{
  "best_metric": 1.143795371055603,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.011536853034913402,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.768426517456701e-05,
      "grad_norm": 8.791138648986816,
      "learning_rate": 1e-05,
      "loss": 5.2556,
      "step": 1
    },
    {
      "epoch": 5.768426517456701e-05,
      "eval_loss": 2.1517772674560547,
      "eval_runtime": 569.1798,
      "eval_samples_per_second": 51.297,
      "eval_steps_per_second": 12.825,
      "step": 1
    },
    {
      "epoch": 0.00011536853034913402,
      "grad_norm": 9.306244850158691,
      "learning_rate": 2e-05,
      "loss": 7.0134,
      "step": 2
    },
    {
      "epoch": 0.00017305279552370101,
      "grad_norm": 10.867076873779297,
      "learning_rate": 3e-05,
      "loss": 7.2109,
      "step": 3
    },
    {
      "epoch": 0.00023073706069826804,
      "grad_norm": 9.395054817199707,
      "learning_rate": 4e-05,
      "loss": 7.2629,
      "step": 4
    },
    {
      "epoch": 0.00028842132587283503,
      "grad_norm": 8.687573432922363,
      "learning_rate": 5e-05,
      "loss": 7.6235,
      "step": 5
    },
    {
      "epoch": 0.00034610559104740203,
      "grad_norm": 7.54335355758667,
      "learning_rate": 6e-05,
      "loss": 6.2341,
      "step": 6
    },
    {
      "epoch": 0.0004037898562219691,
      "grad_norm": 7.846302032470703,
      "learning_rate": 7e-05,
      "loss": 7.5275,
      "step": 7
    },
    {
      "epoch": 0.0004614741213965361,
      "grad_norm": 7.37999963760376,
      "learning_rate": 8e-05,
      "loss": 5.6271,
      "step": 8
    },
    {
      "epoch": 0.0005191583865711031,
      "grad_norm": 7.946924209594727,
      "learning_rate": 9e-05,
      "loss": 6.4196,
      "step": 9
    },
    {
      "epoch": 0.0005768426517456701,
      "grad_norm": 8.877602577209473,
      "learning_rate": 0.0001,
      "loss": 5.8956,
      "step": 10
    },
    {
      "epoch": 0.0006345269169202371,
      "grad_norm": 7.5694475173950195,
      "learning_rate": 9.999316524962345e-05,
      "loss": 5.829,
      "step": 11
    },
    {
      "epoch": 0.0006922111820948041,
      "grad_norm": 6.883509159088135,
      "learning_rate": 9.997266286704631e-05,
      "loss": 5.1214,
      "step": 12
    },
    {
      "epoch": 0.000749895447269371,
      "grad_norm": 7.162557601928711,
      "learning_rate": 9.993849845741524e-05,
      "loss": 5.1232,
      "step": 13
    },
    {
      "epoch": 0.0008075797124439382,
      "grad_norm": 8.084110260009766,
      "learning_rate": 9.989068136093873e-05,
      "loss": 5.9084,
      "step": 14
    },
    {
      "epoch": 0.0008652639776185052,
      "grad_norm": 7.88531494140625,
      "learning_rate": 9.98292246503335e-05,
      "loss": 6.0177,
      "step": 15
    },
    {
      "epoch": 0.0009229482427930721,
      "grad_norm": 9.236763000488281,
      "learning_rate": 9.975414512725057e-05,
      "loss": 6.1991,
      "step": 16
    },
    {
      "epoch": 0.000980632507967639,
      "grad_norm": 7.625688076019287,
      "learning_rate": 9.966546331768191e-05,
      "loss": 5.3697,
      "step": 17
    },
    {
      "epoch": 0.0010383167731422061,
      "grad_norm": 7.455811977386475,
      "learning_rate": 9.956320346634876e-05,
      "loss": 5.2168,
      "step": 18
    },
    {
      "epoch": 0.0010960010383167732,
      "grad_norm": 7.085381507873535,
      "learning_rate": 9.944739353007344e-05,
      "loss": 5.3559,
      "step": 19
    },
    {
      "epoch": 0.0011536853034913401,
      "grad_norm": 7.614997863769531,
      "learning_rate": 9.931806517013612e-05,
      "loss": 5.2088,
      "step": 20
    },
    {
      "epoch": 0.0012113695686659072,
      "grad_norm": 7.585221767425537,
      "learning_rate": 9.917525374361912e-05,
      "loss": 5.522,
      "step": 21
    },
    {
      "epoch": 0.0012690538338404741,
      "grad_norm": 7.2149834632873535,
      "learning_rate": 9.901899829374047e-05,
      "loss": 4.8715,
      "step": 22
    },
    {
      "epoch": 0.0013267380990150412,
      "grad_norm": 7.463019371032715,
      "learning_rate": 9.884934153917997e-05,
      "loss": 5.569,
      "step": 23
    },
    {
      "epoch": 0.0013844223641896081,
      "grad_norm": 7.452372074127197,
      "learning_rate": 9.86663298624003e-05,
      "loss": 5.977,
      "step": 24
    },
    {
      "epoch": 0.0014421066293641752,
      "grad_norm": 6.505922794342041,
      "learning_rate": 9.847001329696653e-05,
      "loss": 5.202,
      "step": 25
    },
    {
      "epoch": 0.001499790894538742,
      "grad_norm": 7.775734901428223,
      "learning_rate": 9.826044551386744e-05,
      "loss": 5.2795,
      "step": 26
    },
    {
      "epoch": 0.0015574751597133092,
      "grad_norm": 5.850821495056152,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.5712,
      "step": 27
    },
    {
      "epoch": 0.0016151594248878763,
      "grad_norm": 6.693484306335449,
      "learning_rate": 9.780178907671789e-05,
      "loss": 5.0794,
      "step": 28
    },
    {
      "epoch": 0.0016728436900624432,
      "grad_norm": 6.86115026473999,
      "learning_rate": 9.755282581475769e-05,
      "loss": 5.1199,
      "step": 29
    },
    {
      "epoch": 0.0017305279552370103,
      "grad_norm": 6.308779239654541,
      "learning_rate": 9.729086208503174e-05,
      "loss": 5.4776,
      "step": 30
    },
    {
      "epoch": 0.0017882122204115772,
      "grad_norm": 6.916812896728516,
      "learning_rate": 9.701596950580806e-05,
      "loss": 4.6694,
      "step": 31
    },
    {
      "epoch": 0.0018458964855861443,
      "grad_norm": 6.394667625427246,
      "learning_rate": 9.672822322997305e-05,
      "loss": 4.8313,
      "step": 32
    },
    {
      "epoch": 0.0019035807507607112,
      "grad_norm": 7.03117036819458,
      "learning_rate": 9.642770192448536e-05,
      "loss": 4.8541,
      "step": 33
    },
    {
      "epoch": 0.001961265015935278,
      "grad_norm": 7.209469795227051,
      "learning_rate": 9.611448774886924e-05,
      "loss": 5.5214,
      "step": 34
    },
    {
      "epoch": 0.0020189492811098454,
      "grad_norm": 6.553521633148193,
      "learning_rate": 9.578866633275288e-05,
      "loss": 4.4674,
      "step": 35
    },
    {
      "epoch": 0.0020766335462844123,
      "grad_norm": 7.645153045654297,
      "learning_rate": 9.545032675245813e-05,
      "loss": 5.4816,
      "step": 36
    },
    {
      "epoch": 0.002134317811458979,
      "grad_norm": 7.72512149810791,
      "learning_rate": 9.509956150664796e-05,
      "loss": 6.149,
      "step": 37
    },
    {
      "epoch": 0.0021920020766335465,
      "grad_norm": 6.975688457489014,
      "learning_rate": 9.473646649103818e-05,
      "loss": 5.0321,
      "step": 38
    },
    {
      "epoch": 0.0022496863418081134,
      "grad_norm": 7.110767841339111,
      "learning_rate": 9.43611409721806e-05,
      "loss": 5.1251,
      "step": 39
    },
    {
      "epoch": 0.0023073706069826803,
      "grad_norm": 7.385117053985596,
      "learning_rate": 9.397368756032445e-05,
      "loss": 5.0338,
      "step": 40
    },
    {
      "epoch": 0.002365054872157247,
      "grad_norm": 8.237039566040039,
      "learning_rate": 9.357421218136386e-05,
      "loss": 5.1602,
      "step": 41
    },
    {
      "epoch": 0.0024227391373318145,
      "grad_norm": 8.620194435119629,
      "learning_rate": 9.316282404787871e-05,
      "loss": 5.9398,
      "step": 42
    },
    {
      "epoch": 0.0024804234025063814,
      "grad_norm": 8.013171195983887,
      "learning_rate": 9.273963562927695e-05,
      "loss": 5.9314,
      "step": 43
    },
    {
      "epoch": 0.0025381076676809482,
      "grad_norm": 8.38439655303955,
      "learning_rate": 9.230476262104677e-05,
      "loss": 4.8618,
      "step": 44
    },
    {
      "epoch": 0.002595791932855515,
      "grad_norm": 7.812241554260254,
      "learning_rate": 9.185832391312644e-05,
      "loss": 5.4533,
      "step": 45
    },
    {
      "epoch": 0.0026534761980300824,
      "grad_norm": 8.033610343933105,
      "learning_rate": 9.140044155740101e-05,
      "loss": 5.4105,
      "step": 46
    },
    {
      "epoch": 0.0027111604632046493,
      "grad_norm": 8.222458839416504,
      "learning_rate": 9.093124073433463e-05,
      "loss": 5.2003,
      "step": 47
    },
    {
      "epoch": 0.0027688447283792162,
      "grad_norm": 8.728143692016602,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.8905,
      "step": 48
    },
    {
      "epoch": 0.0028265289935537835,
      "grad_norm": 9.663448333740234,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.5355,
      "step": 49
    },
    {
      "epoch": 0.0028842132587283504,
      "grad_norm": 9.29330062866211,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.9791,
      "step": 50
    },
    {
      "epoch": 0.0028842132587283504,
      "eval_loss": 1.383618712425232,
      "eval_runtime": 570.1385,
      "eval_samples_per_second": 51.21,
      "eval_steps_per_second": 12.804,
      "step": 50
    },
    {
      "epoch": 0.0029418975239029173,
      "grad_norm": 12.091597557067871,
      "learning_rate": 8.894386393810563e-05,
      "loss": 5.8224,
      "step": 51
    },
    {
      "epoch": 0.002999581789077484,
      "grad_norm": 12.103343963623047,
      "learning_rate": 8.842005554284296e-05,
      "loss": 4.6744,
      "step": 52
    },
    {
      "epoch": 0.0030572660542520515,
      "grad_norm": 8.64933967590332,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.3129,
      "step": 53
    },
    {
      "epoch": 0.0031149503194266184,
      "grad_norm": 5.941163063049316,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.9331,
      "step": 54
    },
    {
      "epoch": 0.0031726345846011853,
      "grad_norm": 5.051603317260742,
      "learning_rate": 8.678619553365659e-05,
      "loss": 4.1336,
      "step": 55
    },
    {
      "epoch": 0.0032303188497757526,
      "grad_norm": 6.365749835968018,
      "learning_rate": 8.622126023955446e-05,
      "loss": 5.1878,
      "step": 56
    },
    {
      "epoch": 0.0032880031149503195,
      "grad_norm": 7.5129075050354,
      "learning_rate": 8.564642241456986e-05,
      "loss": 5.2448,
      "step": 57
    },
    {
      "epoch": 0.0033456873801248864,
      "grad_norm": 6.239485263824463,
      "learning_rate": 8.506183921362443e-05,
      "loss": 5.0404,
      "step": 58
    },
    {
      "epoch": 0.0034033716452994533,
      "grad_norm": 5.955112457275391,
      "learning_rate": 8.44676704559283e-05,
      "loss": 4.5158,
      "step": 59
    },
    {
      "epoch": 0.0034610559104740206,
      "grad_norm": 4.5843610763549805,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.7536,
      "step": 60
    },
    {
      "epoch": 0.0035187401756485875,
      "grad_norm": 4.69848108291626,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.9306,
      "step": 61
    },
    {
      "epoch": 0.0035764244408231544,
      "grad_norm": 5.393914699554443,
      "learning_rate": 8.262928807620843e-05,
      "loss": 5.3387,
      "step": 62
    },
    {
      "epoch": 0.0036341087059977213,
      "grad_norm": 4.38604736328125,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.7064,
      "step": 63
    },
    {
      "epoch": 0.0036917929711722886,
      "grad_norm": 4.157768249511719,
      "learning_rate": 8.135881792367686e-05,
      "loss": 4.1088,
      "step": 64
    },
    {
      "epoch": 0.0037494772363468555,
      "grad_norm": 5.003042697906494,
      "learning_rate": 8.07106356344834e-05,
      "loss": 4.7108,
      "step": 65
    },
    {
      "epoch": 0.0038071615015214224,
      "grad_norm": 4.2943115234375,
      "learning_rate": 8.005405736415126e-05,
      "loss": 4.5644,
      "step": 66
    },
    {
      "epoch": 0.0038648457666959897,
      "grad_norm": 5.018723964691162,
      "learning_rate": 7.938926261462366e-05,
      "loss": 5.4286,
      "step": 67
    },
    {
      "epoch": 0.003922530031870556,
      "grad_norm": 4.28971529006958,
      "learning_rate": 7.871643313414718e-05,
      "loss": 4.2821,
      "step": 68
    },
    {
      "epoch": 0.003980214297045124,
      "grad_norm": 5.047732353210449,
      "learning_rate": 7.803575286758364e-05,
      "loss": 4.4904,
      "step": 69
    },
    {
      "epoch": 0.004037898562219691,
      "grad_norm": 4.882722854614258,
      "learning_rate": 7.734740790612136e-05,
      "loss": 4.5216,
      "step": 70
    },
    {
      "epoch": 0.004095582827394258,
      "grad_norm": 4.752922534942627,
      "learning_rate": 7.66515864363997e-05,
      "loss": 4.5764,
      "step": 71
    },
    {
      "epoch": 0.0041532670925688246,
      "grad_norm": 4.378620624542236,
      "learning_rate": 7.594847868906076e-05,
      "loss": 4.238,
      "step": 72
    },
    {
      "epoch": 0.0042109513577433914,
      "grad_norm": 4.599791526794434,
      "learning_rate": 7.52382768867422e-05,
      "loss": 4.7556,
      "step": 73
    },
    {
      "epoch": 0.004268635622917958,
      "grad_norm": 5.39775276184082,
      "learning_rate": 7.452117519152542e-05,
      "loss": 5.781,
      "step": 74
    },
    {
      "epoch": 0.004326319888092525,
      "grad_norm": 4.7623291015625,
      "learning_rate": 7.379736965185368e-05,
      "loss": 4.9013,
      "step": 75
    },
    {
      "epoch": 0.004384004153267093,
      "grad_norm": 5.231910705566406,
      "learning_rate": 7.30670581489344e-05,
      "loss": 5.157,
      "step": 76
    },
    {
      "epoch": 0.00444168841844166,
      "grad_norm": 5.055672645568848,
      "learning_rate": 7.233044034264034e-05,
      "loss": 5.1334,
      "step": 77
    },
    {
      "epoch": 0.004499372683616227,
      "grad_norm": 5.0222883224487305,
      "learning_rate": 7.158771761692464e-05,
      "loss": 5.3946,
      "step": 78
    },
    {
      "epoch": 0.004557056948790794,
      "grad_norm": 4.748698711395264,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.5471,
      "step": 79
    },
    {
      "epoch": 0.0046147412139653605,
      "grad_norm": 4.822394371032715,
      "learning_rate": 7.008477123264848e-05,
      "loss": 4.7816,
      "step": 80
    },
    {
      "epoch": 0.004672425479139927,
      "grad_norm": 4.734062194824219,
      "learning_rate": 6.932495846462261e-05,
      "loss": 3.9951,
      "step": 81
    },
    {
      "epoch": 0.004730109744314494,
      "grad_norm": 4.876369953155518,
      "learning_rate": 6.855986244591104e-05,
      "loss": 4.2489,
      "step": 82
    },
    {
      "epoch": 0.004787794009489062,
      "grad_norm": 5.225030899047852,
      "learning_rate": 6.778969234612584e-05,
      "loss": 5.0273,
      "step": 83
    },
    {
      "epoch": 0.004845478274663629,
      "grad_norm": 5.081971168518066,
      "learning_rate": 6.701465872208216e-05,
      "loss": 4.6178,
      "step": 84
    },
    {
      "epoch": 0.004903162539838196,
      "grad_norm": 5.925520420074463,
      "learning_rate": 6.623497346023418e-05,
      "loss": 5.6974,
      "step": 85
    },
    {
      "epoch": 0.004960846805012763,
      "grad_norm": 6.199829578399658,
      "learning_rate": 6.545084971874738e-05,
      "loss": 5.3826,
      "step": 86
    },
    {
      "epoch": 0.00501853107018733,
      "grad_norm": 5.986096382141113,
      "learning_rate": 6.466250186922325e-05,
      "loss": 4.9132,
      "step": 87
    },
    {
      "epoch": 0.0050762153353618965,
      "grad_norm": 5.488864898681641,
      "learning_rate": 6.387014543809223e-05,
      "loss": 4.8659,
      "step": 88
    },
    {
      "epoch": 0.005133899600536463,
      "grad_norm": 6.2620530128479,
      "learning_rate": 6.307399704769099e-05,
      "loss": 5.07,
      "step": 89
    },
    {
      "epoch": 0.00519158386571103,
      "grad_norm": 5.246522426605225,
      "learning_rate": 6.227427435703997e-05,
      "loss": 4.7459,
      "step": 90
    },
    {
      "epoch": 0.005249268130885598,
      "grad_norm": 5.883655548095703,
      "learning_rate": 6.147119600233758e-05,
      "loss": 4.9084,
      "step": 91
    },
    {
      "epoch": 0.005306952396060165,
      "grad_norm": 6.195652484893799,
      "learning_rate": 6.066498153718735e-05,
      "loss": 4.7581,
      "step": 92
    },
    {
      "epoch": 0.005364636661234732,
      "grad_norm": 5.9543585777282715,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.3457,
      "step": 93
    },
    {
      "epoch": 0.005422320926409299,
      "grad_norm": 6.17326545715332,
      "learning_rate": 5.90440267166055e-05,
      "loss": 4.8767,
      "step": 94
    },
    {
      "epoch": 0.0054800051915838656,
      "grad_norm": 7.127965927124023,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.8757,
      "step": 95
    },
    {
      "epoch": 0.0055376894567584324,
      "grad_norm": 6.237260341644287,
      "learning_rate": 5.74131823855921e-05,
      "loss": 4.9965,
      "step": 96
    },
    {
      "epoch": 0.005595373721932999,
      "grad_norm": 6.496796131134033,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 5.0044,
      "step": 97
    },
    {
      "epoch": 0.005653057987107567,
      "grad_norm": 5.858653545379639,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.9755,
      "step": 98
    },
    {
      "epoch": 0.005710742252282134,
      "grad_norm": 7.756019115447998,
      "learning_rate": 5.495227651252315e-05,
      "loss": 5.9028,
      "step": 99
    },
    {
      "epoch": 0.005768426517456701,
      "grad_norm": 8.971646308898926,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 4.4486,
      "step": 100
    },
    {
      "epoch": 0.005768426517456701,
      "eval_loss": 1.2152711153030396,
      "eval_runtime": 571.387,
      "eval_samples_per_second": 51.098,
      "eval_steps_per_second": 12.776,
      "step": 100
    },
    {
      "epoch": 0.005826110782631268,
      "grad_norm": 7.063196659088135,
      "learning_rate": 5.330452921628497e-05,
      "loss": 4.0226,
      "step": 101
    },
    {
      "epoch": 0.005883795047805835,
      "grad_norm": 7.931707859039307,
      "learning_rate": 5.247918773366112e-05,
      "loss": 5.2232,
      "step": 102
    },
    {
      "epoch": 0.0059414793129804015,
      "grad_norm": 6.4740681648254395,
      "learning_rate": 5.165316846586541e-05,
      "loss": 4.923,
      "step": 103
    },
    {
      "epoch": 0.005999163578154968,
      "grad_norm": 5.713397979736328,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 5.0176,
      "step": 104
    },
    {
      "epoch": 0.006056847843329536,
      "grad_norm": 4.992979049682617,
      "learning_rate": 5e-05,
      "loss": 4.1034,
      "step": 105
    },
    {
      "epoch": 0.006114532108504103,
      "grad_norm": 5.255120754241943,
      "learning_rate": 4.917330276168208e-05,
      "loss": 4.3639,
      "step": 106
    },
    {
      "epoch": 0.00617221637367867,
      "grad_norm": 4.529256820678711,
      "learning_rate": 4.834683153413459e-05,
      "loss": 4.2105,
      "step": 107
    },
    {
      "epoch": 0.006229900638853237,
      "grad_norm": 4.495347499847412,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 4.6283,
      "step": 108
    },
    {
      "epoch": 0.006287584904027804,
      "grad_norm": 4.299388885498047,
      "learning_rate": 4.669547078371504e-05,
      "loss": 4.5362,
      "step": 109
    },
    {
      "epoch": 0.006345269169202371,
      "grad_norm": 3.4968910217285156,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 3.9904,
      "step": 110
    },
    {
      "epoch": 0.0064029534343769375,
      "grad_norm": 4.361361980438232,
      "learning_rate": 4.504772348747687e-05,
      "loss": 4.7276,
      "step": 111
    },
    {
      "epoch": 0.006460637699551505,
      "grad_norm": 3.8528153896331787,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.3176,
      "step": 112
    },
    {
      "epoch": 0.006518321964726072,
      "grad_norm": 4.828119277954102,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 5.1857,
      "step": 113
    },
    {
      "epoch": 0.006576006229900639,
      "grad_norm": 4.477456569671631,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 5.0375,
      "step": 114
    },
    {
      "epoch": 0.006633690495075206,
      "grad_norm": 5.029172420501709,
      "learning_rate": 4.17702704859633e-05,
      "loss": 5.8694,
      "step": 115
    },
    {
      "epoch": 0.006691374760249773,
      "grad_norm": 4.393564224243164,
      "learning_rate": 4.095597328339452e-05,
      "loss": 4.4881,
      "step": 116
    },
    {
      "epoch": 0.00674905902542434,
      "grad_norm": 4.725098133087158,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 5.0789,
      "step": 117
    },
    {
      "epoch": 0.0068067432905989066,
      "grad_norm": 4.851803779602051,
      "learning_rate": 3.933501846281267e-05,
      "loss": 4.3162,
      "step": 118
    },
    {
      "epoch": 0.006864427555773474,
      "grad_norm": 4.502398490905762,
      "learning_rate": 3.852880399766243e-05,
      "loss": 5.0477,
      "step": 119
    },
    {
      "epoch": 0.006922111820948041,
      "grad_norm": 4.347857475280762,
      "learning_rate": 3.772572564296005e-05,
      "loss": 4.4053,
      "step": 120
    },
    {
      "epoch": 0.006979796086122608,
      "grad_norm": 4.391949653625488,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 4.4583,
      "step": 121
    },
    {
      "epoch": 0.007037480351297175,
      "grad_norm": 4.745469093322754,
      "learning_rate": 3.612985456190778e-05,
      "loss": 4.8135,
      "step": 122
    },
    {
      "epoch": 0.007095164616471742,
      "grad_norm": 4.189981460571289,
      "learning_rate": 3.533749813077677e-05,
      "loss": 4.6287,
      "step": 123
    },
    {
      "epoch": 0.007152848881646309,
      "grad_norm": 4.31459379196167,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 4.6345,
      "step": 124
    },
    {
      "epoch": 0.007210533146820876,
      "grad_norm": 4.6330671310424805,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 5.3061,
      "step": 125
    },
    {
      "epoch": 0.0072682174119954425,
      "grad_norm": 4.118234157562256,
      "learning_rate": 3.298534127791785e-05,
      "loss": 4.6342,
      "step": 126
    },
    {
      "epoch": 0.00732590167717001,
      "grad_norm": 4.425431251525879,
      "learning_rate": 3.221030765387417e-05,
      "loss": 5.0025,
      "step": 127
    },
    {
      "epoch": 0.007383585942344577,
      "grad_norm": 4.177802085876465,
      "learning_rate": 3.144013755408895e-05,
      "loss": 4.5713,
      "step": 128
    },
    {
      "epoch": 0.007441270207519144,
      "grad_norm": 4.990856647491455,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 5.5868,
      "step": 129
    },
    {
      "epoch": 0.007498954472693711,
      "grad_norm": 4.877931118011475,
      "learning_rate": 2.991522876735154e-05,
      "loss": 5.1022,
      "step": 130
    },
    {
      "epoch": 0.007556638737868278,
      "grad_norm": 5.54541540145874,
      "learning_rate": 2.916090697523549e-05,
      "loss": 4.7963,
      "step": 131
    },
    {
      "epoch": 0.007614323003042845,
      "grad_norm": 4.507201671600342,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 4.9663,
      "step": 132
    },
    {
      "epoch": 0.007672007268217412,
      "grad_norm": 4.721153736114502,
      "learning_rate": 2.766955965735968e-05,
      "loss": 4.9097,
      "step": 133
    },
    {
      "epoch": 0.007729691533391979,
      "grad_norm": 5.018895149230957,
      "learning_rate": 2.693294185106562e-05,
      "loss": 4.8714,
      "step": 134
    },
    {
      "epoch": 0.007787375798566546,
      "grad_norm": 4.928877830505371,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 4.7284,
      "step": 135
    },
    {
      "epoch": 0.007845060063741112,
      "grad_norm": 5.144993782043457,
      "learning_rate": 2.547882480847461e-05,
      "loss": 6.1324,
      "step": 136
    },
    {
      "epoch": 0.007902744328915681,
      "grad_norm": 4.932133674621582,
      "learning_rate": 2.476172311325783e-05,
      "loss": 4.8616,
      "step": 137
    },
    {
      "epoch": 0.007960428594090248,
      "grad_norm": 5.834035873413086,
      "learning_rate": 2.405152131093926e-05,
      "loss": 5.4315,
      "step": 138
    },
    {
      "epoch": 0.008018112859264815,
      "grad_norm": 5.277275085449219,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.4591,
      "step": 139
    },
    {
      "epoch": 0.008075797124439382,
      "grad_norm": 5.383817672729492,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 5.1053,
      "step": 140
    },
    {
      "epoch": 0.008133481389613948,
      "grad_norm": 5.679407119750977,
      "learning_rate": 2.196424713241637e-05,
      "loss": 5.9379,
      "step": 141
    },
    {
      "epoch": 0.008191165654788515,
      "grad_norm": 4.832319259643555,
      "learning_rate": 2.128356686585282e-05,
      "loss": 4.6105,
      "step": 142
    },
    {
      "epoch": 0.008248849919963082,
      "grad_norm": 5.268148899078369,
      "learning_rate": 2.061073738537635e-05,
      "loss": 5.3797,
      "step": 143
    },
    {
      "epoch": 0.008306534185137649,
      "grad_norm": 5.340954303741455,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.6882,
      "step": 144
    },
    {
      "epoch": 0.008364218450312216,
      "grad_norm": 6.224842071533203,
      "learning_rate": 1.928936436551661e-05,
      "loss": 5.2727,
      "step": 145
    },
    {
      "epoch": 0.008421902715486783,
      "grad_norm": 5.963640213012695,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 5.1429,
      "step": 146
    },
    {
      "epoch": 0.00847958698066135,
      "grad_norm": 5.485063076019287,
      "learning_rate": 1.800157297483417e-05,
      "loss": 5.098,
      "step": 147
    },
    {
      "epoch": 0.008537271245835917,
      "grad_norm": 6.236100196838379,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.4773,
      "step": 148
    },
    {
      "epoch": 0.008594955511010484,
      "grad_norm": 6.651202201843262,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 5.4952,
      "step": 149
    },
    {
      "epoch": 0.00865263977618505,
      "grad_norm": 9.745405197143555,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 5.0837,
      "step": 150
    },
    {
      "epoch": 0.00865263977618505,
      "eval_loss": 1.150228500366211,
      "eval_runtime": 570.521,
      "eval_samples_per_second": 51.176,
      "eval_steps_per_second": 12.795,
      "step": 150
    },
    {
      "epoch": 0.008710324041359617,
      "grad_norm": 3.696138620376587,
      "learning_rate": 1.553232954407171e-05,
      "loss": 4.5876,
      "step": 151
    },
    {
      "epoch": 0.008768008306534186,
      "grad_norm": 3.629110813140869,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 5.0187,
      "step": 152
    },
    {
      "epoch": 0.008825692571708753,
      "grad_norm": 3.8899803161621094,
      "learning_rate": 1.435357758543015e-05,
      "loss": 4.8667,
      "step": 153
    },
    {
      "epoch": 0.00888337683688332,
      "grad_norm": 3.6859331130981445,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 4.3835,
      "step": 154
    },
    {
      "epoch": 0.008941061102057887,
      "grad_norm": 3.9072530269622803,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 4.8611,
      "step": 155
    },
    {
      "epoch": 0.008998745367232453,
      "grad_norm": 4.271492958068848,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 4.2222,
      "step": 156
    },
    {
      "epoch": 0.00905642963240702,
      "grad_norm": 4.110146999359131,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 4.3338,
      "step": 157
    },
    {
      "epoch": 0.009114113897581587,
      "grad_norm": 3.994717597961426,
      "learning_rate": 1.157994445715706e-05,
      "loss": 4.1897,
      "step": 158
    },
    {
      "epoch": 0.009171798162756154,
      "grad_norm": 3.654979944229126,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.6583,
      "step": 159
    },
    {
      "epoch": 0.009229482427930721,
      "grad_norm": 4.580843925476074,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 4.0071,
      "step": 160
    },
    {
      "epoch": 0.009287166693105288,
      "grad_norm": 3.8958771228790283,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 4.5146,
      "step": 161
    },
    {
      "epoch": 0.009344850958279855,
      "grad_norm": 4.073850631713867,
      "learning_rate": 9.549150281252633e-06,
      "loss": 4.9066,
      "step": 162
    },
    {
      "epoch": 0.009402535223454422,
      "grad_norm": 4.060781955718994,
      "learning_rate": 9.068759265665384e-06,
      "loss": 5.0998,
      "step": 163
    },
    {
      "epoch": 0.009460219488628989,
      "grad_norm": 3.8098297119140625,
      "learning_rate": 8.599558442598998e-06,
      "loss": 4.023,
      "step": 164
    },
    {
      "epoch": 0.009517903753803555,
      "grad_norm": 4.438258647918701,
      "learning_rate": 8.141676086873572e-06,
      "loss": 5.2889,
      "step": 165
    },
    {
      "epoch": 0.009575588018978124,
      "grad_norm": 3.803323745727539,
      "learning_rate": 7.695237378953223e-06,
      "loss": 4.0325,
      "step": 166
    },
    {
      "epoch": 0.009633272284152691,
      "grad_norm": 4.349434852600098,
      "learning_rate": 7.260364370723044e-06,
      "loss": 5.0114,
      "step": 167
    },
    {
      "epoch": 0.009690956549327258,
      "grad_norm": 4.21728515625,
      "learning_rate": 6.837175952121306e-06,
      "loss": 4.7378,
      "step": 168
    },
    {
      "epoch": 0.009748640814501825,
      "grad_norm": 4.619811058044434,
      "learning_rate": 6.425787818636131e-06,
      "loss": 5.6027,
      "step": 169
    },
    {
      "epoch": 0.009806325079676392,
      "grad_norm": 3.8567116260528564,
      "learning_rate": 6.026312439675552e-06,
      "loss": 4.2873,
      "step": 170
    },
    {
      "epoch": 0.009864009344850959,
      "grad_norm": 4.612334728240967,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 5.1547,
      "step": 171
    },
    {
      "epoch": 0.009921693610025525,
      "grad_norm": 4.8865509033203125,
      "learning_rate": 5.263533508961827e-06,
      "loss": 5.1098,
      "step": 172
    },
    {
      "epoch": 0.009979377875200092,
      "grad_norm": 4.453343868255615,
      "learning_rate": 4.900438493352055e-06,
      "loss": 4.5194,
      "step": 173
    },
    {
      "epoch": 0.01003706214037466,
      "grad_norm": 4.374818801879883,
      "learning_rate": 4.549673247541875e-06,
      "loss": 4.5546,
      "step": 174
    },
    {
      "epoch": 0.010094746405549226,
      "grad_norm": 4.641942024230957,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 4.8886,
      "step": 175
    },
    {
      "epoch": 0.010152430670723793,
      "grad_norm": 4.402484893798828,
      "learning_rate": 3.885512251130763e-06,
      "loss": 4.4315,
      "step": 176
    },
    {
      "epoch": 0.01021011493589836,
      "grad_norm": 4.798633575439453,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 4.5846,
      "step": 177
    },
    {
      "epoch": 0.010267799201072927,
      "grad_norm": 5.0984296798706055,
      "learning_rate": 3.271776770026963e-06,
      "loss": 5.719,
      "step": 178
    },
    {
      "epoch": 0.010325483466247494,
      "grad_norm": 4.599299430847168,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 4.2936,
      "step": 179
    },
    {
      "epoch": 0.01038316773142206,
      "grad_norm": 5.3232221603393555,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 5.5233,
      "step": 180
    },
    {
      "epoch": 0.010440851996596629,
      "grad_norm": 4.739865303039551,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 4.1426,
      "step": 181
    },
    {
      "epoch": 0.010498536261771196,
      "grad_norm": 5.614005088806152,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 5.5367,
      "step": 182
    },
    {
      "epoch": 0.010556220526945763,
      "grad_norm": 4.866192817687988,
      "learning_rate": 1.962316193157593e-06,
      "loss": 4.2121,
      "step": 183
    },
    {
      "epoch": 0.01061390479212033,
      "grad_norm": 4.665633678436279,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 4.5529,
      "step": 184
    },
    {
      "epoch": 0.010671589057294897,
      "grad_norm": 4.709635257720947,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 4.7962,
      "step": 185
    },
    {
      "epoch": 0.010729273322469464,
      "grad_norm": 5.334566593170166,
      "learning_rate": 1.333670137599713e-06,
      "loss": 4.6977,
      "step": 186
    },
    {
      "epoch": 0.01078695758764403,
      "grad_norm": 4.877022743225098,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 4.2813,
      "step": 187
    },
    {
      "epoch": 0.010844641852818597,
      "grad_norm": 7.548967361450195,
      "learning_rate": 9.810017062595322e-07,
      "loss": 4.8446,
      "step": 188
    },
    {
      "epoch": 0.010902326117993164,
      "grad_norm": 5.502225875854492,
      "learning_rate": 8.247462563808817e-07,
      "loss": 5.3996,
      "step": 189
    },
    {
      "epoch": 0.010960010383167731,
      "grad_norm": 5.037219524383545,
      "learning_rate": 6.819348298638839e-07,
      "loss": 4.0897,
      "step": 190
    },
    {
      "epoch": 0.011017694648342298,
      "grad_norm": 5.668946743011475,
      "learning_rate": 5.526064699265753e-07,
      "loss": 4.3297,
      "step": 191
    },
    {
      "epoch": 0.011075378913516865,
      "grad_norm": 6.326772212982178,
      "learning_rate": 4.367965336512403e-07,
      "loss": 4.1069,
      "step": 192
    },
    {
      "epoch": 0.011133063178691432,
      "grad_norm": 6.3514814376831055,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 4.8842,
      "step": 193
    },
    {
      "epoch": 0.011190747443865999,
      "grad_norm": 6.059720039367676,
      "learning_rate": 2.458548727494292e-07,
      "loss": 4.7773,
      "step": 194
    },
    {
      "epoch": 0.011248431709040567,
      "grad_norm": 6.349850654602051,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 4.6453,
      "step": 195
    },
    {
      "epoch": 0.011306115974215134,
      "grad_norm": 7.454678058624268,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 5.9683,
      "step": 196
    },
    {
      "epoch": 0.011363800239389701,
      "grad_norm": 6.93421745300293,
      "learning_rate": 6.150154258476315e-08,
      "loss": 4.8673,
      "step": 197
    },
    {
      "epoch": 0.011421484504564268,
      "grad_norm": 7.563024044036865,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 5.5949,
      "step": 198
    },
    {
      "epoch": 0.011479168769738835,
      "grad_norm": 7.1365251541137695,
      "learning_rate": 6.834750376549792e-09,
      "loss": 4.3368,
      "step": 199
    },
    {
      "epoch": 0.011536853034913402,
      "grad_norm": 9.6073637008667,
      "learning_rate": 0.0,
      "loss": 5.0569,
      "step": 200
    },
    {
      "epoch": 0.011536853034913402,
      "eval_loss": 1.143795371055603,
      "eval_runtime": 570.3476,
      "eval_samples_per_second": 51.192,
      "eval_steps_per_second": 12.799,
      "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": 4.97506999861248e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}