{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9972122660294703,
  "eval_steps": 500,
  "global_step": 313,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 10.092560768127441,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.9497,
      "step": 1
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.963287353515625,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.951,
      "step": 2
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.27600383758545,
      "learning_rate": 3e-06,
      "loss": 1.9345,
      "step": 3
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.632405757904053,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.9157,
      "step": 4
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.668580055236816,
      "learning_rate": 5e-06,
      "loss": 1.9026,
      "step": 5
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.8046553134918213,
      "learning_rate": 6e-06,
      "loss": 1.8923,
      "step": 6
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.357985973358154,
      "learning_rate": 7e-06,
      "loss": 1.8241,
      "step": 7
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.685062885284424,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.8467,
      "step": 8
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.768229961395264,
      "learning_rate": 9e-06,
      "loss": 1.8199,
      "step": 9
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.796407699584961,
      "learning_rate": 1e-05,
      "loss": 1.8374,
      "step": 10
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.7536139488220215,
      "learning_rate": 9.999731248679734e-06,
      "loss": 1.779,
      "step": 11
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.202663898468018,
      "learning_rate": 9.99892502360984e-06,
      "loss": 1.7579,
      "step": 12
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9114131927490234,
      "learning_rate": 9.99758141145994e-06,
      "loss": 1.7433,
      "step": 13
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3823723793029785,
      "learning_rate": 9.995700556669052e-06,
      "loss": 1.7212,
      "step": 14
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3254876136779785,
      "learning_rate": 9.993282661430058e-06,
      "loss": 1.7218,
      "step": 15
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.053166151046753,
      "learning_rate": 9.990327985667972e-06,
      "loss": 1.7256,
      "step": 16
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3782012462615967,
      "learning_rate": 9.986836847012001e-06,
      "loss": 1.713,
      "step": 17
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1311683654785156,
      "learning_rate": 9.98280962076139e-06,
      "loss": 1.6785,
      "step": 18
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.325747489929199,
      "learning_rate": 9.978246739845095e-06,
      "loss": 1.7167,
      "step": 19
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1330366134643555,
      "learning_rate": 9.973148694775217e-06,
      "loss": 1.676,
      "step": 20
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2632806301116943,
      "learning_rate": 9.967516033594295e-06,
      "loss": 1.7033,
      "step": 21
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6582744121551514,
      "learning_rate": 9.961349361816384e-06,
      "loss": 1.6957,
      "step": 22
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9663337469100952,
      "learning_rate": 9.954649342361952e-06,
      "loss": 1.6729,
      "step": 23
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2557435035705566,
      "learning_rate": 9.947416695486633e-06,
      "loss": 1.6399,
      "step": 24
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.090054512023926,
      "learning_rate": 9.939652198703785e-06,
      "loss": 1.6792,
      "step": 25
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1648776531219482,
      "learning_rate": 9.93135668670091e-06,
      "loss": 1.6625,
      "step": 26
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1209850311279297,
      "learning_rate": 9.92253105124993e-06,
      "loss": 1.7057,
      "step": 27
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2777392864227295,
      "learning_rate": 9.91317624111132e-06,
      "loss": 1.6052,
      "step": 28
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1739561557769775,
      "learning_rate": 9.903293261932106e-06,
      "loss": 1.6139,
      "step": 29
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4518871307373047,
      "learning_rate": 9.89288317613777e-06,
      "loss": 1.6759,
      "step": 30
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.795519471168518,
      "learning_rate": 9.881947102818036e-06,
      "loss": 1.7036,
      "step": 31
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.150223731994629,
      "learning_rate": 9.870486217606557e-06,
      "loss": 1.6469,
      "step": 32
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.91805899143219,
      "learning_rate": 9.858501752554548e-06,
      "loss": 1.6353,
      "step": 33
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8129810094833374,
      "learning_rate": 9.845994995998332e-06,
      "loss": 1.6551,
      "step": 34
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1308679580688477,
      "learning_rate": 9.83296729242084e-06,
      "loss": 1.617,
      "step": 35
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8321012258529663,
      "learning_rate": 9.819420042307091e-06,
      "loss": 1.6213,
      "step": 36
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9112164974212646,
      "learning_rate": 9.805354701993624e-06,
      "loss": 1.6245,
      "step": 37
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1160471439361572,
      "learning_rate": 9.79077278351195e-06,
      "loss": 1.6405,
      "step": 38
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6318371295928955,
      "learning_rate": 9.77567585442601e-06,
      "loss": 1.6234,
      "step": 39
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.238373279571533,
      "learning_rate": 9.76006553766365e-06,
      "loss": 1.6452,
      "step": 40
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.056736469268799,
      "learning_rate": 9.743943511342168e-06,
      "loss": 1.679,
      "step": 41
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.000826358795166,
      "learning_rate": 9.727311508587907e-06,
      "loss": 1.6904,
      "step": 42
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.089482069015503,
      "learning_rate": 9.710171317349946e-06,
      "loss": 1.62,
      "step": 43
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5748705863952637,
      "learning_rate": 9.692524780207897e-06,
      "loss": 1.6669,
      "step": 44
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9039987325668335,
      "learning_rate": 9.674373794173818e-06,
      "loss": 1.6489,
      "step": 45
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8047181367874146,
      "learning_rate": 9.655720310488298e-06,
      "loss": 1.6227,
      "step": 46
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0552868843078613,
      "learning_rate": 9.636566334410682e-06,
      "loss": 1.5898,
      "step": 47
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.252218008041382,
      "learning_rate": 9.616913925003514e-06,
      "loss": 1.6667,
      "step": 48
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.015887498855591,
      "learning_rate": 9.596765194911182e-06,
      "loss": 1.6668,
      "step": 49
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.582007884979248,
      "learning_rate": 9.576122310132814e-06,
      "loss": 1.6542,
      "step": 50
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9070011377334595,
      "learning_rate": 9.554987489789426e-06,
      "loss": 1.691,
      "step": 51
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9671483039855957,
      "learning_rate": 9.533363005885362e-06,
      "loss": 1.6116,
      "step": 52
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4808287620544434,
      "learning_rate": 9.511251183064068e-06,
      "loss": 1.6653,
      "step": 53
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.237518548965454,
      "learning_rate": 9.48865439835817e-06,
      "loss": 1.6571,
      "step": 54
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.142627477645874,
      "learning_rate": 9.465575080933959e-06,
      "loss": 1.6247,
      "step": 55
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0365779399871826,
      "learning_rate": 9.442015711830246e-06,
      "loss": 1.6323,
      "step": 56
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.427443742752075,
      "learning_rate": 9.417978823691652e-06,
      "loss": 1.6646,
      "step": 57
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.014340877532959,
      "learning_rate": 9.393467000496345e-06,
      "loss": 1.5754,
      "step": 58
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6110949516296387,
      "learning_rate": 9.368482877278264e-06,
      "loss": 1.657,
      "step": 59
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.048231363296509,
      "learning_rate": 9.34302913984385e-06,
      "loss": 1.61,
      "step": 60
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.672330379486084,
      "learning_rate": 9.317108524483319e-06,
      "loss": 1.5894,
      "step": 61
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6745121479034424,
      "learning_rate": 9.29072381767651e-06,
      "loss": 1.5623,
      "step": 62
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1490442752838135,
      "learning_rate": 9.26387785579334e-06,
      "loss": 1.6668,
      "step": 63
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.805147171020508,
      "learning_rate": 9.236573524788888e-06,
      "loss": 1.6274,
      "step": 64
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.243023157119751,
      "learning_rate": 9.208813759893158e-06,
      "loss": 1.6496,
      "step": 65
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.5260114669799805,
      "learning_rate": 9.180601545295535e-06,
      "loss": 1.623,
      "step": 66
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0329606533050537,
      "learning_rate": 9.151939913823988e-06,
      "loss": 1.5873,
      "step": 67
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3269598484039307,
      "learning_rate": 9.122831946619038e-06,
      "loss": 1.6327,
      "step": 68
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1953978538513184,
      "learning_rate": 9.093280772802527e-06,
      "loss": 1.6321,
      "step": 69
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.2001328468322754,
      "learning_rate": 9.063289569141251e-06,
      "loss": 1.6338,
      "step": 70
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.076502799987793,
      "learning_rate": 9.032861559705442e-06,
      "loss": 1.6648,
      "step": 71
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0800352096557617,
      "learning_rate": 9.002000015522182e-06,
      "loss": 1.5694,
      "step": 72
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0901432037353516,
      "learning_rate": 8.970708254223768e-06,
      "loss": 1.6571,
      "step": 73
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1389923095703125,
      "learning_rate": 8.938989639691068e-06,
      "loss": 1.6151,
      "step": 74
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0514352321624756,
      "learning_rate": 8.90684758169189e-06,
      "loss": 1.618,
      "step": 75
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9462100267410278,
      "learning_rate": 8.87428553551445e-06,
      "loss": 1.6367,
      "step": 76
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2343838214874268,
      "learning_rate": 8.841307001595904e-06,
      "loss": 1.6321,
      "step": 77
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.6064207553863525,
      "learning_rate": 8.807915525146065e-06,
      "loss": 1.5913,
      "step": 78
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.589186191558838,
      "learning_rate": 8.774114695766286e-06,
      "loss": 1.6,
      "step": 79
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0082924365997314,
      "learning_rate": 8.739908147063576e-06,
      "loss": 1.6266,
      "step": 80
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9739817380905151,
      "learning_rate": 8.705299556259986e-06,
      "loss": 1.5926,
      "step": 81
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9566433429718018,
      "learning_rate": 8.670292643797302e-06,
      "loss": 1.624,
      "step": 82
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.10595703125,
      "learning_rate": 8.634891172937102e-06,
      "loss": 1.6437,
      "step": 83
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7080891132354736,
      "learning_rate": 8.599098949356201e-06,
      "loss": 1.6131,
      "step": 84
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.358856678009033,
      "learning_rate": 8.562919820737537e-06,
      "loss": 1.5748,
      "step": 85
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.030383825302124,
      "learning_rate": 8.526357676356538e-06,
      "loss": 1.6389,
      "step": 86
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3567280769348145,
      "learning_rate": 8.489416446663037e-06,
      "loss": 1.6288,
      "step": 87
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1183040142059326,
      "learning_rate": 8.452100102858734e-06,
      "loss": 1.5929,
      "step": 88
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.467944860458374,
      "learning_rate": 8.414412656470297e-06,
      "loss": 1.5579,
      "step": 89
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.200274705886841,
      "learning_rate": 8.376358158918114e-06,
      "loss": 1.587,
      "step": 90
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3279902935028076,
      "learning_rate": 8.33794070108077e-06,
      "loss": 1.6496,
      "step": 91
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7403910160064697,
      "learning_rate": 8.299164412855268e-06,
      "loss": 1.5665,
      "step": 92
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5097343921661377,
      "learning_rate": 8.260033462713073e-06,
      "loss": 1.6245,
      "step": 93
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2657604217529297,
      "learning_rate": 8.22055205725199e-06,
      "loss": 1.5785,
      "step": 94
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4038476943969727,
      "learning_rate": 8.180724440743957e-06,
      "loss": 1.6912,
      "step": 95
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.811049461364746,
      "learning_rate": 8.14055489467878e-06,
      "loss": 1.6059,
      "step": 96
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.474153518676758,
      "learning_rate": 8.100047737303877e-06,
      "loss": 1.6423,
      "step": 97
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.351736307144165,
      "learning_rate": 8.059207323160057e-06,
      "loss": 1.6199,
      "step": 98
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3526997566223145,
      "learning_rate": 8.018038042613407e-06,
      "loss": 1.5856,
      "step": 99
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.075406789779663,
      "learning_rate": 7.97654432138333e-06,
      "loss": 1.5901,
      "step": 100
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6973438262939453,
      "learning_rate": 7.93473062006677e-06,
      "loss": 1.5696,
      "step": 101
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6449167728424072,
      "learning_rate": 7.892601433658705e-06,
      "loss": 1.5939,
      "step": 102
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.360849380493164,
      "learning_rate": 7.850161291068915e-06,
      "loss": 1.5449,
      "step": 103
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.467226028442383,
      "learning_rate": 7.807414754635145e-06,
      "loss": 1.5926,
      "step": 104
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.449989080429077,
      "learning_rate": 7.764366419632636e-06,
      "loss": 1.5591,
      "step": 105
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.518803119659424,
      "learning_rate": 7.721020913780137e-06,
      "loss": 1.5406,
      "step": 106
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.6841485500335693,
      "learning_rate": 7.677382896742417e-06,
      "loss": 1.5797,
      "step": 107
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.475085496902466,
      "learning_rate": 7.63345705962935e-06,
      "loss": 1.5653,
      "step": 108
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3905832767486572,
      "learning_rate": 7.589248124491627e-06,
      "loss": 1.5641,
      "step": 109
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7013638019561768,
      "learning_rate": 7.544760843813122e-06,
      "loss": 1.5837,
      "step": 110
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.500148296356201,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.6255,
      "step": 111
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.106887102127075,
      "learning_rate": 7.454970404866612e-06,
      "loss": 1.6069,
      "step": 112
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1840200424194336,
      "learning_rate": 7.409676899118213e-06,
      "loss": 1.5959,
      "step": 113
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.758022308349609,
      "learning_rate": 7.3641243518305915e-06,
      "loss": 1.616,
      "step": 114
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.24625301361084,
      "learning_rate": 7.318317659926637e-06,
      "loss": 1.639,
      "step": 115
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8651440143585205,
      "learning_rate": 7.272261747649922e-06,
      "loss": 1.6287,
      "step": 116
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9262839555740356,
      "learning_rate": 7.225961566035335e-06,
      "loss": 1.5851,
      "step": 117
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4312078952789307,
      "learning_rate": 7.179422092376856e-06,
      "loss": 1.5934,
      "step": 118
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.391693592071533,
      "learning_rate": 7.132648329692478e-06,
      "loss": 1.5719,
      "step": 119
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.10438871383667,
      "learning_rate": 7.085645306186391e-06,
      "loss": 1.5876,
      "step": 120
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.984710931777954,
      "learning_rate": 7.038418074708444e-06,
      "loss": 1.6506,
      "step": 121
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4563817977905273,
      "learning_rate": 6.990971712210966e-06,
      "loss": 1.5578,
      "step": 122
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.856941819190979,
      "learning_rate": 6.943311319202976e-06,
      "loss": 1.5806,
      "step": 123
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.228983163833618,
      "learning_rate": 6.895442019201898e-06,
      "loss": 1.5639,
      "step": 124
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9688351154327393,
      "learning_rate": 6.8473689581827585e-06,
      "loss": 1.5939,
      "step": 125
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.028876304626465,
      "learning_rate": 6.7990973040250055e-06,
      "loss": 1.6096,
      "step": 126
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.385794162750244,
      "learning_rate": 6.750632245956954e-06,
      "loss": 1.5617,
      "step": 127
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0037996768951416,
      "learning_rate": 6.701978993997942e-06,
      "loss": 1.6028,
      "step": 128
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0149664878845215,
      "learning_rate": 6.653142778398247e-06,
      "loss": 1.583,
      "step": 129
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.466585874557495,
      "learning_rate": 6.6041288490768385e-06,
      "loss": 1.6368,
      "step": 130
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.181319236755371,
      "learning_rate": 6.554942475057003e-06,
      "loss": 1.5819,
      "step": 131
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9432387351989746,
      "learning_rate": 6.505588943899923e-06,
      "loss": 1.5551,
      "step": 132
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.277068614959717,
      "learning_rate": 6.456073561136261e-06,
      "loss": 1.5788,
      "step": 133
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9018107652664185,
      "learning_rate": 6.406401649695814e-06,
      "loss": 1.5639,
      "step": 134
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.755958318710327,
      "learning_rate": 6.356578549335295e-06,
      "loss": 1.6015,
      "step": 135
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2761142253875732,
      "learning_rate": 6.306609616064304e-06,
      "loss": 1.6054,
      "step": 136
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.212486505508423,
      "learning_rate": 6.256500221569556e-06,
      "loss": 1.5953,
      "step": 137
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2175703048706055,
      "learning_rate": 6.2062557526374226e-06,
      "loss": 1.5745,
      "step": 138
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9958215951919556,
      "learning_rate": 6.15588161057485e-06,
      "loss": 1.5981,
      "step": 139
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4180896282196045,
      "learning_rate": 6.10538321062871e-06,
      "loss": 1.5836,
      "step": 140
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0187716484069824,
      "learning_rate": 6.0547659814036664e-06,
      "loss": 1.6062,
      "step": 141
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3086986541748047,
      "learning_rate": 6.004035364278593e-06,
      "loss": 1.5764,
      "step": 142
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9949414730072021,
      "learning_rate": 5.953196812821622e-06,
      "loss": 1.6037,
      "step": 143
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8632932901382446,
      "learning_rate": 5.902255792203882e-06,
      "loss": 1.6051,
      "step": 144
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.034745931625366,
      "learning_rate": 5.851217778611994e-06,
      "loss": 1.6386,
      "step": 145
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1993603706359863,
      "learning_rate": 5.800088258659371e-06,
      "loss": 1.6081,
      "step": 146
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0585718154907227,
      "learning_rate": 5.748872728796409e-06,
      "loss": 1.5955,
      "step": 147
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2753782272338867,
      "learning_rate": 5.697576694719616e-06,
      "loss": 1.5655,
      "step": 148
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.237483024597168,
      "learning_rate": 5.646205670779745e-06,
      "loss": 1.6004,
      "step": 149
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.019814968109131,
      "learning_rate": 5.594765179389003e-06,
      "loss": 1.5721,
      "step": 150
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3698084354400635,
      "learning_rate": 5.543260750427373e-06,
      "loss": 1.5946,
      "step": 151
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.441636085510254,
      "learning_rate": 5.4916979206481745e-06,
      "loss": 1.5522,
      "step": 152
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.149003267288208,
      "learning_rate": 5.440082233082837e-06,
      "loss": 1.5935,
      "step": 153
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.03369402885437,
      "learning_rate": 5.388419236445033e-06,
      "loss": 1.5912,
      "step": 154
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9619389772415161,
      "learning_rate": 5.336714484534183e-06,
      "loss": 1.5324,
      "step": 155
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.591277837753296,
      "learning_rate": 5.284973535638424e-06,
      "loss": 1.5662,
      "step": 156
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.212984323501587,
      "learning_rate": 5.233201951937088e-06,
      "loss": 1.6589,
      "step": 157
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.835900902748108,
      "learning_rate": 5.181405298902763e-06,
      "loss": 1.597,
      "step": 158
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.046597480773926,
      "learning_rate": 5.1295891447030056e-06,
      "loss": 1.5742,
      "step": 159
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5161941051483154,
      "learning_rate": 5.077759059601756e-06,
      "loss": 1.5848,
      "step": 160
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1808998584747314,
      "learning_rate": 5.025920615360532e-06,
      "loss": 1.5684,
      "step": 161
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5023722648620605,
      "learning_rate": 4.974079384639469e-06,
      "loss": 1.5815,
      "step": 162
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.056931734085083,
      "learning_rate": 4.922240940398246e-06,
      "loss": 1.5334,
      "step": 163
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1214613914489746,
      "learning_rate": 4.870410855296994e-06,
      "loss": 1.5726,
      "step": 164
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.080734968185425,
      "learning_rate": 4.818594701097239e-06,
      "loss": 1.5869,
      "step": 165
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.13873028755188,
      "learning_rate": 4.766798048062913e-06,
      "loss": 1.6214,
      "step": 166
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4381258487701416,
      "learning_rate": 4.715026464361576e-06,
      "loss": 1.5921,
      "step": 167
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3654768466949463,
      "learning_rate": 4.663285515465818e-06,
      "loss": 1.5598,
      "step": 168
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3201935291290283,
      "learning_rate": 4.611580763554969e-06,
      "loss": 1.5339,
      "step": 169
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3482742309570312,
      "learning_rate": 4.559917766917166e-06,
      "loss": 1.5501,
      "step": 170
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.296869993209839,
      "learning_rate": 4.508302079351827e-06,
      "loss": 1.542,
      "step": 171
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0012331008911133,
      "learning_rate": 4.456739249572628e-06,
      "loss": 1.5223,
      "step": 172
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4259610176086426,
      "learning_rate": 4.405234820611001e-06,
      "loss": 1.5929,
      "step": 173
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1162846088409424,
      "learning_rate": 4.3537943292202555e-06,
      "loss": 1.5213,
      "step": 174
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.398637771606445,
      "learning_rate": 4.3024233052803855e-06,
      "loss": 1.5698,
      "step": 175
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.08495831489563,
      "learning_rate": 4.251127271203593e-06,
      "loss": 1.6086,
      "step": 176
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.053515911102295,
      "learning_rate": 4.199911741340631e-06,
      "loss": 1.6036,
      "step": 177
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.257359266281128,
      "learning_rate": 4.148782221388007e-06,
      "loss": 1.6246,
      "step": 178
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1245572566986084,
      "learning_rate": 4.097744207796119e-06,
      "loss": 1.5544,
      "step": 179
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.556488275527954,
      "learning_rate": 4.04680318717838e-06,
      "loss": 1.5516,
      "step": 180
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.438620090484619,
      "learning_rate": 3.995964635721409e-06,
      "loss": 1.5524,
      "step": 181
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9453068971633911,
      "learning_rate": 3.945234018596335e-06,
      "loss": 1.5657,
      "step": 182
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.2514045238494873,
      "learning_rate": 3.8946167893712916e-06,
      "loss": 1.5916,
      "step": 183
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2144711017608643,
      "learning_rate": 3.844118389425154e-06,
      "loss": 1.6051,
      "step": 184
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4618723392486572,
      "learning_rate": 3.7937442473625787e-06,
      "loss": 1.5841,
      "step": 185
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2561230659484863,
      "learning_rate": 3.743499778430445e-06,
      "loss": 1.5469,
      "step": 186
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2580950260162354,
      "learning_rate": 3.6933903839356983e-06,
      "loss": 1.5618,
      "step": 187
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8674182891845703,
      "learning_rate": 3.6434214506647064e-06,
      "loss": 1.5266,
      "step": 188
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.33337664604187,
      "learning_rate": 3.5935983503041864e-06,
      "loss": 1.5223,
      "step": 189
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.912142038345337,
      "learning_rate": 3.5439264388637407e-06,
      "loss": 1.5331,
      "step": 190
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.8374595642089844,
      "learning_rate": 3.4944110561000785e-06,
      "loss": 1.6068,
      "step": 191
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.329653024673462,
      "learning_rate": 3.4450575249429975e-06,
      "loss": 1.5324,
      "step": 192
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3587117195129395,
      "learning_rate": 3.3958711509231627e-06,
      "loss": 1.5542,
      "step": 193
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.657672643661499,
      "learning_rate": 3.3468572216017536e-06,
      "loss": 1.5755,
      "step": 194
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.17305064201355,
      "learning_rate": 3.29802100600206e-06,
      "loss": 1.5489,
      "step": 195
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5755293369293213,
      "learning_rate": 3.249367754043047e-06,
      "loss": 1.5835,
      "step": 196
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.943556070327759,
      "learning_rate": 3.200902695974995e-06,
      "loss": 1.6003,
      "step": 197
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.425208568572998,
      "learning_rate": 3.152631041817244e-06,
      "loss": 1.5649,
      "step": 198
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0712738037109375,
      "learning_rate": 3.104557980798104e-06,
      "loss": 1.5363,
      "step": 199
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3942067623138428,
      "learning_rate": 3.056688680797024e-06,
      "loss": 1.5522,
      "step": 200
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.139435052871704,
      "learning_rate": 3.0090282877890376e-06,
      "loss": 1.5759,
      "step": 201
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4167027473449707,
      "learning_rate": 2.961581925291557e-06,
      "loss": 1.5254,
      "step": 202
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9939026832580566,
      "learning_rate": 2.9143546938136093e-06,
      "loss": 1.5688,
      "step": 203
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8560909032821655,
      "learning_rate": 2.8673516703075247e-06,
      "loss": 1.5346,
      "step": 204
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.581928730010986,
      "learning_rate": 2.820577907623145e-06,
      "loss": 1.5447,
      "step": 205
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.128434419631958,
      "learning_rate": 2.7740384339646655e-06,
      "loss": 1.5557,
      "step": 206
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.027505397796631,
      "learning_rate": 2.7277382523500804e-06,
      "loss": 1.527,
      "step": 207
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5379273891448975,
      "learning_rate": 2.6816823400733628e-06,
      "loss": 1.5794,
      "step": 208
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.262667655944824,
      "learning_rate": 2.6358756481694115e-06,
      "loss": 1.5452,
      "step": 209
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3182244300842285,
      "learning_rate": 2.5903231008817888e-06,
      "loss": 1.5354,
      "step": 210
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3638322353363037,
      "learning_rate": 2.5450295951333896e-06,
      "loss": 1.5699,
      "step": 211
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2807657718658447,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 1.5958,
      "step": 212
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4435348510742188,
      "learning_rate": 2.4552391561868783e-06,
      "loss": 1.5601,
      "step": 213
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2643187046051025,
      "learning_rate": 2.410751875508373e-06,
      "loss": 1.5474,
      "step": 214
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2094953060150146,
      "learning_rate": 2.3665429403706506e-06,
      "loss": 1.5888,
      "step": 215
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.147650718688965,
      "learning_rate": 2.3226171032575856e-06,
      "loss": 1.5937,
      "step": 216
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9609376192092896,
      "learning_rate": 2.278979086219863e-06,
      "loss": 1.5873,
      "step": 217
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.404101610183716,
      "learning_rate": 2.2356335803673655e-06,
      "loss": 1.5707,
      "step": 218
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1139962673187256,
      "learning_rate": 2.192585245364856e-06,
      "loss": 1.5839,
      "step": 219
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.373765707015991,
      "learning_rate": 2.149838708931087e-06,
      "loss": 1.5289,
      "step": 220
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1613967418670654,
      "learning_rate": 2.1073985663412984e-06,
      "loss": 1.5599,
      "step": 221
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.6613214015960693,
      "learning_rate": 2.0652693799332286e-06,
      "loss": 1.5657,
      "step": 222
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8200939893722534,
      "learning_rate": 2.0234556786166715e-06,
      "loss": 1.4937,
      "step": 223
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8578108549118042,
      "learning_rate": 1.9819619573865932e-06,
      "loss": 1.5268,
      "step": 224
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.355733633041382,
      "learning_rate": 1.9407926768399456e-06,
      "loss": 1.5658,
      "step": 225
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.7907326221466064,
      "learning_rate": 1.8999522626961254e-06,
      "loss": 1.5777,
      "step": 226
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2081241607666016,
      "learning_rate": 1.859445105321221e-06,
      "loss": 1.6039,
      "step": 227
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.7307095527648926,
      "learning_rate": 1.8192755592560446e-06,
      "loss": 1.554,
      "step": 228
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.100717306137085,
      "learning_rate": 1.7794479427480115e-06,
      "loss": 1.5687,
      "step": 229
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.900230884552002,
      "learning_rate": 1.739966537286929e-06,
      "loss": 1.5894,
      "step": 230
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.7056214809417725,
      "learning_rate": 1.7008355871447345e-06,
      "loss": 1.5262,
      "step": 231
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2790305614471436,
      "learning_rate": 1.6620592989192318e-06,
      "loss": 1.5334,
      "step": 232
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.336102247238159,
      "learning_rate": 1.6236418410818872e-06,
      "loss": 1.6231,
      "step": 233
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4769365787506104,
      "learning_rate": 1.5855873435297042e-06,
      "loss": 1.5542,
      "step": 234
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2314493656158447,
      "learning_rate": 1.5478998971412669e-06,
      "loss": 1.582,
      "step": 235
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.37534236907959,
      "learning_rate": 1.510583553336964e-06,
      "loss": 1.5618,
      "step": 236
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2861831188201904,
      "learning_rate": 1.473642323643465e-06,
      "loss": 1.5557,
      "step": 237
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0637292861938477,
      "learning_rate": 1.4370801792624656e-06,
      "loss": 1.5812,
      "step": 238
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.4701573848724365,
      "learning_rate": 1.4009010506437997e-06,
      "loss": 1.5535,
      "step": 239
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5278735160827637,
      "learning_rate": 1.3651088270628992e-06,
      "loss": 1.5792,
      "step": 240
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0601749420166016,
      "learning_rate": 1.3297073562026992e-06,
      "loss": 1.5739,
      "step": 241
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.125802755355835,
      "learning_rate": 1.2947004437400161e-06,
      "loss": 1.579,
      "step": 242
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.429272413253784,
      "learning_rate": 1.2600918529364253e-06,
      "loss": 1.5575,
      "step": 243
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.331495761871338,
      "learning_rate": 1.225885304233716e-06,
      "loss": 1.5796,
      "step": 244
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5032925605773926,
      "learning_rate": 1.1920844748539373e-06,
      "loss": 1.5205,
      "step": 245
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1888771057128906,
      "learning_rate": 1.1586929984040974e-06,
      "loss": 1.5877,
      "step": 246
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.270721673965454,
      "learning_rate": 1.125714464485551e-06,
      "loss": 1.5382,
      "step": 247
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.260922431945801,
      "learning_rate": 1.0931524183081105e-06,
      "loss": 1.585,
      "step": 248
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.52512526512146,
      "learning_rate": 1.0610103603089345e-06,
      "loss": 1.5353,
      "step": 249
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.214529275894165,
      "learning_rate": 1.0292917457762325e-06,
      "loss": 1.6081,
      "step": 250
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.466500997543335,
      "learning_rate": 9.979999844778203e-07,
      "loss": 1.5661,
      "step": 251
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2533483505249023,
      "learning_rate": 9.671384402945588e-07,
      "loss": 1.5718,
      "step": 252
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9847800731658936,
      "learning_rate": 9.367104308587493e-07,
      "loss": 1.561,
      "step": 253
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.6065824031829834,
      "learning_rate": 9.06719227197474e-07,
      "loss": 1.5482,
      "step": 254
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5435142517089844,
      "learning_rate": 8.771680533809634e-07,
      "loss": 1.5918,
      "step": 255
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.8875584602355957,
      "learning_rate": 8.480600861760124e-07,
      "loss": 1.6082,
      "step": 256
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1082193851470947,
      "learning_rate": 8.193984547044659e-07,
      "loss": 1.594,
      "step": 257
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4410855770111084,
      "learning_rate": 7.911862401068431e-07,
      "loss": 1.5938,
      "step": 258
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.8396708965301514,
      "learning_rate": 7.634264752111131e-07,
      "loss": 1.6215,
      "step": 259
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.493769407272339,
      "learning_rate": 7.361221442066607e-07,
      "loss": 1.5734,
      "step": 260
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8893070220947266,
      "learning_rate": 7.092761823234911e-07,
      "loss": 1.5264,
      "step": 261
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.5776591300964355,
      "learning_rate": 6.828914755166826e-07,
      "loss": 1.5733,
      "step": 262
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.7408130168914795,
      "learning_rate": 6.569708601561515e-07,
      "loss": 1.5617,
      "step": 263
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9031057357788086,
      "learning_rate": 6.315171227217365e-07,
      "loss": 1.5979,
      "step": 264
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.713923931121826,
      "learning_rate": 6.065329995036573e-07,
      "loss": 1.5364,
      "step": 265
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8765623569488525,
      "learning_rate": 5.820211763083494e-07,
      "loss": 1.5547,
      "step": 266
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.5959739685058594,
      "learning_rate": 5.579842881697556e-07,
      "loss": 1.5372,
      "step": 267
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1984755992889404,
      "learning_rate": 5.344249190660427e-07,
      "loss": 1.6001,
      "step": 268
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3963303565979004,
      "learning_rate": 5.113456016418305e-07,
      "loss": 1.6344,
      "step": 269
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.728706359863281,
      "learning_rate": 4.88748816935934e-07,
      "loss": 1.5381,
      "step": 270
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.736025333404541,
      "learning_rate": 4.666369941146376e-07,
      "loss": 1.5458,
      "step": 271
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.1046762466430664,
      "learning_rate": 4.4501251021057566e-07,
      "loss": 1.5871,
      "step": 272
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.1005470752716064,
      "learning_rate": 4.2387768986718644e-07,
      "loss": 1.545,
      "step": 273
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.466526985168457,
      "learning_rate": 4.03234805088818e-07,
      "loss": 1.5337,
      "step": 274
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.6507673263549805,
      "learning_rate": 3.8308607499648765e-07,
      "loss": 1.5436,
      "step": 275
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5536324977874756,
      "learning_rate": 3.634336655893189e-07,
      "loss": 1.5853,
      "step": 276
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.4438483715057373,
      "learning_rate": 3.4427968951170287e-07,
      "loss": 1.5655,
      "step": 277
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.799037456512451,
      "learning_rate": 3.256262058261816e-07,
      "loss": 1.6025,
      "step": 278
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.272508144378662,
      "learning_rate": 3.0747521979210436e-07,
      "loss": 1.5339,
      "step": 279
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.6367502212524414,
      "learning_rate": 2.8982868265005457e-07,
      "loss": 1.487,
      "step": 280
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.486656665802002,
      "learning_rate": 2.726884914120936e-07,
      "loss": 1.5738,
      "step": 281
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.579956293106079,
      "learning_rate": 2.5605648865783315e-07,
      "loss": 1.5955,
      "step": 282
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.506580114364624,
      "learning_rate": 2.399344623363503e-07,
      "loss": 1.5778,
      "step": 283
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4960708618164062,
      "learning_rate": 2.2432414557399197e-07,
      "loss": 1.5922,
      "step": 284
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.1219966411590576,
      "learning_rate": 2.0922721648805045e-07,
      "loss": 1.5548,
      "step": 285
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.500851631164551,
      "learning_rate": 1.9464529800637731e-07,
      "loss": 1.5491,
      "step": 286
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.394024133682251,
      "learning_rate": 1.805799576929107e-07,
      "loss": 1.5433,
      "step": 287
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6824209690093994,
      "learning_rate": 1.6703270757916e-07,
      "loss": 1.5465,
      "step": 288
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.4885001182556152,
      "learning_rate": 1.540050040016694e-07,
      "loss": 1.5682,
      "step": 289
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.021599769592285,
      "learning_rate": 1.414982474454524e-07,
      "loss": 1.5516,
      "step": 290
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3498406410217285,
      "learning_rate": 1.2951378239344337e-07,
      "loss": 1.5837,
      "step": 291
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3095269203186035,
      "learning_rate": 1.1805289718196499e-07,
      "loss": 1.5519,
      "step": 292
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.113964080810547,
      "learning_rate": 1.0711682386222943e-07,
      "loss": 1.5897,
      "step": 293
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6248152256011963,
      "learning_rate": 9.670673806789543e-08,
      "loss": 1.5596,
      "step": 294
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.7190983295440674,
      "learning_rate": 8.682375888868167e-08,
      "loss": 1.5829,
      "step": 295
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4610188007354736,
      "learning_rate": 7.746894875007016e-08,
      "loss": 1.6058,
      "step": 296
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.32197904586792,
      "learning_rate": 6.864331329909102e-08,
      "loss": 1.6013,
      "step": 297
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.1023738384246826,
      "learning_rate": 6.034780129621664e-08,
      "loss": 1.5767,
      "step": 298
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.545488119125366,
      "learning_rate": 5.258330451336724e-08,
      "loss": 1.5267,
      "step": 299
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.106576442718506,
      "learning_rate": 4.535065763804802e-08,
      "loss": 1.5447,
      "step": 300
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0030553340911865,
      "learning_rate": 3.8650638183617695e-08,
      "loss": 1.5714,
      "step": 301
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.5803191661834717,
      "learning_rate": 3.248396640570528e-08,
      "loss": 1.5273,
      "step": 302
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.743903160095215,
      "learning_rate": 2.685130522478485e-08,
      "loss": 1.5455,
      "step": 303
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4540319442749023,
      "learning_rate": 2.1753260154906973e-08,
      "loss": 1.5372,
      "step": 304
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.002195358276367,
      "learning_rate": 1.7190379238609666e-08,
      "loss": 1.5892,
      "step": 305
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3258426189422607,
      "learning_rate": 1.3163152988000527e-08,
      "loss": 1.5832,
      "step": 306
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.36910343170166,
      "learning_rate": 9.672014332028357e-09,
      "loss": 1.5319,
      "step": 307
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2311582565307617,
      "learning_rate": 6.717338569942611e-09,
      "loss": 1.5475,
      "step": 308
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.657569646835327,
      "learning_rate": 4.299443330947895e-09,
      "loss": 1.5227,
      "step": 309
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1514527797698975,
      "learning_rate": 2.4185885400596076e-09,
      "loss": 1.5543,
      "step": 310
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7922263145446777,
      "learning_rate": 1.0749763901607425e-09,
      "loss": 1.575,
      "step": 311
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.211779832839966,
      "learning_rate": 2.6875132026760173e-10,
      "loss": 1.5689,
      "step": 312
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.225865125656128,
      "learning_rate": 0.0,
      "loss": 1.5786,
      "step": 313
    },
    {
      "epoch": 1.0,
      "step": 313,
      "total_flos": 399880339259392.0,
      "train_loss": 1.6020259019284964,
      "train_runtime": 68378.6303,
      "train_samples_per_second": 0.588,
      "train_steps_per_second": 0.005
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 313,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "total_flos": 399880339259392.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}