{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.27279497818916904,
  "eval_steps": 2000,
  "global_step": 641,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00042557718906266625,
      "grad_norm": 0.3652644157409668,
      "learning_rate": 1e-05,
      "loss": 1.7802,
      "step": 1
    },
    {
      "epoch": 0.00042557718906266625,
      "eval_loss": 1.6649078130722046,
      "eval_runtime": 273.5839,
      "eval_samples_per_second": 7.234,
      "eval_steps_per_second": 1.809,
      "step": 1
    },
    {
      "epoch": 0.0008511543781253325,
      "grad_norm": 0.4177471697330475,
      "learning_rate": 2e-05,
      "loss": 1.579,
      "step": 2
    },
    {
      "epoch": 0.0012767315671879987,
      "grad_norm": 0.30818939208984375,
      "learning_rate": 3e-05,
      "loss": 1.6122,
      "step": 3
    },
    {
      "epoch": 0.001702308756250665,
      "grad_norm": 0.37045714259147644,
      "learning_rate": 4e-05,
      "loss": 1.72,
      "step": 4
    },
    {
      "epoch": 0.002127885945313331,
      "grad_norm": 0.34079983830451965,
      "learning_rate": 5e-05,
      "loss": 1.5969,
      "step": 5
    },
    {
      "epoch": 0.0025534631343759975,
      "grad_norm": 0.381185919046402,
      "learning_rate": 6e-05,
      "loss": 1.7075,
      "step": 6
    },
    {
      "epoch": 0.0029790403234386635,
      "grad_norm": 0.43642061948776245,
      "learning_rate": 7e-05,
      "loss": 1.5826,
      "step": 7
    },
    {
      "epoch": 0.00340461751250133,
      "grad_norm": 0.395150750875473,
      "learning_rate": 8e-05,
      "loss": 1.666,
      "step": 8
    },
    {
      "epoch": 0.003830194701563996,
      "grad_norm": 0.4904192090034485,
      "learning_rate": 9e-05,
      "loss": 1.5554,
      "step": 9
    },
    {
      "epoch": 0.004255771890626662,
      "grad_norm": 0.4711903929710388,
      "learning_rate": 0.0001,
      "loss": 1.6409,
      "step": 10
    },
    {
      "epoch": 0.0046813490796893285,
      "grad_norm": 0.443033367395401,
      "learning_rate": 9.99993803019373e-05,
      "loss": 1.5268,
      "step": 11
    },
    {
      "epoch": 0.005106926268751995,
      "grad_norm": 0.4341345727443695,
      "learning_rate": 9.999752122311022e-05,
      "loss": 1.6879,
      "step": 12
    },
    {
      "epoch": 0.005532503457814661,
      "grad_norm": 0.36973270773887634,
      "learning_rate": 9.999442280960142e-05,
      "loss": 1.5406,
      "step": 13
    },
    {
      "epoch": 0.005958080646877327,
      "grad_norm": 0.41089990735054016,
      "learning_rate": 9.99900851382142e-05,
      "loss": 1.3971,
      "step": 14
    },
    {
      "epoch": 0.0063836578359399935,
      "grad_norm": 0.37346774339675903,
      "learning_rate": 9.998450831647039e-05,
      "loss": 1.5966,
      "step": 15
    },
    {
      "epoch": 0.00680923502500266,
      "grad_norm": 0.37496069073677063,
      "learning_rate": 9.99776924826078e-05,
      "loss": 1.4682,
      "step": 16
    },
    {
      "epoch": 0.007234812214065326,
      "grad_norm": 0.496907114982605,
      "learning_rate": 9.996963780557683e-05,
      "loss": 1.4213,
      "step": 17
    },
    {
      "epoch": 0.007660389403127992,
      "grad_norm": 0.39572206139564514,
      "learning_rate": 9.996034448503616e-05,
      "loss": 1.3925,
      "step": 18
    },
    {
      "epoch": 0.00808596659219066,
      "grad_norm": 0.3716890215873718,
      "learning_rate": 9.994981275134791e-05,
      "loss": 1.3852,
      "step": 19
    },
    {
      "epoch": 0.008511543781253324,
      "grad_norm": 0.39378899335861206,
      "learning_rate": 9.993804286557188e-05,
      "loss": 1.4922,
      "step": 20
    },
    {
      "epoch": 0.00893712097031599,
      "grad_norm": 0.394651859998703,
      "learning_rate": 9.992503511945907e-05,
      "loss": 1.4906,
      "step": 21
    },
    {
      "epoch": 0.009362698159378657,
      "grad_norm": 0.43118777871131897,
      "learning_rate": 9.991078983544452e-05,
      "loss": 1.355,
      "step": 22
    },
    {
      "epoch": 0.009788275348441323,
      "grad_norm": 0.4793911278247833,
      "learning_rate": 9.989530736663918e-05,
      "loss": 1.5297,
      "step": 23
    },
    {
      "epoch": 0.01021385253750399,
      "grad_norm": 0.4128536283969879,
      "learning_rate": 9.987858809682132e-05,
      "loss": 1.4856,
      "step": 24
    },
    {
      "epoch": 0.010639429726566656,
      "grad_norm": 0.33675417304039,
      "learning_rate": 9.986063244042689e-05,
      "loss": 1.4352,
      "step": 25
    },
    {
      "epoch": 0.011065006915629323,
      "grad_norm": 0.35233640670776367,
      "learning_rate": 9.984144084253932e-05,
      "loss": 1.3988,
      "step": 26
    },
    {
      "epoch": 0.01149058410469199,
      "grad_norm": 0.37231576442718506,
      "learning_rate": 9.982101377887845e-05,
      "loss": 1.4834,
      "step": 27
    },
    {
      "epoch": 0.011916161293754654,
      "grad_norm": 0.35900095105171204,
      "learning_rate": 9.979935175578873e-05,
      "loss": 1.441,
      "step": 28
    },
    {
      "epoch": 0.01234173848281732,
      "grad_norm": 0.33582940697669983,
      "learning_rate": 9.977645531022672e-05,
      "loss": 1.3407,
      "step": 29
    },
    {
      "epoch": 0.012767315671879987,
      "grad_norm": 0.3340405225753784,
      "learning_rate": 9.975232500974776e-05,
      "loss": 1.3519,
      "step": 30
    },
    {
      "epoch": 0.013192892860942653,
      "grad_norm": 0.32448646426200867,
      "learning_rate": 9.972696145249185e-05,
      "loss": 1.3666,
      "step": 31
    },
    {
      "epoch": 0.01361847005000532,
      "grad_norm": 0.31764093041419983,
      "learning_rate": 9.970036526716889e-05,
      "loss": 1.3873,
      "step": 32
    },
    {
      "epoch": 0.014044047239067986,
      "grad_norm": 0.37046271562576294,
      "learning_rate": 9.967253711304304e-05,
      "loss": 1.4668,
      "step": 33
    },
    {
      "epoch": 0.014469624428130653,
      "grad_norm": 0.35011404752731323,
      "learning_rate": 9.964347767991644e-05,
      "loss": 1.4616,
      "step": 34
    },
    {
      "epoch": 0.01489520161719332,
      "grad_norm": 0.37187859416007996,
      "learning_rate": 9.961318768811209e-05,
      "loss": 1.5379,
      "step": 35
    },
    {
      "epoch": 0.015320778806255984,
      "grad_norm": 0.37234461307525635,
      "learning_rate": 9.958166788845592e-05,
      "loss": 1.5205,
      "step": 36
    },
    {
      "epoch": 0.01574635599531865,
      "grad_norm": 0.3146509528160095,
      "learning_rate": 9.954891906225832e-05,
      "loss": 1.3043,
      "step": 37
    },
    {
      "epoch": 0.01617193318438132,
      "grad_norm": 0.29514676332473755,
      "learning_rate": 9.951494202129461e-05,
      "loss": 1.2737,
      "step": 38
    },
    {
      "epoch": 0.016597510373443983,
      "grad_norm": 0.3454630672931671,
      "learning_rate": 9.94797376077851e-05,
      "loss": 1.4062,
      "step": 39
    },
    {
      "epoch": 0.017023087562506648,
      "grad_norm": 0.32935354113578796,
      "learning_rate": 9.944330669437402e-05,
      "loss": 1.3869,
      "step": 40
    },
    {
      "epoch": 0.017448664751569316,
      "grad_norm": 0.3183695077896118,
      "learning_rate": 9.940565018410805e-05,
      "loss": 1.4048,
      "step": 41
    },
    {
      "epoch": 0.01787424194063198,
      "grad_norm": 0.3441618084907532,
      "learning_rate": 9.936676901041386e-05,
      "loss": 1.4645,
      "step": 42
    },
    {
      "epoch": 0.01829981912969465,
      "grad_norm": 0.2902175784111023,
      "learning_rate": 9.932666413707494e-05,
      "loss": 1.4638,
      "step": 43
    },
    {
      "epoch": 0.018725396318757314,
      "grad_norm": 0.3078022003173828,
      "learning_rate": 9.92853365582078e-05,
      "loss": 1.4624,
      "step": 44
    },
    {
      "epoch": 0.019150973507819982,
      "grad_norm": 0.34494903683662415,
      "learning_rate": 9.924278729823727e-05,
      "loss": 1.458,
      "step": 45
    },
    {
      "epoch": 0.019576550696882647,
      "grad_norm": 0.3399023413658142,
      "learning_rate": 9.91990174118711e-05,
      "loss": 1.4537,
      "step": 46
    },
    {
      "epoch": 0.02000212788594531,
      "grad_norm": 0.3627009391784668,
      "learning_rate": 9.915402798407383e-05,
      "loss": 1.366,
      "step": 47
    },
    {
      "epoch": 0.02042770507500798,
      "grad_norm": 0.3227439224720001,
      "learning_rate": 9.910782013003994e-05,
      "loss": 1.4666,
      "step": 48
    },
    {
      "epoch": 0.020853282264070645,
      "grad_norm": 0.37096184492111206,
      "learning_rate": 9.906039499516611e-05,
      "loss": 1.3548,
      "step": 49
    },
    {
      "epoch": 0.021278859453133313,
      "grad_norm": 0.3019040822982788,
      "learning_rate": 9.90117537550229e-05,
      "loss": 1.4061,
      "step": 50
    },
    {
      "epoch": 0.021704436642195978,
      "grad_norm": 0.4224420487880707,
      "learning_rate": 9.896189761532563e-05,
      "loss": 1.4288,
      "step": 51
    },
    {
      "epoch": 0.022130013831258646,
      "grad_norm": 0.3436116874217987,
      "learning_rate": 9.89108278119044e-05,
      "loss": 1.4487,
      "step": 52
    },
    {
      "epoch": 0.02255559102032131,
      "grad_norm": 0.3421087861061096,
      "learning_rate": 9.885854561067357e-05,
      "loss": 1.4266,
      "step": 53
    },
    {
      "epoch": 0.02298116820938398,
      "grad_norm": 0.3183712661266327,
      "learning_rate": 9.880505230760027e-05,
      "loss": 1.4702,
      "step": 54
    },
    {
      "epoch": 0.023406745398446643,
      "grad_norm": 0.4090493321418762,
      "learning_rate": 9.875034922867236e-05,
      "loss": 1.5018,
      "step": 55
    },
    {
      "epoch": 0.023832322587509308,
      "grad_norm": 0.3179483115673065,
      "learning_rate": 9.869443772986551e-05,
      "loss": 1.4036,
      "step": 56
    },
    {
      "epoch": 0.024257899776571976,
      "grad_norm": 0.29585397243499756,
      "learning_rate": 9.863731919710963e-05,
      "loss": 1.3401,
      "step": 57
    },
    {
      "epoch": 0.02468347696563464,
      "grad_norm": 0.33189573884010315,
      "learning_rate": 9.85789950462545e-05,
      "loss": 1.4556,
      "step": 58
    },
    {
      "epoch": 0.02510905415469731,
      "grad_norm": 0.3170922100543976,
      "learning_rate": 9.85194667230346e-05,
      "loss": 1.3966,
      "step": 59
    },
    {
      "epoch": 0.025534631343759974,
      "grad_norm": 0.3412818908691406,
      "learning_rate": 9.845873570303346e-05,
      "loss": 1.4879,
      "step": 60
    },
    {
      "epoch": 0.025960208532822642,
      "grad_norm": 0.32975345849990845,
      "learning_rate": 9.839680349164684e-05,
      "loss": 1.4782,
      "step": 61
    },
    {
      "epoch": 0.026385785721885307,
      "grad_norm": 0.35211431980133057,
      "learning_rate": 9.833367162404563e-05,
      "loss": 1.4164,
      "step": 62
    },
    {
      "epoch": 0.02681136291094797,
      "grad_norm": 0.3229023516178131,
      "learning_rate": 9.826934166513767e-05,
      "loss": 1.4521,
      "step": 63
    },
    {
      "epoch": 0.02723694010001064,
      "grad_norm": 0.3026021718978882,
      "learning_rate": 9.8203815209529e-05,
      "loss": 1.4134,
      "step": 64
    },
    {
      "epoch": 0.027662517289073305,
      "grad_norm": 0.33294665813446045,
      "learning_rate": 9.813709388148429e-05,
      "loss": 1.4614,
      "step": 65
    },
    {
      "epoch": 0.028088094478135973,
      "grad_norm": 0.32732123136520386,
      "learning_rate": 9.806917933488668e-05,
      "loss": 1.4745,
      "step": 66
    },
    {
      "epoch": 0.028513671667198637,
      "grad_norm": 0.3315070569515228,
      "learning_rate": 9.800007325319666e-05,
      "loss": 1.4696,
      "step": 67
    },
    {
      "epoch": 0.028939248856261306,
      "grad_norm": 0.3497024476528168,
      "learning_rate": 9.792977734941049e-05,
      "loss": 1.4456,
      "step": 68
    },
    {
      "epoch": 0.02936482604532397,
      "grad_norm": 0.3212997615337372,
      "learning_rate": 9.785829336601751e-05,
      "loss": 1.4609,
      "step": 69
    },
    {
      "epoch": 0.02979040323438664,
      "grad_norm": 0.3074348270893097,
      "learning_rate": 9.778562307495722e-05,
      "loss": 1.4099,
      "step": 70
    },
    {
      "epoch": 0.030215980423449303,
      "grad_norm": 0.3627077043056488,
      "learning_rate": 9.771176827757512e-05,
      "loss": 1.4311,
      "step": 71
    },
    {
      "epoch": 0.030641557612511968,
      "grad_norm": 0.2996334135532379,
      "learning_rate": 9.763673080457823e-05,
      "loss": 1.3821,
      "step": 72
    },
    {
      "epoch": 0.031067134801574636,
      "grad_norm": 0.3352314233779907,
      "learning_rate": 9.756051251598962e-05,
      "loss": 1.4393,
      "step": 73
    },
    {
      "epoch": 0.0314927119906373,
      "grad_norm": 0.34438130259513855,
      "learning_rate": 9.748311530110229e-05,
      "loss": 1.4367,
      "step": 74
    },
    {
      "epoch": 0.031918289179699966,
      "grad_norm": 0.3071272373199463,
      "learning_rate": 9.740454107843243e-05,
      "loss": 1.3883,
      "step": 75
    },
    {
      "epoch": 0.03234386636876264,
      "grad_norm": 0.32912176847457886,
      "learning_rate": 9.732479179567177e-05,
      "loss": 1.4138,
      "step": 76
    },
    {
      "epoch": 0.0327694435578253,
      "grad_norm": 0.3019627332687378,
      "learning_rate": 9.724386942963938e-05,
      "loss": 1.3916,
      "step": 77
    },
    {
      "epoch": 0.03319502074688797,
      "grad_norm": 0.35912489891052246,
      "learning_rate": 9.716177598623257e-05,
      "loss": 1.3278,
      "step": 78
    },
    {
      "epoch": 0.03362059793595063,
      "grad_norm": 0.31681278347969055,
      "learning_rate": 9.707851350037726e-05,
      "loss": 1.2639,
      "step": 79
    },
    {
      "epoch": 0.034046175125013296,
      "grad_norm": 0.37369856238365173,
      "learning_rate": 9.699408403597749e-05,
      "loss": 1.4526,
      "step": 80
    },
    {
      "epoch": 0.03447175231407597,
      "grad_norm": 0.3292621076107025,
      "learning_rate": 9.69084896858643e-05,
      "loss": 1.3093,
      "step": 81
    },
    {
      "epoch": 0.03489732950313863,
      "grad_norm": 0.32593628764152527,
      "learning_rate": 9.68217325717438e-05,
      "loss": 1.4091,
      "step": 82
    },
    {
      "epoch": 0.0353229066922013,
      "grad_norm": 0.34049278497695923,
      "learning_rate": 9.67338148441446e-05,
      "loss": 1.5005,
      "step": 83
    },
    {
      "epoch": 0.03574848388126396,
      "grad_norm": 0.3360632061958313,
      "learning_rate": 9.664473868236452e-05,
      "loss": 1.4722,
      "step": 84
    },
    {
      "epoch": 0.036174061070326634,
      "grad_norm": 0.3554811477661133,
      "learning_rate": 9.655450629441659e-05,
      "loss": 1.4261,
      "step": 85
    },
    {
      "epoch": 0.0365996382593893,
      "grad_norm": 0.35133641958236694,
      "learning_rate": 9.646311991697419e-05,
      "loss": 1.3752,
      "step": 86
    },
    {
      "epoch": 0.03702521544845196,
      "grad_norm": 0.29005640745162964,
      "learning_rate": 9.637058181531582e-05,
      "loss": 1.2862,
      "step": 87
    },
    {
      "epoch": 0.03745079263751463,
      "grad_norm": 0.3389727771282196,
      "learning_rate": 9.627689428326873e-05,
      "loss": 1.3316,
      "step": 88
    },
    {
      "epoch": 0.03787636982657729,
      "grad_norm": 0.35122182965278625,
      "learning_rate": 9.618205964315223e-05,
      "loss": 1.4339,
      "step": 89
    },
    {
      "epoch": 0.038301947015639964,
      "grad_norm": 0.3106866776943207,
      "learning_rate": 9.608608024572004e-05,
      "loss": 1.4314,
      "step": 90
    },
    {
      "epoch": 0.03872752420470263,
      "grad_norm": 0.32559552788734436,
      "learning_rate": 9.598895847010198e-05,
      "loss": 1.4047,
      "step": 91
    },
    {
      "epoch": 0.039153101393765294,
      "grad_norm": 0.3118344843387604,
      "learning_rate": 9.589069672374515e-05,
      "loss": 1.4316,
      "step": 92
    },
    {
      "epoch": 0.03957867858282796,
      "grad_norm": 0.33870643377304077,
      "learning_rate": 9.579129744235408e-05,
      "loss": 1.4269,
      "step": 93
    },
    {
      "epoch": 0.04000425577189062,
      "grad_norm": 0.3164644241333008,
      "learning_rate": 9.569076308983045e-05,
      "loss": 1.4331,
      "step": 94
    },
    {
      "epoch": 0.040429832960953295,
      "grad_norm": 0.3030961751937866,
      "learning_rate": 9.5589096158212e-05,
      "loss": 1.3635,
      "step": 95
    },
    {
      "epoch": 0.04085541015001596,
      "grad_norm": 0.3118191063404083,
      "learning_rate": 9.548629916761076e-05,
      "loss": 1.3997,
      "step": 96
    },
    {
      "epoch": 0.041280987339078624,
      "grad_norm": 0.2993659973144531,
      "learning_rate": 9.538237466615058e-05,
      "loss": 1.3914,
      "step": 97
    },
    {
      "epoch": 0.04170656452814129,
      "grad_norm": 0.3075704872608185,
      "learning_rate": 9.527732522990391e-05,
      "loss": 1.413,
      "step": 98
    },
    {
      "epoch": 0.04213214171720396,
      "grad_norm": 0.3316209316253662,
      "learning_rate": 9.517115346282807e-05,
      "loss": 1.4021,
      "step": 99
    },
    {
      "epoch": 0.042557718906266626,
      "grad_norm": 0.2849251627922058,
      "learning_rate": 9.50638619967006e-05,
      "loss": 1.3836,
      "step": 100
    },
    {
      "epoch": 0.04298329609532929,
      "grad_norm": 0.3636033535003662,
      "learning_rate": 9.495545349105401e-05,
      "loss": 1.3604,
      "step": 101
    },
    {
      "epoch": 0.043408873284391955,
      "grad_norm": 0.3679890036582947,
      "learning_rate": 9.484593063310998e-05,
      "loss": 1.4271,
      "step": 102
    },
    {
      "epoch": 0.04383445047345462,
      "grad_norm": 0.36696523427963257,
      "learning_rate": 9.47352961377126e-05,
      "loss": 1.4735,
      "step": 103
    },
    {
      "epoch": 0.04426002766251729,
      "grad_norm": 0.33296412229537964,
      "learning_rate": 9.462355274726116e-05,
      "loss": 1.3739,
      "step": 104
    },
    {
      "epoch": 0.044685604851579956,
      "grad_norm": 0.3318886160850525,
      "learning_rate": 9.451070323164218e-05,
      "loss": 1.4887,
      "step": 105
    },
    {
      "epoch": 0.04511118204064262,
      "grad_norm": 0.3170441687107086,
      "learning_rate": 9.439675038816072e-05,
      "loss": 1.4224,
      "step": 106
    },
    {
      "epoch": 0.045536759229705286,
      "grad_norm": 0.29584380984306335,
      "learning_rate": 9.428169704147101e-05,
      "loss": 1.3517,
      "step": 107
    },
    {
      "epoch": 0.04596233641876796,
      "grad_norm": 0.36917728185653687,
      "learning_rate": 9.416554604350649e-05,
      "loss": 1.4098,
      "step": 108
    },
    {
      "epoch": 0.04638791360783062,
      "grad_norm": 0.3215009868144989,
      "learning_rate": 9.404830027340912e-05,
      "loss": 1.2738,
      "step": 109
    },
    {
      "epoch": 0.04681349079689329,
      "grad_norm": 0.3498933017253876,
      "learning_rate": 9.392996263745795e-05,
      "loss": 1.3365,
      "step": 110
    },
    {
      "epoch": 0.04723906798595595,
      "grad_norm": 0.30752959847450256,
      "learning_rate": 9.381053606899713e-05,
      "loss": 1.4227,
      "step": 111
    },
    {
      "epoch": 0.047664645175018616,
      "grad_norm": 0.304813951253891,
      "learning_rate": 9.36900235283632e-05,
      "loss": 1.3317,
      "step": 112
    },
    {
      "epoch": 0.04809022236408129,
      "grad_norm": 0.323702871799469,
      "learning_rate": 9.356842800281164e-05,
      "loss": 1.4016,
      "step": 113
    },
    {
      "epoch": 0.04851579955314395,
      "grad_norm": 0.31091248989105225,
      "learning_rate": 9.344575250644297e-05,
      "loss": 1.4067,
      "step": 114
    },
    {
      "epoch": 0.04894137674220662,
      "grad_norm": 0.37255966663360596,
      "learning_rate": 9.332200008012784e-05,
      "loss": 1.4713,
      "step": 115
    },
    {
      "epoch": 0.04936695393126928,
      "grad_norm": 0.3643665015697479,
      "learning_rate": 9.31971737914318e-05,
      "loss": 1.4232,
      "step": 116
    },
    {
      "epoch": 0.04979253112033195,
      "grad_norm": 0.3126155138015747,
      "learning_rate": 9.307127673453927e-05,
      "loss": 1.3248,
      "step": 117
    },
    {
      "epoch": 0.05021810830939462,
      "grad_norm": 0.3057401478290558,
      "learning_rate": 9.29443120301767e-05,
      "loss": 1.4206,
      "step": 118
    },
    {
      "epoch": 0.05064368549845728,
      "grad_norm": 0.3248518109321594,
      "learning_rate": 9.281628282553536e-05,
      "loss": 1.3366,
      "step": 119
    },
    {
      "epoch": 0.05106926268751995,
      "grad_norm": 0.3369111716747284,
      "learning_rate": 9.268719229419325e-05,
      "loss": 1.41,
      "step": 120
    },
    {
      "epoch": 0.05149483987658261,
      "grad_norm": 0.32806459069252014,
      "learning_rate": 9.255704363603645e-05,
      "loss": 1.3577,
      "step": 121
    },
    {
      "epoch": 0.051920417065645284,
      "grad_norm": 0.3423938751220703,
      "learning_rate": 9.242584007717983e-05,
      "loss": 1.4803,
      "step": 122
    },
    {
      "epoch": 0.05234599425470795,
      "grad_norm": 0.2834835350513458,
      "learning_rate": 9.229358486988702e-05,
      "loss": 1.3583,
      "step": 123
    },
    {
      "epoch": 0.052771571443770614,
      "grad_norm": 0.3676677644252777,
      "learning_rate": 9.216028129248986e-05,
      "loss": 1.5176,
      "step": 124
    },
    {
      "epoch": 0.05319714863283328,
      "grad_norm": 0.32076701521873474,
      "learning_rate": 9.202593264930708e-05,
      "loss": 1.3902,
      "step": 125
    },
    {
      "epoch": 0.05362272582189594,
      "grad_norm": 0.31246402859687805,
      "learning_rate": 9.189054227056247e-05,
      "loss": 1.3882,
      "step": 126
    },
    {
      "epoch": 0.054048303010958615,
      "grad_norm": 0.3379463851451874,
      "learning_rate": 9.175411351230222e-05,
      "loss": 1.268,
      "step": 127
    },
    {
      "epoch": 0.05447388020002128,
      "grad_norm": 0.30225270986557007,
      "learning_rate": 9.161664975631184e-05,
      "loss": 1.4595,
      "step": 128
    },
    {
      "epoch": 0.054899457389083944,
      "grad_norm": 0.3265242874622345,
      "learning_rate": 9.147815441003223e-05,
      "loss": 1.3656,
      "step": 129
    },
    {
      "epoch": 0.05532503457814661,
      "grad_norm": 0.3495875597000122,
      "learning_rate": 9.133863090647532e-05,
      "loss": 1.4378,
      "step": 130
    },
    {
      "epoch": 0.05575061176720928,
      "grad_norm": 0.3625856935977936,
      "learning_rate": 9.119808270413891e-05,
      "loss": 1.4344,
      "step": 131
    },
    {
      "epoch": 0.056176188956271945,
      "grad_norm": 0.304872065782547,
      "learning_rate": 9.105651328692093e-05,
      "loss": 1.3909,
      "step": 132
    },
    {
      "epoch": 0.05660176614533461,
      "grad_norm": 0.3018715977668762,
      "learning_rate": 9.091392616403314e-05,
      "loss": 1.3361,
      "step": 133
    },
    {
      "epoch": 0.057027343334397275,
      "grad_norm": 0.27433308959007263,
      "learning_rate": 9.077032486991408e-05,
      "loss": 1.226,
      "step": 134
    },
    {
      "epoch": 0.05745292052345994,
      "grad_norm": 0.3812541663646698,
      "learning_rate": 9.062571296414154e-05,
      "loss": 1.3638,
      "step": 135
    },
    {
      "epoch": 0.05787849771252261,
      "grad_norm": 0.33055293560028076,
      "learning_rate": 9.048009403134417e-05,
      "loss": 1.4088,
      "step": 136
    },
    {
      "epoch": 0.058304074901585276,
      "grad_norm": 0.325934499502182,
      "learning_rate": 9.033347168111282e-05,
      "loss": 1.3748,
      "step": 137
    },
    {
      "epoch": 0.05872965209064794,
      "grad_norm": 0.327552855014801,
      "learning_rate": 9.018584954791096e-05,
      "loss": 1.4807,
      "step": 138
    },
    {
      "epoch": 0.059155229279710606,
      "grad_norm": 0.3454499840736389,
      "learning_rate": 9.003723129098458e-05,
      "loss": 1.4543,
      "step": 139
    },
    {
      "epoch": 0.05958080646877328,
      "grad_norm": 0.312281996011734,
      "learning_rate": 8.98876205942715e-05,
      "loss": 1.4524,
      "step": 140
    },
    {
      "epoch": 0.06000638365783594,
      "grad_norm": 0.31058740615844727,
      "learning_rate": 8.97370211663101e-05,
      "loss": 1.4508,
      "step": 141
    },
    {
      "epoch": 0.06043196084689861,
      "grad_norm": 0.3037337362766266,
      "learning_rate": 8.958543674014732e-05,
      "loss": 1.2849,
      "step": 142
    },
    {
      "epoch": 0.06085753803596127,
      "grad_norm": 0.3011512756347656,
      "learning_rate": 8.943287107324617e-05,
      "loss": 1.3596,
      "step": 143
    },
    {
      "epoch": 0.061283115225023936,
      "grad_norm": 0.3848366141319275,
      "learning_rate": 8.927932794739257e-05,
      "loss": 1.3857,
      "step": 144
    },
    {
      "epoch": 0.06170869241408661,
      "grad_norm": 0.31173983216285706,
      "learning_rate": 8.912481116860166e-05,
      "loss": 1.4048,
      "step": 145
    },
    {
      "epoch": 0.06213426960314927,
      "grad_norm": 0.3007739782333374,
      "learning_rate": 8.896932456702332e-05,
      "loss": 1.2664,
      "step": 146
    },
    {
      "epoch": 0.06255984679221194,
      "grad_norm": 0.2995300889015198,
      "learning_rate": 8.881287199684743e-05,
      "loss": 1.3843,
      "step": 147
    },
    {
      "epoch": 0.0629854239812746,
      "grad_norm": 0.31542861461639404,
      "learning_rate": 8.865545733620815e-05,
      "loss": 1.472,
      "step": 148
    },
    {
      "epoch": 0.06341100117033727,
      "grad_norm": 0.32314345240592957,
      "learning_rate": 8.84970844870879e-05,
      "loss": 1.4709,
      "step": 149
    },
    {
      "epoch": 0.06383657835939993,
      "grad_norm": 0.30198317766189575,
      "learning_rate": 8.83377573752206e-05,
      "loss": 1.4225,
      "step": 150
    },
    {
      "epoch": 0.0642621555484626,
      "grad_norm": 0.29066887497901917,
      "learning_rate": 8.817747994999432e-05,
      "loss": 1.3723,
      "step": 151
    },
    {
      "epoch": 0.06468773273752527,
      "grad_norm": 0.3179444372653961,
      "learning_rate": 8.801625618435351e-05,
      "loss": 1.4489,
      "step": 152
    },
    {
      "epoch": 0.06511330992658794,
      "grad_norm": 0.3465145230293274,
      "learning_rate": 8.785409007470032e-05,
      "loss": 1.3848,
      "step": 153
    },
    {
      "epoch": 0.0655388871156506,
      "grad_norm": 0.27494004368782043,
      "learning_rate": 8.769098564079574e-05,
      "loss": 1.3309,
      "step": 154
    },
    {
      "epoch": 0.06596446430471327,
      "grad_norm": 0.32629844546318054,
      "learning_rate": 8.752694692565986e-05,
      "loss": 1.3737,
      "step": 155
    },
    {
      "epoch": 0.06639004149377593,
      "grad_norm": 0.3132442533969879,
      "learning_rate": 8.736197799547159e-05,
      "loss": 1.4035,
      "step": 156
    },
    {
      "epoch": 0.0668156186828386,
      "grad_norm": 0.31382986903190613,
      "learning_rate": 8.719608293946802e-05,
      "loss": 1.3258,
      "step": 157
    },
    {
      "epoch": 0.06724119587190126,
      "grad_norm": 0.32756349444389343,
      "learning_rate": 8.702926586984294e-05,
      "loss": 1.3506,
      "step": 158
    },
    {
      "epoch": 0.06766677306096393,
      "grad_norm": 0.33079269528388977,
      "learning_rate": 8.686153092164493e-05,
      "loss": 1.4653,
      "step": 159
    },
    {
      "epoch": 0.06809235025002659,
      "grad_norm": 0.30706673860549927,
      "learning_rate": 8.669288225267492e-05,
      "loss": 1.3536,
      "step": 160
    },
    {
      "epoch": 0.06851792743908927,
      "grad_norm": 0.31477653980255127,
      "learning_rate": 8.6523324043383e-05,
      "loss": 1.3129,
      "step": 161
    },
    {
      "epoch": 0.06894350462815194,
      "grad_norm": 0.32898736000061035,
      "learning_rate": 8.635286049676496e-05,
      "loss": 1.3889,
      "step": 162
    },
    {
      "epoch": 0.0693690818172146,
      "grad_norm": 0.35487836599349976,
      "learning_rate": 8.618149583825796e-05,
      "loss": 1.4337,
      "step": 163
    },
    {
      "epoch": 0.06979465900627727,
      "grad_norm": 0.3246158957481384,
      "learning_rate": 8.60092343156359e-05,
      "loss": 1.4514,
      "step": 164
    },
    {
      "epoch": 0.07022023619533993,
      "grad_norm": 0.34181657433509827,
      "learning_rate": 8.583608019890406e-05,
      "loss": 1.429,
      "step": 165
    },
    {
      "epoch": 0.0706458133844026,
      "grad_norm": 0.3031291961669922,
      "learning_rate": 8.566203778019322e-05,
      "loss": 1.3333,
      "step": 166
    },
    {
      "epoch": 0.07107139057346526,
      "grad_norm": 0.32903921604156494,
      "learning_rate": 8.54871113736534e-05,
      "loss": 1.5184,
      "step": 167
    },
    {
      "epoch": 0.07149696776252792,
      "grad_norm": 0.31410080194473267,
      "learning_rate": 8.531130531534683e-05,
      "loss": 1.581,
      "step": 168
    },
    {
      "epoch": 0.07192254495159059,
      "grad_norm": 0.344510555267334,
      "learning_rate": 8.513462396314041e-05,
      "loss": 1.4454,
      "step": 169
    },
    {
      "epoch": 0.07234812214065327,
      "grad_norm": 0.3599744141101837,
      "learning_rate": 8.495707169659786e-05,
      "loss": 1.4448,
      "step": 170
    },
    {
      "epoch": 0.07277369932971593,
      "grad_norm": 0.3286546468734741,
      "learning_rate": 8.477865291687095e-05,
      "loss": 1.3882,
      "step": 171
    },
    {
      "epoch": 0.0731992765187786,
      "grad_norm": 0.30131304264068604,
      "learning_rate": 8.459937204659063e-05,
      "loss": 1.3306,
      "step": 172
    },
    {
      "epoch": 0.07362485370784126,
      "grad_norm": 0.32244452834129333,
      "learning_rate": 8.441923352975716e-05,
      "loss": 1.4304,
      "step": 173
    },
    {
      "epoch": 0.07405043089690393,
      "grad_norm": 0.32980459928512573,
      "learning_rate": 8.423824183163016e-05,
      "loss": 1.4298,
      "step": 174
    },
    {
      "epoch": 0.07447600808596659,
      "grad_norm": 0.2912943661212921,
      "learning_rate": 8.405640143861782e-05,
      "loss": 1.4104,
      "step": 175
    },
    {
      "epoch": 0.07490158527502926,
      "grad_norm": 0.31498202681541443,
      "learning_rate": 8.387371685816572e-05,
      "loss": 1.4688,
      "step": 176
    },
    {
      "epoch": 0.07532716246409192,
      "grad_norm": 0.31826215982437134,
      "learning_rate": 8.369019261864505e-05,
      "loss": 1.3991,
      "step": 177
    },
    {
      "epoch": 0.07575273965315459,
      "grad_norm": 0.3330978453159332,
      "learning_rate": 8.350583326924048e-05,
      "loss": 1.4436,
      "step": 178
    },
    {
      "epoch": 0.07617831684221726,
      "grad_norm": 0.31430912017822266,
      "learning_rate": 8.332064337983725e-05,
      "loss": 1.4602,
      "step": 179
    },
    {
      "epoch": 0.07660389403127993,
      "grad_norm": 0.3026297986507416,
      "learning_rate": 8.3134627540908e-05,
      "loss": 1.3956,
      "step": 180
    },
    {
      "epoch": 0.0770294712203426,
      "grad_norm": 0.33081120252609253,
      "learning_rate": 8.294779036339893e-05,
      "loss": 1.516,
      "step": 181
    },
    {
      "epoch": 0.07745504840940526,
      "grad_norm": 0.299742728471756,
      "learning_rate": 8.27601364786155e-05,
      "loss": 1.3524,
      "step": 182
    },
    {
      "epoch": 0.07788062559846792,
      "grad_norm": 0.3342128396034241,
      "learning_rate": 8.257167053810768e-05,
      "loss": 1.4861,
      "step": 183
    },
    {
      "epoch": 0.07830620278753059,
      "grad_norm": 0.336272269487381,
      "learning_rate": 8.238239721355461e-05,
      "loss": 1.3751,
      "step": 184
    },
    {
      "epoch": 0.07873177997659325,
      "grad_norm": 0.3233320116996765,
      "learning_rate": 8.219232119664877e-05,
      "loss": 1.3365,
      "step": 185
    },
    {
      "epoch": 0.07915735716565592,
      "grad_norm": 0.31845808029174805,
      "learning_rate": 8.200144719897974e-05,
      "loss": 1.416,
      "step": 186
    },
    {
      "epoch": 0.07958293435471858,
      "grad_norm": 0.3381212055683136,
      "learning_rate": 8.180977995191738e-05,
      "loss": 1.3525,
      "step": 187
    },
    {
      "epoch": 0.08000851154378125,
      "grad_norm": 0.30561843514442444,
      "learning_rate": 8.161732420649459e-05,
      "loss": 1.3485,
      "step": 188
    },
    {
      "epoch": 0.08043408873284393,
      "grad_norm": 0.29561662673950195,
      "learning_rate": 8.142408473328945e-05,
      "loss": 1.2745,
      "step": 189
    },
    {
      "epoch": 0.08085966592190659,
      "grad_norm": 0.2992284893989563,
      "learning_rate": 8.123006632230703e-05,
      "loss": 1.3667,
      "step": 190
    },
    {
      "epoch": 0.08128524311096925,
      "grad_norm": 0.3103450834751129,
      "learning_rate": 8.103527378286071e-05,
      "loss": 1.3894,
      "step": 191
    },
    {
      "epoch": 0.08171082030003192,
      "grad_norm": 0.31997600197792053,
      "learning_rate": 8.083971194345282e-05,
      "loss": 1.4296,
      "step": 192
    },
    {
      "epoch": 0.08213639748909458,
      "grad_norm": 0.2979292571544647,
      "learning_rate": 8.06433856516551e-05,
      "loss": 1.4681,
      "step": 193
    },
    {
      "epoch": 0.08256197467815725,
      "grad_norm": 0.35576626658439636,
      "learning_rate": 8.044629977398845e-05,
      "loss": 1.4637,
      "step": 194
    },
    {
      "epoch": 0.08298755186721991,
      "grad_norm": 0.3102336823940277,
      "learning_rate": 8.024845919580235e-05,
      "loss": 1.3598,
      "step": 195
    },
    {
      "epoch": 0.08341312905628258,
      "grad_norm": 0.3177992105484009,
      "learning_rate": 8.004986882115371e-05,
      "loss": 1.3457,
      "step": 196
    },
    {
      "epoch": 0.08383870624534524,
      "grad_norm": 0.32051852345466614,
      "learning_rate": 7.985053357268533e-05,
      "loss": 1.3453,
      "step": 197
    },
    {
      "epoch": 0.08426428343440792,
      "grad_norm": 0.35972169041633606,
      "learning_rate": 7.965045839150394e-05,
      "loss": 1.4343,
      "step": 198
    },
    {
      "epoch": 0.08468986062347059,
      "grad_norm": 0.3172051012516022,
      "learning_rate": 7.94496482370576e-05,
      "loss": 1.2598,
      "step": 199
    },
    {
      "epoch": 0.08511543781253325,
      "grad_norm": 0.315286785364151,
      "learning_rate": 7.924810808701286e-05,
      "loss": 1.2469,
      "step": 200
    },
    {
      "epoch": 0.08554101500159592,
      "grad_norm": 0.34473949670791626,
      "learning_rate": 7.904584293713134e-05,
      "loss": 1.376,
      "step": 201
    },
    {
      "epoch": 0.08596659219065858,
      "grad_norm": 0.3357507884502411,
      "learning_rate": 7.884285780114593e-05,
      "loss": 1.3202,
      "step": 202
    },
    {
      "epoch": 0.08639216937972125,
      "grad_norm": 0.3239094018936157,
      "learning_rate": 7.86391577106364e-05,
      "loss": 1.3751,
      "step": 203
    },
    {
      "epoch": 0.08681774656878391,
      "grad_norm": 0.32212114334106445,
      "learning_rate": 7.843474771490486e-05,
      "loss": 1.3618,
      "step": 204
    },
    {
      "epoch": 0.08724332375784657,
      "grad_norm": 0.3353959321975708,
      "learning_rate": 7.82296328808504e-05,
      "loss": 1.3613,
      "step": 205
    },
    {
      "epoch": 0.08766890094690924,
      "grad_norm": 0.3211444318294525,
      "learning_rate": 7.802381829284366e-05,
      "loss": 1.3828,
      "step": 206
    },
    {
      "epoch": 0.08809447813597192,
      "grad_norm": 0.3303898572921753,
      "learning_rate": 7.78173090526007e-05,
      "loss": 1.4529,
      "step": 207
    },
    {
      "epoch": 0.08852005532503458,
      "grad_norm": 0.3457483947277069,
      "learning_rate": 7.761011027905654e-05,
      "loss": 1.3714,
      "step": 208
    },
    {
      "epoch": 0.08894563251409725,
      "grad_norm": 0.3243163228034973,
      "learning_rate": 7.740222710823837e-05,
      "loss": 1.3595,
      "step": 209
    },
    {
      "epoch": 0.08937120970315991,
      "grad_norm": 0.322988361120224,
      "learning_rate": 7.719366469313806e-05,
      "loss": 1.3353,
      "step": 210
    },
    {
      "epoch": 0.08979678689222258,
      "grad_norm": 0.31282839179039,
      "learning_rate": 7.698442820358463e-05,
      "loss": 1.4834,
      "step": 211
    },
    {
      "epoch": 0.09022236408128524,
      "grad_norm": 0.318215548992157,
      "learning_rate": 7.677452282611594e-05,
      "loss": 1.3521,
      "step": 212
    },
    {
      "epoch": 0.0906479412703479,
      "grad_norm": 0.3105640113353729,
      "learning_rate": 7.656395376385027e-05,
      "loss": 1.4099,
      "step": 213
    },
    {
      "epoch": 0.09107351845941057,
      "grad_norm": 0.3511587083339691,
      "learning_rate": 7.635272623635716e-05,
      "loss": 1.3177,
      "step": 214
    },
    {
      "epoch": 0.09149909564847324,
      "grad_norm": 0.3450727164745331,
      "learning_rate": 7.614084547952824e-05,
      "loss": 1.5545,
      "step": 215
    },
    {
      "epoch": 0.09192467283753591,
      "grad_norm": 0.33173757791519165,
      "learning_rate": 7.592831674544728e-05,
      "loss": 1.4891,
      "step": 216
    },
    {
      "epoch": 0.09235025002659858,
      "grad_norm": 0.29296064376831055,
      "learning_rate": 7.571514530226004e-05,
      "loss": 1.3501,
      "step": 217
    },
    {
      "epoch": 0.09277582721566124,
      "grad_norm": 0.3370741605758667,
      "learning_rate": 7.550133643404377e-05,
      "loss": 1.4028,
      "step": 218
    },
    {
      "epoch": 0.09320140440472391,
      "grad_norm": 0.31852227449417114,
      "learning_rate": 7.528689544067611e-05,
      "loss": 1.341,
      "step": 219
    },
    {
      "epoch": 0.09362698159378657,
      "grad_norm": 0.3008808493614197,
      "learning_rate": 7.507182763770381e-05,
      "loss": 1.3032,
      "step": 220
    },
    {
      "epoch": 0.09405255878284924,
      "grad_norm": 0.3290260434150696,
      "learning_rate": 7.485613835621088e-05,
      "loss": 1.3937,
      "step": 221
    },
    {
      "epoch": 0.0944781359719119,
      "grad_norm": 0.3093232810497284,
      "learning_rate": 7.46398329426865e-05,
      "loss": 1.4153,
      "step": 222
    },
    {
      "epoch": 0.09490371316097457,
      "grad_norm": 0.35814496874809265,
      "learning_rate": 7.442291675889254e-05,
      "loss": 1.4717,
      "step": 223
    },
    {
      "epoch": 0.09532929035003723,
      "grad_norm": 0.33721545338630676,
      "learning_rate": 7.420539518173053e-05,
      "loss": 1.4509,
      "step": 224
    },
    {
      "epoch": 0.09575486753909991,
      "grad_norm": 0.3308022916316986,
      "learning_rate": 7.398727360310848e-05,
      "loss": 1.4284,
      "step": 225
    },
    {
      "epoch": 0.09618044472816258,
      "grad_norm": 0.29898306727409363,
      "learning_rate": 7.376855742980718e-05,
      "loss": 1.3514,
      "step": 226
    },
    {
      "epoch": 0.09660602191722524,
      "grad_norm": 0.2964514493942261,
      "learning_rate": 7.354925208334614e-05,
      "loss": 1.3246,
      "step": 227
    },
    {
      "epoch": 0.0970315991062879,
      "grad_norm": 0.2953067421913147,
      "learning_rate": 7.332936299984937e-05,
      "loss": 1.4034,
      "step": 228
    },
    {
      "epoch": 0.09745717629535057,
      "grad_norm": 0.2890208661556244,
      "learning_rate": 7.310889562991037e-05,
      "loss": 1.2483,
      "step": 229
    },
    {
      "epoch": 0.09788275348441323,
      "grad_norm": 0.31622129678726196,
      "learning_rate": 7.288785543845725e-05,
      "loss": 1.4184,
      "step": 230
    },
    {
      "epoch": 0.0983083306734759,
      "grad_norm": 0.3258267641067505,
      "learning_rate": 7.266624790461713e-05,
      "loss": 1.4013,
      "step": 231
    },
    {
      "epoch": 0.09873390786253856,
      "grad_norm": 0.3144655227661133,
      "learning_rate": 7.244407852158042e-05,
      "loss": 1.4689,
      "step": 232
    },
    {
      "epoch": 0.09915948505160123,
      "grad_norm": 0.30230382084846497,
      "learning_rate": 7.222135279646453e-05,
      "loss": 1.4526,
      "step": 233
    },
    {
      "epoch": 0.0995850622406639,
      "grad_norm": 0.3089566230773926,
      "learning_rate": 7.199807625017749e-05,
      "loss": 1.315,
      "step": 234
    },
    {
      "epoch": 0.10001063942972657,
      "grad_norm": 0.30076929926872253,
      "learning_rate": 7.177425441728103e-05,
      "loss": 1.3653,
      "step": 235
    },
    {
      "epoch": 0.10043621661878924,
      "grad_norm": 0.3337990641593933,
      "learning_rate": 7.154989284585342e-05,
      "loss": 1.4545,
      "step": 236
    },
    {
      "epoch": 0.1008617938078519,
      "grad_norm": 0.33799222111701965,
      "learning_rate": 7.132499709735187e-05,
      "loss": 1.3691,
      "step": 237
    },
    {
      "epoch": 0.10128737099691457,
      "grad_norm": 0.33428463339805603,
      "learning_rate": 7.109957274647478e-05,
      "loss": 1.5371,
      "step": 238
    },
    {
      "epoch": 0.10171294818597723,
      "grad_norm": 0.32650595903396606,
      "learning_rate": 7.08736253810235e-05,
      "loss": 1.5122,
      "step": 239
    },
    {
      "epoch": 0.1021385253750399,
      "grad_norm": 0.3586381673812866,
      "learning_rate": 7.06471606017638e-05,
      "loss": 1.5361,
      "step": 240
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 0.3396207392215729,
      "learning_rate": 7.04201840222871e-05,
      "loss": 1.4595,
      "step": 241
    },
    {
      "epoch": 0.10298967975316523,
      "grad_norm": 0.3792440593242645,
      "learning_rate": 7.019270126887123e-05,
      "loss": 1.31,
      "step": 242
    },
    {
      "epoch": 0.10341525694222789,
      "grad_norm": 0.3366132378578186,
      "learning_rate": 6.996471798034108e-05,
      "loss": 1.4679,
      "step": 243
    },
    {
      "epoch": 0.10384083413129057,
      "grad_norm": 0.282144159078598,
      "learning_rate": 6.973623980792875e-05,
      "loss": 1.2485,
      "step": 244
    },
    {
      "epoch": 0.10426641132035323,
      "grad_norm": 0.3441198766231537,
      "learning_rate": 6.950727241513344e-05,
      "loss": 1.378,
      "step": 245
    },
    {
      "epoch": 0.1046919885094159,
      "grad_norm": 0.298406720161438,
      "learning_rate": 6.927782147758117e-05,
      "loss": 1.37,
      "step": 246
    },
    {
      "epoch": 0.10511756569847856,
      "grad_norm": 0.3125241696834564,
      "learning_rate": 6.904789268288398e-05,
      "loss": 1.3247,
      "step": 247
    },
    {
      "epoch": 0.10554314288754123,
      "grad_norm": 0.31313666701316833,
      "learning_rate": 6.881749173049901e-05,
      "loss": 1.3459,
      "step": 248
    },
    {
      "epoch": 0.10596872007660389,
      "grad_norm": 0.31229567527770996,
      "learning_rate": 6.858662433158724e-05,
      "loss": 1.4175,
      "step": 249
    },
    {
      "epoch": 0.10639429726566656,
      "grad_norm": 0.3220798671245575,
      "learning_rate": 6.835529620887185e-05,
      "loss": 1.3616,
      "step": 250
    },
    {
      "epoch": 0.10681987445472922,
      "grad_norm": 0.30676543712615967,
      "learning_rate": 6.81235130964964e-05,
      "loss": 1.4731,
      "step": 251
    },
    {
      "epoch": 0.10724545164379189,
      "grad_norm": 0.3131166994571686,
      "learning_rate": 6.789128073988276e-05,
      "loss": 1.3409,
      "step": 252
    },
    {
      "epoch": 0.10767102883285457,
      "grad_norm": 0.311882346868515,
      "learning_rate": 6.765860489558856e-05,
      "loss": 1.3375,
      "step": 253
    },
    {
      "epoch": 0.10809660602191723,
      "grad_norm": 0.3434353470802307,
      "learning_rate": 6.74254913311646e-05,
      "loss": 1.3905,
      "step": 254
    },
    {
      "epoch": 0.1085221832109799,
      "grad_norm": 0.2908113896846771,
      "learning_rate": 6.719194582501183e-05,
      "loss": 1.4806,
      "step": 255
    },
    {
      "epoch": 0.10894776040004256,
      "grad_norm": 0.3273056149482727,
      "learning_rate": 6.695797416623821e-05,
      "loss": 1.4091,
      "step": 256
    },
    {
      "epoch": 0.10937333758910522,
      "grad_norm": 0.32346075773239136,
      "learning_rate": 6.672358215451507e-05,
      "loss": 1.4084,
      "step": 257
    },
    {
      "epoch": 0.10979891477816789,
      "grad_norm": 0.3289119601249695,
      "learning_rate": 6.648877559993339e-05,
      "loss": 1.3466,
      "step": 258
    },
    {
      "epoch": 0.11022449196723055,
      "grad_norm": 0.2760949432849884,
      "learning_rate": 6.62535603228599e-05,
      "loss": 1.3571,
      "step": 259
    },
    {
      "epoch": 0.11065006915629322,
      "grad_norm": 0.29785817861557007,
      "learning_rate": 6.601794215379266e-05,
      "loss": 1.3862,
      "step": 260
    },
    {
      "epoch": 0.11107564634535588,
      "grad_norm": 0.31330984830856323,
      "learning_rate": 6.578192693321656e-05,
      "loss": 1.3972,
      "step": 261
    },
    {
      "epoch": 0.11150122353441856,
      "grad_norm": 0.2923828363418579,
      "learning_rate": 6.55455205114586e-05,
      "loss": 1.3496,
      "step": 262
    },
    {
      "epoch": 0.11192680072348123,
      "grad_norm": 0.32257649302482605,
      "learning_rate": 6.530872874854285e-05,
      "loss": 1.3539,
      "step": 263
    },
    {
      "epoch": 0.11235237791254389,
      "grad_norm": 0.32502755522727966,
      "learning_rate": 6.507155751404518e-05,
      "loss": 1.4341,
      "step": 264
    },
    {
      "epoch": 0.11277795510160656,
      "grad_norm": 0.3588520586490631,
      "learning_rate": 6.483401268694777e-05,
      "loss": 1.4789,
      "step": 265
    },
    {
      "epoch": 0.11320353229066922,
      "grad_norm": 0.3285142183303833,
      "learning_rate": 6.45961001554934e-05,
      "loss": 1.3466,
      "step": 266
    },
    {
      "epoch": 0.11362910947973189,
      "grad_norm": 0.3209497928619385,
      "learning_rate": 6.435782581703945e-05,
      "loss": 1.4529,
      "step": 267
    },
    {
      "epoch": 0.11405468666879455,
      "grad_norm": 0.32251620292663574,
      "learning_rate": 6.411919557791176e-05,
      "loss": 1.3964,
      "step": 268
    },
    {
      "epoch": 0.11448026385785721,
      "grad_norm": 0.27929726243019104,
      "learning_rate": 6.388021535325821e-05,
      "loss": 1.3372,
      "step": 269
    },
    {
      "epoch": 0.11490584104691988,
      "grad_norm": 0.35057204961776733,
      "learning_rate": 6.364089106690209e-05,
      "loss": 1.4241,
      "step": 270
    },
    {
      "epoch": 0.11533141823598256,
      "grad_norm": 0.32568198442459106,
      "learning_rate": 6.340122865119524e-05,
      "loss": 1.2808,
      "step": 271
    },
    {
      "epoch": 0.11575699542504522,
      "grad_norm": 0.3379870057106018,
      "learning_rate": 6.316123404687108e-05,
      "loss": 1.4953,
      "step": 272
    },
    {
      "epoch": 0.11618257261410789,
      "grad_norm": 0.297118604183197,
      "learning_rate": 6.292091320289725e-05,
      "loss": 1.327,
      "step": 273
    },
    {
      "epoch": 0.11660814980317055,
      "grad_norm": 0.30609849095344543,
      "learning_rate": 6.268027207632821e-05,
      "loss": 1.2376,
      "step": 274
    },
    {
      "epoch": 0.11703372699223322,
      "grad_norm": 0.3672487735748291,
      "learning_rate": 6.243931663215756e-05,
      "loss": 1.4375,
      "step": 275
    },
    {
      "epoch": 0.11745930418129588,
      "grad_norm": 0.3317699134349823,
      "learning_rate": 6.219805284317019e-05,
      "loss": 1.4371,
      "step": 276
    },
    {
      "epoch": 0.11788488137035855,
      "grad_norm": 0.285065233707428,
      "learning_rate": 6.195648668979417e-05,
      "loss": 1.3968,
      "step": 277
    },
    {
      "epoch": 0.11831045855942121,
      "grad_norm": 0.33966484665870667,
      "learning_rate": 6.171462415995263e-05,
      "loss": 1.4376,
      "step": 278
    },
    {
      "epoch": 0.11873603574848388,
      "grad_norm": 0.32605886459350586,
      "learning_rate": 6.147247124891519e-05,
      "loss": 1.4235,
      "step": 279
    },
    {
      "epoch": 0.11916161293754655,
      "grad_norm": 0.29964199662208557,
      "learning_rate": 6.123003395914945e-05,
      "loss": 1.3581,
      "step": 280
    },
    {
      "epoch": 0.11958719012660922,
      "grad_norm": 0.3049403429031372,
      "learning_rate": 6.098731830017217e-05,
      "loss": 1.398,
      "step": 281
    },
    {
      "epoch": 0.12001276731567188,
      "grad_norm": 0.30261459946632385,
      "learning_rate": 6.074433028840029e-05,
      "loss": 1.3876,
      "step": 282
    },
    {
      "epoch": 0.12043834450473455,
      "grad_norm": 0.3096999526023865,
      "learning_rate": 6.0501075947001816e-05,
      "loss": 1.4192,
      "step": 283
    },
    {
      "epoch": 0.12086392169379721,
      "grad_norm": 0.3477337956428528,
      "learning_rate": 6.025756130574652e-05,
      "loss": 1.2923,
      "step": 284
    },
    {
      "epoch": 0.12128949888285988,
      "grad_norm": 0.2876332998275757,
      "learning_rate": 6.001379240085645e-05,
      "loss": 1.3937,
      "step": 285
    },
    {
      "epoch": 0.12171507607192254,
      "grad_norm": 0.31134235858917236,
      "learning_rate": 5.976977527485633e-05,
      "loss": 1.4417,
      "step": 286
    },
    {
      "epoch": 0.12214065326098521,
      "grad_norm": 0.33744943141937256,
      "learning_rate": 5.9525515976423775e-05,
      "loss": 1.4114,
      "step": 287
    },
    {
      "epoch": 0.12256623045004787,
      "grad_norm": 0.32942283153533936,
      "learning_rate": 5.928102056023935e-05,
      "loss": 1.4425,
      "step": 288
    },
    {
      "epoch": 0.12299180763911054,
      "grad_norm": 0.3516654074192047,
      "learning_rate": 5.903629508683649e-05,
      "loss": 1.3629,
      "step": 289
    },
    {
      "epoch": 0.12341738482817322,
      "grad_norm": 0.31652727723121643,
      "learning_rate": 5.879134562245123e-05,
      "loss": 1.4935,
      "step": 290
    },
    {
      "epoch": 0.12384296201723588,
      "grad_norm": 0.30340665578842163,
      "learning_rate": 5.854617823887196e-05,
      "loss": 1.3113,
      "step": 291
    },
    {
      "epoch": 0.12426853920629854,
      "grad_norm": 0.3198196589946747,
      "learning_rate": 5.8300799013288754e-05,
      "loss": 1.2864,
      "step": 292
    },
    {
      "epoch": 0.12469411639536121,
      "grad_norm": 0.28304675221443176,
      "learning_rate": 5.8055214028142844e-05,
      "loss": 1.3676,
      "step": 293
    },
    {
      "epoch": 0.12511969358442387,
      "grad_norm": 0.2878504693508148,
      "learning_rate": 5.780942937097584e-05,
      "loss": 1.3713,
      "step": 294
    },
    {
      "epoch": 0.12554527077348654,
      "grad_norm": 0.3235960900783539,
      "learning_rate": 5.7563451134278735e-05,
      "loss": 1.4314,
      "step": 295
    },
    {
      "epoch": 0.1259708479625492,
      "grad_norm": 0.3305697739124298,
      "learning_rate": 5.731728541534101e-05,
      "loss": 1.4453,
      "step": 296
    },
    {
      "epoch": 0.12639642515161187,
      "grad_norm": 0.34397658705711365,
      "learning_rate": 5.7070938316099455e-05,
      "loss": 1.3833,
      "step": 297
    },
    {
      "epoch": 0.12682200234067453,
      "grad_norm": 0.31075742840766907,
      "learning_rate": 5.6824415942986844e-05,
      "loss": 1.3919,
      "step": 298
    },
    {
      "epoch": 0.1272475795297372,
      "grad_norm": 0.32299962639808655,
      "learning_rate": 5.65777244067807e-05,
      "loss": 1.3616,
      "step": 299
    },
    {
      "epoch": 0.12767315671879986,
      "grad_norm": 0.31307482719421387,
      "learning_rate": 5.633086982245166e-05,
      "loss": 1.315,
      "step": 300
    },
    {
      "epoch": 0.12809873390786253,
      "grad_norm": 0.31546834111213684,
      "learning_rate": 5.608385830901206e-05,
      "loss": 1.3429,
      "step": 301
    },
    {
      "epoch": 0.1285243110969252,
      "grad_norm": 0.32566624879837036,
      "learning_rate": 5.583669598936414e-05,
      "loss": 1.3693,
      "step": 302
    },
    {
      "epoch": 0.12894988828598788,
      "grad_norm": 0.3248218894004822,
      "learning_rate": 5.5589388990148326e-05,
      "loss": 1.337,
      "step": 303
    },
    {
      "epoch": 0.12937546547505055,
      "grad_norm": 0.29174041748046875,
      "learning_rate": 5.534194344159136e-05,
      "loss": 1.4242,
      "step": 304
    },
    {
      "epoch": 0.12980104266411321,
      "grad_norm": 0.3306146264076233,
      "learning_rate": 5.5094365477354325e-05,
      "loss": 1.3176,
      "step": 305
    },
    {
      "epoch": 0.13022661985317588,
      "grad_norm": 0.34080812335014343,
      "learning_rate": 5.48466612343806e-05,
      "loss": 1.3678,
      "step": 306
    },
    {
      "epoch": 0.13065219704223854,
      "grad_norm": 0.33598506450653076,
      "learning_rate": 5.4598836852743774e-05,
      "loss": 1.3354,
      "step": 307
    },
    {
      "epoch": 0.1310777742313012,
      "grad_norm": 0.3259298801422119,
      "learning_rate": 5.435089847549541e-05,
      "loss": 1.3574,
      "step": 308
    },
    {
      "epoch": 0.13150335142036387,
      "grad_norm": 0.32616904377937317,
      "learning_rate": 5.4102852248512814e-05,
      "loss": 1.4339,
      "step": 309
    },
    {
      "epoch": 0.13192892860942654,
      "grad_norm": 0.34405210614204407,
      "learning_rate": 5.38547043203466e-05,
      "loss": 1.4094,
      "step": 310
    },
    {
      "epoch": 0.1323545057984892,
      "grad_norm": 0.31888607144355774,
      "learning_rate": 5.3606460842068426e-05,
      "loss": 1.4321,
      "step": 311
    },
    {
      "epoch": 0.13278008298755187,
      "grad_norm": 0.3342142403125763,
      "learning_rate": 5.3358127967118354e-05,
      "loss": 1.4156,
      "step": 312
    },
    {
      "epoch": 0.13320566017661453,
      "grad_norm": 0.4066295921802521,
      "learning_rate": 5.3109711851152456e-05,
      "loss": 1.4043,
      "step": 313
    },
    {
      "epoch": 0.1336312373656772,
      "grad_norm": 0.3196505606174469,
      "learning_rate": 5.286121865189017e-05,
      "loss": 1.2663,
      "step": 314
    },
    {
      "epoch": 0.13405681455473986,
      "grad_norm": 0.30326151847839355,
      "learning_rate": 5.261265452896167e-05,
      "loss": 1.4344,
      "step": 315
    },
    {
      "epoch": 0.13448239174380253,
      "grad_norm": 0.3233386278152466,
      "learning_rate": 5.236402564375514e-05,
      "loss": 1.3237,
      "step": 316
    },
    {
      "epoch": 0.1349079689328652,
      "grad_norm": 0.3291011154651642,
      "learning_rate": 5.211533815926417e-05,
      "loss": 1.4302,
      "step": 317
    },
    {
      "epoch": 0.13533354612192786,
      "grad_norm": 0.338759183883667,
      "learning_rate": 5.186659823993482e-05,
      "loss": 1.4143,
      "step": 318
    },
    {
      "epoch": 0.13575912331099052,
      "grad_norm": 0.33174023032188416,
      "learning_rate": 5.161781205151294e-05,
      "loss": 1.4507,
      "step": 319
    },
    {
      "epoch": 0.13618470050005319,
      "grad_norm": 0.3163542151451111,
      "learning_rate": 5.136898576089131e-05,
      "loss": 1.3993,
      "step": 320
    },
    {
      "epoch": 0.13661027768911585,
      "grad_norm": 0.30455371737480164,
      "learning_rate": 5.112012553595671e-05,
      "loss": 1.4941,
      "step": 321
    },
    {
      "epoch": 0.13703585487817854,
      "grad_norm": 0.30642834305763245,
      "learning_rate": 5.0871237545437145e-05,
      "loss": 1.4564,
      "step": 322
    },
    {
      "epoch": 0.1374614320672412,
      "grad_norm": 0.3390425443649292,
      "learning_rate": 5.062232795874879e-05,
      "loss": 1.3943,
      "step": 323
    },
    {
      "epoch": 0.13788700925630387,
      "grad_norm": 0.31367331743240356,
      "learning_rate": 5.037340294584323e-05,
      "loss": 1.2689,
      "step": 324
    },
    {
      "epoch": 0.13831258644536654,
      "grad_norm": 0.3435627520084381,
      "learning_rate": 5.0124468677054384e-05,
      "loss": 1.4298,
      "step": 325
    },
    {
      "epoch": 0.1387381636344292,
      "grad_norm": 0.35670870542526245,
      "learning_rate": 4.987553132294563e-05,
      "loss": 1.321,
      "step": 326
    },
    {
      "epoch": 0.13916374082349187,
      "grad_norm": 0.32847118377685547,
      "learning_rate": 4.962659705415677e-05,
      "loss": 1.3769,
      "step": 327
    },
    {
      "epoch": 0.13958931801255453,
      "grad_norm": 0.33433982729911804,
      "learning_rate": 4.937767204125122e-05,
      "loss": 1.2956,
      "step": 328
    },
    {
      "epoch": 0.1400148952016172,
      "grad_norm": 0.2976665198802948,
      "learning_rate": 4.912876245456288e-05,
      "loss": 1.4297,
      "step": 329
    },
    {
      "epoch": 0.14044047239067986,
      "grad_norm": 0.33189690113067627,
      "learning_rate": 4.88798744640433e-05,
      "loss": 1.3718,
      "step": 330
    },
    {
      "epoch": 0.14086604957974252,
      "grad_norm": 0.33690720796585083,
      "learning_rate": 4.86310142391087e-05,
      "loss": 1.4438,
      "step": 331
    },
    {
      "epoch": 0.1412916267688052,
      "grad_norm": 0.3246925175189972,
      "learning_rate": 4.8382187948487054e-05,
      "loss": 1.427,
      "step": 332
    },
    {
      "epoch": 0.14171720395786785,
      "grad_norm": 0.3402460217475891,
      "learning_rate": 4.813340176006519e-05,
      "loss": 1.3958,
      "step": 333
    },
    {
      "epoch": 0.14214278114693052,
      "grad_norm": 0.263990193605423,
      "learning_rate": 4.788466184073585e-05,
      "loss": 1.2336,
      "step": 334
    },
    {
      "epoch": 0.14256835833599318,
      "grad_norm": 0.330545037984848,
      "learning_rate": 4.7635974356244864e-05,
      "loss": 1.3919,
      "step": 335
    },
    {
      "epoch": 0.14299393552505585,
      "grad_norm": 0.3169040083885193,
      "learning_rate": 4.738734547103836e-05,
      "loss": 1.3512,
      "step": 336
    },
    {
      "epoch": 0.1434195127141185,
      "grad_norm": 0.32559284567832947,
      "learning_rate": 4.713878134810984e-05,
      "loss": 1.4493,
      "step": 337
    },
    {
      "epoch": 0.14384508990318118,
      "grad_norm": 0.34363633394241333,
      "learning_rate": 4.689028814884756e-05,
      "loss": 1.2949,
      "step": 338
    },
    {
      "epoch": 0.14427066709224384,
      "grad_norm": 0.3349243998527527,
      "learning_rate": 4.664187203288167e-05,
      "loss": 1.4964,
      "step": 339
    },
    {
      "epoch": 0.14469624428130654,
      "grad_norm": 0.30426791310310364,
      "learning_rate": 4.6393539157931586e-05,
      "loss": 1.3914,
      "step": 340
    },
    {
      "epoch": 0.1451218214703692,
      "grad_norm": 0.30210012197494507,
      "learning_rate": 4.61452956796534e-05,
      "loss": 1.4269,
      "step": 341
    },
    {
      "epoch": 0.14554739865943186,
      "grad_norm": 0.35527628660202026,
      "learning_rate": 4.589714775148719e-05,
      "loss": 1.4037,
      "step": 342
    },
    {
      "epoch": 0.14597297584849453,
      "grad_norm": 0.30980184674263,
      "learning_rate": 4.564910152450459e-05,
      "loss": 1.3133,
      "step": 343
    },
    {
      "epoch": 0.1463985530375572,
      "grad_norm": 0.3374204635620117,
      "learning_rate": 4.540116314725622e-05,
      "loss": 1.3178,
      "step": 344
    },
    {
      "epoch": 0.14682413022661986,
      "grad_norm": 0.32149648666381836,
      "learning_rate": 4.515333876561941e-05,
      "loss": 1.3854,
      "step": 345
    },
    {
      "epoch": 0.14724970741568252,
      "grad_norm": 0.3345913589000702,
      "learning_rate": 4.4905634522645694e-05,
      "loss": 1.4373,
      "step": 346
    },
    {
      "epoch": 0.1476752846047452,
      "grad_norm": 0.3649419844150543,
      "learning_rate": 4.4658056558408644e-05,
      "loss": 1.418,
      "step": 347
    },
    {
      "epoch": 0.14810086179380785,
      "grad_norm": 0.3086884319782257,
      "learning_rate": 4.4410611009851686e-05,
      "loss": 1.3965,
      "step": 348
    },
    {
      "epoch": 0.14852643898287052,
      "grad_norm": 0.3454095125198364,
      "learning_rate": 4.4163304010635876e-05,
      "loss": 1.5076,
      "step": 349
    },
    {
      "epoch": 0.14895201617193318,
      "grad_norm": 0.34644293785095215,
      "learning_rate": 4.391614169098795e-05,
      "loss": 1.4771,
      "step": 350
    },
    {
      "epoch": 0.14937759336099585,
      "grad_norm": 0.3557009994983673,
      "learning_rate": 4.366913017754836e-05,
      "loss": 1.3823,
      "step": 351
    },
    {
      "epoch": 0.1498031705500585,
      "grad_norm": 0.31497180461883545,
      "learning_rate": 4.342227559321932e-05,
      "loss": 1.421,
      "step": 352
    },
    {
      "epoch": 0.15022874773912118,
      "grad_norm": 0.29452037811279297,
      "learning_rate": 4.317558405701316e-05,
      "loss": 1.3823,
      "step": 353
    },
    {
      "epoch": 0.15065432492818384,
      "grad_norm": 0.35082679986953735,
      "learning_rate": 4.292906168390055e-05,
      "loss": 1.4215,
      "step": 354
    },
    {
      "epoch": 0.1510799021172465,
      "grad_norm": 0.32057732343673706,
      "learning_rate": 4.2682714584659e-05,
      "loss": 1.2923,
      "step": 355
    },
    {
      "epoch": 0.15150547930630917,
      "grad_norm": 0.3253067135810852,
      "learning_rate": 4.243654886572129e-05,
      "loss": 1.3602,
      "step": 356
    },
    {
      "epoch": 0.15193105649537184,
      "grad_norm": 0.3210274875164032,
      "learning_rate": 4.219057062902417e-05,
      "loss": 1.4533,
      "step": 357
    },
    {
      "epoch": 0.15235663368443453,
      "grad_norm": 0.4304807782173157,
      "learning_rate": 4.194478597185716e-05,
      "loss": 1.4197,
      "step": 358
    },
    {
      "epoch": 0.1527822108734972,
      "grad_norm": 0.3173682391643524,
      "learning_rate": 4.169920098671124e-05,
      "loss": 1.4527,
      "step": 359
    },
    {
      "epoch": 0.15320778806255986,
      "grad_norm": 0.3237103819847107,
      "learning_rate": 4.145382176112804e-05,
      "loss": 1.3939,
      "step": 360
    },
    {
      "epoch": 0.15363336525162252,
      "grad_norm": 0.2935871183872223,
      "learning_rate": 4.120865437754877e-05,
      "loss": 1.312,
      "step": 361
    },
    {
      "epoch": 0.1540589424406852,
      "grad_norm": 0.3397037982940674,
      "learning_rate": 4.0963704913163526e-05,
      "loss": 1.5097,
      "step": 362
    },
    {
      "epoch": 0.15448451962974785,
      "grad_norm": 0.31791093945503235,
      "learning_rate": 4.0718979439760665e-05,
      "loss": 1.3923,
      "step": 363
    },
    {
      "epoch": 0.15491009681881052,
      "grad_norm": 0.32730478048324585,
      "learning_rate": 4.047448402357622e-05,
      "loss": 1.3108,
      "step": 364
    },
    {
      "epoch": 0.15533567400787318,
      "grad_norm": 0.3350706398487091,
      "learning_rate": 4.023022472514368e-05,
      "loss": 1.3763,
      "step": 365
    },
    {
      "epoch": 0.15576125119693585,
      "grad_norm": 0.31630778312683105,
      "learning_rate": 3.9986207599143564e-05,
      "loss": 1.4107,
      "step": 366
    },
    {
      "epoch": 0.1561868283859985,
      "grad_norm": 0.32477709650993347,
      "learning_rate": 3.9742438694253484e-05,
      "loss": 1.3997,
      "step": 367
    },
    {
      "epoch": 0.15661240557506118,
      "grad_norm": 0.32294079661369324,
      "learning_rate": 3.9498924052998195e-05,
      "loss": 1.4678,
      "step": 368
    },
    {
      "epoch": 0.15703798276412384,
      "grad_norm": 0.30389007925987244,
      "learning_rate": 3.925566971159971e-05,
      "loss": 1.317,
      "step": 369
    },
    {
      "epoch": 0.1574635599531865,
      "grad_norm": 0.29815390706062317,
      "learning_rate": 3.901268169982784e-05,
      "loss": 1.2426,
      "step": 370
    },
    {
      "epoch": 0.15788913714224917,
      "grad_norm": 0.34378722310066223,
      "learning_rate": 3.8769966040850566e-05,
      "loss": 1.5315,
      "step": 371
    },
    {
      "epoch": 0.15831471433131183,
      "grad_norm": 0.2889562249183655,
      "learning_rate": 3.8527528751084826e-05,
      "loss": 1.2998,
      "step": 372
    },
    {
      "epoch": 0.1587402915203745,
      "grad_norm": 0.31654536724090576,
      "learning_rate": 3.8285375840047395e-05,
      "loss": 1.3312,
      "step": 373
    },
    {
      "epoch": 0.15916586870943716,
      "grad_norm": 0.3028466999530792,
      "learning_rate": 3.804351331020584e-05,
      "loss": 1.3981,
      "step": 374
    },
    {
      "epoch": 0.15959144589849983,
      "grad_norm": 0.31716054677963257,
      "learning_rate": 3.7801947156829834e-05,
      "loss": 1.4174,
      "step": 375
    },
    {
      "epoch": 0.1600170230875625,
      "grad_norm": 0.3307088017463684,
      "learning_rate": 3.7560683367842456e-05,
      "loss": 1.3022,
      "step": 376
    },
    {
      "epoch": 0.16044260027662519,
      "grad_norm": 0.37073442339897156,
      "learning_rate": 3.7319727923671785e-05,
      "loss": 1.2788,
      "step": 377
    },
    {
      "epoch": 0.16086817746568785,
      "grad_norm": 0.3232400119304657,
      "learning_rate": 3.707908679710276e-05,
      "loss": 1.4228,
      "step": 378
    },
    {
      "epoch": 0.16129375465475052,
      "grad_norm": 0.2912347912788391,
      "learning_rate": 3.6838765953128914e-05,
      "loss": 1.3194,
      "step": 379
    },
    {
      "epoch": 0.16171933184381318,
      "grad_norm": 0.35164567828178406,
      "learning_rate": 3.6598771348804766e-05,
      "loss": 1.4528,
      "step": 380
    },
    {
      "epoch": 0.16214490903287584,
      "grad_norm": 0.3369240164756775,
      "learning_rate": 3.635910893309792e-05,
      "loss": 1.3688,
      "step": 381
    },
    {
      "epoch": 0.1625704862219385,
      "grad_norm": 0.3216218054294586,
      "learning_rate": 3.6119784646741804e-05,
      "loss": 1.3827,
      "step": 382
    },
    {
      "epoch": 0.16299606341100117,
      "grad_norm": 0.33442652225494385,
      "learning_rate": 3.5880804422088255e-05,
      "loss": 1.4323,
      "step": 383
    },
    {
      "epoch": 0.16342164060006384,
      "grad_norm": 0.3149116337299347,
      "learning_rate": 3.564217418296055e-05,
      "loss": 1.3756,
      "step": 384
    },
    {
      "epoch": 0.1638472177891265,
      "grad_norm": 0.32829785346984863,
      "learning_rate": 3.540389984450661e-05,
      "loss": 1.4512,
      "step": 385
    },
    {
      "epoch": 0.16427279497818917,
      "grad_norm": 0.37037840485572815,
      "learning_rate": 3.516598731305222e-05,
      "loss": 1.3539,
      "step": 386
    },
    {
      "epoch": 0.16469837216725183,
      "grad_norm": 0.304059773683548,
      "learning_rate": 3.492844248595483e-05,
      "loss": 1.3534,
      "step": 387
    },
    {
      "epoch": 0.1651239493563145,
      "grad_norm": 0.3242076337337494,
      "learning_rate": 3.469127125145717e-05,
      "loss": 1.3389,
      "step": 388
    },
    {
      "epoch": 0.16554952654537716,
      "grad_norm": 0.33115020394325256,
      "learning_rate": 3.4454479488541414e-05,
      "loss": 1.4357,
      "step": 389
    },
    {
      "epoch": 0.16597510373443983,
      "grad_norm": 0.32753780484199524,
      "learning_rate": 3.421807306678346e-05,
      "loss": 1.3659,
      "step": 390
    },
    {
      "epoch": 0.1664006809235025,
      "grad_norm": 0.30588462948799133,
      "learning_rate": 3.398205784620735e-05,
      "loss": 1.3735,
      "step": 391
    },
    {
      "epoch": 0.16682625811256516,
      "grad_norm": 0.30423295497894287,
      "learning_rate": 3.3746439677140106e-05,
      "loss": 1.2956,
      "step": 392
    },
    {
      "epoch": 0.16725183530162782,
      "grad_norm": 0.33183082938194275,
      "learning_rate": 3.351122440006661e-05,
      "loss": 1.338,
      "step": 393
    },
    {
      "epoch": 0.1676774124906905,
      "grad_norm": 0.32961684465408325,
      "learning_rate": 3.327641784548495e-05,
      "loss": 1.4384,
      "step": 394
    },
    {
      "epoch": 0.16810298967975318,
      "grad_norm": 0.2988288700580597,
      "learning_rate": 3.3042025833761806e-05,
      "loss": 1.3986,
      "step": 395
    },
    {
      "epoch": 0.16852856686881584,
      "grad_norm": 0.32923299074172974,
      "learning_rate": 3.280805417498816e-05,
      "loss": 1.4054,
      "step": 396
    },
    {
      "epoch": 0.1689541440578785,
      "grad_norm": 0.31945064663887024,
      "learning_rate": 3.257450866883542e-05,
      "loss": 1.3031,
      "step": 397
    },
    {
      "epoch": 0.16937972124694117,
      "grad_norm": 0.3342922627925873,
      "learning_rate": 3.234139510441146e-05,
      "loss": 1.4884,
      "step": 398
    },
    {
      "epoch": 0.16980529843600384,
      "grad_norm": 0.31479397416114807,
      "learning_rate": 3.210871926011725e-05,
      "loss": 1.3229,
      "step": 399
    },
    {
      "epoch": 0.1702308756250665,
      "grad_norm": 0.32442063093185425,
      "learning_rate": 3.18764869035036e-05,
      "loss": 1.358,
      "step": 400
    },
    {
      "epoch": 0.17065645281412917,
      "grad_norm": 0.3147645592689514,
      "learning_rate": 3.164470379112816e-05,
      "loss": 1.3472,
      "step": 401
    },
    {
      "epoch": 0.17108203000319183,
      "grad_norm": 0.3168991506099701,
      "learning_rate": 3.141337566841277e-05,
      "loss": 1.365,
      "step": 402
    },
    {
      "epoch": 0.1715076071922545,
      "grad_norm": 0.33438199758529663,
      "learning_rate": 3.1182508269500985e-05,
      "loss": 1.3538,
      "step": 403
    },
    {
      "epoch": 0.17193318438131716,
      "grad_norm": 0.31441277265548706,
      "learning_rate": 3.095210731711603e-05,
      "loss": 1.4489,
      "step": 404
    },
    {
      "epoch": 0.17235876157037983,
      "grad_norm": 0.3182431757450104,
      "learning_rate": 3.072217852241884e-05,
      "loss": 1.3436,
      "step": 405
    },
    {
      "epoch": 0.1727843387594425,
      "grad_norm": 0.3269684612751007,
      "learning_rate": 3.0492727584866554e-05,
      "loss": 1.393,
      "step": 406
    },
    {
      "epoch": 0.17320991594850516,
      "grad_norm": 0.3089240789413452,
      "learning_rate": 3.026376019207126e-05,
      "loss": 1.4003,
      "step": 407
    },
    {
      "epoch": 0.17363549313756782,
      "grad_norm": 0.32416653633117676,
      "learning_rate": 3.0035282019658928e-05,
      "loss": 1.4448,
      "step": 408
    },
    {
      "epoch": 0.17406107032663048,
      "grad_norm": 0.35581550002098083,
      "learning_rate": 2.9807298731128773e-05,
      "loss": 1.3682,
      "step": 409
    },
    {
      "epoch": 0.17448664751569315,
      "grad_norm": 0.34638750553131104,
      "learning_rate": 2.957981597771292e-05,
      "loss": 1.3646,
      "step": 410
    },
    {
      "epoch": 0.17491222470475581,
      "grad_norm": 0.3218763470649719,
      "learning_rate": 2.935283939823621e-05,
      "loss": 1.4833,
      "step": 411
    },
    {
      "epoch": 0.17533780189381848,
      "grad_norm": 0.29522332549095154,
      "learning_rate": 2.9126374618976528e-05,
      "loss": 1.2961,
      "step": 412
    },
    {
      "epoch": 0.17576337908288114,
      "grad_norm": 0.30747368931770325,
      "learning_rate": 2.8900427253525248e-05,
      "loss": 1.3874,
      "step": 413
    },
    {
      "epoch": 0.17618895627194384,
      "grad_norm": 0.3579199016094208,
      "learning_rate": 2.8675002902648146e-05,
      "loss": 1.4221,
      "step": 414
    },
    {
      "epoch": 0.1766145334610065,
      "grad_norm": 0.29601120948791504,
      "learning_rate": 2.8450107154146606e-05,
      "loss": 1.4503,
      "step": 415
    },
    {
      "epoch": 0.17704011065006917,
      "grad_norm": 0.32221487164497375,
      "learning_rate": 2.8225745582718965e-05,
      "loss": 1.4458,
      "step": 416
    },
    {
      "epoch": 0.17746568783913183,
      "grad_norm": 0.316237211227417,
      "learning_rate": 2.8001923749822523e-05,
      "loss": 1.3074,
      "step": 417
    },
    {
      "epoch": 0.1778912650281945,
      "grad_norm": 0.374932199716568,
      "learning_rate": 2.7778647203535475e-05,
      "loss": 1.337,
      "step": 418
    },
    {
      "epoch": 0.17831684221725716,
      "grad_norm": 0.3187963366508484,
      "learning_rate": 2.75559214784196e-05,
      "loss": 1.2731,
      "step": 419
    },
    {
      "epoch": 0.17874241940631982,
      "grad_norm": 0.3493146300315857,
      "learning_rate": 2.733375209538288e-05,
      "loss": 1.3924,
      "step": 420
    },
    {
      "epoch": 0.1791679965953825,
      "grad_norm": 0.3364221155643463,
      "learning_rate": 2.7112144561542757e-05,
      "loss": 1.3695,
      "step": 421
    },
    {
      "epoch": 0.17959357378444515,
      "grad_norm": 0.3134714663028717,
      "learning_rate": 2.6891104370089642e-05,
      "loss": 1.4045,
      "step": 422
    },
    {
      "epoch": 0.18001915097350782,
      "grad_norm": 0.32247835397720337,
      "learning_rate": 2.6670637000150633e-05,
      "loss": 1.3614,
      "step": 423
    },
    {
      "epoch": 0.18044472816257048,
      "grad_norm": 0.3357102572917938,
      "learning_rate": 2.6450747916653857e-05,
      "loss": 1.4259,
      "step": 424
    },
    {
      "epoch": 0.18087030535163315,
      "grad_norm": 0.3296355605125427,
      "learning_rate": 2.6231442570192845e-05,
      "loss": 1.4139,
      "step": 425
    },
    {
      "epoch": 0.1812958825406958,
      "grad_norm": 0.3181535005569458,
      "learning_rate": 2.6012726396891518e-05,
      "loss": 1.3069,
      "step": 426
    },
    {
      "epoch": 0.18172145972975848,
      "grad_norm": 0.33894723653793335,
      "learning_rate": 2.579460481826947e-05,
      "loss": 1.4014,
      "step": 427
    },
    {
      "epoch": 0.18214703691882114,
      "grad_norm": 0.33284929394721985,
      "learning_rate": 2.557708324110747e-05,
      "loss": 1.3041,
      "step": 428
    },
    {
      "epoch": 0.1825726141078838,
      "grad_norm": 0.35112953186035156,
      "learning_rate": 2.536016705731351e-05,
      "loss": 1.4356,
      "step": 429
    },
    {
      "epoch": 0.18299819129694647,
      "grad_norm": 0.3994680941104889,
      "learning_rate": 2.514386164378915e-05,
      "loss": 1.412,
      "step": 430
    },
    {
      "epoch": 0.18342376848600914,
      "grad_norm": 0.32337042689323425,
      "learning_rate": 2.4928172362296205e-05,
      "loss": 1.4461,
      "step": 431
    },
    {
      "epoch": 0.18384934567507183,
      "grad_norm": 0.3220783472061157,
      "learning_rate": 2.4713104559323895e-05,
      "loss": 1.4217,
      "step": 432
    },
    {
      "epoch": 0.1842749228641345,
      "grad_norm": 0.3344449996948242,
      "learning_rate": 2.4498663565956233e-05,
      "loss": 1.4241,
      "step": 433
    },
    {
      "epoch": 0.18470050005319716,
      "grad_norm": 0.3385544717311859,
      "learning_rate": 2.428485469773997e-05,
      "loss": 1.3389,
      "step": 434
    },
    {
      "epoch": 0.18512607724225982,
      "grad_norm": 0.36458736658096313,
      "learning_rate": 2.4071683254552752e-05,
      "loss": 1.4758,
      "step": 435
    },
    {
      "epoch": 0.1855516544313225,
      "grad_norm": 0.3004361093044281,
      "learning_rate": 2.3859154520471766e-05,
      "loss": 1.3474,
      "step": 436
    },
    {
      "epoch": 0.18597723162038515,
      "grad_norm": 0.3578473627567291,
      "learning_rate": 2.364727376364285e-05,
      "loss": 1.4717,
      "step": 437
    },
    {
      "epoch": 0.18640280880944782,
      "grad_norm": 0.3373146057128906,
      "learning_rate": 2.343604623614974e-05,
      "loss": 1.3185,
      "step": 438
    },
    {
      "epoch": 0.18682838599851048,
      "grad_norm": 0.33882611989974976,
      "learning_rate": 2.3225477173884063e-05,
      "loss": 1.4053,
      "step": 439
    },
    {
      "epoch": 0.18725396318757315,
      "grad_norm": 0.3465374708175659,
      "learning_rate": 2.3015571796415398e-05,
      "loss": 1.3454,
      "step": 440
    },
    {
      "epoch": 0.1876795403766358,
      "grad_norm": 0.37321579456329346,
      "learning_rate": 2.280633530686195e-05,
      "loss": 1.3604,
      "step": 441
    },
    {
      "epoch": 0.18810511756569848,
      "grad_norm": 0.3368617594242096,
      "learning_rate": 2.2597772891761653e-05,
      "loss": 1.4659,
      "step": 442
    },
    {
      "epoch": 0.18853069475476114,
      "grad_norm": 0.3479175269603729,
      "learning_rate": 2.2389889720943447e-05,
      "loss": 1.4366,
      "step": 443
    },
    {
      "epoch": 0.1889562719438238,
      "grad_norm": 0.31625932455062866,
      "learning_rate": 2.2182690947399304e-05,
      "loss": 1.3762,
      "step": 444
    },
    {
      "epoch": 0.18938184913288647,
      "grad_norm": 0.30867132544517517,
      "learning_rate": 2.1976181707156345e-05,
      "loss": 1.4616,
      "step": 445
    },
    {
      "epoch": 0.18980742632194914,
      "grad_norm": 0.31360673904418945,
      "learning_rate": 2.1770367119149598e-05,
      "loss": 1.4013,
      "step": 446
    },
    {
      "epoch": 0.1902330035110118,
      "grad_norm": 0.3171330690383911,
      "learning_rate": 2.1565252285095156e-05,
      "loss": 1.2904,
      "step": 447
    },
    {
      "epoch": 0.19065858070007446,
      "grad_norm": 0.33907023072242737,
      "learning_rate": 2.1360842289363614e-05,
      "loss": 1.3836,
      "step": 448
    },
    {
      "epoch": 0.19108415788913713,
      "grad_norm": 0.3281523585319519,
      "learning_rate": 2.1157142198854103e-05,
      "loss": 1.4903,
      "step": 449
    },
    {
      "epoch": 0.19150973507819982,
      "grad_norm": 0.32757705450057983,
      "learning_rate": 2.0954157062868668e-05,
      "loss": 1.5003,
      "step": 450
    },
    {
      "epoch": 0.1919353122672625,
      "grad_norm": 0.3444954752922058,
      "learning_rate": 2.075189191298716e-05,
      "loss": 1.3952,
      "step": 451
    },
    {
      "epoch": 0.19236088945632515,
      "grad_norm": 0.3098054826259613,
      "learning_rate": 2.0550351762942428e-05,
      "loss": 1.2589,
      "step": 452
    },
    {
      "epoch": 0.19278646664538782,
      "grad_norm": 0.3291134834289551,
      "learning_rate": 2.0349541608496077e-05,
      "loss": 1.4151,
      "step": 453
    },
    {
      "epoch": 0.19321204383445048,
      "grad_norm": 0.32852277159690857,
      "learning_rate": 2.0149466427314682e-05,
      "loss": 1.4448,
      "step": 454
    },
    {
      "epoch": 0.19363762102351315,
      "grad_norm": 0.3113350570201874,
      "learning_rate": 1.9950131178846303e-05,
      "loss": 1.3029,
      "step": 455
    },
    {
      "epoch": 0.1940631982125758,
      "grad_norm": 0.30784913897514343,
      "learning_rate": 1.9751540804197666e-05,
      "loss": 1.2535,
      "step": 456
    },
    {
      "epoch": 0.19448877540163848,
      "grad_norm": 0.32419663667678833,
      "learning_rate": 1.9553700226011567e-05,
      "loss": 1.2871,
      "step": 457
    },
    {
      "epoch": 0.19491435259070114,
      "grad_norm": 0.31145137548446655,
      "learning_rate": 1.935661434834491e-05,
      "loss": 1.4369,
      "step": 458
    },
    {
      "epoch": 0.1953399297797638,
      "grad_norm": 0.3428165316581726,
      "learning_rate": 1.91602880565472e-05,
      "loss": 1.395,
      "step": 459
    },
    {
      "epoch": 0.19576550696882647,
      "grad_norm": 0.3369043469429016,
      "learning_rate": 1.89647262171393e-05,
      "loss": 1.3816,
      "step": 460
    },
    {
      "epoch": 0.19619108415788913,
      "grad_norm": 0.33101511001586914,
      "learning_rate": 1.876993367769297e-05,
      "loss": 1.4208,
      "step": 461
    },
    {
      "epoch": 0.1966166613469518,
      "grad_norm": 0.34739160537719727,
      "learning_rate": 1.8575915266710564e-05,
      "loss": 1.3579,
      "step": 462
    },
    {
      "epoch": 0.19704223853601446,
      "grad_norm": 0.34199175238609314,
      "learning_rate": 1.8382675793505406e-05,
      "loss": 1.4565,
      "step": 463
    },
    {
      "epoch": 0.19746781572507713,
      "grad_norm": 0.3198068141937256,
      "learning_rate": 1.8190220048082613e-05,
      "loss": 1.4965,
      "step": 464
    },
    {
      "epoch": 0.1978933929141398,
      "grad_norm": 0.324301153421402,
      "learning_rate": 1.7998552801020257e-05,
      "loss": 1.4035,
      "step": 465
    },
    {
      "epoch": 0.19831897010320246,
      "grad_norm": 0.347256600856781,
      "learning_rate": 1.7807678803351237e-05,
      "loss": 1.4661,
      "step": 466
    },
    {
      "epoch": 0.19874454729226512,
      "grad_norm": 0.3176523447036743,
      "learning_rate": 1.7617602786445403e-05,
      "loss": 1.4176,
      "step": 467
    },
    {
      "epoch": 0.1991701244813278,
      "grad_norm": 0.32808732986450195,
      "learning_rate": 1.7428329461892328e-05,
      "loss": 1.4373,
      "step": 468
    },
    {
      "epoch": 0.19959570167039048,
      "grad_norm": 0.3186570405960083,
      "learning_rate": 1.723986352138452e-05,
      "loss": 1.478,
      "step": 469
    },
    {
      "epoch": 0.20002127885945314,
      "grad_norm": 0.31049400568008423,
      "learning_rate": 1.7052209636601087e-05,
      "loss": 1.3014,
      "step": 470
    },
    {
      "epoch": 0.2004468560485158,
      "grad_norm": 0.314045250415802,
      "learning_rate": 1.686537245909201e-05,
      "loss": 1.4446,
      "step": 471
    },
    {
      "epoch": 0.20087243323757847,
      "grad_norm": 0.3211539387702942,
      "learning_rate": 1.6679356620162766e-05,
      "loss": 1.3681,
      "step": 472
    },
    {
      "epoch": 0.20129801042664114,
      "grad_norm": 0.30641838908195496,
      "learning_rate": 1.6494166730759524e-05,
      "loss": 1.3608,
      "step": 473
    },
    {
      "epoch": 0.2017235876157038,
      "grad_norm": 0.33943578600883484,
      "learning_rate": 1.6309807381354958e-05,
      "loss": 1.3157,
      "step": 474
    },
    {
      "epoch": 0.20214916480476647,
      "grad_norm": 0.29183852672576904,
      "learning_rate": 1.6126283141834293e-05,
      "loss": 1.3042,
      "step": 475
    },
    {
      "epoch": 0.20257474199382913,
      "grad_norm": 0.294260710477829,
      "learning_rate": 1.594359856138219e-05,
      "loss": 1.393,
      "step": 476
    },
    {
      "epoch": 0.2030003191828918,
      "grad_norm": 0.3283140957355499,
      "learning_rate": 1.5761758168369862e-05,
      "loss": 1.431,
      "step": 477
    },
    {
      "epoch": 0.20342589637195446,
      "grad_norm": 0.32865697145462036,
      "learning_rate": 1.558076647024285e-05,
      "loss": 1.3548,
      "step": 478
    },
    {
      "epoch": 0.20385147356101713,
      "grad_norm": 0.3092752993106842,
      "learning_rate": 1.5400627953409396e-05,
      "loss": 1.2688,
      "step": 479
    },
    {
      "epoch": 0.2042770507500798,
      "grad_norm": 0.3280682861804962,
      "learning_rate": 1.5221347083129045e-05,
      "loss": 1.3923,
      "step": 480
    },
    {
      "epoch": 0.20470262793914246,
      "grad_norm": 0.3039858639240265,
      "learning_rate": 1.5042928303402155e-05,
      "loss": 1.3433,
      "step": 481
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 0.3173251748085022,
      "learning_rate": 1.4865376036859597e-05,
      "loss": 1.427,
      "step": 482
    },
    {
      "epoch": 0.20555378231726779,
      "grad_norm": 0.31789281964302063,
      "learning_rate": 1.4688694684653181e-05,
      "loss": 1.4348,
      "step": 483
    },
    {
      "epoch": 0.20597935950633045,
      "grad_norm": 0.31420034170150757,
      "learning_rate": 1.4512888626346599e-05,
      "loss": 1.4612,
      "step": 484
    },
    {
      "epoch": 0.20640493669539312,
      "grad_norm": 0.3418427109718323,
      "learning_rate": 1.4337962219806778e-05,
      "loss": 1.527,
      "step": 485
    },
    {
      "epoch": 0.20683051388445578,
      "grad_norm": 0.2978665828704834,
      "learning_rate": 1.4163919801095954e-05,
      "loss": 1.4091,
      "step": 486
    },
    {
      "epoch": 0.20725609107351847,
      "grad_norm": 0.3711923658847809,
      "learning_rate": 1.39907656843641e-05,
      "loss": 1.3739,
      "step": 487
    },
    {
      "epoch": 0.20768166826258114,
      "grad_norm": 0.3134306073188782,
      "learning_rate": 1.381850416174203e-05,
      "loss": 1.4328,
      "step": 488
    },
    {
      "epoch": 0.2081072454516438,
      "grad_norm": 0.32045456767082214,
      "learning_rate": 1.3647139503235046e-05,
      "loss": 1.4792,
      "step": 489
    },
    {
      "epoch": 0.20853282264070647,
      "grad_norm": 0.3508617579936981,
      "learning_rate": 1.3476675956617007e-05,
      "loss": 1.3979,
      "step": 490
    },
    {
      "epoch": 0.20895839982976913,
      "grad_norm": 0.3196643590927124,
      "learning_rate": 1.3307117747325104e-05,
      "loss": 1.311,
      "step": 491
    },
    {
      "epoch": 0.2093839770188318,
      "grad_norm": 0.3366539776325226,
      "learning_rate": 1.313846907835507e-05,
      "loss": 1.4454,
      "step": 492
    },
    {
      "epoch": 0.20980955420789446,
      "grad_norm": 0.31268173456192017,
      "learning_rate": 1.2970734130157069e-05,
      "loss": 1.373,
      "step": 493
    },
    {
      "epoch": 0.21023513139695713,
      "grad_norm": 0.3394613265991211,
      "learning_rate": 1.2803917060531994e-05,
      "loss": 1.4365,
      "step": 494
    },
    {
      "epoch": 0.2106607085860198,
      "grad_norm": 0.3489893674850464,
      "learning_rate": 1.2638022004528416e-05,
      "loss": 1.3662,
      "step": 495
    },
    {
      "epoch": 0.21108628577508246,
      "grad_norm": 0.3316808044910431,
      "learning_rate": 1.2473053074340157e-05,
      "loss": 1.4181,
      "step": 496
    },
    {
      "epoch": 0.21151186296414512,
      "grad_norm": 0.3342049717903137,
      "learning_rate": 1.2309014359204252e-05,
      "loss": 1.2779,
      "step": 497
    },
    {
      "epoch": 0.21193744015320778,
      "grad_norm": 0.33748358488082886,
      "learning_rate": 1.2145909925299687e-05,
      "loss": 1.3213,
      "step": 498
    },
    {
      "epoch": 0.21236301734227045,
      "grad_norm": 0.406751811504364,
      "learning_rate": 1.1983743815646508e-05,
      "loss": 1.3167,
      "step": 499
    },
    {
      "epoch": 0.21278859453133311,
      "grad_norm": 0.30885371565818787,
      "learning_rate": 1.1822520050005675e-05,
      "loss": 1.4627,
      "step": 500
    },
    {
      "epoch": 0.21321417172039578,
      "grad_norm": 0.30035650730133057,
      "learning_rate": 1.1662242624779413e-05,
      "loss": 1.3638,
      "step": 501
    },
    {
      "epoch": 0.21363974890945844,
      "grad_norm": 0.3490166962146759,
      "learning_rate": 1.1502915512912094e-05,
      "loss": 1.3379,
      "step": 502
    },
    {
      "epoch": 0.2140653260985211,
      "grad_norm": 0.31692811846733093,
      "learning_rate": 1.1344542663791851e-05,
      "loss": 1.3315,
      "step": 503
    },
    {
      "epoch": 0.21449090328758377,
      "grad_norm": 0.3350829780101776,
      "learning_rate": 1.118712800315258e-05,
      "loss": 1.4076,
      "step": 504
    },
    {
      "epoch": 0.21491648047664647,
      "grad_norm": 0.32507750391960144,
      "learning_rate": 1.1030675432976679e-05,
      "loss": 1.3802,
      "step": 505
    },
    {
      "epoch": 0.21534205766570913,
      "grad_norm": 0.3646472692489624,
      "learning_rate": 1.0875188831398353e-05,
      "loss": 1.5334,
      "step": 506
    },
    {
      "epoch": 0.2157676348547718,
      "grad_norm": 0.3090904951095581,
      "learning_rate": 1.0720672052607416e-05,
      "loss": 1.3333,
      "step": 507
    },
    {
      "epoch": 0.21619321204383446,
      "grad_norm": 0.3293876349925995,
      "learning_rate": 1.0567128926753827e-05,
      "loss": 1.2744,
      "step": 508
    },
    {
      "epoch": 0.21661878923289712,
      "grad_norm": 0.32312512397766113,
      "learning_rate": 1.0414563259852684e-05,
      "loss": 1.4732,
      "step": 509
    },
    {
      "epoch": 0.2170443664219598,
      "grad_norm": 0.3056824505329132,
      "learning_rate": 1.0262978833689906e-05,
      "loss": 1.3128,
      "step": 510
    },
    {
      "epoch": 0.21746994361102245,
      "grad_norm": 0.321416974067688,
      "learning_rate": 1.0112379405728512e-05,
      "loss": 1.2522,
      "step": 511
    },
    {
      "epoch": 0.21789552080008512,
      "grad_norm": 0.30441486835479736,
      "learning_rate": 9.962768709015435e-06,
      "loss": 1.4061,
      "step": 512
    },
    {
      "epoch": 0.21832109798914778,
      "grad_norm": 0.3073800504207611,
      "learning_rate": 9.814150452089055e-06,
      "loss": 1.312,
      "step": 513
    },
    {
      "epoch": 0.21874667517821045,
      "grad_norm": 0.3266609311103821,
      "learning_rate": 9.666528318887198e-06,
      "loss": 1.2735,
      "step": 514
    },
    {
      "epoch": 0.2191722523672731,
      "grad_norm": 0.30625462532043457,
      "learning_rate": 9.51990596865585e-06,
      "loss": 1.3488,
      "step": 515
    },
    {
      "epoch": 0.21959782955633578,
      "grad_norm": 0.326201856136322,
      "learning_rate": 9.374287035858491e-06,
      "loss": 1.3727,
      "step": 516
    },
    {
      "epoch": 0.22002340674539844,
      "grad_norm": 0.3594081401824951,
      "learning_rate": 9.229675130085918e-06,
      "loss": 1.3612,
      "step": 517
    },
    {
      "epoch": 0.2204489839344611,
      "grad_norm": 0.31369614601135254,
      "learning_rate": 9.086073835966869e-06,
      "loss": 1.4186,
      "step": 518
    },
    {
      "epoch": 0.22087456112352377,
      "grad_norm": 0.33593955636024475,
      "learning_rate": 8.94348671307907e-06,
      "loss": 1.4858,
      "step": 519
    },
    {
      "epoch": 0.22130013831258644,
      "grad_norm": 0.33301830291748047,
      "learning_rate": 8.801917295861101e-06,
      "loss": 1.2687,
      "step": 520
    },
    {
      "epoch": 0.2217257155016491,
      "grad_norm": 0.3187962770462036,
      "learning_rate": 8.66136909352469e-06,
      "loss": 1.3218,
      "step": 521
    },
    {
      "epoch": 0.22215129269071177,
      "grad_norm": 0.31819725036621094,
      "learning_rate": 8.521845589967775e-06,
      "loss": 1.4195,
      "step": 522
    },
    {
      "epoch": 0.22257686987977443,
      "grad_norm": 0.3147396743297577,
      "learning_rate": 8.383350243688175e-06,
      "loss": 1.3723,
      "step": 523
    },
    {
      "epoch": 0.22300244706883712,
      "grad_norm": 0.3238096237182617,
      "learning_rate": 8.24588648769778e-06,
      "loss": 1.3681,
      "step": 524
    },
    {
      "epoch": 0.2234280242578998,
      "grad_norm": 0.34045907855033875,
      "learning_rate": 8.109457729437537e-06,
      "loss": 1.4074,
      "step": 525
    },
    {
      "epoch": 0.22385360144696245,
      "grad_norm": 0.3139466643333435,
      "learning_rate": 7.974067350692921e-06,
      "loss": 1.3848,
      "step": 526
    },
    {
      "epoch": 0.22427917863602512,
      "grad_norm": 0.33936434984207153,
      "learning_rate": 7.839718707510146e-06,
      "loss": 1.471,
      "step": 527
    },
    {
      "epoch": 0.22470475582508778,
      "grad_norm": 0.300231009721756,
      "learning_rate": 7.706415130112993e-06,
      "loss": 1.4524,
      "step": 528
    },
    {
      "epoch": 0.22513033301415045,
      "grad_norm": 0.3271010220050812,
      "learning_rate": 7.574159922820184e-06,
      "loss": 1.3812,
      "step": 529
    },
    {
      "epoch": 0.2255559102032131,
      "grad_norm": 0.3297939896583557,
      "learning_rate": 7.44295636396356e-06,
      "loss": 1.3319,
      "step": 530
    },
    {
      "epoch": 0.22598148739227578,
      "grad_norm": 0.3267982304096222,
      "learning_rate": 7.3128077058067675e-06,
      "loss": 1.3745,
      "step": 531
    },
    {
      "epoch": 0.22640706458133844,
      "grad_norm": 0.3305337727069855,
      "learning_rate": 7.1837171744646494e-06,
      "loss": 1.4081,
      "step": 532
    },
    {
      "epoch": 0.2268326417704011,
      "grad_norm": 0.3130592405796051,
      "learning_rate": 7.05568796982331e-06,
      "loss": 1.3635,
      "step": 533
    },
    {
      "epoch": 0.22725821895946377,
      "grad_norm": 0.31961482763290405,
      "learning_rate": 6.9287232654607345e-06,
      "loss": 1.2938,
      "step": 534
    },
    {
      "epoch": 0.22768379614852643,
      "grad_norm": 0.3059883117675781,
      "learning_rate": 6.802826208568203e-06,
      "loss": 1.2205,
      "step": 535
    },
    {
      "epoch": 0.2281093733375891,
      "grad_norm": 0.3591614067554474,
      "learning_rate": 6.677999919872185e-06,
      "loss": 1.4298,
      "step": 536
    },
    {
      "epoch": 0.22853495052665176,
      "grad_norm": 0.3192720115184784,
      "learning_rate": 6.554247493557047e-06,
      "loss": 1.486,
      "step": 537
    },
    {
      "epoch": 0.22896052771571443,
      "grad_norm": 0.3339605927467346,
      "learning_rate": 6.431571997188363e-06,
      "loss": 1.3664,
      "step": 538
    },
    {
      "epoch": 0.2293861049047771,
      "grad_norm": 0.33562207221984863,
      "learning_rate": 6.309976471636808e-06,
      "loss": 1.3784,
      "step": 539
    },
    {
      "epoch": 0.22981168209383976,
      "grad_norm": 0.3569333553314209,
      "learning_rate": 6.18946393100287e-06,
      "loss": 1.4994,
      "step": 540
    },
    {
      "epoch": 0.23023725928290242,
      "grad_norm": 0.3155967593193054,
      "learning_rate": 6.070037362542058e-06,
      "loss": 1.3339,
      "step": 541
    },
    {
      "epoch": 0.23066283647196512,
      "grad_norm": 0.3392326533794403,
      "learning_rate": 5.951699726590881e-06,
      "loss": 1.2821,
      "step": 542
    },
    {
      "epoch": 0.23108841366102778,
      "grad_norm": 0.31855908036231995,
      "learning_rate": 5.834453956493519e-06,
      "loss": 1.2014,
      "step": 543
    },
    {
      "epoch": 0.23151399085009045,
      "grad_norm": 0.3185470700263977,
      "learning_rate": 5.718302958528998e-06,
      "loss": 1.465,
      "step": 544
    },
    {
      "epoch": 0.2319395680391531,
      "grad_norm": 0.3613238036632538,
      "learning_rate": 5.603249611839295e-06,
      "loss": 1.4258,
      "step": 545
    },
    {
      "epoch": 0.23236514522821577,
      "grad_norm": 0.3471008837223053,
      "learning_rate": 5.489296768357827e-06,
      "loss": 1.2939,
      "step": 546
    },
    {
      "epoch": 0.23279072241727844,
      "grad_norm": 0.33712702989578247,
      "learning_rate": 5.376447252738848e-06,
      "loss": 1.3278,
      "step": 547
    },
    {
      "epoch": 0.2332162996063411,
      "grad_norm": 0.33189454674720764,
      "learning_rate": 5.264703862287418e-06,
      "loss": 1.3793,
      "step": 548
    },
    {
      "epoch": 0.23364187679540377,
      "grad_norm": 0.32526296377182007,
      "learning_rate": 5.1540693668900355e-06,
      "loss": 1.4734,
      "step": 549
    },
    {
      "epoch": 0.23406745398446643,
      "grad_norm": 0.32723116874694824,
      "learning_rate": 5.044546508945996e-06,
      "loss": 1.3883,
      "step": 550
    },
    {
      "epoch": 0.2344930311735291,
      "grad_norm": 0.3163575232028961,
      "learning_rate": 4.936138003299412e-06,
      "loss": 1.3487,
      "step": 551
    },
    {
      "epoch": 0.23491860836259176,
      "grad_norm": 0.3888072371482849,
      "learning_rate": 4.828846537171932e-06,
      "loss": 1.3733,
      "step": 552
    },
    {
      "epoch": 0.23534418555165443,
      "grad_norm": 0.31089290976524353,
      "learning_rate": 4.722674770096097e-06,
      "loss": 1.3349,
      "step": 553
    },
    {
      "epoch": 0.2357697627407171,
      "grad_norm": 0.3057332932949066,
      "learning_rate": 4.617625333849434e-06,
      "loss": 1.4249,
      "step": 554
    },
    {
      "epoch": 0.23619533992977976,
      "grad_norm": 0.34270864725112915,
      "learning_rate": 4.513700832389245e-06,
      "loss": 1.3863,
      "step": 555
    },
    {
      "epoch": 0.23662091711884242,
      "grad_norm": 0.29578831791877747,
      "learning_rate": 4.410903841788e-06,
      "loss": 1.3942,
      "step": 556
    },
    {
      "epoch": 0.2370464943079051,
      "grad_norm": 0.3092251121997833,
      "learning_rate": 4.309236910169562e-06,
      "loss": 1.3332,
      "step": 557
    },
    {
      "epoch": 0.23747207149696775,
      "grad_norm": 0.35976263880729675,
      "learning_rate": 4.208702557645933e-06,
      "loss": 1.3551,
      "step": 558
    },
    {
      "epoch": 0.23789764868603042,
      "grad_norm": 0.332644522190094,
      "learning_rate": 4.10930327625485e-06,
      "loss": 1.3691,
      "step": 559
    },
    {
      "epoch": 0.2383232258750931,
      "grad_norm": 0.384097158908844,
      "learning_rate": 4.011041529898019e-06,
      "loss": 1.367,
      "step": 560
    },
    {
      "epoch": 0.23874880306415577,
      "grad_norm": 0.3527825176715851,
      "learning_rate": 3.913919754279966e-06,
      "loss": 1.546,
      "step": 561
    },
    {
      "epoch": 0.23917438025321844,
      "grad_norm": 0.3396340012550354,
      "learning_rate": 3.817940356847765e-06,
      "loss": 1.3436,
      "step": 562
    },
    {
      "epoch": 0.2395999574422811,
      "grad_norm": 0.3177872896194458,
      "learning_rate": 3.7231057167312676e-06,
      "loss": 1.3255,
      "step": 563
    },
    {
      "epoch": 0.24002553463134377,
      "grad_norm": 0.33055418729782104,
      "learning_rate": 3.6294181846841856e-06,
      "loss": 1.3571,
      "step": 564
    },
    {
      "epoch": 0.24045111182040643,
      "grad_norm": 0.29036635160446167,
      "learning_rate": 3.5368800830258064e-06,
      "loss": 1.3218,
      "step": 565
    },
    {
      "epoch": 0.2408766890094691,
      "grad_norm": 0.2930334806442261,
      "learning_rate": 3.445493705583419e-06,
      "loss": 1.434,
      "step": 566
    },
    {
      "epoch": 0.24130226619853176,
      "grad_norm": 0.3378046751022339,
      "learning_rate": 3.3552613176354717e-06,
      "loss": 1.3088,
      "step": 567
    },
    {
      "epoch": 0.24172784338759443,
      "grad_norm": 0.3452695906162262,
      "learning_rate": 3.2661851558554057e-06,
      "loss": 1.3953,
      "step": 568
    },
    {
      "epoch": 0.2421534205766571,
      "grad_norm": 0.32704198360443115,
      "learning_rate": 3.17826742825621e-06,
      "loss": 1.4465,
      "step": 569
    },
    {
      "epoch": 0.24257899776571976,
      "grad_norm": 0.3004928529262543,
      "learning_rate": 3.0915103141356984e-06,
      "loss": 1.4172,
      "step": 570
    },
    {
      "epoch": 0.24300457495478242,
      "grad_norm": 0.3545885980129242,
      "learning_rate": 3.0059159640225097e-06,
      "loss": 1.3583,
      "step": 571
    },
    {
      "epoch": 0.24343015214384509,
      "grad_norm": 0.34187474846839905,
      "learning_rate": 2.92148649962275e-06,
      "loss": 1.4452,
      "step": 572
    },
    {
      "epoch": 0.24385572933290775,
      "grad_norm": 0.35106217861175537,
      "learning_rate": 2.8382240137674376e-06,
      "loss": 1.3827,
      "step": 573
    },
    {
      "epoch": 0.24428130652197041,
      "grad_norm": 0.29575130343437195,
      "learning_rate": 2.756130570360621e-06,
      "loss": 1.3322,
      "step": 574
    },
    {
      "epoch": 0.24470688371103308,
      "grad_norm": 0.3385341465473175,
      "learning_rate": 2.6752082043282277e-06,
      "loss": 1.3895,
      "step": 575
    },
    {
      "epoch": 0.24513246090009574,
      "grad_norm": 0.32416394352912903,
      "learning_rate": 2.595458921567573e-06,
      "loss": 1.3432,
      "step": 576
    },
    {
      "epoch": 0.2455580380891584,
      "grad_norm": 0.3210682272911072,
      "learning_rate": 2.51688469889772e-06,
      "loss": 1.4158,
      "step": 577
    },
    {
      "epoch": 0.24598361527822107,
      "grad_norm": 0.3364936411380768,
      "learning_rate": 2.4394874840104e-06,
      "loss": 1.4189,
      "step": 578
    },
    {
      "epoch": 0.24640919246728377,
      "grad_norm": 0.2938537001609802,
      "learning_rate": 2.3632691954217746e-06,
      "loss": 1.2996,
      "step": 579
    },
    {
      "epoch": 0.24683476965634643,
      "grad_norm": 0.31600701808929443,
      "learning_rate": 2.2882317224248883e-06,
      "loss": 1.3783,
      "step": 580
    },
    {
      "epoch": 0.2472603468454091,
      "grad_norm": 0.3460082411766052,
      "learning_rate": 2.2143769250427883e-06,
      "loss": 1.3835,
      "step": 581
    },
    {
      "epoch": 0.24768592403447176,
      "grad_norm": 0.3603164255619049,
      "learning_rate": 2.141706633982493e-06,
      "loss": 1.3649,
      "step": 582
    },
    {
      "epoch": 0.24811150122353443,
      "grad_norm": 0.3762056529521942,
      "learning_rate": 2.070222650589526e-06,
      "loss": 1.4519,
      "step": 583
    },
    {
      "epoch": 0.2485370784125971,
      "grad_norm": 0.3702375590801239,
      "learning_rate": 1.9999267468033323e-06,
      "loss": 1.337,
      "step": 584
    },
    {
      "epoch": 0.24896265560165975,
      "grad_norm": 0.3490135371685028,
      "learning_rate": 1.9308206651133365e-06,
      "loss": 1.3851,
      "step": 585
    },
    {
      "epoch": 0.24938823279072242,
      "grad_norm": 0.32408085465431213,
      "learning_rate": 1.8629061185157225e-06,
      "loss": 1.3629,
      "step": 586
    },
    {
      "epoch": 0.24981380997978508,
      "grad_norm": 0.318099707365036,
      "learning_rate": 1.7961847904710227e-06,
      "loss": 1.3742,
      "step": 587
    },
    {
      "epoch": 0.25023938716884775,
      "grad_norm": 0.3147919178009033,
      "learning_rate": 1.730658334862334e-06,
      "loss": 1.4073,
      "step": 588
    },
    {
      "epoch": 0.2506649643579104,
      "grad_norm": 0.37398630380630493,
      "learning_rate": 1.666328375954368e-06,
      "loss": 1.2967,
      "step": 589
    },
    {
      "epoch": 0.2510905415469731,
      "grad_norm": 0.36172211170196533,
      "learning_rate": 1.6031965083531609e-06,
      "loss": 1.4686,
      "step": 590
    },
    {
      "epoch": 0.25151611873603574,
      "grad_norm": 0.39417141675949097,
      "learning_rate": 1.5412642969665546e-06,
      "loss": 1.4736,
      "step": 591
    },
    {
      "epoch": 0.2519416959250984,
      "grad_norm": 0.3256359398365021,
      "learning_rate": 1.4805332769654012e-06,
      "loss": 1.4766,
      "step": 592
    },
    {
      "epoch": 0.2523672731141611,
      "grad_norm": 0.35618171095848083,
      "learning_rate": 1.4210049537455195e-06,
      "loss": 1.2675,
      "step": 593
    },
    {
      "epoch": 0.25279285030322374,
      "grad_norm": 0.31525132060050964,
      "learning_rate": 1.362680802890376e-06,
      "loss": 1.3951,
      "step": 594
    },
    {
      "epoch": 0.2532184274922864,
      "grad_norm": 0.3315390348434448,
      "learning_rate": 1.3055622701344972e-06,
      "loss": 1.4476,
      "step": 595
    },
    {
      "epoch": 0.25364400468134907,
      "grad_norm": 0.3497242033481598,
      "learning_rate": 1.2496507713276483e-06,
      "loss": 1.507,
      "step": 596
    },
    {
      "epoch": 0.25406958187041173,
      "grad_norm": 0.32933562994003296,
      "learning_rate": 1.1949476923997394e-06,
      "loss": 1.3046,
      "step": 597
    },
    {
      "epoch": 0.2544951590594744,
      "grad_norm": 0.340378999710083,
      "learning_rate": 1.141454389326435e-06,
      "loss": 1.264,
      "step": 598
    },
    {
      "epoch": 0.25492073624853706,
      "grad_norm": 0.30960968136787415,
      "learning_rate": 1.0891721880955995e-06,
      "loss": 1.3234,
      "step": 599
    },
    {
      "epoch": 0.2553463134375997,
      "grad_norm": 0.3078996241092682,
      "learning_rate": 1.038102384674383e-06,
      "loss": 1.4485,
      "step": 600
    },
    {
      "epoch": 0.2557718906266624,
      "grad_norm": 0.3417295217514038,
      "learning_rate": 9.882462449771035e-07,
      "loss": 1.3488,
      "step": 601
    },
    {
      "epoch": 0.25619746781572506,
      "grad_norm": 0.34004366397857666,
      "learning_rate": 9.3960500483391e-07,
      "loss": 1.3679,
      "step": 602
    },
    {
      "epoch": 0.2566230450047877,
      "grad_norm": 0.30232709646224976,
      "learning_rate": 8.921798699600692e-07,
      "loss": 1.2706,
      "step": 603
    },
    {
      "epoch": 0.2570486221938504,
      "grad_norm": 0.26610705256462097,
      "learning_rate": 8.459720159261719e-07,
      "loss": 1.4251,
      "step": 604
    },
    {
      "epoch": 0.25747419938291305,
      "grad_norm": 0.3400372564792633,
      "learning_rate": 8.009825881289124e-07,
      "loss": 1.3359,
      "step": 605
    },
    {
      "epoch": 0.25789977657197577,
      "grad_norm": 0.3164251744747162,
      "learning_rate": 7.572127017627329e-07,
      "loss": 1.3374,
      "step": 606
    },
    {
      "epoch": 0.25832535376103843,
      "grad_norm": 0.31121399998664856,
      "learning_rate": 7.146634417922016e-07,
      "loss": 1.4161,
      "step": 607
    },
    {
      "epoch": 0.2587509309501011,
      "grad_norm": 0.3260907828807831,
      "learning_rate": 6.733358629250619e-07,
      "loss": 1.3978,
      "step": 608
    },
    {
      "epoch": 0.25917650813916376,
      "grad_norm": 0.32060369849205017,
      "learning_rate": 6.332309895861533e-07,
      "loss": 1.3607,
      "step": 609
    },
    {
      "epoch": 0.25960208532822643,
      "grad_norm": 0.30628058314323425,
      "learning_rate": 5.943498158919536e-07,
      "loss": 1.3121,
      "step": 610
    },
    {
      "epoch": 0.2600276625172891,
      "grad_norm": 0.33873802423477173,
      "learning_rate": 5.566933056259882e-07,
      "loss": 1.4258,
      "step": 611
    },
    {
      "epoch": 0.26045323970635176,
      "grad_norm": 0.33910998702049255,
      "learning_rate": 5.202623922149152e-07,
      "loss": 1.4206,
      "step": 612
    },
    {
      "epoch": 0.2608788168954144,
      "grad_norm": 0.32269635796546936,
      "learning_rate": 4.850579787053944e-07,
      "loss": 1.4678,
      "step": 613
    },
    {
      "epoch": 0.2613043940844771,
      "grad_norm": 0.3426719903945923,
      "learning_rate": 4.510809377416936e-07,
      "loss": 1.403,
      "step": 614
    },
    {
      "epoch": 0.26172997127353975,
      "grad_norm": 0.3457522690296173,
      "learning_rate": 4.1833211154408414e-07,
      "loss": 1.4181,
      "step": 615
    },
    {
      "epoch": 0.2621555484626024,
      "grad_norm": 0.3166434168815613,
      "learning_rate": 3.8681231188791857e-07,
      "loss": 1.423,
      "step": 616
    },
    {
      "epoch": 0.2625811256516651,
      "grad_norm": 0.357695609331131,
      "learning_rate": 3.565223200835577e-07,
      "loss": 1.506,
      "step": 617
    },
    {
      "epoch": 0.26300670284072775,
      "grad_norm": 0.3549404740333557,
      "learning_rate": 3.2746288695696404e-07,
      "loss": 1.4517,
      "step": 618
    },
    {
      "epoch": 0.2634322800297904,
      "grad_norm": 0.3207859694957733,
      "learning_rate": 2.996347328311222e-07,
      "loss": 1.4455,
      "step": 619
    },
    {
      "epoch": 0.2638578572188531,
      "grad_norm": 0.3387575149536133,
      "learning_rate": 2.7303854750815316e-07,
      "loss": 1.4779,
      "step": 620
    },
    {
      "epoch": 0.26428343440791574,
      "grad_norm": 0.34494006633758545,
      "learning_rate": 2.4767499025223904e-07,
      "loss": 1.3463,
      "step": 621
    },
    {
      "epoch": 0.2647090115969784,
      "grad_norm": 0.33494681119918823,
      "learning_rate": 2.2354468977327514e-07,
      "loss": 1.23,
      "step": 622
    },
    {
      "epoch": 0.26513458878604107,
      "grad_norm": 0.3021242916584015,
      "learning_rate": 2.006482442112767e-07,
      "loss": 1.2947,
      "step": 623
    },
    {
      "epoch": 0.26556016597510373,
      "grad_norm": 0.31967541575431824,
      "learning_rate": 1.7898622112156317e-07,
      "loss": 1.4255,
      "step": 624
    },
    {
      "epoch": 0.2659857431641664,
      "grad_norm": 0.3589042127132416,
      "learning_rate": 1.5855915746068594e-07,
      "loss": 1.3617,
      "step": 625
    },
    {
      "epoch": 0.26641132035322906,
      "grad_norm": 0.3286203444004059,
      "learning_rate": 1.3936755957311143e-07,
      "loss": 1.4144,
      "step": 626
    },
    {
      "epoch": 0.26683689754229173,
      "grad_norm": 0.3432556986808777,
      "learning_rate": 1.214119031786809e-07,
      "loss": 1.4002,
      "step": 627
    },
    {
      "epoch": 0.2672624747313544,
      "grad_norm": 0.32939445972442627,
      "learning_rate": 1.0469263336082002e-07,
      "loss": 1.4409,
      "step": 628
    },
    {
      "epoch": 0.26768805192041706,
      "grad_norm": 0.3337137997150421,
      "learning_rate": 8.921016455548659e-08,
      "loss": 1.346,
      "step": 629
    },
    {
      "epoch": 0.2681136291094797,
      "grad_norm": 0.3303409516811371,
      "learning_rate": 7.496488054092865e-08,
      "loss": 1.4784,
      "step": 630
    },
    {
      "epoch": 0.2685392062985424,
      "grad_norm": 0.32682710886001587,
      "learning_rate": 6.195713442812556e-08,
      "loss": 1.4013,
      "step": 631
    },
    {
      "epoch": 0.26896478348760505,
      "grad_norm": 0.32266002893447876,
      "learning_rate": 5.0187248652094896e-08,
      "loss": 1.3807,
      "step": 632
    },
    {
      "epoch": 0.2693903606766677,
      "grad_norm": 0.37381482124328613,
      "learning_rate": 3.965551496384334e-08,
      "loss": 1.3242,
      "step": 633
    },
    {
      "epoch": 0.2698159378657304,
      "grad_norm": 0.3564024567604065,
      "learning_rate": 3.036219442317245e-08,
      "loss": 1.4303,
      "step": 634
    },
    {
      "epoch": 0.27024151505479305,
      "grad_norm": 0.3509672284126282,
      "learning_rate": 2.2307517392194944e-08,
      "loss": 1.3847,
      "step": 635
    },
    {
      "epoch": 0.2706670922438557,
      "grad_norm": 0.344876229763031,
      "learning_rate": 1.549168352961705e-08,
      "loss": 1.445,
      "step": 636
    },
    {
      "epoch": 0.2710926694329184,
      "grad_norm": 0.3283238708972931,
      "learning_rate": 9.914861785803586e-09,
      "loss": 1.4581,
      "step": 637
    },
    {
      "epoch": 0.27151824662198104,
      "grad_norm": 0.33270278573036194,
      "learning_rate": 5.577190398575738e-09,
      "loss": 1.4358,
      "step": 638
    },
    {
      "epoch": 0.2719438238110437,
      "grad_norm": 0.3181372880935669,
      "learning_rate": 2.478776889797141e-09,
      "loss": 1.3994,
      "step": 639
    },
    {
      "epoch": 0.27236940100010637,
      "grad_norm": 0.33495181798934937,
      "learning_rate": 6.196980627093396e-10,
      "loss": 1.3727,
      "step": 640
    },
    {
      "epoch": 0.27279497818916904,
      "grad_norm": 0.37345388531684875,
      "learning_rate": 0.0,
      "loss": 1.4929,
      "step": 641
    }
  ],
  "logging_steps": 1,
  "max_steps": 641,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.377339310411612e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}