{
  "best_metric": 4.4400410652160645,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.012829353560947448,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.414676780473723e-05,
      "grad_norm": 1.642872929573059,
      "learning_rate": 1e-05,
      "loss": 4.7768,
      "step": 1
    },
    {
      "epoch": 6.414676780473723e-05,
      "eval_loss": 5.596181392669678,
      "eval_runtime": 175.703,
      "eval_samples_per_second": 149.434,
      "eval_steps_per_second": 37.359,
      "step": 1
    },
    {
      "epoch": 0.00012829353560947447,
      "grad_norm": 1.7171331644058228,
      "learning_rate": 2e-05,
      "loss": 4.9218,
      "step": 2
    },
    {
      "epoch": 0.00019244030341421172,
      "grad_norm": 1.7027052640914917,
      "learning_rate": 3e-05,
      "loss": 4.8481,
      "step": 3
    },
    {
      "epoch": 0.00025658707121894894,
      "grad_norm": 1.678208351135254,
      "learning_rate": 4e-05,
      "loss": 4.7427,
      "step": 4
    },
    {
      "epoch": 0.0003207338390236862,
      "grad_norm": 1.685784101486206,
      "learning_rate": 5e-05,
      "loss": 4.9766,
      "step": 5
    },
    {
      "epoch": 0.00038488060682842344,
      "grad_norm": 1.6038089990615845,
      "learning_rate": 6e-05,
      "loss": 4.8053,
      "step": 6
    },
    {
      "epoch": 0.00044902737463316066,
      "grad_norm": 1.4707045555114746,
      "learning_rate": 7e-05,
      "loss": 4.9695,
      "step": 7
    },
    {
      "epoch": 0.0005131741424378979,
      "grad_norm": 1.2065513134002686,
      "learning_rate": 8e-05,
      "loss": 4.8506,
      "step": 8
    },
    {
      "epoch": 0.0005773209102426352,
      "grad_norm": 1.217518925666809,
      "learning_rate": 9e-05,
      "loss": 4.8318,
      "step": 9
    },
    {
      "epoch": 0.0006414676780473724,
      "grad_norm": 1.1238003969192505,
      "learning_rate": 0.0001,
      "loss": 4.7909,
      "step": 10
    },
    {
      "epoch": 0.0007056144458521096,
      "grad_norm": 1.2250131368637085,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.8419,
      "step": 11
    },
    {
      "epoch": 0.0007697612136568469,
      "grad_norm": 1.0958911180496216,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.6394,
      "step": 12
    },
    {
      "epoch": 0.0008339079814615841,
      "grad_norm": 1.075415849685669,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.7929,
      "step": 13
    },
    {
      "epoch": 0.0008980547492663213,
      "grad_norm": 0.9673530459403992,
      "learning_rate": 9.989068136093873e-05,
      "loss": 4.8091,
      "step": 14
    },
    {
      "epoch": 0.0009622015170710586,
      "grad_norm": 1.070448398590088,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.7363,
      "step": 15
    },
    {
      "epoch": 0.0010263482848757958,
      "grad_norm": 1.0438446998596191,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.6623,
      "step": 16
    },
    {
      "epoch": 0.001090495052680533,
      "grad_norm": 1.0157839059829712,
      "learning_rate": 9.966546331768191e-05,
      "loss": 4.6976,
      "step": 17
    },
    {
      "epoch": 0.0011546418204852703,
      "grad_norm": 1.041697382926941,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.6569,
      "step": 18
    },
    {
      "epoch": 0.0012187885882900076,
      "grad_norm": 1.0324170589447021,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.6052,
      "step": 19
    },
    {
      "epoch": 0.0012829353560947449,
      "grad_norm": 1.0508081912994385,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.8329,
      "step": 20
    },
    {
      "epoch": 0.001347082123899482,
      "grad_norm": 1.0439448356628418,
      "learning_rate": 9.917525374361912e-05,
      "loss": 4.6623,
      "step": 21
    },
    {
      "epoch": 0.0014112288917042192,
      "grad_norm": 0.9907127618789673,
      "learning_rate": 9.901899829374047e-05,
      "loss": 4.6505,
      "step": 22
    },
    {
      "epoch": 0.0014753756595089565,
      "grad_norm": 1.0066382884979248,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.5596,
      "step": 23
    },
    {
      "epoch": 0.0015395224273136937,
      "grad_norm": 1.057041883468628,
      "learning_rate": 9.86663298624003e-05,
      "loss": 4.5897,
      "step": 24
    },
    {
      "epoch": 0.001603669195118431,
      "grad_norm": 1.0306063890457153,
      "learning_rate": 9.847001329696653e-05,
      "loss": 4.7666,
      "step": 25
    },
    {
      "epoch": 0.0016678159629231683,
      "grad_norm": 1.0321996212005615,
      "learning_rate": 9.826044551386744e-05,
      "loss": 4.6842,
      "step": 26
    },
    {
      "epoch": 0.0017319627307279054,
      "grad_norm": 1.0303255319595337,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.7573,
      "step": 27
    },
    {
      "epoch": 0.0017961094985326426,
      "grad_norm": 1.052943468093872,
      "learning_rate": 9.780178907671789e-05,
      "loss": 4.5831,
      "step": 28
    },
    {
      "epoch": 0.00186025626633738,
      "grad_norm": 1.0942226648330688,
      "learning_rate": 9.755282581475769e-05,
      "loss": 4.569,
      "step": 29
    },
    {
      "epoch": 0.0019244030341421172,
      "grad_norm": 1.1259537935256958,
      "learning_rate": 9.729086208503174e-05,
      "loss": 4.6559,
      "step": 30
    },
    {
      "epoch": 0.0019885498019468545,
      "grad_norm": 1.1287720203399658,
      "learning_rate": 9.701596950580806e-05,
      "loss": 4.5996,
      "step": 31
    },
    {
      "epoch": 0.0020526965697515915,
      "grad_norm": 1.3546675443649292,
      "learning_rate": 9.672822322997305e-05,
      "loss": 4.5527,
      "step": 32
    },
    {
      "epoch": 0.002116843337556329,
      "grad_norm": 1.4525164365768433,
      "learning_rate": 9.642770192448536e-05,
      "loss": 4.7025,
      "step": 33
    },
    {
      "epoch": 0.002180990105361066,
      "grad_norm": 1.6699098348617554,
      "learning_rate": 9.611448774886924e-05,
      "loss": 4.6607,
      "step": 34
    },
    {
      "epoch": 0.0022451368731658036,
      "grad_norm": 1.2364184856414795,
      "learning_rate": 9.578866633275288e-05,
      "loss": 4.743,
      "step": 35
    },
    {
      "epoch": 0.0023092836409705406,
      "grad_norm": 1.2899703979492188,
      "learning_rate": 9.545032675245813e-05,
      "loss": 4.5895,
      "step": 36
    },
    {
      "epoch": 0.0023734304087752777,
      "grad_norm": 1.2018170356750488,
      "learning_rate": 9.509956150664796e-05,
      "loss": 4.6149,
      "step": 37
    },
    {
      "epoch": 0.002437577176580015,
      "grad_norm": 1.5725171566009521,
      "learning_rate": 9.473646649103818e-05,
      "loss": 4.7538,
      "step": 38
    },
    {
      "epoch": 0.0025017239443847522,
      "grad_norm": 1.5618175268173218,
      "learning_rate": 9.43611409721806e-05,
      "loss": 4.729,
      "step": 39
    },
    {
      "epoch": 0.0025658707121894897,
      "grad_norm": 1.3766765594482422,
      "learning_rate": 9.397368756032445e-05,
      "loss": 4.6127,
      "step": 40
    },
    {
      "epoch": 0.0026300174799942268,
      "grad_norm": 2.1418488025665283,
      "learning_rate": 9.357421218136386e-05,
      "loss": 4.7404,
      "step": 41
    },
    {
      "epoch": 0.002694164247798964,
      "grad_norm": 4.414487361907959,
      "learning_rate": 9.316282404787871e-05,
      "loss": 4.5486,
      "step": 42
    },
    {
      "epoch": 0.0027583110156037013,
      "grad_norm": 1.9374182224273682,
      "learning_rate": 9.273963562927695e-05,
      "loss": 4.5715,
      "step": 43
    },
    {
      "epoch": 0.0028224577834084384,
      "grad_norm": 1.7547005414962769,
      "learning_rate": 9.230476262104677e-05,
      "loss": 4.9717,
      "step": 44
    },
    {
      "epoch": 0.002886604551213176,
      "grad_norm": 2.0398449897766113,
      "learning_rate": 9.185832391312644e-05,
      "loss": 5.0536,
      "step": 45
    },
    {
      "epoch": 0.002950751319017913,
      "grad_norm": 2.199415683746338,
      "learning_rate": 9.140044155740101e-05,
      "loss": 5.0214,
      "step": 46
    },
    {
      "epoch": 0.0030148980868226504,
      "grad_norm": 2.3262555599212646,
      "learning_rate": 9.093124073433463e-05,
      "loss": 5.1181,
      "step": 47
    },
    {
      "epoch": 0.0030790448546273875,
      "grad_norm": 2.498055934906006,
      "learning_rate": 9.045084971874738e-05,
      "loss": 5.01,
      "step": 48
    },
    {
      "epoch": 0.0031431916224321245,
      "grad_norm": 2.4217894077301025,
      "learning_rate": 8.995939984474624e-05,
      "loss": 5.1427,
      "step": 49
    },
    {
      "epoch": 0.003207338390236862,
      "grad_norm": 3.816474199295044,
      "learning_rate": 8.945702546981969e-05,
      "loss": 5.765,
      "step": 50
    },
    {
      "epoch": 0.003207338390236862,
      "eval_loss": 5.211904048919678,
      "eval_runtime": 176.0481,
      "eval_samples_per_second": 149.141,
      "eval_steps_per_second": 37.285,
      "step": 50
    },
    {
      "epoch": 0.003271485158041599,
      "grad_norm": 16.199979782104492,
      "learning_rate": 8.894386393810563e-05,
      "loss": 5.1198,
      "step": 51
    },
    {
      "epoch": 0.0033356319258463366,
      "grad_norm": 16.18631362915039,
      "learning_rate": 8.842005554284296e-05,
      "loss": 5.1413,
      "step": 52
    },
    {
      "epoch": 0.0033997786936510736,
      "grad_norm": 15.308594703674316,
      "learning_rate": 8.788574348801675e-05,
      "loss": 5.1288,
      "step": 53
    },
    {
      "epoch": 0.0034639254614558107,
      "grad_norm": 11.432890892028809,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.6673,
      "step": 54
    },
    {
      "epoch": 0.003528072229260548,
      "grad_norm": 6.996739864349365,
      "learning_rate": 8.678619553365659e-05,
      "loss": 4.5088,
      "step": 55
    },
    {
      "epoch": 0.0035922189970652853,
      "grad_norm": 2.9491331577301025,
      "learning_rate": 8.622126023955446e-05,
      "loss": 4.2026,
      "step": 56
    },
    {
      "epoch": 0.0036563657648700227,
      "grad_norm": 1.1189446449279785,
      "learning_rate": 8.564642241456986e-05,
      "loss": 4.3928,
      "step": 57
    },
    {
      "epoch": 0.00372051253267476,
      "grad_norm": 1.140417456626892,
      "learning_rate": 8.506183921362443e-05,
      "loss": 4.2991,
      "step": 58
    },
    {
      "epoch": 0.0037846593004794973,
      "grad_norm": 1.085638403892517,
      "learning_rate": 8.44676704559283e-05,
      "loss": 4.1921,
      "step": 59
    },
    {
      "epoch": 0.0038488060682842344,
      "grad_norm": 1.130789041519165,
      "learning_rate": 8.386407858128706e-05,
      "loss": 4.3044,
      "step": 60
    },
    {
      "epoch": 0.003912952836088971,
      "grad_norm": 1.2206088304519653,
      "learning_rate": 8.32512286056924e-05,
      "loss": 4.3771,
      "step": 61
    },
    {
      "epoch": 0.003977099603893709,
      "grad_norm": 1.19307279586792,
      "learning_rate": 8.262928807620843e-05,
      "loss": 4.2969,
      "step": 62
    },
    {
      "epoch": 0.004041246371698446,
      "grad_norm": 1.1722559928894043,
      "learning_rate": 8.199842702516583e-05,
      "loss": 4.4744,
      "step": 63
    },
    {
      "epoch": 0.004105393139503183,
      "grad_norm": 1.073339819908142,
      "learning_rate": 8.135881792367686e-05,
      "loss": 4.439,
      "step": 64
    },
    {
      "epoch": 0.0041695399073079205,
      "grad_norm": 0.9737927317619324,
      "learning_rate": 8.07106356344834e-05,
      "loss": 4.3597,
      "step": 65
    },
    {
      "epoch": 0.004233686675112658,
      "grad_norm": 1.1121063232421875,
      "learning_rate": 8.005405736415126e-05,
      "loss": 4.4142,
      "step": 66
    },
    {
      "epoch": 0.004297833442917395,
      "grad_norm": 0.9784128665924072,
      "learning_rate": 7.938926261462366e-05,
      "loss": 4.3532,
      "step": 67
    },
    {
      "epoch": 0.004361980210722132,
      "grad_norm": 0.9304491877555847,
      "learning_rate": 7.871643313414718e-05,
      "loss": 4.3092,
      "step": 68
    },
    {
      "epoch": 0.00442612697852687,
      "grad_norm": 0.9712762832641602,
      "learning_rate": 7.803575286758364e-05,
      "loss": 4.1377,
      "step": 69
    },
    {
      "epoch": 0.004490273746331607,
      "grad_norm": 1.0609962940216064,
      "learning_rate": 7.734740790612136e-05,
      "loss": 4.5251,
      "step": 70
    },
    {
      "epoch": 0.004554420514136344,
      "grad_norm": 1.3250576257705688,
      "learning_rate": 7.66515864363997e-05,
      "loss": 4.4165,
      "step": 71
    },
    {
      "epoch": 0.004618567281941081,
      "grad_norm": 1.514135479927063,
      "learning_rate": 7.594847868906076e-05,
      "loss": 4.3032,
      "step": 72
    },
    {
      "epoch": 0.004682714049745819,
      "grad_norm": 1.4466665983200073,
      "learning_rate": 7.52382768867422e-05,
      "loss": 4.3565,
      "step": 73
    },
    {
      "epoch": 0.004746860817550555,
      "grad_norm": 1.2483808994293213,
      "learning_rate": 7.452117519152542e-05,
      "loss": 4.4922,
      "step": 74
    },
    {
      "epoch": 0.004811007585355293,
      "grad_norm": 1.1569164991378784,
      "learning_rate": 7.379736965185368e-05,
      "loss": 4.354,
      "step": 75
    },
    {
      "epoch": 0.00487515435316003,
      "grad_norm": 1.0108411312103271,
      "learning_rate": 7.30670581489344e-05,
      "loss": 4.3471,
      "step": 76
    },
    {
      "epoch": 0.004939301120964767,
      "grad_norm": 1.0727484226226807,
      "learning_rate": 7.233044034264034e-05,
      "loss": 4.3835,
      "step": 77
    },
    {
      "epoch": 0.0050034478887695044,
      "grad_norm": 1.1690360307693481,
      "learning_rate": 7.158771761692464e-05,
      "loss": 4.2849,
      "step": 78
    },
    {
      "epoch": 0.005067594656574242,
      "grad_norm": 1.0823379755020142,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.3961,
      "step": 79
    },
    {
      "epoch": 0.0051317414243789794,
      "grad_norm": 1.2482712268829346,
      "learning_rate": 7.008477123264848e-05,
      "loss": 4.4605,
      "step": 80
    },
    {
      "epoch": 0.005195888192183716,
      "grad_norm": 1.125679850578308,
      "learning_rate": 6.932495846462261e-05,
      "loss": 4.5634,
      "step": 81
    },
    {
      "epoch": 0.0052600349599884536,
      "grad_norm": 1.0943341255187988,
      "learning_rate": 6.855986244591104e-05,
      "loss": 4.5166,
      "step": 82
    },
    {
      "epoch": 0.005324181727793191,
      "grad_norm": 1.0785305500030518,
      "learning_rate": 6.778969234612584e-05,
      "loss": 4.7178,
      "step": 83
    },
    {
      "epoch": 0.005388328495597928,
      "grad_norm": 1.0343631505966187,
      "learning_rate": 6.701465872208216e-05,
      "loss": 4.4751,
      "step": 84
    },
    {
      "epoch": 0.005452475263402665,
      "grad_norm": 1.005264401435852,
      "learning_rate": 6.623497346023418e-05,
      "loss": 4.5418,
      "step": 85
    },
    {
      "epoch": 0.005516622031207403,
      "grad_norm": 1.1396501064300537,
      "learning_rate": 6.545084971874738e-05,
      "loss": 4.4077,
      "step": 86
    },
    {
      "epoch": 0.00558076879901214,
      "grad_norm": 1.1338545083999634,
      "learning_rate": 6.466250186922325e-05,
      "loss": 4.5811,
      "step": 87
    },
    {
      "epoch": 0.005644915566816877,
      "grad_norm": 1.0844732522964478,
      "learning_rate": 6.387014543809223e-05,
      "loss": 4.4835,
      "step": 88
    },
    {
      "epoch": 0.005709062334621614,
      "grad_norm": 1.0581928491592407,
      "learning_rate": 6.307399704769099e-05,
      "loss": 4.3321,
      "step": 89
    },
    {
      "epoch": 0.005773209102426352,
      "grad_norm": 1.4487022161483765,
      "learning_rate": 6.227427435703997e-05,
      "loss": 4.3464,
      "step": 90
    },
    {
      "epoch": 0.005837355870231088,
      "grad_norm": 1.3548654317855835,
      "learning_rate": 6.147119600233758e-05,
      "loss": 4.6635,
      "step": 91
    },
    {
      "epoch": 0.005901502638035826,
      "grad_norm": 1.4622880220413208,
      "learning_rate": 6.066498153718735e-05,
      "loss": 4.7465,
      "step": 92
    },
    {
      "epoch": 0.005965649405840563,
      "grad_norm": 1.5502005815505981,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.5915,
      "step": 93
    },
    {
      "epoch": 0.006029796173645301,
      "grad_norm": 1.7211005687713623,
      "learning_rate": 5.90440267166055e-05,
      "loss": 4.6409,
      "step": 94
    },
    {
      "epoch": 0.0060939429414500375,
      "grad_norm": 1.6254551410675049,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.8106,
      "step": 95
    },
    {
      "epoch": 0.006158089709254775,
      "grad_norm": 1.7562053203582764,
      "learning_rate": 5.74131823855921e-05,
      "loss": 4.7768,
      "step": 96
    },
    {
      "epoch": 0.0062222364770595125,
      "grad_norm": 1.8948148488998413,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.756,
      "step": 97
    },
    {
      "epoch": 0.006286383244864249,
      "grad_norm": 2.0640652179718018,
      "learning_rate": 5.577423184847932e-05,
      "loss": 5.1927,
      "step": 98
    },
    {
      "epoch": 0.006350530012668987,
      "grad_norm": 2.5090017318725586,
      "learning_rate": 5.495227651252315e-05,
      "loss": 5.2081,
      "step": 99
    },
    {
      "epoch": 0.006414676780473724,
      "grad_norm": 3.904104709625244,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 5.3744,
      "step": 100
    },
    {
      "epoch": 0.006414676780473724,
      "eval_loss": 5.082364082336426,
      "eval_runtime": 175.9755,
      "eval_samples_per_second": 149.203,
      "eval_steps_per_second": 37.301,
      "step": 100
    },
    {
      "epoch": 0.006478823548278461,
      "grad_norm": 14.663159370422363,
      "learning_rate": 5.330452921628497e-05,
      "loss": 5.0622,
      "step": 101
    },
    {
      "epoch": 0.006542970316083198,
      "grad_norm": 16.544841766357422,
      "learning_rate": 5.247918773366112e-05,
      "loss": 5.3057,
      "step": 102
    },
    {
      "epoch": 0.006607117083887936,
      "grad_norm": 15.879215240478516,
      "learning_rate": 5.165316846586541e-05,
      "loss": 5.1625,
      "step": 103
    },
    {
      "epoch": 0.006671263851692673,
      "grad_norm": 14.463156700134277,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 5.1542,
      "step": 104
    },
    {
      "epoch": 0.00673541061949741,
      "grad_norm": 13.046357154846191,
      "learning_rate": 5e-05,
      "loss": 4.8398,
      "step": 105
    },
    {
      "epoch": 0.006799557387302147,
      "grad_norm": 11.176506042480469,
      "learning_rate": 4.917330276168208e-05,
      "loss": 4.6716,
      "step": 106
    },
    {
      "epoch": 0.006863704155106885,
      "grad_norm": 8.23073673248291,
      "learning_rate": 4.834683153413459e-05,
      "loss": 4.6304,
      "step": 107
    },
    {
      "epoch": 0.006927850922911621,
      "grad_norm": 6.0302734375,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 4.451,
      "step": 108
    },
    {
      "epoch": 0.006991997690716359,
      "grad_norm": 3.8446044921875,
      "learning_rate": 4.669547078371504e-05,
      "loss": 4.3728,
      "step": 109
    },
    {
      "epoch": 0.007056144458521096,
      "grad_norm": 1.688712239265442,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 4.2718,
      "step": 110
    },
    {
      "epoch": 0.007120291226325834,
      "grad_norm": 0.8822144865989685,
      "learning_rate": 4.504772348747687e-05,
      "loss": 4.3909,
      "step": 111
    },
    {
      "epoch": 0.0071844379941305705,
      "grad_norm": 0.9888668656349182,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 4.3215,
      "step": 112
    },
    {
      "epoch": 0.007248584761935308,
      "grad_norm": 1.142039179801941,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 4.2201,
      "step": 113
    },
    {
      "epoch": 0.0073127315297400455,
      "grad_norm": 1.1784989833831787,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 4.3019,
      "step": 114
    },
    {
      "epoch": 0.007376878297544782,
      "grad_norm": 1.1900745630264282,
      "learning_rate": 4.17702704859633e-05,
      "loss": 4.2193,
      "step": 115
    },
    {
      "epoch": 0.00744102506534952,
      "grad_norm": 1.2521610260009766,
      "learning_rate": 4.095597328339452e-05,
      "loss": 4.1634,
      "step": 116
    },
    {
      "epoch": 0.007505171833154257,
      "grad_norm": 1.2006127834320068,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 4.2993,
      "step": 117
    },
    {
      "epoch": 0.007569318600958995,
      "grad_norm": 1.3783690929412842,
      "learning_rate": 3.933501846281267e-05,
      "loss": 4.3544,
      "step": 118
    },
    {
      "epoch": 0.007633465368763731,
      "grad_norm": 1.1826130151748657,
      "learning_rate": 3.852880399766243e-05,
      "loss": 4.2801,
      "step": 119
    },
    {
      "epoch": 0.007697612136568469,
      "grad_norm": 1.212864637374878,
      "learning_rate": 3.772572564296005e-05,
      "loss": 4.4208,
      "step": 120
    },
    {
      "epoch": 0.007761758904373206,
      "grad_norm": 1.213317632675171,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 4.4495,
      "step": 121
    },
    {
      "epoch": 0.007825905672177943,
      "grad_norm": 1.2609578371047974,
      "learning_rate": 3.612985456190778e-05,
      "loss": 4.4031,
      "step": 122
    },
    {
      "epoch": 0.00789005243998268,
      "grad_norm": 1.1333434581756592,
      "learning_rate": 3.533749813077677e-05,
      "loss": 4.3654,
      "step": 123
    },
    {
      "epoch": 0.007954199207787418,
      "grad_norm": 1.1164824962615967,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 4.4081,
      "step": 124
    },
    {
      "epoch": 0.008018345975592155,
      "grad_norm": 1.0088897943496704,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 4.4174,
      "step": 125
    },
    {
      "epoch": 0.008082492743396893,
      "grad_norm": 1.0312213897705078,
      "learning_rate": 3.298534127791785e-05,
      "loss": 4.5553,
      "step": 126
    },
    {
      "epoch": 0.008146639511201629,
      "grad_norm": 0.9682738780975342,
      "learning_rate": 3.221030765387417e-05,
      "loss": 4.3147,
      "step": 127
    },
    {
      "epoch": 0.008210786279006366,
      "grad_norm": 1.0135595798492432,
      "learning_rate": 3.144013755408895e-05,
      "loss": 4.5083,
      "step": 128
    },
    {
      "epoch": 0.008274933046811104,
      "grad_norm": 1.106201171875,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 4.3359,
      "step": 129
    },
    {
      "epoch": 0.008339079814615841,
      "grad_norm": 1.158813238143921,
      "learning_rate": 2.991522876735154e-05,
      "loss": 4.4816,
      "step": 130
    },
    {
      "epoch": 0.008403226582420579,
      "grad_norm": 1.0801420211791992,
      "learning_rate": 2.916090697523549e-05,
      "loss": 4.4342,
      "step": 131
    },
    {
      "epoch": 0.008467373350225316,
      "grad_norm": 1.1284626722335815,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 4.5528,
      "step": 132
    },
    {
      "epoch": 0.008531520118030054,
      "grad_norm": 1.1629167795181274,
      "learning_rate": 2.766955965735968e-05,
      "loss": 4.3737,
      "step": 133
    },
    {
      "epoch": 0.00859566688583479,
      "grad_norm": 1.3331410884857178,
      "learning_rate": 2.693294185106562e-05,
      "loss": 4.3553,
      "step": 134
    },
    {
      "epoch": 0.008659813653639527,
      "grad_norm": 1.1465953588485718,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 4.37,
      "step": 135
    },
    {
      "epoch": 0.008723960421444264,
      "grad_norm": 1.135396122932434,
      "learning_rate": 2.547882480847461e-05,
      "loss": 4.474,
      "step": 136
    },
    {
      "epoch": 0.008788107189249002,
      "grad_norm": 1.2194247245788574,
      "learning_rate": 2.476172311325783e-05,
      "loss": 4.3128,
      "step": 137
    },
    {
      "epoch": 0.00885225395705374,
      "grad_norm": 1.1395728588104248,
      "learning_rate": 2.405152131093926e-05,
      "loss": 4.4183,
      "step": 138
    },
    {
      "epoch": 0.008916400724858477,
      "grad_norm": 1.2786668539047241,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.5955,
      "step": 139
    },
    {
      "epoch": 0.008980547492663214,
      "grad_norm": 1.2455207109451294,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 4.3468,
      "step": 140
    },
    {
      "epoch": 0.00904469426046795,
      "grad_norm": 1.4937551021575928,
      "learning_rate": 2.196424713241637e-05,
      "loss": 4.4906,
      "step": 141
    },
    {
      "epoch": 0.009108841028272687,
      "grad_norm": 1.473012089729309,
      "learning_rate": 2.128356686585282e-05,
      "loss": 4.696,
      "step": 142
    },
    {
      "epoch": 0.009172987796077425,
      "grad_norm": 1.81377112865448,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.734,
      "step": 143
    },
    {
      "epoch": 0.009237134563882162,
      "grad_norm": 1.8109852075576782,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.6593,
      "step": 144
    },
    {
      "epoch": 0.0093012813316869,
      "grad_norm": 1.8153804540634155,
      "learning_rate": 1.928936436551661e-05,
      "loss": 4.6856,
      "step": 145
    },
    {
      "epoch": 0.009365428099491637,
      "grad_norm": 2.1597044467926025,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 4.4313,
      "step": 146
    },
    {
      "epoch": 0.009429574867296375,
      "grad_norm": 2.92808198928833,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.9218,
      "step": 147
    },
    {
      "epoch": 0.00949372163510111,
      "grad_norm": 3.4155867099761963,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 5.177,
      "step": 148
    },
    {
      "epoch": 0.009557868402905848,
      "grad_norm": 3.7774338722229004,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 5.1557,
      "step": 149
    },
    {
      "epoch": 0.009622015170710586,
      "grad_norm": 5.2052226066589355,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 5.3052,
      "step": 150
    },
    {
      "epoch": 0.009622015170710586,
      "eval_loss": 4.5528883934021,
      "eval_runtime": 175.8332,
      "eval_samples_per_second": 149.323,
      "eval_steps_per_second": 37.331,
      "step": 150
    },
    {
      "epoch": 0.009686161938515323,
      "grad_norm": 6.3930864334106445,
      "learning_rate": 1.553232954407171e-05,
      "loss": 4.6273,
      "step": 151
    },
    {
      "epoch": 0.00975030870632006,
      "grad_norm": 6.974958419799805,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 4.3823,
      "step": 152
    },
    {
      "epoch": 0.009814455474124798,
      "grad_norm": 7.43309211730957,
      "learning_rate": 1.435357758543015e-05,
      "loss": 4.4558,
      "step": 153
    },
    {
      "epoch": 0.009878602241929534,
      "grad_norm": 7.601620197296143,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 4.5882,
      "step": 154
    },
    {
      "epoch": 0.009942749009734271,
      "grad_norm": 8.028189659118652,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 4.4678,
      "step": 155
    },
    {
      "epoch": 0.010006895777539009,
      "grad_norm": 7.0772881507873535,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 4.5163,
      "step": 156
    },
    {
      "epoch": 0.010071042545343746,
      "grad_norm": 6.785968780517578,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 4.6539,
      "step": 157
    },
    {
      "epoch": 0.010135189313148484,
      "grad_norm": 6.7161054611206055,
      "learning_rate": 1.157994445715706e-05,
      "loss": 4.5375,
      "step": 158
    },
    {
      "epoch": 0.010199336080953221,
      "grad_norm": 5.955432891845703,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 4.5158,
      "step": 159
    },
    {
      "epoch": 0.010263482848757959,
      "grad_norm": 5.452800750732422,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 4.2282,
      "step": 160
    },
    {
      "epoch": 0.010327629616562695,
      "grad_norm": 5.346560478210449,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 4.4541,
      "step": 161
    },
    {
      "epoch": 0.010391776384367432,
      "grad_norm": 4.477380275726318,
      "learning_rate": 9.549150281252633e-06,
      "loss": 4.3145,
      "step": 162
    },
    {
      "epoch": 0.01045592315217217,
      "grad_norm": 3.7970259189605713,
      "learning_rate": 9.068759265665384e-06,
      "loss": 4.3903,
      "step": 163
    },
    {
      "epoch": 0.010520069919976907,
      "grad_norm": 3.247373342514038,
      "learning_rate": 8.599558442598998e-06,
      "loss": 4.3047,
      "step": 164
    },
    {
      "epoch": 0.010584216687781645,
      "grad_norm": 2.714463472366333,
      "learning_rate": 8.141676086873572e-06,
      "loss": 4.3872,
      "step": 165
    },
    {
      "epoch": 0.010648363455586382,
      "grad_norm": 2.536929130554199,
      "learning_rate": 7.695237378953223e-06,
      "loss": 4.1894,
      "step": 166
    },
    {
      "epoch": 0.01071251022339112,
      "grad_norm": 2.198115825653076,
      "learning_rate": 7.260364370723044e-06,
      "loss": 4.441,
      "step": 167
    },
    {
      "epoch": 0.010776656991195855,
      "grad_norm": 1.95339035987854,
      "learning_rate": 6.837175952121306e-06,
      "loss": 4.3528,
      "step": 168
    },
    {
      "epoch": 0.010840803759000593,
      "grad_norm": 1.3755385875701904,
      "learning_rate": 6.425787818636131e-06,
      "loss": 4.2991,
      "step": 169
    },
    {
      "epoch": 0.01090495052680533,
      "grad_norm": 1.2790817022323608,
      "learning_rate": 6.026312439675552e-06,
      "loss": 4.3256,
      "step": 170
    },
    {
      "epoch": 0.010969097294610068,
      "grad_norm": 1.065985083580017,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 4.4564,
      "step": 171
    },
    {
      "epoch": 0.011033244062414805,
      "grad_norm": 1.0836656093597412,
      "learning_rate": 5.263533508961827e-06,
      "loss": 4.1409,
      "step": 172
    },
    {
      "epoch": 0.011097390830219543,
      "grad_norm": 0.975472092628479,
      "learning_rate": 4.900438493352055e-06,
      "loss": 4.2486,
      "step": 173
    },
    {
      "epoch": 0.01116153759802428,
      "grad_norm": 0.9753958582878113,
      "learning_rate": 4.549673247541875e-06,
      "loss": 4.3776,
      "step": 174
    },
    {
      "epoch": 0.011225684365829016,
      "grad_norm": 0.9526723027229309,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 4.3783,
      "step": 175
    },
    {
      "epoch": 0.011289831133633754,
      "grad_norm": 0.9613341093063354,
      "learning_rate": 3.885512251130763e-06,
      "loss": 4.3595,
      "step": 176
    },
    {
      "epoch": 0.011353977901438491,
      "grad_norm": 0.9815852642059326,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 4.3145,
      "step": 177
    },
    {
      "epoch": 0.011418124669243229,
      "grad_norm": 1.0029901266098022,
      "learning_rate": 3.271776770026963e-06,
      "loss": 4.4467,
      "step": 178
    },
    {
      "epoch": 0.011482271437047966,
      "grad_norm": 1.0242493152618408,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 4.4017,
      "step": 179
    },
    {
      "epoch": 0.011546418204852704,
      "grad_norm": 1.1363314390182495,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 4.5128,
      "step": 180
    },
    {
      "epoch": 0.011610564972657441,
      "grad_norm": 1.1976279020309448,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 4.5127,
      "step": 181
    },
    {
      "epoch": 0.011674711740462177,
      "grad_norm": 1.1066765785217285,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 4.3555,
      "step": 182
    },
    {
      "epoch": 0.011738858508266914,
      "grad_norm": 1.334769606590271,
      "learning_rate": 1.962316193157593e-06,
      "loss": 4.505,
      "step": 183
    },
    {
      "epoch": 0.011803005276071652,
      "grad_norm": 1.1973919868469238,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 4.7047,
      "step": 184
    },
    {
      "epoch": 0.01186715204387639,
      "grad_norm": 1.269687533378601,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 4.4973,
      "step": 185
    },
    {
      "epoch": 0.011931298811681127,
      "grad_norm": 1.3963308334350586,
      "learning_rate": 1.333670137599713e-06,
      "loss": 4.4313,
      "step": 186
    },
    {
      "epoch": 0.011995445579485864,
      "grad_norm": 1.4404746294021606,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 4.5762,
      "step": 187
    },
    {
      "epoch": 0.012059592347290602,
      "grad_norm": 1.4484505653381348,
      "learning_rate": 9.810017062595322e-07,
      "loss": 4.5687,
      "step": 188
    },
    {
      "epoch": 0.012123739115095337,
      "grad_norm": 1.5444748401641846,
      "learning_rate": 8.247462563808817e-07,
      "loss": 4.4911,
      "step": 189
    },
    {
      "epoch": 0.012187885882900075,
      "grad_norm": 1.6145284175872803,
      "learning_rate": 6.819348298638839e-07,
      "loss": 4.2873,
      "step": 190
    },
    {
      "epoch": 0.012252032650704812,
      "grad_norm": 1.7238717079162598,
      "learning_rate": 5.526064699265753e-07,
      "loss": 4.5226,
      "step": 191
    },
    {
      "epoch": 0.01231617941850955,
      "grad_norm": 1.9913530349731445,
      "learning_rate": 4.367965336512403e-07,
      "loss": 4.7709,
      "step": 192
    },
    {
      "epoch": 0.012380326186314287,
      "grad_norm": 2.0532760620117188,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 4.5588,
      "step": 193
    },
    {
      "epoch": 0.012444472954119025,
      "grad_norm": 2.565652847290039,
      "learning_rate": 2.458548727494292e-07,
      "loss": 4.4004,
      "step": 194
    },
    {
      "epoch": 0.012508619721923762,
      "grad_norm": 2.7140657901763916,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 4.7269,
      "step": 195
    },
    {
      "epoch": 0.012572766489728498,
      "grad_norm": 3.383547067642212,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.8739,
      "step": 196
    },
    {
      "epoch": 0.012636913257533236,
      "grad_norm": 3.5743050575256348,
      "learning_rate": 6.150154258476315e-08,
      "loss": 4.9864,
      "step": 197
    },
    {
      "epoch": 0.012701060025337973,
      "grad_norm": 3.5662736892700195,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 4.9759,
      "step": 198
    },
    {
      "epoch": 0.01276520679314271,
      "grad_norm": 4.346615314483643,
      "learning_rate": 6.834750376549792e-09,
      "loss": 5.1423,
      "step": 199
    },
    {
      "epoch": 0.012829353560947448,
      "grad_norm": 6.375400543212891,
      "learning_rate": 0.0,
      "loss": 5.7376,
      "step": 200
    },
    {
      "epoch": 0.012829353560947448,
      "eval_loss": 4.4400410652160645,
      "eval_runtime": 175.4399,
      "eval_samples_per_second": 149.658,
      "eval_steps_per_second": 37.415,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5953579646976000.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}