{
  "best_metric": 0.01890052668750286,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.03191701575902653,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00015958507879513265,
      "grad_norm": 91.70167541503906,
      "learning_rate": 1e-05,
      "loss": 12.0695,
      "step": 1
    },
    {
      "epoch": 0.00015958507879513265,
      "eval_loss": 12.273541450500488,
      "eval_runtime": 806.9351,
      "eval_samples_per_second": 13.079,
      "eval_steps_per_second": 3.27,
      "step": 1
    },
    {
      "epoch": 0.0003191701575902653,
      "grad_norm": 102.95516967773438,
      "learning_rate": 2e-05,
      "loss": 11.8707,
      "step": 2
    },
    {
      "epoch": 0.00047875523638539794,
      "grad_norm": 94.14151000976562,
      "learning_rate": 3e-05,
      "loss": 10.347,
      "step": 3
    },
    {
      "epoch": 0.0006383403151805306,
      "grad_norm": 74.24517059326172,
      "learning_rate": 4e-05,
      "loss": 6.7574,
      "step": 4
    },
    {
      "epoch": 0.0007979253939756633,
      "grad_norm": 80.98632049560547,
      "learning_rate": 5e-05,
      "loss": 1.8561,
      "step": 5
    },
    {
      "epoch": 0.0009575104727707959,
      "grad_norm": 2.4279086589813232,
      "learning_rate": 6e-05,
      "loss": 0.0256,
      "step": 6
    },
    {
      "epoch": 0.0011170955515659286,
      "grad_norm": 0.00727560929954052,
      "learning_rate": 7e-05,
      "loss": 0.0001,
      "step": 7
    },
    {
      "epoch": 0.0012766806303610612,
      "grad_norm": 43.05037307739258,
      "learning_rate": 8e-05,
      "loss": 1.1148,
      "step": 8
    },
    {
      "epoch": 0.001436265709156194,
      "grad_norm": 243.4115753173828,
      "learning_rate": 9e-05,
      "loss": 3.5722,
      "step": 9
    },
    {
      "epoch": 0.0015958507879513266,
      "grad_norm": 0.0019942147191613913,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 10
    },
    {
      "epoch": 0.0017554358667464592,
      "grad_norm": 19.9483699798584,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.394,
      "step": 11
    },
    {
      "epoch": 0.0019150209455415918,
      "grad_norm": 6.683404445648193,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.2813,
      "step": 12
    },
    {
      "epoch": 0.0020746060243367246,
      "grad_norm": 23.342052459716797,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.8593,
      "step": 13
    },
    {
      "epoch": 0.002234191103131857,
      "grad_norm": 6.088518142700195,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.2156,
      "step": 14
    },
    {
      "epoch": 0.0023937761819269898,
      "grad_norm": 1.7149337530136108,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.1325,
      "step": 15
    },
    {
      "epoch": 0.0025533612607221224,
      "grad_norm": 0.6034479737281799,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.0729,
      "step": 16
    },
    {
      "epoch": 0.002712946339517255,
      "grad_norm": 1.5753737688064575,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.1833,
      "step": 17
    },
    {
      "epoch": 0.002872531418312388,
      "grad_norm": 1.3809409141540527,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.0813,
      "step": 18
    },
    {
      "epoch": 0.0030321164971075206,
      "grad_norm": 0.7744306325912476,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.061,
      "step": 19
    },
    {
      "epoch": 0.003191701575902653,
      "grad_norm": 3.3979737758636475,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.1148,
      "step": 20
    },
    {
      "epoch": 0.0033512866546977858,
      "grad_norm": 1.116274356842041,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.075,
      "step": 21
    },
    {
      "epoch": 0.0035108717334929184,
      "grad_norm": 2.1139369010925293,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.158,
      "step": 22
    },
    {
      "epoch": 0.003670456812288051,
      "grad_norm": 0.871885359287262,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.0271,
      "step": 23
    },
    {
      "epoch": 0.0038300418910831835,
      "grad_norm": 3.1295361518859863,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.0878,
      "step": 24
    },
    {
      "epoch": 0.003989626969878316,
      "grad_norm": 0.11067342013120651,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.0019,
      "step": 25
    },
    {
      "epoch": 0.004149212048673449,
      "grad_norm": 4.298526287078857,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.0829,
      "step": 26
    },
    {
      "epoch": 0.004308797127468581,
      "grad_norm": 6.691145896911621,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.0746,
      "step": 27
    },
    {
      "epoch": 0.004468382206263714,
      "grad_norm": 0.22768229246139526,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.0034,
      "step": 28
    },
    {
      "epoch": 0.004627967285058847,
      "grad_norm": 2.516052007675171,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.0364,
      "step": 29
    },
    {
      "epoch": 0.0047875523638539795,
      "grad_norm": 6.500372886657715,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.0606,
      "step": 30
    },
    {
      "epoch": 0.0049471374426491126,
      "grad_norm": 5.292369842529297,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.0527,
      "step": 31
    },
    {
      "epoch": 0.005106722521444245,
      "grad_norm": 5.458245754241943,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.0554,
      "step": 32
    },
    {
      "epoch": 0.005266307600239378,
      "grad_norm": 4.49389123916626,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.0532,
      "step": 33
    },
    {
      "epoch": 0.00542589267903451,
      "grad_norm": 1.8657965660095215,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.0099,
      "step": 34
    },
    {
      "epoch": 0.005585477757829643,
      "grad_norm": 10.926817893981934,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.0738,
      "step": 35
    },
    {
      "epoch": 0.005745062836624776,
      "grad_norm": 10.781157493591309,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.0477,
      "step": 36
    },
    {
      "epoch": 0.005904647915419908,
      "grad_norm": 9.783218383789062,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.1485,
      "step": 37
    },
    {
      "epoch": 0.006064232994215041,
      "grad_norm": 1.8303381204605103,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.0364,
      "step": 38
    },
    {
      "epoch": 0.006223818073010173,
      "grad_norm": 2.6647450923919678,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.0676,
      "step": 39
    },
    {
      "epoch": 0.006383403151805306,
      "grad_norm": 0.9746254086494446,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.0147,
      "step": 40
    },
    {
      "epoch": 0.0065429882306004385,
      "grad_norm": 2.5695393085479736,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.2064,
      "step": 41
    },
    {
      "epoch": 0.0067025733093955715,
      "grad_norm": 2.2885403633117676,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.0852,
      "step": 42
    },
    {
      "epoch": 0.0068621583881907046,
      "grad_norm": 2.5603082180023193,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.084,
      "step": 43
    },
    {
      "epoch": 0.007021743466985837,
      "grad_norm": 1.8648403882980347,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.0348,
      "step": 44
    },
    {
      "epoch": 0.00718132854578097,
      "grad_norm": 1.890605092048645,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.0602,
      "step": 45
    },
    {
      "epoch": 0.007340913624576102,
      "grad_norm": 3.5145490169525146,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.1955,
      "step": 46
    },
    {
      "epoch": 0.007500498703371235,
      "grad_norm": 2.490647315979004,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.0714,
      "step": 47
    },
    {
      "epoch": 0.007660083782166367,
      "grad_norm": 1.671553611755371,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.0452,
      "step": 48
    },
    {
      "epoch": 0.0078196688609615,
      "grad_norm": 1.9121739864349365,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.149,
      "step": 49
    },
    {
      "epoch": 0.007979253939756632,
      "grad_norm": 1.8590110540390015,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.0543,
      "step": 50
    },
    {
      "epoch": 0.007979253939756632,
      "eval_loss": 0.04401622712612152,
      "eval_runtime": 811.03,
      "eval_samples_per_second": 13.013,
      "eval_steps_per_second": 3.254,
      "step": 50
    },
    {
      "epoch": 0.008138839018551766,
      "grad_norm": 2.80737566947937,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.2495,
      "step": 51
    },
    {
      "epoch": 0.008298424097346898,
      "grad_norm": 2.0572423934936523,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.1319,
      "step": 52
    },
    {
      "epoch": 0.00845800917614203,
      "grad_norm": 1.2492311000823975,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.0722,
      "step": 53
    },
    {
      "epoch": 0.008617594254937163,
      "grad_norm": 0.6757921576499939,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.0229,
      "step": 54
    },
    {
      "epoch": 0.008777179333732297,
      "grad_norm": 1.376996636390686,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.0408,
      "step": 55
    },
    {
      "epoch": 0.008936764412527429,
      "grad_norm": 0.6346701979637146,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.0167,
      "step": 56
    },
    {
      "epoch": 0.00909634949132256,
      "grad_norm": 1.844874620437622,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.0575,
      "step": 57
    },
    {
      "epoch": 0.009255934570117695,
      "grad_norm": 0.5899195671081543,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.0106,
      "step": 58
    },
    {
      "epoch": 0.009415519648912827,
      "grad_norm": 2.408447504043579,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.0271,
      "step": 59
    },
    {
      "epoch": 0.009575104727707959,
      "grad_norm": 2.0290961265563965,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.0196,
      "step": 60
    },
    {
      "epoch": 0.009734689806503091,
      "grad_norm": 1.2359588146209717,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.012,
      "step": 61
    },
    {
      "epoch": 0.009894274885298225,
      "grad_norm": 2.971074104309082,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.0423,
      "step": 62
    },
    {
      "epoch": 0.010053859964093357,
      "grad_norm": 0.7732223868370056,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.0062,
      "step": 63
    },
    {
      "epoch": 0.01021344504288849,
      "grad_norm": 1.8813352584838867,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.0269,
      "step": 64
    },
    {
      "epoch": 0.010373030121683623,
      "grad_norm": 1.5054543018341064,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.0483,
      "step": 65
    },
    {
      "epoch": 0.010532615200478756,
      "grad_norm": 1.6870179176330566,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.0238,
      "step": 66
    },
    {
      "epoch": 0.010692200279273888,
      "grad_norm": 0.34705960750579834,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.0041,
      "step": 67
    },
    {
      "epoch": 0.01085178535806902,
      "grad_norm": 0.039703771471977234,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.0004,
      "step": 68
    },
    {
      "epoch": 0.011011370436864154,
      "grad_norm": 0.09697467088699341,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.0015,
      "step": 69
    },
    {
      "epoch": 0.011170955515659286,
      "grad_norm": 0.7236990928649902,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.0047,
      "step": 70
    },
    {
      "epoch": 0.011330540594454418,
      "grad_norm": 3.882035970687866,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.1736,
      "step": 71
    },
    {
      "epoch": 0.011490125673249552,
      "grad_norm": 0.02478318102657795,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.0004,
      "step": 72
    },
    {
      "epoch": 0.011649710752044684,
      "grad_norm": 0.7587752938270569,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.0058,
      "step": 73
    },
    {
      "epoch": 0.011809295830839816,
      "grad_norm": 0.25580620765686035,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.0027,
      "step": 74
    },
    {
      "epoch": 0.011968880909634948,
      "grad_norm": 0.5692357420921326,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.0046,
      "step": 75
    },
    {
      "epoch": 0.012128465988430082,
      "grad_norm": 2.0589792728424072,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.0167,
      "step": 76
    },
    {
      "epoch": 0.012288051067225214,
      "grad_norm": 0.18529736995697021,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.0023,
      "step": 77
    },
    {
      "epoch": 0.012447636146020347,
      "grad_norm": 0.2441205531358719,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.0005,
      "step": 78
    },
    {
      "epoch": 0.01260722122481548,
      "grad_norm": 0.5898708701133728,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.0062,
      "step": 79
    },
    {
      "epoch": 0.012766806303610613,
      "grad_norm": 0.183292418718338,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.002,
      "step": 80
    },
    {
      "epoch": 0.012926391382405745,
      "grad_norm": 3.9771249294281006,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.0648,
      "step": 81
    },
    {
      "epoch": 0.013085976461200877,
      "grad_norm": 0.03600252419710159,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.0008,
      "step": 82
    },
    {
      "epoch": 0.013245561539996011,
      "grad_norm": 0.06460081040859222,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.0011,
      "step": 83
    },
    {
      "epoch": 0.013405146618791143,
      "grad_norm": 2.436270236968994,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.0334,
      "step": 84
    },
    {
      "epoch": 0.013564731697586275,
      "grad_norm": 0.20913580060005188,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.0015,
      "step": 85
    },
    {
      "epoch": 0.013724316776381409,
      "grad_norm": 1.4161549806594849,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.0206,
      "step": 86
    },
    {
      "epoch": 0.013883901855176541,
      "grad_norm": 0.4316036105155945,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.0052,
      "step": 87
    },
    {
      "epoch": 0.014043486933971673,
      "grad_norm": 4.485915184020996,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.0499,
      "step": 88
    },
    {
      "epoch": 0.014203072012766806,
      "grad_norm": 2.920517921447754,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.0181,
      "step": 89
    },
    {
      "epoch": 0.01436265709156194,
      "grad_norm": 3.355133533477783,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.1436,
      "step": 90
    },
    {
      "epoch": 0.014522242170357072,
      "grad_norm": 0.17289471626281738,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.0013,
      "step": 91
    },
    {
      "epoch": 0.014681827249152204,
      "grad_norm": 0.33633261919021606,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.0022,
      "step": 92
    },
    {
      "epoch": 0.014841412327947338,
      "grad_norm": 0.49620193243026733,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.0034,
      "step": 93
    },
    {
      "epoch": 0.01500099740674247,
      "grad_norm": 1.4084627628326416,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.058,
      "step": 94
    },
    {
      "epoch": 0.015160582485537602,
      "grad_norm": 0.2431686669588089,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.0012,
      "step": 95
    },
    {
      "epoch": 0.015320167564332734,
      "grad_norm": 0.4875892400741577,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.0059,
      "step": 96
    },
    {
      "epoch": 0.015479752643127868,
      "grad_norm": 0.030450213700532913,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.0007,
      "step": 97
    },
    {
      "epoch": 0.015639337721923,
      "grad_norm": 2.137424945831299,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.0229,
      "step": 98
    },
    {
      "epoch": 0.015798922800718134,
      "grad_norm": 0.02448270097374916,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.0006,
      "step": 99
    },
    {
      "epoch": 0.015958507879513265,
      "grad_norm": 1.3924957513809204,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.0109,
      "step": 100
    },
    {
      "epoch": 0.015958507879513265,
      "eval_loss": 0.04514055326581001,
      "eval_runtime": 810.186,
      "eval_samples_per_second": 13.027,
      "eval_steps_per_second": 3.257,
      "step": 100
    },
    {
      "epoch": 0.0161180929583084,
      "grad_norm": 7.849484443664551,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2547,
      "step": 101
    },
    {
      "epoch": 0.016277678037103532,
      "grad_norm": 4.806069850921631,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.1816,
      "step": 102
    },
    {
      "epoch": 0.016437263115898663,
      "grad_norm": 0.0011642908211797476,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.0,
      "step": 103
    },
    {
      "epoch": 0.016596848194693797,
      "grad_norm": 2.227031946182251,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.0149,
      "step": 104
    },
    {
      "epoch": 0.016756433273488927,
      "grad_norm": 2.6253600120544434,
      "learning_rate": 5e-05,
      "loss": 0.1116,
      "step": 105
    },
    {
      "epoch": 0.01691601835228406,
      "grad_norm": 0.027695702388882637,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.0004,
      "step": 106
    },
    {
      "epoch": 0.017075603431079195,
      "grad_norm": 0.0602019838988781,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.0015,
      "step": 107
    },
    {
      "epoch": 0.017235188509874325,
      "grad_norm": 3.732633590698242,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.0886,
      "step": 108
    },
    {
      "epoch": 0.01739477358866946,
      "grad_norm": 1.5315437316894531,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.0201,
      "step": 109
    },
    {
      "epoch": 0.017554358667464593,
      "grad_norm": 0.48249104619026184,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.0074,
      "step": 110
    },
    {
      "epoch": 0.017713943746259724,
      "grad_norm": 0.7455866932868958,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.0107,
      "step": 111
    },
    {
      "epoch": 0.017873528825054857,
      "grad_norm": 0.8390172123908997,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.0156,
      "step": 112
    },
    {
      "epoch": 0.01803311390384999,
      "grad_norm": 1.6658551692962646,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.0256,
      "step": 113
    },
    {
      "epoch": 0.01819269898264512,
      "grad_norm": 0.5601900219917297,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.0053,
      "step": 114
    },
    {
      "epoch": 0.018352284061440256,
      "grad_norm": 0.8038157820701599,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.0054,
      "step": 115
    },
    {
      "epoch": 0.01851186914023539,
      "grad_norm": 1.9648799896240234,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.0076,
      "step": 116
    },
    {
      "epoch": 0.01867145421903052,
      "grad_norm": 0.047080546617507935,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.0003,
      "step": 117
    },
    {
      "epoch": 0.018831039297825654,
      "grad_norm": 2.6113030910491943,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.0139,
      "step": 118
    },
    {
      "epoch": 0.018990624376620784,
      "grad_norm": 4.533097743988037,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.1584,
      "step": 119
    },
    {
      "epoch": 0.019150209455415918,
      "grad_norm": 3.181793212890625,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.1118,
      "step": 120
    },
    {
      "epoch": 0.019309794534211052,
      "grad_norm": 3.158857822418213,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.1069,
      "step": 121
    },
    {
      "epoch": 0.019469379613006182,
      "grad_norm": 2.2730276584625244,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.0269,
      "step": 122
    },
    {
      "epoch": 0.019628964691801316,
      "grad_norm": 0.06171315908432007,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.0008,
      "step": 123
    },
    {
      "epoch": 0.01978854977059645,
      "grad_norm": 1.5438772439956665,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.1286,
      "step": 124
    },
    {
      "epoch": 0.01994813484939158,
      "grad_norm": 2.403961181640625,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.0436,
      "step": 125
    },
    {
      "epoch": 0.020107719928186715,
      "grad_norm": 0.01171303354203701,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0002,
      "step": 126
    },
    {
      "epoch": 0.02026730500698185,
      "grad_norm": 0.16018010675907135,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.0008,
      "step": 127
    },
    {
      "epoch": 0.02042689008577698,
      "grad_norm": 0.08517315983772278,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.0012,
      "step": 128
    },
    {
      "epoch": 0.020586475164572113,
      "grad_norm": 5.664571285247803,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.0716,
      "step": 129
    },
    {
      "epoch": 0.020746060243367247,
      "grad_norm": 1.4501450061798096,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.0493,
      "step": 130
    },
    {
      "epoch": 0.020905645322162377,
      "grad_norm": 3.897109270095825,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.0709,
      "step": 131
    },
    {
      "epoch": 0.02106523040095751,
      "grad_norm": 0.5194004774093628,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.0075,
      "step": 132
    },
    {
      "epoch": 0.02122481547975264,
      "grad_norm": 1.4536826610565186,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.0648,
      "step": 133
    },
    {
      "epoch": 0.021384400558547775,
      "grad_norm": 0.35453930497169495,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.0063,
      "step": 134
    },
    {
      "epoch": 0.02154398563734291,
      "grad_norm": 1.3657701015472412,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.0444,
      "step": 135
    },
    {
      "epoch": 0.02170357071613804,
      "grad_norm": 0.612852156162262,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.0116,
      "step": 136
    },
    {
      "epoch": 0.021863155794933174,
      "grad_norm": 1.509804606437683,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.018,
      "step": 137
    },
    {
      "epoch": 0.022022740873728307,
      "grad_norm": 0.2642381191253662,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0049,
      "step": 138
    },
    {
      "epoch": 0.022182325952523438,
      "grad_norm": 0.14495153725147247,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0027,
      "step": 139
    },
    {
      "epoch": 0.022341911031318572,
      "grad_norm": 3.297553539276123,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.0411,
      "step": 140
    },
    {
      "epoch": 0.022501496110113706,
      "grad_norm": 1.9078449010849,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.112,
      "step": 141
    },
    {
      "epoch": 0.022661081188908836,
      "grad_norm": 1.2824450731277466,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0153,
      "step": 142
    },
    {
      "epoch": 0.02282066626770397,
      "grad_norm": 0.3859853148460388,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0117,
      "step": 143
    },
    {
      "epoch": 0.022980251346499104,
      "grad_norm": 1.5403869152069092,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0547,
      "step": 144
    },
    {
      "epoch": 0.023139836425294234,
      "grad_norm": 0.2209412157535553,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.0052,
      "step": 145
    },
    {
      "epoch": 0.023299421504089368,
      "grad_norm": 0.6963176727294922,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0083,
      "step": 146
    },
    {
      "epoch": 0.0234590065828845,
      "grad_norm": 1.7133246660232544,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.0639,
      "step": 147
    },
    {
      "epoch": 0.023618591661679633,
      "grad_norm": 3.069453477859497,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.0582,
      "step": 148
    },
    {
      "epoch": 0.023778176740474766,
      "grad_norm": 1.1332793235778809,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.0247,
      "step": 149
    },
    {
      "epoch": 0.023937761819269897,
      "grad_norm": 1.3368282318115234,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.0127,
      "step": 150
    },
    {
      "epoch": 0.023937761819269897,
      "eval_loss": 0.020805498585104942,
      "eval_runtime": 810.7432,
      "eval_samples_per_second": 13.018,
      "eval_steps_per_second": 3.255,
      "step": 150
    },
    {
      "epoch": 0.02409734689806503,
      "grad_norm": 5.319791316986084,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.3695,
      "step": 151
    },
    {
      "epoch": 0.024256931976860165,
      "grad_norm": 0.9884998798370361,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.0205,
      "step": 152
    },
    {
      "epoch": 0.024416517055655295,
      "grad_norm": 0.5342042446136475,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.0126,
      "step": 153
    },
    {
      "epoch": 0.02457610213445043,
      "grad_norm": 0.92693692445755,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0095,
      "step": 154
    },
    {
      "epoch": 0.024735687213245563,
      "grad_norm": 0.8903139233589172,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0185,
      "step": 155
    },
    {
      "epoch": 0.024895272292040693,
      "grad_norm": 0.4049259126186371,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.004,
      "step": 156
    },
    {
      "epoch": 0.025054857370835827,
      "grad_norm": 2.9271440505981445,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.0377,
      "step": 157
    },
    {
      "epoch": 0.02521444244963096,
      "grad_norm": 1.0630238056182861,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.0171,
      "step": 158
    },
    {
      "epoch": 0.02537402752842609,
      "grad_norm": 0.2653367519378662,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.0028,
      "step": 159
    },
    {
      "epoch": 0.025533612607221225,
      "grad_norm": 3.305790901184082,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.0524,
      "step": 160
    },
    {
      "epoch": 0.025693197686016356,
      "grad_norm": 0.7967613339424133,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.0069,
      "step": 161
    },
    {
      "epoch": 0.02585278276481149,
      "grad_norm": 2.3201496601104736,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0165,
      "step": 162
    },
    {
      "epoch": 0.026012367843606624,
      "grad_norm": 1.7085978984832764,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.0816,
      "step": 163
    },
    {
      "epoch": 0.026171952922401754,
      "grad_norm": 0.14983557164669037,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0027,
      "step": 164
    },
    {
      "epoch": 0.026331538001196888,
      "grad_norm": 0.12315572798252106,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.0021,
      "step": 165
    },
    {
      "epoch": 0.026491123079992022,
      "grad_norm": 0.8559834361076355,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.0122,
      "step": 166
    },
    {
      "epoch": 0.026650708158787152,
      "grad_norm": 0.5335274934768677,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.0048,
      "step": 167
    },
    {
      "epoch": 0.026810293237582286,
      "grad_norm": 0.037546735256910324,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.0009,
      "step": 168
    },
    {
      "epoch": 0.02696987831637742,
      "grad_norm": 0.3159119486808777,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.0047,
      "step": 169
    },
    {
      "epoch": 0.02712946339517255,
      "grad_norm": 1.618561863899231,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.0706,
      "step": 170
    },
    {
      "epoch": 0.027289048473967684,
      "grad_norm": 0.5616846084594727,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.0075,
      "step": 171
    },
    {
      "epoch": 0.027448633552762818,
      "grad_norm": 0.34773898124694824,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.0044,
      "step": 172
    },
    {
      "epoch": 0.02760821863155795,
      "grad_norm": 1.9005978107452393,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.0473,
      "step": 173
    },
    {
      "epoch": 0.027767803710353083,
      "grad_norm": 0.9834507703781128,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.0106,
      "step": 174
    },
    {
      "epoch": 0.027927388789148213,
      "grad_norm": 1.5049645900726318,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.0158,
      "step": 175
    },
    {
      "epoch": 0.028086973867943347,
      "grad_norm": 3.9066481590270996,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.0292,
      "step": 176
    },
    {
      "epoch": 0.02824655894673848,
      "grad_norm": 0.17276357114315033,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.0021,
      "step": 177
    },
    {
      "epoch": 0.02840614402553361,
      "grad_norm": 2.098283052444458,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.0825,
      "step": 178
    },
    {
      "epoch": 0.028565729104328745,
      "grad_norm": 0.1188969761133194,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0017,
      "step": 179
    },
    {
      "epoch": 0.02872531418312388,
      "grad_norm": 1.728696346282959,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.0407,
      "step": 180
    },
    {
      "epoch": 0.02888489926191901,
      "grad_norm": 0.3086645305156708,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.005,
      "step": 181
    },
    {
      "epoch": 0.029044484340714143,
      "grad_norm": 0.23279613256454468,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0025,
      "step": 182
    },
    {
      "epoch": 0.029204069419509277,
      "grad_norm": 0.0351664274930954,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.0008,
      "step": 183
    },
    {
      "epoch": 0.029363654498304408,
      "grad_norm": 1.2695133686065674,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0112,
      "step": 184
    },
    {
      "epoch": 0.02952323957709954,
      "grad_norm": 1.0305896997451782,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.007,
      "step": 185
    },
    {
      "epoch": 0.029682824655894675,
      "grad_norm": 1.0030447244644165,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0117,
      "step": 186
    },
    {
      "epoch": 0.029842409734689806,
      "grad_norm": 2.233741044998169,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0504,
      "step": 187
    },
    {
      "epoch": 0.03000199481348494,
      "grad_norm": 0.5133768320083618,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.0116,
      "step": 188
    },
    {
      "epoch": 0.03016157989228007,
      "grad_norm": 1.5447137355804443,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.0383,
      "step": 189
    },
    {
      "epoch": 0.030321164971075204,
      "grad_norm": 3.3091111183166504,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0318,
      "step": 190
    },
    {
      "epoch": 0.030480750049870338,
      "grad_norm": 0.20804284512996674,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.0029,
      "step": 191
    },
    {
      "epoch": 0.03064033512866547,
      "grad_norm": 0.060637764632701874,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0012,
      "step": 192
    },
    {
      "epoch": 0.030799920207460602,
      "grad_norm": 1.176334261894226,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0293,
      "step": 193
    },
    {
      "epoch": 0.030959505286255736,
      "grad_norm": 0.13167013227939606,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0027,
      "step": 194
    },
    {
      "epoch": 0.031119090365050867,
      "grad_norm": 0.6898083090782166,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.0057,
      "step": 195
    },
    {
      "epoch": 0.031278675443846,
      "grad_norm": 2.790844440460205,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.054,
      "step": 196
    },
    {
      "epoch": 0.031438260522641134,
      "grad_norm": 1.2391724586486816,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0111,
      "step": 197
    },
    {
      "epoch": 0.03159784560143627,
      "grad_norm": 0.817627489566803,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.0087,
      "step": 198
    },
    {
      "epoch": 0.031757430680231395,
      "grad_norm": 0.04306378588080406,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0012,
      "step": 199
    },
    {
      "epoch": 0.03191701575902653,
      "grad_norm": 1.692249059677124,
      "learning_rate": 0.0,
      "loss": 0.0239,
      "step": 200
    },
    {
      "epoch": 0.03191701575902653,
      "eval_loss": 0.01890052668750286,
      "eval_runtime": 810.6733,
      "eval_samples_per_second": 13.019,
      "eval_steps_per_second": 3.255,
      "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": 3.0321122932791706e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}