{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9976019184652278,
  "eval_steps": 500,
  "global_step": 312,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0031974420463629096,
      "grad_norm": 0.3446813225746155,
      "learning_rate": 1.875e-05,
      "loss": 2.181,
      "step": 1
    },
    {
      "epoch": 0.006394884092725819,
      "grad_norm": 0.30540090799331665,
      "learning_rate": 3.75e-05,
      "loss": 1.9989,
      "step": 2
    },
    {
      "epoch": 0.009592326139088728,
      "grad_norm": 0.33418309688568115,
      "learning_rate": 5.625e-05,
      "loss": 2.2171,
      "step": 3
    },
    {
      "epoch": 0.012789768185451638,
      "grad_norm": 0.3458963632583618,
      "learning_rate": 7.5e-05,
      "loss": 2.1132,
      "step": 4
    },
    {
      "epoch": 0.01598721023181455,
      "grad_norm": 0.3518303334712982,
      "learning_rate": 9.374999999999999e-05,
      "loss": 1.9712,
      "step": 5
    },
    {
      "epoch": 0.019184652278177457,
      "grad_norm": 0.3810424506664276,
      "learning_rate": 0.0001125,
      "loss": 1.7775,
      "step": 6
    },
    {
      "epoch": 0.02238209432454037,
      "grad_norm": 0.4941970109939575,
      "learning_rate": 0.00013125,
      "loss": 1.9399,
      "step": 7
    },
    {
      "epoch": 0.025579536370903277,
      "grad_norm": 0.3915010094642639,
      "learning_rate": 0.00015,
      "loss": 1.8796,
      "step": 8
    },
    {
      "epoch": 0.02877697841726619,
      "grad_norm": 0.3387204706668854,
      "learning_rate": 0.00016874999999999998,
      "loss": 1.8703,
      "step": 9
    },
    {
      "epoch": 0.0319744204636291,
      "grad_norm": 0.3416554927825928,
      "learning_rate": 0.00018749999999999998,
      "loss": 1.8606,
      "step": 10
    },
    {
      "epoch": 0.035171862509992005,
      "grad_norm": 0.36551880836486816,
      "learning_rate": 0.00020624999999999997,
      "loss": 1.5603,
      "step": 11
    },
    {
      "epoch": 0.03836930455635491,
      "grad_norm": 0.3075932264328003,
      "learning_rate": 0.000225,
      "loss": 1.5475,
      "step": 12
    },
    {
      "epoch": 0.04156674660271783,
      "grad_norm": 0.28144699335098267,
      "learning_rate": 0.00024375,
      "loss": 1.6917,
      "step": 13
    },
    {
      "epoch": 0.04476418864908074,
      "grad_norm": 0.27931058406829834,
      "learning_rate": 0.0002625,
      "loss": 1.4981,
      "step": 14
    },
    {
      "epoch": 0.047961630695443645,
      "grad_norm": 0.24638418853282928,
      "learning_rate": 0.00028125,
      "loss": 1.599,
      "step": 15
    },
    {
      "epoch": 0.051159072741806554,
      "grad_norm": 0.49918419122695923,
      "learning_rate": 0.0003,
      "loss": 1.5411,
      "step": 16
    },
    {
      "epoch": 0.05435651478816946,
      "grad_norm": 0.227300763130188,
      "learning_rate": 0.00029999155161863667,
      "loss": 1.4908,
      "step": 17
    },
    {
      "epoch": 0.05755395683453238,
      "grad_norm": 0.24631308019161224,
      "learning_rate": 0.0002999662074262154,
      "loss": 1.5127,
      "step": 18
    },
    {
      "epoch": 0.060751398880895285,
      "grad_norm": 0.20278117060661316,
      "learning_rate": 0.00029992397027763483,
      "loss": 1.5784,
      "step": 19
    },
    {
      "epoch": 0.0639488409272582,
      "grad_norm": 0.20311613380908966,
      "learning_rate": 0.00029986484493070223,
      "loss": 1.5577,
      "step": 20
    },
    {
      "epoch": 0.0671462829736211,
      "grad_norm": 0.22366106510162354,
      "learning_rate": 0.00029978883804559716,
      "loss": 1.6616,
      "step": 21
    },
    {
      "epoch": 0.07034372501998401,
      "grad_norm": 0.22588945925235748,
      "learning_rate": 0.00029969595818412183,
      "loss": 1.7524,
      "step": 22
    },
    {
      "epoch": 0.07354116706634692,
      "grad_norm": 0.20929686725139618,
      "learning_rate": 0.000299586215808736,
      "loss": 1.5186,
      "step": 23
    },
    {
      "epoch": 0.07673860911270983,
      "grad_norm": 0.2444813847541809,
      "learning_rate": 0.00029945962328137895,
      "loss": 1.5135,
      "step": 24
    },
    {
      "epoch": 0.07993605115907274,
      "grad_norm": 0.21571452915668488,
      "learning_rate": 0.00029931619486207655,
      "loss": 1.4799,
      "step": 25
    },
    {
      "epoch": 0.08313349320543566,
      "grad_norm": 0.2103520780801773,
      "learning_rate": 0.00029915594670733536,
      "loss": 1.6818,
      "step": 26
    },
    {
      "epoch": 0.08633093525179857,
      "grad_norm": 0.24929186701774597,
      "learning_rate": 0.00029897889686832227,
      "loss": 1.4392,
      "step": 27
    },
    {
      "epoch": 0.08952837729816147,
      "grad_norm": 0.24320849776268005,
      "learning_rate": 0.0002987850652888315,
      "loss": 1.5211,
      "step": 28
    },
    {
      "epoch": 0.09272581934452438,
      "grad_norm": 0.23468714952468872,
      "learning_rate": 0.0002985744738030378,
      "loss": 1.5468,
      "step": 29
    },
    {
      "epoch": 0.09592326139088729,
      "grad_norm": 0.2079857587814331,
      "learning_rate": 0.0002983471461330368,
      "loss": 1.5166,
      "step": 30
    },
    {
      "epoch": 0.0991207034372502,
      "grad_norm": 0.21627485752105713,
      "learning_rate": 0.0002981031078861733,
      "loss": 1.5507,
      "step": 31
    },
    {
      "epoch": 0.10231814548361311,
      "grad_norm": 0.23451927304267883,
      "learning_rate": 0.00029784238655215626,
      "loss": 1.508,
      "step": 32
    },
    {
      "epoch": 0.10551558752997602,
      "grad_norm": 0.2220710664987564,
      "learning_rate": 0.0002975650114999625,
      "loss": 1.5164,
      "step": 33
    },
    {
      "epoch": 0.10871302957633892,
      "grad_norm": 0.22487205266952515,
      "learning_rate": 0.00029727101397452834,
      "loss": 1.4938,
      "step": 34
    },
    {
      "epoch": 0.11191047162270183,
      "grad_norm": 0.2187684029340744,
      "learning_rate": 0.00029696042709322995,
      "loss": 1.3007,
      "step": 35
    },
    {
      "epoch": 0.11510791366906475,
      "grad_norm": 0.2184438705444336,
      "learning_rate": 0.00029663328584215293,
      "loss": 1.5204,
      "step": 36
    },
    {
      "epoch": 0.11830535571542766,
      "grad_norm": 0.21176907420158386,
      "learning_rate": 0.00029628962707215124,
      "loss": 1.5017,
      "step": 37
    },
    {
      "epoch": 0.12150279776179057,
      "grad_norm": 0.2055819034576416,
      "learning_rate": 0.00029592948949469614,
      "loss": 1.2755,
      "step": 38
    },
    {
      "epoch": 0.12470023980815348,
      "grad_norm": 0.220439150929451,
      "learning_rate": 0.00029555291367751573,
      "loss": 1.5057,
      "step": 39
    },
    {
      "epoch": 0.1278976818545164,
      "grad_norm": 0.2324652075767517,
      "learning_rate": 0.00029515994204002484,
      "loss": 1.5839,
      "step": 40
    },
    {
      "epoch": 0.1310951239008793,
      "grad_norm": 0.19285540282726288,
      "learning_rate": 0.0002947506188485468,
      "loss": 1.4434,
      "step": 41
    },
    {
      "epoch": 0.1342925659472422,
      "grad_norm": 0.26798316836357117,
      "learning_rate": 0.00029432499021132737,
      "loss": 1.6137,
      "step": 42
    },
    {
      "epoch": 0.1374900079936051,
      "grad_norm": 0.18407316505908966,
      "learning_rate": 0.0002938831040733405,
      "loss": 1.3876,
      "step": 43
    },
    {
      "epoch": 0.14068745003996802,
      "grad_norm": 0.2084178477525711,
      "learning_rate": 0.0002934250102108876,
      "loss": 1.5409,
      "step": 44
    },
    {
      "epoch": 0.14388489208633093,
      "grad_norm": 0.20955117046833038,
      "learning_rate": 0.00029295076022599077,
      "loss": 1.4635,
      "step": 45
    },
    {
      "epoch": 0.14708233413269384,
      "grad_norm": 0.2144644558429718,
      "learning_rate": 0.00029246040754057976,
      "loss": 1.4585,
      "step": 46
    },
    {
      "epoch": 0.15027977617905675,
      "grad_norm": 0.20352163910865784,
      "learning_rate": 0.0002919540073904744,
      "loss": 1.5338,
      "step": 47
    },
    {
      "epoch": 0.15347721822541965,
      "grad_norm": 0.18800681829452515,
      "learning_rate": 0.0002914316168191626,
      "loss": 1.5031,
      "step": 48
    },
    {
      "epoch": 0.15667466027178256,
      "grad_norm": 0.19407911598682404,
      "learning_rate": 0.00029089329467137456,
      "loss": 1.4457,
      "step": 49
    },
    {
      "epoch": 0.15987210231814547,
      "grad_norm": 0.19459669291973114,
      "learning_rate": 0.0002903391015864543,
      "loss": 1.3383,
      "step": 50
    },
    {
      "epoch": 0.1630695443645084,
      "grad_norm": 0.22761711478233337,
      "learning_rate": 0.0002897690999915289,
      "loss": 1.5057,
      "step": 51
    },
    {
      "epoch": 0.16626698641087131,
      "grad_norm": 0.22577515244483948,
      "learning_rate": 0.0002891833540944764,
      "loss": 1.3057,
      "step": 52
    },
    {
      "epoch": 0.16946442845723422,
      "grad_norm": 0.2257939875125885,
      "learning_rate": 0.000288581929876693,
      "loss": 1.4777,
      "step": 53
    },
    {
      "epoch": 0.17266187050359713,
      "grad_norm": 0.20815476775169373,
      "learning_rate": 0.0002879648950856608,
      "loss": 1.4252,
      "step": 54
    },
    {
      "epoch": 0.17585931254996004,
      "grad_norm": 0.20832973718643188,
      "learning_rate": 0.0002873323192273162,
      "loss": 1.5008,
      "step": 55
    },
    {
      "epoch": 0.17905675459632295,
      "grad_norm": 0.2152206003665924,
      "learning_rate": 0.00028668427355822034,
      "loss": 1.6078,
      "step": 56
    },
    {
      "epoch": 0.18225419664268586,
      "grad_norm": 0.18941529095172882,
      "learning_rate": 0.0002860208310775327,
      "loss": 1.4449,
      "step": 57
    },
    {
      "epoch": 0.18545163868904876,
      "grad_norm": 0.23700568079948425,
      "learning_rate": 0.00028534206651878777,
      "loss": 1.5582,
      "step": 58
    },
    {
      "epoch": 0.18864908073541167,
      "grad_norm": 0.2555181384086609,
      "learning_rate": 0.0002846480563414768,
      "loss": 1.5682,
      "step": 59
    },
    {
      "epoch": 0.19184652278177458,
      "grad_norm": 0.18711774051189423,
      "learning_rate": 0.0002839388787224353,
      "loss": 1.5051,
      "step": 60
    },
    {
      "epoch": 0.1950439648281375,
      "grad_norm": 0.2022084891796112,
      "learning_rate": 0.00028321461354703604,
      "loss": 1.4694,
      "step": 61
    },
    {
      "epoch": 0.1982414068745004,
      "grad_norm": 0.1778743863105774,
      "learning_rate": 0.0002824753424001914,
      "loss": 1.3847,
      "step": 62
    },
    {
      "epoch": 0.2014388489208633,
      "grad_norm": 0.1981406807899475,
      "learning_rate": 0.0002817211485571623,
      "loss": 1.3561,
      "step": 63
    },
    {
      "epoch": 0.20463629096722621,
      "grad_norm": 0.19981688261032104,
      "learning_rate": 0.0002809521169741782,
      "loss": 1.4506,
      "step": 64
    },
    {
      "epoch": 0.20783373301358912,
      "grad_norm": 0.20264656841754913,
      "learning_rate": 0.0002801683342788671,
      "loss": 1.5316,
      "step": 65
    },
    {
      "epoch": 0.21103117505995203,
      "grad_norm": 0.18628135323524475,
      "learning_rate": 0.000279369888760497,
      "loss": 1.4879,
      "step": 66
    },
    {
      "epoch": 0.21422861710631494,
      "grad_norm": 0.2130441665649414,
      "learning_rate": 0.00027855687036003134,
      "loss": 1.6192,
      "step": 67
    },
    {
      "epoch": 0.21742605915267785,
      "grad_norm": 0.19949516654014587,
      "learning_rate": 0.00027772937065999667,
      "loss": 1.4773,
      "step": 68
    },
    {
      "epoch": 0.22062350119904076,
      "grad_norm": 0.20962868630886078,
      "learning_rate": 0.0002768874828741669,
      "loss": 1.4617,
      "step": 69
    },
    {
      "epoch": 0.22382094324540366,
      "grad_norm": 0.21659812331199646,
      "learning_rate": 0.00027603130183706314,
      "loss": 1.5065,
      "step": 70
    },
    {
      "epoch": 0.2270183852917666,
      "grad_norm": 0.19917699694633484,
      "learning_rate": 0.00027516092399327094,
      "loss": 1.6265,
      "step": 71
    },
    {
      "epoch": 0.2302158273381295,
      "grad_norm": 0.20580779016017914,
      "learning_rate": 0.0002742764473865763,
      "loss": 1.4508,
      "step": 72
    },
    {
      "epoch": 0.23341326938449242,
      "grad_norm": 0.20578929781913757,
      "learning_rate": 0.0002733779716489217,
      "loss": 1.5362,
      "step": 73
    },
    {
      "epoch": 0.23661071143085532,
      "grad_norm": 0.21730633080005646,
      "learning_rate": 0.0002724655979891828,
      "loss": 1.4373,
      "step": 74
    },
    {
      "epoch": 0.23980815347721823,
      "grad_norm": 0.21635404229164124,
      "learning_rate": 0.000271539429181768,
      "loss": 1.3639,
      "step": 75
    },
    {
      "epoch": 0.24300559552358114,
      "grad_norm": 0.24112968146800995,
      "learning_rate": 0.0002705995695550411,
      "loss": 1.5238,
      "step": 76
    },
    {
      "epoch": 0.24620303756994405,
      "grad_norm": 0.20409514009952545,
      "learning_rate": 0.00026964612497956946,
      "loss": 1.4533,
      "step": 77
    },
    {
      "epoch": 0.24940047961630696,
      "grad_norm": 0.21514864265918732,
      "learning_rate": 0.0002686792028561983,
      "loss": 1.4657,
      "step": 78
    },
    {
      "epoch": 0.25259792166266987,
      "grad_norm": 0.20796911418437958,
      "learning_rate": 0.00026769891210395207,
      "loss": 1.4834,
      "step": 79
    },
    {
      "epoch": 0.2557953637090328,
      "grad_norm": 0.20425471663475037,
      "learning_rate": 0.00026670536314776593,
      "loss": 1.4799,
      "step": 80
    },
    {
      "epoch": 0.2589928057553957,
      "grad_norm": 0.1899542212486267,
      "learning_rate": 0.0002656986679060462,
      "loss": 1.4862,
      "step": 81
    },
    {
      "epoch": 0.2621902478017586,
      "grad_norm": 0.20222659409046173,
      "learning_rate": 0.00026467893977806387,
      "loss": 1.4788,
      "step": 82
    },
    {
      "epoch": 0.2653876898481215,
      "grad_norm": 0.1941121220588684,
      "learning_rate": 0.0002636462936311804,
      "loss": 1.4913,
      "step": 83
    },
    {
      "epoch": 0.2685851318944844,
      "grad_norm": 0.21576811373233795,
      "learning_rate": 0.0002626008457879086,
      "loss": 1.5327,
      "step": 84
    },
    {
      "epoch": 0.2717825739408473,
      "grad_norm": 0.1937507688999176,
      "learning_rate": 0.00026154271401280957,
      "loss": 1.4609,
      "step": 85
    },
    {
      "epoch": 0.2749800159872102,
      "grad_norm": 0.18996623158454895,
      "learning_rate": 0.0002604720174992268,
      "loss": 1.4023,
      "step": 86
    },
    {
      "epoch": 0.27817745803357313,
      "grad_norm": 0.20716165006160736,
      "learning_rate": 0.00025938887685585994,
      "loss": 1.5351,
      "step": 87
    },
    {
      "epoch": 0.28137490007993604,
      "grad_norm": 0.20239269733428955,
      "learning_rate": 0.0002582934140931786,
      "loss": 1.4851,
      "step": 88
    },
    {
      "epoch": 0.28457234212629895,
      "grad_norm": 0.20915232598781586,
      "learning_rate": 0.0002571857526096788,
      "loss": 1.3798,
      "step": 89
    },
    {
      "epoch": 0.28776978417266186,
      "grad_norm": 0.20972570776939392,
      "learning_rate": 0.00025606601717798207,
      "loss": 1.4097,
      "step": 90
    },
    {
      "epoch": 0.29096722621902477,
      "grad_norm": 0.20584455132484436,
      "learning_rate": 0.0002549343339307813,
      "loss": 1.5279,
      "step": 91
    },
    {
      "epoch": 0.2941646682653877,
      "grad_norm": 0.1897670328617096,
      "learning_rate": 0.00025379083034663194,
      "loss": 1.603,
      "step": 92
    },
    {
      "epoch": 0.2973621103117506,
      "grad_norm": 0.19150228798389435,
      "learning_rate": 0.000252635635235592,
      "loss": 1.3939,
      "step": 93
    },
    {
      "epoch": 0.3005595523581135,
      "grad_norm": 0.1970176249742508,
      "learning_rate": 0.00025146887872471303,
      "loss": 1.468,
      "step": 94
    },
    {
      "epoch": 0.3037569944044764,
      "grad_norm": 0.19097474217414856,
      "learning_rate": 0.000250290692243381,
      "loss": 1.4303,
      "step": 95
    },
    {
      "epoch": 0.3069544364508393,
      "grad_norm": 0.21538837254047394,
      "learning_rate": 0.00024910120850851216,
      "loss": 1.5775,
      "step": 96
    },
    {
      "epoch": 0.3101518784972022,
      "grad_norm": 0.1855296939611435,
      "learning_rate": 0.0002479005615096028,
      "loss": 1.413,
      "step": 97
    },
    {
      "epoch": 0.3133493205435651,
      "grad_norm": 0.23258726298809052,
      "learning_rate": 0.00024668888649363583,
      "loss": 1.5517,
      "step": 98
    },
    {
      "epoch": 0.31654676258992803,
      "grad_norm": 0.19402435421943665,
      "learning_rate": 0.0002454663199498463,
      "loss": 1.3835,
      "step": 99
    },
    {
      "epoch": 0.31974420463629094,
      "grad_norm": 0.1976032257080078,
      "learning_rate": 0.00024423299959434636,
      "loss": 1.4637,
      "step": 100
    },
    {
      "epoch": 0.3229416466826539,
      "grad_norm": 0.19951173663139343,
      "learning_rate": 0.0002429890643546119,
      "loss": 1.3731,
      "step": 101
    },
    {
      "epoch": 0.3261390887290168,
      "grad_norm": 0.20681437849998474,
      "learning_rate": 0.0002417346543538337,
      "loss": 1.4865,
      "step": 102
    },
    {
      "epoch": 0.3293365307753797,
      "grad_norm": 0.36958593130111694,
      "learning_rate": 0.00024046991089513267,
      "loss": 1.4612,
      "step": 103
    },
    {
      "epoch": 0.33253397282174263,
      "grad_norm": 0.20621562004089355,
      "learning_rate": 0.00023919497644564298,
      "loss": 1.357,
      "step": 104
    },
    {
      "epoch": 0.33573141486810554,
      "grad_norm": 0.18956023454666138,
      "learning_rate": 0.00023790999462046394,
      "loss": 1.6554,
      "step": 105
    },
    {
      "epoch": 0.33892885691446845,
      "grad_norm": 0.2084682583808899,
      "learning_rate": 0.0002366151101664822,
      "loss": 1.4853,
      "step": 106
    },
    {
      "epoch": 0.34212629896083135,
      "grad_norm": 0.17509467899799347,
      "learning_rate": 0.00023531046894606703,
      "loss": 1.4028,
      "step": 107
    },
    {
      "epoch": 0.34532374100719426,
      "grad_norm": 0.19247236847877502,
      "learning_rate": 0.00023399621792063928,
      "loss": 1.4353,
      "step": 108
    },
    {
      "epoch": 0.34852118305355717,
      "grad_norm": 0.19204045832157135,
      "learning_rate": 0.00023267250513411733,
      "loss": 1.3393,
      "step": 109
    },
    {
      "epoch": 0.3517186250999201,
      "grad_norm": 0.20329782366752625,
      "learning_rate": 0.00023133947969624028,
      "loss": 1.6107,
      "step": 110
    },
    {
      "epoch": 0.354916067146283,
      "grad_norm": 0.2169138640165329,
      "learning_rate": 0.00022999729176577163,
      "loss": 1.4617,
      "step": 111
    },
    {
      "epoch": 0.3581135091926459,
      "grad_norm": 0.22543761134147644,
      "learning_rate": 0.00022864609253358474,
      "loss": 1.4731,
      "step": 112
    },
    {
      "epoch": 0.3613109512390088,
      "grad_norm": 0.19519487023353577,
      "learning_rate": 0.00022728603420563175,
      "loss": 1.597,
      "step": 113
    },
    {
      "epoch": 0.3645083932853717,
      "grad_norm": 0.20843897759914398,
      "learning_rate": 0.00022591726998579843,
      "loss": 1.4963,
      "step": 114
    },
    {
      "epoch": 0.3677058353317346,
      "grad_norm": 0.2149285078048706,
      "learning_rate": 0.00022453995405864638,
      "loss": 1.5095,
      "step": 115
    },
    {
      "epoch": 0.37090327737809753,
      "grad_norm": 0.19521689414978027,
      "learning_rate": 0.00022315424157204518,
      "loss": 1.5709,
      "step": 116
    },
    {
      "epoch": 0.37410071942446044,
      "grad_norm": 0.19614940881729126,
      "learning_rate": 0.00022176028861969535,
      "loss": 1.4573,
      "step": 117
    },
    {
      "epoch": 0.37729816147082335,
      "grad_norm": 0.1948356330394745,
      "learning_rate": 0.00022035825222354552,
      "loss": 1.309,
      "step": 118
    },
    {
      "epoch": 0.38049560351718625,
      "grad_norm": 0.20020437240600586,
      "learning_rate": 0.00021894829031610452,
      "loss": 1.5289,
      "step": 119
    },
    {
      "epoch": 0.38369304556354916,
      "grad_norm": 0.20084881782531738,
      "learning_rate": 0.00021753056172265096,
      "loss": 1.5456,
      "step": 120
    },
    {
      "epoch": 0.38689048760991207,
      "grad_norm": 0.17715269327163696,
      "learning_rate": 0.00021610522614334265,
      "loss": 1.4322,
      "step": 121
    },
    {
      "epoch": 0.390087929656275,
      "grad_norm": 0.2064034342765808,
      "learning_rate": 0.00021467244413522673,
      "loss": 1.5772,
      "step": 122
    },
    {
      "epoch": 0.3932853717026379,
      "grad_norm": 0.19036740064620972,
      "learning_rate": 0.00021323237709415413,
      "loss": 1.5086,
      "step": 123
    },
    {
      "epoch": 0.3964828137490008,
      "grad_norm": 0.19214606285095215,
      "learning_rate": 0.0002117851872365989,
      "loss": 1.5296,
      "step": 124
    },
    {
      "epoch": 0.3996802557953637,
      "grad_norm": 0.20223727822303772,
      "learning_rate": 0.00021033103758138529,
      "loss": 1.5354,
      "step": 125
    },
    {
      "epoch": 0.4028776978417266,
      "grad_norm": 0.18433460593223572,
      "learning_rate": 0.00020887009193132456,
      "loss": 1.532,
      "step": 126
    },
    {
      "epoch": 0.4060751398880895,
      "grad_norm": 0.18365609645843506,
      "learning_rate": 0.00020740251485476345,
      "loss": 1.3326,
      "step": 127
    },
    {
      "epoch": 0.40927258193445243,
      "grad_norm": 0.19547204673290253,
      "learning_rate": 0.0002059284716670463,
      "loss": 1.4566,
      "step": 128
    },
    {
      "epoch": 0.41247002398081534,
      "grad_norm": 0.2268918752670288,
      "learning_rate": 0.00020444812841189294,
      "loss": 1.6165,
      "step": 129
    },
    {
      "epoch": 0.41566746602717825,
      "grad_norm": 0.21848422288894653,
      "learning_rate": 0.0002029616518426951,
      "loss": 1.6039,
      "step": 130
    },
    {
      "epoch": 0.41886490807354115,
      "grad_norm": 0.19918426871299744,
      "learning_rate": 0.00020146920940373195,
      "loss": 1.4602,
      "step": 131
    },
    {
      "epoch": 0.42206235011990406,
      "grad_norm": 0.18590374290943146,
      "learning_rate": 0.00019997096921130862,
      "loss": 1.2925,
      "step": 132
    },
    {
      "epoch": 0.42525979216626697,
      "grad_norm": 0.19987183809280396,
      "learning_rate": 0.00019846710003481875,
      "loss": 1.4157,
      "step": 133
    },
    {
      "epoch": 0.4284572342126299,
      "grad_norm": 0.20987945795059204,
      "learning_rate": 0.00019695777127773332,
      "loss": 1.4424,
      "step": 134
    },
    {
      "epoch": 0.4316546762589928,
      "grad_norm": 0.21076463162899017,
      "learning_rate": 0.00019544315295851825,
      "loss": 1.4946,
      "step": 135
    },
    {
      "epoch": 0.4348521183053557,
      "grad_norm": 0.20848603546619415,
      "learning_rate": 0.00019392341569148252,
      "loss": 1.4393,
      "step": 136
    },
    {
      "epoch": 0.4380495603517186,
      "grad_norm": 0.21943925321102142,
      "learning_rate": 0.00019239873066755964,
      "loss": 1.6161,
      "step": 137
    },
    {
      "epoch": 0.4412470023980815,
      "grad_norm": 0.23087991774082184,
      "learning_rate": 0.0001908692696350234,
      "loss": 1.3502,
      "step": 138
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.20302651822566986,
      "learning_rate": 0.00018933520488014166,
      "loss": 1.3896,
      "step": 139
    },
    {
      "epoch": 0.44764188649080733,
      "grad_norm": 0.19597011804580688,
      "learning_rate": 0.00018779670920776877,
      "loss": 1.4437,
      "step": 140
    },
    {
      "epoch": 0.45083932853717024,
      "grad_norm": 0.21784569323062897,
      "learning_rate": 0.00018625395592188036,
      "loss": 1.5956,
      "step": 141
    },
    {
      "epoch": 0.4540367705835332,
      "grad_norm": 0.20360009372234344,
      "learning_rate": 0.00018470711880605122,
      "loss": 1.2507,
      "step": 142
    },
    {
      "epoch": 0.4572342126298961,
      "grad_norm": 0.1850934773683548,
      "learning_rate": 0.00018315637210387947,
      "loss": 1.477,
      "step": 143
    },
    {
      "epoch": 0.460431654676259,
      "grad_norm": 0.22538472712039948,
      "learning_rate": 0.00018160189049935892,
      "loss": 1.3688,
      "step": 144
    },
    {
      "epoch": 0.4636290967226219,
      "grad_norm": 0.2093997299671173,
      "learning_rate": 0.00018004384909720188,
      "loss": 1.3953,
      "step": 145
    },
    {
      "epoch": 0.46682653876898483,
      "grad_norm": 0.19743283092975616,
      "learning_rate": 0.00017848242340311424,
      "loss": 1.5111,
      "step": 146
    },
    {
      "epoch": 0.47002398081534774,
      "grad_norm": 0.23592239618301392,
      "learning_rate": 0.0001769177893040258,
      "loss": 1.4628,
      "step": 147
    },
    {
      "epoch": 0.47322142286171065,
      "grad_norm": 0.2107086479663849,
      "learning_rate": 0.00017535012304827736,
      "loss": 1.345,
      "step": 148
    },
    {
      "epoch": 0.47641886490807356,
      "grad_norm": 0.212343230843544,
      "learning_rate": 0.00017377960122576732,
      "loss": 1.4294,
      "step": 149
    },
    {
      "epoch": 0.47961630695443647,
      "grad_norm": 0.280923455953598,
      "learning_rate": 0.0001722064007480597,
      "loss": 1.6237,
      "step": 150
    },
    {
      "epoch": 0.4828137490007994,
      "grad_norm": 0.19629351794719696,
      "learning_rate": 0.00017063069882845575,
      "loss": 1.439,
      "step": 151
    },
    {
      "epoch": 0.4860111910471623,
      "grad_norm": 0.2047591209411621,
      "learning_rate": 0.0001690526729620318,
      "loss": 1.3626,
      "step": 152
    },
    {
      "epoch": 0.4892086330935252,
      "grad_norm": 0.18259218335151672,
      "learning_rate": 0.00016747250090564557,
      "loss": 1.3234,
      "step": 153
    },
    {
      "epoch": 0.4924060751398881,
      "grad_norm": 0.20569853484630585,
      "learning_rate": 0.00016589036065791242,
      "loss": 1.4376,
      "step": 154
    },
    {
      "epoch": 0.495603517186251,
      "grad_norm": 0.18437625467777252,
      "learning_rate": 0.0001643064304391547,
      "loss": 1.4705,
      "step": 155
    },
    {
      "epoch": 0.4988009592326139,
      "grad_norm": 0.22610221803188324,
      "learning_rate": 0.00016272088867132637,
      "loss": 1.3045,
      "step": 156
    },
    {
      "epoch": 0.5019984012789768,
      "grad_norm": 0.197098046541214,
      "learning_rate": 0.00016113391395791436,
      "loss": 1.531,
      "step": 157
    },
    {
      "epoch": 0.5051958433253397,
      "grad_norm": 0.2230396866798401,
      "learning_rate": 0.00015954568506381994,
      "loss": 1.5164,
      "step": 158
    },
    {
      "epoch": 0.5083932853717026,
      "grad_norm": 0.19642704725265503,
      "learning_rate": 0.0001579563808952216,
      "loss": 1.4442,
      "step": 159
    },
    {
      "epoch": 0.5115907274180655,
      "grad_norm": 0.21066069602966309,
      "learning_rate": 0.00015636618047942222,
      "loss": 1.4251,
      "step": 160
    },
    {
      "epoch": 0.5147881694644284,
      "grad_norm": 0.18799303472042084,
      "learning_rate": 0.0001547752629446827,
      "loss": 1.3866,
      "step": 161
    },
    {
      "epoch": 0.5179856115107914,
      "grad_norm": 0.20167718827724457,
      "learning_rate": 0.00015318380750004352,
      "loss": 1.471,
      "step": 162
    },
    {
      "epoch": 0.5211830535571543,
      "grad_norm": 0.20787064731121063,
      "learning_rate": 0.00015159199341513845,
      "loss": 1.5312,
      "step": 163
    },
    {
      "epoch": 0.5243804956035172,
      "grad_norm": 0.19502943754196167,
      "learning_rate": 0.00015,
      "loss": 1.5153,
      "step": 164
    },
    {
      "epoch": 0.5275779376498801,
      "grad_norm": 0.18463830649852753,
      "learning_rate": 0.00014840800658486158,
      "loss": 1.62,
      "step": 165
    },
    {
      "epoch": 0.530775379696243,
      "grad_norm": 0.20096978545188904,
      "learning_rate": 0.00014681619249995646,
      "loss": 1.3816,
      "step": 166
    },
    {
      "epoch": 0.533972821742606,
      "grad_norm": 0.20995350182056427,
      "learning_rate": 0.00014522473705531736,
      "loss": 1.4321,
      "step": 167
    },
    {
      "epoch": 0.5371702637889688,
      "grad_norm": 0.1865735948085785,
      "learning_rate": 0.00014363381952057778,
      "loss": 1.4262,
      "step": 168
    },
    {
      "epoch": 0.5403677058353318,
      "grad_norm": 0.1792657971382141,
      "learning_rate": 0.00014204361910477844,
      "loss": 1.5558,
      "step": 169
    },
    {
      "epoch": 0.5435651478816946,
      "grad_norm": 0.2027653157711029,
      "learning_rate": 0.00014045431493618003,
      "loss": 1.3377,
      "step": 170
    },
    {
      "epoch": 0.5467625899280576,
      "grad_norm": 0.19514119625091553,
      "learning_rate": 0.0001388660860420856,
      "loss": 1.3874,
      "step": 171
    },
    {
      "epoch": 0.5499600319744204,
      "grad_norm": 0.17817656695842743,
      "learning_rate": 0.00013727911132867365,
      "loss": 1.3716,
      "step": 172
    },
    {
      "epoch": 0.5531574740207834,
      "grad_norm": 0.23043349385261536,
      "learning_rate": 0.00013569356956084528,
      "loss": 1.464,
      "step": 173
    },
    {
      "epoch": 0.5563549160671463,
      "grad_norm": 0.19135528802871704,
      "learning_rate": 0.00013410963934208759,
      "loss": 1.3154,
      "step": 174
    },
    {
      "epoch": 0.5595523581135092,
      "grad_norm": 0.20745159685611725,
      "learning_rate": 0.0001325274990943544,
      "loss": 1.4785,
      "step": 175
    },
    {
      "epoch": 0.5627498001598721,
      "grad_norm": 0.20532263815402985,
      "learning_rate": 0.00013094732703796818,
      "loss": 1.5137,
      "step": 176
    },
    {
      "epoch": 0.565947242206235,
      "grad_norm": 0.21446797251701355,
      "learning_rate": 0.00012936930117154425,
      "loss": 1.3701,
      "step": 177
    },
    {
      "epoch": 0.5691446842525979,
      "grad_norm": 0.19260822236537933,
      "learning_rate": 0.0001277935992519403,
      "loss": 1.4443,
      "step": 178
    },
    {
      "epoch": 0.5723421262989609,
      "grad_norm": 0.19996041059494019,
      "learning_rate": 0.00012622039877423265,
      "loss": 1.371,
      "step": 179
    },
    {
      "epoch": 0.5755395683453237,
      "grad_norm": 0.19244007766246796,
      "learning_rate": 0.00012464987695172264,
      "loss": 1.3142,
      "step": 180
    },
    {
      "epoch": 0.5787370103916867,
      "grad_norm": 0.19164302945137024,
      "learning_rate": 0.00012308221069597418,
      "loss": 1.4773,
      "step": 181
    },
    {
      "epoch": 0.5819344524380495,
      "grad_norm": 0.20002460479736328,
      "learning_rate": 0.00012151757659688571,
      "loss": 1.4264,
      "step": 182
    },
    {
      "epoch": 0.5851318944844125,
      "grad_norm": 0.21552026271820068,
      "learning_rate": 0.00011995615090279813,
      "loss": 1.4049,
      "step": 183
    },
    {
      "epoch": 0.5883293365307753,
      "grad_norm": 0.19300565123558044,
      "learning_rate": 0.00011839810950064109,
      "loss": 1.3554,
      "step": 184
    },
    {
      "epoch": 0.5915267785771383,
      "grad_norm": 0.19941386580467224,
      "learning_rate": 0.00011684362789612053,
      "loss": 1.5601,
      "step": 185
    },
    {
      "epoch": 0.5947242206235012,
      "grad_norm": 0.18221646547317505,
      "learning_rate": 0.00011529288119394878,
      "loss": 1.4828,
      "step": 186
    },
    {
      "epoch": 0.5979216626698641,
      "grad_norm": 0.1901618093252182,
      "learning_rate": 0.00011374604407811962,
      "loss": 1.5442,
      "step": 187
    },
    {
      "epoch": 0.601119104716227,
      "grad_norm": 0.17420263588428497,
      "learning_rate": 0.00011220329079223123,
      "loss": 1.285,
      "step": 188
    },
    {
      "epoch": 0.60431654676259,
      "grad_norm": 0.23658356070518494,
      "learning_rate": 0.00011066479511985838,
      "loss": 1.2485,
      "step": 189
    },
    {
      "epoch": 0.6075139888089528,
      "grad_norm": 0.20968788862228394,
      "learning_rate": 0.00010913073036497658,
      "loss": 1.3972,
      "step": 190
    },
    {
      "epoch": 0.6107114308553158,
      "grad_norm": 0.2030273675918579,
      "learning_rate": 0.00010760126933244036,
      "loss": 1.6353,
      "step": 191
    },
    {
      "epoch": 0.6139088729016786,
      "grad_norm": 0.1902075558900833,
      "learning_rate": 0.00010607658430851744,
      "loss": 1.2809,
      "step": 192
    },
    {
      "epoch": 0.6171063149480416,
      "grad_norm": 0.20934785902500153,
      "learning_rate": 0.00010455684704148173,
      "loss": 1.3585,
      "step": 193
    },
    {
      "epoch": 0.6203037569944044,
      "grad_norm": 0.2173265963792801,
      "learning_rate": 0.00010304222872226668,
      "loss": 1.2973,
      "step": 194
    },
    {
      "epoch": 0.6235011990407674,
      "grad_norm": 0.19533811509609222,
      "learning_rate": 0.00010153289996518125,
      "loss": 1.4299,
      "step": 195
    },
    {
      "epoch": 0.6266986410871302,
      "grad_norm": 0.2015613615512848,
      "learning_rate": 0.00010002903078869135,
      "loss": 1.4279,
      "step": 196
    },
    {
      "epoch": 0.6298960831334932,
      "grad_norm": 0.20218639075756073,
      "learning_rate": 9.853079059626805e-05,
      "loss": 1.3212,
      "step": 197
    },
    {
      "epoch": 0.6330935251798561,
      "grad_norm": 0.1902882307767868,
      "learning_rate": 9.703834815730487e-05,
      "loss": 1.3939,
      "step": 198
    },
    {
      "epoch": 0.636290967226219,
      "grad_norm": 0.18366214632987976,
      "learning_rate": 9.555187158810702e-05,
      "loss": 1.4403,
      "step": 199
    },
    {
      "epoch": 0.6394884092725819,
      "grad_norm": 0.1821315586566925,
      "learning_rate": 9.407152833295372e-05,
      "loss": 1.372,
      "step": 200
    },
    {
      "epoch": 0.6426858513189448,
      "grad_norm": 0.20973654091358185,
      "learning_rate": 9.259748514523653e-05,
      "loss": 1.4149,
      "step": 201
    },
    {
      "epoch": 0.6458832933653078,
      "grad_norm": 0.18254290521144867,
      "learning_rate": 9.112990806867543e-05,
      "loss": 1.3052,
      "step": 202
    },
    {
      "epoch": 0.6490807354116707,
      "grad_norm": 0.18717211484909058,
      "learning_rate": 8.966896241861473e-05,
      "loss": 1.4061,
      "step": 203
    },
    {
      "epoch": 0.6522781774580336,
      "grad_norm": 0.17621521651744843,
      "learning_rate": 8.821481276340112e-05,
      "loss": 1.6093,
      "step": 204
    },
    {
      "epoch": 0.6554756195043965,
      "grad_norm": 0.1912049949169159,
      "learning_rate": 8.676762290584585e-05,
      "loss": 1.353,
      "step": 205
    },
    {
      "epoch": 0.6586730615507594,
      "grad_norm": 0.2157009094953537,
      "learning_rate": 8.532755586477324e-05,
      "loss": 1.4063,
      "step": 206
    },
    {
      "epoch": 0.6618705035971223,
      "grad_norm": 0.18072722852230072,
      "learning_rate": 8.389477385665732e-05,
      "loss": 1.5591,
      "step": 207
    },
    {
      "epoch": 0.6650679456434853,
      "grad_norm": 0.22034448385238647,
      "learning_rate": 8.246943827734897e-05,
      "loss": 1.4766,
      "step": 208
    },
    {
      "epoch": 0.6682653876898481,
      "grad_norm": 0.21938645839691162,
      "learning_rate": 8.105170968389552e-05,
      "loss": 1.3791,
      "step": 209
    },
    {
      "epoch": 0.6714628297362111,
      "grad_norm": 0.19702577590942383,
      "learning_rate": 7.964174777645448e-05,
      "loss": 1.5582,
      "step": 210
    },
    {
      "epoch": 0.6746602717825739,
      "grad_norm": 0.20586428046226501,
      "learning_rate": 7.823971138030466e-05,
      "loss": 1.4005,
      "step": 211
    },
    {
      "epoch": 0.6778577138289369,
      "grad_norm": 0.1924622356891632,
      "learning_rate": 7.684575842795485e-05,
      "loss": 1.4078,
      "step": 212
    },
    {
      "epoch": 0.6810551558752997,
      "grad_norm": 0.1937723606824875,
      "learning_rate": 7.546004594135356e-05,
      "loss": 1.2821,
      "step": 213
    },
    {
      "epoch": 0.6842525979216627,
      "grad_norm": 0.22969581186771393,
      "learning_rate": 7.408273001420153e-05,
      "loss": 1.2398,
      "step": 214
    },
    {
      "epoch": 0.6874500399680256,
      "grad_norm": 0.19231727719306946,
      "learning_rate": 7.271396579436825e-05,
      "loss": 1.3752,
      "step": 215
    },
    {
      "epoch": 0.6906474820143885,
      "grad_norm": 0.20469219982624054,
      "learning_rate": 7.135390746641526e-05,
      "loss": 1.352,
      "step": 216
    },
    {
      "epoch": 0.6938449240607514,
      "grad_norm": 0.19728676974773407,
      "learning_rate": 7.000270823422837e-05,
      "loss": 1.5623,
      "step": 217
    },
    {
      "epoch": 0.6970423661071143,
      "grad_norm": 0.22052626311779022,
      "learning_rate": 6.866052030375974e-05,
      "loss": 1.4183,
      "step": 218
    },
    {
      "epoch": 0.7002398081534772,
      "grad_norm": 0.19779476523399353,
      "learning_rate": 6.732749486588266e-05,
      "loss": 1.4014,
      "step": 219
    },
    {
      "epoch": 0.7034372501998402,
      "grad_norm": 0.1978594809770584,
      "learning_rate": 6.600378207936069e-05,
      "loss": 1.4317,
      "step": 220
    },
    {
      "epoch": 0.706634692246203,
      "grad_norm": 0.2020850032567978,
      "learning_rate": 6.468953105393297e-05,
      "loss": 1.4208,
      "step": 221
    },
    {
      "epoch": 0.709832134292566,
      "grad_norm": 0.18292494118213654,
      "learning_rate": 6.338488983351777e-05,
      "loss": 1.3283,
      "step": 222
    },
    {
      "epoch": 0.7130295763389288,
      "grad_norm": 0.2223280966281891,
      "learning_rate": 6.209000537953605e-05,
      "loss": 1.4245,
      "step": 223
    },
    {
      "epoch": 0.7162270183852918,
      "grad_norm": 0.22692078351974487,
      "learning_rate": 6.080502355435701e-05,
      "loss": 1.5982,
      "step": 224
    },
    {
      "epoch": 0.7194244604316546,
      "grad_norm": 0.19702717661857605,
      "learning_rate": 5.9530089104867386e-05,
      "loss": 1.3909,
      "step": 225
    },
    {
      "epoch": 0.7226219024780176,
      "grad_norm": 0.22220925986766815,
      "learning_rate": 5.826534564616633e-05,
      "loss": 1.4322,
      "step": 226
    },
    {
      "epoch": 0.7258193445243805,
      "grad_norm": 0.20837551355361938,
      "learning_rate": 5.701093564538806e-05,
      "loss": 1.3919,
      "step": 227
    },
    {
      "epoch": 0.7290167865707434,
      "grad_norm": 0.1905641108751297,
      "learning_rate": 5.5767000405653636e-05,
      "loss": 1.446,
      "step": 228
    },
    {
      "epoch": 0.7322142286171063,
      "grad_norm": 0.20399922132492065,
      "learning_rate": 5.453368005015363e-05,
      "loss": 1.3922,
      "step": 229
    },
    {
      "epoch": 0.7354116706634692,
      "grad_norm": 0.19176483154296875,
      "learning_rate": 5.3311113506364116e-05,
      "loss": 1.3255,
      "step": 230
    },
    {
      "epoch": 0.7386091127098321,
      "grad_norm": 0.21297192573547363,
      "learning_rate": 5.209943849039722e-05,
      "loss": 1.3992,
      "step": 231
    },
    {
      "epoch": 0.7418065547561951,
      "grad_norm": 0.20219087600708008,
      "learning_rate": 5.089879149148781e-05,
      "loss": 1.5462,
      "step": 232
    },
    {
      "epoch": 0.7450039968025579,
      "grad_norm": 0.1977456510066986,
      "learning_rate": 4.9709307756618985e-05,
      "loss": 1.4046,
      "step": 233
    },
    {
      "epoch": 0.7482014388489209,
      "grad_norm": 0.22329548001289368,
      "learning_rate": 4.853112127528698e-05,
      "loss": 1.5767,
      "step": 234
    },
    {
      "epoch": 0.7513988808952837,
      "grad_norm": 0.20563232898712158,
      "learning_rate": 4.736436476440791e-05,
      "loss": 1.6348,
      "step": 235
    },
    {
      "epoch": 0.7545963229416467,
      "grad_norm": 0.19388997554779053,
      "learning_rate": 4.6209169653368086e-05,
      "loss": 1.364,
      "step": 236
    },
    {
      "epoch": 0.7577937649880095,
      "grad_norm": 0.2103840559720993,
      "learning_rate": 4.506566606921864e-05,
      "loss": 1.4538,
      "step": 237
    },
    {
      "epoch": 0.7609912070343725,
      "grad_norm": 0.17306749522686005,
      "learning_rate": 4.3933982822017876e-05,
      "loss": 1.4435,
      "step": 238
    },
    {
      "epoch": 0.7641886490807354,
      "grad_norm": 0.20918579399585724,
      "learning_rate": 4.2814247390321215e-05,
      "loss": 1.2357,
      "step": 239
    },
    {
      "epoch": 0.7673860911270983,
      "grad_norm": 0.21173876523971558,
      "learning_rate": 4.1706585906821334e-05,
      "loss": 1.2602,
      "step": 240
    },
    {
      "epoch": 0.7705835331734612,
      "grad_norm": 0.19886651635169983,
      "learning_rate": 4.0611123144140075e-05,
      "loss": 1.4166,
      "step": 241
    },
    {
      "epoch": 0.7737809752198241,
      "grad_norm": 0.19375504553318024,
      "learning_rate": 3.952798250077317e-05,
      "loss": 1.3777,
      "step": 242
    },
    {
      "epoch": 0.7769784172661871,
      "grad_norm": 0.20145930349826813,
      "learning_rate": 3.84572859871904e-05,
      "loss": 1.3258,
      "step": 243
    },
    {
      "epoch": 0.78017585931255,
      "grad_norm": 0.2076532244682312,
      "learning_rate": 3.739915421209133e-05,
      "loss": 1.3921,
      "step": 244
    },
    {
      "epoch": 0.7833733013589129,
      "grad_norm": 0.19265635311603546,
      "learning_rate": 3.635370636881958e-05,
      "loss": 1.4043,
      "step": 245
    },
    {
      "epoch": 0.7865707434052758,
      "grad_norm": 0.19883492588996887,
      "learning_rate": 3.532106022193615e-05,
      "loss": 1.346,
      "step": 246
    },
    {
      "epoch": 0.7897681854516387,
      "grad_norm": 0.18948738276958466,
      "learning_rate": 3.4301332093953807e-05,
      "loss": 1.4363,
      "step": 247
    },
    {
      "epoch": 0.7929656274980016,
      "grad_norm": 0.18976429104804993,
      "learning_rate": 3.3294636852234105e-05,
      "loss": 1.4316,
      "step": 248
    },
    {
      "epoch": 0.7961630695443646,
      "grad_norm": 0.202013298869133,
      "learning_rate": 3.230108789604792e-05,
      "loss": 1.4532,
      "step": 249
    },
    {
      "epoch": 0.7993605115907274,
      "grad_norm": 0.2116522341966629,
      "learning_rate": 3.132079714380171e-05,
      "loss": 1.5129,
      "step": 250
    },
    {
      "epoch": 0.8025579536370904,
      "grad_norm": 0.19418169558048248,
      "learning_rate": 3.035387502043052e-05,
      "loss": 1.3265,
      "step": 251
    },
    {
      "epoch": 0.8057553956834532,
      "grad_norm": 0.21084119379520416,
      "learning_rate": 2.9400430444958932e-05,
      "loss": 1.3929,
      "step": 252
    },
    {
      "epoch": 0.8089528377298162,
      "grad_norm": 0.23588140308856964,
      "learning_rate": 2.846057081823201e-05,
      "loss": 1.2077,
      "step": 253
    },
    {
      "epoch": 0.812150279776179,
      "grad_norm": 0.21185244619846344,
      "learning_rate": 2.7534402010817157e-05,
      "loss": 1.2874,
      "step": 254
    },
    {
      "epoch": 0.815347721822542,
      "grad_norm": 0.184846431016922,
      "learning_rate": 2.6622028351078277e-05,
      "loss": 1.4785,
      "step": 255
    },
    {
      "epoch": 0.8185451638689049,
      "grad_norm": 0.1995445042848587,
      "learning_rate": 2.5723552613423687e-05,
      "loss": 1.4153,
      "step": 256
    },
    {
      "epoch": 0.8217426059152678,
      "grad_norm": 0.20493745803833008,
      "learning_rate": 2.4839076006729082e-05,
      "loss": 1.448,
      "step": 257
    },
    {
      "epoch": 0.8249400479616307,
      "grad_norm": 0.1989341676235199,
      "learning_rate": 2.3968698162936854e-05,
      "loss": 1.4733,
      "step": 258
    },
    {
      "epoch": 0.8281374900079936,
      "grad_norm": 0.20579148828983307,
      "learning_rate": 2.311251712583307e-05,
      "loss": 1.4746,
      "step": 259
    },
    {
      "epoch": 0.8313349320543565,
      "grad_norm": 0.2025279700756073,
      "learning_rate": 2.2270629340003303e-05,
      "loss": 1.6248,
      "step": 260
    },
    {
      "epoch": 0.8345323741007195,
      "grad_norm": 0.17980627715587616,
      "learning_rate": 2.1443129639968615e-05,
      "loss": 1.3753,
      "step": 261
    },
    {
      "epoch": 0.8377298161470823,
      "grad_norm": 0.21116185188293457,
      "learning_rate": 2.063011123950295e-05,
      "loss": 1.2975,
      "step": 262
    },
    {
      "epoch": 0.8409272581934453,
      "grad_norm": 0.20071591436862946,
      "learning_rate": 1.9831665721132954e-05,
      "loss": 1.444,
      "step": 263
    },
    {
      "epoch": 0.8441247002398081,
      "grad_norm": 0.19569140672683716,
      "learning_rate": 1.9047883025821774e-05,
      "loss": 1.5126,
      "step": 264
    },
    {
      "epoch": 0.8473221422861711,
      "grad_norm": 0.19419822096824646,
      "learning_rate": 1.827885144283769e-05,
      "loss": 1.3867,
      "step": 265
    },
    {
      "epoch": 0.8505195843325339,
      "grad_norm": 0.19556277990341187,
      "learning_rate": 1.75246575998086e-05,
      "loss": 1.3758,
      "step": 266
    },
    {
      "epoch": 0.8537170263788969,
      "grad_norm": 0.20848549902439117,
      "learning_rate": 1.678538645296391e-05,
      "loss": 1.4835,
      "step": 267
    },
    {
      "epoch": 0.8569144684252598,
      "grad_norm": 0.19634144008159637,
      "learning_rate": 1.6061121277564743e-05,
      "loss": 1.4624,
      "step": 268
    },
    {
      "epoch": 0.8601119104716227,
      "grad_norm": 0.19600766897201538,
      "learning_rate": 1.535194365852315e-05,
      "loss": 1.2323,
      "step": 269
    },
    {
      "epoch": 0.8633093525179856,
      "grad_norm": 0.21323877573013306,
      "learning_rate": 1.4657933481212242e-05,
      "loss": 1.5224,
      "step": 270
    },
    {
      "epoch": 0.8665067945643485,
      "grad_norm": 0.18555647134780884,
      "learning_rate": 1.3979168922467298e-05,
      "loss": 1.3663,
      "step": 271
    },
    {
      "epoch": 0.8697042366107114,
      "grad_norm": 0.19477520883083344,
      "learning_rate": 1.3315726441779629e-05,
      "loss": 1.4892,
      "step": 272
    },
    {
      "epoch": 0.8729016786570744,
      "grad_norm": 0.19639001786708832,
      "learning_rate": 1.2667680772683825e-05,
      "loss": 1.2377,
      "step": 273
    },
    {
      "epoch": 0.8760991207034372,
      "grad_norm": 0.21710480749607086,
      "learning_rate": 1.2035104914339188e-05,
      "loss": 1.3991,
      "step": 274
    },
    {
      "epoch": 0.8792965627498002,
      "grad_norm": 0.21137666702270508,
      "learning_rate": 1.1418070123306989e-05,
      "loss": 1.5307,
      "step": 275
    },
    {
      "epoch": 0.882494004796163,
      "grad_norm": 0.19870568811893463,
      "learning_rate": 1.0816645905523597e-05,
      "loss": 1.341,
      "step": 276
    },
    {
      "epoch": 0.885691446842526,
      "grad_norm": 0.2340983897447586,
      "learning_rate": 1.0230900008471072e-05,
      "loss": 1.3578,
      "step": 277
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.18889744579792023,
      "learning_rate": 9.660898413545692e-06,
      "loss": 1.4085,
      "step": 278
    },
    {
      "epoch": 0.8920863309352518,
      "grad_norm": 0.213284432888031,
      "learning_rate": 9.106705328625408e-06,
      "loss": 1.3843,
      "step": 279
    },
    {
      "epoch": 0.8952837729816147,
      "grad_norm": 0.2060411274433136,
      "learning_rate": 8.568383180837368e-06,
      "loss": 1.473,
      "step": 280
    },
    {
      "epoch": 0.8984812150279776,
      "grad_norm": 0.18018406629562378,
      "learning_rate": 8.04599260952557e-06,
      "loss": 1.3782,
      "step": 281
    },
    {
      "epoch": 0.9016786570743405,
      "grad_norm": 0.18678754568099976,
      "learning_rate": 7.539592459420219e-06,
      "loss": 1.4252,
      "step": 282
    },
    {
      "epoch": 0.9048760991207034,
      "grad_norm": 0.2027515172958374,
      "learning_rate": 7.049239774009213e-06,
      "loss": 1.3717,
      "step": 283
    },
    {
      "epoch": 0.9080735411670664,
      "grad_norm": 0.20960167050361633,
      "learning_rate": 6.574989789112372e-06,
      "loss": 1.2815,
      "step": 284
    },
    {
      "epoch": 0.9112709832134293,
      "grad_norm": 0.19627049565315247,
      "learning_rate": 6.11689592665951e-06,
      "loss": 1.4348,
      "step": 285
    },
    {
      "epoch": 0.9144684252597922,
      "grad_norm": 0.20119017362594604,
      "learning_rate": 5.675009788672596e-06,
      "loss": 1.3343,
      "step": 286
    },
    {
      "epoch": 0.9176658673061551,
      "grad_norm": 0.18706481158733368,
      "learning_rate": 5.2493811514531635e-06,
      "loss": 1.3721,
      "step": 287
    },
    {
      "epoch": 0.920863309352518,
      "grad_norm": 0.19794286787509918,
      "learning_rate": 4.840057959975169e-06,
      "loss": 1.3626,
      "step": 288
    },
    {
      "epoch": 0.9240607513988809,
      "grad_norm": 0.1808895319700241,
      "learning_rate": 4.44708632248425e-06,
      "loss": 1.5342,
      "step": 289
    },
    {
      "epoch": 0.9272581934452439,
      "grad_norm": 0.1820111721754074,
      "learning_rate": 4.070510505303814e-06,
      "loss": 1.4357,
      "step": 290
    },
    {
      "epoch": 0.9304556354916067,
      "grad_norm": 0.1756613701581955,
      "learning_rate": 3.710372927848776e-06,
      "loss": 1.328,
      "step": 291
    },
    {
      "epoch": 0.9336530775379697,
      "grad_norm": 0.19259536266326904,
      "learning_rate": 3.366714157847078e-06,
      "loss": 1.2882,
      "step": 292
    },
    {
      "epoch": 0.9368505195843325,
      "grad_norm": 0.20220039784908295,
      "learning_rate": 3.0395729067700324e-06,
      "loss": 1.3903,
      "step": 293
    },
    {
      "epoch": 0.9400479616306955,
      "grad_norm": 0.1991778463125229,
      "learning_rate": 2.728986025471641e-06,
      "loss": 1.3649,
      "step": 294
    },
    {
      "epoch": 0.9432454036770583,
      "grad_norm": 0.20098921656608582,
      "learning_rate": 2.4349885000374657e-06,
      "loss": 1.4128,
      "step": 295
    },
    {
      "epoch": 0.9464428457234213,
      "grad_norm": 0.18276216089725494,
      "learning_rate": 2.1576134478437313e-06,
      "loss": 1.3548,
      "step": 296
    },
    {
      "epoch": 0.9496402877697842,
      "grad_norm": 0.21758389472961426,
      "learning_rate": 1.8968921138267091e-06,
      "loss": 1.4765,
      "step": 297
    },
    {
      "epoch": 0.9528377298161471,
      "grad_norm": 0.18690507113933563,
      "learning_rate": 1.6528538669631997e-06,
      "loss": 1.5375,
      "step": 298
    },
    {
      "epoch": 0.95603517186251,
      "grad_norm": 0.1706872582435608,
      "learning_rate": 1.4255261969622456e-06,
      "loss": 1.2775,
      "step": 299
    },
    {
      "epoch": 0.9592326139088729,
      "grad_norm": 0.2452152669429779,
      "learning_rate": 1.2149347111684749e-06,
      "loss": 1.2828,
      "step": 300
    },
    {
      "epoch": 0.9624300559552358,
      "grad_norm": 0.17894317209720612,
      "learning_rate": 1.0211031316776919e-06,
      "loss": 1.4131,
      "step": 301
    },
    {
      "epoch": 0.9656274980015987,
      "grad_norm": 0.21982480585575104,
      "learning_rate": 8.440532926646315e-07,
      "loss": 1.3501,
      "step": 302
    },
    {
      "epoch": 0.9688249400479616,
      "grad_norm": 0.19508808851242065,
      "learning_rate": 6.838051379234099e-07,
      "loss": 1.3474,
      "step": 303
    },
    {
      "epoch": 0.9720223820943246,
      "grad_norm": 0.1852046549320221,
      "learning_rate": 5.403767186210218e-07,
      "loss": 1.3791,
      "step": 304
    },
    {
      "epoch": 0.9752198241406874,
      "grad_norm": 0.18738119304180145,
      "learning_rate": 4.137841912639328e-07,
      "loss": 1.4893,
      "step": 305
    },
    {
      "epoch": 0.9784172661870504,
      "grad_norm": 0.20034608244895935,
      "learning_rate": 3.0404181587811994e-07,
      "loss": 1.4388,
      "step": 306
    },
    {
      "epoch": 0.9816147082334132,
      "grad_norm": 0.20295751094818115,
      "learning_rate": 2.1116195440278872e-07,
      "loss": 1.4804,
      "step": 307
    },
    {
      "epoch": 0.9848121502797762,
      "grad_norm": 0.207365021109581,
      "learning_rate": 1.3515506929778762e-07,
      "loss": 1.4719,
      "step": 308
    },
    {
      "epoch": 0.988009592326139,
      "grad_norm": 0.2223723828792572,
      "learning_rate": 7.602972236513405e-08,
      "loss": 1.3123,
      "step": 309
    },
    {
      "epoch": 0.991207034372502,
      "grad_norm": 0.2046136111021042,
      "learning_rate": 3.3792573784585665e-08,
      "loss": 1.4272,
      "step": 310
    },
    {
      "epoch": 0.9944044764188649,
      "grad_norm": 0.21449051797389984,
      "learning_rate": 8.448381363307388e-09,
      "loss": 1.3367,
      "step": 311
    },
    {
      "epoch": 0.9976019184652278,
      "grad_norm": 0.21067871153354645,
      "learning_rate": 0.0,
      "loss": 1.4037,
      "step": 312
    }
  ],
  "logging_steps": 1,
  "max_steps": 312,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.138882997433958e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}