{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.761904761904762,
  "eval_steps": 500,
  "global_step": 165,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.016931216931216932,
      "grad_norm": 2322.343994140625,
      "learning_rate": 0.0,
      "loss": 1.2269,
      "step": 1
    },
    {
      "epoch": 0.033862433862433865,
      "grad_norm": 12.033977508544922,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.2059,
      "step": 2
    },
    {
      "epoch": 0.050793650793650794,
      "grad_norm": 10.809624671936035,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.2198,
      "step": 3
    },
    {
      "epoch": 0.06772486772486773,
      "grad_norm": 5.77828311920166,
      "learning_rate": 1e-05,
      "loss": 1.0599,
      "step": 4
    },
    {
      "epoch": 0.08465608465608465,
      "grad_norm": 5.281270503997803,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.9754,
      "step": 5
    },
    {
      "epoch": 0.10158730158730159,
      "grad_norm": 5.149123191833496,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.9476,
      "step": 6
    },
    {
      "epoch": 0.11851851851851852,
      "grad_norm": 6.636379718780518,
      "learning_rate": 2e-05,
      "loss": 1.0585,
      "step": 7
    },
    {
      "epoch": 0.13544973544973546,
      "grad_norm": 3.8481030464172363,
      "learning_rate": 1.9998370105646414e-05,
      "loss": 0.9413,
      "step": 8
    },
    {
      "epoch": 0.1523809523809524,
      "grad_norm": 3.835693120956421,
      "learning_rate": 1.999348095389677e-05,
      "loss": 0.8891,
      "step": 9
    },
    {
      "epoch": 0.1693121693121693,
      "grad_norm": 2.8596138954162598,
      "learning_rate": 1.998533413851124e-05,
      "loss": 0.8795,
      "step": 10
    },
    {
      "epoch": 0.18624338624338624,
      "grad_norm": 2.8256287574768066,
      "learning_rate": 1.9973932315179502e-05,
      "loss": 0.8446,
      "step": 11
    },
    {
      "epoch": 0.20317460317460317,
      "grad_norm": 5.52194356918335,
      "learning_rate": 1.9959279200655044e-05,
      "loss": 0.8027,
      "step": 12
    },
    {
      "epoch": 0.2201058201058201,
      "grad_norm": 2.0485758781433105,
      "learning_rate": 1.9941379571543597e-05,
      "loss": 0.7859,
      "step": 13
    },
    {
      "epoch": 0.23703703703703705,
      "grad_norm": 2.2072129249572754,
      "learning_rate": 1.9920239262746045e-05,
      "loss": 0.7766,
      "step": 14
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 2.154557943344116,
      "learning_rate": 1.9895865165556375e-05,
      "loss": 0.7366,
      "step": 15
    },
    {
      "epoch": 0.2708994708994709,
      "grad_norm": 1.9130762815475464,
      "learning_rate": 1.9868265225415263e-05,
      "loss": 0.7432,
      "step": 16
    },
    {
      "epoch": 0.2878306878306878,
      "grad_norm": 1.6504809856414795,
      "learning_rate": 1.9837448439320027e-05,
      "loss": 0.6905,
      "step": 17
    },
    {
      "epoch": 0.3047619047619048,
      "grad_norm": 1.7797411680221558,
      "learning_rate": 1.9803424852891803e-05,
      "loss": 0.6799,
      "step": 18
    },
    {
      "epoch": 0.3216931216931217,
      "grad_norm": 1.7259820699691772,
      "learning_rate": 1.976620555710087e-05,
      "loss": 0.6706,
      "step": 19
    },
    {
      "epoch": 0.3386243386243386,
      "grad_norm": 1.4618961811065674,
      "learning_rate": 1.9725802684651235e-05,
      "loss": 0.6687,
      "step": 20
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 1.5048692226409912,
      "learning_rate": 1.9682229406025635e-05,
      "loss": 0.6762,
      "step": 21
    },
    {
      "epoch": 0.3724867724867725,
      "grad_norm": 1.4504609107971191,
      "learning_rate": 1.963549992519223e-05,
      "loss": 0.6631,
      "step": 22
    },
    {
      "epoch": 0.38941798941798944,
      "grad_norm": 1.4441351890563965,
      "learning_rate": 1.9585629474974413e-05,
      "loss": 0.6394,
      "step": 23
    },
    {
      "epoch": 0.40634920634920635,
      "grad_norm": 1.2873355150222778,
      "learning_rate": 1.953263431208523e-05,
      "loss": 0.6279,
      "step": 24
    },
    {
      "epoch": 0.42328042328042326,
      "grad_norm": 1.4676573276519775,
      "learning_rate": 1.9476531711828027e-05,
      "loss": 0.6238,
      "step": 25
    },
    {
      "epoch": 0.4402116402116402,
      "grad_norm": 1.3487250804901123,
      "learning_rate": 1.9417339962465084e-05,
      "loss": 0.6253,
      "step": 26
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 23.1207275390625,
      "learning_rate": 1.935507835925601e-05,
      "loss": 0.6166,
      "step": 27
    },
    {
      "epoch": 0.4740740740740741,
      "grad_norm": 1.408158540725708,
      "learning_rate": 1.9289767198167918e-05,
      "loss": 0.5644,
      "step": 28
    },
    {
      "epoch": 0.491005291005291,
      "grad_norm": 1.199977159500122,
      "learning_rate": 1.9221427769259333e-05,
      "loss": 0.5865,
      "step": 29
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 1.1895148754119873,
      "learning_rate": 1.9150082349740123e-05,
      "loss": 0.5619,
      "step": 30
    },
    {
      "epoch": 0.5248677248677248,
      "grad_norm": 1.1904869079589844,
      "learning_rate": 1.9075754196709574e-05,
      "loss": 0.5642,
      "step": 31
    },
    {
      "epoch": 0.5417989417989418,
      "grad_norm": 1.4911994934082031,
      "learning_rate": 1.899846753957507e-05,
      "loss": 0.5684,
      "step": 32
    },
    {
      "epoch": 0.5587301587301587,
      "grad_norm": 1.2807954549789429,
      "learning_rate": 1.8918247572153822e-05,
      "loss": 0.532,
      "step": 33
    },
    {
      "epoch": 0.5756613756613757,
      "grad_norm": 1.1882894039154053,
      "learning_rate": 1.883512044446023e-05,
      "loss": 0.5527,
      "step": 34
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 1.3212125301361084,
      "learning_rate": 1.8749113254181498e-05,
      "loss": 0.5621,
      "step": 35
    },
    {
      "epoch": 0.6095238095238096,
      "grad_norm": 1.4091153144836426,
      "learning_rate": 1.866025403784439e-05,
      "loss": 0.5205,
      "step": 36
    },
    {
      "epoch": 0.6264550264550265,
      "grad_norm": 1.1614974737167358,
      "learning_rate": 1.8568571761675893e-05,
      "loss": 0.5181,
      "step": 37
    },
    {
      "epoch": 0.6433862433862434,
      "grad_norm": 1.0888781547546387,
      "learning_rate": 1.8474096312160866e-05,
      "loss": 0.5359,
      "step": 38
    },
    {
      "epoch": 0.6603174603174603,
      "grad_norm": 1.303357720375061,
      "learning_rate": 1.837685848629965e-05,
      "loss": 0.5256,
      "step": 39
    },
    {
      "epoch": 0.6772486772486772,
      "grad_norm": 1.1001622676849365,
      "learning_rate": 1.827688998156891e-05,
      "loss": 0.5277,
      "step": 40
    },
    {
      "epoch": 0.6941798941798942,
      "grad_norm": 16.82403564453125,
      "learning_rate": 1.817422338558892e-05,
      "loss": 0.5301,
      "step": 41
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 1.2562721967697144,
      "learning_rate": 1.8068892165500704e-05,
      "loss": 0.5019,
      "step": 42
    },
    {
      "epoch": 0.728042328042328,
      "grad_norm": 1.143244981765747,
      "learning_rate": 1.796093065705644e-05,
      "loss": 0.5098,
      "step": 43
    },
    {
      "epoch": 0.744973544973545,
      "grad_norm": 1.2346585988998413,
      "learning_rate": 1.7850374053426725e-05,
      "loss": 0.5079,
      "step": 44
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 1.292527675628662,
      "learning_rate": 1.7737258393728363e-05,
      "loss": 0.4987,
      "step": 45
    },
    {
      "epoch": 0.7788359788359789,
      "grad_norm": 1.0964988470077515,
      "learning_rate": 1.7621620551276366e-05,
      "loss": 0.52,
      "step": 46
    },
    {
      "epoch": 0.7957671957671958,
      "grad_norm": 1.3093878030776978,
      "learning_rate": 1.7503498221564026e-05,
      "loss": 0.4911,
      "step": 47
    },
    {
      "epoch": 0.8126984126984127,
      "grad_norm": 1.2538305521011353,
      "learning_rate": 1.7382929909974988e-05,
      "loss": 0.4805,
      "step": 48
    },
    {
      "epoch": 0.8296296296296296,
      "grad_norm": 1.5297569036483765,
      "learning_rate": 1.725995491923131e-05,
      "loss": 0.4834,
      "step": 49
    },
    {
      "epoch": 0.8465608465608465,
      "grad_norm": 1.126209020614624,
      "learning_rate": 1.7134613336581602e-05,
      "loss": 0.474,
      "step": 50
    },
    {
      "epoch": 0.8634920634920635,
      "grad_norm": 1.2999428510665894,
      "learning_rate": 1.7006946020733426e-05,
      "loss": 0.4662,
      "step": 51
    },
    {
      "epoch": 0.8804232804232804,
      "grad_norm": 3.900702714920044,
      "learning_rate": 1.6876994588534234e-05,
      "loss": 0.4728,
      "step": 52
    },
    {
      "epoch": 0.8973544973544973,
      "grad_norm": 1.1961653232574463,
      "learning_rate": 1.6744801401405138e-05,
      "loss": 0.4638,
      "step": 53
    },
    {
      "epoch": 0.9142857142857143,
      "grad_norm": 1.1492871046066284,
      "learning_rate": 1.6610409551532006e-05,
      "loss": 0.4608,
      "step": 54
    },
    {
      "epoch": 0.9312169312169312,
      "grad_norm": 1.1923617124557495,
      "learning_rate": 1.647386284781828e-05,
      "loss": 0.4742,
      "step": 55
    },
    {
      "epoch": 0.9481481481481482,
      "grad_norm": 1.0757859945297241,
      "learning_rate": 1.6335205801604242e-05,
      "loss": 0.4757,
      "step": 56
    },
    {
      "epoch": 0.9650793650793651,
      "grad_norm": 1.0793986320495605,
      "learning_rate": 1.6194483612157232e-05,
      "loss": 0.4543,
      "step": 57
    },
    {
      "epoch": 0.982010582010582,
      "grad_norm": 1.0309922695159912,
      "learning_rate": 1.6051742151937655e-05,
      "loss": 0.4637,
      "step": 58
    },
    {
      "epoch": 0.9989417989417989,
      "grad_norm": 1.0178518295288086,
      "learning_rate": 1.590702795164551e-05,
      "loss": 0.4845,
      "step": 59
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0178518295288086,
      "learning_rate": 1.57603881850524e-05,
      "loss": 0.4151,
      "step": 60
    },
    {
      "epoch": 1.016931216931217,
      "grad_norm": 6.670537948608398,
      "learning_rate": 1.5611870653623826e-05,
      "loss": 0.4583,
      "step": 61
    },
    {
      "epoch": 1.0338624338624338,
      "grad_norm": 1.1750141382217407,
      "learning_rate": 1.546152377093697e-05,
      "loss": 0.4378,
      "step": 62
    },
    {
      "epoch": 1.0507936507936508,
      "grad_norm": 1.258583903312683,
      "learning_rate": 1.530939654689887e-05,
      "loss": 0.4698,
      "step": 63
    },
    {
      "epoch": 1.0677248677248676,
      "grad_norm": 1.1863641738891602,
      "learning_rate": 1.515553857177022e-05,
      "loss": 0.4493,
      "step": 64
    },
    {
      "epoch": 1.0846560846560847,
      "grad_norm": 1.2087653875350952,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.4495,
      "step": 65
    },
    {
      "epoch": 1.1015873015873017,
      "grad_norm": 1.1390602588653564,
      "learning_rate": 1.4842831533876196e-05,
      "loss": 0.4628,
      "step": 66
    },
    {
      "epoch": 1.1185185185185185,
      "grad_norm": 1.0523955821990967,
      "learning_rate": 1.4684084406997903e-05,
      "loss": 0.4444,
      "step": 67
    },
    {
      "epoch": 1.1354497354497355,
      "grad_norm": 1.3320904970169067,
      "learning_rate": 1.4523810367574271e-05,
      "loss": 0.4487,
      "step": 68
    },
    {
      "epoch": 1.1523809523809523,
      "grad_norm": 2.979928970336914,
      "learning_rate": 1.4362061661555675e-05,
      "loss": 0.4568,
      "step": 69
    },
    {
      "epoch": 1.1693121693121693,
      "grad_norm": 1.2313185930252075,
      "learning_rate": 1.4198891015602648e-05,
      "loss": 0.4359,
      "step": 70
    },
    {
      "epoch": 1.1862433862433863,
      "grad_norm": 1.1036418676376343,
      "learning_rate": 1.4034351619898088e-05,
      "loss": 0.4698,
      "step": 71
    },
    {
      "epoch": 1.2031746031746031,
      "grad_norm": 1.1762608289718628,
      "learning_rate": 1.3868497110808394e-05,
      "loss": 0.4313,
      "step": 72
    },
    {
      "epoch": 1.2201058201058201,
      "grad_norm": 1.033484697341919,
      "learning_rate": 1.3701381553399147e-05,
      "loss": 0.45,
      "step": 73
    },
    {
      "epoch": 1.237037037037037,
      "grad_norm": 2.2186121940612793,
      "learning_rate": 1.3533059423811026e-05,
      "loss": 0.4345,
      "step": 74
    },
    {
      "epoch": 1.253968253968254,
      "grad_norm": 1.0926917791366577,
      "learning_rate": 1.3363585591501751e-05,
      "loss": 0.4317,
      "step": 75
    },
    {
      "epoch": 1.270899470899471,
      "grad_norm": 1.094223976135254,
      "learning_rate": 1.31930153013598e-05,
      "loss": 0.4274,
      "step": 76
    },
    {
      "epoch": 1.2878306878306878,
      "grad_norm": 1.0197666883468628,
      "learning_rate": 1.3021404155695728e-05,
      "loss": 0.4465,
      "step": 77
    },
    {
      "epoch": 1.3047619047619048,
      "grad_norm": 1.0113344192504883,
      "learning_rate": 1.2848808096117003e-05,
      "loss": 0.4256,
      "step": 78
    },
    {
      "epoch": 1.3216931216931216,
      "grad_norm": 0.9916861653327942,
      "learning_rate": 1.2675283385292212e-05,
      "loss": 0.4291,
      "step": 79
    },
    {
      "epoch": 1.3386243386243386,
      "grad_norm": 1.0909327268600464,
      "learning_rate": 1.250088658861063e-05,
      "loss": 0.4387,
      "step": 80
    },
    {
      "epoch": 1.3555555555555556,
      "grad_norm": 0.9891695976257324,
      "learning_rate": 1.2325674555743106e-05,
      "loss": 0.4274,
      "step": 81
    },
    {
      "epoch": 1.3724867724867724,
      "grad_norm": 1.053537368774414,
      "learning_rate": 1.2149704402110243e-05,
      "loss": 0.4243,
      "step": 82
    },
    {
      "epoch": 1.3894179894179894,
      "grad_norm": 1.0343950986862183,
      "learning_rate": 1.1973033490264e-05,
      "loss": 0.4029,
      "step": 83
    },
    {
      "epoch": 1.4063492063492062,
      "grad_norm": 1.0367940664291382,
      "learning_rate": 1.1795719411188717e-05,
      "loss": 0.4154,
      "step": 84
    },
    {
      "epoch": 1.4232804232804233,
      "grad_norm": 0.9142754673957825,
      "learning_rate": 1.161781996552765e-05,
      "loss": 0.4352,
      "step": 85
    },
    {
      "epoch": 1.4402116402116403,
      "grad_norm": 0.9767040610313416,
      "learning_rate": 1.1439393144741192e-05,
      "loss": 0.4131,
      "step": 86
    },
    {
      "epoch": 1.457142857142857,
      "grad_norm": 1.1032475233078003,
      "learning_rate": 1.1260497112202895e-05,
      "loss": 0.4177,
      "step": 87
    },
    {
      "epoch": 1.474074074074074,
      "grad_norm": 0.8925254940986633,
      "learning_rate": 1.1081190184239418e-05,
      "loss": 0.4384,
      "step": 88
    },
    {
      "epoch": 1.491005291005291,
      "grad_norm": 0.9563277959823608,
      "learning_rate": 1.0901530811120655e-05,
      "loss": 0.404,
      "step": 89
    },
    {
      "epoch": 1.507936507936508,
      "grad_norm": 0.8952954411506653,
      "learning_rate": 1.0721577558006164e-05,
      "loss": 0.3869,
      "step": 90
    },
    {
      "epoch": 1.524867724867725,
      "grad_norm": 1.1573145389556885,
      "learning_rate": 1.0541389085854177e-05,
      "loss": 0.3898,
      "step": 91
    },
    {
      "epoch": 1.541798941798942,
      "grad_norm": 0.9079419374465942,
      "learning_rate": 1.0361024132299364e-05,
      "loss": 0.4094,
      "step": 92
    },
    {
      "epoch": 1.5587301587301587,
      "grad_norm": 1.1131538152694702,
      "learning_rate": 1.0180541492505605e-05,
      "loss": 0.415,
      "step": 93
    },
    {
      "epoch": 1.5756613756613755,
      "grad_norm": 0.9412957429885864,
      "learning_rate": 1e-05,
      "loss": 0.4063,
      "step": 94
    },
    {
      "epoch": 1.5925925925925926,
      "grad_norm": 0.9416212439537048,
      "learning_rate": 9.819458507494395e-06,
      "loss": 0.4073,
      "step": 95
    },
    {
      "epoch": 1.6095238095238096,
      "grad_norm": 1791.885498046875,
      "learning_rate": 9.638975867700638e-06,
      "loss": 0.3969,
      "step": 96
    },
    {
      "epoch": 1.6264550264550266,
      "grad_norm": 0.9790716767311096,
      "learning_rate": 9.458610914145826e-06,
      "loss": 0.4027,
      "step": 97
    },
    {
      "epoch": 1.6433862433862434,
      "grad_norm": 1.0255053043365479,
      "learning_rate": 9.278422441993841e-06,
      "loss": 0.3969,
      "step": 98
    },
    {
      "epoch": 1.6603174603174602,
      "grad_norm": 0.9520485997200012,
      "learning_rate": 9.098469188879348e-06,
      "loss": 0.3999,
      "step": 99
    },
    {
      "epoch": 1.6772486772486772,
      "grad_norm": 0.8883378505706787,
      "learning_rate": 8.918809815760585e-06,
      "loss": 0.3932,
      "step": 100
    },
    {
      "epoch": 1.6941798941798942,
      "grad_norm": 0.922511875629425,
      "learning_rate": 8.739502887797108e-06,
      "loss": 0.379,
      "step": 101
    },
    {
      "epoch": 1.7111111111111112,
      "grad_norm": 0.978553295135498,
      "learning_rate": 8.560606855258808e-06,
      "loss": 0.4112,
      "step": 102
    },
    {
      "epoch": 1.728042328042328,
      "grad_norm": 1.0453370809555054,
      "learning_rate": 8.382180034472353e-06,
      "loss": 0.3979,
      "step": 103
    },
    {
      "epoch": 1.7449735449735448,
      "grad_norm": 0.9882745146751404,
      "learning_rate": 8.204280588811283e-06,
      "loss": 0.4074,
      "step": 104
    },
    {
      "epoch": 1.7619047619047619,
      "grad_norm": 0.8945300579071045,
      "learning_rate": 8.026966509736001e-06,
      "loss": 0.3615,
      "step": 105
    },
    {
      "epoch": 1.7788359788359789,
      "grad_norm": 1.0022529363632202,
      "learning_rate": 7.85029559788976e-06,
      "loss": 0.3805,
      "step": 106
    },
    {
      "epoch": 1.795767195767196,
      "grad_norm": 0.8616457581520081,
      "learning_rate": 7.674325444256899e-06,
      "loss": 0.3842,
      "step": 107
    },
    {
      "epoch": 1.8126984126984127,
      "grad_norm": 1.4591399431228638,
      "learning_rate": 7.499113411389371e-06,
      "loss": 0.3871,
      "step": 108
    },
    {
      "epoch": 1.8296296296296295,
      "grad_norm": 1.0600188970565796,
      "learning_rate": 7.324716614707794e-06,
      "loss": 0.3905,
      "step": 109
    },
    {
      "epoch": 1.8465608465608465,
      "grad_norm": 0.9007483124732971,
      "learning_rate": 7.1511919038830016e-06,
      "loss": 0.3775,
      "step": 110
    },
    {
      "epoch": 1.8634920634920635,
      "grad_norm": 0.9454174041748047,
      "learning_rate": 6.978595844304272e-06,
      "loss": 0.3803,
      "step": 111
    },
    {
      "epoch": 1.8804232804232806,
      "grad_norm": 0.996856153011322,
      "learning_rate": 6.806984698640202e-06,
      "loss": 0.3961,
      "step": 112
    },
    {
      "epoch": 1.8973544973544973,
      "grad_norm": 1.1498510837554932,
      "learning_rate": 6.636414408498249e-06,
      "loss": 0.3851,
      "step": 113
    },
    {
      "epoch": 1.9142857142857141,
      "grad_norm": 9.213700294494629,
      "learning_rate": 6.466940576188978e-06,
      "loss": 0.375,
      "step": 114
    },
    {
      "epoch": 1.9312169312169312,
      "grad_norm": 0.8880584836006165,
      "learning_rate": 6.298618446600856e-06,
      "loss": 0.3691,
      "step": 115
    },
    {
      "epoch": 1.9481481481481482,
      "grad_norm": 1.0605814456939697,
      "learning_rate": 6.131502889191611e-06,
      "loss": 0.3892,
      "step": 116
    },
    {
      "epoch": 1.9650793650793652,
      "grad_norm": 0.8566351532936096,
      "learning_rate": 5.965648380101916e-06,
      "loss": 0.381,
      "step": 117
    },
    {
      "epoch": 1.982010582010582,
      "grad_norm": 0.9199567437171936,
      "learning_rate": 5.801108984397355e-06,
      "loss": 0.3899,
      "step": 118
    },
    {
      "epoch": 1.9989417989417988,
      "grad_norm": 0.9430050253868103,
      "learning_rate": 5.637938338444325e-06,
      "loss": 0.3742,
      "step": 119
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9430050253868103,
      "learning_rate": 5.476189632425732e-06,
      "loss": 0.4288,
      "step": 120
    },
    {
      "epoch": 2.016931216931217,
      "grad_norm": 4.022532939910889,
      "learning_rate": 5.3159155930021e-06,
      "loss": 0.3637,
      "step": 121
    },
    {
      "epoch": 2.033862433862434,
      "grad_norm": 0.9256008267402649,
      "learning_rate": 5.1571684661238075e-06,
      "loss": 0.3864,
      "step": 122
    },
    {
      "epoch": 2.0507936507936506,
      "grad_norm": 1.0907351970672607,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.3566,
      "step": 123
    },
    {
      "epoch": 2.0677248677248676,
      "grad_norm": 1.0017610788345337,
      "learning_rate": 4.844461428229782e-06,
      "loss": 0.36,
      "step": 124
    },
    {
      "epoch": 2.0846560846560847,
      "grad_norm": 1.0023646354675293,
      "learning_rate": 4.690603453101134e-06,
      "loss": 0.3729,
      "step": 125
    },
    {
      "epoch": 2.1015873015873017,
      "grad_norm": 1.0901212692260742,
      "learning_rate": 4.53847622906303e-06,
      "loss": 0.3773,
      "step": 126
    },
    {
      "epoch": 2.1185185185185187,
      "grad_norm": 1.1009324789047241,
      "learning_rate": 4.388129346376177e-06,
      "loss": 0.3847,
      "step": 127
    },
    {
      "epoch": 2.1354497354497353,
      "grad_norm": 1.0068461894989014,
      "learning_rate": 4.239611814947605e-06,
      "loss": 0.3634,
      "step": 128
    },
    {
      "epoch": 2.1523809523809523,
      "grad_norm": 0.8794623613357544,
      "learning_rate": 4.092972048354491e-06,
      "loss": 0.3645,
      "step": 129
    },
    {
      "epoch": 2.1693121693121693,
      "grad_norm": 0.9761602282524109,
      "learning_rate": 3.948257848062351e-06,
      "loss": 0.3676,
      "step": 130
    },
    {
      "epoch": 2.1862433862433863,
      "grad_norm": 1.0799708366394043,
      "learning_rate": 3.8055163878427703e-06,
      "loss": 0.3738,
      "step": 131
    },
    {
      "epoch": 2.2031746031746033,
      "grad_norm": 1.0237700939178467,
      "learning_rate": 3.6647941983957647e-06,
      "loss": 0.3714,
      "step": 132
    },
    {
      "epoch": 2.22010582010582,
      "grad_norm": 0.9058141112327576,
      "learning_rate": 3.5261371521817247e-06,
      "loss": 0.3856,
      "step": 133
    },
    {
      "epoch": 2.237037037037037,
      "grad_norm": 0.9451678395271301,
      "learning_rate": 3.3895904484679986e-06,
      "loss": 0.3795,
      "step": 134
    },
    {
      "epoch": 2.253968253968254,
      "grad_norm": 0.9849613904953003,
      "learning_rate": 3.255198598594862e-06,
      "loss": 0.3503,
      "step": 135
    },
    {
      "epoch": 2.270899470899471,
      "grad_norm": 0.9847337603569031,
      "learning_rate": 3.123005411465766e-06,
      "loss": 0.3762,
      "step": 136
    },
    {
      "epoch": 2.287830687830688,
      "grad_norm": 1.853551983833313,
      "learning_rate": 2.9930539792665767e-06,
      "loss": 0.3683,
      "step": 137
    },
    {
      "epoch": 2.3047619047619046,
      "grad_norm": 1.001577615737915,
      "learning_rate": 2.8653866634184e-06,
      "loss": 0.3728,
      "step": 138
    },
    {
      "epoch": 2.3216931216931216,
      "grad_norm": 1.1002743244171143,
      "learning_rate": 2.740045080768694e-06,
      "loss": 0.3481,
      "step": 139
    },
    {
      "epoch": 2.3386243386243386,
      "grad_norm": 0.878426730632782,
      "learning_rate": 2.6170700900250146e-06,
      "loss": 0.3716,
      "step": 140
    },
    {
      "epoch": 2.3555555555555556,
      "grad_norm": 0.9544874429702759,
      "learning_rate": 2.496501778435977e-06,
      "loss": 0.3689,
      "step": 141
    },
    {
      "epoch": 2.3724867724867726,
      "grad_norm": 1.055256724357605,
      "learning_rate": 2.3783794487236367e-06,
      "loss": 0.3583,
      "step": 142
    },
    {
      "epoch": 2.389417989417989,
      "grad_norm": 1.0352602005004883,
      "learning_rate": 2.2627416062716366e-06,
      "loss": 0.3813,
      "step": 143
    },
    {
      "epoch": 2.4063492063492062,
      "grad_norm": 0.9158893823623657,
      "learning_rate": 2.1496259465732783e-06,
      "loss": 0.3503,
      "step": 144
    },
    {
      "epoch": 2.4232804232804233,
      "grad_norm": 1.4428284168243408,
      "learning_rate": 2.0390693429435626e-06,
      "loss": 0.3597,
      "step": 145
    },
    {
      "epoch": 2.4402116402116403,
      "grad_norm": 0.8209134936332703,
      "learning_rate": 1.931107834499296e-06,
      "loss": 0.3683,
      "step": 146
    },
    {
      "epoch": 2.4571428571428573,
      "grad_norm": 0.8735495805740356,
      "learning_rate": 1.8257766144110823e-06,
      "loss": 0.3736,
      "step": 147
    },
    {
      "epoch": 2.474074074074074,
      "grad_norm": 1.0254844427108765,
      "learning_rate": 1.7231100184310955e-06,
      "loss": 0.3507,
      "step": 148
    },
    {
      "epoch": 2.491005291005291,
      "grad_norm": 0.9373979568481445,
      "learning_rate": 1.6231415137003536e-06,
      "loss": 0.3456,
      "step": 149
    },
    {
      "epoch": 2.507936507936508,
      "grad_norm": 0.8155479431152344,
      "learning_rate": 1.5259036878391342e-06,
      "loss": 0.3857,
      "step": 150
    },
    {
      "epoch": 2.524867724867725,
      "grad_norm": 0.9397179484367371,
      "learning_rate": 1.4314282383241097e-06,
      "loss": 0.3469,
      "step": 151
    },
    {
      "epoch": 2.541798941798942,
      "grad_norm": 0.8838076591491699,
      "learning_rate": 1.339745962155613e-06,
      "loss": 0.332,
      "step": 152
    },
    {
      "epoch": 2.5587301587301585,
      "grad_norm": 0.8838992714881897,
      "learning_rate": 1.2508867458185037e-06,
      "loss": 0.3616,
      "step": 153
    },
    {
      "epoch": 2.5756613756613755,
      "grad_norm": 0.833113968372345,
      "learning_rate": 1.1648795555397719e-06,
      "loss": 0.3652,
      "step": 154
    },
    {
      "epoch": 2.5925925925925926,
      "grad_norm": 0.9157512187957764,
      "learning_rate": 1.0817524278461777e-06,
      "loss": 0.3633,
      "step": 155
    },
    {
      "epoch": 2.6095238095238096,
      "grad_norm": 0.9280597567558289,
      "learning_rate": 1.0015324604249343e-06,
      "loss": 0.3639,
      "step": 156
    },
    {
      "epoch": 2.6264550264550266,
      "grad_norm": 0.9176546931266785,
      "learning_rate": 9.242458032904311e-07,
      "loss": 0.3684,
      "step": 157
    },
    {
      "epoch": 2.643386243386243,
      "grad_norm": 0.9126737713813782,
      "learning_rate": 8.499176502598783e-07,
      "loss": 0.3725,
      "step": 158
    },
    {
      "epoch": 2.66031746031746,
      "grad_norm": 0.8442249298095703,
      "learning_rate": 7.785722307406685e-07,
      "loss": 0.359,
      "step": 159
    },
    {
      "epoch": 2.677248677248677,
      "grad_norm": 0.8220723271369934,
      "learning_rate": 7.102328018320859e-07,
      "loss": 0.3799,
      "step": 160
    },
    {
      "epoch": 2.6941798941798942,
      "grad_norm": 0.8401052951812744,
      "learning_rate": 6.449216407439906e-07,
      "loss": 0.3462,
      "step": 161
    },
    {
      "epoch": 2.7111111111111112,
      "grad_norm": 0.8923698663711548,
      "learning_rate": 5.826600375349201e-07,
      "loss": 0.3763,
      "step": 162
    },
    {
      "epoch": 2.728042328042328,
      "grad_norm": 0.9540568590164185,
      "learning_rate": 5.234682881719766e-07,
      "loss": 0.3546,
      "step": 163
    },
    {
      "epoch": 2.744973544973545,
      "grad_norm": 323.1459655761719,
      "learning_rate": 4.6736568791477367e-07,
      "loss": 0.3602,
      "step": 164
    },
    {
      "epoch": 2.761904761904762,
      "grad_norm": 0.8727107644081116,
      "learning_rate": 4.1437052502558693e-07,
      "loss": 0.3777,
      "step": 165
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 180,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 5,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.163631073258845e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}