{
  "best_metric": 2.323329210281372,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.025809781907342882,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001290489095367144,
      "grad_norm": 1.8614425659179688,
      "learning_rate": 1e-05,
      "loss": 8.9881,
      "step": 1
    },
    {
      "epoch": 0.0001290489095367144,
      "eval_loss": 2.636259078979492,
      "eval_runtime": 1413.6971,
      "eval_samples_per_second": 9.232,
      "eval_steps_per_second": 2.308,
      "step": 1
    },
    {
      "epoch": 0.0002580978190734288,
      "grad_norm": 2.7925186157226562,
      "learning_rate": 2e-05,
      "loss": 9.2454,
      "step": 2
    },
    {
      "epoch": 0.00038714672861014324,
      "grad_norm": 2.5410654544830322,
      "learning_rate": 3e-05,
      "loss": 9.0846,
      "step": 3
    },
    {
      "epoch": 0.0005161956381468576,
      "grad_norm": 2.7802281379699707,
      "learning_rate": 4e-05,
      "loss": 9.4966,
      "step": 4
    },
    {
      "epoch": 0.0006452445476835721,
      "grad_norm": 3.771491527557373,
      "learning_rate": 5e-05,
      "loss": 9.7491,
      "step": 5
    },
    {
      "epoch": 0.0007742934572202865,
      "grad_norm": 2.451549768447876,
      "learning_rate": 6e-05,
      "loss": 8.8294,
      "step": 6
    },
    {
      "epoch": 0.0009033423667570009,
      "grad_norm": 2.5755577087402344,
      "learning_rate": 7e-05,
      "loss": 9.3547,
      "step": 7
    },
    {
      "epoch": 0.0010323912762937152,
      "grad_norm": 2.888935089111328,
      "learning_rate": 8e-05,
      "loss": 9.6213,
      "step": 8
    },
    {
      "epoch": 0.0011614401858304297,
      "grad_norm": 2.7784957885742188,
      "learning_rate": 9e-05,
      "loss": 9.0203,
      "step": 9
    },
    {
      "epoch": 0.0012904890953671442,
      "grad_norm": 2.6944925785064697,
      "learning_rate": 0.0001,
      "loss": 9.1302,
      "step": 10
    },
    {
      "epoch": 0.0014195380049038585,
      "grad_norm": 5.104203701019287,
      "learning_rate": 9.999316524962345e-05,
      "loss": 9.2007,
      "step": 11
    },
    {
      "epoch": 0.001548586914440573,
      "grad_norm": 3.175618886947632,
      "learning_rate": 9.997266286704631e-05,
      "loss": 9.1289,
      "step": 12
    },
    {
      "epoch": 0.0016776358239772875,
      "grad_norm": 2.933878183364868,
      "learning_rate": 9.993849845741524e-05,
      "loss": 9.3867,
      "step": 13
    },
    {
      "epoch": 0.0018066847335140017,
      "grad_norm": 2.917396068572998,
      "learning_rate": 9.989068136093873e-05,
      "loss": 9.3555,
      "step": 14
    },
    {
      "epoch": 0.0019357336430507162,
      "grad_norm": 2.844730854034424,
      "learning_rate": 9.98292246503335e-05,
      "loss": 8.9247,
      "step": 15
    },
    {
      "epoch": 0.0020647825525874305,
      "grad_norm": 2.7617626190185547,
      "learning_rate": 9.975414512725057e-05,
      "loss": 9.3332,
      "step": 16
    },
    {
      "epoch": 0.002193831462124145,
      "grad_norm": 2.6766278743743896,
      "learning_rate": 9.966546331768191e-05,
      "loss": 9.3239,
      "step": 17
    },
    {
      "epoch": 0.0023228803716608595,
      "grad_norm": 2.677168130874634,
      "learning_rate": 9.956320346634876e-05,
      "loss": 9.1266,
      "step": 18
    },
    {
      "epoch": 0.0024519292811975737,
      "grad_norm": 2.8756260871887207,
      "learning_rate": 9.944739353007344e-05,
      "loss": 9.2339,
      "step": 19
    },
    {
      "epoch": 0.0025809781907342884,
      "grad_norm": 3.1478431224823,
      "learning_rate": 9.931806517013612e-05,
      "loss": 9.0789,
      "step": 20
    },
    {
      "epoch": 0.0027100271002710027,
      "grad_norm": 2.9869580268859863,
      "learning_rate": 9.917525374361912e-05,
      "loss": 9.2093,
      "step": 21
    },
    {
      "epoch": 0.002839076009807717,
      "grad_norm": 3.0213706493377686,
      "learning_rate": 9.901899829374047e-05,
      "loss": 8.9714,
      "step": 22
    },
    {
      "epoch": 0.0029681249193444317,
      "grad_norm": 3.244584083557129,
      "learning_rate": 9.884934153917997e-05,
      "loss": 9.3416,
      "step": 23
    },
    {
      "epoch": 0.003097173828881146,
      "grad_norm": 3.0729188919067383,
      "learning_rate": 9.86663298624003e-05,
      "loss": 9.3149,
      "step": 24
    },
    {
      "epoch": 0.0032262227384178602,
      "grad_norm": 3.2723147869110107,
      "learning_rate": 9.847001329696653e-05,
      "loss": 9.0973,
      "step": 25
    },
    {
      "epoch": 0.003355271647954575,
      "grad_norm": 3.824782609939575,
      "learning_rate": 9.826044551386744e-05,
      "loss": 9.3632,
      "step": 26
    },
    {
      "epoch": 0.003484320557491289,
      "grad_norm": 7.753170490264893,
      "learning_rate": 9.803768380684242e-05,
      "loss": 9.2341,
      "step": 27
    },
    {
      "epoch": 0.0036133694670280035,
      "grad_norm": 3.6804659366607666,
      "learning_rate": 9.780178907671789e-05,
      "loss": 9.4726,
      "step": 28
    },
    {
      "epoch": 0.003742418376564718,
      "grad_norm": 3.558088779449463,
      "learning_rate": 9.755282581475769e-05,
      "loss": 9.2617,
      "step": 29
    },
    {
      "epoch": 0.0038714672861014324,
      "grad_norm": 3.9723758697509766,
      "learning_rate": 9.729086208503174e-05,
      "loss": 9.6789,
      "step": 30
    },
    {
      "epoch": 0.004000516195638147,
      "grad_norm": 3.5203986167907715,
      "learning_rate": 9.701596950580806e-05,
      "loss": 9.2883,
      "step": 31
    },
    {
      "epoch": 0.004129565105174861,
      "grad_norm": 3.5905189514160156,
      "learning_rate": 9.672822322997305e-05,
      "loss": 9.5355,
      "step": 32
    },
    {
      "epoch": 0.004258614014711576,
      "grad_norm": 3.8688364028930664,
      "learning_rate": 9.642770192448536e-05,
      "loss": 9.2926,
      "step": 33
    },
    {
      "epoch": 0.00438766292424829,
      "grad_norm": 3.8279356956481934,
      "learning_rate": 9.611448774886924e-05,
      "loss": 9.164,
      "step": 34
    },
    {
      "epoch": 0.004516711833785004,
      "grad_norm": 4.065227508544922,
      "learning_rate": 9.578866633275288e-05,
      "loss": 9.7515,
      "step": 35
    },
    {
      "epoch": 0.004645760743321719,
      "grad_norm": 3.701939582824707,
      "learning_rate": 9.545032675245813e-05,
      "loss": 8.7243,
      "step": 36
    },
    {
      "epoch": 0.004774809652858434,
      "grad_norm": 4.188230991363525,
      "learning_rate": 9.509956150664796e-05,
      "loss": 9.0735,
      "step": 37
    },
    {
      "epoch": 0.0049038585623951475,
      "grad_norm": 3.963853359222412,
      "learning_rate": 9.473646649103818e-05,
      "loss": 9.0152,
      "step": 38
    },
    {
      "epoch": 0.005032907471931862,
      "grad_norm": 3.776251792907715,
      "learning_rate": 9.43611409721806e-05,
      "loss": 9.2071,
      "step": 39
    },
    {
      "epoch": 0.005161956381468577,
      "grad_norm": 4.173133850097656,
      "learning_rate": 9.397368756032445e-05,
      "loss": 9.567,
      "step": 40
    },
    {
      "epoch": 0.005291005291005291,
      "grad_norm": 4.319381237030029,
      "learning_rate": 9.357421218136386e-05,
      "loss": 9.3369,
      "step": 41
    },
    {
      "epoch": 0.005420054200542005,
      "grad_norm": 4.087239742279053,
      "learning_rate": 9.316282404787871e-05,
      "loss": 9.2344,
      "step": 42
    },
    {
      "epoch": 0.00554910311007872,
      "grad_norm": 4.893524169921875,
      "learning_rate": 9.273963562927695e-05,
      "loss": 9.6014,
      "step": 43
    },
    {
      "epoch": 0.005678152019615434,
      "grad_norm": 4.79598331451416,
      "learning_rate": 9.230476262104677e-05,
      "loss": 9.2621,
      "step": 44
    },
    {
      "epoch": 0.005807200929152149,
      "grad_norm": 4.756844997406006,
      "learning_rate": 9.185832391312644e-05,
      "loss": 9.0967,
      "step": 45
    },
    {
      "epoch": 0.005936249838688863,
      "grad_norm": 5.568742752075195,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.0427,
      "step": 46
    },
    {
      "epoch": 0.006065298748225577,
      "grad_norm": 6.306331634521484,
      "learning_rate": 9.093124073433463e-05,
      "loss": 9.6031,
      "step": 47
    },
    {
      "epoch": 0.006194347657762292,
      "grad_norm": 6.237867832183838,
      "learning_rate": 9.045084971874738e-05,
      "loss": 9.7357,
      "step": 48
    },
    {
      "epoch": 0.006323396567299007,
      "grad_norm": 6.019530296325684,
      "learning_rate": 8.995939984474624e-05,
      "loss": 9.5239,
      "step": 49
    },
    {
      "epoch": 0.0064524454768357204,
      "grad_norm": 7.326230049133301,
      "learning_rate": 8.945702546981969e-05,
      "loss": 9.6359,
      "step": 50
    },
    {
      "epoch": 0.0064524454768357204,
      "eval_loss": 2.4660184383392334,
      "eval_runtime": 1422.8992,
      "eval_samples_per_second": 9.172,
      "eval_steps_per_second": 2.293,
      "step": 50
    },
    {
      "epoch": 0.006581494386372435,
      "grad_norm": 5.422894477844238,
      "learning_rate": 8.894386393810563e-05,
      "loss": 8.7521,
      "step": 51
    },
    {
      "epoch": 0.00671054329590915,
      "grad_norm": 4.593626499176025,
      "learning_rate": 8.842005554284296e-05,
      "loss": 9.5515,
      "step": 52
    },
    {
      "epoch": 0.006839592205445864,
      "grad_norm": 3.54670786857605,
      "learning_rate": 8.788574348801675e-05,
      "loss": 9.5144,
      "step": 53
    },
    {
      "epoch": 0.006968641114982578,
      "grad_norm": 2.321512222290039,
      "learning_rate": 8.73410738492077e-05,
      "loss": 9.202,
      "step": 54
    },
    {
      "epoch": 0.007097690024519293,
      "grad_norm": 1.8870216608047485,
      "learning_rate": 8.678619553365659e-05,
      "loss": 8.8939,
      "step": 55
    },
    {
      "epoch": 0.007226738934056007,
      "grad_norm": 1.952763557434082,
      "learning_rate": 8.622126023955446e-05,
      "loss": 8.9011,
      "step": 56
    },
    {
      "epoch": 0.007355787843592722,
      "grad_norm": 2.262312889099121,
      "learning_rate": 8.564642241456986e-05,
      "loss": 9.0816,
      "step": 57
    },
    {
      "epoch": 0.007484836753129436,
      "grad_norm": 2.0647146701812744,
      "learning_rate": 8.506183921362443e-05,
      "loss": 9.0659,
      "step": 58
    },
    {
      "epoch": 0.00761388566266615,
      "grad_norm": 2.413372755050659,
      "learning_rate": 8.44676704559283e-05,
      "loss": 8.8529,
      "step": 59
    },
    {
      "epoch": 0.007742934572202865,
      "grad_norm": 2.297088384628296,
      "learning_rate": 8.386407858128706e-05,
      "loss": 9.3466,
      "step": 60
    },
    {
      "epoch": 0.00787198348173958,
      "grad_norm": 2.3841991424560547,
      "learning_rate": 8.32512286056924e-05,
      "loss": 9.1378,
      "step": 61
    },
    {
      "epoch": 0.008001032391276294,
      "grad_norm": 2.2245638370513916,
      "learning_rate": 8.262928807620843e-05,
      "loss": 9.3911,
      "step": 62
    },
    {
      "epoch": 0.008130081300813009,
      "grad_norm": 2.489924192428589,
      "learning_rate": 8.199842702516583e-05,
      "loss": 9.1214,
      "step": 63
    },
    {
      "epoch": 0.008259130210349722,
      "grad_norm": 2.9884145259857178,
      "learning_rate": 8.135881792367686e-05,
      "loss": 9.0205,
      "step": 64
    },
    {
      "epoch": 0.008388179119886437,
      "grad_norm": 3.655379056930542,
      "learning_rate": 8.07106356344834e-05,
      "loss": 9.3441,
      "step": 65
    },
    {
      "epoch": 0.008517228029423151,
      "grad_norm": 2.8182358741760254,
      "learning_rate": 8.005405736415126e-05,
      "loss": 9.2183,
      "step": 66
    },
    {
      "epoch": 0.008646276938959866,
      "grad_norm": 2.783860445022583,
      "learning_rate": 7.938926261462366e-05,
      "loss": 9.2931,
      "step": 67
    },
    {
      "epoch": 0.00877532584849658,
      "grad_norm": 2.467184543609619,
      "learning_rate": 7.871643313414718e-05,
      "loss": 9.272,
      "step": 68
    },
    {
      "epoch": 0.008904374758033295,
      "grad_norm": 2.793034553527832,
      "learning_rate": 7.803575286758364e-05,
      "loss": 9.7043,
      "step": 69
    },
    {
      "epoch": 0.009033423667570008,
      "grad_norm": 3.135387897491455,
      "learning_rate": 7.734740790612136e-05,
      "loss": 8.8145,
      "step": 70
    },
    {
      "epoch": 0.009162472577106723,
      "grad_norm": 2.727604627609253,
      "learning_rate": 7.66515864363997e-05,
      "loss": 9.2702,
      "step": 71
    },
    {
      "epoch": 0.009291521486643438,
      "grad_norm": 3.130295515060425,
      "learning_rate": 7.594847868906076e-05,
      "loss": 9.1941,
      "step": 72
    },
    {
      "epoch": 0.009420570396180153,
      "grad_norm": 2.866556406021118,
      "learning_rate": 7.52382768867422e-05,
      "loss": 8.871,
      "step": 73
    },
    {
      "epoch": 0.009549619305716867,
      "grad_norm": 3.0217981338500977,
      "learning_rate": 7.452117519152542e-05,
      "loss": 9.2911,
      "step": 74
    },
    {
      "epoch": 0.009678668215253582,
      "grad_norm": 3.505121946334839,
      "learning_rate": 7.379736965185368e-05,
      "loss": 9.1199,
      "step": 75
    },
    {
      "epoch": 0.009807717124790295,
      "grad_norm": 3.0213496685028076,
      "learning_rate": 7.30670581489344e-05,
      "loss": 9.8074,
      "step": 76
    },
    {
      "epoch": 0.00993676603432701,
      "grad_norm": 3.1628036499023438,
      "learning_rate": 7.233044034264034e-05,
      "loss": 9.1827,
      "step": 77
    },
    {
      "epoch": 0.010065814943863724,
      "grad_norm": 3.0443100929260254,
      "learning_rate": 7.158771761692464e-05,
      "loss": 8.8275,
      "step": 78
    },
    {
      "epoch": 0.010194863853400439,
      "grad_norm": 3.220184564590454,
      "learning_rate": 7.083909302476453e-05,
      "loss": 9.6909,
      "step": 79
    },
    {
      "epoch": 0.010323912762937154,
      "grad_norm": 3.0642037391662598,
      "learning_rate": 7.008477123264848e-05,
      "loss": 9.4032,
      "step": 80
    },
    {
      "epoch": 0.010452961672473868,
      "grad_norm": 3.376972198486328,
      "learning_rate": 6.932495846462261e-05,
      "loss": 9.7422,
      "step": 81
    },
    {
      "epoch": 0.010582010582010581,
      "grad_norm": 3.520718812942505,
      "learning_rate": 6.855986244591104e-05,
      "loss": 8.8908,
      "step": 82
    },
    {
      "epoch": 0.010711059491547296,
      "grad_norm": 3.679842472076416,
      "learning_rate": 6.778969234612584e-05,
      "loss": 9.2061,
      "step": 83
    },
    {
      "epoch": 0.01084010840108401,
      "grad_norm": 3.495368003845215,
      "learning_rate": 6.701465872208216e-05,
      "loss": 9.42,
      "step": 84
    },
    {
      "epoch": 0.010969157310620726,
      "grad_norm": 3.4645586013793945,
      "learning_rate": 6.623497346023418e-05,
      "loss": 9.8746,
      "step": 85
    },
    {
      "epoch": 0.01109820622015744,
      "grad_norm": 3.3575072288513184,
      "learning_rate": 6.545084971874738e-05,
      "loss": 9.2253,
      "step": 86
    },
    {
      "epoch": 0.011227255129694155,
      "grad_norm": 3.5768520832061768,
      "learning_rate": 6.466250186922325e-05,
      "loss": 9.1085,
      "step": 87
    },
    {
      "epoch": 0.011356304039230868,
      "grad_norm": 3.6088900566101074,
      "learning_rate": 6.387014543809223e-05,
      "loss": 8.9726,
      "step": 88
    },
    {
      "epoch": 0.011485352948767583,
      "grad_norm": 3.789121627807617,
      "learning_rate": 6.307399704769099e-05,
      "loss": 9.6415,
      "step": 89
    },
    {
      "epoch": 0.011614401858304297,
      "grad_norm": 3.8316235542297363,
      "learning_rate": 6.227427435703997e-05,
      "loss": 9.2873,
      "step": 90
    },
    {
      "epoch": 0.011743450767841012,
      "grad_norm": 3.986023187637329,
      "learning_rate": 6.147119600233758e-05,
      "loss": 9.2406,
      "step": 91
    },
    {
      "epoch": 0.011872499677377727,
      "grad_norm": 4.39637565612793,
      "learning_rate": 6.066498153718735e-05,
      "loss": 9.2628,
      "step": 92
    },
    {
      "epoch": 0.012001548586914441,
      "grad_norm": 4.188934326171875,
      "learning_rate": 5.985585137257401e-05,
      "loss": 9.2704,
      "step": 93
    },
    {
      "epoch": 0.012130597496451154,
      "grad_norm": 4.577038764953613,
      "learning_rate": 5.90440267166055e-05,
      "loss": 9.3472,
      "step": 94
    },
    {
      "epoch": 0.012259646405987869,
      "grad_norm": 4.368072032928467,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 8.7526,
      "step": 95
    },
    {
      "epoch": 0.012388695315524584,
      "grad_norm": 4.861010551452637,
      "learning_rate": 5.74131823855921e-05,
      "loss": 9.3367,
      "step": 96
    },
    {
      "epoch": 0.012517744225061299,
      "grad_norm": 5.3851447105407715,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 9.6473,
      "step": 97
    },
    {
      "epoch": 0.012646793134598013,
      "grad_norm": 4.9451165199279785,
      "learning_rate": 5.577423184847932e-05,
      "loss": 9.5851,
      "step": 98
    },
    {
      "epoch": 0.012775842044134728,
      "grad_norm": 5.429501533508301,
      "learning_rate": 5.495227651252315e-05,
      "loss": 9.8293,
      "step": 99
    },
    {
      "epoch": 0.012904890953671441,
      "grad_norm": 6.867251873016357,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 10.4638,
      "step": 100
    },
    {
      "epoch": 0.012904890953671441,
      "eval_loss": 2.42568302154541,
      "eval_runtime": 1421.3573,
      "eval_samples_per_second": 9.182,
      "eval_steps_per_second": 2.296,
      "step": 100
    },
    {
      "epoch": 0.013033939863208156,
      "grad_norm": 6.558551788330078,
      "learning_rate": 5.330452921628497e-05,
      "loss": 9.0319,
      "step": 101
    },
    {
      "epoch": 0.01316298877274487,
      "grad_norm": 6.864365577697754,
      "learning_rate": 5.247918773366112e-05,
      "loss": 9.3867,
      "step": 102
    },
    {
      "epoch": 0.013292037682281585,
      "grad_norm": 6.573170185089111,
      "learning_rate": 5.165316846586541e-05,
      "loss": 9.3356,
      "step": 103
    },
    {
      "epoch": 0.0134210865918183,
      "grad_norm": 4.381434917449951,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 8.9209,
      "step": 104
    },
    {
      "epoch": 0.013550135501355014,
      "grad_norm": 2.770876407623291,
      "learning_rate": 5e-05,
      "loss": 9.1898,
      "step": 105
    },
    {
      "epoch": 0.013679184410891727,
      "grad_norm": 2.151318311691284,
      "learning_rate": 4.917330276168208e-05,
      "loss": 9.1106,
      "step": 106
    },
    {
      "epoch": 0.013808233320428442,
      "grad_norm": 1.8227428197860718,
      "learning_rate": 4.834683153413459e-05,
      "loss": 8.694,
      "step": 107
    },
    {
      "epoch": 0.013937282229965157,
      "grad_norm": 1.8833378553390503,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 9.3368,
      "step": 108
    },
    {
      "epoch": 0.014066331139501871,
      "grad_norm": 2.1978793144226074,
      "learning_rate": 4.669547078371504e-05,
      "loss": 9.0372,
      "step": 109
    },
    {
      "epoch": 0.014195380049038586,
      "grad_norm": 1.9617077112197876,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 9.1136,
      "step": 110
    },
    {
      "epoch": 0.014324428958575301,
      "grad_norm": 2.611269235610962,
      "learning_rate": 4.504772348747687e-05,
      "loss": 9.5521,
      "step": 111
    },
    {
      "epoch": 0.014453477868112014,
      "grad_norm": 2.0450103282928467,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 8.5825,
      "step": 112
    },
    {
      "epoch": 0.014582526777648729,
      "grad_norm": 2.2265167236328125,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 8.8527,
      "step": 113
    },
    {
      "epoch": 0.014711575687185443,
      "grad_norm": 2.497074842453003,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 9.3511,
      "step": 114
    },
    {
      "epoch": 0.014840624596722158,
      "grad_norm": 2.6098930835723877,
      "learning_rate": 4.17702704859633e-05,
      "loss": 9.1412,
      "step": 115
    },
    {
      "epoch": 0.014969673506258873,
      "grad_norm": 2.4353199005126953,
      "learning_rate": 4.095597328339452e-05,
      "loss": 8.9863,
      "step": 116
    },
    {
      "epoch": 0.015098722415795587,
      "grad_norm": 2.5682668685913086,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 9.1443,
      "step": 117
    },
    {
      "epoch": 0.0152277713253323,
      "grad_norm": 2.6961557865142822,
      "learning_rate": 3.933501846281267e-05,
      "loss": 9.6746,
      "step": 118
    },
    {
      "epoch": 0.015356820234869015,
      "grad_norm": 2.621609687805176,
      "learning_rate": 3.852880399766243e-05,
      "loss": 9.1541,
      "step": 119
    },
    {
      "epoch": 0.01548586914440573,
      "grad_norm": 2.6731786727905273,
      "learning_rate": 3.772572564296005e-05,
      "loss": 9.0717,
      "step": 120
    },
    {
      "epoch": 0.015614918053942444,
      "grad_norm": 2.469315767288208,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 8.7475,
      "step": 121
    },
    {
      "epoch": 0.01574396696347916,
      "grad_norm": 2.711573362350464,
      "learning_rate": 3.612985456190778e-05,
      "loss": 9.1526,
      "step": 122
    },
    {
      "epoch": 0.015873015873015872,
      "grad_norm": 2.6181857585906982,
      "learning_rate": 3.533749813077677e-05,
      "loss": 9.2163,
      "step": 123
    },
    {
      "epoch": 0.01600206478255259,
      "grad_norm": 2.6454579830169678,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 8.5263,
      "step": 124
    },
    {
      "epoch": 0.0161311136920893,
      "grad_norm": 2.8446078300476074,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 9.2747,
      "step": 125
    },
    {
      "epoch": 0.016260162601626018,
      "grad_norm": 2.921170949935913,
      "learning_rate": 3.298534127791785e-05,
      "loss": 9.8549,
      "step": 126
    },
    {
      "epoch": 0.01638921151116273,
      "grad_norm": 3.0550904273986816,
      "learning_rate": 3.221030765387417e-05,
      "loss": 8.9324,
      "step": 127
    },
    {
      "epoch": 0.016518260420699444,
      "grad_norm": 2.9599196910858154,
      "learning_rate": 3.144013755408895e-05,
      "loss": 9.792,
      "step": 128
    },
    {
      "epoch": 0.01664730933023616,
      "grad_norm": 2.7958767414093018,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 8.771,
      "step": 129
    },
    {
      "epoch": 0.016776358239772873,
      "grad_norm": 3.115372896194458,
      "learning_rate": 2.991522876735154e-05,
      "loss": 9.5177,
      "step": 130
    },
    {
      "epoch": 0.01690540714930959,
      "grad_norm": 2.960212230682373,
      "learning_rate": 2.916090697523549e-05,
      "loss": 8.8467,
      "step": 131
    },
    {
      "epoch": 0.017034456058846303,
      "grad_norm": 3.0903122425079346,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 9.8255,
      "step": 132
    },
    {
      "epoch": 0.017163504968383016,
      "grad_norm": 3.277925729751587,
      "learning_rate": 2.766955965735968e-05,
      "loss": 9.437,
      "step": 133
    },
    {
      "epoch": 0.017292553877919732,
      "grad_norm": 3.2966110706329346,
      "learning_rate": 2.693294185106562e-05,
      "loss": 9.1051,
      "step": 134
    },
    {
      "epoch": 0.017421602787456445,
      "grad_norm": 3.4716148376464844,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 9.1138,
      "step": 135
    },
    {
      "epoch": 0.01755065169699316,
      "grad_norm": 3.7124650478363037,
      "learning_rate": 2.547882480847461e-05,
      "loss": 9.827,
      "step": 136
    },
    {
      "epoch": 0.017679700606529875,
      "grad_norm": 3.57387375831604,
      "learning_rate": 2.476172311325783e-05,
      "loss": 9.4026,
      "step": 137
    },
    {
      "epoch": 0.01780874951606659,
      "grad_norm": 3.7645928859710693,
      "learning_rate": 2.405152131093926e-05,
      "loss": 8.8658,
      "step": 138
    },
    {
      "epoch": 0.017937798425603304,
      "grad_norm": 3.6762328147888184,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 9.0362,
      "step": 139
    },
    {
      "epoch": 0.018066847335140017,
      "grad_norm": 3.9711761474609375,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 9.7462,
      "step": 140
    },
    {
      "epoch": 0.018195896244676733,
      "grad_norm": 4.019091606140137,
      "learning_rate": 2.196424713241637e-05,
      "loss": 9.2169,
      "step": 141
    },
    {
      "epoch": 0.018324945154213446,
      "grad_norm": 4.222970485687256,
      "learning_rate": 2.128356686585282e-05,
      "loss": 9.4692,
      "step": 142
    },
    {
      "epoch": 0.018453994063750163,
      "grad_norm": 4.119962215423584,
      "learning_rate": 2.061073738537635e-05,
      "loss": 9.2049,
      "step": 143
    },
    {
      "epoch": 0.018583042973286876,
      "grad_norm": 4.283300876617432,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 9.5706,
      "step": 144
    },
    {
      "epoch": 0.01871209188282359,
      "grad_norm": 5.065773963928223,
      "learning_rate": 1.928936436551661e-05,
      "loss": 9.9508,
      "step": 145
    },
    {
      "epoch": 0.018841140792360305,
      "grad_norm": 4.4507670402526855,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 8.9088,
      "step": 146
    },
    {
      "epoch": 0.018970189701897018,
      "grad_norm": 4.944255352020264,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.1089,
      "step": 147
    },
    {
      "epoch": 0.019099238611433735,
      "grad_norm": 4.992981433868408,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 9.816,
      "step": 148
    },
    {
      "epoch": 0.019228287520970447,
      "grad_norm": 6.395650863647461,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 9.399,
      "step": 149
    },
    {
      "epoch": 0.019357336430507164,
      "grad_norm": 7.018031597137451,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 9.5585,
      "step": 150
    },
    {
      "epoch": 0.019357336430507164,
      "eval_loss": 2.3465826511383057,
      "eval_runtime": 1422.2973,
      "eval_samples_per_second": 9.176,
      "eval_steps_per_second": 2.294,
      "step": 150
    },
    {
      "epoch": 0.019486385340043877,
      "grad_norm": 3.8211867809295654,
      "learning_rate": 1.553232954407171e-05,
      "loss": 8.6706,
      "step": 151
    },
    {
      "epoch": 0.01961543424958059,
      "grad_norm": 3.9743099212646484,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 8.5796,
      "step": 152
    },
    {
      "epoch": 0.019744483159117306,
      "grad_norm": 4.149708271026611,
      "learning_rate": 1.435357758543015e-05,
      "loss": 9.1483,
      "step": 153
    },
    {
      "epoch": 0.01987353206865402,
      "grad_norm": 4.119248867034912,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 8.9964,
      "step": 154
    },
    {
      "epoch": 0.020002580978190736,
      "grad_norm": 4.085949420928955,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 9.0426,
      "step": 155
    },
    {
      "epoch": 0.02013162988772745,
      "grad_norm": 3.6565117835998535,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 8.9493,
      "step": 156
    },
    {
      "epoch": 0.02026067879726416,
      "grad_norm": 3.187164068222046,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 8.838,
      "step": 157
    },
    {
      "epoch": 0.020389727706800878,
      "grad_norm": 3.209070920944214,
      "learning_rate": 1.157994445715706e-05,
      "loss": 9.4441,
      "step": 158
    },
    {
      "epoch": 0.02051877661633759,
      "grad_norm": 2.9217028617858887,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 9.1366,
      "step": 159
    },
    {
      "epoch": 0.020647825525874308,
      "grad_norm": 2.853665590286255,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 9.3016,
      "step": 160
    },
    {
      "epoch": 0.02077687443541102,
      "grad_norm": 3.1376054286956787,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 9.1745,
      "step": 161
    },
    {
      "epoch": 0.020905923344947737,
      "grad_norm": 2.9321136474609375,
      "learning_rate": 9.549150281252633e-06,
      "loss": 8.7416,
      "step": 162
    },
    {
      "epoch": 0.02103497225448445,
      "grad_norm": 2.2884633541107178,
      "learning_rate": 9.068759265665384e-06,
      "loss": 9.0116,
      "step": 163
    },
    {
      "epoch": 0.021164021164021163,
      "grad_norm": 2.5829648971557617,
      "learning_rate": 8.599558442598998e-06,
      "loss": 9.2843,
      "step": 164
    },
    {
      "epoch": 0.02129307007355788,
      "grad_norm": 2.2356209754943848,
      "learning_rate": 8.141676086873572e-06,
      "loss": 9.2051,
      "step": 165
    },
    {
      "epoch": 0.021422118983094592,
      "grad_norm": 2.545142412185669,
      "learning_rate": 7.695237378953223e-06,
      "loss": 8.893,
      "step": 166
    },
    {
      "epoch": 0.02155116789263131,
      "grad_norm": 2.289316415786743,
      "learning_rate": 7.260364370723044e-06,
      "loss": 9.1117,
      "step": 167
    },
    {
      "epoch": 0.02168021680216802,
      "grad_norm": 2.5191667079925537,
      "learning_rate": 6.837175952121306e-06,
      "loss": 8.921,
      "step": 168
    },
    {
      "epoch": 0.021809265711704735,
      "grad_norm": 2.4196970462799072,
      "learning_rate": 6.425787818636131e-06,
      "loss": 9.2911,
      "step": 169
    },
    {
      "epoch": 0.02193831462124145,
      "grad_norm": 2.474123239517212,
      "learning_rate": 6.026312439675552e-06,
      "loss": 9.4719,
      "step": 170
    },
    {
      "epoch": 0.022067363530778164,
      "grad_norm": 2.596231698989868,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 9.2503,
      "step": 171
    },
    {
      "epoch": 0.02219641244031488,
      "grad_norm": 2.5076022148132324,
      "learning_rate": 5.263533508961827e-06,
      "loss": 9.5474,
      "step": 172
    },
    {
      "epoch": 0.022325461349851593,
      "grad_norm": 2.6695826053619385,
      "learning_rate": 4.900438493352055e-06,
      "loss": 9.5692,
      "step": 173
    },
    {
      "epoch": 0.02245451025938831,
      "grad_norm": 2.6091346740722656,
      "learning_rate": 4.549673247541875e-06,
      "loss": 8.7727,
      "step": 174
    },
    {
      "epoch": 0.022583559168925023,
      "grad_norm": 2.7914795875549316,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 9.0877,
      "step": 175
    },
    {
      "epoch": 0.022712608078461736,
      "grad_norm": 2.934940814971924,
      "learning_rate": 3.885512251130763e-06,
      "loss": 9.4135,
      "step": 176
    },
    {
      "epoch": 0.022841656987998452,
      "grad_norm": 2.9756340980529785,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 9.1306,
      "step": 177
    },
    {
      "epoch": 0.022970705897535165,
      "grad_norm": 2.8788416385650635,
      "learning_rate": 3.271776770026963e-06,
      "loss": 9.5178,
      "step": 178
    },
    {
      "epoch": 0.02309975480707188,
      "grad_norm": 2.7710723876953125,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 9.208,
      "step": 179
    },
    {
      "epoch": 0.023228803716608595,
      "grad_norm": 2.812511920928955,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 9.2304,
      "step": 180
    },
    {
      "epoch": 0.023357852626145308,
      "grad_norm": 3.071134328842163,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 9.5987,
      "step": 181
    },
    {
      "epoch": 0.023486901535682024,
      "grad_norm": 2.9297544956207275,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 9.1029,
      "step": 182
    },
    {
      "epoch": 0.023615950445218737,
      "grad_norm": 3.159698724746704,
      "learning_rate": 1.962316193157593e-06,
      "loss": 9.5165,
      "step": 183
    },
    {
      "epoch": 0.023744999354755453,
      "grad_norm": 3.182583808898926,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 8.9895,
      "step": 184
    },
    {
      "epoch": 0.023874048264292166,
      "grad_norm": 3.428360939025879,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 9.2093,
      "step": 185
    },
    {
      "epoch": 0.024003097173828883,
      "grad_norm": 8.172533988952637,
      "learning_rate": 1.333670137599713e-06,
      "loss": 9.9388,
      "step": 186
    },
    {
      "epoch": 0.024132146083365596,
      "grad_norm": 3.192690849304199,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 8.8376,
      "step": 187
    },
    {
      "epoch": 0.02426119499290231,
      "grad_norm": 3.5312483310699463,
      "learning_rate": 9.810017062595322e-07,
      "loss": 9.3668,
      "step": 188
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 3.6525990962982178,
      "learning_rate": 8.247462563808817e-07,
      "loss": 9.127,
      "step": 189
    },
    {
      "epoch": 0.024519292811975738,
      "grad_norm": 4.213011264801025,
      "learning_rate": 6.819348298638839e-07,
      "loss": 9.5994,
      "step": 190
    },
    {
      "epoch": 0.024648341721512455,
      "grad_norm": 3.7822132110595703,
      "learning_rate": 5.526064699265753e-07,
      "loss": 9.2404,
      "step": 191
    },
    {
      "epoch": 0.024777390631049168,
      "grad_norm": 3.9459145069122314,
      "learning_rate": 4.367965336512403e-07,
      "loss": 9.3512,
      "step": 192
    },
    {
      "epoch": 0.02490643954058588,
      "grad_norm": 4.3947343826293945,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.1752,
      "step": 193
    },
    {
      "epoch": 0.025035488450122597,
      "grad_norm": 4.368558883666992,
      "learning_rate": 2.458548727494292e-07,
      "loss": 9.8028,
      "step": 194
    },
    {
      "epoch": 0.02516453735965931,
      "grad_norm": 5.303836822509766,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.5595,
      "step": 195
    },
    {
      "epoch": 0.025293586269196026,
      "grad_norm": 5.162618160247803,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 9.8166,
      "step": 196
    },
    {
      "epoch": 0.02542263517873274,
      "grad_norm": 5.16475248336792,
      "learning_rate": 6.150154258476315e-08,
      "loss": 9.6008,
      "step": 197
    },
    {
      "epoch": 0.025551684088269456,
      "grad_norm": 5.391692161560059,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.0645,
      "step": 198
    },
    {
      "epoch": 0.02568073299780617,
      "grad_norm": 6.20759391784668,
      "learning_rate": 6.834750376549792e-09,
      "loss": 10.3312,
      "step": 199
    },
    {
      "epoch": 0.025809781907342882,
      "grad_norm": 7.109989166259766,
      "learning_rate": 0.0,
      "loss": 9.9628,
      "step": 200
    },
    {
      "epoch": 0.025809781907342882,
      "eval_loss": 2.323329210281372,
      "eval_runtime": 1421.2234,
      "eval_samples_per_second": 9.183,
      "eval_steps_per_second": 2.296,
      "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": 4.665029231326003e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}