{
  "best_metric": 0.34753015637397766,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3006388575723412,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001503194287861706,
      "grad_norm": 0.5596352219581604,
      "learning_rate": 1e-05,
      "loss": 0.3511,
      "step": 1
    },
    {
      "epoch": 0.001503194287861706,
      "eval_loss": 0.624755859375,
      "eval_runtime": 54.3198,
      "eval_samples_per_second": 20.637,
      "eval_steps_per_second": 5.173,
      "step": 1
    },
    {
      "epoch": 0.003006388575723412,
      "grad_norm": 0.7194361090660095,
      "learning_rate": 2e-05,
      "loss": 0.4613,
      "step": 2
    },
    {
      "epoch": 0.004509582863585118,
      "grad_norm": 0.7623833417892456,
      "learning_rate": 3e-05,
      "loss": 0.5343,
      "step": 3
    },
    {
      "epoch": 0.006012777151446824,
      "grad_norm": 0.6368038654327393,
      "learning_rate": 4e-05,
      "loss": 0.4856,
      "step": 4
    },
    {
      "epoch": 0.00751597143930853,
      "grad_norm": 0.519997775554657,
      "learning_rate": 5e-05,
      "loss": 0.4821,
      "step": 5
    },
    {
      "epoch": 0.009019165727170236,
      "grad_norm": 0.5378677248954773,
      "learning_rate": 6e-05,
      "loss": 0.4381,
      "step": 6
    },
    {
      "epoch": 0.010522360015031942,
      "grad_norm": 0.6729313731193542,
      "learning_rate": 7e-05,
      "loss": 0.4567,
      "step": 7
    },
    {
      "epoch": 0.012025554302893648,
      "grad_norm": 0.621289074420929,
      "learning_rate": 8e-05,
      "loss": 0.51,
      "step": 8
    },
    {
      "epoch": 0.013528748590755355,
      "grad_norm": 0.5438550114631653,
      "learning_rate": 9e-05,
      "loss": 0.4581,
      "step": 9
    },
    {
      "epoch": 0.01503194287861706,
      "grad_norm": 0.5436606407165527,
      "learning_rate": 0.0001,
      "loss": 0.4384,
      "step": 10
    },
    {
      "epoch": 0.016535137166478767,
      "grad_norm": 0.48347610235214233,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.4526,
      "step": 11
    },
    {
      "epoch": 0.018038331454340473,
      "grad_norm": 0.4883898198604584,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.4152,
      "step": 12
    },
    {
      "epoch": 0.01954152574220218,
      "grad_norm": 0.4870653450489044,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.3998,
      "step": 13
    },
    {
      "epoch": 0.021044720030063885,
      "grad_norm": 0.48256585001945496,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.4152,
      "step": 14
    },
    {
      "epoch": 0.02254791431792559,
      "grad_norm": 0.44895902276039124,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.4247,
      "step": 15
    },
    {
      "epoch": 0.024051108605787297,
      "grad_norm": 0.3959610164165497,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.3762,
      "step": 16
    },
    {
      "epoch": 0.025554302893649003,
      "grad_norm": 0.43920475244522095,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.4128,
      "step": 17
    },
    {
      "epoch": 0.02705749718151071,
      "grad_norm": 0.4507409334182739,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.4574,
      "step": 18
    },
    {
      "epoch": 0.028560691469372415,
      "grad_norm": 0.44753408432006836,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.47,
      "step": 19
    },
    {
      "epoch": 0.03006388575723412,
      "grad_norm": 0.42001616954803467,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.4492,
      "step": 20
    },
    {
      "epoch": 0.03156708004509583,
      "grad_norm": 0.45444992184638977,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.4415,
      "step": 21
    },
    {
      "epoch": 0.03307027433295753,
      "grad_norm": 0.45968878269195557,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.4587,
      "step": 22
    },
    {
      "epoch": 0.03457346862081924,
      "grad_norm": 0.3964838981628418,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.4046,
      "step": 23
    },
    {
      "epoch": 0.036076662908680945,
      "grad_norm": 0.4261048436164856,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.3905,
      "step": 24
    },
    {
      "epoch": 0.03757985719654265,
      "grad_norm": 0.4289442300796509,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.4308,
      "step": 25
    },
    {
      "epoch": 0.03908305148440436,
      "grad_norm": 0.435476154088974,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.4735,
      "step": 26
    },
    {
      "epoch": 0.040586245772266064,
      "grad_norm": 0.42117980122566223,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.4119,
      "step": 27
    },
    {
      "epoch": 0.04208944006012777,
      "grad_norm": 0.44847437739372253,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.4121,
      "step": 28
    },
    {
      "epoch": 0.043592634347989476,
      "grad_norm": 0.422210693359375,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.4041,
      "step": 29
    },
    {
      "epoch": 0.04509582863585118,
      "grad_norm": 0.4026918113231659,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.3983,
      "step": 30
    },
    {
      "epoch": 0.04659902292371289,
      "grad_norm": 0.45083874464035034,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.3989,
      "step": 31
    },
    {
      "epoch": 0.048102217211574594,
      "grad_norm": 0.42770618200302124,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.4144,
      "step": 32
    },
    {
      "epoch": 0.0496054114994363,
      "grad_norm": 0.4121634364128113,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.3996,
      "step": 33
    },
    {
      "epoch": 0.051108605787298006,
      "grad_norm": 0.4511989653110504,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.3884,
      "step": 34
    },
    {
      "epoch": 0.05261180007515971,
      "grad_norm": 0.44728589057922363,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.4136,
      "step": 35
    },
    {
      "epoch": 0.05411499436302142,
      "grad_norm": 0.4282330274581909,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.3864,
      "step": 36
    },
    {
      "epoch": 0.055618188650883124,
      "grad_norm": 0.46324941515922546,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.4413,
      "step": 37
    },
    {
      "epoch": 0.05712138293874483,
      "grad_norm": 0.4187595546245575,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.3852,
      "step": 38
    },
    {
      "epoch": 0.058624577226606536,
      "grad_norm": 0.43247416615486145,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3865,
      "step": 39
    },
    {
      "epoch": 0.06012777151446824,
      "grad_norm": 0.39086800813674927,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.336,
      "step": 40
    },
    {
      "epoch": 0.06163096580232995,
      "grad_norm": 0.4751962125301361,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.4333,
      "step": 41
    },
    {
      "epoch": 0.06313416009019165,
      "grad_norm": 0.4536871314048767,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.3693,
      "step": 42
    },
    {
      "epoch": 0.06463735437805336,
      "grad_norm": 0.5023728013038635,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.3467,
      "step": 43
    },
    {
      "epoch": 0.06614054866591507,
      "grad_norm": 0.4579885005950928,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.3726,
      "step": 44
    },
    {
      "epoch": 0.06764374295377677,
      "grad_norm": 0.5028191208839417,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.3774,
      "step": 45
    },
    {
      "epoch": 0.06914693724163848,
      "grad_norm": 0.5278530120849609,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.4052,
      "step": 46
    },
    {
      "epoch": 0.07065013152950018,
      "grad_norm": 0.5628510117530823,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.4262,
      "step": 47
    },
    {
      "epoch": 0.07215332581736189,
      "grad_norm": 0.6105784773826599,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.415,
      "step": 48
    },
    {
      "epoch": 0.0736565201052236,
      "grad_norm": 0.5921081304550171,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.3439,
      "step": 49
    },
    {
      "epoch": 0.0751597143930853,
      "grad_norm": 0.7702072858810425,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4604,
      "step": 50
    },
    {
      "epoch": 0.0751597143930853,
      "eval_loss": 0.39259007573127747,
      "eval_runtime": 54.842,
      "eval_samples_per_second": 20.441,
      "eval_steps_per_second": 5.124,
      "step": 50
    },
    {
      "epoch": 0.07666290868094701,
      "grad_norm": 0.32153835892677307,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.3509,
      "step": 51
    },
    {
      "epoch": 0.07816610296880872,
      "grad_norm": 0.29875338077545166,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.4147,
      "step": 52
    },
    {
      "epoch": 0.07966929725667042,
      "grad_norm": 0.308470219373703,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.4196,
      "step": 53
    },
    {
      "epoch": 0.08117249154453213,
      "grad_norm": 0.31666040420532227,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.4143,
      "step": 54
    },
    {
      "epoch": 0.08267568583239383,
      "grad_norm": 0.3026100993156433,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.4456,
      "step": 55
    },
    {
      "epoch": 0.08417888012025554,
      "grad_norm": 0.29371339082717896,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.3559,
      "step": 56
    },
    {
      "epoch": 0.08568207440811725,
      "grad_norm": 0.3148328363895416,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3927,
      "step": 57
    },
    {
      "epoch": 0.08718526869597895,
      "grad_norm": 0.30351120233535767,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3777,
      "step": 58
    },
    {
      "epoch": 0.08868846298384066,
      "grad_norm": 0.34232327342033386,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3867,
      "step": 59
    },
    {
      "epoch": 0.09019165727170236,
      "grad_norm": 0.3512055277824402,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.4172,
      "step": 60
    },
    {
      "epoch": 0.09169485155956407,
      "grad_norm": 0.3395233750343323,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.3771,
      "step": 61
    },
    {
      "epoch": 0.09319804584742578,
      "grad_norm": 0.33125266432762146,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.3987,
      "step": 62
    },
    {
      "epoch": 0.09470124013528748,
      "grad_norm": 0.40485480427742004,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.384,
      "step": 63
    },
    {
      "epoch": 0.09620443442314919,
      "grad_norm": 0.36481529474258423,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.401,
      "step": 64
    },
    {
      "epoch": 0.0977076287110109,
      "grad_norm": 0.33970266580581665,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.385,
      "step": 65
    },
    {
      "epoch": 0.0992108229988726,
      "grad_norm": 0.34393301606178284,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.393,
      "step": 66
    },
    {
      "epoch": 0.1007140172867343,
      "grad_norm": 0.36804962158203125,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.3822,
      "step": 67
    },
    {
      "epoch": 0.10221721157459601,
      "grad_norm": 0.3514614701271057,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.405,
      "step": 68
    },
    {
      "epoch": 0.10372040586245772,
      "grad_norm": 0.37134188413619995,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.4007,
      "step": 69
    },
    {
      "epoch": 0.10522360015031942,
      "grad_norm": 0.36093929409980774,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.3821,
      "step": 70
    },
    {
      "epoch": 0.10672679443818113,
      "grad_norm": 0.36206376552581787,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.3791,
      "step": 71
    },
    {
      "epoch": 0.10822998872604284,
      "grad_norm": 0.3646642565727234,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3992,
      "step": 72
    },
    {
      "epoch": 0.10973318301390454,
      "grad_norm": 0.38398653268814087,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.3709,
      "step": 73
    },
    {
      "epoch": 0.11123637730176625,
      "grad_norm": 0.41967472434043884,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.44,
      "step": 74
    },
    {
      "epoch": 0.11273957158962795,
      "grad_norm": 0.410259485244751,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.389,
      "step": 75
    },
    {
      "epoch": 0.11424276587748966,
      "grad_norm": 0.3982982635498047,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.3772,
      "step": 76
    },
    {
      "epoch": 0.11574596016535137,
      "grad_norm": 0.38881319761276245,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.3725,
      "step": 77
    },
    {
      "epoch": 0.11724915445321307,
      "grad_norm": 0.39276328682899475,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.3715,
      "step": 78
    },
    {
      "epoch": 0.11875234874107478,
      "grad_norm": 0.3920992612838745,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.3714,
      "step": 79
    },
    {
      "epoch": 0.12025554302893648,
      "grad_norm": 0.3906751871109009,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3625,
      "step": 80
    },
    {
      "epoch": 0.12175873731679819,
      "grad_norm": 0.40682023763656616,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.3693,
      "step": 81
    },
    {
      "epoch": 0.1232619316046599,
      "grad_norm": 0.3880535960197449,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.3502,
      "step": 82
    },
    {
      "epoch": 0.1247651258925216,
      "grad_norm": 0.423728346824646,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.365,
      "step": 83
    },
    {
      "epoch": 0.1262683201803833,
      "grad_norm": 0.41527223587036133,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.3402,
      "step": 84
    },
    {
      "epoch": 0.12777151446824503,
      "grad_norm": 0.4017101228237152,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3697,
      "step": 85
    },
    {
      "epoch": 0.12927470875610672,
      "grad_norm": 0.4082302451133728,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.3597,
      "step": 86
    },
    {
      "epoch": 0.13077790304396844,
      "grad_norm": 0.4285735785961151,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.37,
      "step": 87
    },
    {
      "epoch": 0.13228109733183013,
      "grad_norm": 0.47520652413368225,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3604,
      "step": 88
    },
    {
      "epoch": 0.13378429161969185,
      "grad_norm": 0.4193938970565796,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.3669,
      "step": 89
    },
    {
      "epoch": 0.13528748590755355,
      "grad_norm": 0.45612838864326477,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.3725,
      "step": 90
    },
    {
      "epoch": 0.13679068019541527,
      "grad_norm": 0.4199351370334625,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.3249,
      "step": 91
    },
    {
      "epoch": 0.13829387448327696,
      "grad_norm": 0.4824385643005371,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.3901,
      "step": 92
    },
    {
      "epoch": 0.13979706877113868,
      "grad_norm": 0.45574283599853516,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.3575,
      "step": 93
    },
    {
      "epoch": 0.14130026305900037,
      "grad_norm": 0.46389901638031006,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3558,
      "step": 94
    },
    {
      "epoch": 0.1428034573468621,
      "grad_norm": 0.44589194655418396,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.315,
      "step": 95
    },
    {
      "epoch": 0.14430665163472378,
      "grad_norm": 0.4838874042034149,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3692,
      "step": 96
    },
    {
      "epoch": 0.1458098459225855,
      "grad_norm": 0.5774380564689636,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.4069,
      "step": 97
    },
    {
      "epoch": 0.1473130402104472,
      "grad_norm": 0.5149058103561401,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3545,
      "step": 98
    },
    {
      "epoch": 0.1488162344983089,
      "grad_norm": 0.5392869710922241,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.3157,
      "step": 99
    },
    {
      "epoch": 0.1503194287861706,
      "grad_norm": 0.7136433720588684,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3207,
      "step": 100
    },
    {
      "epoch": 0.1503194287861706,
      "eval_loss": 0.3683274984359741,
      "eval_runtime": 54.6677,
      "eval_samples_per_second": 20.506,
      "eval_steps_per_second": 5.14,
      "step": 100
    },
    {
      "epoch": 0.15182262307403233,
      "grad_norm": 0.26212289929389954,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2755,
      "step": 101
    },
    {
      "epoch": 0.15332581736189402,
      "grad_norm": 0.3127160370349884,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.4129,
      "step": 102
    },
    {
      "epoch": 0.15482901164975574,
      "grad_norm": 0.3296348452568054,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.419,
      "step": 103
    },
    {
      "epoch": 0.15633220593761743,
      "grad_norm": 0.30321285128593445,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.3779,
      "step": 104
    },
    {
      "epoch": 0.15783540022547915,
      "grad_norm": 0.301196813583374,
      "learning_rate": 5e-05,
      "loss": 0.3451,
      "step": 105
    },
    {
      "epoch": 0.15933859451334084,
      "grad_norm": 0.3540099859237671,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.362,
      "step": 106
    },
    {
      "epoch": 0.16084178880120256,
      "grad_norm": 0.31483331322669983,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.3439,
      "step": 107
    },
    {
      "epoch": 0.16234498308906425,
      "grad_norm": 0.3202732503414154,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.373,
      "step": 108
    },
    {
      "epoch": 0.16384817737692597,
      "grad_norm": 0.2974686920642853,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.3569,
      "step": 109
    },
    {
      "epoch": 0.16535137166478767,
      "grad_norm": 0.30358532071113586,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3645,
      "step": 110
    },
    {
      "epoch": 0.1668545659526494,
      "grad_norm": 0.34163451194763184,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.4031,
      "step": 111
    },
    {
      "epoch": 0.16835776024051108,
      "grad_norm": 0.34163618087768555,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.4144,
      "step": 112
    },
    {
      "epoch": 0.1698609545283728,
      "grad_norm": 0.3094636797904968,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3709,
      "step": 113
    },
    {
      "epoch": 0.1713641488162345,
      "grad_norm": 0.33215829730033875,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3527,
      "step": 114
    },
    {
      "epoch": 0.1728673431040962,
      "grad_norm": 0.33634069561958313,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3784,
      "step": 115
    },
    {
      "epoch": 0.1743705373919579,
      "grad_norm": 0.3472643792629242,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.4201,
      "step": 116
    },
    {
      "epoch": 0.17587373167981962,
      "grad_norm": 0.344387948513031,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3877,
      "step": 117
    },
    {
      "epoch": 0.17737692596768131,
      "grad_norm": 0.34439584612846375,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3717,
      "step": 118
    },
    {
      "epoch": 0.17888012025554303,
      "grad_norm": 0.35252517461776733,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3812,
      "step": 119
    },
    {
      "epoch": 0.18038331454340473,
      "grad_norm": 0.3847930431365967,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.3554,
      "step": 120
    },
    {
      "epoch": 0.18188650883126645,
      "grad_norm": 0.35014215111732483,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.3915,
      "step": 121
    },
    {
      "epoch": 0.18338970311912814,
      "grad_norm": 0.35064560174942017,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3526,
      "step": 122
    },
    {
      "epoch": 0.18489289740698986,
      "grad_norm": 0.40358299016952515,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3984,
      "step": 123
    },
    {
      "epoch": 0.18639609169485155,
      "grad_norm": 0.3744370639324188,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3496,
      "step": 124
    },
    {
      "epoch": 0.18789928598271327,
      "grad_norm": 0.3636492192745209,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3926,
      "step": 125
    },
    {
      "epoch": 0.18940248027057496,
      "grad_norm": 0.385662317276001,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.369,
      "step": 126
    },
    {
      "epoch": 0.19090567455843668,
      "grad_norm": 0.364361047744751,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3635,
      "step": 127
    },
    {
      "epoch": 0.19240886884629838,
      "grad_norm": 0.40278124809265137,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3716,
      "step": 128
    },
    {
      "epoch": 0.1939120631341601,
      "grad_norm": 0.3735451400279999,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.3252,
      "step": 129
    },
    {
      "epoch": 0.1954152574220218,
      "grad_norm": 0.38422390818595886,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3615,
      "step": 130
    },
    {
      "epoch": 0.1969184517098835,
      "grad_norm": 0.3829570710659027,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.3807,
      "step": 131
    },
    {
      "epoch": 0.1984216459977452,
      "grad_norm": 0.3987874686717987,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3772,
      "step": 132
    },
    {
      "epoch": 0.19992484028560692,
      "grad_norm": 0.3898436427116394,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.3396,
      "step": 133
    },
    {
      "epoch": 0.2014280345734686,
      "grad_norm": 0.4023337960243225,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3481,
      "step": 134
    },
    {
      "epoch": 0.20293122886133033,
      "grad_norm": 0.3817198574542999,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3441,
      "step": 135
    },
    {
      "epoch": 0.20443442314919202,
      "grad_norm": 0.38800162076950073,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.3319,
      "step": 136
    },
    {
      "epoch": 0.20593761743705374,
      "grad_norm": 0.40661248564720154,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.3798,
      "step": 137
    },
    {
      "epoch": 0.20744081172491544,
      "grad_norm": 0.4218497574329376,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3564,
      "step": 138
    },
    {
      "epoch": 0.20894400601277716,
      "grad_norm": 0.42600011825561523,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3439,
      "step": 139
    },
    {
      "epoch": 0.21044720030063885,
      "grad_norm": 0.422944575548172,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.3458,
      "step": 140
    },
    {
      "epoch": 0.21195039458850057,
      "grad_norm": 0.4048214554786682,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.3239,
      "step": 141
    },
    {
      "epoch": 0.21345358887636226,
      "grad_norm": 0.4112938940525055,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.3048,
      "step": 142
    },
    {
      "epoch": 0.21495678316422398,
      "grad_norm": 0.4650246500968933,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3565,
      "step": 143
    },
    {
      "epoch": 0.21645997745208567,
      "grad_norm": 0.5013602375984192,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3867,
      "step": 144
    },
    {
      "epoch": 0.2179631717399474,
      "grad_norm": 0.45937857031822205,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3557,
      "step": 145
    },
    {
      "epoch": 0.21946636602780908,
      "grad_norm": 0.5174309015274048,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.3477,
      "step": 146
    },
    {
      "epoch": 0.2209695603156708,
      "grad_norm": 0.476762592792511,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.3355,
      "step": 147
    },
    {
      "epoch": 0.2224727546035325,
      "grad_norm": 0.4924473464488983,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2744,
      "step": 148
    },
    {
      "epoch": 0.22397594889139422,
      "grad_norm": 0.6597243547439575,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3453,
      "step": 149
    },
    {
      "epoch": 0.2254791431792559,
      "grad_norm": 0.6664057970046997,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.3596,
      "step": 150
    },
    {
      "epoch": 0.2254791431792559,
      "eval_loss": 0.3501320481300354,
      "eval_runtime": 54.7026,
      "eval_samples_per_second": 20.493,
      "eval_steps_per_second": 5.137,
      "step": 150
    },
    {
      "epoch": 0.22698233746711763,
      "grad_norm": 0.22929313778877258,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2784,
      "step": 151
    },
    {
      "epoch": 0.22848553175497932,
      "grad_norm": 0.27377232909202576,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.4036,
      "step": 152
    },
    {
      "epoch": 0.22998872604284104,
      "grad_norm": 0.26963168382644653,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3675,
      "step": 153
    },
    {
      "epoch": 0.23149192033070273,
      "grad_norm": 0.3080262839794159,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.4166,
      "step": 154
    },
    {
      "epoch": 0.23299511461856445,
      "grad_norm": 0.2926326394081116,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.363,
      "step": 155
    },
    {
      "epoch": 0.23449830890642615,
      "grad_norm": 0.29955020546913147,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.3161,
      "step": 156
    },
    {
      "epoch": 0.23600150319428786,
      "grad_norm": 0.3129074275493622,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3622,
      "step": 157
    },
    {
      "epoch": 0.23750469748214956,
      "grad_norm": 0.3651416599750519,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.4296,
      "step": 158
    },
    {
      "epoch": 0.23900789177001128,
      "grad_norm": 0.31064069271087646,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.3492,
      "step": 159
    },
    {
      "epoch": 0.24051108605787297,
      "grad_norm": 0.3591364026069641,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.3917,
      "step": 160
    },
    {
      "epoch": 0.2420142803457347,
      "grad_norm": 0.3335144519805908,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3776,
      "step": 161
    },
    {
      "epoch": 0.24351747463359638,
      "grad_norm": 0.30554336309432983,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.3455,
      "step": 162
    },
    {
      "epoch": 0.2450206689214581,
      "grad_norm": 0.3222895860671997,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.3355,
      "step": 163
    },
    {
      "epoch": 0.2465238632093198,
      "grad_norm": 0.3238259255886078,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.3454,
      "step": 164
    },
    {
      "epoch": 0.2480270574971815,
      "grad_norm": 0.3491290509700775,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.3474,
      "step": 165
    },
    {
      "epoch": 0.2495302517850432,
      "grad_norm": 0.3277914822101593,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.3651,
      "step": 166
    },
    {
      "epoch": 0.2510334460729049,
      "grad_norm": 0.33758988976478577,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3741,
      "step": 167
    },
    {
      "epoch": 0.2525366403607666,
      "grad_norm": 0.3350788354873657,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.3499,
      "step": 168
    },
    {
      "epoch": 0.25403983464862834,
      "grad_norm": 0.3243398666381836,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.347,
      "step": 169
    },
    {
      "epoch": 0.25554302893649006,
      "grad_norm": 0.34355881810188293,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.3526,
      "step": 170
    },
    {
      "epoch": 0.2570462232243517,
      "grad_norm": 0.3813783824443817,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3835,
      "step": 171
    },
    {
      "epoch": 0.25854941751221344,
      "grad_norm": 0.388657808303833,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.3828,
      "step": 172
    },
    {
      "epoch": 0.26005261180007516,
      "grad_norm": 0.3673814833164215,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.3438,
      "step": 173
    },
    {
      "epoch": 0.2615558060879369,
      "grad_norm": 0.32948142290115356,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.3048,
      "step": 174
    },
    {
      "epoch": 0.26305900037579855,
      "grad_norm": 0.3570793569087982,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3447,
      "step": 175
    },
    {
      "epoch": 0.26456219466366027,
      "grad_norm": 0.35363292694091797,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.3631,
      "step": 176
    },
    {
      "epoch": 0.266065388951522,
      "grad_norm": 0.37539130449295044,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.3643,
      "step": 177
    },
    {
      "epoch": 0.2675685832393837,
      "grad_norm": 0.38558998703956604,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.3681,
      "step": 178
    },
    {
      "epoch": 0.26907177752724537,
      "grad_norm": 0.3716927170753479,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3671,
      "step": 179
    },
    {
      "epoch": 0.2705749718151071,
      "grad_norm": 0.3771487772464752,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.3309,
      "step": 180
    },
    {
      "epoch": 0.2720781661029688,
      "grad_norm": 0.3573903441429138,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.3246,
      "step": 181
    },
    {
      "epoch": 0.27358136039083053,
      "grad_norm": 0.39243030548095703,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3374,
      "step": 182
    },
    {
      "epoch": 0.2750845546786922,
      "grad_norm": 0.424135684967041,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.3857,
      "step": 183
    },
    {
      "epoch": 0.2765877489665539,
      "grad_norm": 0.3713597357273102,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3332,
      "step": 184
    },
    {
      "epoch": 0.27809094325441563,
      "grad_norm": 0.39152365922927856,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.398,
      "step": 185
    },
    {
      "epoch": 0.27959413754227735,
      "grad_norm": 0.3714972138404846,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.3134,
      "step": 186
    },
    {
      "epoch": 0.281097331830139,
      "grad_norm": 0.3927851915359497,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.3585,
      "step": 187
    },
    {
      "epoch": 0.28260052611800074,
      "grad_norm": 0.4404532313346863,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.362,
      "step": 188
    },
    {
      "epoch": 0.28410372040586246,
      "grad_norm": 0.4045756757259369,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.3221,
      "step": 189
    },
    {
      "epoch": 0.2856069146937242,
      "grad_norm": 0.42240893840789795,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.3702,
      "step": 190
    },
    {
      "epoch": 0.28711010898158584,
      "grad_norm": 0.4267599880695343,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.3191,
      "step": 191
    },
    {
      "epoch": 0.28861330326944756,
      "grad_norm": 0.4583100974559784,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.3849,
      "step": 192
    },
    {
      "epoch": 0.2901164975573093,
      "grad_norm": 0.4387661814689636,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2992,
      "step": 193
    },
    {
      "epoch": 0.291619691845171,
      "grad_norm": 0.43879735469818115,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.3221,
      "step": 194
    },
    {
      "epoch": 0.29312288613303267,
      "grad_norm": 0.4648761451244354,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.3139,
      "step": 195
    },
    {
      "epoch": 0.2946260804208944,
      "grad_norm": 0.4891892373561859,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.3178,
      "step": 196
    },
    {
      "epoch": 0.2961292747087561,
      "grad_norm": 0.47990643978118896,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.3578,
      "step": 197
    },
    {
      "epoch": 0.2976324689966178,
      "grad_norm": 0.5234610438346863,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.305,
      "step": 198
    },
    {
      "epoch": 0.2991356632844795,
      "grad_norm": 0.5420705080032349,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2801,
      "step": 199
    },
    {
      "epoch": 0.3006388575723412,
      "grad_norm": 0.6559951305389404,
      "learning_rate": 0.0,
      "loss": 0.3526,
      "step": 200
    },
    {
      "epoch": 0.3006388575723412,
      "eval_loss": 0.34753015637397766,
      "eval_runtime": 54.7235,
      "eval_samples_per_second": 20.485,
      "eval_steps_per_second": 5.135,
      "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": 8.826038955225907e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}