{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 404,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0024752475247524753,
      "grad_norm": 0.35921207070350647,
      "learning_rate": 1e-05,
      "loss": 1.8027,
      "step": 1
    },
    {
      "epoch": 0.0049504950495049506,
      "grad_norm": 0.39715632796287537,
      "learning_rate": 9.975247524752477e-06,
      "loss": 1.9376,
      "step": 2
    },
    {
      "epoch": 0.007425742574257425,
      "grad_norm": 0.4083423614501953,
      "learning_rate": 9.950495049504951e-06,
      "loss": 1.9841,
      "step": 3
    },
    {
      "epoch": 0.009900990099009901,
      "grad_norm": 0.40308135747909546,
      "learning_rate": 9.925742574257427e-06,
      "loss": 1.9444,
      "step": 4
    },
    {
      "epoch": 0.012376237623762377,
      "grad_norm": 0.370197057723999,
      "learning_rate": 9.900990099009901e-06,
      "loss": 1.7989,
      "step": 5
    },
    {
      "epoch": 0.01485148514851485,
      "grad_norm": 0.3891890347003937,
      "learning_rate": 9.876237623762377e-06,
      "loss": 1.8703,
      "step": 6
    },
    {
      "epoch": 0.017326732673267328,
      "grad_norm": 0.39329928159713745,
      "learning_rate": 9.851485148514852e-06,
      "loss": 1.9423,
      "step": 7
    },
    {
      "epoch": 0.019801980198019802,
      "grad_norm": 0.38817495107650757,
      "learning_rate": 9.826732673267328e-06,
      "loss": 1.8705,
      "step": 8
    },
    {
      "epoch": 0.022277227722772276,
      "grad_norm": 0.3646544814109802,
      "learning_rate": 9.801980198019802e-06,
      "loss": 1.7702,
      "step": 9
    },
    {
      "epoch": 0.024752475247524754,
      "grad_norm": 0.3860485851764679,
      "learning_rate": 9.777227722772278e-06,
      "loss": 1.863,
      "step": 10
    },
    {
      "epoch": 0.027227722772277228,
      "grad_norm": 0.34650540351867676,
      "learning_rate": 9.752475247524754e-06,
      "loss": 1.7751,
      "step": 11
    },
    {
      "epoch": 0.0297029702970297,
      "grad_norm": 0.3583658039569855,
      "learning_rate": 9.727722772277228e-06,
      "loss": 1.8129,
      "step": 12
    },
    {
      "epoch": 0.03217821782178218,
      "grad_norm": 0.3476351797580719,
      "learning_rate": 9.702970297029704e-06,
      "loss": 1.8005,
      "step": 13
    },
    {
      "epoch": 0.034653465346534656,
      "grad_norm": 0.34600403904914856,
      "learning_rate": 9.678217821782178e-06,
      "loss": 1.8254,
      "step": 14
    },
    {
      "epoch": 0.03712871287128713,
      "grad_norm": 0.3049982786178589,
      "learning_rate": 9.653465346534654e-06,
      "loss": 1.6756,
      "step": 15
    },
    {
      "epoch": 0.039603960396039604,
      "grad_norm": 0.3400605618953705,
      "learning_rate": 9.628712871287129e-06,
      "loss": 1.8066,
      "step": 16
    },
    {
      "epoch": 0.04207920792079208,
      "grad_norm": 0.29243192076683044,
      "learning_rate": 9.603960396039604e-06,
      "loss": 1.7034,
      "step": 17
    },
    {
      "epoch": 0.04455445544554455,
      "grad_norm": 0.3139927089214325,
      "learning_rate": 9.579207920792079e-06,
      "loss": 1.7889,
      "step": 18
    },
    {
      "epoch": 0.04702970297029703,
      "grad_norm": 0.29722219705581665,
      "learning_rate": 9.554455445544555e-06,
      "loss": 1.7024,
      "step": 19
    },
    {
      "epoch": 0.04950495049504951,
      "grad_norm": 0.29992449283599854,
      "learning_rate": 9.52970297029703e-06,
      "loss": 1.7444,
      "step": 20
    },
    {
      "epoch": 0.05198019801980198,
      "grad_norm": 0.3000761568546295,
      "learning_rate": 9.504950495049505e-06,
      "loss": 1.6878,
      "step": 21
    },
    {
      "epoch": 0.054455445544554455,
      "grad_norm": 0.2957460284233093,
      "learning_rate": 9.480198019801981e-06,
      "loss": 1.7008,
      "step": 22
    },
    {
      "epoch": 0.05693069306930693,
      "grad_norm": 0.2921546995639801,
      "learning_rate": 9.455445544554455e-06,
      "loss": 1.749,
      "step": 23
    },
    {
      "epoch": 0.0594059405940594,
      "grad_norm": 0.27504685521125793,
      "learning_rate": 9.430693069306931e-06,
      "loss": 1.6845,
      "step": 24
    },
    {
      "epoch": 0.06188118811881188,
      "grad_norm": 0.2481079399585724,
      "learning_rate": 9.405940594059405e-06,
      "loss": 1.5942,
      "step": 25
    },
    {
      "epoch": 0.06435643564356436,
      "grad_norm": 0.2514517903327942,
      "learning_rate": 9.381188118811881e-06,
      "loss": 1.6142,
      "step": 26
    },
    {
      "epoch": 0.06683168316831684,
      "grad_norm": 0.23834620416164398,
      "learning_rate": 9.356435643564357e-06,
      "loss": 1.5875,
      "step": 27
    },
    {
      "epoch": 0.06930693069306931,
      "grad_norm": 0.246104896068573,
      "learning_rate": 9.331683168316833e-06,
      "loss": 1.6119,
      "step": 28
    },
    {
      "epoch": 0.07178217821782178,
      "grad_norm": 0.21701523661613464,
      "learning_rate": 9.306930693069308e-06,
      "loss": 1.531,
      "step": 29
    },
    {
      "epoch": 0.07425742574257425,
      "grad_norm": 0.26559269428253174,
      "learning_rate": 9.282178217821784e-06,
      "loss": 1.6806,
      "step": 30
    },
    {
      "epoch": 0.07673267326732673,
      "grad_norm": 0.23772238194942474,
      "learning_rate": 9.257425742574258e-06,
      "loss": 1.5922,
      "step": 31
    },
    {
      "epoch": 0.07920792079207921,
      "grad_norm": 0.23874586820602417,
      "learning_rate": 9.232673267326734e-06,
      "loss": 1.567,
      "step": 32
    },
    {
      "epoch": 0.08168316831683169,
      "grad_norm": 0.2239362597465515,
      "learning_rate": 9.20792079207921e-06,
      "loss": 1.5437,
      "step": 33
    },
    {
      "epoch": 0.08415841584158416,
      "grad_norm": 0.21884024143218994,
      "learning_rate": 9.183168316831684e-06,
      "loss": 1.485,
      "step": 34
    },
    {
      "epoch": 0.08663366336633663,
      "grad_norm": 0.22681422531604767,
      "learning_rate": 9.15841584158416e-06,
      "loss": 1.5176,
      "step": 35
    },
    {
      "epoch": 0.0891089108910891,
      "grad_norm": 0.23015110194683075,
      "learning_rate": 9.133663366336634e-06,
      "loss": 1.5312,
      "step": 36
    },
    {
      "epoch": 0.09158415841584158,
      "grad_norm": 0.21569128334522247,
      "learning_rate": 9.10891089108911e-06,
      "loss": 1.4879,
      "step": 37
    },
    {
      "epoch": 0.09405940594059406,
      "grad_norm": 0.247847318649292,
      "learning_rate": 9.084158415841585e-06,
      "loss": 1.5873,
      "step": 38
    },
    {
      "epoch": 0.09653465346534654,
      "grad_norm": 0.20600910484790802,
      "learning_rate": 9.05940594059406e-06,
      "loss": 1.476,
      "step": 39
    },
    {
      "epoch": 0.09900990099009901,
      "grad_norm": 0.23611964285373688,
      "learning_rate": 9.034653465346535e-06,
      "loss": 1.5556,
      "step": 40
    },
    {
      "epoch": 0.10148514851485149,
      "grad_norm": 0.22053782641887665,
      "learning_rate": 9.009900990099011e-06,
      "loss": 1.5271,
      "step": 41
    },
    {
      "epoch": 0.10396039603960396,
      "grad_norm": 0.21321026980876923,
      "learning_rate": 8.985148514851487e-06,
      "loss": 1.47,
      "step": 42
    },
    {
      "epoch": 0.10643564356435643,
      "grad_norm": 0.20851102471351624,
      "learning_rate": 8.960396039603961e-06,
      "loss": 1.4747,
      "step": 43
    },
    {
      "epoch": 0.10891089108910891,
      "grad_norm": 0.20043537020683289,
      "learning_rate": 8.935643564356437e-06,
      "loss": 1.4561,
      "step": 44
    },
    {
      "epoch": 0.11138613861386139,
      "grad_norm": 0.2246498167514801,
      "learning_rate": 8.910891089108911e-06,
      "loss": 1.5265,
      "step": 45
    },
    {
      "epoch": 0.11386138613861387,
      "grad_norm": 0.19988679885864258,
      "learning_rate": 8.886138613861387e-06,
      "loss": 1.4764,
      "step": 46
    },
    {
      "epoch": 0.11633663366336634,
      "grad_norm": 0.1900709569454193,
      "learning_rate": 8.861386138613862e-06,
      "loss": 1.3958,
      "step": 47
    },
    {
      "epoch": 0.1188118811881188,
      "grad_norm": 0.18632960319519043,
      "learning_rate": 8.836633663366338e-06,
      "loss": 1.3834,
      "step": 48
    },
    {
      "epoch": 0.12128712871287128,
      "grad_norm": 0.18428552150726318,
      "learning_rate": 8.811881188118812e-06,
      "loss": 1.3753,
      "step": 49
    },
    {
      "epoch": 0.12376237623762376,
      "grad_norm": 0.2110532969236374,
      "learning_rate": 8.787128712871288e-06,
      "loss": 1.3957,
      "step": 50
    },
    {
      "epoch": 0.12623762376237624,
      "grad_norm": 0.17746232450008392,
      "learning_rate": 8.762376237623764e-06,
      "loss": 1.3436,
      "step": 51
    },
    {
      "epoch": 0.12871287128712872,
      "grad_norm": 0.16532298922538757,
      "learning_rate": 8.737623762376238e-06,
      "loss": 1.3575,
      "step": 52
    },
    {
      "epoch": 0.1311881188118812,
      "grad_norm": 0.1659340113401413,
      "learning_rate": 8.712871287128714e-06,
      "loss": 1.3793,
      "step": 53
    },
    {
      "epoch": 0.13366336633663367,
      "grad_norm": 0.16310887038707733,
      "learning_rate": 8.688118811881188e-06,
      "loss": 1.3364,
      "step": 54
    },
    {
      "epoch": 0.13613861386138615,
      "grad_norm": 0.17784424126148224,
      "learning_rate": 8.663366336633664e-06,
      "loss": 1.3824,
      "step": 55
    },
    {
      "epoch": 0.13861386138613863,
      "grad_norm": 0.1612381488084793,
      "learning_rate": 8.638613861386139e-06,
      "loss": 1.3519,
      "step": 56
    },
    {
      "epoch": 0.14108910891089108,
      "grad_norm": 0.16395767033100128,
      "learning_rate": 8.613861386138615e-06,
      "loss": 1.3209,
      "step": 57
    },
    {
      "epoch": 0.14356435643564355,
      "grad_norm": 0.15274788439273834,
      "learning_rate": 8.58910891089109e-06,
      "loss": 1.3076,
      "step": 58
    },
    {
      "epoch": 0.14603960396039603,
      "grad_norm": 0.26918065547943115,
      "learning_rate": 8.564356435643565e-06,
      "loss": 1.3272,
      "step": 59
    },
    {
      "epoch": 0.1485148514851485,
      "grad_norm": 0.1536840796470642,
      "learning_rate": 8.53960396039604e-06,
      "loss": 1.2732,
      "step": 60
    },
    {
      "epoch": 0.15099009900990099,
      "grad_norm": 0.15451501309871674,
      "learning_rate": 8.514851485148515e-06,
      "loss": 1.3142,
      "step": 61
    },
    {
      "epoch": 0.15346534653465346,
      "grad_norm": 0.15294960141181946,
      "learning_rate": 8.490099009900991e-06,
      "loss": 1.3307,
      "step": 62
    },
    {
      "epoch": 0.15594059405940594,
      "grad_norm": 0.1928381323814392,
      "learning_rate": 8.465346534653465e-06,
      "loss": 1.311,
      "step": 63
    },
    {
      "epoch": 0.15841584158415842,
      "grad_norm": 0.15054042637348175,
      "learning_rate": 8.440594059405941e-06,
      "loss": 1.3191,
      "step": 64
    },
    {
      "epoch": 0.1608910891089109,
      "grad_norm": 0.13962224125862122,
      "learning_rate": 8.415841584158416e-06,
      "loss": 1.2381,
      "step": 65
    },
    {
      "epoch": 0.16336633663366337,
      "grad_norm": 0.1442529261112213,
      "learning_rate": 8.391089108910891e-06,
      "loss": 1.2732,
      "step": 66
    },
    {
      "epoch": 0.16584158415841585,
      "grad_norm": 0.14938265085220337,
      "learning_rate": 8.366336633663367e-06,
      "loss": 1.2835,
      "step": 67
    },
    {
      "epoch": 0.16831683168316833,
      "grad_norm": 0.17844296991825104,
      "learning_rate": 8.341584158415842e-06,
      "loss": 1.3027,
      "step": 68
    },
    {
      "epoch": 0.1707920792079208,
      "grad_norm": 0.14419397711753845,
      "learning_rate": 8.316831683168318e-06,
      "loss": 1.3032,
      "step": 69
    },
    {
      "epoch": 0.17326732673267325,
      "grad_norm": 0.16270500421524048,
      "learning_rate": 8.292079207920792e-06,
      "loss": 1.2799,
      "step": 70
    },
    {
      "epoch": 0.17574257425742573,
      "grad_norm": 0.1458730697631836,
      "learning_rate": 8.267326732673268e-06,
      "loss": 1.2522,
      "step": 71
    },
    {
      "epoch": 0.1782178217821782,
      "grad_norm": 0.14367038011550903,
      "learning_rate": 8.242574257425742e-06,
      "loss": 1.2369,
      "step": 72
    },
    {
      "epoch": 0.1806930693069307,
      "grad_norm": 0.1553303301334381,
      "learning_rate": 8.217821782178218e-06,
      "loss": 1.3366,
      "step": 73
    },
    {
      "epoch": 0.18316831683168316,
      "grad_norm": 0.14889273047447205,
      "learning_rate": 8.193069306930692e-06,
      "loss": 1.3084,
      "step": 74
    },
    {
      "epoch": 0.18564356435643564,
      "grad_norm": 0.14996132254600525,
      "learning_rate": 8.168316831683168e-06,
      "loss": 1.2643,
      "step": 75
    },
    {
      "epoch": 0.18811881188118812,
      "grad_norm": 0.14123128354549408,
      "learning_rate": 8.143564356435644e-06,
      "loss": 1.2621,
      "step": 76
    },
    {
      "epoch": 0.1905940594059406,
      "grad_norm": 0.1484861671924591,
      "learning_rate": 8.11881188118812e-06,
      "loss": 1.3082,
      "step": 77
    },
    {
      "epoch": 0.19306930693069307,
      "grad_norm": 0.1556919664144516,
      "learning_rate": 8.094059405940595e-06,
      "loss": 1.2785,
      "step": 78
    },
    {
      "epoch": 0.19554455445544555,
      "grad_norm": 0.1459795981645584,
      "learning_rate": 8.06930693069307e-06,
      "loss": 1.1944,
      "step": 79
    },
    {
      "epoch": 0.19801980198019803,
      "grad_norm": 0.13556146621704102,
      "learning_rate": 8.044554455445545e-06,
      "loss": 1.2475,
      "step": 80
    },
    {
      "epoch": 0.2004950495049505,
      "grad_norm": 0.14718809723854065,
      "learning_rate": 8.019801980198021e-06,
      "loss": 1.2353,
      "step": 81
    },
    {
      "epoch": 0.20297029702970298,
      "grad_norm": 0.14329920709133148,
      "learning_rate": 7.995049504950497e-06,
      "loss": 1.252,
      "step": 82
    },
    {
      "epoch": 0.20544554455445543,
      "grad_norm": 0.14620834589004517,
      "learning_rate": 7.970297029702971e-06,
      "loss": 1.2394,
      "step": 83
    },
    {
      "epoch": 0.2079207920792079,
      "grad_norm": 0.14883743226528168,
      "learning_rate": 7.945544554455447e-06,
      "loss": 1.2461,
      "step": 84
    },
    {
      "epoch": 0.2103960396039604,
      "grad_norm": 0.14748890697956085,
      "learning_rate": 7.920792079207921e-06,
      "loss": 1.2049,
      "step": 85
    },
    {
      "epoch": 0.21287128712871287,
      "grad_norm": 0.1422479897737503,
      "learning_rate": 7.896039603960397e-06,
      "loss": 1.2621,
      "step": 86
    },
    {
      "epoch": 0.21534653465346534,
      "grad_norm": 0.1533108800649643,
      "learning_rate": 7.871287128712872e-06,
      "loss": 1.2211,
      "step": 87
    },
    {
      "epoch": 0.21782178217821782,
      "grad_norm": 0.14151449501514435,
      "learning_rate": 7.846534653465348e-06,
      "loss": 1.2094,
      "step": 88
    },
    {
      "epoch": 0.2202970297029703,
      "grad_norm": 0.14519892632961273,
      "learning_rate": 7.821782178217822e-06,
      "loss": 1.2665,
      "step": 89
    },
    {
      "epoch": 0.22277227722772278,
      "grad_norm": 0.15295307338237762,
      "learning_rate": 7.797029702970298e-06,
      "loss": 1.2438,
      "step": 90
    },
    {
      "epoch": 0.22524752475247525,
      "grad_norm": 0.14052022993564606,
      "learning_rate": 7.772277227722774e-06,
      "loss": 1.2013,
      "step": 91
    },
    {
      "epoch": 0.22772277227722773,
      "grad_norm": 0.208185613155365,
      "learning_rate": 7.747524752475248e-06,
      "loss": 1.2212,
      "step": 92
    },
    {
      "epoch": 0.2301980198019802,
      "grad_norm": 0.14823101460933685,
      "learning_rate": 7.722772277227724e-06,
      "loss": 1.1976,
      "step": 93
    },
    {
      "epoch": 0.23267326732673269,
      "grad_norm": 0.22633716464042664,
      "learning_rate": 7.698019801980198e-06,
      "loss": 1.2397,
      "step": 94
    },
    {
      "epoch": 0.23514851485148514,
      "grad_norm": 0.15597891807556152,
      "learning_rate": 7.673267326732674e-06,
      "loss": 1.2253,
      "step": 95
    },
    {
      "epoch": 0.2376237623762376,
      "grad_norm": 0.14791743457317352,
      "learning_rate": 7.648514851485149e-06,
      "loss": 1.1459,
      "step": 96
    },
    {
      "epoch": 0.2400990099009901,
      "grad_norm": 0.14784985780715942,
      "learning_rate": 7.6237623762376246e-06,
      "loss": 1.2072,
      "step": 97
    },
    {
      "epoch": 0.24257425742574257,
      "grad_norm": 0.1492685228586197,
      "learning_rate": 7.5990099009901e-06,
      "loss": 1.2038,
      "step": 98
    },
    {
      "epoch": 0.24504950495049505,
      "grad_norm": 0.14586341381072998,
      "learning_rate": 7.574257425742575e-06,
      "loss": 1.1988,
      "step": 99
    },
    {
      "epoch": 0.24752475247524752,
      "grad_norm": 0.15133561193943024,
      "learning_rate": 7.54950495049505e-06,
      "loss": 1.217,
      "step": 100
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.14834076166152954,
      "learning_rate": 7.524752475247525e-06,
      "loss": 1.1654,
      "step": 101
    },
    {
      "epoch": 0.2524752475247525,
      "grad_norm": 0.1549738198518753,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.2182,
      "step": 102
    },
    {
      "epoch": 0.25495049504950495,
      "grad_norm": 0.14767996966838837,
      "learning_rate": 7.475247524752476e-06,
      "loss": 1.1842,
      "step": 103
    },
    {
      "epoch": 0.25742574257425743,
      "grad_norm": 0.16162672638893127,
      "learning_rate": 7.450495049504951e-06,
      "loss": 1.1801,
      "step": 104
    },
    {
      "epoch": 0.2599009900990099,
      "grad_norm": 0.1689019352197647,
      "learning_rate": 7.425742574257426e-06,
      "loss": 1.1873,
      "step": 105
    },
    {
      "epoch": 0.2623762376237624,
      "grad_norm": 0.18430471420288086,
      "learning_rate": 7.4009900990099015e-06,
      "loss": 1.2109,
      "step": 106
    },
    {
      "epoch": 0.26485148514851486,
      "grad_norm": 0.15747009217739105,
      "learning_rate": 7.376237623762377e-06,
      "loss": 1.196,
      "step": 107
    },
    {
      "epoch": 0.26732673267326734,
      "grad_norm": 0.16308680176734924,
      "learning_rate": 7.351485148514852e-06,
      "loss": 1.1703,
      "step": 108
    },
    {
      "epoch": 0.2698019801980198,
      "grad_norm": 0.1602238267660141,
      "learning_rate": 7.326732673267327e-06,
      "loss": 1.2002,
      "step": 109
    },
    {
      "epoch": 0.2722772277227723,
      "grad_norm": 0.15777747333049774,
      "learning_rate": 7.301980198019802e-06,
      "loss": 1.1728,
      "step": 110
    },
    {
      "epoch": 0.2747524752475248,
      "grad_norm": 0.1543913632631302,
      "learning_rate": 7.277227722772278e-06,
      "loss": 1.1419,
      "step": 111
    },
    {
      "epoch": 0.27722772277227725,
      "grad_norm": 0.1703396588563919,
      "learning_rate": 7.252475247524753e-06,
      "loss": 1.2023,
      "step": 112
    },
    {
      "epoch": 0.27970297029702973,
      "grad_norm": 0.16434207558631897,
      "learning_rate": 7.227722772277228e-06,
      "loss": 1.2137,
      "step": 113
    },
    {
      "epoch": 0.28217821782178215,
      "grad_norm": 0.17134983837604523,
      "learning_rate": 7.202970297029703e-06,
      "loss": 1.2105,
      "step": 114
    },
    {
      "epoch": 0.28465346534653463,
      "grad_norm": 0.19009283185005188,
      "learning_rate": 7.1782178217821785e-06,
      "loss": 1.1768,
      "step": 115
    },
    {
      "epoch": 0.2871287128712871,
      "grad_norm": 0.16358834505081177,
      "learning_rate": 7.153465346534654e-06,
      "loss": 1.1668,
      "step": 116
    },
    {
      "epoch": 0.2896039603960396,
      "grad_norm": 0.2351948767900467,
      "learning_rate": 7.128712871287129e-06,
      "loss": 1.1706,
      "step": 117
    },
    {
      "epoch": 0.29207920792079206,
      "grad_norm": 0.18060773611068726,
      "learning_rate": 7.103960396039604e-06,
      "loss": 1.1659,
      "step": 118
    },
    {
      "epoch": 0.29455445544554454,
      "grad_norm": 0.19292417168617249,
      "learning_rate": 7.079207920792079e-06,
      "loss": 1.1685,
      "step": 119
    },
    {
      "epoch": 0.297029702970297,
      "grad_norm": 0.18404839932918549,
      "learning_rate": 7.054455445544555e-06,
      "loss": 1.133,
      "step": 120
    },
    {
      "epoch": 0.2995049504950495,
      "grad_norm": 0.17319351434707642,
      "learning_rate": 7.02970297029703e-06,
      "loss": 1.1565,
      "step": 121
    },
    {
      "epoch": 0.30198019801980197,
      "grad_norm": 0.17463432252407074,
      "learning_rate": 7.004950495049505e-06,
      "loss": 1.1614,
      "step": 122
    },
    {
      "epoch": 0.30445544554455445,
      "grad_norm": 0.17117349803447723,
      "learning_rate": 6.98019801980198e-06,
      "loss": 1.1595,
      "step": 123
    },
    {
      "epoch": 0.3069306930693069,
      "grad_norm": 0.17456741631031036,
      "learning_rate": 6.9554455445544555e-06,
      "loss": 1.1249,
      "step": 124
    },
    {
      "epoch": 0.3094059405940594,
      "grad_norm": 0.2406443953514099,
      "learning_rate": 6.930693069306931e-06,
      "loss": 1.1676,
      "step": 125
    },
    {
      "epoch": 0.3118811881188119,
      "grad_norm": 0.16312997043132782,
      "learning_rate": 6.905940594059406e-06,
      "loss": 1.1581,
      "step": 126
    },
    {
      "epoch": 0.31435643564356436,
      "grad_norm": 0.17027762532234192,
      "learning_rate": 6.881188118811881e-06,
      "loss": 1.1484,
      "step": 127
    },
    {
      "epoch": 0.31683168316831684,
      "grad_norm": 0.17338208854198456,
      "learning_rate": 6.856435643564358e-06,
      "loss": 1.1782,
      "step": 128
    },
    {
      "epoch": 0.3193069306930693,
      "grad_norm": 0.16886372864246368,
      "learning_rate": 6.831683168316833e-06,
      "loss": 1.1233,
      "step": 129
    },
    {
      "epoch": 0.3217821782178218,
      "grad_norm": 0.1658564656972885,
      "learning_rate": 6.806930693069308e-06,
      "loss": 1.1433,
      "step": 130
    },
    {
      "epoch": 0.32425742574257427,
      "grad_norm": 0.2258491814136505,
      "learning_rate": 6.782178217821783e-06,
      "loss": 1.1599,
      "step": 131
    },
    {
      "epoch": 0.32673267326732675,
      "grad_norm": 0.16513760387897491,
      "learning_rate": 6.757425742574258e-06,
      "loss": 1.1104,
      "step": 132
    },
    {
      "epoch": 0.3292079207920792,
      "grad_norm": 0.16575312614440918,
      "learning_rate": 6.732673267326733e-06,
      "loss": 1.1428,
      "step": 133
    },
    {
      "epoch": 0.3316831683168317,
      "grad_norm": 0.17043238878250122,
      "learning_rate": 6.707920792079209e-06,
      "loss": 1.1618,
      "step": 134
    },
    {
      "epoch": 0.3341584158415842,
      "grad_norm": 0.22997120022773743,
      "learning_rate": 6.683168316831684e-06,
      "loss": 1.1464,
      "step": 135
    },
    {
      "epoch": 0.33663366336633666,
      "grad_norm": 0.16962502896785736,
      "learning_rate": 6.6584158415841595e-06,
      "loss": 1.134,
      "step": 136
    },
    {
      "epoch": 0.33910891089108913,
      "grad_norm": 0.17967693507671356,
      "learning_rate": 6.633663366336635e-06,
      "loss": 1.1329,
      "step": 137
    },
    {
      "epoch": 0.3415841584158416,
      "grad_norm": 0.17692621052265167,
      "learning_rate": 6.60891089108911e-06,
      "loss": 1.1281,
      "step": 138
    },
    {
      "epoch": 0.34405940594059403,
      "grad_norm": 0.19165442883968353,
      "learning_rate": 6.584158415841585e-06,
      "loss": 1.1643,
      "step": 139
    },
    {
      "epoch": 0.3465346534653465,
      "grad_norm": 0.1611143797636032,
      "learning_rate": 6.55940594059406e-06,
      "loss": 1.099,
      "step": 140
    },
    {
      "epoch": 0.349009900990099,
      "grad_norm": 0.170062854886055,
      "learning_rate": 6.534653465346535e-06,
      "loss": 1.1317,
      "step": 141
    },
    {
      "epoch": 0.35148514851485146,
      "grad_norm": 0.1633332520723343,
      "learning_rate": 6.509900990099011e-06,
      "loss": 1.1481,
      "step": 142
    },
    {
      "epoch": 0.35396039603960394,
      "grad_norm": 0.16619764268398285,
      "learning_rate": 6.485148514851486e-06,
      "loss": 1.1216,
      "step": 143
    },
    {
      "epoch": 0.3564356435643564,
      "grad_norm": 0.16496872901916504,
      "learning_rate": 6.460396039603961e-06,
      "loss": 1.1103,
      "step": 144
    },
    {
      "epoch": 0.3589108910891089,
      "grad_norm": 0.16046729683876038,
      "learning_rate": 6.4356435643564364e-06,
      "loss": 1.1273,
      "step": 145
    },
    {
      "epoch": 0.3613861386138614,
      "grad_norm": 0.15890191495418549,
      "learning_rate": 6.4108910891089116e-06,
      "loss": 1.0927,
      "step": 146
    },
    {
      "epoch": 0.36386138613861385,
      "grad_norm": 0.15759097039699554,
      "learning_rate": 6.386138613861387e-06,
      "loss": 1.1565,
      "step": 147
    },
    {
      "epoch": 0.36633663366336633,
      "grad_norm": 0.21185743808746338,
      "learning_rate": 6.361386138613862e-06,
      "loss": 1.0925,
      "step": 148
    },
    {
      "epoch": 0.3688118811881188,
      "grad_norm": 0.15979894995689392,
      "learning_rate": 6.336633663366337e-06,
      "loss": 1.1229,
      "step": 149
    },
    {
      "epoch": 0.3712871287128713,
      "grad_norm": 0.16039994359016418,
      "learning_rate": 6.311881188118812e-06,
      "loss": 1.1225,
      "step": 150
    },
    {
      "epoch": 0.37376237623762376,
      "grad_norm": 0.16320356726646423,
      "learning_rate": 6.287128712871288e-06,
      "loss": 1.1513,
      "step": 151
    },
    {
      "epoch": 0.37623762376237624,
      "grad_norm": 0.1686553657054901,
      "learning_rate": 6.262376237623763e-06,
      "loss": 1.1172,
      "step": 152
    },
    {
      "epoch": 0.3787128712871287,
      "grad_norm": 0.164052814245224,
      "learning_rate": 6.237623762376238e-06,
      "loss": 1.1046,
      "step": 153
    },
    {
      "epoch": 0.3811881188118812,
      "grad_norm": 0.15217240154743195,
      "learning_rate": 6.212871287128713e-06,
      "loss": 1.1111,
      "step": 154
    },
    {
      "epoch": 0.38366336633663367,
      "grad_norm": 0.1576368808746338,
      "learning_rate": 6.1881188118811885e-06,
      "loss": 1.0526,
      "step": 155
    },
    {
      "epoch": 0.38613861386138615,
      "grad_norm": 0.1566316783428192,
      "learning_rate": 6.163366336633664e-06,
      "loss": 1.1092,
      "step": 156
    },
    {
      "epoch": 0.3886138613861386,
      "grad_norm": 0.1590651422739029,
      "learning_rate": 6.138613861386139e-06,
      "loss": 1.0949,
      "step": 157
    },
    {
      "epoch": 0.3910891089108911,
      "grad_norm": 0.15860077738761902,
      "learning_rate": 6.113861386138614e-06,
      "loss": 1.0919,
      "step": 158
    },
    {
      "epoch": 0.3935643564356436,
      "grad_norm": 0.1543784886598587,
      "learning_rate": 6.08910891089109e-06,
      "loss": 1.1081,
      "step": 159
    },
    {
      "epoch": 0.39603960396039606,
      "grad_norm": 0.17123788595199585,
      "learning_rate": 6.064356435643565e-06,
      "loss": 1.1478,
      "step": 160
    },
    {
      "epoch": 0.39851485148514854,
      "grad_norm": 0.16677404940128326,
      "learning_rate": 6.03960396039604e-06,
      "loss": 1.1128,
      "step": 161
    },
    {
      "epoch": 0.400990099009901,
      "grad_norm": 0.21054880321025848,
      "learning_rate": 6.014851485148515e-06,
      "loss": 1.1037,
      "step": 162
    },
    {
      "epoch": 0.4034653465346535,
      "grad_norm": 0.15982946753501892,
      "learning_rate": 5.99009900990099e-06,
      "loss": 1.0831,
      "step": 163
    },
    {
      "epoch": 0.40594059405940597,
      "grad_norm": 0.14396493136882782,
      "learning_rate": 5.9653465346534655e-06,
      "loss": 1.0845,
      "step": 164
    },
    {
      "epoch": 0.4084158415841584,
      "grad_norm": 0.16700232028961182,
      "learning_rate": 5.940594059405941e-06,
      "loss": 1.0954,
      "step": 165
    },
    {
      "epoch": 0.41089108910891087,
      "grad_norm": 0.15970200300216675,
      "learning_rate": 5.915841584158416e-06,
      "loss": 1.0977,
      "step": 166
    },
    {
      "epoch": 0.41336633663366334,
      "grad_norm": 0.15419846773147583,
      "learning_rate": 5.891089108910891e-06,
      "loss": 1.1063,
      "step": 167
    },
    {
      "epoch": 0.4158415841584158,
      "grad_norm": 0.16501763463020325,
      "learning_rate": 5.866336633663367e-06,
      "loss": 1.1508,
      "step": 168
    },
    {
      "epoch": 0.4183168316831683,
      "grad_norm": 0.15167059004306793,
      "learning_rate": 5.841584158415842e-06,
      "loss": 1.092,
      "step": 169
    },
    {
      "epoch": 0.4207920792079208,
      "grad_norm": 0.15721780061721802,
      "learning_rate": 5.816831683168317e-06,
      "loss": 1.0915,
      "step": 170
    },
    {
      "epoch": 0.42326732673267325,
      "grad_norm": 0.15296679735183716,
      "learning_rate": 5.792079207920792e-06,
      "loss": 1.0453,
      "step": 171
    },
    {
      "epoch": 0.42574257425742573,
      "grad_norm": 0.16495901346206665,
      "learning_rate": 5.767326732673267e-06,
      "loss": 1.1493,
      "step": 172
    },
    {
      "epoch": 0.4282178217821782,
      "grad_norm": 0.1588982343673706,
      "learning_rate": 5.7425742574257425e-06,
      "loss": 1.0593,
      "step": 173
    },
    {
      "epoch": 0.4306930693069307,
      "grad_norm": 0.15599201619625092,
      "learning_rate": 5.717821782178218e-06,
      "loss": 1.0433,
      "step": 174
    },
    {
      "epoch": 0.43316831683168316,
      "grad_norm": 0.1536027193069458,
      "learning_rate": 5.693069306930693e-06,
      "loss": 1.1101,
      "step": 175
    },
    {
      "epoch": 0.43564356435643564,
      "grad_norm": 0.1521487981081009,
      "learning_rate": 5.668316831683169e-06,
      "loss": 1.0905,
      "step": 176
    },
    {
      "epoch": 0.4381188118811881,
      "grad_norm": 0.21931779384613037,
      "learning_rate": 5.643564356435644e-06,
      "loss": 1.1008,
      "step": 177
    },
    {
      "epoch": 0.4405940594059406,
      "grad_norm": 0.1681768000125885,
      "learning_rate": 5.61881188118812e-06,
      "loss": 1.0724,
      "step": 178
    },
    {
      "epoch": 0.4430693069306931,
      "grad_norm": 0.16810664534568787,
      "learning_rate": 5.594059405940595e-06,
      "loss": 1.1137,
      "step": 179
    },
    {
      "epoch": 0.44554455445544555,
      "grad_norm": 0.15952704846858978,
      "learning_rate": 5.56930693069307e-06,
      "loss": 1.0913,
      "step": 180
    },
    {
      "epoch": 0.44801980198019803,
      "grad_norm": 0.25783032178878784,
      "learning_rate": 5.544554455445545e-06,
      "loss": 1.1057,
      "step": 181
    },
    {
      "epoch": 0.4504950495049505,
      "grad_norm": 0.1551341712474823,
      "learning_rate": 5.519801980198021e-06,
      "loss": 1.0744,
      "step": 182
    },
    {
      "epoch": 0.452970297029703,
      "grad_norm": 0.16667252779006958,
      "learning_rate": 5.495049504950496e-06,
      "loss": 1.066,
      "step": 183
    },
    {
      "epoch": 0.45544554455445546,
      "grad_norm": 0.16125273704528809,
      "learning_rate": 5.470297029702971e-06,
      "loss": 1.1511,
      "step": 184
    },
    {
      "epoch": 0.45792079207920794,
      "grad_norm": 0.16407977044582367,
      "learning_rate": 5.4455445544554465e-06,
      "loss": 1.0719,
      "step": 185
    },
    {
      "epoch": 0.4603960396039604,
      "grad_norm": 0.17295385897159576,
      "learning_rate": 5.420792079207922e-06,
      "loss": 1.1174,
      "step": 186
    },
    {
      "epoch": 0.4628712871287129,
      "grad_norm": 0.18729552626609802,
      "learning_rate": 5.396039603960397e-06,
      "loss": 1.0352,
      "step": 187
    },
    {
      "epoch": 0.46534653465346537,
      "grad_norm": 0.1539359837770462,
      "learning_rate": 5.371287128712872e-06,
      "loss": 1.0807,
      "step": 188
    },
    {
      "epoch": 0.46782178217821785,
      "grad_norm": 0.16208483278751373,
      "learning_rate": 5.346534653465347e-06,
      "loss": 1.08,
      "step": 189
    },
    {
      "epoch": 0.47029702970297027,
      "grad_norm": 0.1604655236005783,
      "learning_rate": 5.321782178217822e-06,
      "loss": 1.0715,
      "step": 190
    },
    {
      "epoch": 0.47277227722772275,
      "grad_norm": 0.20206867158412933,
      "learning_rate": 5.297029702970298e-06,
      "loss": 1.0588,
      "step": 191
    },
    {
      "epoch": 0.4752475247524752,
      "grad_norm": 0.15372274816036224,
      "learning_rate": 5.272277227722773e-06,
      "loss": 1.0771,
      "step": 192
    },
    {
      "epoch": 0.4777227722772277,
      "grad_norm": 0.15498343110084534,
      "learning_rate": 5.247524752475248e-06,
      "loss": 1.06,
      "step": 193
    },
    {
      "epoch": 0.4801980198019802,
      "grad_norm": 0.15957452356815338,
      "learning_rate": 5.2227722772277234e-06,
      "loss": 1.0731,
      "step": 194
    },
    {
      "epoch": 0.48267326732673266,
      "grad_norm": 0.1622665375471115,
      "learning_rate": 5.1980198019801986e-06,
      "loss": 1.0962,
      "step": 195
    },
    {
      "epoch": 0.48514851485148514,
      "grad_norm": 0.1674077957868576,
      "learning_rate": 5.173267326732674e-06,
      "loss": 1.105,
      "step": 196
    },
    {
      "epoch": 0.4876237623762376,
      "grad_norm": 0.169521763920784,
      "learning_rate": 5.148514851485149e-06,
      "loss": 1.0811,
      "step": 197
    },
    {
      "epoch": 0.4900990099009901,
      "grad_norm": 0.1688235104084015,
      "learning_rate": 5.123762376237624e-06,
      "loss": 1.0708,
      "step": 198
    },
    {
      "epoch": 0.49257425742574257,
      "grad_norm": 0.15499871969223022,
      "learning_rate": 5.0990099009901e-06,
      "loss": 1.0947,
      "step": 199
    },
    {
      "epoch": 0.49504950495049505,
      "grad_norm": 0.15326109528541565,
      "learning_rate": 5.074257425742575e-06,
      "loss": 1.0676,
      "step": 200
    },
    {
      "epoch": 0.4975247524752475,
      "grad_norm": 0.16365967690944672,
      "learning_rate": 5.04950495049505e-06,
      "loss": 1.0792,
      "step": 201
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1605796068906784,
      "learning_rate": 5.024752475247525e-06,
      "loss": 1.0939,
      "step": 202
    },
    {
      "epoch": 0.5024752475247525,
      "grad_norm": 0.16988039016723633,
      "learning_rate": 5e-06,
      "loss": 1.0996,
      "step": 203
    },
    {
      "epoch": 0.504950495049505,
      "grad_norm": 0.16347338259220123,
      "learning_rate": 4.9752475247524755e-06,
      "loss": 1.0603,
      "step": 204
    },
    {
      "epoch": 0.5074257425742574,
      "grad_norm": 0.16206030547618866,
      "learning_rate": 4.950495049504951e-06,
      "loss": 1.0675,
      "step": 205
    },
    {
      "epoch": 0.5099009900990099,
      "grad_norm": 0.18767081201076508,
      "learning_rate": 4.925742574257426e-06,
      "loss": 1.0597,
      "step": 206
    },
    {
      "epoch": 0.5123762376237624,
      "grad_norm": 0.1522037833929062,
      "learning_rate": 4.900990099009901e-06,
      "loss": 1.0673,
      "step": 207
    },
    {
      "epoch": 0.5148514851485149,
      "grad_norm": 0.15843883156776428,
      "learning_rate": 4.876237623762377e-06,
      "loss": 1.0544,
      "step": 208
    },
    {
      "epoch": 0.5173267326732673,
      "grad_norm": 0.1723291128873825,
      "learning_rate": 4.851485148514852e-06,
      "loss": 1.0837,
      "step": 209
    },
    {
      "epoch": 0.5198019801980198,
      "grad_norm": 0.16350848972797394,
      "learning_rate": 4.826732673267327e-06,
      "loss": 1.0666,
      "step": 210
    },
    {
      "epoch": 0.5222772277227723,
      "grad_norm": 0.15768328309059143,
      "learning_rate": 4.801980198019802e-06,
      "loss": 1.0723,
      "step": 211
    },
    {
      "epoch": 0.5247524752475248,
      "grad_norm": 0.17364096641540527,
      "learning_rate": 4.777227722772277e-06,
      "loss": 1.0667,
      "step": 212
    },
    {
      "epoch": 0.5272277227722773,
      "grad_norm": 0.16839410364627838,
      "learning_rate": 4.7524752475247525e-06,
      "loss": 1.0728,
      "step": 213
    },
    {
      "epoch": 0.5297029702970297,
      "grad_norm": 0.16369131207466125,
      "learning_rate": 4.727722772277228e-06,
      "loss": 1.0732,
      "step": 214
    },
    {
      "epoch": 0.5321782178217822,
      "grad_norm": 0.1695587933063507,
      "learning_rate": 4.702970297029703e-06,
      "loss": 1.0637,
      "step": 215
    },
    {
      "epoch": 0.5346534653465347,
      "grad_norm": 0.16182969510555267,
      "learning_rate": 4.678217821782179e-06,
      "loss": 1.0429,
      "step": 216
    },
    {
      "epoch": 0.5371287128712872,
      "grad_norm": 0.15863147377967834,
      "learning_rate": 4.653465346534654e-06,
      "loss": 1.0736,
      "step": 217
    },
    {
      "epoch": 0.5396039603960396,
      "grad_norm": 0.15679685771465302,
      "learning_rate": 4.628712871287129e-06,
      "loss": 1.0781,
      "step": 218
    },
    {
      "epoch": 0.5420792079207921,
      "grad_norm": 0.16381986439228058,
      "learning_rate": 4.603960396039605e-06,
      "loss": 1.063,
      "step": 219
    },
    {
      "epoch": 0.5445544554455446,
      "grad_norm": 0.1749560832977295,
      "learning_rate": 4.57920792079208e-06,
      "loss": 1.0588,
      "step": 220
    },
    {
      "epoch": 0.5470297029702971,
      "grad_norm": 0.19300749897956848,
      "learning_rate": 4.554455445544555e-06,
      "loss": 1.1099,
      "step": 221
    },
    {
      "epoch": 0.5495049504950495,
      "grad_norm": 0.16599427163600922,
      "learning_rate": 4.52970297029703e-06,
      "loss": 1.0517,
      "step": 222
    },
    {
      "epoch": 0.551980198019802,
      "grad_norm": 0.1542525589466095,
      "learning_rate": 4.5049504950495054e-06,
      "loss": 1.0474,
      "step": 223
    },
    {
      "epoch": 0.5544554455445545,
      "grad_norm": 0.15763726830482483,
      "learning_rate": 4.4801980198019806e-06,
      "loss": 0.9993,
      "step": 224
    },
    {
      "epoch": 0.556930693069307,
      "grad_norm": 0.1836528331041336,
      "learning_rate": 4.455445544554456e-06,
      "loss": 1.1125,
      "step": 225
    },
    {
      "epoch": 0.5594059405940595,
      "grad_norm": 0.16592352092266083,
      "learning_rate": 4.430693069306931e-06,
      "loss": 1.0154,
      "step": 226
    },
    {
      "epoch": 0.5618811881188119,
      "grad_norm": 0.16697534918785095,
      "learning_rate": 4.405940594059406e-06,
      "loss": 1.0328,
      "step": 227
    },
    {
      "epoch": 0.5643564356435643,
      "grad_norm": 0.16123178601264954,
      "learning_rate": 4.381188118811882e-06,
      "loss": 1.0859,
      "step": 228
    },
    {
      "epoch": 0.5668316831683168,
      "grad_norm": 0.17404749989509583,
      "learning_rate": 4.356435643564357e-06,
      "loss": 1.0121,
      "step": 229
    },
    {
      "epoch": 0.5693069306930693,
      "grad_norm": 0.17720270156860352,
      "learning_rate": 4.331683168316832e-06,
      "loss": 1.0757,
      "step": 230
    },
    {
      "epoch": 0.5717821782178217,
      "grad_norm": 0.1586427241563797,
      "learning_rate": 4.306930693069307e-06,
      "loss": 1.0551,
      "step": 231
    },
    {
      "epoch": 0.5742574257425742,
      "grad_norm": 0.1774415373802185,
      "learning_rate": 4.282178217821782e-06,
      "loss": 1.0333,
      "step": 232
    },
    {
      "epoch": 0.5767326732673267,
      "grad_norm": 0.16705653071403503,
      "learning_rate": 4.2574257425742575e-06,
      "loss": 1.0601,
      "step": 233
    },
    {
      "epoch": 0.5792079207920792,
      "grad_norm": 0.18242689967155457,
      "learning_rate": 4.232673267326733e-06,
      "loss": 1.0476,
      "step": 234
    },
    {
      "epoch": 0.5816831683168316,
      "grad_norm": 0.17149695754051208,
      "learning_rate": 4.207920792079208e-06,
      "loss": 1.0589,
      "step": 235
    },
    {
      "epoch": 0.5841584158415841,
      "grad_norm": 0.1735452264547348,
      "learning_rate": 4.183168316831684e-06,
      "loss": 1.0479,
      "step": 236
    },
    {
      "epoch": 0.5866336633663366,
      "grad_norm": 0.1712096780538559,
      "learning_rate": 4.158415841584159e-06,
      "loss": 1.0727,
      "step": 237
    },
    {
      "epoch": 0.5891089108910891,
      "grad_norm": 0.1779618263244629,
      "learning_rate": 4.133663366336634e-06,
      "loss": 1.0847,
      "step": 238
    },
    {
      "epoch": 0.5915841584158416,
      "grad_norm": 0.165634423494339,
      "learning_rate": 4.108910891089109e-06,
      "loss": 1.0318,
      "step": 239
    },
    {
      "epoch": 0.594059405940594,
      "grad_norm": 0.17693260312080383,
      "learning_rate": 4.084158415841584e-06,
      "loss": 1.0602,
      "step": 240
    },
    {
      "epoch": 0.5965346534653465,
      "grad_norm": 0.16443394124507904,
      "learning_rate": 4.05940594059406e-06,
      "loss": 1.0693,
      "step": 241
    },
    {
      "epoch": 0.599009900990099,
      "grad_norm": 0.1829027682542801,
      "learning_rate": 4.034653465346535e-06,
      "loss": 1.0597,
      "step": 242
    },
    {
      "epoch": 0.6014851485148515,
      "grad_norm": 0.17347212135791779,
      "learning_rate": 4.0099009900990104e-06,
      "loss": 1.0901,
      "step": 243
    },
    {
      "epoch": 0.6039603960396039,
      "grad_norm": 0.1738901287317276,
      "learning_rate": 3.9851485148514856e-06,
      "loss": 1.0378,
      "step": 244
    },
    {
      "epoch": 0.6064356435643564,
      "grad_norm": 0.17123974859714508,
      "learning_rate": 3.960396039603961e-06,
      "loss": 1.0449,
      "step": 245
    },
    {
      "epoch": 0.6089108910891089,
      "grad_norm": 0.16725413501262665,
      "learning_rate": 3.935643564356436e-06,
      "loss": 1.07,
      "step": 246
    },
    {
      "epoch": 0.6113861386138614,
      "grad_norm": 0.16899654269218445,
      "learning_rate": 3.910891089108911e-06,
      "loss": 1.0491,
      "step": 247
    },
    {
      "epoch": 0.6138613861386139,
      "grad_norm": 0.16935132443904877,
      "learning_rate": 3.886138613861387e-06,
      "loss": 1.0099,
      "step": 248
    },
    {
      "epoch": 0.6163366336633663,
      "grad_norm": 0.1622443050146103,
      "learning_rate": 3.861386138613862e-06,
      "loss": 1.0529,
      "step": 249
    },
    {
      "epoch": 0.6188118811881188,
      "grad_norm": 0.16781330108642578,
      "learning_rate": 3.836633663366337e-06,
      "loss": 1.0629,
      "step": 250
    },
    {
      "epoch": 0.6212871287128713,
      "grad_norm": 0.17032338678836823,
      "learning_rate": 3.8118811881188123e-06,
      "loss": 1.0624,
      "step": 251
    },
    {
      "epoch": 0.6237623762376238,
      "grad_norm": 0.17583905160427094,
      "learning_rate": 3.7871287128712874e-06,
      "loss": 1.0191,
      "step": 252
    },
    {
      "epoch": 0.6262376237623762,
      "grad_norm": 0.1608443707227707,
      "learning_rate": 3.7623762376237625e-06,
      "loss": 1.0459,
      "step": 253
    },
    {
      "epoch": 0.6287128712871287,
      "grad_norm": 0.16449084877967834,
      "learning_rate": 3.737623762376238e-06,
      "loss": 1.0341,
      "step": 254
    },
    {
      "epoch": 0.6311881188118812,
      "grad_norm": 0.17335855960845947,
      "learning_rate": 3.712871287128713e-06,
      "loss": 1.0492,
      "step": 255
    },
    {
      "epoch": 0.6336633663366337,
      "grad_norm": 0.17544226348400116,
      "learning_rate": 3.6881188118811883e-06,
      "loss": 1.0682,
      "step": 256
    },
    {
      "epoch": 0.6361386138613861,
      "grad_norm": 0.16722352802753448,
      "learning_rate": 3.6633663366336635e-06,
      "loss": 1.0557,
      "step": 257
    },
    {
      "epoch": 0.6386138613861386,
      "grad_norm": 0.19356003403663635,
      "learning_rate": 3.638613861386139e-06,
      "loss": 1.06,
      "step": 258
    },
    {
      "epoch": 0.6410891089108911,
      "grad_norm": 0.18645772337913513,
      "learning_rate": 3.613861386138614e-06,
      "loss": 1.0467,
      "step": 259
    },
    {
      "epoch": 0.6435643564356436,
      "grad_norm": 0.18059493601322174,
      "learning_rate": 3.5891089108910892e-06,
      "loss": 1.0736,
      "step": 260
    },
    {
      "epoch": 0.6460396039603961,
      "grad_norm": 0.19914749264717102,
      "learning_rate": 3.5643564356435644e-06,
      "loss": 1.0792,
      "step": 261
    },
    {
      "epoch": 0.6485148514851485,
      "grad_norm": 0.16404379904270172,
      "learning_rate": 3.5396039603960395e-06,
      "loss": 1.0253,
      "step": 262
    },
    {
      "epoch": 0.650990099009901,
      "grad_norm": 0.16826879978179932,
      "learning_rate": 3.514851485148515e-06,
      "loss": 1.0458,
      "step": 263
    },
    {
      "epoch": 0.6534653465346535,
      "grad_norm": 0.17538203299045563,
      "learning_rate": 3.49009900990099e-06,
      "loss": 1.0224,
      "step": 264
    },
    {
      "epoch": 0.655940594059406,
      "grad_norm": 0.18196958303451538,
      "learning_rate": 3.4653465346534653e-06,
      "loss": 1.0475,
      "step": 265
    },
    {
      "epoch": 0.6584158415841584,
      "grad_norm": 0.1760399043560028,
      "learning_rate": 3.4405940594059404e-06,
      "loss": 1.045,
      "step": 266
    },
    {
      "epoch": 0.6608910891089109,
      "grad_norm": 0.17527367174625397,
      "learning_rate": 3.4158415841584164e-06,
      "loss": 1.0634,
      "step": 267
    },
    {
      "epoch": 0.6633663366336634,
      "grad_norm": 0.18751391768455505,
      "learning_rate": 3.3910891089108915e-06,
      "loss": 1.0073,
      "step": 268
    },
    {
      "epoch": 0.6658415841584159,
      "grad_norm": 0.16496875882148743,
      "learning_rate": 3.3663366336633666e-06,
      "loss": 1.0435,
      "step": 269
    },
    {
      "epoch": 0.6683168316831684,
      "grad_norm": 0.16843590140342712,
      "learning_rate": 3.341584158415842e-06,
      "loss": 1.035,
      "step": 270
    },
    {
      "epoch": 0.6707920792079208,
      "grad_norm": 0.19264374673366547,
      "learning_rate": 3.3168316831683173e-06,
      "loss": 1.0538,
      "step": 271
    },
    {
      "epoch": 0.6732673267326733,
      "grad_norm": 0.19670583307743073,
      "learning_rate": 3.2920792079207924e-06,
      "loss": 1.041,
      "step": 272
    },
    {
      "epoch": 0.6757425742574258,
      "grad_norm": 0.1634487509727478,
      "learning_rate": 3.2673267326732676e-06,
      "loss": 1.0131,
      "step": 273
    },
    {
      "epoch": 0.6782178217821783,
      "grad_norm": 0.1698472797870636,
      "learning_rate": 3.242574257425743e-06,
      "loss": 1.0293,
      "step": 274
    },
    {
      "epoch": 0.6806930693069307,
      "grad_norm": 0.16460570693016052,
      "learning_rate": 3.2178217821782182e-06,
      "loss": 1.0078,
      "step": 275
    },
    {
      "epoch": 0.6831683168316832,
      "grad_norm": 0.17161054909229279,
      "learning_rate": 3.1930693069306933e-06,
      "loss": 1.0465,
      "step": 276
    },
    {
      "epoch": 0.6856435643564357,
      "grad_norm": 0.1728292554616928,
      "learning_rate": 3.1683168316831685e-06,
      "loss": 1.0593,
      "step": 277
    },
    {
      "epoch": 0.6881188118811881,
      "grad_norm": 0.16688892245292664,
      "learning_rate": 3.143564356435644e-06,
      "loss": 1.0343,
      "step": 278
    },
    {
      "epoch": 0.6905940594059405,
      "grad_norm": 0.17009975016117096,
      "learning_rate": 3.118811881188119e-06,
      "loss": 1.0686,
      "step": 279
    },
    {
      "epoch": 0.693069306930693,
      "grad_norm": 0.16594161093235016,
      "learning_rate": 3.0940594059405943e-06,
      "loss": 1.0239,
      "step": 280
    },
    {
      "epoch": 0.6955445544554455,
      "grad_norm": 0.17890632152557373,
      "learning_rate": 3.0693069306930694e-06,
      "loss": 1.079,
      "step": 281
    },
    {
      "epoch": 0.698019801980198,
      "grad_norm": 0.16677552461624146,
      "learning_rate": 3.044554455445545e-06,
      "loss": 0.999,
      "step": 282
    },
    {
      "epoch": 0.7004950495049505,
      "grad_norm": 0.20281478762626648,
      "learning_rate": 3.01980198019802e-06,
      "loss": 1.0477,
      "step": 283
    },
    {
      "epoch": 0.7029702970297029,
      "grad_norm": 0.1814507395029068,
      "learning_rate": 2.995049504950495e-06,
      "loss": 1.0537,
      "step": 284
    },
    {
      "epoch": 0.7054455445544554,
      "grad_norm": 0.2223450392484665,
      "learning_rate": 2.9702970297029703e-06,
      "loss": 1.0815,
      "step": 285
    },
    {
      "epoch": 0.7079207920792079,
      "grad_norm": 0.17586183547973633,
      "learning_rate": 2.9455445544554454e-06,
      "loss": 1.0277,
      "step": 286
    },
    {
      "epoch": 0.7103960396039604,
      "grad_norm": 0.17505447566509247,
      "learning_rate": 2.920792079207921e-06,
      "loss": 1.0055,
      "step": 287
    },
    {
      "epoch": 0.7128712871287128,
      "grad_norm": 0.17361949384212494,
      "learning_rate": 2.896039603960396e-06,
      "loss": 1.0465,
      "step": 288
    },
    {
      "epoch": 0.7153465346534653,
      "grad_norm": 0.1845312863588333,
      "learning_rate": 2.8712871287128712e-06,
      "loss": 1.0676,
      "step": 289
    },
    {
      "epoch": 0.7178217821782178,
      "grad_norm": 0.1974944770336151,
      "learning_rate": 2.8465346534653464e-06,
      "loss": 1.0766,
      "step": 290
    },
    {
      "epoch": 0.7202970297029703,
      "grad_norm": 0.17243847250938416,
      "learning_rate": 2.821782178217822e-06,
      "loss": 1.0387,
      "step": 291
    },
    {
      "epoch": 0.7227722772277227,
      "grad_norm": 0.18797056376934052,
      "learning_rate": 2.7970297029702974e-06,
      "loss": 1.048,
      "step": 292
    },
    {
      "epoch": 0.7252475247524752,
      "grad_norm": 0.1894993633031845,
      "learning_rate": 2.7722772277227726e-06,
      "loss": 1.0352,
      "step": 293
    },
    {
      "epoch": 0.7277227722772277,
      "grad_norm": 0.16774602234363556,
      "learning_rate": 2.747524752475248e-06,
      "loss": 1.052,
      "step": 294
    },
    {
      "epoch": 0.7301980198019802,
      "grad_norm": 0.17973656952381134,
      "learning_rate": 2.7227722772277232e-06,
      "loss": 1.0066,
      "step": 295
    },
    {
      "epoch": 0.7326732673267327,
      "grad_norm": 0.17266245186328888,
      "learning_rate": 2.6980198019801984e-06,
      "loss": 1.0327,
      "step": 296
    },
    {
      "epoch": 0.7351485148514851,
      "grad_norm": 0.19428043067455292,
      "learning_rate": 2.6732673267326735e-06,
      "loss": 1.0479,
      "step": 297
    },
    {
      "epoch": 0.7376237623762376,
      "grad_norm": 0.16848322749137878,
      "learning_rate": 2.648514851485149e-06,
      "loss": 1.0665,
      "step": 298
    },
    {
      "epoch": 0.7400990099009901,
      "grad_norm": 0.1717626452445984,
      "learning_rate": 2.623762376237624e-06,
      "loss": 1.042,
      "step": 299
    },
    {
      "epoch": 0.7425742574257426,
      "grad_norm": 0.16830983757972717,
      "learning_rate": 2.5990099009900993e-06,
      "loss": 1.0313,
      "step": 300
    },
    {
      "epoch": 0.745049504950495,
      "grad_norm": 0.1700887829065323,
      "learning_rate": 2.5742574257425744e-06,
      "loss": 1.0512,
      "step": 301
    },
    {
      "epoch": 0.7475247524752475,
      "grad_norm": 0.17589855194091797,
      "learning_rate": 2.54950495049505e-06,
      "loss": 1.0411,
      "step": 302
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20275838673114777,
      "learning_rate": 2.524752475247525e-06,
      "loss": 1.0412,
      "step": 303
    },
    {
      "epoch": 0.7524752475247525,
      "grad_norm": 0.17624107003211975,
      "learning_rate": 2.5e-06,
      "loss": 1.0484,
      "step": 304
    },
    {
      "epoch": 0.754950495049505,
      "grad_norm": 0.16618701815605164,
      "learning_rate": 2.4752475247524753e-06,
      "loss": 1.0444,
      "step": 305
    },
    {
      "epoch": 0.7574257425742574,
      "grad_norm": 0.17324908077716827,
      "learning_rate": 2.4504950495049505e-06,
      "loss": 1.0126,
      "step": 306
    },
    {
      "epoch": 0.7599009900990099,
      "grad_norm": 0.1791515350341797,
      "learning_rate": 2.425742574257426e-06,
      "loss": 1.0513,
      "step": 307
    },
    {
      "epoch": 0.7623762376237624,
      "grad_norm": 0.16968585550785065,
      "learning_rate": 2.400990099009901e-06,
      "loss": 1.0458,
      "step": 308
    },
    {
      "epoch": 0.7648514851485149,
      "grad_norm": 0.19068896770477295,
      "learning_rate": 2.3762376237623762e-06,
      "loss": 1.0307,
      "step": 309
    },
    {
      "epoch": 0.7673267326732673,
      "grad_norm": 0.1762915998697281,
      "learning_rate": 2.3514851485148514e-06,
      "loss": 1.0266,
      "step": 310
    },
    {
      "epoch": 0.7698019801980198,
      "grad_norm": 0.16937871277332306,
      "learning_rate": 2.326732673267327e-06,
      "loss": 1.0397,
      "step": 311
    },
    {
      "epoch": 0.7722772277227723,
      "grad_norm": 0.17128786444664001,
      "learning_rate": 2.3019801980198025e-06,
      "loss": 1.0479,
      "step": 312
    },
    {
      "epoch": 0.7747524752475248,
      "grad_norm": 0.17716889083385468,
      "learning_rate": 2.2772277227722776e-06,
      "loss": 0.9953,
      "step": 313
    },
    {
      "epoch": 0.7772277227722773,
      "grad_norm": 0.18644240498542786,
      "learning_rate": 2.2524752475247527e-06,
      "loss": 1.0266,
      "step": 314
    },
    {
      "epoch": 0.7797029702970297,
      "grad_norm": 0.18382704257965088,
      "learning_rate": 2.227722772277228e-06,
      "loss": 1.0472,
      "step": 315
    },
    {
      "epoch": 0.7821782178217822,
      "grad_norm": 0.17966952919960022,
      "learning_rate": 2.202970297029703e-06,
      "loss": 1.03,
      "step": 316
    },
    {
      "epoch": 0.7846534653465347,
      "grad_norm": 0.17476266622543335,
      "learning_rate": 2.1782178217821785e-06,
      "loss": 1.0197,
      "step": 317
    },
    {
      "epoch": 0.7871287128712872,
      "grad_norm": 0.16993242502212524,
      "learning_rate": 2.1534653465346536e-06,
      "loss": 1.0365,
      "step": 318
    },
    {
      "epoch": 0.7896039603960396,
      "grad_norm": 0.17047357559204102,
      "learning_rate": 2.1287128712871288e-06,
      "loss": 1.0412,
      "step": 319
    },
    {
      "epoch": 0.7920792079207921,
      "grad_norm": 0.16769839823246002,
      "learning_rate": 2.103960396039604e-06,
      "loss": 1.0082,
      "step": 320
    },
    {
      "epoch": 0.7945544554455446,
      "grad_norm": 0.17701376974582672,
      "learning_rate": 2.0792079207920794e-06,
      "loss": 1.0574,
      "step": 321
    },
    {
      "epoch": 0.7970297029702971,
      "grad_norm": 0.1958608329296112,
      "learning_rate": 2.0544554455445546e-06,
      "loss": 1.0122,
      "step": 322
    },
    {
      "epoch": 0.7995049504950495,
      "grad_norm": 0.19034409523010254,
      "learning_rate": 2.02970297029703e-06,
      "loss": 0.9904,
      "step": 323
    },
    {
      "epoch": 0.801980198019802,
      "grad_norm": 0.17253142595291138,
      "learning_rate": 2.0049504950495052e-06,
      "loss": 1.0229,
      "step": 324
    },
    {
      "epoch": 0.8044554455445545,
      "grad_norm": 0.1845494955778122,
      "learning_rate": 1.9801980198019803e-06,
      "loss": 1.0537,
      "step": 325
    },
    {
      "epoch": 0.806930693069307,
      "grad_norm": 0.17464852333068848,
      "learning_rate": 1.9554455445544555e-06,
      "loss": 1.0293,
      "step": 326
    },
    {
      "epoch": 0.8094059405940595,
      "grad_norm": 0.18944846093654633,
      "learning_rate": 1.930693069306931e-06,
      "loss": 1.0342,
      "step": 327
    },
    {
      "epoch": 0.8118811881188119,
      "grad_norm": 0.20734351873397827,
      "learning_rate": 1.9059405940594061e-06,
      "loss": 1.0481,
      "step": 328
    },
    {
      "epoch": 0.8143564356435643,
      "grad_norm": 0.18267051875591278,
      "learning_rate": 1.8811881188118813e-06,
      "loss": 1.017,
      "step": 329
    },
    {
      "epoch": 0.8168316831683168,
      "grad_norm": 0.16905878484249115,
      "learning_rate": 1.8564356435643566e-06,
      "loss": 0.9996,
      "step": 330
    },
    {
      "epoch": 0.8193069306930693,
      "grad_norm": 0.18889805674552917,
      "learning_rate": 1.8316831683168317e-06,
      "loss": 1.0487,
      "step": 331
    },
    {
      "epoch": 0.8217821782178217,
      "grad_norm": 0.1784394234418869,
      "learning_rate": 1.806930693069307e-06,
      "loss": 1.0301,
      "step": 332
    },
    {
      "epoch": 0.8242574257425742,
      "grad_norm": 0.1732165515422821,
      "learning_rate": 1.7821782178217822e-06,
      "loss": 0.9989,
      "step": 333
    },
    {
      "epoch": 0.8267326732673267,
      "grad_norm": 0.19555354118347168,
      "learning_rate": 1.7574257425742575e-06,
      "loss": 1.0195,
      "step": 334
    },
    {
      "epoch": 0.8292079207920792,
      "grad_norm": 0.17240861058235168,
      "learning_rate": 1.7326732673267326e-06,
      "loss": 1.0076,
      "step": 335
    },
    {
      "epoch": 0.8316831683168316,
      "grad_norm": 0.1675492227077484,
      "learning_rate": 1.7079207920792082e-06,
      "loss": 1.0393,
      "step": 336
    },
    {
      "epoch": 0.8341584158415841,
      "grad_norm": 0.19649028778076172,
      "learning_rate": 1.6831683168316833e-06,
      "loss": 1.0467,
      "step": 337
    },
    {
      "epoch": 0.8366336633663366,
      "grad_norm": 0.1702542006969452,
      "learning_rate": 1.6584158415841587e-06,
      "loss": 1.0208,
      "step": 338
    },
    {
      "epoch": 0.8391089108910891,
      "grad_norm": 0.16906598210334778,
      "learning_rate": 1.6336633663366338e-06,
      "loss": 1.0268,
      "step": 339
    },
    {
      "epoch": 0.8415841584158416,
      "grad_norm": 0.18987883627414703,
      "learning_rate": 1.6089108910891091e-06,
      "loss": 1.0566,
      "step": 340
    },
    {
      "epoch": 0.844059405940594,
      "grad_norm": 0.1939912885427475,
      "learning_rate": 1.5841584158415842e-06,
      "loss": 1.0871,
      "step": 341
    },
    {
      "epoch": 0.8465346534653465,
      "grad_norm": 0.18526384234428406,
      "learning_rate": 1.5594059405940596e-06,
      "loss": 1.0179,
      "step": 342
    },
    {
      "epoch": 0.849009900990099,
      "grad_norm": 0.17309053242206573,
      "learning_rate": 1.5346534653465347e-06,
      "loss": 1.0509,
      "step": 343
    },
    {
      "epoch": 0.8514851485148515,
      "grad_norm": 0.16723985970020294,
      "learning_rate": 1.50990099009901e-06,
      "loss": 1.0034,
      "step": 344
    },
    {
      "epoch": 0.8539603960396039,
      "grad_norm": 0.17107772827148438,
      "learning_rate": 1.4851485148514852e-06,
      "loss": 0.9677,
      "step": 345
    },
    {
      "epoch": 0.8564356435643564,
      "grad_norm": 0.17373721301555634,
      "learning_rate": 1.4603960396039605e-06,
      "loss": 1.0245,
      "step": 346
    },
    {
      "epoch": 0.8589108910891089,
      "grad_norm": 0.1878427267074585,
      "learning_rate": 1.4356435643564356e-06,
      "loss": 1.023,
      "step": 347
    },
    {
      "epoch": 0.8613861386138614,
      "grad_norm": 0.17384839057922363,
      "learning_rate": 1.410891089108911e-06,
      "loss": 1.0414,
      "step": 348
    },
    {
      "epoch": 0.8638613861386139,
      "grad_norm": 0.17436939477920532,
      "learning_rate": 1.3861386138613863e-06,
      "loss": 1.0371,
      "step": 349
    },
    {
      "epoch": 0.8663366336633663,
      "grad_norm": 0.18823057413101196,
      "learning_rate": 1.3613861386138616e-06,
      "loss": 1.0286,
      "step": 350
    },
    {
      "epoch": 0.8688118811881188,
      "grad_norm": 0.17299430072307587,
      "learning_rate": 1.3366336633663367e-06,
      "loss": 1.0117,
      "step": 351
    },
    {
      "epoch": 0.8712871287128713,
      "grad_norm": 0.2107045203447342,
      "learning_rate": 1.311881188118812e-06,
      "loss": 1.0331,
      "step": 352
    },
    {
      "epoch": 0.8737623762376238,
      "grad_norm": 0.17237113416194916,
      "learning_rate": 1.2871287128712872e-06,
      "loss": 1.0022,
      "step": 353
    },
    {
      "epoch": 0.8762376237623762,
      "grad_norm": 0.17842179536819458,
      "learning_rate": 1.2623762376237625e-06,
      "loss": 1.0496,
      "step": 354
    },
    {
      "epoch": 0.8787128712871287,
      "grad_norm": 0.17555266618728638,
      "learning_rate": 1.2376237623762377e-06,
      "loss": 1.0082,
      "step": 355
    },
    {
      "epoch": 0.8811881188118812,
      "grad_norm": 0.22252587974071503,
      "learning_rate": 1.212871287128713e-06,
      "loss": 1.0386,
      "step": 356
    },
    {
      "epoch": 0.8836633663366337,
      "grad_norm": 0.1819303333759308,
      "learning_rate": 1.1881188118811881e-06,
      "loss": 1.0468,
      "step": 357
    },
    {
      "epoch": 0.8861386138613861,
      "grad_norm": 0.18028509616851807,
      "learning_rate": 1.1633663366336635e-06,
      "loss": 0.972,
      "step": 358
    },
    {
      "epoch": 0.8886138613861386,
      "grad_norm": 0.1826130598783493,
      "learning_rate": 1.1386138613861388e-06,
      "loss": 1.0503,
      "step": 359
    },
    {
      "epoch": 0.8910891089108911,
      "grad_norm": 0.17441998422145844,
      "learning_rate": 1.113861386138614e-06,
      "loss": 1.0267,
      "step": 360
    },
    {
      "epoch": 0.8935643564356436,
      "grad_norm": 0.20245307683944702,
      "learning_rate": 1.0891089108910893e-06,
      "loss": 1.0261,
      "step": 361
    },
    {
      "epoch": 0.8960396039603961,
      "grad_norm": 0.170221745967865,
      "learning_rate": 1.0643564356435644e-06,
      "loss": 1.0217,
      "step": 362
    },
    {
      "epoch": 0.8985148514851485,
      "grad_norm": 0.1835111379623413,
      "learning_rate": 1.0396039603960397e-06,
      "loss": 1.0677,
      "step": 363
    },
    {
      "epoch": 0.900990099009901,
      "grad_norm": 0.17305181920528412,
      "learning_rate": 1.014851485148515e-06,
      "loss": 1.0372,
      "step": 364
    },
    {
      "epoch": 0.9034653465346535,
      "grad_norm": 0.17221209406852722,
      "learning_rate": 9.900990099009902e-07,
      "loss": 1.0351,
      "step": 365
    },
    {
      "epoch": 0.905940594059406,
      "grad_norm": 0.2017694115638733,
      "learning_rate": 9.653465346534655e-07,
      "loss": 1.0641,
      "step": 366
    },
    {
      "epoch": 0.9084158415841584,
      "grad_norm": 0.16452528536319733,
      "learning_rate": 9.405940594059406e-07,
      "loss": 1.0297,
      "step": 367
    },
    {
      "epoch": 0.9108910891089109,
      "grad_norm": 0.1696237474679947,
      "learning_rate": 9.158415841584159e-07,
      "loss": 1.0272,
      "step": 368
    },
    {
      "epoch": 0.9133663366336634,
      "grad_norm": 0.17816710472106934,
      "learning_rate": 8.910891089108911e-07,
      "loss": 0.9946,
      "step": 369
    },
    {
      "epoch": 0.9158415841584159,
      "grad_norm": 0.22065281867980957,
      "learning_rate": 8.663366336633663e-07,
      "loss": 1.0485,
      "step": 370
    },
    {
      "epoch": 0.9183168316831684,
      "grad_norm": 0.1876746118068695,
      "learning_rate": 8.415841584158417e-07,
      "loss": 1.0179,
      "step": 371
    },
    {
      "epoch": 0.9207920792079208,
      "grad_norm": 0.28578758239746094,
      "learning_rate": 8.168316831683169e-07,
      "loss": 1.0024,
      "step": 372
    },
    {
      "epoch": 0.9232673267326733,
      "grad_norm": 0.2457706332206726,
      "learning_rate": 7.920792079207921e-07,
      "loss": 0.9973,
      "step": 373
    },
    {
      "epoch": 0.9257425742574258,
      "grad_norm": 0.17008252441883087,
      "learning_rate": 7.673267326732673e-07,
      "loss": 0.9994,
      "step": 374
    },
    {
      "epoch": 0.9282178217821783,
      "grad_norm": 0.1781049519777298,
      "learning_rate": 7.425742574257426e-07,
      "loss": 1.061,
      "step": 375
    },
    {
      "epoch": 0.9306930693069307,
      "grad_norm": 0.18529166281223297,
      "learning_rate": 7.178217821782178e-07,
      "loss": 1.0428,
      "step": 376
    },
    {
      "epoch": 0.9331683168316832,
      "grad_norm": 0.19657132029533386,
      "learning_rate": 6.930693069306931e-07,
      "loss": 1.0448,
      "step": 377
    },
    {
      "epoch": 0.9356435643564357,
      "grad_norm": 0.17185792326927185,
      "learning_rate": 6.683168316831684e-07,
      "loss": 1.0306,
      "step": 378
    },
    {
      "epoch": 0.9381188118811881,
      "grad_norm": 0.18214091658592224,
      "learning_rate": 6.435643564356436e-07,
      "loss": 1.0433,
      "step": 379
    },
    {
      "epoch": 0.9405940594059405,
      "grad_norm": 0.23452341556549072,
      "learning_rate": 6.188118811881188e-07,
      "loss": 0.9751,
      "step": 380
    },
    {
      "epoch": 0.943069306930693,
      "grad_norm": 0.16824421286582947,
      "learning_rate": 5.940594059405941e-07,
      "loss": 1.032,
      "step": 381
    },
    {
      "epoch": 0.9455445544554455,
      "grad_norm": 0.2157716602087021,
      "learning_rate": 5.693069306930694e-07,
      "loss": 1.0361,
      "step": 382
    },
    {
      "epoch": 0.948019801980198,
      "grad_norm": 0.1715572625398636,
      "learning_rate": 5.445544554455446e-07,
      "loss": 1.0135,
      "step": 383
    },
    {
      "epoch": 0.9504950495049505,
      "grad_norm": 0.18207526206970215,
      "learning_rate": 5.198019801980199e-07,
      "loss": 1.008,
      "step": 384
    },
    {
      "epoch": 0.9529702970297029,
      "grad_norm": 0.17518338561058044,
      "learning_rate": 4.950495049504951e-07,
      "loss": 1.0558,
      "step": 385
    },
    {
      "epoch": 0.9554455445544554,
      "grad_norm": 0.16482017934322357,
      "learning_rate": 4.702970297029703e-07,
      "loss": 1.0176,
      "step": 386
    },
    {
      "epoch": 0.9579207920792079,
      "grad_norm": 0.1714717000722885,
      "learning_rate": 4.4554455445544555e-07,
      "loss": 1.0052,
      "step": 387
    },
    {
      "epoch": 0.9603960396039604,
      "grad_norm": 0.2134282886981964,
      "learning_rate": 4.2079207920792083e-07,
      "loss": 0.9946,
      "step": 388
    },
    {
      "epoch": 0.9628712871287128,
      "grad_norm": 0.1706077754497528,
      "learning_rate": 3.9603960396039606e-07,
      "loss": 1.0198,
      "step": 389
    },
    {
      "epoch": 0.9653465346534653,
      "grad_norm": 0.17025628685951233,
      "learning_rate": 3.712871287128713e-07,
      "loss": 1.0387,
      "step": 390
    },
    {
      "epoch": 0.9678217821782178,
      "grad_norm": 0.20762069523334503,
      "learning_rate": 3.4653465346534657e-07,
      "loss": 1.0192,
      "step": 391
    },
    {
      "epoch": 0.9702970297029703,
      "grad_norm": 0.1808956265449524,
      "learning_rate": 3.217821782178218e-07,
      "loss": 1.0094,
      "step": 392
    },
    {
      "epoch": 0.9727722772277227,
      "grad_norm": 0.18868137896060944,
      "learning_rate": 2.9702970297029703e-07,
      "loss": 1.0505,
      "step": 393
    },
    {
      "epoch": 0.9752475247524752,
      "grad_norm": 0.17066539824008942,
      "learning_rate": 2.722772277227723e-07,
      "loss": 1.0439,
      "step": 394
    },
    {
      "epoch": 0.9777227722772277,
      "grad_norm": 0.19031685590744019,
      "learning_rate": 2.4752475247524754e-07,
      "loss": 1.0565,
      "step": 395
    },
    {
      "epoch": 0.9801980198019802,
      "grad_norm": 0.175362229347229,
      "learning_rate": 2.2277227722772277e-07,
      "loss": 1.0129,
      "step": 396
    },
    {
      "epoch": 0.9826732673267327,
      "grad_norm": 0.17563587427139282,
      "learning_rate": 1.9801980198019803e-07,
      "loss": 1.0256,
      "step": 397
    },
    {
      "epoch": 0.9851485148514851,
      "grad_norm": 0.18792986869812012,
      "learning_rate": 1.7326732673267329e-07,
      "loss": 0.9857,
      "step": 398
    },
    {
      "epoch": 0.9876237623762376,
      "grad_norm": 0.16679790616035461,
      "learning_rate": 1.4851485148514852e-07,
      "loss": 1.0258,
      "step": 399
    },
    {
      "epoch": 0.9900990099009901,
      "grad_norm": 0.17717471718788147,
      "learning_rate": 1.2376237623762377e-07,
      "loss": 1.0558,
      "step": 400
    },
    {
      "epoch": 0.9925742574257426,
      "grad_norm": 0.17575222253799438,
      "learning_rate": 9.900990099009901e-08,
      "loss": 1.0095,
      "step": 401
    },
    {
      "epoch": 0.995049504950495,
      "grad_norm": 0.17630964517593384,
      "learning_rate": 7.425742574257426e-08,
      "loss": 1.0431,
      "step": 402
    },
    {
      "epoch": 0.9975247524752475,
      "grad_norm": 0.18719862401485443,
      "learning_rate": 4.950495049504951e-08,
      "loss": 1.0469,
      "step": 403
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1757795810699463,
      "learning_rate": 2.4752475247524754e-08,
      "loss": 1.0161,
      "step": 404
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 404,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.023190572360008e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}