": 0.8275825021934929,
      "grad_norm": 1.5120340585708618,
      "learning_rate": 1.5188361599325817e-05,
      "loss": 1.3614,
      "step": 23109
    },
    {
      "epoch": 0.8276183143230612,
      "grad_norm": 1.5714203119277954,
      "learning_rate": 1.5182216897215984e-05,
      "loss": 1.6772,
      "step": 23110
    },
    {
      "epoch": 0.8276541264526295,
      "grad_norm": 1.667129397392273,
      "learning_rate": 1.5176073336219965e-05,
      "loss": 1.205,
      "step": 23111
    },
    {
      "epoch": 0.8276899385821977,
      "grad_norm": 1.733232855796814,
      "learning_rate": 1.5169930916420516e-05,
      "loss": 1.3543,
      "step": 23112
    },
    {
      "epoch": 0.8277257507117661,
      "grad_norm": 1.5604608058929443,
      "learning_rate": 1.5163789637900194e-05,
      "loss": 1.5258,
      "step": 23113
    },
    {
      "epoch": 0.8277615628413344,
      "grad_norm": 1.5304710865020752,
      "learning_rate": 1.5157649500741678e-05,
      "loss": 1.4189,
      "step": 23114
    },
    {
      "epoch": 0.8277973749709027,
      "grad_norm": 2.146677017211914,
      "learning_rate": 1.5151510505027499e-05,
      "loss": 1.3662,
      "step": 23115
    },
    {
      "epoch": 0.8278331871004709,
      "grad_norm": 2.4017913341522217,
      "learning_rate": 1.5145372650840361e-05,
      "loss": 1.4622,
      "step": 23116
    },
    {
      "epoch": 0.8278689992300392,
      "grad_norm": 1.358140468597412,
      "learning_rate": 1.5139235938262763e-05,
      "loss": 1.4776,
      "step": 23117
    },
    {
      "epoch": 0.8279048113596075,
      "grad_norm": 2.4922754764556885,
      "learning_rate": 1.513310036737724e-05,
      "loss": 1.7687,
      "step": 23118
    },
    {
      "epoch": 0.8279406234891757,
      "grad_norm": 1.3777458667755127,
      "learning_rate": 1.5126965938266436e-05,
      "loss": 1.504,
      "step": 23119
    },
    {
      "epoch": 0.8279764356187441,
      "grad_norm": 1.5722744464874268,
      "learning_rate": 1.5120832651012795e-05,
      "loss": 1.3428,
      "step": 23120
    },
    {
      "epoch": 0.8280122477483124,
      "grad_norm": 2.008540153503418,
      "learning_rate": 1.5114700505698886e-05,
      "loss": 1.3799,
      "step": 23121
    },
    {
      "epoch": 0.8280480598778807,
      "grad_norm": 1.4039058685302734,
      "learning_rate": 1.5108569502407155e-05,
      "loss": 1.3482,
      "step": 23122
    },
    {
      "epoch": 0.8280838720074489,
      "grad_norm": 1.4145723581314087,
      "learning_rate": 1.5102439641220156e-05,
      "loss": 1.3316,
      "step": 23123
    },
    {
      "epoch": 0.8281196841370172,
      "grad_norm": 1.5959590673446655,
      "learning_rate": 1.5096310922220291e-05,
      "loss": 1.7295,
      "step": 23124
    },
    {
      "epoch": 0.8281554962665855,
      "grad_norm": 1.5081450939178467,
      "learning_rate": 1.5090183345490084e-05,
      "loss": 1.2043,
      "step": 23125
    },
    {
      "epoch": 0.8281913083961537,
      "grad_norm": 1.61405348777771,
      "learning_rate": 1.50840569111119e-05,
      "loss": 1.5296,
      "step": 23126
    },
    {
      "epoch": 0.8282271205257221,
      "grad_norm": 2.1915838718414307,
      "learning_rate": 1.5077931619168196e-05,
      "loss": 1.2124,
      "step": 23127
    },
    {
      "epoch": 0.8282629326552904,
      "grad_norm": 1.5725104808807373,
      "learning_rate": 1.5071807469741406e-05,
      "loss": 1.4422,
      "step": 23128
    },
    {
      "epoch": 0.8282987447848587,
      "grad_norm": 1.9400559663772583,
      "learning_rate": 1.5065684462913853e-05,
      "loss": 1.4408,
      "step": 23129
    },
    {
      "epoch": 0.8283345569144269,
      "grad_norm": 1.7734508514404297,
      "learning_rate": 1.5059562598768007e-05,
      "loss": 1.6698,
      "step": 23130
    },
    {
      "epoch": 0.8283703690439952,
      "grad_norm": 1.7858387231826782,
      "learning_rate": 1.5053441877386154e-05,
      "loss": 1.4474,
      "step": 23131
    },
    {
      "epoch": 0.8284061811735635,
      "grad_norm": 1.9878989458084106,
      "learning_rate": 1.5047322298850685e-05,
      "loss": 1.1392,
      "step": 23132
    },
    {
      "epoch": 0.8284419933031317,
      "grad_norm": 1.5624104738235474,
      "learning_rate": 1.504120386324387e-05,
      "loss": 1.5743,
      "step": 23133
    },
    {
      "epoch": 0.8284778054327001,
      "grad_norm": 1.580743432044983,
      "learning_rate": 1.5035086570648115e-05,
      "loss": 1.2758,
      "step": 23134
    },
    {
      "epoch": 0.8285136175622684,
      "grad_norm": 1.7219854593276978,
      "learning_rate": 1.5028970421145684e-05,
      "loss": 1.2629,
      "step": 23135
    },
    {
      "epoch": 0.8285494296918366,
      "grad_norm": 1.9313101768493652,
      "learning_rate": 1.5022855414818816e-05,
      "loss": 1.3722,
      "step": 23136
    },
    {
      "epoch": 0.8285852418214049,
      "grad_norm": 1.6862903833389282,
      "learning_rate": 1.5016741551749813e-05,
      "loss": 1.4906,
      "step": 23137
    },
    {
      "epoch": 0.8286210539509732,
      "grad_norm": 2.225872039794922,
      "learning_rate": 1.5010628832020945e-05,
      "loss": 1.6626,
      "step": 23138
    },
    {
      "epoch": 0.8286568660805415,
      "grad_norm": 1.539825201034546,
      "learning_rate": 1.5004517255714456e-05,
      "loss": 1.7301,
      "step": 23139
    },
    {
      "epoch": 0.8286926782101097,
      "grad_norm": 1.5023062229156494,
      "learning_rate": 1.4998406822912525e-05,
      "loss": 1.6564,
      "step": 23140
    },
    {
      "epoch": 0.8287284903396781,
      "grad_norm": 2.535398483276367,
      "learning_rate": 1.4992297533697387e-05,
      "loss": 1.4908,
      "step": 23141
    },
    {
      "epoch": 0.8287643024692464,
      "grad_norm": 1.7443580627441406,
      "learning_rate": 1.4986189388151229e-05,
      "loss": 1.3793,
      "step": 23142
    },
    {
      "epoch": 0.8288001145988146,
      "grad_norm": 1.728061318397522,
      "learning_rate": 1.4980082386356264e-05,
      "loss": 1.246,
      "step": 23143
    },
    {
      "epoch": 0.8288359267283829,
      "grad_norm": 1.9140900373458862,
      "learning_rate": 1.4973976528394596e-05,
      "loss": 1.2107,
      "step": 23144
    },
    {
      "epoch": 0.8288717388579512,
      "grad_norm": 2.0008881092071533,
      "learning_rate": 1.4967871814348399e-05,
      "loss": 1.4527,
      "step": 23145
    },
    {
      "epoch": 0.8289075509875194,
      "grad_norm": 1.3603802919387817,
      "learning_rate": 1.4961768244299823e-05,
      "loss": 1.2612,
      "step": 23146
    },
    {
      "epoch": 0.8289433631170877,
      "grad_norm": 1.8144201040267944,
      "learning_rate": 1.4955665818330944e-05,
      "loss": 1.5788,
      "step": 23147
    },
    {
      "epoch": 0.8289791752466561,
      "grad_norm": 1.1382317543029785,
      "learning_rate": 1.4949564536523874e-05,
      "loss": 1.1458,
      "step": 23148
    },
    {
      "epoch": 0.8290149873762244,
      "grad_norm": 1.77524995803833,
      "learning_rate": 1.4943464398960716e-05,
      "loss": 1.6568,
      "step": 23149
    },
    {
      "epoch": 0.8290507995057926,
      "grad_norm": 1.7425357103347778,
      "learning_rate": 1.4937365405723547e-05,
      "loss": 1.3626,
      "step": 23150
    },
    {
      "epoch": 0.8290866116353609,
      "grad_norm": 1.9161444902420044,
      "learning_rate": 1.493126755689439e-05,
      "loss": 1.4522,
      "step": 23151
    },
    {
      "epoch": 0.8291224237649292,
      "grad_norm": 1.7542110681533813,
      "learning_rate": 1.4925170852555282e-05,
      "loss": 1.3236,
      "step": 23152
    },
    {
      "epoch": 0.8291582358944974,
      "grad_norm": 1.5818971395492554,
      "learning_rate": 1.4919075292788298e-05,
      "loss": 1.3392,
      "step": 23153
    },
    {
      "epoch": 0.8291940480240657,
      "grad_norm": 1.3822681903839111,
      "learning_rate": 1.4912980877675387e-05,
      "loss": 1.3119,
      "step": 23154
    },
    {
      "epoch": 0.8292298601536341,
      "grad_norm": 1.6355019807815552,
      "learning_rate": 1.4906887607298548e-05,
      "loss": 1.9485,
      "step": 23155
    },
    {
      "epoch": 0.8292656722832024,
      "grad_norm": 1.661881685256958,
      "learning_rate": 1.4900795481739793e-05,
      "loss": 1.6326,
      "step": 23156
    },
    {
      "epoch": 0.8293014844127706,
      "grad_norm": 1.375794768333435,
      "learning_rate": 1.4894704501081069e-05,
      "loss": 1.4931,
      "step": 23157
    },
    {
      "epoch": 0.8293372965423389,
      "grad_norm": 1.861142873764038,
      "learning_rate": 1.488861466540431e-05,
      "loss": 1.466,
      "step": 23158
    },
    {
      "epoch": 0.8293731086719072,
      "grad_norm": 1.3190573453903198,
      "learning_rate": 1.488252597479145e-05,
      "loss": 1.3289,
      "step": 23159
    },
    {
      "epoch": 0.8294089208014754,
      "grad_norm": 1.687562108039856,
      "learning_rate": 1.4876438429324414e-05,
      "loss": 1.1553,
      "step": 23160
    },
    {
      "epoch": 0.8294447329310437,
      "grad_norm": 1.8980015516281128,
      "learning_rate": 1.487035202908511e-05,
      "loss": 1.4203,
      "step": 23161
    },
    {
      "epoch": 0.8294805450606121,
      "grad_norm": 2.008462905883789,
      "learning_rate": 1.4864266774155389e-05,
      "loss": 1.3749,
      "step": 23162
    },
    {
      "epoch": 0.8295163571901804,
      "grad_norm": 1.576193928718567,
      "learning_rate": 1.4858182664617148e-05,
      "loss": 1.2269,
      "step": 23163
    },
    {
      "epoch": 0.8295521693197486,
      "grad_norm": 1.642438530921936,
      "learning_rate": 1.4852099700552259e-05,
      "loss": 1.38,
      "step": 23164
    },
    {
      "epoch": 0.8295879814493169,
      "grad_norm": 1.7365227937698364,
      "learning_rate": 1.4846017882042506e-05,
      "loss": 1.4031,
      "step": 23165
    },
    {
      "epoch": 0.8296237935788852,
      "grad_norm": 1.6109135150909424,
      "learning_rate": 1.4839937209169741e-05,
      "loss": 1.2866,
      "step": 23166
    },
    {
      "epoch": 0.8296596057084534,
      "grad_norm": 1.110206127166748,
      "learning_rate": 1.4833857682015773e-05,
      "loss": 1.3672,
      "step": 23167
    },
    {
      "epoch": 0.8296954178380217,
      "grad_norm": 1.5050301551818848,
      "learning_rate": 1.4827779300662425e-05,
      "loss": 1.4046,
      "step": 23168
    },
    {
      "epoch": 0.8297312299675901,
      "grad_norm": 1.5726349353790283,
      "learning_rate": 1.4821702065191413e-05,
      "loss": 1.3428,
      "step": 23169
    },
    {
      "epoch": 0.8297670420971583,
      "grad_norm": 1.4734699726104736,
      "learning_rate": 1.4815625975684522e-05,
      "loss": 1.3572,
      "step": 23170
    },
    {
      "epoch": 0.8298028542267266,
      "grad_norm": 1.7343571186065674,
      "learning_rate": 1.4809551032223534e-05,
      "loss": 1.4396,
      "step": 23171
    },
    {
      "epoch": 0.8298386663562949,
      "grad_norm": 1.371505618095398,
      "learning_rate": 1.480347723489013e-05,
      "loss": 1.4953,
      "step": 23172
    },
    {
      "epoch": 0.8298744784858632,
      "grad_norm": 1.4669415950775146,
      "learning_rate": 1.4797404583766028e-05,
      "loss": 0.9984,
      "step": 23173
    },
    {
      "epoch": 0.8299102906154314,
      "grad_norm": 1.7243847846984863,
      "learning_rate": 1.4791333078932956e-05,
      "loss": 1.4721,
      "step": 23174
    },
    {
      "epoch": 0.8299461027449997,
      "grad_norm": 1.569770097732544,
      "learning_rate": 1.4785262720472615e-05,
      "loss": 1.4345,
      "step": 23175
    },
    {
      "epoch": 0.829981914874568,
      "grad_norm": 1.4598844051361084,
      "learning_rate": 1.4779193508466604e-05,
      "loss": 1.3252,
      "step": 23176
    },
    {
      "epoch": 0.8300177270041363,
      "grad_norm": 1.655402660369873,
      "learning_rate": 1.4773125442996626e-05,
      "loss": 1.622,
      "step": 23177
    },
    {
      "epoch": 0.8300535391337046,
      "grad_norm": 1.8673683404922485,
      "learning_rate": 1.4767058524144318e-05,
      "loss": 1.3777,
      "step": 23178
    },
    {
      "epoch": 0.8300893512632729,
      "grad_norm": 1.8644596338272095,
      "learning_rate": 1.476099275199131e-05,
      "loss": 1.4903,
      "step": 23179
    },
    {
      "epoch": 0.8301251633928411,
      "grad_norm": 1.84309720993042,
      "learning_rate": 1.4754928126619172e-05,
      "loss": 1.3268,
      "step": 23180
    },
    {
      "epoch": 0.8301609755224094,
      "grad_norm": 1.785307765007019,
      "learning_rate": 1.4748864648109518e-05,
      "loss": 1.3626,
      "step": 23181
    },
    {
      "epoch": 0.8301967876519777,
      "grad_norm": 1.6583218574523926,
      "learning_rate": 1.4742802316543947e-05,
      "loss": 1.3632,
      "step": 23182
    },
    {
      "epoch": 0.830232599781546,
      "grad_norm": 2.204003095626831,
      "learning_rate": 1.4736741132003984e-05,
      "loss": 1.4275,
      "step": 23183
    },
    {
      "epoch": 0.8302684119111143,
      "grad_norm": 1.5974963903427124,
      "learning_rate": 1.4730681094571175e-05,
      "loss": 1.5503,
      "step": 23184
    },
    {
      "epoch": 0.8303042240406826,
      "grad_norm": 1.756859540939331,
      "learning_rate": 1.4724622204327066e-05,
      "loss": 1.56,
      "step": 23185
    },
    {
      "epoch": 0.8303400361702509,
      "grad_norm": 1.429917335510254,
      "learning_rate": 1.471856446135319e-05,
      "loss": 1.5715,
      "step": 23186
    },
    {
      "epoch": 0.8303758482998191,
      "grad_norm": 1.7499074935913086,
      "learning_rate": 1.4712507865730996e-05,
      "loss": 1.5858,
      "step": 23187
    },
    {
      "epoch": 0.8304116604293874,
      "grad_norm": 1.723129153251648,
      "learning_rate": 1.4706452417542006e-05,
      "loss": 1.4264,
      "step": 23188
    },
    {
      "epoch": 0.8304474725589557,
      "grad_norm": 1.859376311302185,
      "learning_rate": 1.4700398116867697e-05,
      "loss": 1.3564,
      "step": 23189
    },
    {
      "epoch": 0.830483284688524,
      "grad_norm": 1.5590065717697144,
      "learning_rate": 1.4694344963789474e-05,
      "loss": 1.5782,
      "step": 23190
    },
    {
      "epoch": 0.8305190968180923,
      "grad_norm": 1.856689214706421,
      "learning_rate": 1.4688292958388816e-05,
      "loss": 1.1994,
      "step": 23191
    },
    {
      "epoch": 0.8305549089476606,
      "grad_norm": 1.3962154388427734,
      "learning_rate": 1.4682242100747123e-05,
      "loss": 1.4828,
      "step": 23192
    },
    {
      "epoch": 0.8305907210772289,
      "grad_norm": 1.6056325435638428,
      "learning_rate": 1.467619239094583e-05,
      "loss": 1.4341,
      "step": 23193
    },
    {
      "epoch": 0.8306265332067971,
      "grad_norm": 2.161923885345459,
      "learning_rate": 1.4670143829066296e-05,
      "loss": 1.4834,
      "step": 23194
    },
    {
      "epoch": 0.8306623453363654,
      "grad_norm": 1.8729352951049805,
      "learning_rate": 1.4664096415189899e-05,
      "loss": 1.5906,
      "step": 23195
    },
    {
      "epoch": 0.8306981574659337,
      "grad_norm": 1.8302282094955444,
      "learning_rate": 1.465805014939804e-05,
      "loss": 1.3676,
      "step": 23196
    },
    {
      "epoch": 0.8307339695955019,
      "grad_norm": 1.3889516592025757,
      "learning_rate": 1.465200503177201e-05,
      "loss": 1.4782,
      "step": 23197
    },
    {
      "epoch": 0.8307697817250703,
      "grad_norm": 2.007490396499634,
      "learning_rate": 1.4645961062393177e-05,
      "loss": 1.4347,
      "step": 23198
    },
    {
      "epoch": 0.8308055938546386,
      "grad_norm": 1.6827696561813354,
      "learning_rate": 1.4639918241342798e-05,
      "loss": 1.4996,
      "step": 23199
    },
    {
      "epoch": 0.8308414059842069,
      "grad_norm": 1.323423981666565,
      "learning_rate": 1.4633876568702254e-05,
      "loss": 1.1966,
      "step": 23200
    },
    {
      "epoch": 0.8308772181137751,
      "grad_norm": 1.5560758113861084,
      "learning_rate": 1.4627836044552767e-05,
      "loss": 1.4059,
      "step": 23201
    },
    {
      "epoch": 0.8309130302433434,
      "grad_norm": 2.4446370601654053,
      "learning_rate": 1.462179666897563e-05,
      "loss": 1.4578,
      "step": 23202
    },
    {
      "epoch": 0.8309488423729117,
      "grad_norm": 1.423201322555542,
      "learning_rate": 1.4615758442052085e-05,
      "loss": 1.1859,
      "step": 23203
    },
    {
      "epoch": 0.8309846545024799,
      "grad_norm": 1.891046166419983,
      "learning_rate": 1.4609721363863393e-05,
      "loss": 1.3277,
      "step": 23204
    },
    {
      "epoch": 0.8310204666320483,
      "grad_norm": 2.2019035816192627,
      "learning_rate": 1.4603685434490756e-05,
      "loss": 1.154,
      "step": 23205
    },
    {
      "epoch": 0.8310562787616166,
      "grad_norm": 1.9172054529190063,
      "learning_rate": 1.4597650654015327e-05,
      "loss": 1.2394,
      "step": 23206
    },
    {
      "epoch": 0.8310920908911849,
      "grad_norm": 1.5021252632141113,
      "learning_rate": 1.45916170225184e-05,
      "loss": 1.5499,
      "step": 23207
    },
    {
      "epoch": 0.8311279030207531,
      "grad_norm": 2.5014986991882324,
      "learning_rate": 1.4585584540081066e-05,
      "loss": 1.4685,
      "step": 23208
    },
    {
      "epoch": 0.8311637151503214,
      "grad_norm": 1.3708525896072388,
      "learning_rate": 1.4579553206784546e-05,
      "loss": 1.3963,
      "step": 23209
    },
    {
      "epoch": 0.8311995272798897,
      "grad_norm": 1.7175711393356323,
      "learning_rate": 1.45735230227099e-05,
      "loss": 1.5637,
      "step": 23210
    },
    {
      "epoch": 0.8312353394094579,
      "grad_norm": 1.8059896230697632,
      "learning_rate": 1.4567493987938364e-05,
      "loss": 1.3623,
      "step": 23211
    },
    {
      "epoch": 0.8312711515390263,
      "grad_norm": 1.5208191871643066,
      "learning_rate": 1.456146610255097e-05,
      "loss": 1.4858,
      "step": 23212
    },
    {
      "epoch": 0.8313069636685946,
      "grad_norm": 2.4135520458221436,
      "learning_rate": 1.4555439366628843e-05,
      "loss": 1.4645,
      "step": 23213
    },
    {
      "epoch": 0.8313427757981628,
      "grad_norm": 1.7877299785614014,
      "learning_rate": 1.4549413780253085e-05,
      "loss": 1.4712,
      "step": 23214
    },
    {
      "epoch": 0.8313785879277311,
      "grad_norm": 1.7445772886276245,
      "learning_rate": 1.454338934350472e-05,
      "loss": 1.6327,
      "step": 23215
    },
    {
      "epoch": 0.8314144000572994,
      "grad_norm": 1.4530085325241089,
      "learning_rate": 1.453736605646484e-05,
      "loss": 1.4826,
      "step": 23216
    },
    {
      "epoch": 0.8314502121868677,
      "grad_norm": 1.793205738067627,
      "learning_rate": 1.4531343919214414e-05,
      "loss": 1.3664,
      "step": 23217
    },
    {
      "epoch": 0.8314860243164359,
      "grad_norm": 1.2321491241455078,
      "learning_rate": 1.4525322931834562e-05,
      "loss": 1.1971,
      "step": 23218
    },
    {
      "epoch": 0.8315218364460043,
      "grad_norm": 1.8643945455551147,
      "learning_rate": 1.4519303094406211e-05,
      "loss": 1.5384,
      "step": 23219
    },
    {
      "epoch": 0.8315576485755726,
      "grad_norm": 1.8356062173843384,
      "learning_rate": 1.4513284407010385e-05,
      "loss": 1.5823,
      "step": 23220
    },
    {
      "epoch": 0.8315934607051408,
      "grad_norm": 1.8508111238479614,
      "learning_rate": 1.450726686972802e-05,
      "loss": 1.5022,
      "step": 23221
    },
    {
      "epoch": 0.8316292728347091,
      "grad_norm": 1.2595747709274292,
      "learning_rate": 1.4501250482640139e-05,
      "loss": 1.4594,
      "step": 23222
    },
    {
      "epoch": 0.8316650849642774,
      "grad_norm": 2.1554951667785645,
      "learning_rate": 1.4495235245827642e-05,
      "loss": 1.2048,
      "step": 23223
    },
    {
      "epoch": 0.8317008970938456,
      "grad_norm": 4.19765567779541,
      "learning_rate": 1.4489221159371447e-05,
      "loss": 1.5112,
      "step": 23224
    },
    {
      "epoch": 0.8317367092234139,
      "grad_norm": 1.5159118175506592,
      "learning_rate": 1.4483208223352474e-05,
      "loss": 1.6076,
      "step": 23225
    },
    {
      "epoch": 0.8317725213529823,
      "grad_norm": 1.4949398040771484,
      "learning_rate": 1.4477196437851625e-05,
      "loss": 1.5265,
      "step": 23226
    },
    {
      "epoch": 0.8318083334825506,
      "grad_norm": 1.6657307147979736,
      "learning_rate": 1.4471185802949816e-05,
      "loss": 1.501,
      "step": 23227
    },
    {
      "epoch": 0.8318441456121188,
      "grad_norm": 1.5736427307128906,
      "learning_rate": 1.4465176318727825e-05,
      "loss": 1.6235,
      "step": 23228
    },
    {
      "epoch": 0.8318799577416871,
      "grad_norm": 1.6373560428619385,
      "learning_rate": 1.4459167985266597e-05,
      "loss": 1.6095,
      "step": 23229
    },
    {
      "epoch": 0.8319157698712554,
      "grad_norm": 1.5352661609649658,
      "learning_rate": 1.4453160802646903e-05,
      "loss": 1.4128,
      "step": 23230
    },
    {
      "epoch": 0.8319515820008236,
      "grad_norm": 1.6256799697875977,
      "learning_rate": 1.444715477094961e-05,
      "loss": 1.4257,
      "step": 23231
    },
    {
      "epoch": 0.8319873941303919,
      "grad_norm": 1.462746262550354,
      "learning_rate": 1.4441149890255467e-05,
      "loss": 1.2893,
      "step": 23232
    },
    {
      "epoch": 0.8320232062599603,
      "grad_norm": 1.7766444683074951,
      "learning_rate": 1.4435146160645285e-05,
      "loss": 1.4783,
      "step": 23233
    },
    {
      "epoch": 0.8320590183895286,
      "grad_norm": 1.7726175785064697,
      "learning_rate": 1.4429143582199866e-05,
      "loss": 1.4622,
      "step": 23234
    },
    {
      "epoch": 0.8320948305190968,
      "grad_norm": 1.3048326969146729,
      "learning_rate": 1.4423142154999925e-05,
      "loss": 1.1044,
      "step": 23235
    },
    {
      "epoch": 0.8321306426486651,
      "grad_norm": 1.4011657238006592,
      "learning_rate": 1.4417141879126218e-05,
      "loss": 1.5252,
      "step": 23236
    },
    {
      "epoch": 0.8321664547782334,
      "grad_norm": 1.5536365509033203,
      "learning_rate": 1.4411142754659468e-05,
      "loss": 1.2663,
      "step": 23237
    },
    {
      "epoch": 0.8322022669078016,
      "grad_norm": 1.8643995523452759,
      "learning_rate": 1.4405144781680424e-05,
      "loss": 1.2179,
      "step": 23238
    },
    {
      "epoch": 0.8322380790373699,
      "grad_norm": 1.3420333862304688,
      "learning_rate": 1.4399147960269688e-05,
      "loss": 1.5695,
      "step": 23239
    },
    {
      "epoch": 0.8322738911669383,
      "grad_norm": 1.519026279449463,
      "learning_rate": 1.439315229050805e-05,
      "loss": 1.4541,
      "step": 23240
    },
    {
      "epoch": 0.8323097032965066,
      "grad_norm": 1.8381019830703735,
      "learning_rate": 1.4387157772476134e-05,
      "loss": 1.4799,
      "step": 23241
    },
    {
      "epoch": 0.8323455154260748,
      "grad_norm": 1.4542158842086792,
      "learning_rate": 1.4381164406254544e-05,
      "loss": 1.4863,
      "step": 23242
    },
    {
      "epoch": 0.8323813275556431,
      "grad_norm": 1.423120141029358,
      "learning_rate": 1.4375172191923947e-05,
      "loss": 1.3069,
      "step": 23243
    },
    {
      "epoch": 0.8324171396852114,
      "grad_norm": 1.5357532501220703,
      "learning_rate": 1.4369181129564957e-05,
      "loss": 1.473,
      "step": 23244
    },
    {
      "epoch": 0.8324529518147796,
      "grad_norm": 1.6310615539550781,
      "learning_rate": 1.4363191219258209e-05,
      "loss": 1.4536,
      "step": 23245
    },
    {
      "epoch": 0.8324887639443479,
      "grad_norm": 1.673390507698059,
      "learning_rate": 1.4357202461084229e-05,
      "loss": 1.3385,
      "step": 23246
    },
    {
      "epoch": 0.8325245760739163,
      "grad_norm": 2.130810022354126,
      "learning_rate": 1.4351214855123629e-05,
      "loss": 1.3291,
      "step": 23247
    },
    {
      "epoch": 0.8325603882034845,
      "grad_norm": 1.4513099193572998,
      "learning_rate": 1.4345228401456945e-05,
      "loss": 1.2643,
      "step": 23248
    },
    {
      "epoch": 0.8325962003330528,
      "grad_norm": 1.4877856969833374,
      "learning_rate": 1.4339243100164757e-05,
      "loss": 1.6895,
      "step": 23249
    },
    {
      "epoch": 0.8326320124626211,
      "grad_norm": 1.3141956329345703,
      "learning_rate": 1.4333258951327534e-05,
      "loss": 1.3533,
      "step": 23250
    },
    {
      "epoch": 0.8326678245921894,
      "grad_norm": 1.7010196447372437,
      "learning_rate": 1.4327275955025798e-05,
      "loss": 1.2068,
      "step": 23251
    },
    {
      "epoch": 0.8327036367217576,
      "grad_norm": 1.50443434715271,
      "learning_rate": 1.4321294111340089e-05,
      "loss": 1.283,
      "step": 23252
    },
    {
      "epoch": 0.8327394488513259,
      "grad_norm": 1.7006819248199463,
      "learning_rate": 1.4315313420350829e-05,
      "loss": 1.3811,
      "step": 23253
    },
    {
      "epoch": 0.8327752609808943,
      "grad_norm": 1.890299916267395,
      "learning_rate": 1.4309333882138488e-05,
      "loss": 1.3671,
      "step": 23254
    },
    {
      "epoch": 0.8328110731104625,
      "grad_norm": 1.7135711908340454,
      "learning_rate": 1.4303355496783544e-05,
      "loss": 1.3083,
      "step": 23255
    },
    {
      "epoch": 0.8328468852400308,
      "grad_norm": 1.436191201210022,
      "learning_rate": 1.4297378264366423e-05,
      "loss": 1.5384,
      "step": 23256
    },
    {
      "epoch": 0.8328826973695991,
      "grad_norm": 1.4963761568069458,
      "learning_rate": 1.4291402184967507e-05,
      "loss": 1.4136,
      "step": 23257
    },
    {
      "epoch": 0.8329185094991673,
      "grad_norm": 2.164684295654297,
      "learning_rate": 1.4285427258667217e-05,
      "loss": 1.5224,
      "step": 23258
    },
    {
      "epoch": 0.8329543216287356,
      "grad_norm": 1.8645986318588257,
      "learning_rate": 1.4279453485545968e-05,
      "loss": 1.9139,
      "step": 23259
    },
    {
      "epoch": 0.8329901337583039,
      "grad_norm": 1.6702115535736084,
      "learning_rate": 1.4273480865684074e-05,
      "loss": 1.3873,
      "step": 23260
    },
    {
      "epoch": 0.8330259458878723,
      "grad_norm": 1.5879836082458496,
      "learning_rate": 1.4267509399161916e-05,
      "loss": 1.2909,
      "step": 23261
    },
    {
      "epoch": 0.8330617580174405,
      "grad_norm": 1.5265380144119263,
      "learning_rate": 1.4261539086059839e-05,
      "loss": 1.5102,
      "step": 23262
    },
    {
      "epoch": 0.8330975701470088,
      "grad_norm": 1.9467709064483643,
      "learning_rate": 1.4255569926458168e-05,
      "loss": 1.4075,
      "step": 23263
    },
    {
      "epoch": 0.8331333822765771,
      "grad_norm": 1.7792093753814697,
      "learning_rate": 1.4249601920437194e-05,
      "loss": 1.3639,
      "step": 23264
    },
    {
      "epoch": 0.8331691944061453,
      "grad_norm": 1.6893055438995361,
      "learning_rate": 1.42436350680772e-05,
      "loss": 1.7997,
      "step": 23265
    },
    {
      "epoch": 0.8332050065357136,
      "grad_norm": 1.7191356420516968,
      "learning_rate": 1.4237669369458495e-05,
      "loss": 1.5456,
      "step": 23266
    },
    {
      "epoch": 0.8332408186652819,
      "grad_norm": 2.7596893310546875,
      "learning_rate": 1.4231704824661329e-05,
      "loss": 1.5299,
      "step": 23267
    },
    {
      "epoch": 0.8332766307948503,
      "grad_norm": 1.2917113304138184,
      "learning_rate": 1.4225741433765927e-05,
      "loss": 1.2855,
      "step": 23268
    },
    {
      "epoch": 0.8333124429244185,
      "grad_norm": 1.6655503511428833,
      "learning_rate": 1.4219779196852534e-05,
      "loss": 1.5177,
      "step": 23269
    },
    {
      "epoch": 0.8333482550539868,
      "grad_norm": 1.3067896366119385,
      "learning_rate": 1.4213818114001387e-05,
      "loss": 1.131,
      "step": 23270
    },
    {
      "epoch": 0.8333840671835551,
      "grad_norm": 1.9188344478607178,
      "learning_rate": 1.4207858185292643e-05,
      "loss": 1.4093,
      "step": 23271
    },
    {
      "epoch": 0.8334198793131233,
      "grad_norm": 1.8857660293579102,
      "learning_rate": 1.4201899410806496e-05,
      "loss": 1.612,
      "step": 23272
    },
    {
      "epoch": 0.8334556914426916,
      "grad_norm": 1.7722171545028687,
      "learning_rate": 1.4195941790623124e-05,
      "loss": 1.6784,
      "step": 23273
    },
    {
      "epoch": 0.8334915035722599,
      "grad_norm": 1.8155393600463867,
      "learning_rate": 1.4189985324822697e-05,
      "loss": 1.6842,
      "step": 23274
    },
    {
      "epoch": 0.8335273157018283,
      "grad_norm": 1.6256564855575562,
      "learning_rate": 1.4184030013485305e-05,
      "loss": 1.5721,
      "step": 23275
    },
    {
      "epoch": 0.8335631278313965,
      "grad_norm": 1.669029712677002,
      "learning_rate": 1.4178075856691097e-05,
      "loss": 1.4226,
      "step": 23276
    },
    {
      "epoch": 0.8335989399609648,
      "grad_norm": 1.8676083087921143,
      "learning_rate": 1.4172122854520198e-05,
      "loss": 1.7837,
      "step": 23277
    },
    {
      "epoch": 0.8336347520905331,
      "grad_norm": 1.5746208429336548,
      "learning_rate": 1.4166171007052653e-05,
      "loss": 1.2264,
      "step": 23278
    },
    {
      "epoch": 0.8336705642201013,
      "grad_norm": 2.2983994483947754,
      "learning_rate": 1.4160220314368555e-05,
      "loss": 1.6691,
      "step": 23279
    },
    {
      "epoch": 0.8337063763496696,
      "grad_norm": 1.4600412845611572,
      "learning_rate": 1.4154270776547974e-05,
      "loss": 1.1893,
      "step": 23280
    },
    {
      "epoch": 0.8337421884792379,
      "grad_norm": 3.8261477947235107,
      "learning_rate": 1.4148322393670976e-05,
      "loss": 1.55,
      "step": 23281
    },
    {
      "epoch": 0.8337780006088062,
      "grad_norm": 1.9612890481948853,
      "learning_rate": 1.4142375165817523e-05,
      "loss": 1.4173,
      "step": 23282
    },
    {
      "epoch": 0.8338138127383745,
      "grad_norm": 2.1106479167938232,
      "learning_rate": 1.413642909306767e-05,
      "loss": 1.5264,
      "step": 23283
    },
    {
      "epoch": 0.8338496248679428,
      "grad_norm": 1.3178765773773193,
      "learning_rate": 1.4130484175501435e-05,
      "loss": 1.2424,
      "step": 23284
    },
    {
      "epoch": 0.833885436997511,
      "grad_norm": 2.719381332397461,
      "learning_rate": 1.412454041319874e-05,
      "loss": 1.6399,
      "step": 23285
    },
    {
      "epoch": 0.8339212491270793,
      "grad_norm": 2.3354897499084473,
      "learning_rate": 1.4118597806239585e-05,
      "loss": 1.4628,
      "step": 23286
    },
    {
      "epoch": 0.8339570612566476,
      "grad_norm": 1.822835922241211,
      "learning_rate": 1.4112656354703924e-05,
      "loss": 1.2929,
      "step": 23287
    },
    {
      "epoch": 0.8339928733862159,
      "grad_norm": 1.2769025564193726,
      "learning_rate": 1.41067160586717e-05,
      "loss": 1.4113,
      "step": 23288
    },
    {
      "epoch": 0.8340286855157842,
      "grad_norm": 1.8172305822372437,
      "learning_rate": 1.4100776918222802e-05,
      "loss": 1.2328,
      "step": 23289
    },
    {
      "epoch": 0.8340644976453525,
      "grad_norm": 1.472387433052063,
      "learning_rate": 1.4094838933437138e-05,
      "loss": 1.4583,
      "step": 23290
    },
    {
      "epoch": 0.8341003097749208,
      "grad_norm": 1.9044989347457886,
      "learning_rate": 1.4088902104394607e-05,
      "loss": 1.7,
      "step": 23291
    },
    {
      "epoch": 0.834136121904489,
      "grad_norm": 2.0929458141326904,
      "learning_rate": 1.4082966431175116e-05,
      "loss": 1.1544,
      "step": 23292
    },
    {
      "epoch": 0.8341719340340573,
      "grad_norm": 1.484946846961975,
      "learning_rate": 1.4077031913858474e-05,
      "loss": 1.5744,
      "step": 23293
    },
    {
      "epoch": 0.8342077461636256,
      "grad_norm": 1.6925033330917358,
      "learning_rate": 1.4071098552524497e-05,
      "loss": 1.4153,
      "step": 23294
    },
    {
      "epoch": 0.8342435582931939,
      "grad_norm": 1.6346595287322998,
      "learning_rate": 1.4065166347253089e-05,
      "loss": 1.2867,
      "step": 23295
    },
    {
      "epoch": 0.8342793704227622,
      "grad_norm": 1.8432413339614868,
      "learning_rate": 1.4059235298124006e-05,
      "loss": 1.3611,
      "step": 23296
    },
    {
      "epoch": 0.8343151825523305,
      "grad_norm": 1.546345591545105,
      "learning_rate": 1.4053305405217044e-05,
      "loss": 1.531,
      "step": 23297
    },
    {
      "epoch": 0.8343509946818988,
      "grad_norm": 1.4019263982772827,
      "learning_rate": 1.4047376668612e-05,
      "loss": 1.3993,
      "step": 23298
    },
    {
      "epoch": 0.834386806811467,
      "grad_norm": 1.7556198835372925,
      "learning_rate": 1.4041449088388659e-05,
      "loss": 1.0783,
      "step": 23299
    },
    {
      "epoch": 0.8344226189410353,
      "grad_norm": 1.5568428039550781,
      "learning_rate": 1.4035522664626721e-05,
      "loss": 1.4352,
      "step": 23300
    },
    {
      "epoch": 0.8344584310706036,
      "grad_norm": 3.3579607009887695,
      "learning_rate": 1.4029597397405925e-05,
      "loss": 1.6769,
      "step": 23301
    },
    {
      "epoch": 0.8344942432001718,
      "grad_norm": 1.5644280910491943,
      "learning_rate": 1.4023673286806039e-05,
      "loss": 1.089,
      "step": 23302
    },
    {
      "epoch": 0.8345300553297402,
      "grad_norm": 1.8802729845046997,
      "learning_rate": 1.4017750332906698e-05,
      "loss": 1.7205,
      "step": 23303
    },
    {
      "epoch": 0.8345658674593085,
      "grad_norm": 1.711993932723999,
      "learning_rate": 1.4011828535787642e-05,
      "loss": 1.5439,
      "step": 23304
    },
    {
      "epoch": 0.8346016795888768,
      "grad_norm": 1.9384276866912842,
      "learning_rate": 1.400590789552847e-05,
      "loss": 1.3889,
      "step": 23305
    },
    {
      "epoch": 0.834637491718445,
      "grad_norm": 2.598912000656128,
      "learning_rate": 1.3999988412208931e-05,
      "loss": 1.5202,
      "step": 23306
    },
    {
      "epoch": 0.8346733038480133,
      "grad_norm": 1.5115162134170532,
      "learning_rate": 1.3994070085908596e-05,
      "loss": 1.5137,
      "step": 23307
    },
    {
      "epoch": 0.8347091159775816,
      "grad_norm": 1.8529272079467773,
      "learning_rate": 1.3988152916707121e-05,
      "loss": 1.2628,
      "step": 23308
    },
    {
      "epoch": 0.8347449281071498,
      "grad_norm": 1.7130944728851318,
      "learning_rate": 1.3982236904684064e-05,
      "loss": 1.2679,
      "step": 23309
    },
    {
      "epoch": 0.8347807402367182,
      "grad_norm": 1.4939451217651367,
      "learning_rate": 1.3976322049919088e-05,
      "loss": 1.5153,
      "step": 23310
    },
    {
      "epoch": 0.8348165523662865,
      "grad_norm": 1.5681949853897095,
      "learning_rate": 1.3970408352491749e-05,
      "loss": 1.4088,
      "step": 23311
    },
    {
      "epoch": 0.8348523644958548,
      "grad_norm": 1.7648473978042603,
      "learning_rate": 1.3964495812481548e-05,
      "loss": 1.3955,
      "step": 23312
    },
    {
      "epoch": 0.834888176625423,
      "grad_norm": 1.4568158388137817,
      "learning_rate": 1.3958584429968124e-05,
      "loss": 1.4164,
      "step": 23313
    },
    {
      "epoch": 0.8349239887549913,
      "grad_norm": 1.6647497415542603,
      "learning_rate": 1.3952674205030935e-05,
      "loss": 1.1866,
      "step": 23314
    },
    {
      "epoch": 0.8349598008845596,
      "grad_norm": 1.7173484563827515,
      "learning_rate": 1.394676513774954e-05,
      "loss": 1.5996,
      "step": 23315
    },
    {
      "epoch": 0.8349956130141278,
      "grad_norm": 1.6371338367462158,
      "learning_rate": 1.3940857228203386e-05,
      "loss": 1.4188,
      "step": 23316
    },
    {
      "epoch": 0.8350314251436962,
      "grad_norm": 1.3010733127593994,
      "learning_rate": 1.393495047647202e-05,
      "loss": 1.516,
      "step": 23317
    },
    {
      "epoch": 0.8350672372732645,
      "grad_norm": 1.5296361446380615,
      "learning_rate": 1.3929044882634867e-05,
      "loss": 1.4748,
      "step": 23318
    },
    {
      "epoch": 0.8351030494028328,
      "grad_norm": 1.7259113788604736,
      "learning_rate": 1.3923140446771409e-05,
      "loss": 1.8208,
      "step": 23319
    },
    {
      "epoch": 0.835138861532401,
      "grad_norm": 1.6335358619689941,
      "learning_rate": 1.3917237168961051e-05,
      "loss": 1.3154,
      "step": 23320
    },
    {
      "epoch": 0.8351746736619693,
      "grad_norm": 2.0975825786590576,
      "learning_rate": 1.3911335049283225e-05,
      "loss": 1.6693,
      "step": 23321
    },
    {
      "epoch": 0.8352104857915376,
      "grad_norm": 1.558631181716919,
      "learning_rate": 1.390543408781736e-05,
      "loss": 1.5984,
      "step": 23322
    },
    {
      "epoch": 0.8352462979211058,
      "grad_norm": 1.3782986402511597,
      "learning_rate": 1.3899534284642779e-05,
      "loss": 1.1026,
      "step": 23323
    },
    {
      "epoch": 0.8352821100506742,
      "grad_norm": 1.7837612628936768,
      "learning_rate": 1.3893635639838942e-05,
      "loss": 1.2879,
      "step": 23324
    },
    {
      "epoch": 0.8353179221802425,
      "grad_norm": 1.4235824346542358,
      "learning_rate": 1.3887738153485153e-05,
      "loss": 1.5469,
      "step": 23325
    },
    {
      "epoch": 0.8353537343098107,
      "grad_norm": 1.4114874601364136,
      "learning_rate": 1.388184182566079e-05,
      "loss": 1.523,
      "step": 23326
    },
    {
      "epoch": 0.835389546439379,
      "grad_norm": 1.7939683198928833,
      "learning_rate": 1.3875946656445126e-05,
      "loss": 1.5065,
      "step": 23327
    },
    {
      "epoch": 0.8354253585689473,
      "grad_norm": 1.2166051864624023,
      "learning_rate": 1.3870052645917542e-05,
      "loss": 1.257,
      "step": 23328
    },
    {
      "epoch": 0.8354611706985156,
      "grad_norm": 2.7868435382843018,
      "learning_rate": 1.3864159794157305e-05,
      "loss": 0.9794,
      "step": 23329
    },
    {
      "epoch": 0.8354969828280838,
      "grad_norm": 2.0114903450012207,
      "learning_rate": 1.3858268101243666e-05,
      "loss": 1.338,
      "step": 23330
    },
    {
      "epoch": 0.8355327949576522,
      "grad_norm": 1.9516043663024902,
      "learning_rate": 1.3852377567255913e-05,
      "loss": 1.3652,
      "step": 23331
    },
    {
      "epoch": 0.8355686070872205,
      "grad_norm": 1.8483436107635498,
      "learning_rate": 1.3846488192273298e-05,
      "loss": 1.8179,
      "step": 23332
    },
    {
      "epoch": 0.8356044192167887,
      "grad_norm": 1.333381175994873,
      "learning_rate": 1.384059997637508e-05,
      "loss": 1.4457,
      "step": 23333
    },
    {
      "epoch": 0.835640231346357,
      "grad_norm": 1.6573841571807861,
      "learning_rate": 1.3834712919640424e-05,
      "loss": 1.1238,
      "step": 23334
    },
    {
      "epoch": 0.8356760434759253,
      "grad_norm": 1.6716316938400269,
      "learning_rate": 1.382882702214856e-05,
      "loss": 1.6315,
      "step": 23335
    },
    {
      "epoch": 0.8357118556054935,
      "grad_norm": 2.7846570014953613,
      "learning_rate": 1.382294228397868e-05,
      "loss": 1.7287,
      "step": 23336
    },
    {
      "epoch": 0.8357476677350618,
      "grad_norm": 1.2655292749404907,
      "learning_rate": 1.3817058705209973e-05,
      "loss": 1.5531,
      "step": 23337
    },
    {
      "epoch": 0.8357834798646302,
      "grad_norm": 1.5303046703338623,
      "learning_rate": 1.3811176285921557e-05,
      "loss": 1.3362,
      "step": 23338
    },
    {
      "epoch": 0.8358192919941985,
      "grad_norm": 2.3918161392211914,
      "learning_rate": 1.3805295026192577e-05,
      "loss": 1.4692,
      "step": 23339
    },
    {
      "epoch": 0.8358551041237667,
      "grad_norm": 1.6694695949554443,
      "learning_rate": 1.3799414926102194e-05,
      "loss": 1.3039,
      "step": 23340
    },
    {
      "epoch": 0.835890916253335,
      "grad_norm": 1.2947691679000854,
      "learning_rate": 1.3793535985729478e-05,
      "loss": 1.5269,
      "step": 23341
    },
    {
      "epoch": 0.8359267283829033,
      "grad_norm": 1.4575728178024292,
      "learning_rate": 1.3787658205153532e-05,
      "loss": 1.3698,
      "step": 23342
    },
    {
      "epoch": 0.8359625405124715,
      "grad_norm": 1.376752257347107,
      "learning_rate": 1.3781781584453435e-05,
      "loss": 1.4291,
      "step": 23343
    },
    {
      "epoch": 0.8359983526420398,
      "grad_norm": 1.3669610023498535,
      "learning_rate": 1.3775906123708282e-05,
      "loss": 1.2259,
      "step": 23344
    },
    {
      "epoch": 0.8360341647716082,
      "grad_norm": 1.7606292963027954,
      "learning_rate": 1.3770031822997064e-05,
      "loss": 1.4199,
      "step": 23345
    },
    {
      "epoch": 0.8360699769011765,
      "grad_norm": 1.6351981163024902,
      "learning_rate": 1.3764158682398843e-05,
      "loss": 1.3858,
      "step": 23346
    },
    {
      "epoch": 0.8361057890307447,
      "grad_norm": 1.4310319423675537,
      "learning_rate": 1.3758286701992651e-05,
      "loss": 1.4934,
      "step": 23347
    },
    {
      "epoch": 0.836141601160313,
      "grad_norm": 1.3345412015914917,
      "learning_rate": 1.375241588185744e-05,
      "loss": 1.3918,
      "step": 23348
    },
    {
      "epoch": 0.8361774132898813,
      "grad_norm": 1.554101586341858,
      "learning_rate": 1.3746546222072232e-05,
      "loss": 1.2231,
      "step": 23349
    },
    {
      "epoch": 0.8362132254194495,
      "grad_norm": 1.5663398504257202,
      "learning_rate": 1.3740677722715977e-05,
      "loss": 1.5471,
      "step": 23350
    },
    {
      "epoch": 0.8362490375490178,
      "grad_norm": 1.472374439239502,
      "learning_rate": 1.373481038386767e-05,
      "loss": 1.4637,
      "step": 23351
    },
    {
      "epoch": 0.8362848496785862,
      "grad_norm": 1.973220705986023,
      "learning_rate": 1.3728944205606186e-05,
      "loss": 1.3186,
      "step": 23352
    },
    {
      "epoch": 0.8363206618081545,
      "grad_norm": 1.8892866373062134,
      "learning_rate": 1.3723079188010469e-05,
      "loss": 1.4246,
      "step": 23353
    },
    {
      "epoch": 0.8363564739377227,
      "grad_norm": 1.4619778394699097,
      "learning_rate": 1.3717215331159439e-05,
      "loss": 1.3644,
      "step": 23354
    },
    {
      "epoch": 0.836392286067291,
      "grad_norm": 1.3195886611938477,
      "learning_rate": 1.3711352635132002e-05,
      "loss": 1.1605,
      "step": 23355
    },
    {
      "epoch": 0.8364280981968593,
      "grad_norm": 1.655963659286499,
      "learning_rate": 1.3705491100006995e-05,
      "loss": 1.3923,
      "step": 23356
    },
    {
      "epoch": 0.8364639103264275,
      "grad_norm": 1.574783444404602,
      "learning_rate": 1.3699630725863289e-05,
      "loss": 1.3442,
      "step": 23357
    },
    {
      "epoch": 0.8364997224559958,
      "grad_norm": 1.387752652168274,
      "learning_rate": 1.3693771512779752e-05,
      "loss": 1.4718,
      "step": 23358
    },
    {
      "epoch": 0.8365355345855642,
      "grad_norm": 1.196673035621643,
      "learning_rate": 1.3687913460835167e-05,
      "loss": 1.6578,
      "step": 23359
    },
    {
      "epoch": 0.8365713467151324,
      "grad_norm": 1.6154292821884155,
      "learning_rate": 1.3682056570108382e-05,
      "loss": 1.5265,
      "step": 23360
    },
    {
      "epoch": 0.8366071588447007,
      "grad_norm": 1.4177491664886475,
      "learning_rate": 1.3676200840678167e-05,
      "loss": 1.1989,
      "step": 23361
    },
    {
      "epoch": 0.836642970974269,
      "grad_norm": 1.6271674633026123,
      "learning_rate": 1.3670346272623357e-05,
      "loss": 1.2113,
      "step": 23362
    },
    {
      "epoch": 0.8366787831038373,
      "grad_norm": 1.4818865060806274,
      "learning_rate": 1.366449286602265e-05,
      "loss": 1.4199,
      "step": 23363
    },
    {
      "epoch": 0.8367145952334055,
      "grad_norm": 1.7398416996002197,
      "learning_rate": 1.3658640620954832e-05,
      "loss": 1.4518,
      "step": 23364
    },
    {
      "epoch": 0.8367504073629738,
      "grad_norm": 1.4655635356903076,
      "learning_rate": 1.3652789537498656e-05,
      "loss": 1.5296,
      "step": 23365
    },
    {
      "epoch": 0.8367862194925422,
      "grad_norm": 1.5625853538513184,
      "learning_rate": 1.364693961573279e-05,
      "loss": 1.3327,
      "step": 23366
    },
    {
      "epoch": 0.8368220316221104,
      "grad_norm": 1.5531121492385864,
      "learning_rate": 1.3641090855735972e-05,
      "loss": 1.0688,
      "step": 23367
    },
    {
      "epoch": 0.8368578437516787,
      "grad_norm": 1.4807175397872925,
      "learning_rate": 1.3635243257586872e-05,
      "loss": 1.5016,
      "step": 23368
    },
    {
      "epoch": 0.836893655881247,
      "grad_norm": 1.3785326480865479,
      "learning_rate": 1.3629396821364193e-05,
      "loss": 1.4082,
      "step": 23369
    },
    {
      "epoch": 0.8369294680108152,
      "grad_norm": 2.476111888885498,
      "learning_rate": 1.3623551547146552e-05,
      "loss": 1.5001,
      "step": 23370
    },
    {
      "epoch": 0.8369652801403835,
      "grad_norm": 1.6185861825942993,
      "learning_rate": 1.3617707435012606e-05,
      "loss": 1.3256,
      "step": 23371
    },
    {
      "epoch": 0.8370010922699518,
      "grad_norm": 1.973078727722168,
      "learning_rate": 1.3611864485040982e-05,
      "loss": 1.6241,
      "step": 23372
    },
    {
      "epoch": 0.8370369043995202,
      "grad_norm": 1.578711748123169,
      "learning_rate": 1.3606022697310316e-05,
      "loss": 1.5142,
      "step": 23373
    },
    {
      "epoch": 0.8370727165290884,
      "grad_norm": 1.5022464990615845,
      "learning_rate": 1.3600182071899148e-05,
      "loss": 1.6072,
      "step": 23374
    },
    {
      "epoch": 0.8371085286586567,
      "grad_norm": 1.498950481414795,
      "learning_rate": 1.359434260888608e-05,
      "loss": 1.4262,
      "step": 23375
    },
    {
      "epoch": 0.837144340788225,
      "grad_norm": 1.854566216468811,
      "learning_rate": 1.3588504308349703e-05,
      "loss": 1.3126,
      "step": 23376
    },
    {
      "epoch": 0.8371801529177932,
      "grad_norm": 1.531659483909607,
      "learning_rate": 1.3582667170368513e-05,
      "loss": 1.4459,
      "step": 23377
    },
    {
      "epoch": 0.8372159650473615,
      "grad_norm": 1.3971465826034546,
      "learning_rate": 1.3576831195021067e-05,
      "loss": 1.5605,
      "step": 23378
    },
    {
      "epoch": 0.8372517771769298,
      "grad_norm": 1.5634061098098755,
      "learning_rate": 1.357099638238587e-05,
      "loss": 1.3963,
      "step": 23379
    },
    {
      "epoch": 0.8372875893064982,
      "grad_norm": 1.8827449083328247,
      "learning_rate": 1.356516273254147e-05,
      "loss": 1.2539,
      "step": 23380
    },
    {
      "epoch": 0.8373234014360664,
      "grad_norm": 2.146848678588867,
      "learning_rate": 1.3559330245566282e-05,
      "loss": 1.4573,
      "step": 23381
    },
    {
      "epoch": 0.8373592135656347,
      "grad_norm": 1.4957455396652222,
      "learning_rate": 1.3553498921538798e-05,
      "loss": 1.4717,
      "step": 23382
    },
    {
      "epoch": 0.837395025695203,
      "grad_norm": 1.9274322986602783,
      "learning_rate": 1.3547668760537514e-05,
      "loss": 1.6203,
      "step": 23383
    },
    {
      "epoch": 0.8374308378247712,
      "grad_norm": 2.2220675945281982,
      "learning_rate": 1.3541839762640796e-05,
      "loss": 1.3541,
      "step": 23384
    },
    {
      "epoch": 0.8374666499543395,
      "grad_norm": 1.7440205812454224,
      "learning_rate": 1.3536011927927117e-05,
      "loss": 1.4781,
      "step": 23385
    },
    {
      "epoch": 0.8375024620839078,
      "grad_norm": 1.5415992736816406,
      "learning_rate": 1.3530185256474848e-05,
      "loss": 1.4387,
      "step": 23386
    },
    {
      "epoch": 0.8375382742134762,
      "grad_norm": 1.6011251211166382,
      "learning_rate": 1.3524359748362437e-05,
      "loss": 1.4087,
      "step": 23387
    },
    {
      "epoch": 0.8375740863430444,
      "grad_norm": 2.023725986480713,
      "learning_rate": 1.3518535403668186e-05,
      "loss": 1.3027,
      "step": 23388
    },
    {
      "epoch": 0.8376098984726127,
      "grad_norm": 1.5329786539077759,
      "learning_rate": 1.3512712222470491e-05,
      "loss": 1.1108,
      "step": 23389
    },
    {
      "epoch": 0.837645710602181,
      "grad_norm": 1.8306118249893188,
      "learning_rate": 1.3506890204847722e-05,
      "loss": 1.6126,
      "step": 23390
    },
    {
      "epoch": 0.8376815227317492,
      "grad_norm": 1.6327990293502808,
      "learning_rate": 1.3501069350878149e-05,
      "loss": 1.2224,
      "step": 23391
    },
    {
      "epoch": 0.8377173348613175,
      "grad_norm": 1.5039691925048828,
      "learning_rate": 1.3495249660640142e-05,
      "loss": 1.1612,
      "step": 23392
    },
    {
      "epoch": 0.8377531469908858,
      "grad_norm": 1.8655650615692139,
      "learning_rate": 1.3489431134211916e-05,
      "loss": 1.4411,
      "step": 23393
    },
    {
      "epoch": 0.8377889591204541,
      "grad_norm": 1.7704803943634033,
      "learning_rate": 1.3483613771671843e-05,
      "loss": 1.5848,
      "step": 23394
    },
    {
      "epoch": 0.8378247712500224,
      "grad_norm": 1.495496153831482,
      "learning_rate": 1.3477797573098128e-05,
      "loss": 1.3929,
      "step": 23395
    },
    {
      "epoch": 0.8378605833795907,
      "grad_norm": 1.5385957956314087,
      "learning_rate": 1.347198253856905e-05,
      "loss": 1.2143,
      "step": 23396
    },
    {
      "epoch": 0.837896395509159,
      "grad_norm": 2.3165247440338135,
      "learning_rate": 1.3466168668162827e-05,
      "loss": 1.2349,
      "step": 23397
    },
    {
      "epoch": 0.8379322076387272,
      "grad_norm": 1.8588618040084839,
      "learning_rate": 1.3460355961957704e-05,
      "loss": 1.4881,
      "step": 23398
    },
    {
      "epoch": 0.8379680197682955,
      "grad_norm": 1.4896241426467896,
      "learning_rate": 1.3454544420031878e-05,
      "loss": 1.6074,
      "step": 23399
    },
    {
      "epoch": 0.8380038318978638,
      "grad_norm": 1.9073976278305054,
      "learning_rate": 1.3448734042463463e-05,
      "loss": 1.7668,
      "step": 23400
    },
    {
      "epoch": 0.8380396440274321,
      "grad_norm": 1.4687809944152832,
      "learning_rate": 1.3442924829330738e-05,
      "loss": 1.6483,
      "step": 23401
    },
    {
      "epoch": 0.8380754561570004,
      "grad_norm": 1.6176726818084717,
      "learning_rate": 1.3437116780711778e-05,
      "loss": 1.2699,
      "step": 23402
    },
    {
      "epoch": 0.8381112682865687,
      "grad_norm": 1.1954487562179565,
      "learning_rate": 1.3431309896684785e-05,
      "loss": 1.2813,
      "step": 23403
    },
    {
      "epoch": 0.838147080416137,
      "grad_norm": 1.294426441192627,
      "learning_rate": 1.3425504177327808e-05,
      "loss": 1.1881,
      "step": 23404
    },
    {
      "epoch": 0.8381828925457052,
      "grad_norm": 1.699479341506958,
      "learning_rate": 1.341969962271904e-05,
      "loss": 1.2483,
      "step": 23405
    },
    {
      "epoch": 0.8382187046752735,
      "grad_norm": 1.4392833709716797,
      "learning_rate": 1.3413896232936506e-05,
      "loss": 1.3719,
      "step": 23406
    },
    {
      "epoch": 0.8382545168048418,
      "grad_norm": 1.4115568399429321,
      "learning_rate": 1.3408094008058314e-05,
      "loss": 1.3079,
      "step": 23407
    },
    {
      "epoch": 0.8382903289344101,
      "grad_norm": 1.4831119775772095,
      "learning_rate": 1.3402292948162554e-05,
      "loss": 1.259,
      "step": 23408
    },
    {
      "epoch": 0.8383261410639784,
      "grad_norm": 1.8980132341384888,
      "learning_rate": 1.3396493053327208e-05,
      "loss": 1.2174,
      "step": 23409
    },
    {
      "epoch": 0.8383619531935467,
      "grad_norm": 1.7400474548339844,
      "learning_rate": 1.339069432363036e-05,
      "loss": 1.3569,
      "step": 23410
    },
    {
      "epoch": 0.8383977653231149,
      "grad_norm": 1.8057924509048462,
      "learning_rate": 1.3384896759149957e-05,
      "loss": 1.4134,
      "step": 23411
    },
    {
      "epoch": 0.8384335774526832,
      "grad_norm": 1.5717723369598389,
      "learning_rate": 1.3379100359964082e-05,
      "loss": 1.6989,
      "step": 23412
    },
    {
      "epoch": 0.8384693895822515,
      "grad_norm": 1.7798326015472412,
      "learning_rate": 1.337330512615066e-05,
      "loss": 1.3593,
      "step": 23413
    },
    {
      "epoch": 0.8385052017118197,
      "grad_norm": 1.4081906080245972,
      "learning_rate": 1.3367511057787707e-05,
      "loss": 1.275,
      "step": 23414
    },
    {
      "epoch": 0.8385410138413881,
      "grad_norm": 1.6106690168380737,
      "learning_rate": 1.3361718154953096e-05,
      "loss": 1.5284,
      "step": 23415
    },
    {
      "epoch": 0.8385768259709564,
      "grad_norm": 1.369853138923645,
      "learning_rate": 1.3355926417724852e-05,
      "loss": 1.051,
      "step": 23416
    },
    {
      "epoch": 0.8386126381005247,
      "grad_norm": 1.8405297994613647,
      "learning_rate": 1.3350135846180856e-05,
      "loss": 1.6593,
      "step": 23417
    },
    {
      "epoch": 0.8386484502300929,
      "grad_norm": 1.8503670692443848,
      "learning_rate": 1.3344346440398992e-05,
      "loss": 1.4476,
      "step": 23418
    },
    {
      "epoch": 0.8386842623596612,
      "grad_norm": 2.0508267879486084,
      "learning_rate": 1.3338558200457174e-05,
      "loss": 1.3094,
      "step": 23419
    },
    {
      "epoch": 0.8387200744892295,
      "grad_norm": 2.294262647628784,
      "learning_rate": 1.3332771126433263e-05,
      "loss": 1.4598,
      "step": 23420
    },
    {
      "epoch": 0.8387558866187977,
      "grad_norm": 1.7309292554855347,
      "learning_rate": 1.3326985218405152e-05,
      "loss": 1.1046,
      "step": 23421
    },
    {
      "epoch": 0.8387916987483661,
      "grad_norm": 1.3808698654174805,
      "learning_rate": 1.3321200476450602e-05,
      "loss": 1.5908,
      "step": 23422
    },
    {
      "epoch": 0.8388275108779344,
      "grad_norm": 1.4959542751312256,
      "learning_rate": 1.3315416900647548e-05,
      "loss": 1.298,
      "step": 23423
    },
    {
      "epoch": 0.8388633230075027,
      "grad_norm": 1.4639296531677246,
      "learning_rate": 1.3309634491073707e-05,
      "loss": 1.5601,
      "step": 23424
    },
    {
      "epoch": 0.8388991351370709,
      "grad_norm": 1.8734369277954102,
      "learning_rate": 1.330385324780694e-05,
      "loss": 1.5761,
      "step": 23425
    },
    {
      "epoch": 0.8389349472666392,
      "grad_norm": 1.7705508470535278,
      "learning_rate": 1.3298073170924986e-05,
      "loss": 1.6338,
      "step": 23426
    },
    {
      "epoch": 0.8389707593962075,
      "grad_norm": 1.5225541591644287,
      "learning_rate": 1.3292294260505611e-05,
      "loss": 1.3579,
      "step": 23427
    },
    {
      "epoch": 0.8390065715257757,
      "grad_norm": 1.517962098121643,
      "learning_rate": 1.328651651662659e-05,
      "loss": 1.0889,
      "step": 23428
    },
    {
      "epoch": 0.8390423836553441,
      "grad_norm": 1.481013536453247,
      "learning_rate": 1.3280739939365617e-05,
      "loss": 1.5388,
      "step": 23429
    },
    {
      "epoch": 0.8390781957849124,
      "grad_norm": 2.0160789489746094,
      "learning_rate": 1.3274964528800437e-05,
      "loss": 1.8554,
      "step": 23430
    },
    {
      "epoch": 0.8391140079144807,
      "grad_norm": 1.6885130405426025,
      "learning_rate": 1.3269190285008737e-05,
      "loss": 1.433,
      "step": 23431
    },
    {
      "epoch": 0.8391498200440489,
      "grad_norm": 1.642545461654663,
      "learning_rate": 1.3263417208068218e-05,
      "loss": 1.3426,
      "step": 23432
    },
    {
      "epoch": 0.8391856321736172,
      "grad_norm": 1.5598057508468628,
      "learning_rate": 1.325764529805651e-05,
      "loss": 1.3445,
      "step": 23433
    },
    {
      "epoch": 0.8392214443031855,
      "grad_norm": 1.5403119325637817,
      "learning_rate": 1.3251874555051336e-05,
      "loss": 1.5846,
      "step": 23434
    },
    {
      "epoch": 0.8392572564327537,
      "grad_norm": 1.8304781913757324,
      "learning_rate": 1.3246104979130281e-05,
      "loss": 1.3213,
      "step": 23435
    },
    {
      "epoch": 0.8392930685623221,
      "grad_norm": 1.7894020080566406,
      "learning_rate": 1.324033657037097e-05,
      "loss": 1.6969,
      "step": 23436
    },
    {
      "epoch": 0.8393288806918904,
      "grad_norm": 1.809398889541626,
      "learning_rate": 1.323456932885101e-05,
      "loss": 1.4837,
      "step": 23437
    },
    {
      "epoch": 0.8393646928214586,
      "grad_norm": 1.6280808448791504,
      "learning_rate": 1.3228803254648004e-05,
      "loss": 1.6715,
      "step": 23438
    },
    {
      "epoch": 0.8394005049510269,
      "grad_norm": 1.8543540239334106,
      "learning_rate": 1.3223038347839544e-05,
      "loss": 1.3243,
      "step": 23439
    },
    {
      "epoch": 0.8394363170805952,
      "grad_norm": 1.5658224821090698,
      "learning_rate": 1.321727460850315e-05,
      "loss": 1.4643,
      "step": 23440
    },
    {
      "epoch": 0.8394721292101635,
      "grad_norm": 1.7527110576629639,
      "learning_rate": 1.321151203671639e-05,
      "loss": 1.2166,
      "step": 23441
    },
    {
      "epoch": 0.8395079413397317,
      "grad_norm": 1.9235097169876099,
      "learning_rate": 1.320575063255678e-05,
      "loss": 1.3466,
      "step": 23442
    },
    {
      "epoch": 0.8395437534693001,
      "grad_norm": 1.8319237232208252,
      "learning_rate": 1.3199990396101858e-05,
      "loss": 1.3571,
      "step": 23443
    },
    {
      "epoch": 0.8395795655988684,
      "grad_norm": 2.8446924686431885,
      "learning_rate": 1.3194231327429085e-05,
      "loss": 1.7009,
      "step": 23444
    },
    {
      "epoch": 0.8396153777284366,
      "grad_norm": 2.0625321865081787,
      "learning_rate": 1.3188473426615956e-05,
      "loss": 1.5693,
      "step": 23445
    },
    {
      "epoch": 0.8396511898580049,
      "grad_norm": 1.4854190349578857,
      "learning_rate": 1.3182716693739949e-05,
      "loss": 1.1659,
      "step": 23446
    },
    {
      "epoch": 0.8396870019875732,
      "grad_norm": 1.87632155418396,
      "learning_rate": 1.3176961128878495e-05,
      "loss": 1.3416,
      "step": 23447
    },
    {
      "epoch": 0.8397228141171414,
      "grad_norm": 1.6211196184158325,
      "learning_rate": 1.3171206732109031e-05,
      "loss": 1.342,
      "step": 23448
    },
    {
      "epoch": 0.8397586262467097,
      "grad_norm": 1.543217658996582,
      "learning_rate": 1.3165453503508984e-05,
      "loss": 1.5375,
      "step": 23449
    },
    {
      "epoch": 0.8397944383762781,
      "grad_norm": 1.9819585084915161,
      "learning_rate": 1.3159701443155759e-05,
      "loss": 1.6551,
      "step": 23450
    },
    {
      "epoch": 0.8398302505058464,
      "grad_norm": 1.3575177192687988,
      "learning_rate": 1.3153950551126725e-05,
      "loss": 1.5489,
      "step": 23451
    },
    {
      "epoch": 0.8398660626354146,
      "grad_norm": 1.597785472869873,
      "learning_rate": 1.3148200827499269e-05,
      "loss": 1.4854,
      "step": 23452
    },
    {
      "epoch": 0.8399018747649829,
      "grad_norm": 1.6049631834030151,
      "learning_rate": 1.3142452272350747e-05,
      "loss": 1.3046,
      "step": 23453
    },
    {
      "epoch": 0.8399376868945512,
      "grad_norm": 1.8065446615219116,
      "learning_rate": 1.3136704885758477e-05,
      "loss": 1.4958,
      "step": 23454
    },
    {
      "epoch": 0.8399734990241194,
      "grad_norm": 1.5930157899856567,
      "learning_rate": 1.3130958667799798e-05,
      "loss": 1.6832,
      "step": 23455
    },
    {
      "epoch": 0.8400093111536877,
      "grad_norm": 1.551112413406372,
      "learning_rate": 1.3125213618552013e-05,
      "loss": 1.4293,
      "step": 23456
    },
    {
      "epoch": 0.8400451232832561,
      "grad_norm": 2.0661537647247314,
      "learning_rate": 1.3119469738092449e-05,
      "loss": 1.3053,
      "step": 23457
    },
    {
      "epoch": 0.8400809354128244,
      "grad_norm": 1.5538454055786133,
      "learning_rate": 1.3113727026498323e-05,
      "loss": 1.2876,
      "step": 23458
    },
    {
      "epoch": 0.8401167475423926,
      "grad_norm": 1.8126628398895264,
      "learning_rate": 1.310798548384693e-05,
      "loss": 1.5399,
      "step": 23459
    },
    {
      "epoch": 0.8401525596719609,
      "grad_norm": 1.6119327545166016,
      "learning_rate": 1.3102245110215495e-05,
      "loss": 1.3692,
      "step": 23460
    },
    {
      "epoch": 0.8401883718015292,
      "grad_norm": 1.673466444015503,
      "learning_rate": 1.30965059056813e-05,
      "loss": 1.1661,
      "step": 23461
    },
    {
      "epoch": 0.8402241839310974,
      "grad_norm": 3.0586817264556885,
      "learning_rate": 1.3090767870321496e-05,
      "loss": 1.6918,
      "step": 23462
    },
    {
      "epoch": 0.8402599960606657,
      "grad_norm": 1.7422840595245361,
      "learning_rate": 1.30850310042133e-05,
      "loss": 1.4367,
      "step": 23463
    },
    {
      "epoch": 0.8402958081902341,
      "grad_norm": 1.521484613418579,
      "learning_rate": 1.3079295307433925e-05,
      "loss": 1.5184,
      "step": 23464
    },
    {
      "epoch": 0.8403316203198024,
      "grad_norm": 1.3086313009262085,
      "learning_rate": 1.307356078006049e-05,
      "loss": 1.2359,
      "step": 23465
    },
    {
      "epoch": 0.8403674324493706,
      "grad_norm": 1.7559353113174438,
      "learning_rate": 1.3067827422170165e-05,
      "loss": 1.593,
      "step": 23466
    },
    {
      "epoch": 0.8404032445789389,
      "grad_norm": 1.462621808052063,
      "learning_rate": 1.3062095233840089e-05,
      "loss": 1.3325,
      "step": 23467
    },
    {
      "epoch": 0.8404390567085072,
      "grad_norm": 1.2271201610565186,
      "learning_rate": 1.30563642151474e-05,
      "loss": 1.4682,
      "step": 23468
    },
    {
      "epoch": 0.8404748688380754,
      "grad_norm": 2.206615924835205,
      "learning_rate": 1.3050634366169156e-05,
      "loss": 1.0847,
      "step": 23469
    },
    {
      "epoch": 0.8405106809676437,
      "grad_norm": 1.6466546058654785,
      "learning_rate": 1.3044905686982479e-05,
      "loss": 1.3145,
      "step": 23470
    },
    {
      "epoch": 0.8405464930972121,
      "grad_norm": 1.7529257535934448,
      "learning_rate": 1.3039178177664458e-05,
      "loss": 1.2314,
      "step": 23471
    },
    {
      "epoch": 0.8405823052267803,
      "grad_norm": 1.4902821779251099,
      "learning_rate": 1.3033451838292088e-05,
      "loss": 1.453,
      "step": 23472
    },
    {
      "epoch": 0.8406181173563486,
      "grad_norm": 1.2909750938415527,
      "learning_rate": 1.3027726668942452e-05,
      "loss": 1.4757,
      "step": 23473
    },
    {
      "epoch": 0.8406539294859169,
      "grad_norm": 1.7654305696487427,
      "learning_rate": 1.3022002669692568e-05,
      "loss": 1.2688,
      "step": 23474
    },
    {
      "epoch": 0.8406897416154852,
      "grad_norm": 1.5162429809570312,
      "learning_rate": 1.3016279840619461e-05,
      "loss": 1.7285,
      "step": 23475
    },
    {
      "epoch": 0.8407255537450534,
      "grad_norm": 1.8618532419204712,
      "learning_rate": 1.3010558181800091e-05,
      "loss": 1.5294,
      "step": 23476
    },
    {
      "epoch": 0.8407613658746217,
      "grad_norm": 1.9252269268035889,
      "learning_rate": 1.3004837693311445e-05,
      "loss": 1.8202,
      "step": 23477
    },
    {
      "epoch": 0.8407971780041901,
      "grad_norm": 1.5172358751296997,
      "learning_rate": 1.2999118375230523e-05,
      "loss": 1.452,
      "step": 23478
    },
    {
      "epoch": 0.8408329901337583,
      "grad_norm": 1.8392778635025024,
      "learning_rate": 1.2993400227634211e-05,
      "loss": 1.2184,
      "step": 23479
    },
    {
      "epoch": 0.8408688022633266,
      "grad_norm": 1.4657130241394043,
      "learning_rate": 1.2987683250599481e-05,
      "loss": 1.2283,
      "step": 23480
    },
    {
      "epoch": 0.8409046143928949,
      "grad_norm": 1.8285713195800781,
      "learning_rate": 1.2981967444203224e-05,
      "loss": 1.2617,
      "step": 23481
    },
    {
      "epoch": 0.8409404265224631,
      "grad_norm": 1.4383978843688965,
      "learning_rate": 1.297625280852237e-05,
      "loss": 1.2994,
      "step": 23482
    },
    {
      "epoch": 0.8409762386520314,
      "grad_norm": 1.733881950378418,
      "learning_rate": 1.297053934363377e-05,
      "loss": 1.5984,
      "step": 23483
    },
    {
      "epoch": 0.8410120507815997,
      "grad_norm": 1.4433633089065552,
      "learning_rate": 1.2964827049614291e-05,
      "loss": 1.3371,
      "step": 23484
    },
    {
      "epoch": 0.8410478629111681,
      "grad_norm": 1.5623021125793457,
      "learning_rate": 1.295911592654081e-05,
      "loss": 1.8593,
      "step": 23485
    },
    {
      "epoch": 0.8410836750407363,
      "grad_norm": 1.315675973892212,
      "learning_rate": 1.2953405974490163e-05,
      "loss": 1.326,
      "step": 23486
    },
    {
      "epoch": 0.8411194871703046,
      "grad_norm": 2.0233829021453857,
      "learning_rate": 1.2947697193539154e-05,
      "loss": 1.499,
      "step": 23487
    },
    {
      "epoch": 0.8411552992998729,
      "grad_norm": 1.2258970737457275,
      "learning_rate": 1.2941989583764547e-05,
      "loss": 1.0147,
      "step": 23488
    },
    {
      "epoch": 0.8411911114294411,
      "grad_norm": 1.2284082174301147,
      "learning_rate": 1.2936283145243222e-05,
      "loss": 1.2718,
      "step": 23489
    },
    {
      "epoch": 0.8412269235590094,
      "grad_norm": 1.7486215829849243,
      "learning_rate": 1.2930577878051887e-05,
      "loss": 1.5571,
      "step": 23490
    },
    {
      "epoch": 0.8412627356885777,
      "grad_norm": 1.4280521869659424,
      "learning_rate": 1.2924873782267322e-05,
      "loss": 1.5505,
      "step": 23491
    },
    {
      "epoch": 0.8412985478181461,
      "grad_norm": 1.516255259513855,
      "learning_rate": 1.2919170857966223e-05,
      "loss": 1.2592,
      "step": 23492
    },
    {
      "epoch": 0.8413343599477143,
      "grad_norm": 1.6720364093780518,
      "learning_rate": 1.2913469105225407e-05,
      "loss": 1.5911,
      "step": 23493
    },
    {
      "epoch": 0.8413701720772826,
      "grad_norm": 1.5660908222198486,
      "learning_rate": 1.29077685241215e-05,
      "loss": 1.3434,
      "step": 23494
    },
    {
      "epoch": 0.8414059842068509,
      "grad_norm": 1.516623616218567,
      "learning_rate": 1.290206911473123e-05,
      "loss": 1.3898,
      "step": 23495
    },
    {
      "epoch": 0.8414417963364191,
      "grad_norm": 1.6085141897201538,
      "learning_rate": 1.2896370877131293e-05,
      "loss": 1.5737,
      "step": 23496
    },
    {
      "epoch": 0.8414776084659874,
      "grad_norm": 1.5538430213928223,
      "learning_rate": 1.2890673811398301e-05,
      "loss": 1.4526,
      "step": 23497
    },
    {
      "epoch": 0.8415134205955557,
      "grad_norm": 1.6788676977157593,
      "learning_rate": 1.2884977917608964e-05,
      "loss": 1.6376,
      "step": 23498
    },
    {
      "epoch": 0.841549232725124,
      "grad_norm": 1.8180345296859741,
      "learning_rate": 1.287928319583983e-05,
      "loss": 1.3496,
      "step": 23499
    },
    {
      "epoch": 0.8415850448546923,
      "grad_norm": 1.3776708841323853,
      "learning_rate": 1.2873589646167605e-05,
      "loss": 1.3057,
      "step": 23500
    },
    {
      "epoch": 0.8416208569842606,
      "grad_norm": 2.0103936195373535,
      "learning_rate": 1.2867897268668826e-05,
      "loss": 1.289,
      "step": 23501
    },
    {
      "epoch": 0.8416566691138289,
      "grad_norm": 2.1895813941955566,
      "learning_rate": 1.2862206063420113e-05,
      "loss": 1.3973,
      "step": 23502
    },
    {
      "epoch": 0.8416924812433971,
      "grad_norm": 1.5710906982421875,
      "learning_rate": 1.2856516030497979e-05,
      "loss": 1.3817,
      "step": 23503
    },
    {
      "epoch": 0.8417282933729654,
      "grad_norm": 1.7813724279403687,
      "learning_rate": 1.2850827169979063e-05,
      "loss": 1.259,
      "step": 23504
    },
    {
      "epoch": 0.8417641055025337,
      "grad_norm": 1.3750540018081665,
      "learning_rate": 1.284513948193985e-05,
      "loss": 1.3579,
      "step": 23505
    },
    {
      "epoch": 0.841799917632102,
      "grad_norm": 1.7597627639770508,
      "learning_rate": 1.2839452966456822e-05,
      "loss": 1.2448,
      "step": 23506
    },
    {
      "epoch": 0.8418357297616703,
      "grad_norm": 1.2781704664230347,
      "learning_rate": 1.2833767623606563e-05,
      "loss": 1.4802,
      "step": 23507
    },
    {
      "epoch": 0.8418715418912386,
      "grad_norm": 1.4671543836593628,
      "learning_rate": 1.28280834534655e-05,
      "loss": 1.1333,
      "step": 23508
    },
    {
      "epoch": 0.8419073540208069,
      "grad_norm": 1.8261942863464355,
      "learning_rate": 1.2822400456110162e-05,
      "loss": 1.3526,
      "step": 23509
    },
    {
      "epoch": 0.8419431661503751,
      "grad_norm": 1.9040741920471191,
      "learning_rate": 1.281671863161693e-05,
      "loss": 1.2746,
      "step": 23510
    },
    {
      "epoch": 0.8419789782799434,
      "grad_norm": 1.4849711656570435,
      "learning_rate": 1.2811037980062324e-05,
      "loss": 1.4179,
      "step": 23511
    },
    {
      "epoch": 0.8420147904095117,
      "grad_norm": 1.4805055856704712,
      "learning_rate": 1.2805358501522724e-05,
      "loss": 1.4121,
      "step": 23512
    },
    {
      "epoch": 0.84205060253908,
      "grad_norm": 1.6600576639175415,
      "learning_rate": 1.279968019607457e-05,
      "loss": 1.4056,
      "step": 23513
    },
    {
      "epoch": 0.8420864146686483,
      "grad_norm": 1.296726107597351,
      "learning_rate": 1.2794003063794225e-05,
      "loss": 1.6649,
      "step": 23514
    },
    {
      "epoch": 0.8421222267982166,
      "grad_norm": 1.372146725654602,
      "learning_rate": 1.2788327104758068e-05,
      "loss": 1.5393,
      "step": 23515
    },
    {
      "epoch": 0.8421580389277848,
      "grad_norm": 1.7383742332458496,
      "learning_rate": 1.278265231904251e-05,
      "loss": 1.5947,
      "step": 23516
    },
    {
      "epoch": 0.8421938510573531,
      "grad_norm": 1.5140732526779175,
      "learning_rate": 1.277697870672383e-05,
      "loss": 1.3892,
      "step": 23517
    },
    {
      "epoch": 0.8422296631869214,
      "grad_norm": 1.4669893980026245,
      "learning_rate": 1.2771306267878392e-05,
      "loss": 1.2139,
      "step": 23518
    },
    {
      "epoch": 0.8422654753164897,
      "grad_norm": 2.693443536758423,
      "learning_rate": 1.2765635002582521e-05,
      "loss": 1.362,
      "step": 23519
    },
    {
      "epoch": 0.842301287446058,
      "grad_norm": 1.3767279386520386,
      "learning_rate": 1.2759964910912524e-05,
      "loss": 1.2912,
      "step": 23520
    },
    {
      "epoch": 0.8423370995756263,
      "grad_norm": 1.9460740089416504,
      "learning_rate": 1.275429599294462e-05,
      "loss": 1.6095,
      "step": 23521
    },
    {
      "epoch": 0.8423729117051946,
      "grad_norm": 1.8773564100265503,
      "learning_rate": 1.2748628248755167e-05,
      "loss": 1.5899,
      "step": 23522
    },
    {
      "epoch": 0.8424087238347628,
      "grad_norm": 1.7842673063278198,
      "learning_rate": 1.2742961678420385e-05,
      "loss": 1.4446,
      "step": 23523
    },
    {
      "epoch": 0.8424445359643311,
      "grad_norm": 1.3329923152923584,
      "learning_rate": 1.2737296282016464e-05,
      "loss": 1.4283,
      "step": 23524
    },
    {
      "epoch": 0.8424803480938994,
      "grad_norm": 1.8569622039794922,
      "learning_rate": 1.2731632059619669e-05,
      "loss": 1.3498,
      "step": 23525
    },
    {
      "epoch": 0.8425161602234676,
      "grad_norm": 2.4255120754241943,
      "learning_rate": 1.2725969011306204e-05,
      "loss": 1.6669,
      "step": 23526
    },
    {
      "epoch": 0.842551972353036,
      "grad_norm": 1.7957884073257446,
      "learning_rate": 1.2720307137152266e-05,
      "loss": 1.6592,
      "step": 23527
    },
    {
      "epoch": 0.8425877844826043,
      "grad_norm": 1.8802419900894165,
      "learning_rate": 1.271464643723399e-05,
      "loss": 1.3524,
      "step": 23528
    },
    {
      "epoch": 0.8426235966121726,
      "grad_norm": 1.8517471551895142,
      "learning_rate": 1.2708986911627551e-05,
      "loss": 1.5125,
      "step": 23529
    },
    {
      "epoch": 0.8426594087417408,
      "grad_norm": 1.721115231513977,
      "learning_rate": 1.27033285604091e-05,
      "loss": 1.3382,
      "step": 23530
    },
    {
      "epoch": 0.8426952208713091,
      "grad_norm": 1.90151846408844,
      "learning_rate": 1.2697671383654786e-05,
      "loss": 1.58,
      "step": 23531
    },
    {
      "epoch": 0.8427310330008774,
      "grad_norm": 1.4302324056625366,
      "learning_rate": 1.2692015381440658e-05,
      "loss": 1.1196,
      "step": 23532
    },
    {
      "epoch": 0.8427668451304456,
      "grad_norm": 1.452170491218567,
      "learning_rate": 1.2686360553842857e-05,
      "loss": 1.4946,
      "step": 23533
    },
    {
      "epoch": 0.842802657260014,
      "grad_norm": 1.3650285005569458,
      "learning_rate": 1.2680706900937455e-05,
      "loss": 1.3334,
      "step": 23534
    },
    {
      "epoch": 0.8428384693895823,
      "grad_norm": 1.7403556108474731,
      "learning_rate": 1.2675054422800503e-05,
      "loss": 1.6425,
      "step": 23535
    },
    {
      "epoch": 0.8428742815191506,
      "grad_norm": 1.6567943096160889,
      "learning_rate": 1.2669403119508039e-05,
      "loss": 1.5255,
      "step": 23536
    },
    {
      "epoch": 0.8429100936487188,
      "grad_norm": 1.5474529266357422,
      "learning_rate": 1.2663752991136112e-05,
      "loss": 1.6668,
      "step": 23537
    },
    {
      "epoch": 0.8429459057782871,
      "grad_norm": 1.6595933437347412,
      "learning_rate": 1.2658104037760753e-05,
      "loss": 1.3727,
      "step": 23538
    },
    {
      "epoch": 0.8429817179078554,
      "grad_norm": 1.8614715337753296,
      "learning_rate": 1.2652456259457924e-05,
      "loss": 1.4968,
      "step": 23539
    },
    {
      "epoch": 0.8430175300374236,
      "grad_norm": 2.24442720413208,
      "learning_rate": 1.2646809656303627e-05,
      "loss": 1.5371,
      "step": 23540
    },
    {
      "epoch": 0.843053342166992,
      "grad_norm": 1.914910912513733,
      "learning_rate": 1.2641164228373847e-05,
      "loss": 1.4012,
      "step": 23541
    },
    {
      "epoch": 0.8430891542965603,
      "grad_norm": 3.1867616176605225,
      "learning_rate": 1.2635519975744503e-05,
      "loss": 1.6351,
      "step": 23542
    },
    {
      "epoch": 0.8431249664261286,
      "grad_norm": 1.5233803987503052,
      "learning_rate": 1.2629876898491532e-05,
      "loss": 1.4552,
      "step": 23543
    },
    {
      "epoch": 0.8431607785556968,
      "grad_norm": 1.6941332817077637,
      "learning_rate": 1.2624234996690875e-05,
      "loss": 1.233,
      "step": 23544
    },
    {
      "epoch": 0.8431965906852651,
      "grad_norm": 1.8111356496810913,
      "learning_rate": 1.2618594270418448e-05,
      "loss": 1.1046,
      "step": 23545
    },
    {
      "epoch": 0.8432324028148334,
      "grad_norm": 1.4550955295562744,
      "learning_rate": 1.2612954719750103e-05,
      "loss": 1.3761,
      "step": 23546
    },
    {
      "epoch": 0.8432682149444016,
      "grad_norm": 1.7182893753051758,
      "learning_rate": 1.2607316344761733e-05,
      "loss": 1.261,
      "step": 23547
    },
    {
      "epoch": 0.84330402707397,
      "grad_norm": 1.9735833406448364,
      "learning_rate": 1.2601679145529189e-05,
      "loss": 1.2702,
      "step": 23548
    },
    {
      "epoch": 0.8433398392035383,
      "grad_norm": 1.8338557481765747,
      "learning_rate": 1.2596043122128343e-05,
      "loss": 1.2931,
      "step": 23549
    },
    {
      "epoch": 0.8433756513331065,
      "grad_norm": 2.4479641914367676,
      "learning_rate": 1.2590408274634969e-05,
      "loss": 1.2256,
      "step": 23550
    },
    {
      "epoch": 0.8434114634626748,
      "grad_norm": 1.5408915281295776,
      "learning_rate": 1.2584774603124905e-05,
      "loss": 1.352,
      "step": 23551
    },
    {
      "epoch": 0.8434472755922431,
      "grad_norm": 1.7759418487548828,
      "learning_rate": 1.2579142107673959e-05,
      "loss": 1.282,
      "step": 23552
    },
    {
      "epoch": 0.8434830877218114,
      "grad_norm": 1.3212677240371704,
      "learning_rate": 1.2573510788357867e-05,
      "loss": 1.3845,
      "step": 23553
    },
    {
      "epoch": 0.8435188998513796,
      "grad_norm": 2.116630792617798,
      "learning_rate": 1.2567880645252417e-05,
      "loss": 1.23,
      "step": 23554
    },
    {
      "epoch": 0.843554711980948,
      "grad_norm": 2.0516796112060547,
      "learning_rate": 1.2562251678433356e-05,
      "loss": 1.6818,
      "step": 23555
    },
    {
      "epoch": 0.8435905241105163,
      "grad_norm": 1.6188548803329468,
      "learning_rate": 1.2556623887976427e-05,
      "loss": 1.6243,
      "step": 23556
    },
    {
      "epoch": 0.8436263362400845,
      "grad_norm": 1.391708254814148,
      "learning_rate": 1.255099727395732e-05,
      "loss": 1.2512,
      "step": 23557
    },
    {
      "epoch": 0.8436621483696528,
      "grad_norm": 1.5969374179840088,
      "learning_rate": 1.2545371836451736e-05,
      "loss": 1.2388,
      "step": 23558
    },
    {
      "epoch": 0.8436979604992211,
      "grad_norm": 1.3789353370666504,
      "learning_rate": 1.2539747575535387e-05,
      "loss": 1.4325,
      "step": 23559
    },
    {
      "epoch": 0.8437337726287893,
      "grad_norm": 2.3373169898986816,
      "learning_rate": 1.2534124491283893e-05,
      "loss": 1.6212,
      "step": 23560
    },
    {
      "epoch": 0.8437695847583576,
      "grad_norm": 2.5310919284820557,
      "learning_rate": 1.2528502583772938e-05,
      "loss": 1.6208,
      "step": 23561
    },
    {
      "epoch": 0.843805396887926,
      "grad_norm": 1.8715249300003052,
      "learning_rate": 1.252288185307815e-05,
      "loss": 1.5006,
      "step": 23562
    },
    {
      "epoch": 0.8438412090174943,
      "grad_norm": 1.8814998865127563,
      "learning_rate": 1.2517262299275167e-05,
      "loss": 1.52,
      "step": 23563
    },
    {
      "epoch": 0.8438770211470625,
      "grad_norm": 1.4319605827331543,
      "learning_rate": 1.2511643922439564e-05,
      "loss": 1.4662,
      "step": 23564
    },
    {
      "epoch": 0.8439128332766308,
      "grad_norm": 1.735001564025879,
      "learning_rate": 1.2506026722646924e-05,
      "loss": 0.974,
      "step": 23565
    },
    {
      "epoch": 0.8439486454061991,
      "grad_norm": 1.460253119468689,
      "learning_rate": 1.2500410699972853e-05,
      "loss": 1.0201,
      "step": 23566
    },
    {
      "epoch": 0.8439844575357673,
      "grad_norm": 2.5098659992218018,
      "learning_rate": 1.2494795854492903e-05,
      "loss": 1.7105,
      "step": 23567
    },
    {
      "epoch": 0.8440202696653356,
      "grad_norm": 1.4424971342086792,
      "learning_rate": 1.2489182186282577e-05,
      "loss": 1.2676,
      "step": 23568
    },
    {
      "epoch": 0.8440560817949039,
      "grad_norm": 1.512639045715332,
      "learning_rate": 1.2483569695417418e-05,
      "loss": 1.4629,
      "step": 23569
    },
    {
      "epoch": 0.8440918939244723,
      "grad_norm": 1.5117937326431274,
      "learning_rate": 1.2477958381972977e-05,
      "loss": 1.133,
      "step": 23570
    },
    {
      "epoch": 0.8441277060540405,
      "grad_norm": 1.7218340635299683,
      "learning_rate": 1.2472348246024679e-05,
      "loss": 1.2329,
      "step": 23571
    },
    {
      "epoch": 0.8441635181836088,
      "grad_norm": 1.6333949565887451,
      "learning_rate": 1.2466739287648032e-05,
      "loss": 1.5322,
      "step": 23572
    },
    {
      "epoch": 0.8441993303131771,
      "grad_norm": 1.4452733993530273,
      "learning_rate": 1.24611315069185e-05,
      "loss": 1.3076,
      "step": 23573
    },
    {
      "epoch": 0.8442351424427453,
      "grad_norm": 1.46990168094635,
      "learning_rate": 1.2455524903911552e-05,
      "loss": 1.3682,
      "step": 23574
    },
    {
      "epoch": 0.8442709545723136,
      "grad_norm": 2.660382032394409,
      "learning_rate": 1.2449919478702587e-05,
      "loss": 1.623,
      "step": 23575
    },
    {
      "epoch": 0.8443067667018819,
      "grad_norm": 2.3135154247283936,
      "learning_rate": 1.2444315231366988e-05,
      "loss": 1.5553,
      "step": 23576
    },
    {
      "epoch": 0.8443425788314503,
      "grad_norm": 1.4820939302444458,
      "learning_rate": 1.2438712161980226e-05,
      "loss": 1.5055,
      "step": 23577
    },
    {
      "epoch": 0.8443783909610185,
      "grad_norm": 1.7251002788543701,
      "learning_rate": 1.2433110270617632e-05,
      "loss": 1.7433,
      "step": 23578
    },
    {
      "epoch": 0.8444142030905868,
      "grad_norm": 1.5508625507354736,
      "learning_rate": 1.2427509557354578e-05,
      "loss": 1.2144,
      "step": 23579
    },
    {
      "epoch": 0.8444500152201551,
      "grad_norm": 1.7173960208892822,
      "learning_rate": 1.2421910022266425e-05,
      "loss": 1.3851,
      "step": 23580
    },
    {
      "epoch": 0.8444858273497233,
      "grad_norm": 1.360556960105896,
      "learning_rate": 1.2416311665428526e-05,
      "loss": 1.4602,
      "step": 23581
    },
    {
      "epoch": 0.8445216394792916,
      "grad_norm": 1.280659794807434,
      "learning_rate": 1.2410714486916164e-05,
      "loss": 1.4802,
      "step": 23582
    },
    {
      "epoch": 0.8445574516088599,
      "grad_norm": 1.5254395008087158,
      "learning_rate": 1.2405118486804646e-05,
      "loss": 1.4026,
      "step": 23583
    },
    {
      "epoch": 0.8445932637384282,
      "grad_norm": 1.5753604173660278,
      "learning_rate": 1.2399523665169298e-05,
      "loss": 1.353,
      "step": 23584
    },
    {
      "epoch": 0.8446290758679965,
      "grad_norm": 3.215534210205078,
      "learning_rate": 1.239393002208533e-05,
      "loss": 1.3249,
      "step": 23585
    },
    {
      "epoch": 0.8446648879975648,
      "grad_norm": 1.6247320175170898,
      "learning_rate": 1.238833755762806e-05,
      "loss": 1.4145,
      "step": 23586
    },
    {
      "epoch": 0.844700700127133,
      "grad_norm": 2.186195135116577,
      "learning_rate": 1.2382746271872658e-05,
      "loss": 1.7662,
      "step": 23587
    },
    {
      "epoch": 0.8447365122567013,
      "grad_norm": 1.3721939325332642,
      "learning_rate": 1.2377156164894422e-05,
      "loss": 1.0364,
      "step": 23588
    },
    {
      "epoch": 0.8447723243862696,
      "grad_norm": 1.7674872875213623,
      "learning_rate": 1.2371567236768511e-05,
      "loss": 1.2248,
      "step": 23589
    },
    {
      "epoch": 0.8448081365158379,
      "grad_norm": 1.807494044303894,
      "learning_rate": 1.2365979487570122e-05,
      "loss": 1.3406,
      "step": 23590
    },
    {
      "epoch": 0.8448439486454062,
      "grad_norm": 3.335813522338867,
      "learning_rate": 1.2360392917374442e-05,
      "loss": 1.4115,
      "step": 23591
    },
    {
      "epoch": 0.8448797607749745,
      "grad_norm": 1.336086392402649,
      "learning_rate": 1.235480752625665e-05,
      "loss": 1.582,
      "step": 23592
    },
    {
      "epoch": 0.8449155729045428,
      "grad_norm": 2.0651471614837646,
      "learning_rate": 1.234922331429188e-05,
      "loss": 1.2444,
      "step": 23593
    },
    {
      "epoch": 0.844951385034111,
      "grad_norm": 1.5631545782089233,
      "learning_rate": 1.2343640281555191e-05,
      "loss": 1.1421,
      "step": 23594
    },
    {
      "epoch": 0.8449871971636793,
      "grad_norm": 2.9039080142974854,
      "learning_rate": 1.2338058428121802e-05,
      "loss": 1.5782,
      "step": 23595
    },
    {
      "epoch": 0.8450230092932476,
      "grad_norm": 1.8071566820144653,
      "learning_rate": 1.233247775406674e-05,
      "loss": 1.4459,
      "step": 23596
    },
    {
      "epoch": 0.8450588214228159,
      "grad_norm": 1.7052816152572632,
      "learning_rate": 1.2326898259465125e-05,
      "loss": 1.6382,
      "step": 23597
    },
    {
      "epoch": 0.8450946335523842,
      "grad_norm": 1.8090574741363525,
      "learning_rate": 1.2321319944391963e-05,
      "loss": 1.3878,
      "step": 23598
    },
    {
      "epoch": 0.8451304456819525,
      "grad_norm": 1.5731558799743652,
      "learning_rate": 1.2315742808922382e-05,
      "loss": 1.2031,
      "step": 23599
    },
    {
      "epoch": 0.8451662578115208,
      "grad_norm": 1.4847115278244019,
      "learning_rate": 1.2310166853131366e-05,
      "loss": 1.4141,
      "step": 23600
    },
    {
      "epoch": 0.845202069941089,
      "grad_norm": 1.2769142389297485,
      "learning_rate": 1.2304592077093958e-05,
      "loss": 1.3135,
      "step": 23601
    },
    {
      "epoch": 0.8452378820706573,
      "grad_norm": 1.4810028076171875,
      "learning_rate": 1.2299018480885117e-05,
      "loss": 1.0614,
      "step": 23602
    },
    {
      "epoch": 0.8452736942002256,
      "grad_norm": 1.6237541437149048,
      "learning_rate": 1.2293446064579873e-05,
      "loss": 1.2855,
      "step": 23603
    },
    {
      "epoch": 0.8453095063297938,
      "grad_norm": 1.9380799531936646,
      "learning_rate": 1.2287874828253187e-05,
      "loss": 1.5237,
      "step": 23604
    },
    {
      "epoch": 0.8453453184593622,
      "grad_norm": 2.556157350540161,
      "learning_rate": 1.2282304771979958e-05,
      "loss": 1.4906,
      "step": 23605
    },
    {
      "epoch": 0.8453811305889305,
      "grad_norm": 2.639732837677002,
      "learning_rate": 1.2276735895835223e-05,
      "loss": 1.5245,
      "step": 23606
    },
    {
      "epoch": 0.8454169427184988,
      "grad_norm": 1.5100208520889282,
      "learning_rate": 1.2271168199893834e-05,
      "loss": 1.6371,
      "step": 23607
    },
    {
      "epoch": 0.845452754848067,
      "grad_norm": 1.5558120012283325,
      "learning_rate": 1.2265601684230732e-05,
      "loss": 1.3295,
      "step": 23608
    },
    {
      "epoch": 0.8454885669776353,
      "grad_norm": 1.4905898571014404,
      "learning_rate": 1.2260036348920745e-05,
      "loss": 1.6187,
      "step": 23609
    },
    {
      "epoch": 0.8455243791072036,
      "grad_norm": 1.5503627061843872,
      "learning_rate": 1.2254472194038835e-05,
      "loss": 1.4804,
      "step": 23610
    },
    {
      "epoch": 0.8455601912367718,
      "grad_norm": 1.5962167978286743,
      "learning_rate": 1.224890921965981e-05,
      "loss": 1.4638,
      "step": 23611
    },
    {
      "epoch": 0.8455960033663402,
      "grad_norm": 1.7058485746383667,
      "learning_rate": 1.2243347425858508e-05,
      "loss": 1.4108,
      "step": 23612
    },
    {
      "epoch": 0.8456318154959085,
      "grad_norm": 2.0142135620117188,
      "learning_rate": 1.2237786812709773e-05,
      "loss": 1.6224,
      "step": 23613
    },
    {
      "epoch": 0.8456676276254768,
      "grad_norm": 1.3440698385238647,
      "learning_rate": 1.2232227380288408e-05,
      "loss": 1.5021,
      "step": 23614
    },
    {
      "epoch": 0.845703439755045,
      "grad_norm": 1.5169098377227783,
      "learning_rate": 1.2226669128669232e-05,
      "loss": 1.3667,
      "step": 23615
    },
    {
      "epoch": 0.8457392518846133,
      "grad_norm": 2.082235813140869,
      "learning_rate": 1.2221112057926954e-05,
      "loss": 1.5047,
      "step": 23616
    },
    {
      "epoch": 0.8457750640141816,
      "grad_norm": 1.3239139318466187,
      "learning_rate": 1.2215556168136443e-05,
      "loss": 1.5601,
      "step": 23617
    },
    {
      "epoch": 0.8458108761437498,
      "grad_norm": 1.6118295192718506,
      "learning_rate": 1.2210001459372355e-05,
      "loss": 1.5155,
      "step": 23618
    },
    {
      "epoch": 0.8458466882733182,
      "grad_norm": 1.7186284065246582,
      "learning_rate": 1.2204447931709484e-05,
      "loss": 1.6944,
      "step": 23619
    },
    {
      "epoch": 0.8458825004028865,
      "grad_norm": 1.6014262437820435,
      "learning_rate": 1.2198895585222503e-05,
      "loss": 1.5028,
      "step": 23620
    },
    {
      "epoch": 0.8459183125324548,
      "grad_norm": 1.8666081428527832,
      "learning_rate": 1.219334441998612e-05,
      "loss": 1.4784,
      "step": 23621
    },
    {
      "epoch": 0.845954124662023,
      "grad_norm": 1.4160361289978027,
      "learning_rate": 1.2187794436075039e-05,
      "loss": 1.0623,
      "step": 23622
    },
    {
      "epoch": 0.8459899367915913,
      "grad_norm": 1.6083838939666748,
      "learning_rate": 1.2182245633563905e-05,
      "loss": 1.2759,
      "step": 23623
    },
    {
      "epoch": 0.8460257489211596,
      "grad_norm": 1.7315397262573242,
      "learning_rate": 1.2176698012527376e-05,
      "loss": 1.3641,
      "step": 23624
    },
    {
      "epoch": 0.8460615610507278,
      "grad_norm": 1.6413108110427856,
      "learning_rate": 1.2171151573040085e-05,
      "loss": 1.5172,
      "step": 23625
    },
    {
      "epoch": 0.8460973731802962,
      "grad_norm": 2.129379987716675,
      "learning_rate": 1.2165606315176691e-05,
      "loss": 1.7346,
      "step": 23626
    },
    {
      "epoch": 0.8461331853098645,
      "grad_norm": 1.699070930480957,
      "learning_rate": 1.2160062239011739e-05,
      "loss": 1.3449,
      "step": 23627
    },
    {
      "epoch": 0.8461689974394327,
      "grad_norm": 1.3669846057891846,
      "learning_rate": 1.2154519344619841e-05,
      "loss": 1.5008,
      "step": 23628
    },
    {
      "epoch": 0.846204809569001,
      "grad_norm": 1.5487329959869385,
      "learning_rate": 1.2148977632075598e-05,
      "loss": 1.569,
      "step": 23629
    },
    {
      "epoch": 0.8462406216985693,
      "grad_norm": 2.4239537715911865,
      "learning_rate": 1.2143437101453514e-05,
      "loss": 1.4715,
      "step": 23630
    },
    {
      "epoch": 0.8462764338281376,
      "grad_norm": 1.408677101135254,
      "learning_rate": 1.2137897752828165e-05,
      "loss": 1.3176,
      "step": 23631
    },
    {
      "epoch": 0.8463122459577058,
      "grad_norm": 1.3192760944366455,
      "learning_rate": 1.2132359586274067e-05,
      "loss": 1.4138,
      "step": 23632
    },
    {
      "epoch": 0.8463480580872742,
      "grad_norm": 1.8194007873535156,
      "learning_rate": 1.212682260186575e-05,
      "loss": 1.7193,
      "step": 23633
    },
    {
      "epoch": 0.8463838702168425,
      "grad_norm": 1.4878827333450317,
      "learning_rate": 1.2121286799677667e-05,
      "loss": 0.9791,
      "step": 23634
    },
    {
      "epoch": 0.8464196823464107,
      "grad_norm": 1.69992995262146,
      "learning_rate": 1.2115752179784312e-05,
      "loss": 1.4171,
      "step": 23635
    },
    {
      "epoch": 0.846455494475979,
      "grad_norm": 1.8349891901016235,
      "learning_rate": 1.211021874226015e-05,
      "loss": 1.8357,
      "step": 23636
    },
    {
      "epoch": 0.8464913066055473,
      "grad_norm": 1.3100531101226807,
      "learning_rate": 1.2104686487179639e-05,
      "loss": 1.4282,
      "step": 23637
    },
    {
      "epoch": 0.8465271187351155,
      "grad_norm": 1.4935340881347656,
      "learning_rate": 1.209915541461718e-05,
      "loss": 1.4017,
      "step": 23638
    },
    {
      "epoch": 0.8465629308646838,
      "grad_norm": 1.7822998762130737,
      "learning_rate": 1.20936255246472e-05,
      "loss": 1.1659,
      "step": 23639
    },
    {
      "epoch": 0.8465987429942522,
      "grad_norm": 1.7286075353622437,
      "learning_rate": 1.2088096817344118e-05,
      "loss": 1.8532,
      "step": 23640
    },
    {
      "epoch": 0.8466345551238205,
      "grad_norm": 1.912163496017456,
      "learning_rate": 1.2082569292782275e-05,
      "loss": 1.4164,
      "step": 23641
    },
    {
      "epoch": 0.8466703672533887,
      "grad_norm": 1.4342498779296875,
      "learning_rate": 1.2077042951036055e-05,
      "loss": 1.1178,
      "step": 23642
    },
    {
      "epoch": 0.846706179382957,
      "grad_norm": 2.4091644287109375,
      "learning_rate": 1.207151779217981e-05,
      "loss": 1.2119,
      "step": 23643
    },
    {
      "epoch": 0.8467419915125253,
      "grad_norm": 1.5198973417282104,
      "learning_rate": 1.2065993816287901e-05,
      "loss": 1.1991,
      "step": 23644
    },
    {
      "epoch": 0.8467778036420935,
      "grad_norm": 1.4124888181686401,
      "learning_rate": 1.2060471023434594e-05,
      "loss": 1.6583,
      "step": 23645
    },
    {
      "epoch": 0.8468136157716618,
      "grad_norm": 1.6898411512374878,
      "learning_rate": 1.2054949413694216e-05,
      "loss": 1.2365,
      "step": 23646
    },
    {
      "epoch": 0.8468494279012302,
      "grad_norm": 1.4667853116989136,
      "learning_rate": 1.2049428987141065e-05,
      "loss": 1.5452,
      "step": 23647
    },
    {
      "epoch": 0.8468852400307985,
      "grad_norm": 1.607649326324463,
      "learning_rate": 1.204390974384939e-05,
      "loss": 1.347,
      "step": 23648
    },
    {
      "epoch": 0.8469210521603667,
      "grad_norm": 1.8424022197723389,
      "learning_rate": 1.2038391683893446e-05,
      "loss": 1.7185,
      "step": 23649
    },
    {
      "epoch": 0.846956864289935,
      "grad_norm": 1.5463377237319946,
      "learning_rate": 1.2032874807347484e-05,
      "loss": 1.5278,
      "step": 23650
    },
    {
      "epoch": 0.8469926764195033,
      "grad_norm": 1.372738242149353,
      "learning_rate": 1.2027359114285741e-05,
      "loss": 1.4912,
      "step": 23651
    },
    {
      "epoch": 0.8470284885490715,
      "grad_norm": 1.7172951698303223,
      "learning_rate": 1.2021844604782384e-05,
      "loss": 1.6273,
      "step": 23652
    },
    {
      "epoch": 0.8470643006786398,
      "grad_norm": 1.8303214311599731,
      "learning_rate": 1.2016331278911619e-05,
      "loss": 1.2777,
      "step": 23653
    },
    {
      "epoch": 0.8471001128082082,
      "grad_norm": 1.78458833694458,
      "learning_rate": 1.201081913674763e-05,
      "loss": 1.6966,
      "step": 23654
    },
    {
      "epoch": 0.8471359249377765,
      "grad_norm": 1.1796941757202148,
      "learning_rate": 1.2005308178364593e-05,
      "loss": 1.457,
      "step": 23655
    },
    {
      "epoch": 0.8471717370673447,
      "grad_norm": 1.6931653022766113,
      "learning_rate": 1.1999798403836615e-05,
      "loss": 1.0998,
      "step": 23656
    },
    {
      "epoch": 0.847207549196913,
      "grad_norm": 1.5502809286117554,
      "learning_rate": 1.1994289813237835e-05,
      "loss": 1.3704,
      "step": 23657
    },
    {
      "epoch": 0.8472433613264813,
      "grad_norm": 1.3517582416534424,
      "learning_rate": 1.1988782406642385e-05,
      "loss": 1.149,
      "step": 23658
    },
    {
      "epoch": 0.8472791734560495,
      "grad_norm": 1.7601218223571777,
      "learning_rate": 1.1983276184124314e-05,
      "loss": 1.4605,
      "step": 23659
    },
    {
      "epoch": 0.8473149855856178,
      "grad_norm": 1.4631924629211426,
      "learning_rate": 1.1977771145757733e-05,
      "loss": 1.5351,
      "step": 23660
    },
    {
      "epoch": 0.8473507977151862,
      "grad_norm": 1.9184998273849487,
      "learning_rate": 1.1972267291616702e-05,
      "loss": 1.5686,
      "step": 23661
    },
    {
      "epoch": 0.8473866098447544,
      "grad_norm": 1.5700645446777344,
      "learning_rate": 1.1966764621775284e-05,
      "loss": 1.1868,
      "step": 23662
    },
    {
      "epoch": 0.8474224219743227,
      "grad_norm": 1.7700568437576294,
      "learning_rate": 1.1961263136307477e-05,
      "loss": 1.2635,
      "step": 23663
    },
    {
      "epoch": 0.847458234103891,
      "grad_norm": 1.7117582559585571,
      "learning_rate": 1.195576283528731e-05,
      "loss": 1.5889,
      "step": 23664
    },
    {
      "epoch": 0.8474940462334593,
      "grad_norm": 1.486767053604126,
      "learning_rate": 1.1950263718788812e-05,
      "loss": 1.6332,
      "step": 23665
    },
    {
      "epoch": 0.8475298583630275,
      "grad_norm": 1.6834611892700195,
      "learning_rate": 1.1944765786885914e-05,
      "loss": 1.4774,
      "step": 23666
    },
    {
      "epoch": 0.8475656704925958,
      "grad_norm": 1.8783506155014038,
      "learning_rate": 1.1939269039652612e-05,
      "loss": 1.5829,
      "step": 23667
    },
    {
      "epoch": 0.8476014826221642,
      "grad_norm": 1.8544740676879883,
      "learning_rate": 1.1933773477162847e-05,
      "loss": 1.6606,
      "step": 23668
    },
    {
      "epoch": 0.8476372947517324,
      "grad_norm": 1.5424895286560059,
      "learning_rate": 1.192827909949059e-05,
      "loss": 1.3291,
      "step": 23669
    },
    {
      "epoch": 0.8476731068813007,
      "grad_norm": 1.9440540075302124,
      "learning_rate": 1.1922785906709711e-05,
      "loss": 1.461,
      "step": 23670
    },
    {
      "epoch": 0.847708919010869,
      "grad_norm": 2.229961395263672,
      "learning_rate": 1.1917293898894145e-05,
      "loss": 1.4525,
      "step": 23671
    },
    {
      "epoch": 0.8477447311404372,
      "grad_norm": 1.5149637460708618,
      "learning_rate": 1.1911803076117777e-05,
      "loss": 1.3611,
      "step": 23672
    },
    {
      "epoch": 0.8477805432700055,
      "grad_norm": 1.4159404039382935,
      "learning_rate": 1.1906313438454464e-05,
      "loss": 1.444,
      "step": 23673
    },
    {
      "epoch": 0.8478163553995738,
      "grad_norm": 1.359679937362671,
      "learning_rate": 1.1900824985978066e-05,
      "loss": 1.1712,
      "step": 23674
    },
    {
      "epoch": 0.8478521675291422,
      "grad_norm": 1.24271821975708,
      "learning_rate": 1.1895337718762422e-05,
      "loss": 1.493,
      "step": 23675
    },
    {
      "epoch": 0.8478879796587104,
      "grad_norm": 1.549537181854248,
      "learning_rate": 1.1889851636881388e-05,
      "loss": 1.3426,
      "step": 23676
    },
    {
      "epoch": 0.8479237917882787,
      "grad_norm": 1.8525686264038086,
      "learning_rate": 1.1884366740408726e-05,
      "loss": 1.4973,
      "step": 23677
    },
    {
      "epoch": 0.847959603917847,
      "grad_norm": 1.6255896091461182,
      "learning_rate": 1.1878883029418253e-05,
      "loss": 1.6222,
      "step": 23678
    },
    {
      "epoch": 0.8479954160474152,
      "grad_norm": 1.3341083526611328,
      "learning_rate": 1.1873400503983733e-05,
      "loss": 1.4786,
      "step": 23679
    },
    {
      "epoch": 0.8480312281769835,
      "grad_norm": 1.8212075233459473,
      "learning_rate": 1.1867919164178964e-05,
      "loss": 1.4576,
      "step": 23680
    },
    {
      "epoch": 0.8480670403065518,
      "grad_norm": 1.6231342554092407,
      "learning_rate": 1.1862439010077653e-05,
      "loss": 1.5524,
      "step": 23681
    },
    {
      "epoch": 0.8481028524361202,
      "grad_norm": 1.5469783544540405,
      "learning_rate": 1.1856960041753495e-05,
      "loss": 1.3759,
      "step": 23682
    },
    {
      "epoch": 0.8481386645656884,
      "grad_norm": 1.431070327758789,
      "learning_rate": 1.185148225928029e-05,
      "loss": 1.7274,
      "step": 23683
    },
    {
      "epoch": 0.8481744766952567,
      "grad_norm": 1.7324550151824951,
      "learning_rate": 1.1846005662731663e-05,
      "loss": 1.6348,
      "step": 23684
    },
    {
      "epoch": 0.848210288824825,
      "grad_norm": 1.7936501502990723,
      "learning_rate": 1.1840530252181336e-05,
      "loss": 1.6526,
      "step": 23685
    },
    {
      "epoch": 0.8482461009543932,
      "grad_norm": 1.3822287321090698,
      "learning_rate": 1.1835056027702918e-05,
      "loss": 1.3794,
      "step": 23686
    },
    {
      "epoch": 0.8482819130839615,
      "grad_norm": 1.904231071472168,
      "learning_rate": 1.1829582989370148e-05,
      "loss": 1.6432,
      "step": 23687
    },
    {
      "epoch": 0.8483177252135298,
      "grad_norm": 2.0496718883514404,
      "learning_rate": 1.1824111137256577e-05,
      "loss": 1.3671,
      "step": 23688
    },
    {
      "epoch": 0.8483535373430982,
      "grad_norm": 2.090599298477173,
      "learning_rate": 1.1818640471435848e-05,
      "loss": 1.6222,
      "step": 23689
    },
    {
      "epoch": 0.8483893494726664,
      "grad_norm": 1.9762262105941772,
      "learning_rate": 1.1813170991981593e-05,
      "loss": 1.5517,
      "step": 23690
    },
    {
      "epoch": 0.8484251616022347,
      "grad_norm": 1.5518031120300293,
      "learning_rate": 1.1807702698967349e-05,
      "loss": 1.4667,
      "step": 23691
    },
    {
      "epoch": 0.848460973731803,
      "grad_norm": 1.2942707538604736,
      "learning_rate": 1.1802235592466727e-05,
      "loss": 1.093,
      "step": 23692
    },
    {
      "epoch": 0.8484967858613712,
      "grad_norm": 2.022648334503174,
      "learning_rate": 1.179676967255321e-05,
      "loss": 1.55,
      "step": 23693
    },
    {
      "epoch": 0.8485325979909395,
      "grad_norm": 1.2496730089187622,
      "learning_rate": 1.1791304939300429e-05,
      "loss": 0.9355,
      "step": 23694
    },
    {
      "epoch": 0.8485684101205078,
      "grad_norm": 1.1895911693572998,
      "learning_rate": 1.1785841392781838e-05,
      "loss": 1.1347,
      "step": 23695
    },
    {
      "epoch": 0.8486042222500761,
      "grad_norm": 1.413276195526123,
      "learning_rate": 1.1780379033070988e-05,
      "loss": 1.4609,
      "step": 23696
    },
    {
      "epoch": 0.8486400343796444,
      "grad_norm": 1.4332022666931152,
      "learning_rate": 1.1774917860241297e-05,
      "loss": 1.478,
      "step": 23697
    },
    {
      "epoch": 0.8486758465092127,
      "grad_norm": 1.54298734664917,
      "learning_rate": 1.1769457874366318e-05,
      "loss": 1.2612,
      "step": 23698
    },
    {
      "epoch": 0.848711658638781,
      "grad_norm": 1.376090168952942,
      "learning_rate": 1.1763999075519482e-05,
      "loss": 1.2319,
      "step": 23699
    },
    {
      "epoch": 0.8487474707683492,
      "grad_norm": 1.784658670425415,
      "learning_rate": 1.1758541463774186e-05,
      "loss": 1.53,
      "step": 23700
    },
    {
      "epoch": 0.8487832828979175,
      "grad_norm": 2.318807363510132,
      "learning_rate": 1.1753085039203926e-05,
      "loss": 1.2602,
      "step": 23701
    },
    {
      "epoch": 0.8488190950274858,
      "grad_norm": 1.3570774793624878,
      "learning_rate": 1.1747629801882054e-05,
      "loss": 1.635,
      "step": 23702
    },
    {
      "epoch": 0.8488549071570541,
      "grad_norm": 2.0229833126068115,
      "learning_rate": 1.1742175751882012e-05,
      "loss": 1.5221,
      "step": 23703
    },
    {
      "epoch": 0.8488907192866224,
      "grad_norm": 1.5566153526306152,
      "learning_rate": 1.1736722889277107e-05,
      "loss": 1.4815,
      "step": 23704
    },
    {
      "epoch": 0.8489265314161907,
      "grad_norm": 2.224531888961792,
      "learning_rate": 1.1731271214140783e-05,
      "loss": 1.5376,
      "step": 23705
    },
    {
      "epoch": 0.848962343545759,
      "grad_norm": 2.213486671447754,
      "learning_rate": 1.1725820726546322e-05,
      "loss": 1.5942,
      "step": 23706
    },
    {
      "epoch": 0.8489981556753272,
      "grad_norm": 1.877007246017456,
      "learning_rate": 1.1720371426567111e-05,
      "loss": 1.4773,
      "step": 23707
    },
    {
      "epoch": 0.8490339678048955,
      "grad_norm": 1.609832763671875,
      "learning_rate": 1.1714923314276405e-05,
      "loss": 1.2998,
      "step": 23708
    },
    {
      "epoch": 0.8490697799344638,
      "grad_norm": 1.7458049058914185,
      "learning_rate": 1.170947638974752e-05,
      "loss": 1.4685,
      "step": 23709
    },
    {
      "epoch": 0.8491055920640321,
      "grad_norm": 2.7425076961517334,
      "learning_rate": 1.1704030653053766e-05,
      "loss": 1.5496,
      "step": 23710
    },
    {
      "epoch": 0.8491414041936004,
      "grad_norm": 1.5261955261230469,
      "learning_rate": 1.1698586104268372e-05,
      "loss": 1.4367,
      "step": 23711
    },
    {
      "epoch": 0.8491772163231687,
      "grad_norm": 1.8349772691726685,
      "learning_rate": 1.169314274346459e-05,
      "loss": 1.3865,
      "step": 23712
    },
    {
      "epoch": 0.8492130284527369,
      "grad_norm": 1.9247404336929321,
      "learning_rate": 1.1687700570715677e-05,
      "loss": 1.3468,
      "step": 23713
    },
    {
      "epoch": 0.8492488405823052,
      "grad_norm": 1.6313281059265137,
      "learning_rate": 1.1682259586094845e-05,
      "loss": 1.3691,
      "step": 23714
    },
    {
      "epoch": 0.8492846527118735,
      "grad_norm": 1.6033432483673096,
      "learning_rate": 1.1676819789675264e-05,
      "loss": 1.3165,
      "step": 23715
    },
    {
      "epoch": 0.8493204648414417,
      "grad_norm": 1.6103150844573975,
      "learning_rate": 1.1671381181530171e-05,
      "loss": 1.4167,
      "step": 23716
    },
    {
      "epoch": 0.8493562769710101,
      "grad_norm": 1.6567761898040771,
      "learning_rate": 1.1665943761732712e-05,
      "loss": 1.4802,
      "step": 23717
    },
    {
      "epoch": 0.8493920891005784,
      "grad_norm": 1.9496479034423828,
      "learning_rate": 1.1660507530356024e-05,
      "loss": 1.4007,
      "step": 23718
    },
    {
      "epoch": 0.8494279012301467,
      "grad_norm": 1.6476153135299683,
      "learning_rate": 1.1655072487473251e-05,
      "loss": 1.3635,
      "step": 23719
    },
    {
      "epoch": 0.8494637133597149,
      "grad_norm": 1.7189626693725586,
      "learning_rate": 1.1649638633157523e-05,
      "loss": 1.6882,
      "step": 23720
    },
    {
      "epoch": 0.8494995254892832,
      "grad_norm": 1.4892523288726807,
      "learning_rate": 1.1644205967481959e-05,
      "loss": 1.2611,
      "step": 23721
    },
    {
      "epoch": 0.8495353376188515,
      "grad_norm": 1.507788896560669,
      "learning_rate": 1.1638774490519622e-05,
      "loss": 1.5566,
      "step": 23722
    },
    {
      "epoch": 0.8495711497484197,
      "grad_norm": 2.1150612831115723,
      "learning_rate": 1.1633344202343587e-05,
      "loss": 1.3369,
      "step": 23723
    },
    {
      "epoch": 0.8496069618779881,
      "grad_norm": 1.6760443449020386,
      "learning_rate": 1.162791510302692e-05,
      "loss": 1.0433,
      "step": 23724
    },
    {
      "epoch": 0.8496427740075564,
      "grad_norm": 1.6764628887176514,
      "learning_rate": 1.1622487192642694e-05,
      "loss": 1.3644,
      "step": 23725
    },
    {
      "epoch": 0.8496785861371247,
      "grad_norm": 1.971871018409729,
      "learning_rate": 1.1617060471263875e-05,
      "loss": 1.2388,
      "step": 23726
    },
    {
      "epoch": 0.8497143982666929,
      "grad_norm": 2.7978687286376953,
      "learning_rate": 1.1611634938963512e-05,
      "loss": 1.2827,
      "step": 23727
    },
    {
      "epoch": 0.8497502103962612,
      "grad_norm": 2.1130940914154053,
      "learning_rate": 1.1606210595814593e-05,
      "loss": 1.4592,
      "step": 23728
    },
    {
      "epoch": 0.8497860225258295,
      "grad_norm": 1.4283808469772339,
      "learning_rate": 1.1600787441890082e-05,
      "loss": 1.4418,
      "step": 23729
    },
    {
      "epoch": 0.8498218346553977,
      "grad_norm": 1.7132954597473145,
      "learning_rate": 1.1595365477262944e-05,
      "loss": 1.5116,
      "step": 23730
    },
    {
      "epoch": 0.8498576467849661,
      "grad_norm": 1.5262160301208496,
      "learning_rate": 1.1589944702006129e-05,
      "loss": 1.1867,
      "step": 23731
    },
    {
      "epoch": 0.8498934589145344,
      "grad_norm": 2.8375978469848633,
      "learning_rate": 1.158452511619259e-05,
      "loss": 1.5645,
      "step": 23732
    },
    {
      "epoch": 0.8499292710441027,
      "grad_norm": 1.3872442245483398,
      "learning_rate": 1.1579106719895205e-05,
      "loss": 1.4656,
      "step": 23733
    },
    {
      "epoch": 0.8499650831736709,
      "grad_norm": 2.849210262298584,
      "learning_rate": 1.157368951318687e-05,
      "loss": 1.7573,
      "step": 23734
    },
    {
      "epoch": 0.8500008953032392,
      "grad_norm": 1.7254695892333984,
      "learning_rate": 1.1568273496140513e-05,
      "loss": 1.4489,
      "step": 23735
    },
    {
      "epoch": 0.8500367074328075,
      "grad_norm": 1.4236689805984497,
      "learning_rate": 1.1562858668828936e-05,
      "loss": 1.6148,
      "step": 23736
    },
    {
      "epoch": 0.8500725195623757,
      "grad_norm": 1.5997804403305054,
      "learning_rate": 1.1557445031325032e-05,
      "loss": 1.3078,
      "step": 23737
    },
    {
      "epoch": 0.8501083316919441,
      "grad_norm": 2.8871943950653076,
      "learning_rate": 1.1552032583701612e-05,
      "loss": 1.547,
      "step": 23738
    },
    {
      "epoch": 0.8501441438215124,
      "grad_norm": 1.5523663759231567,
      "learning_rate": 1.1546621326031526e-05,
      "loss": 1.407,
      "step": 23739
    },
    {
      "epoch": 0.8501799559510806,
      "grad_norm": 1.4825963973999023,
      "learning_rate": 1.154121125838754e-05,
      "loss": 1.4152,
      "step": 23740
    },
    {
      "epoch": 0.8502157680806489,
      "grad_norm": 1.7809697389602661,
      "learning_rate": 1.1535802380842453e-05,
      "loss": 1.2848,
      "step": 23741
    },
    {
      "epoch": 0.8502515802102172,
      "grad_norm": 1.843872308731079,
      "learning_rate": 1.1530394693469026e-05,
      "loss": 1.4454,
      "step": 23742
    },
    {
      "epoch": 0.8502873923397855,
      "grad_norm": 1.6552728414535522,
      "learning_rate": 1.1524988196340048e-05,
      "loss": 1.4238,
      "step": 23743
    },
    {
      "epoch": 0.8503232044693537,
      "grad_norm": 1.1542280912399292,
      "learning_rate": 1.1519582889528202e-05,
      "loss": 1.3864,
      "step": 23744
    },
    {
      "epoch": 0.8503590165989221,
      "grad_norm": 1.427795648574829,
      "learning_rate": 1.1514178773106243e-05,
      "loss": 1.2635,
      "step": 23745
    },
    {
      "epoch": 0.8503948287284904,
      "grad_norm": 1.981183409690857,
      "learning_rate": 1.150877584714689e-05,
      "loss": 1.3967,
      "step": 23746
    },
    {
      "epoch": 0.8504306408580586,
      "grad_norm": 1.5299640893936157,
      "learning_rate": 1.1503374111722786e-05,
      "loss": 1.6025,
      "step": 23747
    },
    {
      "epoch": 0.8504664529876269,
      "grad_norm": 1.5333250761032104,
      "learning_rate": 1.149797356690664e-05,
      "loss": 1.37,
      "step": 23748
    },
    {
      "epoch": 0.8505022651171952,
      "grad_norm": 1.6742618083953857,
      "learning_rate": 1.149257421277109e-05,
      "loss": 1.4453,
      "step": 23749
    },
    {
      "epoch": 0.8505380772467634,
      "grad_norm": 1.8862006664276123,
      "learning_rate": 1.1487176049388814e-05,
      "loss": 1.4248,
      "step": 23750
    },
    {
      "epoch": 0.8505738893763317,
      "grad_norm": 1.4315221309661865,
      "learning_rate": 1.1481779076832388e-05,
      "loss": 1.3406,
      "step": 23751
    },
    {
      "epoch": 0.8506097015059001,
      "grad_norm": 1.759994387626648,
      "learning_rate": 1.1476383295174452e-05,
      "loss": 1.6171,
      "step": 23752
    },
    {
      "epoch": 0.8506455136354684,
      "grad_norm": 2.4337728023529053,
      "learning_rate": 1.1470988704487607e-05,
      "loss": 1.2688,
      "step": 23753
    },
    {
      "epoch": 0.8506813257650366,
      "grad_norm": 1.9323700666427612,
      "learning_rate": 1.146559530484439e-05,
      "loss": 1.4573,
      "step": 23754
    },
    {
      "epoch": 0.8507171378946049,
      "grad_norm": 1.7567484378814697,
      "learning_rate": 1.146020309631739e-05,
      "loss": 1.5855,
      "step": 23755
    },
    {
      "epoch": 0.8507529500241732,
      "grad_norm": 1.5216503143310547,
      "learning_rate": 1.145481207897915e-05,
      "loss": 1.3883,
      "step": 23756
    },
    {
      "epoch": 0.8507887621537414,
      "grad_norm": 1.536760926246643,
      "learning_rate": 1.144942225290222e-05,
      "loss": 1.4571,
      "step": 23757
    },
    {
      "epoch": 0.8508245742833097,
      "grad_norm": 2.1241402626037598,
      "learning_rate": 1.1444033618159068e-05,
      "loss": 1.9264,
      "step": 23758
    },
    {
      "epoch": 0.8508603864128781,
      "grad_norm": 1.6003999710083008,
      "learning_rate": 1.143864617482222e-05,
      "loss": 1.7025,
      "step": 23759
    },
    {
      "epoch": 0.8508961985424464,
      "grad_norm": 1.6267139911651611,
      "learning_rate": 1.1433259922964146e-05,
      "loss": 1.2968,
      "step": 23760
    },
    {
      "epoch": 0.8509320106720146,
      "grad_norm": 1.788483738899231,
      "learning_rate": 1.1427874862657339e-05,
      "loss": 1.4734,
      "step": 23761
    },
    {
      "epoch": 0.8509678228015829,
      "grad_norm": 1.3906162977218628,
      "learning_rate": 1.1422490993974199e-05,
      "loss": 1.0226,
      "step": 23762
    },
    {
      "epoch": 0.8510036349311512,
      "grad_norm": 2.156611204147339,
      "learning_rate": 1.1417108316987201e-05,
      "loss": 1.4735,
      "step": 23763
    },
    {
      "epoch": 0.8510394470607194,
      "grad_norm": 1.2706553936004639,
      "learning_rate": 1.1411726831768754e-05,
      "loss": 1.6045,
      "step": 23764
    },
    {
      "epoch": 0.8510752591902877,
      "grad_norm": 1.6218516826629639,
      "learning_rate": 1.1406346538391243e-05,
      "loss": 1.5444,
      "step": 23765
    },
    {
      "epoch": 0.8511110713198561,
      "grad_norm": 2.312868118286133,
      "learning_rate": 1.1400967436927056e-05,
      "loss": 1.3219,
      "step": 23766
    },
    {
      "epoch": 0.8511468834494244,
      "grad_norm": 1.7224329710006714,
      "learning_rate": 1.1395589527448558e-05,
      "loss": 1.5206,
      "step": 23767
    },
    {
      "epoch": 0.8511826955789926,
      "grad_norm": 1.7545734643936157,
      "learning_rate": 1.1390212810028144e-05,
      "loss": 1.8438,
      "step": 23768
    },
    {
      "epoch": 0.8512185077085609,
      "grad_norm": 1.76412034034729,
      "learning_rate": 1.1384837284738114e-05,
      "loss": 1.1991,
      "step": 23769
    },
    {
      "epoch": 0.8512543198381292,
      "grad_norm": 1.5692112445831299,
      "learning_rate": 1.1379462951650755e-05,
      "loss": 1.4919,
      "step": 23770
    },
    {
      "epoch": 0.8512901319676974,
      "grad_norm": 1.500417947769165,
      "learning_rate": 1.137408981083845e-05,
      "loss": 1.0246,
      "step": 23771
    },
    {
      "epoch": 0.8513259440972657,
      "grad_norm": 2.5152666568756104,
      "learning_rate": 1.1368717862373424e-05,
      "loss": 1.4325,
      "step": 23772
    },
    {
      "epoch": 0.8513617562268341,
      "grad_norm": 1.9641667604446411,
      "learning_rate": 1.136334710632797e-05,
      "loss": 1.3025,
      "step": 23773
    },
    {
      "epoch": 0.8513975683564023,
      "grad_norm": 1.6455731391906738,
      "learning_rate": 1.1357977542774356e-05,
      "loss": 1.2259,
      "step": 23774
    },
    {
      "epoch": 0.8514333804859706,
      "grad_norm": 2.0032389163970947,
      "learning_rate": 1.1352609171784834e-05,
      "loss": 1.2823,
      "step": 23775
    },
    {
      "epoch": 0.8514691926155389,
      "grad_norm": 1.3516676425933838,
      "learning_rate": 1.1347241993431578e-05,
      "loss": 1.4081,
      "step": 23776
    },
    {
      "epoch": 0.8515050047451072,
      "grad_norm": 2.0282421112060547,
      "learning_rate": 1.1341876007786845e-05,
      "loss": 1.5955,
      "step": 23777
    },
    {
      "epoch": 0.8515408168746754,
      "grad_norm": 2.355489492416382,
      "learning_rate": 1.1336511214922819e-05,
      "loss": 1.2484,
      "step": 23778
    },
    {
      "epoch": 0.8515766290042437,
      "grad_norm": 1.3258121013641357,
      "learning_rate": 1.1331147614911641e-05,
      "loss": 1.5587,
      "step": 23779
    },
    {
      "epoch": 0.8516124411338121,
      "grad_norm": 1.586104154586792,
      "learning_rate": 1.1325785207825524e-05,
      "loss": 1.3955,
      "step": 23780
    },
    {
      "epoch": 0.8516482532633803,
      "grad_norm": 1.611124038696289,
      "learning_rate": 1.132042399373654e-05,
      "loss": 1.3186,
      "step": 23781
    },
    {
      "epoch": 0.8516840653929486,
      "grad_norm": 1.5817452669143677,
      "learning_rate": 1.131506397271691e-05,
      "loss": 1.7507,
      "step": 23782
    },
    {
      "epoch": 0.8517198775225169,
      "grad_norm": 1.5858224630355835,
      "learning_rate": 1.1309705144838678e-05,
      "loss": 1.5766,
      "step": 23783
    },
    {
      "epoch": 0.8517556896520851,
      "grad_norm": 2.0559260845184326,
      "learning_rate": 1.1304347510173963e-05,
      "loss": 1.5677,
      "step": 23784
    },
    {
      "epoch": 0.8517915017816534,
      "grad_norm": 1.479744791984558,
      "learning_rate": 1.129899106879484e-05,
      "loss": 1.4073,
      "step": 23785
    },
    {
      "epoch": 0.8518273139112217,
      "grad_norm": 1.3566659688949585,
      "learning_rate": 1.1293635820773397e-05,
      "loss": 1.2683,
      "step": 23786
    },
    {
      "epoch": 0.8518631260407901,
      "grad_norm": 1.4227968454360962,
      "learning_rate": 1.1288281766181651e-05,
      "loss": 1.4598,
      "step": 23787
    },
    {
      "epoch": 0.8518989381703583,
      "grad_norm": 1.8010733127593994,
      "learning_rate": 1.1282928905091616e-05,
      "loss": 1.268,
      "step": 23788
    },
    {
      "epoch": 0.8519347502999266,
      "grad_norm": 1.2809944152832031,
      "learning_rate": 1.1277577237575377e-05,
      "loss": 1.6804,
      "step": 23789
    },
    {
      "epoch": 0.8519705624294949,
      "grad_norm": 1.514971375465393,
      "learning_rate": 1.1272226763704863e-05,
      "loss": 1.7797,
      "step": 23790
    },
    {
      "epoch": 0.8520063745590631,
      "grad_norm": 2.1083321571350098,
      "learning_rate": 1.1266877483552118e-05,
      "loss": 1.374,
      "step": 23791
    },
    {
      "epoch": 0.8520421866886314,
      "grad_norm": 1.2880560159683228,
      "learning_rate": 1.126152939718903e-05,
      "loss": 1.582,
      "step": 23792
    },
    {
      "epoch": 0.8520779988181997,
      "grad_norm": 1.4016788005828857,
      "learning_rate": 1.125618250468764e-05,
      "loss": 1.5074,
      "step": 23793
    },
    {
      "epoch": 0.8521138109477681,
      "grad_norm": 1.4981015920639038,
      "learning_rate": 1.1250836806119824e-05,
      "loss": 1.3985,
      "step": 23794
    },
    {
      "epoch": 0.8521496230773363,
      "grad_norm": 1.716374397277832,
      "learning_rate": 1.1245492301557547e-05,
      "loss": 1.3592,
      "step": 23795
    },
    {
      "epoch": 0.8521854352069046,
      "grad_norm": 1.9721206426620483,
      "learning_rate": 1.1240148991072662e-05,
      "loss": 1.3089,
      "step": 23796
    },
    {
      "epoch": 0.8522212473364729,
      "grad_norm": 1.718814492225647,
      "learning_rate": 1.123480687473708e-05,
      "loss": 1.5038,
      "step": 23797
    },
    {
      "epoch": 0.8522570594660411,
      "grad_norm": 2.706646203994751,
      "learning_rate": 1.1229465952622686e-05,
      "loss": 1.4359,
      "step": 23798
    },
    {
      "epoch": 0.8522928715956094,
      "grad_norm": 1.659623622894287,
      "learning_rate": 1.122412622480129e-05,
      "loss": 1.4396,
      "step": 23799
    },
    {
      "epoch": 0.8523286837251777,
      "grad_norm": 1.6759545803070068,
      "learning_rate": 1.1218787691344801e-05,
      "loss": 1.6162,
      "step": 23800
    },
    {
      "epoch": 0.852364495854746,
      "grad_norm": 2.4348177909851074,
      "learning_rate": 1.1213450352324983e-05,
      "loss": 1.8575,
      "step": 23801
    },
    {
      "epoch": 0.8524003079843143,
      "grad_norm": 1.6588647365570068,
      "learning_rate": 1.1208114207813691e-05,
      "loss": 1.6102,
      "step": 23802
    },
    {
      "epoch": 0.8524361201138826,
      "grad_norm": 1.6332166194915771,
      "learning_rate": 1.1202779257882645e-05,
      "loss": 1.357,
      "step": 23803
    },
    {
      "epoch": 0.8524719322434509,
      "grad_norm": 1.6586291790008545,
      "learning_rate": 1.1197445502603698e-05,
      "loss": 1.4706,
      "step": 23804
    },
    {
      "epoch": 0.8525077443730191,
      "grad_norm": 1.4110736846923828,
      "learning_rate": 1.1192112942048582e-05,
      "loss": 1.311,
      "step": 23805
    },
    {
      "epoch": 0.8525435565025874,
      "grad_norm": 1.5690315961837769,
      "learning_rate": 1.1186781576289007e-05,
      "loss": 1.3135,
      "step": 23806
    },
    {
      "epoch": 0.8525793686321557,
      "grad_norm": 1.8064452409744263,
      "learning_rate": 1.1181451405396725e-05,
      "loss": 1.6223,
      "step": 23807
    },
    {
      "epoch": 0.852615180761724,
      "grad_norm": 1.4428201913833618,
      "learning_rate": 1.1176122429443458e-05,
      "loss": 1.4009,
      "step": 23808
    },
    {
      "epoch": 0.8526509928912923,
      "grad_norm": 1.7909198999404907,
      "learning_rate": 1.1170794648500893e-05,
      "loss": 1.3315,
      "step": 23809
    },
    {
      "epoch": 0.8526868050208606,
      "grad_norm": 1.5551303625106812,
      "learning_rate": 1.116546806264067e-05,
      "loss": 1.4318,
      "step": 23810
    },
    {
      "epoch": 0.8527226171504289,
      "grad_norm": 1.7789667844772339,
      "learning_rate": 1.1160142671934537e-05,
      "loss": 1.3732,
      "step": 23811
    },
    {
      "epoch": 0.8527584292799971,
      "grad_norm": 1.6274534463882446,
      "learning_rate": 1.1154818476454054e-05,
      "loss": 1.4462,
      "step": 23812
    },
    {
      "epoch": 0.8527942414095654,
      "grad_norm": 1.3836264610290527,
      "learning_rate": 1.114949547627091e-05,
      "loss": 1.5112,
      "step": 23813
    },
    {
      "epoch": 0.8528300535391337,
      "grad_norm": 1.9895684719085693,
      "learning_rate": 1.1144173671456682e-05,
      "loss": 1.3913,
      "step": 23814
    },
    {
      "epoch": 0.852865865668702,
      "grad_norm": 2.5720231533050537,
      "learning_rate": 1.1138853062082977e-05,
      "loss": 1.4404,
      "step": 23815
    },
    {
      "epoch": 0.8529016777982703,
      "grad_norm": 1.6841319799423218,
      "learning_rate": 1.1133533648221405e-05,
      "loss": 1.2546,
      "step": 23816
    },
    {
      "epoch": 0.8529374899278386,
      "grad_norm": 1.8937181234359741,
      "learning_rate": 1.1128215429943477e-05,
      "loss": 1.3837,
      "step": 23817
    },
    {
      "epoch": 0.8529733020574068,
      "grad_norm": 1.6849440336227417,
      "learning_rate": 1.1122898407320791e-05,
      "loss": 1.3869,
      "step": 23818
    },
    {
      "epoch": 0.8530091141869751,
      "grad_norm": 1.8310136795043945,
      "learning_rate": 1.1117582580424857e-05,
      "loss": 1.2157,
      "step": 23819
    },
    {
      "epoch": 0.8530449263165434,
      "grad_norm": 1.718644380569458,
      "learning_rate": 1.1112267949327216e-05,
      "loss": 1.3847,
      "step": 23820
    },
    {
      "epoch": 0.8530807384461117,
      "grad_norm": 1.7329232692718506,
      "learning_rate": 1.1106954514099332e-05,
      "loss": 1.3567,
      "step": 23821
    },
    {
      "epoch": 0.85311655057568,
      "grad_norm": 1.7883803844451904,
      "learning_rate": 1.1101642274812706e-05,
      "loss": 1.5081,
      "step": 23822
    },
    {
      "epoch": 0.8531523627052483,
      "grad_norm": 1.6889233589172363,
      "learning_rate": 1.1096331231538847e-05,
      "loss": 1.2767,
      "step": 23823
    },
    {
      "epoch": 0.8531881748348166,
      "grad_norm": 1.7603107690811157,
      "learning_rate": 1.1091021384349143e-05,
      "loss": 1.3097,
      "step": 23824
    },
    {
      "epoch": 0.8532239869643848,
      "grad_norm": 1.8802781105041504,
      "learning_rate": 1.1085712733315068e-05,
      "loss": 1.3812,
      "step": 23825
    },
    {
      "epoch": 0.8532597990939531,
      "grad_norm": 1.5397545099258423,
      "learning_rate": 1.1080405278508033e-05,
      "loss": 1.4916,
      "step": 23826
    },
    {
      "epoch": 0.8532956112235214,
      "grad_norm": 1.7694755792617798,
      "learning_rate": 1.1075099019999468e-05,
      "loss": 1.6139,
      "step": 23827
    },
    {
      "epoch": 0.8533314233530896,
      "grad_norm": 1.587158441543579,
      "learning_rate": 1.106979395786072e-05,
      "loss": 1.7875,
      "step": 23828
    },
    {
      "epoch": 0.853367235482658,
      "grad_norm": 1.5730561017990112,
      "learning_rate": 1.1064490092163181e-05,
      "loss": 1.2947,
      "step": 23829
    },
    {
      "epoch": 0.8534030476122263,
      "grad_norm": 1.4817062616348267,
      "learning_rate": 1.1059187422978211e-05,
      "loss": 1.5571,
      "step": 23830
    },
    {
      "epoch": 0.8534388597417946,
      "grad_norm": 2.2228071689605713,
      "learning_rate": 1.1053885950377174e-05,
      "loss": 1.4434,
      "step": 23831
    },
    {
      "epoch": 0.8534746718713628,
      "grad_norm": 1.7988253831863403,
      "learning_rate": 1.1048585674431345e-05,
      "loss": 1.5946,
      "step": 23832
    },
    {
      "epoch": 0.8535104840009311,
      "grad_norm": 1.7007036209106445,
      "learning_rate": 1.1043286595212054e-05,
      "loss": 1.3761,
      "step": 23833
    },
    {
      "epoch": 0.8535462961304994,
      "grad_norm": 1.7106313705444336,
      "learning_rate": 1.1037988712790626e-05,
      "loss": 1.3455,
      "step": 23834
    },
    {
      "epoch": 0.8535821082600676,
      "grad_norm": 1.68916916847229,
      "learning_rate": 1.1032692027238279e-05,
      "loss": 1.4494,
      "step": 23835
    },
    {
      "epoch": 0.853617920389636,
      "grad_norm": 1.5535471439361572,
      "learning_rate": 1.10273965386263e-05,
      "loss": 1.4831,
      "step": 23836
    },
    {
      "epoch": 0.8536537325192043,
      "grad_norm": 1.5345951318740845,
      "learning_rate": 1.1022102247025934e-05,
      "loss": 1.6866,
      "step": 23837
    },
    {
      "epoch": 0.8536895446487726,
      "grad_norm": 1.2427003383636475,
      "learning_rate": 1.1016809152508434e-05,
      "loss": 1.3729,
      "step": 23838
    },
    {
      "epoch": 0.8537253567783408,
      "grad_norm": 1.175044298171997,
      "learning_rate": 1.1011517255144965e-05,
      "loss": 1.3661,
      "step": 23839
    },
    {
      "epoch": 0.8537611689079091,
      "grad_norm": 1.9435486793518066,
      "learning_rate": 1.1006226555006749e-05,
      "loss": 1.1561,
      "step": 23840
    },
    {
      "epoch": 0.8537969810374774,
      "grad_norm": 1.4957728385925293,
      "learning_rate": 1.1000937052164973e-05,
      "loss": 1.2308,
      "step": 23841
    },
    {
      "epoch": 0.8538327931670456,
      "grad_norm": 1.851515293121338,
      "learning_rate": 1.0995648746690768e-05,
      "loss": 1.1669,
      "step": 23842
    },
    {
      "epoch": 0.853868605296614,
      "grad_norm": 1.4531022310256958,
      "learning_rate": 1.0990361638655311e-05,
      "loss": 1.3593,
      "step": 23843
    },
    {
      "epoch": 0.8539044174261823,
      "grad_norm": 1.6683907508850098,
      "learning_rate": 1.0985075728129712e-05,
      "loss": 1.413,
      "step": 23844
    },
    {
      "epoch": 0.8539402295557506,
      "grad_norm": 2.5751333236694336,
      "learning_rate": 1.0979791015185125e-05,
      "loss": 1.5377,
      "step": 23845
    },
    {
      "epoch": 0.8539760416853188,
      "grad_norm": 1.8338749408721924,
      "learning_rate": 1.0974507499892605e-05,
      "loss": 1.6186,
      "step": 23846
    },
    {
      "epoch": 0.8540118538148871,
      "grad_norm": 1.6571879386901855,
      "learning_rate": 1.0969225182323239e-05,
      "loss": 1.3968,
      "step": 23847
    },
    {
      "epoch": 0.8540476659444554,
      "grad_norm": 1.5902742147445679,
      "learning_rate": 1.0963944062548125e-05,
      "loss": 1.466,
      "step": 23848
    },
    {
      "epoch": 0.8540834780740236,
      "grad_norm": 1.432607650756836,
      "learning_rate": 1.0958664140638297e-05,
      "loss": 1.4103,
      "step": 23849
    },
    {
      "epoch": 0.854119290203592,
      "grad_norm": 1.4626818895339966,
      "learning_rate": 1.0953385416664785e-05,
      "loss": 1.0073,
      "step": 23850
    },
    {
      "epoch": 0.8541551023331603,
      "grad_norm": 2.2217087745666504,
      "learning_rate": 1.09481078906986e-05,
      "loss": 1.4524,
      "step": 23851
    },
    {
      "epoch": 0.8541909144627285,
      "grad_norm": 1.7042936086654663,
      "learning_rate": 1.0942831562810774e-05,
      "loss": 1.5566,
      "step": 23852
    },
    {
      "epoch": 0.8542267265922968,
      "grad_norm": 1.6705306768417358,
      "learning_rate": 1.093755643307226e-05,
      "loss": 1.3809,
      "step": 23853
    },
    {
      "epoch": 0.8542625387218651,
      "grad_norm": 2.15169620513916,
      "learning_rate": 1.0932282501554037e-05,
      "loss": 1.3918,
      "step": 23854
    },
    {
      "epoch": 0.8542983508514334,
      "grad_norm": 1.519851565361023,
      "learning_rate": 1.0927009768327068e-05,
      "loss": 1.2011,
      "step": 23855
    },
    {
      "epoch": 0.8543341629810016,
      "grad_norm": 1.5760105848312378,
      "learning_rate": 1.0921738233462297e-05,
      "loss": 1.5071,
      "step": 23856
    },
    {
      "epoch": 0.85436997511057,
      "grad_norm": 1.225251317024231,
      "learning_rate": 1.0916467897030625e-05,
      "loss": 1.1271,
      "step": 23857
    },
    {
      "epoch": 0.8544057872401383,
      "grad_norm": 1.4910647869110107,
      "learning_rate": 1.091119875910297e-05,
      "loss": 1.2668,
      "step": 23858
    },
    {
      "epoch": 0.8544415993697065,
      "grad_norm": 1.352760910987854,
      "learning_rate": 1.0905930819750232e-05,
      "loss": 1.1062,
      "step": 23859
    },
    {
      "epoch": 0.8544774114992748,
      "grad_norm": 1.927417278289795,
      "learning_rate": 1.0900664079043255e-05,
      "loss": 1.2948,
      "step": 23860
    },
    {
      "epoch": 0.8545132236288431,
      "grad_norm": 2.803440570831299,
      "learning_rate": 1.0895398537052914e-05,
      "loss": 1.2944,
      "step": 23861
    },
    {
      "epoch": 0.8545490357584113,
      "grad_norm": 1.5908805131912231,
      "learning_rate": 1.0890134193850043e-05,
      "loss": 1.4876,
      "step": 23862
    },
    {
      "epoch": 0.8545848478879796,
      "grad_norm": 1.509763240814209,
      "learning_rate": 1.0884871049505507e-05,
      "loss": 1.4407,
      "step": 23863
    },
    {
      "epoch": 0.854620660017548,
      "grad_norm": 1.424763560295105,
      "learning_rate": 1.0879609104090049e-05,
      "loss": 1.6699,
      "step": 23864
    },
    {
      "epoch": 0.8546564721471163,
      "grad_norm": 1.295540452003479,
      "learning_rate": 1.0874348357674492e-05,
      "loss": 1.2205,
      "step": 23865
    },
    {
      "epoch": 0.8546922842766845,
      "grad_norm": 1.5044755935668945,
      "learning_rate": 1.0869088810329642e-05,
      "loss": 1.3277,
      "step": 23866
    },
    {
      "epoch": 0.8547280964062528,
      "grad_norm": 1.0691648721694946,
      "learning_rate": 1.0863830462126202e-05,
      "loss": 1.3522,
      "step": 23867
    },
    {
      "epoch": 0.8547639085358211,
      "grad_norm": 1.2039045095443726,
      "learning_rate": 1.085857331313498e-05,
      "loss": 1.4813,
      "step": 23868
    },
    {
      "epoch": 0.8547997206653893,
      "grad_norm": 1.421851396560669,
      "learning_rate": 1.0853317363426618e-05,
      "loss": 1.3372,
      "step": 23869
    },
    {
      "epoch": 0.8548355327949576,
      "grad_norm": 1.878260612487793,
      "learning_rate": 1.0848062613071918e-05,
      "loss": 1.6804,
      "step": 23870
    },
    {
      "epoch": 0.854871344924526,
      "grad_norm": 1.3401705026626587,
      "learning_rate": 1.0842809062141524e-05,
      "loss": 1.519,
      "step": 23871
    },
    {
      "epoch": 0.8549071570540943,
      "grad_norm": 1.784934163093567,
      "learning_rate": 1.083755671070613e-05,
      "loss": 1.5975,
      "step": 23872
    },
    {
      "epoch": 0.8549429691836625,
      "grad_norm": 1.439996361732483,
      "learning_rate": 1.0832305558836397e-05,
      "loss": 1.2901,
      "step": 23873
    },
    {
      "epoch": 0.8549787813132308,
      "grad_norm": 2.1497843265533447,
      "learning_rate": 1.0827055606602998e-05,
      "loss": 1.6519,
      "step": 23874
    },
    {
      "epoch": 0.8550145934427991,
      "grad_norm": 1.6868973970413208,
      "learning_rate": 1.0821806854076533e-05,
      "loss": 1.6856,
      "step": 23875
    },
    {
      "epoch": 0.8550504055723673,
      "grad_norm": 1.7532434463500977,
      "learning_rate": 1.0816559301327589e-05,
      "loss": 1.3696,
      "step": 23876
    },
    {
      "epoch": 0.8550862177019356,
      "grad_norm": 1.7454650402069092,
      "learning_rate": 1.0811312948426844e-05,
      "loss": 1.4143,
      "step": 23877
    },
    {
      "epoch": 0.855122029831504,
      "grad_norm": 1.5443825721740723,
      "learning_rate": 1.0806067795444818e-05,
      "loss": 1.7368,
      "step": 23878
    },
    {
      "epoch": 0.8551578419610723,
      "grad_norm": 1.5286147594451904,
      "learning_rate": 1.0800823842452113e-05,
      "loss": 1.492,
      "step": 23879
    },
    {
      "epoch": 0.8551936540906405,
      "grad_norm": 2.415445566177368,
      "learning_rate": 1.0795581089519236e-05,
      "loss": 1.6911,
      "step": 23880
    },
    {
      "epoch": 0.8552294662202088,
      "grad_norm": 2.523890733718872,
      "learning_rate": 1.0790339536716776e-05,
      "loss": 1.4355,
      "step": 23881
    },
    {
      "epoch": 0.8552652783497771,
      "grad_norm": 2.055680751800537,
      "learning_rate": 1.078509918411521e-05,
      "loss": 1.2303,
      "step": 23882
    },
    {
      "epoch": 0.8553010904793453,
      "grad_norm": 1.5757873058319092,
      "learning_rate": 1.0779860031785061e-05,
      "loss": 1.5363,
      "step": 23883
    },
    {
      "epoch": 0.8553369026089136,
      "grad_norm": 1.436819314956665,
      "learning_rate": 1.0774622079796826e-05,
      "loss": 1.586,
      "step": 23884
    },
    {
      "epoch": 0.855372714738482,
      "grad_norm": 2.188523769378662,
      "learning_rate": 1.0769385328220938e-05,
      "loss": 1.1043,
      "step": 23885
    },
    {
      "epoch": 0.8554085268680502,
      "grad_norm": 1.9349035024642944,
      "learning_rate": 1.0764149777127897e-05,
      "loss": 1.5145,
      "step": 23886
    },
    {
      "epoch": 0.8554443389976185,
      "grad_norm": 1.4277589321136475,
      "learning_rate": 1.0758915426588068e-05,
      "loss": 1.4414,
      "step": 23887
    },
    {
      "epoch": 0.8554801511271868,
      "grad_norm": 1.9857627153396606,
      "learning_rate": 1.0753682276671961e-05,
      "loss": 1.3229,
      "step": 23888
    },
    {
      "epoch": 0.855515963256755,
      "grad_norm": 1.2367472648620605,
      "learning_rate": 1.074845032744991e-05,
      "loss": 1.3326,
      "step": 23889
    },
    {
      "epoch": 0.8555517753863233,
      "grad_norm": 1.7371845245361328,
      "learning_rate": 1.0743219578992369e-05,
      "loss": 1.3939,
      "step": 23890
    },
    {
      "epoch": 0.8555875875158916,
      "grad_norm": 1.5299841165542603,
      "learning_rate": 1.0737990031369627e-05,
      "loss": 1.4069,
      "step": 23891
    },
    {
      "epoch": 0.85562339964546,
      "grad_norm": 3.103381395339966,
      "learning_rate": 1.0732761684652127e-05,
      "loss": 1.3743,
      "step": 23892
    },
    {
      "epoch": 0.8556592117750282,
      "grad_norm": 1.619565486907959,
      "learning_rate": 1.0727534538910177e-05,
      "loss": 1.7073,
      "step": 23893
    },
    {
      "epoch": 0.8556950239045965,
      "grad_norm": 1.3993152379989624,
      "learning_rate": 1.0722308594214081e-05,
      "loss": 1.3665,
      "step": 23894
    },
    {
      "epoch": 0.8557308360341648,
      "grad_norm": 1.5730746984481812,
      "learning_rate": 1.0717083850634158e-05,
      "loss": 1.4241,
      "step": 23895
    },
    {
      "epoch": 0.855766648163733,
      "grad_norm": 1.6564350128173828,
      "learning_rate": 1.0711860308240706e-05,
      "loss": 1.5138,
      "step": 23896
    },
    {
      "epoch": 0.8558024602933013,
      "grad_norm": 1.9914555549621582,
      "learning_rate": 1.0706637967104016e-05,
      "loss": 1.4409,
      "step": 23897
    },
    {
      "epoch": 0.8558382724228696,
      "grad_norm": 1.646231770515442,
      "learning_rate": 1.0701416827294297e-05,
      "loss": 1.4735,
      "step": 23898
    },
    {
      "epoch": 0.855874084552438,
      "grad_norm": 1.485743522644043,
      "learning_rate": 1.069619688888187e-05,
      "loss": 1.4125,
      "step": 23899
    },
    {
      "epoch": 0.8559098966820062,
      "grad_norm": 1.75136399269104,
      "learning_rate": 1.0690978151936892e-05,
      "loss": 1.4939,
      "step": 23900
    },
    {
      "epoch": 0.8559457088115745,
      "grad_norm": 1.9949253797531128,
      "learning_rate": 1.0685760616529628e-05,
      "loss": 1.5425,
      "step": 23901
    },
    {
      "epoch": 0.8559815209411428,
      "grad_norm": 1.7424404621124268,
      "learning_rate": 1.068054428273022e-05,
      "loss": 1.4624,
      "step": 23902
    },
    {
      "epoch": 0.856017333070711,
      "grad_norm": 1.6785800457000732,
      "learning_rate": 1.0675329150608892e-05,
      "loss": 1.4386,
      "step": 23903
    },
    {
      "epoch": 0.8560531452002793,
      "grad_norm": 1.4667755365371704,
      "learning_rate": 1.0670115220235799e-05,
      "loss": 1.3168,
      "step": 23904
    },
    {
      "epoch": 0.8560889573298476,
      "grad_norm": 1.4324119091033936,
      "learning_rate": 1.0664902491681051e-05,
      "loss": 1.4958,
      "step": 23905
    },
    {
      "epoch": 0.856124769459416,
      "grad_norm": 3.299861192703247,
      "learning_rate": 1.0659690965014813e-05,
      "loss": 1.3856,
      "step": 23906
    },
    {
      "epoch": 0.8561605815889842,
      "grad_norm": 1.4514801502227783,
      "learning_rate": 1.0654480640307195e-05,
      "loss": 1.5554,
      "step": 23907
    },
    {
      "epoch": 0.8561963937185525,
      "grad_norm": 1.6911697387695312,
      "learning_rate": 1.0649271517628313e-05,
      "loss": 1.5904,
      "step": 23908
    },
    {
      "epoch": 0.8562322058481208,
      "grad_norm": 2.0003654956817627,
      "learning_rate": 1.0644063597048182e-05,
      "loss": 1.2916,
      "step": 23909
    },
    {
      "epoch": 0.856268017977689,
      "grad_norm": 2.0778403282165527,
      "learning_rate": 1.063885687863696e-05,
      "loss": 1.2572,
      "step": 23910
    },
    {
      "epoch": 0.8563038301072573,
      "grad_norm": 1.5954288244247437,
      "learning_rate": 1.0633651362464647e-05,
      "loss": 1.4367,
      "step": 23911
    },
    {
      "epoch": 0.8563396422368256,
      "grad_norm": 1.7671774625778198,
      "learning_rate": 1.0628447048601265e-05,
      "loss": 1.7623,
      "step": 23912
    },
    {
      "epoch": 0.856375454366394,
      "grad_norm": 1.8786360025405884,
      "learning_rate": 1.0623243937116845e-05,
      "loss": 1.3735,
      "step": 23913
    },
    {
      "epoch": 0.8564112664959622,
      "grad_norm": 1.2363111972808838,
      "learning_rate": 1.06180420280814e-05,
      "loss": 1.4896,
      "step": 23914
    },
    {
      "epoch": 0.8564470786255305,
      "grad_norm": 1.6070228815078735,
      "learning_rate": 1.0612841321564915e-05,
      "loss": 1.3349,
      "step": 23915
    },
    {
      "epoch": 0.8564828907550988,
      "grad_norm": 1.6431260108947754,
      "learning_rate": 1.0607641817637326e-05,
      "loss": 1.4696,
      "step": 23916
    },
    {
      "epoch": 0.856518702884667,
      "grad_norm": 2.0966637134552,
      "learning_rate": 1.060244351636861e-05,
      "loss": 1.2884,
      "step": 23917
    },
    {
      "epoch": 0.8565545150142353,
      "grad_norm": 1.5325583219528198,
      "learning_rate": 1.0597246417828698e-05,
      "loss": 1.2232,
      "step": 23918
    },
    {
      "epoch": 0.8565903271438036,
      "grad_norm": 2.3397915363311768,
      "learning_rate": 1.0592050522087549e-05,
      "loss": 1.5092,
      "step": 23919
    },
    {
      "epoch": 0.856626139273372,
      "grad_norm": 1.8545010089874268,
      "learning_rate": 1.0586855829215003e-05,
      "loss": 1.2814,
      "step": 23920
    },
    {
      "epoch": 0.8566619514029402,
      "grad_norm": 1.6768451929092407,
      "learning_rate": 1.0581662339280973e-05,
      "loss": 1.5897,
      "step": 23921
    },
    {
      "epoch": 0.8566977635325085,
      "grad_norm": 1.569795846939087,
      "learning_rate": 1.0576470052355358e-05,
      "loss": 1.4008,
      "step": 23922
    },
    {
      "epoch": 0.8567335756620768,
      "grad_norm": 1.2324440479278564,
      "learning_rate": 1.057127896850797e-05,
      "loss": 1.1082,
      "step": 23923
    },
    {
      "epoch": 0.856769387791645,
      "grad_norm": 2.4131672382354736,
      "learning_rate": 1.0566089087808672e-05,
      "loss": 1.3353,
      "step": 23924
    },
    {
      "epoch": 0.8568051999212133,
      "grad_norm": 1.468252420425415,
      "learning_rate": 1.056090041032729e-05,
      "loss": 1.425,
      "step": 23925
    },
    {
      "epoch": 0.8568410120507816,
      "grad_norm": 2.575080156326294,
      "learning_rate": 1.0555712936133633e-05,
      "loss": 1.446,
      "step": 23926
    },
    {
      "epoch": 0.8568768241803499,
      "grad_norm": 1.8932536840438843,
      "learning_rate": 1.0550526665297466e-05,
      "loss": 1.3337,
      "step": 23927
    },
    {
      "epoch": 0.8569126363099182,
      "grad_norm": 2.1886892318725586,
      "learning_rate": 1.0545341597888581e-05,
      "loss": 1.3273,
      "step": 23928
    },
    {
      "epoch": 0.8569484484394865,
      "grad_norm": 1.8264682292938232,
      "learning_rate": 1.0540157733976763e-05,
      "loss": 1.5032,
      "step": 23929
    },
    {
      "epoch": 0.8569842605690547,
      "grad_norm": 1.7340106964111328,
      "learning_rate": 1.0534975073631703e-05,
      "loss": 1.3628,
      "step": 23930
    },
    {
      "epoch": 0.857020072698623,
      "grad_norm": 1.687674641609192,
      "learning_rate": 1.0529793616923157e-05,
      "loss": 1.4686,
      "step": 23931
    },
    {
      "epoch": 0.8570558848281913,
      "grad_norm": 1.507163643836975,
      "learning_rate": 1.052461336392082e-05,
      "loss": 0.9268,
      "step": 23932
    },
    {
      "epoch": 0.8570916969577596,
      "grad_norm": 1.2720131874084473,
      "learning_rate": 1.0519434314694422e-05,
      "loss": 1.4392,
      "step": 23933
    },
    {
      "epoch": 0.8571275090873279,
      "grad_norm": 1.435158133506775,
      "learning_rate": 1.0514256469313588e-05,
      "loss": 1.6138,
      "step": 23934
    },
    {
      "epoch": 0.8571633212168962,
      "grad_norm": 2.085623025894165,
      "learning_rate": 1.0509079827848012e-05,
      "loss": 1.5755,
      "step": 23935
    },
    {
      "epoch": 0.8571991333464645,
      "grad_norm": 1.562156081199646,
      "learning_rate": 1.0503904390367325e-05,
      "loss": 1.3898,
      "step": 23936
    },
    {
      "epoch": 0.8572349454760327,
      "grad_norm": 2.3077609539031982,
      "learning_rate": 1.0498730156941184e-05,
      "loss": 1.6702,
      "step": 23937
    },
    {
      "epoch": 0.857270757605601,
      "grad_norm": 1.676255702972412,
      "learning_rate": 1.0493557127639164e-05,
      "loss": 1.3625,
      "step": 23938
    },
    {
      "epoch": 0.8573065697351693,
      "grad_norm": 1.866253137588501,
      "learning_rate": 1.0488385302530878e-05,
      "loss": 1.3196,
      "step": 23939
    },
    {
      "epoch": 0.8573423818647375,
      "grad_norm": 1.4433153867721558,
      "learning_rate": 1.0483214681685927e-05,
      "loss": 1.6465,
      "step": 23940
    },
    {
      "epoch": 0.8573781939943059,
      "grad_norm": 2.484351873397827,
      "learning_rate": 1.047804526517383e-05,
      "loss": 1.3115,
      "step": 23941
    },
    {
      "epoch": 0.8574140061238742,
      "grad_norm": 1.8021259307861328,
      "learning_rate": 1.0472877053064156e-05,
      "loss": 1.4364,
      "step": 23942
    },
    {
      "epoch": 0.8574498182534425,
      "grad_norm": 1.6426368951797485,
      "learning_rate": 1.0467710045426449e-05,
      "loss": 1.6388,
      "step": 23943
    },
    {
      "epoch": 0.8574856303830107,
      "grad_norm": 1.6121195554733276,
      "learning_rate": 1.046254424233023e-05,
      "loss": 1.1081,
      "step": 23944
    },
    {
      "epoch": 0.857521442512579,
      "grad_norm": 1.808910846710205,
      "learning_rate": 1.0457379643844966e-05,
      "loss": 1.5212,
      "step": 23945
    },
    {
      "epoch": 0.8575572546421473,
      "grad_norm": 1.7684128284454346,
      "learning_rate": 1.0452216250040148e-05,
      "loss": 1.5465,
      "step": 23946
    },
    {
      "epoch": 0.8575930667717155,
      "grad_norm": 1.6211570501327515,
      "learning_rate": 1.0447054060985284e-05,
      "loss": 1.3641,
      "step": 23947
    },
    {
      "epoch": 0.8576288789012839,
      "grad_norm": 1.535387396812439,
      "learning_rate": 1.0441893076749765e-05,
      "loss": 1.6225,
      "step": 23948
    },
    {
      "epoch": 0.8576646910308522,
      "grad_norm": 1.648389220237732,
      "learning_rate": 1.0436733297403056e-05,
      "loss": 1.1475,
      "step": 23949
    },
    {
      "epoch": 0.8577005031604205,
      "grad_norm": 1.5785759687423706,
      "learning_rate": 1.043157472301457e-05,
      "loss": 1.3503,
      "step": 23950
    },
    {
      "epoch": 0.8577363152899887,
      "grad_norm": 1.7339171171188354,
      "learning_rate": 1.0426417353653739e-05,
      "loss": 1.1188,
      "step": 23951
    },
    {
      "epoch": 0.857772127419557,
      "grad_norm": 1.6951241493225098,
      "learning_rate": 1.0421261189389885e-05,
      "loss": 1.3101,
      "step": 23952
    },
    {
      "epoch": 0.8578079395491253,
      "grad_norm": 2.160072088241577,
      "learning_rate": 1.0416106230292432e-05,
      "loss": 1.5236,
      "step": 23953
    },
    {
      "epoch": 0.8578437516786935,
      "grad_norm": 1.4381632804870605,
      "learning_rate": 1.0410952476430703e-05,
      "loss": 1.3926,
      "step": 23954
    },
    {
      "epoch": 0.8578795638082619,
      "grad_norm": 2.0479938983917236,
      "learning_rate": 1.0405799927874072e-05,
      "loss": 1.4182,
      "step": 23955
    },
    {
      "epoch": 0.8579153759378302,
      "grad_norm": 1.3435159921646118,
      "learning_rate": 1.0400648584691808e-05,
      "loss": 1.4873,
      "step": 23956
    },
    {
      "epoch": 0.8579511880673985,
      "grad_norm": 1.5888739824295044,
      "learning_rate": 1.0395498446953245e-05,
      "loss": 1.4416,
      "step": 23957
    },
    {
      "epoch": 0.8579870001969667,
      "grad_norm": 1.7520204782485962,
      "learning_rate": 1.0390349514727694e-05,
      "loss": 1.2782,
      "step": 23958
    },
    {
      "epoch": 0.858022812326535,
      "grad_norm": 1.682873249053955,
      "learning_rate": 1.0385201788084375e-05,
      "loss": 1.6233,
      "step": 23959
    },
    {
      "epoch": 0.8580586244561033,
      "grad_norm": 1.4209063053131104,
      "learning_rate": 1.0380055267092581e-05,
      "loss": 1.7817,
      "step": 23960
    },
    {
      "epoch": 0.8580944365856715,
      "grad_norm": 1.583618402481079,
      "learning_rate": 1.0374909951821532e-05,
      "loss": 1.3119,
      "step": 23961
    },
    {
      "epoch": 0.8581302487152399,
      "grad_norm": 1.6202590465545654,
      "learning_rate": 1.0369765842340484e-05,
      "loss": 1.3546,
      "step": 23962
    },
    {
      "epoch": 0.8581660608448082,
      "grad_norm": 1.4713712930679321,
      "learning_rate": 1.0364622938718627e-05,
      "loss": 1.2915,
      "step": 23963
    },
    {
      "epoch": 0.8582018729743764,
      "grad_norm": 1.4419574737548828,
      "learning_rate": 1.0359481241025105e-05,
      "loss": 1.3699,
      "step": 23964
    },
    {
      "epoch": 0.8582376851039447,
      "grad_norm": 2.112922430038452,
      "learning_rate": 1.0354340749329172e-05,
      "loss": 1.1557,
      "step": 23965
    },
    {
      "epoch": 0.858273497233513,
      "grad_norm": 1.5085670948028564,
      "learning_rate": 1.0349201463699932e-05,
      "loss": 1.4269,
      "step": 23966
    },
    {
      "epoch": 0.8583093093630813,
      "grad_norm": 1.2277408838272095,
      "learning_rate": 1.0344063384206537e-05,
      "loss": 1.3207,
      "step": 23967
    },
    {
      "epoch": 0.8583451214926495,
      "grad_norm": 1.8692991733551025,
      "learning_rate": 1.0338926510918134e-05,
      "loss": 1.8965,
      "step": 23968
    },
    {
      "epoch": 0.8583809336222178,
      "grad_norm": 1.5993348360061646,
      "learning_rate": 1.0333790843903835e-05,
      "loss": 1.3095,
      "step": 23969
    },
    {
      "epoch": 0.8584167457517862,
      "grad_norm": 1.7335530519485474,
      "learning_rate": 1.0328656383232692e-05,
      "loss": 1.3135,
      "step": 23970
    },
    {
      "epoch": 0.8584525578813544,
      "grad_norm": 1.4201031923294067,
      "learning_rate": 1.0323523128973822e-05,
      "loss": 1.2858,
      "step": 23971
    },
    {
      "epoch": 0.8584883700109227,
      "grad_norm": 1.8706300258636475,
      "learning_rate": 1.0318391081196288e-05,
      "loss": 1.4368,
      "step": 23972
    },
    {
      "epoch": 0.858524182140491,
      "grad_norm": 1.5056337118148804,
      "learning_rate": 1.0313260239969102e-05,
      "loss": 1.4255,
      "step": 23973
    },
    {
      "epoch": 0.8585599942700592,
      "grad_norm": 1.665971040725708,
      "learning_rate": 1.0308130605361333e-05,
      "loss": 1.4536,
      "step": 23974
    },
    {
      "epoch": 0.8585958063996275,
      "grad_norm": 1.5636063814163208,
      "learning_rate": 1.0303002177441934e-05,
      "loss": 1.3331,
      "step": 23975
    },
    {
      "epoch": 0.8586316185291958,
      "grad_norm": 1.4892923831939697,
      "learning_rate": 1.0297874956279974e-05,
      "loss": 1.2562,
      "step": 23976
    },
    {
      "epoch": 0.8586674306587642,
      "grad_norm": 1.3059931993484497,
      "learning_rate": 1.0292748941944385e-05,
      "loss": 1.2593,
      "step": 23977
    },
    {
      "epoch": 0.8587032427883324,
      "grad_norm": 1.7474253177642822,
      "learning_rate": 1.0287624134504158e-05,
      "loss": 1.2735,
      "step": 23978
    },
    {
      "epoch": 0.8587390549179007,
      "grad_norm": 1.916402816772461,
      "learning_rate": 1.0282500534028195e-05,
      "loss": 1.5238,
      "step": 23979
    },
    {
      "epoch": 0.858774867047469,
      "grad_norm": 1.8782771825790405,
      "learning_rate": 1.0277378140585491e-05,
      "loss": 1.4562,
      "step": 23980
    },
    {
      "epoch": 0.8588106791770372,
      "grad_norm": 1.3739666938781738,
      "learning_rate": 1.0272256954244941e-05,
      "loss": 1.6322,
      "step": 23981
    },
    {
      "epoch": 0.8588464913066055,
      "grad_norm": 1.7305517196655273,
      "learning_rate": 1.0267136975075386e-05,
      "loss": 1.2822,
      "step": 23982
    },
    {
      "epoch": 0.8588823034361738,
      "grad_norm": 1.8392616510391235,
      "learning_rate": 1.0262018203145796e-05,
      "loss": 1.331,
      "step": 23983
    },
    {
      "epoch": 0.8589181155657422,
      "grad_norm": 1.3434126377105713,
      "learning_rate": 1.0256900638524979e-05,
      "loss": 1.5343,
      "step": 23984
    },
    {
      "epoch": 0.8589539276953104,
      "grad_norm": 1.373840093612671,
      "learning_rate": 1.0251784281281829e-05,
      "loss": 1.4114,
      "step": 23985
    },
    {
      "epoch": 0.8589897398248787,
      "grad_norm": 2.5821125507354736,
      "learning_rate": 1.0246669131485109e-05,
      "loss": 1.411,
      "step": 23986
    },
    {
      "epoch": 0.859025551954447,
      "grad_norm": 1.8943785429000854,
      "learning_rate": 1.0241555189203722e-05,
      "loss": 1.3037,
      "step": 23987
    },
    {
      "epoch": 0.8590613640840152,
      "grad_norm": 1.3766758441925049,
      "learning_rate": 1.0236442454506411e-05,
      "loss": 1.3691,
      "step": 23988
    },
    {
      "epoch": 0.8590971762135835,
      "grad_norm": 2.0536932945251465,
      "learning_rate": 1.0231330927462002e-05,
      "loss": 1.1827,
      "step": 23989
    },
    {
      "epoch": 0.8591329883431518,
      "grad_norm": 1.7797173261642456,
      "learning_rate": 1.0226220608139214e-05,
      "loss": 1.53,
      "step": 23990
    },
    {
      "epoch": 0.8591688004727202,
      "grad_norm": 1.6189501285552979,
      "learning_rate": 1.022111149660684e-05,
      "loss": 1.8422,
      "step": 23991
    },
    {
      "epoch": 0.8592046126022884,
      "grad_norm": 1.5615276098251343,
      "learning_rate": 1.021600359293361e-05,
      "loss": 1.3183,
      "step": 23992
    },
    {
      "epoch": 0.8592404247318567,
      "grad_norm": 1.7850444316864014,
      "learning_rate": 1.0210896897188216e-05,
      "loss": 1.5006,
      "step": 23993
    },
    {
      "epoch": 0.859276236861425,
      "grad_norm": 2.1221442222595215,
      "learning_rate": 1.0205791409439413e-05,
      "loss": 1.2876,
      "step": 23994
    },
    {
      "epoch": 0.8593120489909932,
      "grad_norm": 1.399099349975586,
      "learning_rate": 1.0200687129755837e-05,
      "loss": 1.0976,
      "step": 23995
    },
    {
      "epoch": 0.8593478611205615,
      "grad_norm": 1.5655336380004883,
      "learning_rate": 1.0195584058206209e-05,
      "loss": 1.1475,
      "step": 23996
    },
    {
      "epoch": 0.8593836732501298,
      "grad_norm": 1.4112638235092163,
      "learning_rate": 1.0190482194859119e-05,
      "loss": 1.2891,
      "step": 23997
    },
    {
      "epoch": 0.8594194853796981,
      "grad_norm": 1.8032960891723633,
      "learning_rate": 1.018538153978329e-05,
      "loss": 1.3916,
      "step": 23998
    },
    {
      "epoch": 0.8594552975092664,
      "grad_norm": 1.1614011526107788,
      "learning_rate": 1.0180282093047288e-05,
      "loss": 0.9882,
      "step": 23999
    },
    {
      "epoch": 0.8594911096388347,
      "grad_norm": 1.553151249885559,
      "learning_rate": 1.0175183854719716e-05,
      "loss": 1.6078,
      "step": 24000
    },
    {
      "epoch": 0.859526921768403,
      "grad_norm": 1.4639896154403687,
      "learning_rate": 1.0170086824869184e-05,
      "loss": 1.3421,
      "step": 24001
    },
    {
      "epoch": 0.8595627338979712,
      "grad_norm": 1.3828316926956177,
      "learning_rate": 1.0164991003564261e-05,
      "loss": 1.2658,
      "step": 24002
    },
    {
      "epoch": 0.8595985460275395,
      "grad_norm": 2.3339357376098633,
      "learning_rate": 1.0159896390873524e-05,
      "loss": 1.5719,
      "step": 24003
    },
    {
      "epoch": 0.8596343581571078,
      "grad_norm": 1.4203572273254395,
      "learning_rate": 1.0154802986865475e-05,
      "loss": 1.5003,
      "step": 24004
    },
    {
      "epoch": 0.8596701702866761,
      "grad_norm": 1.435115933418274,
      "learning_rate": 1.0149710791608657e-05,
      "loss": 1.3779,
      "step": 24005
    },
    {
      "epoch": 0.8597059824162444,
      "grad_norm": 1.3784613609313965,
      "learning_rate": 1.0144619805171584e-05,
      "loss": 1.4091,
      "step": 24006
    },
    {
      "epoch": 0.8597417945458127,
      "grad_norm": 1.42812979221344,
      "learning_rate": 1.0139530027622768e-05,
      "loss": 1.2791,
      "step": 24007
    },
    {
      "epoch": 0.859777606675381,
      "grad_norm": 1.809586763381958,
      "learning_rate": 1.0134441459030642e-05,
      "loss": 1.3902,
      "step": 24008
    },
    {
      "epoch": 0.8598134188049492,
      "grad_norm": 1.387447714805603,
      "learning_rate": 1.0129354099463683e-05,
      "loss": 1.4874,
      "step": 24009
    },
    {
      "epoch": 0.8598492309345175,
      "grad_norm": 1.933063268661499,
      "learning_rate": 1.0124267948990363e-05,
      "loss": 1.7334,
      "step": 24010
    },
    {
      "epoch": 0.8598850430640858,
      "grad_norm": 1.4610179662704468,
      "learning_rate": 1.0119183007679067e-05,
      "loss": 1.3521,
      "step": 24011
    },
    {
      "epoch": 0.8599208551936541,
      "grad_norm": 1.5905406475067139,
      "learning_rate": 1.0114099275598232e-05,
      "loss": 1.497,
      "step": 24012
    },
    {
      "epoch": 0.8599566673232224,
      "grad_norm": 1.7359148263931274,
      "learning_rate": 1.0109016752816247e-05,
      "loss": 1.3048,
      "step": 24013
    },
    {
      "epoch": 0.8599924794527907,
      "grad_norm": 2.0931310653686523,
      "learning_rate": 1.0103935439401502e-05,
      "loss": 1.3526,
      "step": 24014
    },
    {
      "epoch": 0.8600282915823589,
      "grad_norm": 1.846803903579712,
      "learning_rate": 1.0098855335422331e-05,
      "loss": 1.521,
      "step": 24015
    },
    {
      "epoch": 0.8600641037119272,
      "grad_norm": 1.602505087852478,
      "learning_rate": 1.00937764409471e-05,
      "loss": 1.6741,
      "step": 24016
    },
    {
      "epoch": 0.8600999158414955,
      "grad_norm": 1.551328420639038,
      "learning_rate": 1.0088698756044146e-05,
      "loss": 1.2546,
      "step": 24017
    },
    {
      "epoch": 0.8601357279710637,
      "grad_norm": 1.5050621032714844,
      "learning_rate": 1.0083622280781769e-05,
      "loss": 1.4149,
      "step": 24018
    },
    {
      "epoch": 0.8601715401006321,
      "grad_norm": 1.488797903060913,
      "learning_rate": 1.0078547015228257e-05,
      "loss": 1.5439,
      "step": 24019
    },
    {
      "epoch": 0.8602073522302004,
      "grad_norm": 1.7439039945602417,
      "learning_rate": 1.0073472959451913e-05,
      "loss": 1.1836,
      "step": 24020
    },
    {
      "epoch": 0.8602431643597687,
      "grad_norm": 1.8317373991012573,
      "learning_rate": 1.0068400113521014e-05,
      "loss": 1.4543,
      "step": 24021
    },
    {
      "epoch": 0.8602789764893369,
      "grad_norm": 1.8038636445999146,
      "learning_rate": 1.0063328477503764e-05,
      "loss": 1.5558,
      "step": 24022
    },
    {
      "epoch": 0.8603147886189052,
      "grad_norm": 2.062847137451172,
      "learning_rate": 1.0058258051468417e-05,
      "loss": 1.2856,
      "step": 24023
    },
    {
      "epoch": 0.8603506007484735,
      "grad_norm": 2.013709306716919,
      "learning_rate": 1.0053188835483197e-05,
      "loss": 1.5779,
      "step": 24024
    },
    {
      "epoch": 0.8603864128780417,
      "grad_norm": 2.6702752113342285,
      "learning_rate": 1.0048120829616314e-05,
      "loss": 1.3096,
      "step": 24025
    },
    {
      "epoch": 0.8604222250076101,
      "grad_norm": 1.617053508758545,
      "learning_rate": 1.0043054033935917e-05,
      "loss": 1.5229,
      "step": 24026
    },
    {
      "epoch": 0.8604580371371784,
      "grad_norm": 1.3419846296310425,
      "learning_rate": 1.0037988448510193e-05,
      "loss": 1.154,
      "step": 24027
    },
    {
      "epoch": 0.8604938492667467,
      "grad_norm": 1.271104097366333,
      "learning_rate": 1.0032924073407313e-05,
      "loss": 1.4895,
      "step": 24028
    },
    {
      "epoch": 0.8605296613963149,
      "grad_norm": 1.682997703552246,
      "learning_rate": 1.0027860908695363e-05,
      "loss": 1.4205,
      "step": 24029
    },
    {
      "epoch": 0.8605654735258832,
      "grad_norm": 1.687303066253662,
      "learning_rate": 1.0022798954442491e-05,
      "loss": 1.2232,
      "step": 24030
    },
    {
      "epoch": 0.8606012856554515,
      "grad_norm": 1.7541818618774414,
      "learning_rate": 1.001773821071681e-05,
      "loss": 1.2612,
      "step": 24031
    },
    {
      "epoch": 0.8606370977850197,
      "grad_norm": 2.01399302482605,
      "learning_rate": 1.0012678677586396e-05,
      "loss": 1.4981,
      "step": 24032
    },
    {
      "epoch": 0.8606729099145881,
      "grad_norm": 1.589170217514038,
      "learning_rate": 1.0007620355119307e-05,
      "loss": 1.3366,
      "step": 24033
    },
    {
      "epoch": 0.8607087220441564,
      "grad_norm": 1.6481248140335083,
      "learning_rate": 1.000256324338359e-05,
      "loss": 1.3699,
      "step": 24034
    },
    {
      "epoch": 0.8607445341737247,
      "grad_norm": 1.8044880628585815,
      "learning_rate": 9.997507342447333e-06,
      "loss": 1.5593,
      "step": 24035
    },
    {
      "epoch": 0.8607803463032929,
      "grad_norm": 1.9255050420761108,
      "learning_rate": 9.992452652378493e-06,
      "loss": 1.5335,
      "step": 24036
    },
    {
      "epoch": 0.8608161584328612,
      "grad_norm": 1.7744263410568237,
      "learning_rate": 9.987399173245093e-06,
      "loss": 1.1219,
      "step": 24037
    },
    {
      "epoch": 0.8608519705624295,
      "grad_norm": 1.6102805137634277,
      "learning_rate": 9.982346905115137e-06,
      "loss": 1.3634,
      "step": 24038
    },
    {
      "epoch": 0.8608877826919977,
      "grad_norm": 1.4857194423675537,
      "learning_rate": 9.977295848056612e-06,
      "loss": 1.3996,
      "step": 24039
    },
    {
      "epoch": 0.8609235948215661,
      "grad_norm": 1.8359849452972412,
      "learning_rate": 9.97224600213742e-06,
      "loss": 1.3318,
      "step": 24040
    },
    {
      "epoch": 0.8609594069511344,
      "grad_norm": 1.50896155834198,
      "learning_rate": 9.96719736742554e-06,
      "loss": 1.6464,
      "step": 24041
    },
    {
      "epoch": 0.8609952190807026,
      "grad_norm": 1.3213436603546143,
      "learning_rate": 9.962149943988885e-06,
      "loss": 1.2947,
      "step": 24042
    },
    {
      "epoch": 0.8610310312102709,
      "grad_norm": 1.540392279624939,
      "learning_rate": 9.957103731895379e-06,
      "loss": 1.5812,
      "step": 24043
    },
    {
      "epoch": 0.8610668433398392,
      "grad_norm": 1.953053593635559,
      "learning_rate": 9.952058731212877e-06,
      "loss": 1.8253,
      "step": 24044
    },
    {
      "epoch": 0.8611026554694075,
      "grad_norm": 1.9430025815963745,
      "learning_rate": 9.947014942009269e-06,
      "loss": 1.4107,
      "step": 24045
    },
    {
      "epoch": 0.8611384675989757,
      "grad_norm": 1.5683186054229736,
      "learning_rate": 9.941972364352436e-06,
      "loss": 1.6623,
      "step": 24046
    },
    {
      "epoch": 0.8611742797285441,
      "grad_norm": 1.603224754333496,
      "learning_rate": 9.936930998310179e-06,
      "loss": 1.4686,
      "step": 24047
    },
    {
      "epoch": 0.8612100918581124,
      "grad_norm": 1.5715124607086182,
      "learning_rate": 9.931890843950342e-06,
      "loss": 1.481,
      "step": 24048
    },
    {
      "epoch": 0.8612459039876806,
      "grad_norm": 1.7702678442001343,
      "learning_rate": 9.92685190134074e-06,
      "loss": 1.6707,
      "step": 24049
    },
    {
      "epoch": 0.8612817161172489,
      "grad_norm": 1.6073265075683594,
      "learning_rate": 9.921814170549171e-06,
      "loss": 1.2719,
      "step": 24050
    },
    {
      "epoch": 0.8613175282468172,
      "grad_norm": 2.0798747539520264,
      "learning_rate": 9.916777651643383e-06,
      "loss": 1.3234,
      "step": 24051
    },
    {
      "epoch": 0.8613533403763854,
      "grad_norm": 2.2637381553649902,
      "learning_rate": 9.911742344691156e-06,
      "loss": 1.5011,
      "step": 24052
    },
    {
      "epoch": 0.8613891525059537,
      "grad_norm": 1.9088646173477173,
      "learning_rate": 9.906708249760244e-06,
      "loss": 1.2568,
      "step": 24053
    },
    {
      "epoch": 0.8614249646355221,
      "grad_norm": 1.5362104177474976,
      "learning_rate": 9.901675366918339e-06,
      "loss": 1.2768,
      "step": 24054
    },
    {
      "epoch": 0.8614607767650904,
      "grad_norm": 1.4941067695617676,
      "learning_rate": 9.896643696233177e-06,
      "loss": 1.0898,
      "step": 24055
    },
    {
      "epoch": 0.8614965888946586,
      "grad_norm": 1.436546802520752,
      "learning_rate": 9.891613237772458e-06,
      "loss": 1.5783,
      "step": 24056
    },
    {
      "epoch": 0.8615324010242269,
      "grad_norm": 1.2099878787994385,
      "learning_rate": 9.88658399160386e-06,
      "loss": 1.3196,
      "step": 24057
    },
    {
      "epoch": 0.8615682131537952,
      "grad_norm": 1.3755364418029785,
      "learning_rate": 9.88155595779502e-06,
      "loss": 1.4722,
      "step": 24058
    },
    {
      "epoch": 0.8616040252833634,
      "grad_norm": 1.5416947603225708,
      "learning_rate": 9.876529136413593e-06,
      "loss": 1.2288,
      "step": 24059
    },
    {
      "epoch": 0.8616398374129317,
      "grad_norm": 1.4445111751556396,
      "learning_rate": 9.871503527527226e-06,
      "loss": 1.4655,
      "step": 24060
    },
    {
      "epoch": 0.8616756495425001,
      "grad_norm": 1.4712949991226196,
      "learning_rate": 9.866479131203544e-06,
      "loss": 1.3464,
      "step": 24061
    },
    {
      "epoch": 0.8617114616720684,
      "grad_norm": 1.454877495765686,
      "learning_rate": 9.861455947510112e-06,
      "loss": 1.4811,
      "step": 24062
    },
    {
      "epoch": 0.8617472738016366,
      "grad_norm": 1.4645787477493286,
      "learning_rate": 9.856433976514479e-06,
      "loss": 1.4101,
      "step": 24063
    },
    {
      "epoch": 0.8617830859312049,
      "grad_norm": 2.1881892681121826,
      "learning_rate": 9.8514132182843e-06,
      "loss": 1.2949,
      "step": 24064
    },
    {
      "epoch": 0.8618188980607732,
      "grad_norm": 1.827864170074463,
      "learning_rate": 9.846393672887044e-06,
      "loss": 1.496,
      "step": 24065
    },
    {
      "epoch": 0.8618547101903414,
      "grad_norm": 1.3500398397445679,
      "learning_rate": 9.841375340390268e-06,
      "loss": 1.4182,
      "step": 24066
    },
    {
      "epoch": 0.8618905223199097,
      "grad_norm": 1.408724069595337,
      "learning_rate": 9.836358220861508e-06,
      "loss": 1.3504,
      "step": 24067
    },
    {
      "epoch": 0.8619263344494781,
      "grad_norm": 1.4061379432678223,
      "learning_rate": 9.831342314368252e-06,
      "loss": 1.1648,
      "step": 24068
    },
    {
      "epoch": 0.8619621465790464,
      "grad_norm": 1.690593957901001,
      "learning_rate": 9.826327620977972e-06,
      "loss": 1.3875,
      "step": 24069
    },
    {
      "epoch": 0.8619979587086146,
      "grad_norm": 1.5156593322753906,
      "learning_rate": 9.82131414075811e-06,
      "loss": 1.1197,
      "step": 24070
    },
    {
      "epoch": 0.8620337708381829,
      "grad_norm": 1.518989086151123,
      "learning_rate": 9.816301873776178e-06,
      "loss": 1.2619,
      "step": 24071
    },
    {
      "epoch": 0.8620695829677512,
      "grad_norm": 1.8217155933380127,
      "learning_rate": 9.81129082009955e-06,
      "loss": 1.471,
      "step": 24072
    },
    {
      "epoch": 0.8621053950973194,
      "grad_norm": 1.4540605545043945,
      "learning_rate": 9.8062809797957e-06,
      "loss": 1.2745,
      "step": 24073
    },
    {
      "epoch": 0.8621412072268877,
      "grad_norm": 1.4706782102584839,
      "learning_rate": 9.801272352931957e-06,
      "loss": 1.4899,
      "step": 24074
    },
    {
      "epoch": 0.8621770193564561,
      "grad_norm": 1.7271860837936401,
      "learning_rate": 9.796264939575784e-06,
      "loss": 1.5868,
      "step": 24075
    },
    {
      "epoch": 0.8622128314860243,
      "grad_norm": 1.2994468212127686,
      "learning_rate": 9.791258739794484e-06,
      "loss": 1.5339,
      "step": 24076
    },
    {
      "epoch": 0.8622486436155926,
      "grad_norm": 1.6923842430114746,
      "learning_rate": 9.78625375365545e-06,
      "loss": 1.3637,
      "step": 24077
    },
    {
      "epoch": 0.8622844557451609,
      "grad_norm": 1.8347572088241577,
      "learning_rate": 9.781249981226015e-06,
      "loss": 1.6562,
      "step": 24078
    },
    {
      "epoch": 0.8623202678747292,
      "grad_norm": 2.0344903469085693,
      "learning_rate": 9.77624742257347e-06,
      "loss": 1.68,
      "step": 24079
    },
    {
      "epoch": 0.8623560800042974,
      "grad_norm": 1.6433473825454712,
      "learning_rate": 9.771246077765151e-06,
      "loss": 1.554,
      "step": 24080
    },
    {
      "epoch": 0.8623918921338657,
      "grad_norm": 1.4841326475143433,
      "learning_rate": 9.766245946868302e-06,
      "loss": 1.5479,
      "step": 24081
    },
    {
      "epoch": 0.8624277042634341,
      "grad_norm": 1.3508626222610474,
      "learning_rate": 9.761247029950249e-06,
      "loss": 1.2383,
      "step": 24082
    },
    {
      "epoch": 0.8624635163930023,
      "grad_norm": 1.4519118070602417,
      "learning_rate": 9.756249327078204e-06,
      "loss": 1.3748,
      "step": 24083
    },
    {
      "epoch": 0.8624993285225706,
      "grad_norm": 2.055274486541748,
      "learning_rate": 9.751252838319436e-06,
      "loss": 1.6204,
      "step": 24084
    },
    {
      "epoch": 0.8625351406521389,
      "grad_norm": 1.8891230821609497,
      "learning_rate": 9.746257563741102e-06,
      "loss": 1.2223,
      "step": 24085
    },
    {
      "epoch": 0.8625709527817071,
      "grad_norm": 1.5189100503921509,
      "learning_rate": 9.741263503410503e-06,
      "loss": 1.4791,
      "step": 24086
    },
    {
      "epoch": 0.8626067649112754,
      "grad_norm": 1.586703896522522,
      "learning_rate": 9.736270657394774e-06,
      "loss": 1.1318,
      "step": 24087
    },
    {
      "epoch": 0.8626425770408437,
      "grad_norm": 1.821293830871582,
      "learning_rate": 9.731279025761076e-06,
      "loss": 1.2319,
      "step": 24088
    },
    {
      "epoch": 0.8626783891704121,
      "grad_norm": 1.7827171087265015,
      "learning_rate": 9.726288608576573e-06,
      "loss": 1.3376,
      "step": 24089
    },
    {
      "epoch": 0.8627142012999803,
      "grad_norm": 1.368043303489685,
      "learning_rate": 9.721299405908412e-06,
      "loss": 1.5547,
      "step": 24090
    },
    {
      "epoch": 0.8627500134295486,
      "grad_norm": 2.012338638305664,
      "learning_rate": 9.716311417823742e-06,
      "loss": 1.7127,
      "step": 24091
    },
    {
      "epoch": 0.8627858255591169,
      "grad_norm": 2.133498191833496,
      "learning_rate": 9.711324644389609e-06,
      "loss": 1.4458,
      "step": 24092
    },
    {
      "epoch": 0.8628216376886851,
      "grad_norm": 1.5990008115768433,
      "learning_rate": 9.706339085673167e-06,
      "loss": 1.2618,
      "step": 24093
    },
    {
      "epoch": 0.8628574498182534,
      "grad_norm": 1.9938685894012451,
      "learning_rate": 9.701354741741454e-06,
      "loss": 1.6879,
      "step": 24094
    },
    {
      "epoch": 0.8628932619478217,
      "grad_norm": 1.8313326835632324,
      "learning_rate": 9.696371612661548e-06,
      "loss": 1.3386,
      "step": 24095
    },
    {
      "epoch": 0.8629290740773901,
      "grad_norm": 1.820915699005127,
      "learning_rate": 9.691389698500463e-06,
      "loss": 1.2029,
      "step": 24096
    },
    {
      "epoch": 0.8629648862069583,
      "grad_norm": 1.3561536073684692,
      "learning_rate": 9.686408999325236e-06,
      "loss": 1.3453,
      "step": 24097
    },
    {
      "epoch": 0.8630006983365266,
      "grad_norm": 2.9096224308013916,
      "learning_rate": 9.6814295152029e-06,
      "loss": 1.2922,
      "step": 24098
    },
    {
      "epoch": 0.8630365104660949,
      "grad_norm": 1.629462718963623,
      "learning_rate": 9.676451246200401e-06,
      "loss": 1.4125,
      "step": 24099
    },
    {
      "epoch": 0.8630723225956631,
      "grad_norm": 1.6376150846481323,
      "learning_rate": 9.671474192384755e-06,
      "loss": 1.2534,
      "step": 24100
    },
    {
      "epoch": 0.8631081347252314,
      "grad_norm": 1.6502249240875244,
      "learning_rate": 9.666498353822905e-06,
      "loss": 1.478,
      "step": 24101
    },
    {
      "epoch": 0.8631439468547997,
      "grad_norm": 2.37195086479187,
      "learning_rate": 9.661523730581813e-06,
      "loss": 1.2485,
      "step": 24102
    },
    {
      "epoch": 0.863179758984368,
      "grad_norm": 1.9842101335525513,
      "learning_rate": 9.656550322728353e-06,
      "loss": 1.3154,
      "step": 24103
    },
    {
      "epoch": 0.8632155711139363,
      "grad_norm": 1.7502880096435547,
      "learning_rate": 9.651578130329508e-06,
      "loss": 1.5923,
      "step": 24104
    },
    {
      "epoch": 0.8632513832435046,
      "grad_norm": 1.3002649545669556,
      "learning_rate": 9.646607153452147e-06,
      "loss": 1.1142,
      "step": 24105
    },
    {
      "epoch": 0.8632871953730729,
      "grad_norm": 1.5864512920379639,
      "learning_rate": 9.641637392163116e-06,
      "loss": 1.1018,
      "step": 24106
    },
    {
      "epoch": 0.8633230075026411,
      "grad_norm": 1.585464596748352,
      "learning_rate": 9.636668846529296e-06,
      "loss": 1.3463,
      "step": 24107
    },
    {
      "epoch": 0.8633588196322094,
      "grad_norm": 1.4023091793060303,
      "learning_rate": 9.631701516617542e-06,
      "loss": 1.2446,
      "step": 24108
    },
    {
      "epoch": 0.8633946317617777,
      "grad_norm": 1.7487534284591675,
      "learning_rate": 9.626735402494703e-06,
      "loss": 1.1232,
      "step": 24109
    },
    {
      "epoch": 0.863430443891346,
      "grad_norm": 2.0945987701416016,
      "learning_rate": 9.621770504227534e-06,
      "loss": 1.4021,
      "step": 24110
    },
    {
      "epoch": 0.8634662560209143,
      "grad_norm": 1.8034226894378662,
      "learning_rate": 9.616806821882873e-06,
      "loss": 1.2245,
      "step": 24111
    },
    {
      "epoch": 0.8635020681504826,
      "grad_norm": 1.5060043334960938,
      "learning_rate": 9.611844355527477e-06,
      "loss": 1.4211,
      "step": 24112
    },
    {
      "epoch": 0.8635378802800509,
      "grad_norm": 1.9010181427001953,
      "learning_rate": 9.60688310522816e-06,
      "loss": 1.5169,
      "step": 24113
    },
    {
      "epoch": 0.8635736924096191,
      "grad_norm": 2.2144103050231934,
      "learning_rate": 9.6019230710516e-06,
      "loss": 1.7569,
      "step": 24114
    },
    {
      "epoch": 0.8636095045391874,
      "grad_norm": 1.660400390625,
      "learning_rate": 9.596964253064567e-06,
      "loss": 1.3721,
      "step": 24115
    },
    {
      "epoch": 0.8636453166687557,
      "grad_norm": 1.677625298500061,
      "learning_rate": 9.592006651333785e-06,
      "loss": 1.4996,
      "step": 24116
    },
    {
      "epoch": 0.863681128798324,
      "grad_norm": 1.848111867904663,
      "learning_rate": 9.587050265925912e-06,
      "loss": 1.7495,
      "step": 24117
    },
    {
      "epoch": 0.8637169409278923,
      "grad_norm": 1.6364637613296509,
      "learning_rate": 9.582095096907651e-06,
      "loss": 1.3123,
      "step": 24118
    },
    {
      "epoch": 0.8637527530574606,
      "grad_norm": 1.3802284002304077,
      "learning_rate": 9.57714114434568e-06,
      "loss": 1.4342,
      "step": 24119
    },
    {
      "epoch": 0.8637885651870288,
      "grad_norm": 1.6368407011032104,
      "learning_rate": 9.572188408306649e-06,
      "loss": 1.496,
      "step": 24120
    },
    {
      "epoch": 0.8638243773165971,
      "grad_norm": 1.535515546798706,
      "learning_rate": 9.567236888857166e-06,
      "loss": 1.4658,
      "step": 24121
    },
    {
      "epoch": 0.8638601894461654,
      "grad_norm": 1.625495195388794,
      "learning_rate": 9.562286586063861e-06,
      "loss": 1.8956,
      "step": 24122
    },
    {
      "epoch": 0.8638960015757337,
      "grad_norm": 1.8822096586227417,
      "learning_rate": 9.557337499993346e-06,
      "loss": 1.4238,
      "step": 24123
    },
    {
      "epoch": 0.863931813705302,
      "grad_norm": 2.1007256507873535,
      "learning_rate": 9.552389630712178e-06,
      "loss": 1.5239,
      "step": 24124
    },
    {
      "epoch": 0.8639676258348703,
      "grad_norm": 2.123950958251953,
      "learning_rate": 9.547442978286946e-06,
      "loss": 1.5102,
      "step": 24125
    },
    {
      "epoch": 0.8640034379644386,
      "grad_norm": 1.6733735799789429,
      "learning_rate": 9.542497542784178e-06,
      "loss": 1.2072,
      "step": 24126
    },
    {
      "epoch": 0.8640392500940068,
      "grad_norm": 1.3590643405914307,
      "learning_rate": 9.537553324270455e-06,
      "loss": 1.7378,
      "step": 24127
    },
    {
      "epoch": 0.8640750622235751,
      "grad_norm": 1.9480410814285278,
      "learning_rate": 9.53261032281224e-06,
      "loss": 1.5894,
      "step": 24128
    },
    {
      "epoch": 0.8641108743531434,
      "grad_norm": 1.5625706911087036,
      "learning_rate": 9.527668538476054e-06,
      "loss": 1.3688,
      "step": 24129
    },
    {
      "epoch": 0.8641466864827116,
      "grad_norm": 1.3763489723205566,
      "learning_rate": 9.522727971328393e-06,
      "loss": 0.9818,
      "step": 24130
    },
    {
      "epoch": 0.86418249861228,
      "grad_norm": 2.1239705085754395,
      "learning_rate": 9.517788621435742e-06,
      "loss": 1.7518,
      "step": 24131
    },
    {
      "epoch": 0.8642183107418483,
      "grad_norm": 2.0525310039520264,
      "learning_rate": 9.512850488864511e-06,
      "loss": 1.6703,
      "step": 24132
    },
    {
      "epoch": 0.8642541228714166,
      "grad_norm": 2.882822036743164,
      "learning_rate": 9.50791357368115e-06,
      "loss": 1.3199,
      "step": 24133
    },
    {
      "epoch": 0.8642899350009848,
      "grad_norm": 1.5467312335968018,
      "learning_rate": 9.502977875952113e-06,
      "loss": 1.4511,
      "step": 24134
    },
    {
      "epoch": 0.8643257471305531,
      "grad_norm": 1.8754770755767822,
      "learning_rate": 9.49804339574375e-06,
      "loss": 1.5401,
      "step": 24135
    },
    {
      "epoch": 0.8643615592601214,
      "grad_norm": 1.4859979152679443,
      "learning_rate": 9.493110133122474e-06,
      "loss": 1.5788,
      "step": 24136
    },
    {
      "epoch": 0.8643973713896896,
      "grad_norm": 1.6180216073989868,
      "learning_rate": 9.488178088154654e-06,
      "loss": 1.0272,
      "step": 24137
    },
    {
      "epoch": 0.864433183519258,
      "grad_norm": 1.5847938060760498,
      "learning_rate": 9.48324726090667e-06,
      "loss": 1.4547,
      "step": 24138
    },
    {
      "epoch": 0.8644689956488263,
      "grad_norm": 1.4334745407104492,
      "learning_rate": 9.478317651444812e-06,
      "loss": 1.2533,
      "step": 24139
    },
    {
      "epoch": 0.8645048077783946,
      "grad_norm": 1.6126441955566406,
      "learning_rate": 9.47338925983543e-06,
      "loss": 1.2693,
      "step": 24140
    },
    {
      "epoch": 0.8645406199079628,
      "grad_norm": 1.5606629848480225,
      "learning_rate": 9.468462086144847e-06,
      "loss": 1.3871,
      "step": 24141
    },
    {
      "epoch": 0.8645764320375311,
      "grad_norm": 1.4781626462936401,
      "learning_rate": 9.46353613043931e-06,
      "loss": 1.239,
      "step": 24142
    },
    {
      "epoch": 0.8646122441670994,
      "grad_norm": 1.2007153034210205,
      "learning_rate": 9.4586113927851e-06,
      "loss": 1.2876,
      "step": 24143
    },
    {
      "epoch": 0.8646480562966676,
      "grad_norm": 1.7525100708007812,
      "learning_rate": 9.453687873248495e-06,
      "loss": 1.5107,
      "step": 24144
    },
    {
      "epoch": 0.864683868426236,
      "grad_norm": 1.4336668252944946,
      "learning_rate": 9.448765571895735e-06,
      "loss": 1.4305,
      "step": 24145
    },
    {
      "epoch": 0.8647196805558043,
      "grad_norm": 2.0794224739074707,
      "learning_rate": 9.443844488793018e-06,
      "loss": 1.7878,
      "step": 24146
    },
    {
      "epoch": 0.8647554926853726,
      "grad_norm": 1.4945629835128784,
      "learning_rate": 9.438924624006563e-06,
      "loss": 1.5938,
      "step": 24147
    },
    {
      "epoch": 0.8647913048149408,
      "grad_norm": 1.2787604331970215,
      "learning_rate": 9.434005977602556e-06,
      "loss": 1.2151,
      "step": 24148
    },
    {
      "epoch": 0.8648271169445091,
      "grad_norm": 1.4332270622253418,
      "learning_rate": 9.429088549647203e-06,
      "loss": 1.2921,
      "step": 24149
    },
    {
      "epoch": 0.8648629290740774,
      "grad_norm": 1.8126541376113892,
      "learning_rate": 9.424172340206616e-06,
      "loss": 1.3874,
      "step": 24150
    },
    {
      "epoch": 0.8648987412036456,
      "grad_norm": 1.3045158386230469,
      "learning_rate": 9.419257349346956e-06,
      "loss": 1.3823,
      "step": 24151
    },
    {
      "epoch": 0.864934553333214,
      "grad_norm": 1.5711522102355957,
      "learning_rate": 9.414343577134355e-06,
      "loss": 1.5328,
      "step": 24152
    },
    {
      "epoch": 0.8649703654627823,
      "grad_norm": 1.600151777267456,
      "learning_rate": 9.409431023634908e-06,
      "loss": 1.39,
      "step": 24153
    },
    {
      "epoch": 0.8650061775923505,
      "grad_norm": 1.6828088760375977,
      "learning_rate": 9.404519688914703e-06,
      "loss": 1.2594,
      "step": 24154
    },
    {
      "epoch": 0.8650419897219188,
      "grad_norm": 1.626774787902832,
      "learning_rate": 9.399609573039836e-06,
      "loss": 1.4374,
      "step": 24155
    },
    {
      "epoch": 0.8650778018514871,
      "grad_norm": 1.709047555923462,
      "learning_rate": 9.394700676076374e-06,
      "loss": 1.2774,
      "step": 24156
    },
    {
      "epoch": 0.8651136139810554,
      "grad_norm": 1.3657279014587402,
      "learning_rate": 9.389792998090319e-06,
      "loss": 1.4303,
      "step": 24157
    },
    {
      "epoch": 0.8651494261106236,
      "grad_norm": 1.8122633695602417,
      "learning_rate": 9.384886539147718e-06,
      "loss": 1.4683,
      "step": 24158
    },
    {
      "epoch": 0.865185238240192,
      "grad_norm": 2.004755973815918,
      "learning_rate": 9.379981299314611e-06,
      "loss": 1.7386,
      "step": 24159
    },
    {
      "epoch": 0.8652210503697603,
      "grad_norm": 1.629216194152832,
      "learning_rate": 9.375077278656941e-06,
      "loss": 1.3271,
      "step": 24160
    },
    {
      "epoch": 0.8652568624993285,
      "grad_norm": 1.6570649147033691,
      "learning_rate": 9.370174477240712e-06,
      "loss": 1.3823,
      "step": 24161
    },
    {
      "epoch": 0.8652926746288968,
      "grad_norm": 1.4738001823425293,
      "learning_rate": 9.36527289513187e-06,
      "loss": 1.4896,
      "step": 24162
    },
    {
      "epoch": 0.8653284867584651,
      "grad_norm": 1.5269134044647217,
      "learning_rate": 9.3603725323964e-06,
      "loss": 1.3027,
      "step": 24163
    },
    {
      "epoch": 0.8653642988880333,
      "grad_norm": 2.1662659645080566,
      "learning_rate": 9.355473389100178e-06,
      "loss": 1.4489,
      "step": 24164
    },
    {
      "epoch": 0.8654001110176016,
      "grad_norm": 1.455833911895752,
      "learning_rate": 9.350575465309142e-06,
      "loss": 1.3106,
      "step": 24165
    },
    {
      "epoch": 0.86543592314717,
      "grad_norm": 1.9036498069763184,
      "learning_rate": 9.345678761089194e-06,
      "loss": 1.6722,
      "step": 24166
    },
    {
      "epoch": 0.8654717352767383,
      "grad_norm": 1.4704736471176147,
      "learning_rate": 9.340783276506193e-06,
      "loss": 1.2371,
      "step": 24167
    },
    {
      "epoch": 0.8655075474063065,
      "grad_norm": 1.6556400060653687,
      "learning_rate": 9.335889011626032e-06,
      "loss": 1.2193,
      "step": 24168
    },
    {
      "epoch": 0.8655433595358748,
      "grad_norm": 1.7995213270187378,
      "learning_rate": 9.330995966514489e-06,
      "loss": 1.4554,
      "step": 24169
    },
    {
      "epoch": 0.8655791716654431,
      "grad_norm": 1.5920368432998657,
      "learning_rate": 9.32610414123748e-06,
      "loss": 1.2939,
      "step": 24170
    },
    {
      "epoch": 0.8656149837950113,
      "grad_norm": 1.5632593631744385,
      "learning_rate": 9.321213535860763e-06,
      "loss": 1.2844,
      "step": 24171
    },
    {
      "epoch": 0.8656507959245796,
      "grad_norm": 1.6625022888183594,
      "learning_rate": 9.316324150450173e-06,
      "loss": 1.2916,
      "step": 24172
    },
    {
      "epoch": 0.865686608054148,
      "grad_norm": 1.8836668729782104,
      "learning_rate": 9.311435985071426e-06,
      "loss": 1.4325,
      "step": 24173
    },
    {
      "epoch": 0.8657224201837163,
      "grad_norm": 1.6699721813201904,
      "learning_rate": 9.30654903979037e-06,
      "loss": 1.4862,
      "step": 24174
    },
    {
      "epoch": 0.8657582323132845,
      "grad_norm": 1.752537488937378,
      "learning_rate": 9.301663314672704e-06,
      "loss": 1.4677,
      "step": 24175
    },
    {
      "epoch": 0.8657940444428528,
      "grad_norm": 2.023043155670166,
      "learning_rate": 9.296778809784123e-06,
      "loss": 1.2724,
      "step": 24176
    },
    {
      "epoch": 0.8658298565724211,
      "grad_norm": 1.663191556930542,
      "learning_rate": 9.29189552519043e-06,
      "loss": 1.4344,
      "step": 24177
    },
    {
      "epoch": 0.8658656687019893,
      "grad_norm": 1.4714888334274292,
      "learning_rate": 9.287013460957261e-06,
      "loss": 1.2581,
      "step": 24178
    },
    {
      "epoch": 0.8659014808315576,
      "grad_norm": 1.4269914627075195,
      "learning_rate": 9.28213261715033e-06,
      "loss": 1.4169,
      "step": 24179
    },
    {
      "epoch": 0.865937292961126,
      "grad_norm": 1.6043585538864136,
      "learning_rate": 9.27725299383525e-06,
      "loss": 1.5219,
      "step": 24180
    },
    {
      "epoch": 0.8659731050906943,
      "grad_norm": 1.789746642112732,
      "learning_rate": 9.272374591077748e-06,
      "loss": 1.2993,
      "step": 24181
    },
    {
      "epoch": 0.8660089172202625,
      "grad_norm": 1.4630588293075562,
      "learning_rate": 9.267497408943393e-06,
      "loss": 1.5637,
      "step": 24182
    },
    {
      "epoch": 0.8660447293498308,
      "grad_norm": 1.9434194564819336,
      "learning_rate": 9.262621447497844e-06,
      "loss": 1.3992,
      "step": 24183
    },
    {
      "epoch": 0.8660805414793991,
      "grad_norm": 1.6045405864715576,
      "learning_rate": 9.257746706806658e-06,
      "loss": 1.2954,
      "step": 24184
    },
    {
      "epoch": 0.8661163536089673,
      "grad_norm": 1.6443135738372803,
      "learning_rate": 9.252873186935452e-06,
      "loss": 1.2575,
      "step": 24185
    },
    {
      "epoch": 0.8661521657385356,
      "grad_norm": 2.0701043605804443,
      "learning_rate": 9.248000887949782e-06,
      "loss": 1.5517,
      "step": 24186
    },
    {
      "epoch": 0.866187977868104,
      "grad_norm": 1.616709589958191,
      "learning_rate": 9.243129809915175e-06,
      "loss": 1.4381,
      "step": 24187
    },
    {
      "epoch": 0.8662237899976722,
      "grad_norm": 1.3892414569854736,
      "learning_rate": 9.238259952897221e-06,
      "loss": 1.4791,
      "step": 24188
    },
    {
      "epoch": 0.8662596021272405,
      "grad_norm": 1.8243262767791748,
      "learning_rate": 9.233391316961393e-06,
      "loss": 1.6,
      "step": 24189
    },
    {
      "epoch": 0.8662954142568088,
      "grad_norm": 1.4055460691452026,
      "learning_rate": 9.228523902173214e-06,
      "loss": 1.2116,
      "step": 24190
    },
    {
      "epoch": 0.866331226386377,
      "grad_norm": 1.6112884283065796,
      "learning_rate": 9.223657708598133e-06,
      "loss": 1.3763,
      "step": 24191
    },
    {
      "epoch": 0.8663670385159453,
      "grad_norm": 1.9256858825683594,
      "learning_rate": 9.218792736301674e-06,
      "loss": 1.5623,
      "step": 24192
    },
    {
      "epoch": 0.8664028506455136,
      "grad_norm": 1.4280561208724976,
      "learning_rate": 9.213928985349252e-06,
      "loss": 1.3904,
      "step": 24193
    },
    {
      "epoch": 0.866438662775082,
      "grad_norm": 1.9041266441345215,
      "learning_rate": 9.209066455806303e-06,
      "loss": 1.5014,
      "step": 24194
    },
    {
      "epoch": 0.8664744749046502,
      "grad_norm": 1.7539561986923218,
      "learning_rate": 9.204205147738254e-06,
      "loss": 1.3906,
      "step": 24195
    },
    {
      "epoch": 0.8665102870342185,
      "grad_norm": 1.7698585987091064,
      "learning_rate": 9.199345061210495e-06,
      "loss": 1.4611,
      "step": 24196
    },
    {
      "epoch": 0.8665460991637868,
      "grad_norm": 1.589356541633606,
      "learning_rate": 9.194486196288454e-06,
      "loss": 1.4255,
      "step": 24197
    },
    {
      "epoch": 0.866581911293355,
      "grad_norm": 1.2249170541763306,
      "learning_rate": 9.189628553037445e-06,
      "loss": 1.4406,
      "step": 24198
    },
    {
      "epoch": 0.8666177234229233,
      "grad_norm": 1.425215482711792,
      "learning_rate": 9.184772131522845e-06,
      "loss": 1.3896,
      "step": 24199
    },
    {
      "epoch": 0.8666535355524916,
      "grad_norm": 1.1758129596710205,
      "learning_rate": 9.179916931809995e-06,
      "loss": 1.323,
      "step": 24200
    },
    {
      "epoch": 0.86668934768206,
      "grad_norm": 1.505043387413025,
      "learning_rate": 9.175062953964242e-06,
      "loss": 1.5766,
      "step": 24201
    },
    {
      "epoch": 0.8667251598116282,
      "grad_norm": 1.791693091392517,
      "learning_rate": 9.170210198050833e-06,
      "loss": 1.1097,
      "step": 24202
    },
    {
      "epoch": 0.8667609719411965,
      "grad_norm": 1.9780690670013428,
      "learning_rate": 9.165358664135082e-06,
      "loss": 1.4566,
      "step": 24203
    },
    {
      "epoch": 0.8667967840707648,
      "grad_norm": 1.7749983072280884,
      "learning_rate": 9.160508352282282e-06,
      "loss": 1.4966,
      "step": 24204
    },
    {
      "epoch": 0.866832596200333,
      "grad_norm": 2.2015771865844727,
      "learning_rate": 9.155659262557648e-06,
      "loss": 1.4599,
      "step": 24205
    },
    {
      "epoch": 0.8668684083299013,
      "grad_norm": 1.5244667530059814,
      "learning_rate": 9.150811395026448e-06,
      "loss": 1.4312,
      "step": 24206
    },
    {
      "epoch": 0.8669042204594696,
      "grad_norm": 2.0955560207366943,
      "learning_rate": 9.145964749753888e-06,
      "loss": 1.7519,
      "step": 24207
    },
    {
      "epoch": 0.866940032589038,
      "grad_norm": 1.8348698616027832,
      "learning_rate": 9.141119326805193e-06,
      "loss": 1.5078,
      "step": 24208
    },
    {
      "epoch": 0.8669758447186062,
      "grad_norm": 1.442132830619812,
      "learning_rate": 9.13627512624552e-06,
      "loss": 1.3298,
      "step": 24209
    },
    {
      "epoch": 0.8670116568481745,
      "grad_norm": 1.7003968954086304,
      "learning_rate": 9.131432148140062e-06,
      "loss": 1.665,
      "step": 24210
    },
    {
      "epoch": 0.8670474689777428,
      "grad_norm": 2.5281410217285156,
      "learning_rate": 9.126590392553992e-06,
      "loss": 1.4594,
      "step": 24211
    },
    {
      "epoch": 0.867083281107311,
      "grad_norm": 2.014244556427002,
      "learning_rate": 9.12174985955241e-06,
      "loss": 1.2994,
      "step": 24212
    },
    {
      "epoch": 0.8671190932368793,
      "grad_norm": 1.3764430284500122,
      "learning_rate": 9.116910549200452e-06,
      "loss": 1.1514,
      "step": 24213
    },
    {
      "epoch": 0.8671549053664476,
      "grad_norm": 2.000931978225708,
      "learning_rate": 9.112072461563248e-06,
      "loss": 1.557,
      "step": 24214
    },
    {
      "epoch": 0.867190717496016,
      "grad_norm": 1.996650218963623,
      "learning_rate": 9.107235596705877e-06,
      "loss": 1.3858,
      "step": 24215
    },
    {
      "epoch": 0.8672265296255842,
      "grad_norm": 1.7024036645889282,
      "learning_rate": 9.102399954693396e-06,
      "loss": 1.5788,
      "step": 24216
    },
    {
      "epoch": 0.8672623417551525,
      "grad_norm": 1.849635362625122,
      "learning_rate": 9.097565535590869e-06,
      "loss": 1.4354,
      "step": 24217
    },
    {
      "epoch": 0.8672981538847208,
      "grad_norm": 1.4759061336517334,
      "learning_rate": 9.092732339463339e-06,
      "loss": 1.2717,
      "step": 24218
    },
    {
      "epoch": 0.867333966014289,
      "grad_norm": 1.4988124370574951,
      "learning_rate": 9.087900366375868e-06,
      "loss": 1.5389,
      "step": 24219
    },
    {
      "epoch": 0.8673697781438573,
      "grad_norm": 1.6259818077087402,
      "learning_rate": 9.083069616393392e-06,
      "loss": 1.3329,
      "step": 24220
    },
    {
      "epoch": 0.8674055902734256,
      "grad_norm": 2.060462713241577,
      "learning_rate": 9.078240089580948e-06,
      "loss": 1.7412,
      "step": 24221
    },
    {
      "epoch": 0.8674414024029939,
      "grad_norm": 1.432573676109314,
      "learning_rate": 9.073411786003527e-06,
      "loss": 1.658,
      "step": 24222
    },
    {
      "epoch": 0.8674772145325622,
      "grad_norm": 1.677588939666748,
      "learning_rate": 9.068584705726035e-06,
      "loss": 1.2352,
      "step": 24223
    },
    {
      "epoch": 0.8675130266621305,
      "grad_norm": 2.123272180557251,
      "learning_rate": 9.063758848813452e-06,
      "loss": 1.4148,
      "step": 24224
    },
    {
      "epoch": 0.8675488387916988,
      "grad_norm": 1.933234453201294,
      "learning_rate": 9.058934215330695e-06,
      "loss": 1.3967,
      "step": 24225
    },
    {
      "epoch": 0.867584650921267,
      "grad_norm": 1.572487711906433,
      "learning_rate": 9.054110805342686e-06,
      "loss": 1.4756,
      "step": 24226
    },
    {
      "epoch": 0.8676204630508353,
      "grad_norm": 2.7359111309051514,
      "learning_rate": 9.049288618914276e-06,
      "loss": 1.6118,
      "step": 24227
    },
    {
      "epoch": 0.8676562751804036,
      "grad_norm": 1.771386981010437,
      "learning_rate": 9.044467656110389e-06,
      "loss": 1.3877,
      "step": 24228
    },
    {
      "epoch": 0.8676920873099719,
      "grad_norm": 1.7388691902160645,
      "learning_rate": 9.039647916995874e-06,
      "loss": 1.2226,
      "step": 24229
    },
    {
      "epoch": 0.8677278994395402,
      "grad_norm": 2.149545192718506,
      "learning_rate": 9.034829401635547e-06,
      "loss": 1.3794,
      "step": 24230
    },
    {
      "epoch": 0.8677637115691085,
      "grad_norm": 1.9506735801696777,
      "learning_rate": 9.030012110094255e-06,
      "loss": 1.5796,
      "step": 24231
    },
    {
      "epoch": 0.8677995236986767,
      "grad_norm": 1.4188599586486816,
      "learning_rate": 9.025196042436802e-06,
      "loss": 1.5882,
      "step": 24232
    },
    {
      "epoch": 0.867835335828245,
      "grad_norm": 1.2901597023010254,
      "learning_rate": 9.020381198728011e-06,
      "loss": 1.6489,
      "step": 24233
    },
    {
      "epoch": 0.8678711479578133,
      "grad_norm": 1.7088090181350708,
      "learning_rate": 9.015567579032614e-06,
      "loss": 1.5318,
      "step": 24234
    },
    {
      "epoch": 0.8679069600873816,
      "grad_norm": 1.7100541591644287,
      "learning_rate": 9.010755183415398e-06,
      "loss": 1.4519,
      "step": 24235
    },
    {
      "epoch": 0.8679427722169499,
      "grad_norm": 1.6215293407440186,
      "learning_rate": 9.005944011941103e-06,
      "loss": 1.4229,
      "step": 24236
    },
    {
      "epoch": 0.8679785843465182,
      "grad_norm": 1.7219916582107544,
      "learning_rate": 9.001134064674476e-06,
      "loss": 1.4134,
      "step": 24237
    },
    {
      "epoch": 0.8680143964760865,
      "grad_norm": 1.3429756164550781,
      "learning_rate": 8.99632534168019e-06,
      "loss": 1.6244,
      "step": 24238
    },
    {
      "epoch": 0.8680502086056547,
      "grad_norm": 1.7451717853546143,
      "learning_rate": 8.991517843022968e-06,
      "loss": 1.5262,
      "step": 24239
    },
    {
      "epoch": 0.868086020735223,
      "grad_norm": 1.512374758720398,
      "learning_rate": 8.986711568767493e-06,
      "loss": 1.2379,
      "step": 24240
    },
    {
      "epoch": 0.8681218328647913,
      "grad_norm": 2.092271089553833,
      "learning_rate": 8.981906518978389e-06,
      "loss": 1.6425,
      "step": 24241
    },
    {
      "epoch": 0.8681576449943595,
      "grad_norm": 1.376625657081604,
      "learning_rate": 8.977102693720341e-06,
      "loss": 1.5389,
      "step": 24242
    },
    {
      "epoch": 0.8681934571239279,
      "grad_norm": 1.661430835723877,
      "learning_rate": 8.97230009305795e-06,
      "loss": 1.507,
      "step": 24243
    },
    {
      "epoch": 0.8682292692534962,
      "grad_norm": 2.269094944000244,
      "learning_rate": 8.967498717055878e-06,
      "loss": 1.596,
      "step": 24244
    },
    {
      "epoch": 0.8682650813830645,
      "grad_norm": 1.464607834815979,
      "learning_rate": 8.96269856577866e-06,
      "loss": 1.4243,
      "step": 24245
    },
    {
      "epoch": 0.8683008935126327,
      "grad_norm": 1.3154370784759521,
      "learning_rate": 8.9578996392909e-06,
      "loss": 1.2494,
      "step": 24246
    },
    {
      "epoch": 0.868336705642201,
      "grad_norm": 1.6771907806396484,
      "learning_rate": 8.953101937657194e-06,
      "loss": 1.4635,
      "step": 24247
    },
    {
      "epoch": 0.8683725177717693,
      "grad_norm": 1.5361515283584595,
      "learning_rate": 8.94830546094203e-06,
      "loss": 1.6276,
      "step": 24248
    },
    {
      "epoch": 0.8684083299013375,
      "grad_norm": 1.7470072507858276,
      "learning_rate": 8.943510209209971e-06,
      "loss": 1.6124,
      "step": 24249
    },
    {
      "epoch": 0.8684441420309059,
      "grad_norm": 1.491672396659851,
      "learning_rate": 8.93871618252553e-06,
      "loss": 1.2757,
      "step": 24250
    },
    {
      "epoch": 0.8684799541604742,
      "grad_norm": 1.5673433542251587,
      "learning_rate": 8.933923380953224e-06,
      "loss": 1.0436,
      "step": 24251
    },
    {
      "epoch": 0.8685157662900425,
      "grad_norm": 1.1775404214859009,
      "learning_rate": 8.92913180455749e-06,
      "loss": 1.2953,
      "step": 24252
    },
    {
      "epoch": 0.8685515784196107,
      "grad_norm": 1.688716173171997,
      "learning_rate": 8.924341453402817e-06,
      "loss": 1.3285,
      "step": 24253
    },
    {
      "epoch": 0.868587390549179,
      "grad_norm": 1.6891595125198364,
      "learning_rate": 8.919552327553648e-06,
      "loss": 1.5579,
      "step": 24254
    },
    {
      "epoch": 0.8686232026787473,
      "grad_norm": 1.63966703414917,
      "learning_rate": 8.914764427074428e-06,
      "loss": 1.4901,
      "step": 24255
    },
    {
      "epoch": 0.8686590148083155,
      "grad_norm": 1.675835371017456,
      "learning_rate": 8.909977752029574e-06,
      "loss": 1.3802,
      "step": 24256
    },
    {
      "epoch": 0.8686948269378839,
      "grad_norm": 2.029541492462158,
      "learning_rate": 8.905192302483433e-06,
      "loss": 1.4674,
      "step": 24257
    },
    {
      "epoch": 0.8687306390674522,
      "grad_norm": 1.7141716480255127,
      "learning_rate": 8.900408078500454e-06,
      "loss": 1.4041,
      "step": 24258
    },
    {
      "epoch": 0.8687664511970205,
      "grad_norm": 1.9804600477218628,
      "learning_rate": 8.895625080144965e-06,
      "loss": 1.4072,
      "step": 24259
    },
    {
      "epoch": 0.8688022633265887,
      "grad_norm": 1.474840760231018,
      "learning_rate": 8.890843307481322e-06,
      "loss": 1.3555,
      "step": 24260
    },
    {
      "epoch": 0.868838075456157,
      "grad_norm": 1.8305799961090088,
      "learning_rate": 8.886062760573854e-06,
      "loss": 1.4299,
      "step": 24261
    },
    {
      "epoch": 0.8688738875857253,
      "grad_norm": 1.6935603618621826,
      "learning_rate": 8.88128343948691e-06,
      "loss": 1.416,
      "step": 24262
    },
    {
      "epoch": 0.8689096997152935,
      "grad_norm": 1.5021482706069946,
      "learning_rate": 8.876505344284758e-06,
      "loss": 1.318,
      "step": 24263
    },
    {
      "epoch": 0.8689455118448619,
      "grad_norm": 2.0541741847991943,
      "learning_rate": 8.871728475031649e-06,
      "loss": 1.6846,
      "step": 24264
    },
    {
      "epoch": 0.8689813239744302,
      "grad_norm": 1.6807490587234497,
      "learning_rate": 8.86695283179192e-06,
      "loss": 1.2177,
      "step": 24265
    },
    {
      "epoch": 0.8690171361039984,
      "grad_norm": 1.686801552772522,
      "learning_rate": 8.862178414629774e-06,
      "loss": 1.6565,
      "step": 24266
    },
    {
      "epoch": 0.8690529482335667,
      "grad_norm": 1.6299697160720825,
      "learning_rate": 8.857405223609472e-06,
      "loss": 1.5448,
      "step": 24267
    },
    {
      "epoch": 0.869088760363135,
      "grad_norm": 2.021028518676758,
      "learning_rate": 8.852633258795185e-06,
      "loss": 1.4393,
      "step": 24268
    },
    {
      "epoch": 0.8691245724927033,
      "grad_norm": 1.4445551633834839,
      "learning_rate": 8.847862520251182e-06,
      "loss": 1.1612,
      "step": 24269
    },
    {
      "epoch": 0.8691603846222715,
      "grad_norm": 1.3987027406692505,
      "learning_rate": 8.843093008041591e-06,
      "loss": 1.5816,
      "step": 24270
    },
    {
      "epoch": 0.8691961967518399,
      "grad_norm": 1.6368407011032104,
      "learning_rate": 8.838324722230595e-06,
      "loss": 1.1717,
      "step": 24271
    },
    {
      "epoch": 0.8692320088814082,
      "grad_norm": 1.2989962100982666,
      "learning_rate": 8.833557662882374e-06,
      "loss": 1.4289,
      "step": 24272
    },
    {
      "epoch": 0.8692678210109764,
      "grad_norm": 1.8490816354751587,
      "learning_rate": 8.828791830061022e-06,
      "loss": 1.2979,
      "step": 24273
    },
    {
      "epoch": 0.8693036331405447,
      "grad_norm": 1.353345274925232,
      "learning_rate": 8.824027223830688e-06,
      "loss": 1.4558,
      "step": 24274
    },
    {
      "epoch": 0.869339445270113,
      "grad_norm": 1.5162136554718018,
      "learning_rate": 8.819263844255432e-06,
      "loss": 1.5938,
      "step": 24275
    },
    {
      "epoch": 0.8693752573996812,
      "grad_norm": 1.5665966272354126,
      "learning_rate": 8.81450169139939e-06,
      "loss": 1.3911,
      "step": 24276
    },
    {
      "epoch": 0.8694110695292495,
      "grad_norm": 1.4457532167434692,
      "learning_rate": 8.809740765326591e-06,
      "loss": 1.2053,
      "step": 24277
    },
    {
      "epoch": 0.8694468816588179,
      "grad_norm": 2.0011956691741943,
      "learning_rate": 8.804981066101126e-06,
      "loss": 1.682,
      "step": 24278
    },
    {
      "epoch": 0.8694826937883862,
      "grad_norm": 1.6146661043167114,
      "learning_rate": 8.800222593786967e-06,
      "loss": 1.6388,
      "step": 24279
    },
    {
      "epoch": 0.8695185059179544,
      "grad_norm": 1.7931370735168457,
      "learning_rate": 8.795465348448218e-06,
      "loss": 1.3895,
      "step": 24280
    },
    {
      "epoch": 0.8695543180475227,
      "grad_norm": 3.107492446899414,
      "learning_rate": 8.790709330148828e-06,
      "loss": 1.2655,
      "step": 24281
    },
    {
      "epoch": 0.869590130177091,
      "grad_norm": 2.4997458457946777,
      "learning_rate": 8.78595453895278e-06,
      "loss": 1.363,
      "step": 24282
    },
    {
      "epoch": 0.8696259423066592,
      "grad_norm": 1.7735134363174438,
      "learning_rate": 8.781200974924053e-06,
      "loss": 1.5793,
      "step": 24283
    },
    {
      "epoch": 0.8696617544362275,
      "grad_norm": 1.5544660091400146,
      "learning_rate": 8.7764486381266e-06,
      "loss": 1.3186,
      "step": 24284
    },
    {
      "epoch": 0.8696975665657959,
      "grad_norm": 1.9983875751495361,
      "learning_rate": 8.77169752862439e-06,
      "loss": 1.5833,
      "step": 24285
    },
    {
      "epoch": 0.8697333786953642,
      "grad_norm": 1.391190767288208,
      "learning_rate": 8.76694764648126e-06,
      "loss": 1.5987,
      "step": 24286
    },
    {
      "epoch": 0.8697691908249324,
      "grad_norm": 1.8695008754730225,
      "learning_rate": 8.762198991761217e-06,
      "loss": 1.2977,
      "step": 24287
    },
    {
      "epoch": 0.8698050029545007,
      "grad_norm": 1.3335654735565186,
      "learning_rate": 8.757451564528074e-06,
      "loss": 1.2562,
      "step": 24288
    },
    {
      "epoch": 0.869840815084069,
      "grad_norm": 1.5261073112487793,
      "learning_rate": 8.752705364845748e-06,
      "loss": 1.3221,
      "step": 24289
    },
    {
      "epoch": 0.8698766272136372,
      "grad_norm": 1.8547850847244263,
      "learning_rate": 8.747960392778053e-06,
      "loss": 1.3653,
      "step": 24290
    },
    {
      "epoch": 0.8699124393432055,
      "grad_norm": 1.607026219367981,
      "learning_rate": 8.74321664838884e-06,
      "loss": 1.3366,
      "step": 24291
    },
    {
      "epoch": 0.8699482514727739,
      "grad_norm": 1.540658950805664,
      "learning_rate": 8.738474131741958e-06,
      "loss": 1.1223,
      "step": 24292
    },
    {
      "epoch": 0.8699840636023422,
      "grad_norm": 2.3103864192962646,
      "learning_rate": 8.733732842901166e-06,
      "loss": 1.3668,
      "step": 24293
    },
    {
      "epoch": 0.8700198757319104,
      "grad_norm": 1.749535322189331,
      "learning_rate": 8.728992781930278e-06,
      "loss": 1.3763,
      "step": 24294
    },
    {
      "epoch": 0.8700556878614787,
      "grad_norm": 1.4846516847610474,
      "learning_rate": 8.724253948893057e-06,
      "loss": 1.4243,
      "step": 24295
    },
    {
      "epoch": 0.870091499991047,
      "grad_norm": 1.5358998775482178,
      "learning_rate": 8.719516343853273e-06,
      "loss": 1.5699,
      "step": 24296
    },
    {
      "epoch": 0.8701273121206152,
      "grad_norm": 1.1911826133728027,
      "learning_rate": 8.71477996687463e-06,
      "loss": 1.5351,
      "step": 24297
    },
    {
      "epoch": 0.8701631242501835,
      "grad_norm": 2.5798017978668213,
      "learning_rate": 8.710044818020902e-06,
      "loss": 1.2529,
      "step": 24298
    },
    {
      "epoch": 0.8701989363797519,
      "grad_norm": 1.7193776369094849,
      "learning_rate": 8.705310897355768e-06,
      "loss": 1.4123,
      "step": 24299
    },
    {
      "epoch": 0.8702347485093201,
      "grad_norm": 1.6127065420150757,
      "learning_rate": 8.700578204942889e-06,
      "loss": 1.6177,
      "step": 24300
    },
    {
      "epoch": 0.8702705606388884,
      "grad_norm": 1.6372747421264648,
      "learning_rate": 8.69584674084597e-06,
      "loss": 0.992,
      "step": 24301
    },
    {
      "epoch": 0.8703063727684567,
      "grad_norm": 1.3636530637741089,
      "learning_rate": 8.69111650512866e-06,
      "loss": 1.3641,
      "step": 24302
    },
    {
      "epoch": 0.870342184898025,
      "grad_norm": 1.8584938049316406,
      "learning_rate": 8.686387497854609e-06,
      "loss": 1.3151,
      "step": 24303
    },
    {
      "epoch": 0.8703779970275932,
      "grad_norm": 1.4046014547348022,
      "learning_rate": 8.681659719087421e-06,
      "loss": 1.7345,
      "step": 24304
    },
    {
      "epoch": 0.8704138091571615,
      "grad_norm": 1.3433245420455933,
      "learning_rate": 8.676933168890699e-06,
      "loss": 1.2366,
      "step": 24305
    },
    {
      "epoch": 0.8704496212867299,
      "grad_norm": 2.0000412464141846,
      "learning_rate": 8.67220784732804e-06,
      "loss": 1.1339,
      "step": 24306
    },
    {
      "epoch": 0.8704854334162981,
      "grad_norm": 1.5002861022949219,
      "learning_rate": 8.667483754463046e-06,
      "loss": 1.4521,
      "step": 24307
    },
    {
      "epoch": 0.8705212455458664,
      "grad_norm": 1.84786856174469,
      "learning_rate": 8.662760890359233e-06,
      "loss": 1.2317,
      "step": 24308
    },
    {
      "epoch": 0.8705570576754347,
      "grad_norm": 1.4311268329620361,
      "learning_rate": 8.658039255080153e-06,
      "loss": 1.4301,
      "step": 24309
    },
    {
      "epoch": 0.870592869805003,
      "grad_norm": 1.6267906427383423,
      "learning_rate": 8.65331884868934e-06,
      "loss": 1.6193,
      "step": 24310
    },
    {
      "epoch": 0.8706286819345712,
      "grad_norm": 1.5863173007965088,
      "learning_rate": 8.64859967125029e-06,
      "loss": 1.4066,
      "step": 24311
    },
    {
      "epoch": 0.8706644940641395,
      "grad_norm": 1.7492663860321045,
      "learning_rate": 8.643881722826486e-06,
      "loss": 1.1955,
      "step": 24312
    },
    {
      "epoch": 0.8707003061937079,
      "grad_norm": 1.6891770362854004,
      "learning_rate": 8.639165003481408e-06,
      "loss": 1.2417,
      "step": 24313
    },
    {
      "epoch": 0.8707361183232761,
      "grad_norm": 1.444863200187683,
      "learning_rate": 8.634449513278553e-06,
      "loss": 1.4051,
      "step": 24314
    },
    {
      "epoch": 0.8707719304528444,
      "grad_norm": 2.0153067111968994,
      "learning_rate": 8.629735252281301e-06,
      "loss": 1.3517,
      "step": 24315
    },
    {
      "epoch": 0.8708077425824127,
      "grad_norm": 1.9217655658721924,
      "learning_rate": 8.625022220553091e-06,
      "loss": 1.282,
      "step": 24316
    },
    {
      "epoch": 0.8708435547119809,
      "grad_norm": 1.494198203086853,
      "learning_rate": 8.620310418157374e-06,
      "loss": 1.2652,
      "step": 24317
    },
    {
      "epoch": 0.8708793668415492,
      "grad_norm": 1.5170859098434448,
      "learning_rate": 8.615599845157484e-06,
      "loss": 1.7015,
      "step": 24318
    },
    {
      "epoch": 0.8709151789711175,
      "grad_norm": 1.4580488204956055,
      "learning_rate": 8.61089050161683e-06,
      "loss": 1.7065,
      "step": 24319
    },
    {
      "epoch": 0.8709509911006859,
      "grad_norm": 2.421257734298706,
      "learning_rate": 8.60618238759875e-06,
      "loss": 1.3544,
      "step": 24320
    },
    {
      "epoch": 0.8709868032302541,
      "grad_norm": 1.232548713684082,
      "learning_rate": 8.601475503166623e-06,
      "loss": 1.237,
      "step": 24321
    },
    {
      "epoch": 0.8710226153598224,
      "grad_norm": 1.703403115272522,
      "learning_rate": 8.596769848383723e-06,
      "loss": 1.4257,
      "step": 24322
    },
    {
      "epoch": 0.8710584274893907,
      "grad_norm": 1.6034023761749268,
      "learning_rate": 8.592065423313378e-06,
      "loss": 1.2383,
      "step": 24323
    },
    {
      "epoch": 0.8710942396189589,
      "grad_norm": 1.508851408958435,
      "learning_rate": 8.587362228018892e-06,
      "loss": 1.5151,
      "step": 24324
    },
    {
      "epoch": 0.8711300517485272,
      "grad_norm": 1.593440055847168,
      "learning_rate": 8.582660262563558e-06,
      "loss": 1.5984,
      "step": 24325
    },
    {
      "epoch": 0.8711658638780955,
      "grad_norm": 2.054398775100708,
      "learning_rate": 8.577959527010582e-06,
      "loss": 1.169,
      "step": 24326
    },
    {
      "epoch": 0.8712016760076638,
      "grad_norm": 1.829206943511963,
      "learning_rate": 8.573260021423236e-06,
      "loss": 1.2653,
      "step": 24327
    },
    {
      "epoch": 0.8712374881372321,
      "grad_norm": 1.868972897529602,
      "learning_rate": 8.568561745864766e-06,
      "loss": 1.5099,
      "step": 24328
    },
    {
      "epoch": 0.8712733002668004,
      "grad_norm": 1.5950393676757812,
      "learning_rate": 8.563864700398338e-06,
      "loss": 1.6054,
      "step": 24329
    },
    {
      "epoch": 0.8713091123963687,
      "grad_norm": 1.582476258277893,
      "learning_rate": 8.559168885087165e-06,
      "loss": 1.603,
      "step": 24330
    },
    {
      "epoch": 0.8713449245259369,
      "grad_norm": 1.48959481716156,
      "learning_rate": 8.554474299994431e-06,
      "loss": 1.4577,
      "step": 24331
    },
    {
      "epoch": 0.8713807366555052,
      "grad_norm": 1.9258126020431519,
      "learning_rate": 8.549780945183306e-06,
      "loss": 1.4982,
      "step": 24332
    },
    {
      "epoch": 0.8714165487850735,
      "grad_norm": 2.175291061401367,
      "learning_rate": 8.545088820716895e-06,
      "loss": 1.5243,
      "step": 24333
    },
    {
      "epoch": 0.8714523609146418,
      "grad_norm": 1.4352703094482422,
      "learning_rate": 8.54039792665835e-06,
      "loss": 1.2539,
      "step": 24334
    },
    {
      "epoch": 0.8714881730442101,
      "grad_norm": 1.5487613677978516,
      "learning_rate": 8.535708263070785e-06,
      "loss": 1.5298,
      "step": 24335
    },
    {
      "epoch": 0.8715239851737784,
      "grad_norm": 1.46132230758667,
      "learning_rate": 8.531019830017272e-06,
      "loss": 1.4797,
      "step": 24336
    },
    {
      "epoch": 0.8715597973033467,
      "grad_norm": 1.3812408447265625,
      "learning_rate": 8.526332627560906e-06,
      "loss": 1.1609,
      "step": 24337
    },
    {
      "epoch": 0.8715956094329149,
      "grad_norm": 1.4614298343658447,
      "learning_rate": 8.521646655764736e-06,
      "loss": 1.2881,
      "step": 24338
    },
    {
      "epoch": 0.8716314215624832,
      "grad_norm": 1.8450669050216675,
      "learning_rate": 8.516961914691835e-06,
      "loss": 1.1969,
      "step": 24339
    },
    {
      "epoch": 0.8716672336920515,
      "grad_norm": 2.6135294437408447,
      "learning_rate": 8.512278404405182e-06,
      "loss": 1.5435,
      "step": 24340
    },
    {
      "epoch": 0.8717030458216198,
      "grad_norm": 1.3671050071716309,
      "learning_rate": 8.507596124967821e-06,
      "loss": 1.4997,
      "step": 24341
    },
    {
      "epoch": 0.8717388579511881,
      "grad_norm": 1.8961228132247925,
      "learning_rate": 8.50291507644273e-06,
      "loss": 1.3494,
      "step": 24342
    },
    {
      "epoch": 0.8717746700807564,
      "grad_norm": 1.6030116081237793,
      "learning_rate": 8.498235258892907e-06,
      "loss": 1.6106,
      "step": 24343
    },
    {
      "epoch": 0.8718104822103246,
      "grad_norm": 1.4362305402755737,
      "learning_rate": 8.493556672381297e-06,
      "loss": 1.4159,
      "step": 24344
    },
    {
      "epoch": 0.8718462943398929,
      "grad_norm": 1.694785237312317,
      "learning_rate": 8.488879316970832e-06,
      "loss": 1.3855,
      "step": 24345
    },
    {
      "epoch": 0.8718821064694612,
      "grad_norm": 1.857426643371582,
      "learning_rate": 8.484203192724482e-06,
      "loss": 1.4179,
      "step": 24346
    },
    {
      "epoch": 0.8719179185990295,
      "grad_norm": 1.4686394929885864,
      "learning_rate": 8.479528299705108e-06,
      "loss": 1.217,
      "step": 24347
    },
    {
      "epoch": 0.8719537307285978,
      "grad_norm": 1.5443724393844604,
      "learning_rate": 8.474854637975638e-06,
      "loss": 1.3838,
      "step": 24348
    },
    {
      "epoch": 0.8719895428581661,
      "grad_norm": 1.604174017906189,
      "learning_rate": 8.47018220759893e-06,
      "loss": 1.8491,
      "step": 24349
    },
    {
      "epoch": 0.8720253549877344,
      "grad_norm": 1.6408330202102661,
      "learning_rate": 8.465511008637872e-06,
      "loss": 1.503,
      "step": 24350
    },
    {
      "epoch": 0.8720611671173026,
      "grad_norm": 2.2034032344818115,
      "learning_rate": 8.460841041155277e-06,
      "loss": 1.7287,
      "step": 24351
    },
    {
      "epoch": 0.8720969792468709,
      "grad_norm": 2.2767324447631836,
      "learning_rate": 8.456172305213995e-06,
      "loss": 1.2505,
      "step": 24352
    },
    {
      "epoch": 0.8721327913764392,
      "grad_norm": 1.8010529279708862,
      "learning_rate": 8.45150480087684e-06,
      "loss": 1.4255,
      "step": 24353
    },
    {
      "epoch": 0.8721686035060074,
      "grad_norm": 1.6472227573394775,
      "learning_rate": 8.44683852820659e-06,
      "loss": 1.5661,
      "step": 24354
    },
    {
      "epoch": 0.8722044156355758,
      "grad_norm": 1.3859483003616333,
      "learning_rate": 8.442173487266047e-06,
      "loss": 1.047,
      "step": 24355
    },
    {
      "epoch": 0.8722402277651441,
      "grad_norm": 1.3888520002365112,
      "learning_rate": 8.437509678117916e-06,
      "loss": 1.2829,
      "step": 24356
    },
    {
      "epoch": 0.8722760398947124,
      "grad_norm": 1.3679659366607666,
      "learning_rate": 8.432847100825025e-06,
      "loss": 1.2947,
      "step": 24357
    },
    {
      "epoch": 0.8723118520242806,
      "grad_norm": 2.372196912765503,
      "learning_rate": 8.428185755450047e-06,
      "loss": 1.318,
      "step": 24358
    },
    {
      "epoch": 0.8723476641538489,
      "grad_norm": 1.549730896949768,
      "learning_rate": 8.423525642055719e-06,
      "loss": 1.5575,
      "step": 24359
    },
    {
      "epoch": 0.8723834762834172,
      "grad_norm": 1.8781365156173706,
      "learning_rate": 8.418866760704735e-06,
      "loss": 1.4678,
      "step": 24360
    },
    {
      "epoch": 0.8724192884129854,
      "grad_norm": 2.044919729232788,
      "learning_rate": 8.414209111459747e-06,
      "loss": 1.4384,
      "step": 24361
    },
    {
      "epoch": 0.8724551005425537,
      "grad_norm": 1.8532249927520752,
      "learning_rate": 8.409552694383472e-06,
      "loss": 1.1847,
      "step": 24362
    },
    {
      "epoch": 0.8724909126721221,
      "grad_norm": 1.6493250131607056,
      "learning_rate": 8.404897509538468e-06,
      "loss": 1.3719,
      "step": 24363
    },
    {
      "epoch": 0.8725267248016904,
      "grad_norm": 1.678415298461914,
      "learning_rate": 8.400243556987464e-06,
      "loss": 1.0674,
      "step": 24364
    },
    {
      "epoch": 0.8725625369312586,
      "grad_norm": 1.7217615842819214,
      "learning_rate": 8.39559083679301e-06,
      "loss": 1.647,
      "step": 24365
    },
    {
      "epoch": 0.8725983490608269,
      "grad_norm": 1.3149148225784302,
      "learning_rate": 8.390939349017735e-06,
      "loss": 1.2332,
      "step": 24366
    },
    {
      "epoch": 0.8726341611903952,
      "grad_norm": 1.6030855178833008,
      "learning_rate": 8.386289093724175e-06,
      "loss": 1.3858,
      "step": 24367
    },
    {
      "epoch": 0.8726699733199634,
      "grad_norm": 1.845257043838501,
      "learning_rate": 8.38164007097495e-06,
      "loss": 1.278,
      "step": 24368
    },
    {
      "epoch": 0.8727057854495317,
      "grad_norm": 1.8243887424468994,
      "learning_rate": 8.376992280832574e-06,
      "loss": 1.4684,
      "step": 24369
    },
    {
      "epoch": 0.8727415975791001,
      "grad_norm": 1.6303426027297974,
      "learning_rate": 8.372345723359553e-06,
      "loss": 1.6383,
      "step": 24370
    },
    {
      "epoch": 0.8727774097086684,
      "grad_norm": 1.4345486164093018,
      "learning_rate": 8.367700398618472e-06,
      "loss": 1.2597,
      "step": 24371
    },
    {
      "epoch": 0.8728132218382366,
      "grad_norm": 1.5594035387039185,
      "learning_rate": 8.363056306671757e-06,
      "loss": 1.1994,
      "step": 24372
    },
    {
      "epoch": 0.8728490339678049,
      "grad_norm": 1.6975489854812622,
      "learning_rate": 8.358413447581937e-06,
      "loss": 1.1322,
      "step": 24373
    },
    {
      "epoch": 0.8728848460973732,
      "grad_norm": 1.6841044425964355,
      "learning_rate": 8.353771821411415e-06,
      "loss": 1.8269,
      "step": 24374
    },
    {
      "epoch": 0.8729206582269414,
      "grad_norm": 1.4348398447036743,
      "learning_rate": 8.349131428222723e-06,
      "loss": 1.4246,
      "step": 24375
    },
    {
      "epoch": 0.8729564703565097,
      "grad_norm": 2.420299768447876,
      "learning_rate": 8.344492268078219e-06,
      "loss": 1.3549,
      "step": 24376
    },
    {
      "epoch": 0.8729922824860781,
      "grad_norm": 1.3469812870025635,
      "learning_rate": 8.339854341040376e-06,
      "loss": 1.4121,
      "step": 24377
    },
    {
      "epoch": 0.8730280946156463,
      "grad_norm": 1.9128682613372803,
      "learning_rate": 8.335217647171533e-06,
      "loss": 1.5137,
      "step": 24378
    },
    {
      "epoch": 0.8730639067452146,
      "grad_norm": 1.6239603757858276,
      "learning_rate": 8.330582186534097e-06,
      "loss": 1.3636,
      "step": 24379
    },
    {
      "epoch": 0.8730997188747829,
      "grad_norm": 1.9807374477386475,
      "learning_rate": 8.32594795919045e-06,
      "loss": 1.7255,
      "step": 24380
    },
    {
      "epoch": 0.8731355310043512,
      "grad_norm": 1.6199429035186768,
      "learning_rate": 8.321314965202898e-06,
      "loss": 1.3875,
      "step": 24381
    },
    {
      "epoch": 0.8731713431339194,
      "grad_norm": 1.5612422227859497,
      "learning_rate": 8.316683204633814e-06,
      "loss": 1.3494,
      "step": 24382
    },
    {
      "epoch": 0.8732071552634877,
      "grad_norm": 2.1652863025665283,
      "learning_rate": 8.312052677545478e-06,
      "loss": 1.4789,
      "step": 24383
    },
    {
      "epoch": 0.8732429673930561,
      "grad_norm": 1.5461417436599731,
      "learning_rate": 8.307423384000224e-06,
      "loss": 1.6285,
      "step": 24384
    },
    {
      "epoch": 0.8732787795226243,
      "grad_norm": 1.2789685726165771,
      "learning_rate": 8.302795324060287e-06,
      "loss": 1.2883,
      "step": 24385
    },
    {
      "epoch": 0.8733145916521926,
      "grad_norm": 2.073131799697876,
      "learning_rate": 8.298168497787984e-06,
      "loss": 1.3102,
      "step": 24386
    },
    {
      "epoch": 0.8733504037817609,
      "grad_norm": 1.3319644927978516,
      "learning_rate": 8.293542905245543e-06,
      "loss": 1.1161,
      "step": 24387
    },
    {
      "epoch": 0.8733862159113291,
      "grad_norm": 1.5022635459899902,
      "learning_rate": 8.288918546495172e-06,
      "loss": 1.5562,
      "step": 24388
    },
    {
      "epoch": 0.8734220280408974,
      "grad_norm": 2.067774534225464,
      "learning_rate": 8.284295421599097e-06,
      "loss": 1.5094,
      "step": 24389
    },
    {
      "epoch": 0.8734578401704657,
      "grad_norm": 1.600555658340454,
      "learning_rate": 8.279673530619525e-06,
      "loss": 1.6114,
      "step": 24390
    },
    {
      "epoch": 0.8734936523000341,
      "grad_norm": 1.613824486732483,
      "learning_rate": 8.27505287361866e-06,
      "loss": 1.443,
      "step": 24391
    },
    {
      "epoch": 0.8735294644296023,
      "grad_norm": 1.7431727647781372,
      "learning_rate": 8.270433450658621e-06,
      "loss": 1.5098,
      "step": 24392
    },
    {
      "epoch": 0.8735652765591706,
      "grad_norm": 1.3850111961364746,
      "learning_rate": 8.265815261801568e-06,
      "loss": 1.5287,
      "step": 24393
    },
    {
      "epoch": 0.8736010886887389,
      "grad_norm": 1.909886360168457,
      "learning_rate": 8.261198307109651e-06,
      "loss": 1.533,
      "step": 24394
    },
    {
      "epoch": 0.8736369008183071,
      "grad_norm": 1.8211636543273926,
      "learning_rate": 8.25658258664499e-06,
      "loss": 1.6334,
      "step": 24395
    },
    {
      "epoch": 0.8736727129478754,
      "grad_norm": 1.3177467584609985,
      "learning_rate": 8.251968100469653e-06,
      "loss": 1.3133,
      "step": 24396
    },
    {
      "epoch": 0.8737085250774437,
      "grad_norm": 1.4866849184036255,
      "learning_rate": 8.247354848645738e-06,
      "loss": 1.5015,
      "step": 24397
    },
    {
      "epoch": 0.8737443372070121,
      "grad_norm": 1.5447543859481812,
      "learning_rate": 8.242742831235339e-06,
      "loss": 1.5953,
      "step": 24398
    },
    {
      "epoch": 0.8737801493365803,
      "grad_norm": 1.414366602897644,
      "learning_rate": 8.23813204830045e-06,
      "loss": 1.1296,
      "step": 24399
    },
    {
      "epoch": 0.8738159614661486,
      "grad_norm": 1.600353479385376,
      "learning_rate": 8.233522499903123e-06,
      "loss": 1.287,
      "step": 24400
    },
    {
      "epoch": 0.8738517735957169,
      "grad_norm": 1.3291378021240234,
      "learning_rate": 8.228914186105397e-06,
      "loss": 1.2978,
      "step": 24401
    },
    {
      "epoch": 0.8738875857252851,
      "grad_norm": 1.3929870128631592,
      "learning_rate": 8.224307106969264e-06,
      "loss": 1.3377,
      "step": 24402
    },
    {
      "epoch": 0.8739233978548534,
      "grad_norm": 1.5553607940673828,
      "learning_rate": 8.219701262556678e-06,
      "loss": 1.3124,
      "step": 24403
    },
    {
      "epoch": 0.8739592099844217,
      "grad_norm": 1.6648629903793335,
      "learning_rate": 8.21509665292962e-06,
      "loss": 1.2963,
      "step": 24404
    },
    {
      "epoch": 0.87399502211399,
      "grad_norm": 1.5509344339370728,
      "learning_rate": 8.210493278150066e-06,
      "loss": 1.3433,
      "step": 24405
    },
    {
      "epoch": 0.8740308342435583,
      "grad_norm": 1.741041660308838,
      "learning_rate": 8.205891138279898e-06,
      "loss": 1.3943,
      "step": 24406
    },
    {
      "epoch": 0.8740666463731266,
      "grad_norm": 1.877867341041565,
      "learning_rate": 8.201290233381075e-06,
      "loss": 1.662,
      "step": 24407
    },
    {
      "epoch": 0.8741024585026949,
      "grad_norm": 1.2536922693252563,
      "learning_rate": 8.196690563515463e-06,
      "loss": 0.8697,
      "step": 24408
    },
    {
      "epoch": 0.8741382706322631,
      "grad_norm": 2.1341769695281982,
      "learning_rate": 8.192092128744988e-06,
      "loss": 1.1999,
      "step": 24409
    },
    {
      "epoch": 0.8741740827618314,
      "grad_norm": 1.4079196453094482,
      "learning_rate": 8.187494929131478e-06,
      "loss": 1.447,
      "step": 24410
    },
    {
      "epoch": 0.8742098948913997,
      "grad_norm": 1.608680009841919,
      "learning_rate": 8.182898964736785e-06,
      "loss": 1.5774,
      "step": 24411
    },
    {
      "epoch": 0.874245707020968,
      "grad_norm": 1.5955612659454346,
      "learning_rate": 8.178304235622758e-06,
      "loss": 1.735,
      "step": 24412
    },
    {
      "epoch": 0.8742815191505363,
      "grad_norm": 1.7357336282730103,
      "learning_rate": 8.173710741851215e-06,
      "loss": 1.4914,
      "step": 24413
    },
    {
      "epoch": 0.8743173312801046,
      "grad_norm": 1.8703807592391968,
      "learning_rate": 8.169118483483928e-06,
      "loss": 1.552,
      "step": 24414
    },
    {
      "epoch": 0.8743531434096729,
      "grad_norm": 1.617150902748108,
      "learning_rate": 8.164527460582705e-06,
      "loss": 1.4209,
      "step": 24415
    },
    {
      "epoch": 0.8743889555392411,
      "grad_norm": 2.064117670059204,
      "learning_rate": 8.159937673209327e-06,
      "loss": 1.6424,
      "step": 24416
    },
    {
      "epoch": 0.8744247676688094,
      "grad_norm": 1.8013323545455933,
      "learning_rate": 8.155349121425504e-06,
      "loss": 1.4639,
      "step": 24417
    },
    {
      "epoch": 0.8744605797983777,
      "grad_norm": 1.68295419216156,
      "learning_rate": 8.150761805292983e-06,
      "loss": 1.3943,
      "step": 24418
    },
    {
      "epoch": 0.874496391927946,
      "grad_norm": 1.8984606266021729,
      "learning_rate": 8.146175724873485e-06,
      "loss": 1.3342,
      "step": 24419
    },
    {
      "epoch": 0.8745322040575143,
      "grad_norm": 1.3739839792251587,
      "learning_rate": 8.141590880228722e-06,
      "loss": 1.4143,
      "step": 24420
    },
    {
      "epoch": 0.8745680161870826,
      "grad_norm": 1.4901645183563232,
      "learning_rate": 8.137007271420349e-06,
      "loss": 1.2584,
      "step": 24421
    },
    {
      "epoch": 0.8746038283166508,
      "grad_norm": 1.6518315076828003,
      "learning_rate": 8.132424898510061e-06,
      "loss": 1.285,
      "step": 24422
    },
    {
      "epoch": 0.8746396404462191,
      "grad_norm": 1.4844005107879639,
      "learning_rate": 8.127843761559506e-06,
      "loss": 1.5776,
      "step": 24423
    },
    {
      "epoch": 0.8746754525757874,
      "grad_norm": 2.039109945297241,
      "learning_rate": 8.123263860630282e-06,
      "loss": 1.5137,
      "step": 24424
    },
    {
      "epoch": 0.8747112647053557,
      "grad_norm": 1.6008793115615845,
      "learning_rate": 8.118685195784037e-06,
      "loss": 1.2386,
      "step": 24425
    },
    {
      "epoch": 0.874747076834924,
      "grad_norm": 1.5261561870574951,
      "learning_rate": 8.114107767082358e-06,
      "loss": 1.5291,
      "step": 24426
    },
    {
      "epoch": 0.8747828889644923,
      "grad_norm": 1.8413728475570679,
      "learning_rate": 8.109531574586859e-06,
      "loss": 1.0062,
      "step": 24427
    },
    {
      "epoch": 0.8748187010940606,
      "grad_norm": 2.2254388332366943,
      "learning_rate": 8.10495661835906e-06,
      "loss": 1.4563,
      "step": 24428
    },
    {
      "epoch": 0.8748545132236288,
      "grad_norm": 2.0652709007263184,
      "learning_rate": 8.100382898460546e-06,
      "loss": 1.7514,
      "step": 24429
    },
    {
      "epoch": 0.8748903253531971,
      "grad_norm": 1.7497395277023315,
      "learning_rate": 8.095810414952832e-06,
      "loss": 1.2252,
      "step": 24430
    },
    {
      "epoch": 0.8749261374827654,
      "grad_norm": 1.9762898683547974,
      "learning_rate": 8.091239167897446e-06,
      "loss": 1.4887,
      "step": 24431
    },
    {
      "epoch": 0.8749619496123336,
      "grad_norm": 1.4887042045593262,
      "learning_rate": 8.086669157355876e-06,
      "loss": 1.6301,
      "step": 24432
    },
    {
      "epoch": 0.874997761741902,
      "grad_norm": 1.460737705230713,
      "learning_rate": 8.082100383389613e-06,
      "loss": 1.2551,
      "step": 24433
    },
    {
      "epoch": 0.8750335738714703,
      "grad_norm": 1.5650736093521118,
      "learning_rate": 8.077532846060143e-06,
      "loss": 1.298,
      "step": 24434
    },
    {
      "epoch": 0.8750693860010386,
      "grad_norm": 1.7930569648742676,
      "learning_rate": 8.072966545428873e-06,
      "loss": 1.467,
      "step": 24435
    },
    {
      "epoch": 0.8751051981306068,
      "grad_norm": 2.4084696769714355,
      "learning_rate": 8.068401481557263e-06,
      "loss": 1.3533,
      "step": 24436
    },
    {
      "epoch": 0.8751410102601751,
      "grad_norm": 1.8376762866973877,
      "learning_rate": 8.063837654506734e-06,
      "loss": 1.3193,
      "step": 24437
    },
    {
      "epoch": 0.8751768223897434,
      "grad_norm": 1.538162350654602,
      "learning_rate": 8.059275064338689e-06,
      "loss": 1.3677,
      "step": 24438
    },
    {
      "epoch": 0.8752126345193116,
      "grad_norm": 1.8487403392791748,
      "learning_rate": 8.054713711114491e-06,
      "loss": 1.4857,
      "step": 24439
    },
    {
      "epoch": 0.87524844664888,
      "grad_norm": 2.232276678085327,
      "learning_rate": 8.050153594895526e-06,
      "loss": 1.6025,
      "step": 24440
    },
    {
      "epoch": 0.8752842587784483,
      "grad_norm": 1.5102007389068604,
      "learning_rate": 8.045594715743144e-06,
      "loss": 0.9663,
      "step": 24441
    },
    {
      "epoch": 0.8753200709080166,
      "grad_norm": 1.7949094772338867,
      "learning_rate": 8.04103707371866e-06,
      "loss": 1.3811,
      "step": 24442
    },
    {
      "epoch": 0.8753558830375848,
      "grad_norm": 1.8622803688049316,
      "learning_rate": 8.036480668883394e-06,
      "loss": 1.3921,
      "step": 24443
    },
    {
      "epoch": 0.8753916951671531,
      "grad_norm": 1.5665391683578491,
      "learning_rate": 8.031925501298666e-06,
      "loss": 1.338,
      "step": 24444
    },
    {
      "epoch": 0.8754275072967214,
      "grad_norm": 1.5506495237350464,
      "learning_rate": 8.027371571025765e-06,
      "loss": 1.5878,
      "step": 24445
    },
    {
      "epoch": 0.8754633194262896,
      "grad_norm": 1.6303693056106567,
      "learning_rate": 8.022818878125926e-06,
      "loss": 1.5915,
      "step": 24446
    },
    {
      "epoch": 0.875499131555858,
      "grad_norm": 1.6259897947311401,
      "learning_rate": 8.018267422660419e-06,
      "loss": 1.731,
      "step": 24447
    },
    {
      "epoch": 0.8755349436854263,
      "grad_norm": 1.2927833795547485,
      "learning_rate": 8.013717204690474e-06,
      "loss": 1.2528,
      "step": 24448
    },
    {
      "epoch": 0.8755707558149946,
      "grad_norm": 1.6778570413589478,
      "learning_rate": 8.00916822427733e-06,
      "loss": 1.3148,
      "step": 24449
    },
    {
      "epoch": 0.8756065679445628,
      "grad_norm": 1.671055555343628,
      "learning_rate": 8.004620481482161e-06,
      "loss": 1.2224,
      "step": 24450
    },
    {
      "epoch": 0.8756423800741311,
      "grad_norm": 2.1273000240325928,
      "learning_rate": 8.00007397636613e-06,
      "loss": 1.5564,
      "step": 24451
    },
    {
      "epoch": 0.8756781922036994,
      "grad_norm": 1.361875295639038,
      "learning_rate": 7.995528708990463e-06,
      "loss": 1.5403,
      "step": 24452
    },
    {
      "epoch": 0.8757140043332676,
      "grad_norm": 1.7198835611343384,
      "learning_rate": 7.990984679416269e-06,
      "loss": 1.28,
      "step": 24453
    },
    {
      "epoch": 0.875749816462836,
      "grad_norm": 1.4493197202682495,
      "learning_rate": 7.986441887704687e-06,
      "loss": 1.2897,
      "step": 24454
    },
    {
      "epoch": 0.8757856285924043,
      "grad_norm": 1.5954997539520264,
      "learning_rate": 7.981900333916848e-06,
      "loss": 1.2537,
      "step": 24455
    },
    {
      "epoch": 0.8758214407219725,
      "grad_norm": 1.679919958114624,
      "learning_rate": 7.977360018113855e-06,
      "loss": 1.438,
      "step": 24456
    },
    {
      "epoch": 0.8758572528515408,
      "grad_norm": 1.6092506647109985,
      "learning_rate": 7.972820940356785e-06,
      "loss": 1.3889,
      "step": 24457
    },
    {
      "epoch": 0.8758930649811091,
      "grad_norm": 2.2504239082336426,
      "learning_rate": 7.968283100706664e-06,
      "loss": 1.4879,
      "step": 24458
    },
    {
      "epoch": 0.8759288771106774,
      "grad_norm": 2.6346192359924316,
      "learning_rate": 7.963746499224611e-06,
      "loss": 1.4057,
      "step": 24459
    },
    {
      "epoch": 0.8759646892402456,
      "grad_norm": 2.1243059635162354,
      "learning_rate": 7.959211135971622e-06,
      "loss": 1.4884,
      "step": 24460
    },
    {
      "epoch": 0.876000501369814,
      "grad_norm": 1.6332237720489502,
      "learning_rate": 7.954677011008749e-06,
      "loss": 1.1818,
      "step": 24461
    },
    {
      "epoch": 0.8760363134993823,
      "grad_norm": 1.5698171854019165,
      "learning_rate": 7.95014412439692e-06,
      "loss": 1.4265,
      "step": 24462
    },
    {
      "epoch": 0.8760721256289505,
      "grad_norm": 1.746675729751587,
      "learning_rate": 7.945612476197207e-06,
      "loss": 1.6762,
      "step": 24463
    },
    {
      "epoch": 0.8761079377585188,
      "grad_norm": 1.5642447471618652,
      "learning_rate": 7.941082066470507e-06,
      "loss": 1.2722,
      "step": 24464
    },
    {
      "epoch": 0.8761437498880871,
      "grad_norm": 2.2350451946258545,
      "learning_rate": 7.936552895277826e-06,
      "loss": 1.7897,
      "step": 24465
    },
    {
      "epoch": 0.8761795620176553,
      "grad_norm": 2.5448694229125977,
      "learning_rate": 7.932024962680062e-06,
      "loss": 1.3325,
      "step": 24466
    },
    {
      "epoch": 0.8762153741472236,
      "grad_norm": 1.8758574724197388,
      "learning_rate": 7.927498268738132e-06,
      "loss": 1.0997,
      "step": 24467
    },
    {
      "epoch": 0.876251186276792,
      "grad_norm": 2.587082624435425,
      "learning_rate": 7.922972813512974e-06,
      "loss": 1.5028,
      "step": 24468
    },
    {
      "epoch": 0.8762869984063603,
      "grad_norm": 1.386791467666626,
      "learning_rate": 7.918448597065408e-06,
      "loss": 1.5839,
      "step": 24469
    },
    {
      "epoch": 0.8763228105359285,
      "grad_norm": 1.7087793350219727,
      "learning_rate": 7.913925619456374e-06,
      "loss": 1.6286,
      "step": 24470
    },
    {
      "epoch": 0.8763586226654968,
      "grad_norm": 1.446853756904602,
      "learning_rate": 7.909403880746669e-06,
      "loss": 1.3785,
      "step": 24471
    },
    {
      "epoch": 0.8763944347950651,
      "grad_norm": 2.251840829849243,
      "learning_rate": 7.904883380997164e-06,
      "loss": 1.453,
      "step": 24472
    },
    {
      "epoch": 0.8764302469246333,
      "grad_norm": 1.5899436473846436,
      "learning_rate": 7.900364120268622e-06,
      "loss": 1.61,
      "step": 24473
    },
    {
      "epoch": 0.8764660590542016,
      "grad_norm": 1.766593098640442,
      "learning_rate": 7.895846098621917e-06,
      "loss": 1.1907,
      "step": 24474
    },
    {
      "epoch": 0.87650187118377,
      "grad_norm": 1.9567400217056274,
      "learning_rate": 7.891329316117801e-06,
      "loss": 1.363,
      "step": 24475
    },
    {
      "epoch": 0.8765376833133383,
      "grad_norm": 1.5550438165664673,
      "learning_rate": 7.886813772817026e-06,
      "loss": 1.1627,
      "step": 24476
    },
    {
      "epoch": 0.8765734954429065,
      "grad_norm": 1.8496496677398682,
      "learning_rate": 7.88229946878034e-06,
      "loss": 1.1868,
      "step": 24477
    },
    {
      "epoch": 0.8766093075724748,
      "grad_norm": 1.919206142425537,
      "learning_rate": 7.877786404068498e-06,
      "loss": 1.7735,
      "step": 24478
    },
    {
      "epoch": 0.8766451197020431,
      "grad_norm": 1.3942233324050903,
      "learning_rate": 7.873274578742229e-06,
      "loss": 1.4318,
      "step": 24479
    },
    {
      "epoch": 0.8766809318316113,
      "grad_norm": 1.6327040195465088,
      "learning_rate": 7.868763992862182e-06,
      "loss": 1.1995,
      "step": 24480
    },
    {
      "epoch": 0.8767167439611796,
      "grad_norm": 1.5736989974975586,
      "learning_rate": 7.864254646489099e-06,
      "loss": 1.1477,
      "step": 24481
    },
    {
      "epoch": 0.876752556090748,
      "grad_norm": 1.382622480392456,
      "learning_rate": 7.859746539683621e-06,
      "loss": 1.565,
      "step": 24482
    },
    {
      "epoch": 0.8767883682203162,
      "grad_norm": 1.465562105178833,
      "learning_rate": 7.855239672506408e-06,
      "loss": 1.4932,
      "step": 24483
    },
    {
      "epoch": 0.8768241803498845,
      "grad_norm": 1.737520456314087,
      "learning_rate": 7.85073404501807e-06,
      "loss": 1.3897,
      "step": 24484
    },
    {
      "epoch": 0.8768599924794528,
      "grad_norm": 1.8517922163009644,
      "learning_rate": 7.846229657279246e-06,
      "loss": 1.6007,
      "step": 24485
    },
    {
      "epoch": 0.8768958046090211,
      "grad_norm": 1.5409523248672485,
      "learning_rate": 7.841726509350545e-06,
      "loss": 1.3853,
      "step": 24486
    },
    {
      "epoch": 0.8769316167385893,
      "grad_norm": 1.7995675802230835,
      "learning_rate": 7.837224601292525e-06,
      "loss": 1.4295,
      "step": 24487
    },
    {
      "epoch": 0.8769674288681576,
      "grad_norm": 2.4543213844299316,
      "learning_rate": 7.832723933165764e-06,
      "loss": 1.4674,
      "step": 24488
    },
    {
      "epoch": 0.877003240997726,
      "grad_norm": 1.7001053094863892,
      "learning_rate": 7.828224505030823e-06,
      "loss": 1.3251,
      "step": 24489
    },
    {
      "epoch": 0.8770390531272942,
      "grad_norm": 1.9298815727233887,
      "learning_rate": 7.823726316948232e-06,
      "loss": 1.528,
      "step": 24490
    },
    {
      "epoch": 0.8770748652568625,
      "grad_norm": 1.8918613195419312,
      "learning_rate": 7.819229368978498e-06,
      "loss": 1.6044,
      "step": 24491
    },
    {
      "epoch": 0.8771106773864308,
      "grad_norm": 1.5881575345993042,
      "learning_rate": 7.814733661182116e-06,
      "loss": 1.6521,
      "step": 24492
    },
    {
      "epoch": 0.877146489515999,
      "grad_norm": 1.3574950695037842,
      "learning_rate": 7.810239193619618e-06,
      "loss": 1.2039,
      "step": 24493
    },
    {
      "epoch": 0.8771823016455673,
      "grad_norm": 1.7643013000488281,
      "learning_rate": 7.805745966351407e-06,
      "loss": 0.9625,
      "step": 24494
    },
    {
      "epoch": 0.8772181137751356,
      "grad_norm": 1.515485167503357,
      "learning_rate": 7.801253979437962e-06,
      "loss": 1.5384,
      "step": 24495
    },
    {
      "epoch": 0.877253925904704,
      "grad_norm": 1.3017157316207886,
      "learning_rate": 7.796763232939719e-06,
      "loss": 1.2687,
      "step": 24496
    },
    {
      "epoch": 0.8772897380342722,
      "grad_norm": 1.8666355609893799,
      "learning_rate": 7.79227372691711e-06,
      "loss": 1.4063,
      "step": 24497
    },
    {
      "epoch": 0.8773255501638405,
      "grad_norm": 1.7395696640014648,
      "learning_rate": 7.787785461430498e-06,
      "loss": 1.4451,
      "step": 24498
    },
    {
      "epoch": 0.8773613622934088,
      "grad_norm": 1.4882618188858032,
      "learning_rate": 7.783298436540288e-06,
      "loss": 1.4995,
      "step": 24499
    },
    {
      "epoch": 0.877397174422977,
      "grad_norm": 1.7713781595230103,
      "learning_rate": 7.778812652306844e-06,
      "loss": 1.5565,
      "step": 24500
    },
    {
      "epoch": 0.8774329865525453,
      "grad_norm": 1.427703619003296,
      "learning_rate": 7.774328108790541e-06,
      "loss": 1.297,
      "step": 24501
    },
    {
      "epoch": 0.8774687986821136,
      "grad_norm": 2.4432625770568848,
      "learning_rate": 7.769844806051674e-06,
      "loss": 1.7429,
      "step": 24502
    },
    {
      "epoch": 0.877504610811682,
      "grad_norm": 1.2416493892669678,
      "learning_rate": 7.765362744150573e-06,
      "loss": 1.3341,
      "step": 24503
    },
    {
      "epoch": 0.8775404229412502,
      "grad_norm": 1.4567826986312866,
      "learning_rate": 7.760881923147567e-06,
      "loss": 1.5861,
      "step": 24504
    },
    {
      "epoch": 0.8775762350708185,
      "grad_norm": 1.5808217525482178,
      "learning_rate": 7.756402343102897e-06,
      "loss": 1.5537,
      "step": 24505
    },
    {
      "epoch": 0.8776120472003868,
      "grad_norm": 1.6409419775009155,
      "learning_rate": 7.751924004076837e-06,
      "loss": 1.1795,
      "step": 24506
    },
    {
      "epoch": 0.877647859329955,
      "grad_norm": 1.580722689628601,
      "learning_rate": 7.747446906129662e-06,
      "loss": 1.4825,
      "step": 24507
    },
    {
      "epoch": 0.8776836714595233,
      "grad_norm": 1.7144967317581177,
      "learning_rate": 7.742971049321601e-06,
      "loss": 1.392,
      "step": 24508
    },
    {
      "epoch": 0.8777194835890916,
      "grad_norm": 1.1489810943603516,
      "learning_rate": 7.738496433712839e-06,
      "loss": 1.3657,
      "step": 24509
    },
    {
      "epoch": 0.87775529571866,
      "grad_norm": 1.4573662281036377,
      "learning_rate": 7.734023059363605e-06,
      "loss": 1.3468,
      "step": 24510
    },
    {
      "epoch": 0.8777911078482282,
      "grad_norm": 1.420721173286438,
      "learning_rate": 7.729550926334094e-06,
      "loss": 1.3038,
      "step": 24511
    },
    {
      "epoch": 0.8778269199777965,
      "grad_norm": 1.3883370161056519,
      "learning_rate": 7.72508003468444e-06,
      "loss": 1.7398,
      "step": 24512
    },
    {
      "epoch": 0.8778627321073648,
      "grad_norm": 1.651623010635376,
      "learning_rate": 7.720610384474802e-06,
      "loss": 1.4708,
      "step": 24513
    },
    {
      "epoch": 0.877898544236933,
      "grad_norm": 1.322274088859558,
      "learning_rate": 7.716141975765322e-06,
      "loss": 1.4895,
      "step": 24514
    },
    {
      "epoch": 0.8779343563665013,
      "grad_norm": 1.9884264469146729,
      "learning_rate": 7.711674808616132e-06,
      "loss": 1.4892,
      "step": 24515
    },
    {
      "epoch": 0.8779701684960696,
      "grad_norm": 1.6073400974273682,
      "learning_rate": 7.70720888308729e-06,
      "loss": 1.5275,
      "step": 24516
    },
    {
      "epoch": 0.878005980625638,
      "grad_norm": 2.0732638835906982,
      "learning_rate": 7.70274419923892e-06,
      "loss": 1.4096,
      "step": 24517
    },
    {
      "epoch": 0.8780417927552062,
      "grad_norm": 1.6035428047180176,
      "learning_rate": 7.69828075713106e-06,
      "loss": 1.4707,
      "step": 24518
    },
    {
      "epoch": 0.8780776048847745,
      "grad_norm": 1.8507529497146606,
      "learning_rate": 7.693818556823784e-06,
      "loss": 1.6353,
      "step": 24519
    },
    {
      "epoch": 0.8781134170143428,
      "grad_norm": 1.507523536682129,
      "learning_rate": 7.6893575983771e-06,
      "loss": 1.3616,
      "step": 24520
    },
    {
      "epoch": 0.878149229143911,
      "grad_norm": 1.8455419540405273,
      "learning_rate": 7.68489788185105e-06,
      "loss": 1.6077,
      "step": 24521
    },
    {
      "epoch": 0.8781850412734793,
      "grad_norm": 1.4134904146194458,
      "learning_rate": 7.680439407305629e-06,
      "loss": 1.493,
      "step": 24522
    },
    {
      "epoch": 0.8782208534030476,
      "grad_norm": 1.6321110725402832,
      "learning_rate": 7.675982174800788e-06,
      "loss": 1.4207,
      "step": 24523
    },
    {
      "epoch": 0.8782566655326159,
      "grad_norm": 1.6989505290985107,
      "learning_rate": 7.671526184396527e-06,
      "loss": 1.5392,
      "step": 24524
    },
    {
      "epoch": 0.8782924776621842,
      "grad_norm": 1.6748425960540771,
      "learning_rate": 7.667071436152784e-06,
      "loss": 1.3084,
      "step": 24525
    },
    {
      "epoch": 0.8783282897917525,
      "grad_norm": 1.3120161294937134,
      "learning_rate": 7.662617930129502e-06,
      "loss": 1.4486,
      "step": 24526
    },
    {
      "epoch": 0.8783641019213208,
      "grad_norm": 1.5598652362823486,
      "learning_rate": 7.658165666386585e-06,
      "loss": 1.503,
      "step": 24527
    },
    {
      "epoch": 0.878399914050889,
      "grad_norm": 1.6046355962753296,
      "learning_rate": 7.653714644983923e-06,
      "loss": 1.3375,
      "step": 24528
    },
    {
      "epoch": 0.8784357261804573,
      "grad_norm": 1.784735918045044,
      "learning_rate": 7.649264865981443e-06,
      "loss": 1.6171,
      "step": 24529
    },
    {
      "epoch": 0.8784715383100256,
      "grad_norm": 1.5161055326461792,
      "learning_rate": 7.644816329438952e-06,
      "loss": 1.4565,
      "step": 24530
    },
    {
      "epoch": 0.8785073504395939,
      "grad_norm": 1.4199373722076416,
      "learning_rate": 7.640369035416339e-06,
      "loss": 1.6551,
      "step": 24531
    },
    {
      "epoch": 0.8785431625691622,
      "grad_norm": 1.6012083292007446,
      "learning_rate": 7.63592298397342e-06,
      "loss": 1.2885,
      "step": 24532
    },
    {
      "epoch": 0.8785789746987305,
      "grad_norm": 1.571999430656433,
      "learning_rate": 7.631478175170026e-06,
      "loss": 1.4313,
      "step": 24533
    },
    {
      "epoch": 0.8786147868282987,
      "grad_norm": 1.8361530303955078,
      "learning_rate": 7.627034609065942e-06,
      "loss": 1.4186,
      "step": 24534
    },
    {
      "epoch": 0.878650598957867,
      "grad_norm": 1.6053346395492554,
      "learning_rate": 7.622592285720942e-06,
      "loss": 1.3284,
      "step": 24535
    },
    {
      "epoch": 0.8786864110874353,
      "grad_norm": 2.1718499660491943,
      "learning_rate": 7.618151205194813e-06,
      "loss": 1.2265,
      "step": 24536
    },
    {
      "epoch": 0.8787222232170036,
      "grad_norm": 1.7951548099517822,
      "learning_rate": 7.613711367547316e-06,
      "loss": 1.5307,
      "step": 24537
    },
    {
      "epoch": 0.8787580353465719,
      "grad_norm": 2.2277722358703613,
      "learning_rate": 7.609272772838138e-06,
      "loss": 1.8082,
      "step": 24538
    },
    {
      "epoch": 0.8787938474761402,
      "grad_norm": 1.3839033842086792,
      "learning_rate": 7.604835421127021e-06,
      "loss": 1.268,
      "step": 24539
    },
    {
      "epoch": 0.8788296596057085,
      "grad_norm": 1.9207768440246582,
      "learning_rate": 7.600399312473683e-06,
      "loss": 1.7225,
      "step": 24540
    },
    {
      "epoch": 0.8788654717352767,
      "grad_norm": 1.5807201862335205,
      "learning_rate": 7.595964446937764e-06,
      "loss": 1.6244,
      "step": 24541
    },
    {
      "epoch": 0.878901283864845,
      "grad_norm": 1.7699261903762817,
      "learning_rate": 7.591530824578952e-06,
      "loss": 1.6804,
      "step": 24542
    },
    {
      "epoch": 0.8789370959944133,
      "grad_norm": 1.3223094940185547,
      "learning_rate": 7.587098445456897e-06,
      "loss": 1.3716,
      "step": 24543
    },
    {
      "epoch": 0.8789729081239815,
      "grad_norm": 1.6015185117721558,
      "learning_rate": 7.582667309631242e-06,
      "loss": 1.3739,
      "step": 24544
    },
    {
      "epoch": 0.8790087202535499,
      "grad_norm": 1.5923868417739868,
      "learning_rate": 7.578237417161571e-06,
      "loss": 1.4783,
      "step": 24545
    },
    {
      "epoch": 0.8790445323831182,
      "grad_norm": 2.426682710647583,
      "learning_rate": 7.573808768107504e-06,
      "loss": 1.4759,
      "step": 24546
    },
    {
      "epoch": 0.8790803445126865,
      "grad_norm": 1.7707291841506958,
      "learning_rate": 7.569381362528638e-06,
      "loss": 1.2952,
      "step": 24547
    },
    {
      "epoch": 0.8791161566422547,
      "grad_norm": 1.452986240386963,
      "learning_rate": 7.5649552004844915e-06,
      "loss": 1.4157,
      "step": 24548
    },
    {
      "epoch": 0.879151968771823,
      "grad_norm": 1.917758822441101,
      "learning_rate": 7.560530282034662e-06,
      "loss": 1.3931,
      "step": 24549
    },
    {
      "epoch": 0.8791877809013913,
      "grad_norm": 1.3515937328338623,
      "learning_rate": 7.556106607238633e-06,
      "loss": 1.4283,
      "step": 24550
    },
    {
      "epoch": 0.8792235930309595,
      "grad_norm": 1.6925004720687866,
      "learning_rate": 7.551684176155971e-06,
      "loss": 1.3824,
      "step": 24551
    },
    {
      "epoch": 0.8792594051605279,
      "grad_norm": 1.9788447618484497,
      "learning_rate": 7.547262988846126e-06,
      "loss": 1.2657,
      "step": 24552
    },
    {
      "epoch": 0.8792952172900962,
      "grad_norm": 1.8088815212249756,
      "learning_rate": 7.542843045368609e-06,
      "loss": 1.6715,
      "step": 24553
    },
    {
      "epoch": 0.8793310294196645,
      "grad_norm": 1.4498738050460815,
      "learning_rate": 7.538424345782902e-06,
      "loss": 1.3569,
      "step": 24554
    },
    {
      "epoch": 0.8793668415492327,
      "grad_norm": 1.5786463022232056,
      "learning_rate": 7.534006890148404e-06,
      "loss": 1.2862,
      "step": 24555
    },
    {
      "epoch": 0.879402653678801,
      "grad_norm": 1.7088158130645752,
      "learning_rate": 7.52959067852459e-06,
      "loss": 1.4563,
      "step": 24556
    },
    {
      "epoch": 0.8794384658083693,
      "grad_norm": 1.9697120189666748,
      "learning_rate": 7.525175710970811e-06,
      "loss": 1.4709,
      "step": 24557
    },
    {
      "epoch": 0.8794742779379375,
      "grad_norm": 1.5277745723724365,
      "learning_rate": 7.520761987546554e-06,
      "loss": 1.2802,
      "step": 24558
    },
    {
      "epoch": 0.8795100900675059,
      "grad_norm": 2.2481420040130615,
      "learning_rate": 7.516349508311138e-06,
      "loss": 1.3251,
      "step": 24559
    },
    {
      "epoch": 0.8795459021970742,
      "grad_norm": 1.6836209297180176,
      "learning_rate": 7.51193827332396e-06,
      "loss": 1.5707,
      "step": 24560
    },
    {
      "epoch": 0.8795817143266424,
      "grad_norm": 1.6443015336990356,
      "learning_rate": 7.507528282644316e-06,
      "loss": 1.592,
      "step": 24561
    },
    {
      "epoch": 0.8796175264562107,
      "grad_norm": 1.5825704336166382,
      "learning_rate": 7.503119536331604e-06,
      "loss": 1.4901,
      "step": 24562
    },
    {
      "epoch": 0.879653338585779,
      "grad_norm": 1.8732868432998657,
      "learning_rate": 7.49871203444511e-06,
      "loss": 1.2958,
      "step": 24563
    },
    {
      "epoch": 0.8796891507153473,
      "grad_norm": 1.4926637411117554,
      "learning_rate": 7.494305777044086e-06,
      "loss": 1.3454,
      "step": 24564
    },
    {
      "epoch": 0.8797249628449155,
      "grad_norm": 1.602117896080017,
      "learning_rate": 7.489900764187896e-06,
      "loss": 1.6786,
      "step": 24565
    },
    {
      "epoch": 0.8797607749744839,
      "grad_norm": 2.2812118530273438,
      "learning_rate": 7.485496995935748e-06,
      "loss": 1.5049,
      "step": 24566
    },
    {
      "epoch": 0.8797965871040522,
      "grad_norm": 1.250343918800354,
      "learning_rate": 7.481094472346905e-06,
      "loss": 1.3796,
      "step": 24567
    },
    {
      "epoch": 0.8798323992336204,
      "grad_norm": 1.6296786069869995,
      "learning_rate": 7.476693193480577e-06,
      "loss": 1.5116,
      "step": 24568
    },
    {
      "epoch": 0.8798682113631887,
      "grad_norm": 1.5637654066085815,
      "learning_rate": 7.472293159396027e-06,
      "loss": 1.5897,
      "step": 24569
    },
    {
      "epoch": 0.879904023492757,
      "grad_norm": 1.8617569208145142,
      "learning_rate": 7.4678943701523954e-06,
      "loss": 1.4735,
      "step": 24570
    },
    {
      "epoch": 0.8799398356223253,
      "grad_norm": 2.2335526943206787,
      "learning_rate": 7.4634968258089135e-06,
      "loss": 1.5358,
      "step": 24571
    },
    {
      "epoch": 0.8799756477518935,
      "grad_norm": 1.9554426670074463,
      "learning_rate": 7.4591005264246895e-06,
      "loss": 1.4661,
      "step": 24572
    },
    {
      "epoch": 0.8800114598814619,
      "grad_norm": 2.1411080360412598,
      "learning_rate": 7.454705472058909e-06,
      "loss": 1.3836,
      "step": 24573
    },
    {
      "epoch": 0.8800472720110302,
      "grad_norm": 2.0509655475616455,
      "learning_rate": 7.450311662770704e-06,
      "loss": 1.215,
      "step": 24574
    },
    {
      "epoch": 0.8800830841405984,
      "grad_norm": 1.5214743614196777,
      "learning_rate": 7.445919098619159e-06,
      "loss": 1.2725,
      "step": 24575
    },
    {
      "epoch": 0.8801188962701667,
      "grad_norm": 1.6003098487854004,
      "learning_rate": 7.441527779663382e-06,
      "loss": 1.5988,
      "step": 24576
    },
    {
      "epoch": 0.880154708399735,
      "grad_norm": 1.7950252294540405,
      "learning_rate": 7.43713770596246e-06,
      "loss": 1.3825,
      "step": 24577
    },
    {
      "epoch": 0.8801905205293032,
      "grad_norm": 1.5984803438186646,
      "learning_rate": 7.4327488775754794e-06,
      "loss": 1.3248,
      "step": 24578
    },
    {
      "epoch": 0.8802263326588715,
      "grad_norm": 1.6045230627059937,
      "learning_rate": 7.428361294561415e-06,
      "loss": 1.2454,
      "step": 24579
    },
    {
      "epoch": 0.8802621447884399,
      "grad_norm": 2.036458969116211,
      "learning_rate": 7.423974956979374e-06,
      "loss": 1.4281,
      "step": 24580
    },
    {
      "epoch": 0.8802979569180082,
      "grad_norm": 1.4843889474868774,
      "learning_rate": 7.419589864888332e-06,
      "loss": 1.4315,
      "step": 24581
    },
    {
      "epoch": 0.8803337690475764,
      "grad_norm": 1.5990513563156128,
      "learning_rate": 7.415206018347287e-06,
      "loss": 1.1032,
      "step": 24582
    },
    {
      "epoch": 0.8803695811771447,
      "grad_norm": 1.3269641399383545,
      "learning_rate": 7.410823417415203e-06,
      "loss": 1.2407,
      "step": 24583
    },
    {
      "epoch": 0.880405393306713,
      "grad_norm": 1.8271068334579468,
      "learning_rate": 7.406442062151064e-06,
      "loss": 1.3426,
      "step": 24584
    },
    {
      "epoch": 0.8804412054362812,
      "grad_norm": 2.1693825721740723,
      "learning_rate": 7.402061952613826e-06,
      "loss": 1.126,
      "step": 24585
    },
    {
      "epoch": 0.8804770175658495,
      "grad_norm": 1.887425422668457,
      "learning_rate": 7.397683088862395e-06,
      "loss": 1.5462,
      "step": 24586
    },
    {
      "epoch": 0.8805128296954179,
      "grad_norm": 1.5298993587493896,
      "learning_rate": 7.393305470955681e-06,
      "loss": 1.1619,
      "step": 24587
    },
    {
      "epoch": 0.8805486418249862,
      "grad_norm": 2.064223051071167,
      "learning_rate": 7.388929098952579e-06,
      "loss": 1.2954,
      "step": 24588
    },
    {
      "epoch": 0.8805844539545544,
      "grad_norm": 1.1978247165679932,
      "learning_rate": 7.384553972912011e-06,
      "loss": 1.537,
      "step": 24589
    },
    {
      "epoch": 0.8806202660841227,
      "grad_norm": 1.5021591186523438,
      "learning_rate": 7.380180092892775e-06,
      "loss": 1.4642,
      "step": 24590
    },
    {
      "epoch": 0.880656078213691,
      "grad_norm": 2.6436407566070557,
      "learning_rate": 7.375807458953743e-06,
      "loss": 1.812,
      "step": 24591
    },
    {
      "epoch": 0.8806918903432592,
      "grad_norm": 1.5209324359893799,
      "learning_rate": 7.37143607115377e-06,
      "loss": 1.3444,
      "step": 24592
    },
    {
      "epoch": 0.8807277024728275,
      "grad_norm": 1.360662579536438,
      "learning_rate": 7.36706592955162e-06,
      "loss": 1.3151,
      "step": 24593
    },
    {
      "epoch": 0.8807635146023959,
      "grad_norm": 1.9590721130371094,
      "learning_rate": 7.362697034206112e-06,
      "loss": 1.177,
      "step": 24594
    },
    {
      "epoch": 0.8807993267319641,
      "grad_norm": 1.7141327857971191,
      "learning_rate": 7.358329385176033e-06,
      "loss": 1.4844,
      "step": 24595
    },
    {
      "epoch": 0.8808351388615324,
      "grad_norm": 1.7014778852462769,
      "learning_rate": 7.353962982520135e-06,
      "loss": 1.5032,
      "step": 24596
    },
    {
      "epoch": 0.8808709509911007,
      "grad_norm": 1.8111387491226196,
      "learning_rate": 7.34959782629715e-06,
      "loss": 1.264,
      "step": 24597
    },
    {
      "epoch": 0.880906763120669,
      "grad_norm": 1.8225723505020142,
      "learning_rate": 7.345233916565808e-06,
      "loss": 1.4492,
      "step": 24598
    },
    {
      "epoch": 0.8809425752502372,
      "grad_norm": 1.3599964380264282,
      "learning_rate": 7.340871253384851e-06,
      "loss": 1.4305,
      "step": 24599
    },
    {
      "epoch": 0.8809783873798055,
      "grad_norm": 1.8249688148498535,
      "learning_rate": 7.336509836812933e-06,
      "loss": 1.1971,
      "step": 24600
    },
    {
      "epoch": 0.8810141995093739,
      "grad_norm": 1.7372807264328003,
      "learning_rate": 7.3321496669087495e-06,
      "loss": 1.4024,
      "step": 24601
    },
    {
      "epoch": 0.8810500116389421,
      "grad_norm": 1.6618596315383911,
      "learning_rate": 7.327790743730956e-06,
      "loss": 1.3195,
      "step": 24602
    },
    {
      "epoch": 0.8810858237685104,
      "grad_norm": 1.5446751117706299,
      "learning_rate": 7.323433067338214e-06,
      "loss": 1.426,
      "step": 24603
    },
    {
      "epoch": 0.8811216358980787,
      "grad_norm": 1.6365025043487549,
      "learning_rate": 7.319076637789124e-06,
      "loss": 1.1681,
      "step": 24604
    },
    {
      "epoch": 0.881157448027647,
      "grad_norm": 1.3218605518341064,
      "learning_rate": 7.314721455142304e-06,
      "loss": 1.3863,
      "step": 24605
    },
    {
      "epoch": 0.8811932601572152,
      "grad_norm": 1.4129111766815186,
      "learning_rate": 7.310367519456352e-06,
      "loss": 1.5447,
      "step": 24606
    },
    {
      "epoch": 0.8812290722867835,
      "grad_norm": 2.340679407119751,
      "learning_rate": 7.306014830789865e-06,
      "loss": 1.4607,
      "step": 24607
    },
    {
      "epoch": 0.8812648844163519,
      "grad_norm": 2.0600128173828125,
      "learning_rate": 7.3016633892013634e-06,
      "loss": 1.5905,
      "step": 24608
    },
    {
      "epoch": 0.8813006965459201,
      "grad_norm": 1.7693760395050049,
      "learning_rate": 7.2973131947494e-06,
      "loss": 1.3652,
      "step": 24609
    },
    {
      "epoch": 0.8813365086754884,
      "grad_norm": 1.6240615844726562,
      "learning_rate": 7.292964247492539e-06,
      "loss": 1.3207,
      "step": 24610
    },
    {
      "epoch": 0.8813723208050567,
      "grad_norm": 1.4567763805389404,
      "learning_rate": 7.288616547489235e-06,
      "loss": 1.3986,
      "step": 24611
    },
    {
      "epoch": 0.8814081329346249,
      "grad_norm": 1.3678109645843506,
      "learning_rate": 7.284270094798018e-06,
      "loss": 1.2433,
      "step": 24612
    },
    {
      "epoch": 0.8814439450641932,
      "grad_norm": 1.4017517566680908,
      "learning_rate": 7.279924889477341e-06,
      "loss": 1.4941,
      "step": 24613
    },
    {
      "epoch": 0.8814797571937615,
      "grad_norm": 2.248542308807373,
      "learning_rate": 7.27558093158569e-06,
      "loss": 1.5972,
      "step": 24614
    },
    {
      "epoch": 0.8815155693233299,
      "grad_norm": 1.4761277437210083,
      "learning_rate": 7.2712382211814865e-06,
      "loss": 1.4937,
      "step": 24615
    },
    {
      "epoch": 0.8815513814528981,
      "grad_norm": 1.3250051736831665,
      "learning_rate": 7.266896758323149e-06,
      "loss": 1.1396,
      "step": 24616
    },
    {
      "epoch": 0.8815871935824664,
      "grad_norm": 2.3315348625183105,
      "learning_rate": 7.2625565430691214e-06,
      "loss": 1.128,
      "step": 24617
    },
    {
      "epoch": 0.8816230057120347,
      "grad_norm": 1.9104201793670654,
      "learning_rate": 7.258217575477755e-06,
      "loss": 1.364,
      "step": 24618
    },
    {
      "epoch": 0.8816588178416029,
      "grad_norm": 1.6324379444122314,
      "learning_rate": 7.253879855607437e-06,
      "loss": 1.4386,
      "step": 24619
    },
    {
      "epoch": 0.8816946299711712,
      "grad_norm": 1.622747540473938,
      "learning_rate": 7.249543383516544e-06,
      "loss": 1.3758,
      "step": 24620
    },
    {
      "epoch": 0.8817304421007395,
      "grad_norm": 1.622939109802246,
      "learning_rate": 7.245208159263417e-06,
      "loss": 1.5202,
      "step": 24621
    },
    {
      "epoch": 0.8817662542303079,
      "grad_norm": 1.5521178245544434,
      "learning_rate": 7.240874182906343e-06,
      "loss": 1.3341,
      "step": 24622
    },
    {
      "epoch": 0.8818020663598761,
      "grad_norm": 1.466652512550354,
      "learning_rate": 7.236541454503664e-06,
      "loss": 1.5546,
      "step": 24623
    },
    {
      "epoch": 0.8818378784894444,
      "grad_norm": 1.4532678127288818,
      "learning_rate": 7.232209974113668e-06,
      "loss": 1.3809,
      "step": 24624
    },
    {
      "epoch": 0.8818736906190127,
      "grad_norm": 2.0897300243377686,
      "learning_rate": 7.2278797417946405e-06,
      "loss": 1.6313,
      "step": 24625
    },
    {
      "epoch": 0.8819095027485809,
      "grad_norm": 1.290476679801941,
      "learning_rate": 7.2235507576048024e-06,
      "loss": 0.8929,
      "step": 24626
    },
    {
      "epoch": 0.8819453148781492,
      "grad_norm": 1.5610119104385376,
      "learning_rate": 7.219223021602417e-06,
      "loss": 1.1235,
      "step": 24627
    },
    {
      "epoch": 0.8819811270077175,
      "grad_norm": 1.5145996809005737,
      "learning_rate": 7.214896533845716e-06,
      "loss": 1.5866,
      "step": 24628
    },
    {
      "epoch": 0.8820169391372858,
      "grad_norm": 1.9009678363800049,
      "learning_rate": 7.210571294392898e-06,
      "loss": 1.5812,
      "step": 24629
    },
    {
      "epoch": 0.8820527512668541,
      "grad_norm": 1.3867923021316528,
      "learning_rate": 7.206247303302138e-06,
      "loss": 1.3757,
      "step": 24630
    },
    {
      "epoch": 0.8820885633964224,
      "grad_norm": 2.2626278400421143,
      "learning_rate": 7.201924560631634e-06,
      "loss": 1.6227,
      "step": 24631
    },
    {
      "epoch": 0.8821243755259907,
      "grad_norm": 1.3764382600784302,
      "learning_rate": 7.197603066439551e-06,
      "loss": 1.1527,
      "step": 24632
    },
    {
      "epoch": 0.8821601876555589,
      "grad_norm": 1.6069098711013794,
      "learning_rate": 7.193282820783987e-06,
      "loss": 1.3079,
      "step": 24633
    },
    {
      "epoch": 0.8821959997851272,
      "grad_norm": 1.8032952547073364,
      "learning_rate": 7.188963823723105e-06,
      "loss": 1.4146,
      "step": 24634
    },
    {
      "epoch": 0.8822318119146955,
      "grad_norm": 1.4079662561416626,
      "learning_rate": 7.184646075315005e-06,
      "loss": 1.1088,
      "step": 24635
    },
    {
      "epoch": 0.8822676240442638,
      "grad_norm": 1.5412856340408325,
      "learning_rate": 7.18032957561775e-06,
      "loss": 1.5314,
      "step": 24636
    },
    {
      "epoch": 0.8823034361738321,
      "grad_norm": 1.5019010305404663,
      "learning_rate": 7.176014324689428e-06,
      "loss": 1.5056,
      "step": 24637
    },
    {
      "epoch": 0.8823392483034004,
      "grad_norm": 1.4662113189697266,
      "learning_rate": 7.171700322588115e-06,
      "loss": 1.3196,
      "step": 24638
    },
    {
      "epoch": 0.8823750604329686,
      "grad_norm": 1.510379433631897,
      "learning_rate": 7.167387569371842e-06,
      "loss": 1.2243,
      "step": 24639
    },
    {
      "epoch": 0.8824108725625369,
      "grad_norm": 1.4555983543395996,
      "learning_rate": 7.1630760650986065e-06,
      "loss": 1.353,
      "step": 24640
    },
    {
      "epoch": 0.8824466846921052,
      "grad_norm": 1.6988203525543213,
      "learning_rate": 7.158765809826429e-06,
      "loss": 1.5011,
      "step": 24641
    },
    {
      "epoch": 0.8824824968216735,
      "grad_norm": 1.522375464439392,
      "learning_rate": 7.154456803613297e-06,
      "loss": 1.2669,
      "step": 24642
    },
    {
      "epoch": 0.8825183089512418,
      "grad_norm": 1.9415677785873413,
      "learning_rate": 7.150149046517218e-06,
      "loss": 1.7027,
      "step": 24643
    },
    {
      "epoch": 0.8825541210808101,
      "grad_norm": 1.302003264427185,
      "learning_rate": 7.145842538596104e-06,
      "loss": 1.3639,
      "step": 24644
    },
    {
      "epoch": 0.8825899332103784,
      "grad_norm": 1.48600435256958,
      "learning_rate": 7.141537279907873e-06,
      "loss": 1.3566,
      "step": 24645
    },
    {
      "epoch": 0.8826257453399466,
      "grad_norm": 1.4763420820236206,
      "learning_rate": 7.1372332705105125e-06,
      "loss": 1.1619,
      "step": 24646
    },
    {
      "epoch": 0.8826615574695149,
      "grad_norm": 1.6381981372833252,
      "learning_rate": 7.132930510461889e-06,
      "loss": 1.5885,
      "step": 24647
    },
    {
      "epoch": 0.8826973695990832,
      "grad_norm": 1.9131838083267212,
      "learning_rate": 7.128628999819886e-06,
      "loss": 1.5576,
      "step": 24648
    },
    {
      "epoch": 0.8827331817286515,
      "grad_norm": 2.0828161239624023,
      "learning_rate": 7.1243287386423826e-06,
      "loss": 1.5515,
      "step": 24649
    },
    {
      "epoch": 0.8827689938582198,
      "grad_norm": 1.6835256814956665,
      "learning_rate": 7.120029726987254e-06,
      "loss": 1.5039,
      "step": 24650
    },
    {
      "epoch": 0.8828048059877881,
      "grad_norm": 1.6714760065078735,
      "learning_rate": 7.11573196491232e-06,
      "loss": 1.5961,
      "step": 24651
    },
    {
      "epoch": 0.8828406181173564,
      "grad_norm": 1.5934548377990723,
      "learning_rate": 7.111435452475368e-06,
      "loss": 1.5434,
      "step": 24652
    },
    {
      "epoch": 0.8828764302469246,
      "grad_norm": 1.663367509841919,
      "learning_rate": 7.1071401897342625e-06,
      "loss": 1.3505,
      "step": 24653
    },
    {
      "epoch": 0.8829122423764929,
      "grad_norm": 1.8386807441711426,
      "learning_rate": 7.1028461767467466e-06,
      "loss": 1.7194,
      "step": 24654
    },
    {
      "epoch": 0.8829480545060612,
      "grad_norm": 1.7250741720199585,
      "learning_rate": 7.0985534135706296e-06,
      "loss": 1.4797,
      "step": 24655
    },
    {
      "epoch": 0.8829838666356294,
      "grad_norm": 1.8594763278961182,
      "learning_rate": 7.0942619002635995e-06,
      "loss": 1.5841,
      "step": 24656
    },
    {
      "epoch": 0.8830196787651978,
      "grad_norm": 1.341265320777893,
      "learning_rate": 7.089971636883475e-06,
      "loss": 1.3489,
      "step": 24657
    },
    {
      "epoch": 0.8830554908947661,
      "grad_norm": 1.5680384635925293,
      "learning_rate": 7.085682623487921e-06,
      "loss": 1.3567,
      "step": 24658
    },
    {
      "epoch": 0.8830913030243344,
      "grad_norm": 1.4889854192733765,
      "learning_rate": 7.0813948601346715e-06,
      "loss": 1.5167,
      "step": 24659
    },
    {
      "epoch": 0.8831271151539026,
      "grad_norm": 2.0524439811706543,
      "learning_rate": 7.077108346881378e-06,
      "loss": 1.2639,
      "step": 24660
    },
    {
      "epoch": 0.8831629272834709,
      "grad_norm": 1.2576946020126343,
      "learning_rate": 7.07282308378574e-06,
      "loss": 1.4163,
      "step": 24661
    },
    {
      "epoch": 0.8831987394130392,
      "grad_norm": 1.7365771532058716,
      "learning_rate": 7.068539070905411e-06,
      "loss": 1.8081,
      "step": 24662
    },
    {
      "epoch": 0.8832345515426074,
      "grad_norm": 1.9017252922058105,
      "learning_rate": 7.064256308297978e-06,
      "loss": 1.7369,
      "step": 24663
    },
    {
      "epoch": 0.8832703636721758,
      "grad_norm": 1.9138576984405518,
      "learning_rate": 7.05997479602114e-06,
      "loss": 1.2112,
      "step": 24664
    },
    {
      "epoch": 0.8833061758017441,
      "grad_norm": 1.3498430252075195,
      "learning_rate": 7.0556945341324284e-06,
      "loss": 1.4391,
      "step": 24665
    },
    {
      "epoch": 0.8833419879313124,
      "grad_norm": 1.7585177421569824,
      "learning_rate": 7.051415522689487e-06,
      "loss": 1.632,
      "step": 24666
    },
    {
      "epoch": 0.8833778000608806,
      "grad_norm": 1.7154725790023804,
      "learning_rate": 7.047137761749811e-06,
      "loss": 1.4024,
      "step": 24667
    },
    {
      "epoch": 0.8834136121904489,
      "grad_norm": 1.4533997774124146,
      "learning_rate": 7.042861251371036e-06,
      "loss": 1.1525,
      "step": 24668
    },
    {
      "epoch": 0.8834494243200172,
      "grad_norm": 1.5111123323440552,
      "learning_rate": 7.038585991610647e-06,
      "loss": 1.3345,
      "step": 24669
    },
    {
      "epoch": 0.8834852364495854,
      "grad_norm": 2.691683769226074,
      "learning_rate": 7.034311982526165e-06,
      "loss": 1.4586,
      "step": 24670
    },
    {
      "epoch": 0.8835210485791538,
      "grad_norm": 2.8711907863616943,
      "learning_rate": 7.0300392241751e-06,
      "loss": 1.7666,
      "step": 24671
    },
    {
      "epoch": 0.8835568607087221,
      "grad_norm": 1.3507050275802612,
      "learning_rate": 7.025767716614928e-06,
      "loss": 1.2691,
      "step": 24672
    },
    {
      "epoch": 0.8835926728382903,
      "grad_norm": 1.9385850429534912,
      "learning_rate": 7.021497459903137e-06,
      "loss": 1.6612,
      "step": 24673
    },
    {
      "epoch": 0.8836284849678586,
      "grad_norm": 2.0036303997039795,
      "learning_rate": 7.017228454097136e-06,
      "loss": 1.4667,
      "step": 24674
    },
    {
      "epoch": 0.8836642970974269,
      "grad_norm": 1.8919461965560913,
      "learning_rate": 7.012960699254423e-06,
      "loss": 1.2746,
      "step": 24675
    },
    {
      "epoch": 0.8837001092269952,
      "grad_norm": 1.8504756689071655,
      "learning_rate": 7.0086941954323634e-06,
      "loss": 1.3695,
      "step": 24676
    },
    {
      "epoch": 0.8837359213565634,
      "grad_norm": 1.6944911479949951,
      "learning_rate": 7.004428942688379e-06,
      "loss": 1.3329,
      "step": 24677
    },
    {
      "epoch": 0.8837717334861318,
      "grad_norm": 1.8815547227859497,
      "learning_rate": 7.000164941079846e-06,
      "loss": 1.3605,
      "step": 24678
    },
    {
      "epoch": 0.8838075456157001,
      "grad_norm": 1.4075126647949219,
      "learning_rate": 6.995902190664116e-06,
      "loss": 1.439,
      "step": 24679
    },
    {
      "epoch": 0.8838433577452683,
      "grad_norm": 1.4690585136413574,
      "learning_rate": 6.99164069149858e-06,
      "loss": 1.342,
      "step": 24680
    },
    {
      "epoch": 0.8838791698748366,
      "grad_norm": 1.5376348495483398,
      "learning_rate": 6.9873804436405345e-06,
      "loss": 1.4158,
      "step": 24681
    },
    {
      "epoch": 0.8839149820044049,
      "grad_norm": 1.4911822080612183,
      "learning_rate": 6.9831214471473e-06,
      "loss": 1.1987,
      "step": 24682
    },
    {
      "epoch": 0.8839507941339732,
      "grad_norm": 1.6924875974655151,
      "learning_rate": 6.978863702076188e-06,
      "loss": 1.4808,
      "step": 24683
    },
    {
      "epoch": 0.8839866062635414,
      "grad_norm": 1.6052711009979248,
      "learning_rate": 6.974607208484496e-06,
      "loss": 1.5073,
      "step": 24684
    },
    {
      "epoch": 0.8840224183931098,
      "grad_norm": 1.507482886314392,
      "learning_rate": 6.970351966429445e-06,
      "loss": 1.3385,
      "step": 24685
    },
    {
      "epoch": 0.8840582305226781,
      "grad_norm": 1.743933081626892,
      "learning_rate": 6.966097975968311e-06,
      "loss": 1.159,
      "step": 24686
    },
    {
      "epoch": 0.8840940426522463,
      "grad_norm": 1.7662138938903809,
      "learning_rate": 6.961845237158337e-06,
      "loss": 1.4932,
      "step": 24687
    },
    {
      "epoch": 0.8841298547818146,
      "grad_norm": 1.4913418292999268,
      "learning_rate": 6.957593750056712e-06,
      "loss": 1.1511,
      "step": 24688
    },
    {
      "epoch": 0.8841656669113829,
      "grad_norm": 1.4417868852615356,
      "learning_rate": 6.953343514720656e-06,
      "loss": 1.5253,
      "step": 24689
    },
    {
      "epoch": 0.8842014790409511,
      "grad_norm": 1.7345951795578003,
      "learning_rate": 6.949094531207334e-06,
      "loss": 1.3248,
      "step": 24690
    },
    {
      "epoch": 0.8842372911705194,
      "grad_norm": 1.930124044418335,
      "learning_rate": 6.944846799573934e-06,
      "loss": 1.337,
      "step": 24691
    },
    {
      "epoch": 0.8842731033000878,
      "grad_norm": 2.2706034183502197,
      "learning_rate": 6.940600319877566e-06,
      "loss": 1.4853,
      "step": 24692
    },
    {
      "epoch": 0.8843089154296561,
      "grad_norm": 1.682254672050476,
      "learning_rate": 6.936355092175384e-06,
      "loss": 1.4649,
      "step": 24693
    },
    {
      "epoch": 0.8843447275592243,
      "grad_norm": 1.652011752128601,
      "learning_rate": 6.932111116524509e-06,
      "loss": 1.3665,
      "step": 24694
    },
    {
      "epoch": 0.8843805396887926,
      "grad_norm": 1.7562085390090942,
      "learning_rate": 6.92786839298204e-06,
      "loss": 1.4918,
      "step": 24695
    },
    {
      "epoch": 0.8844163518183609,
      "grad_norm": 1.5120139122009277,
      "learning_rate": 6.923626921605031e-06,
      "loss": 1.2655,
      "step": 24696
    },
    {
      "epoch": 0.8844521639479291,
      "grad_norm": 1.5884032249450684,
      "learning_rate": 6.9193867024505695e-06,
      "loss": 1.5416,
      "step": 24697
    },
    {
      "epoch": 0.8844879760774974,
      "grad_norm": 1.2141194343566895,
      "learning_rate": 6.9151477355757e-06,
      "loss": 1.3824,
      "step": 24698
    },
    {
      "epoch": 0.8845237882070658,
      "grad_norm": 1.8356338739395142,
      "learning_rate": 6.910910021037431e-06,
      "loss": 1.3849,
      "step": 24699
    },
    {
      "epoch": 0.884559600336634,
      "grad_norm": 1.3843994140625,
      "learning_rate": 6.906673558892807e-06,
      "loss": 1.4973,
      "step": 24700
    },
    {
      "epoch": 0.8845954124662023,
      "grad_norm": 1.4231609106063843,
      "learning_rate": 6.902438349198792e-06,
      "loss": 1.7378,
      "step": 24701
    },
    {
      "epoch": 0.8846312245957706,
      "grad_norm": 1.7714993953704834,
      "learning_rate": 6.898204392012408e-06,
      "loss": 1.7244,
      "step": 24702
    },
    {
      "epoch": 0.8846670367253389,
      "grad_norm": 1.7537744045257568,
      "learning_rate": 6.893971687390566e-06,
      "loss": 1.4904,
      "step": 24703
    },
    {
      "epoch": 0.8847028488549071,
      "grad_norm": 1.8601810932159424,
      "learning_rate": 6.889740235390241e-06,
      "loss": 1.3624,
      "step": 24704
    },
    {
      "epoch": 0.8847386609844754,
      "grad_norm": 1.8371527194976807,
      "learning_rate": 6.885510036068377e-06,
      "loss": 1.6462,
      "step": 24705
    },
    {
      "epoch": 0.8847744731140438,
      "grad_norm": 2.1824593544006348,
      "learning_rate": 6.881281089481839e-06,
      "loss": 1.3354,
      "step": 24706
    },
    {
      "epoch": 0.884810285243612,
      "grad_norm": 1.5834816694259644,
      "learning_rate": 6.877053395687561e-06,
      "loss": 1.5739,
      "step": 24707
    },
    {
      "epoch": 0.8848460973731803,
      "grad_norm": 1.9149326086044312,
      "learning_rate": 6.872826954742406e-06,
      "loss": 1.5595,
      "step": 24708
    },
    {
      "epoch": 0.8848819095027486,
      "grad_norm": 2.170591354370117,
      "learning_rate": 6.868601766703253e-06,
      "loss": 1.2154,
      "step": 24709
    },
    {
      "epoch": 0.8849177216323169,
      "grad_norm": 1.5158920288085938,
      "learning_rate": 6.8643778316269226e-06,
      "loss": 1.2864,
      "step": 24710
    },
    {
      "epoch": 0.8849535337618851,
      "grad_norm": 1.6316123008728027,
      "learning_rate": 6.860155149570246e-06,
      "loss": 1.5006,
      "step": 24711
    },
    {
      "epoch": 0.8849893458914534,
      "grad_norm": 1.3980909585952759,
      "learning_rate": 6.855933720590047e-06,
      "loss": 1.4343,
      "step": 24712
    },
    {
      "epoch": 0.8850251580210218,
      "grad_norm": 1.5064276456832886,
      "learning_rate": 6.8517135447431215e-06,
      "loss": 1.3443,
      "step": 24713
    },
    {
      "epoch": 0.88506097015059,
      "grad_norm": 1.580866813659668,
      "learning_rate": 6.847494622086226e-06,
      "loss": 1.3168,
      "step": 24714
    },
    {
      "epoch": 0.8850967822801583,
      "grad_norm": 1.6514077186584473,
      "learning_rate": 6.843276952676125e-06,
      "loss": 1.3983,
      "step": 24715
    },
    {
      "epoch": 0.8851325944097266,
      "grad_norm": 1.7922331094741821,
      "learning_rate": 6.839060536569597e-06,
      "loss": 1.4731,
      "step": 24716
    },
    {
      "epoch": 0.8851684065392948,
      "grad_norm": 1.8965702056884766,
      "learning_rate": 6.834845373823317e-06,
      "loss": 1.2081,
      "step": 24717
    },
    {
      "epoch": 0.8852042186688631,
      "grad_norm": 1.391337275505066,
      "learning_rate": 6.830631464494019e-06,
      "loss": 1.4714,
      "step": 24718
    },
    {
      "epoch": 0.8852400307984314,
      "grad_norm": 1.5723334550857544,
      "learning_rate": 6.826418808638391e-06,
      "loss": 1.647,
      "step": 24719
    },
    {
      "epoch": 0.8852758429279998,
      "grad_norm": 1.616264820098877,
      "learning_rate": 6.82220740631313e-06,
      "loss": 1.3395,
      "step": 24720
    },
    {
      "epoch": 0.885311655057568,
      "grad_norm": 1.7802958488464355,
      "learning_rate": 6.8179972575748706e-06,
      "loss": 1.3974,
      "step": 24721
    },
    {
      "epoch": 0.8853474671871363,
      "grad_norm": 1.9190346002578735,
      "learning_rate": 6.813788362480256e-06,
      "loss": 1.4053,
      "step": 24722
    },
    {
      "epoch": 0.8853832793167046,
      "grad_norm": 1.4406262636184692,
      "learning_rate": 6.809580721085929e-06,
      "loss": 1.5542,
      "step": 24723
    },
    {
      "epoch": 0.8854190914462728,
      "grad_norm": 1.6550371646881104,
      "learning_rate": 6.805374333448478e-06,
      "loss": 1.1518,
      "step": 24724
    },
    {
      "epoch": 0.8854549035758411,
      "grad_norm": 1.6076737642288208,
      "learning_rate": 6.801169199624502e-06,
      "loss": 1.3962,
      "step": 24725
    },
    {
      "epoch": 0.8854907157054094,
      "grad_norm": 1.676985263824463,
      "learning_rate": 6.796965319670568e-06,
      "loss": 1.6268,
      "step": 24726
    },
    {
      "epoch": 0.8855265278349778,
      "grad_norm": 2.0720744132995605,
      "learning_rate": 6.792762693643262e-06,
      "loss": 1.3802,
      "step": 24727
    },
    {
      "epoch": 0.885562339964546,
      "grad_norm": 1.511457085609436,
      "learning_rate": 6.7885613215990965e-06,
      "loss": 1.3425,
      "step": 24728
    },
    {
      "epoch": 0.8855981520941143,
      "grad_norm": 2.5622329711914062,
      "learning_rate": 6.7843612035945915e-06,
      "loss": 1.6844,
      "step": 24729
    },
    {
      "epoch": 0.8856339642236826,
      "grad_norm": 1.8360401391983032,
      "learning_rate": 6.78016233968628e-06,
      "loss": 1.3572,
      "step": 24730
    },
    {
      "epoch": 0.8856697763532508,
      "grad_norm": 1.9595917463302612,
      "learning_rate": 6.775964729930651e-06,
      "loss": 1.4014,
      "step": 24731
    },
    {
      "epoch": 0.8857055884828191,
      "grad_norm": 1.515363097190857,
      "learning_rate": 6.771768374384168e-06,
      "loss": 1.1346,
      "step": 24732
    },
    {
      "epoch": 0.8857414006123874,
      "grad_norm": 1.5928148031234741,
      "learning_rate": 6.767573273103245e-06,
      "loss": 1.3723,
      "step": 24733
    },
    {
      "epoch": 0.8857772127419558,
      "grad_norm": 1.628875970840454,
      "learning_rate": 6.7633794261444005e-06,
      "loss": 1.5718,
      "step": 24734
    },
    {
      "epoch": 0.885813024871524,
      "grad_norm": 1.8851513862609863,
      "learning_rate": 6.7591868335640016e-06,
      "loss": 1.4976,
      "step": 24735
    },
    {
      "epoch": 0.8858488370010923,
      "grad_norm": 1.4460524320602417,
      "learning_rate": 6.754995495418482e-06,
      "loss": 1.467,
      "step": 24736
    },
    {
      "epoch": 0.8858846491306606,
      "grad_norm": 2.1435530185699463,
      "learning_rate": 6.750805411764205e-06,
      "loss": 1.3132,
      "step": 24737
    },
    {
      "epoch": 0.8859204612602288,
      "grad_norm": 1.743093490600586,
      "learning_rate": 6.746616582657583e-06,
      "loss": 1.4094,
      "step": 24738
    },
    {
      "epoch": 0.8859562733897971,
      "grad_norm": 2.133668899536133,
      "learning_rate": 6.742429008154927e-06,
      "loss": 1.4657,
      "step": 24739
    },
    {
      "epoch": 0.8859920855193654,
      "grad_norm": 1.4859719276428223,
      "learning_rate": 6.738242688312602e-06,
      "loss": 1.3478,
      "step": 24740
    },
    {
      "epoch": 0.8860278976489337,
      "grad_norm": 2.430969715118408,
      "learning_rate": 6.734057623186929e-06,
      "loss": 1.3177,
      "step": 24741
    },
    {
      "epoch": 0.886063709778502,
      "grad_norm": 1.748104453086853,
      "learning_rate": 6.729873812834198e-06,
      "loss": 1.2021,
      "step": 24742
    },
    {
      "epoch": 0.8860995219080703,
      "grad_norm": 2.162109375,
      "learning_rate": 6.725691257310718e-06,
      "loss": 1.6393,
      "step": 24743
    },
    {
      "epoch": 0.8861353340376386,
      "grad_norm": 1.8088104724884033,
      "learning_rate": 6.721509956672711e-06,
      "loss": 1.4486,
      "step": 24744
    },
    {
      "epoch": 0.8861711461672068,
      "grad_norm": 1.3711811304092407,
      "learning_rate": 6.7173299109765e-06,
      "loss": 1.4542,
      "step": 24745
    },
    {
      "epoch": 0.8862069582967751,
      "grad_norm": 2.1322035789489746,
      "learning_rate": 6.713151120278283e-06,
      "loss": 1.5574,
      "step": 24746
    },
    {
      "epoch": 0.8862427704263434,
      "grad_norm": 2.123325824737549,
      "learning_rate": 6.7089735846342815e-06,
      "loss": 1.3594,
      "step": 24747
    },
    {
      "epoch": 0.8862785825559117,
      "grad_norm": 1.7514166831970215,
      "learning_rate": 6.704797304100707e-06,
      "loss": 1.2006,
      "step": 24748
    },
    {
      "epoch": 0.88631439468548,
      "grad_norm": 1.8067326545715332,
      "learning_rate": 6.700622278733748e-06,
      "loss": 1.5183,
      "step": 24749
    },
    {
      "epoch": 0.8863502068150483,
      "grad_norm": 1.8331403732299805,
      "learning_rate": 6.69644850858957e-06,
      "loss": 1.411,
      "step": 24750
    },
    {
      "epoch": 0.8863860189446165,
      "grad_norm": 1.604770541191101,
      "learning_rate": 6.692275993724295e-06,
      "loss": 1.3225,
      "step": 24751
    },
    {
      "epoch": 0.8864218310741848,
      "grad_norm": 2.0157501697540283,
      "learning_rate": 6.688104734194123e-06,
      "loss": 1.3786,
      "step": 24752
    },
    {
      "epoch": 0.8864576432037531,
      "grad_norm": 1.6937917470932007,
      "learning_rate": 6.683934730055119e-06,
      "loss": 1.3432,
      "step": 24753
    },
    {
      "epoch": 0.8864934553333214,
      "grad_norm": 1.6693098545074463,
      "learning_rate": 6.679765981363417e-06,
      "loss": 1.6571,
      "step": 24754
    },
    {
      "epoch": 0.8865292674628896,
      "grad_norm": 1.3133947849273682,
      "learning_rate": 6.675598488175061e-06,
      "loss": 1.4155,
      "step": 24755
    },
    {
      "epoch": 0.886565079592458,
      "grad_norm": 1.407610297203064,
      "learning_rate": 6.671432250546184e-06,
      "loss": 1.4084,
      "step": 24756
    },
    {
      "epoch": 0.8866008917220263,
      "grad_norm": 1.6117279529571533,
      "learning_rate": 6.6672672685327955e-06,
      "loss": 1.511,
      "step": 24757
    },
    {
      "epoch": 0.8866367038515945,
      "grad_norm": 1.8112707138061523,
      "learning_rate": 6.663103542190918e-06,
      "loss": 1.7054,
      "step": 24758
    },
    {
      "epoch": 0.8866725159811628,
      "grad_norm": 1.5144332647323608,
      "learning_rate": 6.658941071576597e-06,
      "loss": 1.3524,
      "step": 24759
    },
    {
      "epoch": 0.8867083281107311,
      "grad_norm": 1.8527796268463135,
      "learning_rate": 6.654779856745807e-06,
      "loss": 1.6036,
      "step": 24760
    },
    {
      "epoch": 0.8867441402402993,
      "grad_norm": 1.8971564769744873,
      "learning_rate": 6.650619897754573e-06,
      "loss": 1.2975,
      "step": 24761
    },
    {
      "epoch": 0.8867799523698676,
      "grad_norm": 1.2585480213165283,
      "learning_rate": 6.646461194658804e-06,
      "loss": 1.4554,
      "step": 24762
    },
    {
      "epoch": 0.886815764499436,
      "grad_norm": 1.744014024734497,
      "learning_rate": 6.642303747514511e-06,
      "loss": 1.4196,
      "step": 24763
    },
    {
      "epoch": 0.8868515766290043,
      "grad_norm": 1.7813466787338257,
      "learning_rate": 6.638147556377583e-06,
      "loss": 1.1593,
      "step": 24764
    },
    {
      "epoch": 0.8868873887585725,
      "grad_norm": 1.533890962600708,
      "learning_rate": 6.633992621303975e-06,
      "loss": 1.6984,
      "step": 24765
    },
    {
      "epoch": 0.8869232008881408,
      "grad_norm": 1.2901742458343506,
      "learning_rate": 6.629838942349542e-06,
      "loss": 1.3292,
      "step": 24766
    },
    {
      "epoch": 0.8869590130177091,
      "grad_norm": 1.23209810256958,
      "learning_rate": 6.625686519570184e-06,
      "loss": 1.5023,
      "step": 24767
    },
    {
      "epoch": 0.8869948251472773,
      "grad_norm": 1.7562414407730103,
      "learning_rate": 6.621535353021791e-06,
      "loss": 1.3996,
      "step": 24768
    },
    {
      "epoch": 0.8870306372768456,
      "grad_norm": 1.7048919200897217,
      "learning_rate": 6.617385442760171e-06,
      "loss": 1.4065,
      "step": 24769
    },
    {
      "epoch": 0.887066449406414,
      "grad_norm": 2.8547306060791016,
      "learning_rate": 6.61323678884117e-06,
      "loss": 1.6266,
      "step": 24770
    },
    {
      "epoch": 0.8871022615359823,
      "grad_norm": 1.4367671012878418,
      "learning_rate": 6.6090893913206106e-06,
      "loss": 1.5252,
      "step": 24771
    },
    {
      "epoch": 0.8871380736655505,
      "grad_norm": 1.2508554458618164,
      "learning_rate": 6.604943250254303e-06,
      "loss": 0.8535,
      "step": 24772
    },
    {
      "epoch": 0.8871738857951188,
      "grad_norm": 1.5373388528823853,
      "learning_rate": 6.600798365697991e-06,
      "loss": 1.3155,
      "step": 24773
    },
    {
      "epoch": 0.8872096979246871,
      "grad_norm": 1.792191743850708,
      "learning_rate": 6.596654737707486e-06,
      "loss": 1.4589,
      "step": 24774
    },
    {
      "epoch": 0.8872455100542553,
      "grad_norm": 2.387756109237671,
      "learning_rate": 6.592512366338499e-06,
      "loss": 1.5162,
      "step": 24775
    },
    {
      "epoch": 0.8872813221838236,
      "grad_norm": 1.9498742818832397,
      "learning_rate": 6.588371251646774e-06,
      "loss": 1.6594,
      "step": 24776
    },
    {
      "epoch": 0.887317134313392,
      "grad_norm": 1.4522777795791626,
      "learning_rate": 6.584231393688012e-06,
      "loss": 1.5712,
      "step": 24777
    },
    {
      "epoch": 0.8873529464429603,
      "grad_norm": 1.6063610315322876,
      "learning_rate": 6.5800927925179115e-06,
      "loss": 1.3801,
      "step": 24778
    },
    {
      "epoch": 0.8873887585725285,
      "grad_norm": 1.822343111038208,
      "learning_rate": 6.575955448192184e-06,
      "loss": 1.4519,
      "step": 24779
    },
    {
      "epoch": 0.8874245707020968,
      "grad_norm": 2.020034074783325,
      "learning_rate": 6.5718193607664516e-06,
      "loss": 1.1206,
      "step": 24780
    },
    {
      "epoch": 0.8874603828316651,
      "grad_norm": 2.021859884262085,
      "learning_rate": 6.5676845302963805e-06,
      "loss": 1.403,
      "step": 24781
    },
    {
      "epoch": 0.8874961949612333,
      "grad_norm": 1.5246386528015137,
      "learning_rate": 6.563550956837594e-06,
      "loss": 1.1573,
      "step": 24782
    },
    {
      "epoch": 0.8875320070908016,
      "grad_norm": 1.686280369758606,
      "learning_rate": 6.559418640445714e-06,
      "loss": 1.4514,
      "step": 24783
    },
    {
      "epoch": 0.88756781922037,
      "grad_norm": 1.4774922132492065,
      "learning_rate": 6.555287581176317e-06,
      "loss": 1.606,
      "step": 24784
    },
    {
      "epoch": 0.8876036313499382,
      "grad_norm": 1.6550049781799316,
      "learning_rate": 6.551157779084982e-06,
      "loss": 1.2363,
      "step": 24785
    },
    {
      "epoch": 0.8876394434795065,
      "grad_norm": 1.3732515573501587,
      "learning_rate": 6.547029234227298e-06,
      "loss": 1.4708,
      "step": 24786
    },
    {
      "epoch": 0.8876752556090748,
      "grad_norm": 1.4818493127822876,
      "learning_rate": 6.5429019466587745e-06,
      "loss": 1.5955,
      "step": 24787
    },
    {
      "epoch": 0.8877110677386431,
      "grad_norm": 1.542306900024414,
      "learning_rate": 6.5387759164349585e-06,
      "loss": 1.4008,
      "step": 24788
    },
    {
      "epoch": 0.8877468798682113,
      "grad_norm": 1.4538143873214722,
      "learning_rate": 6.5346511436113585e-06,
      "loss": 1.2013,
      "step": 24789
    },
    {
      "epoch": 0.8877826919977796,
      "grad_norm": 2.1333580017089844,
      "learning_rate": 6.5305276282434765e-06,
      "loss": 1.4146,
      "step": 24790
    },
    {
      "epoch": 0.887818504127348,
      "grad_norm": 1.75994873046875,
      "learning_rate": 6.526405370386757e-06,
      "loss": 1.4529,
      "step": 24791
    },
    {
      "epoch": 0.8878543162569162,
      "grad_norm": 1.6889221668243408,
      "learning_rate": 6.522284370096687e-06,
      "loss": 1.5275,
      "step": 24792
    },
    {
      "epoch": 0.8878901283864845,
      "grad_norm": 1.9283156394958496,
      "learning_rate": 6.518164627428724e-06,
      "loss": 1.5469,
      "step": 24793
    },
    {
      "epoch": 0.8879259405160528,
      "grad_norm": 1.4188671112060547,
      "learning_rate": 6.514046142438246e-06,
      "loss": 1.3232,
      "step": 24794
    },
    {
      "epoch": 0.887961752645621,
      "grad_norm": 2.01699161529541,
      "learning_rate": 6.509928915180697e-06,
      "loss": 1.5934,
      "step": 24795
    },
    {
      "epoch": 0.8879975647751893,
      "grad_norm": 1.3128671646118164,
      "learning_rate": 6.505812945711454e-06,
      "loss": 1.2986,
      "step": 24796
    },
    {
      "epoch": 0.8880333769047576,
      "grad_norm": 1.3702200651168823,
      "learning_rate": 6.501698234085929e-06,
      "loss": 1.4046,
      "step": 24797
    },
    {
      "epoch": 0.888069189034326,
      "grad_norm": 1.5460312366485596,
      "learning_rate": 6.497584780359423e-06,
      "loss": 1.3257,
      "step": 24798
    },
    {
      "epoch": 0.8881050011638942,
      "grad_norm": 1.7780214548110962,
      "learning_rate": 6.4934725845873016e-06,
      "loss": 1.5938,
      "step": 24799
    },
    {
      "epoch": 0.8881408132934625,
      "grad_norm": 1.7738583087921143,
      "learning_rate": 6.489361646824898e-06,
      "loss": 1.3187,
      "step": 24800
    },
    {
      "epoch": 0.8881766254230308,
      "grad_norm": 1.6108421087265015,
      "learning_rate": 6.485251967127526e-06,
      "loss": 1.6039,
      "step": 24801
    },
    {
      "epoch": 0.888212437552599,
      "grad_norm": 1.5561243295669556,
      "learning_rate": 6.48114354555045e-06,
      "loss": 1.3836,
      "step": 24802
    },
    {
      "epoch": 0.8882482496821673,
      "grad_norm": 1.284730076789856,
      "learning_rate": 6.47703638214896e-06,
      "loss": 1.3728,
      "step": 24803
    },
    {
      "epoch": 0.8882840618117356,
      "grad_norm": 1.3639391660690308,
      "learning_rate": 6.4729304769783225e-06,
      "loss": 1.3553,
      "step": 24804
    },
    {
      "epoch": 0.888319873941304,
      "grad_norm": 1.3871612548828125,
      "learning_rate": 6.468825830093739e-06,
      "loss": 1.626,
      "step": 24805
    },
    {
      "epoch": 0.8883556860708722,
      "grad_norm": 1.5964213609695435,
      "learning_rate": 6.4647224415504745e-06,
      "loss": 1.4741,
      "step": 24806
    },
    {
      "epoch": 0.8883914982004405,
      "grad_norm": 1.4523588418960571,
      "learning_rate": 6.460620311403709e-06,
      "loss": 1.5063,
      "step": 24807
    },
    {
      "epoch": 0.8884273103300088,
      "grad_norm": 1.8651169538497925,
      "learning_rate": 6.456519439708653e-06,
      "loss": 1.5239,
      "step": 24808
    },
    {
      "epoch": 0.888463122459577,
      "grad_norm": 1.986107349395752,
      "learning_rate": 6.452419826520451e-06,
      "loss": 1.4542,
      "step": 24809
    },
    {
      "epoch": 0.8884989345891453,
      "grad_norm": 1.5596935749053955,
      "learning_rate": 6.44832147189427e-06,
      "loss": 1.439,
      "step": 24810
    },
    {
      "epoch": 0.8885347467187136,
      "grad_norm": 1.6743347644805908,
      "learning_rate": 6.444224375885277e-06,
      "loss": 1.5254,
      "step": 24811
    },
    {
      "epoch": 0.888570558848282,
      "grad_norm": 1.4157236814498901,
      "learning_rate": 6.44012853854854e-06,
      "loss": 1.3591,
      "step": 24812
    },
    {
      "epoch": 0.8886063709778502,
      "grad_norm": 1.396809458732605,
      "learning_rate": 6.436033959939192e-06,
      "loss": 1.1897,
      "step": 24813
    },
    {
      "epoch": 0.8886421831074185,
      "grad_norm": 1.3337267637252808,
      "learning_rate": 6.431940640112322e-06,
      "loss": 1.7,
      "step": 24814
    },
    {
      "epoch": 0.8886779952369868,
      "grad_norm": 1.819524884223938,
      "learning_rate": 6.4278485791230195e-06,
      "loss": 1.5505,
      "step": 24815
    },
    {
      "epoch": 0.888713807366555,
      "grad_norm": 1.781790018081665,
      "learning_rate": 6.423757777026285e-06,
      "loss": 1.2891,
      "step": 24816
    },
    {
      "epoch": 0.8887496194961233,
      "grad_norm": 1.4497478008270264,
      "learning_rate": 6.419668233877197e-06,
      "loss": 1.448,
      "step": 24817
    },
    {
      "epoch": 0.8887854316256916,
      "grad_norm": 1.353127121925354,
      "learning_rate": 6.415579949730755e-06,
      "loss": 1.514,
      "step": 24818
    },
    {
      "epoch": 0.88882124375526,
      "grad_norm": 1.5938628911972046,
      "learning_rate": 6.411492924641982e-06,
      "loss": 1.3923,
      "step": 24819
    },
    {
      "epoch": 0.8888570558848282,
      "grad_norm": 1.3531911373138428,
      "learning_rate": 6.407407158665846e-06,
      "loss": 1.4075,
      "step": 24820
    },
    {
      "epoch": 0.8888928680143965,
      "grad_norm": 1.8812788724899292,
      "learning_rate": 6.403322651857313e-06,
      "loss": 1.4905,
      "step": 24821
    },
    {
      "epoch": 0.8889286801439648,
      "grad_norm": 1.4424673318862915,
      "learning_rate": 6.399239404271362e-06,
      "loss": 1.4742,
      "step": 24822
    },
    {
      "epoch": 0.888964492273533,
      "grad_norm": 1.4213186502456665,
      "learning_rate": 6.395157415962894e-06,
      "loss": 1.4607,
      "step": 24823
    },
    {
      "epoch": 0.8890003044031013,
      "grad_norm": 1.715586543083191,
      "learning_rate": 6.39107668698683e-06,
      "loss": 1.298,
      "step": 24824
    },
    {
      "epoch": 0.8890361165326696,
      "grad_norm": 1.4657044410705566,
      "learning_rate": 6.386997217398094e-06,
      "loss": 1.6368,
      "step": 24825
    },
    {
      "epoch": 0.8890719286622379,
      "grad_norm": 1.7812143564224243,
      "learning_rate": 6.382919007251575e-06,
      "loss": 1.4923,
      "step": 24826
    },
    {
      "epoch": 0.8891077407918062,
      "grad_norm": 2.4855334758758545,
      "learning_rate": 6.378842056602097e-06,
      "loss": 1.3953,
      "step": 24827
    },
    {
      "epoch": 0.8891435529213745,
      "grad_norm": 1.778143286705017,
      "learning_rate": 6.374766365504547e-06,
      "loss": 1.3601,
      "step": 24828
    },
    {
      "epoch": 0.8891793650509427,
      "grad_norm": 1.8928757905960083,
      "learning_rate": 6.370691934013761e-06,
      "loss": 1.29,
      "step": 24829
    },
    {
      "epoch": 0.889215177180511,
      "grad_norm": 1.5196993350982666,
      "learning_rate": 6.366618762184529e-06,
      "loss": 1.4735,
      "step": 24830
    },
    {
      "epoch": 0.8892509893100793,
      "grad_norm": 1.5085698366165161,
      "learning_rate": 6.36254685007166e-06,
      "loss": 1.7028,
      "step": 24831
    },
    {
      "epoch": 0.8892868014396476,
      "grad_norm": 1.4420183897018433,
      "learning_rate": 6.358476197729934e-06,
      "loss": 1.086,
      "step": 24832
    },
    {
      "epoch": 0.8893226135692159,
      "grad_norm": 1.4737896919250488,
      "learning_rate": 6.3544068052141415e-06,
      "loss": 1.3265,
      "step": 24833
    },
    {
      "epoch": 0.8893584256987842,
      "grad_norm": 1.763348937034607,
      "learning_rate": 6.3503386725790034e-06,
      "loss": 1.3154,
      "step": 24834
    },
    {
      "epoch": 0.8893942378283525,
      "grad_norm": 1.7766228914260864,
      "learning_rate": 6.346271799879244e-06,
      "loss": 1.1381,
      "step": 24835
    },
    {
      "epoch": 0.8894300499579207,
      "grad_norm": 1.7439228296279907,
      "learning_rate": 6.342206187169608e-06,
      "loss": 1.1344,
      "step": 24836
    },
    {
      "epoch": 0.889465862087489,
      "grad_norm": 1.729805588722229,
      "learning_rate": 6.338141834504785e-06,
      "loss": 1.3619,
      "step": 24837
    },
    {
      "epoch": 0.8895016742170573,
      "grad_norm": 2.5667622089385986,
      "learning_rate": 6.3340787419394535e-06,
      "loss": 1.2524,
      "step": 24838
    },
    {
      "epoch": 0.8895374863466255,
      "grad_norm": 1.547900676727295,
      "learning_rate": 6.330016909528236e-06,
      "loss": 1.2293,
      "step": 24839
    },
    {
      "epoch": 0.8895732984761939,
      "grad_norm": 1.8758774995803833,
      "learning_rate": 6.325956337325845e-06,
      "loss": 1.3833,
      "step": 24840
    },
    {
      "epoch": 0.8896091106057622,
      "grad_norm": 1.384765625,
      "learning_rate": 6.321897025386869e-06,
      "loss": 1.1106,
      "step": 24841
    },
    {
      "epoch": 0.8896449227353305,
      "grad_norm": 1.5886365175247192,
      "learning_rate": 6.317838973765944e-06,
      "loss": 1.6064,
      "step": 24842
    },
    {
      "epoch": 0.8896807348648987,
      "grad_norm": 1.6912497282028198,
      "learning_rate": 6.313782182517636e-06,
      "loss": 1.5548,
      "step": 24843
    },
    {
      "epoch": 0.889716546994467,
      "grad_norm": 1.452126145362854,
      "learning_rate": 6.309726651696557e-06,
      "loss": 1.4011,
      "step": 24844
    },
    {
      "epoch": 0.8897523591240353,
      "grad_norm": 1.428489089012146,
      "learning_rate": 6.305672381357264e-06,
      "loss": 1.322,
      "step": 24845
    },
    {
      "epoch": 0.8897881712536035,
      "grad_norm": 1.99895179271698,
      "learning_rate": 6.301619371554257e-06,
      "loss": 1.6756,
      "step": 24846
    },
    {
      "epoch": 0.8898239833831719,
      "grad_norm": 2.284052848815918,
      "learning_rate": 6.297567622342127e-06,
      "loss": 1.495,
      "step": 24847
    },
    {
      "epoch": 0.8898597955127402,
      "grad_norm": 1.3983558416366577,
      "learning_rate": 6.29351713377535e-06,
      "loss": 1.1678,
      "step": 24848
    },
    {
      "epoch": 0.8898956076423085,
      "grad_norm": 1.4451669454574585,
      "learning_rate": 6.289467905908442e-06,
      "loss": 1.4856,
      "step": 24849
    },
    {
      "epoch": 0.8899314197718767,
      "grad_norm": 1.845213770866394,
      "learning_rate": 6.285419938795833e-06,
      "loss": 1.5874,
      "step": 24850
    },
    {
      "epoch": 0.889967231901445,
      "grad_norm": 1.7814226150512695,
      "learning_rate": 6.281373232492038e-06,
      "loss": 1.3611,
      "step": 24851
    },
    {
      "epoch": 0.8900030440310133,
      "grad_norm": 2.072495222091675,
      "learning_rate": 6.2773277870514675e-06,
      "loss": 1.3698,
      "step": 24852
    },
    {
      "epoch": 0.8900388561605815,
      "grad_norm": 1.623978853225708,
      "learning_rate": 6.273283602528579e-06,
      "loss": 1.0922,
      "step": 24853
    },
    {
      "epoch": 0.8900746682901499,
      "grad_norm": 1.4474862813949585,
      "learning_rate": 6.269240678977739e-06,
      "loss": 1.3673,
      "step": 24854
    },
    {
      "epoch": 0.8901104804197182,
      "grad_norm": 1.9575895071029663,
      "learning_rate": 6.265199016453371e-06,
      "loss": 1.5366,
      "step": 24855
    },
    {
      "epoch": 0.8901462925492865,
      "grad_norm": 1.3855613470077515,
      "learning_rate": 6.261158615009843e-06,
      "loss": 1.3095,
      "step": 24856
    },
    {
      "epoch": 0.8901821046788547,
      "grad_norm": 1.46273934841156,
      "learning_rate": 6.25711947470149e-06,
      "loss": 1.7024,
      "step": 24857
    },
    {
      "epoch": 0.890217916808423,
      "grad_norm": 1.435225248336792,
      "learning_rate": 6.253081595582699e-06,
      "loss": 1.3673,
      "step": 24858
    },
    {
      "epoch": 0.8902537289379913,
      "grad_norm": 1.5706956386566162,
      "learning_rate": 6.249044977707763e-06,
      "loss": 1.3358,
      "step": 24859
    },
    {
      "epoch": 0.8902895410675595,
      "grad_norm": 1.7113782167434692,
      "learning_rate": 6.245009621131004e-06,
      "loss": 1.4194,
      "step": 24860
    },
    {
      "epoch": 0.8903253531971279,
      "grad_norm": 1.5109610557556152,
      "learning_rate": 6.2409755259066786e-06,
      "loss": 1.3438,
      "step": 24861
    },
    {
      "epoch": 0.8903611653266962,
      "grad_norm": 1.6783028841018677,
      "learning_rate": 6.23694269208912e-06,
      "loss": 1.2793,
      "step": 24862
    },
    {
      "epoch": 0.8903969774562644,
      "grad_norm": 2.1244373321533203,
      "learning_rate": 6.232911119732554e-06,
      "loss": 1.4257,
      "step": 24863
    },
    {
      "epoch": 0.8904327895858327,
      "grad_norm": 1.8511030673980713,
      "learning_rate": 6.228880808891202e-06,
      "loss": 1.1956,
      "step": 24864
    },
    {
      "epoch": 0.890468601715401,
      "grad_norm": 1.8468077182769775,
      "learning_rate": 6.224851759619299e-06,
      "loss": 1.4459,
      "step": 24865
    },
    {
      "epoch": 0.8905044138449693,
      "grad_norm": 1.4527554512023926,
      "learning_rate": 6.220823971971046e-06,
      "loss": 1.4473,
      "step": 24866
    },
    {
      "epoch": 0.8905402259745375,
      "grad_norm": 1.7244460582733154,
      "learning_rate": 6.216797446000666e-06,
      "loss": 1.3496,
      "step": 24867
    },
    {
      "epoch": 0.8905760381041059,
      "grad_norm": 1.394762635231018,
      "learning_rate": 6.212772181762283e-06,
      "loss": 1.3262,
      "step": 24868
    },
    {
      "epoch": 0.8906118502336742,
      "grad_norm": 1.723497748374939,
      "learning_rate": 6.208748179310087e-06,
      "loss": 1.1949,
      "step": 24869
    },
    {
      "epoch": 0.8906476623632424,
      "grad_norm": 1.4614174365997314,
      "learning_rate": 6.204725438698189e-06,
      "loss": 1.5436,
      "step": 24870
    },
    {
      "epoch": 0.8906834744928107,
      "grad_norm": 1.726501703262329,
      "learning_rate": 6.200703959980747e-06,
      "loss": 1.3518,
      "step": 24871
    },
    {
      "epoch": 0.890719286622379,
      "grad_norm": 1.5753802061080933,
      "learning_rate": 6.196683743211818e-06,
      "loss": 1.4447,
      "step": 24872
    },
    {
      "epoch": 0.8907550987519472,
      "grad_norm": 1.9597927331924438,
      "learning_rate": 6.192664788445513e-06,
      "loss": 1.5194,
      "step": 24873
    },
    {
      "epoch": 0.8907909108815155,
      "grad_norm": 1.322771668434143,
      "learning_rate": 6.188647095735911e-06,
      "loss": 1.6306,
      "step": 24874
    },
    {
      "epoch": 0.8908267230110839,
      "grad_norm": 1.6770081520080566,
      "learning_rate": 6.184630665137048e-06,
      "loss": 1.2971,
      "step": 24875
    },
    {
      "epoch": 0.8908625351406522,
      "grad_norm": 1.794987440109253,
      "learning_rate": 6.180615496702968e-06,
      "loss": 1.5026,
      "step": 24876
    },
    {
      "epoch": 0.8908983472702204,
      "grad_norm": 1.406591534614563,
      "learning_rate": 6.176601590487685e-06,
      "loss": 1.1957,
      "step": 24877
    },
    {
      "epoch": 0.8909341593997887,
      "grad_norm": 1.7108861207962036,
      "learning_rate": 6.17258894654521e-06,
      "loss": 1.3956,
      "step": 24878
    },
    {
      "epoch": 0.890969971529357,
      "grad_norm": 1.513010859489441,
      "learning_rate": 6.168577564929523e-06,
      "loss": 1.6367,
      "step": 24879
    },
    {
      "epoch": 0.8910057836589252,
      "grad_norm": 1.442939043045044,
      "learning_rate": 6.16456744569458e-06,
      "loss": 1.1821,
      "step": 24880
    },
    {
      "epoch": 0.8910415957884935,
      "grad_norm": 2.025557279586792,
      "learning_rate": 6.160558588894361e-06,
      "loss": 1.5068,
      "step": 24881
    },
    {
      "epoch": 0.8910774079180619,
      "grad_norm": 1.4389938116073608,
      "learning_rate": 6.156550994582766e-06,
      "loss": 1.0765,
      "step": 24882
    },
    {
      "epoch": 0.8911132200476302,
      "grad_norm": 1.3525745868682861,
      "learning_rate": 6.1525446628137306e-06,
      "loss": 1.2608,
      "step": 24883
    },
    {
      "epoch": 0.8911490321771984,
      "grad_norm": 1.968949794769287,
      "learning_rate": 6.148539593641156e-06,
      "loss": 1.5984,
      "step": 24884
    },
    {
      "epoch": 0.8911848443067667,
      "grad_norm": 1.5880968570709229,
      "learning_rate": 6.144535787118921e-06,
      "loss": 1.5123,
      "step": 24885
    },
    {
      "epoch": 0.891220656436335,
      "grad_norm": 2.515672206878662,
      "learning_rate": 6.140533243300894e-06,
      "loss": 1.5112,
      "step": 24886
    },
    {
      "epoch": 0.8912564685659032,
      "grad_norm": 1.2918782234191895,
      "learning_rate": 6.13653196224091e-06,
      "loss": 1.0852,
      "step": 24887
    },
    {
      "epoch": 0.8912922806954715,
      "grad_norm": 1.7701398134231567,
      "learning_rate": 6.132531943992826e-06,
      "loss": 1.1541,
      "step": 24888
    },
    {
      "epoch": 0.8913280928250399,
      "grad_norm": 1.8440961837768555,
      "learning_rate": 6.128533188610453e-06,
      "loss": 1.3715,
      "step": 24889
    },
    {
      "epoch": 0.8913639049546082,
      "grad_norm": 2.0885820388793945,
      "learning_rate": 6.124535696147559e-06,
      "loss": 1.8181,
      "step": 24890
    },
    {
      "epoch": 0.8913997170841764,
      "grad_norm": 1.7771596908569336,
      "learning_rate": 6.12053946665796e-06,
      "loss": 1.2956,
      "step": 24891
    },
    {
      "epoch": 0.8914355292137447,
      "grad_norm": 1.7697362899780273,
      "learning_rate": 6.1165445001954095e-06,
      "loss": 1.2343,
      "step": 24892
    },
    {
      "epoch": 0.891471341343313,
      "grad_norm": 2.0563573837280273,
      "learning_rate": 6.112550796813643e-06,
      "loss": 1.5779,
      "step": 24893
    },
    {
      "epoch": 0.8915071534728812,
      "grad_norm": 1.5106770992279053,
      "learning_rate": 6.108558356566396e-06,
      "loss": 1.5096,
      "step": 24894
    },
    {
      "epoch": 0.8915429656024495,
      "grad_norm": 1.5133570432662964,
      "learning_rate": 6.104567179507381e-06,
      "loss": 1.21,
      "step": 24895
    },
    {
      "epoch": 0.8915787777320179,
      "grad_norm": 1.82058584690094,
      "learning_rate": 6.100577265690321e-06,
      "loss": 1.6064,
      "step": 24896
    },
    {
      "epoch": 0.8916145898615861,
      "grad_norm": 1.4742602109909058,
      "learning_rate": 6.096588615168864e-06,
      "loss": 1.2988,
      "step": 24897
    },
    {
      "epoch": 0.8916504019911544,
      "grad_norm": 1.487149953842163,
      "learning_rate": 6.092601227996664e-06,
      "loss": 1.4094,
      "step": 24898
    },
    {
      "epoch": 0.8916862141207227,
      "grad_norm": 1.983708381652832,
      "learning_rate": 6.088615104227413e-06,
      "loss": 1.4038,
      "step": 24899
    },
    {
      "epoch": 0.891722026250291,
      "grad_norm": 1.6666429042816162,
      "learning_rate": 6.084630243914679e-06,
      "loss": 1.379,
      "step": 24900
    },
    {
      "epoch": 0.8917578383798592,
      "grad_norm": 1.616948127746582,
      "learning_rate": 6.080646647112109e-06,
      "loss": 1.3444,
      "step": 24901
    },
    {
      "epoch": 0.8917936505094275,
      "grad_norm": 1.2481621503829956,
      "learning_rate": 6.076664313873293e-06,
      "loss": 1.2012,
      "step": 24902
    },
    {
      "epoch": 0.8918294626389959,
      "grad_norm": 1.9596446752548218,
      "learning_rate": 6.07268324425182e-06,
      "loss": 1.6332,
      "step": 24903
    },
    {
      "epoch": 0.8918652747685641,
      "grad_norm": 2.1639833450317383,
      "learning_rate": 6.068703438301226e-06,
      "loss": 1.6922,
      "step": 24904
    },
    {
      "epoch": 0.8919010868981324,
      "grad_norm": 1.326301097869873,
      "learning_rate": 6.064724896075058e-06,
      "loss": 1.5365,
      "step": 24905
    },
    {
      "epoch": 0.8919368990277007,
      "grad_norm": 1.8309438228607178,
      "learning_rate": 6.06074761762685e-06,
      "loss": 1.2234,
      "step": 24906
    },
    {
      "epoch": 0.891972711157269,
      "grad_norm": 1.4915207624435425,
      "learning_rate": 6.056771603010125e-06,
      "loss": 1.4147,
      "step": 24907
    },
    {
      "epoch": 0.8920085232868372,
      "grad_norm": 1.525928020477295,
      "learning_rate": 6.052796852278353e-06,
      "loss": 1.5172,
      "step": 24908
    },
    {
      "epoch": 0.8920443354164055,
      "grad_norm": 1.383877158164978,
      "learning_rate": 6.048823365485012e-06,
      "loss": 1.2705,
      "step": 24909
    },
    {
      "epoch": 0.8920801475459739,
      "grad_norm": 1.3349719047546387,
      "learning_rate": 6.044851142683572e-06,
      "loss": 1.3585,
      "step": 24910
    },
    {
      "epoch": 0.8921159596755421,
      "grad_norm": 2.2004175186157227,
      "learning_rate": 6.040880183927455e-06,
      "loss": 1.3931,
      "step": 24911
    },
    {
      "epoch": 0.8921517718051104,
      "grad_norm": 1.7021312713623047,
      "learning_rate": 6.036910489270098e-06,
      "loss": 1.2756,
      "step": 24912
    },
    {
      "epoch": 0.8921875839346787,
      "grad_norm": 1.871277093887329,
      "learning_rate": 6.0329420587649124e-06,
      "loss": 1.5018,
      "step": 24913
    },
    {
      "epoch": 0.8922233960642469,
      "grad_norm": 2.3825483322143555,
      "learning_rate": 6.028974892465289e-06,
      "loss": 1.2814,
      "step": 24914
    },
    {
      "epoch": 0.8922592081938152,
      "grad_norm": 2.0105133056640625,
      "learning_rate": 6.025008990424585e-06,
      "loss": 1.2743,
      "step": 24915
    },
    {
      "epoch": 0.8922950203233835,
      "grad_norm": 2.2031428813934326,
      "learning_rate": 6.021044352696159e-06,
      "loss": 1.8541,
      "step": 24916
    },
    {
      "epoch": 0.8923308324529519,
      "grad_norm": 1.5223588943481445,
      "learning_rate": 6.017080979333378e-06,
      "loss": 1.491,
      "step": 24917
    },
    {
      "epoch": 0.8923666445825201,
      "grad_norm": 4.320333957672119,
      "learning_rate": 6.013118870389523e-06,
      "loss": 1.7306,
      "step": 24918
    },
    {
      "epoch": 0.8924024567120884,
      "grad_norm": 1.5868854522705078,
      "learning_rate": 6.009158025917927e-06,
      "loss": 1.2724,
      "step": 24919
    },
    {
      "epoch": 0.8924382688416567,
      "grad_norm": 1.3187925815582275,
      "learning_rate": 6.00519844597186e-06,
      "loss": 1.2816,
      "step": 24920
    },
    {
      "epoch": 0.8924740809712249,
      "grad_norm": 1.4020674228668213,
      "learning_rate": 6.001240130604624e-06,
      "loss": 1.484,
      "step": 24921
    },
    {
      "epoch": 0.8925098931007932,
      "grad_norm": 1.5411213636398315,
      "learning_rate": 5.997283079869442e-06,
      "loss": 1.261,
      "step": 24922
    },
    {
      "epoch": 0.8925457052303615,
      "grad_norm": 2.140064001083374,
      "learning_rate": 5.993327293819562e-06,
      "loss": 1.439,
      "step": 24923
    },
    {
      "epoch": 0.8925815173599299,
      "grad_norm": 1.97054922580719,
      "learning_rate": 5.989372772508195e-06,
      "loss": 1.3858,
      "step": 24924
    },
    {
      "epoch": 0.8926173294894981,
      "grad_norm": 2.1375653743743896,
      "learning_rate": 5.985419515988566e-06,
      "loss": 1.3219,
      "step": 24925
    },
    {
      "epoch": 0.8926531416190664,
      "grad_norm": 1.6442660093307495,
      "learning_rate": 5.981467524313855e-06,
      "loss": 1.1661,
      "step": 24926
    },
    {
      "epoch": 0.8926889537486347,
      "grad_norm": 1.7716608047485352,
      "learning_rate": 5.977516797537186e-06,
      "loss": 1.5978,
      "step": 24927
    },
    {
      "epoch": 0.8927247658782029,
      "grad_norm": 1.8038569688796997,
      "learning_rate": 5.973567335711783e-06,
      "loss": 1.3623,
      "step": 24928
    },
    {
      "epoch": 0.8927605780077712,
      "grad_norm": 2.1996359825134277,
      "learning_rate": 5.969619138890737e-06,
      "loss": 1.1365,
      "step": 24929
    },
    {
      "epoch": 0.8927963901373395,
      "grad_norm": 1.420465350151062,
      "learning_rate": 5.965672207127171e-06,
      "loss": 1.3239,
      "step": 24930
    },
    {
      "epoch": 0.8928322022669078,
      "grad_norm": 1.2334009408950806,
      "learning_rate": 5.961726540474189e-06,
      "loss": 1.3553,
      "step": 24931
    },
    {
      "epoch": 0.8928680143964761,
      "grad_norm": 1.63217294216156,
      "learning_rate": 5.95778213898488e-06,
      "loss": 1.382,
      "step": 24932
    },
    {
      "epoch": 0.8929038265260444,
      "grad_norm": 1.6627821922302246,
      "learning_rate": 5.9538390027123025e-06,
      "loss": 1.3492,
      "step": 24933
    },
    {
      "epoch": 0.8929396386556127,
      "grad_norm": 1.619146466255188,
      "learning_rate": 5.949897131709514e-06,
      "loss": 1.6053,
      "step": 24934
    },
    {
      "epoch": 0.8929754507851809,
      "grad_norm": 1.4433724880218506,
      "learning_rate": 5.94595652602955e-06,
      "loss": 1.4331,
      "step": 24935
    },
    {
      "epoch": 0.8930112629147492,
      "grad_norm": 1.6781110763549805,
      "learning_rate": 5.9420171857254126e-06,
      "loss": 1.6084,
      "step": 24936
    },
    {
      "epoch": 0.8930470750443175,
      "grad_norm": 1.698456883430481,
      "learning_rate": 5.938079110850114e-06,
      "loss": 1.4175,
      "step": 24937
    },
    {
      "epoch": 0.8930828871738858,
      "grad_norm": 2.07592511177063,
      "learning_rate": 5.934142301456613e-06,
      "loss": 1.6356,
      "step": 24938
    },
    {
      "epoch": 0.8931186993034541,
      "grad_norm": 1.7842410802841187,
      "learning_rate": 5.9302067575979115e-06,
      "loss": 1.3476,
      "step": 24939
    },
    {
      "epoch": 0.8931545114330224,
      "grad_norm": 1.0301146507263184,
      "learning_rate": 5.926272479326922e-06,
      "loss": 1.4211,
      "step": 24940
    },
    {
      "epoch": 0.8931903235625906,
      "grad_norm": 1.4850590229034424,
      "learning_rate": 5.922339466696591e-06,
      "loss": 1.1895,
      "step": 24941
    },
    {
      "epoch": 0.8932261356921589,
      "grad_norm": 1.8592171669006348,
      "learning_rate": 5.918407719759844e-06,
      "loss": 1.6507,
      "step": 24942
    },
    {
      "epoch": 0.8932619478217272,
      "grad_norm": 1.9317047595977783,
      "learning_rate": 5.914477238569549e-06,
      "loss": 1.2963,
      "step": 24943
    },
    {
      "epoch": 0.8932977599512955,
      "grad_norm": 2.370089292526245,
      "learning_rate": 5.91054802317862e-06,
      "loss": 1.7081,
      "step": 24944
    },
    {
      "epoch": 0.8933335720808638,
      "grad_norm": 2.0202910900115967,
      "learning_rate": 5.906620073639868e-06,
      "loss": 1.5157,
      "step": 24945
    },
    {
      "epoch": 0.8933693842104321,
      "grad_norm": 1.8120547533035278,
      "learning_rate": 5.902693390006209e-06,
      "loss": 1.6128,
      "step": 24946
    },
    {
      "epoch": 0.8934051963400004,
      "grad_norm": 1.7174150943756104,
      "learning_rate": 5.89876797233041e-06,
      "loss": 1.5787,
      "step": 24947
    },
    {
      "epoch": 0.8934410084695686,
      "grad_norm": 2.4649229049682617,
      "learning_rate": 5.894843820665319e-06,
      "loss": 1.7147,
      "step": 24948
    },
    {
      "epoch": 0.8934768205991369,
      "grad_norm": 1.6594107151031494,
      "learning_rate": 5.890920935063693e-06,
      "loss": 1.503,
      "step": 24949
    },
    {
      "epoch": 0.8935126327287052,
      "grad_norm": 2.6280345916748047,
      "learning_rate": 5.8869993155783675e-06,
      "loss": 1.2833,
      "step": 24950
    },
    {
      "epoch": 0.8935484448582734,
      "grad_norm": 1.1953978538513184,
      "learning_rate": 5.883078962262056e-06,
      "loss": 1.6845,
      "step": 24951
    },
    {
      "epoch": 0.8935842569878418,
      "grad_norm": 1.5348491668701172,
      "learning_rate": 5.879159875167517e-06,
      "loss": 1.4533,
      "step": 24952
    },
    {
      "epoch": 0.8936200691174101,
      "grad_norm": 1.3968113660812378,
      "learning_rate": 5.875242054347463e-06,
      "loss": 1.3589,
      "step": 24953
    },
    {
      "epoch": 0.8936558812469784,
      "grad_norm": 1.4299107789993286,
      "learning_rate": 5.871325499854618e-06,
      "loss": 1.2363,
      "step": 24954
    },
    {
      "epoch": 0.8936916933765466,
      "grad_norm": 1.3905788660049438,
      "learning_rate": 5.867410211741686e-06,
      "loss": 1.6219,
      "step": 24955
    },
    {
      "epoch": 0.8937275055061149,
      "grad_norm": 1.5503168106079102,
      "learning_rate": 5.863496190061302e-06,
      "loss": 1.6683,
      "step": 24956
    },
    {
      "epoch": 0.8937633176356832,
      "grad_norm": 2.048081159591675,
      "learning_rate": 5.859583434866167e-06,
      "loss": 1.3548,
      "step": 24957
    },
    {
      "epoch": 0.8937991297652514,
      "grad_norm": 1.719864010810852,
      "learning_rate": 5.855671946208896e-06,
      "loss": 1.3932,
      "step": 24958
    },
    {
      "epoch": 0.8938349418948198,
      "grad_norm": 1.683786153793335,
      "learning_rate": 5.851761724142147e-06,
      "loss": 1.3284,
      "step": 24959
    },
    {
      "epoch": 0.8938707540243881,
      "grad_norm": 1.8406994342803955,
      "learning_rate": 5.8478527687184755e-06,
      "loss": 1.3094,
      "step": 24960
    },
    {
      "epoch": 0.8939065661539564,
      "grad_norm": 1.4304991960525513,
      "learning_rate": 5.843945079990498e-06,
      "loss": 1.4482,
      "step": 24961
    },
    {
      "epoch": 0.8939423782835246,
      "grad_norm": 1.7396273612976074,
      "learning_rate": 5.840038658010805e-06,
      "loss": 1.2779,
      "step": 24962
    },
    {
      "epoch": 0.8939781904130929,
      "grad_norm": 1.7192277908325195,
      "learning_rate": 5.83613350283192e-06,
      "loss": 1.3281,
      "step": 24963
    },
    {
      "epoch": 0.8940140025426612,
      "grad_norm": 2.3847055435180664,
      "learning_rate": 5.83222961450639e-06,
      "loss": 1.6135,
      "step": 24964
    },
    {
      "epoch": 0.8940498146722294,
      "grad_norm": 2.184673309326172,
      "learning_rate": 5.828326993086741e-06,
      "loss": 1.3081,
      "step": 24965
    },
    {
      "epoch": 0.8940856268017978,
      "grad_norm": 1.49175226688385,
      "learning_rate": 5.824425638625508e-06,
      "loss": 1.1297,
      "step": 24966
    },
    {
      "epoch": 0.8941214389313661,
      "grad_norm": 1.4161489009857178,
      "learning_rate": 5.820525551175104e-06,
      "loss": 1.3759,
      "step": 24967
    },
    {
      "epoch": 0.8941572510609344,
      "grad_norm": 1.6693317890167236,
      "learning_rate": 5.8166267307880885e-06,
      "loss": 1.653,
      "step": 24968
    },
    {
      "epoch": 0.8941930631905026,
      "grad_norm": 1.2906261682510376,
      "learning_rate": 5.812729177516874e-06,
      "loss": 1.375,
      "step": 24969
    },
    {
      "epoch": 0.8942288753200709,
      "grad_norm": 1.3217500448226929,
      "learning_rate": 5.808832891413873e-06,
      "loss": 1.1146,
      "step": 24970
    },
    {
      "epoch": 0.8942646874496392,
      "grad_norm": 1.2847282886505127,
      "learning_rate": 5.804937872531524e-06,
      "loss": 1.1061,
      "step": 24971
    },
    {
      "epoch": 0.8943004995792074,
      "grad_norm": 1.6961958408355713,
      "learning_rate": 5.8010441209222384e-06,
      "loss": 1.4849,
      "step": 24972
    },
    {
      "epoch": 0.8943363117087758,
      "grad_norm": 1.6735053062438965,
      "learning_rate": 5.797151636638409e-06,
      "loss": 1.1823,
      "step": 24973
    },
    {
      "epoch": 0.8943721238383441,
      "grad_norm": 1.3020800352096558,
      "learning_rate": 5.7932604197323826e-06,
      "loss": 1.0688,
      "step": 24974
    },
    {
      "epoch": 0.8944079359679123,
      "grad_norm": 1.8242229223251343,
      "learning_rate": 5.789370470256517e-06,
      "loss": 1.2799,
      "step": 24975
    },
    {
      "epoch": 0.8944437480974806,
      "grad_norm": 1.5822815895080566,
      "learning_rate": 5.785481788263147e-06,
      "loss": 1.4551,
      "step": 24976
    },
    {
      "epoch": 0.8944795602270489,
      "grad_norm": 1.5726370811462402,
      "learning_rate": 5.7815943738046e-06,
      "loss": 1.1166,
      "step": 24977
    },
    {
      "epoch": 0.8945153723566172,
      "grad_norm": 1.3955146074295044,
      "learning_rate": 5.777708226933165e-06,
      "loss": 1.4049,
      "step": 24978
    },
    {
      "epoch": 0.8945511844861854,
      "grad_norm": 1.5697683095932007,
      "learning_rate": 5.773823347701124e-06,
      "loss": 1.5281,
      "step": 24979
    },
    {
      "epoch": 0.8945869966157538,
      "grad_norm": 1.8382532596588135,
      "learning_rate": 5.7699397361607564e-06,
      "loss": 1.2936,
      "step": 24980
    },
    {
      "epoch": 0.8946228087453221,
      "grad_norm": 1.5816271305084229,
      "learning_rate": 5.766057392364288e-06,
      "loss": 1.4626,
      "step": 24981
    },
    {
      "epoch": 0.8946586208748903,
      "grad_norm": 1.3491520881652832,
      "learning_rate": 5.7621763163639655e-06,
      "loss": 1.4557,
      "step": 24982
    },
    {
      "epoch": 0.8946944330044586,
      "grad_norm": 1.6579393148422241,
      "learning_rate": 5.758296508212013e-06,
      "loss": 1.3494,
      "step": 24983
    },
    {
      "epoch": 0.8947302451340269,
      "grad_norm": 1.8803688287734985,
      "learning_rate": 5.7544179679606234e-06,
      "loss": 1.9863,
      "step": 24984
    },
    {
      "epoch": 0.8947660572635951,
      "grad_norm": 1.9124130010604858,
      "learning_rate": 5.750540695661955e-06,
      "loss": 1.3755,
      "step": 24985
    },
    {
      "epoch": 0.8948018693931634,
      "grad_norm": 2.12723970413208,
      "learning_rate": 5.746664691368187e-06,
      "loss": 1.3116,
      "step": 24986
    },
    {
      "epoch": 0.8948376815227318,
      "grad_norm": 1.3383265733718872,
      "learning_rate": 5.742789955131489e-06,
      "loss": 1.3311,
      "step": 24987
    },
    {
      "epoch": 0.8948734936523001,
      "grad_norm": 2.026008367538452,
      "learning_rate": 5.7389164870039535e-06,
      "loss": 1.4909,
      "step": 24988
    },
    {
      "epoch": 0.8949093057818683,
      "grad_norm": 1.573212742805481,
      "learning_rate": 5.735044287037705e-06,
      "loss": 1.3221,
      "step": 24989
    },
    {
      "epoch": 0.8949451179114366,
      "grad_norm": 1.6352729797363281,
      "learning_rate": 5.7311733552848355e-06,
      "loss": 1.5761,
      "step": 24990
    },
    {
      "epoch": 0.8949809300410049,
      "grad_norm": 1.3640257120132446,
      "learning_rate": 5.727303691797459e-06,
      "loss": 1.2875,
      "step": 24991
    },
    {
      "epoch": 0.8950167421705731,
      "grad_norm": 2.3444406986236572,
      "learning_rate": 5.723435296627588e-06,
      "loss": 1.3894,
      "step": 24992
    },
    {
      "epoch": 0.8950525543001414,
      "grad_norm": 1.89201021194458,
      "learning_rate": 5.719568169827283e-06,
      "loss": 1.1353,
      "step": 24993
    },
    {
      "epoch": 0.8950883664297098,
      "grad_norm": 1.320304274559021,
      "learning_rate": 5.71570231144859e-06,
      "loss": 1.4152,
      "step": 24994
    },
    {
      "epoch": 0.8951241785592781,
      "grad_norm": 2.161402940750122,
      "learning_rate": 5.7118377215435e-06,
      "loss": 1.3054,
      "step": 24995
    },
    {
      "epoch": 0.8951599906888463,
      "grad_norm": 1.4582648277282715,
      "learning_rate": 5.7079744001640065e-06,
      "loss": 1.5407,
      "step": 24996
    },
    {
      "epoch": 0.8951958028184146,
      "grad_norm": 1.544952630996704,
      "learning_rate": 5.70411234736209e-06,
      "loss": 1.2509,
      "step": 24997
    },
    {
      "epoch": 0.8952316149479829,
      "grad_norm": 1.4822784662246704,
      "learning_rate": 5.700251563189718e-06,
      "loss": 1.6312,
      "step": 24998
    },
    {
      "epoch": 0.8952674270775511,
      "grad_norm": 2.0058014392852783,
      "learning_rate": 5.696392047698817e-06,
      "loss": 1.6654,
      "step": 24999
    },
    {
      "epoch": 0.8953032392071194,
      "grad_norm": 2.085139751434326,
      "learning_rate": 5.6925338009413136e-06,
      "loss": 1.3925,
      "step": 25000
    },
    {
      "epoch": 0.8953390513366878,
      "grad_norm": 1.3141660690307617,
      "learning_rate": 5.688676822969119e-06,
      "loss": 1.3244,
      "step": 25001
    },
    {
      "epoch": 0.895374863466256,
      "grad_norm": 1.5696529150009155,
      "learning_rate": 5.684821113834138e-06,
      "loss": 1.462,
      "step": 25002
    },
    {
      "epoch": 0.8954106755958243,
      "grad_norm": 1.421932578086853,
      "learning_rate": 5.680966673588217e-06,
      "loss": 1.2906,
      "step": 25003
    },
    {
      "epoch": 0.8954464877253926,
      "grad_norm": 2.3099539279937744,
      "learning_rate": 5.677113502283227e-06,
      "loss": 1.2604,
      "step": 25004
    },
    {
      "epoch": 0.8954822998549609,
      "grad_norm": 1.683809757232666,
      "learning_rate": 5.673261599971025e-06,
      "loss": 1.7746,
      "step": 25005
    },
    {
      "epoch": 0.8955181119845291,
      "grad_norm": 1.5996848344802856,
      "learning_rate": 5.669410966703393e-06,
      "loss": 1.1726,
      "step": 25006
    },
    {
      "epoch": 0.8955539241140974,
      "grad_norm": 1.568306803703308,
      "learning_rate": 5.665561602532165e-06,
      "loss": 1.6927,
      "step": 25007
    },
    {
      "epoch": 0.8955897362436658,
      "grad_norm": 1.9789416790008545,
      "learning_rate": 5.661713507509126e-06,
      "loss": 1.327,
      "step": 25008
    },
    {
      "epoch": 0.895625548373234,
      "grad_norm": 1.8262815475463867,
      "learning_rate": 5.657866681686053e-06,
      "loss": 1.6479,
      "step": 25009
    },
    {
      "epoch": 0.8956613605028023,
      "grad_norm": 1.8193621635437012,
      "learning_rate": 5.654021125114672e-06,
      "loss": 1.9024,
      "step": 25010
    },
    {
      "epoch": 0.8956971726323706,
      "grad_norm": 2.2621870040893555,
      "learning_rate": 5.6501768378467546e-06,
      "loss": 1.37,
      "step": 25011
    },
    {
      "epoch": 0.8957329847619389,
      "grad_norm": 2.0445046424865723,
      "learning_rate": 5.646333819933991e-06,
      "loss": 1.2324,
      "step": 25012
    },
    {
      "epoch": 0.8957687968915071,
      "grad_norm": 2.021339178085327,
      "learning_rate": 5.642492071428118e-06,
      "loss": 1.6428,
      "step": 25013
    },
    {
      "epoch": 0.8958046090210754,
      "grad_norm": 2.2916150093078613,
      "learning_rate": 5.638651592380795e-06,
      "loss": 1.6459,
      "step": 25014
    },
    {
      "epoch": 0.8958404211506438,
      "grad_norm": 1.9969441890716553,
      "learning_rate": 5.63481238284368e-06,
      "loss": 1.1157,
      "step": 25015
    },
    {
      "epoch": 0.895876233280212,
      "grad_norm": 1.4788435697555542,
      "learning_rate": 5.630974442868475e-06,
      "loss": 1.5839,
      "step": 25016
    },
    {
      "epoch": 0.8959120454097803,
      "grad_norm": 1.4995172023773193,
      "learning_rate": 5.627137772506752e-06,
      "loss": 1.339,
      "step": 25017
    },
    {
      "epoch": 0.8959478575393486,
      "grad_norm": 1.5286364555358887,
      "learning_rate": 5.623302371810169e-06,
      "loss": 1.2414,
      "step": 25018
    },
    {
      "epoch": 0.8959836696689168,
      "grad_norm": 1.7818827629089355,
      "learning_rate": 5.619468240830306e-06,
      "loss": 1.4386,
      "step": 25019
    },
    {
      "epoch": 0.8960194817984851,
      "grad_norm": 2.339745044708252,
      "learning_rate": 5.615635379618778e-06,
      "loss": 1.3534,
      "step": 25020
    },
    {
      "epoch": 0.8960552939280534,
      "grad_norm": 1.4172555208206177,
      "learning_rate": 5.61180378822711e-06,
      "loss": 1.5372,
      "step": 25021
    },
    {
      "epoch": 0.8960911060576218,
      "grad_norm": 1.7020219564437866,
      "learning_rate": 5.607973466706873e-06,
      "loss": 1.2331,
      "step": 25022
    },
    {
      "epoch": 0.89612691818719,
      "grad_norm": 1.4076426029205322,
      "learning_rate": 5.604144415109614e-06,
      "loss": 1.187,
      "step": 25023
    },
    {
      "epoch": 0.8961627303167583,
      "grad_norm": 1.5349396467208862,
      "learning_rate": 5.600316633486802e-06,
      "loss": 1.3395,
      "step": 25024
    },
    {
      "epoch": 0.8961985424463266,
      "grad_norm": 2.8167307376861572,
      "learning_rate": 5.596490121889975e-06,
      "loss": 1.4081,
      "step": 25025
    },
    {
      "epoch": 0.8962343545758948,
      "grad_norm": 1.624639630317688,
      "learning_rate": 5.592664880370602e-06,
      "loss": 1.3656,
      "step": 25026
    },
    {
      "epoch": 0.8962701667054631,
      "grad_norm": 2.108041524887085,
      "learning_rate": 5.588840908980153e-06,
      "loss": 1.2949,
      "step": 25027
    },
    {
      "epoch": 0.8963059788350314,
      "grad_norm": 1.5697460174560547,
      "learning_rate": 5.585018207770054e-06,
      "loss": 1.7335,
      "step": 25028
    },
    {
      "epoch": 0.8963417909645998,
      "grad_norm": 1.555113434791565,
      "learning_rate": 5.581196776791752e-06,
      "loss": 1.3761,
      "step": 25029
    },
    {
      "epoch": 0.896377603094168,
      "grad_norm": 1.8838539123535156,
      "learning_rate": 5.5773766160966634e-06,
      "loss": 1.6462,
      "step": 25030
    },
    {
      "epoch": 0.8964134152237363,
      "grad_norm": 1.76872718334198,
      "learning_rate": 5.5735577257361785e-06,
      "loss": 1.1874,
      "step": 25031
    },
    {
      "epoch": 0.8964492273533046,
      "grad_norm": 1.5243675708770752,
      "learning_rate": 5.569740105761679e-06,
      "loss": 1.387,
      "step": 25032
    },
    {
      "epoch": 0.8964850394828728,
      "grad_norm": 1.6961188316345215,
      "learning_rate": 5.565923756224489e-06,
      "loss": 1.4076,
      "step": 25033
    },
    {
      "epoch": 0.8965208516124411,
      "grad_norm": 2.217806816101074,
      "learning_rate": 5.562108677176015e-06,
      "loss": 1.3709,
      "step": 25034
    },
    {
      "epoch": 0.8965566637420094,
      "grad_norm": 1.590507984161377,
      "learning_rate": 5.558294868667535e-06,
      "loss": 1.444,
      "step": 25035
    },
    {
      "epoch": 0.8965924758715778,
      "grad_norm": 1.7941250801086426,
      "learning_rate": 5.554482330750388e-06,
      "loss": 1.5058,
      "step": 25036
    },
    {
      "epoch": 0.896628288001146,
      "grad_norm": 1.739349365234375,
      "learning_rate": 5.550671063475832e-06,
      "loss": 1.4063,
      "step": 25037
    },
    {
      "epoch": 0.8966641001307143,
      "grad_norm": 1.7744945287704468,
      "learning_rate": 5.546861066895193e-06,
      "loss": 1.6937,
      "step": 25038
    },
    {
      "epoch": 0.8966999122602826,
      "grad_norm": 1.3879646062850952,
      "learning_rate": 5.543052341059707e-06,
      "loss": 1.3204,
      "step": 25039
    },
    {
      "epoch": 0.8967357243898508,
      "grad_norm": 1.5593581199645996,
      "learning_rate": 5.5392448860205785e-06,
      "loss": 1.4807,
      "step": 25040
    },
    {
      "epoch": 0.8967715365194191,
      "grad_norm": 2.365830659866333,
      "learning_rate": 5.535438701829088e-06,
      "loss": 1.6088,
      "step": 25041
    },
    {
      "epoch": 0.8968073486489874,
      "grad_norm": 2.006086587905884,
      "learning_rate": 5.5316337885364165e-06,
      "loss": 1.7096,
      "step": 25042
    },
    {
      "epoch": 0.8968431607785557,
      "grad_norm": 2.7186625003814697,
      "learning_rate": 5.527830146193758e-06,
      "loss": 1.707,
      "step": 25043
    },
    {
      "epoch": 0.896878972908124,
      "grad_norm": 1.4928075075149536,
      "learning_rate": 5.5240277748522694e-06,
      "loss": 1.0195,
      "step": 25044
    },
    {
      "epoch": 0.8969147850376923,
      "grad_norm": 2.0089306831359863,
      "learning_rate": 5.520226674563145e-06,
      "loss": 1.4221,
      "step": 25045
    },
    {
      "epoch": 0.8969505971672606,
      "grad_norm": 1.3066225051879883,
      "learning_rate": 5.516426845377476e-06,
      "loss": 1.3159,
      "step": 25046
    },
    {
      "epoch": 0.8969864092968288,
      "grad_norm": 1.5065250396728516,
      "learning_rate": 5.512628287346433e-06,
      "loss": 1.3258,
      "step": 25047
    },
    {
      "epoch": 0.8970222214263971,
      "grad_norm": 1.5593777894973755,
      "learning_rate": 5.5088310005210865e-06,
      "loss": 1.1823,
      "step": 25048
    },
    {
      "epoch": 0.8970580335559654,
      "grad_norm": 1.4893614053726196,
      "learning_rate": 5.505034984952529e-06,
      "loss": 1.4447,
      "step": 25049
    },
    {
      "epoch": 0.8970938456855337,
      "grad_norm": 1.4334756135940552,
      "learning_rate": 5.501240240691852e-06,
      "loss": 1.6804,
      "step": 25050
    },
    {
      "epoch": 0.897129657815102,
      "grad_norm": 1.6654868125915527,
      "learning_rate": 5.49744676779006e-06,
      "loss": 1.3822,
      "step": 25051
    },
    {
      "epoch": 0.8971654699446703,
      "grad_norm": 1.5863392353057861,
      "learning_rate": 5.4936545662982455e-06,
      "loss": 1.5038,
      "step": 25052
    },
    {
      "epoch": 0.8972012820742385,
      "grad_norm": 1.6635081768035889,
      "learning_rate": 5.4898636362674e-06,
      "loss": 1.2242,
      "step": 25053
    },
    {
      "epoch": 0.8972370942038068,
      "grad_norm": 1.979265570640564,
      "learning_rate": 5.486073977748541e-06,
      "loss": 1.1738,
      "step": 25054
    },
    {
      "epoch": 0.8972729063333751,
      "grad_norm": 1.9149836301803589,
      "learning_rate": 5.482285590792613e-06,
      "loss": 1.4048,
      "step": 25055
    },
    {
      "epoch": 0.8973087184629434,
      "grad_norm": 1.5873132944107056,
      "learning_rate": 5.478498475450644e-06,
      "loss": 1.4913,
      "step": 25056
    },
    {
      "epoch": 0.8973445305925117,
      "grad_norm": 1.8850102424621582,
      "learning_rate": 5.47471263177356e-06,
      "loss": 1.5126,
      "step": 25057
    },
    {
      "epoch": 0.89738034272208,
      "grad_norm": 1.7988669872283936,
      "learning_rate": 5.470928059812264e-06,
      "loss": 1.1521,
      "step": 25058
    },
    {
      "epoch": 0.8974161548516483,
      "grad_norm": 1.4897023439407349,
      "learning_rate": 5.467144759617704e-06,
      "loss": 1.529,
      "step": 25059
    },
    {
      "epoch": 0.8974519669812165,
      "grad_norm": 1.9250378608703613,
      "learning_rate": 5.463362731240773e-06,
      "loss": 1.5413,
      "step": 25060
    },
    {
      "epoch": 0.8974877791107848,
      "grad_norm": 1.302215576171875,
      "learning_rate": 5.4595819747323636e-06,
      "loss": 1.2577,
      "step": 25061
    },
    {
      "epoch": 0.8975235912403531,
      "grad_norm": 1.791074514389038,
      "learning_rate": 5.455802490143314e-06,
      "loss": 1.4449,
      "step": 25062
    },
    {
      "epoch": 0.8975594033699213,
      "grad_norm": 2.012291669845581,
      "learning_rate": 5.4520242775244925e-06,
      "loss": 1.218,
      "step": 25063
    },
    {
      "epoch": 0.8975952154994897,
      "grad_norm": 1.5739734172821045,
      "learning_rate": 5.4482473369267264e-06,
      "loss": 1.1135,
      "step": 25064
    },
    {
      "epoch": 0.897631027629058,
      "grad_norm": 1.5240012407302856,
      "learning_rate": 5.444471668400841e-06,
      "loss": 1.5621,
      "step": 25065
    },
    {
      "epoch": 0.8976668397586263,
      "grad_norm": 1.944368600845337,
      "learning_rate": 5.440697271997608e-06,
      "loss": 1.0039,
      "step": 25066
    },
    {
      "epoch": 0.8977026518881945,
      "grad_norm": 1.4575976133346558,
      "learning_rate": 5.436924147767819e-06,
      "loss": 1.547,
      "step": 25067
    },
    {
      "epoch": 0.8977384640177628,
      "grad_norm": 2.7136669158935547,
      "learning_rate": 5.433152295762256e-06,
      "loss": 1.402,
      "step": 25068
    },
    {
      "epoch": 0.8977742761473311,
      "grad_norm": 2.263066291809082,
      "learning_rate": 5.429381716031634e-06,
      "loss": 1.2947,
      "step": 25069
    },
    {
      "epoch": 0.8978100882768993,
      "grad_norm": 1.668306589126587,
      "learning_rate": 5.42561240862669e-06,
      "loss": 1.6464,
      "step": 25070
    },
    {
      "epoch": 0.8978459004064677,
      "grad_norm": 1.28193998336792,
      "learning_rate": 5.421844373598139e-06,
      "loss": 1.2046,
      "step": 25071
    },
    {
      "epoch": 0.897881712536036,
      "grad_norm": 1.325032353401184,
      "learning_rate": 5.418077610996686e-06,
      "loss": 1.4681,
      "step": 25072
    },
    {
      "epoch": 0.8979175246656043,
      "grad_norm": 1.4316115379333496,
      "learning_rate": 5.4143121208729885e-06,
      "loss": 1.579,
      "step": 25073
    },
    {
      "epoch": 0.8979533367951725,
      "grad_norm": 1.4661067724227905,
      "learning_rate": 5.410547903277707e-06,
      "loss": 1.2959,
      "step": 25074
    },
    {
      "epoch": 0.8979891489247408,
      "grad_norm": 1.6015124320983887,
      "learning_rate": 5.4067849582615124e-06,
      "loss": 1.5803,
      "step": 25075
    },
    {
      "epoch": 0.8980249610543091,
      "grad_norm": 1.9564155340194702,
      "learning_rate": 5.403023285874997e-06,
      "loss": 1.4142,
      "step": 25076
    },
    {
      "epoch": 0.8980607731838773,
      "grad_norm": 1.7000601291656494,
      "learning_rate": 5.399262886168777e-06,
      "loss": 1.4036,
      "step": 25077
    },
    {
      "epoch": 0.8980965853134457,
      "grad_norm": 2.1915886402130127,
      "learning_rate": 5.395503759193454e-06,
      "loss": 1.7342,
      "step": 25078
    },
    {
      "epoch": 0.898132397443014,
      "grad_norm": 1.5303142070770264,
      "learning_rate": 5.391745904999601e-06,
      "loss": 1.4086,
      "step": 25079
    },
    {
      "epoch": 0.8981682095725823,
      "grad_norm": 1.7111073732376099,
      "learning_rate": 5.387989323637765e-06,
      "loss": 1.5384,
      "step": 25080
    },
    {
      "epoch": 0.8982040217021505,
      "grad_norm": 1.3334734439849854,
      "learning_rate": 5.384234015158495e-06,
      "loss": 1.4625,
      "step": 25081
    },
    {
      "epoch": 0.8982398338317188,
      "grad_norm": 2.577977418899536,
      "learning_rate": 5.380479979612307e-06,
      "loss": 1.4246,
      "step": 25082
    },
    {
      "epoch": 0.8982756459612871,
      "grad_norm": 1.6402453184127808,
      "learning_rate": 5.376727217049726e-06,
      "loss": 1.4184,
      "step": 25083
    },
    {
      "epoch": 0.8983114580908553,
      "grad_norm": 1.353309154510498,
      "learning_rate": 5.372975727521201e-06,
      "loss": 1.2795,
      "step": 25084
    },
    {
      "epoch": 0.8983472702204237,
      "grad_norm": 1.778639793395996,
      "learning_rate": 5.369225511077236e-06,
      "loss": 1.47,
      "step": 25085
    },
    {
      "epoch": 0.898383082349992,
      "grad_norm": 1.7196940183639526,
      "learning_rate": 5.36547656776829e-06,
      "loss": 1.4584,
      "step": 25086
    },
    {
      "epoch": 0.8984188944795602,
      "grad_norm": 1.3862847089767456,
      "learning_rate": 5.36172889764478e-06,
      "loss": 1.7245,
      "step": 25087
    },
    {
      "epoch": 0.8984547066091285,
      "grad_norm": 1.9796829223632812,
      "learning_rate": 5.357982500757119e-06,
      "loss": 1.4531,
      "step": 25088
    },
    {
      "epoch": 0.8984905187386968,
      "grad_norm": 1.9251147508621216,
      "learning_rate": 5.354237377155735e-06,
      "loss": 1.4222,
      "step": 25089
    },
    {
      "epoch": 0.898526330868265,
      "grad_norm": 1.603308916091919,
      "learning_rate": 5.3504935268910095e-06,
      "loss": 1.2329,
      "step": 25090
    },
    {
      "epoch": 0.8985621429978333,
      "grad_norm": 1.7304174900054932,
      "learning_rate": 5.346750950013301e-06,
      "loss": 1.6948,
      "step": 25091
    },
    {
      "epoch": 0.8985979551274017,
      "grad_norm": 1.523800253868103,
      "learning_rate": 5.343009646572949e-06,
      "loss": 1.5058,
      "step": 25092
    },
    {
      "epoch": 0.89863376725697,
      "grad_norm": 1.9301091432571411,
      "learning_rate": 5.3392696166203345e-06,
      "loss": 1.6271,
      "step": 25093
    },
    {
      "epoch": 0.8986695793865382,
      "grad_norm": 1.908660650253296,
      "learning_rate": 5.335530860205718e-06,
      "loss": 1.2108,
      "step": 25094
    },
    {
      "epoch": 0.8987053915161065,
      "grad_norm": 1.437114953994751,
      "learning_rate": 5.331793377379435e-06,
      "loss": 1.5768,
      "step": 25095
    },
    {
      "epoch": 0.8987412036456748,
      "grad_norm": 1.4844969511032104,
      "learning_rate": 5.328057168191747e-06,
      "loss": 1.2316,
      "step": 25096
    },
    {
      "epoch": 0.898777015775243,
      "grad_norm": 1.9871184825897217,
      "learning_rate": 5.324322232692947e-06,
      "loss": 1.3748,
      "step": 25097
    },
    {
      "epoch": 0.8988128279048113,
      "grad_norm": 1.5690678358078003,
      "learning_rate": 5.32058857093326e-06,
      "loss": 1.5002,
      "step": 25098
    },
    {
      "epoch": 0.8988486400343797,
      "grad_norm": 2.270082712173462,
      "learning_rate": 5.316856182962926e-06,
      "loss": 1.6875,
      "step": 25099
    },
    {
      "epoch": 0.898884452163948,
      "grad_norm": 1.716614007949829,
      "learning_rate": 5.313125068832159e-06,
      "loss": 1.4479,
      "step": 25100
    },
    {
      "epoch": 0.8989202642935162,
      "grad_norm": 2.0033087730407715,
      "learning_rate": 5.309395228591174e-06,
      "loss": 1.2221,
      "step": 25101
    },
    {
      "epoch": 0.8989560764230845,
      "grad_norm": 1.4407968521118164,
      "learning_rate": 5.305666662290121e-06,
      "loss": 1.3376,
      "step": 25102
    },
    {
      "epoch": 0.8989918885526528,
      "grad_norm": 1.4618561267852783,
      "learning_rate": 5.30193936997917e-06,
      "loss": 1.6145,
      "step": 25103
    },
    {
      "epoch": 0.899027700682221,
      "grad_norm": 2.3181114196777344,
      "learning_rate": 5.298213351708492e-06,
      "loss": 1.4054,
      "step": 25104
    },
    {
      "epoch": 0.8990635128117893,
      "grad_norm": 1.5716500282287598,
      "learning_rate": 5.29448860752817e-06,
      "loss": 1.3271,
      "step": 25105
    },
    {
      "epoch": 0.8990993249413577,
      "grad_norm": 1.612802505493164,
      "learning_rate": 5.290765137488351e-06,
      "loss": 1.4278,
      "step": 25106
    },
    {
      "epoch": 0.899135137070926,
      "grad_norm": 1.4087631702423096,
      "learning_rate": 5.287042941639131e-06,
      "loss": 1.6085,
      "step": 25107
    },
    {
      "epoch": 0.8991709492004942,
      "grad_norm": 2.212252616882324,
      "learning_rate": 5.2833220200305785e-06,
      "loss": 1.8866,
      "step": 25108
    },
    {
      "epoch": 0.8992067613300625,
      "grad_norm": 1.4179940223693848,
      "learning_rate": 5.279602372712744e-06,
      "loss": 1.5334,
      "step": 25109
    },
    {
      "epoch": 0.8992425734596308,
      "grad_norm": 1.4929925203323364,
      "learning_rate": 5.275883999735676e-06,
      "loss": 1.3277,
      "step": 25110
    },
    {
      "epoch": 0.899278385589199,
      "grad_norm": 2.1377246379852295,
      "learning_rate": 5.272166901149423e-06,
      "loss": 1.1745,
      "step": 25111
    },
    {
      "epoch": 0.8993141977187673,
      "grad_norm": 1.96725332736969,
      "learning_rate": 5.2684510770039556e-06,
      "loss": 1.696,
      "step": 25112
    },
    {
      "epoch": 0.8993500098483357,
      "grad_norm": 2.2955400943756104,
      "learning_rate": 5.264736527349279e-06,
      "loss": 1.2985,
      "step": 25113
    },
    {
      "epoch": 0.899385821977904,
      "grad_norm": 1.76557195186615,
      "learning_rate": 5.261023252235386e-06,
      "loss": 1.5453,
      "step": 25114
    },
    {
      "epoch": 0.8994216341074722,
      "grad_norm": 1.7912245988845825,
      "learning_rate": 5.257311251712227e-06,
      "loss": 1.6088,
      "step": 25115
    },
    {
      "epoch": 0.8994574462370405,
      "grad_norm": 1.81741201877594,
      "learning_rate": 5.253600525829716e-06,
      "loss": 1.8192,
      "step": 25116
    },
    {
      "epoch": 0.8994932583666088,
      "grad_norm": 1.5481551885604858,
      "learning_rate": 5.249891074637803e-06,
      "loss": 1.2055,
      "step": 25117
    },
    {
      "epoch": 0.899529070496177,
      "grad_norm": 2.0782535076141357,
      "learning_rate": 5.2461828981863916e-06,
      "loss": 1.502,
      "step": 25118
    },
    {
      "epoch": 0.8995648826257453,
      "grad_norm": 1.717686653137207,
      "learning_rate": 5.2424759965253645e-06,
      "loss": 1.6172,
      "step": 25119
    },
    {
      "epoch": 0.8996006947553137,
      "grad_norm": 2.288208246231079,
      "learning_rate": 5.2387703697046045e-06,
      "loss": 1.2769,
      "step": 25120
    },
    {
      "epoch": 0.899636506884882,
      "grad_norm": 1.6821223497390747,
      "learning_rate": 5.235066017773926e-06,
      "loss": 1.1108,
      "step": 25121
    },
    {
      "epoch": 0.8996723190144502,
      "grad_norm": 1.282433271408081,
      "learning_rate": 5.2313629407832355e-06,
      "loss": 1.3598,
      "step": 25122
    },
    {
      "epoch": 0.8997081311440185,
      "grad_norm": 1.8116391897201538,
      "learning_rate": 5.227661138782281e-06,
      "loss": 1.7734,
      "step": 25123
    },
    {
      "epoch": 0.8997439432735868,
      "grad_norm": 1.6620230674743652,
      "learning_rate": 5.22396061182091e-06,
      "loss": 1.3506,
      "step": 25124
    },
    {
      "epoch": 0.899779755403155,
      "grad_norm": 1.840793251991272,
      "learning_rate": 5.220261359948897e-06,
      "loss": 1.705,
      "step": 25125
    },
    {
      "epoch": 0.8998155675327233,
      "grad_norm": 1.7171930074691772,
      "learning_rate": 5.216563383216022e-06,
      "loss": 1.5466,
      "step": 25126
    },
    {
      "epoch": 0.8998513796622917,
      "grad_norm": 1.6956909894943237,
      "learning_rate": 5.2128666816720015e-06,
      "loss": 1.7422,
      "step": 25127
    },
    {
      "epoch": 0.8998871917918599,
      "grad_norm": 2.0536861419677734,
      "learning_rate": 5.209171255366607e-06,
      "loss": 1.3849,
      "step": 25128
    },
    {
      "epoch": 0.8999230039214282,
      "grad_norm": 1.4225544929504395,
      "learning_rate": 5.205477104349554e-06,
      "loss": 1.457,
      "step": 25129
    },
    {
      "epoch": 0.8999588160509965,
      "grad_norm": 1.5567337274551392,
      "learning_rate": 5.2017842286705145e-06,
      "loss": 1.4507,
      "step": 25130
    },
    {
      "epoch": 0.8999946281805647,
      "grad_norm": 1.5060020685195923,
      "learning_rate": 5.198092628379192e-06,
      "loss": 1.6578,
      "step": 25131
    },
    {
      "epoch": 0.900030440310133,
      "grad_norm": 2.0625991821289062,
      "learning_rate": 5.194402303525225e-06,
      "loss": 1.5786,
      "step": 25132
    },
    {
      "epoch": 0.9000662524397013,
      "grad_norm": 1.730444073677063,
      "learning_rate": 5.190713254158319e-06,
      "loss": 1.2594,
      "step": 25133
    },
    {
      "epoch": 0.9001020645692697,
      "grad_norm": 1.809962272644043,
      "learning_rate": 5.187025480328056e-06,
      "loss": 1.4024,
      "step": 25134
    },
    {
      "epoch": 0.9001378766988379,
      "grad_norm": 1.6739686727523804,
      "learning_rate": 5.183338982084074e-06,
      "loss": 1.762,
      "step": 25135
    },
    {
      "epoch": 0.9001736888284062,
      "grad_norm": 1.4611245393753052,
      "learning_rate": 5.179653759475933e-06,
      "loss": 1.3352,
      "step": 25136
    },
    {
      "epoch": 0.9002095009579745,
      "grad_norm": 2.249504804611206,
      "learning_rate": 5.175969812553272e-06,
      "loss": 1.1882,
      "step": 25137
    },
    {
      "epoch": 0.9002453130875427,
      "grad_norm": 1.9460619688034058,
      "learning_rate": 5.172287141365628e-06,
      "loss": 1.3776,
      "step": 25138
    },
    {
      "epoch": 0.900281125217111,
      "grad_norm": 1.6175462007522583,
      "learning_rate": 5.168605745962507e-06,
      "loss": 1.5004,
      "step": 25139
    },
    {
      "epoch": 0.9003169373466793,
      "grad_norm": 1.803956151008606,
      "learning_rate": 5.164925626393502e-06,
      "loss": 1.5692,
      "step": 25140
    },
    {
      "epoch": 0.9003527494762477,
      "grad_norm": 1.1598551273345947,
      "learning_rate": 5.161246782708073e-06,
      "loss": 1.3922,
      "step": 25141
    },
    {
      "epoch": 0.9003885616058159,
      "grad_norm": 1.4358941316604614,
      "learning_rate": 5.15756921495576e-06,
      "loss": 1.4219,
      "step": 25142
    },
    {
      "epoch": 0.9004243737353842,
      "grad_norm": 1.7593889236450195,
      "learning_rate": 5.153892923185977e-06,
      "loss": 1.4586,
      "step": 25143
    },
    {
      "epoch": 0.9004601858649525,
      "grad_norm": 1.4885627031326294,
      "learning_rate": 5.150217907448263e-06,
      "loss": 1.5213,
      "step": 25144
    },
    {
      "epoch": 0.9004959979945207,
      "grad_norm": 1.5121746063232422,
      "learning_rate": 5.146544167792011e-06,
      "loss": 1.3499,
      "step": 25145
    },
    {
      "epoch": 0.900531810124089,
      "grad_norm": 1.7629821300506592,
      "learning_rate": 5.1428717042666385e-06,
      "loss": 1.2924,
      "step": 25146
    },
    {
      "epoch": 0.9005676222536573,
      "grad_norm": 1.4092707633972168,
      "learning_rate": 5.1392005169215825e-06,
      "loss": 1.3375,
      "step": 25147
    },
    {
      "epoch": 0.9006034343832257,
      "grad_norm": 1.5190919637680054,
      "learning_rate": 5.1355306058062044e-06,
      "loss": 1.626,
      "step": 25148
    },
    {
      "epoch": 0.9006392465127939,
      "grad_norm": 1.983694314956665,
      "learning_rate": 5.13186197096992e-06,
      "loss": 1.2005,
      "step": 25149
    },
    {
      "epoch": 0.9006750586423622,
      "grad_norm": 1.4387457370758057,
      "learning_rate": 5.128194612462034e-06,
      "loss": 1.4169,
      "step": 25150
    },
    {
      "epoch": 0.9007108707719305,
      "grad_norm": 2.042126178741455,
      "learning_rate": 5.12452853033194e-06,
      "loss": 1.8489,
      "step": 25151
    },
    {
      "epoch": 0.9007466829014987,
      "grad_norm": 1.567436933517456,
      "learning_rate": 5.120863724628922e-06,
      "loss": 1.6615,
      "step": 25152
    },
    {
      "epoch": 0.900782495031067,
      "grad_norm": 1.5410945415496826,
      "learning_rate": 5.117200195402316e-06,
      "loss": 1.1489,
      "step": 25153
    },
    {
      "epoch": 0.9008183071606353,
      "grad_norm": 1.5345313549041748,
      "learning_rate": 5.113537942701363e-06,
      "loss": 1.6927,
      "step": 25154
    },
    {
      "epoch": 0.9008541192902035,
      "grad_norm": 1.8208950757980347,
      "learning_rate": 5.109876966575377e-06,
      "loss": 1.3238,
      "step": 25155
    },
    {
      "epoch": 0.9008899314197719,
      "grad_norm": 1.6119062900543213,
      "learning_rate": 5.106217267073598e-06,
      "loss": 1.6702,
      "step": 25156
    },
    {
      "epoch": 0.9009257435493402,
      "grad_norm": 1.6705838441848755,
      "learning_rate": 5.102558844245265e-06,
      "loss": 1.004,
      "step": 25157
    },
    {
      "epoch": 0.9009615556789085,
      "grad_norm": 1.38338041305542,
      "learning_rate": 5.09890169813958e-06,
      "loss": 1.4428,
      "step": 25158
    },
    {
      "epoch": 0.9009973678084767,
      "grad_norm": 1.6136221885681152,
      "learning_rate": 5.095245828805761e-06,
      "loss": 1.3344,
      "step": 25159
    },
    {
      "epoch": 0.901033179938045,
      "grad_norm": 1.925017237663269,
      "learning_rate": 5.091591236293003e-06,
      "loss": 1.5366,
      "step": 25160
    },
    {
      "epoch": 0.9010689920676133,
      "grad_norm": 1.5999585390090942,
      "learning_rate": 5.087937920650454e-06,
      "loss": 1.3421,
      "step": 25161
    },
    {
      "epoch": 0.9011048041971815,
      "grad_norm": 1.429749608039856,
      "learning_rate": 5.0842858819272644e-06,
      "loss": 1.6059,
      "step": 25162
    },
    {
      "epoch": 0.9011406163267499,
      "grad_norm": 1.6445558071136475,
      "learning_rate": 5.0806351201725944e-06,
      "loss": 1.3776,
      "step": 25163
    },
    {
      "epoch": 0.9011764284563182,
      "grad_norm": 1.7813189029693604,
      "learning_rate": 5.076985635435527e-06,
      "loss": 1.43,
      "step": 25164
    },
    {
      "epoch": 0.9012122405858864,
      "grad_norm": 1.552821159362793,
      "learning_rate": 5.073337427765179e-06,
      "loss": 1.5426,
      "step": 25165
    },
    {
      "epoch": 0.9012480527154547,
      "grad_norm": 1.7873458862304688,
      "learning_rate": 5.069690497210633e-06,
      "loss": 1.3311,
      "step": 25166
    },
    {
      "epoch": 0.901283864845023,
      "grad_norm": 1.6734850406646729,
      "learning_rate": 5.06604484382095e-06,
      "loss": 1.394,
      "step": 25167
    },
    {
      "epoch": 0.9013196769745913,
      "grad_norm": 1.4095349311828613,
      "learning_rate": 5.062400467645178e-06,
      "loss": 1.2577,
      "step": 25168
    },
    {
      "epoch": 0.9013554891041595,
      "grad_norm": 1.764383316040039,
      "learning_rate": 5.058757368732336e-06,
      "loss": 1.3601,
      "step": 25169
    },
    {
      "epoch": 0.9013913012337279,
      "grad_norm": 1.9552977085113525,
      "learning_rate": 5.055115547131462e-06,
      "loss": 1.6717,
      "step": 25170
    },
    {
      "epoch": 0.9014271133632962,
      "grad_norm": 1.2154018878936768,
      "learning_rate": 5.051475002891537e-06,
      "loss": 1.1555,
      "step": 25171
    },
    {
      "epoch": 0.9014629254928644,
      "grad_norm": 2.186868906021118,
      "learning_rate": 5.047835736061535e-06,
      "loss": 1.2166,
      "step": 25172
    },
    {
      "epoch": 0.9014987376224327,
      "grad_norm": 1.7284328937530518,
      "learning_rate": 5.044197746690427e-06,
      "loss": 1.236,
      "step": 25173
    },
    {
      "epoch": 0.901534549752001,
      "grad_norm": 1.353801965713501,
      "learning_rate": 5.040561034827163e-06,
      "loss": 1.3683,
      "step": 25174
    },
    {
      "epoch": 0.9015703618815692,
      "grad_norm": 1.7040997743606567,
      "learning_rate": 5.036925600520648e-06,
      "loss": 1.4311,
      "step": 25175
    },
    {
      "epoch": 0.9016061740111375,
      "grad_norm": 1.961835503578186,
      "learning_rate": 5.0332914438197984e-06,
      "loss": 1.1263,
      "step": 25176
    },
    {
      "epoch": 0.9016419861407059,
      "grad_norm": 1.8802769184112549,
      "learning_rate": 5.029658564773521e-06,
      "loss": 1.5408,
      "step": 25177
    },
    {
      "epoch": 0.9016777982702742,
      "grad_norm": 1.668648600578308,
      "learning_rate": 5.026026963430697e-06,
      "loss": 1.2913,
      "step": 25178
    },
    {
      "epoch": 0.9017136103998424,
      "grad_norm": 1.8066505193710327,
      "learning_rate": 5.022396639840166e-06,
      "loss": 1.5094,
      "step": 25179
    },
    {
      "epoch": 0.9017494225294107,
      "grad_norm": 1.4674502611160278,
      "learning_rate": 5.018767594050766e-06,
      "loss": 1.4396,
      "step": 25180
    },
    {
      "epoch": 0.901785234658979,
      "grad_norm": 1.7392138242721558,
      "learning_rate": 5.015139826111348e-06,
      "loss": 1.8019,
      "step": 25181
    },
    {
      "epoch": 0.9018210467885472,
      "grad_norm": 1.5244306325912476,
      "learning_rate": 5.0115133360706945e-06,
      "loss": 1.6719,
      "step": 25182
    },
    {
      "epoch": 0.9018568589181155,
      "grad_norm": 1.4753122329711914,
      "learning_rate": 5.0078881239776e-06,
      "loss": 1.09,
      "step": 25183
    },
    {
      "epoch": 0.9018926710476839,
      "grad_norm": 1.8521908521652222,
      "learning_rate": 5.0042641898808364e-06,
      "loss": 1.6567,
      "step": 25184
    },
    {
      "epoch": 0.9019284831772522,
      "grad_norm": 1.287305235862732,
      "learning_rate": 5.000641533829176e-06,
      "loss": 1.3805,
      "step": 25185
    },
    {
      "epoch": 0.9019642953068204,
      "grad_norm": 1.5233205556869507,
      "learning_rate": 4.9970201558713345e-06,
      "loss": 1.5226,
      "step": 25186
    },
    {
      "epoch": 0.9020001074363887,
      "grad_norm": 1.4274674654006958,
      "learning_rate": 4.99340005605603e-06,
      "loss": 1.4153,
      "step": 25187
    },
    {
      "epoch": 0.902035919565957,
      "grad_norm": 1.5530496835708618,
      "learning_rate": 4.98978123443199e-06,
      "loss": 1.6574,
      "step": 25188
    },
    {
      "epoch": 0.9020717316955252,
      "grad_norm": 1.6646677255630493,
      "learning_rate": 4.986163691047896e-06,
      "loss": 1.177,
      "step": 25189
    },
    {
      "epoch": 0.9021075438250935,
      "grad_norm": 1.5656930208206177,
      "learning_rate": 4.982547425952399e-06,
      "loss": 1.5505,
      "step": 25190
    },
    {
      "epoch": 0.9021433559546619,
      "grad_norm": 1.369274377822876,
      "learning_rate": 4.9789324391941615e-06,
      "loss": 1.3201,
      "step": 25191
    },
    {
      "epoch": 0.9021791680842302,
      "grad_norm": 1.6000550985336304,
      "learning_rate": 4.975318730821843e-06,
      "loss": 1.2981,
      "step": 25192
    },
    {
      "epoch": 0.9022149802137984,
      "grad_norm": 1.3364497423171997,
      "learning_rate": 4.971706300884016e-06,
      "loss": 1.3928,
      "step": 25193
    },
    {
      "epoch": 0.9022507923433667,
      "grad_norm": 1.555607557296753,
      "learning_rate": 4.9680951494292975e-06,
      "loss": 1.4731,
      "step": 25194
    },
    {
      "epoch": 0.902286604472935,
      "grad_norm": 1.5311139822006226,
      "learning_rate": 4.964485276506281e-06,
      "loss": 1.416,
      "step": 25195
    },
    {
      "epoch": 0.9023224166025032,
      "grad_norm": 2.2002294063568115,
      "learning_rate": 4.960876682163551e-06,
      "loss": 1.5436,
      "step": 25196
    },
    {
      "epoch": 0.9023582287320715,
      "grad_norm": 1.4345999956130981,
      "learning_rate": 4.957269366449613e-06,
      "loss": 1.2206,
      "step": 25197
    },
    {
      "epoch": 0.9023940408616399,
      "grad_norm": 1.821091651916504,
      "learning_rate": 4.953663329413017e-06,
      "loss": 1.3636,
      "step": 25198
    },
    {
      "epoch": 0.9024298529912081,
      "grad_norm": 2.0151352882385254,
      "learning_rate": 4.950058571102289e-06,
      "loss": 1.5156,
      "step": 25199
    },
    {
      "epoch": 0.9024656651207764,
      "grad_norm": 2.489975929260254,
      "learning_rate": 4.946455091565916e-06,
      "loss": 1.3341,
      "step": 25200
    },
    {
      "epoch": 0.9025014772503447,
      "grad_norm": 1.6598349809646606,
      "learning_rate": 4.942852890852367e-06,
      "loss": 1.5651,
      "step": 25201
    },
    {
      "epoch": 0.902537289379913,
      "grad_norm": 1.6895561218261719,
      "learning_rate": 4.939251969010128e-06,
      "loss": 1.4153,
      "step": 25202
    },
    {
      "epoch": 0.9025731015094812,
      "grad_norm": 2.297715902328491,
      "learning_rate": 4.935652326087648e-06,
      "loss": 1.6323,
      "step": 25203
    },
    {
      "epoch": 0.9026089136390495,
      "grad_norm": 1.411095142364502,
      "learning_rate": 4.932053962133321e-06,
      "loss": 1.3637,
      "step": 25204
    },
    {
      "epoch": 0.9026447257686179,
      "grad_norm": 1.6230731010437012,
      "learning_rate": 4.928456877195586e-06,
      "loss": 1.3658,
      "step": 25205
    },
    {
      "epoch": 0.9026805378981861,
      "grad_norm": 2.2268002033233643,
      "learning_rate": 4.924861071322817e-06,
      "loss": 1.602,
      "step": 25206
    },
    {
      "epoch": 0.9027163500277544,
      "grad_norm": 1.9682823419570923,
      "learning_rate": 4.92126654456343e-06,
      "loss": 1.4974,
      "step": 25207
    },
    {
      "epoch": 0.9027521621573227,
      "grad_norm": 1.9790271520614624,
      "learning_rate": 4.917673296965741e-06,
      "loss": 1.635,
      "step": 25208
    },
    {
      "epoch": 0.902787974286891,
      "grad_norm": 1.872644066810608,
      "learning_rate": 4.914081328578113e-06,
      "loss": 1.2021,
      "step": 25209
    },
    {
      "epoch": 0.9028237864164592,
      "grad_norm": 2.065763235092163,
      "learning_rate": 4.910490639448884e-06,
      "loss": 1.497,
      "step": 25210
    },
    {
      "epoch": 0.9028595985460275,
      "grad_norm": 1.9289997816085815,
      "learning_rate": 4.906901229626326e-06,
      "loss": 1.7655,
      "step": 25211
    },
    {
      "epoch": 0.9028954106755959,
      "grad_norm": 1.6984025239944458,
      "learning_rate": 4.903313099158757e-06,
      "loss": 1.4767,
      "step": 25212
    },
    {
      "epoch": 0.9029312228051641,
      "grad_norm": 1.4198287725448608,
      "learning_rate": 4.8997262480944385e-06,
      "loss": 1.3188,
      "step": 25213
    },
    {
      "epoch": 0.9029670349347324,
      "grad_norm": 1.8730920553207397,
      "learning_rate": 4.896140676481653e-06,
      "loss": 1.0944,
      "step": 25214
    },
    {
      "epoch": 0.9030028470643007,
      "grad_norm": 1.4049091339111328,
      "learning_rate": 4.892556384368607e-06,
      "loss": 1.4781,
      "step": 25215
    },
    {
      "epoch": 0.9030386591938689,
      "grad_norm": 1.9881552457809448,
      "learning_rate": 4.8889733718035295e-06,
      "loss": 1.507,
      "step": 25216
    },
    {
      "epoch": 0.9030744713234372,
      "grad_norm": 1.4669917821884155,
      "learning_rate": 4.885391638834646e-06,
      "loss": 1.2178,
      "step": 25217
    },
    {
      "epoch": 0.9031102834530055,
      "grad_norm": 2.0855116844177246,
      "learning_rate": 4.88181118551011e-06,
      "loss": 1.1911,
      "step": 25218
    },
    {
      "epoch": 0.9031460955825739,
      "grad_norm": 1.5934829711914062,
      "learning_rate": 4.878232011878136e-06,
      "loss": 1.4744,
      "step": 25219
    },
    {
      "epoch": 0.9031819077121421,
      "grad_norm": 2.0535361766815186,
      "learning_rate": 4.874654117986821e-06,
      "loss": 1.5789,
      "step": 25220
    },
    {
      "epoch": 0.9032177198417104,
      "grad_norm": 1.54163658618927,
      "learning_rate": 4.871077503884358e-06,
      "loss": 1.5965,
      "step": 25221
    },
    {
      "epoch": 0.9032535319712787,
      "grad_norm": 1.860435962677002,
      "learning_rate": 4.86750216961882e-06,
      "loss": 1.2296,
      "step": 25222
    },
    {
      "epoch": 0.9032893441008469,
      "grad_norm": 1.6049903631210327,
      "learning_rate": 4.863928115238336e-06,
      "loss": 1.3,
      "step": 25223
    },
    {
      "epoch": 0.9033251562304152,
      "grad_norm": 1.5540401935577393,
      "learning_rate": 4.860355340790978e-06,
      "loss": 1.128,
      "step": 25224
    },
    {
      "epoch": 0.9033609683599835,
      "grad_norm": 1.7751628160476685,
      "learning_rate": 4.85678384632483e-06,
      "loss": 1.462,
      "step": 25225
    },
    {
      "epoch": 0.9033967804895519,
      "grad_norm": 2.4625301361083984,
      "learning_rate": 4.8532136318879315e-06,
      "loss": 1.7404,
      "step": 25226
    },
    {
      "epoch": 0.9034325926191201,
      "grad_norm": 1.419631838798523,
      "learning_rate": 4.8496446975282885e-06,
      "loss": 1.5578,
      "step": 25227
    },
    {
      "epoch": 0.9034684047486884,
      "grad_norm": 1.7044492959976196,
      "learning_rate": 4.846077043293973e-06,
      "loss": 1.1315,
      "step": 25228
    },
    {
      "epoch": 0.9035042168782567,
      "grad_norm": 1.758029818534851,
      "learning_rate": 4.842510669232925e-06,
      "loss": 1.3711,
      "step": 25229
    },
    {
      "epoch": 0.9035400290078249,
      "grad_norm": 2.1265738010406494,
      "learning_rate": 4.8389455753931726e-06,
      "loss": 1.2978,
      "step": 25230
    },
    {
      "epoch": 0.9035758411373932,
      "grad_norm": 1.8212213516235352,
      "learning_rate": 4.835381761822633e-06,
      "loss": 1.5178,
      "step": 25231
    },
    {
      "epoch": 0.9036116532669615,
      "grad_norm": 1.3521546125411987,
      "learning_rate": 4.831819228569301e-06,
      "loss": 1.1085,
      "step": 25232
    },
    {
      "epoch": 0.9036474653965298,
      "grad_norm": 1.5324915647506714,
      "learning_rate": 4.828257975681072e-06,
      "loss": 1.7123,
      "step": 25233
    },
    {
      "epoch": 0.9036832775260981,
      "grad_norm": 2.087040901184082,
      "learning_rate": 4.824698003205863e-06,
      "loss": 1.4467,
      "step": 25234
    },
    {
      "epoch": 0.9037190896556664,
      "grad_norm": 1.3491966724395752,
      "learning_rate": 4.8211393111915915e-06,
      "loss": 1.2438,
      "step": 25235
    },
    {
      "epoch": 0.9037549017852347,
      "grad_norm": 1.7652667760849,
      "learning_rate": 4.817581899686108e-06,
      "loss": 1.6547,
      "step": 25236
    },
    {
      "epoch": 0.9037907139148029,
      "grad_norm": 1.6787374019622803,
      "learning_rate": 4.814025768737296e-06,
      "loss": 1.2073,
      "step": 25237
    },
    {
      "epoch": 0.9038265260443712,
      "grad_norm": 1.5756950378417969,
      "learning_rate": 4.810470918392962e-06,
      "loss": 1.3766,
      "step": 25238
    },
    {
      "epoch": 0.9038623381739395,
      "grad_norm": 1.498803973197937,
      "learning_rate": 4.8069173487009785e-06,
      "loss": 1.3329,
      "step": 25239
    },
    {
      "epoch": 0.9038981503035078,
      "grad_norm": 1.668433427810669,
      "learning_rate": 4.803365059709131e-06,
      "loss": 1.4101,
      "step": 25240
    },
    {
      "epoch": 0.9039339624330761,
      "grad_norm": 1.7079111337661743,
      "learning_rate": 4.799814051465212e-06,
      "loss": 1.6624,
      "step": 25241
    },
    {
      "epoch": 0.9039697745626444,
      "grad_norm": 1.6564069986343384,
      "learning_rate": 4.7962643240169854e-06,
      "loss": 1.2615,
      "step": 25242
    },
    {
      "epoch": 0.9040055866922126,
      "grad_norm": 1.8872987031936646,
      "learning_rate": 4.792715877412213e-06,
      "loss": 1.494,
      "step": 25243
    },
    {
      "epoch": 0.9040413988217809,
      "grad_norm": 1.6588809490203857,
      "learning_rate": 4.789168711698655e-06,
      "loss": 1.5164,
      "step": 25244
    },
    {
      "epoch": 0.9040772109513492,
      "grad_norm": 1.7760930061340332,
      "learning_rate": 4.785622826924019e-06,
      "loss": 1.64,
      "step": 25245
    },
    {
      "epoch": 0.9041130230809175,
      "grad_norm": 1.5673993825912476,
      "learning_rate": 4.782078223135999e-06,
      "loss": 1.6067,
      "step": 25246
    },
    {
      "epoch": 0.9041488352104858,
      "grad_norm": 1.5011767148971558,
      "learning_rate": 4.778534900382292e-06,
      "loss": 1.441,
      "step": 25247
    },
    {
      "epoch": 0.9041846473400541,
      "grad_norm": 1.4967342615127563,
      "learning_rate": 4.774992858710581e-06,
      "loss": 1.2915,
      "step": 25248
    },
    {
      "epoch": 0.9042204594696224,
      "grad_norm": 1.8950409889221191,
      "learning_rate": 4.771452098168494e-06,
      "loss": 1.6408,
      "step": 25249
    },
    {
      "epoch": 0.9042562715991906,
      "grad_norm": 1.2963480949401855,
      "learning_rate": 4.767912618803705e-06,
      "loss": 1.1706,
      "step": 25250
    },
    {
      "epoch": 0.9042920837287589,
      "grad_norm": 1.2724201679229736,
      "learning_rate": 4.764374420663808e-06,
      "loss": 1.3548,
      "step": 25251
    },
    {
      "epoch": 0.9043278958583272,
      "grad_norm": 2.205700159072876,
      "learning_rate": 4.7608375037964e-06,
      "loss": 1.2672,
      "step": 25252
    },
    {
      "epoch": 0.9043637079878954,
      "grad_norm": 1.8525762557983398,
      "learning_rate": 4.757301868249076e-06,
      "loss": 1.2589,
      "step": 25253
    },
    {
      "epoch": 0.9043995201174638,
      "grad_norm": 1.4828543663024902,
      "learning_rate": 4.753767514069396e-06,
      "loss": 1.5784,
      "step": 25254
    },
    {
      "epoch": 0.9044353322470321,
      "grad_norm": 1.468957543373108,
      "learning_rate": 4.750234441304924e-06,
      "loss": 1.479,
      "step": 25255
    },
    {
      "epoch": 0.9044711443766004,
      "grad_norm": 1.6229407787322998,
      "learning_rate": 4.746702650003176e-06,
      "loss": 1.2722,
      "step": 25256
    },
    {
      "epoch": 0.9045069565061686,
      "grad_norm": 2.103856086730957,
      "learning_rate": 4.743172140211683e-06,
      "loss": 1.305,
      "step": 25257
    },
    {
      "epoch": 0.9045427686357369,
      "grad_norm": 1.640134572982788,
      "learning_rate": 4.7396429119779265e-06,
      "loss": 1.3592,
      "step": 25258
    },
    {
      "epoch": 0.9045785807653052,
      "grad_norm": 1.485317349433899,
      "learning_rate": 4.736114965349414e-06,
      "loss": 1.4747,
      "step": 25259
    },
    {
      "epoch": 0.9046143928948734,
      "grad_norm": 1.4240471124649048,
      "learning_rate": 4.732588300373586e-06,
      "loss": 1.2655,
      "step": 25260
    },
    {
      "epoch": 0.9046502050244418,
      "grad_norm": 1.6139092445373535,
      "learning_rate": 4.729062917097882e-06,
      "loss": 1.0557,
      "step": 25261
    },
    {
      "epoch": 0.9046860171540101,
      "grad_norm": 2.000537633895874,
      "learning_rate": 4.725538815569774e-06,
      "loss": 1.6511,
      "step": 25262
    },
    {
      "epoch": 0.9047218292835784,
      "grad_norm": 1.554819941520691,
      "learning_rate": 4.722015995836626e-06,
      "loss": 1.5654,
      "step": 25263
    },
    {
      "epoch": 0.9047576414131466,
      "grad_norm": 2.3554747104644775,
      "learning_rate": 4.718494457945855e-06,
      "loss": 1.5196,
      "step": 25264
    },
    {
      "epoch": 0.9047934535427149,
      "grad_norm": 1.2216877937316895,
      "learning_rate": 4.714974201944833e-06,
      "loss": 1.3913,
      "step": 25265
    },
    {
      "epoch": 0.9048292656722832,
      "grad_norm": 1.3515695333480835,
      "learning_rate": 4.711455227880935e-06,
      "loss": 1.4451,
      "step": 25266
    },
    {
      "epoch": 0.9048650778018514,
      "grad_norm": 1.557699203491211,
      "learning_rate": 4.707937535801488e-06,
      "loss": 1.023,
      "step": 25267
    },
    {
      "epoch": 0.9049008899314198,
      "grad_norm": 1.7648674249649048,
      "learning_rate": 4.704421125753822e-06,
      "loss": 1.4819,
      "step": 25268
    },
    {
      "epoch": 0.9049367020609881,
      "grad_norm": 2.089319944381714,
      "learning_rate": 4.700905997785254e-06,
      "loss": 1.3192,
      "step": 25269
    },
    {
      "epoch": 0.9049725141905564,
      "grad_norm": 1.6241894960403442,
      "learning_rate": 4.697392151943059e-06,
      "loss": 1.2065,
      "step": 25270
    },
    {
      "epoch": 0.9050083263201246,
      "grad_norm": 1.50771164894104,
      "learning_rate": 4.693879588274519e-06,
      "loss": 1.334,
      "step": 25271
    },
    {
      "epoch": 0.9050441384496929,
      "grad_norm": 1.4906578063964844,
      "learning_rate": 4.690368306826898e-06,
      "loss": 1.547,
      "step": 25272
    },
    {
      "epoch": 0.9050799505792612,
      "grad_norm": 1.5478224754333496,
      "learning_rate": 4.686858307647446e-06,
      "loss": 1.4141,
      "step": 25273
    },
    {
      "epoch": 0.9051157627088294,
      "grad_norm": 1.7823578119277954,
      "learning_rate": 4.683349590783348e-06,
      "loss": 1.5894,
      "step": 25274
    },
    {
      "epoch": 0.9051515748383978,
      "grad_norm": 1.496527075767517,
      "learning_rate": 4.679842156281844e-06,
      "loss": 1.2226,
      "step": 25275
    },
    {
      "epoch": 0.9051873869679661,
      "grad_norm": 1.2077003717422485,
      "learning_rate": 4.676336004190096e-06,
      "loss": 1.1825,
      "step": 25276
    },
    {
      "epoch": 0.9052231990975343,
      "grad_norm": 1.6871800422668457,
      "learning_rate": 4.6728311345553115e-06,
      "loss": 1.2391,
      "step": 25277
    },
    {
      "epoch": 0.9052590112271026,
      "grad_norm": 1.8817455768585205,
      "learning_rate": 4.669327547424607e-06,
      "loss": 1.5434,
      "step": 25278
    },
    {
      "epoch": 0.9052948233566709,
      "grad_norm": 1.5902183055877686,
      "learning_rate": 4.665825242845134e-06,
      "loss": 1.4759,
      "step": 25279
    },
    {
      "epoch": 0.9053306354862392,
      "grad_norm": 1.7828710079193115,
      "learning_rate": 4.662324220864011e-06,
      "loss": 1.3709,
      "step": 25280
    },
    {
      "epoch": 0.9053664476158074,
      "grad_norm": 1.9528136253356934,
      "learning_rate": 4.658824481528335e-06,
      "loss": 1.3893,
      "step": 25281
    },
    {
      "epoch": 0.9054022597453758,
      "grad_norm": 2.271791934967041,
      "learning_rate": 4.655326024885198e-06,
      "loss": 1.5958,
      "step": 25282
    },
    {
      "epoch": 0.9054380718749441,
      "grad_norm": 1.7135944366455078,
      "learning_rate": 4.651828850981654e-06,
      "loss": 1.7616,
      "step": 25283
    },
    {
      "epoch": 0.9054738840045123,
      "grad_norm": 2.0489044189453125,
      "learning_rate": 4.6483329598647874e-06,
      "loss": 1.4232,
      "step": 25284
    },
    {
      "epoch": 0.9055096961340806,
      "grad_norm": 1.7245614528656006,
      "learning_rate": 4.644838351581582e-06,
      "loss": 1.6448,
      "step": 25285
    },
    {
      "epoch": 0.9055455082636489,
      "grad_norm": 1.375653624534607,
      "learning_rate": 4.6413450261790894e-06,
      "loss": 1.3573,
      "step": 25286
    },
    {
      "epoch": 0.9055813203932171,
      "grad_norm": 1.45327889919281,
      "learning_rate": 4.637852983704294e-06,
      "loss": 1.5115,
      "step": 25287
    },
    {
      "epoch": 0.9056171325227854,
      "grad_norm": 1.4247119426727295,
      "learning_rate": 4.63436222420417e-06,
      "loss": 1.488,
      "step": 25288
    },
    {
      "epoch": 0.9056529446523538,
      "grad_norm": 1.779223918914795,
      "learning_rate": 4.630872747725701e-06,
      "loss": 1.5267,
      "step": 25289
    },
    {
      "epoch": 0.9056887567819221,
      "grad_norm": 1.4207184314727783,
      "learning_rate": 4.627384554315806e-06,
      "loss": 1.3127,
      "step": 25290
    },
    {
      "epoch": 0.9057245689114903,
      "grad_norm": 1.3641339540481567,
      "learning_rate": 4.623897644021446e-06,
      "loss": 1.1716,
      "step": 25291
    },
    {
      "epoch": 0.9057603810410586,
      "grad_norm": 1.7971229553222656,
      "learning_rate": 4.6204120168895085e-06,
      "loss": 1.7832,
      "step": 25292
    },
    {
      "epoch": 0.9057961931706269,
      "grad_norm": 1.6615642309188843,
      "learning_rate": 4.616927672966898e-06,
      "loss": 1.2579,
      "step": 25293
    },
    {
      "epoch": 0.9058320053001951,
      "grad_norm": 1.582353949546814,
      "learning_rate": 4.6134446123004885e-06,
      "loss": 1.2602,
      "step": 25294
    },
    {
      "epoch": 0.9058678174297634,
      "grad_norm": 1.9554626941680908,
      "learning_rate": 4.609962834937153e-06,
      "loss": 1.2429,
      "step": 25295
    },
    {
      "epoch": 0.9059036295593318,
      "grad_norm": 1.6370717287063599,
      "learning_rate": 4.606482340923712e-06,
      "loss": 1.3997,
      "step": 25296
    },
    {
      "epoch": 0.9059394416889001,
      "grad_norm": 1.5962556600570679,
      "learning_rate": 4.6030031303070045e-06,
      "loss": 1.5322,
      "step": 25297
    },
    {
      "epoch": 0.9059752538184683,
      "grad_norm": 1.372740387916565,
      "learning_rate": 4.599525203133848e-06,
      "loss": 1.2582,
      "step": 25298
    },
    {
      "epoch": 0.9060110659480366,
      "grad_norm": 1.5764113664627075,
      "learning_rate": 4.596048559451005e-06,
      "loss": 1.4412,
      "step": 25299
    },
    {
      "epoch": 0.9060468780776049,
      "grad_norm": 1.6428806781768799,
      "learning_rate": 4.592573199305272e-06,
      "loss": 1.5331,
      "step": 25300
    },
    {
      "epoch": 0.9060826902071731,
      "grad_norm": 2.063018560409546,
      "learning_rate": 4.58909912274339e-06,
      "loss": 1.4299,
      "step": 25301
    },
    {
      "epoch": 0.9061185023367414,
      "grad_norm": 1.7012205123901367,
      "learning_rate": 4.585626329812132e-06,
      "loss": 1.6607,
      "step": 25302
    },
    {
      "epoch": 0.9061543144663098,
      "grad_norm": 1.4773954153060913,
      "learning_rate": 4.582154820558182e-06,
      "loss": 1.2731,
      "step": 25303
    },
    {
      "epoch": 0.906190126595878,
      "grad_norm": 1.781326413154602,
      "learning_rate": 4.5786845950282486e-06,
      "loss": 1.2667,
      "step": 25304
    },
    {
      "epoch": 0.9062259387254463,
      "grad_norm": 1.574438452720642,
      "learning_rate": 4.575215653269061e-06,
      "loss": 1.4534,
      "step": 25305
    },
    {
      "epoch": 0.9062617508550146,
      "grad_norm": 1.2895307540893555,
      "learning_rate": 4.571747995327224e-06,
      "loss": 1.4442,
      "step": 25306
    },
    {
      "epoch": 0.9062975629845829,
      "grad_norm": 1.6740641593933105,
      "learning_rate": 4.568281621249437e-06,
      "loss": 1.269,
      "step": 25307
    },
    {
      "epoch": 0.9063333751141511,
      "grad_norm": 2.1443235874176025,
      "learning_rate": 4.564816531082316e-06,
      "loss": 1.6352,
      "step": 25308
    },
    {
      "epoch": 0.9063691872437194,
      "grad_norm": 1.7951942682266235,
      "learning_rate": 4.561352724872503e-06,
      "loss": 1.3721,
      "step": 25309
    },
    {
      "epoch": 0.9064049993732878,
      "grad_norm": 1.5101568698883057,
      "learning_rate": 4.557890202666571e-06,
      "loss": 1.2677,
      "step": 25310
    },
    {
      "epoch": 0.906440811502856,
      "grad_norm": 1.488024353981018,
      "learning_rate": 4.5544289645111145e-06,
      "loss": 1.3814,
      "step": 25311
    },
    {
      "epoch": 0.9064766236324243,
      "grad_norm": 1.546871304512024,
      "learning_rate": 4.5509690104526995e-06,
      "loss": 1.1841,
      "step": 25312
    },
    {
      "epoch": 0.9065124357619926,
      "grad_norm": 1.6033934354782104,
      "learning_rate": 4.547510340537886e-06,
      "loss": 1.4059,
      "step": 25313
    },
    {
      "epoch": 0.9065482478915609,
      "grad_norm": 1.8143798112869263,
      "learning_rate": 4.544052954813194e-06,
      "loss": 1.2067,
      "step": 25314
    },
    {
      "epoch": 0.9065840600211291,
      "grad_norm": 1.500676155090332,
      "learning_rate": 4.540596853325119e-06,
      "loss": 1.2754,
      "step": 25315
    },
    {
      "epoch": 0.9066198721506974,
      "grad_norm": 1.7909812927246094,
      "learning_rate": 4.537142036120212e-06,
      "loss": 1.3087,
      "step": 25316
    },
    {
      "epoch": 0.9066556842802658,
      "grad_norm": 1.3690929412841797,
      "learning_rate": 4.533688503244893e-06,
      "loss": 1.2588,
      "step": 25317
    },
    {
      "epoch": 0.906691496409834,
      "grad_norm": 1.785294532775879,
      "learning_rate": 4.5302362547456565e-06,
      "loss": 1.3569,
      "step": 25318
    },
    {
      "epoch": 0.9067273085394023,
      "grad_norm": 1.5315146446228027,
      "learning_rate": 4.5267852906689555e-06,
      "loss": 1.4828,
      "step": 25319
    },
    {
      "epoch": 0.9067631206689706,
      "grad_norm": 1.328297734260559,
      "learning_rate": 4.523335611061208e-06,
      "loss": 1.5227,
      "step": 25320
    },
    {
      "epoch": 0.9067989327985388,
      "grad_norm": 1.671176791191101,
      "learning_rate": 4.51988721596881e-06,
      "loss": 1.3934,
      "step": 25321
    },
    {
      "epoch": 0.9068347449281071,
      "grad_norm": 1.6849404573440552,
      "learning_rate": 4.51644010543818e-06,
      "loss": 1.3083,
      "step": 25322
    },
    {
      "epoch": 0.9068705570576754,
      "grad_norm": 1.6850334405899048,
      "learning_rate": 4.512994279515692e-06,
      "loss": 1.5317,
      "step": 25323
    },
    {
      "epoch": 0.9069063691872438,
      "grad_norm": 1.4295352697372437,
      "learning_rate": 4.509549738247676e-06,
      "loss": 1.378,
      "step": 25324
    },
    {
      "epoch": 0.906942181316812,
      "grad_norm": 2.1510770320892334,
      "learning_rate": 4.5061064816805165e-06,
      "loss": 1.4521,
      "step": 25325
    },
    {
      "epoch": 0.9069779934463803,
      "grad_norm": 1.9867370128631592,
      "learning_rate": 4.502664509860488e-06,
      "loss": 1.4834,
      "step": 25326
    },
    {
      "epoch": 0.9070138055759486,
      "grad_norm": 1.2240649461746216,
      "learning_rate": 4.499223822833942e-06,
      "loss": 1.3368,
      "step": 25327
    },
    {
      "epoch": 0.9070496177055168,
      "grad_norm": 1.844240665435791,
      "learning_rate": 4.4957844206471535e-06,
      "loss": 1.3575,
      "step": 25328
    },
    {
      "epoch": 0.9070854298350851,
      "grad_norm": 1.304735779762268,
      "learning_rate": 4.492346303346395e-06,
      "loss": 1.1249,
      "step": 25329
    },
    {
      "epoch": 0.9071212419646534,
      "grad_norm": 1.8168039321899414,
      "learning_rate": 4.488909470977909e-06,
      "loss": 1.3287,
      "step": 25330
    },
    {
      "epoch": 0.9071570540942218,
      "grad_norm": 1.2686028480529785,
      "learning_rate": 4.485473923587957e-06,
      "loss": 1.5049,
      "step": 25331
    },
    {
      "epoch": 0.90719286622379,
      "grad_norm": 1.6690572500228882,
      "learning_rate": 4.482039661222759e-06,
      "loss": 1.43,
      "step": 25332
    },
    {
      "epoch": 0.9072286783533583,
      "grad_norm": 1.5077745914459229,
      "learning_rate": 4.478606683928476e-06,
      "loss": 1.6013,
      "step": 25333
    },
    {
      "epoch": 0.9072644904829266,
      "grad_norm": 1.62128484249115,
      "learning_rate": 4.475174991751352e-06,
      "loss": 1.1891,
      "step": 25334
    },
    {
      "epoch": 0.9073003026124948,
      "grad_norm": 1.4367783069610596,
      "learning_rate": 4.471744584737525e-06,
      "loss": 1.5777,
      "step": 25335
    },
    {
      "epoch": 0.9073361147420631,
      "grad_norm": 2.2138447761535645,
      "learning_rate": 4.468315462933159e-06,
      "loss": 1.4724,
      "step": 25336
    },
    {
      "epoch": 0.9073719268716314,
      "grad_norm": 1.362805724143982,
      "learning_rate": 4.464887626384362e-06,
      "loss": 1.3973,
      "step": 25337
    },
    {
      "epoch": 0.9074077390011998,
      "grad_norm": 1.5935183763504028,
      "learning_rate": 4.461461075137285e-06,
      "loss": 1.6762,
      "step": 25338
    },
    {
      "epoch": 0.907443551130768,
      "grad_norm": 1.413731575012207,
      "learning_rate": 4.458035809238026e-06,
      "loss": 1.4027,
      "step": 25339
    },
    {
      "epoch": 0.9074793632603363,
      "grad_norm": 1.4532414674758911,
      "learning_rate": 4.454611828732636e-06,
      "loss": 1.385,
      "step": 25340
    },
    {
      "epoch": 0.9075151753899046,
      "grad_norm": 1.9259858131408691,
      "learning_rate": 4.4511891336671885e-06,
      "loss": 1.6331,
      "step": 25341
    },
    {
      "epoch": 0.9075509875194728,
      "grad_norm": 1.7889636754989624,
      "learning_rate": 4.447767724087759e-06,
      "loss": 1.1333,
      "step": 25342
    },
    {
      "epoch": 0.9075867996490411,
      "grad_norm": 1.47054922580719,
      "learning_rate": 4.444347600040366e-06,
      "loss": 1.4389,
      "step": 25343
    },
    {
      "epoch": 0.9076226117786094,
      "grad_norm": 1.6787251234054565,
      "learning_rate": 4.440928761570995e-06,
      "loss": 1.1389,
      "step": 25344
    },
    {
      "epoch": 0.9076584239081777,
      "grad_norm": 1.4817570447921753,
      "learning_rate": 4.4375112087256864e-06,
      "loss": 1.1824,
      "step": 25345
    },
    {
      "epoch": 0.907694236037746,
      "grad_norm": 2.3525238037109375,
      "learning_rate": 4.434094941550393e-06,
      "loss": 1.3532,
      "step": 25346
    },
    {
      "epoch": 0.9077300481673143,
      "grad_norm": 1.610073208808899,
      "learning_rate": 4.430679960091089e-06,
      "loss": 1.6939,
      "step": 25347
    },
    {
      "epoch": 0.9077658602968826,
      "grad_norm": 2.0152218341827393,
      "learning_rate": 4.427266264393693e-06,
      "loss": 1.2251,
      "step": 25348
    },
    {
      "epoch": 0.9078016724264508,
      "grad_norm": 1.7770185470581055,
      "learning_rate": 4.423853854504156e-06,
      "loss": 1.3707,
      "step": 25349
    },
    {
      "epoch": 0.9078374845560191,
      "grad_norm": 2.1795554161071777,
      "learning_rate": 4.420442730468388e-06,
      "loss": 1.2333,
      "step": 25350
    },
    {
      "epoch": 0.9078732966855874,
      "grad_norm": 1.550978660583496,
      "learning_rate": 4.417032892332263e-06,
      "loss": 1.528,
      "step": 25351
    },
    {
      "epoch": 0.9079091088151557,
      "grad_norm": 1.501692533493042,
      "learning_rate": 4.413624340141676e-06,
      "loss": 1.2372,
      "step": 25352
    },
    {
      "epoch": 0.907944920944724,
      "grad_norm": 1.6500540971755981,
      "learning_rate": 4.410217073942468e-06,
      "loss": 1.4474,
      "step": 25353
    },
    {
      "epoch": 0.9079807330742923,
      "grad_norm": 1.5187865495681763,
      "learning_rate": 4.4068110937805055e-06,
      "loss": 1.1773,
      "step": 25354
    },
    {
      "epoch": 0.9080165452038605,
      "grad_norm": 1.3688457012176514,
      "learning_rate": 4.40340639970157e-06,
      "loss": 1.2746,
      "step": 25355
    },
    {
      "epoch": 0.9080523573334288,
      "grad_norm": 1.6650950908660889,
      "learning_rate": 4.400002991751495e-06,
      "loss": 1.6117,
      "step": 25356
    },
    {
      "epoch": 0.9080881694629971,
      "grad_norm": 1.6353851556777954,
      "learning_rate": 4.396600869976086e-06,
      "loss": 1.2558,
      "step": 25357
    },
    {
      "epoch": 0.9081239815925654,
      "grad_norm": 1.6381622552871704,
      "learning_rate": 4.393200034421074e-06,
      "loss": 1.4325,
      "step": 25358
    },
    {
      "epoch": 0.9081597937221337,
      "grad_norm": 2.1839332580566406,
      "learning_rate": 4.3898004851322335e-06,
      "loss": 1.5382,
      "step": 25359
    },
    {
      "epoch": 0.908195605851702,
      "grad_norm": 1.539484977722168,
      "learning_rate": 4.386402222155295e-06,
      "loss": 1.6129,
      "step": 25360
    },
    {
      "epoch": 0.9082314179812703,
      "grad_norm": 1.4183225631713867,
      "learning_rate": 4.383005245535998e-06,
      "loss": 1.3707,
      "step": 25361
    },
    {
      "epoch": 0.9082672301108385,
      "grad_norm": 1.6400363445281982,
      "learning_rate": 4.379609555320008e-06,
      "loss": 0.9586,
      "step": 25362
    },
    {
      "epoch": 0.9083030422404068,
      "grad_norm": 1.407865047454834,
      "learning_rate": 4.376215151553042e-06,
      "loss": 1.3569,
      "step": 25363
    },
    {
      "epoch": 0.9083388543699751,
      "grad_norm": 1.3456802368164062,
      "learning_rate": 4.372822034280744e-06,
      "loss": 0.978,
      "step": 25364
    },
    {
      "epoch": 0.9083746664995433,
      "grad_norm": 2.59515380859375,
      "learning_rate": 4.3694302035487965e-06,
      "loss": 1.2745,
      "step": 25365
    },
    {
      "epoch": 0.9084104786291117,
      "grad_norm": 1.7671518325805664,
      "learning_rate": 4.366039659402798e-06,
      "loss": 1.4961,
      "step": 25366
    },
    {
      "epoch": 0.90844629075868,
      "grad_norm": 2.6446104049682617,
      "learning_rate": 4.362650401888369e-06,
      "loss": 1.2699,
      "step": 25367
    },
    {
      "epoch": 0.9084821028882483,
      "grad_norm": 1.3323699235916138,
      "learning_rate": 4.359262431051137e-06,
      "loss": 1.4321,
      "step": 25368
    },
    {
      "epoch": 0.9085179150178165,
      "grad_norm": 1.728229284286499,
      "learning_rate": 4.355875746936644e-06,
      "loss": 1.5239,
      "step": 25369
    },
    {
      "epoch": 0.9085537271473848,
      "grad_norm": 1.4732418060302734,
      "learning_rate": 4.352490349590477e-06,
      "loss": 1.4759,
      "step": 25370
    },
    {
      "epoch": 0.9085895392769531,
      "grad_norm": 1.4603962898254395,
      "learning_rate": 4.349106239058165e-06,
      "loss": 1.6496,
      "step": 25371
    },
    {
      "epoch": 0.9086253514065213,
      "grad_norm": 1.721230149269104,
      "learning_rate": 4.345723415385272e-06,
      "loss": 1.7606,
      "step": 25372
    },
    {
      "epoch": 0.9086611635360897,
      "grad_norm": 1.6769062280654907,
      "learning_rate": 4.342341878617262e-06,
      "loss": 1.7661,
      "step": 25373
    },
    {
      "epoch": 0.908696975665658,
      "grad_norm": 2.0991551876068115,
      "learning_rate": 4.338961628799665e-06,
      "loss": 1.2793,
      "step": 25374
    },
    {
      "epoch": 0.9087327877952263,
      "grad_norm": 1.581172227859497,
      "learning_rate": 4.335582665977944e-06,
      "loss": 1.4508,
      "step": 25375
    },
    {
      "epoch": 0.9087685999247945,
      "grad_norm": 1.404631495475769,
      "learning_rate": 4.332204990197564e-06,
      "loss": 1.4273,
      "step": 25376
    },
    {
      "epoch": 0.9088044120543628,
      "grad_norm": 1.1999675035476685,
      "learning_rate": 4.328828601503943e-06,
      "loss": 1.2674,
      "step": 25377
    },
    {
      "epoch": 0.9088402241839311,
      "grad_norm": 1.3518636226654053,
      "learning_rate": 4.325453499942545e-06,
      "loss": 1.3804,
      "step": 25378
    },
    {
      "epoch": 0.9088760363134993,
      "grad_norm": 1.5245866775512695,
      "learning_rate": 4.322079685558755e-06,
      "loss": 1.2018,
      "step": 25379
    },
    {
      "epoch": 0.9089118484430677,
      "grad_norm": 1.940627932548523,
      "learning_rate": 4.318707158397972e-06,
      "loss": 1.5514,
      "step": 25380
    },
    {
      "epoch": 0.908947660572636,
      "grad_norm": 1.9791686534881592,
      "learning_rate": 4.3153359185055474e-06,
      "loss": 1.3945,
      "step": 25381
    },
    {
      "epoch": 0.9089834727022043,
      "grad_norm": 2.057812452316284,
      "learning_rate": 4.311965965926867e-06,
      "loss": 1.1848,
      "step": 25382
    },
    {
      "epoch": 0.9090192848317725,
      "grad_norm": 1.4718939065933228,
      "learning_rate": 4.308597300707262e-06,
      "loss": 1.4995,
      "step": 25383
    },
    {
      "epoch": 0.9090550969613408,
      "grad_norm": 1.5255402326583862,
      "learning_rate": 4.305229922892029e-06,
      "loss": 1.418,
      "step": 25384
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 1.564485788345337,
      "learning_rate": 4.301863832526498e-06,
      "loss": 1.1865,
      "step": 25385
    },
    {
      "epoch": 0.9091267212204773,
      "grad_norm": 1.8961577415466309,
      "learning_rate": 4.298499029655967e-06,
      "loss": 1.4887,
      "step": 25386
    },
    {
      "epoch": 0.9091625333500457,
      "grad_norm": 1.5364948511123657,
      "learning_rate": 4.295135514325654e-06,
      "loss": 1.5881,
      "step": 25387
    },
    {
      "epoch": 0.909198345479614,
      "grad_norm": 2.002300977706909,
      "learning_rate": 4.291773286580858e-06,
      "loss": 1.6364,
      "step": 25388
    },
    {
      "epoch": 0.9092341576091822,
      "grad_norm": 2.0127434730529785,
      "learning_rate": 4.288412346466797e-06,
      "loss": 1.3339,
      "step": 25389
    },
    {
      "epoch": 0.9092699697387505,
      "grad_norm": 1.5489211082458496,
      "learning_rate": 4.28505269402869e-06,
      "loss": 1.3824,
      "step": 25390
    },
    {
      "epoch": 0.9093057818683188,
      "grad_norm": 1.6152890920639038,
      "learning_rate": 4.281694329311736e-06,
      "loss": 1.3801,
      "step": 25391
    },
    {
      "epoch": 0.909341593997887,
      "grad_norm": 2.6852145195007324,
      "learning_rate": 4.278337252361109e-06,
      "loss": 1.5435,
      "step": 25392
    },
    {
      "epoch": 0.9093774061274553,
      "grad_norm": 2.5400712490081787,
      "learning_rate": 4.2749814632219946e-06,
      "loss": 1.4603,
      "step": 25393
    },
    {
      "epoch": 0.9094132182570237,
      "grad_norm": 2.176903486251831,
      "learning_rate": 4.271626961939524e-06,
      "loss": 1.4666,
      "step": 25394
    },
    {
      "epoch": 0.909449030386592,
      "grad_norm": 1.6950504779815674,
      "learning_rate": 4.268273748558815e-06,
      "loss": 1.3297,
      "step": 25395
    },
    {
      "epoch": 0.9094848425161602,
      "grad_norm": 1.5055739879608154,
      "learning_rate": 4.264921823125012e-06,
      "loss": 1.4014,
      "step": 25396
    },
    {
      "epoch": 0.9095206546457285,
      "grad_norm": 1.2444562911987305,
      "learning_rate": 4.261571185683211e-06,
      "loss": 1.4675,
      "step": 25397
    },
    {
      "epoch": 0.9095564667752968,
      "grad_norm": 2.0132944583892822,
      "learning_rate": 4.258221836278453e-06,
      "loss": 1.6443,
      "step": 25398
    },
    {
      "epoch": 0.909592278904865,
      "grad_norm": 1.8058809041976929,
      "learning_rate": 4.2548737749558255e-06,
      "loss": 1.347,
      "step": 25399
    },
    {
      "epoch": 0.9096280910344333,
      "grad_norm": 1.5747233629226685,
      "learning_rate": 4.2515270017603695e-06,
      "loss": 1.6271,
      "step": 25400
    },
    {
      "epoch": 0.9096639031640017,
      "grad_norm": 1.754483699798584,
      "learning_rate": 4.248181516737127e-06,
      "loss": 1.4832,
      "step": 25401
    },
    {
      "epoch": 0.90969971529357,
      "grad_norm": 1.453294038772583,
      "learning_rate": 4.244837319931072e-06,
      "loss": 1.5246,
      "step": 25402
    },
    {
      "epoch": 0.9097355274231382,
      "grad_norm": 1.683763027191162,
      "learning_rate": 4.2414944113872255e-06,
      "loss": 1.3641,
      "step": 25403
    },
    {
      "epoch": 0.9097713395527065,
      "grad_norm": 1.6361896991729736,
      "learning_rate": 4.2381527911505625e-06,
      "loss": 1.8167,
      "step": 25404
    },
    {
      "epoch": 0.9098071516822748,
      "grad_norm": 1.424181580543518,
      "learning_rate": 4.234812459266013e-06,
      "loss": 1.5335,
      "step": 25405
    },
    {
      "epoch": 0.909842963811843,
      "grad_norm": 2.2374167442321777,
      "learning_rate": 4.231473415778531e-06,
      "loss": 1.4462,
      "step": 25406
    },
    {
      "epoch": 0.9098787759414113,
      "grad_norm": 1.852335810661316,
      "learning_rate": 4.228135660733046e-06,
      "loss": 1.648,
      "step": 25407
    },
    {
      "epoch": 0.9099145880709797,
      "grad_norm": 2.4960708618164062,
      "learning_rate": 4.224799194174467e-06,
      "loss": 1.3632,
      "step": 25408
    },
    {
      "epoch": 0.909950400200548,
      "grad_norm": 1.384843349456787,
      "learning_rate": 4.221464016147669e-06,
      "loss": 1.5738,
      "step": 25409
    },
    {
      "epoch": 0.9099862123301162,
      "grad_norm": 1.7134809494018555,
      "learning_rate": 4.218130126697517e-06,
      "loss": 1.5194,
      "step": 25410
    },
    {
      "epoch": 0.9100220244596845,
      "grad_norm": 1.2886848449707031,
      "learning_rate": 4.214797525868897e-06,
      "loss": 1.2627,
      "step": 25411
    },
    {
      "epoch": 0.9100578365892528,
      "grad_norm": 1.9155702590942383,
      "learning_rate": 4.2114662137066055e-06,
      "loss": 1.8567,
      "step": 25412
    },
    {
      "epoch": 0.910093648718821,
      "grad_norm": 1.4528340101242065,
      "learning_rate": 4.208136190255485e-06,
      "loss": 1.4138,
      "step": 25413
    },
    {
      "epoch": 0.9101294608483893,
      "grad_norm": 1.6072170734405518,
      "learning_rate": 4.204807455560311e-06,
      "loss": 1.4734,
      "step": 25414
    },
    {
      "epoch": 0.9101652729779577,
      "grad_norm": 1.2965353727340698,
      "learning_rate": 4.201480009665915e-06,
      "loss": 1.3501,
      "step": 25415
    },
    {
      "epoch": 0.910201085107526,
      "grad_norm": 1.2253295183181763,
      "learning_rate": 4.198153852617015e-06,
      "loss": 1.2337,
      "step": 25416
    },
    {
      "epoch": 0.9102368972370942,
      "grad_norm": 2.0200018882751465,
      "learning_rate": 4.194828984458376e-06,
      "loss": 1.5565,
      "step": 25417
    },
    {
      "epoch": 0.9102727093666625,
      "grad_norm": 1.6863843202590942,
      "learning_rate": 4.191505405234741e-06,
      "loss": 1.4327,
      "step": 25418
    },
    {
      "epoch": 0.9103085214962308,
      "grad_norm": 1.7367192506790161,
      "learning_rate": 4.188183114990829e-06,
      "loss": 1.4891,
      "step": 25419
    },
    {
      "epoch": 0.910344333625799,
      "grad_norm": 2.484041452407837,
      "learning_rate": 4.1848621137713154e-06,
      "loss": 1.5881,
      "step": 25420
    },
    {
      "epoch": 0.9103801457553673,
      "grad_norm": 1.6672288179397583,
      "learning_rate": 4.181542401620875e-06,
      "loss": 1.2391,
      "step": 25421
    },
    {
      "epoch": 0.9104159578849357,
      "grad_norm": 1.5806713104248047,
      "learning_rate": 4.178223978584206e-06,
      "loss": 1.6583,
      "step": 25422
    },
    {
      "epoch": 0.910451770014504,
      "grad_norm": 1.8388357162475586,
      "learning_rate": 4.174906844705917e-06,
      "loss": 1.246,
      "step": 25423
    },
    {
      "epoch": 0.9104875821440722,
      "grad_norm": 1.3316192626953125,
      "learning_rate": 4.171591000030672e-06,
      "loss": 1.4997,
      "step": 25424
    },
    {
      "epoch": 0.9105233942736405,
      "grad_norm": 1.8472051620483398,
      "learning_rate": 4.168276444603026e-06,
      "loss": 1.4753,
      "step": 25425
    },
    {
      "epoch": 0.9105592064032088,
      "grad_norm": 1.7919559478759766,
      "learning_rate": 4.164963178467629e-06,
      "loss": 1.5098,
      "step": 25426
    },
    {
      "epoch": 0.910595018532777,
      "grad_norm": 1.5775041580200195,
      "learning_rate": 4.161651201669036e-06,
      "loss": 1.531,
      "step": 25427
    },
    {
      "epoch": 0.9106308306623453,
      "grad_norm": 2.2557876110076904,
      "learning_rate": 4.1583405142517906e-06,
      "loss": 1.5838,
      "step": 25428
    },
    {
      "epoch": 0.9106666427919137,
      "grad_norm": 1.9684809446334839,
      "learning_rate": 4.155031116260466e-06,
      "loss": 1.2069,
      "step": 25429
    },
    {
      "epoch": 0.9107024549214819,
      "grad_norm": 1.215849757194519,
      "learning_rate": 4.15172300773955e-06,
      "loss": 1.4853,
      "step": 25430
    },
    {
      "epoch": 0.9107382670510502,
      "grad_norm": 1.6601389646530151,
      "learning_rate": 4.148416188733584e-06,
      "loss": 1.3532,
      "step": 25431
    },
    {
      "epoch": 0.9107740791806185,
      "grad_norm": 1.467976450920105,
      "learning_rate": 4.1451106592869995e-06,
      "loss": 1.3732,
      "step": 25432
    },
    {
      "epoch": 0.9108098913101867,
      "grad_norm": 1.4758421182632446,
      "learning_rate": 4.14180641944435e-06,
      "loss": 1.4902,
      "step": 25433
    },
    {
      "epoch": 0.910845703439755,
      "grad_norm": 1.355146884918213,
      "learning_rate": 4.138503469250021e-06,
      "loss": 1.3033,
      "step": 25434
    },
    {
      "epoch": 0.9108815155693233,
      "grad_norm": 1.4203933477401733,
      "learning_rate": 4.13520180874849e-06,
      "loss": 1.3695,
      "step": 25435
    },
    {
      "epoch": 0.9109173276988917,
      "grad_norm": 2.0352320671081543,
      "learning_rate": 4.131901437984153e-06,
      "loss": 1.651,
      "step": 25436
    },
    {
      "epoch": 0.9109531398284599,
      "grad_norm": 1.4367496967315674,
      "learning_rate": 4.128602357001421e-06,
      "loss": 1.3241,
      "step": 25437
    },
    {
      "epoch": 0.9109889519580282,
      "grad_norm": 1.3705084323883057,
      "learning_rate": 4.12530456584469e-06,
      "loss": 1.5884,
      "step": 25438
    },
    {
      "epoch": 0.9110247640875965,
      "grad_norm": 1.4604884386062622,
      "learning_rate": 4.122008064558313e-06,
      "loss": 1.295,
      "step": 25439
    },
    {
      "epoch": 0.9110605762171647,
      "grad_norm": 1.8145109415054321,
      "learning_rate": 4.118712853186634e-06,
      "loss": 1.5224,
      "step": 25440
    },
    {
      "epoch": 0.911096388346733,
      "grad_norm": 1.4539945125579834,
      "learning_rate": 4.115418931773996e-06,
      "loss": 1.6827,
      "step": 25441
    },
    {
      "epoch": 0.9111322004763013,
      "grad_norm": 1.368727445602417,
      "learning_rate": 4.112126300364727e-06,
      "loss": 1.4317,
      "step": 25442
    },
    {
      "epoch": 0.9111680126058697,
      "grad_norm": 1.5504074096679688,
      "learning_rate": 4.108834959003094e-06,
      "loss": 1.4278,
      "step": 25443
    },
    {
      "epoch": 0.9112038247354379,
      "grad_norm": 1.5621628761291504,
      "learning_rate": 4.1055449077334165e-06,
      "loss": 1.2978,
      "step": 25444
    },
    {
      "epoch": 0.9112396368650062,
      "grad_norm": 1.6070566177368164,
      "learning_rate": 4.102256146599936e-06,
      "loss": 1.158,
      "step": 25445
    },
    {
      "epoch": 0.9112754489945745,
      "grad_norm": 2.3168065547943115,
      "learning_rate": 4.098968675646886e-06,
      "loss": 1.3911,
      "step": 25446
    },
    {
      "epoch": 0.9113112611241427,
      "grad_norm": 1.7678879499435425,
      "learning_rate": 4.095682494918507e-06,
      "loss": 1.1655,
      "step": 25447
    },
    {
      "epoch": 0.911347073253711,
      "grad_norm": 2.47255539894104,
      "learning_rate": 4.092397604459019e-06,
      "loss": 1.484,
      "step": 25448
    },
    {
      "epoch": 0.9113828853832793,
      "grad_norm": 1.4481755495071411,
      "learning_rate": 4.089114004312622e-06,
      "loss": 1.3727,
      "step": 25449
    },
    {
      "epoch": 0.9114186975128477,
      "grad_norm": 1.532240390777588,
      "learning_rate": 4.085831694523456e-06,
      "loss": 1.5036,
      "step": 25450
    },
    {
      "epoch": 0.9114545096424159,
      "grad_norm": 1.6303825378417969,
      "learning_rate": 4.082550675135721e-06,
      "loss": 1.2653,
      "step": 25451
    },
    {
      "epoch": 0.9114903217719842,
      "grad_norm": 2.0714213848114014,
      "learning_rate": 4.079270946193525e-06,
      "loss": 1.2285,
      "step": 25452
    },
    {
      "epoch": 0.9115261339015525,
      "grad_norm": 2.202430486679077,
      "learning_rate": 4.075992507741033e-06,
      "loss": 1.3034,
      "step": 25453
    },
    {
      "epoch": 0.9115619460311207,
      "grad_norm": 2.01389741897583,
      "learning_rate": 4.07271535982231e-06,
      "loss": 1.6597,
      "step": 25454
    },
    {
      "epoch": 0.911597758160689,
      "grad_norm": 1.577217698097229,
      "learning_rate": 4.0694395024814754e-06,
      "loss": 1.4017,
      "step": 25455
    },
    {
      "epoch": 0.9116335702902573,
      "grad_norm": 1.35993230342865,
      "learning_rate": 4.066164935762595e-06,
      "loss": 1.406,
      "step": 25456
    },
    {
      "epoch": 0.9116693824198256,
      "grad_norm": 1.9778926372528076,
      "learning_rate": 4.062891659709711e-06,
      "loss": 1.3003,
      "step": 25457
    },
    {
      "epoch": 0.9117051945493939,
      "grad_norm": 1.6684564352035522,
      "learning_rate": 4.059619674366866e-06,
      "loss": 1.5999,
      "step": 25458
    },
    {
      "epoch": 0.9117410066789622,
      "grad_norm": 2.0282845497131348,
      "learning_rate": 4.05634897977808e-06,
      "loss": 1.3524,
      "step": 25459
    },
    {
      "epoch": 0.9117768188085305,
      "grad_norm": 2.274070978164673,
      "learning_rate": 4.053079575987384e-06,
      "loss": 1.3631,
      "step": 25460
    },
    {
      "epoch": 0.9118126309380987,
      "grad_norm": 1.6663633584976196,
      "learning_rate": 4.049811463038722e-06,
      "loss": 1.5455,
      "step": 25461
    },
    {
      "epoch": 0.911848443067667,
      "grad_norm": 1.4195297956466675,
      "learning_rate": 4.0465446409760795e-06,
      "loss": 1.4995,
      "step": 25462
    },
    {
      "epoch": 0.9118842551972353,
      "grad_norm": 1.5641138553619385,
      "learning_rate": 4.043279109843412e-06,
      "loss": 1.4892,
      "step": 25463
    },
    {
      "epoch": 0.9119200673268036,
      "grad_norm": 1.4835494756698608,
      "learning_rate": 4.04001486968465e-06,
      "loss": 1.4001,
      "step": 25464
    },
    {
      "epoch": 0.9119558794563719,
      "grad_norm": 1.8441171646118164,
      "learning_rate": 4.036751920543702e-06,
      "loss": 1.6182,
      "step": 25465
    },
    {
      "epoch": 0.9119916915859402,
      "grad_norm": 2.2617461681365967,
      "learning_rate": 4.033490262464468e-06,
      "loss": 1.2301,
      "step": 25466
    },
    {
      "epoch": 0.9120275037155084,
      "grad_norm": 1.4524611234664917,
      "learning_rate": 4.030229895490856e-06,
      "loss": 1.5047,
      "step": 25467
    },
    {
      "epoch": 0.9120633158450767,
      "grad_norm": 1.821845531463623,
      "learning_rate": 4.026970819666698e-06,
      "loss": 1.2156,
      "step": 25468
    },
    {
      "epoch": 0.912099127974645,
      "grad_norm": 1.517996072769165,
      "learning_rate": 4.023713035035836e-06,
      "loss": 1.4621,
      "step": 25469
    },
    {
      "epoch": 0.9121349401042133,
      "grad_norm": 2.0149309635162354,
      "learning_rate": 4.020456541642126e-06,
      "loss": 1.4772,
      "step": 25470
    },
    {
      "epoch": 0.9121707522337816,
      "grad_norm": 1.5749101638793945,
      "learning_rate": 4.017201339529386e-06,
      "loss": 1.3722,
      "step": 25471
    },
    {
      "epoch": 0.9122065643633499,
      "grad_norm": 1.673729658126831,
      "learning_rate": 4.013947428741372e-06,
      "loss": 1.6295,
      "step": 25472
    },
    {
      "epoch": 0.9122423764929182,
      "grad_norm": 1.4756914377212524,
      "learning_rate": 4.01069480932188e-06,
      "loss": 1.5144,
      "step": 25473
    },
    {
      "epoch": 0.9122781886224864,
      "grad_norm": 1.3629252910614014,
      "learning_rate": 4.007443481314699e-06,
      "loss": 1.3908,
      "step": 25474
    },
    {
      "epoch": 0.9123140007520547,
      "grad_norm": 1.5156805515289307,
      "learning_rate": 4.0041934447635156e-06,
      "loss": 1.3617,
      "step": 25475
    },
    {
      "epoch": 0.912349812881623,
      "grad_norm": 1.570890188217163,
      "learning_rate": 4.000944699712094e-06,
      "loss": 1.2795,
      "step": 25476
    },
    {
      "epoch": 0.9123856250111912,
      "grad_norm": 1.760048508644104,
      "learning_rate": 3.997697246204124e-06,
      "loss": 1.4411,
      "step": 25477
    },
    {
      "epoch": 0.9124214371407596,
      "grad_norm": 1.7072031497955322,
      "learning_rate": 3.994451084283324e-06,
      "loss": 1.4699,
      "step": 25478
    },
    {
      "epoch": 0.9124572492703279,
      "grad_norm": 1.4796770811080933,
      "learning_rate": 3.991206213993326e-06,
      "loss": 1.326,
      "step": 25479
    },
    {
      "epoch": 0.9124930613998962,
      "grad_norm": 1.6207607984542847,
      "learning_rate": 3.987962635377806e-06,
      "loss": 1.2875,
      "step": 25480
    },
    {
      "epoch": 0.9125288735294644,
      "grad_norm": 1.4966524839401245,
      "learning_rate": 3.98472034848042e-06,
      "loss": 1.3703,
      "step": 25481
    },
    {
      "epoch": 0.9125646856590327,
      "grad_norm": 1.418821096420288,
      "learning_rate": 3.9814793533447635e-06,
      "loss": 1.2243,
      "step": 25482
    },
    {
      "epoch": 0.912600497788601,
      "grad_norm": 1.7261139154434204,
      "learning_rate": 3.978239650014437e-06,
      "loss": 1.4594,
      "step": 25483
    },
    {
      "epoch": 0.9126363099181692,
      "grad_norm": 1.719135046005249,
      "learning_rate": 3.975001238533038e-06,
      "loss": 1.5768,
      "step": 25484
    },
    {
      "epoch": 0.9126721220477376,
      "grad_norm": 1.6824172735214233,
      "learning_rate": 3.971764118944155e-06,
      "loss": 1.45,
      "step": 25485
    },
    {
      "epoch": 0.9127079341773059,
      "grad_norm": 1.4959074258804321,
      "learning_rate": 3.968528291291296e-06,
      "loss": 1.1142,
      "step": 25486
    },
    {
      "epoch": 0.9127437463068742,
      "grad_norm": 1.5593359470367432,
      "learning_rate": 3.965293755618027e-06,
      "loss": 1.3833,
      "step": 25487
    },
    {
      "epoch": 0.9127795584364424,
      "grad_norm": 1.45145845413208,
      "learning_rate": 3.962060511967846e-06,
      "loss": 1.1292,
      "step": 25488
    },
    {
      "epoch": 0.9128153705660107,
      "grad_norm": 1.793111801147461,
      "learning_rate": 3.9588285603842755e-06,
      "loss": 1.384,
      "step": 25489
    },
    {
      "epoch": 0.912851182695579,
      "grad_norm": 1.5112842321395874,
      "learning_rate": 3.955597900910768e-06,
      "loss": 1.4585,
      "step": 25490
    },
    {
      "epoch": 0.9128869948251472,
      "grad_norm": 1.887883186340332,
      "learning_rate": 3.9523685335908e-06,
      "loss": 1.5188,
      "step": 25491
    },
    {
      "epoch": 0.9129228069547156,
      "grad_norm": 1.8775691986083984,
      "learning_rate": 3.9491404584678485e-06,
      "loss": 1.4276,
      "step": 25492
    },
    {
      "epoch": 0.9129586190842839,
      "grad_norm": 1.926689863204956,
      "learning_rate": 3.945913675585289e-06,
      "loss": 1.4073,
      "step": 25493
    },
    {
      "epoch": 0.9129944312138522,
      "grad_norm": 1.671813726425171,
      "learning_rate": 3.9426881849865646e-06,
      "loss": 1.4181,
      "step": 25494
    },
    {
      "epoch": 0.9130302433434204,
      "grad_norm": 1.5096828937530518,
      "learning_rate": 3.939463986715064e-06,
      "loss": 1.4566,
      "step": 25495
    },
    {
      "epoch": 0.9130660554729887,
      "grad_norm": 1.3902485370635986,
      "learning_rate": 3.936241080814174e-06,
      "loss": 1.2241,
      "step": 25496
    },
    {
      "epoch": 0.913101867602557,
      "grad_norm": 1.5125646591186523,
      "learning_rate": 3.933019467327248e-06,
      "loss": 1.548,
      "step": 25497
    },
    {
      "epoch": 0.9131376797321252,
      "grad_norm": 1.8359897136688232,
      "learning_rate": 3.9297991462976196e-06,
      "loss": 1.5648,
      "step": 25498
    },
    {
      "epoch": 0.9131734918616936,
      "grad_norm": 1.5971086025238037,
      "learning_rate": 3.92658011776863e-06,
      "loss": 1.3288,
      "step": 25499
    },
    {
      "epoch": 0.9132093039912619,
      "grad_norm": 2.038939952850342,
      "learning_rate": 3.923362381783568e-06,
      "loss": 1.2699,
      "step": 25500
    },
    {
      "epoch": 0.9132451161208301,
      "grad_norm": 2.618192195892334,
      "learning_rate": 3.920145938385744e-06,
      "loss": 1.3613,
      "step": 25501
    },
    {
      "epoch": 0.9132809282503984,
      "grad_norm": 1.5217925310134888,
      "learning_rate": 3.916930787618412e-06,
      "loss": 1.46,
      "step": 25502
    },
    {
      "epoch": 0.9133167403799667,
      "grad_norm": 2.0505995750427246,
      "learning_rate": 3.913716929524857e-06,
      "loss": 1.23,
      "step": 25503
    },
    {
      "epoch": 0.913352552509535,
      "grad_norm": 1.67502760887146,
      "learning_rate": 3.910504364148282e-06,
      "loss": 1.3907,
      "step": 25504
    },
    {
      "epoch": 0.9133883646391032,
      "grad_norm": 1.720183253288269,
      "learning_rate": 3.907293091531927e-06,
      "loss": 1.1388,
      "step": 25505
    },
    {
      "epoch": 0.9134241767686716,
      "grad_norm": 1.7315115928649902,
      "learning_rate": 3.904083111718993e-06,
      "loss": 1.2282,
      "step": 25506
    },
    {
      "epoch": 0.9134599888982399,
      "grad_norm": 1.504867672920227,
      "learning_rate": 3.900874424752677e-06,
      "loss": 1.4048,
      "step": 25507
    },
    {
      "epoch": 0.9134958010278081,
      "grad_norm": 1.2139843702316284,
      "learning_rate": 3.897667030676133e-06,
      "loss": 1.311,
      "step": 25508
    },
    {
      "epoch": 0.9135316131573764,
      "grad_norm": 1.6019784212112427,
      "learning_rate": 3.8944609295324955e-06,
      "loss": 1.394,
      "step": 25509
    },
    {
      "epoch": 0.9135674252869447,
      "grad_norm": 1.34242582321167,
      "learning_rate": 3.89125612136495e-06,
      "loss": 1.1476,
      "step": 25510
    },
    {
      "epoch": 0.913603237416513,
      "grad_norm": 1.3961800336837769,
      "learning_rate": 3.888052606216564e-06,
      "loss": 1.481,
      "step": 25511
    },
    {
      "epoch": 0.9136390495460812,
      "grad_norm": 1.6185824871063232,
      "learning_rate": 3.884850384130456e-06,
      "loss": 1.1685,
      "step": 25512
    },
    {
      "epoch": 0.9136748616756496,
      "grad_norm": 2.0552761554718018,
      "learning_rate": 3.881649455149694e-06,
      "loss": 1.3966,
      "step": 25513
    },
    {
      "epoch": 0.9137106738052179,
      "grad_norm": 1.247956395149231,
      "learning_rate": 3.878449819317376e-06,
      "loss": 1.1779,
      "step": 25514
    },
    {
      "epoch": 0.9137464859347861,
      "grad_norm": 1.512967586517334,
      "learning_rate": 3.875251476676522e-06,
      "loss": 1.199,
      "step": 25515
    },
    {
      "epoch": 0.9137822980643544,
      "grad_norm": 1.5441175699234009,
      "learning_rate": 3.872054427270167e-06,
      "loss": 1.108,
      "step": 25516
    },
    {
      "epoch": 0.9138181101939227,
      "grad_norm": 2.1521899700164795,
      "learning_rate": 3.868858671141329e-06,
      "loss": 1.6401,
      "step": 25517
    },
    {
      "epoch": 0.9138539223234909,
      "grad_norm": 2.341466188430786,
      "learning_rate": 3.865664208332986e-06,
      "loss": 1.3783,
      "step": 25518
    },
    {
      "epoch": 0.9138897344530592,
      "grad_norm": 1.7183805704116821,
      "learning_rate": 3.862471038888138e-06,
      "loss": 1.2954,
      "step": 25519
    },
    {
      "epoch": 0.9139255465826276,
      "grad_norm": 1.7073190212249756,
      "learning_rate": 3.859279162849716e-06,
      "loss": 1.6382,
      "step": 25520
    },
    {
      "epoch": 0.9139613587121959,
      "grad_norm": 1.8298503160476685,
      "learning_rate": 3.856088580260697e-06,
      "loss": 1.6447,
      "step": 25521
    },
    {
      "epoch": 0.9139971708417641,
      "grad_norm": 1.38832688331604,
      "learning_rate": 3.8528992911639806e-06,
      "loss": 1.3147,
      "step": 25522
    },
    {
      "epoch": 0.9140329829713324,
      "grad_norm": 1.9283418655395508,
      "learning_rate": 3.8497112956024875e-06,
      "loss": 1.2106,
      "step": 25523
    },
    {
      "epoch": 0.9140687951009007,
      "grad_norm": 1.6704370975494385,
      "learning_rate": 3.846524593619094e-06,
      "loss": 1.6686,
      "step": 25524
    },
    {
      "epoch": 0.9141046072304689,
      "grad_norm": 2.005181074142456,
      "learning_rate": 3.8433391852567e-06,
      "loss": 1.9029,
      "step": 25525
    },
    {
      "epoch": 0.9141404193600372,
      "grad_norm": 1.8579343557357788,
      "learning_rate": 3.840155070558149e-06,
      "loss": 1.2547,
      "step": 25526
    },
    {
      "epoch": 0.9141762314896056,
      "grad_norm": 1.4794865846633911,
      "learning_rate": 3.836972249566239e-06,
      "loss": 1.1888,
      "step": 25527
    },
    {
      "epoch": 0.9142120436191739,
      "grad_norm": 1.6024411916732788,
      "learning_rate": 3.83379072232386e-06,
      "loss": 1.2923,
      "step": 25528
    },
    {
      "epoch": 0.9142478557487421,
      "grad_norm": 1.8010493516921997,
      "learning_rate": 3.830610488873765e-06,
      "loss": 1.2005,
      "step": 25529
    },
    {
      "epoch": 0.9142836678783104,
      "grad_norm": 1.4514111280441284,
      "learning_rate": 3.827431549258764e-06,
      "loss": 1.6173,
      "step": 25530
    },
    {
      "epoch": 0.9143194800078787,
      "grad_norm": 1.4556541442871094,
      "learning_rate": 3.824253903521602e-06,
      "loss": 1.3869,
      "step": 25531
    },
    {
      "epoch": 0.9143552921374469,
      "grad_norm": 1.7731586694717407,
      "learning_rate": 3.821077551705065e-06,
      "loss": 1.557,
      "step": 25532
    },
    {
      "epoch": 0.9143911042670152,
      "grad_norm": 1.5040415525436401,
      "learning_rate": 3.817902493851877e-06,
      "loss": 1.1799,
      "step": 25533
    },
    {
      "epoch": 0.9144269163965836,
      "grad_norm": 1.508422613143921,
      "learning_rate": 3.814728730004724e-06,
      "loss": 1.3727,
      "step": 25534
    },
    {
      "epoch": 0.9144627285261518,
      "grad_norm": 1.460686206817627,
      "learning_rate": 3.811556260206328e-06,
      "loss": 1.3637,
      "step": 25535
    },
    {
      "epoch": 0.9144985406557201,
      "grad_norm": 2.1639134883880615,
      "learning_rate": 3.808385084499366e-06,
      "loss": 1.4715,
      "step": 25536
    },
    {
      "epoch": 0.9145343527852884,
      "grad_norm": 1.949466586112976,
      "learning_rate": 3.8052152029265154e-06,
      "loss": 1.3316,
      "step": 25537
    },
    {
      "epoch": 0.9145701649148567,
      "grad_norm": 1.9992917776107788,
      "learning_rate": 3.8020466155304078e-06,
      "loss": 1.3247,
      "step": 25538
    },
    {
      "epoch": 0.9146059770444249,
      "grad_norm": 1.9052329063415527,
      "learning_rate": 3.798879322353666e-06,
      "loss": 1.5343,
      "step": 25539
    },
    {
      "epoch": 0.9146417891739932,
      "grad_norm": 1.6159497499465942,
      "learning_rate": 3.7957133234389207e-06,
      "loss": 1.5618,
      "step": 25540
    },
    {
      "epoch": 0.9146776013035616,
      "grad_norm": 1.5084832906723022,
      "learning_rate": 3.7925486188287727e-06,
      "loss": 1.4374,
      "step": 25541
    },
    {
      "epoch": 0.9147134134331298,
      "grad_norm": 1.6207736730575562,
      "learning_rate": 3.7893852085657657e-06,
      "loss": 1.5015,
      "step": 25542
    },
    {
      "epoch": 0.9147492255626981,
      "grad_norm": 1.9216586351394653,
      "learning_rate": 3.786223092692476e-06,
      "loss": 1.6365,
      "step": 25543
    },
    {
      "epoch": 0.9147850376922664,
      "grad_norm": 1.3661115169525146,
      "learning_rate": 3.7830622712514696e-06,
      "loss": 1.3516,
      "step": 25544
    },
    {
      "epoch": 0.9148208498218346,
      "grad_norm": 1.7824134826660156,
      "learning_rate": 3.779902744285224e-06,
      "loss": 1.0715,
      "step": 25545
    },
    {
      "epoch": 0.9148566619514029,
      "grad_norm": 2.196190357208252,
      "learning_rate": 3.7767445118362832e-06,
      "loss": 1.1834,
      "step": 25546
    },
    {
      "epoch": 0.9148924740809712,
      "grad_norm": 1.838174819946289,
      "learning_rate": 3.7735875739471237e-06,
      "loss": 1.417,
      "step": 25547
    },
    {
      "epoch": 0.9149282862105395,
      "grad_norm": 1.4311494827270508,
      "learning_rate": 3.770431930660223e-06,
      "loss": 1.3594,
      "step": 25548
    },
    {
      "epoch": 0.9149640983401078,
      "grad_norm": 1.6186925172805786,
      "learning_rate": 3.767277582018036e-06,
      "loss": 1.2366,
      "step": 25549
    },
    {
      "epoch": 0.9149999104696761,
      "grad_norm": 1.7173895835876465,
      "learning_rate": 3.7641245280629842e-06,
      "loss": 1.3939,
      "step": 25550
    },
    {
      "epoch": 0.9150357225992444,
      "grad_norm": 2.2020442485809326,
      "learning_rate": 3.760972768837523e-06,
      "loss": 1.4908,
      "step": 25551
    },
    {
      "epoch": 0.9150715347288126,
      "grad_norm": 1.694572925567627,
      "learning_rate": 3.757822304384018e-06,
      "loss": 1.3844,
      "step": 25552
    },
    {
      "epoch": 0.9151073468583809,
      "grad_norm": 1.406884789466858,
      "learning_rate": 3.7546731347448685e-06,
      "loss": 1.3992,
      "step": 25553
    },
    {
      "epoch": 0.9151431589879492,
      "grad_norm": 1.7870287895202637,
      "learning_rate": 3.7515252599624516e-06,
      "loss": 1.3392,
      "step": 25554
    },
    {
      "epoch": 0.9151789711175174,
      "grad_norm": 1.6366087198257446,
      "learning_rate": 3.748378680079112e-06,
      "loss": 1.3854,
      "step": 25555
    },
    {
      "epoch": 0.9152147832470858,
      "grad_norm": 1.4431771039962769,
      "learning_rate": 3.745233395137182e-06,
      "loss": 1.4213,
      "step": 25556
    },
    {
      "epoch": 0.9152505953766541,
      "grad_norm": 1.887716293334961,
      "learning_rate": 3.7420894051789723e-06,
      "loss": 1.394,
      "step": 25557
    },
    {
      "epoch": 0.9152864075062224,
      "grad_norm": 1.6417644023895264,
      "learning_rate": 3.7389467102467823e-06,
      "loss": 1.3629,
      "step": 25558
    },
    {
      "epoch": 0.9153222196357906,
      "grad_norm": 1.8466694355010986,
      "learning_rate": 3.7358053103829117e-06,
      "loss": 1.4253,
      "step": 25559
    },
    {
      "epoch": 0.9153580317653589,
      "grad_norm": 1.1880935430526733,
      "learning_rate": 3.732665205629593e-06,
      "loss": 1.306,
      "step": 25560
    },
    {
      "epoch": 0.9153938438949272,
      "grad_norm": 1.6200544834136963,
      "learning_rate": 3.7295263960290927e-06,
      "loss": 1.3328,
      "step": 25561
    },
    {
      "epoch": 0.9154296560244954,
      "grad_norm": 1.4610018730163574,
      "learning_rate": 3.7263888816236435e-06,
      "loss": 0.9073,
      "step": 25562
    },
    {
      "epoch": 0.9154654681540638,
      "grad_norm": 1.679741621017456,
      "learning_rate": 3.7232526624554344e-06,
      "loss": 1.6995,
      "step": 25563
    },
    {
      "epoch": 0.9155012802836321,
      "grad_norm": 1.569359540939331,
      "learning_rate": 3.720117738566675e-06,
      "loss": 1.2822,
      "step": 25564
    },
    {
      "epoch": 0.9155370924132004,
      "grad_norm": 2.0695765018463135,
      "learning_rate": 3.7169841099995438e-06,
      "loss": 1.6033,
      "step": 25565
    },
    {
      "epoch": 0.9155729045427686,
      "grad_norm": 1.6272145509719849,
      "learning_rate": 3.7138517767961954e-06,
      "loss": 1.2955,
      "step": 25566
    },
    {
      "epoch": 0.9156087166723369,
      "grad_norm": 1.9181761741638184,
      "learning_rate": 3.710720738998774e-06,
      "loss": 1.4108,
      "step": 25567
    },
    {
      "epoch": 0.9156445288019052,
      "grad_norm": 2.273002862930298,
      "learning_rate": 3.7075909966493903e-06,
      "loss": 1.6432,
      "step": 25568
    },
    {
      "epoch": 0.9156803409314734,
      "grad_norm": 1.7246898412704468,
      "learning_rate": 3.7044625497901774e-06,
      "loss": 1.2398,
      "step": 25569
    },
    {
      "epoch": 0.9157161530610418,
      "grad_norm": 1.4257093667984009,
      "learning_rate": 3.7013353984631906e-06,
      "loss": 1.7232,
      "step": 25570
    },
    {
      "epoch": 0.9157519651906101,
      "grad_norm": 1.7329950332641602,
      "learning_rate": 3.698209542710529e-06,
      "loss": 1.4783,
      "step": 25571
    },
    {
      "epoch": 0.9157877773201784,
      "grad_norm": 1.9357683658599854,
      "learning_rate": 3.6950849825742375e-06,
      "loss": 1.4423,
      "step": 25572
    },
    {
      "epoch": 0.9158235894497466,
      "grad_norm": 2.051203489303589,
      "learning_rate": 3.6919617180963595e-06,
      "loss": 1.6041,
      "step": 25573
    },
    {
      "epoch": 0.9158594015793149,
      "grad_norm": 1.6284784078598022,
      "learning_rate": 3.6888397493188954e-06,
      "loss": 1.5151,
      "step": 25574
    },
    {
      "epoch": 0.9158952137088832,
      "grad_norm": 1.5720276832580566,
      "learning_rate": 3.685719076283867e-06,
      "loss": 1.398,
      "step": 25575
    },
    {
      "epoch": 0.9159310258384514,
      "grad_norm": 1.6514610052108765,
      "learning_rate": 3.68259969903324e-06,
      "loss": 1.3815,
      "step": 25576
    },
    {
      "epoch": 0.9159668379680198,
      "grad_norm": 1.2488956451416016,
      "learning_rate": 3.6794816176090152e-06,
      "loss": 1.3639,
      "step": 25577
    },
    {
      "epoch": 0.9160026500975881,
      "grad_norm": 2.5675435066223145,
      "learning_rate": 3.676364832053103e-06,
      "loss": 1.2767,
      "step": 25578
    },
    {
      "epoch": 0.9160384622271563,
      "grad_norm": 1.3877888917922974,
      "learning_rate": 3.6732493424074587e-06,
      "loss": 1.2629,
      "step": 25579
    },
    {
      "epoch": 0.9160742743567246,
      "grad_norm": 1.5578382015228271,
      "learning_rate": 3.6701351487140046e-06,
      "loss": 1.5932,
      "step": 25580
    },
    {
      "epoch": 0.9161100864862929,
      "grad_norm": 2.35837459564209,
      "learning_rate": 3.667022251014607e-06,
      "loss": 1.3385,
      "step": 25581
    },
    {
      "epoch": 0.9161458986158612,
      "grad_norm": 1.7006118297576904,
      "learning_rate": 3.6639106493511766e-06,
      "loss": 1.262,
      "step": 25582
    },
    {
      "epoch": 0.9161817107454294,
      "grad_norm": 1.6541268825531006,
      "learning_rate": 3.660800343765547e-06,
      "loss": 1.4949,
      "step": 25583
    },
    {
      "epoch": 0.9162175228749978,
      "grad_norm": 1.5310859680175781,
      "learning_rate": 3.657691334299607e-06,
      "loss": 1.6051,
      "step": 25584
    },
    {
      "epoch": 0.9162533350045661,
      "grad_norm": 1.7938069105148315,
      "learning_rate": 3.6545836209951333e-06,
      "loss": 1.4242,
      "step": 25585
    },
    {
      "epoch": 0.9162891471341343,
      "grad_norm": 1.6601662635803223,
      "learning_rate": 3.6514772038939714e-06,
      "loss": 1.4616,
      "step": 25586
    },
    {
      "epoch": 0.9163249592637026,
      "grad_norm": 2.1517858505249023,
      "learning_rate": 3.64837208303791e-06,
      "loss": 1.5088,
      "step": 25587
    },
    {
      "epoch": 0.9163607713932709,
      "grad_norm": 1.3455365896224976,
      "learning_rate": 3.6452682584687035e-06,
      "loss": 1.12,
      "step": 25588
    },
    {
      "epoch": 0.9163965835228391,
      "grad_norm": 1.8301458358764648,
      "learning_rate": 3.642165730228131e-06,
      "loss": 1.3795,
      "step": 25589
    },
    {
      "epoch": 0.9164323956524074,
      "grad_norm": 1.7565302848815918,
      "learning_rate": 3.6390644983579135e-06,
      "loss": 1.4932,
      "step": 25590
    },
    {
      "epoch": 0.9164682077819758,
      "grad_norm": 1.7920565605163574,
      "learning_rate": 3.6359645628998073e-06,
      "loss": 1.5721,
      "step": 25591
    },
    {
      "epoch": 0.9165040199115441,
      "grad_norm": 1.578942894935608,
      "learning_rate": 3.6328659238954897e-06,
      "loss": 1.3415,
      "step": 25592
    },
    {
      "epoch": 0.9165398320411123,
      "grad_norm": 1.4769006967544556,
      "learning_rate": 3.62976858138665e-06,
      "loss": 1.2929,
      "step": 25593
    },
    {
      "epoch": 0.9165756441706806,
      "grad_norm": 1.8735318183898926,
      "learning_rate": 3.6266725354149656e-06,
      "loss": 1.7607,
      "step": 25594
    },
    {
      "epoch": 0.9166114563002489,
      "grad_norm": 1.9881622791290283,
      "learning_rate": 3.6235777860221033e-06,
      "loss": 1.5311,
      "step": 25595
    },
    {
      "epoch": 0.9166472684298171,
      "grad_norm": 1.5557109117507935,
      "learning_rate": 3.620484333249674e-06,
      "loss": 1.3811,
      "step": 25596
    },
    {
      "epoch": 0.9166830805593854,
      "grad_norm": 1.9754571914672852,
      "learning_rate": 3.6173921771393003e-06,
      "loss": 1.4716,
      "step": 25597
    },
    {
      "epoch": 0.9167188926889538,
      "grad_norm": 1.7173326015472412,
      "learning_rate": 3.6143013177326046e-06,
      "loss": 1.6932,
      "step": 25598
    },
    {
      "epoch": 0.9167547048185221,
      "grad_norm": 1.4891917705535889,
      "learning_rate": 3.611211755071142e-06,
      "loss": 1.4409,
      "step": 25599
    },
    {
      "epoch": 0.9167905169480903,
      "grad_norm": 2.2099905014038086,
      "learning_rate": 3.608123489196502e-06,
      "loss": 1.2602,
      "step": 25600
    },
    {
      "epoch": 0.9168263290776586,
      "grad_norm": 1.5880131721496582,
      "learning_rate": 3.605036520150218e-06,
      "loss": 1.3322,
      "step": 25601
    },
    {
      "epoch": 0.9168621412072269,
      "grad_norm": 1.5003440380096436,
      "learning_rate": 3.601950847973845e-06,
      "loss": 1.5016,
      "step": 25602
    },
    {
      "epoch": 0.9168979533367951,
      "grad_norm": 1.5447492599487305,
      "learning_rate": 3.598866472708862e-06,
      "loss": 1.4572,
      "step": 25603
    },
    {
      "epoch": 0.9169337654663634,
      "grad_norm": 1.4146647453308105,
      "learning_rate": 3.595783394396779e-06,
      "loss": 1.418,
      "step": 25604
    },
    {
      "epoch": 0.9169695775959318,
      "grad_norm": 1.7357875108718872,
      "learning_rate": 3.592701613079097e-06,
      "loss": 1.1391,
      "step": 25605
    },
    {
      "epoch": 0.9170053897255,
      "grad_norm": 1.6281183958053589,
      "learning_rate": 3.5896211287972383e-06,
      "loss": 1.4511,
      "step": 25606
    },
    {
      "epoch": 0.9170412018550683,
      "grad_norm": 1.7715595960617065,
      "learning_rate": 3.5865419415926803e-06,
      "loss": 1.5196,
      "step": 25607
    },
    {
      "epoch": 0.9170770139846366,
      "grad_norm": 1.6238850355148315,
      "learning_rate": 3.583464051506813e-06,
      "loss": 1.3839,
      "step": 25608
    },
    {
      "epoch": 0.9171128261142049,
      "grad_norm": 1.6311407089233398,
      "learning_rate": 3.5803874585811024e-06,
      "loss": 1.4173,
      "step": 25609
    },
    {
      "epoch": 0.9171486382437731,
      "grad_norm": 1.4624392986297607,
      "learning_rate": 3.577312162856883e-06,
      "loss": 1.3834,
      "step": 25610
    },
    {
      "epoch": 0.9171844503733414,
      "grad_norm": 1.55482816696167,
      "learning_rate": 3.574238164375554e-06,
      "loss": 1.433,
      "step": 25611
    },
    {
      "epoch": 0.9172202625029098,
      "grad_norm": 1.4993577003479004,
      "learning_rate": 3.571165463178472e-06,
      "loss": 1.4662,
      "step": 25612
    },
    {
      "epoch": 0.917256074632478,
      "grad_norm": 1.3473643064498901,
      "learning_rate": 3.568094059306981e-06,
      "loss": 1.401,
      "step": 25613
    },
    {
      "epoch": 0.9172918867620463,
      "grad_norm": 2.217761516571045,
      "learning_rate": 3.5650239528024043e-06,
      "loss": 1.266,
      "step": 25614
    },
    {
      "epoch": 0.9173276988916146,
      "grad_norm": 1.6229770183563232,
      "learning_rate": 3.5619551437060083e-06,
      "loss": 1.5986,
      "step": 25615
    },
    {
      "epoch": 0.9173635110211829,
      "grad_norm": 1.9005887508392334,
      "learning_rate": 3.558887632059138e-06,
      "loss": 0.9255,
      "step": 25616
    },
    {
      "epoch": 0.9173993231507511,
      "grad_norm": 1.1790728569030762,
      "learning_rate": 3.555821417903027e-06,
      "loss": 1.1397,
      "step": 25617
    },
    {
      "epoch": 0.9174351352803194,
      "grad_norm": 1.6984502077102661,
      "learning_rate": 3.552756501278931e-06,
      "loss": 1.413,
      "step": 25618
    },
    {
      "epoch": 0.9174709474098878,
      "grad_norm": 1.8751317262649536,
      "learning_rate": 3.549692882228084e-06,
      "loss": 1.6012,
      "step": 25619
    },
    {
      "epoch": 0.917506759539456,
      "grad_norm": 1.5878108739852905,
      "learning_rate": 3.5466305607917195e-06,
      "loss": 1.2097,
      "step": 25620
    },
    {
      "epoch": 0.9175425716690243,
      "grad_norm": 1.529499888420105,
      "learning_rate": 3.5435695370110154e-06,
      "loss": 1.5132,
      "step": 25621
    },
    {
      "epoch": 0.9175783837985926,
      "grad_norm": 1.7148410081863403,
      "learning_rate": 3.540509810927173e-06,
      "loss": 1.4484,
      "step": 25622
    },
    {
      "epoch": 0.9176141959281608,
      "grad_norm": 1.3747161626815796,
      "learning_rate": 3.537451382581336e-06,
      "loss": 1.1541,
      "step": 25623
    },
    {
      "epoch": 0.9176500080577291,
      "grad_norm": 1.7710306644439697,
      "learning_rate": 3.534394252014661e-06,
      "loss": 1.4204,
      "step": 25624
    },
    {
      "epoch": 0.9176858201872974,
      "grad_norm": 1.4869321584701538,
      "learning_rate": 3.531338419268293e-06,
      "loss": 1.6908,
      "step": 25625
    },
    {
      "epoch": 0.9177216323168658,
      "grad_norm": 1.8567776679992676,
      "learning_rate": 3.52828388438331e-06,
      "loss": 1.131,
      "step": 25626
    },
    {
      "epoch": 0.917757444446434,
      "grad_norm": 1.6301764249801636,
      "learning_rate": 3.5252306474008457e-06,
      "loss": 1.3344,
      "step": 25627
    },
    {
      "epoch": 0.9177932565760023,
      "grad_norm": 1.6216933727264404,
      "learning_rate": 3.522178708361956e-06,
      "loss": 1.4663,
      "step": 25628
    },
    {
      "epoch": 0.9178290687055706,
      "grad_norm": 1.4409526586532593,
      "learning_rate": 3.5191280673077086e-06,
      "loss": 1.3619,
      "step": 25629
    },
    {
      "epoch": 0.9178648808351388,
      "grad_norm": 2.590237855911255,
      "learning_rate": 3.516078724279137e-06,
      "loss": 1.1399,
      "step": 25630
    },
    {
      "epoch": 0.9179006929647071,
      "grad_norm": 2.0231406688690186,
      "learning_rate": 3.513030679317264e-06,
      "loss": 1.4057,
      "step": 25631
    },
    {
      "epoch": 0.9179365050942754,
      "grad_norm": 1.7201601266860962,
      "learning_rate": 3.5099839324631233e-06,
      "loss": 1.4519,
      "step": 25632
    },
    {
      "epoch": 0.9179723172238438,
      "grad_norm": 1.846491813659668,
      "learning_rate": 3.506938483757671e-06,
      "loss": 1.2908,
      "step": 25633
    },
    {
      "epoch": 0.918008129353412,
      "grad_norm": 2.1384713649749756,
      "learning_rate": 3.503894333241886e-06,
      "loss": 1.3344,
      "step": 25634
    },
    {
      "epoch": 0.9180439414829803,
      "grad_norm": 1.8844846487045288,
      "learning_rate": 3.500851480956746e-06,
      "loss": 1.228,
      "step": 25635
    },
    {
      "epoch": 0.9180797536125486,
      "grad_norm": 2.2950620651245117,
      "learning_rate": 3.497809926943174e-06,
      "loss": 1.5764,
      "step": 25636
    },
    {
      "epoch": 0.9181155657421168,
      "grad_norm": 1.582805871963501,
      "learning_rate": 3.4947696712420708e-06,
      "loss": 1.4444,
      "step": 25637
    },
    {
      "epoch": 0.9181513778716851,
      "grad_norm": 1.5671573877334595,
      "learning_rate": 3.491730713894381e-06,
      "loss": 1.3715,
      "step": 25638
    },
    {
      "epoch": 0.9181871900012534,
      "grad_norm": 1.5163581371307373,
      "learning_rate": 3.4886930549409724e-06,
      "loss": 1.3593,
      "step": 25639
    },
    {
      "epoch": 0.9182230021308218,
      "grad_norm": 1.5615180730819702,
      "learning_rate": 3.485656694422701e-06,
      "loss": 1.1561,
      "step": 25640
    },
    {
      "epoch": 0.91825881426039,
      "grad_norm": 1.5961110591888428,
      "learning_rate": 3.482621632380412e-06,
      "loss": 1.4103,
      "step": 25641
    },
    {
      "epoch": 0.9182946263899583,
      "grad_norm": 1.2680072784423828,
      "learning_rate": 3.479587868854961e-06,
      "loss": 1.6213,
      "step": 25642
    },
    {
      "epoch": 0.9183304385195266,
      "grad_norm": 1.8912404775619507,
      "learning_rate": 3.4765554038871607e-06,
      "loss": 1.2256,
      "step": 25643
    },
    {
      "epoch": 0.9183662506490948,
      "grad_norm": 1.5284357070922852,
      "learning_rate": 3.4735242375177777e-06,
      "loss": 1.577,
      "step": 25644
    },
    {
      "epoch": 0.9184020627786631,
      "grad_norm": 1.3570201396942139,
      "learning_rate": 3.470494369787636e-06,
      "loss": 1.508,
      "step": 25645
    },
    {
      "epoch": 0.9184378749082314,
      "grad_norm": 2.326610803604126,
      "learning_rate": 3.4674658007374683e-06,
      "loss": 1.3932,
      "step": 25646
    },
    {
      "epoch": 0.9184736870377997,
      "grad_norm": 1.5897072553634644,
      "learning_rate": 3.464438530408043e-06,
      "loss": 1.6262,
      "step": 25647
    },
    {
      "epoch": 0.918509499167368,
      "grad_norm": 1.651442289352417,
      "learning_rate": 3.46141255884006e-06,
      "loss": 1.0848,
      "step": 25648
    },
    {
      "epoch": 0.9185453112969363,
      "grad_norm": 1.5025795698165894,
      "learning_rate": 3.4583878860742434e-06,
      "loss": 1.5907,
      "step": 25649
    },
    {
      "epoch": 0.9185811234265046,
      "grad_norm": 2.048462152481079,
      "learning_rate": 3.4553645121513046e-06,
      "loss": 1.8224,
      "step": 25650
    },
    {
      "epoch": 0.9186169355560728,
      "grad_norm": 1.4465118646621704,
      "learning_rate": 3.4523424371118885e-06,
      "loss": 1.4116,
      "step": 25651
    },
    {
      "epoch": 0.9186527476856411,
      "grad_norm": 2.2787973880767822,
      "learning_rate": 3.449321660996674e-06,
      "loss": 1.4564,
      "step": 25652
    },
    {
      "epoch": 0.9186885598152094,
      "grad_norm": 1.6836674213409424,
      "learning_rate": 3.446302183846295e-06,
      "loss": 1.1173,
      "step": 25653
    },
    {
      "epoch": 0.9187243719447777,
      "grad_norm": 1.779667615890503,
      "learning_rate": 3.443284005701375e-06,
      "loss": 1.5232,
      "step": 25654
    },
    {
      "epoch": 0.918760184074346,
      "grad_norm": 1.2427458763122559,
      "learning_rate": 3.4402671266025253e-06,
      "loss": 1.2979,
      "step": 25655
    },
    {
      "epoch": 0.9187959962039143,
      "grad_norm": 1.405266523361206,
      "learning_rate": 3.4372515465903145e-06,
      "loss": 1.5639,
      "step": 25656
    },
    {
      "epoch": 0.9188318083334825,
      "grad_norm": 1.3440146446228027,
      "learning_rate": 3.434237265705342e-06,
      "loss": 1.3727,
      "step": 25657
    },
    {
      "epoch": 0.9188676204630508,
      "grad_norm": 1.969915747642517,
      "learning_rate": 3.4312242839881325e-06,
      "loss": 1.0968,
      "step": 25658
    },
    {
      "epoch": 0.9189034325926191,
      "grad_norm": 1.9113439321517944,
      "learning_rate": 3.4282126014792414e-06,
      "loss": 1.6633,
      "step": 25659
    },
    {
      "epoch": 0.9189392447221874,
      "grad_norm": 1.5810271501541138,
      "learning_rate": 3.4252022182191813e-06,
      "loss": 1.4188,
      "step": 25660
    },
    {
      "epoch": 0.9189750568517557,
      "grad_norm": 2.1815831661224365,
      "learning_rate": 3.4221931342484525e-06,
      "loss": 1.7841,
      "step": 25661
    },
    {
      "epoch": 0.919010868981324,
      "grad_norm": 1.472642183303833,
      "learning_rate": 3.4191853496075343e-06,
      "loss": 1.4584,
      "step": 25662
    },
    {
      "epoch": 0.9190466811108923,
      "grad_norm": 1.7633564472198486,
      "learning_rate": 3.4161788643369052e-06,
      "loss": 1.2982,
      "step": 25663
    },
    {
      "epoch": 0.9190824932404605,
      "grad_norm": 2.459702730178833,
      "learning_rate": 3.4131736784769996e-06,
      "loss": 1.5756,
      "step": 25664
    },
    {
      "epoch": 0.9191183053700288,
      "grad_norm": 1.4552972316741943,
      "learning_rate": 3.410169792068263e-06,
      "loss": 1.2331,
      "step": 25665
    },
    {
      "epoch": 0.9191541174995971,
      "grad_norm": 1.558521032333374,
      "learning_rate": 3.407167205151085e-06,
      "loss": 1.194,
      "step": 25666
    },
    {
      "epoch": 0.9191899296291653,
      "grad_norm": 1.9817105531692505,
      "learning_rate": 3.404165917765889e-06,
      "loss": 1.4798,
      "step": 25667
    },
    {
      "epoch": 0.9192257417587337,
      "grad_norm": 1.6500427722930908,
      "learning_rate": 3.401165929953043e-06,
      "loss": 1.4384,
      "step": 25668
    },
    {
      "epoch": 0.919261553888302,
      "grad_norm": 1.5914006233215332,
      "learning_rate": 3.398167241752892e-06,
      "loss": 1.521,
      "step": 25669
    },
    {
      "epoch": 0.9192973660178703,
      "grad_norm": 1.3515881299972534,
      "learning_rate": 3.395169853205793e-06,
      "loss": 1.5902,
      "step": 25670
    },
    {
      "epoch": 0.9193331781474385,
      "grad_norm": 1.7092448472976685,
      "learning_rate": 3.3921737643520803e-06,
      "loss": 1.537,
      "step": 25671
    },
    {
      "epoch": 0.9193689902770068,
      "grad_norm": 1.3170454502105713,
      "learning_rate": 3.3891789752320656e-06,
      "loss": 1.3537,
      "step": 25672
    },
    {
      "epoch": 0.9194048024065751,
      "grad_norm": 2.0973784923553467,
      "learning_rate": 3.3861854858860177e-06,
      "loss": 1.5353,
      "step": 25673
    },
    {
      "epoch": 0.9194406145361433,
      "grad_norm": 1.5266790390014648,
      "learning_rate": 3.3831932963542147e-06,
      "loss": 1.3477,
      "step": 25674
    },
    {
      "epoch": 0.9194764266657117,
      "grad_norm": 1.6230286359786987,
      "learning_rate": 3.3802024066769355e-06,
      "loss": 1.3931,
      "step": 25675
    },
    {
      "epoch": 0.91951223879528,
      "grad_norm": 2.351484775543213,
      "learning_rate": 3.3772128168943816e-06,
      "loss": 1.5749,
      "step": 25676
    },
    {
      "epoch": 0.9195480509248483,
      "grad_norm": 2.194748878479004,
      "learning_rate": 3.37422452704681e-06,
      "loss": 1.4503,
      "step": 25677
    },
    {
      "epoch": 0.9195838630544165,
      "grad_norm": 1.614140510559082,
      "learning_rate": 3.3712375371743987e-06,
      "loss": 1.5758,
      "step": 25678
    },
    {
      "epoch": 0.9196196751839848,
      "grad_norm": 2.11362624168396,
      "learning_rate": 3.3682518473173607e-06,
      "loss": 1.3219,
      "step": 25679
    },
    {
      "epoch": 0.9196554873135531,
      "grad_norm": 2.002917766571045,
      "learning_rate": 3.3652674575158306e-06,
      "loss": 1.2815,
      "step": 25680
    },
    {
      "epoch": 0.9196912994431213,
      "grad_norm": 1.6976182460784912,
      "learning_rate": 3.362284367809976e-06,
      "loss": 1.3635,
      "step": 25681
    },
    {
      "epoch": 0.9197271115726897,
      "grad_norm": 1.6870442628860474,
      "learning_rate": 3.3593025782399424e-06,
      "loss": 1.2197,
      "step": 25682
    },
    {
      "epoch": 0.919762923702258,
      "grad_norm": 2.186999797821045,
      "learning_rate": 3.3563220888458425e-06,
      "loss": 1.3623,
      "step": 25683
    },
    {
      "epoch": 0.9197987358318263,
      "grad_norm": 1.727008581161499,
      "learning_rate": 3.353342899667755e-06,
      "loss": 1.2574,
      "step": 25684
    },
    {
      "epoch": 0.9198345479613945,
      "grad_norm": 1.7052570581436157,
      "learning_rate": 3.3503650107457706e-06,
      "loss": 1.7394,
      "step": 25685
    },
    {
      "epoch": 0.9198703600909628,
      "grad_norm": 1.6224033832550049,
      "learning_rate": 3.347388422119968e-06,
      "loss": 1.3684,
      "step": 25686
    },
    {
      "epoch": 0.9199061722205311,
      "grad_norm": 1.8566479682922363,
      "learning_rate": 3.3444131338303708e-06,
      "loss": 1.127,
      "step": 25687
    },
    {
      "epoch": 0.9199419843500993,
      "grad_norm": 2.9639549255371094,
      "learning_rate": 3.3414391459170134e-06,
      "loss": 1.5899,
      "step": 25688
    },
    {
      "epoch": 0.9199777964796677,
      "grad_norm": 1.5856989622116089,
      "learning_rate": 3.33846645841992e-06,
      "loss": 1.2898,
      "step": 25689
    },
    {
      "epoch": 0.920013608609236,
      "grad_norm": 1.1478488445281982,
      "learning_rate": 3.33549507137908e-06,
      "loss": 1.3804,
      "step": 25690
    },
    {
      "epoch": 0.9200494207388042,
      "grad_norm": 1.6433991193771362,
      "learning_rate": 3.332524984834462e-06,
      "loss": 1.6188,
      "step": 25691
    },
    {
      "epoch": 0.9200852328683725,
      "grad_norm": 1.6301097869873047,
      "learning_rate": 3.3295561988260227e-06,
      "loss": 1.6194,
      "step": 25692
    },
    {
      "epoch": 0.9201210449979408,
      "grad_norm": 3.6255156993865967,
      "learning_rate": 3.326588713393719e-06,
      "loss": 1.531,
      "step": 25693
    },
    {
      "epoch": 0.920156857127509,
      "grad_norm": 1.6354080438613892,
      "learning_rate": 3.3236225285774637e-06,
      "loss": 1.4595,
      "step": 25694
    },
    {
      "epoch": 0.9201926692570773,
      "grad_norm": 1.5773730278015137,
      "learning_rate": 3.3206576444171577e-06,
      "loss": 1.2759,
      "step": 25695
    },
    {
      "epoch": 0.9202284813866457,
      "grad_norm": 1.889205813407898,
      "learning_rate": 3.317694060952692e-06,
      "loss": 1.1222,
      "step": 25696
    },
    {
      "epoch": 0.920264293516214,
      "grad_norm": 1.4370874166488647,
      "learning_rate": 3.314731778223956e-06,
      "loss": 1.443,
      "step": 25697
    },
    {
      "epoch": 0.9203001056457822,
      "grad_norm": 1.4879106283187866,
      "learning_rate": 3.3117707962707746e-06,
      "loss": 1.5812,
      "step": 25698
    },
    {
      "epoch": 0.9203359177753505,
      "grad_norm": 1.411486268043518,
      "learning_rate": 3.308811115133004e-06,
      "loss": 1.5091,
      "step": 25699
    },
    {
      "epoch": 0.9203717299049188,
      "grad_norm": 2.297484874725342,
      "learning_rate": 3.3058527348504455e-06,
      "loss": 1.1828,
      "step": 25700
    },
    {
      "epoch": 0.920407542034487,
      "grad_norm": 1.640698790550232,
      "learning_rate": 3.302895655462934e-06,
      "loss": 1.6501,
      "step": 25701
    },
    {
      "epoch": 0.9204433541640553,
      "grad_norm": 1.9217278957366943,
      "learning_rate": 3.2999398770102276e-06,
      "loss": 1.2598,
      "step": 25702
    },
    {
      "epoch": 0.9204791662936237,
      "grad_norm": 1.5686508417129517,
      "learning_rate": 3.296985399532071e-06,
      "loss": 1.4808,
      "step": 25703
    },
    {
      "epoch": 0.920514978423192,
      "grad_norm": 1.4743748903274536,
      "learning_rate": 3.2940322230682664e-06,
      "loss": 1.2292,
      "step": 25704
    },
    {
      "epoch": 0.9205507905527602,
      "grad_norm": 1.3531376123428345,
      "learning_rate": 3.291080347658504e-06,
      "loss": 1.2794,
      "step": 25705
    },
    {
      "epoch": 0.9205866026823285,
      "grad_norm": 1.6078829765319824,
      "learning_rate": 3.2881297733425188e-06,
      "loss": 1.4727,
      "step": 25706
    },
    {
      "epoch": 0.9206224148118968,
      "grad_norm": 1.4829717874526978,
      "learning_rate": 3.285180500159979e-06,
      "loss": 1.5516,
      "step": 25707
    },
    {
      "epoch": 0.920658226941465,
      "grad_norm": 1.7333753108978271,
      "learning_rate": 3.2822325281505973e-06,
      "loss": 1.3617,
      "step": 25708
    },
    {
      "epoch": 0.9206940390710333,
      "grad_norm": 1.9004656076431274,
      "learning_rate": 3.27928585735402e-06,
      "loss": 1.8051,
      "step": 25709
    },
    {
      "epoch": 0.9207298512006017,
      "grad_norm": 1.6876628398895264,
      "learning_rate": 3.2763404878098815e-06,
      "loss": 1.4722,
      "step": 25710
    },
    {
      "epoch": 0.92076566333017,
      "grad_norm": 1.6359210014343262,
      "learning_rate": 3.273396419557839e-06,
      "loss": 1.2267,
      "step": 25711
    },
    {
      "epoch": 0.9208014754597382,
      "grad_norm": 1.5168014764785767,
      "learning_rate": 3.2704536526374506e-06,
      "loss": 1.3787,
      "step": 25712
    },
    {
      "epoch": 0.9208372875893065,
      "grad_norm": 2.176189661026001,
      "learning_rate": 3.267512187088362e-06,
      "loss": 1.6466,
      "step": 25713
    },
    {
      "epoch": 0.9208730997188748,
      "grad_norm": 1.7817593812942505,
      "learning_rate": 3.2645720229500965e-06,
      "loss": 1.7843,
      "step": 25714
    },
    {
      "epoch": 0.920908911848443,
      "grad_norm": 1.368715524673462,
      "learning_rate": 3.2616331602622565e-06,
      "loss": 1.5336,
      "step": 25715
    },
    {
      "epoch": 0.9209447239780113,
      "grad_norm": 1.853217363357544,
      "learning_rate": 3.2586955990643432e-06,
      "loss": 1.4965,
      "step": 25716
    },
    {
      "epoch": 0.9209805361075797,
      "grad_norm": 2.11053729057312,
      "learning_rate": 3.255759339395903e-06,
      "loss": 1.5349,
      "step": 25717
    },
    {
      "epoch": 0.921016348237148,
      "grad_norm": 1.5958489179611206,
      "learning_rate": 3.2528243812964156e-06,
      "loss": 1.5431,
      "step": 25718
    },
    {
      "epoch": 0.9210521603667162,
      "grad_norm": 1.3022187948226929,
      "learning_rate": 3.2498907248054045e-06,
      "loss": 1.6714,
      "step": 25719
    },
    {
      "epoch": 0.9210879724962845,
      "grad_norm": 1.4858121871948242,
      "learning_rate": 3.2469583699623053e-06,
      "loss": 1.5852,
      "step": 25720
    },
    {
      "epoch": 0.9211237846258528,
      "grad_norm": 2.0671451091766357,
      "learning_rate": 3.2440273168065636e-06,
      "loss": 1.0612,
      "step": 25721
    },
    {
      "epoch": 0.921159596755421,
      "grad_norm": 1.670335054397583,
      "learning_rate": 3.241097565377649e-06,
      "loss": 1.5164,
      "step": 25722
    },
    {
      "epoch": 0.9211954088849893,
      "grad_norm": 1.7597389221191406,
      "learning_rate": 3.2381691157149395e-06,
      "loss": 1.4661,
      "step": 25723
    },
    {
      "epoch": 0.9212312210145577,
      "grad_norm": 1.881030797958374,
      "learning_rate": 3.2352419678578714e-06,
      "loss": 1.3387,
      "step": 25724
    },
    {
      "epoch": 0.921267033144126,
      "grad_norm": 1.9583910703659058,
      "learning_rate": 3.2323161218457796e-06,
      "loss": 1.5953,
      "step": 25725
    },
    {
      "epoch": 0.9213028452736942,
      "grad_norm": 1.6051225662231445,
      "learning_rate": 3.229391577718066e-06,
      "loss": 1.2554,
      "step": 25726
    },
    {
      "epoch": 0.9213386574032625,
      "grad_norm": 1.7156764268875122,
      "learning_rate": 3.226468335514077e-06,
      "loss": 1.4546,
      "step": 25727
    },
    {
      "epoch": 0.9213744695328308,
      "grad_norm": 1.7657434940338135,
      "learning_rate": 3.223546395273114e-06,
      "loss": 1.3154,
      "step": 25728
    },
    {
      "epoch": 0.921410281662399,
      "grad_norm": 1.699459433555603,
      "learning_rate": 3.220625757034501e-06,
      "loss": 1.4964,
      "step": 25729
    },
    {
      "epoch": 0.9214460937919673,
      "grad_norm": 1.7240272760391235,
      "learning_rate": 3.2177064208375298e-06,
      "loss": 1.6393,
      "step": 25730
    },
    {
      "epoch": 0.9214819059215357,
      "grad_norm": 1.9153105020523071,
      "learning_rate": 3.21478838672149e-06,
      "loss": 1.7071,
      "step": 25731
    },
    {
      "epoch": 0.9215177180511039,
      "grad_norm": 2.9040825366973877,
      "learning_rate": 3.211871654725618e-06,
      "loss": 1.7588,
      "step": 25732
    },
    {
      "epoch": 0.9215535301806722,
      "grad_norm": 1.6486643552780151,
      "learning_rate": 3.208956224889159e-06,
      "loss": 1.3715,
      "step": 25733
    },
    {
      "epoch": 0.9215893423102405,
      "grad_norm": 1.6218197345733643,
      "learning_rate": 3.2060420972513494e-06,
      "loss": 1.3616,
      "step": 25734
    },
    {
      "epoch": 0.9216251544398087,
      "grad_norm": 1.4166431427001953,
      "learning_rate": 3.203129271851402e-06,
      "loss": 1.2931,
      "step": 25735
    },
    {
      "epoch": 0.921660966569377,
      "grad_norm": 1.4935784339904785,
      "learning_rate": 3.2002177487284736e-06,
      "loss": 1.3154,
      "step": 25736
    },
    {
      "epoch": 0.9216967786989453,
      "grad_norm": 1.6623226404190063,
      "learning_rate": 3.197307527921756e-06,
      "loss": 1.4607,
      "step": 25737
    },
    {
      "epoch": 0.9217325908285137,
      "grad_norm": 1.6059545278549194,
      "learning_rate": 3.194398609470406e-06,
      "loss": 1.4489,
      "step": 25738
    },
    {
      "epoch": 0.9217684029580819,
      "grad_norm": 1.5388230085372925,
      "learning_rate": 3.1914909934135483e-06,
      "loss": 1.2268,
      "step": 25739
    },
    {
      "epoch": 0.9218042150876502,
      "grad_norm": 1.3560782670974731,
      "learning_rate": 3.1885846797902964e-06,
      "loss": 1.3344,
      "step": 25740
    },
    {
      "epoch": 0.9218400272172185,
      "grad_norm": 1.4934988021850586,
      "learning_rate": 3.185679668639763e-06,
      "loss": 1.4104,
      "step": 25741
    },
    {
      "epoch": 0.9218758393467867,
      "grad_norm": 1.6801812648773193,
      "learning_rate": 3.1827759600010498e-06,
      "loss": 1.6805,
      "step": 25742
    },
    {
      "epoch": 0.921911651476355,
      "grad_norm": 1.7978261709213257,
      "learning_rate": 3.179873553913171e-06,
      "loss": 1.5292,
      "step": 25743
    },
    {
      "epoch": 0.9219474636059233,
      "grad_norm": 1.6324962377548218,
      "learning_rate": 3.1769724504152164e-06,
      "loss": 1.1737,
      "step": 25744
    },
    {
      "epoch": 0.9219832757354917,
      "grad_norm": 1.4262455701828003,
      "learning_rate": 3.1740726495462223e-06,
      "loss": 1.3579,
      "step": 25745
    },
    {
      "epoch": 0.9220190878650599,
      "grad_norm": 1.422607183456421,
      "learning_rate": 3.1711741513451576e-06,
      "loss": 1.2943,
      "step": 25746
    },
    {
      "epoch": 0.9220548999946282,
      "grad_norm": 1.3137279748916626,
      "learning_rate": 3.1682769558510574e-06,
      "loss": 1.6224,
      "step": 25747
    },
    {
      "epoch": 0.9220907121241965,
      "grad_norm": 1.8328073024749756,
      "learning_rate": 3.165381063102879e-06,
      "loss": 1.3839,
      "step": 25748
    },
    {
      "epoch": 0.9221265242537647,
      "grad_norm": 1.4281005859375,
      "learning_rate": 3.162486473139603e-06,
      "loss": 1.2144,
      "step": 25749
    },
    {
      "epoch": 0.922162336383333,
      "grad_norm": 1.3970149755477905,
      "learning_rate": 3.1595931860001536e-06,
      "loss": 1.2464,
      "step": 25750
    },
    {
      "epoch": 0.9221981485129013,
      "grad_norm": 1.5348528623580933,
      "learning_rate": 3.1567012017234553e-06,
      "loss": 1.4653,
      "step": 25751
    },
    {
      "epoch": 0.9222339606424697,
      "grad_norm": 2.080173969268799,
      "learning_rate": 3.1538105203484323e-06,
      "loss": 1.4834,
      "step": 25752
    },
    {
      "epoch": 0.9222697727720379,
      "grad_norm": 1.4271656274795532,
      "learning_rate": 3.150921141913965e-06,
      "loss": 1.6015,
      "step": 25753
    },
    {
      "epoch": 0.9223055849016062,
      "grad_norm": 1.562336802482605,
      "learning_rate": 3.148033066458933e-06,
      "loss": 1.8834,
      "step": 25754
    },
    {
      "epoch": 0.9223413970311745,
      "grad_norm": 1.3763641119003296,
      "learning_rate": 3.145146294022172e-06,
      "loss": 1.3171,
      "step": 25755
    },
    {
      "epoch": 0.9223772091607427,
      "grad_norm": 1.3976991176605225,
      "learning_rate": 3.1422608246425513e-06,
      "loss": 1.185,
      "step": 25756
    },
    {
      "epoch": 0.922413021290311,
      "grad_norm": 1.4210984706878662,
      "learning_rate": 3.1393766583588614e-06,
      "loss": 1.6143,
      "step": 25757
    },
    {
      "epoch": 0.9224488334198793,
      "grad_norm": 1.7500970363616943,
      "learning_rate": 3.136493795209916e-06,
      "loss": 1.702,
      "step": 25758
    },
    {
      "epoch": 0.9224846455494476,
      "grad_norm": 2.02065372467041,
      "learning_rate": 3.1336122352345065e-06,
      "loss": 1.5751,
      "step": 25759
    },
    {
      "epoch": 0.9225204576790159,
      "grad_norm": 1.553234338760376,
      "learning_rate": 3.130731978471402e-06,
      "loss": 1.1833,
      "step": 25760
    },
    {
      "epoch": 0.9225562698085842,
      "grad_norm": 1.5323034524917603,
      "learning_rate": 3.1278530249593372e-06,
      "loss": 1.5451,
      "step": 25761
    },
    {
      "epoch": 0.9225920819381525,
      "grad_norm": 1.3176145553588867,
      "learning_rate": 3.124975374737049e-06,
      "loss": 1.055,
      "step": 25762
    },
    {
      "epoch": 0.9226278940677207,
      "grad_norm": 2.0297014713287354,
      "learning_rate": 3.1220990278432727e-06,
      "loss": 1.3966,
      "step": 25763
    },
    {
      "epoch": 0.922663706197289,
      "grad_norm": 1.19528067111969,
      "learning_rate": 3.119223984316677e-06,
      "loss": 1.4489,
      "step": 25764
    },
    {
      "epoch": 0.9226995183268573,
      "grad_norm": 1.6679394245147705,
      "learning_rate": 3.1163502441959647e-06,
      "loss": 1.1535,
      "step": 25765
    },
    {
      "epoch": 0.9227353304564256,
      "grad_norm": 1.6907655000686646,
      "learning_rate": 3.113477807519782e-06,
      "loss": 1.6175,
      "step": 25766
    },
    {
      "epoch": 0.9227711425859939,
      "grad_norm": 1.7465883493423462,
      "learning_rate": 3.110606674326788e-06,
      "loss": 1.6063,
      "step": 25767
    },
    {
      "epoch": 0.9228069547155622,
      "grad_norm": 1.6468842029571533,
      "learning_rate": 3.1077368446555956e-06,
      "loss": 1.378,
      "step": 25768
    },
    {
      "epoch": 0.9228427668451304,
      "grad_norm": 1.6969002485275269,
      "learning_rate": 3.104868318544818e-06,
      "loss": 1.4272,
      "step": 25769
    },
    {
      "epoch": 0.9228785789746987,
      "grad_norm": 1.4377319812774658,
      "learning_rate": 3.1020010960330583e-06,
      "loss": 1.6473,
      "step": 25770
    },
    {
      "epoch": 0.922914391104267,
      "grad_norm": 2.1896963119506836,
      "learning_rate": 3.0991351771588963e-06,
      "loss": 1.2011,
      "step": 25771
    },
    {
      "epoch": 0.9229502032338353,
      "grad_norm": 1.6838210821151733,
      "learning_rate": 3.0962705619608565e-06,
      "loss": 1.372,
      "step": 25772
    },
    {
      "epoch": 0.9229860153634036,
      "grad_norm": 1.592898964881897,
      "learning_rate": 3.093407250477509e-06,
      "loss": 1.4824,
      "step": 25773
    },
    {
      "epoch": 0.9230218274929719,
      "grad_norm": 1.7946579456329346,
      "learning_rate": 3.0905452427473667e-06,
      "loss": 1.4023,
      "step": 25774
    },
    {
      "epoch": 0.9230576396225402,
      "grad_norm": 2.0756218433380127,
      "learning_rate": 3.0876845388089327e-06,
      "loss": 1.4185,
      "step": 25775
    },
    {
      "epoch": 0.9230934517521084,
      "grad_norm": 1.2983440160751343,
      "learning_rate": 3.084825138700698e-06,
      "loss": 1.4131,
      "step": 25776
    },
    {
      "epoch": 0.9231292638816767,
      "grad_norm": 2.0300850868225098,
      "learning_rate": 3.08196704246112e-06,
      "loss": 1.5474,
      "step": 25777
    },
    {
      "epoch": 0.923165076011245,
      "grad_norm": 2.3360812664031982,
      "learning_rate": 3.0791102501286804e-06,
      "loss": 1.1895,
      "step": 25778
    },
    {
      "epoch": 0.9232008881408132,
      "grad_norm": 1.792782187461853,
      "learning_rate": 3.0762547617417703e-06,
      "loss": 1.308,
      "step": 25779
    },
    {
      "epoch": 0.9232367002703816,
      "grad_norm": 1.6354910135269165,
      "learning_rate": 3.0734005773388364e-06,
      "loss": 1.405,
      "step": 25780
    },
    {
      "epoch": 0.9232725123999499,
      "grad_norm": 1.6714788675308228,
      "learning_rate": 3.0705476969582813e-06,
      "loss": 1.2023,
      "step": 25781
    },
    {
      "epoch": 0.9233083245295182,
      "grad_norm": 1.793521523475647,
      "learning_rate": 3.0676961206384746e-06,
      "loss": 1.3668,
      "step": 25782
    },
    {
      "epoch": 0.9233441366590864,
      "grad_norm": 1.6857377290725708,
      "learning_rate": 3.0648458484177746e-06,
      "loss": 1.5258,
      "step": 25783
    },
    {
      "epoch": 0.9233799487886547,
      "grad_norm": 1.6202548742294312,
      "learning_rate": 3.061996880334539e-06,
      "loss": 1.2652,
      "step": 25784
    },
    {
      "epoch": 0.923415760918223,
      "grad_norm": 1.563083291053772,
      "learning_rate": 3.059149216427104e-06,
      "loss": 1.5697,
      "step": 25785
    },
    {
      "epoch": 0.9234515730477912,
      "grad_norm": 1.3261644840240479,
      "learning_rate": 3.0563028567337614e-06,
      "loss": 1.38,
      "step": 25786
    },
    {
      "epoch": 0.9234873851773596,
      "grad_norm": 1.2361618280410767,
      "learning_rate": 3.053457801292814e-06,
      "loss": 1.3723,
      "step": 25787
    },
    {
      "epoch": 0.9235231973069279,
      "grad_norm": 1.499376893043518,
      "learning_rate": 3.0506140501425417e-06,
      "loss": 1.4285,
      "step": 25788
    },
    {
      "epoch": 0.9235590094364962,
      "grad_norm": 1.281010389328003,
      "learning_rate": 3.0477716033212032e-06,
      "loss": 1.4848,
      "step": 25789
    },
    {
      "epoch": 0.9235948215660644,
      "grad_norm": 2.0112226009368896,
      "learning_rate": 3.044930460867046e-06,
      "loss": 1.4768,
      "step": 25790
    },
    {
      "epoch": 0.9236306336956327,
      "grad_norm": 1.588922142982483,
      "learning_rate": 3.042090622818272e-06,
      "loss": 1.2698,
      "step": 25791
    },
    {
      "epoch": 0.923666445825201,
      "grad_norm": 1.5995523929595947,
      "learning_rate": 3.039252089213118e-06,
      "loss": 1.3193,
      "step": 25792
    },
    {
      "epoch": 0.9237022579547692,
      "grad_norm": 2.97119140625,
      "learning_rate": 3.0364148600897423e-06,
      "loss": 1.5305,
      "step": 25793
    },
    {
      "epoch": 0.9237380700843376,
      "grad_norm": 2.8016178607940674,
      "learning_rate": 3.0335789354863362e-06,
      "loss": 1.5246,
      "step": 25794
    },
    {
      "epoch": 0.9237738822139059,
      "grad_norm": 1.5737606287002563,
      "learning_rate": 3.0307443154410365e-06,
      "loss": 1.42,
      "step": 25795
    },
    {
      "epoch": 0.9238096943434742,
      "grad_norm": 1.4103105068206787,
      "learning_rate": 3.027910999992012e-06,
      "loss": 1.4139,
      "step": 25796
    },
    {
      "epoch": 0.9238455064730424,
      "grad_norm": 1.6987563371658325,
      "learning_rate": 3.0250789891773433e-06,
      "loss": 1.6168,
      "step": 25797
    },
    {
      "epoch": 0.9238813186026107,
      "grad_norm": 1.6376010179519653,
      "learning_rate": 3.022248283035156e-06,
      "loss": 1.3091,
      "step": 25798
    },
    {
      "epoch": 0.923917130732179,
      "grad_norm": 1.1662794351577759,
      "learning_rate": 3.0194188816035305e-06,
      "loss": 1.2575,
      "step": 25799
    },
    {
      "epoch": 0.9239529428617472,
      "grad_norm": 1.8269308805465698,
      "learning_rate": 3.0165907849205254e-06,
      "loss": 1.2645,
      "step": 25800
    },
    {
      "epoch": 0.9239887549913156,
      "grad_norm": 1.7312307357788086,
      "learning_rate": 3.013763993024188e-06,
      "loss": 1.2861,
      "step": 25801
    },
    {
      "epoch": 0.9240245671208839,
      "grad_norm": 1.7468078136444092,
      "learning_rate": 3.010938505952543e-06,
      "loss": 1.4326,
      "step": 25802
    },
    {
      "epoch": 0.9240603792504521,
      "grad_norm": 1.40182363986969,
      "learning_rate": 3.008114323743627e-06,
      "loss": 1.2857,
      "step": 25803
    },
    {
      "epoch": 0.9240961913800204,
      "grad_norm": 1.6253925561904907,
      "learning_rate": 3.005291446435421e-06,
      "loss": 1.4106,
      "step": 25804
    },
    {
      "epoch": 0.9241320035095887,
      "grad_norm": 1.5819063186645508,
      "learning_rate": 3.002469874065894e-06,
      "loss": 1.2463,
      "step": 25805
    },
    {
      "epoch": 0.924167815639157,
      "grad_norm": 1.8966995477676392,
      "learning_rate": 2.999649606673027e-06,
      "loss": 1.4093,
      "step": 25806
    },
    {
      "epoch": 0.9242036277687252,
      "grad_norm": 1.4328267574310303,
      "learning_rate": 2.996830644294757e-06,
      "loss": 1.2878,
      "step": 25807
    },
    {
      "epoch": 0.9242394398982936,
      "grad_norm": 2.1859099864959717,
      "learning_rate": 2.994012986969008e-06,
      "loss": 1.3069,
      "step": 25808
    },
    {
      "epoch": 0.9242752520278619,
      "grad_norm": 1.6977571249008179,
      "learning_rate": 2.991196634733662e-06,
      "loss": 1.162,
      "step": 25809
    },
    {
      "epoch": 0.9243110641574301,
      "grad_norm": 1.3135548830032349,
      "learning_rate": 2.9883815876266653e-06,
      "loss": 1.3733,
      "step": 25810
    },
    {
      "epoch": 0.9243468762869984,
      "grad_norm": 1.7272981405258179,
      "learning_rate": 2.985567845685833e-06,
      "loss": 1.5297,
      "step": 25811
    },
    {
      "epoch": 0.9243826884165667,
      "grad_norm": 1.472185492515564,
      "learning_rate": 2.982755408949067e-06,
      "loss": 1.3625,
      "step": 25812
    },
    {
      "epoch": 0.924418500546135,
      "grad_norm": 1.6928966045379639,
      "learning_rate": 2.97994427745415e-06,
      "loss": 1.4356,
      "step": 25813
    },
    {
      "epoch": 0.9244543126757032,
      "grad_norm": 1.5978893041610718,
      "learning_rate": 2.977134451238972e-06,
      "loss": 1.322,
      "step": 25814
    },
    {
      "epoch": 0.9244901248052716,
      "grad_norm": 1.2723480463027954,
      "learning_rate": 2.9743259303412707e-06,
      "loss": 1.3138,
      "step": 25815
    },
    {
      "epoch": 0.9245259369348399,
      "grad_norm": 1.316504716873169,
      "learning_rate": 2.9715187147988823e-06,
      "loss": 1.177,
      "step": 25816
    },
    {
      "epoch": 0.9245617490644081,
      "grad_norm": 1.8215168714523315,
      "learning_rate": 2.968712804649543e-06,
      "loss": 1.5501,
      "step": 25817
    },
    {
      "epoch": 0.9245975611939764,
      "grad_norm": 1.5576127767562866,
      "learning_rate": 2.9659081999310112e-06,
      "loss": 1.3463,
      "step": 25818
    },
    {
      "epoch": 0.9246333733235447,
      "grad_norm": 1.40639328956604,
      "learning_rate": 2.9631049006810243e-06,
      "loss": 1.1919,
      "step": 25819
    },
    {
      "epoch": 0.9246691854531129,
      "grad_norm": 1.4003286361694336,
      "learning_rate": 2.9603029069372733e-06,
      "loss": 1.5546,
      "step": 25820
    },
    {
      "epoch": 0.9247049975826812,
      "grad_norm": 1.6173608303070068,
      "learning_rate": 2.9575022187374958e-06,
      "loss": 1.5284,
      "step": 25821
    },
    {
      "epoch": 0.9247408097122496,
      "grad_norm": 1.644357442855835,
      "learning_rate": 2.9547028361193495e-06,
      "loss": 1.3983,
      "step": 25822
    },
    {
      "epoch": 0.9247766218418179,
      "grad_norm": 1.6090519428253174,
      "learning_rate": 2.951904759120494e-06,
      "loss": 1.4281,
      "step": 25823
    },
    {
      "epoch": 0.9248124339713861,
      "grad_norm": 1.6314555406570435,
      "learning_rate": 2.9491079877785767e-06,
      "loss": 1.3319,
      "step": 25824
    },
    {
      "epoch": 0.9248482461009544,
      "grad_norm": 1.707886815071106,
      "learning_rate": 2.9463125221312117e-06,
      "loss": 1.2501,
      "step": 25825
    },
    {
      "epoch": 0.9248840582305227,
      "grad_norm": 1.4632911682128906,
      "learning_rate": 2.9435183622160465e-06,
      "loss": 1.526,
      "step": 25826
    },
    {
      "epoch": 0.9249198703600909,
      "grad_norm": 1.7945195436477661,
      "learning_rate": 2.9407255080706297e-06,
      "loss": 1.3357,
      "step": 25827
    },
    {
      "epoch": 0.9249556824896592,
      "grad_norm": 1.7915374040603638,
      "learning_rate": 2.937933959732553e-06,
      "loss": 1.5678,
      "step": 25828
    },
    {
      "epoch": 0.9249914946192276,
      "grad_norm": 1.4077473878860474,
      "learning_rate": 2.9351437172393746e-06,
      "loss": 1.5492,
      "step": 25829
    },
    {
      "epoch": 0.9250273067487959,
      "grad_norm": 1.4841781854629517,
      "learning_rate": 2.9323547806286432e-06,
      "loss": 1.391,
      "step": 25830
    },
    {
      "epoch": 0.9250631188783641,
      "grad_norm": 2.442631721496582,
      "learning_rate": 2.9295671499378506e-06,
      "loss": 1.6424,
      "step": 25831
    },
    {
      "epoch": 0.9250989310079324,
      "grad_norm": 1.5331733226776123,
      "learning_rate": 2.9267808252045338e-06,
      "loss": 1.4525,
      "step": 25832
    },
    {
      "epoch": 0.9251347431375007,
      "grad_norm": 2.0141773223876953,
      "learning_rate": 2.923995806466173e-06,
      "loss": 1.367,
      "step": 25833
    },
    {
      "epoch": 0.9251705552670689,
      "grad_norm": 1.460148572921753,
      "learning_rate": 2.9212120937602174e-06,
      "loss": 1.5702,
      "step": 25834
    },
    {
      "epoch": 0.9252063673966372,
      "grad_norm": 1.5255372524261475,
      "learning_rate": 2.9184296871241357e-06,
      "loss": 1.3125,
      "step": 25835
    },
    {
      "epoch": 0.9252421795262056,
      "grad_norm": 1.992004156112671,
      "learning_rate": 2.9156485865953544e-06,
      "loss": 1.5373,
      "step": 25836
    },
    {
      "epoch": 0.9252779916557738,
      "grad_norm": 1.4714744091033936,
      "learning_rate": 2.9128687922112987e-06,
      "loss": 1.3717,
      "step": 25837
    },
    {
      "epoch": 0.9253138037853421,
      "grad_norm": 1.6506719589233398,
      "learning_rate": 2.91009030400935e-06,
      "loss": 1.3876,
      "step": 25838
    },
    {
      "epoch": 0.9253496159149104,
      "grad_norm": 1.959061622619629,
      "learning_rate": 2.9073131220269e-06,
      "loss": 1.2011,
      "step": 25839
    },
    {
      "epoch": 0.9253854280444787,
      "grad_norm": 1.3772485256195068,
      "learning_rate": 2.9045372463013088e-06,
      "loss": 1.3579,
      "step": 25840
    },
    {
      "epoch": 0.9254212401740469,
      "grad_norm": 2.8980002403259277,
      "learning_rate": 2.9017626768699346e-06,
      "loss": 1.3286,
      "step": 25841
    },
    {
      "epoch": 0.9254570523036152,
      "grad_norm": 2.0866899490356445,
      "learning_rate": 2.8989894137700924e-06,
      "loss": 1.7091,
      "step": 25842
    },
    {
      "epoch": 0.9254928644331836,
      "grad_norm": 1.37162446975708,
      "learning_rate": 2.8962174570390965e-06,
      "loss": 1.5254,
      "step": 25843
    },
    {
      "epoch": 0.9255286765627518,
      "grad_norm": 1.817568063735962,
      "learning_rate": 2.8934468067142396e-06,
      "loss": 1.1529,
      "step": 25844
    },
    {
      "epoch": 0.9255644886923201,
      "grad_norm": 1.9801509380340576,
      "learning_rate": 2.8906774628327917e-06,
      "loss": 1.5762,
      "step": 25845
    },
    {
      "epoch": 0.9256003008218884,
      "grad_norm": 1.6975454092025757,
      "learning_rate": 2.8879094254320225e-06,
      "loss": 1.2415,
      "step": 25846
    },
    {
      "epoch": 0.9256361129514566,
      "grad_norm": 2.336732864379883,
      "learning_rate": 2.8851426945491588e-06,
      "loss": 1.4693,
      "step": 25847
    },
    {
      "epoch": 0.9256719250810249,
      "grad_norm": 1.9778311252593994,
      "learning_rate": 2.882377270221448e-06,
      "loss": 1.4825,
      "step": 25848
    },
    {
      "epoch": 0.9257077372105932,
      "grad_norm": 1.873709797859192,
      "learning_rate": 2.8796131524860603e-06,
      "loss": 1.408,
      "step": 25849
    },
    {
      "epoch": 0.9257435493401616,
      "grad_norm": 1.574170470237732,
      "learning_rate": 2.8768503413802108e-06,
      "loss": 1.3283,
      "step": 25850
    },
    {
      "epoch": 0.9257793614697298,
      "grad_norm": 1.7731971740722656,
      "learning_rate": 2.8740888369410577e-06,
      "loss": 1.6502,
      "step": 25851
    },
    {
      "epoch": 0.9258151735992981,
      "grad_norm": 1.3365644216537476,
      "learning_rate": 2.8713286392057614e-06,
      "loss": 1.4837,
      "step": 25852
    },
    {
      "epoch": 0.9258509857288664,
      "grad_norm": 1.4238344430923462,
      "learning_rate": 2.868569748211436e-06,
      "loss": 1.3521,
      "step": 25853
    },
    {
      "epoch": 0.9258867978584346,
      "grad_norm": 1.5015554428100586,
      "learning_rate": 2.8658121639952297e-06,
      "loss": 1.5199,
      "step": 25854
    },
    {
      "epoch": 0.9259226099880029,
      "grad_norm": 1.2220364809036255,
      "learning_rate": 2.8630558865942237e-06,
      "loss": 1.33,
      "step": 25855
    },
    {
      "epoch": 0.9259584221175712,
      "grad_norm": 1.678745985031128,
      "learning_rate": 2.8603009160454995e-06,
      "loss": 1.8247,
      "step": 25856
    },
    {
      "epoch": 0.9259942342471396,
      "grad_norm": 1.7295626401901245,
      "learning_rate": 2.857547252386117e-06,
      "loss": 1.4945,
      "step": 25857
    },
    {
      "epoch": 0.9260300463767078,
      "grad_norm": 1.4009100198745728,
      "learning_rate": 2.854794895653146e-06,
      "loss": 1.1137,
      "step": 25858
    },
    {
      "epoch": 0.9260658585062761,
      "grad_norm": 1.4563788175582886,
      "learning_rate": 2.8520438458836007e-06,
      "loss": 1.5361,
      "step": 25859
    },
    {
      "epoch": 0.9261016706358444,
      "grad_norm": 1.6020429134368896,
      "learning_rate": 2.849294103114486e-06,
      "loss": 1.329,
      "step": 25860
    },
    {
      "epoch": 0.9261374827654126,
      "grad_norm": 1.4788800477981567,
      "learning_rate": 2.846545667382805e-06,
      "loss": 1.3567,
      "step": 25861
    },
    {
      "epoch": 0.9261732948949809,
      "grad_norm": 1.8151326179504395,
      "learning_rate": 2.8437985387255394e-06,
      "loss": 1.7524,
      "step": 25862
    },
    {
      "epoch": 0.9262091070245492,
      "grad_norm": 1.3136143684387207,
      "learning_rate": 2.8410527171796376e-06,
      "loss": 1.109,
      "step": 25863
    },
    {
      "epoch": 0.9262449191541176,
      "grad_norm": 3.2092204093933105,
      "learning_rate": 2.838308202782036e-06,
      "loss": 1.8549,
      "step": 25864
    },
    {
      "epoch": 0.9262807312836858,
      "grad_norm": 1.5936020612716675,
      "learning_rate": 2.835564995569684e-06,
      "loss": 1.6723,
      "step": 25865
    },
    {
      "epoch": 0.9263165434132541,
      "grad_norm": 1.55735445022583,
      "learning_rate": 2.8328230955794733e-06,
      "loss": 1.3075,
      "step": 25866
    },
    {
      "epoch": 0.9263523555428224,
      "grad_norm": 1.6565905809402466,
      "learning_rate": 2.8300825028482748e-06,
      "loss": 1.3222,
      "step": 25867
    },
    {
      "epoch": 0.9263881676723906,
      "grad_norm": 1.7441405057907104,
      "learning_rate": 2.827343217412981e-06,
      "loss": 1.2376,
      "step": 25868
    },
    {
      "epoch": 0.9264239798019589,
      "grad_norm": 1.8766648769378662,
      "learning_rate": 2.8246052393104516e-06,
      "loss": 1.4782,
      "step": 25869
    },
    {
      "epoch": 0.9264597919315272,
      "grad_norm": 1.460390329360962,
      "learning_rate": 2.8218685685775015e-06,
      "loss": 1.4786,
      "step": 25870
    },
    {
      "epoch": 0.9264956040610955,
      "grad_norm": 1.593667984008789,
      "learning_rate": 2.8191332052509567e-06,
      "loss": 1.6359,
      "step": 25871
    },
    {
      "epoch": 0.9265314161906638,
      "grad_norm": 1.744284987449646,
      "learning_rate": 2.8163991493676212e-06,
      "loss": 1.2058,
      "step": 25872
    },
    {
      "epoch": 0.9265672283202321,
      "grad_norm": 1.8081164360046387,
      "learning_rate": 2.8136664009642877e-06,
      "loss": 1.678,
      "step": 25873
    },
    {
      "epoch": 0.9266030404498004,
      "grad_norm": 1.7997514009475708,
      "learning_rate": 2.8109349600777045e-06,
      "loss": 1.6136,
      "step": 25874
    },
    {
      "epoch": 0.9266388525793686,
      "grad_norm": 1.647782802581787,
      "learning_rate": 2.8082048267446203e-06,
      "loss": 1.1104,
      "step": 25875
    },
    {
      "epoch": 0.9266746647089369,
      "grad_norm": 1.5916073322296143,
      "learning_rate": 2.805476001001772e-06,
      "loss": 1.5106,
      "step": 25876
    },
    {
      "epoch": 0.9267104768385052,
      "grad_norm": 1.8126418590545654,
      "learning_rate": 2.802748482885886e-06,
      "loss": 1.3683,
      "step": 25877
    },
    {
      "epoch": 0.9267462889680735,
      "grad_norm": 1.6614654064178467,
      "learning_rate": 2.800022272433633e-06,
      "loss": 1.3221,
      "step": 25878
    },
    {
      "epoch": 0.9267821010976418,
      "grad_norm": 1.9782456159591675,
      "learning_rate": 2.797297369681706e-06,
      "loss": 1.3235,
      "step": 25879
    },
    {
      "epoch": 0.9268179132272101,
      "grad_norm": 1.6730729341506958,
      "learning_rate": 2.7945737746667643e-06,
      "loss": 1.6093,
      "step": 25880
    },
    {
      "epoch": 0.9268537253567783,
      "grad_norm": 1.5324561595916748,
      "learning_rate": 2.7918514874254454e-06,
      "loss": 1.3519,
      "step": 25881
    },
    {
      "epoch": 0.9268895374863466,
      "grad_norm": 1.5539942979812622,
      "learning_rate": 2.789130507994364e-06,
      "loss": 1.1705,
      "step": 25882
    },
    {
      "epoch": 0.9269253496159149,
      "grad_norm": 1.2986265420913696,
      "learning_rate": 2.786410836410147e-06,
      "loss": 1.4491,
      "step": 25883
    },
    {
      "epoch": 0.9269611617454832,
      "grad_norm": 1.458767056465149,
      "learning_rate": 2.783692472709376e-06,
      "loss": 1.5184,
      "step": 25884
    },
    {
      "epoch": 0.9269969738750515,
      "grad_norm": 1.4174643754959106,
      "learning_rate": 2.7809754169286216e-06,
      "loss": 1.3685,
      "step": 25885
    },
    {
      "epoch": 0.9270327860046198,
      "grad_norm": 1.9586244821548462,
      "learning_rate": 2.7782596691044327e-06,
      "loss": 1.3631,
      "step": 25886
    },
    {
      "epoch": 0.9270685981341881,
      "grad_norm": 1.8806778192520142,
      "learning_rate": 2.7755452292733684e-06,
      "loss": 1.5604,
      "step": 25887
    },
    {
      "epoch": 0.9271044102637563,
      "grad_norm": 1.5468672513961792,
      "learning_rate": 2.7728320974719225e-06,
      "loss": 1.0677,
      "step": 25888
    },
    {
      "epoch": 0.9271402223933246,
      "grad_norm": 1.300863265991211,
      "learning_rate": 2.7701202737366096e-06,
      "loss": 1.379,
      "step": 25889
    },
    {
      "epoch": 0.9271760345228929,
      "grad_norm": 1.3959417343139648,
      "learning_rate": 2.7674097581039004e-06,
      "loss": 1.4749,
      "step": 25890
    },
    {
      "epoch": 0.9272118466524611,
      "grad_norm": 1.5664010047912598,
      "learning_rate": 2.7647005506102886e-06,
      "loss": 1.2282,
      "step": 25891
    },
    {
      "epoch": 0.9272476587820295,
      "grad_norm": 2.319263458251953,
      "learning_rate": 2.7619926512921888e-06,
      "loss": 1.4032,
      "step": 25892
    },
    {
      "epoch": 0.9272834709115978,
      "grad_norm": 1.7327048778533936,
      "learning_rate": 2.7592860601860616e-06,
      "loss": 1.7905,
      "step": 25893
    },
    {
      "epoch": 0.9273192830411661,
      "grad_norm": 1.7181236743927002,
      "learning_rate": 2.7565807773282994e-06,
      "loss": 1.3992,
      "step": 25894
    },
    {
      "epoch": 0.9273550951707343,
      "grad_norm": 1.821083426475525,
      "learning_rate": 2.7538768027553174e-06,
      "loss": 1.6069,
      "step": 25895
    },
    {
      "epoch": 0.9273909073003026,
      "grad_norm": 1.5232264995574951,
      "learning_rate": 2.751174136503498e-06,
      "loss": 1.3917,
      "step": 25896
    },
    {
      "epoch": 0.9274267194298709,
      "grad_norm": 1.4377104043960571,
      "learning_rate": 2.748472778609157e-06,
      "loss": 1.4063,
      "step": 25897
    },
    {
      "epoch": 0.9274625315594391,
      "grad_norm": 1.4285274744033813,
      "learning_rate": 2.7457727291086867e-06,
      "loss": 1.4704,
      "step": 25898
    },
    {
      "epoch": 0.9274983436890075,
      "grad_norm": 1.573488473892212,
      "learning_rate": 2.7430739880383915e-06,
      "loss": 1.3264,
      "step": 25899
    },
    {
      "epoch": 0.9275341558185758,
      "grad_norm": 1.4215755462646484,
      "learning_rate": 2.7403765554345984e-06,
      "loss": 1.3079,
      "step": 25900
    },
    {
      "epoch": 0.9275699679481441,
      "grad_norm": 1.3711589574813843,
      "learning_rate": 2.737680431333556e-06,
      "loss": 1.3841,
      "step": 25901
    },
    {
      "epoch": 0.9276057800777123,
      "grad_norm": 1.6720585823059082,
      "learning_rate": 2.7349856157715793e-06,
      "loss": 1.4348,
      "step": 25902
    },
    {
      "epoch": 0.9276415922072806,
      "grad_norm": 1.738786220550537,
      "learning_rate": 2.7322921087849063e-06,
      "loss": 1.5032,
      "step": 25903
    },
    {
      "epoch": 0.9276774043368489,
      "grad_norm": 2.6275978088378906,
      "learning_rate": 2.7295999104097746e-06,
      "loss": 1.3313,
      "step": 25904
    },
    {
      "epoch": 0.9277132164664171,
      "grad_norm": 1.735266923904419,
      "learning_rate": 2.726909020682422e-06,
      "loss": 1.2613,
      "step": 25905
    },
    {
      "epoch": 0.9277490285959855,
      "grad_norm": 1.6922903060913086,
      "learning_rate": 2.72421943963902e-06,
      "loss": 1.3072,
      "step": 25906
    },
    {
      "epoch": 0.9277848407255538,
      "grad_norm": 1.5648711919784546,
      "learning_rate": 2.7215311673157715e-06,
      "loss": 1.5856,
      "step": 25907
    },
    {
      "epoch": 0.927820652855122,
      "grad_norm": 1.460918664932251,
      "learning_rate": 2.718844203748827e-06,
      "loss": 1.3441,
      "step": 25908
    },
    {
      "epoch": 0.9278564649846903,
      "grad_norm": 1.5979174375534058,
      "learning_rate": 2.716158548974379e-06,
      "loss": 1.6612,
      "step": 25909
    },
    {
      "epoch": 0.9278922771142586,
      "grad_norm": 1.7928675413131714,
      "learning_rate": 2.71347420302851e-06,
      "loss": 1.5751,
      "step": 25910
    },
    {
      "epoch": 0.9279280892438269,
      "grad_norm": 1.5402860641479492,
      "learning_rate": 2.7107911659473682e-06,
      "loss": 1.2454,
      "step": 25911
    },
    {
      "epoch": 0.9279639013733951,
      "grad_norm": 1.4214240312576294,
      "learning_rate": 2.708109437767015e-06,
      "loss": 1.1149,
      "step": 25912
    },
    {
      "epoch": 0.9279997135029635,
      "grad_norm": 1.4117375612258911,
      "learning_rate": 2.705429018523575e-06,
      "loss": 1.102,
      "step": 25913
    },
    {
      "epoch": 0.9280355256325318,
      "grad_norm": 1.7786318063735962,
      "learning_rate": 2.702749908253077e-06,
      "loss": 1.2132,
      "step": 25914
    },
    {
      "epoch": 0.9280713377621,
      "grad_norm": 1.6047890186309814,
      "learning_rate": 2.70007210699158e-06,
      "loss": 1.266,
      "step": 25915
    },
    {
      "epoch": 0.9281071498916683,
      "grad_norm": 2.034290075302124,
      "learning_rate": 2.697395614775089e-06,
      "loss": 1.2727,
      "step": 25916
    },
    {
      "epoch": 0.9281429620212366,
      "grad_norm": 1.5755672454833984,
      "learning_rate": 2.694720431639641e-06,
      "loss": 1.0504,
      "step": 25917
    },
    {
      "epoch": 0.9281787741508049,
      "grad_norm": 1.8133633136749268,
      "learning_rate": 2.6920465576212195e-06,
      "loss": 1.6142,
      "step": 25918
    },
    {
      "epoch": 0.9282145862803731,
      "grad_norm": 2.0033183097839355,
      "learning_rate": 2.6893739927557725e-06,
      "loss": 1.4867,
      "step": 25919
    },
    {
      "epoch": 0.9282503984099415,
      "grad_norm": 1.8075355291366577,
      "learning_rate": 2.6867027370793053e-06,
      "loss": 1.2702,
      "step": 25920
    },
    {
      "epoch": 0.9282862105395098,
      "grad_norm": 1.8477706909179688,
      "learning_rate": 2.684032790627722e-06,
      "loss": 1.1577,
      "step": 25921
    },
    {
      "epoch": 0.928322022669078,
      "grad_norm": 1.8060771226882935,
      "learning_rate": 2.6813641534369383e-06,
      "loss": 1.4247,
      "step": 25922
    },
    {
      "epoch": 0.9283578347986463,
      "grad_norm": 1.3779622316360474,
      "learning_rate": 2.678696825542859e-06,
      "loss": 1.5116,
      "step": 25923
    },
    {
      "epoch": 0.9283936469282146,
      "grad_norm": 1.5192153453826904,
      "learning_rate": 2.676030806981389e-06,
      "loss": 1.442,
      "step": 25924
    },
    {
      "epoch": 0.9284294590577828,
      "grad_norm": 1.7140536308288574,
      "learning_rate": 2.673366097788399e-06,
      "loss": 1.4663,
      "step": 25925
    },
    {
      "epoch": 0.9284652711873511,
      "grad_norm": 1.3808923959732056,
      "learning_rate": 2.670702697999705e-06,
      "loss": 1.4116,
      "step": 25926
    },
    {
      "epoch": 0.9285010833169195,
      "grad_norm": 1.517737865447998,
      "learning_rate": 2.6680406076511677e-06,
      "loss": 1.3334,
      "step": 25927
    },
    {
      "epoch": 0.9285368954464878,
      "grad_norm": 1.7509211301803589,
      "learning_rate": 2.6653798267785912e-06,
      "loss": 1.2938,
      "step": 25928
    },
    {
      "epoch": 0.928572707576056,
      "grad_norm": 1.7517247200012207,
      "learning_rate": 2.6627203554177916e-06,
      "loss": 1.1616,
      "step": 25929
    },
    {
      "epoch": 0.9286085197056243,
      "grad_norm": 1.3450393676757812,
      "learning_rate": 2.660062193604518e-06,
      "loss": 1.4951,
      "step": 25930
    },
    {
      "epoch": 0.9286443318351926,
      "grad_norm": 1.2545347213745117,
      "learning_rate": 2.6574053413745524e-06,
      "loss": 1.2913,
      "step": 25931
    },
    {
      "epoch": 0.9286801439647608,
      "grad_norm": 1.784175992012024,
      "learning_rate": 2.654749798763645e-06,
      "loss": 1.8439,
      "step": 25932
    },
    {
      "epoch": 0.9287159560943291,
      "grad_norm": 1.5474541187286377,
      "learning_rate": 2.6520955658074997e-06,
      "loss": 1.6509,
      "step": 25933
    },
    {
      "epoch": 0.9287517682238975,
      "grad_norm": 1.63936448097229,
      "learning_rate": 2.649442642541833e-06,
      "loss": 1.5111,
      "step": 25934
    },
    {
      "epoch": 0.9287875803534658,
      "grad_norm": 1.4332921504974365,
      "learning_rate": 2.646791029002349e-06,
      "loss": 1.5307,
      "step": 25935
    },
    {
      "epoch": 0.928823392483034,
      "grad_norm": 1.8409286737442017,
      "learning_rate": 2.6441407252247306e-06,
      "loss": 1.5314,
      "step": 25936
    },
    {
      "epoch": 0.9288592046126023,
      "grad_norm": 1.276275396347046,
      "learning_rate": 2.641491731244605e-06,
      "loss": 1.4106,
      "step": 25937
    },
    {
      "epoch": 0.9288950167421706,
      "grad_norm": 1.8311176300048828,
      "learning_rate": 2.6388440470976217e-06,
      "loss": 1.326,
      "step": 25938
    },
    {
      "epoch": 0.9289308288717388,
      "grad_norm": 2.0258560180664062,
      "learning_rate": 2.6361976728194183e-06,
      "loss": 1.3223,
      "step": 25939
    },
    {
      "epoch": 0.9289666410013071,
      "grad_norm": 1.7356675863265991,
      "learning_rate": 2.6335526084455665e-06,
      "loss": 1.4339,
      "step": 25940
    },
    {
      "epoch": 0.9290024531308755,
      "grad_norm": 1.5074106454849243,
      "learning_rate": 2.630908854011682e-06,
      "loss": 1.4021,
      "step": 25941
    },
    {
      "epoch": 0.9290382652604438,
      "grad_norm": 1.8826628923416138,
      "learning_rate": 2.628266409553315e-06,
      "loss": 1.5917,
      "step": 25942
    },
    {
      "epoch": 0.929074077390012,
      "grad_norm": 1.2302558422088623,
      "learning_rate": 2.625625275106036e-06,
      "loss": 1.2319,
      "step": 25943
    },
    {
      "epoch": 0.9291098895195803,
      "grad_norm": 1.6267286539077759,
      "learning_rate": 2.6229854507053507e-06,
      "loss": 1.6428,
      "step": 25944
    },
    {
      "epoch": 0.9291457016491486,
      "grad_norm": 1.582241415977478,
      "learning_rate": 2.6203469363867973e-06,
      "loss": 1.4682,
      "step": 25945
    },
    {
      "epoch": 0.9291815137787168,
      "grad_norm": 1.7281001806259155,
      "learning_rate": 2.6177097321858578e-06,
      "loss": 1.4495,
      "step": 25946
    },
    {
      "epoch": 0.9292173259082851,
      "grad_norm": 2.26235294342041,
      "learning_rate": 2.615073838138027e-06,
      "loss": 1.5354,
      "step": 25947
    },
    {
      "epoch": 0.9292531380378534,
      "grad_norm": 1.591934084892273,
      "learning_rate": 2.6124392542787645e-06,
      "loss": 1.4247,
      "step": 25948
    },
    {
      "epoch": 0.9292889501674217,
      "grad_norm": 2.2034246921539307,
      "learning_rate": 2.609805980643498e-06,
      "loss": 1.4845,
      "step": 25949
    },
    {
      "epoch": 0.92932476229699,
      "grad_norm": 2.1909966468811035,
      "learning_rate": 2.607174017267677e-06,
      "loss": 1.3116,
      "step": 25950
    },
    {
      "epoch": 0.9293605744265583,
      "grad_norm": 1.33592689037323,
      "learning_rate": 2.6045433641866958e-06,
      "loss": 1.4352,
      "step": 25951
    },
    {
      "epoch": 0.9293963865561266,
      "grad_norm": 1.6754401922225952,
      "learning_rate": 2.6019140214359585e-06,
      "loss": 1.4147,
      "step": 25952
    },
    {
      "epoch": 0.9294321986856948,
      "grad_norm": 2.1295597553253174,
      "learning_rate": 2.599285989050826e-06,
      "loss": 1.6047,
      "step": 25953
    },
    {
      "epoch": 0.9294680108152631,
      "grad_norm": 2.042235851287842,
      "learning_rate": 2.59665926706667e-06,
      "loss": 1.5889,
      "step": 25954
    },
    {
      "epoch": 0.9295038229448314,
      "grad_norm": 1.2845739126205444,
      "learning_rate": 2.594033855518818e-06,
      "loss": 1.4005,
      "step": 25955
    },
    {
      "epoch": 0.9295396350743997,
      "grad_norm": 1.5340960025787354,
      "learning_rate": 2.5914097544425975e-06,
      "loss": 1.5359,
      "step": 25956
    },
    {
      "epoch": 0.929575447203968,
      "grad_norm": 1.4688477516174316,
      "learning_rate": 2.588786963873313e-06,
      "loss": 1.603,
      "step": 25957
    },
    {
      "epoch": 0.9296112593335363,
      "grad_norm": 1.4088783264160156,
      "learning_rate": 2.586165483846248e-06,
      "loss": 1.496,
      "step": 25958
    },
    {
      "epoch": 0.9296470714631045,
      "grad_norm": 1.6192882061004639,
      "learning_rate": 2.5835453143966627e-06,
      "loss": 1.5361,
      "step": 25959
    },
    {
      "epoch": 0.9296828835926728,
      "grad_norm": 1.3497220277786255,
      "learning_rate": 2.580926455559829e-06,
      "loss": 1.3451,
      "step": 25960
    },
    {
      "epoch": 0.9297186957222411,
      "grad_norm": 1.6116409301757812,
      "learning_rate": 2.5783089073709633e-06,
      "loss": 1.3246,
      "step": 25961
    },
    {
      "epoch": 0.9297545078518094,
      "grad_norm": 1.3760795593261719,
      "learning_rate": 2.5756926698652816e-06,
      "loss": 1.3463,
      "step": 25962
    },
    {
      "epoch": 0.9297903199813777,
      "grad_norm": 1.4690639972686768,
      "learning_rate": 2.5730777430779895e-06,
      "loss": 1.2777,
      "step": 25963
    },
    {
      "epoch": 0.929826132110946,
      "grad_norm": 1.1338255405426025,
      "learning_rate": 2.57046412704427e-06,
      "loss": 1.4689,
      "step": 25964
    },
    {
      "epoch": 0.9298619442405143,
      "grad_norm": 2.178570032119751,
      "learning_rate": 2.567851821799283e-06,
      "loss": 1.7218,
      "step": 25965
    },
    {
      "epoch": 0.9298977563700825,
      "grad_norm": 1.6850600242614746,
      "learning_rate": 2.565240827378157e-06,
      "loss": 1.2855,
      "step": 25966
    },
    {
      "epoch": 0.9299335684996508,
      "grad_norm": 1.398482322692871,
      "learning_rate": 2.562631143816041e-06,
      "loss": 1.212,
      "step": 25967
    },
    {
      "epoch": 0.9299693806292191,
      "grad_norm": 1.6965020895004272,
      "learning_rate": 2.560022771148052e-06,
      "loss": 1.4466,
      "step": 25968
    },
    {
      "epoch": 0.9300051927587873,
      "grad_norm": 2.532001256942749,
      "learning_rate": 2.55741570940925e-06,
      "loss": 1.3208,
      "step": 25969
    },
    {
      "epoch": 0.9300410048883557,
      "grad_norm": 1.4785244464874268,
      "learning_rate": 2.5548099586347296e-06,
      "loss": 1.2717,
      "step": 25970
    },
    {
      "epoch": 0.930076817017924,
      "grad_norm": 1.5374987125396729,
      "learning_rate": 2.552205518859552e-06,
      "loss": 1.4995,
      "step": 25971
    },
    {
      "epoch": 0.9301126291474923,
      "grad_norm": 1.8516207933425903,
      "learning_rate": 2.549602390118755e-06,
      "loss": 1.3219,
      "step": 25972
    },
    {
      "epoch": 0.9301484412770605,
      "grad_norm": 1.6778855323791504,
      "learning_rate": 2.5470005724473447e-06,
      "loss": 1.4766,
      "step": 25973
    },
    {
      "epoch": 0.9301842534066288,
      "grad_norm": 1.8335471153259277,
      "learning_rate": 2.544400065880337e-06,
      "loss": 1.6164,
      "step": 25974
    },
    {
      "epoch": 0.9302200655361971,
      "grad_norm": 1.762180209159851,
      "learning_rate": 2.5418008704527263e-06,
      "loss": 1.3349,
      "step": 25975
    },
    {
      "epoch": 0.9302558776657653,
      "grad_norm": 1.5323717594146729,
      "learning_rate": 2.5392029861994625e-06,
      "loss": 1.361,
      "step": 25976
    },
    {
      "epoch": 0.9302916897953337,
      "grad_norm": 1.7938995361328125,
      "learning_rate": 2.5366064131555066e-06,
      "loss": 1.4333,
      "step": 25977
    },
    {
      "epoch": 0.930327501924902,
      "grad_norm": 3.3588671684265137,
      "learning_rate": 2.534011151355797e-06,
      "loss": 1.7233,
      "step": 25978
    },
    {
      "epoch": 0.9303633140544703,
      "grad_norm": 1.9261466264724731,
      "learning_rate": 2.531417200835251e-06,
      "loss": 1.1339,
      "step": 25979
    },
    {
      "epoch": 0.9303991261840385,
      "grad_norm": 2.6531336307525635,
      "learning_rate": 2.528824561628762e-06,
      "loss": 1.3105,
      "step": 25980
    },
    {
      "epoch": 0.9304349383136068,
      "grad_norm": 1.9853758811950684,
      "learning_rate": 2.5262332337712025e-06,
      "loss": 1.5984,
      "step": 25981
    },
    {
      "epoch": 0.9304707504431751,
      "grad_norm": 2.0931293964385986,
      "learning_rate": 2.5236432172974333e-06,
      "loss": 1.4613,
      "step": 25982
    },
    {
      "epoch": 0.9305065625727433,
      "grad_norm": 1.676774263381958,
      "learning_rate": 2.521054512242338e-06,
      "loss": 1.3516,
      "step": 25983
    },
    {
      "epoch": 0.9305423747023117,
      "grad_norm": 2.231872797012329,
      "learning_rate": 2.5184671186406996e-06,
      "loss": 1.4991,
      "step": 25984
    },
    {
      "epoch": 0.93057818683188,
      "grad_norm": 1.5376390218734741,
      "learning_rate": 2.5158810365273345e-06,
      "loss": 1.341,
      "step": 25985
    },
    {
      "epoch": 0.9306139989614483,
      "grad_norm": 1.4266811609268188,
      "learning_rate": 2.5132962659370595e-06,
      "loss": 1.5016,
      "step": 25986
    },
    {
      "epoch": 0.9306498110910165,
      "grad_norm": 1.5460529327392578,
      "learning_rate": 2.510712806904625e-06,
      "loss": 1.1129,
      "step": 25987
    },
    {
      "epoch": 0.9306856232205848,
      "grad_norm": 1.3211203813552856,
      "learning_rate": 2.5081306594647912e-06,
      "loss": 1.3368,
      "step": 25988
    },
    {
      "epoch": 0.9307214353501531,
      "grad_norm": 1.648240566253662,
      "learning_rate": 2.505549823652309e-06,
      "loss": 1.2672,
      "step": 25989
    },
    {
      "epoch": 0.9307572474797213,
      "grad_norm": 1.4202256202697754,
      "learning_rate": 2.5029702995019055e-06,
      "loss": 1.5001,
      "step": 25990
    },
    {
      "epoch": 0.9307930596092897,
      "grad_norm": 1.5760564804077148,
      "learning_rate": 2.5003920870482644e-06,
      "loss": 1.4344,
      "step": 25991
    },
    {
      "epoch": 0.930828871738858,
      "grad_norm": 1.6077864170074463,
      "learning_rate": 2.4978151863260914e-06,
      "loss": 1.4363,
      "step": 25992
    },
    {
      "epoch": 0.9308646838684262,
      "grad_norm": 1.4821381568908691,
      "learning_rate": 2.495239597370047e-06,
      "loss": 1.5983,
      "step": 25993
    },
    {
      "epoch": 0.9309004959979945,
      "grad_norm": 1.120290994644165,
      "learning_rate": 2.492665320214771e-06,
      "loss": 1.4057,
      "step": 25994
    },
    {
      "epoch": 0.9309363081275628,
      "grad_norm": 1.2401329278945923,
      "learning_rate": 2.490092354894913e-06,
      "loss": 1.1569,
      "step": 25995
    },
    {
      "epoch": 0.930972120257131,
      "grad_norm": 1.77287757396698,
      "learning_rate": 2.4875207014450785e-06,
      "loss": 1.5907,
      "step": 25996
    },
    {
      "epoch": 0.9310079323866993,
      "grad_norm": 1.4243831634521484,
      "learning_rate": 2.4849503598998738e-06,
      "loss": 1.4151,
      "step": 25997
    },
    {
      "epoch": 0.9310437445162677,
      "grad_norm": 1.5774353742599487,
      "learning_rate": 2.4823813302938814e-06,
      "loss": 1.3764,
      "step": 25998
    },
    {
      "epoch": 0.931079556645836,
      "grad_norm": 1.5646772384643555,
      "learning_rate": 2.4798136126616634e-06,
      "loss": 1.5245,
      "step": 25999
    },
    {
      "epoch": 0.9311153687754042,
      "grad_norm": 1.583769679069519,
      "learning_rate": 2.477247207037736e-06,
      "loss": 1.7049,
      "step": 26000
    },
    {
      "epoch": 0.9311511809049725,
      "grad_norm": 1.5972886085510254,
      "learning_rate": 2.4746821134566833e-06,
      "loss": 1.2635,
      "step": 26001
    },
    {
      "epoch": 0.9311869930345408,
      "grad_norm": 1.2284495830535889,
      "learning_rate": 2.4721183319529774e-06,
      "loss": 1.3457,
      "step": 26002
    },
    {
      "epoch": 0.931222805164109,
      "grad_norm": 1.774373173713684,
      "learning_rate": 2.4695558625611015e-06,
      "loss": 1.2893,
      "step": 26003
    },
    {
      "epoch": 0.9312586172936773,
      "grad_norm": 2.132659435272217,
      "learning_rate": 2.4669947053155617e-06,
      "loss": 1.2126,
      "step": 26004
    },
    {
      "epoch": 0.9312944294232457,
      "grad_norm": 1.3666845560073853,
      "learning_rate": 2.464434860250786e-06,
      "loss": 1.5706,
      "step": 26005
    },
    {
      "epoch": 0.931330241552814,
      "grad_norm": 1.9008538722991943,
      "learning_rate": 2.461876327401247e-06,
      "loss": 1.3145,
      "step": 26006
    },
    {
      "epoch": 0.9313660536823822,
      "grad_norm": 1.4239145517349243,
      "learning_rate": 2.4593191068013164e-06,
      "loss": 1.3138,
      "step": 26007
    },
    {
      "epoch": 0.9314018658119505,
      "grad_norm": 1.7623815536499023,
      "learning_rate": 2.4567631984854566e-06,
      "loss": 1.5899,
      "step": 26008
    },
    {
      "epoch": 0.9314376779415188,
      "grad_norm": 1.5793737173080444,
      "learning_rate": 2.4542086024880174e-06,
      "loss": 1.5624,
      "step": 26009
    },
    {
      "epoch": 0.931473490071087,
      "grad_norm": 1.4286556243896484,
      "learning_rate": 2.4516553188433823e-06,
      "loss": 0.9605,
      "step": 26010
    },
    {
      "epoch": 0.9315093022006553,
      "grad_norm": 1.671534538269043,
      "learning_rate": 2.4491033475858795e-06,
      "loss": 1.5661,
      "step": 26011
    },
    {
      "epoch": 0.9315451143302237,
      "grad_norm": 2.992291212081909,
      "learning_rate": 2.44655268874987e-06,
      "loss": 1.1977,
      "step": 26012
    },
    {
      "epoch": 0.931580926459792,
      "grad_norm": 1.70440673828125,
      "learning_rate": 2.4440033423696717e-06,
      "loss": 1.4343,
      "step": 26013
    },
    {
      "epoch": 0.9316167385893602,
      "grad_norm": 2.031369924545288,
      "learning_rate": 2.4414553084795455e-06,
      "loss": 1.4702,
      "step": 26014
    },
    {
      "epoch": 0.9316525507189285,
      "grad_norm": 1.6914634704589844,
      "learning_rate": 2.4389085871138086e-06,
      "loss": 1.4862,
      "step": 26015
    },
    {
      "epoch": 0.9316883628484968,
      "grad_norm": 2.1513466835021973,
      "learning_rate": 2.4363631783067108e-06,
      "loss": 1.523,
      "step": 26016
    },
    {
      "epoch": 0.931724174978065,
      "grad_norm": 1.6219446659088135,
      "learning_rate": 2.4338190820925145e-06,
      "loss": 1.4038,
      "step": 26017
    },
    {
      "epoch": 0.9317599871076333,
      "grad_norm": 1.5040446519851685,
      "learning_rate": 2.4312762985054137e-06,
      "loss": 1.3775,
      "step": 26018
    },
    {
      "epoch": 0.9317957992372017,
      "grad_norm": 3.2530477046966553,
      "learning_rate": 2.4287348275796373e-06,
      "loss": 1.0825,
      "step": 26019
    },
    {
      "epoch": 0.93183161136677,
      "grad_norm": 1.5604746341705322,
      "learning_rate": 2.4261946693493797e-06,
      "loss": 1.0338,
      "step": 26020
    },
    {
      "epoch": 0.9318674234963382,
      "grad_norm": 2.0306684970855713,
      "learning_rate": 2.4236558238488025e-06,
      "loss": 1.4578,
      "step": 26021
    },
    {
      "epoch": 0.9319032356259065,
      "grad_norm": 2.1134817600250244,
      "learning_rate": 2.421118291112079e-06,
      "loss": 1.5503,
      "step": 26022
    },
    {
      "epoch": 0.9319390477554748,
      "grad_norm": 1.4790269136428833,
      "learning_rate": 2.4185820711733363e-06,
      "loss": 1.5024,
      "step": 26023
    },
    {
      "epoch": 0.931974859885043,
      "grad_norm": 1.7622898817062378,
      "learning_rate": 2.4160471640667147e-06,
      "loss": 1.5454,
      "step": 26024
    },
    {
      "epoch": 0.9320106720146113,
      "grad_norm": 1.9339481592178345,
      "learning_rate": 2.413513569826298e-06,
      "loss": 1.3688,
      "step": 26025
    },
    {
      "epoch": 0.9320464841441797,
      "grad_norm": 1.6316418647766113,
      "learning_rate": 2.410981288486169e-06,
      "loss": 1.4478,
      "step": 26026
    },
    {
      "epoch": 0.932082296273748,
      "grad_norm": 1.73930823802948,
      "learning_rate": 2.408450320080413e-06,
      "loss": 1.5425,
      "step": 26027
    },
    {
      "epoch": 0.9321181084033162,
      "grad_norm": 1.9473265409469604,
      "learning_rate": 2.40592066464308e-06,
      "loss": 1.5209,
      "step": 26028
    },
    {
      "epoch": 0.9321539205328845,
      "grad_norm": 1.8710750341415405,
      "learning_rate": 2.4033923222081868e-06,
      "loss": 1.3462,
      "step": 26029
    },
    {
      "epoch": 0.9321897326624528,
      "grad_norm": 1.5108715295791626,
      "learning_rate": 2.400865292809762e-06,
      "loss": 1.4425,
      "step": 26030
    },
    {
      "epoch": 0.932225544792021,
      "grad_norm": 1.6120551824569702,
      "learning_rate": 2.3983395764818008e-06,
      "loss": 1.2536,
      "step": 26031
    },
    {
      "epoch": 0.9322613569215893,
      "grad_norm": 1.3649541139602661,
      "learning_rate": 2.395815173258287e-06,
      "loss": 1.34,
      "step": 26032
    },
    {
      "epoch": 0.9322971690511577,
      "grad_norm": 1.4536439180374146,
      "learning_rate": 2.393292083173171e-06,
      "loss": 1.5061,
      "step": 26033
    },
    {
      "epoch": 0.9323329811807259,
      "grad_norm": 1.6911566257476807,
      "learning_rate": 2.390770306260415e-06,
      "loss": 1.3117,
      "step": 26034
    },
    {
      "epoch": 0.9323687933102942,
      "grad_norm": 1.714871883392334,
      "learning_rate": 2.388249842553936e-06,
      "loss": 1.5929,
      "step": 26035
    },
    {
      "epoch": 0.9324046054398625,
      "grad_norm": 1.709179162979126,
      "learning_rate": 2.385730692087651e-06,
      "loss": 1.5818,
      "step": 26036
    },
    {
      "epoch": 0.9324404175694307,
      "grad_norm": 1.9201855659484863,
      "learning_rate": 2.3832128548954334e-06,
      "loss": 1.3567,
      "step": 26037
    },
    {
      "epoch": 0.932476229698999,
      "grad_norm": 1.8978859186172485,
      "learning_rate": 2.3806963310111786e-06,
      "loss": 1.6552,
      "step": 26038
    },
    {
      "epoch": 0.9325120418285673,
      "grad_norm": 1.9330658912658691,
      "learning_rate": 2.3781811204687367e-06,
      "loss": 1.3294,
      "step": 26039
    },
    {
      "epoch": 0.9325478539581357,
      "grad_norm": 1.352783441543579,
      "learning_rate": 2.375667223301936e-06,
      "loss": 1.5671,
      "step": 26040
    },
    {
      "epoch": 0.9325836660877039,
      "grad_norm": 1.6923365592956543,
      "learning_rate": 2.3731546395446056e-06,
      "loss": 1.4242,
      "step": 26041
    },
    {
      "epoch": 0.9326194782172722,
      "grad_norm": 1.456566333770752,
      "learning_rate": 2.370643369230563e-06,
      "loss": 1.8199,
      "step": 26042
    },
    {
      "epoch": 0.9326552903468405,
      "grad_norm": 1.8373398780822754,
      "learning_rate": 2.3681334123935805e-06,
      "loss": 1.5232,
      "step": 26043
    },
    {
      "epoch": 0.9326911024764087,
      "grad_norm": 1.4119781255722046,
      "learning_rate": 2.3656247690674092e-06,
      "loss": 1.5827,
      "step": 26044
    },
    {
      "epoch": 0.932726914605977,
      "grad_norm": 1.4931371212005615,
      "learning_rate": 2.3631174392858335e-06,
      "loss": 1.5398,
      "step": 26045
    },
    {
      "epoch": 0.9327627267355453,
      "grad_norm": 2.504749298095703,
      "learning_rate": 2.3606114230825704e-06,
      "loss": 1.3982,
      "step": 26046
    },
    {
      "epoch": 0.9327985388651137,
      "grad_norm": 1.844650149345398,
      "learning_rate": 2.3581067204913267e-06,
      "loss": 1.3822,
      "step": 26047
    },
    {
      "epoch": 0.9328343509946819,
      "grad_norm": 1.961634635925293,
      "learning_rate": 2.355603331545808e-06,
      "loss": 1.5726,
      "step": 26048
    },
    {
      "epoch": 0.9328701631242502,
      "grad_norm": 1.4602668285369873,
      "learning_rate": 2.3531012562796995e-06,
      "loss": 1.3054,
      "step": 26049
    },
    {
      "epoch": 0.9329059752538185,
      "grad_norm": 1.6050527095794678,
      "learning_rate": 2.3506004947266512e-06,
      "loss": 1.5448,
      "step": 26050
    },
    {
      "epoch": 0.9329417873833867,
      "grad_norm": 1.3830983638763428,
      "learning_rate": 2.3481010469203256e-06,
      "loss": 1.141,
      "step": 26051
    },
    {
      "epoch": 0.932977599512955,
      "grad_norm": 1.515015721321106,
      "learning_rate": 2.345602912894329e-06,
      "loss": 1.4587,
      "step": 26052
    },
    {
      "epoch": 0.9330134116425233,
      "grad_norm": 2.136948347091675,
      "learning_rate": 2.3431060926822903e-06,
      "loss": 1.4007,
      "step": 26053
    },
    {
      "epoch": 0.9330492237720917,
      "grad_norm": 1.7469984292984009,
      "learning_rate": 2.340610586317782e-06,
      "loss": 1.6419,
      "step": 26054
    },
    {
      "epoch": 0.9330850359016599,
      "grad_norm": 1.5936490297317505,
      "learning_rate": 2.3381163938343776e-06,
      "loss": 1.5191,
      "step": 26055
    },
    {
      "epoch": 0.9331208480312282,
      "grad_norm": 1.4409581422805786,
      "learning_rate": 2.3356235152656613e-06,
      "loss": 1.4645,
      "step": 26056
    },
    {
      "epoch": 0.9331566601607965,
      "grad_norm": 1.6994661092758179,
      "learning_rate": 2.33313195064514e-06,
      "loss": 1.4291,
      "step": 26057
    },
    {
      "epoch": 0.9331924722903647,
      "grad_norm": 2.277144193649292,
      "learning_rate": 2.330641700006353e-06,
      "loss": 1.7529,
      "step": 26058
    },
    {
      "epoch": 0.933228284419933,
      "grad_norm": 1.867384910583496,
      "learning_rate": 2.328152763382796e-06,
      "loss": 1.3866,
      "step": 26059
    },
    {
      "epoch": 0.9332640965495013,
      "grad_norm": 1.7690753936767578,
      "learning_rate": 2.325665140807964e-06,
      "loss": 1.6213,
      "step": 26060
    },
    {
      "epoch": 0.9332999086790696,
      "grad_norm": 1.4293922185897827,
      "learning_rate": 2.323178832315298e-06,
      "loss": 1.381,
      "step": 26061
    },
    {
      "epoch": 0.9333357208086379,
      "grad_norm": 1.8195722103118896,
      "learning_rate": 2.3206938379382813e-06,
      "loss": 1.5304,
      "step": 26062
    },
    {
      "epoch": 0.9333715329382062,
      "grad_norm": 1.5737478733062744,
      "learning_rate": 2.318210157710332e-06,
      "loss": 1.3585,
      "step": 26063
    },
    {
      "epoch": 0.9334073450677745,
      "grad_norm": 1.6605415344238281,
      "learning_rate": 2.3157277916648567e-06,
      "loss": 1.1337,
      "step": 26064
    },
    {
      "epoch": 0.9334431571973427,
      "grad_norm": 1.6333032846450806,
      "learning_rate": 2.313246739835262e-06,
      "loss": 1.4887,
      "step": 26065
    },
    {
      "epoch": 0.933478969326911,
      "grad_norm": 1.589296579360962,
      "learning_rate": 2.3107670022549323e-06,
      "loss": 1.6372,
      "step": 26066
    },
    {
      "epoch": 0.9335147814564793,
      "grad_norm": 1.6948356628417969,
      "learning_rate": 2.3082885789572182e-06,
      "loss": 1.316,
      "step": 26067
    },
    {
      "epoch": 0.9335505935860476,
      "grad_norm": 1.5151407718658447,
      "learning_rate": 2.305811469975472e-06,
      "loss": 1.4232,
      "step": 26068
    },
    {
      "epoch": 0.9335864057156159,
      "grad_norm": 1.8046801090240479,
      "learning_rate": 2.30333567534301e-06,
      "loss": 1.536,
      "step": 26069
    },
    {
      "epoch": 0.9336222178451842,
      "grad_norm": 1.2985812425613403,
      "learning_rate": 2.3008611950931404e-06,
      "loss": 1.5431,
      "step": 26070
    },
    {
      "epoch": 0.9336580299747524,
      "grad_norm": 1.8768430948257446,
      "learning_rate": 2.2983880292591798e-06,
      "loss": 1.6994,
      "step": 26071
    },
    {
      "epoch": 0.9336938421043207,
      "grad_norm": 2.3779830932617188,
      "learning_rate": 2.295916177874369e-06,
      "loss": 1.2774,
      "step": 26072
    },
    {
      "epoch": 0.933729654233889,
      "grad_norm": 2.449038028717041,
      "learning_rate": 2.2934456409719698e-06,
      "loss": 1.2834,
      "step": 26073
    },
    {
      "epoch": 0.9337654663634573,
      "grad_norm": 1.443924069404602,
      "learning_rate": 2.2909764185852447e-06,
      "loss": 1.166,
      "step": 26074
    },
    {
      "epoch": 0.9338012784930256,
      "grad_norm": 1.5654816627502441,
      "learning_rate": 2.288508510747389e-06,
      "loss": 1.4546,
      "step": 26075
    },
    {
      "epoch": 0.9338370906225939,
      "grad_norm": 1.7187449932098389,
      "learning_rate": 2.2860419174916104e-06,
      "loss": 1.4493,
      "step": 26076
    },
    {
      "epoch": 0.9338729027521622,
      "grad_norm": 1.4201176166534424,
      "learning_rate": 2.2835766388510926e-06,
      "loss": 1.2746,
      "step": 26077
    },
    {
      "epoch": 0.9339087148817304,
      "grad_norm": 1.4719351530075073,
      "learning_rate": 2.2811126748590207e-06,
      "loss": 1.6015,
      "step": 26078
    },
    {
      "epoch": 0.9339445270112987,
      "grad_norm": 1.7975047826766968,
      "learning_rate": 2.278650025548512e-06,
      "loss": 1.4907,
      "step": 26079
    },
    {
      "epoch": 0.933980339140867,
      "grad_norm": 2.0674078464508057,
      "learning_rate": 2.2761886909527187e-06,
      "loss": 1.4675,
      "step": 26080
    },
    {
      "epoch": 0.9340161512704352,
      "grad_norm": 1.5718308687210083,
      "learning_rate": 2.273728671104769e-06,
      "loss": 1.4185,
      "step": 26081
    },
    {
      "epoch": 0.9340519634000036,
      "grad_norm": 1.3587315082550049,
      "learning_rate": 2.271269966037726e-06,
      "loss": 1.3692,
      "step": 26082
    },
    {
      "epoch": 0.9340877755295719,
      "grad_norm": 2.1399447917938232,
      "learning_rate": 2.2688125757846957e-06,
      "loss": 1.3626,
      "step": 26083
    },
    {
      "epoch": 0.9341235876591402,
      "grad_norm": 2.119439125061035,
      "learning_rate": 2.2663565003787078e-06,
      "loss": 1.5303,
      "step": 26084
    },
    {
      "epoch": 0.9341593997887084,
      "grad_norm": 1.2838094234466553,
      "learning_rate": 2.263901739852847e-06,
      "loss": 1.4867,
      "step": 26085
    },
    {
      "epoch": 0.9341952119182767,
      "grad_norm": 1.8111953735351562,
      "learning_rate": 2.2614482942400984e-06,
      "loss": 1.2078,
      "step": 26086
    },
    {
      "epoch": 0.934231024047845,
      "grad_norm": 1.9275093078613281,
      "learning_rate": 2.2589961635735015e-06,
      "loss": 1.512,
      "step": 26087
    },
    {
      "epoch": 0.9342668361774132,
      "grad_norm": 1.3588461875915527,
      "learning_rate": 2.2565453478860297e-06,
      "loss": 1.1154,
      "step": 26088
    },
    {
      "epoch": 0.9343026483069816,
      "grad_norm": 1.3044812679290771,
      "learning_rate": 2.254095847210669e-06,
      "loss": 1.4511,
      "step": 26089
    },
    {
      "epoch": 0.9343384604365499,
      "grad_norm": 1.3384042978286743,
      "learning_rate": 2.2516476615803694e-06,
      "loss": 1.2449,
      "step": 26090
    },
    {
      "epoch": 0.9343742725661182,
      "grad_norm": 1.543078899383545,
      "learning_rate": 2.249200791028039e-06,
      "loss": 1.4121,
      "step": 26091
    },
    {
      "epoch": 0.9344100846956864,
      "grad_norm": 1.5723681449890137,
      "learning_rate": 2.2467552355866505e-06,
      "loss": 1.5712,
      "step": 26092
    },
    {
      "epoch": 0.9344458968252547,
      "grad_norm": 1.4810062646865845,
      "learning_rate": 2.2443109952890674e-06,
      "loss": 1.3736,
      "step": 26093
    },
    {
      "epoch": 0.934481708954823,
      "grad_norm": 1.5201414823532104,
      "learning_rate": 2.241868070168185e-06,
      "loss": 1.3559,
      "step": 26094
    },
    {
      "epoch": 0.9345175210843912,
      "grad_norm": 1.811244010925293,
      "learning_rate": 2.239426460256855e-06,
      "loss": 1.2936,
      "step": 26095
    },
    {
      "epoch": 0.9345533332139596,
      "grad_norm": 1.8156101703643799,
      "learning_rate": 2.236986165587951e-06,
      "loss": 1.601,
      "step": 26096
    },
    {
      "epoch": 0.9345891453435279,
      "grad_norm": 1.2325727939605713,
      "learning_rate": 2.2345471861942914e-06,
      "loss": 1.2936,
      "step": 26097
    },
    {
      "epoch": 0.9346249574730962,
      "grad_norm": 1.9929440021514893,
      "learning_rate": 2.232109522108694e-06,
      "loss": 1.3793,
      "step": 26098
    },
    {
      "epoch": 0.9346607696026644,
      "grad_norm": 1.516374111175537,
      "learning_rate": 2.2296731733639552e-06,
      "loss": 1.2078,
      "step": 26099
    },
    {
      "epoch": 0.9346965817322327,
      "grad_norm": 3.2850093841552734,
      "learning_rate": 2.227238139992849e-06,
      "loss": 1.2199,
      "step": 26100
    },
    {
      "epoch": 0.934732393861801,
      "grad_norm": 1.9196382761001587,
      "learning_rate": 2.224804422028137e-06,
      "loss": 1.4492,
      "step": 26101
    },
    {
      "epoch": 0.9347682059913692,
      "grad_norm": 1.7088524103164673,
      "learning_rate": 2.2223720195025386e-06,
      "loss": 1.8744,
      "step": 26102
    },
    {
      "epoch": 0.9348040181209376,
      "grad_norm": 1.5560927391052246,
      "learning_rate": 2.2199409324488275e-06,
      "loss": 1.2936,
      "step": 26103
    },
    {
      "epoch": 0.9348398302505059,
      "grad_norm": 1.8463906049728394,
      "learning_rate": 2.2175111608996657e-06,
      "loss": 1.4824,
      "step": 26104
    },
    {
      "epoch": 0.9348756423800741,
      "grad_norm": 1.8507603406906128,
      "learning_rate": 2.215082704887772e-06,
      "loss": 1.5644,
      "step": 26105
    },
    {
      "epoch": 0.9349114545096424,
      "grad_norm": 1.5813714265823364,
      "learning_rate": 2.212655564445798e-06,
      "loss": 1.2692,
      "step": 26106
    },
    {
      "epoch": 0.9349472666392107,
      "grad_norm": 1.5400943756103516,
      "learning_rate": 2.2102297396064176e-06,
      "loss": 1.3857,
      "step": 26107
    },
    {
      "epoch": 0.934983078768779,
      "grad_norm": 1.7825454473495483,
      "learning_rate": 2.20780523040226e-06,
      "loss": 1.1498,
      "step": 26108
    },
    {
      "epoch": 0.9350188908983472,
      "grad_norm": 1.7498632669448853,
      "learning_rate": 2.2053820368659215e-06,
      "loss": 1.4762,
      "step": 26109
    },
    {
      "epoch": 0.9350547030279156,
      "grad_norm": 1.3450285196304321,
      "learning_rate": 2.202960159030032e-06,
      "loss": 1.3703,
      "step": 26110
    },
    {
      "epoch": 0.9350905151574839,
      "grad_norm": 1.6363308429718018,
      "learning_rate": 2.200539596927165e-06,
      "loss": 1.3916,
      "step": 26111
    },
    {
      "epoch": 0.9351263272870521,
      "grad_norm": 1.3470873832702637,
      "learning_rate": 2.1981203505898827e-06,
      "loss": 1.4387,
      "step": 26112
    },
    {
      "epoch": 0.9351621394166204,
      "grad_norm": 1.7252334356307983,
      "learning_rate": 2.195702420050727e-06,
      "loss": 1.7303,
      "step": 26113
    },
    {
      "epoch": 0.9351979515461887,
      "grad_norm": 1.8090825080871582,
      "learning_rate": 2.19328580534226e-06,
      "loss": 1.1771,
      "step": 26114
    },
    {
      "epoch": 0.935233763675757,
      "grad_norm": 1.476894736289978,
      "learning_rate": 2.190870506496956e-06,
      "loss": 1.4577,
      "step": 26115
    },
    {
      "epoch": 0.9352695758053252,
      "grad_norm": 1.3571605682373047,
      "learning_rate": 2.188456523547322e-06,
      "loss": 1.389,
      "step": 26116
    },
    {
      "epoch": 0.9353053879348936,
      "grad_norm": 1.636551856994629,
      "learning_rate": 2.1860438565258433e-06,
      "loss": 1.1224,
      "step": 26117
    },
    {
      "epoch": 0.9353412000644619,
      "grad_norm": 1.9286612272262573,
      "learning_rate": 2.183632505464972e-06,
      "loss": 1.4867,
      "step": 26118
    },
    {
      "epoch": 0.9353770121940301,
      "grad_norm": 1.419366717338562,
      "learning_rate": 2.1812224703971597e-06,
      "loss": 1.399,
      "step": 26119
    },
    {
      "epoch": 0.9354128243235984,
      "grad_norm": 1.7759507894515991,
      "learning_rate": 2.1788137513548134e-06,
      "loss": 1.1488,
      "step": 26120
    },
    {
      "epoch": 0.9354486364531667,
      "grad_norm": 1.542161464691162,
      "learning_rate": 2.176406348370341e-06,
      "loss": 1.4903,
      "step": 26121
    },
    {
      "epoch": 0.9354844485827349,
      "grad_norm": 1.8545513153076172,
      "learning_rate": 2.174000261476139e-06,
      "loss": 1.3911,
      "step": 26122
    },
    {
      "epoch": 0.9355202607123032,
      "grad_norm": 1.7891281843185425,
      "learning_rate": 2.171595490704592e-06,
      "loss": 1.4406,
      "step": 26123
    },
    {
      "epoch": 0.9355560728418716,
      "grad_norm": 1.7108030319213867,
      "learning_rate": 2.1691920360880303e-06,
      "loss": 1.1197,
      "step": 26124
    },
    {
      "epoch": 0.9355918849714399,
      "grad_norm": 1.6328608989715576,
      "learning_rate": 2.166789897658794e-06,
      "loss": 1.5987,
      "step": 26125
    },
    {
      "epoch": 0.9356276971010081,
      "grad_norm": 1.4279072284698486,
      "learning_rate": 2.1643890754492136e-06,
      "loss": 1.3825,
      "step": 26126
    },
    {
      "epoch": 0.9356635092305764,
      "grad_norm": 1.4188889265060425,
      "learning_rate": 2.1619895694915624e-06,
      "loss": 1.826,
      "step": 26127
    },
    {
      "epoch": 0.9356993213601447,
      "grad_norm": 1.5467236042022705,
      "learning_rate": 2.159591379818149e-06,
      "loss": 1.4663,
      "step": 26128
    },
    {
      "epoch": 0.9357351334897129,
      "grad_norm": 1.9561902284622192,
      "learning_rate": 2.1571945064612243e-06,
      "loss": 1.3251,
      "step": 26129
    },
    {
      "epoch": 0.9357709456192812,
      "grad_norm": 1.4215346574783325,
      "learning_rate": 2.1547989494530517e-06,
      "loss": 1.3927,
      "step": 26130
    },
    {
      "epoch": 0.9358067577488496,
      "grad_norm": 1.427872896194458,
      "learning_rate": 2.1524047088258394e-06,
      "loss": 1.5345,
      "step": 26131
    },
    {
      "epoch": 0.9358425698784179,
      "grad_norm": 1.6102819442749023,
      "learning_rate": 2.1500117846118053e-06,
      "loss": 1.3043,
      "step": 26132
    },
    {
      "epoch": 0.9358783820079861,
      "grad_norm": 1.5477509498596191,
      "learning_rate": 2.147620176843157e-06,
      "loss": 1.0714,
      "step": 26133
    },
    {
      "epoch": 0.9359141941375544,
      "grad_norm": 1.82809317111969,
      "learning_rate": 2.145229885552047e-06,
      "loss": 1.4888,
      "step": 26134
    },
    {
      "epoch": 0.9359500062671227,
      "grad_norm": 2.383084535598755,
      "learning_rate": 2.142840910770638e-06,
      "loss": 1.1927,
      "step": 26135
    },
    {
      "epoch": 0.9359858183966909,
      "grad_norm": 2.7783987522125244,
      "learning_rate": 2.140453252531083e-06,
      "loss": 1.5051,
      "step": 26136
    },
    {
      "epoch": 0.9360216305262592,
      "grad_norm": 1.496717929840088,
      "learning_rate": 2.1380669108655105e-06,
      "loss": 1.25,
      "step": 26137
    },
    {
      "epoch": 0.9360574426558276,
      "grad_norm": 1.7770413160324097,
      "learning_rate": 2.135681885806007e-06,
      "loss": 1.5141,
      "step": 26138
    },
    {
      "epoch": 0.9360932547853958,
      "grad_norm": 1.5932000875473022,
      "learning_rate": 2.133298177384668e-06,
      "loss": 1.4161,
      "step": 26139
    },
    {
      "epoch": 0.9361290669149641,
      "grad_norm": 1.4183354377746582,
      "learning_rate": 2.1309157856335694e-06,
      "loss": 1.6704,
      "step": 26140
    },
    {
      "epoch": 0.9361648790445324,
      "grad_norm": 2.0836100578308105,
      "learning_rate": 2.128534710584751e-06,
      "loss": 1.1774,
      "step": 26141
    },
    {
      "epoch": 0.9362006911741007,
      "grad_norm": 1.5669517517089844,
      "learning_rate": 2.126154952270254e-06,
      "loss": 1.4349,
      "step": 26142
    },
    {
      "epoch": 0.9362365033036689,
      "grad_norm": 1.7547202110290527,
      "learning_rate": 2.1237765107220973e-06,
      "loss": 1.235,
      "step": 26143
    },
    {
      "epoch": 0.9362723154332372,
      "grad_norm": 1.7867295742034912,
      "learning_rate": 2.121399385972278e-06,
      "loss": 1.0921,
      "step": 26144
    },
    {
      "epoch": 0.9363081275628056,
      "grad_norm": 1.2874763011932373,
      "learning_rate": 2.11902357805277e-06,
      "loss": 1.058,
      "step": 26145
    },
    {
      "epoch": 0.9363439396923738,
      "grad_norm": 1.6540254354476929,
      "learning_rate": 2.116649086995537e-06,
      "loss": 1.5833,
      "step": 26146
    },
    {
      "epoch": 0.9363797518219421,
      "grad_norm": 1.5691769123077393,
      "learning_rate": 2.1142759128325306e-06,
      "loss": 1.2805,
      "step": 26147
    },
    {
      "epoch": 0.9364155639515104,
      "grad_norm": 2.247567653656006,
      "learning_rate": 2.1119040555956925e-06,
      "loss": 1.2188,
      "step": 26148
    },
    {
      "epoch": 0.9364513760810786,
      "grad_norm": 1.526218295097351,
      "learning_rate": 2.109533515316908e-06,
      "loss": 1.198,
      "step": 26149
    },
    {
      "epoch": 0.9364871882106469,
      "grad_norm": 1.5749701261520386,
      "learning_rate": 2.1071642920280855e-06,
      "loss": 1.3176,
      "step": 26150
    },
    {
      "epoch": 0.9365230003402152,
      "grad_norm": 1.655874490737915,
      "learning_rate": 2.1047963857610986e-06,
      "loss": 1.3957,
      "step": 26151
    },
    {
      "epoch": 0.9365588124697836,
      "grad_norm": 1.9014919996261597,
      "learning_rate": 2.102429796547789e-06,
      "loss": 1.6384,
      "step": 26152
    },
    {
      "epoch": 0.9365946245993518,
      "grad_norm": 1.5637357234954834,
      "learning_rate": 2.100064524420009e-06,
      "loss": 1.303,
      "step": 26153
    },
    {
      "epoch": 0.9366304367289201,
      "grad_norm": 2.0295560359954834,
      "learning_rate": 2.0977005694095774e-06,
      "loss": 1.3744,
      "step": 26154
    },
    {
      "epoch": 0.9366662488584884,
      "grad_norm": 2.1295666694641113,
      "learning_rate": 2.0953379315483134e-06,
      "loss": 1.2995,
      "step": 26155
    },
    {
      "epoch": 0.9367020609880566,
      "grad_norm": 1.3736835718154907,
      "learning_rate": 2.0929766108679803e-06,
      "loss": 1.4379,
      "step": 26156
    },
    {
      "epoch": 0.9367378731176249,
      "grad_norm": 1.850738763809204,
      "learning_rate": 2.0906166074003532e-06,
      "loss": 1.3111,
      "step": 26157
    },
    {
      "epoch": 0.9367736852471932,
      "grad_norm": 1.3870395421981812,
      "learning_rate": 2.0882579211771837e-06,
      "loss": 1.3912,
      "step": 26158
    },
    {
      "epoch": 0.9368094973767616,
      "grad_norm": 1.872261881828308,
      "learning_rate": 2.0859005522302245e-06,
      "loss": 1.5057,
      "step": 26159
    },
    {
      "epoch": 0.9368453095063298,
      "grad_norm": 1.5025333166122437,
      "learning_rate": 2.0835445005911503e-06,
      "loss": 1.2798,
      "step": 26160
    },
    {
      "epoch": 0.9368811216358981,
      "grad_norm": 1.5420435667037964,
      "learning_rate": 2.081189766291691e-06,
      "loss": 1.303,
      "step": 26161
    },
    {
      "epoch": 0.9369169337654664,
      "grad_norm": 1.7748241424560547,
      "learning_rate": 2.0788363493635333e-06,
      "loss": 1.4483,
      "step": 26162
    },
    {
      "epoch": 0.9369527458950346,
      "grad_norm": 1.7625503540039062,
      "learning_rate": 2.0764842498383063e-06,
      "loss": 1.348,
      "step": 26163
    },
    {
      "epoch": 0.9369885580246029,
      "grad_norm": 1.4400793313980103,
      "learning_rate": 2.074133467747663e-06,
      "loss": 1.6231,
      "step": 26164
    },
    {
      "epoch": 0.9370243701541712,
      "grad_norm": 1.3040224313735962,
      "learning_rate": 2.071784003123256e-06,
      "loss": 1.4208,
      "step": 26165
    },
    {
      "epoch": 0.9370601822837396,
      "grad_norm": 1.9658088684082031,
      "learning_rate": 2.069435855996671e-06,
      "loss": 1.5414,
      "step": 26166
    },
    {
      "epoch": 0.9370959944133078,
      "grad_norm": 1.4681401252746582,
      "learning_rate": 2.0670890263995047e-06,
      "loss": 1.1518,
      "step": 26167
    },
    {
      "epoch": 0.9371318065428761,
      "grad_norm": 1.5593922138214111,
      "learning_rate": 2.0647435143633322e-06,
      "loss": 1.3844,
      "step": 26168
    },
    {
      "epoch": 0.9371676186724444,
      "grad_norm": 1.4829758405685425,
      "learning_rate": 2.0623993199197055e-06,
      "loss": 1.4184,
      "step": 26169
    },
    {
      "epoch": 0.9372034308020126,
      "grad_norm": 1.5997012853622437,
      "learning_rate": 2.0600564431001668e-06,
      "loss": 1.2306,
      "step": 26170
    },
    {
      "epoch": 0.9372392429315809,
      "grad_norm": 1.61320960521698,
      "learning_rate": 2.057714883936235e-06,
      "loss": 1.3906,
      "step": 26171
    },
    {
      "epoch": 0.9372750550611492,
      "grad_norm": 2.0251548290252686,
      "learning_rate": 2.0553746424594065e-06,
      "loss": 1.2477,
      "step": 26172
    },
    {
      "epoch": 0.9373108671907175,
      "grad_norm": 1.9987709522247314,
      "learning_rate": 2.0530357187011907e-06,
      "loss": 1.3705,
      "step": 26173
    },
    {
      "epoch": 0.9373466793202858,
      "grad_norm": 1.7930504083633423,
      "learning_rate": 2.050698112693028e-06,
      "loss": 1.1294,
      "step": 26174
    },
    {
      "epoch": 0.9373824914498541,
      "grad_norm": 2.174814462661743,
      "learning_rate": 2.0483618244663714e-06,
      "loss": 1.3507,
      "step": 26175
    },
    {
      "epoch": 0.9374183035794224,
      "grad_norm": 2.3176262378692627,
      "learning_rate": 2.0460268540526518e-06,
      "loss": 1.405,
      "step": 26176
    },
    {
      "epoch": 0.9374541157089906,
      "grad_norm": 2.033170223236084,
      "learning_rate": 2.04369320148331e-06,
      "loss": 1.5684,
      "step": 26177
    },
    {
      "epoch": 0.9374899278385589,
      "grad_norm": 1.7996870279312134,
      "learning_rate": 2.041360866789721e-06,
      "loss": 1.1019,
      "step": 26178
    },
    {
      "epoch": 0.9375257399681272,
      "grad_norm": 2.1224141120910645,
      "learning_rate": 2.0390298500032377e-06,
      "loss": 1.1435,
      "step": 26179
    },
    {
      "epoch": 0.9375615520976955,
      "grad_norm": 1.8272570371627808,
      "learning_rate": 2.0367001511552685e-06,
      "loss": 1.5015,
      "step": 26180
    },
    {
      "epoch": 0.9375973642272638,
      "grad_norm": 1.5879367589950562,
      "learning_rate": 2.0343717702771325e-06,
      "loss": 1.3781,
      "step": 26181
    },
    {
      "epoch": 0.9376331763568321,
      "grad_norm": 1.2769107818603516,
      "learning_rate": 2.0320447074001492e-06,
      "loss": 1.6663,
      "step": 26182
    },
    {
      "epoch": 0.9376689884864003,
      "grad_norm": 1.972424864768982,
      "learning_rate": 2.0297189625556377e-06,
      "loss": 1.2255,
      "step": 26183
    },
    {
      "epoch": 0.9377048006159686,
      "grad_norm": 1.7590620517730713,
      "learning_rate": 2.027394535774896e-06,
      "loss": 1.394,
      "step": 26184
    },
    {
      "epoch": 0.9377406127455369,
      "grad_norm": 1.6972074508666992,
      "learning_rate": 2.0250714270891757e-06,
      "loss": 1.5186,
      "step": 26185
    },
    {
      "epoch": 0.9377764248751052,
      "grad_norm": 1.7677464485168457,
      "learning_rate": 2.0227496365297304e-06,
      "loss": 1.415,
      "step": 26186
    },
    {
      "epoch": 0.9378122370046735,
      "grad_norm": 1.1106623411178589,
      "learning_rate": 2.020429164127835e-06,
      "loss": 0.9381,
      "step": 26187
    },
    {
      "epoch": 0.9378480491342418,
      "grad_norm": 1.5314291715621948,
      "learning_rate": 2.0181100099146533e-06,
      "loss": 1.1579,
      "step": 26188
    },
    {
      "epoch": 0.9378838612638101,
      "grad_norm": 1.614284873008728,
      "learning_rate": 2.015792173921438e-06,
      "loss": 1.1576,
      "step": 26189
    },
    {
      "epoch": 0.9379196733933783,
      "grad_norm": 1.5244933366775513,
      "learning_rate": 2.01347565617932e-06,
      "loss": 1.3892,
      "step": 26190
    },
    {
      "epoch": 0.9379554855229466,
      "grad_norm": 1.5229711532592773,
      "learning_rate": 2.0111604567195185e-06,
      "loss": 1.249,
      "step": 26191
    },
    {
      "epoch": 0.9379912976525149,
      "grad_norm": 1.5737578868865967,
      "learning_rate": 2.008846575573142e-06,
      "loss": 1.0932,
      "step": 26192
    },
    {
      "epoch": 0.9380271097820831,
      "grad_norm": 1.5744316577911377,
      "learning_rate": 2.006534012771344e-06,
      "loss": 1.4432,
      "step": 26193
    },
    {
      "epoch": 0.9380629219116515,
      "grad_norm": 1.4843591451644897,
      "learning_rate": 2.004222768345221e-06,
      "loss": 1.482,
      "step": 26194
    },
    {
      "epoch": 0.9380987340412198,
      "grad_norm": 1.2316222190856934,
      "learning_rate": 2.0019128423258816e-06,
      "loss": 1.3807,
      "step": 26195
    },
    {
      "epoch": 0.9381345461707881,
      "grad_norm": 1.4942682981491089,
      "learning_rate": 1.999604234744401e-06,
      "loss": 1.4272,
      "step": 26196
    },
    {
      "epoch": 0.9381703583003563,
      "grad_norm": 2.2705743312835693,
      "learning_rate": 1.99729694563181e-06,
      "loss": 1.5428,
      "step": 26197
    },
    {
      "epoch": 0.9382061704299246,
      "grad_norm": 1.8797887563705444,
      "learning_rate": 1.9949909750192064e-06,
      "loss": 1.4058,
      "step": 26198
    },
    {
      "epoch": 0.9382419825594929,
      "grad_norm": 1.439017653465271,
      "learning_rate": 1.992686322937565e-06,
      "loss": 1.211,
      "step": 26199
    },
    {
      "epoch": 0.9382777946890611,
      "grad_norm": 1.3723461627960205,
      "learning_rate": 1.990382989417916e-06,
      "loss": 1.4947,
      "step": 26200
    },
    {
      "epoch": 0.9383136068186295,
      "grad_norm": 1.789353847503662,
      "learning_rate": 1.9880809744912244e-06,
      "loss": 1.4519,
      "step": 26201
    },
    {
      "epoch": 0.9383494189481978,
      "grad_norm": 1.4479377269744873,
      "learning_rate": 1.985780278188487e-06,
      "loss": 1.5526,
      "step": 26202
    },
    {
      "epoch": 0.9383852310777661,
      "grad_norm": 1.693623661994934,
      "learning_rate": 1.983480900540646e-06,
      "loss": 1.5865,
      "step": 26203
    },
    {
      "epoch": 0.9384210432073343,
      "grad_norm": 1.6015986204147339,
      "learning_rate": 1.981182841578644e-06,
      "loss": 1.5839,
      "step": 26204
    },
    {
      "epoch": 0.9384568553369026,
      "grad_norm": 2.0890917778015137,
      "learning_rate": 1.978886101333388e-06,
      "loss": 1.1135,
      "step": 26205
    },
    {
      "epoch": 0.9384926674664709,
      "grad_norm": 1.5381704568862915,
      "learning_rate": 1.9765906798357767e-06,
      "loss": 1.5057,
      "step": 26206
    },
    {
      "epoch": 0.9385284795960391,
      "grad_norm": 1.5179535150527954,
      "learning_rate": 1.9742965771167077e-06,
      "loss": 1.5197,
      "step": 26207
    },
    {
      "epoch": 0.9385642917256075,
      "grad_norm": 1.7586171627044678,
      "learning_rate": 1.972003793207011e-06,
      "loss": 1.5069,
      "step": 26208
    },
    {
      "epoch": 0.9386001038551758,
      "grad_norm": 1.4683053493499756,
      "learning_rate": 1.969712328137574e-06,
      "loss": 1.2305,
      "step": 26209
    },
    {
      "epoch": 0.938635915984744,
      "grad_norm": 1.755323886871338,
      "learning_rate": 1.967422181939205e-06,
      "loss": 1.4299,
      "step": 26210
    },
    {
      "epoch": 0.9386717281143123,
      "grad_norm": 1.5153175592422485,
      "learning_rate": 1.9651333546427232e-06,
      "loss": 1.5045,
      "step": 26211
    },
    {
      "epoch": 0.9387075402438806,
      "grad_norm": 1.4256856441497803,
      "learning_rate": 1.9628458462789044e-06,
      "loss": 1.1516,
      "step": 26212
    },
    {
      "epoch": 0.9387433523734489,
      "grad_norm": 1.3613686561584473,
      "learning_rate": 1.960559656878547e-06,
      "loss": 1.4651,
      "step": 26213
    },
    {
      "epoch": 0.9387791645030171,
      "grad_norm": 1.3513787984848022,
      "learning_rate": 1.9582747864723917e-06,
      "loss": 1.3124,
      "step": 26214
    },
    {
      "epoch": 0.9388149766325855,
      "grad_norm": 1.5042704343795776,
      "learning_rate": 1.9559912350911925e-06,
      "loss": 1.398,
      "step": 26215
    },
    {
      "epoch": 0.9388507887621538,
      "grad_norm": 1.5108270645141602,
      "learning_rate": 1.953709002765647e-06,
      "loss": 1.3385,
      "step": 26216
    },
    {
      "epoch": 0.938886600891722,
      "grad_norm": 1.5499955415725708,
      "learning_rate": 1.951428089526486e-06,
      "loss": 1.5028,
      "step": 26217
    },
    {
      "epoch": 0.9389224130212903,
      "grad_norm": 1.7364864349365234,
      "learning_rate": 1.949148495404396e-06,
      "loss": 1.306,
      "step": 26218
    },
    {
      "epoch": 0.9389582251508586,
      "grad_norm": 1.5313931703567505,
      "learning_rate": 1.9468702204300195e-06,
      "loss": 1.5107,
      "step": 26219
    },
    {
      "epoch": 0.9389940372804269,
      "grad_norm": 1.9169384241104126,
      "learning_rate": 1.9445932646340314e-06,
      "loss": 1.3873,
      "step": 26220
    },
    {
      "epoch": 0.9390298494099951,
      "grad_norm": 1.4949196577072144,
      "learning_rate": 1.9423176280470633e-06,
      "loss": 1.5275,
      "step": 26221
    },
    {
      "epoch": 0.9390656615395635,
      "grad_norm": 1.7438608407974243,
      "learning_rate": 1.940043310699724e-06,
      "loss": 1.3416,
      "step": 26222
    },
    {
      "epoch": 0.9391014736691318,
      "grad_norm": 2.022535562515259,
      "learning_rate": 1.937770312622611e-06,
      "loss": 1.5157,
      "step": 26223
    },
    {
      "epoch": 0.9391372857987,
      "grad_norm": 1.9582380056381226,
      "learning_rate": 1.9354986338463e-06,
      "loss": 1.2197,
      "step": 26224
    },
    {
      "epoch": 0.9391730979282683,
      "grad_norm": 1.4199358224868774,
      "learning_rate": 1.9332282744013774e-06,
      "loss": 1.5727,
      "step": 26225
    },
    {
      "epoch": 0.9392089100578366,
      "grad_norm": 1.7500290870666504,
      "learning_rate": 1.9309592343183636e-06,
      "loss": 1.5048,
      "step": 26226
    },
    {
      "epoch": 0.9392447221874048,
      "grad_norm": 1.1745381355285645,
      "learning_rate": 1.9286915136277894e-06,
      "loss": 1.3728,
      "step": 26227
    },
    {
      "epoch": 0.9392805343169731,
      "grad_norm": 1.6411765813827515,
      "learning_rate": 1.926425112360164e-06,
      "loss": 1.1032,
      "step": 26228
    },
    {
      "epoch": 0.9393163464465415,
      "grad_norm": 2.0992393493652344,
      "learning_rate": 1.924160030545996e-06,
      "loss": 1.5346,
      "step": 26229
    },
    {
      "epoch": 0.9393521585761098,
      "grad_norm": 1.2511874437332153,
      "learning_rate": 1.9218962682157395e-06,
      "loss": 1.4908,
      "step": 26230
    },
    {
      "epoch": 0.939387970705678,
      "grad_norm": 1.5121225118637085,
      "learning_rate": 1.919633825399858e-06,
      "loss": 1.4154,
      "step": 26231
    },
    {
      "epoch": 0.9394237828352463,
      "grad_norm": 1.8958724737167358,
      "learning_rate": 1.9173727021287947e-06,
      "loss": 1.4142,
      "step": 26232
    },
    {
      "epoch": 0.9394595949648146,
      "grad_norm": 1.8122069835662842,
      "learning_rate": 1.915112898432947e-06,
      "loss": 1.5462,
      "step": 26233
    },
    {
      "epoch": 0.9394954070943828,
      "grad_norm": 1.5994408130645752,
      "learning_rate": 1.9128544143427463e-06,
      "loss": 1.5741,
      "step": 26234
    },
    {
      "epoch": 0.9395312192239511,
      "grad_norm": 1.965086817741394,
      "learning_rate": 1.910597249888568e-06,
      "loss": 1.4839,
      "step": 26235
    },
    {
      "epoch": 0.9395670313535195,
      "grad_norm": 1.5724587440490723,
      "learning_rate": 1.9083414051007776e-06,
      "loss": 1.6046,
      "step": 26236
    },
    {
      "epoch": 0.9396028434830878,
      "grad_norm": 1.785568118095398,
      "learning_rate": 1.9060868800097164e-06,
      "loss": 1.3867,
      "step": 26237
    },
    {
      "epoch": 0.939638655612656,
      "grad_norm": 1.5689623355865479,
      "learning_rate": 1.9038336746457276e-06,
      "loss": 1.2745,
      "step": 26238
    },
    {
      "epoch": 0.9396744677422243,
      "grad_norm": 2.110036611557007,
      "learning_rate": 1.9015817890391308e-06,
      "loss": 1.2477,
      "step": 26239
    },
    {
      "epoch": 0.9397102798717926,
      "grad_norm": 2.069141149520874,
      "learning_rate": 1.8993312232202021e-06,
      "loss": 1.5739,
      "step": 26240
    },
    {
      "epoch": 0.9397460920013608,
      "grad_norm": 1.2748523950576782,
      "learning_rate": 1.8970819772192394e-06,
      "loss": 1.4898,
      "step": 26241
    },
    {
      "epoch": 0.9397819041309291,
      "grad_norm": 1.6000792980194092,
      "learning_rate": 1.8948340510664853e-06,
      "loss": 1.1472,
      "step": 26242
    },
    {
      "epoch": 0.9398177162604975,
      "grad_norm": 1.5243643522262573,
      "learning_rate": 1.8925874447922044e-06,
      "loss": 1.3172,
      "step": 26243
    },
    {
      "epoch": 0.9398535283900658,
      "grad_norm": 1.70756995677948,
      "learning_rate": 1.8903421584266056e-06,
      "loss": 1.4195,
      "step": 26244
    },
    {
      "epoch": 0.939889340519634,
      "grad_norm": 1.3219189643859863,
      "learning_rate": 1.8880981919998875e-06,
      "loss": 1.3242,
      "step": 26245
    },
    {
      "epoch": 0.9399251526492023,
      "grad_norm": 1.4109432697296143,
      "learning_rate": 1.8858555455422699e-06,
      "loss": 1.4621,
      "step": 26246
    },
    {
      "epoch": 0.9399609647787706,
      "grad_norm": 1.9342769384384155,
      "learning_rate": 1.8836142190839067e-06,
      "loss": 1.5312,
      "step": 26247
    },
    {
      "epoch": 0.9399967769083388,
      "grad_norm": 1.6864076852798462,
      "learning_rate": 1.8813742126549404e-06,
      "loss": 1.4521,
      "step": 26248
    },
    {
      "epoch": 0.9400325890379071,
      "grad_norm": 1.7248731851577759,
      "learning_rate": 1.879135526285525e-06,
      "loss": 1.3108,
      "step": 26249
    },
    {
      "epoch": 0.9400684011674755,
      "grad_norm": 1.3078269958496094,
      "learning_rate": 1.876898160005791e-06,
      "loss": 1.2327,
      "step": 26250
    },
    {
      "epoch": 0.9401042132970437,
      "grad_norm": 1.7912778854370117,
      "learning_rate": 1.8746621138458042e-06,
      "loss": 1.3538,
      "step": 26251
    },
    {
      "epoch": 0.940140025426612,
      "grad_norm": 1.9527921676635742,
      "learning_rate": 1.8724273878356624e-06,
      "loss": 1.23,
      "step": 26252
    },
    {
      "epoch": 0.9401758375561803,
      "grad_norm": 1.850545883178711,
      "learning_rate": 1.8701939820054414e-06,
      "loss": 1.5436,
      "step": 26253
    },
    {
      "epoch": 0.9402116496857486,
      "grad_norm": 1.3803719282150269,
      "learning_rate": 1.8679618963851952e-06,
      "loss": 1.543,
      "step": 26254
    },
    {
      "epoch": 0.9402474618153168,
      "grad_norm": 1.3144088983535767,
      "learning_rate": 1.8657311310049218e-06,
      "loss": 1.5007,
      "step": 26255
    },
    {
      "epoch": 0.9402832739448851,
      "grad_norm": 1.282601237297058,
      "learning_rate": 1.863501685894664e-06,
      "loss": 1.3551,
      "step": 26256
    },
    {
      "epoch": 0.9403190860744535,
      "grad_norm": 2.010690927505493,
      "learning_rate": 1.861273561084398e-06,
      "loss": 1.3937,
      "step": 26257
    },
    {
      "epoch": 0.9403548982040217,
      "grad_norm": 2.084581136703491,
      "learning_rate": 1.8590467566041104e-06,
      "loss": 1.7239,
      "step": 26258
    },
    {
      "epoch": 0.94039071033359,
      "grad_norm": 1.5342340469360352,
      "learning_rate": 1.8568212724837442e-06,
      "loss": 1.2138,
      "step": 26259
    },
    {
      "epoch": 0.9404265224631583,
      "grad_norm": 1.4099143743515015,
      "learning_rate": 1.8545971087532644e-06,
      "loss": 1.7193,
      "step": 26260
    },
    {
      "epoch": 0.9404623345927265,
      "grad_norm": 1.9763034582138062,
      "learning_rate": 1.8523742654425802e-06,
      "loss": 1.4201,
      "step": 26261
    },
    {
      "epoch": 0.9404981467222948,
      "grad_norm": 1.699539065361023,
      "learning_rate": 1.8501527425816012e-06,
      "loss": 1.6167,
      "step": 26262
    },
    {
      "epoch": 0.9405339588518631,
      "grad_norm": 1.9624556303024292,
      "learning_rate": 1.8479325402002034e-06,
      "loss": 1.5734,
      "step": 26263
    },
    {
      "epoch": 0.9405697709814315,
      "grad_norm": 2.5679054260253906,
      "learning_rate": 1.8457136583282741e-06,
      "loss": 1.3815,
      "step": 26264
    },
    {
      "epoch": 0.9406055831109997,
      "grad_norm": 1.602765440940857,
      "learning_rate": 1.8434960969956561e-06,
      "loss": 1.489,
      "step": 26265
    },
    {
      "epoch": 0.940641395240568,
      "grad_norm": 2.7537906169891357,
      "learning_rate": 1.8412798562321809e-06,
      "loss": 1.2926,
      "step": 26266
    },
    {
      "epoch": 0.9406772073701363,
      "grad_norm": 1.3616611957550049,
      "learning_rate": 1.8390649360676692e-06,
      "loss": 1.625,
      "step": 26267
    },
    {
      "epoch": 0.9407130194997045,
      "grad_norm": 1.3584762811660767,
      "learning_rate": 1.8368513365319306e-06,
      "loss": 1.11,
      "step": 26268
    },
    {
      "epoch": 0.9407488316292728,
      "grad_norm": 1.5779452323913574,
      "learning_rate": 1.83463905765473e-06,
      "loss": 1.6435,
      "step": 26269
    },
    {
      "epoch": 0.9407846437588411,
      "grad_norm": 1.533232569694519,
      "learning_rate": 1.8324280994658327e-06,
      "loss": 1.4304,
      "step": 26270
    },
    {
      "epoch": 0.9408204558884095,
      "grad_norm": 1.4318612813949585,
      "learning_rate": 1.8302184619949925e-06,
      "loss": 1.3363,
      "step": 26271
    },
    {
      "epoch": 0.9408562680179777,
      "grad_norm": 1.623886227607727,
      "learning_rate": 1.8280101452719412e-06,
      "loss": 1.2529,
      "step": 26272
    },
    {
      "epoch": 0.940892080147546,
      "grad_norm": 1.5533499717712402,
      "learning_rate": 1.825803149326366e-06,
      "loss": 1.488,
      "step": 26273
    },
    {
      "epoch": 0.9409278922771143,
      "grad_norm": 1.4572827816009521,
      "learning_rate": 1.8235974741879769e-06,
      "loss": 1.2201,
      "step": 26274
    },
    {
      "epoch": 0.9409637044066825,
      "grad_norm": 1.7187683582305908,
      "learning_rate": 1.8213931198864608e-06,
      "loss": 1.6117,
      "step": 26275
    },
    {
      "epoch": 0.9409995165362508,
      "grad_norm": 1.6530053615570068,
      "learning_rate": 1.8191900864514388e-06,
      "loss": 1.2012,
      "step": 26276
    },
    {
      "epoch": 0.9410353286658191,
      "grad_norm": 1.9581012725830078,
      "learning_rate": 1.816988373912587e-06,
      "loss": 1.4425,
      "step": 26277
    },
    {
      "epoch": 0.9410711407953875,
      "grad_norm": 1.6734392642974854,
      "learning_rate": 1.8147879822994928e-06,
      "loss": 1.3796,
      "step": 26278
    },
    {
      "epoch": 0.9411069529249557,
      "grad_norm": 1.4807915687561035,
      "learning_rate": 1.8125889116417883e-06,
      "loss": 1.314,
      "step": 26279
    },
    {
      "epoch": 0.941142765054524,
      "grad_norm": 2.1756036281585693,
      "learning_rate": 1.8103911619690384e-06,
      "loss": 1.6318,
      "step": 26280
    },
    {
      "epoch": 0.9411785771840923,
      "grad_norm": 1.8008071184158325,
      "learning_rate": 1.8081947333108195e-06,
      "loss": 1.4568,
      "step": 26281
    },
    {
      "epoch": 0.9412143893136605,
      "grad_norm": 1.7400346994400024,
      "learning_rate": 1.805999625696686e-06,
      "loss": 1.4965,
      "step": 26282
    },
    {
      "epoch": 0.9412502014432288,
      "grad_norm": 1.8043440580368042,
      "learning_rate": 1.8038058391561697e-06,
      "loss": 1.1718,
      "step": 26283
    },
    {
      "epoch": 0.9412860135727971,
      "grad_norm": 2.3016085624694824,
      "learning_rate": 1.8016133737187913e-06,
      "loss": 1.3647,
      "step": 26284
    },
    {
      "epoch": 0.9413218257023654,
      "grad_norm": 1.9413868188858032,
      "learning_rate": 1.799422229414016e-06,
      "loss": 1.1064,
      "step": 26285
    },
    {
      "epoch": 0.9413576378319337,
      "grad_norm": 1.3793493509292603,
      "learning_rate": 1.7972324062713652e-06,
      "loss": 1.5471,
      "step": 26286
    },
    {
      "epoch": 0.941393449961502,
      "grad_norm": 1.6569875478744507,
      "learning_rate": 1.7950439043202593e-06,
      "loss": 1.3871,
      "step": 26287
    },
    {
      "epoch": 0.9414292620910703,
      "grad_norm": 2.83836030960083,
      "learning_rate": 1.7928567235901861e-06,
      "loss": 1.6417,
      "step": 26288
    },
    {
      "epoch": 0.9414650742206385,
      "grad_norm": 1.4256877899169922,
      "learning_rate": 1.790670864110522e-06,
      "loss": 1.3258,
      "step": 26289
    },
    {
      "epoch": 0.9415008863502068,
      "grad_norm": 1.5324336290359497,
      "learning_rate": 1.7884863259107209e-06,
      "loss": 1.4885,
      "step": 26290
    },
    {
      "epoch": 0.9415366984797751,
      "grad_norm": 1.494587779045105,
      "learning_rate": 1.7863031090201377e-06,
      "loss": 1.5611,
      "step": 26291
    },
    {
      "epoch": 0.9415725106093434,
      "grad_norm": 1.7873804569244385,
      "learning_rate": 1.7841212134681705e-06,
      "loss": 1.4419,
      "step": 26292
    },
    {
      "epoch": 0.9416083227389117,
      "grad_norm": 1.5246162414550781,
      "learning_rate": 1.781940639284163e-06,
      "loss": 1.4035,
      "step": 26293
    },
    {
      "epoch": 0.94164413486848,
      "grad_norm": 2.191568613052368,
      "learning_rate": 1.7797613864974472e-06,
      "loss": 1.4957,
      "step": 26294
    },
    {
      "epoch": 0.9416799469980482,
      "grad_norm": 1.926516056060791,
      "learning_rate": 1.7775834551373548e-06,
      "loss": 0.9744,
      "step": 26295
    },
    {
      "epoch": 0.9417157591276165,
      "grad_norm": 2.023942470550537,
      "learning_rate": 1.775406845233163e-06,
      "loss": 1.1816,
      "step": 26296
    },
    {
      "epoch": 0.9417515712571848,
      "grad_norm": 1.6293940544128418,
      "learning_rate": 1.7732315568141811e-06,
      "loss": 1.2553,
      "step": 26297
    },
    {
      "epoch": 0.941787383386753,
      "grad_norm": 1.5346230268478394,
      "learning_rate": 1.7710575899096637e-06,
      "loss": 1.3784,
      "step": 26298
    },
    {
      "epoch": 0.9418231955163214,
      "grad_norm": 1.4494526386260986,
      "learning_rate": 1.7688849445488654e-06,
      "loss": 1.2135,
      "step": 26299
    },
    {
      "epoch": 0.9418590076458897,
      "grad_norm": 1.3800164461135864,
      "learning_rate": 1.7667136207609958e-06,
      "loss": 1.4168,
      "step": 26300
    },
    {
      "epoch": 0.941894819775458,
      "grad_norm": 1.7155061960220337,
      "learning_rate": 1.7645436185753095e-06,
      "loss": 1.5418,
      "step": 26301
    },
    {
      "epoch": 0.9419306319050262,
      "grad_norm": 1.9509527683258057,
      "learning_rate": 1.7623749380209609e-06,
      "loss": 1.2594,
      "step": 26302
    },
    {
      "epoch": 0.9419664440345945,
      "grad_norm": 1.7195106744766235,
      "learning_rate": 1.7602075791271377e-06,
      "loss": 1.451,
      "step": 26303
    },
    {
      "epoch": 0.9420022561641628,
      "grad_norm": 1.8029755353927612,
      "learning_rate": 1.7580415419229946e-06,
      "loss": 1.1763,
      "step": 26304
    },
    {
      "epoch": 0.942038068293731,
      "grad_norm": 1.6189782619476318,
      "learning_rate": 1.7558768264376856e-06,
      "loss": 1.4517,
      "step": 26305
    },
    {
      "epoch": 0.9420738804232994,
      "grad_norm": 1.5579745769500732,
      "learning_rate": 1.7537134327003324e-06,
      "loss": 1.4916,
      "step": 26306
    },
    {
      "epoch": 0.9421096925528677,
      "grad_norm": 2.151827573776245,
      "learning_rate": 1.7515513607400225e-06,
      "loss": 1.6583,
      "step": 26307
    },
    {
      "epoch": 0.942145504682436,
      "grad_norm": 1.8219877481460571,
      "learning_rate": 1.749390610585877e-06,
      "loss": 1.2741,
      "step": 26308
    },
    {
      "epoch": 0.9421813168120042,
      "grad_norm": 1.3775928020477295,
      "learning_rate": 1.7472311822669397e-06,
      "loss": 1.3416,
      "step": 26309
    },
    {
      "epoch": 0.9422171289415725,
      "grad_norm": 1.1605829000473022,
      "learning_rate": 1.7450730758122757e-06,
      "loss": 1.4166,
      "step": 26310
    },
    {
      "epoch": 0.9422529410711408,
      "grad_norm": 1.6979670524597168,
      "learning_rate": 1.7429162912508956e-06,
      "loss": 1.4178,
      "step": 26311
    },
    {
      "epoch": 0.942288753200709,
      "grad_norm": 1.6193804740905762,
      "learning_rate": 1.7407608286118427e-06,
      "loss": 1.5039,
      "step": 26312
    },
    {
      "epoch": 0.9423245653302774,
      "grad_norm": 2.2689478397369385,
      "learning_rate": 1.7386066879241159e-06,
      "loss": 1.6966,
      "step": 26313
    },
    {
      "epoch": 0.9423603774598457,
      "grad_norm": 1.6678130626678467,
      "learning_rate": 1.736453869216681e-06,
      "loss": 1.4604,
      "step": 26314
    },
    {
      "epoch": 0.942396189589414,
      "grad_norm": 1.554015040397644,
      "learning_rate": 1.7343023725185038e-06,
      "loss": 1.471,
      "step": 26315
    },
    {
      "epoch": 0.9424320017189822,
      "grad_norm": 1.548634648323059,
      "learning_rate": 1.7321521978585387e-06,
      "loss": 1.5321,
      "step": 26316
    },
    {
      "epoch": 0.9424678138485505,
      "grad_norm": 1.383250117301941,
      "learning_rate": 1.7300033452657184e-06,
      "loss": 1.3117,
      "step": 26317
    },
    {
      "epoch": 0.9425036259781188,
      "grad_norm": 1.8333631753921509,
      "learning_rate": 1.7278558147689306e-06,
      "loss": 1.4467,
      "step": 26318
    },
    {
      "epoch": 0.942539438107687,
      "grad_norm": 2.024153232574463,
      "learning_rate": 1.7257096063970856e-06,
      "loss": 1.4771,
      "step": 26319
    },
    {
      "epoch": 0.9425752502372554,
      "grad_norm": 2.0110878944396973,
      "learning_rate": 1.7235647201790605e-06,
      "loss": 1.9133,
      "step": 26320
    },
    {
      "epoch": 0.9426110623668237,
      "grad_norm": 2.2840211391448975,
      "learning_rate": 1.7214211561436987e-06,
      "loss": 1.576,
      "step": 26321
    },
    {
      "epoch": 0.942646874496392,
      "grad_norm": 1.5875118970870972,
      "learning_rate": 1.719278914319844e-06,
      "loss": 1.4262,
      "step": 26322
    },
    {
      "epoch": 0.9426826866259602,
      "grad_norm": 1.5386816263198853,
      "learning_rate": 1.7171379947363175e-06,
      "loss": 1.3028,
      "step": 26323
    },
    {
      "epoch": 0.9427184987555285,
      "grad_norm": 1.6080418825149536,
      "learning_rate": 1.7149983974219297e-06,
      "loss": 1.4332,
      "step": 26324
    },
    {
      "epoch": 0.9427543108850968,
      "grad_norm": 1.3982714414596558,
      "learning_rate": 1.7128601224054464e-06,
      "loss": 1.2564,
      "step": 26325
    },
    {
      "epoch": 0.942790123014665,
      "grad_norm": 1.6409122943878174,
      "learning_rate": 1.7107231697156557e-06,
      "loss": 1.3616,
      "step": 26326
    },
    {
      "epoch": 0.9428259351442334,
      "grad_norm": 1.7633566856384277,
      "learning_rate": 1.7085875393813123e-06,
      "loss": 1.3169,
      "step": 26327
    },
    {
      "epoch": 0.9428617472738017,
      "grad_norm": 1.4810881614685059,
      "learning_rate": 1.7064532314311266e-06,
      "loss": 1.3493,
      "step": 26328
    },
    {
      "epoch": 0.94289755940337,
      "grad_norm": 1.8232383728027344,
      "learning_rate": 1.70432024589382e-06,
      "loss": 1.5052,
      "step": 26329
    },
    {
      "epoch": 0.9429333715329382,
      "grad_norm": 2.7344441413879395,
      "learning_rate": 1.702188582798092e-06,
      "loss": 1.6025,
      "step": 26330
    },
    {
      "epoch": 0.9429691836625065,
      "grad_norm": 2.21616792678833,
      "learning_rate": 1.7000582421726308e-06,
      "loss": 1.5006,
      "step": 26331
    },
    {
      "epoch": 0.9430049957920748,
      "grad_norm": 1.2851320505142212,
      "learning_rate": 1.6979292240460799e-06,
      "loss": 1.2972,
      "step": 26332
    },
    {
      "epoch": 0.943040807921643,
      "grad_norm": 1.8838286399841309,
      "learning_rate": 1.695801528447094e-06,
      "loss": 1.398,
      "step": 26333
    },
    {
      "epoch": 0.9430766200512114,
      "grad_norm": 1.508453130722046,
      "learning_rate": 1.6936751554042951e-06,
      "loss": 1.4436,
      "step": 26334
    },
    {
      "epoch": 0.9431124321807797,
      "grad_norm": 1.5754175186157227,
      "learning_rate": 1.6915501049462934e-06,
      "loss": 1.2449,
      "step": 26335
    },
    {
      "epoch": 0.9431482443103479,
      "grad_norm": 1.53984797000885,
      "learning_rate": 1.6894263771016661e-06,
      "loss": 1.8434,
      "step": 26336
    },
    {
      "epoch": 0.9431840564399162,
      "grad_norm": 1.8990575075149536,
      "learning_rate": 1.6873039718990014e-06,
      "loss": 1.317,
      "step": 26337
    },
    {
      "epoch": 0.9432198685694845,
      "grad_norm": 1.4001402854919434,
      "learning_rate": 1.6851828893668543e-06,
      "loss": 1.2721,
      "step": 26338
    },
    {
      "epoch": 0.9432556806990527,
      "grad_norm": 1.814524531364441,
      "learning_rate": 1.6830631295337462e-06,
      "loss": 1.4002,
      "step": 26339
    },
    {
      "epoch": 0.943291492828621,
      "grad_norm": 1.9315727949142456,
      "learning_rate": 1.68094469242821e-06,
      "loss": 1.3913,
      "step": 26340
    },
    {
      "epoch": 0.9433273049581893,
      "grad_norm": 1.8088985681533813,
      "learning_rate": 1.6788275780787343e-06,
      "loss": 1.4572,
      "step": 26341
    },
    {
      "epoch": 0.9433631170877577,
      "grad_norm": 1.4374310970306396,
      "learning_rate": 1.6767117865138182e-06,
      "loss": 1.3947,
      "step": 26342
    },
    {
      "epoch": 0.9433989292173259,
      "grad_norm": 1.5694729089736938,
      "learning_rate": 1.6745973177619056e-06,
      "loss": 1.3992,
      "step": 26343
    },
    {
      "epoch": 0.9434347413468942,
      "grad_norm": 1.4492532014846802,
      "learning_rate": 1.6724841718514629e-06,
      "loss": 1.266,
      "step": 26344
    },
    {
      "epoch": 0.9434705534764625,
      "grad_norm": 1.6183035373687744,
      "learning_rate": 1.6703723488109112e-06,
      "loss": 1.5549,
      "step": 26345
    },
    {
      "epoch": 0.9435063656060307,
      "grad_norm": 1.5488311052322388,
      "learning_rate": 1.6682618486686619e-06,
      "loss": 1.3828,
      "step": 26346
    },
    {
      "epoch": 0.943542177735599,
      "grad_norm": 1.724896788597107,
      "learning_rate": 1.6661526714531029e-06,
      "loss": 1.1049,
      "step": 26347
    },
    {
      "epoch": 0.9435779898651673,
      "grad_norm": 1.9380396604537964,
      "learning_rate": 1.6640448171926226e-06,
      "loss": 1.3431,
      "step": 26348
    },
    {
      "epoch": 0.9436138019947357,
      "grad_norm": 2.2302026748657227,
      "learning_rate": 1.6619382859155873e-06,
      "loss": 1.6598,
      "step": 26349
    },
    {
      "epoch": 0.9436496141243039,
      "grad_norm": 2.1377158164978027,
      "learning_rate": 1.659833077650319e-06,
      "loss": 1.3672,
      "step": 26350
    },
    {
      "epoch": 0.9436854262538722,
      "grad_norm": 1.8395555019378662,
      "learning_rate": 1.6577291924251392e-06,
      "loss": 1.4068,
      "step": 26351
    },
    {
      "epoch": 0.9437212383834405,
      "grad_norm": 1.6605585813522339,
      "learning_rate": 1.6556266302683588e-06,
      "loss": 1.1427,
      "step": 26352
    },
    {
      "epoch": 0.9437570505130087,
      "grad_norm": 1.5564028024673462,
      "learning_rate": 1.6535253912082772e-06,
      "loss": 1.1415,
      "step": 26353
    },
    {
      "epoch": 0.943792862642577,
      "grad_norm": 1.9182465076446533,
      "learning_rate": 1.6514254752731494e-06,
      "loss": 1.3955,
      "step": 26354
    },
    {
      "epoch": 0.9438286747721453,
      "grad_norm": 1.4966611862182617,
      "learning_rate": 1.6493268824912312e-06,
      "loss": 1.3372,
      "step": 26355
    },
    {
      "epoch": 0.9438644869017137,
      "grad_norm": 1.8412766456604004,
      "learning_rate": 1.647229612890766e-06,
      "loss": 1.5312,
      "step": 26356
    },
    {
      "epoch": 0.9439002990312819,
      "grad_norm": 1.7466552257537842,
      "learning_rate": 1.6451336664999539e-06,
      "loss": 1.268,
      "step": 26357
    },
    {
      "epoch": 0.9439361111608502,
      "grad_norm": 1.5257654190063477,
      "learning_rate": 1.6430390433469945e-06,
      "loss": 1.4011,
      "step": 26358
    },
    {
      "epoch": 0.9439719232904185,
      "grad_norm": 1.8344718217849731,
      "learning_rate": 1.640945743460065e-06,
      "loss": 1.5296,
      "step": 26359
    },
    {
      "epoch": 0.9440077354199867,
      "grad_norm": 1.9049286842346191,
      "learning_rate": 1.6388537668673542e-06,
      "loss": 1.1641,
      "step": 26360
    },
    {
      "epoch": 0.944043547549555,
      "grad_norm": 1.3105257749557495,
      "learning_rate": 1.636763113596984e-06,
      "loss": 1.3407,
      "step": 26361
    },
    {
      "epoch": 0.9440793596791233,
      "grad_norm": 2.12705659866333,
      "learning_rate": 1.6346737836770875e-06,
      "loss": 1.4898,
      "step": 26362
    },
    {
      "epoch": 0.9441151718086916,
      "grad_norm": 1.5113617181777954,
      "learning_rate": 1.6325857771357756e-06,
      "loss": 1.5729,
      "step": 26363
    },
    {
      "epoch": 0.9441509839382599,
      "grad_norm": 1.867129921913147,
      "learning_rate": 1.6304990940011255e-06,
      "loss": 1.3769,
      "step": 26364
    },
    {
      "epoch": 0.9441867960678282,
      "grad_norm": 1.3271534442901611,
      "learning_rate": 1.6284137343012263e-06,
      "loss": 1.2326,
      "step": 26365
    },
    {
      "epoch": 0.9442226081973965,
      "grad_norm": 1.6503703594207764,
      "learning_rate": 1.6263296980641328e-06,
      "loss": 1.2799,
      "step": 26366
    },
    {
      "epoch": 0.9442584203269647,
      "grad_norm": 1.6071670055389404,
      "learning_rate": 1.62424698531789e-06,
      "loss": 1.2021,
      "step": 26367
    },
    {
      "epoch": 0.944294232456533,
      "grad_norm": 1.499289870262146,
      "learning_rate": 1.6221655960904968e-06,
      "loss": 1.0335,
      "step": 26368
    },
    {
      "epoch": 0.9443300445861013,
      "grad_norm": 1.6527241468429565,
      "learning_rate": 1.620085530409965e-06,
      "loss": 1.3493,
      "step": 26369
    },
    {
      "epoch": 0.9443658567156696,
      "grad_norm": 1.5437915325164795,
      "learning_rate": 1.6180067883042937e-06,
      "loss": 1.468,
      "step": 26370
    },
    {
      "epoch": 0.9444016688452379,
      "grad_norm": 2.0119853019714355,
      "learning_rate": 1.6159293698014278e-06,
      "loss": 1.4788,
      "step": 26371
    },
    {
      "epoch": 0.9444374809748062,
      "grad_norm": 1.936726689338684,
      "learning_rate": 1.6138532749293335e-06,
      "loss": 1.4597,
      "step": 26372
    },
    {
      "epoch": 0.9444732931043744,
      "grad_norm": 1.780268907546997,
      "learning_rate": 1.6117785037159216e-06,
      "loss": 1.1247,
      "step": 26373
    },
    {
      "epoch": 0.9445091052339427,
      "grad_norm": 1.8385742902755737,
      "learning_rate": 1.6097050561891369e-06,
      "loss": 1.5127,
      "step": 26374
    },
    {
      "epoch": 0.944544917363511,
      "grad_norm": 1.4780793190002441,
      "learning_rate": 1.6076329323768347e-06,
      "loss": 1.4763,
      "step": 26375
    },
    {
      "epoch": 0.9445807294930793,
      "grad_norm": 1.8092213869094849,
      "learning_rate": 1.605562132306937e-06,
      "loss": 1.3985,
      "step": 26376
    },
    {
      "epoch": 0.9446165416226476,
      "grad_norm": 1.405903935432434,
      "learning_rate": 1.6034926560072549e-06,
      "loss": 1.1074,
      "step": 26377
    },
    {
      "epoch": 0.9446523537522159,
      "grad_norm": 1.8108632564544678,
      "learning_rate": 1.6014245035056775e-06,
      "loss": 1.3274,
      "step": 26378
    },
    {
      "epoch": 0.9446881658817842,
      "grad_norm": 1.407546877861023,
      "learning_rate": 1.5993576748300043e-06,
      "loss": 1.4492,
      "step": 26379
    },
    {
      "epoch": 0.9447239780113524,
      "grad_norm": 1.6472501754760742,
      "learning_rate": 1.5972921700080357e-06,
      "loss": 1.2341,
      "step": 26380
    },
    {
      "epoch": 0.9447597901409207,
      "grad_norm": 1.5353707075119019,
      "learning_rate": 1.5952279890675826e-06,
      "loss": 1.1901,
      "step": 26381
    },
    {
      "epoch": 0.944795602270489,
      "grad_norm": 1.5727671384811401,
      "learning_rate": 1.5931651320364006e-06,
      "loss": 1.5107,
      "step": 26382
    },
    {
      "epoch": 0.9448314144000572,
      "grad_norm": 1.323904037475586,
      "learning_rate": 1.5911035989422562e-06,
      "loss": 1.3362,
      "step": 26383
    },
    {
      "epoch": 0.9448672265296256,
      "grad_norm": 1.4275002479553223,
      "learning_rate": 1.5890433898128498e-06,
      "loss": 1.4968,
      "step": 26384
    },
    {
      "epoch": 0.9449030386591939,
      "grad_norm": 1.5640442371368408,
      "learning_rate": 1.5869845046759369e-06,
      "loss": 1.4972,
      "step": 26385
    },
    {
      "epoch": 0.9449388507887622,
      "grad_norm": 2.0282034873962402,
      "learning_rate": 1.5849269435592061e-06,
      "loss": 1.6468,
      "step": 26386
    },
    {
      "epoch": 0.9449746629183304,
      "grad_norm": 1.7705632448196411,
      "learning_rate": 1.5828707064903359e-06,
      "loss": 1.3092,
      "step": 26387
    },
    {
      "epoch": 0.9450104750478987,
      "grad_norm": 1.7389479875564575,
      "learning_rate": 1.5808157934969813e-06,
      "loss": 1.0817,
      "step": 26388
    },
    {
      "epoch": 0.945046287177467,
      "grad_norm": 1.5240728855133057,
      "learning_rate": 1.5787622046068207e-06,
      "loss": 1.3589,
      "step": 26389
    },
    {
      "epoch": 0.9450820993070352,
      "grad_norm": 1.9323331117630005,
      "learning_rate": 1.576709939847454e-06,
      "loss": 1.2756,
      "step": 26390
    },
    {
      "epoch": 0.9451179114366036,
      "grad_norm": 1.5090250968933105,
      "learning_rate": 1.574658999246481e-06,
      "loss": 1.1489,
      "step": 26391
    },
    {
      "epoch": 0.9451537235661719,
      "grad_norm": 2.539374828338623,
      "learning_rate": 1.5726093828315248e-06,
      "loss": 1.2725,
      "step": 26392
    },
    {
      "epoch": 0.9451895356957402,
      "grad_norm": 1.6672533750534058,
      "learning_rate": 1.5705610906301404e-06,
      "loss": 1.317,
      "step": 26393
    },
    {
      "epoch": 0.9452253478253084,
      "grad_norm": 1.4400562047958374,
      "learning_rate": 1.5685141226699064e-06,
      "loss": 1.4577,
      "step": 26394
    },
    {
      "epoch": 0.9452611599548767,
      "grad_norm": 1.512329339981079,
      "learning_rate": 1.5664684789783224e-06,
      "loss": 1.2943,
      "step": 26395
    },
    {
      "epoch": 0.945296972084445,
      "grad_norm": 1.447803258895874,
      "learning_rate": 1.5644241595829557e-06,
      "loss": 1.2944,
      "step": 26396
    },
    {
      "epoch": 0.9453327842140132,
      "grad_norm": 1.6319823265075684,
      "learning_rate": 1.562381164511284e-06,
      "loss": 1.2474,
      "step": 26397
    },
    {
      "epoch": 0.9453685963435816,
      "grad_norm": 1.9061074256896973,
      "learning_rate": 1.5603394937907967e-06,
      "loss": 1.4807,
      "step": 26398
    },
    {
      "epoch": 0.9454044084731499,
      "grad_norm": 1.5398566722869873,
      "learning_rate": 1.5582991474489607e-06,
      "loss": 1.526,
      "step": 26399
    },
    {
      "epoch": 0.9454402206027182,
      "grad_norm": 4.880867958068848,
      "learning_rate": 1.5562601255132314e-06,
      "loss": 1.3848,
      "step": 26400
    },
    {
      "epoch": 0.9454760327322864,
      "grad_norm": 1.713074803352356,
      "learning_rate": 1.554222428011043e-06,
      "loss": 1.3903,
      "step": 26401
    },
    {
      "epoch": 0.9455118448618547,
      "grad_norm": 1.401732325553894,
      "learning_rate": 1.5521860549698063e-06,
      "loss": 1.6819,
      "step": 26402
    },
    {
      "epoch": 0.945547656991423,
      "grad_norm": 1.8747241497039795,
      "learning_rate": 1.550151006416911e-06,
      "loss": 1.5883,
      "step": 26403
    },
    {
      "epoch": 0.9455834691209912,
      "grad_norm": 1.228621244430542,
      "learning_rate": 1.5481172823797463e-06,
      "loss": 1.3525,
      "step": 26404
    },
    {
      "epoch": 0.9456192812505596,
      "grad_norm": 1.479820728302002,
      "learning_rate": 1.5460848828856677e-06,
      "loss": 1.6433,
      "step": 26405
    },
    {
      "epoch": 0.9456550933801279,
      "grad_norm": 1.6731551885604858,
      "learning_rate": 1.5440538079620204e-06,
      "loss": 1.3101,
      "step": 26406
    },
    {
      "epoch": 0.9456909055096961,
      "grad_norm": 1.7374619245529175,
      "learning_rate": 1.5420240576361378e-06,
      "loss": 1.4818,
      "step": 26407
    },
    {
      "epoch": 0.9457267176392644,
      "grad_norm": 1.6798261404037476,
      "learning_rate": 1.5399956319353092e-06,
      "loss": 1.2353,
      "step": 26408
    },
    {
      "epoch": 0.9457625297688327,
      "grad_norm": 1.8133715391159058,
      "learning_rate": 1.5379685308868464e-06,
      "loss": 1.3315,
      "step": 26409
    },
    {
      "epoch": 0.945798341898401,
      "grad_norm": 2.04004168510437,
      "learning_rate": 1.535942754517994e-06,
      "loss": 1.154,
      "step": 26410
    },
    {
      "epoch": 0.9458341540279692,
      "grad_norm": 1.9576140642166138,
      "learning_rate": 1.53391830285603e-06,
      "loss": 1.4195,
      "step": 26411
    },
    {
      "epoch": 0.9458699661575376,
      "grad_norm": 1.7399269342422485,
      "learning_rate": 1.5318951759281885e-06,
      "loss": 1.6082,
      "step": 26412
    },
    {
      "epoch": 0.9459057782871059,
      "grad_norm": 1.564070224761963,
      "learning_rate": 1.529873373761681e-06,
      "loss": 1.7338,
      "step": 26413
    },
    {
      "epoch": 0.9459415904166741,
      "grad_norm": 1.5939946174621582,
      "learning_rate": 1.527852896383708e-06,
      "loss": 1.4318,
      "step": 26414
    },
    {
      "epoch": 0.9459774025462424,
      "grad_norm": 1.428228497505188,
      "learning_rate": 1.5258337438214587e-06,
      "loss": 1.4115,
      "step": 26415
    },
    {
      "epoch": 0.9460132146758107,
      "grad_norm": 1.6301560401916504,
      "learning_rate": 1.5238159161020893e-06,
      "loss": 1.5172,
      "step": 26416
    },
    {
      "epoch": 0.946049026805379,
      "grad_norm": 1.5005348920822144,
      "learning_rate": 1.5217994132527448e-06,
      "loss": 1.5653,
      "step": 26417
    },
    {
      "epoch": 0.9460848389349472,
      "grad_norm": 1.9517136812210083,
      "learning_rate": 1.5197842353005698e-06,
      "loss": 1.4366,
      "step": 26418
    },
    {
      "epoch": 0.9461206510645156,
      "grad_norm": 1.6520404815673828,
      "learning_rate": 1.5177703822726652e-06,
      "loss": 1.4942,
      "step": 26419
    },
    {
      "epoch": 0.9461564631940839,
      "grad_norm": 1.7810132503509521,
      "learning_rate": 1.5157578541961315e-06,
      "loss": 1.1101,
      "step": 26420
    },
    {
      "epoch": 0.9461922753236521,
      "grad_norm": 1.829916000366211,
      "learning_rate": 1.5137466510980357e-06,
      "loss": 1.4551,
      "step": 26421
    },
    {
      "epoch": 0.9462280874532204,
      "grad_norm": 1.324633002281189,
      "learning_rate": 1.5117367730054343e-06,
      "loss": 1.3078,
      "step": 26422
    },
    {
      "epoch": 0.9462638995827887,
      "grad_norm": 1.974655270576477,
      "learning_rate": 1.5097282199453943e-06,
      "loss": 1.5415,
      "step": 26423
    },
    {
      "epoch": 0.9462997117123569,
      "grad_norm": 1.4947298765182495,
      "learning_rate": 1.5077209919449053e-06,
      "loss": 1.2912,
      "step": 26424
    },
    {
      "epoch": 0.9463355238419252,
      "grad_norm": 1.5060384273529053,
      "learning_rate": 1.505715089030979e-06,
      "loss": 1.293,
      "step": 26425
    },
    {
      "epoch": 0.9463713359714936,
      "grad_norm": 2.223021984100342,
      "learning_rate": 1.503710511230616e-06,
      "loss": 1.344,
      "step": 26426
    },
    {
      "epoch": 0.9464071481010619,
      "grad_norm": 1.7746235132217407,
      "learning_rate": 1.5017072585707725e-06,
      "loss": 1.4105,
      "step": 26427
    },
    {
      "epoch": 0.9464429602306301,
      "grad_norm": 1.6269015073776245,
      "learning_rate": 1.4997053310784047e-06,
      "loss": 1.2152,
      "step": 26428
    },
    {
      "epoch": 0.9464787723601984,
      "grad_norm": 1.9540683031082153,
      "learning_rate": 1.497704728780447e-06,
      "loss": 1.4217,
      "step": 26429
    },
    {
      "epoch": 0.9465145844897667,
      "grad_norm": 1.675334095954895,
      "learning_rate": 1.4957054517038106e-06,
      "loss": 1.4094,
      "step": 26430
    },
    {
      "epoch": 0.9465503966193349,
      "grad_norm": 2.8871023654937744,
      "learning_rate": 1.4937074998753965e-06,
      "loss": 1.3651,
      "step": 26431
    },
    {
      "epoch": 0.9465862087489032,
      "grad_norm": 1.911722183227539,
      "learning_rate": 1.491710873322083e-06,
      "loss": 1.441,
      "step": 26432
    },
    {
      "epoch": 0.9466220208784716,
      "grad_norm": 1.7608087062835693,
      "learning_rate": 1.489715572070738e-06,
      "loss": 1.421,
      "step": 26433
    },
    {
      "epoch": 0.9466578330080399,
      "grad_norm": 1.8853119611740112,
      "learning_rate": 1.4877215961482062e-06,
      "loss": 1.2649,
      "step": 26434
    },
    {
      "epoch": 0.9466936451376081,
      "grad_norm": 1.8366423845291138,
      "learning_rate": 1.4857289455812883e-06,
      "loss": 1.5063,
      "step": 26435
    },
    {
      "epoch": 0.9467294572671764,
      "grad_norm": 1.7807245254516602,
      "learning_rate": 1.48373762039683e-06,
      "loss": 1.3553,
      "step": 26436
    },
    {
      "epoch": 0.9467652693967447,
      "grad_norm": 1.6845966577529907,
      "learning_rate": 1.4817476206216096e-06,
      "loss": 1.2403,
      "step": 26437
    },
    {
      "epoch": 0.9468010815263129,
      "grad_norm": 1.5173600912094116,
      "learning_rate": 1.4797589462823836e-06,
      "loss": 1.5802,
      "step": 26438
    },
    {
      "epoch": 0.9468368936558812,
      "grad_norm": 1.406801462173462,
      "learning_rate": 1.4777715974059192e-06,
      "loss": 1.5145,
      "step": 26439
    },
    {
      "epoch": 0.9468727057854496,
      "grad_norm": 1.5290340185165405,
      "learning_rate": 1.4757855740189508e-06,
      "loss": 1.3885,
      "step": 26440
    },
    {
      "epoch": 0.9469085179150178,
      "grad_norm": 2.0321245193481445,
      "learning_rate": 1.4738008761482125e-06,
      "loss": 1.6672,
      "step": 26441
    },
    {
      "epoch": 0.9469443300445861,
      "grad_norm": 1.7161322832107544,
      "learning_rate": 1.471817503820383e-06,
      "loss": 1.3366,
      "step": 26442
    },
    {
      "epoch": 0.9469801421741544,
      "grad_norm": 1.8721601963043213,
      "learning_rate": 1.469835457062163e-06,
      "loss": 1.3743,
      "step": 26443
    },
    {
      "epoch": 0.9470159543037227,
      "grad_norm": 1.5677088499069214,
      "learning_rate": 1.4678547359002092e-06,
      "loss": 1.556,
      "step": 26444
    },
    {
      "epoch": 0.9470517664332909,
      "grad_norm": 1.6195982694625854,
      "learning_rate": 1.465875340361178e-06,
      "loss": 1.4419,
      "step": 26445
    },
    {
      "epoch": 0.9470875785628592,
      "grad_norm": 1.714670181274414,
      "learning_rate": 1.4638972704716814e-06,
      "loss": 1.7476,
      "step": 26446
    },
    {
      "epoch": 0.9471233906924276,
      "grad_norm": 1.6740520000457764,
      "learning_rate": 1.4619205262583536e-06,
      "loss": 1.4975,
      "step": 26447
    },
    {
      "epoch": 0.9471592028219958,
      "grad_norm": 1.2522761821746826,
      "learning_rate": 1.4599451077477844e-06,
      "loss": 1.4487,
      "step": 26448
    },
    {
      "epoch": 0.9471950149515641,
      "grad_norm": 1.4357094764709473,
      "learning_rate": 1.4579710149665416e-06,
      "loss": 1.2878,
      "step": 26449
    },
    {
      "epoch": 0.9472308270811324,
      "grad_norm": 1.617078423500061,
      "learning_rate": 1.4559982479411927e-06,
      "loss": 1.4942,
      "step": 26450
    },
    {
      "epoch": 0.9472666392107006,
      "grad_norm": 1.5373728275299072,
      "learning_rate": 1.4540268066982722e-06,
      "loss": 1.2622,
      "step": 26451
    },
    {
      "epoch": 0.9473024513402689,
      "grad_norm": 2.110975980758667,
      "learning_rate": 1.452056691264303e-06,
      "loss": 1.2574,
      "step": 26452
    },
    {
      "epoch": 0.9473382634698372,
      "grad_norm": 2.1490426063537598,
      "learning_rate": 1.4500879016657865e-06,
      "loss": 1.5997,
      "step": 26453
    },
    {
      "epoch": 0.9473740755994056,
      "grad_norm": 1.6299991607666016,
      "learning_rate": 1.4481204379292234e-06,
      "loss": 1.5127,
      "step": 26454
    },
    {
      "epoch": 0.9474098877289738,
      "grad_norm": 1.6052266359329224,
      "learning_rate": 1.4461543000810929e-06,
      "loss": 1.6573,
      "step": 26455
    },
    {
      "epoch": 0.9474456998585421,
      "grad_norm": 2.261275291442871,
      "learning_rate": 1.4441894881478069e-06,
      "loss": 1.5065,
      "step": 26456
    },
    {
      "epoch": 0.9474815119881104,
      "grad_norm": 1.8761277198791504,
      "learning_rate": 1.4422260021558331e-06,
      "loss": 1.6339,
      "step": 26457
    },
    {
      "epoch": 0.9475173241176786,
      "grad_norm": 1.7935103178024292,
      "learning_rate": 1.440263842131573e-06,
      "loss": 1.5145,
      "step": 26458
    },
    {
      "epoch": 0.9475531362472469,
      "grad_norm": 1.3724159002304077,
      "learning_rate": 1.4383030081014493e-06,
      "loss": 1.3055,
      "step": 26459
    },
    {
      "epoch": 0.9475889483768152,
      "grad_norm": 1.9284459352493286,
      "learning_rate": 1.436343500091808e-06,
      "loss": 1.5151,
      "step": 26460
    },
    {
      "epoch": 0.9476247605063836,
      "grad_norm": 1.7961905002593994,
      "learning_rate": 1.4343853181290168e-06,
      "loss": 1.5256,
      "step": 26461
    },
    {
      "epoch": 0.9476605726359518,
      "grad_norm": 1.7151559591293335,
      "learning_rate": 1.4324284622394547e-06,
      "loss": 1.4779,
      "step": 26462
    },
    {
      "epoch": 0.9476963847655201,
      "grad_norm": 1.4853241443634033,
      "learning_rate": 1.4304729324494115e-06,
      "loss": 1.4664,
      "step": 26463
    },
    {
      "epoch": 0.9477321968950884,
      "grad_norm": 1.5965665578842163,
      "learning_rate": 1.4285187287851997e-06,
      "loss": 1.3899,
      "step": 26464
    },
    {
      "epoch": 0.9477680090246566,
      "grad_norm": 1.4473627805709839,
      "learning_rate": 1.4265658512731316e-06,
      "loss": 1.5785,
      "step": 26465
    },
    {
      "epoch": 0.9478038211542249,
      "grad_norm": 1.9590723514556885,
      "learning_rate": 1.4246142999394751e-06,
      "loss": 1.564,
      "step": 26466
    },
    {
      "epoch": 0.9478396332837932,
      "grad_norm": 1.9028202295303345,
      "learning_rate": 1.4226640748104757e-06,
      "loss": 1.5236,
      "step": 26467
    },
    {
      "epoch": 0.9478754454133616,
      "grad_norm": 1.640310525894165,
      "learning_rate": 1.4207151759123683e-06,
      "loss": 1.5118,
      "step": 26468
    },
    {
      "epoch": 0.9479112575429298,
      "grad_norm": 1.495152473449707,
      "learning_rate": 1.418767603271387e-06,
      "loss": 1.4463,
      "step": 26469
    },
    {
      "epoch": 0.9479470696724981,
      "grad_norm": 1.3607423305511475,
      "learning_rate": 1.4168213569137223e-06,
      "loss": 1.2448,
      "step": 26470
    },
    {
      "epoch": 0.9479828818020664,
      "grad_norm": 1.2964245080947876,
      "learning_rate": 1.4148764368655754e-06,
      "loss": 1.4987,
      "step": 26471
    },
    {
      "epoch": 0.9480186939316346,
      "grad_norm": 1.689813494682312,
      "learning_rate": 1.4129328431530807e-06,
      "loss": 1.2008,
      "step": 26472
    },
    {
      "epoch": 0.9480545060612029,
      "grad_norm": 1.8330782651901245,
      "learning_rate": 1.4109905758024177e-06,
      "loss": 1.8116,
      "step": 26473
    },
    {
      "epoch": 0.9480903181907712,
      "grad_norm": 1.5751451253890991,
      "learning_rate": 1.4090496348397097e-06,
      "loss": 1.2216,
      "step": 26474
    },
    {
      "epoch": 0.9481261303203395,
      "grad_norm": 2.155590057373047,
      "learning_rate": 1.407110020291058e-06,
      "loss": 1.3441,
      "step": 26475
    },
    {
      "epoch": 0.9481619424499078,
      "grad_norm": 2.1611201763153076,
      "learning_rate": 1.4051717321825643e-06,
      "loss": 1.453,
      "step": 26476
    },
    {
      "epoch": 0.9481977545794761,
      "grad_norm": 1.6963903903961182,
      "learning_rate": 1.403234770540307e-06,
      "loss": 1.1672,
      "step": 26477
    },
    {
      "epoch": 0.9482335667090444,
      "grad_norm": 1.572092056274414,
      "learning_rate": 1.4012991353903549e-06,
      "loss": 1.6028,
      "step": 26478
    },
    {
      "epoch": 0.9482693788386126,
      "grad_norm": 1.452835202217102,
      "learning_rate": 1.3993648267587312e-06,
      "loss": 1.4143,
      "step": 26479
    },
    {
      "epoch": 0.9483051909681809,
      "grad_norm": 1.6740903854370117,
      "learning_rate": 1.3974318446714706e-06,
      "loss": 1.61,
      "step": 26480
    },
    {
      "epoch": 0.9483410030977492,
      "grad_norm": 1.7781686782836914,
      "learning_rate": 1.395500189154575e-06,
      "loss": 1.4476,
      "step": 26481
    },
    {
      "epoch": 0.9483768152273175,
      "grad_norm": 1.7080374956130981,
      "learning_rate": 1.3935698602340452e-06,
      "loss": 1.3767,
      "step": 26482
    },
    {
      "epoch": 0.9484126273568858,
      "grad_norm": 1.3993390798568726,
      "learning_rate": 1.3916408579358164e-06,
      "loss": 1.5493,
      "step": 26483
    },
    {
      "epoch": 0.9484484394864541,
      "grad_norm": 1.3815793991088867,
      "learning_rate": 1.3897131822858789e-06,
      "loss": 1.4565,
      "step": 26484
    },
    {
      "epoch": 0.9484842516160223,
      "grad_norm": 2.051628589630127,
      "learning_rate": 1.3877868333101562e-06,
      "loss": 1.5791,
      "step": 26485
    },
    {
      "epoch": 0.9485200637455906,
      "grad_norm": 1.3823293447494507,
      "learning_rate": 1.385861811034561e-06,
      "loss": 1.2821,
      "step": 26486
    },
    {
      "epoch": 0.9485558758751589,
      "grad_norm": 2.022139310836792,
      "learning_rate": 1.383938115484984e-06,
      "loss": 1.4049,
      "step": 26487
    },
    {
      "epoch": 0.9485916880047272,
      "grad_norm": 1.9774534702301025,
      "learning_rate": 1.3820157466873152e-06,
      "loss": 1.4567,
      "step": 26488
    },
    {
      "epoch": 0.9486275001342955,
      "grad_norm": 1.6379338502883911,
      "learning_rate": 1.3800947046674228e-06,
      "loss": 1.175,
      "step": 26489
    },
    {
      "epoch": 0.9486633122638638,
      "grad_norm": 1.9727866649627686,
      "learning_rate": 1.3781749894511308e-06,
      "loss": 1.3028,
      "step": 26490
    },
    {
      "epoch": 0.9486991243934321,
      "grad_norm": 1.7060546875,
      "learning_rate": 1.3762566010642962e-06,
      "loss": 1.3837,
      "step": 26491
    },
    {
      "epoch": 0.9487349365230003,
      "grad_norm": 1.596717119216919,
      "learning_rate": 1.3743395395326985e-06,
      "loss": 1.2626,
      "step": 26492
    },
    {
      "epoch": 0.9487707486525686,
      "grad_norm": 1.3802907466888428,
      "learning_rate": 1.3724238048821615e-06,
      "loss": 1.1592,
      "step": 26493
    },
    {
      "epoch": 0.9488065607821369,
      "grad_norm": 1.8134254217147827,
      "learning_rate": 1.370509397138431e-06,
      "loss": 1.674,
      "step": 26494
    },
    {
      "epoch": 0.9488423729117051,
      "grad_norm": 1.4546838998794556,
      "learning_rate": 1.3685963163272752e-06,
      "loss": 1.401,
      "step": 26495
    },
    {
      "epoch": 0.9488781850412735,
      "grad_norm": 1.6962960958480835,
      "learning_rate": 1.3666845624744406e-06,
      "loss": 1.3322,
      "step": 26496
    },
    {
      "epoch": 0.9489139971708418,
      "grad_norm": 1.3423304557800293,
      "learning_rate": 1.3647741356056287e-06,
      "loss": 1.2208,
      "step": 26497
    },
    {
      "epoch": 0.9489498093004101,
      "grad_norm": 1.656837821006775,
      "learning_rate": 1.3628650357465522e-06,
      "loss": 1.3385,
      "step": 26498
    },
    {
      "epoch": 0.9489856214299783,
      "grad_norm": 1.3497300148010254,
      "learning_rate": 1.3609572629228906e-06,
      "loss": 1.2634,
      "step": 26499
    },
    {
      "epoch": 0.9490214335595466,
      "grad_norm": 2.2240593433380127,
      "learning_rate": 1.3590508171603233e-06,
      "loss": 1.6153,
      "step": 26500
    },
    {
      "epoch": 0.9490572456891149,
      "grad_norm": 1.696808934211731,
      "learning_rate": 1.3571456984844743e-06,
      "loss": 1.33,
      "step": 26501
    },
    {
      "epoch": 0.9490930578186831,
      "grad_norm": 1.5843878984451294,
      "learning_rate": 1.3552419069210009e-06,
      "loss": 1.2396,
      "step": 26502
    },
    {
      "epoch": 0.9491288699482515,
      "grad_norm": 1.4548217058181763,
      "learning_rate": 1.3533394424954937e-06,
      "loss": 1.2981,
      "step": 26503
    },
    {
      "epoch": 0.9491646820778198,
      "grad_norm": 1.916221022605896,
      "learning_rate": 1.3514383052335766e-06,
      "loss": 1.3667,
      "step": 26504
    },
    {
      "epoch": 0.9492004942073881,
      "grad_norm": 1.5517075061798096,
      "learning_rate": 1.3495384951607958e-06,
      "loss": 1.2612,
      "step": 26505
    },
    {
      "epoch": 0.9492363063369563,
      "grad_norm": 3.5113589763641357,
      "learning_rate": 1.3476400123027312e-06,
      "loss": 1.6406,
      "step": 26506
    },
    {
      "epoch": 0.9492721184665246,
      "grad_norm": 1.6516262292861938,
      "learning_rate": 1.3457428566849173e-06,
      "loss": 1.6534,
      "step": 26507
    },
    {
      "epoch": 0.9493079305960929,
      "grad_norm": 1.575160264968872,
      "learning_rate": 1.3438470283328785e-06,
      "loss": 1.4022,
      "step": 26508
    },
    {
      "epoch": 0.9493437427256611,
      "grad_norm": 1.442672610282898,
      "learning_rate": 1.3419525272721168e-06,
      "loss": 1.4947,
      "step": 26509
    },
    {
      "epoch": 0.9493795548552295,
      "grad_norm": 1.832108736038208,
      "learning_rate": 1.3400593535281224e-06,
      "loss": 1.4931,
      "step": 26510
    },
    {
      "epoch": 0.9494153669847978,
      "grad_norm": 1.702636957168579,
      "learning_rate": 1.3381675071263755e-06,
      "loss": 1.564,
      "step": 26511
    },
    {
      "epoch": 0.949451179114366,
      "grad_norm": 1.6425777673721313,
      "learning_rate": 1.3362769880923221e-06,
      "loss": 1.2737,
      "step": 26512
    },
    {
      "epoch": 0.9494869912439343,
      "grad_norm": 1.3201589584350586,
      "learning_rate": 1.3343877964513863e-06,
      "loss": 1.3811,
      "step": 26513
    },
    {
      "epoch": 0.9495228033735026,
      "grad_norm": 2.1531147956848145,
      "learning_rate": 1.3324999322290033e-06,
      "loss": 1.2665,
      "step": 26514
    },
    {
      "epoch": 0.9495586155030709,
      "grad_norm": 1.7345833778381348,
      "learning_rate": 1.330613395450553e-06,
      "loss": 1.6657,
      "step": 26515
    },
    {
      "epoch": 0.9495944276326391,
      "grad_norm": 1.5731521844863892,
      "learning_rate": 1.3287281861414258e-06,
      "loss": 1.4453,
      "step": 26516
    },
    {
      "epoch": 0.9496302397622075,
      "grad_norm": 1.4536832571029663,
      "learning_rate": 1.3268443043269796e-06,
      "loss": 1.5805,
      "step": 26517
    },
    {
      "epoch": 0.9496660518917758,
      "grad_norm": 1.422829031944275,
      "learning_rate": 1.3249617500325718e-06,
      "loss": 1.7652,
      "step": 26518
    },
    {
      "epoch": 0.949701864021344,
      "grad_norm": 1.5982391834259033,
      "learning_rate": 1.3230805232835153e-06,
      "loss": 1.4507,
      "step": 26519
    },
    {
      "epoch": 0.9497376761509123,
      "grad_norm": 1.574022650718689,
      "learning_rate": 1.3212006241051345e-06,
      "loss": 1.5129,
      "step": 26520
    },
    {
      "epoch": 0.9497734882804806,
      "grad_norm": 1.3596420288085938,
      "learning_rate": 1.319322052522709e-06,
      "loss": 1.2358,
      "step": 26521
    },
    {
      "epoch": 0.9498093004100489,
      "grad_norm": 1.462693214416504,
      "learning_rate": 1.3174448085615187e-06,
      "loss": 1.4235,
      "step": 26522
    },
    {
      "epoch": 0.9498451125396171,
      "grad_norm": 1.5312095880508423,
      "learning_rate": 1.3155688922468101e-06,
      "loss": 1.4207,
      "step": 26523
    },
    {
      "epoch": 0.9498809246691855,
      "grad_norm": 1.4311823844909668,
      "learning_rate": 1.3136943036038297e-06,
      "loss": 1.3391,
      "step": 26524
    },
    {
      "epoch": 0.9499167367987538,
      "grad_norm": 1.4623321294784546,
      "learning_rate": 1.3118210426578015e-06,
      "loss": 1.2909,
      "step": 26525
    },
    {
      "epoch": 0.949952548928322,
      "grad_norm": 1.7718008756637573,
      "learning_rate": 1.3099491094339279e-06,
      "loss": 1.3087,
      "step": 26526
    },
    {
      "epoch": 0.9499883610578903,
      "grad_norm": 1.7378346920013428,
      "learning_rate": 1.3080785039573773e-06,
      "loss": 1.4943,
      "step": 26527
    },
    {
      "epoch": 0.9500241731874586,
      "grad_norm": 1.7058058977127075,
      "learning_rate": 1.3062092262533189e-06,
      "loss": 1.4128,
      "step": 26528
    },
    {
      "epoch": 0.9500599853170268,
      "grad_norm": 1.367444396018982,
      "learning_rate": 1.304341276346932e-06,
      "loss": 1.0962,
      "step": 26529
    },
    {
      "epoch": 0.9500957974465951,
      "grad_norm": 1.446695327758789,
      "learning_rate": 1.3024746542633082e-06,
      "loss": 1.2731,
      "step": 26530
    },
    {
      "epoch": 0.9501316095761635,
      "grad_norm": 1.3863316774368286,
      "learning_rate": 1.3006093600275825e-06,
      "loss": 1.5057,
      "step": 26531
    },
    {
      "epoch": 0.9501674217057318,
      "grad_norm": 2.1868419647216797,
      "learning_rate": 1.2987453936648575e-06,
      "loss": 1.3829,
      "step": 26532
    },
    {
      "epoch": 0.9502032338353,
      "grad_norm": 1.467094898223877,
      "learning_rate": 1.2968827552001793e-06,
      "loss": 1.5164,
      "step": 26533
    },
    {
      "epoch": 0.9502390459648683,
      "grad_norm": 1.3066911697387695,
      "learning_rate": 1.2950214446586284e-06,
      "loss": 1.3691,
      "step": 26534
    },
    {
      "epoch": 0.9502748580944366,
      "grad_norm": 1.5992457866668701,
      "learning_rate": 1.2931614620652511e-06,
      "loss": 1.4246,
      "step": 26535
    },
    {
      "epoch": 0.9503106702240048,
      "grad_norm": 1.3781858682632446,
      "learning_rate": 1.2913028074450607e-06,
      "loss": 1.4019,
      "step": 26536
    },
    {
      "epoch": 0.9503464823535731,
      "grad_norm": 1.9242557287216187,
      "learning_rate": 1.2894454808230593e-06,
      "loss": 1.3886,
      "step": 26537
    },
    {
      "epoch": 0.9503822944831415,
      "grad_norm": 1.6685802936553955,
      "learning_rate": 1.2875894822242496e-06,
      "loss": 1.3555,
      "step": 26538
    },
    {
      "epoch": 0.9504181066127098,
      "grad_norm": 1.6586769819259644,
      "learning_rate": 1.2857348116736002e-06,
      "loss": 1.3652,
      "step": 26539
    },
    {
      "epoch": 0.950453918742278,
      "grad_norm": 1.9376729726791382,
      "learning_rate": 1.2838814691960355e-06,
      "loss": 1.5063,
      "step": 26540
    },
    {
      "epoch": 0.9504897308718463,
      "grad_norm": 1.4433695077896118,
      "learning_rate": 1.2820294548165246e-06,
      "loss": 1.2929,
      "step": 26541
    },
    {
      "epoch": 0.9505255430014146,
      "grad_norm": 1.3399428129196167,
      "learning_rate": 1.2801787685599698e-06,
      "loss": 1.3715,
      "step": 26542
    },
    {
      "epoch": 0.9505613551309828,
      "grad_norm": 1.332889437675476,
      "learning_rate": 1.2783294104512734e-06,
      "loss": 1.4932,
      "step": 26543
    },
    {
      "epoch": 0.9505971672605511,
      "grad_norm": 1.9041041135787964,
      "learning_rate": 1.2764813805153041e-06,
      "loss": 1.6289,
      "step": 26544
    },
    {
      "epoch": 0.9506329793901195,
      "grad_norm": 1.5188629627227783,
      "learning_rate": 1.2746346787769425e-06,
      "loss": 1.08,
      "step": 26545
    },
    {
      "epoch": 0.9506687915196878,
      "grad_norm": 1.580101490020752,
      "learning_rate": 1.272789305261013e-06,
      "loss": 1.5524,
      "step": 26546
    },
    {
      "epoch": 0.950704603649256,
      "grad_norm": 1.4802852869033813,
      "learning_rate": 1.2709452599923731e-06,
      "loss": 1.4638,
      "step": 26547
    },
    {
      "epoch": 0.9507404157788243,
      "grad_norm": 3.1193196773529053,
      "learning_rate": 1.2691025429958037e-06,
      "loss": 1.2083,
      "step": 26548
    },
    {
      "epoch": 0.9507762279083926,
      "grad_norm": 1.8259837627410889,
      "learning_rate": 1.2672611542960954e-06,
      "loss": 1.3745,
      "step": 26549
    },
    {
      "epoch": 0.9508120400379608,
      "grad_norm": 1.8224256038665771,
      "learning_rate": 1.2654210939180511e-06,
      "loss": 1.4162,
      "step": 26550
    },
    {
      "epoch": 0.9508478521675291,
      "grad_norm": 2.090353488922119,
      "learning_rate": 1.2635823618863951e-06,
      "loss": 1.4225,
      "step": 26551
    },
    {
      "epoch": 0.9508836642970975,
      "grad_norm": 1.8558623790740967,
      "learning_rate": 1.2617449582258744e-06,
      "loss": 1.6497,
      "step": 26552
    },
    {
      "epoch": 0.9509194764266657,
      "grad_norm": 1.5263234376907349,
      "learning_rate": 1.2599088829612249e-06,
      "loss": 1.4847,
      "step": 26553
    },
    {
      "epoch": 0.950955288556234,
      "grad_norm": 1.3935396671295166,
      "learning_rate": 1.2580741361171267e-06,
      "loss": 1.5371,
      "step": 26554
    },
    {
      "epoch": 0.9509911006858023,
      "grad_norm": 1.8329626321792603,
      "learning_rate": 1.2562407177182712e-06,
      "loss": 1.3644,
      "step": 26555
    },
    {
      "epoch": 0.9510269128153706,
      "grad_norm": 1.6709609031677246,
      "learning_rate": 1.2544086277893386e-06,
      "loss": 1.6079,
      "step": 26556
    },
    {
      "epoch": 0.9510627249449388,
      "grad_norm": 1.656237244606018,
      "learning_rate": 1.2525778663549537e-06,
      "loss": 1.2923,
      "step": 26557
    },
    {
      "epoch": 0.9510985370745071,
      "grad_norm": 1.7023342847824097,
      "learning_rate": 1.2507484334397634e-06,
      "loss": 1.4957,
      "step": 26558
    },
    {
      "epoch": 0.9511343492040755,
      "grad_norm": 1.3861737251281738,
      "learning_rate": 1.2489203290683703e-06,
      "loss": 1.3171,
      "step": 26559
    },
    {
      "epoch": 0.9511701613336437,
      "grad_norm": 1.4072065353393555,
      "learning_rate": 1.2470935532653772e-06,
      "loss": 1.5584,
      "step": 26560
    },
    {
      "epoch": 0.951205973463212,
      "grad_norm": 1.6545718908309937,
      "learning_rate": 1.2452681060553639e-06,
      "loss": 1.6098,
      "step": 26561
    },
    {
      "epoch": 0.9512417855927803,
      "grad_norm": 1.359649658203125,
      "learning_rate": 1.243443987462878e-06,
      "loss": 1.4323,
      "step": 26562
    },
    {
      "epoch": 0.9512775977223485,
      "grad_norm": 1.870990514755249,
      "learning_rate": 1.2416211975124658e-06,
      "loss": 1.4793,
      "step": 26563
    },
    {
      "epoch": 0.9513134098519168,
      "grad_norm": 1.6119709014892578,
      "learning_rate": 1.2397997362286528e-06,
      "loss": 1.1674,
      "step": 26564
    },
    {
      "epoch": 0.9513492219814851,
      "grad_norm": 1.4036765098571777,
      "learning_rate": 1.2379796036359526e-06,
      "loss": 1.4654,
      "step": 26565
    },
    {
      "epoch": 0.9513850341110535,
      "grad_norm": 1.458703875541687,
      "learning_rate": 1.2361607997588343e-06,
      "loss": 1.538,
      "step": 26566
    },
    {
      "epoch": 0.9514208462406217,
      "grad_norm": 1.934006690979004,
      "learning_rate": 1.2343433246217673e-06,
      "loss": 1.5755,
      "step": 26567
    },
    {
      "epoch": 0.95145665837019,
      "grad_norm": 1.8799941539764404,
      "learning_rate": 1.232527178249232e-06,
      "loss": 1.5967,
      "step": 26568
    },
    {
      "epoch": 0.9514924704997583,
      "grad_norm": 2.1603283882141113,
      "learning_rate": 1.2307123606656312e-06,
      "loss": 1.2299,
      "step": 26569
    },
    {
      "epoch": 0.9515282826293265,
      "grad_norm": 2.093066692352295,
      "learning_rate": 1.2288988718953897e-06,
      "loss": 1.441,
      "step": 26570
    },
    {
      "epoch": 0.9515640947588948,
      "grad_norm": 2.1063077449798584,
      "learning_rate": 1.2270867119629103e-06,
      "loss": 1.376,
      "step": 26571
    },
    {
      "epoch": 0.9515999068884631,
      "grad_norm": 1.4114335775375366,
      "learning_rate": 1.2252758808925736e-06,
      "loss": 1.3026,
      "step": 26572
    },
    {
      "epoch": 0.9516357190180315,
      "grad_norm": 1.5860474109649658,
      "learning_rate": 1.2234663787087375e-06,
      "loss": 1.3799,
      "step": 26573
    },
    {
      "epoch": 0.9516715311475997,
      "grad_norm": 1.4082696437835693,
      "learning_rate": 1.2216582054357495e-06,
      "loss": 1.3476,
      "step": 26574
    },
    {
      "epoch": 0.951707343277168,
      "grad_norm": 1.7118854522705078,
      "learning_rate": 1.2198513610979346e-06,
      "loss": 1.3482,
      "step": 26575
    },
    {
      "epoch": 0.9517431554067363,
      "grad_norm": 2.112502336502075,
      "learning_rate": 1.2180458457196064e-06,
      "loss": 1.2543,
      "step": 26576
    },
    {
      "epoch": 0.9517789675363045,
      "grad_norm": 1.6754275560379028,
      "learning_rate": 1.2162416593250569e-06,
      "loss": 1.5346,
      "step": 26577
    },
    {
      "epoch": 0.9518147796658728,
      "grad_norm": 1.9799768924713135,
      "learning_rate": 1.2144388019385333e-06,
      "loss": 1.2265,
      "step": 26578
    },
    {
      "epoch": 0.9518505917954411,
      "grad_norm": 1.3142305612564087,
      "learning_rate": 1.2126372735843272e-06,
      "loss": 1.4418,
      "step": 26579
    },
    {
      "epoch": 0.9518864039250094,
      "grad_norm": 1.5378390550613403,
      "learning_rate": 1.2108370742866526e-06,
      "loss": 1.7047,
      "step": 26580
    },
    {
      "epoch": 0.9519222160545777,
      "grad_norm": 1.4842941761016846,
      "learning_rate": 1.2090382040697456e-06,
      "loss": 1.3786,
      "step": 26581
    },
    {
      "epoch": 0.951958028184146,
      "grad_norm": 1.9399491548538208,
      "learning_rate": 1.2072406629577871e-06,
      "loss": 1.6141,
      "step": 26582
    },
    {
      "epoch": 0.9519938403137143,
      "grad_norm": 2.0363988876342773,
      "learning_rate": 1.2054444509749906e-06,
      "loss": 1.5161,
      "step": 26583
    },
    {
      "epoch": 0.9520296524432825,
      "grad_norm": 1.677929162979126,
      "learning_rate": 1.203649568145493e-06,
      "loss": 1.4207,
      "step": 26584
    },
    {
      "epoch": 0.9520654645728508,
      "grad_norm": 1.5547820329666138,
      "learning_rate": 1.201856014493441e-06,
      "loss": 1.4561,
      "step": 26585
    },
    {
      "epoch": 0.9521012767024191,
      "grad_norm": 1.590218186378479,
      "learning_rate": 1.2000637900429934e-06,
      "loss": 1.3997,
      "step": 26586
    },
    {
      "epoch": 0.9521370888319874,
      "grad_norm": 1.4837863445281982,
      "learning_rate": 1.1982728948182308e-06,
      "loss": 1.3099,
      "step": 26587
    },
    {
      "epoch": 0.9521729009615557,
      "grad_norm": 1.7251883745193481,
      "learning_rate": 1.1964833288432674e-06,
      "loss": 1.3323,
      "step": 26588
    },
    {
      "epoch": 0.952208713091124,
      "grad_norm": 2.328374147415161,
      "learning_rate": 1.194695092142173e-06,
      "loss": 1.6759,
      "step": 26589
    },
    {
      "epoch": 0.9522445252206923,
      "grad_norm": 1.6122950315475464,
      "learning_rate": 1.1929081847390056e-06,
      "loss": 1.2681,
      "step": 26590
    },
    {
      "epoch": 0.9522803373502605,
      "grad_norm": 1.4836642742156982,
      "learning_rate": 1.191122606657813e-06,
      "loss": 1.2356,
      "step": 26591
    },
    {
      "epoch": 0.9523161494798288,
      "grad_norm": 1.837382435798645,
      "learning_rate": 1.1893383579226091e-06,
      "loss": 1.5641,
      "step": 26592
    },
    {
      "epoch": 0.9523519616093971,
      "grad_norm": 2.236088275909424,
      "learning_rate": 1.1875554385573972e-06,
      "loss": 1.538,
      "step": 26593
    },
    {
      "epoch": 0.9523877737389654,
      "grad_norm": 1.5058979988098145,
      "learning_rate": 1.185773848586158e-06,
      "loss": 1.551,
      "step": 26594
    },
    {
      "epoch": 0.9524235858685337,
      "grad_norm": 2.073533773422241,
      "learning_rate": 1.1839935880328946e-06,
      "loss": 1.7001,
      "step": 26595
    },
    {
      "epoch": 0.952459397998102,
      "grad_norm": 2.2686805725097656,
      "learning_rate": 1.1822146569215097e-06,
      "loss": 1.5851,
      "step": 26596
    },
    {
      "epoch": 0.9524952101276702,
      "grad_norm": 1.4711614847183228,
      "learning_rate": 1.1804370552759735e-06,
      "loss": 1.4615,
      "step": 26597
    },
    {
      "epoch": 0.9525310222572385,
      "grad_norm": 1.4939398765563965,
      "learning_rate": 1.178660783120189e-06,
      "loss": 1.4739,
      "step": 26598
    },
    {
      "epoch": 0.9525668343868068,
      "grad_norm": 1.7867990732192993,
      "learning_rate": 1.176885840478048e-06,
      "loss": 1.6565,
      "step": 26599
    },
    {
      "epoch": 0.952602646516375,
      "grad_norm": 1.488075613975525,
      "learning_rate": 1.1751122273734316e-06,
      "loss": 1.131,
      "step": 26600
    },
    {
      "epoch": 0.9526384586459434,
      "grad_norm": 2.2243800163269043,
      "learning_rate": 1.1733399438302206e-06,
      "loss": 1.2857,
      "step": 26601
    },
    {
      "epoch": 0.9526742707755117,
      "grad_norm": 1.2374006509780884,
      "learning_rate": 1.1715689898722404e-06,
      "loss": 1.6162,
      "step": 26602
    },
    {
      "epoch": 0.95271008290508,
      "grad_norm": 1.3920713663101196,
      "learning_rate": 1.1697993655233164e-06,
      "loss": 1.394,
      "step": 26603
    },
    {
      "epoch": 0.9527458950346482,
      "grad_norm": 1.5225958824157715,
      "learning_rate": 1.1680310708072518e-06,
      "loss": 1.46,
      "step": 26604
    },
    {
      "epoch": 0.9527817071642165,
      "grad_norm": 1.5914231538772583,
      "learning_rate": 1.1662641057478497e-06,
      "loss": 1.4585,
      "step": 26605
    },
    {
      "epoch": 0.9528175192937848,
      "grad_norm": 1.5339897871017456,
      "learning_rate": 1.1644984703688799e-06,
      "loss": 1.4752,
      "step": 26606
    },
    {
      "epoch": 0.952853331423353,
      "grad_norm": 1.8416153192520142,
      "learning_rate": 1.1627341646941015e-06,
      "loss": 1.706,
      "step": 26607
    },
    {
      "epoch": 0.9528891435529214,
      "grad_norm": 1.4199974536895752,
      "learning_rate": 1.1609711887472286e-06,
      "loss": 1.3469,
      "step": 26608
    },
    {
      "epoch": 0.9529249556824897,
      "grad_norm": 1.518681287765503,
      "learning_rate": 1.1592095425520088e-06,
      "loss": 1.392,
      "step": 26609
    },
    {
      "epoch": 0.952960767812058,
      "grad_norm": 1.9634442329406738,
      "learning_rate": 1.1574492261321236e-06,
      "loss": 1.1609,
      "step": 26610
    },
    {
      "epoch": 0.9529965799416262,
      "grad_norm": 1.7711156606674194,
      "learning_rate": 1.1556902395112645e-06,
      "loss": 1.3154,
      "step": 26611
    },
    {
      "epoch": 0.9530323920711945,
      "grad_norm": 1.8715670108795166,
      "learning_rate": 1.1539325827130799e-06,
      "loss": 1.1835,
      "step": 26612
    },
    {
      "epoch": 0.9530682042007628,
      "grad_norm": 1.6497482061386108,
      "learning_rate": 1.1521762557612502e-06,
      "loss": 1.1409,
      "step": 26613
    },
    {
      "epoch": 0.953104016330331,
      "grad_norm": 1.8769314289093018,
      "learning_rate": 1.1504212586793683e-06,
      "loss": 1.4843,
      "step": 26614
    },
    {
      "epoch": 0.9531398284598994,
      "grad_norm": 1.9313020706176758,
      "learning_rate": 1.1486675914910705e-06,
      "loss": 1.6296,
      "step": 26615
    },
    {
      "epoch": 0.9531756405894677,
      "grad_norm": 1.3665918111801147,
      "learning_rate": 1.1469152542199379e-06,
      "loss": 1.3802,
      "step": 26616
    },
    {
      "epoch": 0.953211452719036,
      "grad_norm": 1.4282692670822144,
      "learning_rate": 1.1451642468895518e-06,
      "loss": 1.2054,
      "step": 26617
    },
    {
      "epoch": 0.9532472648486042,
      "grad_norm": 1.5573898553848267,
      "learning_rate": 1.14341456952346e-06,
      "loss": 1.4843,
      "step": 26618
    },
    {
      "epoch": 0.9532830769781725,
      "grad_norm": 1.994338870048523,
      "learning_rate": 1.1416662221452211e-06,
      "loss": 1.3083,
      "step": 26619
    },
    {
      "epoch": 0.9533188891077408,
      "grad_norm": 1.591489315032959,
      "learning_rate": 1.139919204778339e-06,
      "loss": 1.3245,
      "step": 26620
    },
    {
      "epoch": 0.953354701237309,
      "grad_norm": 1.2546374797821045,
      "learning_rate": 1.1381735174463283e-06,
      "loss": 0.9723,
      "step": 26621
    },
    {
      "epoch": 0.9533905133668774,
      "grad_norm": 1.652453064918518,
      "learning_rate": 1.1364291601726585e-06,
      "loss": 1.4873,
      "step": 26622
    },
    {
      "epoch": 0.9534263254964457,
      "grad_norm": 1.4358993768692017,
      "learning_rate": 1.1346861329808112e-06,
      "loss": 1.5944,
      "step": 26623
    },
    {
      "epoch": 0.953462137626014,
      "grad_norm": 1.5629358291625977,
      "learning_rate": 1.1329444358942454e-06,
      "loss": 1.2247,
      "step": 26624
    },
    {
      "epoch": 0.9534979497555822,
      "grad_norm": 1.5762356519699097,
      "learning_rate": 1.1312040689363757e-06,
      "loss": 1.4513,
      "step": 26625
    },
    {
      "epoch": 0.9535337618851505,
      "grad_norm": 1.2871313095092773,
      "learning_rate": 1.1294650321306277e-06,
      "loss": 1.2398,
      "step": 26626
    },
    {
      "epoch": 0.9535695740147188,
      "grad_norm": 1.5916223526000977,
      "learning_rate": 1.127727325500394e-06,
      "loss": 1.5777,
      "step": 26627
    },
    {
      "epoch": 0.953605386144287,
      "grad_norm": 1.6760125160217285,
      "learning_rate": 1.1259909490690556e-06,
      "loss": 1.4228,
      "step": 26628
    },
    {
      "epoch": 0.9536411982738554,
      "grad_norm": 1.6149671077728271,
      "learning_rate": 1.1242559028599609e-06,
      "loss": 1.4672,
      "step": 26629
    },
    {
      "epoch": 0.9536770104034237,
      "grad_norm": 1.6717442274093628,
      "learning_rate": 1.1225221868964686e-06,
      "loss": 1.2505,
      "step": 26630
    },
    {
      "epoch": 0.9537128225329919,
      "grad_norm": 1.4332804679870605,
      "learning_rate": 1.1207898012018936e-06,
      "loss": 1.3423,
      "step": 26631
    },
    {
      "epoch": 0.9537486346625602,
      "grad_norm": 1.3053473234176636,
      "learning_rate": 1.1190587457995506e-06,
      "loss": 1.7562,
      "step": 26632
    },
    {
      "epoch": 0.9537844467921285,
      "grad_norm": 1.356052279472351,
      "learning_rate": 1.1173290207127207e-06,
      "loss": 1.47,
      "step": 26633
    },
    {
      "epoch": 0.9538202589216968,
      "grad_norm": 1.9330263137817383,
      "learning_rate": 1.1156006259646856e-06,
      "loss": 1.6352,
      "step": 26634
    },
    {
      "epoch": 0.953856071051265,
      "grad_norm": 2.3237826824188232,
      "learning_rate": 1.1138735615786933e-06,
      "loss": 1.331,
      "step": 26635
    },
    {
      "epoch": 0.9538918831808334,
      "grad_norm": 1.5061534643173218,
      "learning_rate": 1.1121478275779696e-06,
      "loss": 1.2172,
      "step": 26636
    },
    {
      "epoch": 0.9539276953104017,
      "grad_norm": 1.64857816696167,
      "learning_rate": 1.1104234239857402e-06,
      "loss": 1.6107,
      "step": 26637
    },
    {
      "epoch": 0.9539635074399699,
      "grad_norm": 1.9496674537658691,
      "learning_rate": 1.1087003508252202e-06,
      "loss": 1.3248,
      "step": 26638
    },
    {
      "epoch": 0.9539993195695382,
      "grad_norm": 1.5698601007461548,
      "learning_rate": 1.1069786081195687e-06,
      "loss": 1.4504,
      "step": 26639
    },
    {
      "epoch": 0.9540351316991065,
      "grad_norm": 1.6455700397491455,
      "learning_rate": 1.105258195891945e-06,
      "loss": 1.368,
      "step": 26640
    },
    {
      "epoch": 0.9540709438286747,
      "grad_norm": 1.2808388471603394,
      "learning_rate": 1.1035391141655195e-06,
      "loss": 1.5479,
      "step": 26641
    },
    {
      "epoch": 0.954106755958243,
      "grad_norm": 2.0161030292510986,
      "learning_rate": 1.1018213629634178e-06,
      "loss": 1.6358,
      "step": 26642
    },
    {
      "epoch": 0.9541425680878114,
      "grad_norm": 1.532594919204712,
      "learning_rate": 1.1001049423087217e-06,
      "loss": 1.046,
      "step": 26643
    },
    {
      "epoch": 0.9541783802173797,
      "grad_norm": 1.7464717626571655,
      "learning_rate": 1.098389852224546e-06,
      "loss": 1.1881,
      "step": 26644
    },
    {
      "epoch": 0.9542141923469479,
      "grad_norm": 1.4636871814727783,
      "learning_rate": 1.0966760927339726e-06,
      "loss": 1.4925,
      "step": 26645
    },
    {
      "epoch": 0.9542500044765162,
      "grad_norm": 1.6371227502822876,
      "learning_rate": 1.094963663860027e-06,
      "loss": 1.5365,
      "step": 26646
    },
    {
      "epoch": 0.9542858166060845,
      "grad_norm": 1.4008408784866333,
      "learning_rate": 1.0932525656257796e-06,
      "loss": 1.4636,
      "step": 26647
    },
    {
      "epoch": 0.9543216287356527,
      "grad_norm": 1.474956750869751,
      "learning_rate": 1.0915427980542348e-06,
      "loss": 1.5885,
      "step": 26648
    },
    {
      "epoch": 0.954357440865221,
      "grad_norm": 1.310798168182373,
      "learning_rate": 1.089834361168407e-06,
      "loss": 1.1127,
      "step": 26649
    },
    {
      "epoch": 0.9543932529947894,
      "grad_norm": 1.807181715965271,
      "learning_rate": 1.088127254991267e-06,
      "loss": 1.4589,
      "step": 26650
    },
    {
      "epoch": 0.9544290651243577,
      "grad_norm": 1.9590734243392944,
      "learning_rate": 1.086421479545785e-06,
      "loss": 1.5383,
      "step": 26651
    },
    {
      "epoch": 0.9544648772539259,
      "grad_norm": 1.3808904886245728,
      "learning_rate": 1.0847170348549096e-06,
      "loss": 1.3616,
      "step": 26652
    },
    {
      "epoch": 0.9545006893834942,
      "grad_norm": 2.0818872451782227,
      "learning_rate": 1.0830139209415779e-06,
      "loss": 1.6911,
      "step": 26653
    },
    {
      "epoch": 0.9545365015130625,
      "grad_norm": 1.6958940029144287,
      "learning_rate": 1.081312137828716e-06,
      "loss": 1.8023,
      "step": 26654
    },
    {
      "epoch": 0.9545723136426307,
      "grad_norm": 1.6296740770339966,
      "learning_rate": 1.0796116855391724e-06,
      "loss": 1.0263,
      "step": 26655
    },
    {
      "epoch": 0.954608125772199,
      "grad_norm": 1.9934402704238892,
      "learning_rate": 1.0779125640958843e-06,
      "loss": 1.5194,
      "step": 26656
    },
    {
      "epoch": 0.9546439379017674,
      "grad_norm": 1.6104828119277954,
      "learning_rate": 1.0762147735216665e-06,
      "loss": 1.5336,
      "step": 26657
    },
    {
      "epoch": 0.9546797500313356,
      "grad_norm": 1.7793383598327637,
      "learning_rate": 1.0745183138393788e-06,
      "loss": 1.5556,
      "step": 26658
    },
    {
      "epoch": 0.9547155621609039,
      "grad_norm": 1.319137454032898,
      "learning_rate": 1.0728231850718363e-06,
      "loss": 1.6508,
      "step": 26659
    },
    {
      "epoch": 0.9547513742904722,
      "grad_norm": 1.5470972061157227,
      "learning_rate": 1.071129387241865e-06,
      "loss": 1.492,
      "step": 26660
    },
    {
      "epoch": 0.9547871864200405,
      "grad_norm": 1.630874752998352,
      "learning_rate": 1.0694369203722354e-06,
      "loss": 1.7498,
      "step": 26661
    },
    {
      "epoch": 0.9548229985496087,
      "grad_norm": 2.3035964965820312,
      "learning_rate": 1.0677457844857186e-06,
      "loss": 1.5807,
      "step": 26662
    },
    {
      "epoch": 0.954858810679177,
      "grad_norm": 1.6051185131072998,
      "learning_rate": 1.0660559796050739e-06,
      "loss": 1.4429,
      "step": 26663
    },
    {
      "epoch": 0.9548946228087454,
      "grad_norm": 2.025474786758423,
      "learning_rate": 1.0643675057530166e-06,
      "loss": 1.2662,
      "step": 26664
    },
    {
      "epoch": 0.9549304349383136,
      "grad_norm": 1.6414517164230347,
      "learning_rate": 1.0626803629522951e-06,
      "loss": 0.994,
      "step": 26665
    },
    {
      "epoch": 0.9549662470678819,
      "grad_norm": 1.5560110807418823,
      "learning_rate": 1.0609945512255692e-06,
      "loss": 1.2765,
      "step": 26666
    },
    {
      "epoch": 0.9550020591974502,
      "grad_norm": 1.2948863506317139,
      "learning_rate": 1.0593100705955538e-06,
      "loss": 1.4118,
      "step": 26667
    },
    {
      "epoch": 0.9550378713270185,
      "grad_norm": 1.3893499374389648,
      "learning_rate": 1.0576269210848867e-06,
      "loss": 1.5091,
      "step": 26668
    },
    {
      "epoch": 0.9550736834565867,
      "grad_norm": 1.5306798219680786,
      "learning_rate": 1.055945102716227e-06,
      "loss": 1.2156,
      "step": 26669
    },
    {
      "epoch": 0.955109495586155,
      "grad_norm": 1.7268040180206299,
      "learning_rate": 1.0542646155122015e-06,
      "loss": 1.7405,
      "step": 26670
    },
    {
      "epoch": 0.9551453077157234,
      "grad_norm": 2.0746474266052246,
      "learning_rate": 1.0525854594954143e-06,
      "loss": 1.3419,
      "step": 26671
    },
    {
      "epoch": 0.9551811198452916,
      "grad_norm": 1.7448663711547852,
      "learning_rate": 1.0509076346884583e-06,
      "loss": 1.827,
      "step": 26672
    },
    {
      "epoch": 0.9552169319748599,
      "grad_norm": 1.811957597732544,
      "learning_rate": 1.0492311411138934e-06,
      "loss": 1.6032,
      "step": 26673
    },
    {
      "epoch": 0.9552527441044282,
      "grad_norm": 2.3352155685424805,
      "learning_rate": 1.0475559787943012e-06,
      "loss": 1.2959,
      "step": 26674
    },
    {
      "epoch": 0.9552885562339964,
      "grad_norm": 1.8025147914886475,
      "learning_rate": 1.0458821477521974e-06,
      "loss": 1.5484,
      "step": 26675
    },
    {
      "epoch": 0.9553243683635647,
      "grad_norm": 1.6638069152832031,
      "learning_rate": 1.0442096480101082e-06,
      "loss": 1.1597,
      "step": 26676
    },
    {
      "epoch": 0.955360180493133,
      "grad_norm": 1.7823617458343506,
      "learning_rate": 1.042538479590527e-06,
      "loss": 1.7342,
      "step": 26677
    },
    {
      "epoch": 0.9553959926227014,
      "grad_norm": 1.538569450378418,
      "learning_rate": 1.0408686425159574e-06,
      "loss": 1.3907,
      "step": 26678
    },
    {
      "epoch": 0.9554318047522696,
      "grad_norm": 1.5950886011123657,
      "learning_rate": 1.0392001368088377e-06,
      "loss": 1.4772,
      "step": 26679
    },
    {
      "epoch": 0.9554676168818379,
      "grad_norm": 2.025176525115967,
      "learning_rate": 1.0375329624916386e-06,
      "loss": 1.4865,
      "step": 26680
    },
    {
      "epoch": 0.9555034290114062,
      "grad_norm": 1.6462210416793823,
      "learning_rate": 1.0358671195867865e-06,
      "loss": 1.2826,
      "step": 26681
    },
    {
      "epoch": 0.9555392411409744,
      "grad_norm": 1.741531491279602,
      "learning_rate": 1.0342026081166745e-06,
      "loss": 1.3633,
      "step": 26682
    },
    {
      "epoch": 0.9555750532705427,
      "grad_norm": 1.3028637170791626,
      "learning_rate": 1.0325394281037293e-06,
      "loss": 1.3505,
      "step": 26683
    },
    {
      "epoch": 0.955610865400111,
      "grad_norm": 1.8471804857254028,
      "learning_rate": 1.0308775795702775e-06,
      "loss": 1.2296,
      "step": 26684
    },
    {
      "epoch": 0.9556466775296794,
      "grad_norm": 1.38747239112854,
      "learning_rate": 1.0292170625387342e-06,
      "loss": 1.4824,
      "step": 26685
    },
    {
      "epoch": 0.9556824896592476,
      "grad_norm": 1.3638286590576172,
      "learning_rate": 1.0275578770313933e-06,
      "loss": 1.3719,
      "step": 26686
    },
    {
      "epoch": 0.9557183017888159,
      "grad_norm": 1.4321017265319824,
      "learning_rate": 1.025900023070614e-06,
      "loss": 1.4655,
      "step": 26687
    },
    {
      "epoch": 0.9557541139183842,
      "grad_norm": 1.3580139875411987,
      "learning_rate": 1.0242435006786677e-06,
      "loss": 1.5321,
      "step": 26688
    },
    {
      "epoch": 0.9557899260479524,
      "grad_norm": 1.8781472444534302,
      "learning_rate": 1.0225883098778588e-06,
      "loss": 1.3467,
      "step": 26689
    },
    {
      "epoch": 0.9558257381775207,
      "grad_norm": 1.5805599689483643,
      "learning_rate": 1.0209344506904694e-06,
      "loss": 1.4534,
      "step": 26690
    },
    {
      "epoch": 0.955861550307089,
      "grad_norm": 1.7332732677459717,
      "learning_rate": 1.019281923138715e-06,
      "loss": 1.4556,
      "step": 26691
    },
    {
      "epoch": 0.9558973624366573,
      "grad_norm": 1.5968459844589233,
      "learning_rate": 1.0176307272448448e-06,
      "loss": 1.3131,
      "step": 26692
    },
    {
      "epoch": 0.9559331745662256,
      "grad_norm": 1.7155627012252808,
      "learning_rate": 1.015980863031074e-06,
      "loss": 1.657,
      "step": 26693
    },
    {
      "epoch": 0.9559689866957939,
      "grad_norm": 1.7028433084487915,
      "learning_rate": 1.0143323305196184e-06,
      "loss": 1.4293,
      "step": 26694
    },
    {
      "epoch": 0.9560047988253622,
      "grad_norm": 1.3736257553100586,
      "learning_rate": 1.0126851297326157e-06,
      "loss": 1.1688,
      "step": 26695
    },
    {
      "epoch": 0.9560406109549304,
      "grad_norm": 1.3911411762237549,
      "learning_rate": 1.0110392606922703e-06,
      "loss": 1.3592,
      "step": 26696
    },
    {
      "epoch": 0.9560764230844987,
      "grad_norm": 1.8647515773773193,
      "learning_rate": 1.0093947234206868e-06,
      "loss": 1.5904,
      "step": 26697
    },
    {
      "epoch": 0.956112235214067,
      "grad_norm": 2.0917210578918457,
      "learning_rate": 1.0077515179400254e-06,
      "loss": 1.5137,
      "step": 26698
    },
    {
      "epoch": 0.9561480473436353,
      "grad_norm": 1.5692753791809082,
      "learning_rate": 1.0061096442723683e-06,
      "loss": 1.1976,
      "step": 26699
    },
    {
      "epoch": 0.9561838594732036,
      "grad_norm": 1.5351765155792236,
      "learning_rate": 1.004469102439809e-06,
      "loss": 1.2061,
      "step": 26700
    },
    {
      "epoch": 0.9562196716027719,
      "grad_norm": 1.3395408391952515,
      "learning_rate": 1.0028298924644408e-06,
      "loss": 1.1257,
      "step": 26701
    },
    {
      "epoch": 0.9562554837323402,
      "grad_norm": 1.8311903476715088,
      "learning_rate": 1.0011920143682796e-06,
      "loss": 1.3361,
      "step": 26702
    },
    {
      "epoch": 0.9562912958619084,
      "grad_norm": 1.7517163753509521,
      "learning_rate": 9.995554681733855e-07,
      "loss": 1.0912,
      "step": 26703
    },
    {
      "epoch": 0.9563271079914767,
      "grad_norm": 1.5621646642684937,
      "learning_rate": 9.97920253901774e-07,
      "loss": 1.7053,
      "step": 26704
    },
    {
      "epoch": 0.956362920121045,
      "grad_norm": 1.5236084461212158,
      "learning_rate": 9.96286371575439e-07,
      "loss": 1.6629,
      "step": 26705
    },
    {
      "epoch": 0.9563987322506133,
      "grad_norm": 2.2787911891937256,
      "learning_rate": 9.946538212163736e-07,
      "loss": 1.6588,
      "step": 26706
    },
    {
      "epoch": 0.9564345443801816,
      "grad_norm": 1.7489440441131592,
      "learning_rate": 9.930226028465272e-07,
      "loss": 1.5958,
      "step": 26707
    },
    {
      "epoch": 0.9564703565097499,
      "grad_norm": 2.500220775604248,
      "learning_rate": 9.913927164878488e-07,
      "loss": 1.5689,
      "step": 26708
    },
    {
      "epoch": 0.9565061686393181,
      "grad_norm": 1.724913239479065,
      "learning_rate": 9.897641621622765e-07,
      "loss": 1.1326,
      "step": 26709
    },
    {
      "epoch": 0.9565419807688864,
      "grad_norm": 2.303943634033203,
      "learning_rate": 9.88136939891704e-07,
      "loss": 1.4504,
      "step": 26710
    },
    {
      "epoch": 0.9565777928984547,
      "grad_norm": 2.162876844406128,
      "learning_rate": 9.865110496980356e-07,
      "loss": 1.5306,
      "step": 26711
    },
    {
      "epoch": 0.956613605028023,
      "grad_norm": 1.4706625938415527,
      "learning_rate": 9.84886491603154e-07,
      "loss": 1.3547,
      "step": 26712
    },
    {
      "epoch": 0.9566494171575913,
      "grad_norm": 1.7020697593688965,
      "learning_rate": 9.832632656288864e-07,
      "loss": 1.5613,
      "step": 26713
    },
    {
      "epoch": 0.9566852292871596,
      "grad_norm": 1.5223006010055542,
      "learning_rate": 9.81641371797104e-07,
      "loss": 1.5805,
      "step": 26714
    },
    {
      "epoch": 0.9567210414167279,
      "grad_norm": 1.1612199544906616,
      "learning_rate": 9.800208101296115e-07,
      "loss": 1.5998,
      "step": 26715
    },
    {
      "epoch": 0.9567568535462961,
      "grad_norm": 1.6327825784683228,
      "learning_rate": 9.784015806482028e-07,
      "loss": 1.1863,
      "step": 26716
    },
    {
      "epoch": 0.9567926656758644,
      "grad_norm": 1.9279059171676636,
      "learning_rate": 9.767836833746714e-07,
      "loss": 1.4357,
      "step": 26717
    },
    {
      "epoch": 0.9568284778054327,
      "grad_norm": 1.5725735425949097,
      "learning_rate": 9.751671183307888e-07,
      "loss": 1.4529,
      "step": 26718
    },
    {
      "epoch": 0.956864289935001,
      "grad_norm": 1.6223695278167725,
      "learning_rate": 9.735518855383152e-07,
      "loss": 1.4732,
      "step": 26719
    },
    {
      "epoch": 0.9569001020645693,
      "grad_norm": 2.1366162300109863,
      "learning_rate": 9.719379850189447e-07,
      "loss": 1.4565,
      "step": 26720
    },
    {
      "epoch": 0.9569359141941376,
      "grad_norm": 2.1586503982543945,
      "learning_rate": 9.703254167944154e-07,
      "loss": 1.2536,
      "step": 26721
    },
    {
      "epoch": 0.9569717263237059,
      "grad_norm": 2.3456790447235107,
      "learning_rate": 9.68714180886421e-07,
      "loss": 1.3688,
      "step": 26722
    },
    {
      "epoch": 0.9570075384532741,
      "grad_norm": 1.5606224536895752,
      "learning_rate": 9.67104277316644e-07,
      "loss": 1.6539,
      "step": 26723
    },
    {
      "epoch": 0.9570433505828424,
      "grad_norm": 1.9346030950546265,
      "learning_rate": 9.654957061067228e-07,
      "loss": 1.4414,
      "step": 26724
    },
    {
      "epoch": 0.9570791627124107,
      "grad_norm": 1.4974099397659302,
      "learning_rate": 9.638884672783176e-07,
      "loss": 1.3323,
      "step": 26725
    },
    {
      "epoch": 0.9571149748419789,
      "grad_norm": 1.7413378953933716,
      "learning_rate": 9.622825608530561e-07,
      "loss": 1.1622,
      "step": 26726
    },
    {
      "epoch": 0.9571507869715473,
      "grad_norm": 1.3514119386672974,
      "learning_rate": 9.606779868525206e-07,
      "loss": 1.0968,
      "step": 26727
    },
    {
      "epoch": 0.9571865991011156,
      "grad_norm": 1.4090226888656616,
      "learning_rate": 9.590747452983161e-07,
      "loss": 1.6259,
      "step": 26728
    },
    {
      "epoch": 0.9572224112306839,
      "grad_norm": 1.9629863500595093,
      "learning_rate": 9.574728362120033e-07,
      "loss": 1.0267,
      "step": 26729
    },
    {
      "epoch": 0.9572582233602521,
      "grad_norm": 1.4839998483657837,
      "learning_rate": 9.558722596151425e-07,
      "loss": 1.3511,
      "step": 26730
    },
    {
      "epoch": 0.9572940354898204,
      "grad_norm": 1.6294034719467163,
      "learning_rate": 9.5427301552925e-07,
      "loss": 1.4625,
      "step": 26731
    },
    {
      "epoch": 0.9573298476193887,
      "grad_norm": 1.5633814334869385,
      "learning_rate": 9.526751039758641e-07,
      "loss": 1.6409,
      "step": 26732
    },
    {
      "epoch": 0.9573656597489569,
      "grad_norm": 1.6685106754302979,
      "learning_rate": 9.510785249764786e-07,
      "loss": 1.4664,
      "step": 26733
    },
    {
      "epoch": 0.9574014718785253,
      "grad_norm": 1.2557251453399658,
      "learning_rate": 9.494832785525653e-07,
      "loss": 1.4488,
      "step": 26734
    },
    {
      "epoch": 0.9574372840080936,
      "grad_norm": 1.9086263179779053,
      "learning_rate": 9.478893647255849e-07,
      "loss": 1.2848,
      "step": 26735
    },
    {
      "epoch": 0.9574730961376618,
      "grad_norm": 1.5768110752105713,
      "learning_rate": 9.462967835169756e-07,
      "loss": 1.4434,
      "step": 26736
    },
    {
      "epoch": 0.9575089082672301,
      "grad_norm": 1.5403549671173096,
      "learning_rate": 9.44705534948187e-07,
      "loss": 1.4612,
      "step": 26737
    },
    {
      "epoch": 0.9575447203967984,
      "grad_norm": 1.6581268310546875,
      "learning_rate": 9.431156190406131e-07,
      "loss": 1.2959,
      "step": 26738
    },
    {
      "epoch": 0.9575805325263667,
      "grad_norm": 1.4390212297439575,
      "learning_rate": 9.41527035815637e-07,
      "loss": 1.2648,
      "step": 26739
    },
    {
      "epoch": 0.9576163446559349,
      "grad_norm": 1.7185239791870117,
      "learning_rate": 9.399397852946413e-07,
      "loss": 1.5569,
      "step": 26740
    },
    {
      "epoch": 0.9576521567855032,
      "grad_norm": 2.195068359375,
      "learning_rate": 9.383538674989756e-07,
      "loss": 1.7491,
      "step": 26741
    },
    {
      "epoch": 0.9576879689150716,
      "grad_norm": 1.4795840978622437,
      "learning_rate": 9.367692824499786e-07,
      "loss": 1.5406,
      "step": 26742
    },
    {
      "epoch": 0.9577237810446398,
      "grad_norm": 1.736876130104065,
      "learning_rate": 9.351860301689775e-07,
      "loss": 1.2074,
      "step": 26743
    },
    {
      "epoch": 0.9577595931742081,
      "grad_norm": 1.4114105701446533,
      "learning_rate": 9.336041106772553e-07,
      "loss": 1.3731,
      "step": 26744
    },
    {
      "epoch": 0.9577954053037764,
      "grad_norm": 1.434982180595398,
      "learning_rate": 9.320235239961061e-07,
      "loss": 1.257,
      "step": 26745
    },
    {
      "epoch": 0.9578312174333447,
      "grad_norm": 2.827622413635254,
      "learning_rate": 9.304442701467908e-07,
      "loss": 1.6111,
      "step": 26746
    },
    {
      "epoch": 0.9578670295629129,
      "grad_norm": 2.319575786590576,
      "learning_rate": 9.288663491505478e-07,
      "loss": 1.4424,
      "step": 26747
    },
    {
      "epoch": 0.9579028416924812,
      "grad_norm": 1.5229041576385498,
      "learning_rate": 9.27289761028638e-07,
      "loss": 1.2918,
      "step": 26748
    },
    {
      "epoch": 0.9579386538220496,
      "grad_norm": 1.6973974704742432,
      "learning_rate": 9.257145058022331e-07,
      "loss": 1.3539,
      "step": 26749
    },
    {
      "epoch": 0.9579744659516178,
      "grad_norm": 1.4195101261138916,
      "learning_rate": 9.241405834925388e-07,
      "loss": 1.3205,
      "step": 26750
    },
    {
      "epoch": 0.9580102780811861,
      "grad_norm": 1.6695009469985962,
      "learning_rate": 9.225679941207488e-07,
      "loss": 1.4648,
      "step": 26751
    },
    {
      "epoch": 0.9580460902107544,
      "grad_norm": 1.8575265407562256,
      "learning_rate": 9.20996737708002e-07,
      "loss": 1.4749,
      "step": 26752
    },
    {
      "epoch": 0.9580819023403226,
      "grad_norm": 1.3980778455734253,
      "learning_rate": 9.19426814275437e-07,
      "loss": 1.5525,
      "step": 26753
    },
    {
      "epoch": 0.9581177144698909,
      "grad_norm": 1.5089133977890015,
      "learning_rate": 9.178582238441702e-07,
      "loss": 1.1665,
      "step": 26754
    },
    {
      "epoch": 0.9581535265994592,
      "grad_norm": 1.390755534172058,
      "learning_rate": 9.162909664353292e-07,
      "loss": 1.6808,
      "step": 26755
    },
    {
      "epoch": 0.9581893387290276,
      "grad_norm": 1.4605809450149536,
      "learning_rate": 9.14725042069986e-07,
      "loss": 1.5083,
      "step": 26756
    },
    {
      "epoch": 0.9582251508585958,
      "grad_norm": 1.6914252042770386,
      "learning_rate": 9.131604507691904e-07,
      "loss": 1.635,
      "step": 26757
    },
    {
      "epoch": 0.9582609629881641,
      "grad_norm": 1.5005120038986206,
      "learning_rate": 9.115971925540257e-07,
      "loss": 1.3975,
      "step": 26758
    },
    {
      "epoch": 0.9582967751177324,
      "grad_norm": 1.4641746282577515,
      "learning_rate": 9.100352674454971e-07,
      "loss": 1.4,
      "step": 26759
    },
    {
      "epoch": 0.9583325872473006,
      "grad_norm": 1.9565989971160889,
      "learning_rate": 9.084746754646323e-07,
      "loss": 1.4052,
      "step": 26760
    },
    {
      "epoch": 0.9583683993768689,
      "grad_norm": 1.6255415678024292,
      "learning_rate": 9.069154166324146e-07,
      "loss": 1.3676,
      "step": 26761
    },
    {
      "epoch": 0.9584042115064372,
      "grad_norm": 2.551762104034424,
      "learning_rate": 9.053574909698381e-07,
      "loss": 1.4617,
      "step": 26762
    },
    {
      "epoch": 0.9584400236360056,
      "grad_norm": 1.3324257135391235,
      "learning_rate": 9.038008984978419e-07,
      "loss": 1.3645,
      "step": 26763
    },
    {
      "epoch": 0.9584758357655738,
      "grad_norm": 3.2408790588378906,
      "learning_rate": 9.022456392373868e-07,
      "loss": 1.3834,
      "step": 26764
    },
    {
      "epoch": 0.9585116478951421,
      "grad_norm": 1.4632230997085571,
      "learning_rate": 9.006917132093895e-07,
      "loss": 1.0799,
      "step": 26765
    },
    {
      "epoch": 0.9585474600247104,
      "grad_norm": 1.6773384809494019,
      "learning_rate": 8.991391204347555e-07,
      "loss": 1.4687,
      "step": 26766
    },
    {
      "epoch": 0.9585832721542786,
      "grad_norm": 1.8148605823516846,
      "learning_rate": 8.97587860934368e-07,
      "loss": 1.2124,
      "step": 26767
    },
    {
      "epoch": 0.9586190842838469,
      "grad_norm": 1.7556577920913696,
      "learning_rate": 8.960379347291103e-07,
      "loss": 1.6248,
      "step": 26768
    },
    {
      "epoch": 0.9586548964134152,
      "grad_norm": 2.0786924362182617,
      "learning_rate": 8.944893418398326e-07,
      "loss": 1.5726,
      "step": 26769
    },
    {
      "epoch": 0.9586907085429835,
      "grad_norm": 1.6416964530944824,
      "learning_rate": 8.929420822873513e-07,
      "loss": 1.4571,
      "step": 26770
    },
    {
      "epoch": 0.9587265206725518,
      "grad_norm": 1.4926072359085083,
      "learning_rate": 8.913961560925055e-07,
      "loss": 1.3523,
      "step": 26771
    },
    {
      "epoch": 0.9587623328021201,
      "grad_norm": 1.511873483657837,
      "learning_rate": 8.898515632760784e-07,
      "loss": 1.0218,
      "step": 26772
    },
    {
      "epoch": 0.9587981449316884,
      "grad_norm": 1.5736192464828491,
      "learning_rate": 8.883083038588536e-07,
      "loss": 1.7893,
      "step": 26773
    },
    {
      "epoch": 0.9588339570612566,
      "grad_norm": 1.9865953922271729,
      "learning_rate": 8.867663778616031e-07,
      "loss": 1.5569,
      "step": 26774
    },
    {
      "epoch": 0.9588697691908249,
      "grad_norm": 1.5352931022644043,
      "learning_rate": 8.852257853050661e-07,
      "loss": 1.7685,
      "step": 26775
    },
    {
      "epoch": 0.9589055813203932,
      "grad_norm": 2.0710299015045166,
      "learning_rate": 8.836865262099481e-07,
      "loss": 1.5833,
      "step": 26776
    },
    {
      "epoch": 0.9589413934499615,
      "grad_norm": 1.8323180675506592,
      "learning_rate": 8.821486005969992e-07,
      "loss": 1.1984,
      "step": 26777
    },
    {
      "epoch": 0.9589772055795298,
      "grad_norm": 1.1376245021820068,
      "learning_rate": 8.806120084868807e-07,
      "loss": 1.3374,
      "step": 26778
    },
    {
      "epoch": 0.9590130177090981,
      "grad_norm": 1.5147979259490967,
      "learning_rate": 8.79076749900265e-07,
      "loss": 1.4927,
      "step": 26779
    },
    {
      "epoch": 0.9590488298386664,
      "grad_norm": 1.8162474632263184,
      "learning_rate": 8.775428248578243e-07,
      "loss": 1.8344,
      "step": 26780
    },
    {
      "epoch": 0.9590846419682346,
      "grad_norm": 1.5417896509170532,
      "learning_rate": 8.760102333801756e-07,
      "loss": 1.3234,
      "step": 26781
    },
    {
      "epoch": 0.9591204540978029,
      "grad_norm": 1.8801335096359253,
      "learning_rate": 8.744789754879579e-07,
      "loss": 1.6405,
      "step": 26782
    },
    {
      "epoch": 0.9591562662273712,
      "grad_norm": 1.4927922487258911,
      "learning_rate": 8.729490512017547e-07,
      "loss": 1.3414,
      "step": 26783
    },
    {
      "epoch": 0.9591920783569395,
      "grad_norm": 1.7957075834274292,
      "learning_rate": 8.714204605421716e-07,
      "loss": 1.4618,
      "step": 26784
    },
    {
      "epoch": 0.9592278904865078,
      "grad_norm": 1.5633784532546997,
      "learning_rate": 8.69893203529748e-07,
      "loss": 1.208,
      "step": 26785
    },
    {
      "epoch": 0.9592637026160761,
      "grad_norm": 2.0784788131713867,
      "learning_rate": 8.683672801850451e-07,
      "loss": 1.613,
      "step": 26786
    },
    {
      "epoch": 0.9592995147456443,
      "grad_norm": 1.31596040725708,
      "learning_rate": 8.668426905285909e-07,
      "loss": 0.8767,
      "step": 26787
    },
    {
      "epoch": 0.9593353268752126,
      "grad_norm": 1.5244179964065552,
      "learning_rate": 8.653194345808913e-07,
      "loss": 1.4995,
      "step": 26788
    },
    {
      "epoch": 0.9593711390047809,
      "grad_norm": 1.6521975994110107,
      "learning_rate": 8.63797512362452e-07,
      "loss": 1.2428,
      "step": 26789
    },
    {
      "epoch": 0.9594069511343492,
      "grad_norm": 1.871695637702942,
      "learning_rate": 8.622769238937345e-07,
      "loss": 1.2734,
      "step": 26790
    },
    {
      "epoch": 0.9594427632639175,
      "grad_norm": 1.8507094383239746,
      "learning_rate": 8.607576691952002e-07,
      "loss": 1.5353,
      "step": 26791
    },
    {
      "epoch": 0.9594785753934858,
      "grad_norm": 1.6126267910003662,
      "learning_rate": 8.592397482872993e-07,
      "loss": 1.3408,
      "step": 26792
    },
    {
      "epoch": 0.9595143875230541,
      "grad_norm": 1.529656171798706,
      "learning_rate": 8.577231611904379e-07,
      "loss": 1.6628,
      "step": 26793
    },
    {
      "epoch": 0.9595501996526223,
      "grad_norm": 1.551055908203125,
      "learning_rate": 8.562079079250219e-07,
      "loss": 1.3079,
      "step": 26794
    },
    {
      "epoch": 0.9595860117821906,
      "grad_norm": 1.8348814249038696,
      "learning_rate": 8.546939885114569e-07,
      "loss": 1.394,
      "step": 26795
    },
    {
      "epoch": 0.9596218239117589,
      "grad_norm": 1.4194611310958862,
      "learning_rate": 8.531814029700935e-07,
      "loss": 1.3471,
      "step": 26796
    },
    {
      "epoch": 0.9596576360413271,
      "grad_norm": 1.5185151100158691,
      "learning_rate": 8.516701513212821e-07,
      "loss": 1.5099,
      "step": 26797
    },
    {
      "epoch": 0.9596934481708955,
      "grad_norm": 1.589985728263855,
      "learning_rate": 8.501602335853509e-07,
      "loss": 1.2476,
      "step": 26798
    },
    {
      "epoch": 0.9597292603004638,
      "grad_norm": 2.1401360034942627,
      "learning_rate": 8.48651649782628e-07,
      "loss": 1.6254,
      "step": 26799
    },
    {
      "epoch": 0.9597650724300321,
      "grad_norm": 1.7402527332305908,
      "learning_rate": 8.471443999333972e-07,
      "loss": 1.423,
      "step": 26800
    },
    {
      "epoch": 0.9598008845596003,
      "grad_norm": 1.600137710571289,
      "learning_rate": 8.456384840579423e-07,
      "loss": 1.5019,
      "step": 26801
    },
    {
      "epoch": 0.9598366966891686,
      "grad_norm": 1.6320627927780151,
      "learning_rate": 8.441339021765138e-07,
      "loss": 1.3925,
      "step": 26802
    },
    {
      "epoch": 0.9598725088187369,
      "grad_norm": 2.637521266937256,
      "learning_rate": 8.426306543093732e-07,
      "loss": 1.6728,
      "step": 26803
    },
    {
      "epoch": 0.9599083209483051,
      "grad_norm": 1.9787664413452148,
      "learning_rate": 8.411287404767265e-07,
      "loss": 1.6501,
      "step": 26804
    },
    {
      "epoch": 0.9599441330778735,
      "grad_norm": 1.341884970664978,
      "learning_rate": 8.396281606987799e-07,
      "loss": 1.4126,
      "step": 26805
    },
    {
      "epoch": 0.9599799452074418,
      "grad_norm": 1.7115846872329712,
      "learning_rate": 8.381289149957395e-07,
      "loss": 1.5408,
      "step": 26806
    },
    {
      "epoch": 0.9600157573370101,
      "grad_norm": 1.4531255960464478,
      "learning_rate": 8.366310033877667e-07,
      "loss": 1.589,
      "step": 26807
    },
    {
      "epoch": 0.9600515694665783,
      "grad_norm": 2.3292057514190674,
      "learning_rate": 8.351344258950123e-07,
      "loss": 1.8701,
      "step": 26808
    },
    {
      "epoch": 0.9600873815961466,
      "grad_norm": 1.7623049020767212,
      "learning_rate": 8.336391825376044e-07,
      "loss": 1.3654,
      "step": 26809
    },
    {
      "epoch": 0.9601231937257149,
      "grad_norm": 1.5028655529022217,
      "learning_rate": 8.321452733356605e-07,
      "loss": 1.4138,
      "step": 26810
    },
    {
      "epoch": 0.9601590058552831,
      "grad_norm": 1.9508914947509766,
      "learning_rate": 8.306526983092977e-07,
      "loss": 1.3763,
      "step": 26811
    },
    {
      "epoch": 0.9601948179848515,
      "grad_norm": 1.7204222679138184,
      "learning_rate": 8.291614574785777e-07,
      "loss": 1.3373,
      "step": 26812
    },
    {
      "epoch": 0.9602306301144198,
      "grad_norm": 1.9033406972885132,
      "learning_rate": 8.276715508635624e-07,
      "loss": 1.3409,
      "step": 26813
    },
    {
      "epoch": 0.960266442243988,
      "grad_norm": 1.5977619886398315,
      "learning_rate": 8.261829784843133e-07,
      "loss": 1.2249,
      "step": 26814
    },
    {
      "epoch": 0.9603022543735563,
      "grad_norm": 1.9996955394744873,
      "learning_rate": 8.246957403608479e-07,
      "loss": 1.5399,
      "step": 26815
    },
    {
      "epoch": 0.9603380665031246,
      "grad_norm": 1.686150312423706,
      "learning_rate": 8.232098365131613e-07,
      "loss": 1.2805,
      "step": 26816
    },
    {
      "epoch": 0.9603738786326929,
      "grad_norm": 1.4930355548858643,
      "learning_rate": 8.217252669612708e-07,
      "loss": 1.4063,
      "step": 26817
    },
    {
      "epoch": 0.9604096907622611,
      "grad_norm": 1.8459813594818115,
      "learning_rate": 8.20242031725138e-07,
      "loss": 1.5286,
      "step": 26818
    },
    {
      "epoch": 0.9604455028918295,
      "grad_norm": 1.9831894636154175,
      "learning_rate": 8.187601308247028e-07,
      "loss": 1.4814,
      "step": 26819
    },
    {
      "epoch": 0.9604813150213978,
      "grad_norm": 1.6175930500030518,
      "learning_rate": 8.172795642799269e-07,
      "loss": 1.24,
      "step": 26820
    },
    {
      "epoch": 0.960517127150966,
      "grad_norm": 1.5015861988067627,
      "learning_rate": 8.158003321107167e-07,
      "loss": 1.2342,
      "step": 26821
    },
    {
      "epoch": 0.9605529392805343,
      "grad_norm": 1.2254098653793335,
      "learning_rate": 8.143224343369671e-07,
      "loss": 1.3663,
      "step": 26822
    },
    {
      "epoch": 0.9605887514101026,
      "grad_norm": 1.7844117879867554,
      "learning_rate": 8.128458709785736e-07,
      "loss": 1.5133,
      "step": 26823
    },
    {
      "epoch": 0.9606245635396709,
      "grad_norm": 1.9229958057403564,
      "learning_rate": 8.113706420553868e-07,
      "loss": 1.7314,
      "step": 26824
    },
    {
      "epoch": 0.9606603756692391,
      "grad_norm": 1.4775229692459106,
      "learning_rate": 8.098967475872798e-07,
      "loss": 1.4693,
      "step": 26825
    },
    {
      "epoch": 0.9606961877988075,
      "grad_norm": 1.6402254104614258,
      "learning_rate": 8.084241875940591e-07,
      "loss": 1.3664,
      "step": 26826
    },
    {
      "epoch": 0.9607319999283758,
      "grad_norm": 1.7152255773544312,
      "learning_rate": 8.069529620955418e-07,
      "loss": 1.3495,
      "step": 26827
    },
    {
      "epoch": 0.960767812057944,
      "grad_norm": 1.5333672761917114,
      "learning_rate": 8.054830711115236e-07,
      "loss": 1.533,
      "step": 26828
    },
    {
      "epoch": 0.9608036241875123,
      "grad_norm": 1.5828067064285278,
      "learning_rate": 8.040145146617883e-07,
      "loss": 1.7531,
      "step": 26829
    },
    {
      "epoch": 0.9608394363170806,
      "grad_norm": 1.8916761875152588,
      "learning_rate": 8.025472927660649e-07,
      "loss": 1.0839,
      "step": 26830
    },
    {
      "epoch": 0.9608752484466488,
      "grad_norm": 1.5456327199935913,
      "learning_rate": 8.010814054441262e-07,
      "loss": 1.5393,
      "step": 26831
    },
    {
      "epoch": 0.9609110605762171,
      "grad_norm": 1.50007164478302,
      "learning_rate": 7.996168527156789e-07,
      "loss": 1.049,
      "step": 26832
    },
    {
      "epoch": 0.9609468727057855,
      "grad_norm": 1.552908182144165,
      "learning_rate": 7.981536346004292e-07,
      "loss": 1.4244,
      "step": 26833
    },
    {
      "epoch": 0.9609826848353538,
      "grad_norm": 1.4380766153335571,
      "learning_rate": 7.966917511180505e-07,
      "loss": 1.1892,
      "step": 26834
    },
    {
      "epoch": 0.961018496964922,
      "grad_norm": 2.1194546222686768,
      "learning_rate": 7.952312022882269e-07,
      "loss": 1.5398,
      "step": 26835
    },
    {
      "epoch": 0.9610543090944903,
      "grad_norm": 1.35401451587677,
      "learning_rate": 7.937719881306094e-07,
      "loss": 1.566,
      "step": 26836
    },
    {
      "epoch": 0.9610901212240586,
      "grad_norm": 2.0986156463623047,
      "learning_rate": 7.923141086648156e-07,
      "loss": 1.3092,
      "step": 26837
    },
    {
      "epoch": 0.9611259333536268,
      "grad_norm": 1.8315749168395996,
      "learning_rate": 7.908575639104631e-07,
      "loss": 1.5848,
      "step": 26838
    },
    {
      "epoch": 0.9611617454831951,
      "grad_norm": 1.6822844743728638,
      "learning_rate": 7.894023538871587e-07,
      "loss": 1.5722,
      "step": 26839
    },
    {
      "epoch": 0.9611975576127635,
      "grad_norm": 1.2145322561264038,
      "learning_rate": 7.879484786144753e-07,
      "loss": 1.2355,
      "step": 26840
    },
    {
      "epoch": 0.9612333697423318,
      "grad_norm": 1.5983326435089111,
      "learning_rate": 7.864959381119641e-07,
      "loss": 1.6364,
      "step": 26841
    },
    {
      "epoch": 0.9612691818719,
      "grad_norm": 1.8771876096725464,
      "learning_rate": 7.85044732399165e-07,
      "loss": 1.4767,
      "step": 26842
    },
    {
      "epoch": 0.9613049940014683,
      "grad_norm": 2.0207412242889404,
      "learning_rate": 7.83594861495629e-07,
      "loss": 1.4324,
      "step": 26843
    },
    {
      "epoch": 0.9613408061310366,
      "grad_norm": 1.887495517730713,
      "learning_rate": 7.821463254208405e-07,
      "loss": 1.4604,
      "step": 26844
    },
    {
      "epoch": 0.9613766182606048,
      "grad_norm": 1.4683293104171753,
      "learning_rate": 7.80699124194284e-07,
      "loss": 1.5779,
      "step": 26845
    },
    {
      "epoch": 0.9614124303901731,
      "grad_norm": 1.590532660484314,
      "learning_rate": 7.792532578354439e-07,
      "loss": 1.4709,
      "step": 26846
    },
    {
      "epoch": 0.9614482425197415,
      "grad_norm": 1.5941145420074463,
      "learning_rate": 7.778087263637601e-07,
      "loss": 1.5715,
      "step": 26847
    },
    {
      "epoch": 0.9614840546493097,
      "grad_norm": 2.0739402770996094,
      "learning_rate": 7.763655297986839e-07,
      "loss": 1.6764,
      "step": 26848
    },
    {
      "epoch": 0.961519866778878,
      "grad_norm": 1.8480732440948486,
      "learning_rate": 7.749236681595995e-07,
      "loss": 1.6399,
      "step": 26849
    },
    {
      "epoch": 0.9615556789084463,
      "grad_norm": 1.648714542388916,
      "learning_rate": 7.734831414659471e-07,
      "loss": 1.5086,
      "step": 26850
    },
    {
      "epoch": 0.9615914910380146,
      "grad_norm": 1.796379804611206,
      "learning_rate": 7.720439497370668e-07,
      "loss": 1.3986,
      "step": 26851
    },
    {
      "epoch": 0.9616273031675828,
      "grad_norm": 1.7396949529647827,
      "learning_rate": 7.706060929923542e-07,
      "loss": 1.1852,
      "step": 26852
    },
    {
      "epoch": 0.9616631152971511,
      "grad_norm": 1.374097228050232,
      "learning_rate": 7.691695712511382e-07,
      "loss": 1.4309,
      "step": 26853
    },
    {
      "epoch": 0.9616989274267195,
      "grad_norm": 1.4233494997024536,
      "learning_rate": 7.677343845327478e-07,
      "loss": 1.1654,
      "step": 26854
    },
    {
      "epoch": 0.9617347395562877,
      "grad_norm": 2.057396173477173,
      "learning_rate": 7.663005328564787e-07,
      "loss": 1.6129,
      "step": 26855
    },
    {
      "epoch": 0.961770551685856,
      "grad_norm": 2.265347480773926,
      "learning_rate": 7.648680162416489e-07,
      "loss": 1.3166,
      "step": 26856
    },
    {
      "epoch": 0.9618063638154243,
      "grad_norm": 1.4549821615219116,
      "learning_rate": 7.634368347075093e-07,
      "loss": 1.2215,
      "step": 26857
    },
    {
      "epoch": 0.9618421759449926,
      "grad_norm": 1.7157262563705444,
      "learning_rate": 7.620069882733227e-07,
      "loss": 1.5253,
      "step": 26858
    },
    {
      "epoch": 0.9618779880745608,
      "grad_norm": 2.227339506149292,
      "learning_rate": 7.605784769583291e-07,
      "loss": 1.5059,
      "step": 26859
    },
    {
      "epoch": 0.9619138002041291,
      "grad_norm": 1.5298230648040771,
      "learning_rate": 7.591513007817242e-07,
      "loss": 1.3268,
      "step": 26860
    },
    {
      "epoch": 0.9619496123336975,
      "grad_norm": 1.7910478115081787,
      "learning_rate": 7.577254597627481e-07,
      "loss": 1.3596,
      "step": 26861
    },
    {
      "epoch": 0.9619854244632657,
      "grad_norm": 1.3827205896377563,
      "learning_rate": 7.563009539205524e-07,
      "loss": 1.4843,
      "step": 26862
    },
    {
      "epoch": 0.962021236592834,
      "grad_norm": 1.5558743476867676,
      "learning_rate": 7.548777832743214e-07,
      "loss": 1.2898,
      "step": 26863
    },
    {
      "epoch": 0.9620570487224023,
      "grad_norm": 2.3799028396606445,
      "learning_rate": 7.534559478431735e-07,
      "loss": 1.2981,
      "step": 26864
    },
    {
      "epoch": 0.9620928608519705,
      "grad_norm": 1.3155205249786377,
      "learning_rate": 7.52035447646271e-07,
      "loss": 1.3632,
      "step": 26865
    },
    {
      "epoch": 0.9621286729815388,
      "grad_norm": 1.3187897205352783,
      "learning_rate": 7.506162827027097e-07,
      "loss": 1.3126,
      "step": 26866
    },
    {
      "epoch": 0.9621644851111071,
      "grad_norm": 1.7988380193710327,
      "learning_rate": 7.491984530315854e-07,
      "loss": 1.6965,
      "step": 26867
    },
    {
      "epoch": 0.9622002972406755,
      "grad_norm": 1.1581627130508423,
      "learning_rate": 7.477819586519719e-07,
      "loss": 1.1106,
      "step": 26868
    },
    {
      "epoch": 0.9622361093702437,
      "grad_norm": 1.507202386856079,
      "learning_rate": 7.463667995829205e-07,
      "loss": 1.4695,
      "step": 26869
    },
    {
      "epoch": 0.962271921499812,
      "grad_norm": 2.0441157817840576,
      "learning_rate": 7.449529758434826e-07,
      "loss": 2.2557,
      "step": 26870
    },
    {
      "epoch": 0.9623077336293803,
      "grad_norm": 1.7305978536605835,
      "learning_rate": 7.435404874526542e-07,
      "loss": 1.238,
      "step": 26871
    },
    {
      "epoch": 0.9623435457589485,
      "grad_norm": 1.8210771083831787,
      "learning_rate": 7.421293344294755e-07,
      "loss": 1.5418,
      "step": 26872
    },
    {
      "epoch": 0.9623793578885168,
      "grad_norm": 1.6746025085449219,
      "learning_rate": 7.407195167929093e-07,
      "loss": 1.2805,
      "step": 26873
    },
    {
      "epoch": 0.9624151700180851,
      "grad_norm": 1.6006801128387451,
      "learning_rate": 7.393110345619291e-07,
      "loss": 1.269,
      "step": 26874
    },
    {
      "epoch": 0.9624509821476535,
      "grad_norm": 2.0808916091918945,
      "learning_rate": 7.379038877554755e-07,
      "loss": 1.2495,
      "step": 26875
    },
    {
      "epoch": 0.9624867942772217,
      "grad_norm": 1.9244236946105957,
      "learning_rate": 7.364980763924889e-07,
      "loss": 1.3046,
      "step": 26876
    },
    {
      "epoch": 0.96252260640679,
      "grad_norm": 2.2387166023254395,
      "learning_rate": 7.350936004918873e-07,
      "loss": 1.4992,
      "step": 26877
    },
    {
      "epoch": 0.9625584185363583,
      "grad_norm": 1.9266955852508545,
      "learning_rate": 7.336904600725447e-07,
      "loss": 1.306,
      "step": 26878
    },
    {
      "epoch": 0.9625942306659265,
      "grad_norm": 1.6920135021209717,
      "learning_rate": 7.322886551533681e-07,
      "loss": 1.6659,
      "step": 26879
    },
    {
      "epoch": 0.9626300427954948,
      "grad_norm": 1.9262080192565918,
      "learning_rate": 7.308881857531869e-07,
      "loss": 1.4505,
      "step": 26880
    },
    {
      "epoch": 0.9626658549250631,
      "grad_norm": 1.438104271888733,
      "learning_rate": 7.294890518908748e-07,
      "loss": 1.3511,
      "step": 26881
    },
    {
      "epoch": 0.9627016670546314,
      "grad_norm": 1.2074106931686401,
      "learning_rate": 7.280912535852169e-07,
      "loss": 1.4729,
      "step": 26882
    },
    {
      "epoch": 0.9627374791841997,
      "grad_norm": 1.594930648803711,
      "learning_rate": 7.266947908550536e-07,
      "loss": 1.2189,
      "step": 26883
    },
    {
      "epoch": 0.962773291313768,
      "grad_norm": 1.5803894996643066,
      "learning_rate": 7.252996637191589e-07,
      "loss": 1.0278,
      "step": 26884
    },
    {
      "epoch": 0.9628091034433363,
      "grad_norm": 1.9029607772827148,
      "learning_rate": 7.239058721962954e-07,
      "loss": 1.3136,
      "step": 26885
    },
    {
      "epoch": 0.9628449155729045,
      "grad_norm": 2.055635452270508,
      "learning_rate": 7.22513416305226e-07,
      "loss": 1.294,
      "step": 26886
    },
    {
      "epoch": 0.9628807277024728,
      "grad_norm": 1.8848527669906616,
      "learning_rate": 7.211222960646691e-07,
      "loss": 1.4366,
      "step": 26887
    },
    {
      "epoch": 0.9629165398320411,
      "grad_norm": 1.2585186958312988,
      "learning_rate": 7.197325114933651e-07,
      "loss": 1.1339,
      "step": 26888
    },
    {
      "epoch": 0.9629523519616094,
      "grad_norm": 1.8972463607788086,
      "learning_rate": 7.18344062609988e-07,
      "loss": 1.5986,
      "step": 26889
    },
    {
      "epoch": 0.9629881640911777,
      "grad_norm": 2.0489518642425537,
      "learning_rate": 7.16956949433234e-07,
      "loss": 1.325,
      "step": 26890
    },
    {
      "epoch": 0.963023976220746,
      "grad_norm": 1.7141057252883911,
      "learning_rate": 7.155711719817548e-07,
      "loss": 1.3843,
      "step": 26891
    },
    {
      "epoch": 0.9630597883503142,
      "grad_norm": 1.6005711555480957,
      "learning_rate": 7.141867302742023e-07,
      "loss": 1.265,
      "step": 26892
    },
    {
      "epoch": 0.9630956004798825,
      "grad_norm": 1.6023467779159546,
      "learning_rate": 7.128036243291947e-07,
      "loss": 1.2884,
      "step": 26893
    },
    {
      "epoch": 0.9631314126094508,
      "grad_norm": 1.2686173915863037,
      "learning_rate": 7.114218541653395e-07,
      "loss": 1.1523,
      "step": 26894
    },
    {
      "epoch": 0.9631672247390191,
      "grad_norm": 1.4661974906921387,
      "learning_rate": 7.100414198012439e-07,
      "loss": 1.2727,
      "step": 26895
    },
    {
      "epoch": 0.9632030368685874,
      "grad_norm": 1.7659021615982056,
      "learning_rate": 7.086623212554488e-07,
      "loss": 1.5836,
      "step": 26896
    },
    {
      "epoch": 0.9632388489981557,
      "grad_norm": 1.6870405673980713,
      "learning_rate": 7.072845585465282e-07,
      "loss": 1.3479,
      "step": 26897
    },
    {
      "epoch": 0.963274661127724,
      "grad_norm": 1.9497534036636353,
      "learning_rate": 7.059081316930227e-07,
      "loss": 1.334,
      "step": 26898
    },
    {
      "epoch": 0.9633104732572922,
      "grad_norm": 1.5201635360717773,
      "learning_rate": 7.045330407134398e-07,
      "loss": 1.5913,
      "step": 26899
    },
    {
      "epoch": 0.9633462853868605,
      "grad_norm": 1.6073867082595825,
      "learning_rate": 7.03159285626287e-07,
      "loss": 1.531,
      "step": 26900
    },
    {
      "epoch": 0.9633820975164288,
      "grad_norm": 2.0254600048065186,
      "learning_rate": 7.017868664500382e-07,
      "loss": 1.5583,
      "step": 26901
    },
    {
      "epoch": 0.963417909645997,
      "grad_norm": 2.1424214839935303,
      "learning_rate": 7.004157832031677e-07,
      "loss": 1.5673,
      "step": 26902
    },
    {
      "epoch": 0.9634537217755654,
      "grad_norm": 1.7584158182144165,
      "learning_rate": 6.990460359041051e-07,
      "loss": 1.17,
      "step": 26903
    },
    {
      "epoch": 0.9634895339051337,
      "grad_norm": 1.9335455894470215,
      "learning_rate": 6.976776245712913e-07,
      "loss": 1.464,
      "step": 26904
    },
    {
      "epoch": 0.963525346034702,
      "grad_norm": 1.4011025428771973,
      "learning_rate": 6.963105492231336e-07,
      "loss": 1.5505,
      "step": 26905
    },
    {
      "epoch": 0.9635611581642702,
      "grad_norm": 1.5605167150497437,
      "learning_rate": 6.949448098780398e-07,
      "loss": 1.2234,
      "step": 26906
    },
    {
      "epoch": 0.9635969702938385,
      "grad_norm": 1.5061067342758179,
      "learning_rate": 6.935804065543505e-07,
      "loss": 1.1857,
      "step": 26907
    },
    {
      "epoch": 0.9636327824234068,
      "grad_norm": 1.4534320831298828,
      "learning_rate": 6.922173392704512e-07,
      "loss": 1.625,
      "step": 26908
    },
    {
      "epoch": 0.963668594552975,
      "grad_norm": 1.4374990463256836,
      "learning_rate": 6.908556080446715e-07,
      "loss": 1.3349,
      "step": 26909
    },
    {
      "epoch": 0.9637044066825434,
      "grad_norm": 1.656003475189209,
      "learning_rate": 6.894952128953191e-07,
      "loss": 1.3548,
      "step": 26910
    },
    {
      "epoch": 0.9637402188121117,
      "grad_norm": 1.9220056533813477,
      "learning_rate": 6.881361538407127e-07,
      "loss": 1.3714,
      "step": 26911
    },
    {
      "epoch": 0.96377603094168,
      "grad_norm": 1.139539361000061,
      "learning_rate": 6.867784308991266e-07,
      "loss": 1.3443,
      "step": 26912
    },
    {
      "epoch": 0.9638118430712482,
      "grad_norm": 1.2987382411956787,
      "learning_rate": 6.854220440888459e-07,
      "loss": 1.4791,
      "step": 26913
    },
    {
      "epoch": 0.9638476552008165,
      "grad_norm": 1.2175135612487793,
      "learning_rate": 6.840669934280897e-07,
      "loss": 1.3745,
      "step": 26914
    },
    {
      "epoch": 0.9638834673303848,
      "grad_norm": 1.5563571453094482,
      "learning_rate": 6.827132789351098e-07,
      "loss": 1.6409,
      "step": 26915
    },
    {
      "epoch": 0.963919279459953,
      "grad_norm": 1.1882492303848267,
      "learning_rate": 6.813609006281141e-07,
      "loss": 1.41,
      "step": 26916
    },
    {
      "epoch": 0.9639550915895214,
      "grad_norm": 1.3806182146072388,
      "learning_rate": 6.800098585252989e-07,
      "loss": 1.3614,
      "step": 26917
    },
    {
      "epoch": 0.9639909037190897,
      "grad_norm": 1.8224741220474243,
      "learning_rate": 6.786601526448277e-07,
      "loss": 1.3087,
      "step": 26918
    },
    {
      "epoch": 0.964026715848658,
      "grad_norm": 1.6787079572677612,
      "learning_rate": 6.773117830048747e-07,
      "loss": 1.5826,
      "step": 26919
    },
    {
      "epoch": 0.9640625279782262,
      "grad_norm": 2.0062367916107178,
      "learning_rate": 6.75964749623581e-07,
      "loss": 1.6547,
      "step": 26920
    },
    {
      "epoch": 0.9640983401077945,
      "grad_norm": 1.6232973337173462,
      "learning_rate": 6.746190525190543e-07,
      "loss": 1.529,
      "step": 26921
    },
    {
      "epoch": 0.9641341522373628,
      "grad_norm": 1.5797841548919678,
      "learning_rate": 6.732746917094135e-07,
      "loss": 1.5343,
      "step": 26922
    },
    {
      "epoch": 0.964169964366931,
      "grad_norm": 1.5757808685302734,
      "learning_rate": 6.719316672127329e-07,
      "loss": 1.5117,
      "step": 26923
    },
    {
      "epoch": 0.9642057764964994,
      "grad_norm": 1.458806037902832,
      "learning_rate": 6.70589979047087e-07,
      "loss": 1.3089,
      "step": 26924
    },
    {
      "epoch": 0.9642415886260677,
      "grad_norm": 1.3439996242523193,
      "learning_rate": 6.692496272305282e-07,
      "loss": 1.4445,
      "step": 26925
    },
    {
      "epoch": 0.964277400755636,
      "grad_norm": 1.373887538909912,
      "learning_rate": 6.679106117810974e-07,
      "loss": 1.4416,
      "step": 26926
    },
    {
      "epoch": 0.9643132128852042,
      "grad_norm": 1.9046086072921753,
      "learning_rate": 6.665729327167913e-07,
      "loss": 1.7455,
      "step": 26927
    },
    {
      "epoch": 0.9643490250147725,
      "grad_norm": 1.4031591415405273,
      "learning_rate": 6.652365900556179e-07,
      "loss": 1.5515,
      "step": 26928
    },
    {
      "epoch": 0.9643848371443408,
      "grad_norm": 1.617029070854187,
      "learning_rate": 6.639015838155515e-07,
      "loss": 1.713,
      "step": 26929
    },
    {
      "epoch": 0.964420649273909,
      "grad_norm": 1.4018628597259521,
      "learning_rate": 6.625679140145557e-07,
      "loss": 0.8941,
      "step": 26930
    },
    {
      "epoch": 0.9644564614034774,
      "grad_norm": 1.6945066452026367,
      "learning_rate": 6.612355806705828e-07,
      "loss": 1.3377,
      "step": 26931
    },
    {
      "epoch": 0.9644922735330457,
      "grad_norm": 1.6313774585723877,
      "learning_rate": 6.599045838015294e-07,
      "loss": 1.0908,
      "step": 26932
    },
    {
      "epoch": 0.9645280856626139,
      "grad_norm": 1.6346851587295532,
      "learning_rate": 6.585749234253258e-07,
      "loss": 1.3156,
      "step": 26933
    },
    {
      "epoch": 0.9645638977921822,
      "grad_norm": 2.169135570526123,
      "learning_rate": 6.572465995598575e-07,
      "loss": 1.5391,
      "step": 26934
    },
    {
      "epoch": 0.9645997099217505,
      "grad_norm": 1.4126479625701904,
      "learning_rate": 6.559196122229994e-07,
      "loss": 1.5252,
      "step": 26935
    },
    {
      "epoch": 0.9646355220513188,
      "grad_norm": 2.0156967639923096,
      "learning_rate": 6.545939614325924e-07,
      "loss": 1.5874,
      "step": 26936
    },
    {
      "epoch": 0.964671334180887,
      "grad_norm": 1.959639310836792,
      "learning_rate": 6.532696472064781e-07,
      "loss": 1.4451,
      "step": 26937
    },
    {
      "epoch": 0.9647071463104554,
      "grad_norm": 1.6670535802841187,
      "learning_rate": 6.519466695624755e-07,
      "loss": 1.7676,
      "step": 26938
    },
    {
      "epoch": 0.9647429584400237,
      "grad_norm": 1.3304088115692139,
      "learning_rate": 6.506250285183812e-07,
      "loss": 1.3935,
      "step": 26939
    },
    {
      "epoch": 0.9647787705695919,
      "grad_norm": 1.6989744901657104,
      "learning_rate": 6.493047240919703e-07,
      "loss": 1.3568,
      "step": 26940
    },
    {
      "epoch": 0.9648145826991602,
      "grad_norm": 1.3269007205963135,
      "learning_rate": 6.479857563010062e-07,
      "loss": 1.2906,
      "step": 26941
    },
    {
      "epoch": 0.9648503948287285,
      "grad_norm": 1.288854956626892,
      "learning_rate": 6.466681251632522e-07,
      "loss": 1.3047,
      "step": 26942
    },
    {
      "epoch": 0.9648862069582967,
      "grad_norm": 1.3249443769454956,
      "learning_rate": 6.453518306964168e-07,
      "loss": 1.4381,
      "step": 26943
    },
    {
      "epoch": 0.964922019087865,
      "grad_norm": 1.3801206350326538,
      "learning_rate": 6.440368729182078e-07,
      "loss": 1.4704,
      "step": 26944
    },
    {
      "epoch": 0.9649578312174334,
      "grad_norm": 1.7521178722381592,
      "learning_rate": 6.427232518463333e-07,
      "loss": 1.3048,
      "step": 26945
    },
    {
      "epoch": 0.9649936433470017,
      "grad_norm": 1.3953019380569458,
      "learning_rate": 6.414109674984458e-07,
      "loss": 1.4156,
      "step": 26946
    },
    {
      "epoch": 0.9650294554765699,
      "grad_norm": 1.5526511669158936,
      "learning_rate": 6.401000198922202e-07,
      "loss": 1.5402,
      "step": 26947
    },
    {
      "epoch": 0.9650652676061382,
      "grad_norm": 1.5012935400009155,
      "learning_rate": 6.387904090452757e-07,
      "loss": 1.5017,
      "step": 26948
    },
    {
      "epoch": 0.9651010797357065,
      "grad_norm": 1.2707685232162476,
      "learning_rate": 6.374821349752424e-07,
      "loss": 1.1305,
      "step": 26949
    },
    {
      "epoch": 0.9651368918652747,
      "grad_norm": 1.609204649925232,
      "learning_rate": 6.361751976997177e-07,
      "loss": 1.5067,
      "step": 26950
    },
    {
      "epoch": 0.965172703994843,
      "grad_norm": 1.312595009803772,
      "learning_rate": 6.348695972362872e-07,
      "loss": 1.4279,
      "step": 26951
    },
    {
      "epoch": 0.9652085161244114,
      "grad_norm": 1.449548602104187,
      "learning_rate": 6.33565333602515e-07,
      "loss": 1.6088,
      "step": 26952
    },
    {
      "epoch": 0.9652443282539797,
      "grad_norm": 1.653937816619873,
      "learning_rate": 6.322624068159421e-07,
      "loss": 1.2204,
      "step": 26953
    },
    {
      "epoch": 0.9652801403835479,
      "grad_norm": 2.3628926277160645,
      "learning_rate": 6.309608168941217e-07,
      "loss": 1.6656,
      "step": 26954
    },
    {
      "epoch": 0.9653159525131162,
      "grad_norm": 1.80828857421875,
      "learning_rate": 6.296605638545172e-07,
      "loss": 1.3403,
      "step": 26955
    },
    {
      "epoch": 0.9653517646426845,
      "grad_norm": 1.901632308959961,
      "learning_rate": 6.283616477146703e-07,
      "loss": 1.1962,
      "step": 26956
    },
    {
      "epoch": 0.9653875767722527,
      "grad_norm": 1.6647350788116455,
      "learning_rate": 6.270640684920337e-07,
      "loss": 1.7878,
      "step": 26957
    },
    {
      "epoch": 0.965423388901821,
      "grad_norm": 1.296832799911499,
      "learning_rate": 6.257678262040712e-07,
      "loss": 1.3981,
      "step": 26958
    },
    {
      "epoch": 0.9654592010313894,
      "grad_norm": 1.8085285425186157,
      "learning_rate": 6.244729208682131e-07,
      "loss": 1.1554,
      "step": 26959
    },
    {
      "epoch": 0.9654950131609576,
      "grad_norm": 1.1620607376098633,
      "learning_rate": 6.231793525018903e-07,
      "loss": 1.0353,
      "step": 26960
    },
    {
      "epoch": 0.9655308252905259,
      "grad_norm": 1.9427391290664673,
      "learning_rate": 6.218871211224997e-07,
      "loss": 1.4748,
      "step": 26961
    },
    {
      "epoch": 0.9655666374200942,
      "grad_norm": 1.4789810180664062,
      "learning_rate": 6.205962267474386e-07,
      "loss": 1.1432,
      "step": 26962
    },
    {
      "epoch": 0.9656024495496625,
      "grad_norm": 1.6948950290679932,
      "learning_rate": 6.193066693940597e-07,
      "loss": 1.1085,
      "step": 26963
    },
    {
      "epoch": 0.9656382616792307,
      "grad_norm": 1.3911237716674805,
      "learning_rate": 6.180184490797158e-07,
      "loss": 1.3466,
      "step": 26964
    },
    {
      "epoch": 0.965674073808799,
      "grad_norm": 1.7155574560165405,
      "learning_rate": 6.167315658217376e-07,
      "loss": 1.575,
      "step": 26965
    },
    {
      "epoch": 0.9657098859383674,
      "grad_norm": 2.408755302429199,
      "learning_rate": 6.154460196374445e-07,
      "loss": 1.6934,
      "step": 26966
    },
    {
      "epoch": 0.9657456980679356,
      "grad_norm": 1.5642789602279663,
      "learning_rate": 6.141618105441227e-07,
      "loss": 1.4295,
      "step": 26967
    },
    {
      "epoch": 0.9657815101975039,
      "grad_norm": 1.8017868995666504,
      "learning_rate": 6.128789385590583e-07,
      "loss": 1.1765,
      "step": 26968
    },
    {
      "epoch": 0.9658173223270722,
      "grad_norm": 2.101372241973877,
      "learning_rate": 6.115974036995154e-07,
      "loss": 1.272,
      "step": 26969
    },
    {
      "epoch": 0.9658531344566404,
      "grad_norm": 1.5753008127212524,
      "learning_rate": 6.103172059827134e-07,
      "loss": 1.1554,
      "step": 26970
    },
    {
      "epoch": 0.9658889465862087,
      "grad_norm": 1.654207468032837,
      "learning_rate": 6.090383454259052e-07,
      "loss": 1.7734,
      "step": 26971
    },
    {
      "epoch": 0.965924758715777,
      "grad_norm": 2.283719539642334,
      "learning_rate": 6.077608220462771e-07,
      "loss": 0.987,
      "step": 26972
    },
    {
      "epoch": 0.9659605708453454,
      "grad_norm": 1.7849880456924438,
      "learning_rate": 6.064846358610154e-07,
      "loss": 1.5324,
      "step": 26973
    },
    {
      "epoch": 0.9659963829749136,
      "grad_norm": 2.171529769897461,
      "learning_rate": 6.052097868872953e-07,
      "loss": 1.4802,
      "step": 26974
    },
    {
      "epoch": 0.9660321951044819,
      "grad_norm": 2.1995201110839844,
      "learning_rate": 6.039362751422695e-07,
      "loss": 1.3182,
      "step": 26975
    },
    {
      "epoch": 0.9660680072340502,
      "grad_norm": 1.4685053825378418,
      "learning_rate": 6.026641006430689e-07,
      "loss": 1.4164,
      "step": 26976
    },
    {
      "epoch": 0.9661038193636184,
      "grad_norm": 1.7886309623718262,
      "learning_rate": 6.013932634068021e-07,
      "loss": 1.6115,
      "step": 26977
    },
    {
      "epoch": 0.9661396314931867,
      "grad_norm": 1.714840054512024,
      "learning_rate": 6.001237634505885e-07,
      "loss": 1.8374,
      "step": 26978
    },
    {
      "epoch": 0.966175443622755,
      "grad_norm": 1.6897437572479248,
      "learning_rate": 5.988556007914814e-07,
      "loss": 1.9828,
      "step": 26979
    },
    {
      "epoch": 0.9662112557523234,
      "grad_norm": 1.6048061847686768,
      "learning_rate": 5.975887754465559e-07,
      "loss": 1.5878,
      "step": 26980
    },
    {
      "epoch": 0.9662470678818916,
      "grad_norm": 1.753430724143982,
      "learning_rate": 5.96323287432854e-07,
      "loss": 1.4545,
      "step": 26981
    },
    {
      "epoch": 0.9662828800114599,
      "grad_norm": 1.6798697710037231,
      "learning_rate": 5.950591367674064e-07,
      "loss": 1.2394,
      "step": 26982
    },
    {
      "epoch": 0.9663186921410282,
      "grad_norm": 2.274351119995117,
      "learning_rate": 5.937963234672106e-07,
      "loss": 1.3658,
      "step": 26983
    },
    {
      "epoch": 0.9663545042705964,
      "grad_norm": 1.8976236581802368,
      "learning_rate": 5.925348475492643e-07,
      "loss": 1.0295,
      "step": 26984
    },
    {
      "epoch": 0.9663903164001647,
      "grad_norm": 1.966462254524231,
      "learning_rate": 5.912747090305315e-07,
      "loss": 1.659,
      "step": 26985
    },
    {
      "epoch": 0.966426128529733,
      "grad_norm": 1.4895193576812744,
      "learning_rate": 5.900159079279654e-07,
      "loss": 1.377,
      "step": 26986
    },
    {
      "epoch": 0.9664619406593014,
      "grad_norm": 1.6803325414657593,
      "learning_rate": 5.887584442585081e-07,
      "loss": 1.2961,
      "step": 26987
    },
    {
      "epoch": 0.9664977527888696,
      "grad_norm": 1.4784119129180908,
      "learning_rate": 5.875023180390793e-07,
      "loss": 1.6773,
      "step": 26988
    },
    {
      "epoch": 0.9665335649184379,
      "grad_norm": 1.7596663236618042,
      "learning_rate": 5.862475292865655e-07,
      "loss": 1.3413,
      "step": 26989
    },
    {
      "epoch": 0.9665693770480062,
      "grad_norm": 2.026500940322876,
      "learning_rate": 5.849940780178642e-07,
      "loss": 1.1128,
      "step": 26990
    },
    {
      "epoch": 0.9666051891775744,
      "grad_norm": 1.9572885036468506,
      "learning_rate": 5.837419642498288e-07,
      "loss": 1.7062,
      "step": 26991
    },
    {
      "epoch": 0.9666410013071427,
      "grad_norm": 1.2944083213806152,
      "learning_rate": 5.824911879993123e-07,
      "loss": 1.2943,
      "step": 26992
    },
    {
      "epoch": 0.966676813436711,
      "grad_norm": 2.0377883911132812,
      "learning_rate": 5.812417492831346e-07,
      "loss": 1.4307,
      "step": 26993
    },
    {
      "epoch": 0.9667126255662793,
      "grad_norm": 1.5536547899246216,
      "learning_rate": 5.799936481181045e-07,
      "loss": 1.448,
      "step": 26994
    },
    {
      "epoch": 0.9667484376958476,
      "grad_norm": 2.0070462226867676,
      "learning_rate": 5.787468845210198e-07,
      "loss": 1.2658,
      "step": 26995
    },
    {
      "epoch": 0.9667842498254159,
      "grad_norm": 1.606011986732483,
      "learning_rate": 5.775014585086446e-07,
      "loss": 1.7443,
      "step": 26996
    },
    {
      "epoch": 0.9668200619549842,
      "grad_norm": 2.032684326171875,
      "learning_rate": 5.762573700977547e-07,
      "loss": 1.5145,
      "step": 26997
    },
    {
      "epoch": 0.9668558740845524,
      "grad_norm": 1.5862538814544678,
      "learning_rate": 5.750146193050698e-07,
      "loss": 1.3515,
      "step": 26998
    },
    {
      "epoch": 0.9668916862141207,
      "grad_norm": 1.840734839439392,
      "learning_rate": 5.7377320614731e-07,
      "loss": 1.6445,
      "step": 26999
    },
    {
      "epoch": 0.966927498343689,
      "grad_norm": 2.2183516025543213,
      "learning_rate": 5.725331306411841e-07,
      "loss": 1.6061,
      "step": 27000
    },
    {
      "epoch": 0.9669633104732573,
      "grad_norm": 1.8486427068710327,
      "learning_rate": 5.712943928033787e-07,
      "loss": 1.2393,
      "step": 27001
    },
    {
      "epoch": 0.9669991226028256,
      "grad_norm": 1.6504371166229248,
      "learning_rate": 5.700569926505361e-07,
      "loss": 1.1869,
      "step": 27002
    },
    {
      "epoch": 0.9670349347323939,
      "grad_norm": 1.9643383026123047,
      "learning_rate": 5.688209301993319e-07,
      "loss": 1.5403,
      "step": 27003
    },
    {
      "epoch": 0.9670707468619621,
      "grad_norm": 1.8288096189498901,
      "learning_rate": 5.675862054663861e-07,
      "loss": 1.4967,
      "step": 27004
    },
    {
      "epoch": 0.9671065589915304,
      "grad_norm": 2.0646116733551025,
      "learning_rate": 5.663528184683186e-07,
      "loss": 1.5351,
      "step": 27005
    },
    {
      "epoch": 0.9671423711210987,
      "grad_norm": 2.1061975955963135,
      "learning_rate": 5.651207692216942e-07,
      "loss": 1.564,
      "step": 27006
    },
    {
      "epoch": 0.967178183250667,
      "grad_norm": 1.4611119031906128,
      "learning_rate": 5.638900577431216e-07,
      "loss": 1.3468,
      "step": 27007
    },
    {
      "epoch": 0.9672139953802353,
      "grad_norm": 1.4998737573623657,
      "learning_rate": 5.626606840491433e-07,
      "loss": 1.6063,
      "step": 27008
    },
    {
      "epoch": 0.9672498075098036,
      "grad_norm": 1.3977317810058594,
      "learning_rate": 5.614326481562904e-07,
      "loss": 1.4851,
      "step": 27009
    },
    {
      "epoch": 0.9672856196393719,
      "grad_norm": 1.8437548875808716,
      "learning_rate": 5.602059500811052e-07,
      "loss": 1.705,
      "step": 27010
    },
    {
      "epoch": 0.9673214317689401,
      "grad_norm": 1.445892572402954,
      "learning_rate": 5.589805898400746e-07,
      "loss": 1.3189,
      "step": 27011
    },
    {
      "epoch": 0.9673572438985084,
      "grad_norm": 1.510178804397583,
      "learning_rate": 5.577565674496965e-07,
      "loss": 1.3705,
      "step": 27012
    },
    {
      "epoch": 0.9673930560280767,
      "grad_norm": 1.8084083795547485,
      "learning_rate": 5.565338829264355e-07,
      "loss": 1.4164,
      "step": 27013
    },
    {
      "epoch": 0.967428868157645,
      "grad_norm": 1.2820135354995728,
      "learning_rate": 5.553125362867228e-07,
      "loss": 1.2558,
      "step": 27014
    },
    {
      "epoch": 0.9674646802872133,
      "grad_norm": 1.898093581199646,
      "learning_rate": 5.540925275470232e-07,
      "loss": 1.796,
      "step": 27015
    },
    {
      "epoch": 0.9675004924167816,
      "grad_norm": 1.603390097618103,
      "learning_rate": 5.528738567237235e-07,
      "loss": 1.697,
      "step": 27016
    },
    {
      "epoch": 0.9675363045463499,
      "grad_norm": 1.477070927619934,
      "learning_rate": 5.516565238332328e-07,
      "loss": 1.2186,
      "step": 27017
    },
    {
      "epoch": 0.9675721166759181,
      "grad_norm": 1.664985179901123,
      "learning_rate": 5.504405288919156e-07,
      "loss": 1.1862,
      "step": 27018
    },
    {
      "epoch": 0.9676079288054864,
      "grad_norm": 1.6416716575622559,
      "learning_rate": 5.492258719161481e-07,
      "loss": 1.4381,
      "step": 27019
    },
    {
      "epoch": 0.9676437409350547,
      "grad_norm": 1.1065994501113892,
      "learning_rate": 5.480125529222613e-07,
      "loss": 1.0456,
      "step": 27020
    },
    {
      "epoch": 0.9676795530646229,
      "grad_norm": 1.6238538026809692,
      "learning_rate": 5.468005719265868e-07,
      "loss": 1.6424,
      "step": 27021
    },
    {
      "epoch": 0.9677153651941913,
      "grad_norm": 1.3962935209274292,
      "learning_rate": 5.455899289454225e-07,
      "loss": 1.6138,
      "step": 27022
    },
    {
      "epoch": 0.9677511773237596,
      "grad_norm": 1.931966781616211,
      "learning_rate": 5.443806239950555e-07,
      "loss": 1.2321,
      "step": 27023
    },
    {
      "epoch": 0.9677869894533279,
      "grad_norm": 1.9834296703338623,
      "learning_rate": 5.431726570917617e-07,
      "loss": 1.26,
      "step": 27024
    },
    {
      "epoch": 0.9678228015828961,
      "grad_norm": 1.65738844871521,
      "learning_rate": 5.419660282517836e-07,
      "loss": 1.153,
      "step": 27025
    },
    {
      "epoch": 0.9678586137124644,
      "grad_norm": 1.481104850769043,
      "learning_rate": 5.407607374913748e-07,
      "loss": 1.4851,
      "step": 27026
    },
    {
      "epoch": 0.9678944258420327,
      "grad_norm": 1.8398284912109375,
      "learning_rate": 5.395567848267225e-07,
      "loss": 1.6348,
      "step": 27027
    },
    {
      "epoch": 0.9679302379716009,
      "grad_norm": 1.63754141330719,
      "learning_rate": 5.383541702740469e-07,
      "loss": 1.5278,
      "step": 27028
    },
    {
      "epoch": 0.9679660501011693,
      "grad_norm": 1.5657932758331299,
      "learning_rate": 5.37152893849513e-07,
      "loss": 1.4017,
      "step": 27029
    },
    {
      "epoch": 0.9680018622307376,
      "grad_norm": 1.9750888347625732,
      "learning_rate": 5.359529555692966e-07,
      "loss": 1.5823,
      "step": 27030
    },
    {
      "epoch": 0.9680376743603059,
      "grad_norm": 1.7874226570129395,
      "learning_rate": 5.347543554495293e-07,
      "loss": 1.4156,
      "step": 27031
    },
    {
      "epoch": 0.9680734864898741,
      "grad_norm": 2.025090456008911,
      "learning_rate": 5.335570935063427e-07,
      "loss": 1.6055,
      "step": 27032
    },
    {
      "epoch": 0.9681092986194424,
      "grad_norm": 1.950339913368225,
      "learning_rate": 5.323611697558462e-07,
      "loss": 1.2355,
      "step": 27033
    },
    {
      "epoch": 0.9681451107490107,
      "grad_norm": 1.6537014245986938,
      "learning_rate": 5.311665842141155e-07,
      "loss": 1.3523,
      "step": 27034
    },
    {
      "epoch": 0.9681809228785789,
      "grad_norm": 1.566712498664856,
      "learning_rate": 5.299733368972492e-07,
      "loss": 1.5859,
      "step": 27035
    },
    {
      "epoch": 0.9682167350081473,
      "grad_norm": 2.611342668533325,
      "learning_rate": 5.287814278212677e-07,
      "loss": 1.6179,
      "step": 27036
    },
    {
      "epoch": 0.9682525471377156,
      "grad_norm": 1.9200516939163208,
      "learning_rate": 5.275908570022359e-07,
      "loss": 1.6782,
      "step": 27037
    },
    {
      "epoch": 0.9682883592672838,
      "grad_norm": 1.4160640239715576,
      "learning_rate": 5.26401624456152e-07,
      "loss": 1.287,
      "step": 27038
    },
    {
      "epoch": 0.9683241713968521,
      "grad_norm": 1.3941360712051392,
      "learning_rate": 5.252137301990256e-07,
      "loss": 1.3371,
      "step": 27039
    },
    {
      "epoch": 0.9683599835264204,
      "grad_norm": 3.1313459873199463,
      "learning_rate": 5.240271742468328e-07,
      "loss": 1.3915,
      "step": 27040
    },
    {
      "epoch": 0.9683957956559887,
      "grad_norm": 2.499171018600464,
      "learning_rate": 5.228419566155385e-07,
      "loss": 1.3878,
      "step": 27041
    },
    {
      "epoch": 0.9684316077855569,
      "grad_norm": 1.5517430305480957,
      "learning_rate": 5.216580773210966e-07,
      "loss": 1.3956,
      "step": 27042
    },
    {
      "epoch": 0.9684674199151253,
      "grad_norm": 1.5848878622055054,
      "learning_rate": 5.204755363794167e-07,
      "loss": 1.3835,
      "step": 27043
    },
    {
      "epoch": 0.9685032320446936,
      "grad_norm": 1.4164799451828003,
      "learning_rate": 5.192943338064305e-07,
      "loss": 1.5956,
      "step": 27044
    },
    {
      "epoch": 0.9685390441742618,
      "grad_norm": 1.7449172735214233,
      "learning_rate": 5.18114469618014e-07,
      "loss": 1.7308,
      "step": 27045
    },
    {
      "epoch": 0.9685748563038301,
      "grad_norm": 1.5101077556610107,
      "learning_rate": 5.169359438300436e-07,
      "loss": 1.1332,
      "step": 27046
    },
    {
      "epoch": 0.9686106684333984,
      "grad_norm": 1.4586726427078247,
      "learning_rate": 5.157587564583733e-07,
      "loss": 1.5053,
      "step": 27047
    },
    {
      "epoch": 0.9686464805629666,
      "grad_norm": 1.5485001802444458,
      "learning_rate": 5.145829075188457e-07,
      "loss": 1.1668,
      "step": 27048
    },
    {
      "epoch": 0.9686822926925349,
      "grad_norm": 1.5960198640823364,
      "learning_rate": 5.134083970272819e-07,
      "loss": 1.6091,
      "step": 27049
    },
    {
      "epoch": 0.9687181048221033,
      "grad_norm": 1.9195680618286133,
      "learning_rate": 5.1223522499948e-07,
      "loss": 1.3194,
      "step": 27050
    },
    {
      "epoch": 0.9687539169516716,
      "grad_norm": 1.7722783088684082,
      "learning_rate": 5.110633914512164e-07,
      "loss": 1.5024,
      "step": 27051
    },
    {
      "epoch": 0.9687897290812398,
      "grad_norm": 1.3769675493240356,
      "learning_rate": 5.098928963982674e-07,
      "loss": 1.5481,
      "step": 27052
    },
    {
      "epoch": 0.9688255412108081,
      "grad_norm": 1.214339256286621,
      "learning_rate": 5.08723739856376e-07,
      "loss": 1.1588,
      "step": 27053
    },
    {
      "epoch": 0.9688613533403764,
      "grad_norm": 1.6598423719406128,
      "learning_rate": 5.07555921841274e-07,
      "loss": 1.4385,
      "step": 27054
    },
    {
      "epoch": 0.9688971654699446,
      "grad_norm": 1.1218611001968384,
      "learning_rate": 5.06389442368671e-07,
      "loss": 1.3616,
      "step": 27055
    },
    {
      "epoch": 0.9689329775995129,
      "grad_norm": 1.3781205415725708,
      "learning_rate": 5.052243014542546e-07,
      "loss": 1.5814,
      "step": 27056
    },
    {
      "epoch": 0.9689687897290813,
      "grad_norm": 1.7898406982421875,
      "learning_rate": 5.040604991137121e-07,
      "loss": 1.1487,
      "step": 27057
    },
    {
      "epoch": 0.9690046018586496,
      "grad_norm": 1.6360304355621338,
      "learning_rate": 5.028980353626866e-07,
      "loss": 1.4937,
      "step": 27058
    },
    {
      "epoch": 0.9690404139882178,
      "grad_norm": 1.2866710424423218,
      "learning_rate": 5.017369102168435e-07,
      "loss": 1.2457,
      "step": 27059
    },
    {
      "epoch": 0.9690762261177861,
      "grad_norm": 1.3128433227539062,
      "learning_rate": 5.005771236917811e-07,
      "loss": 1.4053,
      "step": 27060
    },
    {
      "epoch": 0.9691120382473544,
      "grad_norm": 1.6207622289657593,
      "learning_rate": 4.994186758030983e-07,
      "loss": 1.5096,
      "step": 27061
    },
    {
      "epoch": 0.9691478503769226,
      "grad_norm": 1.527236819267273,
      "learning_rate": 4.982615665663937e-07,
      "loss": 1.4559,
      "step": 27062
    },
    {
      "epoch": 0.9691836625064909,
      "grad_norm": 1.528676152229309,
      "learning_rate": 4.971057959972325e-07,
      "loss": 1.1819,
      "step": 27063
    },
    {
      "epoch": 0.9692194746360593,
      "grad_norm": 1.549570083618164,
      "learning_rate": 4.95951364111169e-07,
      "loss": 1.344,
      "step": 27064
    },
    {
      "epoch": 0.9692552867656276,
      "grad_norm": 2.239349603652954,
      "learning_rate": 4.947982709237131e-07,
      "loss": 1.349,
      "step": 27065
    },
    {
      "epoch": 0.9692910988951958,
      "grad_norm": 1.8288278579711914,
      "learning_rate": 4.936465164504079e-07,
      "loss": 1.8213,
      "step": 27066
    },
    {
      "epoch": 0.9693269110247641,
      "grad_norm": 1.5280758142471313,
      "learning_rate": 4.924961007067408e-07,
      "loss": 1.7326,
      "step": 27067
    },
    {
      "epoch": 0.9693627231543324,
      "grad_norm": 1.4166117906570435,
      "learning_rate": 4.913470237081774e-07,
      "loss": 1.5967,
      "step": 27068
    },
    {
      "epoch": 0.9693985352839006,
      "grad_norm": 1.780877947807312,
      "learning_rate": 4.90199285470172e-07,
      "loss": 1.6305,
      "step": 27069
    },
    {
      "epoch": 0.9694343474134689,
      "grad_norm": 1.342950463294983,
      "learning_rate": 4.8905288600819e-07,
      "loss": 1.2657,
      "step": 27070
    },
    {
      "epoch": 0.9694701595430373,
      "grad_norm": 1.7711330652236938,
      "learning_rate": 4.879078253376412e-07,
      "loss": 1.4819,
      "step": 27071
    },
    {
      "epoch": 0.9695059716726055,
      "grad_norm": 1.4958521127700806,
      "learning_rate": 4.867641034739134e-07,
      "loss": 1.4194,
      "step": 27072
    },
    {
      "epoch": 0.9695417838021738,
      "grad_norm": 1.9198914766311646,
      "learning_rate": 4.856217204324275e-07,
      "loss": 1.6569,
      "step": 27073
    },
    {
      "epoch": 0.9695775959317421,
      "grad_norm": 1.47645902633667,
      "learning_rate": 4.844806762285381e-07,
      "loss": 1.2356,
      "step": 27074
    },
    {
      "epoch": 0.9696134080613104,
      "grad_norm": 1.583561658859253,
      "learning_rate": 4.833409708775882e-07,
      "loss": 1.4539,
      "step": 27075
    },
    {
      "epoch": 0.9696492201908786,
      "grad_norm": 1.3329954147338867,
      "learning_rate": 4.822026043949213e-07,
      "loss": 1.4844,
      "step": 27076
    },
    {
      "epoch": 0.9696850323204469,
      "grad_norm": 1.3243921995162964,
      "learning_rate": 4.810655767958583e-07,
      "loss": 1.1535,
      "step": 27077
    },
    {
      "epoch": 0.9697208444500153,
      "grad_norm": 1.422972559928894,
      "learning_rate": 4.799298880956759e-07,
      "loss": 1.4926,
      "step": 27078
    },
    {
      "epoch": 0.9697566565795835,
      "grad_norm": 1.622788667678833,
      "learning_rate": 4.787955383096731e-07,
      "loss": 1.1752,
      "step": 27079
    },
    {
      "epoch": 0.9697924687091518,
      "grad_norm": 1.558379888534546,
      "learning_rate": 4.776625274530933e-07,
      "loss": 1.1288,
      "step": 27080
    },
    {
      "epoch": 0.9698282808387201,
      "grad_norm": 1.6241610050201416,
      "learning_rate": 4.765308555411907e-07,
      "loss": 1.1744,
      "step": 27081
    },
    {
      "epoch": 0.9698640929682883,
      "grad_norm": 1.4816008806228638,
      "learning_rate": 4.754005225891978e-07,
      "loss": 1.344,
      "step": 27082
    },
    {
      "epoch": 0.9698999050978566,
      "grad_norm": 1.7448887825012207,
      "learning_rate": 4.742715286123134e-07,
      "loss": 1.3445,
      "step": 27083
    },
    {
      "epoch": 0.9699357172274249,
      "grad_norm": 1.9753953218460083,
      "learning_rate": 4.7314387362572545e-07,
      "loss": 1.3754,
      "step": 27084
    },
    {
      "epoch": 0.9699715293569933,
      "grad_norm": 1.8347922563552856,
      "learning_rate": 4.7201755764459953e-07,
      "loss": 1.4986,
      "step": 27085
    },
    {
      "epoch": 0.9700073414865615,
      "grad_norm": 1.6232060194015503,
      "learning_rate": 4.7089258068410133e-07,
      "loss": 1.4232,
      "step": 27086
    },
    {
      "epoch": 0.9700431536161298,
      "grad_norm": 2.0615720748901367,
      "learning_rate": 4.6976894275935215e-07,
      "loss": 1.9353,
      "step": 27087
    },
    {
      "epoch": 0.9700789657456981,
      "grad_norm": 1.7156115770339966,
      "learning_rate": 4.6864664388548417e-07,
      "loss": 1.4939,
      "step": 27088
    },
    {
      "epoch": 0.9701147778752663,
      "grad_norm": 1.5696758031845093,
      "learning_rate": 4.6752568407759655e-07,
      "loss": 1.2644,
      "step": 27089
    },
    {
      "epoch": 0.9701505900048346,
      "grad_norm": 1.4447704553604126,
      "learning_rate": 4.66406063350755e-07,
      "loss": 1.4013,
      "step": 27090
    },
    {
      "epoch": 0.9701864021344029,
      "grad_norm": 1.5259318351745605,
      "learning_rate": 4.652877817200252e-07,
      "loss": 1.5891,
      "step": 27091
    },
    {
      "epoch": 0.9702222142639713,
      "grad_norm": 1.3538051843643188,
      "learning_rate": 4.6417083920046176e-07,
      "loss": 1.4276,
      "step": 27092
    },
    {
      "epoch": 0.9702580263935395,
      "grad_norm": 1.2154916524887085,
      "learning_rate": 4.630552358070972e-07,
      "loss": 1.2274,
      "step": 27093
    },
    {
      "epoch": 0.9702938385231078,
      "grad_norm": 1.5083569288253784,
      "learning_rate": 4.6194097155491944e-07,
      "loss": 1.7358,
      "step": 27094
    },
    {
      "epoch": 0.9703296506526761,
      "grad_norm": 1.5696214437484741,
      "learning_rate": 4.6082804645893874e-07,
      "loss": 1.4356,
      "step": 27095
    },
    {
      "epoch": 0.9703654627822443,
      "grad_norm": 1.5140817165374756,
      "learning_rate": 4.597164605341209e-07,
      "loss": 1.3379,
      "step": 27096
    },
    {
      "epoch": 0.9704012749118126,
      "grad_norm": 1.9681334495544434,
      "learning_rate": 4.5860621379540944e-07,
      "loss": 1.3589,
      "step": 27097
    },
    {
      "epoch": 0.9704370870413809,
      "grad_norm": 1.6079471111297607,
      "learning_rate": 4.574973062577592e-07,
      "loss": 1.2374,
      "step": 27098
    },
    {
      "epoch": 0.9704728991709493,
      "grad_norm": 1.9617375135421753,
      "learning_rate": 4.5638973793608043e-07,
      "loss": 1.429,
      "step": 27099
    },
    {
      "epoch": 0.9705087113005175,
      "grad_norm": 1.4626727104187012,
      "learning_rate": 4.5528350884528335e-07,
      "loss": 1.5835,
      "step": 27100
    },
    {
      "epoch": 0.9705445234300858,
      "grad_norm": 1.6523990631103516,
      "learning_rate": 4.5417861900023397e-07,
      "loss": 1.3469,
      "step": 27101
    },
    {
      "epoch": 0.9705803355596541,
      "grad_norm": 1.3021047115325928,
      "learning_rate": 4.5307506841580914e-07,
      "loss": 1.2336,
      "step": 27102
    },
    {
      "epoch": 0.9706161476892223,
      "grad_norm": 1.75713050365448,
      "learning_rate": 4.5197285710685265e-07,
      "loss": 1.8113,
      "step": 27103
    },
    {
      "epoch": 0.9706519598187906,
      "grad_norm": 1.3806419372558594,
      "learning_rate": 4.5087198508819705e-07,
      "loss": 1.3937,
      "step": 27104
    },
    {
      "epoch": 0.9706877719483589,
      "grad_norm": 1.4847077131271362,
      "learning_rate": 4.497724523746416e-07,
      "loss": 1.3979,
      "step": 27105
    },
    {
      "epoch": 0.9707235840779272,
      "grad_norm": 1.5948593616485596,
      "learning_rate": 4.486742589809967e-07,
      "loss": 1.5002,
      "step": 27106
    },
    {
      "epoch": 0.9707593962074955,
      "grad_norm": 1.5786889791488647,
      "learning_rate": 4.4757740492201717e-07,
      "loss": 1.5142,
      "step": 27107
    },
    {
      "epoch": 0.9707952083370638,
      "grad_norm": 2.1769702434539795,
      "learning_rate": 4.464818902124801e-07,
      "loss": 1.2433,
      "step": 27108
    },
    {
      "epoch": 0.970831020466632,
      "grad_norm": 1.6344908475875854,
      "learning_rate": 4.4538771486710706e-07,
      "loss": 1.8311,
      "step": 27109
    },
    {
      "epoch": 0.9708668325962003,
      "grad_norm": 2.086514472961426,
      "learning_rate": 4.442948789006307e-07,
      "loss": 1.4124,
      "step": 27110
    },
    {
      "epoch": 0.9709026447257686,
      "grad_norm": 2.050281286239624,
      "learning_rate": 4.432033823277504e-07,
      "loss": 1.2815,
      "step": 27111
    },
    {
      "epoch": 0.9709384568553369,
      "grad_norm": 2.2314565181732178,
      "learning_rate": 4.4211322516314324e-07,
      "loss": 1.7093,
      "step": 27112
    },
    {
      "epoch": 0.9709742689849052,
      "grad_norm": 1.5358484983444214,
      "learning_rate": 4.410244074214864e-07,
      "loss": 1.4834,
      "step": 27113
    },
    {
      "epoch": 0.9710100811144735,
      "grad_norm": 1.5115602016448975,
      "learning_rate": 4.399369291174349e-07,
      "loss": 1.2563,
      "step": 27114
    },
    {
      "epoch": 0.9710458932440418,
      "grad_norm": 2.1405441761016846,
      "learning_rate": 4.388507902655881e-07,
      "loss": 1.3515,
      "step": 27115
    },
    {
      "epoch": 0.97108170537361,
      "grad_norm": 1.8947888612747192,
      "learning_rate": 4.377659908805898e-07,
      "loss": 1.5495,
      "step": 27116
    },
    {
      "epoch": 0.9711175175031783,
      "grad_norm": 1.3705615997314453,
      "learning_rate": 4.366825309770284e-07,
      "loss": 1.0637,
      "step": 27117
    },
    {
      "epoch": 0.9711533296327466,
      "grad_norm": 2.133995532989502,
      "learning_rate": 4.3560041056947e-07,
      "loss": 1.4022,
      "step": 27118
    },
    {
      "epoch": 0.9711891417623149,
      "grad_norm": 2.3694558143615723,
      "learning_rate": 4.345196296724807e-07,
      "loss": 1.4344,
      "step": 27119
    },
    {
      "epoch": 0.9712249538918832,
      "grad_norm": 1.50997793674469,
      "learning_rate": 4.334401883005934e-07,
      "loss": 1.4774,
      "step": 27120
    },
    {
      "epoch": 0.9712607660214515,
      "grad_norm": 2.032027244567871,
      "learning_rate": 4.32362086468352e-07,
      "loss": 1.5578,
      "step": 27121
    },
    {
      "epoch": 0.9712965781510198,
      "grad_norm": 1.652375340461731,
      "learning_rate": 4.312853241902337e-07,
      "loss": 1.4457,
      "step": 27122
    },
    {
      "epoch": 0.971332390280588,
      "grad_norm": 1.819873332977295,
      "learning_rate": 4.3020990148073815e-07,
      "loss": 1.6141,
      "step": 27123
    },
    {
      "epoch": 0.9713682024101563,
      "grad_norm": 1.587869644165039,
      "learning_rate": 4.2913581835433147e-07,
      "loss": 1.6089,
      "step": 27124
    },
    {
      "epoch": 0.9714040145397246,
      "grad_norm": 1.382434368133545,
      "learning_rate": 4.2806307482546883e-07,
      "loss": 1.5386,
      "step": 27125
    },
    {
      "epoch": 0.9714398266692928,
      "grad_norm": 1.4083807468414307,
      "learning_rate": 4.2699167090858303e-07,
      "loss": 1.2496,
      "step": 27126
    },
    {
      "epoch": 0.9714756387988612,
      "grad_norm": 1.9262927770614624,
      "learning_rate": 4.259216066180738e-07,
      "loss": 1.4704,
      "step": 27127
    },
    {
      "epoch": 0.9715114509284295,
      "grad_norm": 1.735276460647583,
      "learning_rate": 4.248528819683517e-07,
      "loss": 1.3874,
      "step": 27128
    },
    {
      "epoch": 0.9715472630579978,
      "grad_norm": 1.3945090770721436,
      "learning_rate": 4.2378549697380533e-07,
      "loss": 1.4935,
      "step": 27129
    },
    {
      "epoch": 0.971583075187566,
      "grad_norm": 1.7358595132827759,
      "learning_rate": 4.2271945164876756e-07,
      "loss": 1.4826,
      "step": 27130
    },
    {
      "epoch": 0.9716188873171343,
      "grad_norm": 1.6000685691833496,
      "learning_rate": 4.216547460075937e-07,
      "loss": 1.4768,
      "step": 27131
    },
    {
      "epoch": 0.9716546994467026,
      "grad_norm": 1.8158823251724243,
      "learning_rate": 4.2059138006460554e-07,
      "loss": 1.4642,
      "step": 27132
    },
    {
      "epoch": 0.9716905115762708,
      "grad_norm": 1.5935328006744385,
      "learning_rate": 4.1952935383412494e-07,
      "loss": 1.3549,
      "step": 27133
    },
    {
      "epoch": 0.9717263237058391,
      "grad_norm": 1.1625258922576904,
      "learning_rate": 4.1846866733041834e-07,
      "loss": 1.3621,
      "step": 27134
    },
    {
      "epoch": 0.9717621358354075,
      "grad_norm": 1.8553330898284912,
      "learning_rate": 4.1740932056776317e-07,
      "loss": 1.5953,
      "step": 27135
    },
    {
      "epoch": 0.9717979479649758,
      "grad_norm": 1.6651084423065186,
      "learning_rate": 4.1635131356041467e-07,
      "loss": 1.6272,
      "step": 27136
    },
    {
      "epoch": 0.971833760094544,
      "grad_norm": 1.5465548038482666,
      "learning_rate": 4.1529464632260597e-07,
      "loss": 1.3804,
      "step": 27137
    },
    {
      "epoch": 0.9718695722241123,
      "grad_norm": 1.316346526145935,
      "learning_rate": 4.1423931886854785e-07,
      "loss": 1.5989,
      "step": 27138
    },
    {
      "epoch": 0.9719053843536806,
      "grad_norm": 1.3082847595214844,
      "learning_rate": 4.131853312124512e-07,
      "loss": 1.5898,
      "step": 27139
    },
    {
      "epoch": 0.9719411964832488,
      "grad_norm": 1.3995599746704102,
      "learning_rate": 4.1213268336849355e-07,
      "loss": 1.4302,
      "step": 27140
    },
    {
      "epoch": 0.9719770086128171,
      "grad_norm": 1.8354229927062988,
      "learning_rate": 4.1108137535081914e-07,
      "loss": 1.3678,
      "step": 27141
    },
    {
      "epoch": 0.9720128207423855,
      "grad_norm": 1.4459917545318604,
      "learning_rate": 4.1003140717358336e-07,
      "loss": 1.4699,
      "step": 27142
    },
    {
      "epoch": 0.9720486328719538,
      "grad_norm": 1.557350993156433,
      "learning_rate": 4.089827788509304e-07,
      "loss": 1.2917,
      "step": 27143
    },
    {
      "epoch": 0.972084445001522,
      "grad_norm": 1.7156989574432373,
      "learning_rate": 4.079354903969379e-07,
      "loss": 1.4024,
      "step": 27144
    },
    {
      "epoch": 0.9721202571310903,
      "grad_norm": 2.2056097984313965,
      "learning_rate": 4.068895418257057e-07,
      "loss": 1.2329,
      "step": 27145
    },
    {
      "epoch": 0.9721560692606586,
      "grad_norm": 1.616719126701355,
      "learning_rate": 4.0584493315131146e-07,
      "loss": 1.4484,
      "step": 27146
    },
    {
      "epoch": 0.9721918813902268,
      "grad_norm": 1.549475908279419,
      "learning_rate": 4.048016643878105e-07,
      "loss": 1.6468,
      "step": 27147
    },
    {
      "epoch": 0.9722276935197951,
      "grad_norm": 1.5446029901504517,
      "learning_rate": 4.037597355492362e-07,
      "loss": 1.5179,
      "step": 27148
    },
    {
      "epoch": 0.9722635056493635,
      "grad_norm": 1.9477657079696655,
      "learning_rate": 4.0271914664959944e-07,
      "loss": 1.5566,
      "step": 27149
    },
    {
      "epoch": 0.9722993177789317,
      "grad_norm": 1.5786957740783691,
      "learning_rate": 4.016798977029113e-07,
      "loss": 1.2298,
      "step": 27150
    },
    {
      "epoch": 0.9723351299085,
      "grad_norm": 1.673638939857483,
      "learning_rate": 4.006419887231383e-07,
      "loss": 1.5531,
      "step": 27151
    },
    {
      "epoch": 0.9723709420380683,
      "grad_norm": 1.7010632753372192,
      "learning_rate": 3.9960541972426936e-07,
      "loss": 1.5984,
      "step": 27152
    },
    {
      "epoch": 0.9724067541676366,
      "grad_norm": 1.7504063844680786,
      "learning_rate": 3.985701907202155e-07,
      "loss": 1.3279,
      "step": 27153
    },
    {
      "epoch": 0.9724425662972048,
      "grad_norm": 1.5936665534973145,
      "learning_rate": 3.975363017249323e-07,
      "loss": 1.2734,
      "step": 27154
    },
    {
      "epoch": 0.9724783784267731,
      "grad_norm": 1.544960856437683,
      "learning_rate": 3.9650375275231967e-07,
      "loss": 1.4678,
      "step": 27155
    },
    {
      "epoch": 0.9725141905563415,
      "grad_norm": 1.6201308965682983,
      "learning_rate": 3.9547254381626653e-07,
      "loss": 1.3237,
      "step": 27156
    },
    {
      "epoch": 0.9725500026859097,
      "grad_norm": 1.5041776895523071,
      "learning_rate": 3.944426749306507e-07,
      "loss": 1.4167,
      "step": 27157
    },
    {
      "epoch": 0.972585814815478,
      "grad_norm": 1.9468683004379272,
      "learning_rate": 3.934141461093277e-07,
      "loss": 1.2188,
      "step": 27158
    },
    {
      "epoch": 0.9726216269450463,
      "grad_norm": 1.457405686378479,
      "learning_rate": 3.9238695736614207e-07,
      "loss": 1.3617,
      "step": 27159
    },
    {
      "epoch": 0.9726574390746145,
      "grad_norm": 1.630591630935669,
      "learning_rate": 3.913611087148938e-07,
      "loss": 1.438,
      "step": 27160
    },
    {
      "epoch": 0.9726932512041828,
      "grad_norm": 1.6744720935821533,
      "learning_rate": 3.903366001694053e-07,
      "loss": 1.4143,
      "step": 27161
    },
    {
      "epoch": 0.9727290633337511,
      "grad_norm": 1.620065689086914,
      "learning_rate": 3.8931343174344324e-07,
      "loss": 1.2351,
      "step": 27162
    },
    {
      "epoch": 0.9727648754633195,
      "grad_norm": 1.7324904203414917,
      "learning_rate": 3.882916034507855e-07,
      "loss": 1.212,
      "step": 27163
    },
    {
      "epoch": 0.9728006875928877,
      "grad_norm": 1.5588440895080566,
      "learning_rate": 3.8727111530516556e-07,
      "loss": 1.4532,
      "step": 27164
    },
    {
      "epoch": 0.972836499722456,
      "grad_norm": 1.5138558149337769,
      "learning_rate": 3.86251967320328e-07,
      "loss": 1.4652,
      "step": 27165
    },
    {
      "epoch": 0.9728723118520243,
      "grad_norm": 1.6585930585861206,
      "learning_rate": 3.85234159509984e-07,
      "loss": 1.5071,
      "step": 27166
    },
    {
      "epoch": 0.9729081239815925,
      "grad_norm": 1.572977900505066,
      "learning_rate": 3.842176918878115e-07,
      "loss": 1.5052,
      "step": 27167
    },
    {
      "epoch": 0.9729439361111608,
      "grad_norm": 1.2999725341796875,
      "learning_rate": 3.832025644674886e-07,
      "loss": 1.2422,
      "step": 27168
    },
    {
      "epoch": 0.9729797482407291,
      "grad_norm": 2.5449912548065186,
      "learning_rate": 3.821887772626931e-07,
      "loss": 1.3325,
      "step": 27169
    },
    {
      "epoch": 0.9730155603702975,
      "grad_norm": 1.7839765548706055,
      "learning_rate": 3.8117633028704745e-07,
      "loss": 1.0184,
      "step": 27170
    },
    {
      "epoch": 0.9730513724998657,
      "grad_norm": 1.3288191556930542,
      "learning_rate": 3.801652235541631e-07,
      "loss": 1.4608,
      "step": 27171
    },
    {
      "epoch": 0.973087184629434,
      "grad_norm": 1.5566009283065796,
      "learning_rate": 3.791554570776734e-07,
      "loss": 1.4342,
      "step": 27172
    },
    {
      "epoch": 0.9731229967590023,
      "grad_norm": 1.7924531698226929,
      "learning_rate": 3.781470308711343e-07,
      "loss": 1.2537,
      "step": 27173
    },
    {
      "epoch": 0.9731588088885705,
      "grad_norm": 1.666527271270752,
      "learning_rate": 3.771399449481239e-07,
      "loss": 1.0901,
      "step": 27174
    },
    {
      "epoch": 0.9731946210181388,
      "grad_norm": 1.79993736743927,
      "learning_rate": 3.761341993221867e-07,
      "loss": 1.2884,
      "step": 27175
    },
    {
      "epoch": 0.9732304331477071,
      "grad_norm": 1.9205312728881836,
      "learning_rate": 3.7512979400686763e-07,
      "loss": 1.389,
      "step": 27176
    },
    {
      "epoch": 0.9732662452772755,
      "grad_norm": 1.901877999305725,
      "learning_rate": 3.741267290156669e-07,
      "loss": 1.6964,
      "step": 27177
    },
    {
      "epoch": 0.9733020574068437,
      "grad_norm": 1.5668498277664185,
      "learning_rate": 3.7312500436208487e-07,
      "loss": 1.1559,
      "step": 27178
    },
    {
      "epoch": 0.973337869536412,
      "grad_norm": 1.4472907781600952,
      "learning_rate": 3.7212462005959957e-07,
      "loss": 1.306,
      "step": 27179
    },
    {
      "epoch": 0.9733736816659803,
      "grad_norm": 1.4478232860565186,
      "learning_rate": 3.7112557612165586e-07,
      "loss": 1.3558,
      "step": 27180
    },
    {
      "epoch": 0.9734094937955485,
      "grad_norm": 1.8058240413665771,
      "learning_rate": 3.7012787256172075e-07,
      "loss": 1.0638,
      "step": 27181
    },
    {
      "epoch": 0.9734453059251168,
      "grad_norm": 1.9508631229400635,
      "learning_rate": 3.6913150939318353e-07,
      "loss": 1.5124,
      "step": 27182
    },
    {
      "epoch": 0.9734811180546851,
      "grad_norm": 1.470751404762268,
      "learning_rate": 3.6813648662947785e-07,
      "loss": 1.427,
      "step": 27183
    },
    {
      "epoch": 0.9735169301842534,
      "grad_norm": 1.1770071983337402,
      "learning_rate": 3.67142804283982e-07,
      "loss": 1.2635,
      "step": 27184
    },
    {
      "epoch": 0.9735527423138217,
      "grad_norm": 1.5016038417816162,
      "learning_rate": 3.661504623700629e-07,
      "loss": 1.2942,
      "step": 27185
    },
    {
      "epoch": 0.97358855444339,
      "grad_norm": 1.4237912893295288,
      "learning_rate": 3.6515946090106557e-07,
      "loss": 1.7626,
      "step": 27186
    },
    {
      "epoch": 0.9736243665729583,
      "grad_norm": 1.517599105834961,
      "learning_rate": 3.641697998903237e-07,
      "loss": 1.4293,
      "step": 27187
    },
    {
      "epoch": 0.9736601787025265,
      "grad_norm": 1.4845151901245117,
      "learning_rate": 3.631814793511712e-07,
      "loss": 1.1693,
      "step": 27188
    },
    {
      "epoch": 0.9736959908320948,
      "grad_norm": 1.8167911767959595,
      "learning_rate": 3.621944992968751e-07,
      "loss": 1.401,
      "step": 27189
    },
    {
      "epoch": 0.9737318029616631,
      "grad_norm": 1.7373158931732178,
      "learning_rate": 3.612088597407359e-07,
      "loss": 1.4933,
      "step": 27190
    },
    {
      "epoch": 0.9737676150912314,
      "grad_norm": 1.7631866931915283,
      "learning_rate": 3.6022456069600973e-07,
      "loss": 1.3376,
      "step": 27191
    },
    {
      "epoch": 0.9738034272207997,
      "grad_norm": 1.3551874160766602,
      "learning_rate": 3.592416021759304e-07,
      "loss": 1.0676,
      "step": 27192
    },
    {
      "epoch": 0.973839239350368,
      "grad_norm": 1.2745553255081177,
      "learning_rate": 3.5825998419372065e-07,
      "loss": 1.3036,
      "step": 27193
    },
    {
      "epoch": 0.9738750514799362,
      "grad_norm": 1.5684278011322021,
      "learning_rate": 3.5727970676260327e-07,
      "loss": 1.1635,
      "step": 27194
    },
    {
      "epoch": 0.9739108636095045,
      "grad_norm": 1.5761404037475586,
      "learning_rate": 3.563007698957566e-07,
      "loss": 1.2785,
      "step": 27195
    },
    {
      "epoch": 0.9739466757390728,
      "grad_norm": 1.679574728012085,
      "learning_rate": 3.553231736063589e-07,
      "loss": 1.723,
      "step": 27196
    },
    {
      "epoch": 0.9739824878686411,
      "grad_norm": 1.4595870971679688,
      "learning_rate": 3.5434691790754427e-07,
      "loss": 1.6485,
      "step": 27197
    },
    {
      "epoch": 0.9740182999982094,
      "grad_norm": 1.6029869318008423,
      "learning_rate": 3.5337200281245765e-07,
      "loss": 1.4157,
      "step": 27198
    },
    {
      "epoch": 0.9740541121277777,
      "grad_norm": 1.6887636184692383,
      "learning_rate": 3.52398428334233e-07,
      "loss": 1.6353,
      "step": 27199
    },
    {
      "epoch": 0.974089924257346,
      "grad_norm": 1.629212737083435,
      "learning_rate": 3.514261944859376e-07,
      "loss": 1.4481,
      "step": 27200
    },
    {
      "epoch": 0.9741257363869142,
      "grad_norm": 1.597827672958374,
      "learning_rate": 3.5045530128066106e-07,
      "loss": 1.5889,
      "step": 27201
    },
    {
      "epoch": 0.9741615485164825,
      "grad_norm": 1.69936203956604,
      "learning_rate": 3.4948574873148174e-07,
      "loss": 1.5639,
      "step": 27202
    },
    {
      "epoch": 0.9741973606460508,
      "grad_norm": 2.234874963760376,
      "learning_rate": 3.4851753685142265e-07,
      "loss": 1.637,
      "step": 27203
    },
    {
      "epoch": 0.974233172775619,
      "grad_norm": 1.872274398803711,
      "learning_rate": 3.4755066565351767e-07,
      "loss": 1.4049,
      "step": 27204
    },
    {
      "epoch": 0.9742689849051874,
      "grad_norm": 1.5312186479568481,
      "learning_rate": 3.465851351507787e-07,
      "loss": 1.2672,
      "step": 27205
    },
    {
      "epoch": 0.9743047970347557,
      "grad_norm": 1.6697921752929688,
      "learning_rate": 3.456209453561954e-07,
      "loss": 1.6174,
      "step": 27206
    },
    {
      "epoch": 0.974340609164324,
      "grad_norm": 1.3625354766845703,
      "learning_rate": 3.4465809628273504e-07,
      "loss": 0.9797,
      "step": 27207
    },
    {
      "epoch": 0.9743764212938922,
      "grad_norm": 1.308997631072998,
      "learning_rate": 3.4369658794335403e-07,
      "loss": 1.3414,
      "step": 27208
    },
    {
      "epoch": 0.9744122334234605,
      "grad_norm": 1.4477964639663696,
      "learning_rate": 3.4273642035099753e-07,
      "loss": 1.2255,
      "step": 27209
    },
    {
      "epoch": 0.9744480455530288,
      "grad_norm": 1.5210626125335693,
      "learning_rate": 3.417775935185663e-07,
      "loss": 1.4795,
      "step": 27210
    },
    {
      "epoch": 0.974483857682597,
      "grad_norm": 1.6373701095581055,
      "learning_rate": 3.408201074589612e-07,
      "loss": 1.6343,
      "step": 27211
    },
    {
      "epoch": 0.9745196698121654,
      "grad_norm": 1.727810263633728,
      "learning_rate": 3.39863962185083e-07,
      "loss": 1.565,
      "step": 27212
    },
    {
      "epoch": 0.9745554819417337,
      "grad_norm": 1.3922851085662842,
      "learning_rate": 3.3890915770977694e-07,
      "loss": 1.2344,
      "step": 27213
    },
    {
      "epoch": 0.974591294071302,
      "grad_norm": 1.5838536024093628,
      "learning_rate": 3.379556940458883e-07,
      "loss": 1.3403,
      "step": 27214
    },
    {
      "epoch": 0.9746271062008702,
      "grad_norm": 1.6230381727218628,
      "learning_rate": 3.3700357120626247e-07,
      "loss": 1.4572,
      "step": 27215
    },
    {
      "epoch": 0.9746629183304385,
      "grad_norm": 1.2696282863616943,
      "learning_rate": 3.360527892036891e-07,
      "loss": 1.6312,
      "step": 27216
    },
    {
      "epoch": 0.9746987304600068,
      "grad_norm": 1.652672529220581,
      "learning_rate": 3.351033480509691e-07,
      "loss": 1.4796,
      "step": 27217
    },
    {
      "epoch": 0.974734542589575,
      "grad_norm": 1.8868895769119263,
      "learning_rate": 3.3415524776088116e-07,
      "loss": 1.3838,
      "step": 27218
    },
    {
      "epoch": 0.9747703547191434,
      "grad_norm": 1.9636809825897217,
      "learning_rate": 3.332084883461706e-07,
      "loss": 1.42,
      "step": 27219
    },
    {
      "epoch": 0.9748061668487117,
      "grad_norm": 1.3785717487335205,
      "learning_rate": 3.3226306981957166e-07,
      "loss": 1.6404,
      "step": 27220
    },
    {
      "epoch": 0.97484197897828,
      "grad_norm": 1.4908267259597778,
      "learning_rate": 3.313189921938187e-07,
      "loss": 1.2639,
      "step": 27221
    },
    {
      "epoch": 0.9748777911078482,
      "grad_norm": 2.0432732105255127,
      "learning_rate": 3.3037625548160143e-07,
      "loss": 1.8091,
      "step": 27222
    },
    {
      "epoch": 0.9749136032374165,
      "grad_norm": 2.1694438457489014,
      "learning_rate": 3.294348596956098e-07,
      "loss": 1.3754,
      "step": 27223
    },
    {
      "epoch": 0.9749494153669848,
      "grad_norm": 1.6928741931915283,
      "learning_rate": 3.2849480484851145e-07,
      "loss": 1.213,
      "step": 27224
    },
    {
      "epoch": 0.974985227496553,
      "grad_norm": 1.616788625717163,
      "learning_rate": 3.275560909529407e-07,
      "loss": 1.5529,
      "step": 27225
    },
    {
      "epoch": 0.9750210396261214,
      "grad_norm": 1.5182127952575684,
      "learning_rate": 3.2661871802154296e-07,
      "loss": 1.4226,
      "step": 27226
    },
    {
      "epoch": 0.9750568517556897,
      "grad_norm": 1.6872198581695557,
      "learning_rate": 3.256826860669193e-07,
      "loss": 1.3533,
      "step": 27227
    },
    {
      "epoch": 0.975092663885258,
      "grad_norm": 2.066617012023926,
      "learning_rate": 3.2474799510165965e-07,
      "loss": 1.5938,
      "step": 27228
    },
    {
      "epoch": 0.9751284760148262,
      "grad_norm": 1.7410343885421753,
      "learning_rate": 3.238146451383428e-07,
      "loss": 1.4438,
      "step": 27229
    },
    {
      "epoch": 0.9751642881443945,
      "grad_norm": 1.7189652919769287,
      "learning_rate": 3.228826361895254e-07,
      "loss": 1.6799,
      "step": 27230
    },
    {
      "epoch": 0.9752001002739628,
      "grad_norm": 1.2681914567947388,
      "learning_rate": 3.21951968267753e-07,
      "loss": 1.3594,
      "step": 27231
    },
    {
      "epoch": 0.975235912403531,
      "grad_norm": 1.4830594062805176,
      "learning_rate": 3.2102264138553774e-07,
      "loss": 1.3071,
      "step": 27232
    },
    {
      "epoch": 0.9752717245330994,
      "grad_norm": 1.574739694595337,
      "learning_rate": 3.200946555553919e-07,
      "loss": 1.2429,
      "step": 27233
    },
    {
      "epoch": 0.9753075366626677,
      "grad_norm": 2.1808180809020996,
      "learning_rate": 3.191680107897943e-07,
      "loss": 1.2584,
      "step": 27234
    },
    {
      "epoch": 0.9753433487922359,
      "grad_norm": 1.646176815032959,
      "learning_rate": 3.1824270710121286e-07,
      "loss": 1.3929,
      "step": 27235
    },
    {
      "epoch": 0.9753791609218042,
      "grad_norm": 1.216214656829834,
      "learning_rate": 3.173187445020931e-07,
      "loss": 1.1705,
      "step": 27236
    },
    {
      "epoch": 0.9754149730513725,
      "grad_norm": 1.5510247945785522,
      "learning_rate": 3.1639612300485844e-07,
      "loss": 1.4461,
      "step": 27237
    },
    {
      "epoch": 0.9754507851809407,
      "grad_norm": 1.6966277360916138,
      "learning_rate": 3.1547484262194336e-07,
      "loss": 1.5542,
      "step": 27238
    },
    {
      "epoch": 0.975486597310509,
      "grad_norm": 2.0955851078033447,
      "learning_rate": 3.1455490336572693e-07,
      "loss": 1.1453,
      "step": 27239
    },
    {
      "epoch": 0.9755224094400774,
      "grad_norm": 1.7406843900680542,
      "learning_rate": 3.13636305248588e-07,
      "loss": 1.4391,
      "step": 27240
    },
    {
      "epoch": 0.9755582215696457,
      "grad_norm": 1.6218078136444092,
      "learning_rate": 3.1271904828288343e-07,
      "loss": 1.4322,
      "step": 27241
    },
    {
      "epoch": 0.9755940336992139,
      "grad_norm": 1.6305664777755737,
      "learning_rate": 3.1180313248097004e-07,
      "loss": 1.2418,
      "step": 27242
    },
    {
      "epoch": 0.9756298458287822,
      "grad_norm": 1.9506369829177856,
      "learning_rate": 3.10888557855149e-07,
      "loss": 1.5543,
      "step": 27243
    },
    {
      "epoch": 0.9756656579583505,
      "grad_norm": 1.932220697402954,
      "learning_rate": 3.099753244177217e-07,
      "loss": 1.4439,
      "step": 27244
    },
    {
      "epoch": 0.9757014700879187,
      "grad_norm": 1.4722483158111572,
      "learning_rate": 3.090634321810004e-07,
      "loss": 1.3055,
      "step": 27245
    },
    {
      "epoch": 0.975737282217487,
      "grad_norm": 2.48748779296875,
      "learning_rate": 3.0815288115723095e-07,
      "loss": 1.7239,
      "step": 27246
    },
    {
      "epoch": 0.9757730943470554,
      "grad_norm": 1.9032803773880005,
      "learning_rate": 3.0724367135868126e-07,
      "loss": 1.4195,
      "step": 27247
    },
    {
      "epoch": 0.9758089064766237,
      "grad_norm": 2.151289939880371,
      "learning_rate": 3.063358027975638e-07,
      "loss": 1.3273,
      "step": 27248
    },
    {
      "epoch": 0.9758447186061919,
      "grad_norm": 1.8209279775619507,
      "learning_rate": 3.054292754861021e-07,
      "loss": 1.611,
      "step": 27249
    },
    {
      "epoch": 0.9758805307357602,
      "grad_norm": 1.8769649267196655,
      "learning_rate": 3.0452408943649756e-07,
      "loss": 1.5656,
      "step": 27250
    },
    {
      "epoch": 0.9759163428653285,
      "grad_norm": 1.3402371406555176,
      "learning_rate": 3.0362024466092933e-07,
      "loss": 1.4473,
      "step": 27251
    },
    {
      "epoch": 0.9759521549948967,
      "grad_norm": 1.6566824913024902,
      "learning_rate": 3.0271774117153207e-07,
      "loss": 1.3018,
      "step": 27252
    },
    {
      "epoch": 0.975987967124465,
      "grad_norm": 1.9619542360305786,
      "learning_rate": 3.01816578980485e-07,
      "loss": 1.405,
      "step": 27253
    },
    {
      "epoch": 0.9760237792540334,
      "grad_norm": 1.8266921043395996,
      "learning_rate": 3.009167580998895e-07,
      "loss": 1.2748,
      "step": 27254
    },
    {
      "epoch": 0.9760595913836017,
      "grad_norm": 2.3018181324005127,
      "learning_rate": 3.0001827854184704e-07,
      "loss": 1.6089,
      "step": 27255
    },
    {
      "epoch": 0.9760954035131699,
      "grad_norm": 1.7781779766082764,
      "learning_rate": 2.9912114031847015e-07,
      "loss": 1.3481,
      "step": 27256
    },
    {
      "epoch": 0.9761312156427382,
      "grad_norm": 1.4521387815475464,
      "learning_rate": 2.9822534344180475e-07,
      "loss": 1.3735,
      "step": 27257
    },
    {
      "epoch": 0.9761670277723065,
      "grad_norm": 2.076186180114746,
      "learning_rate": 2.9733088792391894e-07,
      "loss": 1.0925,
      "step": 27258
    },
    {
      "epoch": 0.9762028399018747,
      "grad_norm": 1.6280510425567627,
      "learning_rate": 2.9643777377682535e-07,
      "loss": 1.4911,
      "step": 27259
    },
    {
      "epoch": 0.976238652031443,
      "grad_norm": 1.9344412088394165,
      "learning_rate": 2.955460010125699e-07,
      "loss": 1.3981,
      "step": 27260
    },
    {
      "epoch": 0.9762744641610114,
      "grad_norm": 1.447930932044983,
      "learning_rate": 2.946555696431208e-07,
      "loss": 1.571,
      "step": 27261
    },
    {
      "epoch": 0.9763102762905796,
      "grad_norm": 1.7467442750930786,
      "learning_rate": 2.9376647968047954e-07,
      "loss": 1.5898,
      "step": 27262
    },
    {
      "epoch": 0.9763460884201479,
      "grad_norm": 1.490739107131958,
      "learning_rate": 2.928787311365921e-07,
      "loss": 1.4754,
      "step": 27263
    },
    {
      "epoch": 0.9763819005497162,
      "grad_norm": 1.6217472553253174,
      "learning_rate": 2.919923240234046e-07,
      "loss": 1.4527,
      "step": 27264
    },
    {
      "epoch": 0.9764177126792845,
      "grad_norm": 1.551711916923523,
      "learning_rate": 2.911072583528518e-07,
      "loss": 1.3942,
      "step": 27265
    },
    {
      "epoch": 0.9764535248088527,
      "grad_norm": 1.4863595962524414,
      "learning_rate": 2.9022353413683534e-07,
      "loss": 1.371,
      "step": 27266
    },
    {
      "epoch": 0.976489336938421,
      "grad_norm": 1.9599124193191528,
      "learning_rate": 2.893411513872457e-07,
      "loss": 1.4827,
      "step": 27267
    },
    {
      "epoch": 0.9765251490679894,
      "grad_norm": 1.4526926279067993,
      "learning_rate": 2.884601101159512e-07,
      "loss": 1.252,
      "step": 27268
    },
    {
      "epoch": 0.9765609611975576,
      "grad_norm": 1.6005815267562866,
      "learning_rate": 2.87580410334809e-07,
      "loss": 1.4749,
      "step": 27269
    },
    {
      "epoch": 0.9765967733271259,
      "grad_norm": 1.9566835165023804,
      "learning_rate": 2.8670205205565406e-07,
      "loss": 1.6584,
      "step": 27270
    },
    {
      "epoch": 0.9766325854566942,
      "grad_norm": 1.6158761978149414,
      "learning_rate": 2.8582503529029916e-07,
      "loss": 1.0847,
      "step": 27271
    },
    {
      "epoch": 0.9766683975862624,
      "grad_norm": 1.6226449012756348,
      "learning_rate": 2.84949360050546e-07,
      "loss": 1.2105,
      "step": 27272
    },
    {
      "epoch": 0.9767042097158307,
      "grad_norm": 1.3435142040252686,
      "learning_rate": 2.8407502634817395e-07,
      "loss": 1.3259,
      "step": 27273
    },
    {
      "epoch": 0.976740021845399,
      "grad_norm": 1.224449872970581,
      "learning_rate": 2.8320203419495153e-07,
      "loss": 1.0536,
      "step": 27274
    },
    {
      "epoch": 0.9767758339749674,
      "grad_norm": 1.4458800554275513,
      "learning_rate": 2.8233038360262474e-07,
      "loss": 1.2663,
      "step": 27275
    },
    {
      "epoch": 0.9768116461045356,
      "grad_norm": 1.431313395500183,
      "learning_rate": 2.814600745829177e-07,
      "loss": 1.3087,
      "step": 27276
    },
    {
      "epoch": 0.9768474582341039,
      "grad_norm": 1.2949119806289673,
      "learning_rate": 2.805911071475209e-07,
      "loss": 1.3999,
      "step": 27277
    },
    {
      "epoch": 0.9768832703636722,
      "grad_norm": 1.5865280628204346,
      "learning_rate": 2.797234813081584e-07,
      "loss": 1.1732,
      "step": 27278
    },
    {
      "epoch": 0.9769190824932404,
      "grad_norm": 1.7895091772079468,
      "learning_rate": 2.788571970764764e-07,
      "loss": 1.3668,
      "step": 27279
    },
    {
      "epoch": 0.9769548946228087,
      "grad_norm": 1.5917913913726807,
      "learning_rate": 2.7799225446414334e-07,
      "loss": 1.3931,
      "step": 27280
    },
    {
      "epoch": 0.976990706752377,
      "grad_norm": 1.47417414188385,
      "learning_rate": 2.7712865348279436e-07,
      "loss": 1.3361,
      "step": 27281
    },
    {
      "epoch": 0.9770265188819454,
      "grad_norm": 1.7814379930496216,
      "learning_rate": 2.762663941440424e-07,
      "loss": 1.6861,
      "step": 27282
    },
    {
      "epoch": 0.9770623310115136,
      "grad_norm": 1.5276187658309937,
      "learning_rate": 2.7540547645950045e-07,
      "loss": 1.3426,
      "step": 27283
    },
    {
      "epoch": 0.9770981431410819,
      "grad_norm": 1.551092505455017,
      "learning_rate": 2.745459004407369e-07,
      "loss": 1.2497,
      "step": 27284
    },
    {
      "epoch": 0.9771339552706502,
      "grad_norm": 1.6632046699523926,
      "learning_rate": 2.736876660993204e-07,
      "loss": 1.4969,
      "step": 27285
    },
    {
      "epoch": 0.9771697674002184,
      "grad_norm": 1.940461277961731,
      "learning_rate": 2.728307734467972e-07,
      "loss": 1.3537,
      "step": 27286
    },
    {
      "epoch": 0.9772055795297867,
      "grad_norm": 2.302395820617676,
      "learning_rate": 2.719752224947025e-07,
      "loss": 1.3458,
      "step": 27287
    },
    {
      "epoch": 0.977241391659355,
      "grad_norm": 1.845481038093567,
      "learning_rate": 2.711210132545383e-07,
      "loss": 1.3436,
      "step": 27288
    },
    {
      "epoch": 0.9772772037889234,
      "grad_norm": 1.4163119792938232,
      "learning_rate": 2.7026814573779534e-07,
      "loss": 1.3536,
      "step": 27289
    },
    {
      "epoch": 0.9773130159184916,
      "grad_norm": 1.6288304328918457,
      "learning_rate": 2.6941661995596446e-07,
      "loss": 1.4439,
      "step": 27290
    },
    {
      "epoch": 0.9773488280480599,
      "grad_norm": 1.5382457971572876,
      "learning_rate": 2.6856643592048093e-07,
      "loss": 1.5094,
      "step": 27291
    },
    {
      "epoch": 0.9773846401776282,
      "grad_norm": 1.2881473302841187,
      "learning_rate": 2.6771759364279114e-07,
      "loss": 1.5564,
      "step": 27292
    },
    {
      "epoch": 0.9774204523071964,
      "grad_norm": 1.9076626300811768,
      "learning_rate": 2.668700931343082e-07,
      "loss": 1.5091,
      "step": 27293
    },
    {
      "epoch": 0.9774562644367647,
      "grad_norm": 2.171661376953125,
      "learning_rate": 2.6602393440645636e-07,
      "loss": 1.4619,
      "step": 27294
    },
    {
      "epoch": 0.977492076566333,
      "grad_norm": 1.776262879371643,
      "learning_rate": 2.651791174706042e-07,
      "loss": 1.5386,
      "step": 27295
    },
    {
      "epoch": 0.9775278886959013,
      "grad_norm": 1.5984774827957153,
      "learning_rate": 2.6433564233810936e-07,
      "loss": 1.5978,
      "step": 27296
    },
    {
      "epoch": 0.9775637008254696,
      "grad_norm": 1.9452379941940308,
      "learning_rate": 2.6349350902032944e-07,
      "loss": 1.668,
      "step": 27297
    },
    {
      "epoch": 0.9775995129550379,
      "grad_norm": 1.268587589263916,
      "learning_rate": 2.6265271752859975e-07,
      "loss": 1.3742,
      "step": 27298
    },
    {
      "epoch": 0.9776353250846062,
      "grad_norm": 1.4393203258514404,
      "learning_rate": 2.618132678742224e-07,
      "loss": 1.2159,
      "step": 27299
    },
    {
      "epoch": 0.9776711372141744,
      "grad_norm": 1.389753818511963,
      "learning_rate": 2.609751600684995e-07,
      "loss": 1.1791,
      "step": 27300
    },
    {
      "epoch": 0.9777069493437427,
      "grad_norm": 1.9547994136810303,
      "learning_rate": 2.601383941226998e-07,
      "loss": 1.4618,
      "step": 27301
    },
    {
      "epoch": 0.977742761473311,
      "grad_norm": 1.6610933542251587,
      "learning_rate": 2.593029700480698e-07,
      "loss": 1.6401,
      "step": 27302
    },
    {
      "epoch": 0.9777785736028793,
      "grad_norm": 2.492784261703491,
      "learning_rate": 2.584688878558783e-07,
      "loss": 1.4826,
      "step": 27303
    },
    {
      "epoch": 0.9778143857324476,
      "grad_norm": 1.273895263671875,
      "learning_rate": 2.576361475573275e-07,
      "loss": 1.4765,
      "step": 27304
    },
    {
      "epoch": 0.9778501978620159,
      "grad_norm": 1.622200846672058,
      "learning_rate": 2.568047491636194e-07,
      "loss": 1.1659,
      "step": 27305
    },
    {
      "epoch": 0.9778860099915841,
      "grad_norm": 1.9652822017669678,
      "learning_rate": 2.559746926859452e-07,
      "loss": 1.4729,
      "step": 27306
    },
    {
      "epoch": 0.9779218221211524,
      "grad_norm": 1.710400938987732,
      "learning_rate": 2.551459781354737e-07,
      "loss": 1.716,
      "step": 27307
    },
    {
      "epoch": 0.9779576342507207,
      "grad_norm": 2.0867536067962646,
      "learning_rate": 2.543186055233515e-07,
      "loss": 1.2675,
      "step": 27308
    },
    {
      "epoch": 0.977993446380289,
      "grad_norm": 1.5373716354370117,
      "learning_rate": 2.534925748607031e-07,
      "loss": 1.2282,
      "step": 27309
    },
    {
      "epoch": 0.9780292585098573,
      "grad_norm": 1.6054054498672485,
      "learning_rate": 2.526678861586529e-07,
      "loss": 1.6797,
      "step": 27310
    },
    {
      "epoch": 0.9780650706394256,
      "grad_norm": 1.2186475992202759,
      "learning_rate": 2.51844539428292e-07,
      "loss": 1.5985,
      "step": 27311
    },
    {
      "epoch": 0.9781008827689939,
      "grad_norm": 1.5311107635498047,
      "learning_rate": 2.5102253468070047e-07,
      "loss": 1.2037,
      "step": 27312
    },
    {
      "epoch": 0.9781366948985621,
      "grad_norm": 2.0063748359680176,
      "learning_rate": 2.50201871926925e-07,
      "loss": 1.2174,
      "step": 27313
    },
    {
      "epoch": 0.9781725070281304,
      "grad_norm": 1.6210482120513916,
      "learning_rate": 2.4938255117802347e-07,
      "loss": 1.3572,
      "step": 27314
    },
    {
      "epoch": 0.9782083191576987,
      "grad_norm": 1.8170969486236572,
      "learning_rate": 2.485645724450092e-07,
      "loss": 1.2536,
      "step": 27315
    },
    {
      "epoch": 0.978244131287267,
      "grad_norm": 1.7604585886001587,
      "learning_rate": 2.4774793573888453e-07,
      "loss": 1.336,
      "step": 27316
    },
    {
      "epoch": 0.9782799434168353,
      "grad_norm": 1.353798747062683,
      "learning_rate": 2.4693264107064075e-07,
      "loss": 1.3545,
      "step": 27317
    },
    {
      "epoch": 0.9783157555464036,
      "grad_norm": 1.6163965463638306,
      "learning_rate": 2.4611868845124673e-07,
      "loss": 1.1518,
      "step": 27318
    },
    {
      "epoch": 0.9783515676759719,
      "grad_norm": 1.461172342300415,
      "learning_rate": 2.453060778916605e-07,
      "loss": 1.5765,
      "step": 27319
    },
    {
      "epoch": 0.9783873798055401,
      "grad_norm": 1.6993408203125,
      "learning_rate": 2.4449480940279547e-07,
      "loss": 1.5854,
      "step": 27320
    },
    {
      "epoch": 0.9784231919351084,
      "grad_norm": 1.431376576423645,
      "learning_rate": 2.436848829955762e-07,
      "loss": 1.1842,
      "step": 27321
    },
    {
      "epoch": 0.9784590040646767,
      "grad_norm": 2.1553871631622314,
      "learning_rate": 2.428762986809052e-07,
      "loss": 1.4902,
      "step": 27322
    },
    {
      "epoch": 0.9784948161942449,
      "grad_norm": 1.702622652053833,
      "learning_rate": 2.420690564696626e-07,
      "loss": 1.5358,
      "step": 27323
    },
    {
      "epoch": 0.9785306283238133,
      "grad_norm": 1.2670756578445435,
      "learning_rate": 2.4126315637269523e-07,
      "loss": 0.9199,
      "step": 27324
    },
    {
      "epoch": 0.9785664404533816,
      "grad_norm": 1.6474324464797974,
      "learning_rate": 2.4045859840085005e-07,
      "loss": 1.3944,
      "step": 27325
    },
    {
      "epoch": 0.9786022525829499,
      "grad_norm": 1.6244606971740723,
      "learning_rate": 2.3965538256496276e-07,
      "loss": 1.1658,
      "step": 27326
    },
    {
      "epoch": 0.9786380647125181,
      "grad_norm": 1.3171576261520386,
      "learning_rate": 2.388535088758248e-07,
      "loss": 1.4406,
      "step": 27327
    },
    {
      "epoch": 0.9786738768420864,
      "grad_norm": 1.325938105583191,
      "learning_rate": 2.3805297734422748e-07,
      "loss": 1.3195,
      "step": 27328
    },
    {
      "epoch": 0.9787096889716547,
      "grad_norm": 1.6224009990692139,
      "learning_rate": 2.3725378798095112e-07,
      "loss": 1.7065,
      "step": 27329
    },
    {
      "epoch": 0.9787455011012229,
      "grad_norm": 1.4370075464248657,
      "learning_rate": 2.364559407967426e-07,
      "loss": 1.4062,
      "step": 27330
    },
    {
      "epoch": 0.9787813132307913,
      "grad_norm": 1.7838104963302612,
      "learning_rate": 2.3565943580232676e-07,
      "loss": 1.6156,
      "step": 27331
    },
    {
      "epoch": 0.9788171253603596,
      "grad_norm": 1.8117674589157104,
      "learning_rate": 2.3486427300841717e-07,
      "loss": 1.8238,
      "step": 27332
    },
    {
      "epoch": 0.9788529374899279,
      "grad_norm": 1.5283007621765137,
      "learning_rate": 2.3407045242573867e-07,
      "loss": 1.1369,
      "step": 27333
    },
    {
      "epoch": 0.9788887496194961,
      "grad_norm": 1.715983510017395,
      "learning_rate": 2.3327797406494934e-07,
      "loss": 1.3713,
      "step": 27334
    },
    {
      "epoch": 0.9789245617490644,
      "grad_norm": 1.5859873294830322,
      "learning_rate": 2.3248683793670735e-07,
      "loss": 1.4738,
      "step": 27335
    },
    {
      "epoch": 0.9789603738786327,
      "grad_norm": 1.5380550622940063,
      "learning_rate": 2.316970440516708e-07,
      "loss": 1.3575,
      "step": 27336
    },
    {
      "epoch": 0.9789961860082009,
      "grad_norm": 1.4396132230758667,
      "learning_rate": 2.309085924204535e-07,
      "loss": 1.3689,
      "step": 27337
    },
    {
      "epoch": 0.9790319981377693,
      "grad_norm": 1.9657173156738281,
      "learning_rate": 2.301214830536691e-07,
      "loss": 1.7597,
      "step": 27338
    },
    {
      "epoch": 0.9790678102673376,
      "grad_norm": 2.931874990463257,
      "learning_rate": 2.2933571596190918e-07,
      "loss": 1.9935,
      "step": 27339
    },
    {
      "epoch": 0.9791036223969058,
      "grad_norm": 1.8927358388900757,
      "learning_rate": 2.2855129115574304e-07,
      "loss": 1.2429,
      "step": 27340
    },
    {
      "epoch": 0.9791394345264741,
      "grad_norm": 1.5039610862731934,
      "learning_rate": 2.277682086457289e-07,
      "loss": 1.6117,
      "step": 27341
    },
    {
      "epoch": 0.9791752466560424,
      "grad_norm": 1.5220850706100464,
      "learning_rate": 2.269864684423917e-07,
      "loss": 1.2156,
      "step": 27342
    },
    {
      "epoch": 0.9792110587856107,
      "grad_norm": 1.7066388130187988,
      "learning_rate": 2.262060705562452e-07,
      "loss": 1.4429,
      "step": 27343
    },
    {
      "epoch": 0.9792468709151789,
      "grad_norm": 1.3088314533233643,
      "learning_rate": 2.2542701499780327e-07,
      "loss": 1.3944,
      "step": 27344
    },
    {
      "epoch": 0.9792826830447473,
      "grad_norm": 2.0127079486846924,
      "learning_rate": 2.246493017775353e-07,
      "loss": 1.0906,
      "step": 27345
    },
    {
      "epoch": 0.9793184951743156,
      "grad_norm": 1.8624136447906494,
      "learning_rate": 2.2387293090592177e-07,
      "loss": 1.7316,
      "step": 27346
    },
    {
      "epoch": 0.9793543073038838,
      "grad_norm": 2.0756258964538574,
      "learning_rate": 2.230979023933877e-07,
      "loss": 1.341,
      "step": 27347
    },
    {
      "epoch": 0.9793901194334521,
      "grad_norm": 1.8070365190505981,
      "learning_rate": 2.2232421625036914e-07,
      "loss": 1.7985,
      "step": 27348
    },
    {
      "epoch": 0.9794259315630204,
      "grad_norm": 1.252737283706665,
      "learning_rate": 2.2155187248728004e-07,
      "loss": 1.3545,
      "step": 27349
    },
    {
      "epoch": 0.9794617436925886,
      "grad_norm": 1.4469175338745117,
      "learning_rate": 2.2078087111450097e-07,
      "loss": 1.5457,
      "step": 27350
    },
    {
      "epoch": 0.9794975558221569,
      "grad_norm": 1.5265394449234009,
      "learning_rate": 2.200112121424125e-07,
      "loss": 1.4557,
      "step": 27351
    },
    {
      "epoch": 0.9795333679517253,
      "grad_norm": 1.837325096130371,
      "learning_rate": 2.192428955813619e-07,
      "loss": 1.3282,
      "step": 27352
    },
    {
      "epoch": 0.9795691800812936,
      "grad_norm": 2.001613140106201,
      "learning_rate": 2.184759214416854e-07,
      "loss": 1.3861,
      "step": 27353
    },
    {
      "epoch": 0.9796049922108618,
      "grad_norm": 2.0825603008270264,
      "learning_rate": 2.177102897337191e-07,
      "loss": 1.5685,
      "step": 27354
    },
    {
      "epoch": 0.9796408043404301,
      "grad_norm": 1.7808622121810913,
      "learning_rate": 2.169460004677437e-07,
      "loss": 1.4419,
      "step": 27355
    },
    {
      "epoch": 0.9796766164699984,
      "grad_norm": 1.5627222061157227,
      "learning_rate": 2.16183053654051e-07,
      "loss": 1.3342,
      "step": 27356
    },
    {
      "epoch": 0.9797124285995666,
      "grad_norm": 1.7511463165283203,
      "learning_rate": 2.1542144930289943e-07,
      "loss": 1.1166,
      "step": 27357
    },
    {
      "epoch": 0.9797482407291349,
      "grad_norm": 1.368848204612732,
      "learning_rate": 2.1466118742453634e-07,
      "loss": 1.4296,
      "step": 27358
    },
    {
      "epoch": 0.9797840528587033,
      "grad_norm": 1.8091198205947876,
      "learning_rate": 2.139022680292091e-07,
      "loss": 1.3872,
      "step": 27359
    },
    {
      "epoch": 0.9798198649882716,
      "grad_norm": 1.5368421077728271,
      "learning_rate": 2.1314469112709844e-07,
      "loss": 1.1766,
      "step": 27360
    },
    {
      "epoch": 0.9798556771178398,
      "grad_norm": 1.9813166856765747,
      "learning_rate": 2.1238845672841845e-07,
      "loss": 1.4277,
      "step": 27361
    },
    {
      "epoch": 0.9798914892474081,
      "grad_norm": 2.2372472286224365,
      "learning_rate": 2.1163356484332764e-07,
      "loss": 1.2097,
      "step": 27362
    },
    {
      "epoch": 0.9799273013769764,
      "grad_norm": 1.3528220653533936,
      "learning_rate": 2.1088001548199565e-07,
      "loss": 1.6458,
      "step": 27363
    },
    {
      "epoch": 0.9799631135065446,
      "grad_norm": 1.6533170938491821,
      "learning_rate": 2.1012780865454773e-07,
      "loss": 1.089,
      "step": 27364
    },
    {
      "epoch": 0.9799989256361129,
      "grad_norm": 1.8356963396072388,
      "learning_rate": 2.093769443711091e-07,
      "loss": 1.4844,
      "step": 27365
    },
    {
      "epoch": 0.9800347377656813,
      "grad_norm": 1.6935428380966187,
      "learning_rate": 2.086274226417939e-07,
      "loss": 1.5054,
      "step": 27366
    },
    {
      "epoch": 0.9800705498952496,
      "grad_norm": 1.528594732284546,
      "learning_rate": 2.0787924347666076e-07,
      "loss": 1.5082,
      "step": 27367
    },
    {
      "epoch": 0.9801063620248178,
      "grad_norm": 1.4918920993804932,
      "learning_rate": 2.071324068858016e-07,
      "loss": 1.2325,
      "step": 27368
    },
    {
      "epoch": 0.9801421741543861,
      "grad_norm": 1.6513475179672241,
      "learning_rate": 2.0638691287925282e-07,
      "loss": 1.7074,
      "step": 27369
    },
    {
      "epoch": 0.9801779862839544,
      "grad_norm": 1.2635741233825684,
      "learning_rate": 2.0564276146703977e-07,
      "loss": 1.2669,
      "step": 27370
    },
    {
      "epoch": 0.9802137984135226,
      "grad_norm": 2.1320393085479736,
      "learning_rate": 2.048999526591766e-07,
      "loss": 1.4007,
      "step": 27371
    },
    {
      "epoch": 0.9802496105430909,
      "grad_norm": 1.2138698101043701,
      "learning_rate": 2.041584864656554e-07,
      "loss": 1.3064,
      "step": 27372
    },
    {
      "epoch": 0.9802854226726593,
      "grad_norm": 1.551811695098877,
      "learning_rate": 2.03418362896457e-07,
      "loss": 1.4349,
      "step": 27373
    },
    {
      "epoch": 0.9803212348022275,
      "grad_norm": 1.5845892429351807,
      "learning_rate": 2.0267958196154013e-07,
      "loss": 1.3266,
      "step": 27374
    },
    {
      "epoch": 0.9803570469317958,
      "grad_norm": 1.8183881044387817,
      "learning_rate": 2.019421436708413e-07,
      "loss": 1.7351,
      "step": 27375
    },
    {
      "epoch": 0.9803928590613641,
      "grad_norm": 1.50227952003479,
      "learning_rate": 2.012060480342748e-07,
      "loss": 1.6177,
      "step": 27376
    },
    {
      "epoch": 0.9804286711909324,
      "grad_norm": 1.61687171459198,
      "learning_rate": 2.0047129506175488e-07,
      "loss": 1.2517,
      "step": 27377
    },
    {
      "epoch": 0.9804644833205006,
      "grad_norm": 1.2359267473220825,
      "learning_rate": 1.9973788476315147e-07,
      "loss": 1.3682,
      "step": 27378
    },
    {
      "epoch": 0.9805002954500689,
      "grad_norm": 1.4068397283554077,
      "learning_rate": 1.9900581714835666e-07,
      "loss": 1.4976,
      "step": 27379
    },
    {
      "epoch": 0.9805361075796373,
      "grad_norm": 1.6050434112548828,
      "learning_rate": 1.9827509222719587e-07,
      "loss": 1.3544,
      "step": 27380
    },
    {
      "epoch": 0.9805719197092055,
      "grad_norm": 1.5304006338119507,
      "learning_rate": 1.9754571000950572e-07,
      "loss": 1.4646,
      "step": 27381
    },
    {
      "epoch": 0.9806077318387738,
      "grad_norm": 1.6975919008255005,
      "learning_rate": 1.968176705051117e-07,
      "loss": 1.4701,
      "step": 27382
    },
    {
      "epoch": 0.9806435439683421,
      "grad_norm": 1.4890451431274414,
      "learning_rate": 1.960909737237837e-07,
      "loss": 0.9229,
      "step": 27383
    },
    {
      "epoch": 0.9806793560979103,
      "grad_norm": 1.4120088815689087,
      "learning_rate": 1.9536561967532507e-07,
      "loss": 1.7422,
      "step": 27384
    },
    {
      "epoch": 0.9807151682274786,
      "grad_norm": 1.6301904916763306,
      "learning_rate": 1.9464160836948354e-07,
      "loss": 1.2318,
      "step": 27385
    },
    {
      "epoch": 0.9807509803570469,
      "grad_norm": 1.9923317432403564,
      "learning_rate": 1.9391893981599575e-07,
      "loss": 1.6143,
      "step": 27386
    },
    {
      "epoch": 0.9807867924866153,
      "grad_norm": 1.7062000036239624,
      "learning_rate": 1.9319761402458726e-07,
      "loss": 1.5476,
      "step": 27387
    },
    {
      "epoch": 0.9808226046161835,
      "grad_norm": 2.0269036293029785,
      "learning_rate": 1.924776310049725e-07,
      "loss": 1.3775,
      "step": 27388
    },
    {
      "epoch": 0.9808584167457518,
      "grad_norm": 2.484912395477295,
      "learning_rate": 1.9175899076682158e-07,
      "loss": 1.5035,
      "step": 27389
    },
    {
      "epoch": 0.9808942288753201,
      "grad_norm": 1.5835357904434204,
      "learning_rate": 1.9104169331981558e-07,
      "loss": 1.264,
      "step": 27390
    },
    {
      "epoch": 0.9809300410048883,
      "grad_norm": 1.6880030632019043,
      "learning_rate": 1.9032573867359126e-07,
      "loss": 1.5247,
      "step": 27391
    },
    {
      "epoch": 0.9809658531344566,
      "grad_norm": 1.7247785329818726,
      "learning_rate": 1.8961112683778536e-07,
      "loss": 1.5215,
      "step": 27392
    },
    {
      "epoch": 0.9810016652640249,
      "grad_norm": 1.686508297920227,
      "learning_rate": 1.8889785782202352e-07,
      "loss": 1.4663,
      "step": 27393
    },
    {
      "epoch": 0.9810374773935933,
      "grad_norm": 1.2703536748886108,
      "learning_rate": 1.8818593163589805e-07,
      "loss": 1.281,
      "step": 27394
    },
    {
      "epoch": 0.9810732895231615,
      "grad_norm": 2.6047658920288086,
      "learning_rate": 1.8747534828897905e-07,
      "loss": 1.5477,
      "step": 27395
    },
    {
      "epoch": 0.9811091016527298,
      "grad_norm": 1.490902304649353,
      "learning_rate": 1.867661077908256e-07,
      "loss": 1.2813,
      "step": 27396
    },
    {
      "epoch": 0.9811449137822981,
      "grad_norm": 1.706141710281372,
      "learning_rate": 1.8605821015098556e-07,
      "loss": 1.3684,
      "step": 27397
    },
    {
      "epoch": 0.9811807259118663,
      "grad_norm": 3.6071617603302,
      "learning_rate": 1.853516553789847e-07,
      "loss": 1.553,
      "step": 27398
    },
    {
      "epoch": 0.9812165380414346,
      "grad_norm": 1.1728358268737793,
      "learning_rate": 1.8464644348432647e-07,
      "loss": 1.2748,
      "step": 27399
    },
    {
      "epoch": 0.9812523501710029,
      "grad_norm": 1.6108249425888062,
      "learning_rate": 1.8394257447650332e-07,
      "loss": 1.4349,
      "step": 27400
    },
    {
      "epoch": 0.9812881623005713,
      "grad_norm": 1.6908518075942993,
      "learning_rate": 1.832400483649632e-07,
      "loss": 1.6162,
      "step": 27401
    },
    {
      "epoch": 0.9813239744301395,
      "grad_norm": 1.55814528465271,
      "learning_rate": 1.825388651591875e-07,
      "loss": 1.6483,
      "step": 27402
    },
    {
      "epoch": 0.9813597865597078,
      "grad_norm": 1.2762830257415771,
      "learning_rate": 1.8183902486859083e-07,
      "loss": 1.5125,
      "step": 27403
    },
    {
      "epoch": 0.9813955986892761,
      "grad_norm": 1.6804869174957275,
      "learning_rate": 1.8114052750259902e-07,
      "loss": 1.4573,
      "step": 27404
    },
    {
      "epoch": 0.9814314108188443,
      "grad_norm": 1.4806815385818481,
      "learning_rate": 1.8044337307059345e-07,
      "loss": 1.2894,
      "step": 27405
    },
    {
      "epoch": 0.9814672229484126,
      "grad_norm": 1.408532738685608,
      "learning_rate": 1.7974756158196658e-07,
      "loss": 1.4177,
      "step": 27406
    },
    {
      "epoch": 0.9815030350779809,
      "grad_norm": 1.4657418727874756,
      "learning_rate": 1.7905309304608876e-07,
      "loss": 1.3174,
      "step": 27407
    },
    {
      "epoch": 0.9815388472075492,
      "grad_norm": 1.338862657546997,
      "learning_rate": 1.7835996747228578e-07,
      "loss": 1.3876,
      "step": 27408
    },
    {
      "epoch": 0.9815746593371175,
      "grad_norm": 1.70529305934906,
      "learning_rate": 1.7766818486988357e-07,
      "loss": 1.5092,
      "step": 27409
    },
    {
      "epoch": 0.9816104714666858,
      "grad_norm": 1.8063136339187622,
      "learning_rate": 1.769777452481969e-07,
      "loss": 1.3785,
      "step": 27410
    },
    {
      "epoch": 0.981646283596254,
      "grad_norm": 1.7511066198349,
      "learning_rate": 1.7628864861651827e-07,
      "loss": 1.334,
      "step": 27411
    },
    {
      "epoch": 0.9816820957258223,
      "grad_norm": 2.083026647567749,
      "learning_rate": 1.7560089498410704e-07,
      "loss": 1.3795,
      "step": 27412
    },
    {
      "epoch": 0.9817179078553906,
      "grad_norm": 1.639681339263916,
      "learning_rate": 1.749144843602224e-07,
      "loss": 1.3635,
      "step": 27413
    },
    {
      "epoch": 0.9817537199849589,
      "grad_norm": 1.5684409141540527,
      "learning_rate": 1.7422941675410143e-07,
      "loss": 1.3592,
      "step": 27414
    },
    {
      "epoch": 0.9817895321145272,
      "grad_norm": 2.3734753131866455,
      "learning_rate": 1.7354569217494788e-07,
      "loss": 1.4517,
      "step": 27415
    },
    {
      "epoch": 0.9818253442440955,
      "grad_norm": 1.4742026329040527,
      "learning_rate": 1.728633106319766e-07,
      "loss": 1.4555,
      "step": 27416
    },
    {
      "epoch": 0.9818611563736638,
      "grad_norm": 1.7053145170211792,
      "learning_rate": 1.721822721343691e-07,
      "loss": 1.4405,
      "step": 27417
    },
    {
      "epoch": 0.981896968503232,
      "grad_norm": 1.586531400680542,
      "learning_rate": 1.7150257669127367e-07,
      "loss": 1.1004,
      "step": 27418
    },
    {
      "epoch": 0.9819327806328003,
      "grad_norm": 1.7251858711242676,
      "learning_rate": 1.7082422431183853e-07,
      "loss": 1.5102,
      "step": 27419
    },
    {
      "epoch": 0.9819685927623686,
      "grad_norm": 1.6602956056594849,
      "learning_rate": 1.701472150051897e-07,
      "loss": 1.6386,
      "step": 27420
    },
    {
      "epoch": 0.9820044048919369,
      "grad_norm": 1.4878696203231812,
      "learning_rate": 1.6947154878045324e-07,
      "loss": 1.6375,
      "step": 27421
    },
    {
      "epoch": 0.9820402170215052,
      "grad_norm": 2.1307990550994873,
      "learning_rate": 1.6879722564669964e-07,
      "loss": 1.5758,
      "step": 27422
    },
    {
      "epoch": 0.9820760291510735,
      "grad_norm": 1.5953096151351929,
      "learning_rate": 1.6812424561299943e-07,
      "loss": 1.3952,
      "step": 27423
    },
    {
      "epoch": 0.9821118412806418,
      "grad_norm": 1.6359837055206299,
      "learning_rate": 1.6745260868841207e-07,
      "loss": 1.5067,
      "step": 27424
    },
    {
      "epoch": 0.98214765341021,
      "grad_norm": 1.5047430992126465,
      "learning_rate": 1.667823148819858e-07,
      "loss": 1.2965,
      "step": 27425
    },
    {
      "epoch": 0.9821834655397783,
      "grad_norm": 1.7252825498580933,
      "learning_rate": 1.661133642027246e-07,
      "loss": 1.1945,
      "step": 27426
    },
    {
      "epoch": 0.9822192776693466,
      "grad_norm": 1.6850330829620361,
      "learning_rate": 1.6544575665963236e-07,
      "loss": 1.2893,
      "step": 27427
    },
    {
      "epoch": 0.9822550897989148,
      "grad_norm": 1.569167971611023,
      "learning_rate": 1.6477949226167967e-07,
      "loss": 1.4836,
      "step": 27428
    },
    {
      "epoch": 0.9822909019284832,
      "grad_norm": 1.9602831602096558,
      "learning_rate": 1.6411457101784822e-07,
      "loss": 1.337,
      "step": 27429
    },
    {
      "epoch": 0.9823267140580515,
      "grad_norm": 1.4004954099655151,
      "learning_rate": 1.6345099293708644e-07,
      "loss": 1.324,
      "step": 27430
    },
    {
      "epoch": 0.9823625261876198,
      "grad_norm": 3.534090042114258,
      "learning_rate": 1.627887580282983e-07,
      "loss": 1.4684,
      "step": 27431
    },
    {
      "epoch": 0.982398338317188,
      "grad_norm": 1.5962355136871338,
      "learning_rate": 1.6212786630041e-07,
      "loss": 1.2013,
      "step": 27432
    },
    {
      "epoch": 0.9824341504467563,
      "grad_norm": 1.6603145599365234,
      "learning_rate": 1.6146831776231442e-07,
      "loss": 1.3004,
      "step": 27433
    },
    {
      "epoch": 0.9824699625763246,
      "grad_norm": 1.6587244272232056,
      "learning_rate": 1.6081011242287115e-07,
      "loss": 1.5779,
      "step": 27434
    },
    {
      "epoch": 0.9825057747058928,
      "grad_norm": 1.366642951965332,
      "learning_rate": 1.6015325029095084e-07,
      "loss": 1.3437,
      "step": 27435
    },
    {
      "epoch": 0.9825415868354612,
      "grad_norm": 1.6663930416107178,
      "learning_rate": 1.5949773137537982e-07,
      "loss": 1.0643,
      "step": 27436
    },
    {
      "epoch": 0.9825773989650295,
      "grad_norm": 1.70339834690094,
      "learning_rate": 1.588435556849843e-07,
      "loss": 1.2619,
      "step": 27437
    },
    {
      "epoch": 0.9826132110945978,
      "grad_norm": 1.9648959636688232,
      "learning_rate": 1.5819072322856842e-07,
      "loss": 1.2753,
      "step": 27438
    },
    {
      "epoch": 0.982649023224166,
      "grad_norm": 1.460438847541809,
      "learning_rate": 1.575392340149029e-07,
      "loss": 1.3353,
      "step": 27439
    },
    {
      "epoch": 0.9826848353537343,
      "grad_norm": 1.61161208152771,
      "learning_rate": 1.5688908805275848e-07,
      "loss": 1.6739,
      "step": 27440
    },
    {
      "epoch": 0.9827206474833026,
      "grad_norm": 1.4673449993133545,
      "learning_rate": 1.5624028535088375e-07,
      "loss": 1.4056,
      "step": 27441
    },
    {
      "epoch": 0.9827564596128708,
      "grad_norm": 1.7691820859909058,
      "learning_rate": 1.5559282591801617e-07,
      "loss": 1.679,
      "step": 27442
    },
    {
      "epoch": 0.9827922717424392,
      "grad_norm": 1.7717576026916504,
      "learning_rate": 1.5494670976284875e-07,
      "loss": 1.574,
      "step": 27443
    },
    {
      "epoch": 0.9828280838720075,
      "grad_norm": 1.412382960319519,
      "learning_rate": 1.543019368940857e-07,
      "loss": 1.3202,
      "step": 27444
    },
    {
      "epoch": 0.9828638960015758,
      "grad_norm": 1.745245099067688,
      "learning_rate": 1.5365850732039778e-07,
      "loss": 1.6919,
      "step": 27445
    },
    {
      "epoch": 0.982899708131144,
      "grad_norm": 1.7831408977508545,
      "learning_rate": 1.5301642105043368e-07,
      "loss": 1.3777,
      "step": 27446
    },
    {
      "epoch": 0.9829355202607123,
      "grad_norm": 1.3212556838989258,
      "learning_rate": 1.5237567809285314e-07,
      "loss": 1.1704,
      "step": 27447
    },
    {
      "epoch": 0.9829713323902806,
      "grad_norm": 1.4446337223052979,
      "learning_rate": 1.5173627845624927e-07,
      "loss": 1.2402,
      "step": 27448
    },
    {
      "epoch": 0.9830071445198488,
      "grad_norm": 1.6273493766784668,
      "learning_rate": 1.510982221492485e-07,
      "loss": 1.616,
      "step": 27449
    },
    {
      "epoch": 0.9830429566494172,
      "grad_norm": 1.3335165977478027,
      "learning_rate": 1.5046150918042178e-07,
      "loss": 1.4248,
      "step": 27450
    },
    {
      "epoch": 0.9830787687789855,
      "grad_norm": 1.4034610986709595,
      "learning_rate": 1.4982613955834001e-07,
      "loss": 1.2438,
      "step": 27451
    },
    {
      "epoch": 0.9831145809085537,
      "grad_norm": 1.4101903438568115,
      "learning_rate": 1.4919211329156302e-07,
      "loss": 1.3619,
      "step": 27452
    },
    {
      "epoch": 0.983150393038122,
      "grad_norm": 2.1397769451141357,
      "learning_rate": 1.4855943038858399e-07,
      "loss": 1.52,
      "step": 27453
    },
    {
      "epoch": 0.9831862051676903,
      "grad_norm": 1.4390918016433716,
      "learning_rate": 1.4792809085795166e-07,
      "loss": 1.2698,
      "step": 27454
    },
    {
      "epoch": 0.9832220172972586,
      "grad_norm": 1.4676564931869507,
      "learning_rate": 1.4729809470814815e-07,
      "loss": 1.3267,
      "step": 27455
    },
    {
      "epoch": 0.9832578294268268,
      "grad_norm": 2.3807873725891113,
      "learning_rate": 1.4666944194764443e-07,
      "loss": 1.358,
      "step": 27456
    },
    {
      "epoch": 0.9832936415563952,
      "grad_norm": 1.7312041521072388,
      "learning_rate": 1.4604213258491152e-07,
      "loss": 1.5644,
      "step": 27457
    },
    {
      "epoch": 0.9833294536859635,
      "grad_norm": 1.8110259771347046,
      "learning_rate": 1.4541616662836488e-07,
      "loss": 1.3413,
      "step": 27458
    },
    {
      "epoch": 0.9833652658155317,
      "grad_norm": 1.3456915616989136,
      "learning_rate": 1.4479154408645335e-07,
      "loss": 1.2207,
      "step": 27459
    },
    {
      "epoch": 0.9834010779451,
      "grad_norm": 1.6277025938034058,
      "learning_rate": 1.441682649675591e-07,
      "loss": 1.4059,
      "step": 27460
    },
    {
      "epoch": 0.9834368900746683,
      "grad_norm": 1.6478482484817505,
      "learning_rate": 1.435463292800754e-07,
      "loss": 1.4681,
      "step": 27461
    },
    {
      "epoch": 0.9834727022042365,
      "grad_norm": 2.0339913368225098,
      "learning_rate": 1.4292573703237333e-07,
      "loss": 1.4551,
      "step": 27462
    },
    {
      "epoch": 0.9835085143338048,
      "grad_norm": 2.6009304523468018,
      "learning_rate": 1.423064882328018e-07,
      "loss": 1.6065,
      "step": 27463
    },
    {
      "epoch": 0.9835443264633732,
      "grad_norm": 2.1348941326141357,
      "learning_rate": 1.4168858288968745e-07,
      "loss": 1.4424,
      "step": 27464
    },
    {
      "epoch": 0.9835801385929415,
      "grad_norm": 2.100189208984375,
      "learning_rate": 1.4107202101134588e-07,
      "loss": 1.8112,
      "step": 27465
    },
    {
      "epoch": 0.9836159507225097,
      "grad_norm": 1.6087077856063843,
      "learning_rate": 1.404568026060704e-07,
      "loss": 1.2046,
      "step": 27466
    },
    {
      "epoch": 0.983651762852078,
      "grad_norm": 1.3044302463531494,
      "learning_rate": 1.3984292768213225e-07,
      "loss": 1.7624,
      "step": 27467
    },
    {
      "epoch": 0.9836875749816463,
      "grad_norm": 1.8005906343460083,
      "learning_rate": 1.3923039624780255e-07,
      "loss": 1.6117,
      "step": 27468
    },
    {
      "epoch": 0.9837233871112145,
      "grad_norm": 1.8799329996109009,
      "learning_rate": 1.3861920831131914e-07,
      "loss": 1.4647,
      "step": 27469
    },
    {
      "epoch": 0.9837591992407828,
      "grad_norm": 1.3361802101135254,
      "learning_rate": 1.380093638808977e-07,
      "loss": 1.2524,
      "step": 27470
    },
    {
      "epoch": 0.9837950113703512,
      "grad_norm": 1.5945254564285278,
      "learning_rate": 1.3740086296475385e-07,
      "loss": 1.6172,
      "step": 27471
    },
    {
      "epoch": 0.9838308234999195,
      "grad_norm": 1.4752191305160522,
      "learning_rate": 1.3679370557106997e-07,
      "loss": 1.24,
      "step": 27472
    },
    {
      "epoch": 0.9838666356294877,
      "grad_norm": 1.4537187814712524,
      "learning_rate": 1.3618789170800618e-07,
      "loss": 1.3304,
      "step": 27473
    },
    {
      "epoch": 0.983902447759056,
      "grad_norm": 1.3634134531021118,
      "learning_rate": 1.355834213837226e-07,
      "loss": 1.4036,
      "step": 27474
    },
    {
      "epoch": 0.9839382598886243,
      "grad_norm": 1.8052500486373901,
      "learning_rate": 1.349802946063461e-07,
      "loss": 1.4053,
      "step": 27475
    },
    {
      "epoch": 0.9839740720181925,
      "grad_norm": 1.618231177330017,
      "learning_rate": 1.3437851138399237e-07,
      "loss": 1.4643,
      "step": 27476
    },
    {
      "epoch": 0.9840098841477608,
      "grad_norm": 1.6881462335586548,
      "learning_rate": 1.33778071724755e-07,
      "loss": 1.4012,
      "step": 27477
    },
    {
      "epoch": 0.9840456962773292,
      "grad_norm": 1.7199472188949585,
      "learning_rate": 1.3317897563671633e-07,
      "loss": 1.4276,
      "step": 27478
    },
    {
      "epoch": 0.9840815084068975,
      "grad_norm": 1.4663841724395752,
      "learning_rate": 1.3258122312793663e-07,
      "loss": 1.4161,
      "step": 27479
    },
    {
      "epoch": 0.9841173205364657,
      "grad_norm": 1.3081108331680298,
      "learning_rate": 1.3198481420646504e-07,
      "loss": 0.9065,
      "step": 27480
    },
    {
      "epoch": 0.984153132666034,
      "grad_norm": 1.7379041910171509,
      "learning_rate": 1.313897488803062e-07,
      "loss": 1.4373,
      "step": 27481
    },
    {
      "epoch": 0.9841889447956023,
      "grad_norm": 1.5058616399765015,
      "learning_rate": 1.3079602715748706e-07,
      "loss": 1.3416,
      "step": 27482
    },
    {
      "epoch": 0.9842247569251705,
      "grad_norm": 1.4991620779037476,
      "learning_rate": 1.3020364904597903e-07,
      "loss": 1.4064,
      "step": 27483
    },
    {
      "epoch": 0.9842605690547388,
      "grad_norm": 2.0297024250030518,
      "learning_rate": 1.296126145537646e-07,
      "loss": 1.5304,
      "step": 27484
    },
    {
      "epoch": 0.9842963811843072,
      "grad_norm": 1.7815202474594116,
      "learning_rate": 1.2902292368878188e-07,
      "loss": 1.3314,
      "step": 27485
    },
    {
      "epoch": 0.9843321933138754,
      "grad_norm": 1.5582129955291748,
      "learning_rate": 1.2843457645896895e-07,
      "loss": 1.2736,
      "step": 27486
    },
    {
      "epoch": 0.9843680054434437,
      "grad_norm": 1.263076901435852,
      "learning_rate": 1.278475728722528e-07,
      "loss": 1.4625,
      "step": 27487
    },
    {
      "epoch": 0.984403817573012,
      "grad_norm": 2.047459602355957,
      "learning_rate": 1.27261912936516e-07,
      "loss": 1.7214,
      "step": 27488
    },
    {
      "epoch": 0.9844396297025803,
      "grad_norm": 1.9874540567398071,
      "learning_rate": 1.2667759665964118e-07,
      "loss": 1.1864,
      "step": 27489
    },
    {
      "epoch": 0.9844754418321485,
      "grad_norm": 1.2950546741485596,
      "learning_rate": 1.260946240494998e-07,
      "loss": 1.496,
      "step": 27490
    },
    {
      "epoch": 0.9845112539617168,
      "grad_norm": 1.5544168949127197,
      "learning_rate": 1.255129951139189e-07,
      "loss": 1.14,
      "step": 27491
    },
    {
      "epoch": 0.9845470660912852,
      "grad_norm": 1.3458187580108643,
      "learning_rate": 1.249327098607367e-07,
      "loss": 1.3636,
      "step": 27492
    },
    {
      "epoch": 0.9845828782208534,
      "grad_norm": 1.8006720542907715,
      "learning_rate": 1.2435376829775803e-07,
      "loss": 1.6198,
      "step": 27493
    },
    {
      "epoch": 0.9846186903504217,
      "grad_norm": 1.344710111618042,
      "learning_rate": 1.2377617043276556e-07,
      "loss": 1.2155,
      "step": 27494
    },
    {
      "epoch": 0.98465450247999,
      "grad_norm": 1.3209199905395508,
      "learning_rate": 1.231999162735309e-07,
      "loss": 1.2898,
      "step": 27495
    },
    {
      "epoch": 0.9846903146095582,
      "grad_norm": 1.2484784126281738,
      "learning_rate": 1.2262500582781445e-07,
      "loss": 1.2969,
      "step": 27496
    },
    {
      "epoch": 0.9847261267391265,
      "grad_norm": 1.6258513927459717,
      "learning_rate": 1.2205143910334338e-07,
      "loss": 1.3481,
      "step": 27497
    },
    {
      "epoch": 0.9847619388686948,
      "grad_norm": 2.1914570331573486,
      "learning_rate": 1.2147921610783374e-07,
      "loss": 1.7681,
      "step": 27498
    },
    {
      "epoch": 0.9847977509982632,
      "grad_norm": 1.3298579454421997,
      "learning_rate": 1.209083368490016e-07,
      "loss": 0.9676,
      "step": 27499
    },
    {
      "epoch": 0.9848335631278314,
      "grad_norm": 2.284987211227417,
      "learning_rate": 1.2033880133449638e-07,
      "loss": 1.3902,
      "step": 27500
    },
    {
      "epoch": 0.9848693752573997,
      "grad_norm": 2.084764003753662,
      "learning_rate": 1.1977060957200083e-07,
      "loss": 1.6581,
      "step": 27501
    },
    {
      "epoch": 0.984905187386968,
      "grad_norm": 1.331100344657898,
      "learning_rate": 1.1920376156916435e-07,
      "loss": 1.7012,
      "step": 27502
    },
    {
      "epoch": 0.9849409995165362,
      "grad_norm": 1.537713885307312,
      "learning_rate": 1.1863825733359201e-07,
      "loss": 1.6487,
      "step": 27503
    },
    {
      "epoch": 0.9849768116461045,
      "grad_norm": 1.940346360206604,
      "learning_rate": 1.1807409687291104e-07,
      "loss": 1.5181,
      "step": 27504
    },
    {
      "epoch": 0.9850126237756728,
      "grad_norm": 1.2605899572372437,
      "learning_rate": 1.1751128019470426e-07,
      "loss": 1.3682,
      "step": 27505
    },
    {
      "epoch": 0.9850484359052412,
      "grad_norm": 1.9707603454589844,
      "learning_rate": 1.1694980730654337e-07,
      "loss": 1.4205,
      "step": 27506
    },
    {
      "epoch": 0.9850842480348094,
      "grad_norm": 2.1587588787078857,
      "learning_rate": 1.163896782159779e-07,
      "loss": 1.4563,
      "step": 27507
    },
    {
      "epoch": 0.9851200601643777,
      "grad_norm": 2.022172451019287,
      "learning_rate": 1.1583089293055738e-07,
      "loss": 1.4855,
      "step": 27508
    },
    {
      "epoch": 0.985155872293946,
      "grad_norm": 1.5751782655715942,
      "learning_rate": 1.152734514577869e-07,
      "loss": 1.4265,
      "step": 27509
    },
    {
      "epoch": 0.9851916844235142,
      "grad_norm": 1.86588454246521,
      "learning_rate": 1.1471735380517156e-07,
      "loss": 1.5442,
      "step": 27510
    },
    {
      "epoch": 0.9852274965530825,
      "grad_norm": 1.3863067626953125,
      "learning_rate": 1.141625999801943e-07,
      "loss": 1.5659,
      "step": 27511
    },
    {
      "epoch": 0.9852633086826508,
      "grad_norm": 1.3101366758346558,
      "learning_rate": 1.1360918999030467e-07,
      "loss": 1.3462,
      "step": 27512
    },
    {
      "epoch": 0.9852991208122192,
      "grad_norm": 2.0936365127563477,
      "learning_rate": 1.1305712384297451e-07,
      "loss": 1.4808,
      "step": 27513
    },
    {
      "epoch": 0.9853349329417874,
      "grad_norm": 1.8865522146224976,
      "learning_rate": 1.1250640154560898e-07,
      "loss": 1.5132,
      "step": 27514
    },
    {
      "epoch": 0.9853707450713557,
      "grad_norm": 1.5039201974868774,
      "learning_rate": 1.1195702310561329e-07,
      "loss": 1.5329,
      "step": 27515
    },
    {
      "epoch": 0.985406557200924,
      "grad_norm": 1.6145402193069458,
      "learning_rate": 1.1140898853040372e-07,
      "loss": 1.1151,
      "step": 27516
    },
    {
      "epoch": 0.9854423693304922,
      "grad_norm": 1.550848126411438,
      "learning_rate": 1.1086229782734103e-07,
      "loss": 1.1668,
      "step": 27517
    },
    {
      "epoch": 0.9854781814600605,
      "grad_norm": 1.7086671590805054,
      "learning_rate": 1.1031695100376382e-07,
      "loss": 1.6588,
      "step": 27518
    },
    {
      "epoch": 0.9855139935896288,
      "grad_norm": 1.6773920059204102,
      "learning_rate": 1.0977294806703286e-07,
      "loss": 1.6271,
      "step": 27519
    },
    {
      "epoch": 0.9855498057191971,
      "grad_norm": 1.6794071197509766,
      "learning_rate": 1.0923028902446453e-07,
      "loss": 1.5406,
      "step": 27520
    },
    {
      "epoch": 0.9855856178487654,
      "grad_norm": 1.3817123174667358,
      "learning_rate": 1.0868897388334187e-07,
      "loss": 1.7427,
      "step": 27521
    },
    {
      "epoch": 0.9856214299783337,
      "grad_norm": 1.4821956157684326,
      "learning_rate": 1.0814900265095907e-07,
      "loss": 1.5651,
      "step": 27522
    },
    {
      "epoch": 0.985657242107902,
      "grad_norm": 1.8491703271865845,
      "learning_rate": 1.0761037533457696e-07,
      "loss": 1.5515,
      "step": 27523
    },
    {
      "epoch": 0.9856930542374702,
      "grad_norm": 2.7087044715881348,
      "learning_rate": 1.0707309194145643e-07,
      "loss": 1.3486,
      "step": 27524
    },
    {
      "epoch": 0.9857288663670385,
      "grad_norm": 1.8126064538955688,
      "learning_rate": 1.0653715247881391e-07,
      "loss": 1.4137,
      "step": 27525
    },
    {
      "epoch": 0.9857646784966068,
      "grad_norm": 2.0652072429656982,
      "learning_rate": 1.0600255695385475e-07,
      "loss": 1.2903,
      "step": 27526
    },
    {
      "epoch": 0.9858004906261751,
      "grad_norm": 1.4882497787475586,
      "learning_rate": 1.054693053737843e-07,
      "loss": 1.6498,
      "step": 27527
    },
    {
      "epoch": 0.9858363027557434,
      "grad_norm": 1.413586139678955,
      "learning_rate": 1.049373977457635e-07,
      "loss": 1.3082,
      "step": 27528
    },
    {
      "epoch": 0.9858721148853117,
      "grad_norm": 2.0597503185272217,
      "learning_rate": 1.0440683407695328e-07,
      "loss": 1.4598,
      "step": 27529
    },
    {
      "epoch": 0.98590792701488,
      "grad_norm": 1.3489118814468384,
      "learning_rate": 1.0387761437449239e-07,
      "loss": 1.4809,
      "step": 27530
    },
    {
      "epoch": 0.9859437391444482,
      "grad_norm": 1.6404932737350464,
      "learning_rate": 1.0334973864550845e-07,
      "loss": 1.1483,
      "step": 27531
    },
    {
      "epoch": 0.9859795512740165,
      "grad_norm": 1.574805736541748,
      "learning_rate": 1.0282320689708469e-07,
      "loss": 1.4865,
      "step": 27532
    },
    {
      "epoch": 0.9860153634035848,
      "grad_norm": 1.9894275665283203,
      "learning_rate": 1.0229801913632653e-07,
      "loss": 1.5721,
      "step": 27533
    },
    {
      "epoch": 0.986051175533153,
      "grad_norm": 1.813362717628479,
      "learning_rate": 1.0177417537028389e-07,
      "loss": 1.6005,
      "step": 27534
    },
    {
      "epoch": 0.9860869876627214,
      "grad_norm": 1.4725996255874634,
      "learning_rate": 1.0125167560601778e-07,
      "loss": 1.4278,
      "step": 27535
    },
    {
      "epoch": 0.9861227997922897,
      "grad_norm": 1.5005663633346558,
      "learning_rate": 1.0073051985054482e-07,
      "loss": 1.2329,
      "step": 27536
    },
    {
      "epoch": 0.9861586119218579,
      "grad_norm": 1.8271814584732056,
      "learning_rate": 1.0021070811088162e-07,
      "loss": 1.3513,
      "step": 27537
    },
    {
      "epoch": 0.9861944240514262,
      "grad_norm": 1.5297101736068726,
      "learning_rate": 9.969224039403369e-08,
      "loss": 1.5614,
      "step": 27538
    },
    {
      "epoch": 0.9862302361809945,
      "grad_norm": 1.3022998571395874,
      "learning_rate": 9.9175116706951e-08,
      "loss": 1.0406,
      "step": 27539
    },
    {
      "epoch": 0.9862660483105627,
      "grad_norm": 1.8046320676803589,
      "learning_rate": 9.86593370566058e-08,
      "loss": 1.3661,
      "step": 27540
    },
    {
      "epoch": 0.986301860440131,
      "grad_norm": 1.635322093963623,
      "learning_rate": 9.814490144993693e-08,
      "loss": 1.5341,
      "step": 27541
    },
    {
      "epoch": 0.9863376725696994,
      "grad_norm": 1.8201141357421875,
      "learning_rate": 9.763180989386112e-08,
      "loss": 1.3124,
      "step": 27542
    },
    {
      "epoch": 0.9863734846992677,
      "grad_norm": 1.5187281370162964,
      "learning_rate": 9.712006239529503e-08,
      "loss": 1.2908,
      "step": 27543
    },
    {
      "epoch": 0.9864092968288359,
      "grad_norm": 1.995166540145874,
      "learning_rate": 9.660965896111095e-08,
      "loss": 1.5077,
      "step": 27544
    },
    {
      "epoch": 0.9864451089584042,
      "grad_norm": 1.9507112503051758,
      "learning_rate": 9.610059959817008e-08,
      "loss": 1.1559,
      "step": 27545
    },
    {
      "epoch": 0.9864809210879725,
      "grad_norm": 1.3299881219863892,
      "learning_rate": 9.559288431333357e-08,
      "loss": 1.5751,
      "step": 27546
    },
    {
      "epoch": 0.9865167332175407,
      "grad_norm": 1.6809051036834717,
      "learning_rate": 9.50865131134182e-08,
      "loss": 1.2238,
      "step": 27547
    },
    {
      "epoch": 0.986552545347109,
      "grad_norm": 1.5759165287017822,
      "learning_rate": 9.458148600525185e-08,
      "loss": 1.2708,
      "step": 27548
    },
    {
      "epoch": 0.9865883574766774,
      "grad_norm": 1.5264739990234375,
      "learning_rate": 9.407780299562908e-08,
      "loss": 1.8756,
      "step": 27549
    },
    {
      "epoch": 0.9866241696062457,
      "grad_norm": 1.4205260276794434,
      "learning_rate": 9.357546409132223e-08,
      "loss": 1.5195,
      "step": 27550
    },
    {
      "epoch": 0.9866599817358139,
      "grad_norm": 1.4577481746673584,
      "learning_rate": 9.307446929908148e-08,
      "loss": 1.8085,
      "step": 27551
    },
    {
      "epoch": 0.9866957938653822,
      "grad_norm": 1.9873889684677124,
      "learning_rate": 9.257481862564587e-08,
      "loss": 1.3263,
      "step": 27552
    },
    {
      "epoch": 0.9867316059949505,
      "grad_norm": 1.4352340698242188,
      "learning_rate": 9.207651207775448e-08,
      "loss": 1.5041,
      "step": 27553
    },
    {
      "epoch": 0.9867674181245187,
      "grad_norm": 1.8972949981689453,
      "learning_rate": 9.157954966210192e-08,
      "loss": 1.5607,
      "step": 27554
    },
    {
      "epoch": 0.986803230254087,
      "grad_norm": 2.068582057952881,
      "learning_rate": 9.108393138536064e-08,
      "loss": 1.3331,
      "step": 27555
    },
    {
      "epoch": 0.9868390423836554,
      "grad_norm": 1.2161344289779663,
      "learning_rate": 9.05896572542253e-08,
      "loss": 1.1665,
      "step": 27556
    },
    {
      "epoch": 0.9868748545132237,
      "grad_norm": 2.187922239303589,
      "learning_rate": 9.009672727533503e-08,
      "loss": 1.7574,
      "step": 27557
    },
    {
      "epoch": 0.9869106666427919,
      "grad_norm": 1.6480915546417236,
      "learning_rate": 8.960514145530674e-08,
      "loss": 1.1512,
      "step": 27558
    },
    {
      "epoch": 0.9869464787723602,
      "grad_norm": 1.5355759859085083,
      "learning_rate": 8.911489980076848e-08,
      "loss": 1.5087,
      "step": 27559
    },
    {
      "epoch": 0.9869822909019285,
      "grad_norm": 2.0616066455841064,
      "learning_rate": 8.862600231832607e-08,
      "loss": 1.545,
      "step": 27560
    },
    {
      "epoch": 0.9870181030314967,
      "grad_norm": 1.3449019193649292,
      "learning_rate": 8.813844901452984e-08,
      "loss": 1.6202,
      "step": 27561
    },
    {
      "epoch": 0.987053915161065,
      "grad_norm": 1.6275995969772339,
      "learning_rate": 8.765223989596338e-08,
      "loss": 1.4442,
      "step": 27562
    },
    {
      "epoch": 0.9870897272906334,
      "grad_norm": 1.3992468118667603,
      "learning_rate": 8.716737496915483e-08,
      "loss": 1.2691,
      "step": 27563
    },
    {
      "epoch": 0.9871255394202016,
      "grad_norm": 1.3481507301330566,
      "learning_rate": 8.66838542406212e-08,
      "loss": 1.29,
      "step": 27564
    },
    {
      "epoch": 0.9871613515497699,
      "grad_norm": 1.4527891874313354,
      "learning_rate": 8.620167771689058e-08,
      "loss": 1.2298,
      "step": 27565
    },
    {
      "epoch": 0.9871971636793382,
      "grad_norm": 1.4806122779846191,
      "learning_rate": 8.572084540443558e-08,
      "loss": 1.4566,
      "step": 27566
    },
    {
      "epoch": 0.9872329758089065,
      "grad_norm": 1.745713472366333,
      "learning_rate": 8.524135730971772e-08,
      "loss": 1.4321,
      "step": 27567
    },
    {
      "epoch": 0.9872687879384747,
      "grad_norm": 1.633641242980957,
      "learning_rate": 8.476321343920957e-08,
      "loss": 1.3979,
      "step": 27568
    },
    {
      "epoch": 0.987304600068043,
      "grad_norm": 1.8932297229766846,
      "learning_rate": 8.428641379931713e-08,
      "loss": 1.4096,
      "step": 27569
    },
    {
      "epoch": 0.9873404121976114,
      "grad_norm": 1.2509734630584717,
      "learning_rate": 8.381095839647967e-08,
      "loss": 1.2951,
      "step": 27570
    },
    {
      "epoch": 0.9873762243271796,
      "grad_norm": 1.5911204814910889,
      "learning_rate": 8.333684723708102e-08,
      "loss": 1.2692,
      "step": 27571
    },
    {
      "epoch": 0.9874120364567479,
      "grad_norm": 1.4687579870224,
      "learning_rate": 8.286408032749382e-08,
      "loss": 1.694,
      "step": 27572
    },
    {
      "epoch": 0.9874478485863162,
      "grad_norm": 1.4162245988845825,
      "learning_rate": 8.239265767410187e-08,
      "loss": 1.3073,
      "step": 27573
    },
    {
      "epoch": 0.9874836607158844,
      "grad_norm": 1.4807811975479126,
      "learning_rate": 8.192257928322233e-08,
      "loss": 1.3811,
      "step": 27574
    },
    {
      "epoch": 0.9875194728454527,
      "grad_norm": 2.3424911499023438,
      "learning_rate": 8.145384516118349e-08,
      "loss": 1.4198,
      "step": 27575
    },
    {
      "epoch": 0.987555284975021,
      "grad_norm": 1.6639134883880615,
      "learning_rate": 8.098645531431359e-08,
      "loss": 1.3878,
      "step": 27576
    },
    {
      "epoch": 0.9875910971045894,
      "grad_norm": 1.885317087173462,
      "learning_rate": 8.052040974887432e-08,
      "loss": 1.669,
      "step": 27577
    },
    {
      "epoch": 0.9876269092341576,
      "grad_norm": 1.4320241212844849,
      "learning_rate": 8.005570847113841e-08,
      "loss": 1.1812,
      "step": 27578
    },
    {
      "epoch": 0.9876627213637259,
      "grad_norm": 1.843773603439331,
      "learning_rate": 7.959235148737865e-08,
      "loss": 1.4705,
      "step": 27579
    },
    {
      "epoch": 0.9876985334932942,
      "grad_norm": 1.7358323335647583,
      "learning_rate": 7.913033880381226e-08,
      "loss": 1.5676,
      "step": 27580
    },
    {
      "epoch": 0.9877343456228624,
      "grad_norm": 1.5030070543289185,
      "learning_rate": 7.866967042665651e-08,
      "loss": 1.2817,
      "step": 27581
    },
    {
      "epoch": 0.9877701577524307,
      "grad_norm": 2.4306750297546387,
      "learning_rate": 7.821034636211755e-08,
      "loss": 1.5123,
      "step": 27582
    },
    {
      "epoch": 0.987805969881999,
      "grad_norm": 2.3857903480529785,
      "learning_rate": 7.77523666163571e-08,
      "loss": 1.4726,
      "step": 27583
    },
    {
      "epoch": 0.9878417820115674,
      "grad_norm": 1.7818009853363037,
      "learning_rate": 7.729573119555911e-08,
      "loss": 1.5097,
      "step": 27584
    },
    {
      "epoch": 0.9878775941411356,
      "grad_norm": 1.879576325416565,
      "learning_rate": 7.684044010585201e-08,
      "loss": 1.3873,
      "step": 27585
    },
    {
      "epoch": 0.9879134062707039,
      "grad_norm": 1.4474228620529175,
      "learning_rate": 7.638649335336423e-08,
      "loss": 1.5264,
      "step": 27586
    },
    {
      "epoch": 0.9879492184002722,
      "grad_norm": 1.6005494594573975,
      "learning_rate": 7.593389094420201e-08,
      "loss": 1.217,
      "step": 27587
    },
    {
      "epoch": 0.9879850305298404,
      "grad_norm": 1.4687864780426025,
      "learning_rate": 7.548263288446045e-08,
      "loss": 1.5201,
      "step": 27588
    },
    {
      "epoch": 0.9880208426594087,
      "grad_norm": 1.9801404476165771,
      "learning_rate": 7.503271918020138e-08,
      "loss": 1.2031,
      "step": 27589
    },
    {
      "epoch": 0.988056654788977,
      "grad_norm": 2.481457233428955,
      "learning_rate": 7.458414983748663e-08,
      "loss": 1.7782,
      "step": 27590
    },
    {
      "epoch": 0.9880924669185454,
      "grad_norm": 1.6557172536849976,
      "learning_rate": 7.41369248623447e-08,
      "loss": 1.2114,
      "step": 27591
    },
    {
      "epoch": 0.9881282790481136,
      "grad_norm": 2.5019419193267822,
      "learning_rate": 7.369104426080409e-08,
      "loss": 1.3993,
      "step": 27592
    },
    {
      "epoch": 0.9881640911776819,
      "grad_norm": 1.443387508392334,
      "learning_rate": 7.324650803884891e-08,
      "loss": 1.6107,
      "step": 27593
    },
    {
      "epoch": 0.9881999033072502,
      "grad_norm": 1.4562711715698242,
      "learning_rate": 7.280331620246328e-08,
      "loss": 1.4164,
      "step": 27594
    },
    {
      "epoch": 0.9882357154368184,
      "grad_norm": 1.7285951375961304,
      "learning_rate": 7.236146875762017e-08,
      "loss": 1.7147,
      "step": 27595
    },
    {
      "epoch": 0.9882715275663867,
      "grad_norm": 1.6666792631149292,
      "learning_rate": 7.19209657102482e-08,
      "loss": 1.2269,
      "step": 27596
    },
    {
      "epoch": 0.988307339695955,
      "grad_norm": 1.4787660837173462,
      "learning_rate": 7.148180706628704e-08,
      "loss": 1.088,
      "step": 27597
    },
    {
      "epoch": 0.9883431518255233,
      "grad_norm": 1.591696858406067,
      "learning_rate": 7.104399283163199e-08,
      "loss": 1.429,
      "step": 27598
    },
    {
      "epoch": 0.9883789639550916,
      "grad_norm": 1.709412932395935,
      "learning_rate": 7.060752301218942e-08,
      "loss": 1.2527,
      "step": 27599
    },
    {
      "epoch": 0.9884147760846599,
      "grad_norm": 1.553296446800232,
      "learning_rate": 7.017239761381022e-08,
      "loss": 1.4105,
      "step": 27600
    },
    {
      "epoch": 0.9884505882142282,
      "grad_norm": 1.3799681663513184,
      "learning_rate": 6.973861664237857e-08,
      "loss": 1.441,
      "step": 27601
    },
    {
      "epoch": 0.9884864003437964,
      "grad_norm": 1.5649609565734863,
      "learning_rate": 6.930618010370094e-08,
      "loss": 1.4126,
      "step": 27602
    },
    {
      "epoch": 0.9885222124733647,
      "grad_norm": 1.5194275379180908,
      "learning_rate": 6.887508800361708e-08,
      "loss": 1.1752,
      "step": 27603
    },
    {
      "epoch": 0.988558024602933,
      "grad_norm": 1.4459376335144043,
      "learning_rate": 6.844534034791128e-08,
      "loss": 1.2249,
      "step": 27604
    },
    {
      "epoch": 0.9885938367325013,
      "grad_norm": 1.5532654523849487,
      "learning_rate": 6.801693714236779e-08,
      "loss": 1.541,
      "step": 27605
    },
    {
      "epoch": 0.9886296488620696,
      "grad_norm": 1.2949588298797607,
      "learning_rate": 6.758987839275976e-08,
      "loss": 1.4817,
      "step": 27606
    },
    {
      "epoch": 0.9886654609916379,
      "grad_norm": 1.7721437215805054,
      "learning_rate": 6.716416410481596e-08,
      "loss": 1.6125,
      "step": 27607
    },
    {
      "epoch": 0.9887012731212061,
      "grad_norm": 1.6902053356170654,
      "learning_rate": 6.673979428428733e-08,
      "loss": 1.4175,
      "step": 27608
    },
    {
      "epoch": 0.9887370852507744,
      "grad_norm": 1.6286303997039795,
      "learning_rate": 6.631676893685823e-08,
      "loss": 1.6024,
      "step": 27609
    },
    {
      "epoch": 0.9887728973803427,
      "grad_norm": 1.3310688734054565,
      "learning_rate": 6.589508806823518e-08,
      "loss": 1.2111,
      "step": 27610
    },
    {
      "epoch": 0.988808709509911,
      "grad_norm": 2.158609628677368,
      "learning_rate": 6.547475168409145e-08,
      "loss": 1.1975,
      "step": 27611
    },
    {
      "epoch": 0.9888445216394793,
      "grad_norm": 1.4116407632827759,
      "learning_rate": 6.505575979007805e-08,
      "loss": 1.2926,
      "step": 27612
    },
    {
      "epoch": 0.9888803337690476,
      "grad_norm": 1.7177308797836304,
      "learning_rate": 6.463811239183492e-08,
      "loss": 1.272,
      "step": 27613
    },
    {
      "epoch": 0.9889161458986159,
      "grad_norm": 1.2889899015426636,
      "learning_rate": 6.42218094949798e-08,
      "loss": 1.3474,
      "step": 27614
    },
    {
      "epoch": 0.9889519580281841,
      "grad_norm": 2.355888605117798,
      "learning_rate": 6.38068511051082e-08,
      "loss": 1.6049,
      "step": 27615
    },
    {
      "epoch": 0.9889877701577524,
      "grad_norm": 1.743425726890564,
      "learning_rate": 6.339323722780455e-08,
      "loss": 1.738,
      "step": 27616
    },
    {
      "epoch": 0.9890235822873207,
      "grad_norm": 1.74456787109375,
      "learning_rate": 6.298096786864216e-08,
      "loss": 1.4269,
      "step": 27617
    },
    {
      "epoch": 0.989059394416889,
      "grad_norm": 1.743873119354248,
      "learning_rate": 6.257004303316106e-08,
      "loss": 1.6638,
      "step": 27618
    },
    {
      "epoch": 0.9890952065464573,
      "grad_norm": 2.2863898277282715,
      "learning_rate": 6.216046272687904e-08,
      "loss": 1.3663,
      "step": 27619
    },
    {
      "epoch": 0.9891310186760256,
      "grad_norm": 2.197343111038208,
      "learning_rate": 6.1752226955325e-08,
      "loss": 1.6005,
      "step": 27620
    },
    {
      "epoch": 0.9891668308055939,
      "grad_norm": 1.486518144607544,
      "learning_rate": 6.134533572398349e-08,
      "loss": 1.5046,
      "step": 27621
    },
    {
      "epoch": 0.9892026429351621,
      "grad_norm": 1.5760600566864014,
      "learning_rate": 6.093978903833897e-08,
      "loss": 1.3962,
      "step": 27622
    },
    {
      "epoch": 0.9892384550647304,
      "grad_norm": 1.749154806137085,
      "learning_rate": 6.053558690382045e-08,
      "loss": 1.4331,
      "step": 27623
    },
    {
      "epoch": 0.9892742671942987,
      "grad_norm": 1.6556254625320435,
      "learning_rate": 6.013272932590131e-08,
      "loss": 1.5677,
      "step": 27624
    },
    {
      "epoch": 0.9893100793238669,
      "grad_norm": 1.8295544385910034,
      "learning_rate": 5.973121630996615e-08,
      "loss": 1.1101,
      "step": 27625
    },
    {
      "epoch": 0.9893458914534353,
      "grad_norm": 1.4694961309432983,
      "learning_rate": 5.9331047861443944e-08,
      "loss": 1.3744,
      "step": 27626
    },
    {
      "epoch": 0.9893817035830036,
      "grad_norm": 1.7429505586624146,
      "learning_rate": 5.893222398569709e-08,
      "loss": 1.4978,
      "step": 27627
    },
    {
      "epoch": 0.9894175157125719,
      "grad_norm": 1.6086716651916504,
      "learning_rate": 5.8534744688110156e-08,
      "loss": 1.007,
      "step": 27628
    },
    {
      "epoch": 0.9894533278421401,
      "grad_norm": 1.5418668985366821,
      "learning_rate": 5.8138609974023316e-08,
      "loss": 1.2586,
      "step": 27629
    },
    {
      "epoch": 0.9894891399717084,
      "grad_norm": 1.3214435577392578,
      "learning_rate": 5.774381984876565e-08,
      "loss": 1.4815,
      "step": 27630
    },
    {
      "epoch": 0.9895249521012767,
      "grad_norm": 1.4706875085830688,
      "learning_rate": 5.735037431765511e-08,
      "loss": 1.1197,
      "step": 27631
    },
    {
      "epoch": 0.9895607642308449,
      "grad_norm": 1.5423040390014648,
      "learning_rate": 5.6958273385965264e-08,
      "loss": 1.1352,
      "step": 27632
    },
    {
      "epoch": 0.9895965763604133,
      "grad_norm": 1.6856582164764404,
      "learning_rate": 5.656751705899188e-08,
      "loss": 1.5453,
      "step": 27633
    },
    {
      "epoch": 0.9896323884899816,
      "grad_norm": 1.7213730812072754,
      "learning_rate": 5.617810534198631e-08,
      "loss": 1.136,
      "step": 27634
    },
    {
      "epoch": 0.9896682006195499,
      "grad_norm": 2.0866880416870117,
      "learning_rate": 5.57900382401777e-08,
      "loss": 1.1906,
      "step": 27635
    },
    {
      "epoch": 0.9897040127491181,
      "grad_norm": 2.5806922912597656,
      "learning_rate": 5.540331575880631e-08,
      "loss": 1.7527,
      "step": 27636
    },
    {
      "epoch": 0.9897398248786864,
      "grad_norm": 1.7330875396728516,
      "learning_rate": 5.501793790305687e-08,
      "loss": 1.3741,
      "step": 27637
    },
    {
      "epoch": 0.9897756370082547,
      "grad_norm": 1.4673677682876587,
      "learning_rate": 5.4633904678125234e-08,
      "loss": 1.5148,
      "step": 27638
    },
    {
      "epoch": 0.9898114491378229,
      "grad_norm": 1.8655990362167358,
      "learning_rate": 5.425121608917394e-08,
      "loss": 1.6041,
      "step": 27639
    },
    {
      "epoch": 0.9898472612673913,
      "grad_norm": 1.5611985921859741,
      "learning_rate": 5.3869872141343313e-08,
      "loss": 1.3083,
      "step": 27640
    },
    {
      "epoch": 0.9898830733969596,
      "grad_norm": 1.6236423254013062,
      "learning_rate": 5.348987283978479e-08,
      "loss": 1.3885,
      "step": 27641
    },
    {
      "epoch": 0.9899188855265278,
      "grad_norm": 1.4711804389953613,
      "learning_rate": 5.3111218189594304e-08,
      "loss": 1.2077,
      "step": 27642
    },
    {
      "epoch": 0.9899546976560961,
      "grad_norm": 1.3890304565429688,
      "learning_rate": 5.2733908195867764e-08,
      "loss": 1.526,
      "step": 27643
    },
    {
      "epoch": 0.9899905097856644,
      "grad_norm": 1.4793809652328491,
      "learning_rate": 5.23579428636789e-08,
      "loss": 1.4713,
      "step": 27644
    },
    {
      "epoch": 0.9900263219152327,
      "grad_norm": 1.3004274368286133,
      "learning_rate": 5.1983322198101425e-08,
      "loss": 1.5026,
      "step": 27645
    },
    {
      "epoch": 0.9900621340448009,
      "grad_norm": 2.2040960788726807,
      "learning_rate": 5.161004620416465e-08,
      "loss": 1.1282,
      "step": 27646
    },
    {
      "epoch": 0.9900979461743693,
      "grad_norm": 1.7686691284179688,
      "learning_rate": 5.1238114886875685e-08,
      "loss": 1.3961,
      "step": 27647
    },
    {
      "epoch": 0.9901337583039376,
      "grad_norm": 1.619179606437683,
      "learning_rate": 5.086752825126384e-08,
      "loss": 1.5506,
      "step": 27648
    },
    {
      "epoch": 0.9901695704335058,
      "grad_norm": 1.7559400796890259,
      "learning_rate": 5.049828630230291e-08,
      "loss": 1.3118,
      "step": 27649
    },
    {
      "epoch": 0.9902053825630741,
      "grad_norm": 1.9877136945724487,
      "learning_rate": 5.01303890449667e-08,
      "loss": 1.5718,
      "step": 27650
    },
    {
      "epoch": 0.9902411946926424,
      "grad_norm": 1.8758115768432617,
      "learning_rate": 4.976383648419569e-08,
      "loss": 1.306,
      "step": 27651
    },
    {
      "epoch": 0.9902770068222106,
      "grad_norm": 2.0945935249328613,
      "learning_rate": 4.9398628624930385e-08,
      "loss": 1.5006,
      "step": 27652
    },
    {
      "epoch": 0.9903128189517789,
      "grad_norm": 1.786147117614746,
      "learning_rate": 4.903476547206687e-08,
      "loss": 1.273,
      "step": 27653
    },
    {
      "epoch": 0.9903486310813473,
      "grad_norm": 1.7926175594329834,
      "learning_rate": 4.8672247030523425e-08,
      "loss": 1.3818,
      "step": 27654
    },
    {
      "epoch": 0.9903844432109156,
      "grad_norm": 2.316483497619629,
      "learning_rate": 4.8311073305162825e-08,
      "loss": 1.248,
      "step": 27655
    },
    {
      "epoch": 0.9904202553404838,
      "grad_norm": 2.2332239151000977,
      "learning_rate": 4.795124430085896e-08,
      "loss": 1.1146,
      "step": 27656
    },
    {
      "epoch": 0.9904560674700521,
      "grad_norm": 1.80547297000885,
      "learning_rate": 4.7592760022430185e-08,
      "loss": 1.372,
      "step": 27657
    },
    {
      "epoch": 0.9904918795996204,
      "grad_norm": 1.2292410135269165,
      "learning_rate": 4.723562047471708e-08,
      "loss": 1.6469,
      "step": 27658
    },
    {
      "epoch": 0.9905276917291886,
      "grad_norm": 1.5700558423995972,
      "learning_rate": 4.687982566251581e-08,
      "loss": 1.3877,
      "step": 27659
    },
    {
      "epoch": 0.9905635038587569,
      "grad_norm": 1.5725685358047485,
      "learning_rate": 4.652537559062253e-08,
      "loss": 1.337,
      "step": 27660
    },
    {
      "epoch": 0.9905993159883253,
      "grad_norm": 1.444566249847412,
      "learning_rate": 4.617227026378901e-08,
      "loss": 1.2462,
      "step": 27661
    },
    {
      "epoch": 0.9906351281178936,
      "grad_norm": 1.5620014667510986,
      "learning_rate": 4.582050968677809e-08,
      "loss": 1.3293,
      "step": 27662
    },
    {
      "epoch": 0.9906709402474618,
      "grad_norm": 1.445263147354126,
      "learning_rate": 4.5470093864330435e-08,
      "loss": 1.4857,
      "step": 27663
    },
    {
      "epoch": 0.9907067523770301,
      "grad_norm": 2.2626137733459473,
      "learning_rate": 4.5121022801142275e-08,
      "loss": 1.2967,
      "step": 27664
    },
    {
      "epoch": 0.9907425645065984,
      "grad_norm": 1.5922274589538574,
      "learning_rate": 4.477329650192097e-08,
      "loss": 1.2494,
      "step": 27665
    },
    {
      "epoch": 0.9907783766361666,
      "grad_norm": 1.4853841066360474,
      "learning_rate": 4.442691497134055e-08,
      "loss": 1.2511,
      "step": 27666
    },
    {
      "epoch": 0.9908141887657349,
      "grad_norm": 1.8890957832336426,
      "learning_rate": 4.408187821406395e-08,
      "loss": 1.6964,
      "step": 27667
    },
    {
      "epoch": 0.9908500008953033,
      "grad_norm": 1.2731983661651611,
      "learning_rate": 4.373818623473191e-08,
      "loss": 1.3958,
      "step": 27668
    },
    {
      "epoch": 0.9908858130248716,
      "grad_norm": 1.4172587394714355,
      "learning_rate": 4.3395839037962956e-08,
      "loss": 1.3414,
      "step": 27669
    },
    {
      "epoch": 0.9909216251544398,
      "grad_norm": 1.5718883275985718,
      "learning_rate": 4.305483662837562e-08,
      "loss": 1.5334,
      "step": 27670
    },
    {
      "epoch": 0.9909574372840081,
      "grad_norm": 1.6942057609558105,
      "learning_rate": 4.2715179010555106e-08,
      "loss": 1.4271,
      "step": 27671
    },
    {
      "epoch": 0.9909932494135764,
      "grad_norm": 1.4615954160690308,
      "learning_rate": 4.2376866189053346e-08,
      "loss": 1.4432,
      "step": 27672
    },
    {
      "epoch": 0.9910290615431446,
      "grad_norm": 1.955811858177185,
      "learning_rate": 4.2039898168444445e-08,
      "loss": 1.4947,
      "step": 27673
    },
    {
      "epoch": 0.9910648736727129,
      "grad_norm": 1.6968096494674683,
      "learning_rate": 4.170427495324702e-08,
      "loss": 1.558,
      "step": 27674
    },
    {
      "epoch": 0.9911006858022813,
      "grad_norm": 1.3976668119430542,
      "learning_rate": 4.1369996547979685e-08,
      "loss": 1.142,
      "step": 27675
    },
    {
      "epoch": 0.9911364979318495,
      "grad_norm": 2.15129017829895,
      "learning_rate": 4.1037062957138825e-08,
      "loss": 1.7194,
      "step": 27676
    },
    {
      "epoch": 0.9911723100614178,
      "grad_norm": 1.6736570596694946,
      "learning_rate": 4.070547418522086e-08,
      "loss": 1.307,
      "step": 27677
    },
    {
      "epoch": 0.9912081221909861,
      "grad_norm": 1.7805536985397339,
      "learning_rate": 4.037523023666667e-08,
      "loss": 1.5194,
      "step": 27678
    },
    {
      "epoch": 0.9912439343205544,
      "grad_norm": 1.7508223056793213,
      "learning_rate": 4.0046331115917157e-08,
      "loss": 1.5008,
      "step": 27679
    },
    {
      "epoch": 0.9912797464501226,
      "grad_norm": 1.3880499601364136,
      "learning_rate": 3.9718776827413204e-08,
      "loss": 1.5748,
      "step": 27680
    },
    {
      "epoch": 0.9913155585796909,
      "grad_norm": 2.6532273292541504,
      "learning_rate": 3.9392567375551306e-08,
      "loss": 1.1765,
      "step": 27681
    },
    {
      "epoch": 0.9913513707092593,
      "grad_norm": 1.741824984550476,
      "learning_rate": 3.906770276471683e-08,
      "loss": 1.1866,
      "step": 27682
    },
    {
      "epoch": 0.9913871828388275,
      "grad_norm": 1.3869082927703857,
      "learning_rate": 3.8744182999295164e-08,
      "loss": 1.4337,
      "step": 27683
    },
    {
      "epoch": 0.9914229949683958,
      "grad_norm": 2.356285333633423,
      "learning_rate": 3.842200808362728e-08,
      "loss": 1.7948,
      "step": 27684
    },
    {
      "epoch": 0.9914588070979641,
      "grad_norm": 1.4748954772949219,
      "learning_rate": 3.810117802204305e-08,
      "loss": 1.323,
      "step": 27685
    },
    {
      "epoch": 0.9914946192275323,
      "grad_norm": 1.8216158151626587,
      "learning_rate": 3.778169281887234e-08,
      "loss": 1.4063,
      "step": 27686
    },
    {
      "epoch": 0.9915304313571006,
      "grad_norm": 1.3165158033370972,
      "learning_rate": 3.746355247841171e-08,
      "loss": 1.1767,
      "step": 27687
    },
    {
      "epoch": 0.9915662434866689,
      "grad_norm": 2.005211114883423,
      "learning_rate": 3.7146757004924425e-08,
      "loss": 1.1474,
      "step": 27688
    },
    {
      "epoch": 0.9916020556162373,
      "grad_norm": 1.54912531375885,
      "learning_rate": 3.683130640269594e-08,
      "loss": 1.4143,
      "step": 27689
    },
    {
      "epoch": 0.9916378677458055,
      "grad_norm": 1.9291484355926514,
      "learning_rate": 3.651720067595621e-08,
      "loss": 1.4354,
      "step": 27690
    },
    {
      "epoch": 0.9916736798753738,
      "grad_norm": 1.4543896913528442,
      "learning_rate": 3.620443982892407e-08,
      "loss": 1.5983,
      "step": 27691
    },
    {
      "epoch": 0.9917094920049421,
      "grad_norm": 2.140312910079956,
      "learning_rate": 3.589302386582949e-08,
      "loss": 1.352,
      "step": 27692
    },
    {
      "epoch": 0.9917453041345103,
      "grad_norm": 1.8118711709976196,
      "learning_rate": 3.558295279084689e-08,
      "loss": 1.4931,
      "step": 27693
    },
    {
      "epoch": 0.9917811162640786,
      "grad_norm": 1.5152671337127686,
      "learning_rate": 3.527422660815072e-08,
      "loss": 1.3262,
      "step": 27694
    },
    {
      "epoch": 0.9918169283936469,
      "grad_norm": 1.843450903892517,
      "learning_rate": 3.4966845321893204e-08,
      "loss": 1.3335,
      "step": 27695
    },
    {
      "epoch": 0.9918527405232153,
      "grad_norm": 2.2369720935821533,
      "learning_rate": 3.4660808936215485e-08,
      "loss": 1.5118,
      "step": 27696
    },
    {
      "epoch": 0.9918885526527835,
      "grad_norm": 1.4145985841751099,
      "learning_rate": 3.435611745522538e-08,
      "loss": 1.5992,
      "step": 27697
    },
    {
      "epoch": 0.9919243647823518,
      "grad_norm": 1.7589112520217896,
      "learning_rate": 3.405277088301961e-08,
      "loss": 1.3502,
      "step": 27698
    },
    {
      "epoch": 0.9919601769119201,
      "grad_norm": 1.60627281665802,
      "learning_rate": 3.375076922370601e-08,
      "loss": 1.2305,
      "step": 27699
    },
    {
      "epoch": 0.9919959890414883,
      "grad_norm": 1.4741889238357544,
      "learning_rate": 3.345011248131469e-08,
      "loss": 1.3674,
      "step": 27700
    },
    {
      "epoch": 0.9920318011710566,
      "grad_norm": 1.848053216934204,
      "learning_rate": 3.3150800659909055e-08,
      "loss": 1.4557,
      "step": 27701
    },
    {
      "epoch": 0.9920676133006249,
      "grad_norm": 1.6344389915466309,
      "learning_rate": 3.285283376350812e-08,
      "loss": 1.3267,
      "step": 27702
    },
    {
      "epoch": 0.9921034254301933,
      "grad_norm": 1.2263892889022827,
      "learning_rate": 3.255621179613089e-08,
      "loss": 1.3394,
      "step": 27703
    },
    {
      "epoch": 0.9921392375597615,
      "grad_norm": 1.8323966264724731,
      "learning_rate": 3.226093476175196e-08,
      "loss": 1.4297,
      "step": 27704
    },
    {
      "epoch": 0.9921750496893298,
      "grad_norm": 1.2270833253860474,
      "learning_rate": 3.1967002664357036e-08,
      "loss": 1.1901,
      "step": 27705
    },
    {
      "epoch": 0.9922108618188981,
      "grad_norm": 1.5533965826034546,
      "learning_rate": 3.167441550789851e-08,
      "loss": 1.455,
      "step": 27706
    },
    {
      "epoch": 0.9922466739484663,
      "grad_norm": 1.9076077938079834,
      "learning_rate": 3.138317329630658e-08,
      "loss": 1.3796,
      "step": 27707
    },
    {
      "epoch": 0.9922824860780346,
      "grad_norm": 1.721117377281189,
      "learning_rate": 3.109327603351142e-08,
      "loss": 1.2874,
      "step": 27708
    },
    {
      "epoch": 0.9923182982076029,
      "grad_norm": 1.3826229572296143,
      "learning_rate": 3.080472372339882e-08,
      "loss": 1.524,
      "step": 27709
    },
    {
      "epoch": 0.9923541103371712,
      "grad_norm": 1.7644851207733154,
      "learning_rate": 3.0517516369865665e-08,
      "loss": 1.3418,
      "step": 27710
    },
    {
      "epoch": 0.9923899224667395,
      "grad_norm": 1.6510146856307983,
      "learning_rate": 3.0231653976764415e-08,
      "loss": 1.2241,
      "step": 27711
    },
    {
      "epoch": 0.9924257345963078,
      "grad_norm": 1.384408950805664,
      "learning_rate": 2.994713654793646e-08,
      "loss": 1.5097,
      "step": 27712
    },
    {
      "epoch": 0.992461546725876,
      "grad_norm": 1.7258027791976929,
      "learning_rate": 2.966396408722316e-08,
      "loss": 1.3808,
      "step": 27713
    },
    {
      "epoch": 0.9924973588554443,
      "grad_norm": 2.7897722721099854,
      "learning_rate": 2.9382136598432587e-08,
      "loss": 0.9909,
      "step": 27714
    },
    {
      "epoch": 0.9925331709850126,
      "grad_norm": 1.8314100503921509,
      "learning_rate": 2.9101654085350594e-08,
      "loss": 1.4235,
      "step": 27715
    },
    {
      "epoch": 0.9925689831145809,
      "grad_norm": 1.441307783126831,
      "learning_rate": 2.8822516551751942e-08,
      "loss": 1.1951,
      "step": 27716
    },
    {
      "epoch": 0.9926047952441492,
      "grad_norm": 1.8378783464431763,
      "learning_rate": 2.854472400138919e-08,
      "loss": 1.2973,
      "step": 27717
    },
    {
      "epoch": 0.9926406073737175,
      "grad_norm": 1.8195862770080566,
      "learning_rate": 2.8268276438003782e-08,
      "loss": 1.4349,
      "step": 27718
    },
    {
      "epoch": 0.9926764195032858,
      "grad_norm": 1.6150768995285034,
      "learning_rate": 2.799317386531497e-08,
      "loss": 1.547,
      "step": 27719
    },
    {
      "epoch": 0.992712231632854,
      "grad_norm": 1.6209663152694702,
      "learning_rate": 2.7719416287030897e-08,
      "loss": 1.4109,
      "step": 27720
    },
    {
      "epoch": 0.9927480437624223,
      "grad_norm": 1.6728167533874512,
      "learning_rate": 2.74470037068264e-08,
      "loss": 1.4065,
      "step": 27721
    },
    {
      "epoch": 0.9927838558919906,
      "grad_norm": 1.3521865606307983,
      "learning_rate": 2.717593612835412e-08,
      "loss": 1.4566,
      "step": 27722
    },
    {
      "epoch": 0.9928196680215589,
      "grad_norm": 1.5996589660644531,
      "learning_rate": 2.6906213555288884e-08,
      "loss": 1.4181,
      "step": 27723
    },
    {
      "epoch": 0.9928554801511272,
      "grad_norm": 1.476045846939087,
      "learning_rate": 2.6637835991238924e-08,
      "loss": 1.2518,
      "step": 27724
    },
    {
      "epoch": 0.9928912922806955,
      "grad_norm": 2.2209970951080322,
      "learning_rate": 2.6370803439812463e-08,
      "loss": 1.0453,
      "step": 27725
    },
    {
      "epoch": 0.9929271044102638,
      "grad_norm": 1.3425819873809814,
      "learning_rate": 2.6105115904617726e-08,
      "loss": 1.5999,
      "step": 27726
    },
    {
      "epoch": 0.992962916539832,
      "grad_norm": 1.6703921556472778,
      "learning_rate": 2.584077338921853e-08,
      "loss": 1.3171,
      "step": 27727
    },
    {
      "epoch": 0.9929987286694003,
      "grad_norm": 1.6346490383148193,
      "learning_rate": 2.557777589717869e-08,
      "loss": 1.4642,
      "step": 27728
    },
    {
      "epoch": 0.9930345407989686,
      "grad_norm": 1.7949005365371704,
      "learning_rate": 2.5316123432028714e-08,
      "loss": 1.6685,
      "step": 27729
    },
    {
      "epoch": 0.9930703529285368,
      "grad_norm": 1.5301461219787598,
      "learning_rate": 2.5055815997299113e-08,
      "loss": 1.7001,
      "step": 27730
    },
    {
      "epoch": 0.9931061650581052,
      "grad_norm": 1.6735132932662964,
      "learning_rate": 2.479685359647599e-08,
      "loss": 1.4314,
      "step": 27731
    },
    {
      "epoch": 0.9931419771876735,
      "grad_norm": 1.8682321310043335,
      "learning_rate": 2.453923623305654e-08,
      "loss": 1.4695,
      "step": 27732
    },
    {
      "epoch": 0.9931777893172418,
      "grad_norm": 1.7532639503479004,
      "learning_rate": 2.4282963910504664e-08,
      "loss": 1.6481,
      "step": 27733
    },
    {
      "epoch": 0.99321360144681,
      "grad_norm": 1.6439052820205688,
      "learning_rate": 2.4028036632262053e-08,
      "loss": 1.3373,
      "step": 27734
    },
    {
      "epoch": 0.9932494135763783,
      "grad_norm": 2.574580192565918,
      "learning_rate": 2.3774454401770396e-08,
      "loss": 1.5889,
      "step": 27735
    },
    {
      "epoch": 0.9932852257059466,
      "grad_norm": 1.586375117301941,
      "learning_rate": 2.3522217222426978e-08,
      "loss": 1.4725,
      "step": 27736
    },
    {
      "epoch": 0.9933210378355148,
      "grad_norm": 1.4359043836593628,
      "learning_rate": 2.3271325097629082e-08,
      "loss": 1.3233,
      "step": 27737
    },
    {
      "epoch": 0.9933568499650832,
      "grad_norm": 1.3824795484542847,
      "learning_rate": 2.3021778030751784e-08,
      "loss": 1.4131,
      "step": 27738
    },
    {
      "epoch": 0.9933926620946515,
      "grad_norm": 1.66568124294281,
      "learning_rate": 2.2773576025170163e-08,
      "loss": 1.3049,
      "step": 27739
    },
    {
      "epoch": 0.9934284742242198,
      "grad_norm": 1.4649953842163086,
      "learning_rate": 2.2526719084192683e-08,
      "loss": 1.4202,
      "step": 27740
    },
    {
      "epoch": 0.993464286353788,
      "grad_norm": 1.3860225677490234,
      "learning_rate": 2.2281207211172218e-08,
      "loss": 1.4507,
      "step": 27741
    },
    {
      "epoch": 0.9935000984833563,
      "grad_norm": 1.6990655660629272,
      "learning_rate": 2.2037040409383924e-08,
      "loss": 1.3812,
      "step": 27742
    },
    {
      "epoch": 0.9935359106129246,
      "grad_norm": 1.6962313652038574,
      "learning_rate": 2.1794218682125168e-08,
      "loss": 1.4086,
      "step": 27743
    },
    {
      "epoch": 0.9935717227424928,
      "grad_norm": 1.4338451623916626,
      "learning_rate": 2.15527420326711e-08,
      "loss": 1.5719,
      "step": 27744
    },
    {
      "epoch": 0.9936075348720612,
      "grad_norm": 1.4438395500183105,
      "learning_rate": 2.131261046425248e-08,
      "loss": 1.2907,
      "step": 27745
    },
    {
      "epoch": 0.9936433470016295,
      "grad_norm": 1.768540620803833,
      "learning_rate": 2.107382398011115e-08,
      "loss": 1.33,
      "step": 27746
    },
    {
      "epoch": 0.9936791591311978,
      "grad_norm": 1.6516530513763428,
      "learning_rate": 2.0836382583466762e-08,
      "loss": 1.4927,
      "step": 27747
    },
    {
      "epoch": 0.993714971260766,
      "grad_norm": 1.5076415538787842,
      "learning_rate": 2.0600286277494552e-08,
      "loss": 1.2049,
      "step": 27748
    },
    {
      "epoch": 0.9937507833903343,
      "grad_norm": 1.4209827184677124,
      "learning_rate": 2.0365535065391962e-08,
      "loss": 1.297,
      "step": 27749
    },
    {
      "epoch": 0.9937865955199026,
      "grad_norm": 1.4648021459579468,
      "learning_rate": 2.013212895030092e-08,
      "loss": 1.286,
      "step": 27750
    },
    {
      "epoch": 0.9938224076494708,
      "grad_norm": 2.422774314880371,
      "learning_rate": 1.9900067935363364e-08,
      "loss": 1.6095,
      "step": 27751
    },
    {
      "epoch": 0.9938582197790392,
      "grad_norm": 1.4855307340621948,
      "learning_rate": 1.966935202371012e-08,
      "loss": 1.5097,
      "step": 27752
    },
    {
      "epoch": 0.9938940319086075,
      "grad_norm": 2.235036849975586,
      "learning_rate": 1.9439981218438708e-08,
      "loss": 1.4405,
      "step": 27753
    },
    {
      "epoch": 0.9939298440381757,
      "grad_norm": 1.6031259298324585,
      "learning_rate": 1.921195552263555e-08,
      "loss": 1.4121,
      "step": 27754
    },
    {
      "epoch": 0.993965656167744,
      "grad_norm": 1.9243065118789673,
      "learning_rate": 1.8985274939375962e-08,
      "loss": 1.5876,
      "step": 27755
    },
    {
      "epoch": 0.9940014682973123,
      "grad_norm": 1.6919931173324585,
      "learning_rate": 1.8759939471690858e-08,
      "loss": 1.453,
      "step": 27756
    },
    {
      "epoch": 0.9940372804268806,
      "grad_norm": 1.5720360279083252,
      "learning_rate": 1.8535949122633346e-08,
      "loss": 1.4789,
      "step": 27757
    },
    {
      "epoch": 0.9940730925564488,
      "grad_norm": 1.6576461791992188,
      "learning_rate": 1.831330389521213e-08,
      "loss": 1.417,
      "step": 27758
    },
    {
      "epoch": 0.9941089046860172,
      "grad_norm": 1.631563425064087,
      "learning_rate": 1.8092003792413714e-08,
      "loss": 1.3115,
      "step": 27759
    },
    {
      "epoch": 0.9941447168155855,
      "grad_norm": 1.270306944847107,
      "learning_rate": 1.7872048817213495e-08,
      "loss": 1.5715,
      "step": 27760
    },
    {
      "epoch": 0.9941805289451537,
      "grad_norm": 1.9433504343032837,
      "learning_rate": 1.7653438972586868e-08,
      "loss": 1.4304,
      "step": 27761
    },
    {
      "epoch": 0.994216341074722,
      "grad_norm": 2.1544501781463623,
      "learning_rate": 1.743617426145372e-08,
      "loss": 1.4941,
      "step": 27762
    },
    {
      "epoch": 0.9942521532042903,
      "grad_norm": 1.2280709743499756,
      "learning_rate": 1.7220254686756142e-08,
      "loss": 1.0029,
      "step": 27763
    },
    {
      "epoch": 0.9942879653338585,
      "grad_norm": 1.3527542352676392,
      "learning_rate": 1.700568025139182e-08,
      "loss": 1.4559,
      "step": 27764
    },
    {
      "epoch": 0.9943237774634268,
      "grad_norm": 1.9199059009552002,
      "learning_rate": 1.679245095824733e-08,
      "loss": 1.3947,
      "step": 27765
    },
    {
      "epoch": 0.9943595895929952,
      "grad_norm": 2.5638864040374756,
      "learning_rate": 1.658056681019815e-08,
      "loss": 1.5441,
      "step": 27766
    },
    {
      "epoch": 0.9943954017225635,
      "grad_norm": 1.6788527965545654,
      "learning_rate": 1.637002781007535e-08,
      "loss": 1.3188,
      "step": 27767
    },
    {
      "epoch": 0.9944312138521317,
      "grad_norm": 1.2428195476531982,
      "learning_rate": 1.6160833960732203e-08,
      "loss": 1.4178,
      "step": 27768
    },
    {
      "epoch": 0.9944670259817,
      "grad_norm": 1.7041432857513428,
      "learning_rate": 1.595298526496647e-08,
      "loss": 1.5002,
      "step": 27769
    },
    {
      "epoch": 0.9945028381112683,
      "grad_norm": 1.5376256704330444,
      "learning_rate": 1.5746481725598117e-08,
      "loss": 1.6361,
      "step": 27770
    },
    {
      "epoch": 0.9945386502408365,
      "grad_norm": 1.9216423034667969,
      "learning_rate": 1.5541323345380497e-08,
      "loss": 1.2145,
      "step": 27771
    },
    {
      "epoch": 0.9945744623704048,
      "grad_norm": 1.9160414934158325,
      "learning_rate": 1.533751012707807e-08,
      "loss": 1.5186,
      "step": 27772
    },
    {
      "epoch": 0.9946102744999732,
      "grad_norm": 1.4614883661270142,
      "learning_rate": 1.5135042073444182e-08,
      "loss": 1.6713,
      "step": 27773
    },
    {
      "epoch": 0.9946460866295415,
      "grad_norm": 1.514203667640686,
      "learning_rate": 1.4933919187198884e-08,
      "loss": 1.041,
      "step": 27774
    },
    {
      "epoch": 0.9946818987591097,
      "grad_norm": 1.3184031248092651,
      "learning_rate": 1.4734141471051122e-08,
      "loss": 1.3827,
      "step": 27775
    },
    {
      "epoch": 0.994717710888678,
      "grad_norm": 1.7681211233139038,
      "learning_rate": 1.4535708927676529e-08,
      "loss": 1.5363,
      "step": 27776
    },
    {
      "epoch": 0.9947535230182463,
      "grad_norm": 1.4674324989318848,
      "learning_rate": 1.4338621559750742e-08,
      "loss": 1.5941,
      "step": 27777
    },
    {
      "epoch": 0.9947893351478145,
      "grad_norm": 1.7696936130523682,
      "learning_rate": 1.4142879369927198e-08,
      "loss": 1.4344,
      "step": 27778
    },
    {
      "epoch": 0.9948251472773828,
      "grad_norm": 1.9830656051635742,
      "learning_rate": 1.3948482360848225e-08,
      "loss": 1.5663,
      "step": 27779
    },
    {
      "epoch": 0.9948609594069512,
      "grad_norm": 1.470342993736267,
      "learning_rate": 1.3755430535111747e-08,
      "loss": 1.6629,
      "step": 27780
    },
    {
      "epoch": 0.9948967715365195,
      "grad_norm": 1.7345906496047974,
      "learning_rate": 1.3563723895326785e-08,
      "loss": 1.4955,
      "step": 27781
    },
    {
      "epoch": 0.9949325836660877,
      "grad_norm": 1.6475701332092285,
      "learning_rate": 1.3373362444057957e-08,
      "loss": 1.443,
      "step": 27782
    },
    {
      "epoch": 0.994968395795656,
      "grad_norm": 1.7012711763381958,
      "learning_rate": 1.3184346183892082e-08,
      "loss": 1.341,
      "step": 27783
    },
    {
      "epoch": 0.9950042079252243,
      "grad_norm": 1.4471287727355957,
      "learning_rate": 1.2996675117349367e-08,
      "loss": 1.3683,
      "step": 27784
    },
    {
      "epoch": 0.9950400200547925,
      "grad_norm": 2.2064197063446045,
      "learning_rate": 1.2810349246961117e-08,
      "loss": 1.3698,
      "step": 27785
    },
    {
      "epoch": 0.9950758321843608,
      "grad_norm": 2.3040659427642822,
      "learning_rate": 1.262536857523644e-08,
      "loss": 1.364,
      "step": 27786
    },
    {
      "epoch": 0.9951116443139292,
      "grad_norm": 1.5879874229431152,
      "learning_rate": 1.2441733104662234e-08,
      "loss": 1.3092,
      "step": 27787
    },
    {
      "epoch": 0.9951474564434974,
      "grad_norm": 1.4612466096878052,
      "learning_rate": 1.2259442837714297e-08,
      "loss": 1.4385,
      "step": 27788
    },
    {
      "epoch": 0.9951832685730657,
      "grad_norm": 2.061544418334961,
      "learning_rate": 1.2078497776835119e-08,
      "loss": 1.4193,
      "step": 27789
    },
    {
      "epoch": 0.995219080702634,
      "grad_norm": 1.4527156352996826,
      "learning_rate": 1.1898897924467189e-08,
      "loss": 1.4357,
      "step": 27790
    },
    {
      "epoch": 0.9952548928322023,
      "grad_norm": 1.664737343788147,
      "learning_rate": 1.1720643283019694e-08,
      "loss": 1.6887,
      "step": 27791
    },
    {
      "epoch": 0.9952907049617705,
      "grad_norm": 1.4561351537704468,
      "learning_rate": 1.1543733854901817e-08,
      "loss": 1.2292,
      "step": 27792
    },
    {
      "epoch": 0.9953265170913388,
      "grad_norm": 1.2238926887512207,
      "learning_rate": 1.1368169642489435e-08,
      "loss": 1.2025,
      "step": 27793
    },
    {
      "epoch": 0.9953623292209072,
      "grad_norm": 1.3671455383300781,
      "learning_rate": 1.119395064813622e-08,
      "loss": 1.3286,
      "step": 27794
    },
    {
      "epoch": 0.9953981413504754,
      "grad_norm": 1.3853384256362915,
      "learning_rate": 1.1021076874195846e-08,
      "loss": 1.3408,
      "step": 27795
    },
    {
      "epoch": 0.9954339534800437,
      "grad_norm": 1.7308851480484009,
      "learning_rate": 1.0849548322988679e-08,
      "loss": 1.3314,
      "step": 27796
    },
    {
      "epoch": 0.995469765609612,
      "grad_norm": 1.6281728744506836,
      "learning_rate": 1.0679364996823982e-08,
      "loss": 1.4538,
      "step": 27797
    },
    {
      "epoch": 0.9955055777391802,
      "grad_norm": 1.8174667358398438,
      "learning_rate": 1.0510526897988815e-08,
      "loss": 1.2637,
      "step": 27798
    },
    {
      "epoch": 0.9955413898687485,
      "grad_norm": 2.2247347831726074,
      "learning_rate": 1.0343034028759136e-08,
      "loss": 1.6667,
      "step": 27799
    },
    {
      "epoch": 0.9955772019983168,
      "grad_norm": 1.9621455669403076,
      "learning_rate": 1.0176886391388695e-08,
      "loss": 1.5749,
      "step": 27800
    },
    {
      "epoch": 0.9956130141278852,
      "grad_norm": 1.6727039813995361,
      "learning_rate": 1.0012083988109045e-08,
      "loss": 1.2988,
      "step": 27801
    },
    {
      "epoch": 0.9956488262574534,
      "grad_norm": 1.5761189460754395,
      "learning_rate": 9.848626821140627e-09,
      "loss": 1.5141,
      "step": 27802
    },
    {
      "epoch": 0.9956846383870217,
      "grad_norm": 1.3723686933517456,
      "learning_rate": 9.686514892681687e-09,
      "loss": 1.2416,
      "step": 27803
    },
    {
      "epoch": 0.99572045051659,
      "grad_norm": 1.3970915079116821,
      "learning_rate": 9.525748204908258e-09,
      "loss": 1.2696,
      "step": 27804
    },
    {
      "epoch": 0.9957562626461582,
      "grad_norm": 1.6122543811798096,
      "learning_rate": 9.366326759985278e-09,
      "loss": 1.5499,
      "step": 27805
    },
    {
      "epoch": 0.9957920747757265,
      "grad_norm": 1.3238275051116943,
      "learning_rate": 9.208250560066578e-09,
      "loss": 1.0434,
      "step": 27806
    },
    {
      "epoch": 0.9958278869052948,
      "grad_norm": 1.784379243850708,
      "learning_rate": 9.051519607272684e-09,
      "loss": 1.6537,
      "step": 27807
    },
    {
      "epoch": 0.9958636990348632,
      "grad_norm": 1.6680328845977783,
      "learning_rate": 8.89613390370192e-09,
      "loss": 1.658,
      "step": 27808
    },
    {
      "epoch": 0.9958995111644314,
      "grad_norm": 1.6057300567626953,
      "learning_rate": 8.742093451463707e-09,
      "loss": 1.3095,
      "step": 27809
    },
    {
      "epoch": 0.9959353232939997,
      "grad_norm": 1.358614444732666,
      "learning_rate": 8.589398252611957e-09,
      "loss": 1.3424,
      "step": 27810
    },
    {
      "epoch": 0.995971135423568,
      "grad_norm": 1.5714879035949707,
      "learning_rate": 8.43804830922279e-09,
      "loss": 1.2419,
      "step": 27811
    },
    {
      "epoch": 0.9960069475531362,
      "grad_norm": 1.897676706314087,
      "learning_rate": 8.28804362331681e-09,
      "loss": 1.6724,
      "step": 27812
    },
    {
      "epoch": 0.9960427596827045,
      "grad_norm": 1.658821940422058,
      "learning_rate": 8.139384196903522e-09,
      "loss": 1.2458,
      "step": 27813
    },
    {
      "epoch": 0.9960785718122728,
      "grad_norm": 1.9782476425170898,
      "learning_rate": 7.992070032003529e-09,
      "loss": 1.3115,
      "step": 27814
    },
    {
      "epoch": 0.9961143839418412,
      "grad_norm": 1.8454557657241821,
      "learning_rate": 7.84610113059303e-09,
      "loss": 1.4479,
      "step": 27815
    },
    {
      "epoch": 0.9961501960714094,
      "grad_norm": 1.320586085319519,
      "learning_rate": 7.70147749462602e-09,
      "loss": 1.4129,
      "step": 27816
    },
    {
      "epoch": 0.9961860082009777,
      "grad_norm": 1.5092272758483887,
      "learning_rate": 7.558199126056487e-09,
      "loss": 1.5169,
      "step": 27817
    },
    {
      "epoch": 0.996221820330546,
      "grad_norm": 1.7253291606903076,
      "learning_rate": 7.416266026816221e-09,
      "loss": 1.6159,
      "step": 27818
    },
    {
      "epoch": 0.9962576324601142,
      "grad_norm": 1.819915771484375,
      "learning_rate": 7.275678198803703e-09,
      "loss": 1.2604,
      "step": 27819
    },
    {
      "epoch": 0.9962934445896825,
      "grad_norm": 1.2910900115966797,
      "learning_rate": 7.136435643917416e-09,
      "loss": 1.3797,
      "step": 27820
    },
    {
      "epoch": 0.9963292567192508,
      "grad_norm": 1.5493773221969604,
      "learning_rate": 6.998538364022533e-09,
      "loss": 1.6762,
      "step": 27821
    },
    {
      "epoch": 0.9963650688488191,
      "grad_norm": 1.7264066934585571,
      "learning_rate": 6.861986360995332e-09,
      "loss": 1.2949,
      "step": 27822
    },
    {
      "epoch": 0.9964008809783874,
      "grad_norm": 2.5256896018981934,
      "learning_rate": 6.726779636645475e-09,
      "loss": 1.688,
      "step": 27823
    },
    {
      "epoch": 0.9964366931079557,
      "grad_norm": 1.2364569902420044,
      "learning_rate": 6.592918192804831e-09,
      "loss": 1.2282,
      "step": 27824
    },
    {
      "epoch": 0.996472505237524,
      "grad_norm": 1.384142518043518,
      "learning_rate": 6.460402031283064e-09,
      "loss": 1.3266,
      "step": 27825
    },
    {
      "epoch": 0.9965083173670922,
      "grad_norm": 1.398099660873413,
      "learning_rate": 6.329231153845427e-09,
      "loss": 1.0723,
      "step": 27826
    },
    {
      "epoch": 0.9965441294966605,
      "grad_norm": 2.250969886779785,
      "learning_rate": 6.199405562268279e-09,
      "loss": 1.545,
      "step": 27827
    },
    {
      "epoch": 0.9965799416262288,
      "grad_norm": 1.5381693840026855,
      "learning_rate": 6.070925258294668e-09,
      "loss": 1.4875,
      "step": 27828
    },
    {
      "epoch": 0.9966157537557971,
      "grad_norm": 1.3616868257522583,
      "learning_rate": 5.943790243656544e-09,
      "loss": 1.3771,
      "step": 27829
    },
    {
      "epoch": 0.9966515658853654,
      "grad_norm": 1.404511570930481,
      "learning_rate": 5.818000520052547e-09,
      "loss": 1.3444,
      "step": 27830
    },
    {
      "epoch": 0.9966873780149337,
      "grad_norm": 1.3971214294433594,
      "learning_rate": 5.69355608919242e-09,
      "loss": 1.3699,
      "step": 27831
    },
    {
      "epoch": 0.996723190144502,
      "grad_norm": 1.6311243772506714,
      "learning_rate": 5.570456952741499e-09,
      "loss": 1.3255,
      "step": 27832
    },
    {
      "epoch": 0.9967590022740702,
      "grad_norm": 1.9845330715179443,
      "learning_rate": 5.448703112365117e-09,
      "loss": 1.5725,
      "step": 27833
    },
    {
      "epoch": 0.9967948144036385,
      "grad_norm": 1.6047090291976929,
      "learning_rate": 5.328294569673098e-09,
      "loss": 1.4876,
      "step": 27834
    },
    {
      "epoch": 0.9968306265332068,
      "grad_norm": 1.8364373445510864,
      "learning_rate": 5.209231326319674e-09,
      "loss": 1.3304,
      "step": 27835
    },
    {
      "epoch": 0.9968664386627751,
      "grad_norm": 1.2628228664398193,
      "learning_rate": 5.0915133838924656e-09,
      "loss": 1.21,
      "step": 27836
    },
    {
      "epoch": 0.9969022507923434,
      "grad_norm": 1.866013765335083,
      "learning_rate": 4.975140743967987e-09,
      "loss": 1.4851,
      "step": 27837
    },
    {
      "epoch": 0.9969380629219117,
      "grad_norm": 1.4663585424423218,
      "learning_rate": 4.860113408122757e-09,
      "loss": 1.7521,
      "step": 27838
    },
    {
      "epoch": 0.9969738750514799,
      "grad_norm": 1.7713894844055176,
      "learning_rate": 4.746431377899985e-09,
      "loss": 1.4532,
      "step": 27839
    },
    {
      "epoch": 0.9970096871810482,
      "grad_norm": 1.5363188982009888,
      "learning_rate": 4.634094654820675e-09,
      "loss": 1.2711,
      "step": 27840
    },
    {
      "epoch": 0.9970454993106165,
      "grad_norm": 1.7956669330596924,
      "learning_rate": 4.523103240416937e-09,
      "loss": 1.5147,
      "step": 27841
    },
    {
      "epoch": 0.9970813114401847,
      "grad_norm": 2.0350286960601807,
      "learning_rate": 4.413457136165367e-09,
      "loss": 1.5831,
      "step": 27842
    },
    {
      "epoch": 0.9971171235697531,
      "grad_norm": 1.6121965646743774,
      "learning_rate": 4.3051563435425605e-09,
      "loss": 1.3732,
      "step": 27843
    },
    {
      "epoch": 0.9971529356993214,
      "grad_norm": 1.6830921173095703,
      "learning_rate": 4.198200864014012e-09,
      "loss": 1.6102,
      "step": 27844
    },
    {
      "epoch": 0.9971887478288897,
      "grad_norm": 1.550490379333496,
      "learning_rate": 4.092590699011911e-09,
      "loss": 1.4481,
      "step": 27845
    },
    {
      "epoch": 0.9972245599584579,
      "grad_norm": 1.875200629234314,
      "learning_rate": 3.988325849957342e-09,
      "loss": 1.3372,
      "step": 27846
    },
    {
      "epoch": 0.9972603720880262,
      "grad_norm": 1.4787503480911255,
      "learning_rate": 3.885406318260288e-09,
      "loss": 1.4715,
      "step": 27847
    },
    {
      "epoch": 0.9972961842175945,
      "grad_norm": 1.6989432573318481,
      "learning_rate": 3.783832105286322e-09,
      "loss": 1.3271,
      "step": 27848
    },
    {
      "epoch": 0.9973319963471627,
      "grad_norm": 1.4826751947402954,
      "learning_rate": 3.6836032124232256e-09,
      "loss": 1.3739,
      "step": 27849
    },
    {
      "epoch": 0.9973678084767311,
      "grad_norm": 1.7975443601608276,
      "learning_rate": 3.5847196410143667e-09,
      "loss": 1.6923,
      "step": 27850
    },
    {
      "epoch": 0.9974036206062994,
      "grad_norm": 1.5432265996932983,
      "learning_rate": 3.4871813923809117e-09,
      "loss": 1.3351,
      "step": 27851
    },
    {
      "epoch": 0.9974394327358677,
      "grad_norm": 1.5838642120361328,
      "learning_rate": 3.390988467844025e-09,
      "loss": 1.579,
      "step": 27852
    },
    {
      "epoch": 0.9974752448654359,
      "grad_norm": 1.9075052738189697,
      "learning_rate": 3.2961408686915662e-09,
      "loss": 1.3764,
      "step": 27853
    },
    {
      "epoch": 0.9975110569950042,
      "grad_norm": 1.5979747772216797,
      "learning_rate": 3.2026385962113937e-09,
      "loss": 1.4625,
      "step": 27854
    },
    {
      "epoch": 0.9975468691245725,
      "grad_norm": 1.5051318407058716,
      "learning_rate": 3.110481651646957e-09,
      "loss": 1.3817,
      "step": 27855
    },
    {
      "epoch": 0.9975826812541407,
      "grad_norm": 2.1045610904693604,
      "learning_rate": 3.0196700362417065e-09,
      "loss": 1.4953,
      "step": 27856
    },
    {
      "epoch": 0.9976184933837091,
      "grad_norm": 1.5052860975265503,
      "learning_rate": 2.930203751227989e-09,
      "loss": 1.3494,
      "step": 27857
    },
    {
      "epoch": 0.9976543055132774,
      "grad_norm": 1.5225681066513062,
      "learning_rate": 2.8420827977937437e-09,
      "loss": 1.5778,
      "step": 27858
    },
    {
      "epoch": 0.9976901176428457,
      "grad_norm": 1.5422025918960571,
      "learning_rate": 2.7553071771380112e-09,
      "loss": 1.3814,
      "step": 27859
    },
    {
      "epoch": 0.9977259297724139,
      "grad_norm": 3.553717851638794,
      "learning_rate": 2.6698768904154236e-09,
      "loss": 1.4703,
      "step": 27860
    },
    {
      "epoch": 0.9977617419019822,
      "grad_norm": 1.6214388608932495,
      "learning_rate": 2.585791938791715e-09,
      "loss": 1.1495,
      "step": 27861
    },
    {
      "epoch": 0.9977975540315505,
      "grad_norm": 1.3043503761291504,
      "learning_rate": 2.5030523233771087e-09,
      "loss": 1.4878,
      "step": 27862
    },
    {
      "epoch": 0.9978333661611187,
      "grad_norm": 1.5366240739822388,
      "learning_rate": 2.4216580453040314e-09,
      "loss": 1.0653,
      "step": 27863
    },
    {
      "epoch": 0.9978691782906871,
      "grad_norm": 1.7023508548736572,
      "learning_rate": 2.3416091056605027e-09,
      "loss": 1.5414,
      "step": 27864
    },
    {
      "epoch": 0.9979049904202554,
      "grad_norm": 1.317268967628479,
      "learning_rate": 2.2629055055234384e-09,
      "loss": 1.4184,
      "step": 27865
    },
    {
      "epoch": 0.9979408025498236,
      "grad_norm": 1.655472755432129,
      "learning_rate": 2.18554724594755e-09,
      "loss": 1.6407,
      "step": 27866
    },
    {
      "epoch": 0.9979766146793919,
      "grad_norm": 1.6678001880645752,
      "learning_rate": 2.1095343279764477e-09,
      "loss": 1.4937,
      "step": 27867
    },
    {
      "epoch": 0.9980124268089602,
      "grad_norm": 1.3210474252700806,
      "learning_rate": 2.0348667526426392e-09,
      "loss": 1.3464,
      "step": 27868
    },
    {
      "epoch": 0.9980482389385285,
      "grad_norm": 1.340619444847107,
      "learning_rate": 1.961544520934222e-09,
      "loss": 1.1373,
      "step": 27869
    },
    {
      "epoch": 0.9980840510680967,
      "grad_norm": 2.0693037509918213,
      "learning_rate": 1.8895676338392952e-09,
      "loss": 1.6612,
      "step": 27870
    },
    {
      "epoch": 0.9981198631976651,
      "grad_norm": 1.7016184329986572,
      "learning_rate": 1.8189360923459575e-09,
      "loss": 1.3875,
      "step": 27871
    },
    {
      "epoch": 0.9981556753272334,
      "grad_norm": 1.4674826860427856,
      "learning_rate": 1.7496498973756936e-09,
      "loss": 1.4655,
      "step": 27872
    },
    {
      "epoch": 0.9981914874568016,
      "grad_norm": 1.7653722763061523,
      "learning_rate": 1.6817090498832954e-09,
      "loss": 1.4544,
      "step": 27873
    },
    {
      "epoch": 0.9982272995863699,
      "grad_norm": 1.617550015449524,
      "learning_rate": 1.615113550779146e-09,
      "loss": 1.4976,
      "step": 27874
    },
    {
      "epoch": 0.9982631117159382,
      "grad_norm": 1.6793699264526367,
      "learning_rate": 1.5498634009514235e-09,
      "loss": 1.6882,
      "step": 27875
    },
    {
      "epoch": 0.9982989238455064,
      "grad_norm": 1.7984838485717773,
      "learning_rate": 1.4859586012772042e-09,
      "loss": 1.6261,
      "step": 27876
    },
    {
      "epoch": 0.9983347359750747,
      "grad_norm": 1.3389443159103394,
      "learning_rate": 1.4233991526224622e-09,
      "loss": 1.2497,
      "step": 27877
    },
    {
      "epoch": 0.9983705481046431,
      "grad_norm": 1.4950339794158936,
      "learning_rate": 1.3621850558309668e-09,
      "loss": 1.4324,
      "step": 27878
    },
    {
      "epoch": 0.9984063602342114,
      "grad_norm": 1.9988292455673218,
      "learning_rate": 1.3023163117242832e-09,
      "loss": 1.3851,
      "step": 27879
    },
    {
      "epoch": 0.9984421723637796,
      "grad_norm": 1.5393435955047607,
      "learning_rate": 1.2437929211017718e-09,
      "loss": 1.4378,
      "step": 27880
    },
    {
      "epoch": 0.9984779844933479,
      "grad_norm": 1.34359872341156,
      "learning_rate": 1.1866148847516912e-09,
      "loss": 1.4509,
      "step": 27881
    },
    {
      "epoch": 0.9985137966229162,
      "grad_norm": 1.7370821237564087,
      "learning_rate": 1.1307822034511973e-09,
      "loss": 1.3819,
      "step": 27882
    },
    {
      "epoch": 0.9985496087524844,
      "grad_norm": 2.0988402366638184,
      "learning_rate": 1.0762948779441396e-09,
      "loss": 1.3734,
      "step": 27883
    },
    {
      "epoch": 0.9985854208820527,
      "grad_norm": 1.5026785135269165,
      "learning_rate": 1.0231529089632652e-09,
      "loss": 1.4251,
      "step": 27884
    },
    {
      "epoch": 0.9986212330116211,
      "grad_norm": 2.722470283508301,
      "learning_rate": 9.713562972302193e-10,
      "loss": 1.5023,
      "step": 27885
    },
    {
      "epoch": 0.9986570451411894,
      "grad_norm": 1.7294663190841675,
      "learning_rate": 9.209050434333399e-10,
      "loss": 1.3209,
      "step": 27886
    },
    {
      "epoch": 0.9986928572707576,
      "grad_norm": 2.4848201274871826,
      "learning_rate": 8.717991482609655e-10,
      "loss": 1.7707,
      "step": 27887
    },
    {
      "epoch": 0.9987286694003259,
      "grad_norm": 1.535007357597351,
      "learning_rate": 8.240386123681276e-10,
      "loss": 1.4537,
      "step": 27888
    },
    {
      "epoch": 0.9987644815298942,
      "grad_norm": 1.3857157230377197,
      "learning_rate": 7.776234363987555e-10,
      "loss": 1.79,
      "step": 27889
    },
    {
      "epoch": 0.9988002936594624,
      "grad_norm": 1.4100427627563477,
      "learning_rate": 7.325536209856765e-10,
      "loss": 1.578,
      "step": 27890
    },
    {
      "epoch": 0.9988361057890307,
      "grad_norm": 1.3705559968948364,
      "learning_rate": 6.888291667173085e-10,
      "loss": 1.3029,
      "step": 27891
    },
    {
      "epoch": 0.9988719179185991,
      "grad_norm": 1.7069154977798462,
      "learning_rate": 6.464500741820701e-10,
      "loss": 1.4618,
      "step": 27892
    },
    {
      "epoch": 0.9989077300481674,
      "grad_norm": 2.2990357875823975,
      "learning_rate": 6.054163439683791e-10,
      "loss": 1.4391,
      "step": 27893
    },
    {
      "epoch": 0.9989435421777356,
      "grad_norm": 1.4286106824874878,
      "learning_rate": 5.65727976620245e-10,
      "loss": 1.5029,
      "step": 27894
    },
    {
      "epoch": 0.9989793543073039,
      "grad_norm": 1.4237838983535767,
      "learning_rate": 5.273849726705748e-10,
      "loss": 1.4531,
      "step": 27895
    },
    {
      "epoch": 0.9990151664368722,
      "grad_norm": 1.7203515768051147,
      "learning_rate": 4.903873326300712e-10,
      "loss": 1.3107,
      "step": 27896
    },
    {
      "epoch": 0.9990509785664404,
      "grad_norm": 1.3520474433898926,
      "learning_rate": 4.5473505700943664e-10,
      "loss": 1.3599,
      "step": 27897
    },
    {
      "epoch": 0.9990867906960087,
      "grad_norm": 1.73139488697052,
      "learning_rate": 4.2042814627496483e-10,
      "loss": 1.4747,
      "step": 27898
    },
    {
      "epoch": 0.9991226028255771,
      "grad_norm": 1.3812834024429321,
      "learning_rate": 3.874666008929495e-10,
      "loss": 1.3831,
      "step": 27899
    },
    {
      "epoch": 0.9991584149551453,
      "grad_norm": 1.7007373571395874,
      "learning_rate": 3.558504213074798e-10,
      "loss": 1.186,
      "step": 27900
    },
    {
      "epoch": 0.9991942270847136,
      "grad_norm": 2.0018959045410156,
      "learning_rate": 3.255796079404405e-10,
      "loss": 1.5158,
      "step": 27901
    },
    {
      "epoch": 0.9992300392142819,
      "grad_norm": 1.486700177192688,
      "learning_rate": 2.9665416120261414e-10,
      "loss": 1.6186,
      "step": 27902
    },
    {
      "epoch": 0.9992658513438502,
      "grad_norm": 2.046440601348877,
      "learning_rate": 2.69074081493681e-10,
      "loss": 1.1622,
      "step": 27903
    },
    {
      "epoch": 0.9993016634734184,
      "grad_norm": 1.8606266975402832,
      "learning_rate": 2.4283936915781014e-10,
      "loss": 1.4256,
      "step": 27904
    },
    {
      "epoch": 0.9993374756029867,
      "grad_norm": 1.6825835704803467,
      "learning_rate": 2.1795002457247748e-10,
      "loss": 1.3446,
      "step": 27905
    },
    {
      "epoch": 0.9993732877325551,
      "grad_norm": 1.6700379848480225,
      "learning_rate": 1.9440604807074992e-10,
      "loss": 1.4342,
      "step": 27906
    },
    {
      "epoch": 0.9994090998621233,
      "grad_norm": 1.4203354120254517,
      "learning_rate": 1.7220743995238763e-10,
      "loss": 1.5311,
      "step": 27907
    },
    {
      "epoch": 0.9994449119916916,
      "grad_norm": 1.6370415687561035,
      "learning_rate": 1.5135420052825312e-10,
      "loss": 1.5626,
      "step": 27908
    },
    {
      "epoch": 0.9994807241212599,
      "grad_norm": 1.8386884927749634,
      "learning_rate": 1.318463300870043e-10,
      "loss": 1.1738,
      "step": 27909
    },
    {
      "epoch": 0.9995165362508281,
      "grad_norm": 2.393831253051758,
      "learning_rate": 1.136838288728903e-10,
      "loss": 1.4101,
      "step": 27910
    },
    {
      "epoch": 0.9995523483803964,
      "grad_norm": 1.738539695739746,
      "learning_rate": 9.68666971412624e-11,
      "loss": 1.7001,
      "step": 27911
    },
    {
      "epoch": 0.9995881605099647,
      "grad_norm": 1.6437656879425049,
      "learning_rate": 8.139493511416518e-11,
      "loss": 1.5699,
      "step": 27912
    },
    {
      "epoch": 0.9996239726395331,
      "grad_norm": 1.6513583660125732,
      "learning_rate": 6.726854300254104e-11,
      "loss": 1.7211,
      "step": 27913
    },
    {
      "epoch": 0.9996597847691013,
      "grad_norm": 1.3128859996795654,
      "learning_rate": 5.448752098402565e-11,
      "loss": 1.0227,
      "step": 27914
    },
    {
      "epoch": 0.9996955968986696,
      "grad_norm": 1.9902230501174927,
      "learning_rate": 4.3051869258459163e-11,
      "loss": 1.5862,
      "step": 27915
    },
    {
      "epoch": 0.9997314090282379,
      "grad_norm": 2.3277649879455566,
      "learning_rate": 3.296158795906834e-11,
      "loss": 1.4745,
      "step": 27916
    },
    {
      "epoch": 0.9997672211578061,
      "grad_norm": 1.625291109085083,
      "learning_rate": 2.421667721907994e-11,
      "loss": 1.3652,
      "step": 27917
    },
    {
      "epoch": 0.9998030332873744,
      "grad_norm": 1.4065899848937988,
      "learning_rate": 1.68171371606185e-11,
      "loss": 1.5924,
      "step": 27918
    },
    {
      "epoch": 0.9998388454169427,
      "grad_norm": 1.4588338136672974,
      "learning_rate": 1.0762967894706321e-11,
      "loss": 1.3983,
      "step": 27919
    },
    {
      "epoch": 0.9998746575465111,
      "grad_norm": 1.98700749874115,
      "learning_rate": 6.054169487956784e-12,
      "loss": 1.338,
      "step": 27920
    },
    {
      "epoch": 0.9999104696760793,
      "grad_norm": 1.688843846321106,
      "learning_rate": 2.690742006983271e-12,
      "loss": 1.1956,
      "step": 27921
    },
    {
      "epoch": 0.9999462818056476,
      "grad_norm": 1.4015154838562012,
      "learning_rate": 6.726855072969329e-13,
      "loss": 1.3722,
      "step": 27922
    },
    {
      "epoch": 0.9999820939352159,
      "grad_norm": 1.3305832147598267,
      "learning_rate": 0.0,
      "loss": 1.4175,
      "step": 27923
    },
    {
      "epoch": 0.9999820939352159,
      "step": 27923,
      "total_flos": 2.3643443035111424e+18,
      "train_loss": 1.5265222442264104,
      "train_runtime": 42480.1433,
      "train_samples_per_second": 5.259,
      "train_steps_per_second": 0.657
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 27923,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.3643443035111424e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}